summaryrefslogtreecommitdiffstats
path: root/src
diff options
context:
space:
mode:
Diffstat (limited to 'src')
-rw-r--r--src/audio_core/audio_renderer.cpp43
-rw-r--r--src/audio_core/audio_renderer.h15
-rw-r--r--src/audio_core/stream.cpp2
-rw-r--r--src/audio_core/stream.h25
-rw-r--r--src/common/assert.h5
-rw-r--r--src/common/common_funcs.h12
-rw-r--r--src/common/hash.h35
-rw-r--r--src/common/logging/backend.cpp8
-rw-r--r--src/common/threadsafe_queue.h11
-rw-r--r--src/core/CMakeLists.txt19
-rw-r--r--src/core/arm/arm_interface.cpp27
-rw-r--r--src/core/arm/arm_interface.h8
-rw-r--r--src/core/arm/dynarmic/arm_dynarmic.cpp50
-rw-r--r--src/core/arm/dynarmic/arm_dynarmic.h8
-rw-r--r--src/core/arm/unicorn/arm_unicorn.cpp16
-rw-r--r--src/core/arm/unicorn/arm_unicorn.h3
-rw-r--r--src/core/core.cpp26
-rw-r--r--src/core/core.h15
-rw-r--r--src/core/core_cpu.cpp7
-rw-r--r--src/core/core_cpu.h18
-rw-r--r--src/core/core_timing.cpp53
-rw-r--r--src/core/core_timing.h45
-rw-r--r--src/core/cpu_core_manager.cpp2
-rw-r--r--src/core/crypto/key_manager.cpp44
-rw-r--r--src/core/crypto/partition_data_manager.cpp11
-rw-r--r--src/core/file_sys/directory.h2
-rw-r--r--src/core/file_sys/kernel_executable.cpp2
-rw-r--r--src/core/file_sys/patch_manager.cpp16
-rw-r--r--src/core/file_sys/program_metadata.cpp6
-rw-r--r--src/core/file_sys/program_metadata.h4
-rw-r--r--src/core/file_sys/registered_cache.cpp10
-rw-r--r--src/core/file_sys/romfs.cpp7
-rw-r--r--src/core/file_sys/romfs.h4
-rw-r--r--src/core/file_sys/romfs_factory.cpp4
-rw-r--r--src/core/file_sys/savedata_factory.cpp4
-rw-r--r--src/core/file_sys/vfs_libzip.cpp2
-rw-r--r--src/core/file_sys/xts_archive.cpp17
-rw-r--r--src/core/gdbstub/gdbstub.cpp39
-rw-r--r--src/core/hardware_interrupt_manager.cpp13
-rw-r--r--src/core/hardware_interrupt_manager.h4
-rw-r--r--src/core/hle/ipc_helpers.h22
-rw-r--r--src/core/hle/kernel/address_arbiter.cpp110
-rw-r--r--src/core/hle/kernel/address_arbiter.h21
-rw-r--r--src/core/hle/kernel/client_port.cpp18
-rw-r--r--src/core/hle/kernel/client_port.h14
-rw-r--r--src/core/hle/kernel/client_session.cpp41
-rw-r--r--src/core/hle/kernel/client_session.h29
-rw-r--r--src/core/hle/kernel/handle_table.cpp10
-rw-r--r--src/core/hle/kernel/handle_table.h10
-rw-r--r--src/core/hle/kernel/hle_ipc.cpp43
-rw-r--r--src/core/hle/kernel/hle_ipc.h51
-rw-r--r--src/core/hle/kernel/kernel.cpp62
-rw-r--r--src/core/hle/kernel/kernel.h16
-rw-r--r--src/core/hle/kernel/mutex.cpp29
-rw-r--r--src/core/hle/kernel/object.cpp1
-rw-r--r--src/core/hle/kernel/object.h31
-rw-r--r--src/core/hle/kernel/process.cpp55
-rw-r--r--src/core/hle/kernel/process.h26
-rw-r--r--src/core/hle/kernel/resource_limit.cpp4
-rw-r--r--src/core/hle/kernel/resource_limit.h10
-rw-r--r--src/core/hle/kernel/scheduler.cpp57
-rw-r--r--src/core/hle/kernel/scheduler.h53
-rw-r--r--src/core/hle/kernel/server_port.cpp8
-rw-r--r--src/core/hle/kernel/server_port.h14
-rw-r--r--src/core/hle/kernel/server_session.cpp118
-rw-r--r--src/core/hle/kernel/server_session.h66
-rw-r--r--src/core/hle/kernel/session.cpp32
-rw-r--r--src/core/hle/kernel/session.h57
-rw-r--r--src/core/hle/kernel/shared_memory.cpp15
-rw-r--r--src/core/hle/kernel/shared_memory.h20
-rw-r--r--src/core/hle/kernel/svc.cpp164
-rw-r--r--src/core/hle/kernel/svc_wrap.h21
-rw-r--r--src/core/hle/kernel/thread.cpp102
-rw-r--r--src/core/hle/kernel/thread.h48
-rw-r--r--src/core/hle/kernel/transfer_memory.cpp6
-rw-r--r--src/core/hle/kernel/transfer_memory.h11
-rw-r--r--src/core/hle/kernel/vm_manager.cpp15
-rw-r--r--src/core/hle/kernel/wait_object.cpp17
-rw-r--r--src/core/hle/kernel/wait_object.h19
-rw-r--r--src/core/hle/kernel/writable_event.cpp6
-rw-r--r--src/core/hle/kernel/writable_event.h10
-rw-r--r--src/core/hle/result.h10
-rw-r--r--src/core/hle/service/acc/acc.cpp6
-rw-r--r--src/core/hle/service/acc/acc_su.cpp3
-rw-r--r--src/core/hle/service/acc/acc_u1.cpp1
-rw-r--r--src/core/hle/service/acc/profile_manager.cpp4
-rw-r--r--src/core/hle/service/am/am.cpp73
-rw-r--r--src/core/hle/service/am/am.h6
-rw-r--r--src/core/hle/service/am/applets/applets.cpp6
-rw-r--r--src/core/hle/service/am/applets/applets.h6
-rw-r--r--src/core/hle/service/am/applets/web_browser.cpp16
-rw-r--r--src/core/hle/service/am/idle.cpp2
-rw-r--r--src/core/hle/service/am/omm.cpp2
-rw-r--r--src/core/hle/service/aoc/aoc_u.cpp3
-rw-r--r--src/core/hle/service/audio/audctl.cpp1
-rw-r--r--src/core/hle/service/audio/audout_u.cpp6
-rw-r--r--src/core/hle/service/audio/audren_u.cpp5
-rw-r--r--src/core/hle/service/audio/hwopus.cpp4
-rw-r--r--src/core/hle/service/bcat/backend/backend.cpp2
-rw-r--r--src/core/hle/service/bcat/backend/backend.h2
-rw-r--r--src/core/hle/service/bcat/backend/boxcat.cpp4
-rw-r--r--src/core/hle/service/bcat/module.cpp12
-rw-r--r--src/core/hle/service/btdrv/btdrv.cpp1
-rw-r--r--src/core/hle/service/erpt/erpt.cpp3
-rw-r--r--src/core/hle/service/es/es.cpp2
-rw-r--r--src/core/hle/service/filesystem/filesystem.cpp26
-rw-r--r--src/core/hle/service/filesystem/fsp_srv.cpp23
-rw-r--r--src/core/hle/service/friend/friend.cpp5
-rw-r--r--src/core/hle/service/hid/controllers/npad.cpp3
-rw-r--r--src/core/hle/service/hid/controllers/npad.h2
-rw-r--r--src/core/hle/service/hid/hid.cpp29
-rw-r--r--src/core/hle/service/hid/hid.h4
-rw-r--r--src/core/hle/service/hid/irs.h2
-rw-r--r--src/core/hle/service/ldr/ldr.cpp7
-rw-r--r--src/core/hle/service/lm/lm.cpp23
-rw-r--r--src/core/hle/service/mii/mii.cpp8
-rw-r--r--src/core/hle/service/ncm/ncm.cpp6
-rw-r--r--src/core/hle/service/nfc/nfc.cpp1
-rw-r--r--src/core/hle/service/nfp/nfp.cpp7
-rw-r--r--src/core/hle/service/nfp/nfp.h2
-rw-r--r--src/core/hle/service/nifm/nifm.cpp8
-rw-r--r--src/core/hle/service/nim/nim.cpp2
-rw-r--r--src/core/hle/service/npns/npns.cpp7
-rw-r--r--src/core/hle/service/ns/ns.cpp31
-rw-r--r--src/core/hle/service/ns/pl_u.cpp12
-rw-r--r--src/core/hle/service/nvdrv/devices/nvhost_as_gpu.cpp6
-rw-r--r--src/core/hle/service/nvdrv/devices/nvhost_as_gpu.h2
-rw-r--r--src/core/hle/service/nvdrv/devices/nvhost_gpu.cpp4
-rw-r--r--src/core/hle/service/nvdrv/interface.cpp2
-rw-r--r--src/core/hle/service/nvdrv/nvdrv.cpp4
-rw-r--r--src/core/hle/service/nvdrv/nvdrv.h4
-rw-r--r--src/core/hle/service/nvflinger/buffer_queue.cpp4
-rw-r--r--src/core/hle/service/nvflinger/buffer_queue.h4
-rw-r--r--src/core/hle/service/nvflinger/nvflinger.cpp6
-rw-r--r--src/core/hle/service/nvflinger/nvflinger.h4
-rw-r--r--src/core/hle/service/pm/pm.cpp14
-rw-r--r--src/core/hle/service/prepo/prepo.cpp1
-rw-r--r--src/core/hle/service/service.cpp6
-rw-r--r--src/core/hle/service/service.h2
-rw-r--r--src/core/hle/service/set/set.cpp1
-rw-r--r--src/core/hle/service/set/set_cal.cpp11
-rw-r--r--src/core/hle/service/set/set_fd.cpp4
-rw-r--r--src/core/hle/service/set/set_sys.cpp11
-rw-r--r--src/core/hle/service/sm/controller.cpp5
-rw-r--r--src/core/hle/service/sm/sm.cpp39
-rw-r--r--src/core/hle/service/sm/sm.h16
-rw-r--r--src/core/hle/service/sockets/nsd.cpp5
-rw-r--r--src/core/hle/service/ssl/ssl.cpp5
-rw-r--r--src/core/hle/service/time/interface.cpp1
-rw-r--r--src/core/hle/service/time/time.cpp39
-rw-r--r--src/core/hle/service/time/time_sharedmemory.cpp2
-rw-r--r--src/core/hle/service/time/time_sharedmemory.h4
-rw-r--r--src/core/hle/service/vi/display/vi_display.cpp2
-rw-r--r--src/core/hle/service/vi/display/vi_display.h2
-rw-r--r--src/core/hle/service/vi/vi.cpp10
-rw-r--r--src/core/memory.cpp1008
-rw-r--r--src/core/memory.h396
-rw-r--r--src/core/memory/cheat_engine.cpp9
-rw-r--r--src/core/memory/cheat_engine.h7
-rw-r--r--src/core/memory_setup.h43
-rw-r--r--src/core/perf_stats.cpp2
-rw-r--r--src/core/reporter.cpp16
-rw-r--r--src/core/telemetry_session.cpp28
-rw-r--r--src/core/tools/freezer.cpp32
-rw-r--r--src/core/tools/freezer.h10
-rw-r--r--src/tests/core/arm/arm_test_common.cpp15
-rw-r--r--src/tests/core/core_timing.cpp84
-rw-r--r--src/video_core/CMakeLists.txt17
-rw-r--r--src/video_core/buffer_cache/buffer_cache.h24
-rw-r--r--src/video_core/debug_utils/debug_utils.cpp49
-rw-r--r--src/video_core/debug_utils/debug_utils.h157
-rw-r--r--src/video_core/engines/const_buffer_engine_interface.h10
-rw-r--r--src/video_core/engines/kepler_compute.cpp1
-rw-r--r--src/video_core/engines/kepler_compute.h10
-rw-r--r--src/video_core/engines/maxwell_3d.cpp81
-rw-r--r--src/video_core/engines/maxwell_3d.h87
-rw-r--r--src/video_core/engines/shader_bytecode.h106
-rw-r--r--src/video_core/engines/shader_type.h21
-rw-r--r--src/video_core/gpu_thread.cpp35
-rw-r--r--src/video_core/memory_manager.cpp4
-rw-r--r--src/video_core/rasterizer_accelerated.cpp8
-rw-r--r--src/video_core/rasterizer_accelerated.h9
-rw-r--r--src/video_core/renderer_opengl/gl_device.cpp141
-rw-r--r--src/video_core/renderer_opengl/gl_device.h32
-rw-r--r--src/video_core/renderer_opengl/gl_framebuffer_cache.cpp57
-rw-r--r--src/video_core/renderer_opengl/gl_framebuffer_cache.h19
-rw-r--r--src/video_core/renderer_opengl/gl_rasterizer.cpp353
-rw-r--r--src/video_core/renderer_opengl/gl_rasterizer.h32
-rw-r--r--src/video_core/renderer_opengl/gl_shader_cache.cpp368
-rw-r--r--src/video_core/renderer_opengl/gl_shader_cache.h19
-rw-r--r--src/video_core/renderer_opengl/gl_shader_decompiler.cpp405
-rw-r--r--src/video_core/renderer_opengl/gl_shader_decompiler.h17
-rw-r--r--src/video_core/renderer_opengl/gl_shader_disk_cache.cpp78
-rw-r--r--src/video_core/renderer_opengl/gl_shader_disk_cache.h102
-rw-r--r--src/video_core/renderer_opengl/gl_shader_gen.cpp71
-rw-r--r--src/video_core/renderer_opengl/gl_shader_manager.cpp20
-rw-r--r--src/video_core/renderer_opengl/gl_shader_manager.h19
-rw-r--r--src/video_core/renderer_opengl/gl_state.cpp31
-rw-r--r--src/video_core/renderer_opengl/gl_state.h18
-rw-r--r--src/video_core/renderer_opengl/gl_texture_cache.cpp186
-rw-r--r--src/video_core/renderer_opengl/maxwell_to_gl.h22
-rw-r--r--src/video_core/renderer_opengl/renderer_opengl.cpp232
-rw-r--r--src/video_core/renderer_opengl/renderer_opengl.h22
-rw-r--r--src/video_core/renderer_opengl/utils.cpp32
-rw-r--r--src/video_core/renderer_opengl/utils.h18
-rw-r--r--src/video_core/renderer_vulkan/declarations.h13
-rw-r--r--src/video_core/renderer_vulkan/fixed_pipeline_state.cpp296
-rw-r--r--src/video_core/renderer_vulkan/fixed_pipeline_state.h282
-rw-r--r--src/video_core/renderer_vulkan/maxwell_to_vk.cpp351
-rw-r--r--src/video_core/renderer_vulkan/maxwell_to_vk.h21
-rw-r--r--src/video_core/renderer_vulkan/shaders/blit.frag24
-rw-r--r--src/video_core/renderer_vulkan/shaders/blit.vert28
-rw-r--r--src/video_core/renderer_vulkan/shaders/quad_array.comp37
-rw-r--r--src/video_core/renderer_vulkan/shaders/uint8.comp33
-rw-r--r--src/video_core/renderer_vulkan/vk_buffer_cache.cpp10
-rw-r--r--src/video_core/renderer_vulkan/vk_buffer_cache.h7
-rw-r--r--src/video_core/renderer_vulkan/vk_device.cpp324
-rw-r--r--src/video_core/renderer_vulkan/vk_device.h103
-rw-r--r--src/video_core/renderer_vulkan/vk_image.cpp106
-rw-r--r--src/video_core/renderer_vulkan/vk_image.h84
-rw-r--r--src/video_core/renderer_vulkan/vk_resource_manager.cpp36
-rw-r--r--src/video_core/renderer_vulkan/vk_resource_manager.h16
-rw-r--r--src/video_core/renderer_vulkan/vk_sampler_cache.cpp7
-rw-r--r--src/video_core/renderer_vulkan/vk_scheduler.cpp150
-rw-r--r--src/video_core/renderer_vulkan/vk_scheduler.h198
-rw-r--r--src/video_core/renderer_vulkan/vk_shader_decompiler.cpp2376
-rw-r--r--src/video_core/renderer_vulkan/vk_shader_decompiler.h78
-rw-r--r--src/video_core/renderer_vulkan/vk_staging_buffer_pool.cpp127
-rw-r--r--src/video_core/renderer_vulkan/vk_staging_buffer_pool.h83
-rw-r--r--src/video_core/renderer_vulkan/vk_swapchain.cpp45
-rw-r--r--src/video_core/renderer_vulkan/vk_swapchain.h10
-rw-r--r--src/video_core/shader/const_buffer_locker.cpp11
-rw-r--r--src/video_core/shader/const_buffer_locker.h4
-rw-r--r--src/video_core/shader/control_flow.cpp214
-rw-r--r--src/video_core/shader/decode.cpp4
-rw-r--r--src/video_core/shader/decode/arithmetic.cpp11
-rw-r--r--src/video_core/shader/decode/arithmetic_half.cpp4
-rw-r--r--src/video_core/shader/decode/arithmetic_half_immediate.cpp8
-rw-r--r--src/video_core/shader/decode/arithmetic_integer.cpp19
-rw-r--r--src/video_core/shader/decode/conversion.cpp15
-rw-r--r--src/video_core/shader/decode/ffma.cpp4
-rw-r--r--src/video_core/shader/decode/half_set.cpp4
-rw-r--r--src/video_core/shader/decode/half_set_predicate.cpp4
-rw-r--r--src/video_core/shader/decode/memory.cpp72
-rw-r--r--src/video_core/shader/decode/other.cpp10
-rw-r--r--src/video_core/shader/decode/register_set_predicate.cpp60
-rw-r--r--src/video_core/shader/decode/texture.cpp365
-rw-r--r--src/video_core/shader/decode/warp.cpp82
-rw-r--r--src/video_core/shader/node.h78
-rw-r--r--src/video_core/shader/shader_ir.cpp6
-rw-r--r--src/video_core/shader/shader_ir.h41
-rw-r--r--src/video_core/shader/track.cpp1
-rw-r--r--src/video_core/surface.cpp319
-rw-r--r--src/video_core/surface.h20
-rw-r--r--src/video_core/texture_cache/format_lookup_table.cpp208
-rw-r--r--src/video_core/texture_cache/format_lookup_table.h51
-rw-r--r--src/video_core/texture_cache/surface_base.h11
-rw-r--r--src/video_core/texture_cache/surface_params.cpp80
-rw-r--r--src/video_core/texture_cache/surface_params.h31
-rw-r--r--src/video_core/texture_cache/texture_cache.h220
-rw-r--r--src/video_core/textures/astc.cpp73
-rw-r--r--src/video_core/textures/decoders.h4
-rw-r--r--src/video_core/textures/texture.h7
-rw-r--r--src/video_core/video_core.cpp2
-rw-r--r--src/web_service/web_backend.cpp52
-rw-r--r--src/web_service/web_backend.h23
-rw-r--r--src/yuzu/CMakeLists.txt5
-rw-r--r--src/yuzu/configuration/configure_web.cpp73
-rw-r--r--src/yuzu/configuration/configure_web.ui12
-rw-r--r--src/yuzu/debugger/graphics/graphics_breakpoint_observer.cpp27
-rw-r--r--src/yuzu/debugger/graphics/graphics_breakpoint_observer.h33
-rw-r--r--src/yuzu/debugger/graphics/graphics_breakpoints.cpp221
-rw-r--r--src/yuzu/debugger/graphics/graphics_breakpoints.h45
-rw-r--r--src/yuzu/debugger/graphics/graphics_breakpoints_p.h37
-rw-r--r--src/yuzu/debugger/wait_tree.cpp15
-rw-r--r--src/yuzu/debugger/wait_tree.h11
-rw-r--r--src/yuzu/main.cpp14
-rw-r--r--src/yuzu/main.h8
-rw-r--r--src/yuzu/main.ui13
-rw-r--r--src/yuzu_cmd/emu_window/emu_window_sdl2_gl.cpp8
280 files changed, 9225 insertions, 5535 deletions
diff --git a/src/audio_core/audio_renderer.cpp b/src/audio_core/audio_renderer.cpp
index e6f38d600..c187d8ac5 100644
--- a/src/audio_core/audio_renderer.cpp
+++ b/src/audio_core/audio_renderer.cpp
@@ -36,9 +36,9 @@ public:
}
void SetWaveIndex(std::size_t index);
- std::vector<s16> DequeueSamples(std::size_t sample_count);
+ std::vector<s16> DequeueSamples(std::size_t sample_count, Memory::Memory& memory);
void UpdateState();
- void RefreshBuffer();
+ void RefreshBuffer(Memory::Memory& memory);
private:
bool is_in_use{};
@@ -66,17 +66,18 @@ public:
return info;
}
- void UpdateState();
+ void UpdateState(Memory::Memory& memory);
private:
EffectOutStatus out_status{};
EffectInStatus info{};
};
-AudioRenderer::AudioRenderer(Core::Timing::CoreTiming& core_timing, AudioRendererParameter params,
- Kernel::SharedPtr<Kernel::WritableEvent> buffer_event,
+AudioRenderer::AudioRenderer(Core::Timing::CoreTiming& core_timing, Memory::Memory& memory_,
+ AudioRendererParameter params,
+ std::shared_ptr<Kernel::WritableEvent> buffer_event,
std::size_t instance_number)
: worker_params{params}, buffer_event{buffer_event}, voices(params.voice_count),
- effects(params.effect_count) {
+ effects(params.effect_count), memory{memory_} {
audio_out = std::make_unique<AudioCore::AudioOut>();
stream = audio_out->OpenStream(core_timing, STREAM_SAMPLE_RATE, STREAM_NUM_CHANNELS,
@@ -162,7 +163,7 @@ std::vector<u8> AudioRenderer::UpdateAudioRenderer(const std::vector<u8>& input_
}
for (auto& effect : effects) {
- effect.UpdateState();
+ effect.UpdateState(memory);
}
// Release previous buffers and queue next ones for playback
@@ -206,13 +207,14 @@ void AudioRenderer::VoiceState::SetWaveIndex(std::size_t index) {
is_refresh_pending = true;
}
-std::vector<s16> AudioRenderer::VoiceState::DequeueSamples(std::size_t sample_count) {
+std::vector<s16> AudioRenderer::VoiceState::DequeueSamples(std::size_t sample_count,
+ Memory::Memory& memory) {
if (!IsPlaying()) {
return {};
}
if (is_refresh_pending) {
- RefreshBuffer();
+ RefreshBuffer(memory);
}
const std::size_t max_size{samples.size() - offset};
@@ -256,10 +258,11 @@ void AudioRenderer::VoiceState::UpdateState() {
is_in_use = info.is_in_use;
}
-void AudioRenderer::VoiceState::RefreshBuffer() {
- std::vector<s16> new_samples(info.wave_buffer[wave_index].buffer_sz / sizeof(s16));
- Memory::ReadBlock(info.wave_buffer[wave_index].buffer_addr, new_samples.data(),
- info.wave_buffer[wave_index].buffer_sz);
+void AudioRenderer::VoiceState::RefreshBuffer(Memory::Memory& memory) {
+ const auto wave_buffer_address = info.wave_buffer[wave_index].buffer_addr;
+ const auto wave_buffer_size = info.wave_buffer[wave_index].buffer_sz;
+ std::vector<s16> new_samples(wave_buffer_size / sizeof(s16));
+ memory.ReadBlock(wave_buffer_address, new_samples.data(), wave_buffer_size);
switch (static_cast<Codec::PcmFormat>(info.sample_format)) {
case Codec::PcmFormat::Int16: {
@@ -269,7 +272,7 @@ void AudioRenderer::VoiceState::RefreshBuffer() {
case Codec::PcmFormat::Adpcm: {
// Decode ADPCM to PCM16
Codec::ADPCM_Coeff coeffs;
- Memory::ReadBlock(info.additional_params_addr, coeffs.data(), sizeof(Codec::ADPCM_Coeff));
+ memory.ReadBlock(info.additional_params_addr, coeffs.data(), sizeof(Codec::ADPCM_Coeff));
new_samples = Codec::DecodeADPCM(reinterpret_cast<u8*>(new_samples.data()),
new_samples.size() * sizeof(s16), coeffs, adpcm_state);
break;
@@ -307,18 +310,18 @@ void AudioRenderer::VoiceState::RefreshBuffer() {
is_refresh_pending = false;
}
-void AudioRenderer::EffectState::UpdateState() {
+void AudioRenderer::EffectState::UpdateState(Memory::Memory& memory) {
if (info.is_new) {
out_status.state = EffectStatus::New;
} else {
if (info.type == Effect::Aux) {
- ASSERT_MSG(Memory::Read32(info.aux_info.return_buffer_info) == 0,
+ ASSERT_MSG(memory.Read32(info.aux_info.return_buffer_info) == 0,
"Aux buffers tried to update");
- ASSERT_MSG(Memory::Read32(info.aux_info.send_buffer_info) == 0,
+ ASSERT_MSG(memory.Read32(info.aux_info.send_buffer_info) == 0,
"Aux buffers tried to update");
- ASSERT_MSG(Memory::Read32(info.aux_info.return_buffer_base) == 0,
+ ASSERT_MSG(memory.Read32(info.aux_info.return_buffer_base) == 0,
"Aux buffers tried to update");
- ASSERT_MSG(Memory::Read32(info.aux_info.send_buffer_base) == 0,
+ ASSERT_MSG(memory.Read32(info.aux_info.send_buffer_base) == 0,
"Aux buffers tried to update");
}
}
@@ -340,7 +343,7 @@ void AudioRenderer::QueueMixedBuffer(Buffer::Tag tag) {
std::size_t offset{};
s64 samples_remaining{BUFFER_SIZE};
while (samples_remaining > 0) {
- const std::vector<s16> samples{voice.DequeueSamples(samples_remaining)};
+ const std::vector<s16> samples{voice.DequeueSamples(samples_remaining, memory)};
if (samples.empty()) {
break;
diff --git a/src/audio_core/audio_renderer.h b/src/audio_core/audio_renderer.h
index 4f14b91cd..be1b019f1 100644
--- a/src/audio_core/audio_renderer.h
+++ b/src/audio_core/audio_renderer.h
@@ -22,6 +22,10 @@ namespace Kernel {
class WritableEvent;
}
+namespace Memory {
+class Memory;
+}
+
namespace AudioCore {
class AudioOut;
@@ -217,9 +221,9 @@ static_assert(sizeof(UpdateDataHeader) == 0x40, "UpdateDataHeader has wrong size
class AudioRenderer {
public:
- AudioRenderer(Core::Timing::CoreTiming& core_timing, AudioRendererParameter params,
- Kernel::SharedPtr<Kernel::WritableEvent> buffer_event,
- std::size_t instance_number);
+ AudioRenderer(Core::Timing::CoreTiming& core_timing, Memory::Memory& memory_,
+ AudioRendererParameter params,
+ std::shared_ptr<Kernel::WritableEvent> buffer_event, std::size_t instance_number);
~AudioRenderer();
std::vector<u8> UpdateAudioRenderer(const std::vector<u8>& input_params);
@@ -235,11 +239,12 @@ private:
class VoiceState;
AudioRendererParameter worker_params;
- Kernel::SharedPtr<Kernel::WritableEvent> buffer_event;
+ std::shared_ptr<Kernel::WritableEvent> buffer_event;
std::vector<VoiceState> voices;
std::vector<EffectState> effects;
std::unique_ptr<AudioOut> audio_out;
- AudioCore::StreamPtr stream;
+ StreamPtr stream;
+ Memory::Memory& memory;
};
} // namespace AudioCore
diff --git a/src/audio_core/stream.cpp b/src/audio_core/stream.cpp
index 6a5f53a57..4ca98f8ea 100644
--- a/src/audio_core/stream.cpp
+++ b/src/audio_core/stream.cpp
@@ -37,7 +37,7 @@ Stream::Stream(Core::Timing::CoreTiming& core_timing, u32 sample_rate, Format fo
: sample_rate{sample_rate}, format{format}, release_callback{std::move(release_callback)},
sink_stream{sink_stream}, core_timing{core_timing}, name{std::move(name_)} {
- release_event = core_timing.RegisterEvent(
+ release_event = Core::Timing::CreateEvent(
name, [this](u64 userdata, s64 cycles_late) { ReleaseActiveBuffer(); });
}
diff --git a/src/audio_core/stream.h b/src/audio_core/stream.h
index 8106cea43..1708a4d98 100644
--- a/src/audio_core/stream.h
+++ b/src/audio_core/stream.h
@@ -98,18 +98,19 @@ private:
/// Gets the number of core cycles when the specified buffer will be released
s64 GetBufferReleaseCycles(const Buffer& buffer) const;
- u32 sample_rate; ///< Sample rate of the stream
- Format format; ///< Format of the stream
- float game_volume = 1.0f; ///< The volume the game currently has set
- ReleaseCallback release_callback; ///< Buffer release callback for the stream
- State state{State::Stopped}; ///< Playback state of the stream
- Core::Timing::EventType* release_event{}; ///< Core timing release event for the stream
- BufferPtr active_buffer; ///< Actively playing buffer in the stream
- std::queue<BufferPtr> queued_buffers; ///< Buffers queued to be played in the stream
- std::queue<BufferPtr> released_buffers; ///< Buffers recently released from the stream
- SinkStream& sink_stream; ///< Output sink for the stream
- Core::Timing::CoreTiming& core_timing; ///< Core timing instance.
- std::string name; ///< Name of the stream, must be unique
+ u32 sample_rate; ///< Sample rate of the stream
+ Format format; ///< Format of the stream
+ float game_volume = 1.0f; ///< The volume the game currently has set
+ ReleaseCallback release_callback; ///< Buffer release callback for the stream
+ State state{State::Stopped}; ///< Playback state of the stream
+ std::shared_ptr<Core::Timing::EventType>
+ release_event; ///< Core timing release event for the stream
+ BufferPtr active_buffer; ///< Actively playing buffer in the stream
+ std::queue<BufferPtr> queued_buffers; ///< Buffers queued to be played in the stream
+ std::queue<BufferPtr> released_buffers; ///< Buffers recently released from the stream
+ SinkStream& sink_stream; ///< Output sink for the stream
+ Core::Timing::CoreTiming& core_timing; ///< Core timing instance.
+ std::string name; ///< Name of the stream, must be unique
};
using StreamPtr = std::shared_ptr<Stream>;
diff --git a/src/common/assert.h b/src/common/assert.h
index 4b0e3f64e..5b67c5c52 100644
--- a/src/common/assert.h
+++ b/src/common/assert.h
@@ -41,8 +41,9 @@ __declspec(noinline, noreturn)
} \
while (0)
-#define UNREACHABLE() ASSERT_MSG(false, "Unreachable code!")
-#define UNREACHABLE_MSG(...) ASSERT_MSG(false, __VA_ARGS__)
+#define UNREACHABLE() assert_noinline_call([] { LOG_CRITICAL(Debug, "Unreachable code!"); })
+#define UNREACHABLE_MSG(...) \
+ assert_noinline_call([&] { LOG_CRITICAL(Debug, "Unreachable code!\n" __VA_ARGS__); })
#ifdef _DEBUG
#define DEBUG_ASSERT(_a_) ASSERT(_a_)
diff --git a/src/common/common_funcs.h b/src/common/common_funcs.h
index c029dc7b3..052254678 100644
--- a/src/common/common_funcs.h
+++ b/src/common/common_funcs.h
@@ -19,13 +19,15 @@
/// Helper macros to insert unused bytes or words to properly align structs. These values will be
/// zero-initialized.
-#define INSERT_PADDING_BYTES(num_bytes) std::array<u8, num_bytes> CONCAT2(pad, __LINE__){};
-#define INSERT_PADDING_WORDS(num_words) std::array<u32, num_words> CONCAT2(pad, __LINE__){};
+#define INSERT_PADDING_BYTES(num_bytes) \
+ std::array<u8, num_bytes> CONCAT2(pad, __LINE__) {}
+#define INSERT_PADDING_WORDS(num_words) \
+ std::array<u32, num_words> CONCAT2(pad, __LINE__) {}
/// These are similar to the INSERT_PADDING_* macros, but are needed for padding unions. This is
/// because unions can only be initialized by one member.
-#define INSERT_UNION_PADDING_BYTES(num_bytes) std::array<u8, num_bytes> CONCAT2(pad, __LINE__);
-#define INSERT_UNION_PADDING_WORDS(num_words) std::array<u32, num_words> CONCAT2(pad, __LINE__);
+#define INSERT_UNION_PADDING_BYTES(num_bytes) std::array<u8, num_bytes> CONCAT2(pad, __LINE__)
+#define INSERT_UNION_PADDING_WORDS(num_words) std::array<u32, num_words> CONCAT2(pad, __LINE__)
#ifndef _MSC_VER
@@ -56,7 +58,7 @@ std::string GetLastErrorMsg();
namespace Common {
constexpr u32 MakeMagic(char a, char b, char c, char d) {
- return a | b << 8 | c << 16 | d << 24;
+ return u32(a) | u32(b) << 8 | u32(c) << 16 | u32(d) << 24;
}
} // namespace Common
diff --git a/src/common/hash.h b/src/common/hash.h
index ebd4125e2..b2538f3ea 100644
--- a/src/common/hash.h
+++ b/src/common/hash.h
@@ -35,41 +35,6 @@ static inline u64 ComputeStructHash64(const T& data) {
return ComputeHash64(&data, sizeof(data));
}
-/// A helper template that ensures the padding in a struct is initialized by memsetting to 0.
-template <typename T>
-struct HashableStruct {
- // In addition to being trivially copyable, T must also have a trivial default constructor,
- // because any member initialization would be overridden by memset
- static_assert(std::is_trivial_v<T>, "Type passed to HashableStruct must be trivial");
- /*
- * We use a union because "implicitly-defined copy/move constructor for a union X copies the
- * object representation of X." and "implicitly-defined copy assignment operator for a union X
- * copies the object representation (3.9) of X." = Bytewise copy instead of memberwise copy.
- * This is important because the padding bytes are included in the hash and comparison between
- * objects.
- */
- union {
- T state;
- };
-
- HashableStruct() {
- // Memset structure to zero padding bits, so that they will be deterministic when hashing
- std::memset(&state, 0, sizeof(T));
- }
-
- bool operator==(const HashableStruct<T>& o) const {
- return std::memcmp(&state, &o.state, sizeof(T)) == 0;
- };
-
- bool operator!=(const HashableStruct<T>& o) const {
- return !(*this == o);
- };
-
- std::size_t Hash() const {
- return Common::ComputeStructHash64(state);
- }
-};
-
struct PairHash {
template <class T1, class T2>
std::size_t operator()(const std::pair<T1, T2>& pair) const noexcept {
diff --git a/src/common/logging/backend.cpp b/src/common/logging/backend.cpp
index 1111cfbad..8f2591d53 100644
--- a/src/common/logging/backend.cpp
+++ b/src/common/logging/backend.cpp
@@ -272,8 +272,10 @@ const char* GetLogClassName(Class log_class) {
#undef CLS
#undef SUB
case Class::Count:
- UNREACHABLE();
+ break;
}
+ UNREACHABLE();
+ return "Invalid";
}
const char* GetLevelName(Level log_level) {
@@ -288,9 +290,11 @@ const char* GetLevelName(Level log_level) {
LVL(Error);
LVL(Critical);
case Level::Count:
- UNREACHABLE();
+ break;
}
#undef LVL
+ UNREACHABLE();
+ return "Invalid";
}
void SetGlobalFilter(const Filter& filter) {
diff --git a/src/common/threadsafe_queue.h b/src/common/threadsafe_queue.h
index e714ba5b3..8268bbd5c 100644
--- a/src/common/threadsafe_queue.h
+++ b/src/common/threadsafe_queue.h
@@ -46,9 +46,16 @@ public:
ElementPtr* new_ptr = new ElementPtr();
write_ptr->next.store(new_ptr, std::memory_order_release);
write_ptr = new_ptr;
- cv.notify_one();
- ++size;
+ const size_t previous_size{size++};
+
+ // Acquire the mutex and then immediately release it as a fence.
+ // TODO(bunnei): This can be replaced with C++20 waitable atomics when properly supported.
+ // See discussion on https://github.com/yuzu-emu/yuzu/pull/3173 for details.
+ if (previous_size == 0) {
+ std::lock_guard lock{cv_mutex};
+ }
+ cv.notify_one();
}
void Pop() {
diff --git a/src/core/CMakeLists.txt b/src/core/CMakeLists.txt
index 4f6a87b0a..7fd226050 100644
--- a/src/core/CMakeLists.txt
+++ b/src/core/CMakeLists.txt
@@ -170,6 +170,7 @@ add_library(core STATIC
hle/kernel/server_port.h
hle/kernel/server_session.cpp
hle/kernel/server_session.h
+ hle/kernel/session.cpp
hle/kernel/session.h
hle/kernel/shared_memory.cpp
hle/kernel/shared_memory.h
@@ -509,7 +510,6 @@ add_library(core STATIC
memory/dmnt_cheat_vm.h
memory.cpp
memory.h
- memory_setup.h
perf_stats.cpp
perf_stats.h
reporter.cpp
@@ -522,6 +522,23 @@ add_library(core STATIC
tools/freezer.h
)
+if (MSVC)
+ target_compile_options(core PRIVATE
+ # 'expression' : signed/unsigned mismatch
+ /we4018
+ # 'argument' : conversion from 'type1' to 'type2', possible loss of data (floating-point)
+ /we4244
+ # 'conversion' : conversion from 'type1' to 'type2', signed/unsigned mismatch
+ /we4245
+ # 'operator': conversion from 'type1:field_bits' to 'type2:field_bits', possible loss of data
+ /we4254
+ # 'var' : conversion from 'size_t' to 'type', possible loss of data
+ /we4267
+ # 'context' : truncation from 'type1' to 'type2'
+ /we4305
+ )
+endif()
+
create_target_directory_groups(core)
target_link_libraries(core PUBLIC common PRIVATE audio_core video_core)
diff --git a/src/core/arm/arm_interface.cpp b/src/core/arm/arm_interface.cpp
index 372612c9b..7e846ddd5 100644
--- a/src/core/arm/arm_interface.cpp
+++ b/src/core/arm/arm_interface.cpp
@@ -13,7 +13,6 @@
#include "core/memory.h"
namespace Core {
-
namespace {
constexpr u64 ELF_DYNAMIC_TAG_NULL = 0;
@@ -61,15 +60,15 @@ static_assert(sizeof(ELFSymbol) == 0x18, "ELFSymbol has incorrect size.");
using Symbols = std::vector<std::pair<ELFSymbol, std::string>>;
-Symbols GetSymbols(VAddr text_offset) {
- const auto mod_offset = text_offset + Memory::Read32(text_offset + 4);
+Symbols GetSymbols(VAddr text_offset, Memory::Memory& memory) {
+ const auto mod_offset = text_offset + memory.Read32(text_offset + 4);
if (mod_offset < text_offset || (mod_offset & 0b11) != 0 ||
- Memory::Read32(mod_offset) != Common::MakeMagic('M', 'O', 'D', '0')) {
+ memory.Read32(mod_offset) != Common::MakeMagic('M', 'O', 'D', '0')) {
return {};
}
- const auto dynamic_offset = Memory::Read32(mod_offset + 0x4) + mod_offset;
+ const auto dynamic_offset = memory.Read32(mod_offset + 0x4) + mod_offset;
VAddr string_table_offset{};
VAddr symbol_table_offset{};
@@ -77,8 +76,8 @@ Symbols GetSymbols(VAddr text_offset) {
VAddr dynamic_index = dynamic_offset;
while (true) {
- const auto tag = Memory::Read64(dynamic_index);
- const auto value = Memory::Read64(dynamic_index + 0x8);
+ const u64 tag = memory.Read64(dynamic_index);
+ const u64 value = memory.Read64(dynamic_index + 0x8);
dynamic_index += 0x10;
if (tag == ELF_DYNAMIC_TAG_NULL) {
@@ -106,11 +105,11 @@ Symbols GetSymbols(VAddr text_offset) {
VAddr symbol_index = symbol_table_address;
while (symbol_index < string_table_address) {
ELFSymbol symbol{};
- Memory::ReadBlock(symbol_index, &symbol, sizeof(ELFSymbol));
+ memory.ReadBlock(symbol_index, &symbol, sizeof(ELFSymbol));
VAddr string_offset = string_table_address + symbol.name_index;
std::string name;
- for (u8 c = Memory::Read8(string_offset); c != 0; c = Memory::Read8(++string_offset)) {
+ for (u8 c = memory.Read8(string_offset); c != 0; c = memory.Read8(++string_offset)) {
name += static_cast<char>(c);
}
@@ -142,28 +141,28 @@ constexpr u64 SEGMENT_BASE = 0x7100000000ull;
std::vector<ARM_Interface::BacktraceEntry> ARM_Interface::GetBacktrace() const {
std::vector<BacktraceEntry> out;
+ auto& memory = system.Memory();
auto fp = GetReg(29);
auto lr = GetReg(30);
-
while (true) {
out.push_back({"", 0, lr, 0});
if (!fp) {
break;
}
- lr = Memory::Read64(fp + 8) - 4;
- fp = Memory::Read64(fp);
+ lr = memory.Read64(fp + 8) - 4;
+ fp = memory.Read64(fp);
}
std::map<VAddr, std::string> modules;
- auto& loader{System::GetInstance().GetAppLoader()};
+ auto& loader{system.GetAppLoader()};
if (loader.ReadNSOModules(modules) != Loader::ResultStatus::Success) {
return {};
}
std::map<std::string, Symbols> symbols;
for (const auto& module : modules) {
- symbols.insert_or_assign(module.second, GetSymbols(module.first));
+ symbols.insert_or_assign(module.second, GetSymbols(module.first, memory));
}
for (auto& entry : out) {
diff --git a/src/core/arm/arm_interface.h b/src/core/arm/arm_interface.h
index 45e94e625..47b964eb7 100644
--- a/src/core/arm/arm_interface.h
+++ b/src/core/arm/arm_interface.h
@@ -17,11 +17,13 @@ enum class VMAPermission : u8;
}
namespace Core {
+class System;
/// Generic ARMv8 CPU interface
class ARM_Interface : NonCopyable {
public:
- virtual ~ARM_Interface() {}
+ explicit ARM_Interface(System& system_) : system{system_} {}
+ virtual ~ARM_Interface() = default;
struct ThreadContext {
std::array<u64, 31> cpu_registers;
@@ -163,6 +165,10 @@ public:
/// fp+0 : pointer to previous frame record
/// fp+8 : value of lr for frame
void LogBacktrace() const;
+
+protected:
+ /// System context that this ARM interface is running under.
+ System& system;
};
} // namespace Core
diff --git a/src/core/arm/dynarmic/arm_dynarmic.cpp b/src/core/arm/dynarmic/arm_dynarmic.cpp
index 700c4afff..f8c7f0efd 100644
--- a/src/core/arm/dynarmic/arm_dynarmic.cpp
+++ b/src/core/arm/dynarmic/arm_dynarmic.cpp
@@ -28,36 +28,38 @@ public:
explicit ARM_Dynarmic_Callbacks(ARM_Dynarmic& parent) : parent(parent) {}
u8 MemoryRead8(u64 vaddr) override {
- return Memory::Read8(vaddr);
+ return parent.system.Memory().Read8(vaddr);
}
u16 MemoryRead16(u64 vaddr) override {
- return Memory::Read16(vaddr);
+ return parent.system.Memory().Read16(vaddr);
}
u32 MemoryRead32(u64 vaddr) override {
- return Memory::Read32(vaddr);
+ return parent.system.Memory().Read32(vaddr);
}
u64 MemoryRead64(u64 vaddr) override {
- return Memory::Read64(vaddr);
+ return parent.system.Memory().Read64(vaddr);
}
Vector MemoryRead128(u64 vaddr) override {
- return {Memory::Read64(vaddr), Memory::Read64(vaddr + 8)};
+ auto& memory = parent.system.Memory();
+ return {memory.Read64(vaddr), memory.Read64(vaddr + 8)};
}
void MemoryWrite8(u64 vaddr, u8 value) override {
- Memory::Write8(vaddr, value);
+ parent.system.Memory().Write8(vaddr, value);
}
void MemoryWrite16(u64 vaddr, u16 value) override {
- Memory::Write16(vaddr, value);
+ parent.system.Memory().Write16(vaddr, value);
}
void MemoryWrite32(u64 vaddr, u32 value) override {
- Memory::Write32(vaddr, value);
+ parent.system.Memory().Write32(vaddr, value);
}
void MemoryWrite64(u64 vaddr, u64 value) override {
- Memory::Write64(vaddr, value);
+ parent.system.Memory().Write64(vaddr, value);
}
void MemoryWrite128(u64 vaddr, Vector value) override {
- Memory::Write64(vaddr, value[0]);
- Memory::Write64(vaddr + 8, value[1]);
+ auto& memory = parent.system.Memory();
+ memory.Write64(vaddr, value[0]);
+ memory.Write64(vaddr + 8, value[1]);
}
void InterpreterFallback(u64 pc, std::size_t num_instructions) override {
@@ -67,7 +69,7 @@ public:
ARM_Interface::ThreadContext ctx;
parent.SaveContext(ctx);
parent.inner_unicorn.LoadContext(ctx);
- parent.inner_unicorn.ExecuteInstructions(static_cast<int>(num_instructions));
+ parent.inner_unicorn.ExecuteInstructions(num_instructions);
parent.inner_unicorn.SaveContext(ctx);
parent.LoadContext(ctx);
num_interpreted_instructions += num_instructions;
@@ -171,9 +173,10 @@ void ARM_Dynarmic::Step() {
ARM_Dynarmic::ARM_Dynarmic(System& system, ExclusiveMonitor& exclusive_monitor,
std::size_t core_index)
- : cb(std::make_unique<ARM_Dynarmic_Callbacks>(*this)), inner_unicorn{system},
- core_index{core_index}, system{system},
- exclusive_monitor{dynamic_cast<DynarmicExclusiveMonitor&>(exclusive_monitor)} {}
+ : ARM_Interface{system},
+ cb(std::make_unique<ARM_Dynarmic_Callbacks>(*this)), inner_unicorn{system},
+ core_index{core_index}, exclusive_monitor{
+ dynamic_cast<DynarmicExclusiveMonitor&>(exclusive_monitor)} {}
ARM_Dynarmic::~ARM_Dynarmic() = default;
@@ -264,7 +267,9 @@ void ARM_Dynarmic::PageTableChanged(Common::PageTable& page_table,
jit = MakeJit(page_table, new_address_space_size_in_bits);
}
-DynarmicExclusiveMonitor::DynarmicExclusiveMonitor(std::size_t core_count) : monitor(core_count) {}
+DynarmicExclusiveMonitor::DynarmicExclusiveMonitor(Memory::Memory& memory_, std::size_t core_count)
+ : monitor(core_count), memory{memory_} {}
+
DynarmicExclusiveMonitor::~DynarmicExclusiveMonitor() = default;
void DynarmicExclusiveMonitor::SetExclusive(std::size_t core_index, VAddr addr) {
@@ -277,29 +282,28 @@ void DynarmicExclusiveMonitor::ClearExclusive() {
}
bool DynarmicExclusiveMonitor::ExclusiveWrite8(std::size_t core_index, VAddr vaddr, u8 value) {
- return monitor.DoExclusiveOperation(core_index, vaddr, 1,
- [&] { Memory::Write8(vaddr, value); });
+ return monitor.DoExclusiveOperation(core_index, vaddr, 1, [&] { memory.Write8(vaddr, value); });
}
bool DynarmicExclusiveMonitor::ExclusiveWrite16(std::size_t core_index, VAddr vaddr, u16 value) {
return monitor.DoExclusiveOperation(core_index, vaddr, 2,
- [&] { Memory::Write16(vaddr, value); });
+ [&] { memory.Write16(vaddr, value); });
}
bool DynarmicExclusiveMonitor::ExclusiveWrite32(std::size_t core_index, VAddr vaddr, u32 value) {
return monitor.DoExclusiveOperation(core_index, vaddr, 4,
- [&] { Memory::Write32(vaddr, value); });
+ [&] { memory.Write32(vaddr, value); });
}
bool DynarmicExclusiveMonitor::ExclusiveWrite64(std::size_t core_index, VAddr vaddr, u64 value) {
return monitor.DoExclusiveOperation(core_index, vaddr, 8,
- [&] { Memory::Write64(vaddr, value); });
+ [&] { memory.Write64(vaddr, value); });
}
bool DynarmicExclusiveMonitor::ExclusiveWrite128(std::size_t core_index, VAddr vaddr, u128 value) {
return monitor.DoExclusiveOperation(core_index, vaddr, 16, [&] {
- Memory::Write64(vaddr + 0, value[0]);
- Memory::Write64(vaddr + 8, value[1]);
+ memory.Write64(vaddr + 0, value[0]);
+ memory.Write64(vaddr + 8, value[1]);
});
}
diff --git a/src/core/arm/dynarmic/arm_dynarmic.h b/src/core/arm/dynarmic/arm_dynarmic.h
index 504d46c68..9cd475cfb 100644
--- a/src/core/arm/dynarmic/arm_dynarmic.h
+++ b/src/core/arm/dynarmic/arm_dynarmic.h
@@ -12,6 +12,10 @@
#include "core/arm/exclusive_monitor.h"
#include "core/arm/unicorn/arm_unicorn.h"
+namespace Memory {
+class Memory;
+}
+
namespace Core {
class ARM_Dynarmic_Callbacks;
@@ -58,13 +62,12 @@ private:
ARM_Unicorn inner_unicorn;
std::size_t core_index;
- System& system;
DynarmicExclusiveMonitor& exclusive_monitor;
};
class DynarmicExclusiveMonitor final : public ExclusiveMonitor {
public:
- explicit DynarmicExclusiveMonitor(std::size_t core_count);
+ explicit DynarmicExclusiveMonitor(Memory::Memory& memory_, std::size_t core_count);
~DynarmicExclusiveMonitor() override;
void SetExclusive(std::size_t core_index, VAddr addr) override;
@@ -79,6 +82,7 @@ public:
private:
friend class ARM_Dynarmic;
Dynarmic::A64::ExclusiveMonitor monitor;
+ Memory::Memory& memory;
};
} // namespace Core
diff --git a/src/core/arm/unicorn/arm_unicorn.cpp b/src/core/arm/unicorn/arm_unicorn.cpp
index d4f41bfc1..48182c99a 100644
--- a/src/core/arm/unicorn/arm_unicorn.cpp
+++ b/src/core/arm/unicorn/arm_unicorn.cpp
@@ -60,17 +60,18 @@ static bool UnmappedMemoryHook(uc_engine* uc, uc_mem_type type, u64 addr, int si
return false;
}
-ARM_Unicorn::ARM_Unicorn(System& system) : system{system} {
+ARM_Unicorn::ARM_Unicorn(System& system) : ARM_Interface{system} {
CHECKED(uc_open(UC_ARCH_ARM64, UC_MODE_ARM, &uc));
auto fpv = 3 << 20;
CHECKED(uc_reg_write(uc, UC_ARM64_REG_CPACR_EL1, &fpv));
uc_hook hook{};
- CHECKED(uc_hook_add(uc, &hook, UC_HOOK_INTR, (void*)InterruptHook, this, 0, -1));
- CHECKED(uc_hook_add(uc, &hook, UC_HOOK_MEM_INVALID, (void*)UnmappedMemoryHook, &system, 0, -1));
+ CHECKED(uc_hook_add(uc, &hook, UC_HOOK_INTR, (void*)InterruptHook, this, 0, UINT64_MAX));
+ CHECKED(uc_hook_add(uc, &hook, UC_HOOK_MEM_INVALID, (void*)UnmappedMemoryHook, &system, 0,
+ UINT64_MAX));
if (GDBStub::IsServerEnabled()) {
- CHECKED(uc_hook_add(uc, &hook, UC_HOOK_CODE, (void*)CodeHook, this, 0, -1));
+ CHECKED(uc_hook_add(uc, &hook, UC_HOOK_CODE, (void*)CodeHook, this, 0, UINT64_MAX));
last_bkpt_hit = false;
}
}
@@ -154,9 +155,10 @@ void ARM_Unicorn::SetTPIDR_EL0(u64 value) {
void ARM_Unicorn::Run() {
if (GDBStub::IsServerEnabled()) {
- ExecuteInstructions(std::max(4000000, 0));
+ ExecuteInstructions(std::max(4000000U, 0U));
} else {
- ExecuteInstructions(std::max(system.CoreTiming().GetDowncount(), s64{0}));
+ ExecuteInstructions(
+ std::max(std::size_t(system.CoreTiming().GetDowncount()), std::size_t{0}));
}
}
@@ -166,7 +168,7 @@ void ARM_Unicorn::Step() {
MICROPROFILE_DEFINE(ARM_Jit_Unicorn, "ARM JIT", "Unicorn", MP_RGB(255, 64, 64));
-void ARM_Unicorn::ExecuteInstructions(int num_instructions) {
+void ARM_Unicorn::ExecuteInstructions(std::size_t num_instructions) {
MICROPROFILE_SCOPE(ARM_Jit_Unicorn);
CHECKED(uc_emu_start(uc, GetPC(), 1ULL << 63, 0, num_instructions));
system.CoreTiming().AddTicks(num_instructions);
diff --git a/src/core/arm/unicorn/arm_unicorn.h b/src/core/arm/unicorn/arm_unicorn.h
index fe2ffd70c..3c5b155f9 100644
--- a/src/core/arm/unicorn/arm_unicorn.h
+++ b/src/core/arm/unicorn/arm_unicorn.h
@@ -34,7 +34,7 @@ public:
void LoadContext(const ThreadContext& ctx) override;
void PrepareReschedule() override;
void ClearExclusiveState() override;
- void ExecuteInstructions(int num_instructions);
+ void ExecuteInstructions(std::size_t num_instructions);
void Run() override;
void Step() override;
void ClearInstructionCache() override;
@@ -45,7 +45,6 @@ private:
static void InterruptHook(uc_engine* uc, u32 int_no, void* user_data);
uc_engine* uc{};
- System& system;
GDBStub::BreakpointAddress last_bkpt{};
bool last_bkpt_hit = false;
};
diff --git a/src/core/core.cpp b/src/core/core.cpp
index eba17218a..d697b80ef 100644
--- a/src/core/core.cpp
+++ b/src/core/core.cpp
@@ -39,13 +39,13 @@
#include "core/hle/service/service.h"
#include "core/hle/service/sm/sm.h"
#include "core/loader/loader.h"
+#include "core/memory.h"
#include "core/memory/cheat_engine.h"
#include "core/perf_stats.h"
#include "core/reporter.h"
#include "core/settings.h"
#include "core/telemetry_session.h"
#include "core/tools/freezer.h"
-#include "video_core/debug_utils/debug_utils.h"
#include "video_core/renderer_base.h"
#include "video_core/video_core.h"
@@ -112,8 +112,8 @@ FileSys::VirtualFile GetGameFileFromPath(const FileSys::VirtualFilesystem& vfs,
}
struct System::Impl {
explicit Impl(System& system)
- : kernel{system}, fs_controller{system}, cpu_core_manager{system}, reporter{system},
- applet_manager{system} {}
+ : kernel{system}, fs_controller{system}, memory{system},
+ cpu_core_manager{system}, reporter{system}, applet_manager{system} {}
Cpu& CurrentCpuCore() {
return cpu_core_manager.GetCurrentCore();
@@ -340,8 +340,8 @@ struct System::Impl {
std::unique_ptr<Loader::AppLoader> app_loader;
std::unique_ptr<VideoCore::RendererBase> renderer;
std::unique_ptr<Tegra::GPU> gpu_core;
- std::shared_ptr<Tegra::DebugContext> debug_context;
- std::unique_ptr<Core::Hardware::InterruptManager> interrupt_manager;
+ std::unique_ptr<Hardware::InterruptManager> interrupt_manager;
+ Memory::Memory memory;
CpuCoreManager cpu_core_manager;
bool is_powered_on = false;
bool exit_lock = false;
@@ -498,6 +498,14 @@ const ExclusiveMonitor& System::Monitor() const {
return impl->cpu_core_manager.GetExclusiveMonitor();
}
+Memory::Memory& System::Memory() {
+ return impl->memory;
+}
+
+const Memory::Memory& System::Memory() const {
+ return impl->memory;
+}
+
Tegra::GPU& System::GPU() {
return *impl->gpu_core;
}
@@ -570,14 +578,6 @@ Loader::AppLoader& System::GetAppLoader() const {
return *impl->app_loader;
}
-void System::SetGPUDebugContext(std::shared_ptr<Tegra::DebugContext> context) {
- impl->debug_context = std::move(context);
-}
-
-Tegra::DebugContext* System::GetGPUDebugContext() const {
- return impl->debug_context.get();
-}
-
void System::SetFilesystem(std::shared_ptr<FileSys::VfsFilesystem> vfs) {
impl->virtual_filesystem = std::move(vfs);
}
diff --git a/src/core/core.h b/src/core/core.h
index 984074ce3..e240c5c58 100644
--- a/src/core/core.h
+++ b/src/core/core.h
@@ -7,6 +7,7 @@
#include <cstddef>
#include <memory>
#include <string>
+#include <vector>
#include "common/common_types.h"
#include "core/file_sys/vfs_types.h"
@@ -85,6 +86,10 @@ namespace Core::Hardware {
class InterruptManager;
}
+namespace Memory {
+class Memory;
+}
+
namespace Core {
class ARM_Interface;
@@ -224,6 +229,12 @@ public:
/// Gets a constant reference to the exclusive monitor
const ExclusiveMonitor& Monitor() const;
+ /// Gets a mutable reference to the system memory instance.
+ Memory::Memory& Memory();
+
+ /// Gets a constant reference to the system memory instance.
+ const Memory::Memory& Memory() const;
+
/// Gets a mutable reference to the GPU interface
Tegra::GPU& GPU();
@@ -296,10 +307,6 @@ public:
Service::SM::ServiceManager& ServiceManager();
const Service::SM::ServiceManager& ServiceManager() const;
- void SetGPUDebugContext(std::shared_ptr<Tegra::DebugContext> context);
-
- Tegra::DebugContext* GetGPUDebugContext() const;
-
void SetFilesystem(std::shared_ptr<FileSys::VfsFilesystem> vfs);
std::shared_ptr<FileSys::VfsFilesystem> GetFilesystem() const;
diff --git a/src/core/core_cpu.cpp b/src/core/core_cpu.cpp
index 233ea572c..630cd4feb 100644
--- a/src/core/core_cpu.cpp
+++ b/src/core/core_cpu.cpp
@@ -66,9 +66,10 @@ Cpu::Cpu(System& system, ExclusiveMonitor& exclusive_monitor, CpuBarrier& cpu_ba
Cpu::~Cpu() = default;
-std::unique_ptr<ExclusiveMonitor> Cpu::MakeExclusiveMonitor(std::size_t num_cores) {
+std::unique_ptr<ExclusiveMonitor> Cpu::MakeExclusiveMonitor(
+ [[maybe_unused]] Memory::Memory& memory, [[maybe_unused]] std::size_t num_cores) {
#ifdef ARCHITECTURE_x86_64
- return std::make_unique<DynarmicExclusiveMonitor>(num_cores);
+ return std::make_unique<DynarmicExclusiveMonitor>(memory, num_cores);
#else
// TODO(merry): Passthrough exclusive monitor
return nullptr;
@@ -95,6 +96,8 @@ void Cpu::RunLoop(bool tight_loop) {
} else {
arm_interface->Step();
}
+ // We are stopping a run, exclusive state must be cleared
+ arm_interface->ClearExclusiveState();
}
core_timing.Advance();
diff --git a/src/core/core_cpu.h b/src/core/core_cpu.h
index cafca8df7..78f5021a2 100644
--- a/src/core/core_cpu.h
+++ b/src/core/core_cpu.h
@@ -24,6 +24,10 @@ namespace Core::Timing {
class CoreTiming;
}
+namespace Memory {
+class Memory;
+}
+
namespace Core {
class ARM_Interface;
@@ -86,7 +90,19 @@ public:
void Shutdown();
- static std::unique_ptr<ExclusiveMonitor> MakeExclusiveMonitor(std::size_t num_cores);
+ /**
+ * Creates an exclusive monitor to handle exclusive reads/writes.
+ *
+ * @param memory The current memory subsystem that the monitor may wish
+ * to keep track of.
+ *
+ * @param num_cores The number of cores to assume about the CPU.
+ *
+ * @returns The constructed exclusive monitor instance, or nullptr if the current
+ * CPU backend is unable to use an exclusive monitor.
+ */
+ static std::unique_ptr<ExclusiveMonitor> MakeExclusiveMonitor(Memory::Memory& memory,
+ std::size_t num_cores);
private:
void Reschedule();
diff --git a/src/core/core_timing.cpp b/src/core/core_timing.cpp
index 0e9570685..aa09fa453 100644
--- a/src/core/core_timing.cpp
+++ b/src/core/core_timing.cpp
@@ -17,11 +17,15 @@ namespace Core::Timing {
constexpr int MAX_SLICE_LENGTH = 10000;
+std::shared_ptr<EventType> CreateEvent(std::string name, TimedCallback&& callback) {
+ return std::make_shared<EventType>(std::move(callback), std::move(name));
+}
+
struct CoreTiming::Event {
s64 time;
u64 fifo_order;
u64 userdata;
- const EventType* type;
+ std::weak_ptr<EventType> type;
// Sort by time, unless the times are the same, in which case sort by
// the order added to the queue
@@ -54,36 +58,15 @@ void CoreTiming::Initialize() {
event_fifo_id = 0;
const auto empty_timed_callback = [](u64, s64) {};
- ev_lost = RegisterEvent("_lost_event", empty_timed_callback);
+ ev_lost = CreateEvent("_lost_event", empty_timed_callback);
}
void CoreTiming::Shutdown() {
ClearPendingEvents();
- UnregisterAllEvents();
-}
-
-EventType* CoreTiming::RegisterEvent(const std::string& name, TimedCallback callback) {
- std::lock_guard guard{inner_mutex};
- // check for existing type with same name.
- // we want event type names to remain unique so that we can use them for serialization.
- ASSERT_MSG(event_types.find(name) == event_types.end(),
- "CoreTiming Event \"{}\" is already registered. Events should only be registered "
- "during Init to avoid breaking save states.",
- name.c_str());
-
- auto info = event_types.emplace(name, EventType{callback, nullptr});
- EventType* event_type = &info.first->second;
- event_type->name = &info.first->first;
- return event_type;
-}
-
-void CoreTiming::UnregisterAllEvents() {
- ASSERT_MSG(event_queue.empty(), "Cannot unregister events with events pending");
- event_types.clear();
}
-void CoreTiming::ScheduleEvent(s64 cycles_into_future, const EventType* event_type, u64 userdata) {
- ASSERT(event_type != nullptr);
+void CoreTiming::ScheduleEvent(s64 cycles_into_future, const std::shared_ptr<EventType>& event_type,
+ u64 userdata) {
std::lock_guard guard{inner_mutex};
const s64 timeout = GetTicks() + cycles_into_future;
@@ -93,13 +76,15 @@ void CoreTiming::ScheduleEvent(s64 cycles_into_future, const EventType* event_ty
}
event_queue.emplace_back(Event{timeout, event_fifo_id++, userdata, event_type});
+
std::push_heap(event_queue.begin(), event_queue.end(), std::greater<>());
}
-void CoreTiming::UnscheduleEvent(const EventType* event_type, u64 userdata) {
+void CoreTiming::UnscheduleEvent(const std::shared_ptr<EventType>& event_type, u64 userdata) {
std::lock_guard guard{inner_mutex};
+
const auto itr = std::remove_if(event_queue.begin(), event_queue.end(), [&](const Event& e) {
- return e.type == event_type && e.userdata == userdata;
+ return e.type.lock().get() == event_type.get() && e.userdata == userdata;
});
// Removing random items breaks the invariant so we have to re-establish it.
@@ -130,10 +115,12 @@ void CoreTiming::ClearPendingEvents() {
event_queue.clear();
}
-void CoreTiming::RemoveEvent(const EventType* event_type) {
+void CoreTiming::RemoveEvent(const std::shared_ptr<EventType>& event_type) {
std::lock_guard guard{inner_mutex};
- const auto itr = std::remove_if(event_queue.begin(), event_queue.end(),
- [&](const Event& e) { return e.type == event_type; });
+
+ const auto itr = std::remove_if(event_queue.begin(), event_queue.end(), [&](const Event& e) {
+ return e.type.lock().get() == event_type.get();
+ });
// Removing random items breaks the invariant so we have to re-establish it.
if (itr != event_queue.end()) {
@@ -181,7 +168,11 @@ void CoreTiming::Advance() {
std::pop_heap(event_queue.begin(), event_queue.end(), std::greater<>());
event_queue.pop_back();
inner_mutex.unlock();
- evt.type->callback(evt.userdata, global_timer - evt.time);
+
+ if (auto event_type{evt.type.lock()}) {
+ event_type->callback(evt.userdata, global_timer - evt.time);
+ }
+
inner_mutex.lock();
}
diff --git a/src/core/core_timing.h b/src/core/core_timing.h
index 3bb88c810..d50f4eb8a 100644
--- a/src/core/core_timing.h
+++ b/src/core/core_timing.h
@@ -6,11 +6,12 @@
#include <chrono>
#include <functional>
+#include <memory>
#include <mutex>
#include <optional>
#include <string>
-#include <unordered_map>
#include <vector>
+
#include "common/common_types.h"
#include "common/threadsafe_queue.h"
@@ -21,10 +22,13 @@ using TimedCallback = std::function<void(u64 userdata, s64 cycles_late)>;
/// Contains the characteristics of a particular event.
struct EventType {
+ EventType(TimedCallback&& callback, std::string&& name)
+ : callback{std::move(callback)}, name{std::move(name)} {}
+
/// The event's callback function.
TimedCallback callback;
/// A pointer to the name of the event.
- const std::string* name;
+ const std::string name;
};
/**
@@ -57,31 +61,17 @@ public:
/// Tears down all timing related functionality.
void Shutdown();
- /// Registers a core timing event with the given name and callback.
- ///
- /// @param name The name of the core timing event to register.
- /// @param callback The callback to execute for the event.
- ///
- /// @returns An EventType instance representing the registered event.
- ///
- /// @pre The name of the event being registered must be unique among all
- /// registered events.
- ///
- EventType* RegisterEvent(const std::string& name, TimedCallback callback);
-
- /// Unregisters all registered events thus far. Note: not thread unsafe
- void UnregisterAllEvents();
-
/// After the first Advance, the slice lengths and the downcount will be reduced whenever an
/// event is scheduled earlier than the current values.
///
/// Scheduling from a callback will not update the downcount until the Advance() completes.
- void ScheduleEvent(s64 cycles_into_future, const EventType* event_type, u64 userdata = 0);
+ void ScheduleEvent(s64 cycles_into_future, const std::shared_ptr<EventType>& event_type,
+ u64 userdata = 0);
- void UnscheduleEvent(const EventType* event_type, u64 userdata);
+ void UnscheduleEvent(const std::shared_ptr<EventType>& event_type, u64 userdata);
/// We only permit one event of each type in the queue at a time.
- void RemoveEvent(const EventType* event_type);
+ void RemoveEvent(const std::shared_ptr<EventType>& event_type);
void ForceExceptionCheck(s64 cycles);
@@ -148,13 +138,18 @@ private:
std::vector<Event> event_queue;
u64 event_fifo_id = 0;
- // Stores each element separately as a linked list node so pointers to elements
- // remain stable regardless of rehashes/resizing.
- std::unordered_map<std::string, EventType> event_types;
-
- EventType* ev_lost = nullptr;
+ std::shared_ptr<EventType> ev_lost;
std::mutex inner_mutex;
};
+/// Creates a core timing event with the given name and callback.
+///
+/// @param name The name of the core timing event to create.
+/// @param callback The callback to execute for the event.
+///
+/// @returns An EventType instance representing the created event.
+///
+std::shared_ptr<EventType> CreateEvent(std::string name, TimedCallback&& callback);
+
} // namespace Core::Timing
diff --git a/src/core/cpu_core_manager.cpp b/src/core/cpu_core_manager.cpp
index 8efd410bb..f04a34133 100644
--- a/src/core/cpu_core_manager.cpp
+++ b/src/core/cpu_core_manager.cpp
@@ -25,7 +25,7 @@ CpuCoreManager::~CpuCoreManager() = default;
void CpuCoreManager::Initialize() {
barrier = std::make_unique<CpuBarrier>();
- exclusive_monitor = Cpu::MakeExclusiveMonitor(cores.size());
+ exclusive_monitor = Cpu::MakeExclusiveMonitor(system.Memory(), cores.size());
for (std::size_t index = 0; index < cores.size(); ++index) {
cores[index] = std::make_unique<Cpu>(system, *exclusive_monitor, *barrier, index);
diff --git a/src/core/crypto/key_manager.cpp b/src/core/crypto/key_manager.cpp
index 222fc95ba..87e6a1fd3 100644
--- a/src/core/crypto/key_manager.cpp
+++ b/src/core/crypto/key_manager.cpp
@@ -22,6 +22,7 @@
#include "common/file_util.h"
#include "common/hex_util.h"
#include "common/logging/log.h"
+#include "common/string_util.h"
#include "core/core.h"
#include "core/crypto/aes_util.h"
#include "core/crypto/key_manager.h"
@@ -378,8 +379,9 @@ std::vector<Ticket> GetTicketblob(const FileUtil::IOFile& ticket_save) {
template <size_t size>
static std::array<u8, size> operator^(const std::array<u8, size>& lhs,
const std::array<u8, size>& rhs) {
- std::array<u8, size> out{};
- std::transform(lhs.begin(), lhs.end(), rhs.begin(), out.begin(), std::bit_xor<>());
+ std::array<u8, size> out;
+ std::transform(lhs.begin(), lhs.end(), rhs.begin(), out.begin(),
+ [](u8 lhs, u8 rhs) { return u8(lhs ^ rhs); });
return out;
}
@@ -396,7 +398,7 @@ static std::array<u8, target_size> MGF1(const std::array<u8, in_size>& seed) {
while (out.size() < target_size) {
out.resize(out.size() + 0x20);
seed_exp[in_size + 3] = static_cast<u8>(i);
- mbedtls_sha256(seed_exp.data(), seed_exp.size(), out.data() + out.size() - 0x20, 0);
+ mbedtls_sha256_ret(seed_exp.data(), seed_exp.size(), out.data() + out.size() - 0x20, 0);
++i;
}
@@ -538,7 +540,7 @@ void KeyManager::LoadFromFile(const std::string& filename, bool is_title_keys) {
Key128 key = Common::HexStringToArray<16>(out[1]);
s128_keys[{S128KeyType::Titlekey, rights_id[1], rights_id[0]}] = key;
} else {
- std::transform(out[0].begin(), out[0].end(), out[0].begin(), ::tolower);
+ out[0] = Common::ToLower(out[0]);
if (s128_file_id.find(out[0]) != s128_file_id.end()) {
const auto index = s128_file_id.at(out[0]);
Key128 key = Common::HexStringToArray<16>(out[1]);
@@ -668,23 +670,27 @@ void KeyManager::WriteKeyToFile(KeyCategory category, std::string_view keyname,
const std::array<u8, Size>& key) {
const std::string yuzu_keys_dir = FileUtil::GetUserPath(FileUtil::UserPath::KeysDir);
std::string filename = "title.keys_autogenerated";
- if (category == KeyCategory::Standard)
+ if (category == KeyCategory::Standard) {
filename = dev_mode ? "dev.keys_autogenerated" : "prod.keys_autogenerated";
- else if (category == KeyCategory::Console)
+ } else if (category == KeyCategory::Console) {
filename = "console.keys_autogenerated";
- const auto add_info_text = !FileUtil::Exists(yuzu_keys_dir + DIR_SEP + filename);
- FileUtil::CreateFullPath(yuzu_keys_dir + DIR_SEP + filename);
- std::ofstream file(yuzu_keys_dir + DIR_SEP + filename, std::ios::app);
- if (!file.is_open())
+ }
+
+ const auto path = yuzu_keys_dir + DIR_SEP + filename;
+ const auto add_info_text = !FileUtil::Exists(path);
+ FileUtil::CreateFullPath(path);
+ FileUtil::IOFile file{path, "a"};
+ if (!file.IsOpen()) {
return;
+ }
if (add_info_text) {
- file
- << "# This file is autogenerated by Yuzu\n"
- << "# It serves to store keys that were automatically generated from the normal keys\n"
- << "# If you are experiencing issues involving keys, it may help to delete this file\n";
+ file.WriteString(
+ "# This file is autogenerated by Yuzu\n"
+ "# It serves to store keys that were automatically generated from the normal keys\n"
+ "# If you are experiencing issues involving keys, it may help to delete this file\n");
}
- file << fmt::format("\n{} = {}", keyname, Common::HexToString(key));
+ file.WriteString(fmt::format("\n{} = {}", keyname, Common::HexToString(key)));
AttemptLoadKeyFile(yuzu_keys_dir, yuzu_keys_dir, filename, category == KeyCategory::Title);
}
@@ -944,12 +950,10 @@ void KeyManager::DeriveETicket(PartitionDataManager& data) {
return;
}
- Key128 rsa_oaep_kek{};
- std::transform(seed3.begin(), seed3.end(), mask0.begin(), rsa_oaep_kek.begin(),
- std::bit_xor<>());
-
- if (rsa_oaep_kek == Key128{})
+ const Key128 rsa_oaep_kek = seed3 ^ mask0;
+ if (rsa_oaep_kek == Key128{}) {
return;
+ }
SetKey(S128KeyType::Source, rsa_oaep_kek,
static_cast<u64>(SourceKeyType::RSAOaepKekGeneration));
diff --git a/src/core/crypto/partition_data_manager.cpp b/src/core/crypto/partition_data_manager.cpp
index 594cd82c5..d64302f2e 100644
--- a/src/core/crypto/partition_data_manager.cpp
+++ b/src/core/crypto/partition_data_manager.cpp
@@ -161,7 +161,7 @@ std::array<u8, key_size> FindKeyFromHex(const std::vector<u8>& binary,
std::array<u8, 0x20> temp{};
for (size_t i = 0; i < binary.size() - key_size; ++i) {
- mbedtls_sha256(binary.data() + i, key_size, temp.data(), 0);
+ mbedtls_sha256_ret(binary.data() + i, key_size, temp.data(), 0);
if (temp != hash)
continue;
@@ -189,7 +189,7 @@ static std::array<Key128, 0x20> FindEncryptedMasterKeyFromHex(const std::vector<
AESCipher<Key128> cipher(key, Mode::ECB);
for (size_t i = 0; i < binary.size() - 0x10; ++i) {
cipher.Transcode(binary.data() + i, dec_temp.size(), dec_temp.data(), Op::Decrypt);
- mbedtls_sha256(dec_temp.data(), dec_temp.size(), temp.data(), 0);
+ mbedtls_sha256_ret(dec_temp.data(), dec_temp.size(), temp.data(), 0);
for (size_t k = 0; k < out.size(); ++k) {
if (temp == master_key_hashes[k]) {
@@ -204,11 +204,12 @@ static std::array<Key128, 0x20> FindEncryptedMasterKeyFromHex(const std::vector<
FileSys::VirtualFile FindFileInDirWithNames(const FileSys::VirtualDir& dir,
const std::string& name) {
- auto upper = name;
- std::transform(upper.begin(), upper.end(), upper.begin(), [](u8 c) { return std::toupper(c); });
+ const auto upper = Common::ToUpper(name);
+
for (const auto& fname : {name, name + ".bin", upper, upper + ".BIN"}) {
- if (dir->GetFile(fname) != nullptr)
+ if (dir->GetFile(fname) != nullptr) {
return dir->GetFile(fname);
+ }
}
return nullptr;
diff --git a/src/core/file_sys/directory.h b/src/core/file_sys/directory.h
index 7b5c509fb..0d73eecc9 100644
--- a/src/core/file_sys/directory.h
+++ b/src/core/file_sys/directory.h
@@ -15,7 +15,7 @@
namespace FileSys {
-enum EntryType : u8 {
+enum class EntryType : u8 {
Directory = 0,
File = 1,
};
diff --git a/src/core/file_sys/kernel_executable.cpp b/src/core/file_sys/kernel_executable.cpp
index 371300684..76313679d 100644
--- a/src/core/file_sys/kernel_executable.cpp
+++ b/src/core/file_sys/kernel_executable.cpp
@@ -147,7 +147,7 @@ std::vector<u32> KIP::GetKernelCapabilities() const {
}
s32 KIP::GetMainThreadPriority() const {
- return header.main_thread_priority;
+ return static_cast<s32>(header.main_thread_priority);
}
u32 KIP::GetMainThreadStackSize() const {
diff --git a/src/core/file_sys/patch_manager.cpp b/src/core/file_sys/patch_manager.cpp
index df0ecb15c..e226e9711 100644
--- a/src/core/file_sys/patch_manager.cpp
+++ b/src/core/file_sys/patch_manager.cpp
@@ -76,7 +76,7 @@ VirtualDir PatchManager::PatchExeFS(VirtualDir exefs) const {
const auto& disabled = Settings::values.disabled_addons[title_id];
const auto update_disabled =
- std::find(disabled.begin(), disabled.end(), "Update") != disabled.end();
+ std::find(disabled.cbegin(), disabled.cend(), "Update") != disabled.cend();
// Game Updates
const auto update_tid = GetUpdateTitleID(title_id);
@@ -127,7 +127,7 @@ std::vector<VirtualFile> PatchManager::CollectPatches(const std::vector<VirtualD
std::vector<VirtualFile> out;
out.reserve(patch_dirs.size());
for (const auto& subdir : patch_dirs) {
- if (std::find(disabled.begin(), disabled.end(), subdir->GetName()) != disabled.end())
+ if (std::find(disabled.cbegin(), disabled.cend(), subdir->GetName()) != disabled.cend())
continue;
auto exefs_dir = subdir->GetSubdirectory("exefs");
@@ -284,12 +284,17 @@ std::vector<Memory::CheatEntry> PatchManager::CreateCheatList(
return {};
}
+ const auto& disabled = Settings::values.disabled_addons[title_id];
auto patch_dirs = load_dir->GetSubdirectories();
std::sort(patch_dirs.begin(), patch_dirs.end(),
[](const VirtualDir& l, const VirtualDir& r) { return l->GetName() < r->GetName(); });
std::vector<Memory::CheatEntry> out;
for (const auto& subdir : patch_dirs) {
+ if (std::find(disabled.cbegin(), disabled.cend(), subdir->GetName()) != disabled.cend()) {
+ continue;
+ }
+
auto cheats_dir = subdir->GetSubdirectory("cheats");
if (cheats_dir != nullptr) {
auto res = ReadCheatFileFromFolder(system, title_id, build_id_, cheats_dir, true);
@@ -331,8 +336,9 @@ static void ApplyLayeredFS(VirtualFile& romfs, u64 title_id, ContentRecordType t
layers.reserve(patch_dirs.size() + 1);
layers_ext.reserve(patch_dirs.size() + 1);
for (const auto& subdir : patch_dirs) {
- if (std::find(disabled.begin(), disabled.end(), subdir->GetName()) != disabled.end())
+ if (std::find(disabled.cbegin(), disabled.cend(), subdir->GetName()) != disabled.cend()) {
continue;
+ }
auto romfs_dir = subdir->GetSubdirectory("romfs");
if (romfs_dir != nullptr)
@@ -381,7 +387,7 @@ VirtualFile PatchManager::PatchRomFS(VirtualFile romfs, u64 ivfc_offset, Content
const auto& disabled = Settings::values.disabled_addons[title_id];
const auto update_disabled =
- std::find(disabled.begin(), disabled.end(), "Update") != disabled.end();
+ std::find(disabled.cbegin(), disabled.cend(), "Update") != disabled.cend();
if (!update_disabled && update != nullptr) {
const auto new_nca = std::make_shared<NCA>(update, romfs, ivfc_offset);
@@ -431,7 +437,7 @@ std::map<std::string, std::string, std::less<>> PatchManager::GetPatchVersionNam
auto [nacp, discard_icon_file] = update.GetControlMetadata();
const auto update_disabled =
- std::find(disabled.begin(), disabled.end(), "Update") != disabled.end();
+ std::find(disabled.cbegin(), disabled.cend(), "Update") != disabled.cend();
const auto update_label = update_disabled ? "[D] Update" : "Update";
if (nacp != nullptr) {
diff --git a/src/core/file_sys/program_metadata.cpp b/src/core/file_sys/program_metadata.cpp
index 7310b3602..1d6c30962 100644
--- a/src/core/file_sys/program_metadata.cpp
+++ b/src/core/file_sys/program_metadata.cpp
@@ -52,14 +52,14 @@ Loader::ResultStatus ProgramMetadata::Load(VirtualFile file) {
}
void ProgramMetadata::LoadManual(bool is_64_bit, ProgramAddressSpaceType address_space,
- u8 main_thread_prio, u8 main_thread_core,
+ s32 main_thread_prio, u32 main_thread_core,
u32 main_thread_stack_size, u64 title_id,
u64 filesystem_permissions,
KernelCapabilityDescriptors capabilities) {
npdm_header.has_64_bit_instructions.Assign(is_64_bit);
npdm_header.address_space_type.Assign(address_space);
- npdm_header.main_thread_priority = main_thread_prio;
- npdm_header.main_thread_cpu = main_thread_core;
+ npdm_header.main_thread_priority = static_cast<u8>(main_thread_prio);
+ npdm_header.main_thread_cpu = static_cast<u8>(main_thread_core);
npdm_header.main_stack_size = main_thread_stack_size;
aci_header.title_id = title_id;
aci_file_access.permissions = filesystem_permissions;
diff --git a/src/core/file_sys/program_metadata.h b/src/core/file_sys/program_metadata.h
index 88ec97d85..f8759a396 100644
--- a/src/core/file_sys/program_metadata.h
+++ b/src/core/file_sys/program_metadata.h
@@ -47,8 +47,8 @@ public:
Loader::ResultStatus Load(VirtualFile file);
// Load from parameters instead of NPDM file, used for KIP
- void LoadManual(bool is_64_bit, ProgramAddressSpaceType address_space, u8 main_thread_prio,
- u8 main_thread_core, u32 main_thread_stack_size, u64 title_id,
+ void LoadManual(bool is_64_bit, ProgramAddressSpaceType address_space, s32 main_thread_prio,
+ u32 main_thread_core, u32 main_thread_stack_size, u64 title_id,
u64 filesystem_permissions, KernelCapabilityDescriptors capabilities);
bool Is64BitProgram() const;
diff --git a/src/core/file_sys/registered_cache.cpp b/src/core/file_sys/registered_cache.cpp
index ac3fbd849..6e9cf67ef 100644
--- a/src/core/file_sys/registered_cache.cpp
+++ b/src/core/file_sys/registered_cache.cpp
@@ -62,7 +62,7 @@ static std::string GetRelativePathFromNcaID(const std::array<u8, 16>& nca_id, bo
Common::HexToString(nca_id, second_hex_upper));
Core::Crypto::SHA256Hash hash{};
- mbedtls_sha256(nca_id.data(), nca_id.size(), hash.data(), 0);
+ mbedtls_sha256_ret(nca_id.data(), nca_id.size(), hash.data(), 0);
return fmt::format(cnmt_suffix ? "/000000{:02X}/{}.cnmt.nca" : "/000000{:02X}/{}.nca", hash[0],
Common::HexToString(nca_id, second_hex_upper));
}
@@ -141,7 +141,7 @@ bool PlaceholderCache::Create(const NcaID& id, u64 size) const {
}
Core::Crypto::SHA256Hash hash{};
- mbedtls_sha256(id.data(), id.size(), hash.data(), 0);
+ mbedtls_sha256_ret(id.data(), id.size(), hash.data(), 0);
const auto dirname = fmt::format("000000{:02X}", hash[0]);
const auto dir2 = GetOrCreateDirectoryRelative(dir, dirname);
@@ -165,7 +165,7 @@ bool PlaceholderCache::Delete(const NcaID& id) const {
}
Core::Crypto::SHA256Hash hash{};
- mbedtls_sha256(id.data(), id.size(), hash.data(), 0);
+ mbedtls_sha256_ret(id.data(), id.size(), hash.data(), 0);
const auto dirname = fmt::format("000000{:02X}", hash[0]);
const auto dir2 = GetOrCreateDirectoryRelative(dir, dirname);
@@ -603,7 +603,7 @@ InstallResult RegisteredCache::InstallEntry(const NCA& nca, TitleType type,
OptionalHeader opt_header{0, 0};
ContentRecord c_rec{{}, {}, {}, GetCRTypeFromNCAType(nca.GetType()), {}};
const auto& data = nca.GetBaseFile()->ReadBytes(0x100000);
- mbedtls_sha256(data.data(), data.size(), c_rec.hash.data(), 0);
+ mbedtls_sha256_ret(data.data(), data.size(), c_rec.hash.data(), 0);
memcpy(&c_rec.nca_id, &c_rec.hash, 16);
const CNMT new_cnmt(header, opt_header, {c_rec}, {});
if (!RawInstallYuzuMeta(new_cnmt))
@@ -626,7 +626,7 @@ InstallResult RegisteredCache::RawInstallNCA(const NCA& nca, const VfsCopyFuncti
id = *override_id;
} else {
const auto& data = in->ReadBytes(0x100000);
- mbedtls_sha256(data.data(), data.size(), hash.data(), 0);
+ mbedtls_sha256_ret(data.data(), data.size(), hash.data(), 0);
memcpy(id.data(), hash.data(), 16);
}
diff --git a/src/core/file_sys/romfs.cpp b/src/core/file_sys/romfs.cpp
index ebbdf081e..c909d1ce4 100644
--- a/src/core/file_sys/romfs.cpp
+++ b/src/core/file_sys/romfs.cpp
@@ -2,6 +2,8 @@
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
+#include <memory>
+
#include "common/common_types.h"
#include "common/swap.h"
#include "core/file_sys/fsmitm_romfsbuild.h"
@@ -12,7 +14,7 @@
#include "core/file_sys/vfs_vector.h"
namespace FileSys {
-
+namespace {
constexpr u32 ROMFS_ENTRY_EMPTY = 0xFFFFFFFF;
struct TableLocation {
@@ -51,7 +53,7 @@ struct FileEntry {
static_assert(sizeof(FileEntry) == 0x20, "FileEntry has incorrect size.");
template <typename Entry>
-static std::pair<Entry, std::string> GetEntry(const VirtualFile& file, std::size_t offset) {
+std::pair<Entry, std::string> GetEntry(const VirtualFile& file, std::size_t offset) {
Entry entry{};
if (file->ReadObject(&entry, offset) != sizeof(Entry))
return {};
@@ -99,6 +101,7 @@ void ProcessDirectory(VirtualFile file, std::size_t dir_offset, std::size_t file
this_dir_offset = entry.first.sibling;
}
}
+} // Anonymous namespace
VirtualDir ExtractRomFS(VirtualFile file, RomFSExtractionType type) {
RomFSHeader header{};
diff --git a/src/core/file_sys/romfs.h b/src/core/file_sys/romfs.h
index 1c89be8a4..2fd07ed04 100644
--- a/src/core/file_sys/romfs.h
+++ b/src/core/file_sys/romfs.h
@@ -5,10 +5,6 @@
#pragma once
#include <array>
-#include <map>
-#include "common/common_funcs.h"
-#include "common/common_types.h"
-#include "common/swap.h"
#include "core/file_sys/vfs.h"
namespace FileSys {
diff --git a/src/core/file_sys/romfs_factory.cpp b/src/core/file_sys/romfs_factory.cpp
index 4bd2e6183..418a39a7e 100644
--- a/src/core/file_sys/romfs_factory.cpp
+++ b/src/core/file_sys/romfs_factory.cpp
@@ -71,12 +71,12 @@ ResultVal<VirtualFile> RomFSFactory::Open(u64 title_id, StorageId storage,
if (res == nullptr) {
// TODO(DarkLordZach): Find the right error code to use here
- return ResultCode(-1);
+ return RESULT_UNKNOWN;
}
const auto romfs = res->GetRomFS();
if (romfs == nullptr) {
// TODO(DarkLordZach): Find the right error code to use here
- return ResultCode(-1);
+ return RESULT_UNKNOWN;
}
return MakeResult<VirtualFile>(romfs);
}
diff --git a/src/core/file_sys/savedata_factory.cpp b/src/core/file_sys/savedata_factory.cpp
index e2a7eaf7b..f3def93ab 100644
--- a/src/core/file_sys/savedata_factory.cpp
+++ b/src/core/file_sys/savedata_factory.cpp
@@ -90,7 +90,7 @@ ResultVal<VirtualDir> SaveDataFactory::Create(SaveDataSpaceId space,
// Return an error if the save data doesn't actually exist.
if (out == nullptr) {
// TODO(DarkLordZach): Find out correct error code.
- return ResultCode(-1);
+ return RESULT_UNKNOWN;
}
return MakeResult<VirtualDir>(std::move(out));
@@ -111,7 +111,7 @@ ResultVal<VirtualDir> SaveDataFactory::Open(SaveDataSpaceId space,
// Return an error if the save data doesn't actually exist.
if (out == nullptr) {
// TODO(Subv): Find out correct error code.
- return ResultCode(-1);
+ return RESULT_UNKNOWN;
}
return MakeResult<VirtualDir>(std::move(out));
diff --git a/src/core/file_sys/vfs_libzip.cpp b/src/core/file_sys/vfs_libzip.cpp
index 8bdaa7e4a..11d1978ea 100644
--- a/src/core/file_sys/vfs_libzip.cpp
+++ b/src/core/file_sys/vfs_libzip.cpp
@@ -27,7 +27,7 @@ VirtualDir ExtractZIP(VirtualFile file) {
std::shared_ptr<VectorVfsDirectory> out = std::make_shared<VectorVfsDirectory>();
- const auto num_entries = zip_get_num_entries(zip.get(), 0);
+ const auto num_entries = static_cast<std::size_t>(zip_get_num_entries(zip.get(), 0));
zip_stat_t stat{};
zip_stat_init(&stat);
diff --git a/src/core/file_sys/xts_archive.cpp b/src/core/file_sys/xts_archive.cpp
index 4bc5cb2ee..86e06ccb9 100644
--- a/src/core/file_sys/xts_archive.cpp
+++ b/src/core/file_sys/xts_archive.cpp
@@ -7,12 +7,13 @@
#include <cstring>
#include <regex>
#include <string>
+
#include <mbedtls/md.h>
#include <mbedtls/sha256.h>
-#include "common/assert.h"
+
#include "common/file_util.h"
#include "common/hex_util.h"
-#include "common/logging/log.h"
+#include "common/string_util.h"
#include "core/crypto/aes_util.h"
#include "core/crypto/xts_encryption_layer.h"
#include "core/file_sys/partition_filesystem.h"
@@ -53,18 +54,15 @@ NAX::NAX(VirtualFile file_) : header(std::make_unique<NAXHeader>()), file(std::m
return;
}
- std::string two_dir = match[1];
- std::string nca_id = match[2];
- std::transform(two_dir.begin(), two_dir.end(), two_dir.begin(), ::toupper);
- std::transform(nca_id.begin(), nca_id.end(), nca_id.begin(), ::tolower);
-
+ const std::string two_dir = Common::ToUpper(match[1]);
+ const std::string nca_id = Common::ToLower(match[2]);
status = Parse(fmt::format("/registered/{}/{}.nca", two_dir, nca_id));
}
NAX::NAX(VirtualFile file_, std::array<u8, 0x10> nca_id)
: header(std::make_unique<NAXHeader>()), file(std::move(file_)) {
Core::Crypto::SHA256Hash hash{};
- mbedtls_sha256(nca_id.data(), nca_id.size(), hash.data(), 0);
+ mbedtls_sha256_ret(nca_id.data(), nca_id.size(), hash.data(), 0);
status = Parse(fmt::format("/registered/000000{:02X}/{}.nca", hash[0],
Common::HexToString(nca_id, false)));
}
@@ -93,8 +91,7 @@ Loader::ResultStatus NAX::Parse(std::string_view path) {
std::size_t i = 0;
for (; i < sd_keys.size(); ++i) {
std::array<Core::Crypto::Key128, 2> nax_keys{};
- if (!CalculateHMAC256(nax_keys.data(), sd_keys[i].data(), 0x10, std::string(path).c_str(),
- path.size())) {
+ if (!CalculateHMAC256(nax_keys.data(), sd_keys[i].data(), 0x10, path.data(), path.size())) {
return Loader::ResultStatus::ErrorNAXKeyHMACFailed;
}
diff --git a/src/core/gdbstub/gdbstub.cpp b/src/core/gdbstub/gdbstub.cpp
index 20bb50868..37cb28848 100644
--- a/src/core/gdbstub/gdbstub.cpp
+++ b/src/core/gdbstub/gdbstub.cpp
@@ -468,7 +468,8 @@ static u8 ReadByte() {
/// Calculate the checksum of the current command buffer.
static u8 CalculateChecksum(const u8* buffer, std::size_t length) {
- return static_cast<u8>(std::accumulate(buffer, buffer + length, 0, std::plus<u8>()));
+ return static_cast<u8>(std::accumulate(buffer, buffer + length, u8{0},
+ [](u8 lhs, u8 rhs) { return u8(lhs + rhs); }));
}
/**
@@ -507,8 +508,9 @@ static void RemoveBreakpoint(BreakpointType type, VAddr addr) {
bp->second.len, bp->second.addr, static_cast<int>(type));
if (type == BreakpointType::Execute) {
- Memory::WriteBlock(bp->second.addr, bp->second.inst.data(), bp->second.inst.size());
- Core::System::GetInstance().InvalidateCpuInstructionCaches();
+ auto& system = Core::System::GetInstance();
+ system.Memory().WriteBlock(bp->second.addr, bp->second.inst.data(), bp->second.inst.size());
+ system.InvalidateCpuInstructionCaches();
}
p.erase(addr);
}
@@ -968,12 +970,13 @@ static void ReadMemory() {
SendReply("E01");
}
- if (!Memory::IsValidVirtualAddress(addr)) {
+ auto& memory = Core::System::GetInstance().Memory();
+ if (!memory.IsValidVirtualAddress(addr)) {
return SendReply("E00");
}
std::vector<u8> data(len);
- Memory::ReadBlock(addr, data.data(), len);
+ memory.ReadBlock(addr, data.data(), len);
MemToGdbHex(reply, data.data(), len);
reply[len * 2] = '\0';
@@ -983,22 +986,23 @@ static void ReadMemory() {
/// Modify location in memory with data received from the gdb client.
static void WriteMemory() {
auto start_offset = command_buffer + 1;
- auto addr_pos = std::find(start_offset, command_buffer + command_length, ',');
- VAddr addr = HexToLong(start_offset, static_cast<u64>(addr_pos - start_offset));
+ const auto addr_pos = std::find(start_offset, command_buffer + command_length, ',');
+ const VAddr addr = HexToLong(start_offset, static_cast<u64>(addr_pos - start_offset));
start_offset = addr_pos + 1;
- auto len_pos = std::find(start_offset, command_buffer + command_length, ':');
- u64 len = HexToLong(start_offset, static_cast<u64>(len_pos - start_offset));
+ const auto len_pos = std::find(start_offset, command_buffer + command_length, ':');
+ const u64 len = HexToLong(start_offset, static_cast<u64>(len_pos - start_offset));
- if (!Memory::IsValidVirtualAddress(addr)) {
+ auto& system = Core::System::GetInstance();
+ auto& memory = system.Memory();
+ if (!memory.IsValidVirtualAddress(addr)) {
return SendReply("E00");
}
std::vector<u8> data(len);
-
GdbHexToMem(data.data(), len_pos + 1, len);
- Memory::WriteBlock(addr, data.data(), len);
- Core::System::GetInstance().InvalidateCpuInstructionCaches();
+ memory.WriteBlock(addr, data.data(), len);
+ system.InvalidateCpuInstructionCaches();
SendReply("OK");
}
@@ -1054,12 +1058,15 @@ static bool CommitBreakpoint(BreakpointType type, VAddr addr, u64 len) {
breakpoint.active = true;
breakpoint.addr = addr;
breakpoint.len = len;
- Memory::ReadBlock(addr, breakpoint.inst.data(), breakpoint.inst.size());
+
+ auto& system = Core::System::GetInstance();
+ auto& memory = system.Memory();
+ memory.ReadBlock(addr, breakpoint.inst.data(), breakpoint.inst.size());
static constexpr std::array<u8, 4> btrap{0x00, 0x7d, 0x20, 0xd4};
if (type == BreakpointType::Execute) {
- Memory::WriteBlock(addr, btrap.data(), btrap.size());
- Core::System::GetInstance().InvalidateCpuInstructionCaches();
+ memory.WriteBlock(addr, btrap.data(), btrap.size());
+ system.InvalidateCpuInstructionCaches();
}
p.insert({addr, breakpoint});
diff --git a/src/core/hardware_interrupt_manager.cpp b/src/core/hardware_interrupt_manager.cpp
index c2115db2d..c629d9fa1 100644
--- a/src/core/hardware_interrupt_manager.cpp
+++ b/src/core/hardware_interrupt_manager.cpp
@@ -11,13 +11,12 @@
namespace Core::Hardware {
InterruptManager::InterruptManager(Core::System& system_in) : system(system_in) {
- gpu_interrupt_event =
- system.CoreTiming().RegisterEvent("GPUInterrupt", [this](u64 message, s64) {
- auto nvdrv = system.ServiceManager().GetService<Service::Nvidia::NVDRV>("nvdrv");
- const u32 syncpt = static_cast<u32>(message >> 32);
- const u32 value = static_cast<u32>(message);
- nvdrv->SignalGPUInterruptSyncpt(syncpt, value);
- });
+ gpu_interrupt_event = Core::Timing::CreateEvent("GPUInterrupt", [this](u64 message, s64) {
+ auto nvdrv = system.ServiceManager().GetService<Service::Nvidia::NVDRV>("nvdrv");
+ const u32 syncpt = static_cast<u32>(message >> 32);
+ const u32 value = static_cast<u32>(message);
+ nvdrv->SignalGPUInterruptSyncpt(syncpt, value);
+ });
}
InterruptManager::~InterruptManager() = default;
diff --git a/src/core/hardware_interrupt_manager.h b/src/core/hardware_interrupt_manager.h
index 494db883a..5fa306ae0 100644
--- a/src/core/hardware_interrupt_manager.h
+++ b/src/core/hardware_interrupt_manager.h
@@ -4,6 +4,8 @@
#pragma once
+#include <memory>
+
#include "common/common_types.h"
namespace Core {
@@ -25,7 +27,7 @@ public:
private:
Core::System& system;
- Core::Timing::EventType* gpu_interrupt_event{};
+ std::shared_ptr<Core::Timing::EventType> gpu_interrupt_event;
};
} // namespace Core::Hardware
diff --git a/src/core/hle/ipc_helpers.h b/src/core/hle/ipc_helpers.h
index 5bb139483..0dc6a4a43 100644
--- a/src/core/hle/ipc_helpers.h
+++ b/src/core/hle/ipc_helpers.h
@@ -19,6 +19,7 @@
#include "core/hle/kernel/hle_ipc.h"
#include "core/hle/kernel/object.h"
#include "core/hle/kernel/server_session.h"
+#include "core/hle/kernel/session.h"
#include "core/hle/result.h"
namespace IPC {
@@ -139,10 +140,9 @@ public:
context->AddDomainObject(std::move(iface));
} else {
auto& kernel = Core::System::GetInstance().Kernel();
- auto [server, client] =
- Kernel::ServerSession::CreateSessionPair(kernel, iface->GetServiceName());
- iface->ClientConnected(server);
+ auto [client, server] = Kernel::Session::Create(kernel, iface->GetServiceName());
context->AddMoveObject(std::move(client));
+ iface->ClientConnected(std::move(server));
}
}
@@ -203,10 +203,10 @@ public:
void PushRaw(const T& value);
template <typename... O>
- void PushMoveObjects(Kernel::SharedPtr<O>... pointers);
+ void PushMoveObjects(std::shared_ptr<O>... pointers);
template <typename... O>
- void PushCopyObjects(Kernel::SharedPtr<O>... pointers);
+ void PushCopyObjects(std::shared_ptr<O>... pointers);
private:
u32 normal_params_size{};
@@ -298,7 +298,7 @@ void ResponseBuilder::Push(const First& first_value, const Other&... other_value
}
template <typename... O>
-inline void ResponseBuilder::PushCopyObjects(Kernel::SharedPtr<O>... pointers) {
+inline void ResponseBuilder::PushCopyObjects(std::shared_ptr<O>... pointers) {
auto objects = {pointers...};
for (auto& object : objects) {
context->AddCopyObject(std::move(object));
@@ -306,7 +306,7 @@ inline void ResponseBuilder::PushCopyObjects(Kernel::SharedPtr<O>... pointers) {
}
template <typename... O>
-inline void ResponseBuilder::PushMoveObjects(Kernel::SharedPtr<O>... pointers) {
+inline void ResponseBuilder::PushMoveObjects(std::shared_ptr<O>... pointers) {
auto objects = {pointers...};
for (auto& object : objects) {
context->AddMoveObject(std::move(object));
@@ -357,10 +357,10 @@ public:
T PopRaw();
template <typename T>
- Kernel::SharedPtr<T> GetMoveObject(std::size_t index);
+ std::shared_ptr<T> GetMoveObject(std::size_t index);
template <typename T>
- Kernel::SharedPtr<T> GetCopyObject(std::size_t index);
+ std::shared_ptr<T> GetCopyObject(std::size_t index);
template <class T>
std::shared_ptr<T> PopIpcInterface() {
@@ -465,12 +465,12 @@ void RequestParser::Pop(First& first_value, Other&... other_values) {
}
template <typename T>
-Kernel::SharedPtr<T> RequestParser::GetMoveObject(std::size_t index) {
+std::shared_ptr<T> RequestParser::GetMoveObject(std::size_t index) {
return context->GetMoveObject<T>(index);
}
template <typename T>
-Kernel::SharedPtr<T> RequestParser::GetCopyObject(std::size_t index) {
+std::shared_ptr<T> RequestParser::GetCopyObject(std::size_t index) {
return context->GetCopyObject<T>(index);
}
diff --git a/src/core/hle/kernel/address_arbiter.cpp b/src/core/hle/kernel/address_arbiter.cpp
index de0a9064e..db189c8e3 100644
--- a/src/core/hle/kernel/address_arbiter.cpp
+++ b/src/core/hle/kernel/address_arbiter.cpp
@@ -11,18 +11,16 @@
#include "core/core_cpu.h"
#include "core/hle/kernel/address_arbiter.h"
#include "core/hle/kernel/errors.h"
-#include "core/hle/kernel/object.h"
-#include "core/hle/kernel/process.h"
#include "core/hle/kernel/scheduler.h"
#include "core/hle/kernel/thread.h"
#include "core/hle/result.h"
#include "core/memory.h"
namespace Kernel {
-namespace {
+
// Wake up num_to_wake (or all) threads in a vector.
-void WakeThreads(const std::vector<SharedPtr<Thread>>& waiting_threads, s32 num_to_wake) {
- auto& system = Core::System::GetInstance();
+void AddressArbiter::WakeThreads(const std::vector<std::shared_ptr<Thread>>& waiting_threads,
+ s32 num_to_wake) {
// Only process up to 'target' threads, unless 'target' is <= 0, in which case process
// them all.
std::size_t last = waiting_threads.size();
@@ -34,12 +32,12 @@ void WakeThreads(const std::vector<SharedPtr<Thread>>& waiting_threads, s32 num_
for (std::size_t i = 0; i < last; i++) {
ASSERT(waiting_threads[i]->GetStatus() == ThreadStatus::WaitArb);
waiting_threads[i]->SetWaitSynchronizationResult(RESULT_SUCCESS);
+ RemoveThread(waiting_threads[i]);
waiting_threads[i]->SetArbiterWaitAddress(0);
waiting_threads[i]->ResumeFromWait();
system.PrepareReschedule(waiting_threads[i]->GetProcessorID());
}
}
-} // Anonymous namespace
AddressArbiter::AddressArbiter(Core::System& system) : system{system} {}
AddressArbiter::~AddressArbiter() = default;
@@ -59,35 +57,41 @@ ResultCode AddressArbiter::SignalToAddress(VAddr address, SignalType type, s32 v
}
ResultCode AddressArbiter::SignalToAddressOnly(VAddr address, s32 num_to_wake) {
- const std::vector<SharedPtr<Thread>> waiting_threads = GetThreadsWaitingOnAddress(address);
+ const std::vector<std::shared_ptr<Thread>> waiting_threads =
+ GetThreadsWaitingOnAddress(address);
WakeThreads(waiting_threads, num_to_wake);
return RESULT_SUCCESS;
}
ResultCode AddressArbiter::IncrementAndSignalToAddressIfEqual(VAddr address, s32 value,
s32 num_to_wake) {
+ auto& memory = system.Memory();
+
// Ensure that we can write to the address.
- if (!Memory::IsValidVirtualAddress(address)) {
+ if (!memory.IsValidVirtualAddress(address)) {
return ERR_INVALID_ADDRESS_STATE;
}
- if (static_cast<s32>(Memory::Read32(address)) != value) {
+ if (static_cast<s32>(memory.Read32(address)) != value) {
return ERR_INVALID_STATE;
}
- Memory::Write32(address, static_cast<u32>(value + 1));
+ memory.Write32(address, static_cast<u32>(value + 1));
return SignalToAddressOnly(address, num_to_wake);
}
ResultCode AddressArbiter::ModifyByWaitingCountAndSignalToAddressIfEqual(VAddr address, s32 value,
s32 num_to_wake) {
+ auto& memory = system.Memory();
+
// Ensure that we can write to the address.
- if (!Memory::IsValidVirtualAddress(address)) {
+ if (!memory.IsValidVirtualAddress(address)) {
return ERR_INVALID_ADDRESS_STATE;
}
// Get threads waiting on the address.
- const std::vector<SharedPtr<Thread>> waiting_threads = GetThreadsWaitingOnAddress(address);
+ const std::vector<std::shared_ptr<Thread>> waiting_threads =
+ GetThreadsWaitingOnAddress(address);
// Determine the modified value depending on the waiting count.
s32 updated_value;
@@ -107,11 +111,11 @@ ResultCode AddressArbiter::ModifyByWaitingCountAndSignalToAddressIfEqual(VAddr a
}
}
- if (static_cast<s32>(Memory::Read32(address)) != value) {
+ if (static_cast<s32>(memory.Read32(address)) != value) {
return ERR_INVALID_STATE;
}
- Memory::Write32(address, static_cast<u32>(updated_value));
+ memory.Write32(address, static_cast<u32>(updated_value));
WakeThreads(waiting_threads, num_to_wake);
return RESULT_SUCCESS;
}
@@ -132,18 +136,20 @@ ResultCode AddressArbiter::WaitForAddress(VAddr address, ArbitrationType type, s
ResultCode AddressArbiter::WaitForAddressIfLessThan(VAddr address, s32 value, s64 timeout,
bool should_decrement) {
+ auto& memory = system.Memory();
+
// Ensure that we can read the address.
- if (!Memory::IsValidVirtualAddress(address)) {
+ if (!memory.IsValidVirtualAddress(address)) {
return ERR_INVALID_ADDRESS_STATE;
}
- const s32 cur_value = static_cast<s32>(Memory::Read32(address));
+ const s32 cur_value = static_cast<s32>(memory.Read32(address));
if (cur_value >= value) {
return ERR_INVALID_STATE;
}
if (should_decrement) {
- Memory::Write32(address, static_cast<u32>(cur_value - 1));
+ memory.Write32(address, static_cast<u32>(cur_value - 1));
}
// Short-circuit without rescheduling, if timeout is zero.
@@ -155,15 +161,19 @@ ResultCode AddressArbiter::WaitForAddressIfLessThan(VAddr address, s32 value, s6
}
ResultCode AddressArbiter::WaitForAddressIfEqual(VAddr address, s32 value, s64 timeout) {
+ auto& memory = system.Memory();
+
// Ensure that we can read the address.
- if (!Memory::IsValidVirtualAddress(address)) {
+ if (!memory.IsValidVirtualAddress(address)) {
return ERR_INVALID_ADDRESS_STATE;
}
+
// Only wait for the address if equal.
- if (static_cast<s32>(Memory::Read32(address)) != value) {
+ if (static_cast<s32>(memory.Read32(address)) != value) {
return ERR_INVALID_STATE;
}
- // Short-circuit without rescheduling, if timeout is zero.
+
+ // Short-circuit without rescheduling if timeout is zero.
if (timeout == 0) {
return RESULT_TIMEOUT;
}
@@ -172,36 +182,62 @@ ResultCode AddressArbiter::WaitForAddressIfEqual(VAddr address, s32 value, s64 t
}
ResultCode AddressArbiter::WaitForAddressImpl(VAddr address, s64 timeout) {
- SharedPtr<Thread> current_thread = system.CurrentScheduler().GetCurrentThread();
+ Thread* current_thread = system.CurrentScheduler().GetCurrentThread();
current_thread->SetArbiterWaitAddress(address);
+ InsertThread(SharedFrom(current_thread));
current_thread->SetStatus(ThreadStatus::WaitArb);
current_thread->InvalidateWakeupCallback();
-
current_thread->WakeAfterDelay(timeout);
system.PrepareReschedule(current_thread->GetProcessorID());
return RESULT_TIMEOUT;
}
-std::vector<SharedPtr<Thread>> AddressArbiter::GetThreadsWaitingOnAddress(VAddr address) const {
-
- // Retrieve all threads that are waiting for this address.
- std::vector<SharedPtr<Thread>> threads;
- const auto& scheduler = system.GlobalScheduler();
- const auto& thread_list = scheduler.GetThreadList();
+void AddressArbiter::HandleWakeupThread(std::shared_ptr<Thread> thread) {
+ ASSERT(thread->GetStatus() == ThreadStatus::WaitArb);
+ RemoveThread(thread);
+ thread->SetArbiterWaitAddress(0);
+}
- for (const auto& thread : thread_list) {
- if (thread->GetArbiterWaitAddress() == address) {
- threads.push_back(thread);
+void AddressArbiter::InsertThread(std::shared_ptr<Thread> thread) {
+ const VAddr arb_addr = thread->GetArbiterWaitAddress();
+ std::list<std::shared_ptr<Thread>>& thread_list = arb_threads[arb_addr];
+ auto it = thread_list.begin();
+ while (it != thread_list.end()) {
+ const std::shared_ptr<Thread>& current_thread = *it;
+ if (current_thread->GetPriority() >= thread->GetPriority()) {
+ thread_list.insert(it, thread);
+ return;
}
+ ++it;
}
+ thread_list.push_back(std::move(thread));
+}
- // Sort them by priority, such that the highest priority ones come first.
- std::sort(threads.begin(), threads.end(),
- [](const SharedPtr<Thread>& lhs, const SharedPtr<Thread>& rhs) {
- return lhs->GetPriority() < rhs->GetPriority();
- });
+void AddressArbiter::RemoveThread(std::shared_ptr<Thread> thread) {
+ const VAddr arb_addr = thread->GetArbiterWaitAddress();
+ std::list<std::shared_ptr<Thread>>& thread_list = arb_threads[arb_addr];
+ auto it = thread_list.begin();
+ while (it != thread_list.end()) {
+ const std::shared_ptr<Thread>& current_thread = *it;
+ if (current_thread.get() == thread.get()) {
+ thread_list.erase(it);
+ return;
+ }
+ ++it;
+ }
+ UNREACHABLE();
+}
- return threads;
+std::vector<std::shared_ptr<Thread>> AddressArbiter::GetThreadsWaitingOnAddress(VAddr address) {
+ std::vector<std::shared_ptr<Thread>> result;
+ std::list<std::shared_ptr<Thread>>& thread_list = arb_threads[address];
+ auto it = thread_list.begin();
+ while (it != thread_list.end()) {
+ std::shared_ptr<Thread> current_thread = *it;
+ result.push_back(std::move(current_thread));
+ ++it;
+ }
+ return result;
}
} // namespace Kernel
diff --git a/src/core/hle/kernel/address_arbiter.h b/src/core/hle/kernel/address_arbiter.h
index ed0d0e69f..386983e54 100644
--- a/src/core/hle/kernel/address_arbiter.h
+++ b/src/core/hle/kernel/address_arbiter.h
@@ -4,10 +4,12 @@
#pragma once
+#include <list>
+#include <memory>
+#include <unordered_map>
#include <vector>
#include "common/common_types.h"
-#include "core/hle/kernel/object.h"
union ResultCode;
@@ -48,6 +50,9 @@ public:
/// Waits on an address with a particular arbitration type.
ResultCode WaitForAddress(VAddr address, ArbitrationType type, s32 value, s64 timeout_ns);
+ /// Removes a thread from the container and resets its address arbiter adress to 0
+ void HandleWakeupThread(std::shared_ptr<Thread> thread);
+
private:
/// Signals an address being waited on.
ResultCode SignalToAddressOnly(VAddr address, s32 num_to_wake);
@@ -71,8 +76,20 @@ private:
// Waits on the given address with a timeout in nanoseconds
ResultCode WaitForAddressImpl(VAddr address, s64 timeout);
+ /// Wake up num_to_wake (or all) threads in a vector.
+ void WakeThreads(const std::vector<std::shared_ptr<Thread>>& waiting_threads, s32 num_to_wake);
+
+ /// Insert a thread into the address arbiter container
+ void InsertThread(std::shared_ptr<Thread> thread);
+
+ /// Removes a thread from the address arbiter container
+ void RemoveThread(std::shared_ptr<Thread> thread);
+
// Gets the threads waiting on an address.
- std::vector<SharedPtr<Thread>> GetThreadsWaitingOnAddress(VAddr address) const;
+ std::vector<std::shared_ptr<Thread>> GetThreadsWaitingOnAddress(VAddr address);
+
+ /// List of threads waiting for a address arbiter
+ std::unordered_map<VAddr, std::list<std::shared_ptr<Thread>>> arb_threads;
Core::System& system;
};
diff --git a/src/core/hle/kernel/client_port.cpp b/src/core/hle/kernel/client_port.cpp
index 744b1697d..5498fd313 100644
--- a/src/core/hle/kernel/client_port.cpp
+++ b/src/core/hle/kernel/client_port.cpp
@@ -8,39 +8,35 @@
#include "core/hle/kernel/hle_ipc.h"
#include "core/hle/kernel/object.h"
#include "core/hle/kernel/server_port.h"
-#include "core/hle/kernel/server_session.h"
+#include "core/hle/kernel/session.h"
namespace Kernel {
ClientPort::ClientPort(KernelCore& kernel) : Object{kernel} {}
ClientPort::~ClientPort() = default;
-SharedPtr<ServerPort> ClientPort::GetServerPort() const {
+std::shared_ptr<ServerPort> ClientPort::GetServerPort() const {
return server_port;
}
-ResultVal<SharedPtr<ClientSession>> ClientPort::Connect() {
- // Note: Threads do not wait for the server endpoint to call
- // AcceptSession before returning from this call.
-
+ResultVal<std::shared_ptr<ClientSession>> ClientPort::Connect() {
if (active_sessions >= max_sessions) {
return ERR_MAX_CONNECTIONS_REACHED;
}
active_sessions++;
- // Create a new session pair, let the created sessions inherit the parent port's HLE handler.
- auto [server, client] = ServerSession::CreateSessionPair(kernel, server_port->GetName(), this);
+ auto [client, server] = Kernel::Session::Create(kernel, name);
if (server_port->HasHLEHandler()) {
- server_port->GetHLEHandler()->ClientConnected(server);
+ server_port->GetHLEHandler()->ClientConnected(std::move(server));
} else {
- server_port->AppendPendingSession(server);
+ server_port->AppendPendingSession(std::move(server));
}
// Wake the threads waiting on the ServerPort
server_port->WakeupAllWaitingThreads();
- return MakeResult(client);
+ return MakeResult(std::move(client));
}
void ClientPort::ConnectionClosed() {
diff --git a/src/core/hle/kernel/client_port.h b/src/core/hle/kernel/client_port.h
index 4921ad4f0..9762bbf0d 100644
--- a/src/core/hle/kernel/client_port.h
+++ b/src/core/hle/kernel/client_port.h
@@ -4,7 +4,9 @@
#pragma once
+#include <memory>
#include <string>
+
#include "common/common_types.h"
#include "core/hle/kernel/object.h"
#include "core/hle/result.h"
@@ -17,6 +19,9 @@ class ServerPort;
class ClientPort final : public Object {
public:
+ explicit ClientPort(KernelCore& kernel);
+ ~ClientPort() override;
+
friend class ServerPort;
std::string GetTypeName() const override {
return "ClientPort";
@@ -30,7 +35,7 @@ public:
return HANDLE_TYPE;
}
- SharedPtr<ServerPort> GetServerPort() const;
+ std::shared_ptr<ServerPort> GetServerPort() const;
/**
* Creates a new Session pair, adds the created ServerSession to the associated ServerPort's
@@ -38,7 +43,7 @@ public:
* waiting on it to awake.
* @returns ClientSession The client endpoint of the created Session pair, or error code.
*/
- ResultVal<SharedPtr<ClientSession>> Connect();
+ ResultVal<std::shared_ptr<ClientSession>> Connect();
/**
* Signifies that a previously active connection has been closed,
@@ -47,10 +52,7 @@ public:
void ConnectionClosed();
private:
- explicit ClientPort(KernelCore& kernel);
- ~ClientPort() override;
-
- SharedPtr<ServerPort> server_port; ///< ServerPort associated with this client port.
+ std::shared_ptr<ServerPort> server_port; ///< ServerPort associated with this client port.
u32 max_sessions = 0; ///< Maximum number of simultaneous sessions the port can have
u32 active_sessions = 0; ///< Number of currently open sessions to this port
std::string name; ///< Name of client port (optional)
diff --git a/src/core/hle/kernel/client_session.cpp b/src/core/hle/kernel/client_session.cpp
index c17baa50a..4669a14ad 100644
--- a/src/core/hle/kernel/client_session.cpp
+++ b/src/core/hle/kernel/client_session.cpp
@@ -1,4 +1,4 @@
-// Copyright 2016 Citra Emulator Project
+// Copyright 2019 yuzu emulator team
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
@@ -12,29 +12,44 @@
namespace Kernel {
-ClientSession::ClientSession(KernelCore& kernel) : Object{kernel} {}
+ClientSession::ClientSession(KernelCore& kernel) : WaitObject{kernel} {}
+
ClientSession::~ClientSession() {
// This destructor will be called automatically when the last ClientSession handle is closed by
// the emulated application.
-
- // A local reference to the ServerSession is necessary to guarantee it
- // will be kept alive until after ClientDisconnected() returns.
- SharedPtr<ServerSession> server = parent->server;
- if (server) {
- server->ClientDisconnected();
+ if (parent->Server()) {
+ parent->Server()->ClientDisconnected();
}
+}
+
+bool ClientSession::ShouldWait(const Thread* thread) const {
+ UNIMPLEMENTED();
+ return {};
+}
- parent->client = nullptr;
+void ClientSession::Acquire(Thread* thread) {
+ UNIMPLEMENTED();
}
-ResultCode ClientSession::SendSyncRequest(SharedPtr<Thread> thread) {
+ResultVal<std::shared_ptr<ClientSession>> ClientSession::Create(KernelCore& kernel,
+ std::shared_ptr<Session> parent,
+ std::string name) {
+ std::shared_ptr<ClientSession> client_session{std::make_shared<ClientSession>(kernel)};
+
+ client_session->name = std::move(name);
+ client_session->parent = std::move(parent);
+
+ return MakeResult(std::move(client_session));
+}
+
+ResultCode ClientSession::SendSyncRequest(std::shared_ptr<Thread> thread, Memory::Memory& memory) {
// Keep ServerSession alive until we're done working with it.
- SharedPtr<ServerSession> server = parent->server;
- if (server == nullptr)
+ if (!parent->Server()) {
return ERR_SESSION_CLOSED_BY_REMOTE;
+ }
// Signal the server session that new data is available
- return server->HandleSyncRequest(std::move(thread));
+ return parent->Server()->HandleSyncRequest(std::move(thread), memory);
}
} // namespace Kernel
diff --git a/src/core/hle/kernel/client_session.h b/src/core/hle/kernel/client_session.h
index 09cdff588..b4289a9a8 100644
--- a/src/core/hle/kernel/client_session.h
+++ b/src/core/hle/kernel/client_session.h
@@ -1,4 +1,4 @@
-// Copyright 2016 Citra Emulator Project
+// Copyright 2019 yuzu emulator team
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
@@ -6,20 +6,28 @@
#include <memory>
#include <string>
-#include "core/hle/kernel/object.h"
+
+#include "core/hle/kernel/wait_object.h"
+#include "core/hle/result.h"
union ResultCode;
+namespace Memory {
+class Memory;
+}
+
namespace Kernel {
class KernelCore;
class Session;
-class ServerSession;
class Thread;
-class ClientSession final : public Object {
+class ClientSession final : public WaitObject {
public:
- friend class ServerSession;
+ explicit ClientSession(KernelCore& kernel);
+ ~ClientSession() override;
+
+ friend class Session;
std::string GetTypeName() const override {
return "ClientSession";
@@ -34,11 +42,16 @@ public:
return HANDLE_TYPE;
}
- ResultCode SendSyncRequest(SharedPtr<Thread> thread);
+ ResultCode SendSyncRequest(std::shared_ptr<Thread> thread, Memory::Memory& memory);
+
+ bool ShouldWait(const Thread* thread) const override;
+
+ void Acquire(Thread* thread) override;
private:
- explicit ClientSession(KernelCore& kernel);
- ~ClientSession() override;
+ static ResultVal<std::shared_ptr<ClientSession>> Create(KernelCore& kernel,
+ std::shared_ptr<Session> parent,
+ std::string name = "Unknown");
/// The parent session, which links to the server endpoint.
std::shared_ptr<Session> parent;
diff --git a/src/core/hle/kernel/handle_table.cpp b/src/core/hle/kernel/handle_table.cpp
index 2cc5d536b..e441a27fc 100644
--- a/src/core/hle/kernel/handle_table.cpp
+++ b/src/core/hle/kernel/handle_table.cpp
@@ -44,7 +44,7 @@ ResultCode HandleTable::SetSize(s32 handle_table_size) {
return RESULT_SUCCESS;
}
-ResultVal<Handle> HandleTable::Create(SharedPtr<Object> obj) {
+ResultVal<Handle> HandleTable::Create(std::shared_ptr<Object> obj) {
DEBUG_ASSERT(obj != nullptr);
const u16 slot = next_free_slot;
@@ -70,7 +70,7 @@ ResultVal<Handle> HandleTable::Create(SharedPtr<Object> obj) {
}
ResultVal<Handle> HandleTable::Duplicate(Handle handle) {
- SharedPtr<Object> object = GetGeneric(handle);
+ std::shared_ptr<Object> object = GetGeneric(handle);
if (object == nullptr) {
LOG_ERROR(Kernel, "Tried to duplicate invalid handle: {:08X}", handle);
return ERR_INVALID_HANDLE;
@@ -99,11 +99,11 @@ bool HandleTable::IsValid(Handle handle) const {
return slot < table_size && objects[slot] != nullptr && generations[slot] == generation;
}
-SharedPtr<Object> HandleTable::GetGeneric(Handle handle) const {
+std::shared_ptr<Object> HandleTable::GetGeneric(Handle handle) const {
if (handle == CurrentThread) {
- return GetCurrentThread();
+ return SharedFrom(GetCurrentThread());
} else if (handle == CurrentProcess) {
- return Core::System::GetInstance().CurrentProcess();
+ return SharedFrom(Core::System::GetInstance().CurrentProcess());
}
if (!IsValid(handle)) {
diff --git a/src/core/hle/kernel/handle_table.h b/src/core/hle/kernel/handle_table.h
index 44901391b..8029660ed 100644
--- a/src/core/hle/kernel/handle_table.h
+++ b/src/core/hle/kernel/handle_table.h
@@ -6,6 +6,8 @@
#include <array>
#include <cstddef>
+#include <memory>
+
#include "common/common_types.h"
#include "core/hle/kernel/object.h"
#include "core/hle/result.h"
@@ -68,7 +70,7 @@ public:
* @return The created Handle or one of the following errors:
* - `ERR_HANDLE_TABLE_FULL`: the maximum number of handles has been exceeded.
*/
- ResultVal<Handle> Create(SharedPtr<Object> obj);
+ ResultVal<Handle> Create(std::shared_ptr<Object> obj);
/**
* Returns a new handle that points to the same object as the passed in handle.
@@ -92,7 +94,7 @@ public:
* Looks up a handle.
* @return Pointer to the looked-up object, or `nullptr` if the handle is not valid.
*/
- SharedPtr<Object> GetGeneric(Handle handle) const;
+ std::shared_ptr<Object> GetGeneric(Handle handle) const;
/**
* Looks up a handle while verifying its type.
@@ -100,7 +102,7 @@ public:
* type differs from the requested one.
*/
template <class T>
- SharedPtr<T> Get(Handle handle) const {
+ std::shared_ptr<T> Get(Handle handle) const {
return DynamicObjectCast<T>(GetGeneric(handle));
}
@@ -109,7 +111,7 @@ public:
private:
/// Stores the Object referenced by the handle or null if the slot is empty.
- std::array<SharedPtr<Object>, MAX_COUNT> objects;
+ std::array<std::shared_ptr<Object>, MAX_COUNT> objects;
/**
* The value of `next_generation` when the handle was created, used to check for validity. For
diff --git a/src/core/hle/kernel/hle_ipc.cpp b/src/core/hle/kernel/hle_ipc.cpp
index a7b5849b0..2db28dcf0 100644
--- a/src/core/hle/kernel/hle_ipc.cpp
+++ b/src/core/hle/kernel/hle_ipc.cpp
@@ -32,23 +32,25 @@ SessionRequestHandler::SessionRequestHandler() = default;
SessionRequestHandler::~SessionRequestHandler() = default;
-void SessionRequestHandler::ClientConnected(SharedPtr<ServerSession> server_session) {
+void SessionRequestHandler::ClientConnected(std::shared_ptr<ServerSession> server_session) {
server_session->SetHleHandler(shared_from_this());
connected_sessions.push_back(std::move(server_session));
}
-void SessionRequestHandler::ClientDisconnected(const SharedPtr<ServerSession>& server_session) {
+void SessionRequestHandler::ClientDisconnected(
+ const std::shared_ptr<ServerSession>& server_session) {
server_session->SetHleHandler(nullptr);
boost::range::remove_erase(connected_sessions, server_session);
}
-SharedPtr<WritableEvent> HLERequestContext::SleepClientThread(
+std::shared_ptr<WritableEvent> HLERequestContext::SleepClientThread(
const std::string& reason, u64 timeout, WakeupCallback&& callback,
- SharedPtr<WritableEvent> writable_event) {
+ std::shared_ptr<WritableEvent> writable_event) {
// Put the client thread to sleep until the wait event is signaled or the timeout expires.
- thread->SetWakeupCallback([context = *this, callback](
- ThreadWakeupReason reason, SharedPtr<Thread> thread,
- SharedPtr<WaitObject> object, std::size_t index) mutable -> bool {
+ thread->SetWakeupCallback([context = *this, callback](ThreadWakeupReason reason,
+ std::shared_ptr<Thread> thread,
+ std::shared_ptr<WaitObject> object,
+ std::size_t index) mutable -> bool {
ASSERT(thread->GetStatus() == ThreadStatus::WaitHLEEvent);
callback(thread, context, reason);
context.WriteToOutgoingCommandBuffer(*thread);
@@ -72,11 +74,13 @@ SharedPtr<WritableEvent> HLERequestContext::SleepClientThread(
thread->WakeAfterDelay(timeout);
}
+ is_thread_waiting = true;
+
return writable_event;
}
-HLERequestContext::HLERequestContext(SharedPtr<Kernel::ServerSession> server_session,
- SharedPtr<Thread> thread)
+HLERequestContext::HLERequestContext(std::shared_ptr<Kernel::ServerSession> server_session,
+ std::shared_ptr<Thread> thread)
: server_session(std::move(server_session)), thread(std::move(thread)) {
cmd_buf[0] = 0;
}
@@ -212,10 +216,11 @@ ResultCode HLERequestContext::PopulateFromIncomingCommandBuffer(const HandleTabl
ResultCode HLERequestContext::WriteToOutgoingCommandBuffer(Thread& thread) {
auto& owner_process = *thread.GetOwnerProcess();
auto& handle_table = owner_process.GetHandleTable();
+ auto& memory = Core::System::GetInstance().Memory();
std::array<u32, IPC::COMMAND_BUFFER_LENGTH> dst_cmdbuf;
- Memory::ReadBlock(owner_process, thread.GetTLSAddress(), dst_cmdbuf.data(),
- dst_cmdbuf.size() * sizeof(u32));
+ memory.ReadBlock(owner_process, thread.GetTLSAddress(), dst_cmdbuf.data(),
+ dst_cmdbuf.size() * sizeof(u32));
// The header was already built in the internal command buffer. Attempt to parse it to verify
// the integrity and then copy it over to the target command buffer.
@@ -271,8 +276,8 @@ ResultCode HLERequestContext::WriteToOutgoingCommandBuffer(Thread& thread) {
}
// Copy the translated command buffer back into the thread's command buffer area.
- Memory::WriteBlock(owner_process, thread.GetTLSAddress(), dst_cmdbuf.data(),
- dst_cmdbuf.size() * sizeof(u32));
+ memory.WriteBlock(owner_process, thread.GetTLSAddress(), dst_cmdbuf.data(),
+ dst_cmdbuf.size() * sizeof(u32));
return RESULT_SUCCESS;
}
@@ -280,15 +285,14 @@ ResultCode HLERequestContext::WriteToOutgoingCommandBuffer(Thread& thread) {
std::vector<u8> HLERequestContext::ReadBuffer(int buffer_index) const {
std::vector<u8> buffer;
const bool is_buffer_a{BufferDescriptorA().size() && BufferDescriptorA()[buffer_index].Size()};
+ auto& memory = Core::System::GetInstance().Memory();
if (is_buffer_a) {
buffer.resize(BufferDescriptorA()[buffer_index].Size());
- Memory::ReadBlock(BufferDescriptorA()[buffer_index].Address(), buffer.data(),
- buffer.size());
+ memory.ReadBlock(BufferDescriptorA()[buffer_index].Address(), buffer.data(), buffer.size());
} else {
buffer.resize(BufferDescriptorX()[buffer_index].Size());
- Memory::ReadBlock(BufferDescriptorX()[buffer_index].Address(), buffer.data(),
- buffer.size());
+ memory.ReadBlock(BufferDescriptorX()[buffer_index].Address(), buffer.data(), buffer.size());
}
return buffer;
@@ -309,10 +313,11 @@ std::size_t HLERequestContext::WriteBuffer(const void* buffer, std::size_t size,
size = buffer_size; // TODO(bunnei): This needs to be HW tested
}
+ auto& memory = Core::System::GetInstance().Memory();
if (is_buffer_b) {
- Memory::WriteBlock(BufferDescriptorB()[buffer_index].Address(), buffer, size);
+ memory.WriteBlock(BufferDescriptorB()[buffer_index].Address(), buffer, size);
} else {
- Memory::WriteBlock(BufferDescriptorC()[buffer_index].Address(), buffer, size);
+ memory.WriteBlock(BufferDescriptorC()[buffer_index].Address(), buffer, size);
}
return size;
diff --git a/src/core/hle/kernel/hle_ipc.h b/src/core/hle/kernel/hle_ipc.h
index ccf5e56aa..050ad8fd7 100644
--- a/src/core/hle/kernel/hle_ipc.h
+++ b/src/core/hle/kernel/hle_ipc.h
@@ -5,6 +5,7 @@
#pragma once
#include <array>
+#include <functional>
#include <memory>
#include <optional>
#include <string>
@@ -60,20 +61,20 @@ public:
* associated ServerSession alive for the duration of the connection.
* @param server_session Owning pointer to the ServerSession associated with the connection.
*/
- void ClientConnected(SharedPtr<ServerSession> server_session);
+ void ClientConnected(std::shared_ptr<ServerSession> server_session);
/**
* Signals that a client has just disconnected from this HLE handler and releases the
* associated ServerSession.
* @param server_session ServerSession associated with the connection.
*/
- void ClientDisconnected(const SharedPtr<ServerSession>& server_session);
+ void ClientDisconnected(const std::shared_ptr<ServerSession>& server_session);
protected:
/// List of sessions that are connected to this handler.
/// A ServerSession whose server endpoint is an HLE implementation is kept alive by this list
/// for the duration of the connection.
- std::vector<SharedPtr<ServerSession>> connected_sessions;
+ std::vector<std::shared_ptr<ServerSession>> connected_sessions;
};
/**
@@ -97,7 +98,8 @@ protected:
*/
class HLERequestContext {
public:
- explicit HLERequestContext(SharedPtr<ServerSession> session, SharedPtr<Thread> thread);
+ explicit HLERequestContext(std::shared_ptr<ServerSession> session,
+ std::shared_ptr<Thread> thread);
~HLERequestContext();
/// Returns a pointer to the IPC command buffer for this request.
@@ -109,12 +111,12 @@ public:
* Returns the session through which this request was made. This can be used as a map key to
* access per-client data on services.
*/
- const SharedPtr<Kernel::ServerSession>& Session() const {
+ const std::shared_ptr<Kernel::ServerSession>& Session() const {
return server_session;
}
- using WakeupCallback = std::function<void(SharedPtr<Thread> thread, HLERequestContext& context,
- ThreadWakeupReason reason)>;
+ using WakeupCallback = std::function<void(
+ std::shared_ptr<Thread> thread, HLERequestContext& context, ThreadWakeupReason reason)>;
/**
* Puts the specified guest thread to sleep until the returned event is signaled or until the
@@ -129,9 +131,9 @@ public:
* created.
* @returns Event that when signaled will resume the thread and call the callback function.
*/
- SharedPtr<WritableEvent> SleepClientThread(const std::string& reason, u64 timeout,
- WakeupCallback&& callback,
- SharedPtr<WritableEvent> writable_event = nullptr);
+ std::shared_ptr<WritableEvent> SleepClientThread(
+ const std::string& reason, u64 timeout, WakeupCallback&& callback,
+ std::shared_ptr<WritableEvent> writable_event = nullptr);
/// Populates this context with data from the requesting process/thread.
ResultCode PopulateFromIncomingCommandBuffer(const HandleTable& handle_table,
@@ -209,20 +211,20 @@ public:
std::size_t GetWriteBufferSize(int buffer_index = 0) const;
template <typename T>
- SharedPtr<T> GetCopyObject(std::size_t index) {
+ std::shared_ptr<T> GetCopyObject(std::size_t index) {
return DynamicObjectCast<T>(copy_objects.at(index));
}
template <typename T>
- SharedPtr<T> GetMoveObject(std::size_t index) {
+ std::shared_ptr<T> GetMoveObject(std::size_t index) {
return DynamicObjectCast<T>(move_objects.at(index));
}
- void AddMoveObject(SharedPtr<Object> object) {
+ void AddMoveObject(std::shared_ptr<Object> object) {
move_objects.emplace_back(std::move(object));
}
- void AddCopyObject(SharedPtr<Object> object) {
+ void AddCopyObject(std::shared_ptr<Object> object) {
copy_objects.emplace_back(std::move(object));
}
@@ -262,15 +264,27 @@ public:
std::string Description() const;
+ Thread& GetThread() {
+ return *thread;
+ }
+
+ const Thread& GetThread() const {
+ return *thread;
+ }
+
+ bool IsThreadWaiting() const {
+ return is_thread_waiting;
+ }
+
private:
void ParseCommandBuffer(const HandleTable& handle_table, u32_le* src_cmdbuf, bool incoming);
std::array<u32, IPC::COMMAND_BUFFER_LENGTH> cmd_buf;
- SharedPtr<Kernel::ServerSession> server_session;
- SharedPtr<Thread> thread;
+ std::shared_ptr<Kernel::ServerSession> server_session;
+ std::shared_ptr<Thread> thread;
// TODO(yuriks): Check common usage of this and optimize size accordingly
- boost::container::small_vector<SharedPtr<Object>, 8> move_objects;
- boost::container::small_vector<SharedPtr<Object>, 8> copy_objects;
+ boost::container::small_vector<std::shared_ptr<Object>, 8> move_objects;
+ boost::container::small_vector<std::shared_ptr<Object>, 8> copy_objects;
boost::container::small_vector<std::shared_ptr<SessionRequestHandler>, 8> domain_objects;
std::optional<IPC::CommandHeader> command_header;
@@ -288,6 +302,7 @@ private:
u32_le command{};
std::vector<std::shared_ptr<SessionRequestHandler>> domain_request_handlers;
+ bool is_thread_waiting{};
};
} // namespace Kernel
diff --git a/src/core/hle/kernel/kernel.cpp b/src/core/hle/kernel/kernel.cpp
index f94ac150d..1d0783bd3 100644
--- a/src/core/hle/kernel/kernel.cpp
+++ b/src/core/hle/kernel/kernel.cpp
@@ -13,7 +13,6 @@
#include "core/core.h"
#include "core/core_timing.h"
#include "core/core_timing_util.h"
-#include "core/hle/kernel/address_arbiter.h"
#include "core/hle/kernel/client_port.h"
#include "core/hle/kernel/errors.h"
#include "core/hle/kernel/handle_table.h"
@@ -40,7 +39,7 @@ static void ThreadWakeupCallback(u64 thread_handle, [[maybe_unused]] s64 cycles_
// Lock the global kernel mutex when we enter the kernel HLE.
std::lock_guard lock{HLE::g_hle_lock};
- SharedPtr<Thread> thread =
+ std::shared_ptr<Thread> thread =
system.Kernel().RetrieveThreadFromWakeupCallbackHandleTable(proper_handle);
if (thread == nullptr) {
LOG_CRITICAL(Kernel, "Callback fired for invalid thread {:08X}", proper_handle);
@@ -53,7 +52,7 @@ static void ThreadWakeupCallback(u64 thread_handle, [[maybe_unused]] s64 cycles_
thread->GetStatus() == ThreadStatus::WaitHLEEvent) {
// Remove the thread from each of its waiting objects' waitlists
for (const auto& object : thread->GetWaitObjects()) {
- object->RemoveWaitingThread(thread.get());
+ object->RemoveWaitingThread(thread);
}
thread->ClearWaitObjects();
@@ -64,8 +63,11 @@ static void ThreadWakeupCallback(u64 thread_handle, [[maybe_unused]] s64 cycles_
} else if (thread->GetStatus() == ThreadStatus::WaitMutex ||
thread->GetStatus() == ThreadStatus::WaitCondVar) {
thread->SetMutexWaitAddress(0);
- thread->SetCondVarWaitAddress(0);
thread->SetWaitHandle(0);
+ if (thread->GetStatus() == ThreadStatus::WaitCondVar) {
+ thread->GetOwnerProcess()->RemoveConditionVariableThread(thread);
+ thread->SetCondVarWaitAddress(0);
+ }
auto* const lock_owner = thread->GetLockOwner();
// Threads waking up by timeout from WaitProcessWideKey do not perform priority inheritance
@@ -76,9 +78,9 @@ static void ThreadWakeupCallback(u64 thread_handle, [[maybe_unused]] s64 cycles_
}
}
- if (thread->GetArbiterWaitAddress() != 0) {
- ASSERT(thread->GetStatus() == ThreadStatus::WaitArb);
- thread->SetArbiterWaitAddress(0);
+ if (thread->GetStatus() == ThreadStatus::WaitArb) {
+ auto& address_arbiter = thread->GetOwnerProcess()->GetAddressArbiter();
+ address_arbiter.HandleWakeupThread(thread);
}
if (resume) {
@@ -136,12 +138,12 @@ struct KernelCore::Impl {
void InitializeThreads() {
thread_wakeup_event_type =
- system.CoreTiming().RegisterEvent("ThreadWakeupCallback", ThreadWakeupCallback);
+ Core::Timing::CreateEvent("ThreadWakeupCallback", ThreadWakeupCallback);
}
void InitializePreemption() {
- preemption_event = system.CoreTiming().RegisterEvent(
- "PreemptionCallback", [this](u64 userdata, s64 cycles_late) {
+ preemption_event =
+ Core::Timing::CreateEvent("PreemptionCallback", [this](u64 userdata, s64 cycles_late) {
global_scheduler.PreemptThreads();
s64 time_interval = Core::Timing::msToCycles(std::chrono::milliseconds(10));
system.CoreTiming().ScheduleEvent(time_interval, preemption_event);
@@ -151,20 +153,31 @@ struct KernelCore::Impl {
system.CoreTiming().ScheduleEvent(time_interval, preemption_event);
}
+ void MakeCurrentProcess(Process* process) {
+ current_process = process;
+
+ if (process == nullptr) {
+ return;
+ }
+
+ system.Memory().SetCurrentPageTable(*process);
+ }
+
std::atomic<u32> next_object_id{0};
std::atomic<u64> next_kernel_process_id{Process::InitialKIPIDMin};
std::atomic<u64> next_user_process_id{Process::ProcessIDMin};
std::atomic<u64> next_thread_id{1};
// Lists all processes that exist in the current session.
- std::vector<SharedPtr<Process>> process_list;
+ std::vector<std::shared_ptr<Process>> process_list;
Process* current_process = nullptr;
Kernel::GlobalScheduler global_scheduler;
- SharedPtr<ResourceLimit> system_resource_limit;
+ std::shared_ptr<ResourceLimit> system_resource_limit;
+
+ std::shared_ptr<Core::Timing::EventType> thread_wakeup_event_type;
+ std::shared_ptr<Core::Timing::EventType> preemption_event;
- Core::Timing::EventType* thread_wakeup_event_type = nullptr;
- Core::Timing::EventType* preemption_event = nullptr;
// TODO(yuriks): This can be removed if Thread objects are explicitly pooled in the future,
// allowing us to simply use a pool index or similar.
Kernel::HandleTable thread_wakeup_callback_handle_table;
@@ -190,26 +203,21 @@ void KernelCore::Shutdown() {
impl->Shutdown();
}
-SharedPtr<ResourceLimit> KernelCore::GetSystemResourceLimit() const {
+std::shared_ptr<ResourceLimit> KernelCore::GetSystemResourceLimit() const {
return impl->system_resource_limit;
}
-SharedPtr<Thread> KernelCore::RetrieveThreadFromWakeupCallbackHandleTable(Handle handle) const {
+std::shared_ptr<Thread> KernelCore::RetrieveThreadFromWakeupCallbackHandleTable(
+ Handle handle) const {
return impl->thread_wakeup_callback_handle_table.Get<Thread>(handle);
}
-void KernelCore::AppendNewProcess(SharedPtr<Process> process) {
+void KernelCore::AppendNewProcess(std::shared_ptr<Process> process) {
impl->process_list.push_back(std::move(process));
}
void KernelCore::MakeCurrentProcess(Process* process) {
- impl->current_process = process;
-
- if (process == nullptr) {
- return;
- }
-
- Memory::SetCurrentPageTable(*process);
+ impl->MakeCurrentProcess(process);
}
Process* KernelCore::CurrentProcess() {
@@ -220,7 +228,7 @@ const Process* KernelCore::CurrentProcess() const {
return impl->current_process;
}
-const std::vector<SharedPtr<Process>>& KernelCore::GetProcessList() const {
+const std::vector<std::shared_ptr<Process>>& KernelCore::GetProcessList() const {
return impl->process_list;
}
@@ -232,7 +240,7 @@ const Kernel::GlobalScheduler& KernelCore::GlobalScheduler() const {
return impl->global_scheduler;
}
-void KernelCore::AddNamedPort(std::string name, SharedPtr<ClientPort> port) {
+void KernelCore::AddNamedPort(std::string name, std::shared_ptr<ClientPort> port) {
impl->named_ports.emplace(std::move(name), std::move(port));
}
@@ -265,7 +273,7 @@ u64 KernelCore::CreateNewUserProcessID() {
return impl->next_user_process_id++;
}
-Core::Timing::EventType* KernelCore::ThreadWakeupCallbackEventType() const {
+const std::shared_ptr<Core::Timing::EventType>& KernelCore::ThreadWakeupCallbackEventType() const {
return impl->thread_wakeup_event_type;
}
diff --git a/src/core/hle/kernel/kernel.h b/src/core/hle/kernel/kernel.h
index c4397fc77..3bf0068ed 100644
--- a/src/core/hle/kernel/kernel.h
+++ b/src/core/hle/kernel/kernel.h
@@ -4,8 +4,10 @@
#pragma once
+#include <memory>
#include <string>
#include <unordered_map>
+#include <vector>
#include "core/hle/kernel/object.h"
namespace Core {
@@ -30,7 +32,7 @@ class Thread;
/// Represents a single instance of the kernel.
class KernelCore {
private:
- using NamedPortTable = std::unordered_map<std::string, SharedPtr<ClientPort>>;
+ using NamedPortTable = std::unordered_map<std::string, std::shared_ptr<ClientPort>>;
public:
/// Constructs an instance of the kernel using the given System
@@ -56,13 +58,13 @@ public:
void Shutdown();
/// Retrieves a shared pointer to the system resource limit instance.
- SharedPtr<ResourceLimit> GetSystemResourceLimit() const;
+ std::shared_ptr<ResourceLimit> GetSystemResourceLimit() const;
/// Retrieves a shared pointer to a Thread instance within the thread wakeup handle table.
- SharedPtr<Thread> RetrieveThreadFromWakeupCallbackHandleTable(Handle handle) const;
+ std::shared_ptr<Thread> RetrieveThreadFromWakeupCallbackHandleTable(Handle handle) const;
/// Adds the given shared pointer to an internal list of active processes.
- void AppendNewProcess(SharedPtr<Process> process);
+ void AppendNewProcess(std::shared_ptr<Process> process);
/// Makes the given process the new current process.
void MakeCurrentProcess(Process* process);
@@ -74,7 +76,7 @@ public:
const Process* CurrentProcess() const;
/// Retrieves the list of processes.
- const std::vector<SharedPtr<Process>>& GetProcessList() const;
+ const std::vector<std::shared_ptr<Process>>& GetProcessList() const;
/// Gets the sole instance of the global scheduler
Kernel::GlobalScheduler& GlobalScheduler();
@@ -83,7 +85,7 @@ public:
const Kernel::GlobalScheduler& GlobalScheduler() const;
/// Adds a port to the named port table
- void AddNamedPort(std::string name, SharedPtr<ClientPort> port);
+ void AddNamedPort(std::string name, std::shared_ptr<ClientPort> port);
/// Finds a port within the named port table with the given name.
NamedPortTable::iterator FindNamedPort(const std::string& name);
@@ -112,7 +114,7 @@ private:
u64 CreateNewThreadID();
/// Retrieves the event type used for thread wakeup callbacks.
- Core::Timing::EventType* ThreadWakeupCallbackEventType() const;
+ const std::shared_ptr<Core::Timing::EventType>& ThreadWakeupCallbackEventType() const;
/// Provides a reference to the thread wakeup callback handle table.
Kernel::HandleTable& ThreadWakeupCallbackHandleTable();
diff --git a/src/core/hle/kernel/mutex.cpp b/src/core/hle/kernel/mutex.cpp
index 663d0f4b6..eff4e45b0 100644
--- a/src/core/hle/kernel/mutex.cpp
+++ b/src/core/hle/kernel/mutex.cpp
@@ -2,6 +2,7 @@
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
+#include <memory>
#include <utility>
#include <vector>
@@ -22,10 +23,10 @@ namespace Kernel {
/// Returns the number of threads that are waiting for a mutex, and the highest priority one among
/// those.
-static std::pair<SharedPtr<Thread>, u32> GetHighestPriorityMutexWaitingThread(
- const SharedPtr<Thread>& current_thread, VAddr mutex_addr) {
+static std::pair<std::shared_ptr<Thread>, u32> GetHighestPriorityMutexWaitingThread(
+ const std::shared_ptr<Thread>& current_thread, VAddr mutex_addr) {
- SharedPtr<Thread> highest_priority_thread;
+ std::shared_ptr<Thread> highest_priority_thread;
u32 num_waiters = 0;
for (const auto& thread : current_thread->GetMutexWaitingThreads()) {
@@ -45,14 +46,14 @@ static std::pair<SharedPtr<Thread>, u32> GetHighestPriorityMutexWaitingThread(
}
/// Update the mutex owner field of all threads waiting on the mutex to point to the new owner.
-static void TransferMutexOwnership(VAddr mutex_addr, SharedPtr<Thread> current_thread,
- SharedPtr<Thread> new_owner) {
+static void TransferMutexOwnership(VAddr mutex_addr, std::shared_ptr<Thread> current_thread,
+ std::shared_ptr<Thread> new_owner) {
const auto threads = current_thread->GetMutexWaitingThreads();
for (const auto& thread : threads) {
if (thread->GetMutexWaitAddress() != mutex_addr)
continue;
- ASSERT(thread->GetLockOwner() == current_thread);
+ ASSERT(thread->GetLockOwner() == current_thread.get());
current_thread->RemoveMutexWaiter(thread);
if (new_owner != thread)
new_owner->AddMutexWaiter(thread);
@@ -70,15 +71,16 @@ ResultCode Mutex::TryAcquire(VAddr address, Handle holding_thread_handle,
}
const auto& handle_table = system.Kernel().CurrentProcess()->GetHandleTable();
- Thread* const current_thread = system.CurrentScheduler().GetCurrentThread();
- SharedPtr<Thread> holding_thread = handle_table.Get<Thread>(holding_thread_handle);
- SharedPtr<Thread> requesting_thread = handle_table.Get<Thread>(requesting_thread_handle);
+ std::shared_ptr<Thread> current_thread =
+ SharedFrom(system.CurrentScheduler().GetCurrentThread());
+ std::shared_ptr<Thread> holding_thread = handle_table.Get<Thread>(holding_thread_handle);
+ std::shared_ptr<Thread> requesting_thread = handle_table.Get<Thread>(requesting_thread_handle);
// TODO(Subv): It is currently unknown if it is possible to lock a mutex in behalf of another
// thread.
ASSERT(requesting_thread == current_thread);
- const u32 addr_value = Memory::Read32(address);
+ const u32 addr_value = system.Memory().Read32(address);
// If the mutex isn't being held, just return success.
if (addr_value != (holding_thread_handle | Mutex::MutexHasWaitersFlag)) {
@@ -110,12 +112,13 @@ ResultCode Mutex::Release(VAddr address) {
return ERR_INVALID_ADDRESS;
}
- auto* const current_thread = system.CurrentScheduler().GetCurrentThread();
+ std::shared_ptr<Thread> current_thread =
+ SharedFrom(system.CurrentScheduler().GetCurrentThread());
auto [thread, num_waiters] = GetHighestPriorityMutexWaitingThread(current_thread, address);
// There are no more threads waiting for the mutex, release it completely.
if (thread == nullptr) {
- Memory::Write32(address, 0);
+ system.Memory().Write32(address, 0);
return RESULT_SUCCESS;
}
@@ -130,7 +133,7 @@ ResultCode Mutex::Release(VAddr address) {
}
// Grant the mutex to the next waiting thread and resume it.
- Memory::Write32(address, mutex_value);
+ system.Memory().Write32(address, mutex_value);
ASSERT(thread->GetStatus() == ThreadStatus::WaitMutex);
thread->ResumeFromWait();
diff --git a/src/core/hle/kernel/object.cpp b/src/core/hle/kernel/object.cpp
index 10431e94c..2c571792b 100644
--- a/src/core/hle/kernel/object.cpp
+++ b/src/core/hle/kernel/object.cpp
@@ -27,6 +27,7 @@ bool Object::IsWaitable() const {
case HandleType::ResourceLimit:
case HandleType::ClientPort:
case HandleType::ClientSession:
+ case HandleType::Session:
return false;
}
diff --git a/src/core/hle/kernel/object.h b/src/core/hle/kernel/object.h
index a6faeb83b..e3391e2af 100644
--- a/src/core/hle/kernel/object.h
+++ b/src/core/hle/kernel/object.h
@@ -5,10 +5,9 @@
#pragma once
#include <atomic>
+#include <memory>
#include <string>
-#include <boost/smart_ptr/intrusive_ptr.hpp>
-
#include "common/common_types.h"
namespace Kernel {
@@ -30,9 +29,10 @@ enum class HandleType : u32 {
ServerPort,
ClientSession,
ServerSession,
+ Session,
};
-class Object : NonCopyable {
+class Object : NonCopyable, public std::enable_shared_from_this<Object> {
public:
explicit Object(KernelCore& kernel);
virtual ~Object();
@@ -61,35 +61,24 @@ protected:
KernelCore& kernel;
private:
- friend void intrusive_ptr_add_ref(Object*);
- friend void intrusive_ptr_release(Object*);
-
- std::atomic<u32> ref_count{0};
std::atomic<u32> object_id{0};
};
-// Special functions used by boost::instrusive_ptr to do automatic ref-counting
-inline void intrusive_ptr_add_ref(Object* object) {
- object->ref_count.fetch_add(1, std::memory_order_relaxed);
-}
-
-inline void intrusive_ptr_release(Object* object) {
- if (object->ref_count.fetch_sub(1, std::memory_order_acq_rel) == 1) {
- delete object;
- }
-}
-
template <typename T>
-using SharedPtr = boost::intrusive_ptr<T>;
+std::shared_ptr<T> SharedFrom(T* raw) {
+ if (raw == nullptr)
+ return nullptr;
+ return std::static_pointer_cast<T>(raw->shared_from_this());
+}
/**
* Attempts to downcast the given Object pointer to a pointer to T.
* @return Derived pointer to the object, or `nullptr` if `object` isn't of type T.
*/
template <typename T>
-inline SharedPtr<T> DynamicObjectCast(SharedPtr<Object> object) {
+inline std::shared_ptr<T> DynamicObjectCast(std::shared_ptr<Object> object) {
if (object != nullptr && object->GetHandleType() == T::HANDLE_TYPE) {
- return boost::static_pointer_cast<T>(object);
+ return std::static_pointer_cast<T>(object);
}
return nullptr;
}
diff --git a/src/core/hle/kernel/process.cpp b/src/core/hle/kernel/process.cpp
index 12a900bcc..12ea4ebe3 100644
--- a/src/core/hle/kernel/process.cpp
+++ b/src/core/hle/kernel/process.cpp
@@ -38,7 +38,7 @@ void SetupMainThread(Process& owner_process, KernelCore& kernel, u32 priority) {
auto thread_res = Thread::Create(kernel, "main", entry_point, priority, 0,
owner_process.GetIdealCore(), stack_top, owner_process);
- SharedPtr<Thread> thread = std::move(thread_res).Unwrap();
+ std::shared_ptr<Thread> thread = std::move(thread_res).Unwrap();
// Register 1 must be a handle to the main thread
const Handle thread_handle = owner_process.GetHandleTable().Create(thread).Unwrap();
@@ -100,10 +100,10 @@ private:
std::bitset<num_slot_entries> is_slot_used;
};
-SharedPtr<Process> Process::Create(Core::System& system, std::string name, ProcessType type) {
+std::shared_ptr<Process> Process::Create(Core::System& system, std::string name, ProcessType type) {
auto& kernel = system.Kernel();
- SharedPtr<Process> process(new Process(system));
+ std::shared_ptr<Process> process = std::make_shared<Process>(system);
process->name = std::move(name);
process->resource_limit = kernel.GetSystemResourceLimit();
process->status = ProcessStatus::Created;
@@ -121,7 +121,7 @@ SharedPtr<Process> Process::Create(Core::System& system, std::string name, Proce
return process;
}
-SharedPtr<ResourceLimit> Process::GetResourceLimit() const {
+std::shared_ptr<ResourceLimit> Process::GetResourceLimit() const {
return resource_limit;
}
@@ -142,6 +142,49 @@ u64 Process::GetTotalPhysicalMemoryUsedWithoutSystemResource() const {
return GetTotalPhysicalMemoryUsed() - GetSystemResourceUsage();
}
+void Process::InsertConditionVariableThread(std::shared_ptr<Thread> thread) {
+ VAddr cond_var_addr = thread->GetCondVarWaitAddress();
+ std::list<std::shared_ptr<Thread>>& thread_list = cond_var_threads[cond_var_addr];
+ auto it = thread_list.begin();
+ while (it != thread_list.end()) {
+ const std::shared_ptr<Thread> current_thread = *it;
+ if (current_thread->GetPriority() > thread->GetPriority()) {
+ thread_list.insert(it, thread);
+ return;
+ }
+ ++it;
+ }
+ thread_list.push_back(thread);
+}
+
+void Process::RemoveConditionVariableThread(std::shared_ptr<Thread> thread) {
+ VAddr cond_var_addr = thread->GetCondVarWaitAddress();
+ std::list<std::shared_ptr<Thread>>& thread_list = cond_var_threads[cond_var_addr];
+ auto it = thread_list.begin();
+ while (it != thread_list.end()) {
+ const std::shared_ptr<Thread> current_thread = *it;
+ if (current_thread.get() == thread.get()) {
+ thread_list.erase(it);
+ return;
+ }
+ ++it;
+ }
+ UNREACHABLE();
+}
+
+std::vector<std::shared_ptr<Thread>> Process::GetConditionVariableThreads(
+ const VAddr cond_var_addr) {
+ std::vector<std::shared_ptr<Thread>> result{};
+ std::list<std::shared_ptr<Thread>>& thread_list = cond_var_threads[cond_var_addr];
+ auto it = thread_list.begin();
+ while (it != thread_list.end()) {
+ std::shared_ptr<Thread> current_thread = *it;
+ result.push_back(current_thread);
+ ++it;
+ }
+ return result;
+}
+
void Process::RegisterThread(const Thread* thread) {
thread_list.push_back(thread);
}
@@ -197,12 +240,12 @@ void Process::Run(s32 main_thread_priority, u64 stack_size) {
void Process::PrepareForTermination() {
ChangeStatus(ProcessStatus::Exiting);
- const auto stop_threads = [this](const std::vector<SharedPtr<Thread>>& thread_list) {
+ const auto stop_threads = [this](const std::vector<std::shared_ptr<Thread>>& thread_list) {
for (auto& thread : thread_list) {
if (thread->GetOwnerProcess() != this)
continue;
- if (thread == system.CurrentScheduler().GetCurrentThread())
+ if (thread.get() == system.CurrentScheduler().GetCurrentThread())
continue;
// TODO(Subv): When are the other running/ready threads terminated?
diff --git a/src/core/hle/kernel/process.h b/src/core/hle/kernel/process.h
index c2df451f3..3483fa19d 100644
--- a/src/core/hle/kernel/process.h
+++ b/src/core/hle/kernel/process.h
@@ -8,6 +8,7 @@
#include <cstddef>
#include <list>
#include <string>
+#include <unordered_map>
#include <vector>
#include "common/common_types.h"
#include "core/hle/kernel/address_arbiter.h"
@@ -61,6 +62,9 @@ enum class ProcessStatus {
class Process final : public WaitObject {
public:
+ explicit Process(Core::System& system);
+ ~Process() override;
+
enum : u64 {
/// Lowest allowed process ID for a kernel initial process.
InitialKIPIDMin = 1,
@@ -81,7 +85,8 @@ public:
static constexpr std::size_t RANDOM_ENTROPY_SIZE = 4;
- static SharedPtr<Process> Create(Core::System& system, std::string name, ProcessType type);
+ static std::shared_ptr<Process> Create(Core::System& system, std::string name,
+ ProcessType type);
std::string GetTypeName() const override {
return "Process";
@@ -156,7 +161,7 @@ public:
}
/// Gets the resource limit descriptor for this process
- SharedPtr<ResourceLimit> GetResourceLimit() const;
+ std::shared_ptr<ResourceLimit> GetResourceLimit() const;
/// Gets the ideal CPU core ID for this process
u8 GetIdealCore() const {
@@ -232,6 +237,15 @@ public:
return thread_list;
}
+ /// Insert a thread into the condition variable wait container
+ void InsertConditionVariableThread(std::shared_ptr<Thread> thread);
+
+ /// Remove a thread from the condition variable wait container
+ void RemoveConditionVariableThread(std::shared_ptr<Thread> thread);
+
+ /// Obtain all condition variable threads waiting for some address
+ std::vector<std::shared_ptr<Thread>> GetConditionVariableThreads(VAddr cond_var_addr);
+
/// Registers a thread as being created under this process,
/// adding it to this process' thread list.
void RegisterThread(const Thread* thread);
@@ -287,9 +301,6 @@ public:
void FreeTLSRegion(VAddr tls_address);
private:
- explicit Process(Core::System& system);
- ~Process() override;
-
/// Checks if the specified thread should wait until this process is available.
bool ShouldWait(const Thread* thread) const override;
@@ -328,7 +339,7 @@ private:
u32 system_resource_size = 0;
/// Resource limit descriptor for this process
- SharedPtr<ResourceLimit> resource_limit;
+ std::shared_ptr<ResourceLimit> resource_limit;
/// The ideal CPU core for this process, threads are scheduled on this core by default.
u8 ideal_core = 0;
@@ -375,6 +386,9 @@ private:
/// List of threads that are running with this process as their owner.
std::list<const Thread*> thread_list;
+ /// List of threads waiting for a condition variable
+ std::unordered_map<VAddr, std::list<std::shared_ptr<Thread>>> cond_var_threads;
+
/// System context
Core::System& system;
diff --git a/src/core/hle/kernel/resource_limit.cpp b/src/core/hle/kernel/resource_limit.cpp
index 173f69915..b53423462 100644
--- a/src/core/hle/kernel/resource_limit.cpp
+++ b/src/core/hle/kernel/resource_limit.cpp
@@ -16,8 +16,8 @@ constexpr std::size_t ResourceTypeToIndex(ResourceType type) {
ResourceLimit::ResourceLimit(KernelCore& kernel) : Object{kernel} {}
ResourceLimit::~ResourceLimit() = default;
-SharedPtr<ResourceLimit> ResourceLimit::Create(KernelCore& kernel) {
- return new ResourceLimit(kernel);
+std::shared_ptr<ResourceLimit> ResourceLimit::Create(KernelCore& kernel) {
+ return std::make_shared<ResourceLimit>(kernel);
}
s64 ResourceLimit::GetCurrentResourceValue(ResourceType resource) const {
diff --git a/src/core/hle/kernel/resource_limit.h b/src/core/hle/kernel/resource_limit.h
index 2613a6bb5..53b89e621 100644
--- a/src/core/hle/kernel/resource_limit.h
+++ b/src/core/hle/kernel/resource_limit.h
@@ -5,6 +5,8 @@
#pragma once
#include <array>
+#include <memory>
+
#include "common/common_types.h"
#include "core/hle/kernel/object.h"
@@ -31,8 +33,11 @@ constexpr bool IsValidResourceType(ResourceType type) {
class ResourceLimit final : public Object {
public:
+ explicit ResourceLimit(KernelCore& kernel);
+ ~ResourceLimit() override;
+
/// Creates a resource limit object.
- static SharedPtr<ResourceLimit> Create(KernelCore& kernel);
+ static std::shared_ptr<ResourceLimit> Create(KernelCore& kernel);
std::string GetTypeName() const override {
return "ResourceLimit";
@@ -76,9 +81,6 @@ public:
ResultCode SetLimitValue(ResourceType resource, s64 value);
private:
- explicit ResourceLimit(KernelCore& kernel);
- ~ResourceLimit() override;
-
// TODO(Subv): Increment resource limit current values in their respective Kernel::T::Create
// functions
//
diff --git a/src/core/hle/kernel/scheduler.cpp b/src/core/hle/kernel/scheduler.cpp
index 0e2dbf13e..d36fcd7d9 100644
--- a/src/core/hle/kernel/scheduler.cpp
+++ b/src/core/hle/kernel/scheduler.cpp
@@ -26,27 +26,27 @@ GlobalScheduler::GlobalScheduler(Core::System& system) : system{system} {}
GlobalScheduler::~GlobalScheduler() = default;
-void GlobalScheduler::AddThread(SharedPtr<Thread> thread) {
+void GlobalScheduler::AddThread(std::shared_ptr<Thread> thread) {
thread_list.push_back(std::move(thread));
}
-void GlobalScheduler::RemoveThread(const Thread* thread) {
+void GlobalScheduler::RemoveThread(std::shared_ptr<Thread> thread) {
thread_list.erase(std::remove(thread_list.begin(), thread_list.end(), thread),
thread_list.end());
}
-void GlobalScheduler::UnloadThread(s32 core) {
+void GlobalScheduler::UnloadThread(std::size_t core) {
Scheduler& sched = system.Scheduler(core);
sched.UnloadThread();
}
-void GlobalScheduler::SelectThread(u32 core) {
+void GlobalScheduler::SelectThread(std::size_t core) {
const auto update_thread = [](Thread* thread, Scheduler& sched) {
- if (thread != sched.selected_thread) {
+ if (thread != sched.selected_thread.get()) {
if (thread == nullptr) {
++sched.idle_selection_count;
}
- sched.selected_thread = thread;
+ sched.selected_thread = SharedFrom(thread);
}
sched.is_context_switch_pending = sched.selected_thread != sched.current_thread;
std::atomic_thread_fence(std::memory_order_seq_cst);
@@ -77,9 +77,9 @@ void GlobalScheduler::SelectThread(u32 core) {
// if we got a suggested thread, select it, else do a second pass.
if (winner && winner->GetPriority() > 2) {
if (winner->IsRunning()) {
- UnloadThread(winner->GetProcessorID());
+ UnloadThread(static_cast<u32>(winner->GetProcessorID()));
}
- TransferToCore(winner->GetPriority(), core, winner);
+ TransferToCore(winner->GetPriority(), static_cast<s32>(core), winner);
update_thread(winner, sched);
return;
}
@@ -91,9 +91,9 @@ void GlobalScheduler::SelectThread(u32 core) {
Thread* thread_on_core = scheduled_queue[src_core].front();
Thread* to_change = *it;
if (thread_on_core->IsRunning() || to_change->IsRunning()) {
- UnloadThread(src_core);
+ UnloadThread(static_cast<u32>(src_core));
}
- TransferToCore(thread_on_core->GetPriority(), core, thread_on_core);
+ TransferToCore(thread_on_core->GetPriority(), static_cast<s32>(core), thread_on_core);
current_thread = thread_on_core;
break;
}
@@ -154,9 +154,9 @@ bool GlobalScheduler::YieldThreadAndBalanceLoad(Thread* yielding_thread) {
if (winner != nullptr) {
if (winner != yielding_thread) {
if (winner->IsRunning()) {
- UnloadThread(winner->GetProcessorID());
+ UnloadThread(static_cast<u32>(winner->GetProcessorID()));
}
- TransferToCore(winner->GetPriority(), core_id, winner);
+ TransferToCore(winner->GetPriority(), s32(core_id), winner);
}
} else {
winner = next_thread;
@@ -196,9 +196,9 @@ bool GlobalScheduler::YieldThreadAndWaitForLoadBalancing(Thread* yielding_thread
if (winner != nullptr) {
if (winner != yielding_thread) {
if (winner->IsRunning()) {
- UnloadThread(winner->GetProcessorID());
+ UnloadThread(static_cast<u32>(winner->GetProcessorID()));
}
- TransferToCore(winner->GetPriority(), core_id, winner);
+ TransferToCore(winner->GetPriority(), static_cast<s32>(core_id), winner);
}
} else {
winner = yielding_thread;
@@ -248,7 +248,7 @@ void GlobalScheduler::PreemptThreads() {
if (winner != nullptr) {
if (winner->IsRunning()) {
- UnloadThread(winner->GetProcessorID());
+ UnloadThread(static_cast<u32>(winner->GetProcessorID()));
}
TransferToCore(winner->GetPriority(), s32(core_id), winner);
current_thread =
@@ -281,7 +281,7 @@ void GlobalScheduler::PreemptThreads() {
if (winner != nullptr) {
if (winner->IsRunning()) {
- UnloadThread(winner->GetProcessorID());
+ UnloadThread(static_cast<u32>(winner->GetProcessorID()));
}
TransferToCore(winner->GetPriority(), s32(core_id), winner);
current_thread = winner;
@@ -292,30 +292,30 @@ void GlobalScheduler::PreemptThreads() {
}
}
-void GlobalScheduler::Suggest(u32 priority, u32 core, Thread* thread) {
+void GlobalScheduler::Suggest(u32 priority, std::size_t core, Thread* thread) {
suggested_queue[core].add(thread, priority);
}
-void GlobalScheduler::Unsuggest(u32 priority, u32 core, Thread* thread) {
+void GlobalScheduler::Unsuggest(u32 priority, std::size_t core, Thread* thread) {
suggested_queue[core].remove(thread, priority);
}
-void GlobalScheduler::Schedule(u32 priority, u32 core, Thread* thread) {
+void GlobalScheduler::Schedule(u32 priority, std::size_t core, Thread* thread) {
ASSERT_MSG(thread->GetProcessorID() == s32(core), "Thread must be assigned to this core.");
scheduled_queue[core].add(thread, priority);
}
-void GlobalScheduler::SchedulePrepend(u32 priority, u32 core, Thread* thread) {
+void GlobalScheduler::SchedulePrepend(u32 priority, std::size_t core, Thread* thread) {
ASSERT_MSG(thread->GetProcessorID() == s32(core), "Thread must be assigned to this core.");
scheduled_queue[core].add(thread, priority, false);
}
-void GlobalScheduler::Reschedule(u32 priority, u32 core, Thread* thread) {
+void GlobalScheduler::Reschedule(u32 priority, std::size_t core, Thread* thread) {
scheduled_queue[core].remove(thread, priority);
scheduled_queue[core].add(thread, priority);
}
-void GlobalScheduler::Unschedule(u32 priority, u32 core, Thread* thread) {
+void GlobalScheduler::Unschedule(u32 priority, std::size_t core, Thread* thread) {
scheduled_queue[core].remove(thread, priority);
}
@@ -327,14 +327,14 @@ void GlobalScheduler::TransferToCore(u32 priority, s32 destination_core, Thread*
}
thread->SetProcessorID(destination_core);
if (source_core >= 0) {
- Unschedule(priority, source_core, thread);
+ Unschedule(priority, static_cast<u32>(source_core), thread);
}
if (destination_core >= 0) {
- Unsuggest(priority, destination_core, thread);
- Schedule(priority, destination_core, thread);
+ Unsuggest(priority, static_cast<u32>(destination_core), thread);
+ Schedule(priority, static_cast<u32>(destination_core), thread);
}
if (source_core >= 0) {
- Suggest(priority, source_core, thread);
+ Suggest(priority, static_cast<u32>(source_core), thread);
}
}
@@ -357,7 +357,7 @@ void GlobalScheduler::Shutdown() {
thread_list.clear();
}
-Scheduler::Scheduler(Core::System& system, Core::ARM_Interface& cpu_core, u32 core_id)
+Scheduler::Scheduler(Core::System& system, Core::ARM_Interface& cpu_core, std::size_t core_id)
: system(system), cpu_core(cpu_core), core_id(core_id) {}
Scheduler::~Scheduler() = default;
@@ -446,7 +446,7 @@ void Scheduler::SwitchContext() {
// Cancel any outstanding wakeup events for this thread
new_thread->CancelWakeupTimer();
- current_thread = new_thread;
+ current_thread = SharedFrom(new_thread);
new_thread->SetStatus(ThreadStatus::Running);
new_thread->SetIsRunning(true);
@@ -458,7 +458,6 @@ void Scheduler::SwitchContext() {
cpu_core.LoadContext(new_thread->GetContext());
cpu_core.SetTlsAddress(new_thread->GetTLSAddress());
cpu_core.SetTPIDR_EL0(new_thread->GetTPIDR_EL0());
- cpu_core.ClearExclusiveState();
} else {
current_thread = nullptr;
// Note: We do not reset the current process and current page table when idling because
diff --git a/src/core/hle/kernel/scheduler.h b/src/core/hle/kernel/scheduler.h
index f2d6311b8..14b77960a 100644
--- a/src/core/hle/kernel/scheduler.h
+++ b/src/core/hle/kernel/scheduler.h
@@ -4,11 +4,12 @@
#pragma once
-#include <mutex>
+#include <atomic>
+#include <memory>
#include <vector>
+
#include "common/common_types.h"
#include "common/multi_level_queue.h"
-#include "core/hle/kernel/object.h"
#include "core/hle/kernel/thread.h"
namespace Core {
@@ -28,13 +29,13 @@ public:
~GlobalScheduler();
/// Adds a new thread to the scheduler
- void AddThread(SharedPtr<Thread> thread);
+ void AddThread(std::shared_ptr<Thread> thread);
/// Removes a thread from the scheduler
- void RemoveThread(const Thread* thread);
+ void RemoveThread(std::shared_ptr<Thread> thread);
/// Returns a list of all threads managed by the scheduler
- const std::vector<SharedPtr<Thread>>& GetThreadList() const {
+ const std::vector<std::shared_ptr<Thread>>& GetThreadList() const {
return thread_list;
}
@@ -42,41 +43,34 @@ public:
* Add a thread to the suggested queue of a cpu core. Suggested threads may be
* picked if no thread is scheduled to run on the core.
*/
- void Suggest(u32 priority, u32 core, Thread* thread);
+ void Suggest(u32 priority, std::size_t core, Thread* thread);
/**
* Remove a thread to the suggested queue of a cpu core. Suggested threads may be
* picked if no thread is scheduled to run on the core.
*/
- void Unsuggest(u32 priority, u32 core, Thread* thread);
+ void Unsuggest(u32 priority, std::size_t core, Thread* thread);
/**
* Add a thread to the scheduling queue of a cpu core. The thread is added at the
* back the queue in its priority level.
*/
- void Schedule(u32 priority, u32 core, Thread* thread);
+ void Schedule(u32 priority, std::size_t core, Thread* thread);
/**
* Add a thread to the scheduling queue of a cpu core. The thread is added at the
* front the queue in its priority level.
*/
- void SchedulePrepend(u32 priority, u32 core, Thread* thread);
+ void SchedulePrepend(u32 priority, std::size_t core, Thread* thread);
/// Reschedule an already scheduled thread based on a new priority
- void Reschedule(u32 priority, u32 core, Thread* thread);
+ void Reschedule(u32 priority, std::size_t core, Thread* thread);
/// Unschedules a thread.
- void Unschedule(u32 priority, u32 core, Thread* thread);
-
- /**
- * Transfers a thread into an specific core. If the destination_core is -1
- * it will be unscheduled from its source code and added into its suggested
- * queue.
- */
- void TransferToCore(u32 priority, s32 destination_core, Thread* thread);
+ void Unschedule(u32 priority, std::size_t core, Thread* thread);
/// Selects a core and forces it to unload its current thread's context
- void UnloadThread(s32 core);
+ void UnloadThread(std::size_t core);
/**
* Takes care of selecting the new scheduled thread in three steps:
@@ -90,9 +84,9 @@ public:
* 3. Third is no suggested thread is found, we do a second pass and pick a running
* thread in another core and swap it with its current thread.
*/
- void SelectThread(u32 core);
+ void SelectThread(std::size_t core);
- bool HaveReadyThreads(u32 core_id) const {
+ bool HaveReadyThreads(std::size_t core_id) const {
return !scheduled_queue[core_id].empty();
}
@@ -145,6 +139,13 @@ public:
void Shutdown();
private:
+ /**
+ * Transfers a thread into an specific core. If the destination_core is -1
+ * it will be unscheduled from its source code and added into its suggested
+ * queue.
+ */
+ void TransferToCore(u32 priority, s32 destination_core, Thread* thread);
+
bool AskForReselectionOrMarkRedundant(Thread* current_thread, const Thread* winner);
static constexpr u32 min_regular_priority = 2;
@@ -157,13 +158,13 @@ private:
std::array<u32, NUM_CPU_CORES> preemption_priorities = {59, 59, 59, 62};
/// Lists all thread ids that aren't deleted/etc.
- std::vector<SharedPtr<Thread>> thread_list;
+ std::vector<std::shared_ptr<Thread>> thread_list;
Core::System& system;
};
class Scheduler final {
public:
- explicit Scheduler(Core::System& system, Core::ARM_Interface& cpu_core, u32 core_id);
+ explicit Scheduler(Core::System& system, Core::ARM_Interface& cpu_core, std::size_t core_id);
~Scheduler();
/// Returns whether there are any threads that are ready to run.
@@ -213,14 +214,14 @@ private:
*/
void UpdateLastContextSwitchTime(Thread* thread, Process* process);
- SharedPtr<Thread> current_thread = nullptr;
- SharedPtr<Thread> selected_thread = nullptr;
+ std::shared_ptr<Thread> current_thread = nullptr;
+ std::shared_ptr<Thread> selected_thread = nullptr;
Core::System& system;
Core::ARM_Interface& cpu_core;
u64 last_context_switch_time = 0;
u64 idle_selection_count = 0;
- const u32 core_id;
+ const std::size_t core_id;
bool is_context_switch_pending = false;
};
diff --git a/src/core/hle/kernel/server_port.cpp b/src/core/hle/kernel/server_port.cpp
index 02e7c60e6..a4ccfa35e 100644
--- a/src/core/hle/kernel/server_port.cpp
+++ b/src/core/hle/kernel/server_port.cpp
@@ -16,7 +16,7 @@ namespace Kernel {
ServerPort::ServerPort(KernelCore& kernel) : WaitObject{kernel} {}
ServerPort::~ServerPort() = default;
-ResultVal<SharedPtr<ServerSession>> ServerPort::Accept() {
+ResultVal<std::shared_ptr<ServerSession>> ServerPort::Accept() {
if (pending_sessions.empty()) {
return ERR_NOT_FOUND;
}
@@ -26,7 +26,7 @@ ResultVal<SharedPtr<ServerSession>> ServerPort::Accept() {
return MakeResult(std::move(session));
}
-void ServerPort::AppendPendingSession(SharedPtr<ServerSession> pending_session) {
+void ServerPort::AppendPendingSession(std::shared_ptr<ServerSession> pending_session) {
pending_sessions.push_back(std::move(pending_session));
}
@@ -41,8 +41,8 @@ void ServerPort::Acquire(Thread* thread) {
ServerPort::PortPair ServerPort::CreatePortPair(KernelCore& kernel, u32 max_sessions,
std::string name) {
- SharedPtr<ServerPort> server_port(new ServerPort(kernel));
- SharedPtr<ClientPort> client_port(new ClientPort(kernel));
+ std::shared_ptr<ServerPort> server_port = std::make_shared<ServerPort>(kernel);
+ std::shared_ptr<ClientPort> client_port = std::make_shared<ClientPort>(kernel);
server_port->name = name + "_Server";
client_port->name = name + "_Client";
diff --git a/src/core/hle/kernel/server_port.h b/src/core/hle/kernel/server_port.h
index dc88a1ebd..8be8a75ea 100644
--- a/src/core/hle/kernel/server_port.h
+++ b/src/core/hle/kernel/server_port.h
@@ -22,8 +22,11 @@ class SessionRequestHandler;
class ServerPort final : public WaitObject {
public:
+ explicit ServerPort(KernelCore& kernel);
+ ~ServerPort() override;
+
using HLEHandler = std::shared_ptr<SessionRequestHandler>;
- using PortPair = std::pair<SharedPtr<ServerPort>, SharedPtr<ClientPort>>;
+ using PortPair = std::pair<std::shared_ptr<ServerPort>, std::shared_ptr<ClientPort>>;
/**
* Creates a pair of ServerPort and an associated ClientPort.
@@ -52,7 +55,7 @@ public:
* Accepts a pending incoming connection on this port. If there are no pending sessions, will
* return ERR_NO_PENDING_SESSIONS.
*/
- ResultVal<SharedPtr<ServerSession>> Accept();
+ ResultVal<std::shared_ptr<ServerSession>> Accept();
/// Whether or not this server port has an HLE handler available.
bool HasHLEHandler() const {
@@ -74,17 +77,14 @@ public:
/// Appends a ServerSession to the collection of ServerSessions
/// waiting to be accepted by this port.
- void AppendPendingSession(SharedPtr<ServerSession> pending_session);
+ void AppendPendingSession(std::shared_ptr<ServerSession> pending_session);
bool ShouldWait(const Thread* thread) const override;
void Acquire(Thread* thread) override;
private:
- explicit ServerPort(KernelCore& kernel);
- ~ServerPort() override;
-
/// ServerSessions waiting to be accepted by the port
- std::vector<SharedPtr<ServerSession>> pending_sessions;
+ std::vector<std::shared_ptr<ServerSession>> pending_sessions;
/// This session's HLE request handler template (optional)
/// ServerSessions created from this port inherit a reference to this handler.
diff --git a/src/core/hle/kernel/server_session.cpp b/src/core/hle/kernel/server_session.cpp
index 30b2bfb5a..7825e1ec4 100644
--- a/src/core/hle/kernel/server_session.cpp
+++ b/src/core/hle/kernel/server_session.cpp
@@ -1,4 +1,4 @@
-// Copyright 2016 Citra Emulator Project
+// Copyright 2019 yuzu emulator team
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
@@ -9,6 +9,7 @@
#include "common/common_types.h"
#include "common/logging/log.h"
#include "core/core.h"
+#include "core/core_timing.h"
#include "core/hle/ipc_helpers.h"
#include "core/hle/kernel/client_port.h"
#include "core/hle/kernel/client_session.h"
@@ -19,35 +20,32 @@
#include "core/hle/kernel/server_session.h"
#include "core/hle/kernel/session.h"
#include "core/hle/kernel/thread.h"
+#include "core/memory.h"
namespace Kernel {
ServerSession::ServerSession(KernelCore& kernel) : WaitObject{kernel} {}
-ServerSession::~ServerSession() {
- // This destructor will be called automatically when the last ServerSession handle is closed by
- // the emulated application.
+ServerSession::~ServerSession() = default;
- // Decrease the port's connection count.
- if (parent->port) {
- parent->port->ConnectionClosed();
- }
-
- parent->server = nullptr;
-}
+ResultVal<std::shared_ptr<ServerSession>> ServerSession::Create(KernelCore& kernel,
+ std::shared_ptr<Session> parent,
+ std::string name) {
+ std::shared_ptr<ServerSession> session{std::make_shared<ServerSession>(kernel)};
-ResultVal<SharedPtr<ServerSession>> ServerSession::Create(KernelCore& kernel, std::string name) {
- SharedPtr<ServerSession> server_session(new ServerSession(kernel));
+ session->request_event = Core::Timing::CreateEvent(
+ name, [session](u64 userdata, s64 cycles_late) { session->CompleteSyncRequest(); });
+ session->name = std::move(name);
+ session->parent = std::move(parent);
- server_session->name = std::move(name);
- server_session->parent = nullptr;
-
- return MakeResult(std::move(server_session));
+ return MakeResult(std::move(session));
}
bool ServerSession::ShouldWait(const Thread* thread) const {
// Closed sessions should never wait, an error will be returned from svcReplyAndReceive.
- if (parent->client == nullptr)
+ if (!parent->Client()) {
return false;
+ }
+
// Wait if we have no pending requests, or if we're currently handling a request.
return pending_requesting_threads.empty() || currently_handling != nullptr;
}
@@ -69,7 +67,7 @@ void ServerSession::ClientDisconnected() {
if (handler) {
// Note that after this returns, this server session's hle_handler is
// invalidated (set to null).
- handler->ClientDisconnected(this);
+ handler->ClientDisconnected(SharedFrom(this));
}
// Clean up the list of client threads with pending requests, they are unneeded now that the
@@ -126,13 +124,21 @@ ResultCode ServerSession::HandleDomainSyncRequest(Kernel::HLERequestContext& con
return RESULT_SUCCESS;
}
-ResultCode ServerSession::HandleSyncRequest(SharedPtr<Thread> thread) {
- // The ServerSession received a sync request, this means that there's new data available
- // from its ClientSession, so wake up any threads that may be waiting on a svcReplyAndReceive or
- // similar.
- Kernel::HLERequestContext context(this, thread);
- u32* cmd_buf = (u32*)Memory::GetPointer(thread->GetTLSAddress());
- context.PopulateFromIncomingCommandBuffer(kernel.CurrentProcess()->GetHandleTable(), cmd_buf);
+ResultCode ServerSession::QueueSyncRequest(std::shared_ptr<Thread> thread, Memory::Memory& memory) {
+ u32* cmd_buf{reinterpret_cast<u32*>(memory.GetPointer(thread->GetTLSAddress()))};
+ std::shared_ptr<Kernel::HLERequestContext> context{
+ std::make_shared<Kernel::HLERequestContext>(SharedFrom(this), std::move(thread))};
+
+ context->PopulateFromIncomingCommandBuffer(kernel.CurrentProcess()->GetHandleTable(), cmd_buf);
+ request_queue.Push(std::move(context));
+
+ return RESULT_SUCCESS;
+}
+
+ResultCode ServerSession::CompleteSyncRequest() {
+ ASSERT(!request_queue.Empty());
+
+ auto& context = *request_queue.Front();
ResultCode result = RESULT_SUCCESS;
// If the session has been converted to a domain, handle the domain request
@@ -144,61 +150,27 @@ ResultCode ServerSession::HandleSyncRequest(SharedPtr<Thread> thread) {
result = hle_handler->HandleSyncRequest(context);
}
- if (thread->GetStatus() == ThreadStatus::Running) {
- // Put the thread to sleep until the server replies, it will be awoken in
- // svcReplyAndReceive for LLE servers.
- thread->SetStatus(ThreadStatus::WaitIPC);
-
- if (hle_handler != nullptr) {
- // For HLE services, we put the request threads to sleep for a short duration to
- // simulate IPC overhead, but only if the HLE handler didn't put the thread to sleep for
- // other reasons like an async callback. The IPC overhead is needed to prevent
- // starvation when a thread only does sync requests to HLE services while a
- // lower-priority thread is waiting to run.
-
- // This delay was approximated in a homebrew application by measuring the average time
- // it takes for svcSendSyncRequest to return when performing the SetLcdForceBlack IPC
- // request to the GSP:GPU service in a n3DS with firmware 11.6. The measured values have
- // a high variance and vary between models.
- static constexpr u64 IPCDelayNanoseconds = 39000;
- thread->WakeAfterDelay(IPCDelayNanoseconds);
- } else {
- // Add the thread to the list of threads that have issued a sync request with this
- // server.
- pending_requesting_threads.push_back(std::move(thread));
- }
- }
-
- // If this ServerSession does not have an HLE implementation, just wake up the threads waiting
- // on it.
- WakeupAllWaitingThreads();
-
- // Handle scenario when ConvertToDomain command was issued, as we must do the conversion at the
- // end of the command such that only commands following this one are handled as domains
if (convert_to_domain) {
ASSERT_MSG(IsSession(), "ServerSession is already a domain instance.");
domain_request_handlers = {hle_handler};
convert_to_domain = false;
}
- return result;
-}
-
-ServerSession::SessionPair ServerSession::CreateSessionPair(KernelCore& kernel,
- const std::string& name,
- SharedPtr<ClientPort> port) {
- auto server_session = ServerSession::Create(kernel, name + "_Server").Unwrap();
- SharedPtr<ClientSession> client_session(new ClientSession(kernel));
- client_session->name = name + "_Client";
+ // Some service requests require the thread to block
+ if (!context.IsThreadWaiting()) {
+ context.GetThread().ResumeFromWait();
+ context.GetThread().SetWaitSynchronizationResult(result);
+ }
- std::shared_ptr<Session> parent(new Session);
- parent->client = client_session.get();
- parent->server = server_session.get();
- parent->port = std::move(port);
+ request_queue.Pop();
- client_session->parent = parent;
- server_session->parent = parent;
+ return result;
+}
- return std::make_pair(std::move(server_session), std::move(client_session));
+ResultCode ServerSession::HandleSyncRequest(std::shared_ptr<Thread> thread,
+ Memory::Memory& memory) {
+ Core::System::GetInstance().CoreTiming().ScheduleEvent(20000, request_event, {});
+ return QueueSyncRequest(std::move(thread), memory);
}
+
} // namespace Kernel
diff --git a/src/core/hle/kernel/server_session.h b/src/core/hle/kernel/server_session.h
index 738df30f8..d6e48109e 100644
--- a/src/core/hle/kernel/server_session.h
+++ b/src/core/hle/kernel/server_session.h
@@ -1,4 +1,4 @@
-// Copyright 2014 Citra Emulator Project
+// Copyright 2019 yuzu emulator team
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
@@ -9,17 +9,22 @@
#include <utility>
#include <vector>
-#include "core/hle/kernel/object.h"
+#include "common/threadsafe_queue.h"
#include "core/hle/kernel/wait_object.h"
#include "core/hle/result.h"
+namespace Memory {
+class Memory;
+}
+
+namespace Core::Timing {
+struct EventType;
+}
+
namespace Kernel {
-class ClientPort;
-class ClientSession;
class HLERequestContext;
class KernelCore;
-class ServerSession;
class Session;
class SessionRequestHandler;
class Thread;
@@ -38,6 +43,15 @@ class Thread;
*/
class ServerSession final : public WaitObject {
public:
+ explicit ServerSession(KernelCore& kernel);
+ ~ServerSession() override;
+
+ friend class Session;
+
+ static ResultVal<std::shared_ptr<ServerSession>> Create(KernelCore& kernel,
+ std::shared_ptr<Session> parent,
+ std::string name = "Unknown");
+
std::string GetTypeName() const override {
return "ServerSession";
}
@@ -59,18 +73,6 @@ public:
return parent.get();
}
- using SessionPair = std::pair<SharedPtr<ServerSession>, SharedPtr<ClientSession>>;
-
- /**
- * Creates a pair of ServerSession and an associated ClientSession.
- * @param kernel The kernal instance to create the session pair under.
- * @param name Optional name of the ports.
- * @param client_port Optional The ClientPort that spawned this session.
- * @return The created session tuple
- */
- static SessionPair CreateSessionPair(KernelCore& kernel, const std::string& name = "Unknown",
- SharedPtr<ClientPort> client_port = nullptr);
-
/**
* Sets the HLE handler for the session. This handler will be called to service IPC requests
* instead of the regular IPC machinery. (The regular IPC machinery is currently not
@@ -82,10 +84,13 @@ public:
/**
* Handle a sync request from the emulated application.
+ *
* @param thread Thread that initiated the request.
+ * @param memory Memory context to handle the sync request under.
+ *
* @returns ResultCode from the operation.
*/
- ResultCode HandleSyncRequest(SharedPtr<Thread> thread);
+ ResultCode HandleSyncRequest(std::shared_ptr<Thread> thread, Memory::Memory& memory);
bool ShouldWait(const Thread* thread) const override;
@@ -118,18 +123,11 @@ public:
}
private:
- explicit ServerSession(KernelCore& kernel);
- ~ServerSession() override;
+ /// Queues a sync request from the emulated application.
+ ResultCode QueueSyncRequest(std::shared_ptr<Thread> thread, Memory::Memory& memory);
- /**
- * Creates a server session. The server session can have an optional HLE handler,
- * which will be invoked to handle the IPC requests that this session receives.
- * @param kernel The kernel instance to create this server session under.
- * @param name Optional name of the server session.
- * @return The created server session
- */
- static ResultVal<SharedPtr<ServerSession>> Create(KernelCore& kernel,
- std::string name = "Unknown");
+ /// Completes a sync request from the emulated application.
+ ResultCode CompleteSyncRequest();
/// Handles a SyncRequest to a domain, forwarding the request to the proper object or closing an
/// object handle.
@@ -147,18 +145,24 @@ private:
/// List of threads that are pending a response after a sync request. This list is processed in
/// a LIFO manner, thus, the last request will be dispatched first.
/// TODO(Subv): Verify if this is indeed processed in LIFO using a hardware test.
- std::vector<SharedPtr<Thread>> pending_requesting_threads;
+ std::vector<std::shared_ptr<Thread>> pending_requesting_threads;
/// Thread whose request is currently being handled. A request is considered "handled" when a
/// response is sent via svcReplyAndReceive.
/// TODO(Subv): Find a better name for this.
- SharedPtr<Thread> currently_handling;
+ std::shared_ptr<Thread> currently_handling;
/// When set to True, converts the session to a domain at the end of the command
bool convert_to_domain{};
/// The name of this session (optional)
std::string name;
+
+ /// Core timing event used to schedule the service request at some point in the future
+ std::shared_ptr<Core::Timing::EventType> request_event;
+
+ /// Queue of scheduled service requests
+ Common::MPSCQueue<std::shared_ptr<Kernel::HLERequestContext>> request_queue;
};
} // namespace Kernel
diff --git a/src/core/hle/kernel/session.cpp b/src/core/hle/kernel/session.cpp
index 642914744..dee6e2b72 100644
--- a/src/core/hle/kernel/session.cpp
+++ b/src/core/hle/kernel/session.cpp
@@ -1,12 +1,36 @@
-// Copyright 2015 Citra Emulator Project
+// Copyright 2019 yuzu emulator team
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
+#include "common/assert.h"
+#include "core/hle/kernel/client_session.h"
+#include "core/hle/kernel/server_session.h"
#include "core/hle/kernel/session.h"
-#include "core/hle/kernel/thread.h"
namespace Kernel {
-Session::Session() {}
-Session::~Session() {}
+Session::Session(KernelCore& kernel) : WaitObject{kernel} {}
+Session::~Session() = default;
+
+Session::SessionPair Session::Create(KernelCore& kernel, std::string name) {
+ auto session{std::make_shared<Session>(kernel)};
+ auto client_session{Kernel::ClientSession::Create(kernel, session, name + "_Client").Unwrap()};
+ auto server_session{Kernel::ServerSession::Create(kernel, session, name + "_Server").Unwrap()};
+
+ session->name = std::move(name);
+ session->client = client_session;
+ session->server = server_session;
+
+ return std::make_pair(std::move(client_session), std::move(server_session));
+}
+
+bool Session::ShouldWait(const Thread* thread) const {
+ UNIMPLEMENTED();
+ return {};
+}
+
+void Session::Acquire(Thread* thread) {
+ UNIMPLEMENTED();
+}
+
} // namespace Kernel
diff --git a/src/core/hle/kernel/session.h b/src/core/hle/kernel/session.h
index 7a551f5e4..15a5ac15f 100644
--- a/src/core/hle/kernel/session.h
+++ b/src/core/hle/kernel/session.h
@@ -1,27 +1,64 @@
-// Copyright 2018 yuzu emulator team
+// Copyright 2019 yuzu emulator team
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
#pragma once
-#include "core/hle/kernel/object.h"
+#include <memory>
+#include <string>
+#include <utility>
+
+#include "core/hle/kernel/wait_object.h"
namespace Kernel {
class ClientSession;
-class ClientPort;
class ServerSession;
/**
* Parent structure to link the client and server endpoints of a session with their associated
- * client port. The client port need not exist, as is the case for portless sessions like the
- * FS File and Directory sessions. When one of the endpoints of a session is destroyed, its
- * corresponding field in this structure will be set to nullptr.
+ * client port.
*/
-class Session final {
+class Session final : public WaitObject {
public:
- ClientSession* client = nullptr; ///< The client endpoint of the session.
- ServerSession* server = nullptr; ///< The server endpoint of the session.
- SharedPtr<ClientPort> port; ///< The port that this session is associated with (optional).
+ explicit Session(KernelCore& kernel);
+ ~Session() override;
+
+ using SessionPair = std::pair<std::shared_ptr<ClientSession>, std::shared_ptr<ServerSession>>;
+
+ static SessionPair Create(KernelCore& kernel, std::string name = "Unknown");
+
+ std::string GetName() const override {
+ return name;
+ }
+
+ static constexpr HandleType HANDLE_TYPE = HandleType::Session;
+ HandleType GetHandleType() const override {
+ return HANDLE_TYPE;
+ }
+
+ bool ShouldWait(const Thread* thread) const override;
+
+ void Acquire(Thread* thread) override;
+
+ std::shared_ptr<ClientSession> Client() {
+ if (auto result{client.lock()}) {
+ return result;
+ }
+ return {};
+ }
+
+ std::shared_ptr<ServerSession> Server() {
+ if (auto result{server.lock()}) {
+ return result;
+ }
+ return {};
+ }
+
+private:
+ std::string name;
+ std::weak_ptr<ClientSession> client;
+ std::weak_ptr<ServerSession> server;
};
+
} // namespace Kernel
diff --git a/src/core/hle/kernel/shared_memory.cpp b/src/core/hle/kernel/shared_memory.cpp
index a815c4eea..afb2e3fc2 100644
--- a/src/core/hle/kernel/shared_memory.cpp
+++ b/src/core/hle/kernel/shared_memory.cpp
@@ -15,11 +15,12 @@ namespace Kernel {
SharedMemory::SharedMemory(KernelCore& kernel) : Object{kernel} {}
SharedMemory::~SharedMemory() = default;
-SharedPtr<SharedMemory> SharedMemory::Create(KernelCore& kernel, Process* owner_process, u64 size,
- MemoryPermission permissions,
- MemoryPermission other_permissions, VAddr address,
- MemoryRegion region, std::string name) {
- SharedPtr<SharedMemory> shared_memory(new SharedMemory(kernel));
+std::shared_ptr<SharedMemory> SharedMemory::Create(KernelCore& kernel, Process* owner_process,
+ u64 size, MemoryPermission permissions,
+ MemoryPermission other_permissions,
+ VAddr address, MemoryRegion region,
+ std::string name) {
+ std::shared_ptr<SharedMemory> shared_memory = std::make_shared<SharedMemory>(kernel);
shared_memory->owner_process = owner_process;
shared_memory->name = std::move(name);
@@ -58,10 +59,10 @@ SharedPtr<SharedMemory> SharedMemory::Create(KernelCore& kernel, Process* owner_
return shared_memory;
}
-SharedPtr<SharedMemory> SharedMemory::CreateForApplet(
+std::shared_ptr<SharedMemory> SharedMemory::CreateForApplet(
KernelCore& kernel, std::shared_ptr<Kernel::PhysicalMemory> heap_block, std::size_t offset,
u64 size, MemoryPermission permissions, MemoryPermission other_permissions, std::string name) {
- SharedPtr<SharedMemory> shared_memory(new SharedMemory(kernel));
+ std::shared_ptr<SharedMemory> shared_memory = std::make_shared<SharedMemory>(kernel);
shared_memory->owner_process = nullptr;
shared_memory->name = std::move(name);
diff --git a/src/core/hle/kernel/shared_memory.h b/src/core/hle/kernel/shared_memory.h
index 01ca6dcd2..014951d82 100644
--- a/src/core/hle/kernel/shared_memory.h
+++ b/src/core/hle/kernel/shared_memory.h
@@ -6,7 +6,6 @@
#include <memory>
#include <string>
-#include <vector>
#include "common/common_types.h"
#include "core/hle/kernel/object.h"
@@ -33,6 +32,9 @@ enum class MemoryPermission : u32 {
class SharedMemory final : public Object {
public:
+ explicit SharedMemory(KernelCore& kernel);
+ ~SharedMemory() override;
+
/**
* Creates a shared memory object.
* @param kernel The kernel instance to create a shared memory instance under.
@@ -46,11 +48,12 @@ public:
* linear heap.
* @param name Optional object name, used for debugging purposes.
*/
- static SharedPtr<SharedMemory> Create(KernelCore& kernel, Process* owner_process, u64 size,
- MemoryPermission permissions,
- MemoryPermission other_permissions, VAddr address = 0,
- MemoryRegion region = MemoryRegion::BASE,
- std::string name = "Unknown");
+ static std::shared_ptr<SharedMemory> Create(KernelCore& kernel, Process* owner_process,
+ u64 size, MemoryPermission permissions,
+ MemoryPermission other_permissions,
+ VAddr address = 0,
+ MemoryRegion region = MemoryRegion::BASE,
+ std::string name = "Unknown");
/**
* Creates a shared memory object from a block of memory managed by an HLE applet.
@@ -63,7 +66,7 @@ public:
* block.
* @param name Optional object name, used for debugging purposes.
*/
- static SharedPtr<SharedMemory> CreateForApplet(
+ static std::shared_ptr<SharedMemory> CreateForApplet(
KernelCore& kernel, std::shared_ptr<Kernel::PhysicalMemory> heap_block, std::size_t offset,
u64 size, MemoryPermission permissions, MemoryPermission other_permissions,
std::string name = "Unknown Applet");
@@ -130,9 +133,6 @@ public:
const u8* GetPointer(std::size_t offset = 0) const;
private:
- explicit SharedMemory(KernelCore& kernel);
- ~SharedMemory() override;
-
/// Backing memory for this shared memory block.
std::shared_ptr<PhysicalMemory> backing_block;
/// Offset into the backing block for this shared memory.
diff --git a/src/core/hle/kernel/svc.cpp b/src/core/hle/kernel/svc.cpp
index c63a9ba8b..dbcdb0b88 100644
--- a/src/core/hle/kernel/svc.cpp
+++ b/src/core/hle/kernel/svc.cpp
@@ -17,6 +17,7 @@
#include "core/core.h"
#include "core/core_cpu.h"
#include "core/core_timing.h"
+#include "core/core_timing_util.h"
#include "core/hle/kernel/address_arbiter.h"
#include "core/hle/kernel/client_port.h"
#include "core/hle/kernel/client_session.h"
@@ -331,7 +332,9 @@ static ResultCode UnmapMemory(Core::System& system, VAddr dst_addr, VAddr src_ad
/// Connect to an OS service given the port name, returns the handle to the port to out
static ResultCode ConnectToNamedPort(Core::System& system, Handle* out_handle,
VAddr port_name_address) {
- if (!Memory::IsValidVirtualAddress(port_name_address)) {
+ auto& memory = system.Memory();
+
+ if (!memory.IsValidVirtualAddress(port_name_address)) {
LOG_ERROR(Kernel_SVC,
"Port Name Address is not a valid virtual address, port_name_address=0x{:016X}",
port_name_address);
@@ -340,7 +343,7 @@ static ResultCode ConnectToNamedPort(Core::System& system, Handle* out_handle,
static constexpr std::size_t PortNameMaxLength = 11;
// Read 1 char beyond the max allowed port name to detect names that are too long.
- std::string port_name = Memory::ReadCString(port_name_address, PortNameMaxLength + 1);
+ const std::string port_name = memory.ReadCString(port_name_address, PortNameMaxLength + 1);
if (port_name.size() > PortNameMaxLength) {
LOG_ERROR(Kernel_SVC, "Port name is too long, expected {} but got {}", PortNameMaxLength,
port_name.size());
@@ -358,7 +361,7 @@ static ResultCode ConnectToNamedPort(Core::System& system, Handle* out_handle,
auto client_port = it->second;
- SharedPtr<ClientSession> client_session;
+ std::shared_ptr<ClientSession> client_session;
CASCADE_RESULT(client_session, client_port->Connect());
// Return the client session
@@ -370,7 +373,7 @@ static ResultCode ConnectToNamedPort(Core::System& system, Handle* out_handle,
/// Makes a blocking IPC call to an OS service.
static ResultCode SendSyncRequest(Core::System& system, Handle handle) {
const auto& handle_table = system.Kernel().CurrentProcess()->GetHandleTable();
- SharedPtr<ClientSession> session = handle_table.Get<ClientSession>(handle);
+ std::shared_ptr<ClientSession> session = handle_table.Get<ClientSession>(handle);
if (!session) {
LOG_ERROR(Kernel_SVC, "called with invalid handle=0x{:08X}", handle);
return ERR_INVALID_HANDLE;
@@ -378,11 +381,12 @@ static ResultCode SendSyncRequest(Core::System& system, Handle handle) {
LOG_TRACE(Kernel_SVC, "called handle=0x{:08X}({})", handle, session->GetName());
- system.PrepareReschedule();
+ auto thread = system.CurrentScheduler().GetCurrentThread();
+ thread->InvalidateWakeupCallback();
+ thread->SetStatus(ThreadStatus::WaitIPC);
+ system.PrepareReschedule(thread->GetProcessorID());
- // TODO(Subv): svcSendSyncRequest should put the caller thread to sleep while the server
- // responds and cause a reschedule.
- return session->SendSyncRequest(system.CurrentScheduler().GetCurrentThread());
+ return session->SendSyncRequest(SharedFrom(thread), system.Memory());
}
/// Get the ID for the specified thread.
@@ -390,7 +394,7 @@ static ResultCode GetThreadId(Core::System& system, u64* thread_id, Handle threa
LOG_TRACE(Kernel_SVC, "called thread=0x{:08X}", thread_handle);
const auto& handle_table = system.Kernel().CurrentProcess()->GetHandleTable();
- const SharedPtr<Thread> thread = handle_table.Get<Thread>(thread_handle);
+ const std::shared_ptr<Thread> thread = handle_table.Get<Thread>(thread_handle);
if (!thread) {
LOG_ERROR(Kernel_SVC, "Thread handle does not exist, handle=0x{:08X}", thread_handle);
return ERR_INVALID_HANDLE;
@@ -405,13 +409,13 @@ static ResultCode GetProcessId(Core::System& system, u64* process_id, Handle han
LOG_DEBUG(Kernel_SVC, "called handle=0x{:08X}", handle);
const auto& handle_table = system.Kernel().CurrentProcess()->GetHandleTable();
- const SharedPtr<Process> process = handle_table.Get<Process>(handle);
+ const std::shared_ptr<Process> process = handle_table.Get<Process>(handle);
if (process) {
*process_id = process->GetProcessID();
return RESULT_SUCCESS;
}
- const SharedPtr<Thread> thread = handle_table.Get<Thread>(handle);
+ const std::shared_ptr<Thread> thread = handle_table.Get<Thread>(handle);
if (thread) {
const Process* const owner_process = thread->GetOwnerProcess();
if (!owner_process) {
@@ -430,8 +434,8 @@ static ResultCode GetProcessId(Core::System& system, u64* process_id, Handle han
}
/// Default thread wakeup callback for WaitSynchronization
-static bool DefaultThreadWakeupCallback(ThreadWakeupReason reason, SharedPtr<Thread> thread,
- SharedPtr<WaitObject> object, std::size_t index) {
+static bool DefaultThreadWakeupCallback(ThreadWakeupReason reason, std::shared_ptr<Thread> thread,
+ std::shared_ptr<WaitObject> object, std::size_t index) {
ASSERT(thread->GetStatus() == ThreadStatus::WaitSynch);
if (reason == ThreadWakeupReason::Timeout) {
@@ -451,7 +455,8 @@ static ResultCode WaitSynchronization(Core::System& system, Handle* index, VAddr
LOG_TRACE(Kernel_SVC, "called handles_address=0x{:X}, handle_count={}, nano_seconds={}",
handles_address, handle_count, nano_seconds);
- if (!Memory::IsValidVirtualAddress(handles_address)) {
+ auto& memory = system.Memory();
+ if (!memory.IsValidVirtualAddress(handles_address)) {
LOG_ERROR(Kernel_SVC,
"Handle address is not a valid virtual address, handle_address=0x{:016X}",
handles_address);
@@ -473,7 +478,7 @@ static ResultCode WaitSynchronization(Core::System& system, Handle* index, VAddr
const auto& handle_table = system.Kernel().CurrentProcess()->GetHandleTable();
for (u64 i = 0; i < handle_count; ++i) {
- const Handle handle = Memory::Read32(handles_address + i * sizeof(Handle));
+ const Handle handle = memory.Read32(handles_address + i * sizeof(Handle));
const auto object = handle_table.Get<WaitObject>(handle);
if (object == nullptr) {
@@ -505,8 +510,13 @@ static ResultCode WaitSynchronization(Core::System& system, Handle* index, VAddr
return RESULT_TIMEOUT;
}
+ if (thread->IsSyncCancelled()) {
+ thread->SetSyncCancelled(false);
+ return ERR_SYNCHRONIZATION_CANCELED;
+ }
+
for (auto& object : objects) {
- object->AddWaitingThread(thread);
+ object->AddWaitingThread(SharedFrom(thread));
}
thread->SetWaitObjects(std::move(objects));
@@ -526,7 +536,7 @@ static ResultCode CancelSynchronization(Core::System& system, Handle thread_hand
LOG_TRACE(Kernel_SVC, "called thread=0x{:X}", thread_handle);
const auto& handle_table = system.Kernel().CurrentProcess()->GetHandleTable();
- SharedPtr<Thread> thread = handle_table.Get<Thread>(thread_handle);
+ std::shared_ptr<Thread> thread = handle_table.Get<Thread>(thread_handle);
if (!thread) {
LOG_ERROR(Kernel_SVC, "Thread handle does not exist, thread_handle=0x{:08X}",
thread_handle);
@@ -610,13 +620,15 @@ static void Break(Core::System& system, u32 reason, u64 info1, u64 info2) {
return;
}
+ auto& memory = system.Memory();
+
// This typically is an error code so we're going to assume this is the case
if (sz == sizeof(u32)) {
- LOG_CRITICAL(Debug_Emulated, "debug_buffer_err_code={:X}", Memory::Read32(addr));
+ LOG_CRITICAL(Debug_Emulated, "debug_buffer_err_code={:X}", memory.Read32(addr));
} else {
// We don't know what's in here so we'll hexdump it
debug_buffer.resize(sz);
- Memory::ReadBlock(addr, debug_buffer.data(), sz);
+ memory.ReadBlock(addr, debug_buffer.data(), sz);
std::string hexdump;
for (std::size_t i = 0; i < debug_buffer.size(); i++) {
hexdump += fmt::format("{:02X} ", debug_buffer[i]);
@@ -706,7 +718,7 @@ static void OutputDebugString([[maybe_unused]] Core::System& system, VAddr addre
}
std::string str(len, '\0');
- Memory::ReadBlock(address, str.data(), str.size());
+ system.Memory().ReadBlock(address, str.data(), str.size());
LOG_DEBUG(Debug_Emulated, "{}", str);
}
@@ -935,7 +947,7 @@ static ResultCode GetInfo(Core::System& system, u64* result, u64 info_id, u64 ha
const auto& core_timing = system.CoreTiming();
const auto& scheduler = system.CurrentScheduler();
const auto* const current_thread = scheduler.GetCurrentThread();
- const bool same_thread = current_thread == thread;
+ const bool same_thread = current_thread == thread.get();
const u64 prev_ctx_ticks = scheduler.GetLastContextSwitchTicks();
u64 out_ticks = 0;
@@ -1045,7 +1057,7 @@ static ResultCode SetThreadActivity(Core::System& system, Handle handle, u32 act
}
const auto* current_process = system.Kernel().CurrentProcess();
- const SharedPtr<Thread> thread = current_process->GetHandleTable().Get<Thread>(handle);
+ const std::shared_ptr<Thread> thread = current_process->GetHandleTable().Get<Thread>(handle);
if (!thread) {
LOG_ERROR(Kernel_SVC, "Thread handle does not exist, handle=0x{:08X}", handle);
return ERR_INVALID_HANDLE;
@@ -1061,7 +1073,7 @@ static ResultCode SetThreadActivity(Core::System& system, Handle handle, u32 act
return ERR_INVALID_HANDLE;
}
- if (thread == system.CurrentScheduler().GetCurrentThread()) {
+ if (thread.get() == system.CurrentScheduler().GetCurrentThread()) {
LOG_ERROR(Kernel_SVC, "The thread handle specified is the current running thread");
return ERR_BUSY;
}
@@ -1077,7 +1089,7 @@ static ResultCode GetThreadContext(Core::System& system, VAddr thread_context, H
LOG_DEBUG(Kernel_SVC, "called, context=0x{:08X}, thread=0x{:X}", thread_context, handle);
const auto* current_process = system.Kernel().CurrentProcess();
- const SharedPtr<Thread> thread = current_process->GetHandleTable().Get<Thread>(handle);
+ const std::shared_ptr<Thread> thread = current_process->GetHandleTable().Get<Thread>(handle);
if (!thread) {
LOG_ERROR(Kernel_SVC, "Thread handle does not exist, handle=0x{:08X}", handle);
return ERR_INVALID_HANDLE;
@@ -1093,7 +1105,7 @@ static ResultCode GetThreadContext(Core::System& system, VAddr thread_context, H
return ERR_INVALID_HANDLE;
}
- if (thread == system.CurrentScheduler().GetCurrentThread()) {
+ if (thread.get() == system.CurrentScheduler().GetCurrentThread()) {
LOG_ERROR(Kernel_SVC, "The thread handle specified is the current running thread");
return ERR_BUSY;
}
@@ -1109,7 +1121,7 @@ static ResultCode GetThreadContext(Core::System& system, VAddr thread_context, H
std::fill(ctx.vector_registers.begin() + 16, ctx.vector_registers.end(), u128{});
}
- Memory::WriteBlock(thread_context, &ctx, sizeof(ctx));
+ system.Memory().WriteBlock(thread_context, &ctx, sizeof(ctx));
return RESULT_SUCCESS;
}
@@ -1118,7 +1130,7 @@ static ResultCode GetThreadPriority(Core::System& system, u32* priority, Handle
LOG_TRACE(Kernel_SVC, "called");
const auto& handle_table = system.Kernel().CurrentProcess()->GetHandleTable();
- const SharedPtr<Thread> thread = handle_table.Get<Thread>(handle);
+ const std::shared_ptr<Thread> thread = handle_table.Get<Thread>(handle);
if (!thread) {
LOG_ERROR(Kernel_SVC, "Thread handle does not exist, handle=0x{:08X}", handle);
return ERR_INVALID_HANDLE;
@@ -1142,7 +1154,7 @@ static ResultCode SetThreadPriority(Core::System& system, Handle handle, u32 pri
const auto* const current_process = system.Kernel().CurrentProcess();
- SharedPtr<Thread> thread = current_process->GetHandleTable().Get<Thread>(handle);
+ std::shared_ptr<Thread> thread = current_process->GetHandleTable().Get<Thread>(handle);
if (!thread) {
LOG_ERROR(Kernel_SVC, "Thread handle does not exist, handle=0x{:08X}", handle);
return ERR_INVALID_HANDLE;
@@ -1262,27 +1274,28 @@ static ResultCode QueryProcessMemory(Core::System& system, VAddr memory_info_add
VAddr address) {
LOG_TRACE(Kernel_SVC, "called process=0x{:08X} address={:X}", process_handle, address);
const auto& handle_table = system.Kernel().CurrentProcess()->GetHandleTable();
- SharedPtr<Process> process = handle_table.Get<Process>(process_handle);
+ std::shared_ptr<Process> process = handle_table.Get<Process>(process_handle);
if (!process) {
LOG_ERROR(Kernel_SVC, "Process handle does not exist, process_handle=0x{:08X}",
process_handle);
return ERR_INVALID_HANDLE;
}
+ auto& memory = system.Memory();
const auto& vm_manager = process->VMManager();
const MemoryInfo memory_info = vm_manager.QueryMemory(address);
- Memory::Write64(memory_info_address, memory_info.base_address);
- Memory::Write64(memory_info_address + 8, memory_info.size);
- Memory::Write32(memory_info_address + 16, memory_info.state);
- Memory::Write32(memory_info_address + 20, memory_info.attributes);
- Memory::Write32(memory_info_address + 24, memory_info.permission);
- Memory::Write32(memory_info_address + 32, memory_info.ipc_ref_count);
- Memory::Write32(memory_info_address + 28, memory_info.device_ref_count);
- Memory::Write32(memory_info_address + 36, 0);
+ memory.Write64(memory_info_address, memory_info.base_address);
+ memory.Write64(memory_info_address + 8, memory_info.size);
+ memory.Write32(memory_info_address + 16, memory_info.state);
+ memory.Write32(memory_info_address + 20, memory_info.attributes);
+ memory.Write32(memory_info_address + 24, memory_info.permission);
+ memory.Write32(memory_info_address + 32, memory_info.ipc_ref_count);
+ memory.Write32(memory_info_address + 28, memory_info.device_ref_count);
+ memory.Write32(memory_info_address + 36, 0);
// Page info appears to be currently unused by the kernel and is always set to zero.
- Memory::Write32(page_info_address, 0);
+ memory.Write32(page_info_address, 0);
return RESULT_SUCCESS;
}
@@ -1490,7 +1503,7 @@ static ResultCode CreateThread(Core::System& system, Handle* out_handle, VAddr e
}
auto& kernel = system.Kernel();
- CASCADE_RESULT(SharedPtr<Thread> thread,
+ CASCADE_RESULT(std::shared_ptr<Thread> thread,
Thread::Create(kernel, "", entry_point, priority, arg, processor_id, stack_top,
*current_process));
@@ -1516,7 +1529,7 @@ static ResultCode StartThread(Core::System& system, Handle thread_handle) {
LOG_DEBUG(Kernel_SVC, "called thread=0x{:08X}", thread_handle);
const auto& handle_table = system.Kernel().CurrentProcess()->GetHandleTable();
- const SharedPtr<Thread> thread = handle_table.Get<Thread>(thread_handle);
+ const std::shared_ptr<Thread> thread = handle_table.Get<Thread>(thread_handle);
if (!thread) {
LOG_ERROR(Kernel_SVC, "Thread handle does not exist, thread_handle=0x{:08X}",
thread_handle);
@@ -1540,7 +1553,7 @@ static void ExitThread(Core::System& system) {
auto* const current_thread = system.CurrentScheduler().GetCurrentThread();
current_thread->Stop();
- system.GlobalScheduler().RemoveThread(current_thread);
+ system.GlobalScheduler().RemoveThread(SharedFrom(current_thread));
system.PrepareReschedule();
}
@@ -1612,7 +1625,7 @@ static ResultCode WaitProcessWideKeyAtomic(Core::System& system, VAddr mutex_add
auto* const current_process = system.Kernel().CurrentProcess();
const auto& handle_table = current_process->GetHandleTable();
- SharedPtr<Thread> thread = handle_table.Get<Thread>(thread_handle);
+ std::shared_ptr<Thread> thread = handle_table.Get<Thread>(thread_handle);
ASSERT(thread);
const auto release_result = current_process->GetMutex().Release(mutex_addr);
@@ -1620,12 +1633,13 @@ static ResultCode WaitProcessWideKeyAtomic(Core::System& system, VAddr mutex_add
return release_result;
}
- SharedPtr<Thread> current_thread = system.CurrentScheduler().GetCurrentThread();
+ Thread* current_thread = system.CurrentScheduler().GetCurrentThread();
current_thread->SetCondVarWaitAddress(condition_variable_addr);
current_thread->SetMutexWaitAddress(mutex_addr);
current_thread->SetWaitHandle(thread_handle);
current_thread->SetStatus(ThreadStatus::WaitCondVar);
current_thread->InvalidateWakeupCallback();
+ current_process->InsertConditionVariableThread(SharedFrom(current_thread));
current_thread->WakeAfterDelay(nano_seconds);
@@ -1636,50 +1650,35 @@ static ResultCode WaitProcessWideKeyAtomic(Core::System& system, VAddr mutex_add
}
/// Signal process wide key
-static ResultCode SignalProcessWideKey(Core::System& system, VAddr condition_variable_addr,
- s32 target) {
+static void SignalProcessWideKey(Core::System& system, VAddr condition_variable_addr, s32 target) {
LOG_TRACE(Kernel_SVC, "called, condition_variable_addr=0x{:X}, target=0x{:08X}",
condition_variable_addr, target);
ASSERT(condition_variable_addr == Common::AlignDown(condition_variable_addr, 4));
// Retrieve a list of all threads that are waiting for this condition variable.
- std::vector<SharedPtr<Thread>> waiting_threads;
- const auto& scheduler = system.GlobalScheduler();
- const auto& thread_list = scheduler.GetThreadList();
-
- for (const auto& thread : thread_list) {
- if (thread->GetCondVarWaitAddress() == condition_variable_addr) {
- waiting_threads.push_back(thread);
- }
- }
-
- // Sort them by priority, such that the highest priority ones come first.
- std::sort(waiting_threads.begin(), waiting_threads.end(),
- [](const SharedPtr<Thread>& lhs, const SharedPtr<Thread>& rhs) {
- return lhs->GetPriority() < rhs->GetPriority();
- });
+ auto* const current_process = system.Kernel().CurrentProcess();
+ std::vector<std::shared_ptr<Thread>> waiting_threads =
+ current_process->GetConditionVariableThreads(condition_variable_addr);
- // Only process up to 'target' threads, unless 'target' is -1, in which case process
+ // Only process up to 'target' threads, unless 'target' is less equal 0, in which case process
// them all.
std::size_t last = waiting_threads.size();
- if (target != -1)
+ if (target > 0)
last = std::min(waiting_threads.size(), static_cast<std::size_t>(target));
- // If there are no threads waiting on this condition variable, just exit
- if (last == 0)
- return RESULT_SUCCESS;
-
for (std::size_t index = 0; index < last; ++index) {
auto& thread = waiting_threads[index];
ASSERT(thread->GetCondVarWaitAddress() == condition_variable_addr);
// liberate Cond Var Thread.
+ current_process->RemoveConditionVariableThread(thread);
thread->SetCondVarWaitAddress(0);
const std::size_t current_core = system.CurrentCoreIndex();
auto& monitor = system.Monitor();
+ auto& memory = system.Memory();
// Atomically read the value of the mutex.
u32 mutex_val = 0;
@@ -1689,7 +1688,7 @@ static ResultCode SignalProcessWideKey(Core::System& system, VAddr condition_var
monitor.SetExclusive(current_core, mutex_address);
// If the mutex is not yet acquired, acquire it.
- mutex_val = Memory::Read32(mutex_address);
+ mutex_val = memory.Read32(mutex_address);
if (mutex_val != 0) {
update_val = mutex_val | Mutex::MutexHasWaitersFlag;
@@ -1726,8 +1725,6 @@ static ResultCode SignalProcessWideKey(Core::System& system, VAddr condition_var
system.PrepareReschedule(thread->GetProcessorID());
}
}
-
- return RESULT_SUCCESS;
}
// Wait for an address (via Address Arbiter)
@@ -1781,12 +1778,25 @@ static ResultCode SignalToAddress(Core::System& system, VAddr address, u32 type,
return address_arbiter.SignalToAddress(address, signal_type, value, num_to_wake);
}
+static void KernelDebug([[maybe_unused]] Core::System& system,
+ [[maybe_unused]] u32 kernel_debug_type, [[maybe_unused]] u64 param1,
+ [[maybe_unused]] u64 param2, [[maybe_unused]] u64 param3) {
+ // Intentionally do nothing, as this does nothing in released kernel binaries.
+}
+
+static void ChangeKernelTraceState([[maybe_unused]] Core::System& system,
+ [[maybe_unused]] u32 trace_state) {
+ // Intentionally do nothing, as this does nothing in released kernel binaries.
+}
+
/// This returns the total CPU ticks elapsed since the CPU was powered-on
static u64 GetSystemTick(Core::System& system) {
LOG_TRACE(Kernel_SVC, "called");
auto& core_timing = system.CoreTiming();
- const u64 result{core_timing.GetTicks()};
+
+ // Returns the value of cntpct_el0 (https://switchbrew.org/wiki/SVC#svcGetSystemTick)
+ const u64 result{Core::Timing::CpuCyclesToClockCycles(system.CoreTiming().GetTicks())};
// Advance time to defeat dumb games that busy-wait for the frame to end.
core_timing.AddTicks(400);
@@ -1975,7 +1985,7 @@ static ResultCode GetThreadCoreMask(Core::System& system, Handle thread_handle,
LOG_TRACE(Kernel_SVC, "called, handle=0x{:08X}", thread_handle);
const auto& handle_table = system.Kernel().CurrentProcess()->GetHandleTable();
- const SharedPtr<Thread> thread = handle_table.Get<Thread>(thread_handle);
+ const std::shared_ptr<Thread> thread = handle_table.Get<Thread>(thread_handle);
if (!thread) {
LOG_ERROR(Kernel_SVC, "Thread handle does not exist, thread_handle=0x{:08X}",
thread_handle);
@@ -2034,7 +2044,7 @@ static ResultCode SetThreadCoreMask(Core::System& system, Handle thread_handle,
}
const auto& handle_table = current_process->GetHandleTable();
- const SharedPtr<Thread> thread = handle_table.Get<Thread>(thread_handle);
+ const std::shared_ptr<Thread> thread = handle_table.Get<Thread>(thread_handle);
if (!thread) {
LOG_ERROR(Kernel_SVC, "Thread handle does not exist, thread_handle=0x{:08X}",
thread_handle);
@@ -2290,12 +2300,13 @@ static ResultCode GetProcessList(Core::System& system, u32* out_num_processes,
return ERR_INVALID_ADDRESS_STATE;
}
+ auto& memory = system.Memory();
const auto& process_list = kernel.GetProcessList();
const auto num_processes = process_list.size();
const auto copy_amount = std::min(std::size_t{out_process_ids_size}, num_processes);
for (std::size_t i = 0; i < copy_amount; ++i) {
- Memory::Write64(out_process_ids, process_list[i]->GetProcessID());
+ memory.Write64(out_process_ids, process_list[i]->GetProcessID());
out_process_ids += sizeof(u64);
}
@@ -2329,13 +2340,14 @@ static ResultCode GetThreadList(Core::System& system, u32* out_num_threads, VAdd
return ERR_INVALID_ADDRESS_STATE;
}
+ auto& memory = system.Memory();
const auto& thread_list = current_process->GetThreadList();
const auto num_threads = thread_list.size();
const auto copy_amount = std::min(std::size_t{out_thread_ids_size}, num_threads);
auto list_iter = thread_list.cbegin();
for (std::size_t i = 0; i < copy_amount; ++i, ++list_iter) {
- Memory::Write64(out_thread_ids, (*list_iter)->GetThreadID());
+ memory.Write64(out_thread_ids, (*list_iter)->GetThreadID());
out_thread_ids += sizeof(u64);
}
@@ -2414,8 +2426,8 @@ static const FunctionDef SVC_Table[] = {
{0x39, nullptr, "Unknown"},
{0x3A, nullptr, "Unknown"},
{0x3B, nullptr, "Unknown"},
- {0x3C, nullptr, "DumpInfo"},
- {0x3D, nullptr, "DumpInfoNew"},
+ {0x3C, SvcWrap<KernelDebug>, "KernelDebug"},
+ {0x3D, SvcWrap<ChangeKernelTraceState>, "ChangeKernelTraceState"},
{0x3E, nullptr, "Unknown"},
{0x3F, nullptr, "Unknown"},
{0x40, nullptr, "CreateSession"},
diff --git a/src/core/hle/kernel/svc_wrap.h b/src/core/hle/kernel/svc_wrap.h
index c2d8d0dc3..29a2cfa9d 100644
--- a/src/core/hle/kernel/svc_wrap.h
+++ b/src/core/hle/kernel/svc_wrap.h
@@ -112,11 +112,6 @@ void SvcWrap(Core::System& system) {
FuncReturn(system, retval);
}
-template <ResultCode func(Core::System&, u64, s32)>
-void SvcWrap(Core::System& system) {
- FuncReturn(system, func(system, Param(system, 0), static_cast<s32>(Param(system, 1))).raw);
-}
-
template <ResultCode func(Core::System&, u64, u32)>
void SvcWrap(Core::System& system) {
FuncReturn(system, func(system, Param(system, 0), static_cast<u32>(Param(system, 1))).raw);
@@ -311,11 +306,27 @@ void SvcWrap(Core::System& system) {
func(system);
}
+template <void func(Core::System&, u32)>
+void SvcWrap(Core::System& system) {
+ func(system, static_cast<u32>(Param(system, 0)));
+}
+
+template <void func(Core::System&, u32, u64, u64, u64)>
+void SvcWrap(Core::System& system) {
+ func(system, static_cast<u32>(Param(system, 0)), Param(system, 1), Param(system, 2),
+ Param(system, 3));
+}
+
template <void func(Core::System&, s64)>
void SvcWrap(Core::System& system) {
func(system, static_cast<s64>(Param(system, 0)));
}
+template <void func(Core::System&, u64, s32)>
+void SvcWrap(Core::System& system) {
+ func(system, Param(system, 0), static_cast<s32>(Param(system, 1)));
+}
+
template <void func(Core::System&, u64, u64)>
void SvcWrap(Core::System& system) {
func(system, Param(system, 0), Param(system, 1));
diff --git a/src/core/hle/kernel/thread.cpp b/src/core/hle/kernel/thread.cpp
index 962530d2d..e84e5ce0d 100644
--- a/src/core/hle/kernel/thread.cpp
+++ b/src/core/hle/kernel/thread.cpp
@@ -50,7 +50,7 @@ void Thread::Stop() {
// Clean up any dangling references in objects that this thread was waiting for
for (auto& wait_object : wait_objects) {
- wait_object->RemoveWaitingThread(this);
+ wait_object->RemoveWaitingThread(SharedFrom(this));
}
wait_objects.clear();
@@ -77,18 +77,6 @@ void Thread::CancelWakeupTimer() {
callback_handle);
}
-static std::optional<s32> GetNextProcessorId(u64 mask) {
- for (s32 index = 0; index < Core::NUM_CPU_CORES; ++index) {
- if (mask & (1ULL << index)) {
- if (!Core::System::GetInstance().Scheduler(index).GetCurrentThread()) {
- // Core is enabled and not running any threads, use this one
- return index;
- }
- }
- }
- return {};
-}
-
void Thread::ResumeFromWait() {
ASSERT_MSG(wait_objects.empty(), "Thread is waking up while waiting for objects");
@@ -132,8 +120,11 @@ void Thread::ResumeFromWait() {
}
void Thread::CancelWait() {
- ASSERT(GetStatus() == ThreadStatus::WaitSynch);
- ClearWaitObjects();
+ if (GetSchedulingStatus() != ThreadSchedStatus::Paused) {
+ is_sync_cancelled = true;
+ return;
+ }
+ is_sync_cancelled = false;
SetWaitSynchronizationResult(ERR_SYNCHRONIZATION_CANCELED);
ResumeFromWait();
}
@@ -156,9 +147,10 @@ static void ResetThreadContext(Core::ARM_Interface::ThreadContext& context, VAdd
context.fpcr = 0x03C00000;
}
-ResultVal<SharedPtr<Thread>> Thread::Create(KernelCore& kernel, std::string name, VAddr entry_point,
- u32 priority, u64 arg, s32 processor_id,
- VAddr stack_top, Process& owner_process) {
+ResultVal<std::shared_ptr<Thread>> Thread::Create(KernelCore& kernel, std::string name,
+ VAddr entry_point, u32 priority, u64 arg,
+ s32 processor_id, VAddr stack_top,
+ Process& owner_process) {
// Check if priority is in ranged. Lowest priority -> highest priority id.
if (priority > THREADPRIO_LOWEST) {
LOG_ERROR(Kernel_SVC, "Invalid thread priority: {}", priority);
@@ -170,14 +162,14 @@ ResultVal<SharedPtr<Thread>> Thread::Create(KernelCore& kernel, std::string name
return ERR_INVALID_PROCESSOR_ID;
}
- if (!Memory::IsValidVirtualAddress(owner_process, entry_point)) {
+ auto& system = Core::System::GetInstance();
+ if (!system.Memory().IsValidVirtualAddress(owner_process, entry_point)) {
LOG_ERROR(Kernel_SVC, "(name={}): invalid entry {:016X}", name, entry_point);
// TODO (bunnei): Find the correct error code to use here
- return ResultCode(-1);
+ return RESULT_UNKNOWN;
}
- auto& system = Core::System::GetInstance();
- SharedPtr<Thread> thread(new Thread(kernel));
+ std::shared_ptr<Thread> thread = std::make_shared<Thread>(kernel);
thread->thread_id = kernel.CreateNewThreadID();
thread->status = ThreadStatus::Dormant;
@@ -206,7 +198,7 @@ ResultVal<SharedPtr<Thread>> Thread::Create(KernelCore& kernel, std::string name
// to initialize the context
ResetThreadContext(thread->context, stack_top, entry_point, arg);
- return MakeResult<SharedPtr<Thread>>(std::move(thread));
+ return MakeResult<std::shared_ptr<Thread>>(std::move(thread));
}
void Thread::SetPriority(u32 priority) {
@@ -224,7 +216,7 @@ void Thread::SetWaitSynchronizationOutput(s32 output) {
context.cpu_registers[1] = output;
}
-s32 Thread::GetWaitObjectIndex(const WaitObject* object) const {
+s32 Thread::GetWaitObjectIndex(std::shared_ptr<WaitObject> object) const {
ASSERT_MSG(!wait_objects.empty(), "Thread is not waiting for anything");
const auto match = std::find(wait_objects.rbegin(), wait_objects.rend(), object);
return static_cast<s32>(std::distance(match, wait_objects.rend()) - 1);
@@ -264,8 +256,8 @@ void Thread::SetStatus(ThreadStatus new_status) {
status = new_status;
}
-void Thread::AddMutexWaiter(SharedPtr<Thread> thread) {
- if (thread->lock_owner == this) {
+void Thread::AddMutexWaiter(std::shared_ptr<Thread> thread) {
+ if (thread->lock_owner.get() == this) {
// If the thread is already waiting for this thread to release the mutex, ensure that the
// waiters list is consistent and return without doing anything.
const auto iter = std::find(wait_mutex_threads.begin(), wait_mutex_threads.end(), thread);
@@ -285,13 +277,13 @@ void Thread::AddMutexWaiter(SharedPtr<Thread> thread) {
wait_mutex_threads.begin(), wait_mutex_threads.end(),
[&thread](const auto& entry) { return entry->GetPriority() > thread->GetPriority(); });
wait_mutex_threads.insert(insertion_point, thread);
- thread->lock_owner = this;
+ thread->lock_owner = SharedFrom(this);
UpdatePriority();
}
-void Thread::RemoveMutexWaiter(SharedPtr<Thread> thread) {
- ASSERT(thread->lock_owner == this);
+void Thread::RemoveMutexWaiter(std::shared_ptr<Thread> thread) {
+ ASSERT(thread->lock_owner.get() == this);
// Ensure that the thread is in the list of mutex waiters
const auto iter = std::find(wait_mutex_threads.begin(), wait_mutex_threads.end(), thread);
@@ -318,16 +310,24 @@ void Thread::UpdatePriority() {
return;
}
+ if (GetStatus() == ThreadStatus::WaitCondVar) {
+ owner_process->RemoveConditionVariableThread(SharedFrom(this));
+ }
+
SetCurrentPriority(new_priority);
+ if (GetStatus() == ThreadStatus::WaitCondVar) {
+ owner_process->InsertConditionVariableThread(SharedFrom(this));
+ }
+
if (!lock_owner) {
return;
}
// Ensure that the thread is within the correct location in the waiting list.
auto old_owner = lock_owner;
- lock_owner->RemoveMutexWaiter(this);
- old_owner->AddMutexWaiter(this);
+ lock_owner->RemoveMutexWaiter(SharedFrom(this));
+ old_owner->AddMutexWaiter(SharedFrom(this));
// Recursively update the priority of the thread that depends on the priority of this one.
lock_owner->UpdatePriority();
@@ -340,11 +340,11 @@ void Thread::ChangeCore(u32 core, u64 mask) {
bool Thread::AllWaitObjectsReady() const {
return std::none_of(
wait_objects.begin(), wait_objects.end(),
- [this](const SharedPtr<WaitObject>& object) { return object->ShouldWait(this); });
+ [this](const std::shared_ptr<WaitObject>& object) { return object->ShouldWait(this); });
}
-bool Thread::InvokeWakeupCallback(ThreadWakeupReason reason, SharedPtr<Thread> thread,
- SharedPtr<WaitObject> object, std::size_t index) {
+bool Thread::InvokeWakeupCallback(ThreadWakeupReason reason, std::shared_ptr<Thread> thread,
+ std::shared_ptr<WaitObject> object, std::size_t index) {
ASSERT(wakeup_callback);
return wakeup_callback(reason, std::move(thread), std::move(object), index);
}
@@ -401,7 +401,7 @@ void Thread::SetCurrentPriority(u32 new_priority) {
ResultCode Thread::SetCoreAndAffinityMask(s32 new_core, u64 new_affinity_mask) {
const auto HighestSetCore = [](u64 mask, u32 max_cores) {
- for (s32 core = max_cores - 1; core >= 0; core--) {
+ for (s32 core = static_cast<s32>(max_cores - 1); core >= 0; core--) {
if (((mask >> core) & 1) != 0) {
return core;
}
@@ -425,7 +425,7 @@ ResultCode Thread::SetCoreAndAffinityMask(s32 new_core, u64 new_affinity_mask) {
if (old_affinity_mask != new_affinity_mask) {
const s32 old_core = processor_id;
if (processor_id >= 0 && ((affinity_mask >> processor_id) & 1) == 0) {
- if (ideal_core < 0) {
+ if (static_cast<s32>(ideal_core) < 0) {
processor_id = HighestSetCore(affinity_mask, GlobalScheduler::NUM_CPU_CORES);
} else {
processor_id = ideal_core;
@@ -447,23 +447,23 @@ void Thread::AdjustSchedulingOnStatus(u32 old_flags) {
ThreadSchedStatus::Runnable) {
// In this case the thread was running, now it's pausing/exitting
if (processor_id >= 0) {
- scheduler.Unschedule(current_priority, processor_id, this);
+ scheduler.Unschedule(current_priority, static_cast<u32>(processor_id), this);
}
- for (s32 core = 0; core < GlobalScheduler::NUM_CPU_CORES; core++) {
- if (core != processor_id && ((affinity_mask >> core) & 1) != 0) {
- scheduler.Unsuggest(current_priority, static_cast<u32>(core), this);
+ for (u32 core = 0; core < GlobalScheduler::NUM_CPU_CORES; core++) {
+ if (core != static_cast<u32>(processor_id) && ((affinity_mask >> core) & 1) != 0) {
+ scheduler.Unsuggest(current_priority, core, this);
}
}
} else if (GetSchedulingStatus() == ThreadSchedStatus::Runnable) {
// The thread is now set to running from being stopped
if (processor_id >= 0) {
- scheduler.Schedule(current_priority, processor_id, this);
+ scheduler.Schedule(current_priority, static_cast<u32>(processor_id), this);
}
- for (s32 core = 0; core < GlobalScheduler::NUM_CPU_CORES; core++) {
- if (core != processor_id && ((affinity_mask >> core) & 1) != 0) {
- scheduler.Suggest(current_priority, static_cast<u32>(core), this);
+ for (u32 core = 0; core < GlobalScheduler::NUM_CPU_CORES; core++) {
+ if (core != static_cast<u32>(processor_id) && ((affinity_mask >> core) & 1) != 0) {
+ scheduler.Suggest(current_priority, core, this);
}
}
}
@@ -477,11 +477,11 @@ void Thread::AdjustSchedulingOnPriority(u32 old_priority) {
}
auto& scheduler = Core::System::GetInstance().GlobalScheduler();
if (processor_id >= 0) {
- scheduler.Unschedule(old_priority, processor_id, this);
+ scheduler.Unschedule(old_priority, static_cast<u32>(processor_id), this);
}
for (u32 core = 0; core < GlobalScheduler::NUM_CPU_CORES; core++) {
- if (core != processor_id && ((affinity_mask >> core) & 1) != 0) {
+ if (core != static_cast<u32>(processor_id) && ((affinity_mask >> core) & 1) != 0) {
scheduler.Unsuggest(old_priority, core, this);
}
}
@@ -491,14 +491,14 @@ void Thread::AdjustSchedulingOnPriority(u32 old_priority) {
if (processor_id >= 0) {
if (current_thread == this) {
- scheduler.SchedulePrepend(current_priority, processor_id, this);
+ scheduler.SchedulePrepend(current_priority, static_cast<u32>(processor_id), this);
} else {
- scheduler.Schedule(current_priority, processor_id, this);
+ scheduler.Schedule(current_priority, static_cast<u32>(processor_id), this);
}
}
for (u32 core = 0; core < GlobalScheduler::NUM_CPU_CORES; core++) {
- if (core != processor_id && ((affinity_mask >> core) & 1) != 0) {
+ if (core != static_cast<u32>(processor_id) && ((affinity_mask >> core) & 1) != 0) {
scheduler.Suggest(current_priority, core, this);
}
}
@@ -515,7 +515,7 @@ void Thread::AdjustSchedulingOnAffinity(u64 old_affinity_mask, s32 old_core) {
for (u32 core = 0; core < GlobalScheduler::NUM_CPU_CORES; core++) {
if (((old_affinity_mask >> core) & 1) != 0) {
- if (core == old_core) {
+ if (core == static_cast<u32>(old_core)) {
scheduler.Unschedule(current_priority, core, this);
} else {
scheduler.Unsuggest(current_priority, core, this);
@@ -525,7 +525,7 @@ void Thread::AdjustSchedulingOnAffinity(u64 old_affinity_mask, s32 old_core) {
for (u32 core = 0; core < GlobalScheduler::NUM_CPU_CORES; core++) {
if (((affinity_mask >> core) & 1) != 0) {
- if (core == processor_id) {
+ if (core == static_cast<u32>(processor_id)) {
scheduler.Schedule(current_priority, core, this);
} else {
scheduler.Suggest(current_priority, core, this);
diff --git a/src/core/hle/kernel/thread.h b/src/core/hle/kernel/thread.h
index c9870873d..3bcf9e137 100644
--- a/src/core/hle/kernel/thread.h
+++ b/src/core/hle/kernel/thread.h
@@ -97,14 +97,18 @@ enum class ThreadSchedMasks : u32 {
class Thread final : public WaitObject {
public:
- using MutexWaitingThreads = std::vector<SharedPtr<Thread>>;
+ explicit Thread(KernelCore& kernel);
+ ~Thread() override;
+
+ using MutexWaitingThreads = std::vector<std::shared_ptr<Thread>>;
using ThreadContext = Core::ARM_Interface::ThreadContext;
- using ThreadWaitObjects = std::vector<SharedPtr<WaitObject>>;
+ using ThreadWaitObjects = std::vector<std::shared_ptr<WaitObject>>;
- using WakeupCallback = std::function<bool(ThreadWakeupReason reason, SharedPtr<Thread> thread,
- SharedPtr<WaitObject> object, std::size_t index)>;
+ using WakeupCallback =
+ std::function<bool(ThreadWakeupReason reason, std::shared_ptr<Thread> thread,
+ std::shared_ptr<WaitObject> object, std::size_t index)>;
/**
* Creates and returns a new thread. The new thread is immediately scheduled
@@ -118,10 +122,10 @@ public:
* @param owner_process The parent process for the thread
* @return A shared pointer to the newly created thread
*/
- static ResultVal<SharedPtr<Thread>> Create(KernelCore& kernel, std::string name,
- VAddr entry_point, u32 priority, u64 arg,
- s32 processor_id, VAddr stack_top,
- Process& owner_process);
+ static ResultVal<std::shared_ptr<Thread>> Create(KernelCore& kernel, std::string name,
+ VAddr entry_point, u32 priority, u64 arg,
+ s32 processor_id, VAddr stack_top,
+ Process& owner_process);
std::string GetName() const override {
return name;
@@ -166,10 +170,10 @@ public:
void SetPriority(u32 priority);
/// Adds a thread to the list of threads that are waiting for a lock held by this thread.
- void AddMutexWaiter(SharedPtr<Thread> thread);
+ void AddMutexWaiter(std::shared_ptr<Thread> thread);
/// Removes a thread from the list of threads that are waiting for a lock held by this thread.
- void RemoveMutexWaiter(SharedPtr<Thread> thread);
+ void RemoveMutexWaiter(std::shared_ptr<Thread> thread);
/// Recalculates the current priority taking into account priority inheritance.
void UpdatePriority();
@@ -229,7 +233,7 @@ public:
*
* @param object Object to query the index of.
*/
- s32 GetWaitObjectIndex(const WaitObject* object) const;
+ s32 GetWaitObjectIndex(std::shared_ptr<WaitObject> object) const;
/**
* Stops a thread, invalidating it from further use
@@ -320,7 +324,7 @@ public:
void ClearWaitObjects() {
for (const auto& waiting_object : wait_objects) {
- waiting_object->RemoveWaitingThread(this);
+ waiting_object->RemoveWaitingThread(SharedFrom(this));
}
wait_objects.clear();
}
@@ -336,7 +340,7 @@ public:
return lock_owner.get();
}
- void SetLockOwner(SharedPtr<Thread> owner) {
+ void SetLockOwner(std::shared_ptr<Thread> owner) {
lock_owner = std::move(owner);
}
@@ -390,8 +394,8 @@ public:
* @pre A valid wakeup callback has been set. Violating this precondition
* will cause an assertion to trigger.
*/
- bool InvokeWakeupCallback(ThreadWakeupReason reason, SharedPtr<Thread> thread,
- SharedPtr<WaitObject> object, std::size_t index);
+ bool InvokeWakeupCallback(ThreadWakeupReason reason, std::shared_ptr<Thread> thread,
+ std::shared_ptr<WaitObject> object, std::size_t index);
u32 GetIdealCore() const {
return ideal_core;
@@ -440,10 +444,15 @@ public:
is_running = value;
}
-private:
- explicit Thread(KernelCore& kernel);
- ~Thread() override;
+ bool IsSyncCancelled() const {
+ return is_sync_cancelled;
+ }
+ void SetSyncCancelled(bool value) {
+ is_sync_cancelled = value;
+ }
+
+private:
void SetSchedulingStatus(ThreadSchedStatus new_status);
void SetCurrentPriority(u32 new_priority);
ResultCode SetCoreAndAffinityMask(s32 new_core, u64 new_affinity_mask);
@@ -491,7 +500,7 @@ private:
MutexWaitingThreads wait_mutex_threads;
/// Thread that owns the lock that this thread is waiting for.
- SharedPtr<Thread> lock_owner;
+ std::shared_ptr<Thread> lock_owner;
/// If waiting on a ConditionVariable, this is the ConditionVariable address
VAddr condvar_wait_address = 0;
@@ -524,6 +533,7 @@ private:
u32 scheduling_state = 0;
bool is_running = false;
+ bool is_sync_cancelled = false;
std::string name;
};
diff --git a/src/core/hle/kernel/transfer_memory.cpp b/src/core/hle/kernel/transfer_memory.cpp
index 1113c815e..f0e73f57b 100644
--- a/src/core/hle/kernel/transfer_memory.cpp
+++ b/src/core/hle/kernel/transfer_memory.cpp
@@ -14,9 +14,9 @@ namespace Kernel {
TransferMemory::TransferMemory(KernelCore& kernel) : Object{kernel} {}
TransferMemory::~TransferMemory() = default;
-SharedPtr<TransferMemory> TransferMemory::Create(KernelCore& kernel, VAddr base_address, u64 size,
- MemoryPermission permissions) {
- SharedPtr<TransferMemory> transfer_memory{new TransferMemory(kernel)};
+std::shared_ptr<TransferMemory> TransferMemory::Create(KernelCore& kernel, VAddr base_address,
+ u64 size, MemoryPermission permissions) {
+ std::shared_ptr<TransferMemory> transfer_memory{std::make_shared<TransferMemory>(kernel)};
transfer_memory->base_address = base_address;
transfer_memory->memory_size = size;
diff --git a/src/core/hle/kernel/transfer_memory.h b/src/core/hle/kernel/transfer_memory.h
index 6be9dc094..0a6e15d18 100644
--- a/src/core/hle/kernel/transfer_memory.h
+++ b/src/core/hle/kernel/transfer_memory.h
@@ -5,7 +5,6 @@
#pragma once
#include <memory>
-#include <vector>
#include "core/hle/kernel/object.h"
#include "core/hle/kernel/physical_memory.h"
@@ -27,10 +26,13 @@ enum class MemoryPermission : u32;
///
class TransferMemory final : public Object {
public:
+ explicit TransferMemory(KernelCore& kernel);
+ ~TransferMemory() override;
+
static constexpr HandleType HANDLE_TYPE = HandleType::TransferMemory;
- static SharedPtr<TransferMemory> Create(KernelCore& kernel, VAddr base_address, u64 size,
- MemoryPermission permissions);
+ static std::shared_ptr<TransferMemory> Create(KernelCore& kernel, VAddr base_address, u64 size,
+ MemoryPermission permissions);
TransferMemory(const TransferMemory&) = delete;
TransferMemory& operator=(const TransferMemory&) = delete;
@@ -79,9 +81,6 @@ public:
ResultCode UnmapMemory(VAddr address, u64 size);
private:
- explicit TransferMemory(KernelCore& kernel);
- ~TransferMemory() override;
-
/// Memory block backing this instance.
std::shared_ptr<PhysicalMemory> backing_block;
diff --git a/src/core/hle/kernel/vm_manager.cpp b/src/core/hle/kernel/vm_manager.cpp
index c7af87073..a9a20ef76 100644
--- a/src/core/hle/kernel/vm_manager.cpp
+++ b/src/core/hle/kernel/vm_manager.cpp
@@ -16,7 +16,6 @@
#include "core/hle/kernel/resource_limit.h"
#include "core/hle/kernel/vm_manager.h"
#include "core/memory.h"
-#include "core/memory_setup.h"
namespace Kernel {
namespace {
@@ -167,7 +166,7 @@ ResultVal<VAddr> VMManager::FindFreeRegion(VAddr begin, VAddr end, u64 size) con
if (vma_handle == vma_map.cend()) {
// TODO(Subv): Find the correct error code here.
- return ResultCode(-1);
+ return RESULT_UNKNOWN;
}
const VAddr target = std::max(begin, vma_handle->second.base);
@@ -786,19 +785,21 @@ void VMManager::MergeAdjacentVMA(VirtualMemoryArea& left, const VirtualMemoryAre
}
void VMManager::UpdatePageTableForVMA(const VirtualMemoryArea& vma) {
+ auto& memory = system.Memory();
+
switch (vma.type) {
case VMAType::Free:
- Memory::UnmapRegion(page_table, vma.base, vma.size);
+ memory.UnmapRegion(page_table, vma.base, vma.size);
break;
case VMAType::AllocatedMemoryBlock:
- Memory::MapMemoryRegion(page_table, vma.base, vma.size,
- vma.backing_block->data() + vma.offset);
+ memory.MapMemoryRegion(page_table, vma.base, vma.size,
+ vma.backing_block->data() + vma.offset);
break;
case VMAType::BackingMemory:
- Memory::MapMemoryRegion(page_table, vma.base, vma.size, vma.backing_memory);
+ memory.MapMemoryRegion(page_table, vma.base, vma.size, vma.backing_memory);
break;
case VMAType::MMIO:
- Memory::MapIoRegion(page_table, vma.base, vma.size, vma.mmio_handler);
+ memory.MapIoRegion(page_table, vma.base, vma.size, vma.mmio_handler);
break;
}
}
diff --git a/src/core/hle/kernel/wait_object.cpp b/src/core/hle/kernel/wait_object.cpp
index c00cef062..745f2c4e8 100644
--- a/src/core/hle/kernel/wait_object.cpp
+++ b/src/core/hle/kernel/wait_object.cpp
@@ -18,13 +18,13 @@ namespace Kernel {
WaitObject::WaitObject(KernelCore& kernel) : Object{kernel} {}
WaitObject::~WaitObject() = default;
-void WaitObject::AddWaitingThread(SharedPtr<Thread> thread) {
+void WaitObject::AddWaitingThread(std::shared_ptr<Thread> thread) {
auto itr = std::find(waiting_threads.begin(), waiting_threads.end(), thread);
if (itr == waiting_threads.end())
waiting_threads.push_back(std::move(thread));
}
-void WaitObject::RemoveWaitingThread(Thread* thread) {
+void WaitObject::RemoveWaitingThread(std::shared_ptr<Thread> thread) {
auto itr = std::find(waiting_threads.begin(), waiting_threads.end(), thread);
// If a thread passed multiple handles to the same object,
// the kernel might attempt to remove the thread from the object's
@@ -33,7 +33,7 @@ void WaitObject::RemoveWaitingThread(Thread* thread) {
waiting_threads.erase(itr);
}
-SharedPtr<Thread> WaitObject::GetHighestPriorityReadyThread() const {
+std::shared_ptr<Thread> WaitObject::GetHighestPriorityReadyThread() const {
Thread* candidate = nullptr;
u32 candidate_priority = THREADPRIO_LOWEST + 1;
@@ -64,10 +64,10 @@ SharedPtr<Thread> WaitObject::GetHighestPriorityReadyThread() const {
}
}
- return candidate;
+ return SharedFrom(candidate);
}
-void WaitObject::WakeupWaitingThread(SharedPtr<Thread> thread) {
+void WaitObject::WakeupWaitingThread(std::shared_ptr<Thread> thread) {
ASSERT(!ShouldWait(thread.get()));
if (!thread) {
@@ -83,7 +83,7 @@ void WaitObject::WakeupWaitingThread(SharedPtr<Thread> thread) {
Acquire(thread.get());
}
- const std::size_t index = thread->GetWaitObjectIndex(this);
+ const std::size_t index = thread->GetWaitObjectIndex(SharedFrom(this));
thread->ClearWaitObjects();
@@ -91,7 +91,8 @@ void WaitObject::WakeupWaitingThread(SharedPtr<Thread> thread) {
bool resume = true;
if (thread->HasWakeupCallback()) {
- resume = thread->InvokeWakeupCallback(ThreadWakeupReason::Signal, thread, this, index);
+ resume = thread->InvokeWakeupCallback(ThreadWakeupReason::Signal, thread, SharedFrom(this),
+ index);
}
if (resume) {
thread->ResumeFromWait();
@@ -105,7 +106,7 @@ void WaitObject::WakeupAllWaitingThreads() {
}
}
-const std::vector<SharedPtr<Thread>>& WaitObject::GetWaitingThreads() const {
+const std::vector<std::shared_ptr<Thread>>& WaitObject::GetWaitingThreads() const {
return waiting_threads;
}
diff --git a/src/core/hle/kernel/wait_object.h b/src/core/hle/kernel/wait_object.h
index 3271a30a7..9a17958a4 100644
--- a/src/core/hle/kernel/wait_object.h
+++ b/src/core/hle/kernel/wait_object.h
@@ -4,8 +4,9 @@
#pragma once
+#include <memory>
#include <vector>
-#include <boost/smart_ptr/intrusive_ptr.hpp>
+
#include "core/hle/kernel/object.h"
namespace Kernel {
@@ -33,13 +34,13 @@ public:
* Add a thread to wait on this object
* @param thread Pointer to thread to add
*/
- void AddWaitingThread(SharedPtr<Thread> thread);
+ void AddWaitingThread(std::shared_ptr<Thread> thread);
/**
* Removes a thread from waiting on this object (e.g. if it was resumed already)
* @param thread Pointer to thread to remove
*/
- void RemoveWaitingThread(Thread* thread);
+ void RemoveWaitingThread(std::shared_ptr<Thread> thread);
/**
* Wake up all threads waiting on this object that can be awoken, in priority order,
@@ -51,24 +52,24 @@ public:
* Wakes up a single thread waiting on this object.
* @param thread Thread that is waiting on this object to wakeup.
*/
- void WakeupWaitingThread(SharedPtr<Thread> thread);
+ void WakeupWaitingThread(std::shared_ptr<Thread> thread);
/// Obtains the highest priority thread that is ready to run from this object's waiting list.
- SharedPtr<Thread> GetHighestPriorityReadyThread() const;
+ std::shared_ptr<Thread> GetHighestPriorityReadyThread() const;
/// Get a const reference to the waiting threads list for debug use
- const std::vector<SharedPtr<Thread>>& GetWaitingThreads() const;
+ const std::vector<std::shared_ptr<Thread>>& GetWaitingThreads() const;
private:
/// Threads waiting for this object to become available
- std::vector<SharedPtr<Thread>> waiting_threads;
+ std::vector<std::shared_ptr<Thread>> waiting_threads;
};
// Specialization of DynamicObjectCast for WaitObjects
template <>
-inline SharedPtr<WaitObject> DynamicObjectCast<WaitObject>(SharedPtr<Object> object) {
+inline std::shared_ptr<WaitObject> DynamicObjectCast<WaitObject>(std::shared_ptr<Object> object) {
if (object != nullptr && object->IsWaitable()) {
- return boost::static_pointer_cast<WaitObject>(object);
+ return std::static_pointer_cast<WaitObject>(object);
}
return nullptr;
}
diff --git a/src/core/hle/kernel/writable_event.cpp b/src/core/hle/kernel/writable_event.cpp
index c783a34ee..c9332e3e1 100644
--- a/src/core/hle/kernel/writable_event.cpp
+++ b/src/core/hle/kernel/writable_event.cpp
@@ -16,8 +16,8 @@ WritableEvent::WritableEvent(KernelCore& kernel) : Object{kernel} {}
WritableEvent::~WritableEvent() = default;
EventPair WritableEvent::CreateEventPair(KernelCore& kernel, std::string name) {
- SharedPtr<WritableEvent> writable_event(new WritableEvent(kernel));
- SharedPtr<ReadableEvent> readable_event(new ReadableEvent(kernel));
+ std::shared_ptr<WritableEvent> writable_event(new WritableEvent(kernel));
+ std::shared_ptr<ReadableEvent> readable_event(new ReadableEvent(kernel));
writable_event->name = name + ":Writable";
writable_event->readable = readable_event;
@@ -27,7 +27,7 @@ EventPair WritableEvent::CreateEventPair(KernelCore& kernel, std::string name) {
return {std::move(readable_event), std::move(writable_event)};
}
-SharedPtr<ReadableEvent> WritableEvent::GetReadableEvent() const {
+std::shared_ptr<ReadableEvent> WritableEvent::GetReadableEvent() const {
return readable;
}
diff --git a/src/core/hle/kernel/writable_event.h b/src/core/hle/kernel/writable_event.h
index f46cf1dd8..6189cf65c 100644
--- a/src/core/hle/kernel/writable_event.h
+++ b/src/core/hle/kernel/writable_event.h
@@ -4,6 +4,8 @@
#pragma once
+#include <memory>
+
#include "core/hle/kernel/object.h"
namespace Kernel {
@@ -13,8 +15,8 @@ class ReadableEvent;
class WritableEvent;
struct EventPair {
- SharedPtr<ReadableEvent> readable;
- SharedPtr<WritableEvent> writable;
+ std::shared_ptr<ReadableEvent> readable;
+ std::shared_ptr<WritableEvent> writable;
};
class WritableEvent final : public Object {
@@ -40,7 +42,7 @@ public:
return HANDLE_TYPE;
}
- SharedPtr<ReadableEvent> GetReadableEvent() const;
+ std::shared_ptr<ReadableEvent> GetReadableEvent() const;
void Signal();
void Clear();
@@ -49,7 +51,7 @@ public:
private:
explicit WritableEvent(KernelCore& kernel);
- SharedPtr<ReadableEvent> readable;
+ std::shared_ptr<ReadableEvent> readable;
std::string name; ///< Name of event (optional)
};
diff --git a/src/core/hle/result.h b/src/core/hle/result.h
index 8a3701151..450f61fea 100644
--- a/src/core/hle/result.h
+++ b/src/core/hle/result.h
@@ -147,6 +147,14 @@ constexpr bool operator!=(const ResultCode& a, const ResultCode& b) {
constexpr ResultCode RESULT_SUCCESS(0);
/**
+ * Placeholder result code used for unknown error codes.
+ *
+ * @note This should only be used when a particular error code
+ * is not known yet.
+ */
+constexpr ResultCode RESULT_UNKNOWN(UINT32_MAX);
+
+/**
* This is an optional value type. It holds a `ResultCode` and, if that code is a success code,
* also holds a result of type `T`. If the code is an error code then trying to access the inner
* value fails, thus ensuring that the ResultCode of functions is always checked properly before
@@ -183,7 +191,7 @@ class ResultVal {
public:
/// Constructs an empty `ResultVal` with the given error code. The code must not be a success
/// code.
- ResultVal(ResultCode error_code = ResultCode(-1)) : result_code(error_code) {
+ ResultVal(ResultCode error_code = RESULT_UNKNOWN) : result_code(error_code) {
ASSERT(error_code.IsError());
}
diff --git a/src/core/hle/service/acc/acc.cpp b/src/core/hle/service/acc/acc.cpp
index 0c0f7ed6e..7e3e311fb 100644
--- a/src/core/hle/service/acc/acc.cpp
+++ b/src/core/hle/service/acc/acc.cpp
@@ -84,7 +84,7 @@ protected:
LOG_ERROR(Service_ACC, "Failed to get profile base and data for user={}",
user_id.Format());
IPC::ResponseBuilder rb{ctx, 2};
- rb.Push(ResultCode(-1)); // TODO(ogniK): Get actual error code
+ rb.Push(RESULT_UNKNOWN); // TODO(ogniK): Get actual error code
}
}
@@ -98,7 +98,7 @@ protected:
} else {
LOG_ERROR(Service_ACC, "Failed to get profile base for user={}", user_id.Format());
IPC::ResponseBuilder rb{ctx, 2};
- rb.Push(ResultCode(-1)); // TODO(ogniK): Get actual error code
+ rb.Push(RESULT_UNKNOWN); // TODO(ogniK): Get actual error code
}
}
@@ -442,7 +442,7 @@ void Module::Interface::TrySelectUserWithoutInteraction(Kernel::HLERequestContex
const auto user_list = profile_manager->GetAllUsers();
if (std::all_of(user_list.begin(), user_list.end(),
[](const auto& user) { return user.uuid == Common::INVALID_UUID; })) {
- rb.Push(ResultCode(-1)); // TODO(ogniK): Find the correct error code
+ rb.Push(RESULT_UNKNOWN); // TODO(ogniK): Find the correct error code
rb.PushRaw<u128>(Common::INVALID_UUID);
return;
}
diff --git a/src/core/hle/service/acc/acc_su.cpp b/src/core/hle/service/acc/acc_su.cpp
index 0d1663657..b941c260b 100644
--- a/src/core/hle/service/acc/acc_su.cpp
+++ b/src/core/hle/service/acc/acc_su.cpp
@@ -28,6 +28,7 @@ ACC_SU::ACC_SU(std::shared_ptr<Module> module, std::shared_ptr<ProfileManager> p
{103, nullptr, "GetBaasUserAvailabilityChangeNotifier"},
{104, nullptr, "GetProfileUpdateNotifier"},
{105, nullptr, "CheckNetworkServiceAvailabilityAsync"},
+ {106, nullptr, "GetProfileSyncNotifier"},
{110, nullptr, "StoreSaveDataThumbnail"},
{111, nullptr, "ClearSaveDataThumbnail"},
{112, nullptr, "LoadSaveDataThumbnail"},
@@ -44,6 +45,8 @@ ACC_SU::ACC_SU(std::shared_ptr<Module> module, std::shared_ptr<ProfileManager> p
{205, &ACC_SU::GetProfileEditor, "GetProfileEditor"},
{206, nullptr, "CompleteUserRegistrationForcibly"},
{210, nullptr, "CreateFloatingRegistrationRequest"},
+ {211, nullptr, "CreateProcedureToRegisterUserWithNintendoAccount"},
+ {212, nullptr, "ResumeProcedureToRegisterUserWithNintendoAccount"},
{230, nullptr, "AuthenticateServiceAsync"},
{250, nullptr, "GetBaasAccountAdministrator"},
{290, nullptr, "ProxyProcedureForGuestLoginWithNintendoAccount"},
diff --git a/src/core/hle/service/acc/acc_u1.cpp b/src/core/hle/service/acc/acc_u1.cpp
index 6520b3968..858e91dde 100644
--- a/src/core/hle/service/acc/acc_u1.cpp
+++ b/src/core/hle/service/acc/acc_u1.cpp
@@ -28,6 +28,7 @@ ACC_U1::ACC_U1(std::shared_ptr<Module> module, std::shared_ptr<ProfileManager> p
{103, nullptr, "GetProfileUpdateNotifier"},
{104, nullptr, "CheckNetworkServiceAvailabilityAsync"},
{105, nullptr, "GetBaasUserAvailabilityChangeNotifier"},
+ {106, nullptr, "GetProfileSyncNotifier"},
{110, nullptr, "StoreSaveDataThumbnail"},
{111, nullptr, "ClearSaveDataThumbnail"},
{112, nullptr, "LoadSaveDataThumbnail"},
diff --git a/src/core/hle/service/acc/profile_manager.cpp b/src/core/hle/service/acc/profile_manager.cpp
index 8f9986326..3e756e59e 100644
--- a/src/core/hle/service/acc/profile_manager.cpp
+++ b/src/core/hle/service/acc/profile_manager.cpp
@@ -31,8 +31,8 @@ struct ProfileDataRaw {
static_assert(sizeof(ProfileDataRaw) == 0x650, "ProfileDataRaw has incorrect size.");
// TODO(ogniK): Get actual error codes
-constexpr ResultCode ERROR_TOO_MANY_USERS(ErrorModule::Account, -1);
-constexpr ResultCode ERROR_USER_ALREADY_EXISTS(ErrorModule::Account, -2);
+constexpr ResultCode ERROR_TOO_MANY_USERS(ErrorModule::Account, u32(-1));
+constexpr ResultCode ERROR_USER_ALREADY_EXISTS(ErrorModule::Account, u32(-2));
constexpr ResultCode ERROR_ARGUMENT_IS_NULL(ErrorModule::Account, 20);
constexpr char ACC_SAVE_AVATORS_BASE_PATH[] = "/system/save/8000000000000010/su/avators/";
diff --git a/src/core/hle/service/am/am.cpp b/src/core/hle/service/am/am.cpp
index ba54b3040..95aa5d23d 100644
--- a/src/core/hle/service/am/am.cpp
+++ b/src/core/hle/service/am/am.cpp
@@ -229,7 +229,15 @@ IDebugFunctions::IDebugFunctions() : ServiceFramework{"IDebugFunctions"} {
{20, nullptr, "InvalidateTransitionLayer"},
{30, nullptr, "RequestLaunchApplicationWithUserAndArgumentForDebug"},
{40, nullptr, "GetAppletResourceUsageInfo"},
- {41, nullptr, "SetCpuBoostModeForApplet"},
+ {100, nullptr, "SetCpuBoostModeForApplet"},
+ {110, nullptr, "PushToAppletBoundChannelForDebug"},
+ {111, nullptr, "TryPopFromAppletBoundChannelForDebug"},
+ {120, nullptr, "AlarmSettingNotificationEnableAppEventReserve"},
+ {121, nullptr, "AlarmSettingNotificationDisableAppEventReserve"},
+ {122, nullptr, "AlarmSettingNotificationPushAppEventNotify"},
+ {130, nullptr, "FriendInvitationSetApplicationParameter"},
+ {131, nullptr, "FriendInvitationClearApplicationParameter"},
+ {132, nullptr, "FriendInvitationPushApplicationParameter"},
};
// clang-format on
@@ -278,10 +286,12 @@ ISelfController::ISelfController(Core::System& system,
{69, &ISelfController::IsAutoSleepDisabled, "IsAutoSleepDisabled"},
{70, nullptr, "ReportMultimediaError"},
{71, nullptr, "GetCurrentIlluminanceEx"},
+ {72, nullptr, "SetInputDetectionPolicy"},
{80, nullptr, "SetWirelessPriorityMode"},
{90, &ISelfController::GetAccumulatedSuspendedTickValue, "GetAccumulatedSuspendedTickValue"},
{91, &ISelfController::GetAccumulatedSuspendedTickChangedEvent, "GetAccumulatedSuspendedTickChangedEvent"},
{100, nullptr, "SetAlbumImageTakenNotificationEnabled"},
+ {110, nullptr, "SetApplicationAlbumUserData"},
{1000, nullptr, "GetDebugStorageChannel"},
};
// clang-format on
@@ -531,12 +541,11 @@ AppletMessageQueue::AppletMessageQueue(Kernel::KernelCore& kernel) {
AppletMessageQueue::~AppletMessageQueue() = default;
-const Kernel::SharedPtr<Kernel::ReadableEvent>& AppletMessageQueue::GetMesssageRecieveEvent()
- const {
+const std::shared_ptr<Kernel::ReadableEvent>& AppletMessageQueue::GetMesssageRecieveEvent() const {
return on_new_message.readable;
}
-const Kernel::SharedPtr<Kernel::ReadableEvent>& AppletMessageQueue::GetOperationModeChangedEvent()
+const std::shared_ptr<Kernel::ReadableEvent>& AppletMessageQueue::GetOperationModeChangedEvent()
const {
return on_operation_mode_changed.readable;
}
@@ -613,6 +622,7 @@ ICommonStateGetter::ICommonStateGetter(Core::System& system,
{90, nullptr, "SetPerformanceConfigurationChangedNotification"},
{91, nullptr, "GetCurrentPerformanceConfiguration"},
{200, nullptr, "GetOperationModeSystemInfo"},
+ {300, nullptr, "GetSettingsPlatformRegion"},
};
// clang-format on
@@ -991,7 +1001,7 @@ void ILibraryAppletCreator::CreateLibraryApplet(Kernel::HLERequestContext& ctx)
LOG_ERROR(Service_AM, "Applet doesn't exist! applet_id={}", static_cast<u32>(applet_id));
IPC::ResponseBuilder rb{ctx, 2};
- rb.Push(ResultCode(-1));
+ rb.Push(RESULT_UNKNOWN);
return;
}
@@ -1027,7 +1037,7 @@ void ILibraryAppletCreator::CreateTransferMemoryStorage(Kernel::HLERequestContex
if (transfer_mem == nullptr) {
LOG_ERROR(Service_AM, "shared_mem is a nullpr for handle={:08X}", handle);
IPC::ResponseBuilder rb{ctx, 2};
- rb.Push(ResultCode(-1));
+ rb.Push(RESULT_UNKNOWN);
return;
}
@@ -1076,11 +1086,18 @@ IApplicationFunctions::IApplicationFunctions(Core::System& system_)
{100, &IApplicationFunctions::InitializeApplicationCopyrightFrameBuffer, "InitializeApplicationCopyrightFrameBuffer"},
{101, &IApplicationFunctions::SetApplicationCopyrightImage, "SetApplicationCopyrightImage"},
{102, &IApplicationFunctions::SetApplicationCopyrightVisibility, "SetApplicationCopyrightVisibility"},
- {110, nullptr, "QueryApplicationPlayStatistics"},
+ {110, &IApplicationFunctions::QueryApplicationPlayStatistics, "QueryApplicationPlayStatistics"},
+ {111, &IApplicationFunctions::QueryApplicationPlayStatisticsByUid, "QueryApplicationPlayStatisticsByUid"},
{120, nullptr, "ExecuteProgram"},
{121, nullptr, "ClearUserChannel"},
{122, nullptr, "UnpopToUserChannel"},
{130, &IApplicationFunctions::GetGpuErrorDetectedSystemEvent, "GetGpuErrorDetectedSystemEvent"},
+ {140, nullptr, "GetFriendInvitationStorageChannelEvent"},
+ {141, nullptr, "TryPopFromFriendInvitationStorageChannel"},
+ {150, nullptr, "GetNotificationStorageChannelEvent"},
+ {151, nullptr, "TryPopFromNotificationStorageChannel"},
+ {160, nullptr, "GetHealthWarningDisappearedSystemEvent"},
+ {170, nullptr, "SetHdcpAuthenticationActivated"},
{500, nullptr, "StartContinuousRecordingFlushForDebug"},
{1000, nullptr, "CreateMovieMaker"},
{1001, nullptr, "PrepareForJit"},
@@ -1335,12 +1352,16 @@ void IApplicationFunctions::GetPseudoDeviceId(Kernel::HLERequestContext& ctx) {
}
void IApplicationFunctions::ExtendSaveData(Kernel::HLERequestContext& ctx) {
+ struct Parameters {
+ FileSys::SaveDataType type;
+ u128 user_id;
+ u64 new_normal_size;
+ u64 new_journal_size;
+ };
+ static_assert(sizeof(Parameters) == 40);
+
IPC::RequestParser rp{ctx};
- const auto type{rp.PopRaw<FileSys::SaveDataType>()};
- rp.Skip(1, false);
- const auto user_id{rp.PopRaw<u128>()};
- const auto new_normal_size{rp.PopRaw<u64>()};
- const auto new_journal_size{rp.PopRaw<u64>()};
+ const auto [type, user_id, new_normal_size, new_journal_size] = rp.PopRaw<Parameters>();
LOG_DEBUG(Service_AM,
"called with type={:02X}, user_id={:016X}{:016X}, new_normal={:016X}, "
@@ -1359,10 +1380,14 @@ void IApplicationFunctions::ExtendSaveData(Kernel::HLERequestContext& ctx) {
}
void IApplicationFunctions::GetSaveDataSize(Kernel::HLERequestContext& ctx) {
+ struct Parameters {
+ FileSys::SaveDataType type;
+ u128 user_id;
+ };
+ static_assert(sizeof(Parameters) == 24);
+
IPC::RequestParser rp{ctx};
- const auto type{rp.PopRaw<FileSys::SaveDataType>()};
- rp.Skip(1, false);
- const auto user_id{rp.PopRaw<u128>()};
+ const auto [type, user_id] = rp.PopRaw<Parameters>();
LOG_DEBUG(Service_AM, "called with type={:02X}, user_id={:016X}{:016X}", static_cast<u8>(type),
user_id[1], user_id[0]);
@@ -1376,6 +1401,22 @@ void IApplicationFunctions::GetSaveDataSize(Kernel::HLERequestContext& ctx) {
rb.Push(size.journal);
}
+void IApplicationFunctions::QueryApplicationPlayStatistics(Kernel::HLERequestContext& ctx) {
+ LOG_WARNING(Service_AM, "(STUBBED) called");
+
+ IPC::ResponseBuilder rb{ctx, 3};
+ rb.Push(RESULT_SUCCESS);
+ rb.Push<u32>(0);
+}
+
+void IApplicationFunctions::QueryApplicationPlayStatisticsByUid(Kernel::HLERequestContext& ctx) {
+ LOG_WARNING(Service_AM, "(STUBBED) called");
+
+ IPC::ResponseBuilder rb{ctx, 3};
+ rb.Push(RESULT_SUCCESS);
+ rb.Push<u32>(0);
+}
+
void IApplicationFunctions::GetGpuErrorDetectedSystemEvent(Kernel::HLERequestContext& ctx) {
LOG_WARNING(Service_AM, "(STUBBED) called");
@@ -1409,6 +1450,8 @@ IHomeMenuFunctions::IHomeMenuFunctions() : ServiceFramework("IHomeMenuFunctions"
{30, nullptr, "GetHomeButtonWriterLockAccessor"},
{31, nullptr, "GetWriterLockAccessorEx"},
{100, nullptr, "PopRequestLaunchApplicationForDebug"},
+ {110, nullptr, "IsForceTerminateApplicationDisabledForDebug"},
+ {200, nullptr, "LaunchDevMenu"},
};
// clang-format on
diff --git a/src/core/hle/service/am/am.h b/src/core/hle/service/am/am.h
index 2ae9402a8..448817be9 100644
--- a/src/core/hle/service/am/am.h
+++ b/src/core/hle/service/am/am.h
@@ -54,8 +54,8 @@ public:
explicit AppletMessageQueue(Kernel::KernelCore& kernel);
~AppletMessageQueue();
- const Kernel::SharedPtr<Kernel::ReadableEvent>& GetMesssageRecieveEvent() const;
- const Kernel::SharedPtr<Kernel::ReadableEvent>& GetOperationModeChangedEvent() const;
+ const std::shared_ptr<Kernel::ReadableEvent>& GetMesssageRecieveEvent() const;
+ const std::shared_ptr<Kernel::ReadableEvent>& GetOperationModeChangedEvent() const;
void PushMessage(AppletMessage msg);
AppletMessage PopMessage();
std::size_t GetMessageCount() const;
@@ -255,6 +255,8 @@ private:
void InitializeApplicationCopyrightFrameBuffer(Kernel::HLERequestContext& ctx);
void SetApplicationCopyrightImage(Kernel::HLERequestContext& ctx);
void SetApplicationCopyrightVisibility(Kernel::HLERequestContext& ctx);
+ void QueryApplicationPlayStatistics(Kernel::HLERequestContext& ctx);
+ void QueryApplicationPlayStatisticsByUid(Kernel::HLERequestContext& ctx);
void GetGpuErrorDetectedSystemEvent(Kernel::HLERequestContext& ctx);
bool launch_popped_application_specific = false;
diff --git a/src/core/hle/service/am/applets/applets.cpp b/src/core/hle/service/am/applets/applets.cpp
index 673ad1f7f..92f995f8f 100644
--- a/src/core/hle/service/am/applets/applets.cpp
+++ b/src/core/hle/service/am/applets/applets.cpp
@@ -108,15 +108,15 @@ void AppletDataBroker::SignalStateChanged() const {
state_changed_event.writable->Signal();
}
-Kernel::SharedPtr<Kernel::ReadableEvent> AppletDataBroker::GetNormalDataEvent() const {
+std::shared_ptr<Kernel::ReadableEvent> AppletDataBroker::GetNormalDataEvent() const {
return pop_out_data_event.readable;
}
-Kernel::SharedPtr<Kernel::ReadableEvent> AppletDataBroker::GetInteractiveDataEvent() const {
+std::shared_ptr<Kernel::ReadableEvent> AppletDataBroker::GetInteractiveDataEvent() const {
return pop_interactive_out_data_event.readable;
}
-Kernel::SharedPtr<Kernel::ReadableEvent> AppletDataBroker::GetStateChangedEvent() const {
+std::shared_ptr<Kernel::ReadableEvent> AppletDataBroker::GetStateChangedEvent() const {
return state_changed_event.readable;
}
diff --git a/src/core/hle/service/am/applets/applets.h b/src/core/hle/service/am/applets/applets.h
index 226be88b1..16e61fc6f 100644
--- a/src/core/hle/service/am/applets/applets.h
+++ b/src/core/hle/service/am/applets/applets.h
@@ -86,9 +86,9 @@ public:
void SignalStateChanged() const;
- Kernel::SharedPtr<Kernel::ReadableEvent> GetNormalDataEvent() const;
- Kernel::SharedPtr<Kernel::ReadableEvent> GetInteractiveDataEvent() const;
- Kernel::SharedPtr<Kernel::ReadableEvent> GetStateChangedEvent() const;
+ std::shared_ptr<Kernel::ReadableEvent> GetNormalDataEvent() const;
+ std::shared_ptr<Kernel::ReadableEvent> GetInteractiveDataEvent() const;
+ std::shared_ptr<Kernel::ReadableEvent> GetStateChangedEvent() const;
private:
// Queues are named from applet's perspective
diff --git a/src/core/hle/service/am/applets/web_browser.cpp b/src/core/hle/service/am/applets/web_browser.cpp
index 32283e819..5546ef6e8 100644
--- a/src/core/hle/service/am/applets/web_browser.cpp
+++ b/src/core/hle/service/am/applets/web_browser.cpp
@@ -337,7 +337,7 @@ void WebBrowser::ExecuteInternal() {
void WebBrowser::InitializeShop() {
if (frontend_e_commerce == nullptr) {
LOG_ERROR(Service_AM, "Missing ECommerce Applet frontend!");
- status = ResultCode(-1);
+ status = RESULT_UNKNOWN;
return;
}
@@ -353,7 +353,7 @@ void WebBrowser::InitializeShop() {
if (url == args.end()) {
LOG_ERROR(Service_AM, "Missing EShop Arguments URL for initialization!");
- status = ResultCode(-1);
+ status = RESULT_UNKNOWN;
return;
}
@@ -366,7 +366,7 @@ void WebBrowser::InitializeShop() {
// Less is missing info, More is malformed
if (split_query.size() != 2) {
LOG_ERROR(Service_AM, "EShop Arguments has more than one question mark, malformed");
- status = ResultCode(-1);
+ status = RESULT_UNKNOWN;
return;
}
@@ -390,7 +390,7 @@ void WebBrowser::InitializeShop() {
if (scene == shop_query.end()) {
LOG_ERROR(Service_AM, "No scene parameter was passed via shop query!");
- status = ResultCode(-1);
+ status = RESULT_UNKNOWN;
return;
}
@@ -406,7 +406,7 @@ void WebBrowser::InitializeShop() {
const auto target = target_map.find(scene->second);
if (target == target_map.end()) {
LOG_ERROR(Service_AM, "Scene for shop query is invalid! (scene={})", scene->second);
- status = ResultCode(-1);
+ status = RESULT_UNKNOWN;
return;
}
@@ -427,7 +427,7 @@ void WebBrowser::InitializeOffline() {
if (args.find(WebArgTLVType::DocumentPath) == args.end() ||
args.find(WebArgTLVType::DocumentKind) == args.end() ||
args.find(WebArgTLVType::ApplicationID) == args.end()) {
- status = ResultCode(-1);
+ status = RESULT_UNKNOWN;
LOG_ERROR(Service_AM, "Missing necessary parameters for initialization!");
}
@@ -476,7 +476,7 @@ void WebBrowser::InitializeOffline() {
offline_romfs = GetApplicationRomFS(system, title_id, type);
if (offline_romfs == nullptr) {
- status = ResultCode(-1);
+ status = RESULT_UNKNOWN;
LOG_ERROR(Service_AM, "Failed to find offline data for request!");
}
@@ -496,7 +496,7 @@ void WebBrowser::ExecuteShop() {
const auto check_optional_parameter = [this](const auto& p) {
if (!p.has_value()) {
LOG_ERROR(Service_AM, "Missing one or more necessary parameters for execution!");
- status = ResultCode(-1);
+ status = RESULT_UNKNOWN;
return false;
}
diff --git a/src/core/hle/service/am/idle.cpp b/src/core/hle/service/am/idle.cpp
index f814fe2c0..d256d57c8 100644
--- a/src/core/hle/service/am/idle.cpp
+++ b/src/core/hle/service/am/idle.cpp
@@ -10,7 +10,7 @@ IdleSys::IdleSys() : ServiceFramework{"idle:sys"} {
// clang-format off
static const FunctionInfo functions[] = {
{0, nullptr, "GetAutoPowerDownEvent"},
- {1, nullptr, "Unknown1"},
+ {1, nullptr, "IsAutoPowerDownRequested"},
{2, nullptr, "Unknown2"},
{3, nullptr, "SetHandlingContext"},
{4, nullptr, "LoadAndApplySettings"},
diff --git a/src/core/hle/service/am/omm.cpp b/src/core/hle/service/am/omm.cpp
index 6ab3fb906..37389ccda 100644
--- a/src/core/hle/service/am/omm.cpp
+++ b/src/core/hle/service/am/omm.cpp
@@ -35,6 +35,8 @@ OMM::OMM() : ServiceFramework{"omm"} {
{23, nullptr, "GetHdcpState"},
{24, nullptr, "ShowCardUpdateProcessing"},
{25, nullptr, "SetApplicationCecSettingsAndNotifyChanged"},
+ {26, nullptr, "GetOperationModeSystemInfo"},
+ {27, nullptr, "GetAppletFullAwakingSystemEvent"},
};
// clang-format on
diff --git a/src/core/hle/service/aoc/aoc_u.cpp b/src/core/hle/service/aoc/aoc_u.cpp
index f36ccbc49..4227a4adf 100644
--- a/src/core/hle/service/aoc/aoc_u.cpp
+++ b/src/core/hle/service/aoc/aoc_u.cpp
@@ -61,6 +61,7 @@ AOC_U::AOC_U(Core::System& system)
{7, &AOC_U::PrepareAddOnContent, "PrepareAddOnContent"},
{8, &AOC_U::GetAddOnContentListChangedEvent, "GetAddOnContentListChangedEvent"},
{100, nullptr, "CreateEcPurchasedEventManager"},
+ {101, nullptr, "CreatePermanentEcPurchasedEventManager"},
};
// clang-format on
@@ -131,7 +132,7 @@ void AOC_U::ListAddOnContent(Kernel::HLERequestContext& ctx) {
if (out.size() < offset) {
IPC::ResponseBuilder rb{ctx, 2};
// TODO(DarkLordZach): Find the correct error code.
- rb.Push(ResultCode(-1));
+ rb.Push(RESULT_UNKNOWN);
return;
}
diff --git a/src/core/hle/service/audio/audctl.cpp b/src/core/hle/service/audio/audctl.cpp
index 6a01d4d29..9e08e5346 100644
--- a/src/core/hle/service/audio/audctl.cpp
+++ b/src/core/hle/service/audio/audctl.cpp
@@ -38,6 +38,7 @@ AudCtl::AudCtl() : ServiceFramework{"audctl"} {
{24, nullptr, "GetSystemOutputMasterVolume"},
{25, nullptr, "GetAudioVolumeDataForPlayReport"},
{26, nullptr, "UpdateHeadphoneSettings"},
+ {27, nullptr, "SetVolumeMappingTableForDev"},
};
// clang-format on
diff --git a/src/core/hle/service/audio/audout_u.cpp b/src/core/hle/service/audio/audout_u.cpp
index 6a29377e3..4fb2cbc4b 100644
--- a/src/core/hle/service/audio/audout_u.cpp
+++ b/src/core/hle/service/audio/audout_u.cpp
@@ -43,7 +43,8 @@ public:
IAudioOut(Core::System& system, AudoutParams audio_params, AudioCore::AudioOut& audio_core,
std::string&& device_name, std::string&& unique_name)
: ServiceFramework("IAudioOut"), audio_core(audio_core),
- device_name(std::move(device_name)), audio_params(audio_params) {
+ device_name(std::move(device_name)),
+ audio_params(audio_params), main_memory{system.Memory()} {
// clang-format off
static const FunctionInfo functions[] = {
{0, &IAudioOut::GetAudioOutState, "GetAudioOutState"},
@@ -137,7 +138,7 @@ private:
const u64 tag{rp.Pop<u64>()};
std::vector<s16> samples(audio_buffer.buffer_size / sizeof(s16));
- Memory::ReadBlock(audio_buffer.buffer, samples.data(), audio_buffer.buffer_size);
+ main_memory.ReadBlock(audio_buffer.buffer, samples.data(), audio_buffer.buffer_size);
if (!audio_core.QueueBuffer(stream, tag, std::move(samples))) {
IPC::ResponseBuilder rb{ctx, 2};
@@ -209,6 +210,7 @@ private:
/// This is the event handle used to check if the audio buffer was released
Kernel::EventPair buffer_event;
+ Memory::Memory& main_memory;
};
AudOutU::AudOutU(Core::System& system_) : ServiceFramework("audout:u"), system{system_} {
diff --git a/src/core/hle/service/audio/audren_u.cpp b/src/core/hle/service/audio/audren_u.cpp
index 4ea7ade6e..82a5dbf14 100644
--- a/src/core/hle/service/audio/audren_u.cpp
+++ b/src/core/hle/service/audio/audren_u.cpp
@@ -49,8 +49,9 @@ public:
system_event =
Kernel::WritableEvent::CreateEventPair(system.Kernel(), "IAudioRenderer:SystemEvent");
- renderer = std::make_unique<AudioCore::AudioRenderer>(
- system.CoreTiming(), audren_params, system_event.writable, instance_number);
+ renderer = std::make_unique<AudioCore::AudioRenderer>(system.CoreTiming(), system.Memory(),
+ audren_params, system_event.writable,
+ instance_number);
}
private:
diff --git a/src/core/hle/service/audio/hwopus.cpp b/src/core/hle/service/audio/hwopus.cpp
index cb4a1160d..cb839e4a2 100644
--- a/src/core/hle/service/audio/hwopus.cpp
+++ b/src/core/hle/service/audio/hwopus.cpp
@@ -80,7 +80,7 @@ private:
LOG_ERROR(Audio, "Failed to decode opus data");
IPC::ResponseBuilder rb{ctx, 2};
// TODO(ogniK): Use correct error code
- rb.Push(ResultCode(-1));
+ rb.Push(RESULT_UNKNOWN);
return;
}
@@ -278,7 +278,7 @@ void HwOpus::OpenOpusDecoder(Kernel::HLERequestContext& ctx) {
LOG_ERROR(Audio, "Failed to create Opus decoder (error={}).", error);
IPC::ResponseBuilder rb{ctx, 2};
// TODO(ogniK): Use correct error code
- rb.Push(ResultCode(-1));
+ rb.Push(RESULT_UNKNOWN);
return;
}
diff --git a/src/core/hle/service/bcat/backend/backend.cpp b/src/core/hle/service/bcat/backend/backend.cpp
index dec0849b8..6f5ea095a 100644
--- a/src/core/hle/service/bcat/backend/backend.cpp
+++ b/src/core/hle/service/bcat/backend/backend.cpp
@@ -16,7 +16,7 @@ ProgressServiceBackend::ProgressServiceBackend(Kernel::KernelCore& kernel,
kernel, std::string("ProgressServiceBackend:UpdateEvent:").append(event_name));
}
-Kernel::SharedPtr<Kernel::ReadableEvent> ProgressServiceBackend::GetEvent() const {
+std::shared_ptr<Kernel::ReadableEvent> ProgressServiceBackend::GetEvent() const {
return event.readable;
}
diff --git a/src/core/hle/service/bcat/backend/backend.h b/src/core/hle/service/bcat/backend/backend.h
index ea4b16ad0..48bbbe66f 100644
--- a/src/core/hle/service/bcat/backend/backend.h
+++ b/src/core/hle/service/bcat/backend/backend.h
@@ -98,7 +98,7 @@ public:
private:
explicit ProgressServiceBackend(Kernel::KernelCore& kernel, std::string_view event_name);
- Kernel::SharedPtr<Kernel::ReadableEvent> GetEvent() const;
+ std::shared_ptr<Kernel::ReadableEvent> GetEvent() const;
DeliveryCacheProgressImpl& GetImpl();
void SignalUpdate() const;
diff --git a/src/core/hle/service/bcat/backend/boxcat.cpp b/src/core/hle/service/bcat/backend/boxcat.cpp
index 918159e11..67e39a5c4 100644
--- a/src/core/hle/service/bcat/backend/boxcat.cpp
+++ b/src/core/hle/service/bcat/backend/boxcat.cpp
@@ -114,7 +114,7 @@ void HandleDownloadDisplayResult(const AM::Applets::AppletManager& applet_manage
const auto& frontend{applet_manager.GetAppletFrontendSet()};
frontend.error->ShowCustomErrorText(
- ResultCode(-1), "There was an error while attempting to use Boxcat.",
+ RESULT_UNKNOWN, "There was an error while attempting to use Boxcat.",
DOWNLOAD_RESULT_LOG_MESSAGES[static_cast<std::size_t>(res)], [] {});
}
@@ -255,7 +255,7 @@ private:
using Digest = std::array<u8, 0x20>;
static Digest DigestFile(std::vector<u8> bytes) {
Digest out{};
- mbedtls_sha256(bytes.data(), bytes.size(), out.data(), 0);
+ mbedtls_sha256_ret(bytes.data(), bytes.size(), out.data(), 0);
return out;
}
diff --git a/src/core/hle/service/bcat/module.cpp b/src/core/hle/service/bcat/module.cpp
index 6d9d1527d..7ada67130 100644
--- a/src/core/hle/service/bcat/module.cpp
+++ b/src/core/hle/service/bcat/module.cpp
@@ -46,7 +46,7 @@ u64 GetCurrentBuildID(const Core::System::CurrentBuildProcessID& id) {
BCATDigest DigestFile(const FileSys::VirtualFile& file) {
BCATDigest out{};
const auto bytes = file->ReadAllBytes();
- mbedtls_md5(bytes.data(), bytes.size(), out.data());
+ mbedtls_md5_ret(bytes.data(), bytes.size(), out.data());
return out;
}
@@ -87,7 +87,7 @@ struct DeliveryCacheDirectoryEntry {
class IDeliveryCacheProgressService final : public ServiceFramework<IDeliveryCacheProgressService> {
public:
- IDeliveryCacheProgressService(Kernel::SharedPtr<Kernel::ReadableEvent> event,
+ IDeliveryCacheProgressService(std::shared_ptr<Kernel::ReadableEvent> event,
const DeliveryCacheProgressImpl& impl)
: ServiceFramework{"IDeliveryCacheProgressService"}, event(std::move(event)), impl(impl) {
// clang-format off
@@ -118,7 +118,7 @@ private:
rb.Push(RESULT_SUCCESS);
}
- Kernel::SharedPtr<Kernel::ReadableEvent> event;
+ std::shared_ptr<Kernel::ReadableEvent> event;
const DeliveryCacheProgressImpl& impl;
};
@@ -137,14 +137,20 @@ public:
{10200, nullptr, "CancelSyncDeliveryCacheRequest"},
{20100, nullptr, "RequestSyncDeliveryCacheWithApplicationId"},
{20101, nullptr, "RequestSyncDeliveryCacheWithApplicationIdAndDirectoryName"},
+ {20300, nullptr, "GetDeliveryCacheStorageUpdateNotifier"},
+ {20301, nullptr, "RequestSuspendDeliveryTask"},
+ {20400, nullptr, "RegisterSystemApplicationDeliveryTask"},
+ {20401, nullptr, "UnregisterSystemApplicationDeliveryTask"},
{30100, &IBcatService::SetPassphrase, "SetPassphrase"},
{30200, nullptr, "RegisterBackgroundDeliveryTask"},
{30201, nullptr, "UnregisterBackgroundDeliveryTask"},
{30202, nullptr, "BlockDeliveryTask"},
{30203, nullptr, "UnblockDeliveryTask"},
+ {30300, nullptr, "RegisterSystemApplicationDeliveryTasks"},
{90100, nullptr, "EnumerateBackgroundDeliveryTask"},
{90200, nullptr, "GetDeliveryList"},
{90201, &IBcatService::ClearDeliveryCacheStorage, "ClearDeliveryCacheStorage"},
+ {90202, nullptr, "ClearDeliveryTaskSubscriptionStatus"},
{90300, nullptr, "GetPushNotificationLog"},
};
// clang-format on
diff --git a/src/core/hle/service/btdrv/btdrv.cpp b/src/core/hle/service/btdrv/btdrv.cpp
index 4574d9572..40a06c9fd 100644
--- a/src/core/hle/service/btdrv/btdrv.cpp
+++ b/src/core/hle/service/btdrv/btdrv.cpp
@@ -155,6 +155,7 @@ public:
{98, nullptr, "SetLeScanParameter"},
{256, nullptr, "GetIsManufacturingMode"},
{257, nullptr, "EmulateBluetoothCrash"},
+ {258, nullptr, "GetBleChannelMap"},
};
// clang-format on
diff --git a/src/core/hle/service/erpt/erpt.cpp b/src/core/hle/service/erpt/erpt.cpp
index d9b32954e..4ec8c3093 100644
--- a/src/core/hle/service/erpt/erpt.cpp
+++ b/src/core/hle/service/erpt/erpt.cpp
@@ -24,6 +24,8 @@ public:
{6, nullptr, "SubmitMultipleCategoryContext"},
{7, nullptr, "UpdateApplicationLaunchTime"},
{8, nullptr, "ClearApplicationLaunchTime"},
+ {9, nullptr, "SubmitAttachment"},
+ {10, nullptr, "CreateReportWithAttachments"},
};
// clang-format on
@@ -38,6 +40,7 @@ public:
static const FunctionInfo functions[] = {
{0, nullptr, "OpenReport"},
{1, nullptr, "OpenManager"},
+ {2, nullptr, "OpenAttachment"},
};
// clang-format on
diff --git a/src/core/hle/service/es/es.cpp b/src/core/hle/service/es/es.cpp
index f77ddd739..df00ae625 100644
--- a/src/core/hle/service/es/es.cpp
+++ b/src/core/hle/service/es/es.cpp
@@ -52,6 +52,8 @@ public:
{34, nullptr, "GetEncryptedTicketSize"},
{35, nullptr, "GetEncryptedTicketData"},
{36, nullptr, "DeleteAllInactiveELicenseRequiredPersonalizedTicket"},
+ {37, nullptr, "OwnTicket2"},
+ {38, nullptr, "OwnTicket3"},
{503, nullptr, "GetTitleKey"},
};
// clang-format on
diff --git a/src/core/hle/service/filesystem/filesystem.cpp b/src/core/hle/service/filesystem/filesystem.cpp
index 11e5c56b7..102017d73 100644
--- a/src/core/hle/service/filesystem/filesystem.cpp
+++ b/src/core/hle/service/filesystem/filesystem.cpp
@@ -58,11 +58,11 @@ ResultCode VfsDirectoryServiceWrapper::CreateFile(const std::string& path_, u64
auto file = dir->CreateFile(FileUtil::GetFilename(path));
if (file == nullptr) {
// TODO(DarkLordZach): Find a better error code for this
- return ResultCode(-1);
+ return RESULT_UNKNOWN;
}
if (!file->Resize(size)) {
// TODO(DarkLordZach): Find a better error code for this
- return ResultCode(-1);
+ return RESULT_UNKNOWN;
}
return RESULT_SUCCESS;
}
@@ -80,7 +80,7 @@ ResultCode VfsDirectoryServiceWrapper::DeleteFile(const std::string& path_) cons
}
if (!dir->DeleteFile(FileUtil::GetFilename(path))) {
// TODO(DarkLordZach): Find a better error code for this
- return ResultCode(-1);
+ return RESULT_UNKNOWN;
}
return RESULT_SUCCESS;
@@ -94,7 +94,7 @@ ResultCode VfsDirectoryServiceWrapper::CreateDirectory(const std::string& path_)
auto new_dir = dir->CreateSubdirectory(FileUtil::GetFilename(path));
if (new_dir == nullptr) {
// TODO(DarkLordZach): Find a better error code for this
- return ResultCode(-1);
+ return RESULT_UNKNOWN;
}
return RESULT_SUCCESS;
}
@@ -104,7 +104,7 @@ ResultCode VfsDirectoryServiceWrapper::DeleteDirectory(const std::string& path_)
auto dir = GetDirectoryRelativeWrapped(backing, FileUtil::GetParentPath(path));
if (!dir->DeleteSubdirectory(FileUtil::GetFilename(path))) {
// TODO(DarkLordZach): Find a better error code for this
- return ResultCode(-1);
+ return RESULT_UNKNOWN;
}
return RESULT_SUCCESS;
}
@@ -114,7 +114,7 @@ ResultCode VfsDirectoryServiceWrapper::DeleteDirectoryRecursively(const std::str
auto dir = GetDirectoryRelativeWrapped(backing, FileUtil::GetParentPath(path));
if (!dir->DeleteSubdirectoryRecursive(FileUtil::GetFilename(path))) {
// TODO(DarkLordZach): Find a better error code for this
- return ResultCode(-1);
+ return RESULT_UNKNOWN;
}
return RESULT_SUCCESS;
}
@@ -125,7 +125,7 @@ ResultCode VfsDirectoryServiceWrapper::CleanDirectoryRecursively(const std::stri
if (!dir->CleanSubdirectoryRecursive(FileUtil::GetFilename(sanitized_path))) {
// TODO(DarkLordZach): Find a better error code for this
- return ResultCode(-1);
+ return RESULT_UNKNOWN;
}
return RESULT_SUCCESS;
@@ -142,7 +142,7 @@ ResultCode VfsDirectoryServiceWrapper::RenameFile(const std::string& src_path_,
return FileSys::ERROR_PATH_NOT_FOUND;
if (!src->Rename(FileUtil::GetFilename(dest_path))) {
// TODO(DarkLordZach): Find a better error code for this
- return ResultCode(-1);
+ return RESULT_UNKNOWN;
}
return RESULT_SUCCESS;
}
@@ -160,7 +160,7 @@ ResultCode VfsDirectoryServiceWrapper::RenameFile(const std::string& src_path_,
if (!src->GetContainingDirectory()->DeleteFile(FileUtil::GetFilename(src_path))) {
// TODO(DarkLordZach): Find a better error code for this
- return ResultCode(-1);
+ return RESULT_UNKNOWN;
}
return RESULT_SUCCESS;
@@ -177,7 +177,7 @@ ResultCode VfsDirectoryServiceWrapper::RenameDirectory(const std::string& src_pa
return FileSys::ERROR_PATH_NOT_FOUND;
if (!src->Rename(FileUtil::GetFilename(dest_path))) {
// TODO(DarkLordZach): Find a better error code for this
- return ResultCode(-1);
+ return RESULT_UNKNOWN;
}
return RESULT_SUCCESS;
}
@@ -189,7 +189,7 @@ ResultCode VfsDirectoryServiceWrapper::RenameDirectory(const std::string& src_pa
src_path, dest_path);
// TODO(DarkLordZach): Find a better error code for this
- return ResultCode(-1);
+ return RESULT_UNKNOWN;
}
ResultVal<FileSys::VirtualFile> VfsDirectoryServiceWrapper::OpenFile(const std::string& path_,
@@ -287,7 +287,7 @@ ResultVal<FileSys::VirtualFile> FileSystemController::OpenRomFSCurrentProcess()
if (romfs_factory == nullptr) {
// TODO(bunnei): Find a better error code for this
- return ResultCode(-1);
+ return RESULT_UNKNOWN;
}
return romfs_factory->OpenCurrentProcess(system.CurrentProcess()->GetTitleID());
@@ -300,7 +300,7 @@ ResultVal<FileSys::VirtualFile> FileSystemController::OpenRomFS(
if (romfs_factory == nullptr) {
// TODO(bunnei): Find a better error code for this
- return ResultCode(-1);
+ return RESULT_UNKNOWN;
}
return romfs_factory->Open(title_id, storage_id, type);
diff --git a/src/core/hle/service/filesystem/fsp_srv.cpp b/src/core/hle/service/filesystem/fsp_srv.cpp
index cbd5466c1..55d62fc5e 100644
--- a/src/core/hle/service/filesystem/fsp_srv.cpp
+++ b/src/core/hle/service/filesystem/fsp_srv.cpp
@@ -256,8 +256,8 @@ public:
// TODO(DarkLordZach): Verify that this is the correct behavior.
// Build entry index now to save time later.
- BuildEntryIndex(entries, backend->GetFiles(), FileSys::File);
- BuildEntryIndex(entries, backend->GetSubdirectories(), FileSys::Directory);
+ BuildEntryIndex(entries, backend->GetFiles(), FileSys::EntryType::File);
+ BuildEntryIndex(entries, backend->GetSubdirectories(), FileSys::EntryType::Directory);
}
private:
@@ -391,13 +391,10 @@ public:
}
void RenameFile(Kernel::HLERequestContext& ctx) {
- std::vector<u8> buffer;
- buffer.resize(ctx.BufferDescriptorX()[0].Size());
- Memory::ReadBlock(ctx.BufferDescriptorX()[0].Address(), buffer.data(), buffer.size());
+ std::vector<u8> buffer = ctx.ReadBuffer(0);
const std::string src_name = Common::StringFromBuffer(buffer);
- buffer.resize(ctx.BufferDescriptorX()[1].Size());
- Memory::ReadBlock(ctx.BufferDescriptorX()[1].Address(), buffer.data(), buffer.size());
+ buffer = ctx.ReadBuffer(1);
const std::string dst_name = Common::StringFromBuffer(buffer);
LOG_DEBUG(Service_FS, "called. file '{}' to file '{}'", src_name, dst_name);
@@ -680,6 +677,7 @@ FSP_SRV::FSP_SRV(FileSystemController& fsc, const Core::Reporter& reporter)
{33, nullptr, "DeleteCacheStorage"},
{34, nullptr, "GetCacheStorageSize"},
{35, nullptr, "CreateSaveDataFileSystemByHashSalt"},
+ {36, nullptr, "OpenHostFileSystemWithOption"},
{51, &FSP_SRV::OpenSaveDataFileSystem, "OpenSaveDataFileSystem"},
{52, nullptr, "OpenSaveDataFileSystemBySystemSaveDataId"},
{53, &FSP_SRV::OpenReadOnlySaveDataFileSystem, "OpenReadOnlySaveDataFileSystem"},
@@ -694,11 +692,14 @@ FSP_SRV::FSP_SRV(FileSystemController& fsc, const Core::Reporter& reporter)
{66, nullptr, "WriteSaveDataFileSystemExtraData2"},
{67, nullptr, "FindSaveDataWithFilter"},
{68, nullptr, "OpenSaveDataInfoReaderBySaveDataFilter"},
+ {69, nullptr, "ReadSaveDataFileSystemExtraDataBySaveDataAttribute"},
+ {70, nullptr, "WriteSaveDataFileSystemExtraDataBySaveDataAttribute"},
{80, nullptr, "OpenSaveDataMetaFile"},
{81, nullptr, "OpenSaveDataTransferManager"},
{82, nullptr, "OpenSaveDataTransferManagerVersion2"},
{83, nullptr, "OpenSaveDataTransferProhibiterForCloudBackUp"},
{84, nullptr, "ListApplicationAccessibleSaveDataOwnerId"},
+ {85, nullptr, "OpenSaveDataTransferManagerForSaveDataRepair"},
{100, nullptr, "OpenImageDirectoryFileSystem"},
{110, nullptr, "OpenContentStorageFileSystem"},
{120, nullptr, "OpenCloudBackupWorkStorageFileSystem"},
@@ -756,6 +757,8 @@ FSP_SRV::FSP_SRV(FileSystemController& fsc, const Core::Reporter& reporter)
{1009, nullptr, "GetAndClearMemoryReportInfo"},
{1010, nullptr, "SetDataStorageRedirectTarget"},
{1011, &FSP_SRV::GetAccessLogVersionInfo, "GetAccessLogVersionInfo"},
+ {1012, nullptr, "GetFsStackUsage"},
+ {1013, nullptr, "UnsetSaveDataRootPath"},
{1100, nullptr, "OverrideSaveDataTransferTokenSignVerificationKey"},
{1110, nullptr, "CorruptSaveDataFileSystemBySaveDataSpaceId2"},
{1200, nullptr, "OpenMultiCommitManager"},
@@ -785,7 +788,7 @@ void FSP_SRV::OpenFileSystemWithPatch(Kernel::HLERequestContext& ctx) {
static_cast<u8>(type), title_id);
IPC::ResponseBuilder rb{ctx, 2, 0, 0};
- rb.Push(ResultCode(-1));
+ rb.Push(RESULT_UNKNOWN);
}
void FSP_SRV::OpenSdCardFileSystem(Kernel::HLERequestContext& ctx) {
@@ -891,7 +894,7 @@ void FSP_SRV::OpenDataStorageByCurrentProcess(Kernel::HLERequestContext& ctx) {
// TODO (bunnei): Find the right error code to use here
LOG_CRITICAL(Service_FS, "no file system interface available!");
IPC::ResponseBuilder rb{ctx, 2};
- rb.Push(ResultCode(-1));
+ rb.Push(RESULT_UNKNOWN);
return;
}
@@ -928,7 +931,7 @@ void FSP_SRV::OpenDataStorageByDataId(Kernel::HLERequestContext& ctx) {
"could not open data storage with title_id={:016X}, storage_id={:02X}", title_id,
static_cast<u8>(storage_id));
IPC::ResponseBuilder rb{ctx, 2};
- rb.Push(ResultCode(-1));
+ rb.Push(RESULT_UNKNOWN);
return;
}
diff --git a/src/core/hle/service/friend/friend.cpp b/src/core/hle/service/friend/friend.cpp
index 1a0214f08..219176c31 100644
--- a/src/core/hle/service/friend/friend.cpp
+++ b/src/core/hle/service/friend/friend.cpp
@@ -60,6 +60,9 @@ public:
{20801, nullptr, "SyncUserSetting"},
{20900, nullptr, "RequestListSummaryOverlayNotification"},
{21000, nullptr, "GetExternalApplicationCatalog"},
+ {22000, nullptr, "GetReceivedFriendInvitationList"},
+ {22001, nullptr, "GetReceivedFriendInvitationDetailedInfo"},
+ {22010, nullptr, "GetReceivedFriendInvitationCountCache"},
{30100, nullptr, "DropFriendNewlyFlags"},
{30101, nullptr, "DeleteFriend"},
{30110, nullptr, "DropFriendNewlyFlag"},
@@ -91,6 +94,8 @@ public:
{30812, nullptr, "ChangePlayLogPermission"},
{30820, nullptr, "IssueFriendCode"},
{30830, nullptr, "ClearPlayLog"},
+ {30900, nullptr, "SendFriendInvitation"},
+ {30910, nullptr, "ReadFriendInvitation"},
{49900, nullptr, "DeleteNetworkServiceAccountCache"},
};
// clang-format on
diff --git a/src/core/hle/service/hid/controllers/npad.cpp b/src/core/hle/service/hid/controllers/npad.cpp
index 79fff517e..4d952adc0 100644
--- a/src/core/hle/service/hid/controllers/npad.cpp
+++ b/src/core/hle/service/hid/controllers/npad.cpp
@@ -501,8 +501,7 @@ void Controller_NPad::VibrateController(const std::vector<u32>& controller_ids,
last_processed_vibration = vibrations.back();
}
-Kernel::SharedPtr<Kernel::ReadableEvent> Controller_NPad::GetStyleSetChangedEvent(
- u32 npad_id) const {
+std::shared_ptr<Kernel::ReadableEvent> Controller_NPad::GetStyleSetChangedEvent(u32 npad_id) const {
// TODO(ogniK): Figure out the best time to signal this event. This event seems that it should
// be signalled at least once, and signaled after a new controller is connected?
const auto& styleset_event = styleset_changed_events[NPadIdToIndex(npad_id)];
diff --git a/src/core/hle/service/hid/controllers/npad.h b/src/core/hle/service/hid/controllers/npad.h
index 16c4caa1f..931f03430 100644
--- a/src/core/hle/service/hid/controllers/npad.h
+++ b/src/core/hle/service/hid/controllers/npad.h
@@ -109,7 +109,7 @@ public:
void VibrateController(const std::vector<u32>& controller_ids,
const std::vector<Vibration>& vibrations);
- Kernel::SharedPtr<Kernel::ReadableEvent> GetStyleSetChangedEvent(u32 npad_id) const;
+ std::shared_ptr<Kernel::ReadableEvent> GetStyleSetChangedEvent(u32 npad_id) const;
Vibration GetLastVibration() const;
void AddNewController(NPadControllerType controller);
diff --git a/src/core/hle/service/hid/hid.cpp b/src/core/hle/service/hid/hid.cpp
index ecc130f6c..89bf8b815 100644
--- a/src/core/hle/service/hid/hid.cpp
+++ b/src/core/hle/service/hid/hid.cpp
@@ -77,15 +77,14 @@ IAppletResource::IAppletResource(Core::System& system)
GetController<Controller_Stubbed>(HidController::Unknown3).SetCommonHeaderOffset(0x5000);
// Register update callbacks
- auto& core_timing = system.CoreTiming();
pad_update_event =
- core_timing.RegisterEvent("HID::UpdatePadCallback", [this](u64 userdata, s64 cycles_late) {
+ Core::Timing::CreateEvent("HID::UpdatePadCallback", [this](u64 userdata, s64 cycles_late) {
UpdateControllers(userdata, cycles_late);
});
// TODO(shinyquagsire23): Other update callbacks? (accel, gyro?)
- core_timing.ScheduleEvent(pad_update_ticks, pad_update_event);
+ system.CoreTiming().ScheduleEvent(pad_update_ticks, pad_update_event);
ReloadInputDevices();
}
@@ -215,6 +214,8 @@ Hid::Hid(Core::System& system) : ServiceFramework("hid"), system(system) {
{132, nullptr, "EnableUnintendedHomeButtonInputProtection"},
{133, nullptr, "SetNpadJoyAssignmentModeSingleWithDestination"},
{134, nullptr, "SetNpadAnalogStickUseCenterClamp"},
+ {135, nullptr, "SetNpadCaptureButtonAssignment"},
+ {136, nullptr, "ClearNpadCaptureButtonAssignment"},
{200, &Hid::GetVibrationDeviceInfo, "GetVibrationDeviceInfo"},
{201, &Hid::SendVibrationValue, "SendVibrationValue"},
{202, &Hid::GetActualVibrationValue, "GetActualVibrationValue"},
@@ -245,6 +246,8 @@ Hid::Hid(Core::System& system) : ServiceFramework("hid"), system(system) {
{404, nullptr, "HasLeftRightBattery"},
{405, nullptr, "GetNpadInterfaceType"},
{406, nullptr, "GetNpadLeftRightInterfaceType"},
+ {407, nullptr, "GetNpadOfHighestBatteryLevelForJoyLeft"},
+ {408, nullptr, "GetNpadOfHighestBatteryLevelForJoyRight"},
{500, nullptr, "GetPalmaConnectionHandle"},
{501, nullptr, "InitializePalma"},
{502, nullptr, "AcquirePalmaOperationCompleteEvent"},
@@ -272,8 +275,13 @@ Hid::Hid(Core::System& system) : ServiceFramework("hid"), system(system) {
{524, nullptr, "PairPalma"},
{525, &Hid::SetPalmaBoostMode, "SetPalmaBoostMode"},
{526, nullptr, "CancelWritePalmaWaveEntry"},
+ {527, nullptr, "EnablePalmaBoostMode"},
+ {528, nullptr, "GetPalmaBluetoothAddress"},
+ {529, nullptr, "SetDisallowedPalmaConnection"},
{1000, nullptr, "SetNpadCommunicationMode"},
{1001, nullptr, "GetNpadCommunicationMode"},
+ {1002, nullptr, "SetTouchScreenConfiguration"},
+ {1003, nullptr, "IsFirmwareUpdateNeededForNotification"},
};
// clang-format on
@@ -969,6 +977,9 @@ public:
{310, nullptr, "GetMaskedSupportedNpadStyleSet"},
{311, nullptr, "SetNpadPlayerLedBlinkingDevice"},
{312, nullptr, "SetSupportedNpadStyleSetAll"},
+ {313, nullptr, "GetNpadCaptureButtonAssignment"},
+ {314, nullptr, "GetAppletFooterUiType"},
+ {315, nullptr, "GetAppletDetailedUiType"},
{321, nullptr, "GetUniquePadsFromNpad"},
{322, nullptr, "GetIrSensorState"},
{323, nullptr, "GetXcdHandleForNpadWithIrSensor"},
@@ -984,6 +995,8 @@ public:
{513, nullptr, "EndPermitVibrationSession"},
{520, nullptr, "EnableHandheldHids"},
{521, nullptr, "DisableHandheldHids"},
+ {522, nullptr, "SetJoyConRailEnabled"},
+ {523, nullptr, "IsJoyConRailEnabled"},
{540, nullptr, "AcquirePlayReportControllerUsageUpdateEvent"},
{541, nullptr, "GetPlayReportControllerUsages"},
{542, nullptr, "AcquirePlayReportRegisteredDeviceUpdateEvent"},
@@ -1010,6 +1023,7 @@ public:
{809, nullptr, "GetUniquePadSerialNumber"},
{810, nullptr, "GetUniquePadControllerNumber"},
{811, nullptr, "GetSixAxisSensorUserCalibrationStage"},
+ {812, nullptr, "GetConsoleUniqueSixAxisSensorHandle"},
{821, nullptr, "StartAnalogStickManualCalibration"},
{822, nullptr, "RetryCurrentAnalogStickManualCalibrationStage"},
{823, nullptr, "CancelAnalogStickManualCalibration"},
@@ -1020,6 +1034,8 @@ public:
{828, nullptr, "IsAnalogStickInReleasePosition"},
{829, nullptr, "IsAnalogStickInCircumference"},
{830, nullptr, "SetNotificationLedPattern"},
+ {831, nullptr, "SetNotificationLedPatternWithTimeout"},
+ {832, nullptr, "PrepareHidsForNotificationWake"},
{850, nullptr, "IsUsbFullKeyControllerEnabled"},
{851, nullptr, "EnableUsbFullKeyController"},
{852, nullptr, "IsUsbConnected"},
@@ -1049,6 +1065,13 @@ public:
{1132, nullptr, "CheckUsbFirmwareUpdateRequired"},
{1133, nullptr, "StartUsbFirmwareUpdate"},
{1134, nullptr, "GetUsbFirmwareUpdateState"},
+ {1150, nullptr, "SetTouchScreenMagnification"},
+ {1151, nullptr, "GetTouchScreenFirmwareVersion"},
+ {1152, nullptr, "SetTouchScreenDefaultConfiguration"},
+ {1153, nullptr, "GetTouchScreenDefaultConfiguration"},
+ {1154, nullptr, "IsFirmwareAvailableForNotification"},
+ {1155, nullptr, "SetForceHandheldStyleVibration"},
+ {1156, nullptr, "SendConnectionTriggerWithoutTimeoutEvent"},
};
// clang-format on
diff --git a/src/core/hle/service/hid/hid.h b/src/core/hle/service/hid/hid.h
index f08e036a3..ad20f147c 100644
--- a/src/core/hle/service/hid/hid.h
+++ b/src/core/hle/service/hid/hid.h
@@ -67,9 +67,9 @@ private:
void GetSharedMemoryHandle(Kernel::HLERequestContext& ctx);
void UpdateControllers(u64 userdata, s64 cycles_late);
- Kernel::SharedPtr<Kernel::SharedMemory> shared_mem;
+ std::shared_ptr<Kernel::SharedMemory> shared_mem;
- Core::Timing::EventType* pad_update_event;
+ std::shared_ptr<Core::Timing::EventType> pad_update_event;
Core::System& system;
std::array<std::unique_ptr<ControllerBase>, static_cast<size_t>(HidController::MaxControllers)>
diff --git a/src/core/hle/service/hid/irs.h b/src/core/hle/service/hid/irs.h
index eb4e898dd..8918ad6ca 100644
--- a/src/core/hle/service/hid/irs.h
+++ b/src/core/hle/service/hid/irs.h
@@ -37,7 +37,7 @@ private:
void RunIrLedProcessor(Kernel::HLERequestContext& ctx);
void StopImageProcessorAsync(Kernel::HLERequestContext& ctx);
void ActivateIrsensorWithFunctionLevel(Kernel::HLERequestContext& ctx);
- Kernel::SharedPtr<Kernel::SharedMemory> shared_mem;
+ std::shared_ptr<Kernel::SharedMemory> shared_mem;
const u32 device_handle{0xABCD};
Core::System& system;
};
diff --git a/src/core/hle/service/ldr/ldr.cpp b/src/core/hle/service/ldr/ldr.cpp
index 499376bfc..157aeec88 100644
--- a/src/core/hle/service/ldr/ldr.cpp
+++ b/src/core/hle/service/ldr/ldr.cpp
@@ -140,9 +140,10 @@ public:
rb.Push(ERROR_INVALID_SIZE);
return;
}
+
// Read NRR data from memory
std::vector<u8> nrr_data(nrr_size);
- Memory::ReadBlock(nrr_address, nrr_data.data(), nrr_size);
+ system.Memory().ReadBlock(nrr_address, nrr_data.data(), nrr_size);
NRRHeader header;
std::memcpy(&header, nrr_data.data(), sizeof(NRRHeader));
@@ -291,10 +292,10 @@ public:
// Read NRO data from memory
std::vector<u8> nro_data(nro_size);
- Memory::ReadBlock(nro_address, nro_data.data(), nro_size);
+ system.Memory().ReadBlock(nro_address, nro_data.data(), nro_size);
SHA256Hash hash{};
- mbedtls_sha256(nro_data.data(), nro_data.size(), hash.data(), 0);
+ mbedtls_sha256_ret(nro_data.data(), nro_data.size(), hash.data(), 0);
// NRO Hash is already loaded
if (std::any_of(nro.begin(), nro.end(), [&hash](const std::pair<VAddr, NROInfo>& info) {
diff --git a/src/core/hle/service/lm/lm.cpp b/src/core/hle/service/lm/lm.cpp
index 435f2d286..346c8f899 100644
--- a/src/core/hle/service/lm/lm.cpp
+++ b/src/core/hle/service/lm/lm.cpp
@@ -17,7 +17,8 @@ namespace Service::LM {
class ILogger final : public ServiceFramework<ILogger> {
public:
- ILogger(Manager& manager) : ServiceFramework("ILogger"), manager(manager) {
+ explicit ILogger(Manager& manager_, Memory::Memory& memory_)
+ : ServiceFramework("ILogger"), manager{manager_}, memory{memory_} {
static const FunctionInfo functions[] = {
{0, &ILogger::Log, "Log"},
{1, &ILogger::SetDestination, "SetDestination"},
@@ -35,15 +36,15 @@ private:
MessageHeader header{};
VAddr addr{ctx.BufferDescriptorX()[0].Address()};
const VAddr end_addr{addr + ctx.BufferDescriptorX()[0].size};
- Memory::ReadBlock(addr, &header, sizeof(MessageHeader));
+ memory.ReadBlock(addr, &header, sizeof(MessageHeader));
addr += sizeof(MessageHeader);
FieldMap fields;
while (addr < end_addr) {
- const auto field = static_cast<Field>(Memory::Read8(addr++));
- const auto length = Memory::Read8(addr++);
+ const auto field = static_cast<Field>(memory.Read8(addr++));
+ const auto length = memory.Read8(addr++);
- if (static_cast<Field>(Memory::Read8(addr)) == Field::Skip) {
+ if (static_cast<Field>(memory.Read8(addr)) == Field::Skip) {
++addr;
}
@@ -54,7 +55,7 @@ private:
}
std::vector<u8> data(length);
- Memory::ReadBlock(addr, data.data(), length);
+ memory.ReadBlock(addr, data.data(), length);
fields.emplace(field, std::move(data));
}
@@ -74,11 +75,13 @@ private:
}
Manager& manager;
+ Memory::Memory& memory;
};
class LM final : public ServiceFramework<LM> {
public:
- explicit LM(Manager& manager) : ServiceFramework{"lm"}, manager(manager) {
+ explicit LM(Manager& manager_, Memory::Memory& memory_)
+ : ServiceFramework{"lm"}, manager{manager_}, memory{memory_} {
// clang-format off
static const FunctionInfo functions[] = {
{0, &LM::OpenLogger, "OpenLogger"},
@@ -94,14 +97,16 @@ private:
IPC::ResponseBuilder rb{ctx, 2, 0, 1};
rb.Push(RESULT_SUCCESS);
- rb.PushIpcInterface<ILogger>(manager);
+ rb.PushIpcInterface<ILogger>(manager, memory);
}
Manager& manager;
+ Memory::Memory& memory;
};
void InstallInterfaces(Core::System& system) {
- std::make_shared<LM>(system.GetLogManager())->InstallAsService(system.ServiceManager());
+ std::make_shared<LM>(system.GetLogManager(), system.Memory())
+ ->InstallAsService(system.ServiceManager());
}
} // namespace Service::LM
diff --git a/src/core/hle/service/mii/mii.cpp b/src/core/hle/service/mii/mii.cpp
index 0b3923ad9..a128edb43 100644
--- a/src/core/hle/service/mii/mii.cpp
+++ b/src/core/hle/service/mii/mii.cpp
@@ -50,6 +50,8 @@ public:
{21, &IDatabaseService::GetIndex, "GetIndex"},
{22, &IDatabaseService::SetInterfaceVersion, "SetInterfaceVersion"},
{23, nullptr, "Convert"},
+ {24, nullptr, "ConvertCoreDataToCharInfo"},
+ {25, nullptr, "ConvertCharInfoToCoreData"},
};
// clang-format on
@@ -242,7 +244,7 @@ private:
const auto index = db.IndexOf(uuid);
if (index > MAX_MIIS) {
// TODO(DarkLordZach): Find a better error code
- rb.Push(ResultCode(-1));
+ rb.Push(RESULT_UNKNOWN);
rb.Push(index);
} else {
rb.Push(RESULT_SUCCESS);
@@ -268,7 +270,7 @@ private:
IPC::ResponseBuilder rb{ctx, 2};
// TODO(DarkLordZach): Find a better error code
- rb.Push(success ? RESULT_SUCCESS : ResultCode(-1));
+ rb.Push(success ? RESULT_SUCCESS : RESULT_UNKNOWN);
}
void AddOrReplace(Kernel::HLERequestContext& ctx) {
@@ -282,7 +284,7 @@ private:
IPC::ResponseBuilder rb{ctx, 2};
// TODO(DarkLordZach): Find a better error code
- rb.Push(success ? RESULT_SUCCESS : ResultCode(-1));
+ rb.Push(success ? RESULT_SUCCESS : RESULT_UNKNOWN);
}
void Delete(Kernel::HLERequestContext& ctx) {
diff --git a/src/core/hle/service/ncm/ncm.cpp b/src/core/hle/service/ncm/ncm.cpp
index b405a4b66..89e283ca5 100644
--- a/src/core/hle/service/ncm/ncm.cpp
+++ b/src/core/hle/service/ncm/ncm.cpp
@@ -61,7 +61,8 @@ public:
{5, nullptr, "RegisterHtmlDocumentPath"},
{6, nullptr, "UnregisterHtmlDocumentPath"},
{7, nullptr, "RedirectHtmlDocumentPath"},
- {8, nullptr, ""},
+ {8, nullptr, "Refresh"},
+ {9, nullptr, "RefreshExcluding"},
};
// clang-format on
@@ -77,6 +78,8 @@ public:
{0, nullptr, "ResolveAddOnContentPath"},
{1, nullptr, "RegisterAddOnContentStorage"},
{2, nullptr, "UnregisterAllAddOnContentPath"},
+ {3, nullptr, "RefreshApplicationAddOnContent"},
+ {4, nullptr, "UnregisterApplicationAddOnContent"},
};
// clang-format on
@@ -118,6 +121,7 @@ public:
{10, nullptr, "InactivateContentStorage"},
{11, nullptr, "ActivateContentMetaDatabase"},
{12, nullptr, "InactivateContentMetaDatabase"},
+ {13, nullptr, "InvalidateRightsIdCache"},
};
// clang-format on
diff --git a/src/core/hle/service/nfc/nfc.cpp b/src/core/hle/service/nfc/nfc.cpp
index ca88bf97f..b7b34ce7e 100644
--- a/src/core/hle/service/nfc/nfc.cpp
+++ b/src/core/hle/service/nfc/nfc.cpp
@@ -215,6 +215,7 @@ public:
{411, nullptr, "AttachActivateEvent"},
{412, nullptr, "AttachDeactivateEvent"},
{500, nullptr, "SetNfcEnabled"},
+ {510, nullptr, "OutputTestWave"},
{1000, nullptr, "ReadMifare"},
{1001, nullptr, "WriteMifare"},
{1300, nullptr, "SendCommandByPassThrough"},
diff --git a/src/core/hle/service/nfp/nfp.cpp b/src/core/hle/service/nfp/nfp.cpp
index 795d7b716..4b79eb81d 100644
--- a/src/core/hle/service/nfp/nfp.cpp
+++ b/src/core/hle/service/nfp/nfp.cpp
@@ -16,10 +16,7 @@
#include "core/hle/service/nfp/nfp_user.h"
namespace Service::NFP {
-
namespace ErrCodes {
-[[maybe_unused]] constexpr ResultCode ERR_TAG_FAILED(ErrorModule::NFP,
- -1); // TODO(ogniK): Find the actual error code
constexpr ResultCode ERR_NO_APPLICATION_AREA(ErrorModule::NFP, 152);
} // namespace ErrCodes
@@ -192,7 +189,7 @@ private:
LOG_DEBUG(Service_NFP, "called");
auto nfc_event = nfp_interface.GetNFCEvent();
- if (!nfc_event->ShouldWait(Kernel::GetCurrentThread()) && !has_attached_handle) {
+ if (!nfc_event->ShouldWait(&ctx.GetThread()) && !has_attached_handle) {
device_state = DeviceState::TagFound;
nfc_event->Clear();
}
@@ -345,7 +342,7 @@ bool Module::Interface::LoadAmiibo(const std::vector<u8>& buffer) {
return true;
}
-const Kernel::SharedPtr<Kernel::ReadableEvent>& Module::Interface::GetNFCEvent() const {
+const std::shared_ptr<Kernel::ReadableEvent>& Module::Interface::GetNFCEvent() const {
return nfc_tag_load.readable;
}
diff --git a/src/core/hle/service/nfp/nfp.h b/src/core/hle/service/nfp/nfp.h
index 9718ef745..200013795 100644
--- a/src/core/hle/service/nfp/nfp.h
+++ b/src/core/hle/service/nfp/nfp.h
@@ -34,7 +34,7 @@ public:
void CreateUserInterface(Kernel::HLERequestContext& ctx);
bool LoadAmiibo(const std::vector<u8>& buffer);
- const Kernel::SharedPtr<Kernel::ReadableEvent>& GetNFCEvent() const;
+ const std::shared_ptr<Kernel::ReadableEvent>& GetNFCEvent() const;
const AmiiboFile& GetAmiiboBuffer() const;
private:
diff --git a/src/core/hle/service/nifm/nifm.cpp b/src/core/hle/service/nifm/nifm.cpp
index feb176668..767158444 100644
--- a/src/core/hle/service/nifm/nifm.cpp
+++ b/src/core/hle/service/nifm/nifm.cpp
@@ -222,6 +222,7 @@ private:
IGeneralService::IGeneralService(Core::System& system)
: ServiceFramework("IGeneralService"), system(system) {
+ // clang-format off
static const FunctionInfo functions[] = {
{1, &IGeneralService::GetClientId, "GetClientId"},
{2, &IGeneralService::CreateScanRequest, "CreateScanRequest"},
@@ -260,7 +261,14 @@ IGeneralService::IGeneralService(Core::System& system)
{36, nullptr, "GetCurrentAccessPoint"},
{37, nullptr, "Shutdown"},
{38, nullptr, "GetAllowedChannels"},
+ {39, nullptr, "NotifyApplicationSuspended"},
+ {40, nullptr, "SetAcceptableNetworkTypeFlag"},
+ {41, nullptr, "GetAcceptableNetworkTypeFlag"},
+ {42, nullptr, "NotifyConnectionStateChanged"},
+ {43, nullptr, "SetWowlDelayedWakeTime"},
};
+ // clang-format on
+
RegisterHandlers(functions);
}
diff --git a/src/core/hle/service/nim/nim.cpp b/src/core/hle/service/nim/nim.cpp
index 7d6cf2070..e85f123e2 100644
--- a/src/core/hle/service/nim/nim.cpp
+++ b/src/core/hle/service/nim/nim.cpp
@@ -116,6 +116,8 @@ public:
{500, nullptr, "RequestSyncTicket"},
{501, nullptr, "RequestDownloadTicket"},
{502, nullptr, "RequestDownloadTicketForPrepurchasedContents"},
+ {503, nullptr, "RequestSyncTicket"},
+ {504, nullptr, "RequestDownloadTicketForPrepurchasedContents2"},
};
// clang-format on
diff --git a/src/core/hle/service/npns/npns.cpp b/src/core/hle/service/npns/npns.cpp
index 8751522ca..aa171473b 100644
--- a/src/core/hle/service/npns/npns.cpp
+++ b/src/core/hle/service/npns/npns.cpp
@@ -44,6 +44,10 @@ public:
{113, nullptr, "DestroyJid"},
{114, nullptr, "AttachJid"},
{115, nullptr, "DetachJid"},
+ {120, nullptr, "CreateNotificationReceiver"},
+ {151, nullptr, "GetStateWithHandover"},
+ {152, nullptr, "GetStateChangeEventWithHandover"},
+ {153, nullptr, "GetDropEventWithHandover"},
{201, nullptr, "RequestChangeStateForceTimed"},
{202, nullptr, "RequestChangeStateForceAsync"},
};
@@ -74,6 +78,9 @@ public:
{104, nullptr, "GetStatistics"},
{111, nullptr, "GetJid"},
{120, nullptr, "CreateNotificationReceiver"},
+ {151, nullptr, "GetStateWithHandover"},
+ {152, nullptr, "GetStateChangeEventWithHandover"},
+ {153, nullptr, "GetDropEventWithHandover"},
};
// clang-format on
diff --git a/src/core/hle/service/ns/ns.cpp b/src/core/hle/service/ns/ns.cpp
index 15c156ce1..fdab3cf78 100644
--- a/src/core/hle/service/ns/ns.cpp
+++ b/src/core/hle/service/ns/ns.cpp
@@ -106,6 +106,7 @@ IApplicationManagerInterface::IApplicationManagerInterface()
{96, nullptr, "AcquireApplicationLaunchInfo"},
{97, nullptr, "GetMainApplicationProgramIndex2"},
{98, nullptr, "EnableApplicationAllThreadDumpOnCrash"},
+ {99, nullptr, "LaunchDevMenu"},
{100, nullptr, "ResetToFactorySettings"},
{101, nullptr, "ResetToFactorySettingsWithoutUserSaveData"},
{102, nullptr, "ResetToFactorySettingsForRefurbishment"},
@@ -130,6 +131,8 @@ IApplicationManagerInterface::IApplicationManagerInterface()
{404, nullptr, "InvalidateApplicationControlCache"},
{405, nullptr, "ListApplicationControlCacheEntryInfo"},
{406, nullptr, "GetApplicationControlProperty"},
+ {407, nullptr, "ListApplicationTitle"},
+ {408, nullptr, "ListApplicationIcon"},
{502, nullptr, "RequestCheckGameCardRegistration"},
{503, nullptr, "RequestGameCardRegistrationGoldPoint"},
{504, nullptr, "RequestRegisterGameCard"},
@@ -138,6 +141,7 @@ IApplicationManagerInterface::IApplicationManagerInterface()
{507, nullptr, "EnsureGameCardAccess"},
{508, nullptr, "GetLastGameCardMountFailureResult"},
{509, nullptr, "ListApplicationIdOnGameCard"},
+ {510, nullptr, "GetGameCardPlatformRegion"},
{600, nullptr, "CountApplicationContentMeta"},
{601, nullptr, "ListApplicationContentMetaStatus"},
{602, nullptr, "ListAvailableAddOnContent"},
@@ -168,6 +172,9 @@ IApplicationManagerInterface::IApplicationManagerInterface()
{910, nullptr, "HasApplicationRecord"},
{911, nullptr, "SetPreInstalledApplication"},
{912, nullptr, "ClearPreInstalledApplicationFlag"},
+ {913, nullptr, "ListAllApplicationRecord"},
+ {914, nullptr, "HideApplicationRecord"},
+ {915, nullptr, "ShowApplicationRecord"},
{1000, nullptr, "RequestVerifyApplicationDeprecated"},
{1001, nullptr, "CorruptApplicationForDebug"},
{1002, nullptr, "RequestVerifyAddOnContentsRights"},
@@ -190,12 +197,14 @@ IApplicationManagerInterface::IApplicationManagerInterface()
{1502, nullptr, "GetLastSdCardFormatUnexpectedResult"},
{1504, nullptr, "InsertSdCard"},
{1505, nullptr, "RemoveSdCard"},
+ {1506, nullptr, "GetSdCardStartupStatus"},
{1600, nullptr, "GetSystemSeedForPseudoDeviceId"},
{1601, nullptr, "ResetSystemSeedForPseudoDeviceId"},
{1700, nullptr, "ListApplicationDownloadingContentMeta"},
{1701, nullptr, "GetApplicationView"},
{1702, nullptr, "GetApplicationDownloadTaskStatus"},
{1703, nullptr, "GetApplicationViewDownloadErrorContext"},
+ {1704, nullptr, "GetApplicationViewWithPromotionInfo"},
{1800, nullptr, "IsNotificationSetupCompleted"},
{1801, nullptr, "GetLastNotificationInfoCount"},
{1802, nullptr, "ListLastNotificationInfo"},
@@ -223,6 +232,7 @@ IApplicationManagerInterface::IApplicationManagerInterface()
{2017, nullptr, "CreateDownloadTask"},
{2018, nullptr, "GetApplicationDeliveryInfoHash"},
{2050, nullptr, "GetApplicationRightsOnClient"},
+ {2051, nullptr, "InvalidateRightsIdCache"},
{2100, nullptr, "GetApplicationTerminateResult"},
{2101, nullptr, "GetRawApplicationTerminateResult"},
{2150, nullptr, "CreateRightsEnvironment"},
@@ -230,6 +240,8 @@ IApplicationManagerInterface::IApplicationManagerInterface()
{2152, nullptr, "ActivateRightsEnvironment"},
{2153, nullptr, "DeactivateRightsEnvironment"},
{2154, nullptr, "ForceActivateRightsContextForExit"},
+ {2155, nullptr, "UpdateRightsEnvironmentStatus"},
+ {2156, nullptr, "CreateRightsEnvironmentForPreomia"},
{2160, nullptr, "AddTargetApplicationToRightsEnvironment"},
{2161, nullptr, "SetUsersToRightsEnvironment"},
{2170, nullptr, "GetRightsEnvironmentStatus"},
@@ -243,6 +255,20 @@ IApplicationManagerInterface::IApplicationManagerInterface()
{2201, nullptr, "GetInstalledApplicationCopyIdentifier"},
{2250, nullptr, "RequestReportActiveELicence"},
{2300, nullptr, "ListEventLog"},
+ {2350, nullptr, "PerformAutoUpdateByApplicationId"},
+ {2351, nullptr, "RequestNoDownloadRightsErrorResolution"},
+ {2352, nullptr, "RequestResolveNoDownloadRightsError"},
+ {2400, nullptr, "GetPromotionInfo"},
+ {2401, nullptr, "CountPromotionInfo"},
+ {2402, nullptr, "ListPromotionInfo"},
+ {2403, nullptr, "ImportPromotionJsonForDebug"},
+ {2404, nullptr, "ClearPromotionInfoForDebug"},
+ {2500, nullptr, "ConfirmAvailableTime"},
+ {2510, nullptr, "CreateApplicationResource"},
+ {2511, nullptr, "GetApplicationResource"},
+ {2513, nullptr, "LaunchPreomia"},
+ {2514, nullptr, "ClearTaskOfAsyncTaskManager"},
+ {2800, nullptr, "GetApplicationIdOfPreomia"},
};
// clang-format on
@@ -271,7 +297,7 @@ void IApplicationManagerInterface::GetApplicationControlData(Kernel::HLERequestC
"output buffer is too small! (actual={:016X}, expected_min=0x4000)", size);
IPC::ResponseBuilder rb{ctx, 2};
// TODO(DarkLordZach): Find a better error code for this.
- rb.Push(ResultCode(-1));
+ rb.Push(RESULT_UNKNOWN);
return;
}
@@ -291,7 +317,7 @@ void IApplicationManagerInterface::GetApplicationControlData(Kernel::HLERequestC
0x4000 + control.second->GetSize());
IPC::ResponseBuilder rb{ctx, 2};
// TODO(DarkLordZach): Find a better error code for this.
- rb.Push(ResultCode(-1));
+ rb.Push(RESULT_UNKNOWN);
return;
}
@@ -463,6 +489,7 @@ IECommerceInterface::IECommerceInterface() : ServiceFramework{"IECommerceInterfa
{3, nullptr, "RequestSyncRights"},
{4, nullptr, "RequestUnlinkDevice"},
{5, nullptr, "RequestRevokeAllELicense"},
+ {6, nullptr, "RequestSyncRightsBasedOnAssignedELicenses"},
};
// clang-format on
diff --git a/src/core/hle/service/ns/pl_u.cpp b/src/core/hle/service/ns/pl_u.cpp
index 23477315f..8da4e52c5 100644
--- a/src/core/hle/service/ns/pl_u.cpp
+++ b/src/core/hle/service/ns/pl_u.cpp
@@ -97,7 +97,7 @@ void EncryptSharedFont(const std::vector<u32>& input, std::vector<u8>& output,
const auto key = Common::swap32(EXPECTED_RESULT ^ EXPECTED_MAGIC);
std::vector<u32> transformed_font(input.size() + 2);
transformed_font[0] = Common::swap32(EXPECTED_MAGIC);
- transformed_font[1] = Common::swap32(input.size() * sizeof(u32)) ^ key;
+ transformed_font[1] = Common::swap32(static_cast<u32>(input.size() * sizeof(u32))) ^ key;
std::transform(input.begin(), input.end(), transformed_font.begin() + 2,
[key](u32 in) { return in ^ key; });
std::memcpy(output.data() + offset, transformed_font.data(),
@@ -141,7 +141,7 @@ struct PL_U::Impl {
}
/// Handle to shared memory region designated for a shared font
- Kernel::SharedPtr<Kernel::SharedMemory> shared_font_mem;
+ std::shared_ptr<Kernel::SharedMemory> shared_font_mem;
/// Backing memory for the shared font data
std::shared_ptr<Kernel::PhysicalMemory> shared_font;
@@ -152,7 +152,7 @@ struct PL_U::Impl {
PL_U::PL_U(Core::System& system)
: ServiceFramework("pl:u"), impl{std::make_unique<Impl>()}, system(system) {
-
+ // clang-format off
static const FunctionInfo functions[] = {
{0, &PL_U::RequestLoad, "RequestLoad"},
{1, &PL_U::GetLoadState, "GetLoadState"},
@@ -160,7 +160,13 @@ PL_U::PL_U(Core::System& system)
{3, &PL_U::GetSharedMemoryAddressOffset, "GetSharedMemoryAddressOffset"},
{4, &PL_U::GetSharedMemoryNativeHandle, "GetSharedMemoryNativeHandle"},
{5, &PL_U::GetSharedFontInOrderOfPriority, "GetSharedFontInOrderOfPriority"},
+ {6, nullptr, "GetSharedFontInOrderOfPriorityForSystem"},
+ {100, nullptr, "RequestApplicationFunctionAuthorization"},
+ {101, nullptr, "RequestApplicationFunctionAuthorizationForSystem"},
+ {1000, nullptr, "LoadNgWordDataForPlatformRegionChina"},
+ {1001, nullptr, "GetNgWordDataSizeForPlatformRegionChina"},
};
+ // clang-format on
RegisterHandlers(functions);
auto& fsc = system.GetFileSystemController();
diff --git a/src/core/hle/service/nvdrv/devices/nvhost_as_gpu.cpp b/src/core/hle/service/nvdrv/devices/nvhost_as_gpu.cpp
index 07c88465e..195421cc0 100644
--- a/src/core/hle/service/nvdrv/devices/nvhost_as_gpu.cpp
+++ b/src/core/hle/service/nvdrv/devices/nvhost_as_gpu.cpp
@@ -104,10 +104,12 @@ u32 nvhost_as_gpu::Remap(const std::vector<u8>& input, std::vector<u8>& output)
ASSERT(object->status == nvmap::Object::Status::Allocated);
- u64 size = static_cast<u64>(entry.pages) << 0x10;
+ const u64 size = static_cast<u64>(entry.pages) << 0x10;
ASSERT(size <= object->size);
+ const u64 map_offset = static_cast<u64>(entry.map_offset) << 0x10;
- GPUVAddr returned = gpu.MemoryManager().MapBufferEx(object->addr, offset, size);
+ const GPUVAddr returned =
+ gpu.MemoryManager().MapBufferEx(object->addr + map_offset, offset, size);
ASSERT(returned == offset);
}
std::memcpy(output.data(), entries.data(), output.size());
diff --git a/src/core/hle/service/nvdrv/devices/nvhost_as_gpu.h b/src/core/hle/service/nvdrv/devices/nvhost_as_gpu.h
index 169fb8f0e..f79fcc065 100644
--- a/src/core/hle/service/nvdrv/devices/nvhost_as_gpu.h
+++ b/src/core/hle/service/nvdrv/devices/nvhost_as_gpu.h
@@ -62,7 +62,7 @@ private:
u16_le flags;
u16_le kind;
u32_le nvmap_handle;
- INSERT_PADDING_WORDS(1);
+ u32_le map_offset;
u32_le offset;
u32_le pages;
};
diff --git a/src/core/hle/service/nvdrv/devices/nvhost_gpu.cpp b/src/core/hle/service/nvdrv/devices/nvhost_gpu.cpp
index 9de0ace22..6d8bca8bb 100644
--- a/src/core/hle/service/nvdrv/devices/nvhost_gpu.cpp
+++ b/src/core/hle/service/nvdrv/devices/nvhost_gpu.cpp
@@ -191,8 +191,8 @@ u32 nvhost_gpu::KickoffPB(const std::vector<u8>& input, std::vector<u8>& output,
std::memcpy(entries.data(), input2.data(),
params.num_entries * sizeof(Tegra::CommandListHeader));
} else {
- Memory::ReadBlock(params.address, entries.data(),
- params.num_entries * sizeof(Tegra::CommandListHeader));
+ system.Memory().ReadBlock(params.address, entries.data(),
+ params.num_entries * sizeof(Tegra::CommandListHeader));
}
UNIMPLEMENTED_IF(params.flags.add_wait.Value() != 0);
UNIMPLEMENTED_IF(params.flags.add_increment.Value() != 0);
diff --git a/src/core/hle/service/nvdrv/interface.cpp b/src/core/hle/service/nvdrv/interface.cpp
index 68d139cfb..c8ea6c661 100644
--- a/src/core/hle/service/nvdrv/interface.cpp
+++ b/src/core/hle/service/nvdrv/interface.cpp
@@ -61,7 +61,7 @@ void NVDRV::IoctlBase(Kernel::HLERequestContext& ctx, IoctlVersion version) {
if (ctrl.must_delay) {
ctrl.fresh_call = false;
ctx.SleepClientThread("NVServices::DelayedResponse", ctrl.timeout,
- [=](Kernel::SharedPtr<Kernel::Thread> thread,
+ [=](std::shared_ptr<Kernel::Thread> thread,
Kernel::HLERequestContext& ctx,
Kernel::ThreadWakeupReason reason) {
IoctlCtrl ctrl2{ctrl};
diff --git a/src/core/hle/service/nvdrv/nvdrv.cpp b/src/core/hle/service/nvdrv/nvdrv.cpp
index cc9cd3fd1..197c77db0 100644
--- a/src/core/hle/service/nvdrv/nvdrv.cpp
+++ b/src/core/hle/service/nvdrv/nvdrv.cpp
@@ -100,11 +100,11 @@ void Module::SignalSyncpt(const u32 syncpoint_id, const u32 value) {
}
}
-Kernel::SharedPtr<Kernel::ReadableEvent> Module::GetEvent(const u32 event_id) const {
+std::shared_ptr<Kernel::ReadableEvent> Module::GetEvent(const u32 event_id) const {
return events_interface.events[event_id].readable;
}
-Kernel::SharedPtr<Kernel::WritableEvent> Module::GetEventWriteable(const u32 event_id) const {
+std::shared_ptr<Kernel::WritableEvent> Module::GetEventWriteable(const u32 event_id) const {
return events_interface.events[event_id].writable;
}
diff --git a/src/core/hle/service/nvdrv/nvdrv.h b/src/core/hle/service/nvdrv/nvdrv.h
index f8bb28969..d7a1bef91 100644
--- a/src/core/hle/service/nvdrv/nvdrv.h
+++ b/src/core/hle/service/nvdrv/nvdrv.h
@@ -114,9 +114,9 @@ public:
void SignalSyncpt(const u32 syncpoint_id, const u32 value);
- Kernel::SharedPtr<Kernel::ReadableEvent> GetEvent(u32 event_id) const;
+ std::shared_ptr<Kernel::ReadableEvent> GetEvent(u32 event_id) const;
- Kernel::SharedPtr<Kernel::WritableEvent> GetEventWriteable(u32 event_id) const;
+ std::shared_ptr<Kernel::WritableEvent> GetEventWriteable(u32 event_id) const;
private:
/// Id to use for the next open file descriptor.
diff --git a/src/core/hle/service/nvflinger/buffer_queue.cpp b/src/core/hle/service/nvflinger/buffer_queue.cpp
index 1af11e80c..32b6f4b27 100644
--- a/src/core/hle/service/nvflinger/buffer_queue.cpp
+++ b/src/core/hle/service/nvflinger/buffer_queue.cpp
@@ -117,11 +117,11 @@ u32 BufferQueue::Query(QueryType type) {
return 0;
}
-Kernel::SharedPtr<Kernel::WritableEvent> BufferQueue::GetWritableBufferWaitEvent() const {
+std::shared_ptr<Kernel::WritableEvent> BufferQueue::GetWritableBufferWaitEvent() const {
return buffer_wait_event.writable;
}
-Kernel::SharedPtr<Kernel::ReadableEvent> BufferQueue::GetBufferWaitEvent() const {
+std::shared_ptr<Kernel::ReadableEvent> BufferQueue::GetBufferWaitEvent() const {
return buffer_wait_event.readable;
}
diff --git a/src/core/hle/service/nvflinger/buffer_queue.h b/src/core/hle/service/nvflinger/buffer_queue.h
index 8f9b18547..f4bbfd945 100644
--- a/src/core/hle/service/nvflinger/buffer_queue.h
+++ b/src/core/hle/service/nvflinger/buffer_queue.h
@@ -93,9 +93,9 @@ public:
return id;
}
- Kernel::SharedPtr<Kernel::WritableEvent> GetWritableBufferWaitEvent() const;
+ std::shared_ptr<Kernel::WritableEvent> GetWritableBufferWaitEvent() const;
- Kernel::SharedPtr<Kernel::ReadableEvent> GetBufferWaitEvent() const;
+ std::shared_ptr<Kernel::ReadableEvent> GetBufferWaitEvent() const;
private:
u32 id;
diff --git a/src/core/hle/service/nvflinger/nvflinger.cpp b/src/core/hle/service/nvflinger/nvflinger.cpp
index cc9522aad..52623cf89 100644
--- a/src/core/hle/service/nvflinger/nvflinger.cpp
+++ b/src/core/hle/service/nvflinger/nvflinger.cpp
@@ -37,8 +37,8 @@ NVFlinger::NVFlinger(Core::System& system) : system(system) {
displays.emplace_back(4, "Null", system);
// Schedule the screen composition events
- composition_event = system.CoreTiming().RegisterEvent(
- "ScreenComposition", [this](u64 userdata, s64 cycles_late) {
+ composition_event =
+ Core::Timing::CreateEvent("ScreenComposition", [this](u64 userdata, s64 cycles_late) {
Compose();
const auto ticks =
Settings::values.force_30fps_mode ? frame_ticks_30fps : GetNextTicks();
@@ -98,7 +98,7 @@ std::optional<u32> NVFlinger::FindBufferQueueId(u64 display_id, u64 layer_id) co
return layer->GetBufferQueue().GetId();
}
-Kernel::SharedPtr<Kernel::ReadableEvent> NVFlinger::FindVsyncEvent(u64 display_id) const {
+std::shared_ptr<Kernel::ReadableEvent> NVFlinger::FindVsyncEvent(u64 display_id) const {
auto* const display = FindDisplay(display_id);
if (display == nullptr) {
diff --git a/src/core/hle/service/nvflinger/nvflinger.h b/src/core/hle/service/nvflinger/nvflinger.h
index 5d7e3bfb8..e3cc14bdc 100644
--- a/src/core/hle/service/nvflinger/nvflinger.h
+++ b/src/core/hle/service/nvflinger/nvflinger.h
@@ -62,7 +62,7 @@ public:
/// Gets the vsync event for the specified display.
///
/// If an invalid display ID is provided, then nullptr is returned.
- Kernel::SharedPtr<Kernel::ReadableEvent> FindVsyncEvent(u64 display_id) const;
+ std::shared_ptr<Kernel::ReadableEvent> FindVsyncEvent(u64 display_id) const;
/// Obtains a buffer queue identified by the ID.
BufferQueue& FindBufferQueue(u32 id);
@@ -103,7 +103,7 @@ private:
u32 swap_interval = 1;
/// Event that handles screen composition.
- Core::Timing::EventType* composition_event;
+ std::shared_ptr<Core::Timing::EventType> composition_event;
Core::System& system;
};
diff --git a/src/core/hle/service/pm/pm.cpp b/src/core/hle/service/pm/pm.cpp
index fe6b5f798..809eca0ab 100644
--- a/src/core/hle/service/pm/pm.cpp
+++ b/src/core/hle/service/pm/pm.cpp
@@ -16,9 +16,9 @@ constexpr ResultCode ERROR_PROCESS_NOT_FOUND{ErrorModule::PM, 1};
constexpr u64 NO_PROCESS_FOUND_PID{0};
-std::optional<Kernel::SharedPtr<Kernel::Process>> SearchProcessList(
- const std::vector<Kernel::SharedPtr<Kernel::Process>>& process_list,
- std::function<bool(const Kernel::SharedPtr<Kernel::Process>&)> predicate) {
+std::optional<std::shared_ptr<Kernel::Process>> SearchProcessList(
+ const std::vector<std::shared_ptr<Kernel::Process>>& process_list,
+ std::function<bool(const std::shared_ptr<Kernel::Process>&)> predicate) {
const auto iter = std::find_if(process_list.begin(), process_list.end(), predicate);
if (iter == process_list.end()) {
@@ -29,7 +29,7 @@ std::optional<Kernel::SharedPtr<Kernel::Process>> SearchProcessList(
}
void GetApplicationPidGeneric(Kernel::HLERequestContext& ctx,
- const std::vector<Kernel::SharedPtr<Kernel::Process>>& process_list) {
+ const std::vector<std::shared_ptr<Kernel::Process>>& process_list) {
const auto process = SearchProcessList(process_list, [](const auto& process) {
return process->GetProcessID() == Kernel::Process::ProcessIDMin;
});
@@ -124,7 +124,7 @@ private:
class Info final : public ServiceFramework<Info> {
public:
- explicit Info(const std::vector<Kernel::SharedPtr<Kernel::Process>>& process_list)
+ explicit Info(const std::vector<std::shared_ptr<Kernel::Process>>& process_list)
: ServiceFramework{"pm:info"}, process_list(process_list) {
static const FunctionInfo functions[] = {
{0, &Info::GetTitleId, "GetTitleId"},
@@ -154,7 +154,7 @@ private:
rb.Push((*process)->GetTitleID());
}
- const std::vector<Kernel::SharedPtr<Kernel::Process>>& process_list;
+ const std::vector<std::shared_ptr<Kernel::Process>>& process_list;
};
class Shell final : public ServiceFramework<Shell> {
@@ -172,7 +172,7 @@ public:
{6, &Shell::GetApplicationPid, "GetApplicationPid"},
{7, nullptr, "BoostSystemMemoryResourceLimit"},
{8, nullptr, "EnableAdditionalSystemThreads"},
- {9, nullptr, "GetUnimplementedEventHandle"},
+ {9, nullptr, "GetBootFinishedEventHandle"},
};
// clang-format on
diff --git a/src/core/hle/service/prepo/prepo.cpp b/src/core/hle/service/prepo/prepo.cpp
index 18d895263..5eb26caf8 100644
--- a/src/core/hle/service/prepo/prepo.cpp
+++ b/src/core/hle/service/prepo/prepo.cpp
@@ -25,6 +25,7 @@ public:
{10103, &PlayReport::SaveReportWithUser<Core::Reporter::PlayReportType::New>, "SaveReportWithUser"},
{10200, nullptr, "RequestImmediateTransmission"},
{10300, nullptr, "GetTransmissionStatus"},
+ {10400, nullptr, "GetSystemSessionId"},
{20100, &PlayReport::SaveSystemReport, "SaveSystemReport"},
{20101, &PlayReport::SaveSystemReportWithUser, "SaveSystemReportWithUser"},
{20200, nullptr, "SetOperationMode"},
diff --git a/src/core/hle/service/service.cpp b/src/core/hle/service/service.cpp
index 7c5302017..fa5347af9 100644
--- a/src/core/hle/service/service.cpp
+++ b/src/core/hle/service/service.cpp
@@ -116,7 +116,7 @@ void ServiceFrameworkBase::InstallAsNamedPort() {
port_installed = true;
}
-Kernel::SharedPtr<Kernel::ClientPort> ServiceFrameworkBase::CreatePort() {
+std::shared_ptr<Kernel::ClientPort> ServiceFrameworkBase::CreatePort() {
ASSERT(!port_installed);
auto& kernel = Core::System::GetInstance().Kernel();
@@ -186,7 +186,7 @@ ResultCode ServiceFrameworkBase::HandleSyncRequest(Kernel::HLERequestContext& co
UNIMPLEMENTED_MSG("command_type={}", static_cast<int>(context.GetCommandType()));
}
- context.WriteToOutgoingCommandBuffer(*Kernel::GetCurrentThread());
+ context.WriteToOutgoingCommandBuffer(context.GetThread());
return RESULT_SUCCESS;
}
@@ -201,7 +201,7 @@ void Init(std::shared_ptr<SM::ServiceManager>& sm, Core::System& system) {
auto nv_flinger = std::make_shared<NVFlinger::NVFlinger>(system);
system.GetFileSystemController().CreateFactories(*system.GetFilesystem(), false);
- SM::ServiceManager::InstallInterfaces(sm);
+ SM::ServiceManager::InstallInterfaces(sm, system.Kernel());
Account::InstallInterfaces(system);
AM::InstallInterfaces(*sm, nv_flinger, system);
diff --git a/src/core/hle/service/service.h b/src/core/hle/service/service.h
index aef964861..022d885b6 100644
--- a/src/core/hle/service/service.h
+++ b/src/core/hle/service/service.h
@@ -65,7 +65,7 @@ public:
/// Creates a port pair and registers it on the kernel's global port registry.
void InstallAsNamedPort();
/// Creates and returns an unregistered port for the service.
- Kernel::SharedPtr<Kernel::ClientPort> CreatePort();
+ std::shared_ptr<Kernel::ClientPort> CreatePort();
void InvokeRequest(Kernel::HLERequestContext& ctx);
diff --git a/src/core/hle/service/set/set.cpp b/src/core/hle/service/set/set.cpp
index b54214421..5bcc0b588 100644
--- a/src/core/hle/service/set/set.cpp
+++ b/src/core/hle/service/set/set.cpp
@@ -124,6 +124,7 @@ SET::SET() : ServiceFramework("set") {
{7, nullptr, "GetKeyCodeMap"},
{8, &SET::GetQuestFlag, "GetQuestFlag"},
{9, nullptr, "GetKeyCodeMap2"},
+ {10, nullptr, "GetFirmwareVersionForDebug"},
};
// clang-format on
diff --git a/src/core/hle/service/set/set_cal.cpp b/src/core/hle/service/set/set_cal.cpp
index 5981c575c..1398a4a48 100644
--- a/src/core/hle/service/set/set_cal.cpp
+++ b/src/core/hle/service/set/set_cal.cpp
@@ -7,6 +7,7 @@
namespace Service::Set {
SET_CAL::SET_CAL() : ServiceFramework("set:cal") {
+ // clang-format off
static const FunctionInfo functions[] = {
{0, nullptr, "GetBluetoothBdAddress"},
{1, nullptr, "GetConfigurationId1"},
@@ -40,8 +41,18 @@ SET_CAL::SET_CAL() : ServiceFramework("set:cal") {
{30, nullptr, "GetAmiiboEcqvBlsCertificate"},
{31, nullptr, "GetAmiiboEcqvBlsRootCertificate"},
{32, nullptr, "GetUsbTypeCPowerSourceCircuitVersion"},
+ {33, nullptr, "GetAnalogStickModuleTypeL"},
+ {34, nullptr, "GetAnalogStickModelParameterL"},
+ {35, nullptr, "GetAnalogStickFactoryCalibrationL"},
+ {36, nullptr, "GetAnalogStickModuleTypeR"},
+ {37, nullptr, "GetAnalogStickModelParameterR"},
+ {38, nullptr, "GetAnalogStickFactoryCalibrationR"},
+ {39, nullptr, "GetConsoleSixAxisSensorModuleType"},
+ {40, nullptr, "GetConsoleSixAxisSensorHorizontalOffset"},
{41, nullptr, "GetBatteryVersion"},
};
+ // clang-format on
+
RegisterHandlers(functions);
}
diff --git a/src/core/hle/service/set/set_fd.cpp b/src/core/hle/service/set/set_fd.cpp
index cac6af86d..565882a31 100644
--- a/src/core/hle/service/set/set_fd.cpp
+++ b/src/core/hle/service/set/set_fd.cpp
@@ -7,6 +7,7 @@
namespace Service::Set {
SET_FD::SET_FD() : ServiceFramework("set:fd") {
+ // clang-format off
static const FunctionInfo functions[] = {
{2, nullptr, "SetSettingsItemValue"},
{3, nullptr, "ResetSettingsItemValue"},
@@ -16,7 +17,10 @@ SET_FD::SET_FD() : ServiceFramework("set:fd") {
{20, nullptr, "SetWebInspectorFlag"},
{21, nullptr, "SetAllowedSslHosts"},
{22, nullptr, "SetHostFsMountPoint"},
+ {23, nullptr, "SetMemoryUsageRateFlag"},
};
+ // clang-format on
+
RegisterHandlers(functions);
}
diff --git a/src/core/hle/service/set/set_sys.cpp b/src/core/hle/service/set/set_sys.cpp
index 98d0cfdfd..b7c9ea74b 100644
--- a/src/core/hle/service/set/set_sys.cpp
+++ b/src/core/hle/service/set/set_sys.cpp
@@ -273,10 +273,21 @@ SET_SYS::SET_SYS() : ServiceFramework("set:sys") {
{171, nullptr, "SetChineseTraditionalInputMethod"},
{172, nullptr, "GetPtmCycleCountReliability"},
{173, nullptr, "SetPtmCycleCountReliability"},
+ {174, nullptr, "GetHomeMenuScheme"},
{175, nullptr, "GetThemeSettings"},
{176, nullptr, "SetThemeSettings"},
{177, nullptr, "GetThemeKey"},
{178, nullptr, "SetThemeKey"},
+ {179, nullptr, "GetZoomFlag"},
+ {180, nullptr, "SetZoomFlag"},
+ {181, nullptr, "GetT"},
+ {182, nullptr, "SetT"},
+ {183, nullptr, "GetPlatformRegion"},
+ {184, nullptr, "SetPlatformRegion"},
+ {185, nullptr, "GetHomeMenuSchemeModel"},
+ {186, nullptr, "GetMemoryUsageRateFlag"},
+ {187, nullptr, "GetTouchScreenMode"},
+ {188, nullptr, "SetTouchScreenMode"},
};
// clang-format on
diff --git a/src/core/hle/service/sm/controller.cpp b/src/core/hle/service/sm/controller.cpp
index e9ee73710..c45b285f8 100644
--- a/src/core/hle/service/sm/controller.cpp
+++ b/src/core/hle/service/sm/controller.cpp
@@ -30,10 +30,7 @@ void Controller::DuplicateSession(Kernel::HLERequestContext& ctx) {
IPC::ResponseBuilder rb{ctx, 2, 0, 1, IPC::ResponseBuilder::Flags::AlwaysMoveHandles};
rb.Push(RESULT_SUCCESS);
- Kernel::SharedPtr<Kernel::ClientSession> session{ctx.Session()->GetParent()->client};
- rb.PushMoveObjects(session);
-
- LOG_DEBUG(Service, "session={}", session->GetObjectId());
+ rb.PushMoveObjects(ctx.Session()->GetParent()->Client());
}
void Controller::DuplicateSessionEx(Kernel::HLERequestContext& ctx) {
diff --git a/src/core/hle/service/sm/sm.cpp b/src/core/hle/service/sm/sm.cpp
index 142929124..88909504d 100644
--- a/src/core/hle/service/sm/sm.cpp
+++ b/src/core/hle/service/sm/sm.cpp
@@ -36,16 +36,17 @@ static ResultCode ValidateServiceName(const std::string& name) {
return RESULT_SUCCESS;
}
-void ServiceManager::InstallInterfaces(std::shared_ptr<ServiceManager> self) {
+void ServiceManager::InstallInterfaces(std::shared_ptr<ServiceManager> self,
+ Kernel::KernelCore& kernel) {
ASSERT(self->sm_interface.expired());
- auto sm = std::make_shared<SM>(self);
+ auto sm = std::make_shared<SM>(self, kernel);
sm->InstallAsNamedPort();
self->sm_interface = sm;
self->controller_interface = std::make_unique<Controller>();
}
-ResultVal<Kernel::SharedPtr<Kernel::ServerPort>> ServiceManager::RegisterService(
+ResultVal<std::shared_ptr<Kernel::ServerPort>> ServiceManager::RegisterService(
std::string name, unsigned int max_sessions) {
CASCADE_CODE(ValidateServiceName(name));
@@ -72,7 +73,7 @@ ResultCode ServiceManager::UnregisterService(const std::string& name) {
return RESULT_SUCCESS;
}
-ResultVal<Kernel::SharedPtr<Kernel::ClientPort>> ServiceManager::GetServicePort(
+ResultVal<std::shared_ptr<Kernel::ClientPort>> ServiceManager::GetServicePort(
const std::string& name) {
CASCADE_CODE(ValidateServiceName(name));
@@ -84,7 +85,7 @@ ResultVal<Kernel::SharedPtr<Kernel::ClientPort>> ServiceManager::GetServicePort(
return MakeResult(it->second);
}
-ResultVal<Kernel::SharedPtr<Kernel::ClientSession>> ServiceManager::ConnectToService(
+ResultVal<std::shared_ptr<Kernel::ClientSession>> ServiceManager::ConnectToService(
const std::string& name) {
CASCADE_RESULT(auto client_port, GetServicePort(name));
@@ -114,8 +115,6 @@ void SM::GetService(Kernel::HLERequestContext& ctx) {
std::string name(name_buf.begin(), end);
- // TODO(yuriks): Permission checks go here
-
auto client_port = service_manager->GetServicePort(name);
if (client_port.Failed()) {
IPC::ResponseBuilder rb{ctx, 2};
@@ -127,14 +126,22 @@ void SM::GetService(Kernel::HLERequestContext& ctx) {
return;
}
- auto session = client_port.Unwrap()->Connect();
- ASSERT(session.Succeeded());
- if (session.Succeeded()) {
- LOG_DEBUG(Service_SM, "called service={} -> session={}", name, (*session)->GetObjectId());
- IPC::ResponseBuilder rb{ctx, 2, 0, 1, IPC::ResponseBuilder::Flags::AlwaysMoveHandles};
- rb.Push(session.Code());
- rb.PushMoveObjects(std::move(session).Unwrap());
+ auto [client, server] = Kernel::Session::Create(kernel, name);
+
+ const auto& server_port = client_port.Unwrap()->GetServerPort();
+ if (server_port->GetHLEHandler()) {
+ server_port->GetHLEHandler()->ClientConnected(server);
+ } else {
+ server_port->AppendPendingSession(server);
}
+
+ // Wake the threads waiting on the ServerPort
+ server_port->WakeupAllWaitingThreads();
+
+ LOG_DEBUG(Service_SM, "called service={} -> session={}", name, client->GetObjectId());
+ IPC::ResponseBuilder rb{ctx, 2, 0, 1, IPC::ResponseBuilder::Flags::AlwaysMoveHandles};
+ rb.Push(RESULT_SUCCESS);
+ rb.PushMoveObjects(std::move(client));
}
void SM::RegisterService(Kernel::HLERequestContext& ctx) {
@@ -178,8 +185,8 @@ void SM::UnregisterService(Kernel::HLERequestContext& ctx) {
rb.Push(service_manager->UnregisterService(name));
}
-SM::SM(std::shared_ptr<ServiceManager> service_manager)
- : ServiceFramework("sm:", 4), service_manager(std::move(service_manager)) {
+SM::SM(std::shared_ptr<ServiceManager> service_manager, Kernel::KernelCore& kernel)
+ : ServiceFramework{"sm:", 4}, service_manager{std::move(service_manager)}, kernel{kernel} {
static const FunctionInfo functions[] = {
{0x00000000, &SM::Initialize, "Initialize"},
{0x00000001, &SM::GetService, "GetService"},
diff --git a/src/core/hle/service/sm/sm.h b/src/core/hle/service/sm/sm.h
index b9d6381b4..b06d2f103 100644
--- a/src/core/hle/service/sm/sm.h
+++ b/src/core/hle/service/sm/sm.h
@@ -18,6 +18,7 @@
namespace Kernel {
class ClientPort;
class ClientSession;
+class KernelCore;
class ServerPort;
class SessionRequestHandler;
} // namespace Kernel
@@ -29,7 +30,7 @@ class Controller;
/// Interface to "sm:" service
class SM final : public ServiceFramework<SM> {
public:
- explicit SM(std::shared_ptr<ServiceManager> service_manager);
+ explicit SM(std::shared_ptr<ServiceManager> service_manager, Kernel::KernelCore& kernel);
~SM() override;
private:
@@ -39,20 +40,21 @@ private:
void UnregisterService(Kernel::HLERequestContext& ctx);
std::shared_ptr<ServiceManager> service_manager;
+ Kernel::KernelCore& kernel;
};
class ServiceManager {
public:
- static void InstallInterfaces(std::shared_ptr<ServiceManager> self);
+ static void InstallInterfaces(std::shared_ptr<ServiceManager> self, Kernel::KernelCore& kernel);
ServiceManager();
~ServiceManager();
- ResultVal<Kernel::SharedPtr<Kernel::ServerPort>> RegisterService(std::string name,
- unsigned int max_sessions);
+ ResultVal<std::shared_ptr<Kernel::ServerPort>> RegisterService(std::string name,
+ unsigned int max_sessions);
ResultCode UnregisterService(const std::string& name);
- ResultVal<Kernel::SharedPtr<Kernel::ClientPort>> GetServicePort(const std::string& name);
- ResultVal<Kernel::SharedPtr<Kernel::ClientSession>> ConnectToService(const std::string& name);
+ ResultVal<std::shared_ptr<Kernel::ClientPort>> GetServicePort(const std::string& name);
+ ResultVal<std::shared_ptr<Kernel::ClientSession>> ConnectToService(const std::string& name);
template <typename T>
std::shared_ptr<T> GetService(const std::string& service_name) const {
@@ -77,7 +79,7 @@ private:
std::unique_ptr<Controller> controller_interface;
/// Map of registered services, retrieved using GetServicePort or ConnectToService.
- std::unordered_map<std::string, Kernel::SharedPtr<Kernel::ClientPort>> registered_services;
+ std::unordered_map<std::string, std::shared_ptr<Kernel::ClientPort>> registered_services;
};
} // namespace Service::SM
diff --git a/src/core/hle/service/sockets/nsd.cpp b/src/core/hle/service/sockets/nsd.cpp
index e6d73065e..dc70fd6fe 100644
--- a/src/core/hle/service/sockets/nsd.cpp
+++ b/src/core/hle/service/sockets/nsd.cpp
@@ -7,6 +7,7 @@
namespace Service::Sockets {
NSD::NSD(const char* name) : ServiceFramework(name) {
+ // clang-format off
static const FunctionInfo functions[] = {
{10, nullptr, "GetSettingName"},
{11, nullptr, "GetEnvironmentIdentifier"},
@@ -22,10 +23,14 @@ NSD::NSD(const char* name) : ServiceFramework(name) {
{42, nullptr, "GetNasApiFqdn"},
{43, nullptr, "GetNasApiFqdnEx"},
{50, nullptr, "GetCurrentSetting"},
+ {51, nullptr, "WriteTestParameter"},
+ {52, nullptr, "ReadTestParameter"},
{60, nullptr, "ReadSaveDataFromFsForTest"},
{61, nullptr, "WriteSaveDataToFsForTest"},
{62, nullptr, "DeleteSaveDataOfFsForTest"},
};
+ // clang-format on
+
RegisterHandlers(functions);
}
diff --git a/src/core/hle/service/ssl/ssl.cpp b/src/core/hle/service/ssl/ssl.cpp
index 65040c077..1ba8c19a0 100644
--- a/src/core/hle/service/ssl/ssl.cpp
+++ b/src/core/hle/service/ssl/ssl.cpp
@@ -13,6 +13,7 @@ namespace Service::SSL {
class ISslConnection final : public ServiceFramework<ISslConnection> {
public:
ISslConnection() : ServiceFramework("ISslConnection") {
+ // clang-format off
static const FunctionInfo functions[] = {
{0, nullptr, "SetSocketDescriptor"},
{1, nullptr, "SetHostName"},
@@ -40,7 +41,11 @@ public:
{23, nullptr, "GetOption"},
{24, nullptr, "GetVerifyCertErrors"},
{25, nullptr, "GetCipherInfo"},
+ {26, nullptr, "SetNextAlpnProto"},
+ {27, nullptr, "GetNextAlpnProto"},
};
+ // clang-format on
+
RegisterHandlers(functions);
}
};
diff --git a/src/core/hle/service/time/interface.cpp b/src/core/hle/service/time/interface.cpp
index 9565e7de5..bc74f1e1d 100644
--- a/src/core/hle/service/time/interface.cpp
+++ b/src/core/hle/service/time/interface.cpp
@@ -21,6 +21,7 @@ Time::Time(std::shared_ptr<Module> time, std::shared_ptr<SharedMemory> shared_me
{30, nullptr, "GetStandardNetworkClockOperationEventReadableHandle"},
{31, nullptr, "GetEphemeralNetworkClockOperationEventReadableHandle"},
{50, nullptr, "SetStandardSteadyClockInternalOffset"},
+ {51, nullptr, "GetStandardSteadyClockRtcValue"},
{100, &Time::IsStandardUserSystemClockAutomaticCorrectionEnabled, "IsStandardUserSystemClockAutomaticCorrectionEnabled"},
{101, &Time::SetStandardUserSystemClockAutomaticCorrectionEnabled, "SetStandardUserSystemClockAutomaticCorrectionEnabled"},
{102, nullptr, "GetStandardUserSystemClockInitialYear"},
diff --git a/src/core/hle/service/time/time.cpp b/src/core/hle/service/time/time.cpp
index 1b9ab8401..6ee77c5f9 100644
--- a/src/core/hle/service/time/time.cpp
+++ b/src/core/hle/service/time/time.cpp
@@ -34,12 +34,12 @@ static void PosixToCalendar(u64 posix_time, CalendarTime& calendar_time,
additional_info = {};
return;
}
- calendar_time.year = tm->tm_year + 1900;
- calendar_time.month = tm->tm_mon + 1;
- calendar_time.day = tm->tm_mday;
- calendar_time.hour = tm->tm_hour;
- calendar_time.minute = tm->tm_min;
- calendar_time.second = tm->tm_sec;
+ calendar_time.year = static_cast<u16_le>(tm->tm_year + 1900);
+ calendar_time.month = static_cast<u8>(tm->tm_mon + 1);
+ calendar_time.day = static_cast<u8>(tm->tm_mday);
+ calendar_time.hour = static_cast<u8>(tm->tm_hour);
+ calendar_time.minute = static_cast<u8>(tm->tm_min);
+ calendar_time.second = static_cast<u8>(tm->tm_sec);
additional_info.day_of_week = tm->tm_wday;
additional_info.day_of_year = tm->tm_yday;
@@ -74,15 +74,17 @@ public:
ISystemClock(std::shared_ptr<Service::Time::SharedMemory> shared_memory,
ClockContextType clock_type)
: ServiceFramework("ISystemClock"), shared_memory(shared_memory), clock_type(clock_type) {
+ // clang-format off
static const FunctionInfo functions[] = {
{0, &ISystemClock::GetCurrentTime, "GetCurrentTime"},
{1, nullptr, "SetCurrentTime"},
{2, &ISystemClock::GetSystemClockContext, "GetSystemClockContext"},
{3, nullptr, "SetSystemClockContext"},
-
+ {4, nullptr, "GetOperationEventReadableHandle"},
};
- RegisterHandlers(functions);
+ // clang-format on
+ RegisterHandlers(functions);
UpdateSharedMemoryContext(system_clock_context);
}
@@ -162,6 +164,7 @@ private:
class ITimeZoneService final : public ServiceFramework<ITimeZoneService> {
public:
ITimeZoneService() : ServiceFramework("ITimeZoneService") {
+ // clang-format off
static const FunctionInfo functions[] = {
{0, &ITimeZoneService::GetDeviceLocationName, "GetDeviceLocationName"},
{1, nullptr, "SetDeviceLocationName"},
@@ -169,11 +172,17 @@ public:
{3, nullptr, "LoadLocationNameList"},
{4, &ITimeZoneService::LoadTimeZoneRule, "LoadTimeZoneRule"},
{5, nullptr, "GetTimeZoneRuleVersion"},
+ {6, nullptr, "GetDeviceLocationNameAndUpdatedTime"},
+ {7, nullptr, "SetDeviceLocationNameWithTimeZoneRule"},
+ {8, nullptr, "ParseTimeZoneBinary"},
+ {20, nullptr, "GetDeviceLocationNameOperationEventReadableHandle"},
{100, &ITimeZoneService::ToCalendarTime, "ToCalendarTime"},
{101, &ITimeZoneService::ToCalendarTimeWithMyRule, "ToCalendarTimeWithMyRule"},
{201, &ITimeZoneService::ToPosixTime, "ToPosixTime"},
{202, &ITimeZoneService::ToPosixTimeWithMyRule, "ToPosixTimeWithMyRule"},
};
+ // clang-format on
+
RegisterHandlers(functions);
}
@@ -322,7 +331,7 @@ void Module::Interface::GetClockSnapshot(Kernel::HLERequestContext& ctx) {
if (tm == nullptr) {
LOG_ERROR(Service_Time, "tm is a nullptr");
IPC::ResponseBuilder rb{ctx, 2};
- rb.Push(ResultCode(-1)); // TODO(ogniK): Find appropriate error code
+ rb.Push(RESULT_UNKNOWN); // TODO(ogniK): Find appropriate error code
return;
}
@@ -331,12 +340,12 @@ void Module::Interface::GetClockSnapshot(Kernel::HLERequestContext& ctx) {
const SteadyClockTimePoint steady_clock_time_point{static_cast<u64_le>(ms.count() / 1000), {}};
CalendarTime calendar_time{};
- calendar_time.year = tm->tm_year + 1900;
- calendar_time.month = tm->tm_mon + 1;
- calendar_time.day = tm->tm_mday;
- calendar_time.hour = tm->tm_hour;
- calendar_time.minute = tm->tm_min;
- calendar_time.second = tm->tm_sec;
+ calendar_time.year = static_cast<u16_le>(tm->tm_year + 1900);
+ calendar_time.month = static_cast<u8>(tm->tm_mon + 1);
+ calendar_time.day = static_cast<u8>(tm->tm_mday);
+ calendar_time.hour = static_cast<u8>(tm->tm_hour);
+ calendar_time.minute = static_cast<u8>(tm->tm_min);
+ calendar_time.second = static_cast<u8>(tm->tm_sec);
ClockSnapshot clock_snapshot{};
clock_snapshot.system_posix_time = time_since_epoch;
diff --git a/src/core/hle/service/time/time_sharedmemory.cpp b/src/core/hle/service/time/time_sharedmemory.cpp
index bfc81b83c..4035f5072 100644
--- a/src/core/hle/service/time/time_sharedmemory.cpp
+++ b/src/core/hle/service/time/time_sharedmemory.cpp
@@ -21,7 +21,7 @@ SharedMemory::SharedMemory(Core::System& system) : system(system) {
SharedMemory::~SharedMemory() = default;
-Kernel::SharedPtr<Kernel::SharedMemory> SharedMemory::GetSharedMemoryHolder() const {
+std::shared_ptr<Kernel::SharedMemory> SharedMemory::GetSharedMemoryHolder() const {
return shared_memory_holder;
}
diff --git a/src/core/hle/service/time/time_sharedmemory.h b/src/core/hle/service/time/time_sharedmemory.h
index cb8253541..904a96430 100644
--- a/src/core/hle/service/time/time_sharedmemory.h
+++ b/src/core/hle/service/time/time_sharedmemory.h
@@ -15,7 +15,7 @@ public:
~SharedMemory();
// Return the shared memory handle
- Kernel::SharedPtr<Kernel::SharedMemory> GetSharedMemoryHolder() const;
+ std::shared_ptr<Kernel::SharedMemory> GetSharedMemoryHolder() const;
// Set memory barriers in shared memory and update them
void SetStandardSteadyClockTimepoint(const SteadyClockTimePoint& timepoint);
@@ -66,7 +66,7 @@ public:
static_assert(sizeof(Format) == 0xd8, "Format is an invalid size");
private:
- Kernel::SharedPtr<Kernel::SharedMemory> shared_memory_holder{};
+ std::shared_ptr<Kernel::SharedMemory> shared_memory_holder{};
Core::System& system;
Format shared_memory_format{};
};
diff --git a/src/core/hle/service/vi/display/vi_display.cpp b/src/core/hle/service/vi/display/vi_display.cpp
index 07033fb98..cd18c1610 100644
--- a/src/core/hle/service/vi/display/vi_display.cpp
+++ b/src/core/hle/service/vi/display/vi_display.cpp
@@ -31,7 +31,7 @@ const Layer& Display::GetLayer(std::size_t index) const {
return layers.at(index);
}
-Kernel::SharedPtr<Kernel::ReadableEvent> Display::GetVSyncEvent() const {
+std::shared_ptr<Kernel::ReadableEvent> Display::GetVSyncEvent() const {
return vsync_event.readable;
}
diff --git a/src/core/hle/service/vi/display/vi_display.h b/src/core/hle/service/vi/display/vi_display.h
index f56b5badc..8bb966a85 100644
--- a/src/core/hle/service/vi/display/vi_display.h
+++ b/src/core/hle/service/vi/display/vi_display.h
@@ -57,7 +57,7 @@ public:
const Layer& GetLayer(std::size_t index) const;
/// Gets the readable vsync event.
- Kernel::SharedPtr<Kernel::ReadableEvent> GetVSyncEvent() const;
+ std::shared_ptr<Kernel::ReadableEvent> GetVSyncEvent() const;
/// Signals the internal vsync event.
void SignalVSyncEvent();
diff --git a/src/core/hle/service/vi/vi.cpp b/src/core/hle/service/vi/vi.cpp
index 611cecc20..651c89dc0 100644
--- a/src/core/hle/service/vi/vi.cpp
+++ b/src/core/hle/service/vi/vi.cpp
@@ -541,8 +541,8 @@ private:
} else {
// Wait the current thread until a buffer becomes available
ctx.SleepClientThread(
- "IHOSBinderDriver::DequeueBuffer", -1,
- [=](Kernel::SharedPtr<Kernel::Thread> thread, Kernel::HLERequestContext& ctx,
+ "IHOSBinderDriver::DequeueBuffer", UINT64_MAX,
+ [=](std::shared_ptr<Kernel::Thread> thread, Kernel::HLERequestContext& ctx,
Kernel::ThreadWakeupReason reason) {
// Repeat TransactParcel DequeueBuffer when a buffer is available
auto& buffer_queue = nv_flinger->FindBufferQueue(id);
@@ -731,6 +731,7 @@ class IManagerDisplayService final : public ServiceFramework<IManagerDisplayServ
public:
explicit IManagerDisplayService(std::shared_ptr<NVFlinger::NVFlinger> nv_flinger)
: ServiceFramework("IManagerDisplayService"), nv_flinger(std::move(nv_flinger)) {
+ // clang-format off
static const FunctionInfo functions[] = {
{200, nullptr, "AllocateProcessHeapBlock"},
{201, nullptr, "FreeProcessHeapBlock"},
@@ -766,8 +767,11 @@ public:
{6008, nullptr, "StartLayerPresentationFenceWait"},
{6009, nullptr, "StopLayerPresentationFenceWait"},
{6010, nullptr, "GetLayerPresentationAllFencesExpiredEvent"},
+ {6011, nullptr, "EnableLayerAutoClearTransitionBuffer"},
+ {6012, nullptr, "DisableLayerAutoClearTransitionBuffer"},
{7000, nullptr, "SetContentVisibility"},
{8000, nullptr, "SetConductorLayer"},
+ {8001, nullptr, "SetTimestampTracking"},
{8100, nullptr, "SetIndirectProducerFlipOffset"},
{8200, nullptr, "CreateSharedBufferStaticStorage"},
{8201, nullptr, "CreateSharedBufferTransferMemory"},
@@ -800,6 +804,8 @@ public:
{8297, nullptr, "GetSharedFrameBufferContentParameter"},
{8298, nullptr, "ExpandStartupLogoOnSharedFrameBuffer"},
};
+ // clang-format on
+
RegisterHandlers(functions);
}
diff --git a/src/core/memory.cpp b/src/core/memory.cpp
index fa49f3dd0..91bf07a92 100644
--- a/src/core/memory.cpp
+++ b/src/core/memory.cpp
@@ -17,529 +17,699 @@
#include "core/hle/kernel/process.h"
#include "core/hle/kernel/vm_manager.h"
#include "core/memory.h"
-#include "core/memory_setup.h"
#include "video_core/gpu.h"
namespace Memory {
-static Common::PageTable* current_page_table = nullptr;
+// Implementation class used to keep the specifics of the memory subsystem hidden
+// from outside classes. This also allows modification to the internals of the memory
+// subsystem without needing to rebuild all files that make use of the memory interface.
+struct Memory::Impl {
+ explicit Impl(Core::System& system_) : system{system_} {}
-void SetCurrentPageTable(Kernel::Process& process) {
- current_page_table = &process.VMManager().page_table;
+ void SetCurrentPageTable(Kernel::Process& process) {
+ current_page_table = &process.VMManager().page_table;
- const std::size_t address_space_width = process.VMManager().GetAddressSpaceWidth();
+ const std::size_t address_space_width = process.VMManager().GetAddressSpaceWidth();
- auto& system = Core::System::GetInstance();
- system.ArmInterface(0).PageTableChanged(*current_page_table, address_space_width);
- system.ArmInterface(1).PageTableChanged(*current_page_table, address_space_width);
- system.ArmInterface(2).PageTableChanged(*current_page_table, address_space_width);
- system.ArmInterface(3).PageTableChanged(*current_page_table, address_space_width);
-}
+ system.ArmInterface(0).PageTableChanged(*current_page_table, address_space_width);
+ system.ArmInterface(1).PageTableChanged(*current_page_table, address_space_width);
+ system.ArmInterface(2).PageTableChanged(*current_page_table, address_space_width);
+ system.ArmInterface(3).PageTableChanged(*current_page_table, address_space_width);
+ }
-static void MapPages(Common::PageTable& page_table, VAddr base, u64 size, u8* memory,
- Common::PageType type) {
- LOG_DEBUG(HW_Memory, "Mapping {} onto {:016X}-{:016X}", fmt::ptr(memory), base * PAGE_SIZE,
- (base + size) * PAGE_SIZE);
-
- // During boot, current_page_table might not be set yet, in which case we need not flush
- if (Core::System::GetInstance().IsPoweredOn()) {
- auto& gpu = Core::System::GetInstance().GPU();
- for (u64 i = 0; i < size; i++) {
- const auto page = base + i;
- if (page_table.attributes[page] == Common::PageType::RasterizerCachedMemory) {
- gpu.FlushAndInvalidateRegion(page << PAGE_BITS, PAGE_SIZE);
- }
- }
+ void MapMemoryRegion(Common::PageTable& page_table, VAddr base, u64 size, u8* target) {
+ ASSERT_MSG((size & PAGE_MASK) == 0, "non-page aligned size: {:016X}", size);
+ ASSERT_MSG((base & PAGE_MASK) == 0, "non-page aligned base: {:016X}", base);
+ MapPages(page_table, base / PAGE_SIZE, size / PAGE_SIZE, target, Common::PageType::Memory);
}
- VAddr end = base + size;
- ASSERT_MSG(end <= page_table.pointers.size(), "out of range mapping at {:016X}",
- base + page_table.pointers.size());
+ void MapIoRegion(Common::PageTable& page_table, VAddr base, u64 size,
+ Common::MemoryHookPointer mmio_handler) {
+ ASSERT_MSG((size & PAGE_MASK) == 0, "non-page aligned size: {:016X}", size);
+ ASSERT_MSG((base & PAGE_MASK) == 0, "non-page aligned base: {:016X}", base);
+ MapPages(page_table, base / PAGE_SIZE, size / PAGE_SIZE, nullptr,
+ Common::PageType::Special);
+
+ const auto interval = boost::icl::discrete_interval<VAddr>::closed(base, base + size - 1);
+ const Common::SpecialRegion region{Common::SpecialRegion::Type::IODevice,
+ std::move(mmio_handler)};
+ page_table.special_regions.add(
+ std::make_pair(interval, std::set<Common::SpecialRegion>{region}));
+ }
- std::fill(page_table.attributes.begin() + base, page_table.attributes.begin() + end, type);
+ void UnmapRegion(Common::PageTable& page_table, VAddr base, u64 size) {
+ ASSERT_MSG((size & PAGE_MASK) == 0, "non-page aligned size: {:016X}", size);
+ ASSERT_MSG((base & PAGE_MASK) == 0, "non-page aligned base: {:016X}", base);
+ MapPages(page_table, base / PAGE_SIZE, size / PAGE_SIZE, nullptr,
+ Common::PageType::Unmapped);
- if (memory == nullptr) {
- std::fill(page_table.pointers.begin() + base, page_table.pointers.begin() + end, memory);
- } else {
- while (base != end) {
- page_table.pointers[base] = memory;
+ const auto interval = boost::icl::discrete_interval<VAddr>::closed(base, base + size - 1);
+ page_table.special_regions.erase(interval);
+ }
- base += 1;
- memory += PAGE_SIZE;
- }
+ void AddDebugHook(Common::PageTable& page_table, VAddr base, u64 size,
+ Common::MemoryHookPointer hook) {
+ const auto interval = boost::icl::discrete_interval<VAddr>::closed(base, base + size - 1);
+ const Common::SpecialRegion region{Common::SpecialRegion::Type::DebugHook, std::move(hook)};
+ page_table.special_regions.add(
+ std::make_pair(interval, std::set<Common::SpecialRegion>{region}));
}
-}
-void MapMemoryRegion(Common::PageTable& page_table, VAddr base, u64 size, u8* target) {
- ASSERT_MSG((size & PAGE_MASK) == 0, "non-page aligned size: {:016X}", size);
- ASSERT_MSG((base & PAGE_MASK) == 0, "non-page aligned base: {:016X}", base);
- MapPages(page_table, base / PAGE_SIZE, size / PAGE_SIZE, target, Common::PageType::Memory);
-}
+ void RemoveDebugHook(Common::PageTable& page_table, VAddr base, u64 size,
+ Common::MemoryHookPointer hook) {
+ const auto interval = boost::icl::discrete_interval<VAddr>::closed(base, base + size - 1);
+ const Common::SpecialRegion region{Common::SpecialRegion::Type::DebugHook, std::move(hook)};
+ page_table.special_regions.subtract(
+ std::make_pair(interval, std::set<Common::SpecialRegion>{region}));
+ }
-void MapIoRegion(Common::PageTable& page_table, VAddr base, u64 size,
- Common::MemoryHookPointer mmio_handler) {
- ASSERT_MSG((size & PAGE_MASK) == 0, "non-page aligned size: {:016X}", size);
- ASSERT_MSG((base & PAGE_MASK) == 0, "non-page aligned base: {:016X}", base);
- MapPages(page_table, base / PAGE_SIZE, size / PAGE_SIZE, nullptr, Common::PageType::Special);
+ bool IsValidVirtualAddress(const Kernel::Process& process, const VAddr vaddr) const {
+ const auto& page_table = process.VMManager().page_table;
- auto interval = boost::icl::discrete_interval<VAddr>::closed(base, base + size - 1);
- Common::SpecialRegion region{Common::SpecialRegion::Type::IODevice, std::move(mmio_handler)};
- page_table.special_regions.add(
- std::make_pair(interval, std::set<Common::SpecialRegion>{region}));
-}
+ const u8* const page_pointer = page_table.pointers[vaddr >> PAGE_BITS];
+ if (page_pointer != nullptr) {
+ return true;
+ }
-void UnmapRegion(Common::PageTable& page_table, VAddr base, u64 size) {
- ASSERT_MSG((size & PAGE_MASK) == 0, "non-page aligned size: {:016X}", size);
- ASSERT_MSG((base & PAGE_MASK) == 0, "non-page aligned base: {:016X}", base);
- MapPages(page_table, base / PAGE_SIZE, size / PAGE_SIZE, nullptr, Common::PageType::Unmapped);
+ if (page_table.attributes[vaddr >> PAGE_BITS] == Common::PageType::RasterizerCachedMemory) {
+ return true;
+ }
- auto interval = boost::icl::discrete_interval<VAddr>::closed(base, base + size - 1);
- page_table.special_regions.erase(interval);
-}
+ if (page_table.attributes[vaddr >> PAGE_BITS] != Common::PageType::Special) {
+ return false;
+ }
-void AddDebugHook(Common::PageTable& page_table, VAddr base, u64 size,
- Common::MemoryHookPointer hook) {
- auto interval = boost::icl::discrete_interval<VAddr>::closed(base, base + size - 1);
- Common::SpecialRegion region{Common::SpecialRegion::Type::DebugHook, std::move(hook)};
- page_table.special_regions.add(
- std::make_pair(interval, std::set<Common::SpecialRegion>{region}));
-}
+ return false;
+ }
-void RemoveDebugHook(Common::PageTable& page_table, VAddr base, u64 size,
- Common::MemoryHookPointer hook) {
- auto interval = boost::icl::discrete_interval<VAddr>::closed(base, base + size - 1);
- Common::SpecialRegion region{Common::SpecialRegion::Type::DebugHook, std::move(hook)};
- page_table.special_regions.subtract(
- std::make_pair(interval, std::set<Common::SpecialRegion>{region}));
-}
+ bool IsValidVirtualAddress(VAddr vaddr) const {
+ return IsValidVirtualAddress(*system.CurrentProcess(), vaddr);
+ }
-/**
- * Gets a pointer to the exact memory at the virtual address (i.e. not page aligned)
- * using a VMA from the current process
- */
-static u8* GetPointerFromVMA(const Kernel::Process& process, VAddr vaddr) {
- const auto& vm_manager = process.VMManager();
-
- const auto it = vm_manager.FindVMA(vaddr);
- DEBUG_ASSERT(vm_manager.IsValidHandle(it));
-
- u8* direct_pointer = nullptr;
- const auto& vma = it->second;
- switch (vma.type) {
- case Kernel::VMAType::AllocatedMemoryBlock:
- direct_pointer = vma.backing_block->data() + vma.offset;
- break;
- case Kernel::VMAType::BackingMemory:
- direct_pointer = vma.backing_memory;
- break;
- case Kernel::VMAType::Free:
- return nullptr;
- default:
- UNREACHABLE();
+ /**
+ * Gets a pointer to the exact memory at the virtual address (i.e. not page aligned)
+ * using a VMA from the current process
+ */
+ u8* GetPointerFromVMA(const Kernel::Process& process, VAddr vaddr) {
+ const auto& vm_manager = process.VMManager();
+
+ const auto it = vm_manager.FindVMA(vaddr);
+ DEBUG_ASSERT(vm_manager.IsValidHandle(it));
+
+ u8* direct_pointer = nullptr;
+ const auto& vma = it->second;
+ switch (vma.type) {
+ case Kernel::VMAType::AllocatedMemoryBlock:
+ direct_pointer = vma.backing_block->data() + vma.offset;
+ break;
+ case Kernel::VMAType::BackingMemory:
+ direct_pointer = vma.backing_memory;
+ break;
+ case Kernel::VMAType::Free:
+ return nullptr;
+ default:
+ UNREACHABLE();
+ }
+
+ return direct_pointer + (vaddr - vma.base);
}
- return direct_pointer + (vaddr - vma.base);
-}
+ /**
+ * Gets a pointer to the exact memory at the virtual address (i.e. not page aligned)
+ * using a VMA from the current process.
+ */
+ u8* GetPointerFromVMA(VAddr vaddr) {
+ return GetPointerFromVMA(*system.CurrentProcess(), vaddr);
+ }
-/**
- * Gets a pointer to the exact memory at the virtual address (i.e. not page aligned)
- * using a VMA from the current process.
- */
-static u8* GetPointerFromVMA(VAddr vaddr) {
- return GetPointerFromVMA(*Core::System::GetInstance().CurrentProcess(), vaddr);
-}
+ u8* GetPointer(const VAddr vaddr) {
+ u8* const page_pointer = current_page_table->pointers[vaddr >> PAGE_BITS];
+ if (page_pointer != nullptr) {
+ return page_pointer + (vaddr & PAGE_MASK);
+ }
-template <typename T>
-T Read(const VAddr vaddr) {
- const u8* page_pointer = current_page_table->pointers[vaddr >> PAGE_BITS];
- if (page_pointer) {
- // NOTE: Avoid adding any extra logic to this fast-path block
- T value;
- std::memcpy(&value, &page_pointer[vaddr & PAGE_MASK], sizeof(T));
- return value;
- }
-
- Common::PageType type = current_page_table->attributes[vaddr >> PAGE_BITS];
- switch (type) {
- case Common::PageType::Unmapped:
- LOG_ERROR(HW_Memory, "Unmapped Read{} @ 0x{:08X}", sizeof(T) * 8, vaddr);
- return 0;
- case Common::PageType::Memory:
- ASSERT_MSG(false, "Mapped memory page without a pointer @ {:016X}", vaddr);
- break;
- case Common::PageType::RasterizerCachedMemory: {
- auto host_ptr{GetPointerFromVMA(vaddr)};
- Core::System::GetInstance().GPU().FlushRegion(ToCacheAddr(host_ptr), sizeof(T));
- T value;
- std::memcpy(&value, host_ptr, sizeof(T));
- return value;
- }
- default:
- UNREACHABLE();
- }
- return {};
-}
+ if (current_page_table->attributes[vaddr >> PAGE_BITS] ==
+ Common::PageType::RasterizerCachedMemory) {
+ return GetPointerFromVMA(vaddr);
+ }
-template <typename T>
-void Write(const VAddr vaddr, const T data) {
- u8* page_pointer = current_page_table->pointers[vaddr >> PAGE_BITS];
- if (page_pointer) {
- // NOTE: Avoid adding any extra logic to this fast-path block
- std::memcpy(&page_pointer[vaddr & PAGE_MASK], &data, sizeof(T));
- return;
- }
-
- Common::PageType type = current_page_table->attributes[vaddr >> PAGE_BITS];
- switch (type) {
- case Common::PageType::Unmapped:
- LOG_ERROR(HW_Memory, "Unmapped Write{} 0x{:08X} @ 0x{:016X}", sizeof(data) * 8,
- static_cast<u32>(data), vaddr);
- return;
- case Common::PageType::Memory:
- ASSERT_MSG(false, "Mapped memory page without a pointer @ {:016X}", vaddr);
- break;
- case Common::PageType::RasterizerCachedMemory: {
- auto host_ptr{GetPointerFromVMA(vaddr)};
- Core::System::GetInstance().GPU().InvalidateRegion(ToCacheAddr(host_ptr), sizeof(T));
- std::memcpy(host_ptr, &data, sizeof(T));
- break;
- }
- default:
- UNREACHABLE();
+ LOG_ERROR(HW_Memory, "Unknown GetPointer @ 0x{:016X}", vaddr);
+ return nullptr;
}
-}
-bool IsValidVirtualAddress(const Kernel::Process& process, const VAddr vaddr) {
- const auto& page_table = process.VMManager().page_table;
+ u8 Read8(const VAddr addr) {
+ return Read<u8>(addr);
+ }
- const u8* page_pointer = page_table.pointers[vaddr >> PAGE_BITS];
- if (page_pointer)
- return true;
+ u16 Read16(const VAddr addr) {
+ return Read<u16_le>(addr);
+ }
- if (page_table.attributes[vaddr >> PAGE_BITS] == Common::PageType::RasterizerCachedMemory)
- return true;
+ u32 Read32(const VAddr addr) {
+ return Read<u32_le>(addr);
+ }
- if (page_table.attributes[vaddr >> PAGE_BITS] != Common::PageType::Special)
- return false;
+ u64 Read64(const VAddr addr) {
+ return Read<u64_le>(addr);
+ }
- return false;
-}
+ void Write8(const VAddr addr, const u8 data) {
+ Write<u8>(addr, data);
+ }
-bool IsValidVirtualAddress(const VAddr vaddr) {
- return IsValidVirtualAddress(*Core::System::GetInstance().CurrentProcess(), vaddr);
-}
+ void Write16(const VAddr addr, const u16 data) {
+ Write<u16_le>(addr, data);
+ }
-bool IsKernelVirtualAddress(const VAddr vaddr) {
- return KERNEL_REGION_VADDR <= vaddr && vaddr < KERNEL_REGION_END;
-}
+ void Write32(const VAddr addr, const u32 data) {
+ Write<u32_le>(addr, data);
+ }
-u8* GetPointer(const VAddr vaddr) {
- u8* page_pointer = current_page_table->pointers[vaddr >> PAGE_BITS];
- if (page_pointer) {
- return page_pointer + (vaddr & PAGE_MASK);
+ void Write64(const VAddr addr, const u64 data) {
+ Write<u64_le>(addr, data);
}
- if (current_page_table->attributes[vaddr >> PAGE_BITS] ==
- Common::PageType::RasterizerCachedMemory) {
- return GetPointerFromVMA(vaddr);
+ std::string ReadCString(VAddr vaddr, std::size_t max_length) {
+ std::string string;
+ string.reserve(max_length);
+ for (std::size_t i = 0; i < max_length; ++i) {
+ const char c = Read8(vaddr);
+ if (c == '\0') {
+ break;
+ }
+ string.push_back(c);
+ ++vaddr;
+ }
+ string.shrink_to_fit();
+ return string;
}
- LOG_ERROR(HW_Memory, "Unknown GetPointer @ 0x{:016X}", vaddr);
- return nullptr;
-}
+ void ReadBlock(const Kernel::Process& process, const VAddr src_addr, void* dest_buffer,
+ const std::size_t size) {
+ const auto& page_table = process.VMManager().page_table;
+
+ std::size_t remaining_size = size;
+ std::size_t page_index = src_addr >> PAGE_BITS;
+ std::size_t page_offset = src_addr & PAGE_MASK;
+
+ while (remaining_size > 0) {
+ const std::size_t copy_amount =
+ std::min(static_cast<std::size_t>(PAGE_SIZE) - page_offset, remaining_size);
+ const auto current_vaddr = static_cast<VAddr>((page_index << PAGE_BITS) + page_offset);
+
+ switch (page_table.attributes[page_index]) {
+ case Common::PageType::Unmapped: {
+ LOG_ERROR(HW_Memory,
+ "Unmapped ReadBlock @ 0x{:016X} (start address = 0x{:016X}, size = {})",
+ current_vaddr, src_addr, size);
+ std::memset(dest_buffer, 0, copy_amount);
+ break;
+ }
+ case Common::PageType::Memory: {
+ DEBUG_ASSERT(page_table.pointers[page_index]);
-std::string ReadCString(VAddr vaddr, std::size_t max_length) {
- std::string string;
- string.reserve(max_length);
- for (std::size_t i = 0; i < max_length; ++i) {
- char c = Read8(vaddr);
- if (c == '\0')
- break;
- string.push_back(c);
- ++vaddr;
+ const u8* const src_ptr = page_table.pointers[page_index] + page_offset;
+ std::memcpy(dest_buffer, src_ptr, copy_amount);
+ break;
+ }
+ case Common::PageType::RasterizerCachedMemory: {
+ const u8* const host_ptr = GetPointerFromVMA(process, current_vaddr);
+ system.GPU().FlushRegion(ToCacheAddr(host_ptr), copy_amount);
+ std::memcpy(dest_buffer, host_ptr, copy_amount);
+ break;
+ }
+ default:
+ UNREACHABLE();
+ }
+
+ page_index++;
+ page_offset = 0;
+ dest_buffer = static_cast<u8*>(dest_buffer) + copy_amount;
+ remaining_size -= copy_amount;
+ }
}
- string.shrink_to_fit();
- return string;
-}
-void RasterizerMarkRegionCached(VAddr vaddr, u64 size, bool cached) {
- if (vaddr == 0) {
- return;
+ void ReadBlock(const VAddr src_addr, void* dest_buffer, const std::size_t size) {
+ ReadBlock(*system.CurrentProcess(), src_addr, dest_buffer, size);
}
- // Iterate over a contiguous CPU address space, which corresponds to the specified GPU address
- // space, marking the region as un/cached. The region is marked un/cached at a granularity of
- // CPU pages, hence why we iterate on a CPU page basis (note: GPU page size is different). This
- // assumes the specified GPU address region is contiguous as well.
+ void WriteBlock(const Kernel::Process& process, const VAddr dest_addr, const void* src_buffer,
+ const std::size_t size) {
+ const auto& page_table = process.VMManager().page_table;
+ std::size_t remaining_size = size;
+ std::size_t page_index = dest_addr >> PAGE_BITS;
+ std::size_t page_offset = dest_addr & PAGE_MASK;
+
+ while (remaining_size > 0) {
+ const std::size_t copy_amount =
+ std::min(static_cast<std::size_t>(PAGE_SIZE) - page_offset, remaining_size);
+ const auto current_vaddr = static_cast<VAddr>((page_index << PAGE_BITS) + page_offset);
+
+ switch (page_table.attributes[page_index]) {
+ case Common::PageType::Unmapped: {
+ LOG_ERROR(HW_Memory,
+ "Unmapped WriteBlock @ 0x{:016X} (start address = 0x{:016X}, size = {})",
+ current_vaddr, dest_addr, size);
+ break;
+ }
+ case Common::PageType::Memory: {
+ DEBUG_ASSERT(page_table.pointers[page_index]);
+
+ u8* const dest_ptr = page_table.pointers[page_index] + page_offset;
+ std::memcpy(dest_ptr, src_buffer, copy_amount);
+ break;
+ }
+ case Common::PageType::RasterizerCachedMemory: {
+ u8* const host_ptr = GetPointerFromVMA(process, current_vaddr);
+ system.GPU().InvalidateRegion(ToCacheAddr(host_ptr), copy_amount);
+ std::memcpy(host_ptr, src_buffer, copy_amount);
+ break;
+ }
+ default:
+ UNREACHABLE();
+ }
+
+ page_index++;
+ page_offset = 0;
+ src_buffer = static_cast<const u8*>(src_buffer) + copy_amount;
+ remaining_size -= copy_amount;
+ }
+ }
- u64 num_pages = ((vaddr + size - 1) >> PAGE_BITS) - (vaddr >> PAGE_BITS) + 1;
- for (unsigned i = 0; i < num_pages; ++i, vaddr += PAGE_SIZE) {
- Common::PageType& page_type = current_page_table->attributes[vaddr >> PAGE_BITS];
+ void WriteBlock(const VAddr dest_addr, const void* src_buffer, const std::size_t size) {
+ WriteBlock(*system.CurrentProcess(), dest_addr, src_buffer, size);
+ }
- if (cached) {
- // Switch page type to cached if now cached
- switch (page_type) {
- case Common::PageType::Unmapped:
- // It is not necessary for a process to have this region mapped into its address
- // space, for example, a system module need not have a VRAM mapping.
+ void ZeroBlock(const Kernel::Process& process, const VAddr dest_addr, const std::size_t size) {
+ const auto& page_table = process.VMManager().page_table;
+ std::size_t remaining_size = size;
+ std::size_t page_index = dest_addr >> PAGE_BITS;
+ std::size_t page_offset = dest_addr & PAGE_MASK;
+
+ while (remaining_size > 0) {
+ const std::size_t copy_amount =
+ std::min(static_cast<std::size_t>(PAGE_SIZE) - page_offset, remaining_size);
+ const auto current_vaddr = static_cast<VAddr>((page_index << PAGE_BITS) + page_offset);
+
+ switch (page_table.attributes[page_index]) {
+ case Common::PageType::Unmapped: {
+ LOG_ERROR(HW_Memory,
+ "Unmapped ZeroBlock @ 0x{:016X} (start address = 0x{:016X}, size = {})",
+ current_vaddr, dest_addr, size);
break;
- case Common::PageType::Memory:
- page_type = Common::PageType::RasterizerCachedMemory;
- current_page_table->pointers[vaddr >> PAGE_BITS] = nullptr;
+ }
+ case Common::PageType::Memory: {
+ DEBUG_ASSERT(page_table.pointers[page_index]);
+
+ u8* dest_ptr = page_table.pointers[page_index] + page_offset;
+ std::memset(dest_ptr, 0, copy_amount);
break;
- case Common::PageType::RasterizerCachedMemory:
- // There can be more than one GPU region mapped per CPU region, so it's common that
- // this area is already marked as cached.
+ }
+ case Common::PageType::RasterizerCachedMemory: {
+ u8* const host_ptr = GetPointerFromVMA(process, current_vaddr);
+ system.GPU().InvalidateRegion(ToCacheAddr(host_ptr), copy_amount);
+ std::memset(host_ptr, 0, copy_amount);
break;
+ }
default:
UNREACHABLE();
}
- } else {
- // Switch page type to uncached if now uncached
- switch (page_type) {
- case Common::PageType::Unmapped:
- // It is not necessary for a process to have this region mapped into its address
- // space, for example, a system module need not have a VRAM mapping.
+
+ page_index++;
+ page_offset = 0;
+ remaining_size -= copy_amount;
+ }
+ }
+
+ void ZeroBlock(const VAddr dest_addr, const std::size_t size) {
+ ZeroBlock(*system.CurrentProcess(), dest_addr, size);
+ }
+
+ void CopyBlock(const Kernel::Process& process, VAddr dest_addr, VAddr src_addr,
+ const std::size_t size) {
+ const auto& page_table = process.VMManager().page_table;
+ std::size_t remaining_size = size;
+ std::size_t page_index = src_addr >> PAGE_BITS;
+ std::size_t page_offset = src_addr & PAGE_MASK;
+
+ while (remaining_size > 0) {
+ const std::size_t copy_amount =
+ std::min(static_cast<std::size_t>(PAGE_SIZE) - page_offset, remaining_size);
+ const auto current_vaddr = static_cast<VAddr>((page_index << PAGE_BITS) + page_offset);
+
+ switch (page_table.attributes[page_index]) {
+ case Common::PageType::Unmapped: {
+ LOG_ERROR(HW_Memory,
+ "Unmapped CopyBlock @ 0x{:016X} (start address = 0x{:016X}, size = {})",
+ current_vaddr, src_addr, size);
+ ZeroBlock(process, dest_addr, copy_amount);
break;
- case Common::PageType::Memory:
- // There can be more than one GPU region mapped per CPU region, so it's common that
- // this area is already unmarked as cached.
+ }
+ case Common::PageType::Memory: {
+ DEBUG_ASSERT(page_table.pointers[page_index]);
+ const u8* src_ptr = page_table.pointers[page_index] + page_offset;
+ WriteBlock(process, dest_addr, src_ptr, copy_amount);
break;
+ }
case Common::PageType::RasterizerCachedMemory: {
- u8* pointer = GetPointerFromVMA(vaddr & ~PAGE_MASK);
- if (pointer == nullptr) {
- // It's possible that this function has been called while updating the pagetable
- // after unmapping a VMA. In that case the underlying VMA will no longer exist,
- // and we should just leave the pagetable entry blank.
- page_type = Common::PageType::Unmapped;
- } else {
- page_type = Common::PageType::Memory;
- current_page_table->pointers[vaddr >> PAGE_BITS] = pointer;
- }
+ const u8* const host_ptr = GetPointerFromVMA(process, current_vaddr);
+ system.GPU().FlushRegion(ToCacheAddr(host_ptr), copy_amount);
+ WriteBlock(process, dest_addr, host_ptr, copy_amount);
break;
}
default:
UNREACHABLE();
}
+
+ page_index++;
+ page_offset = 0;
+ dest_addr += static_cast<VAddr>(copy_amount);
+ src_addr += static_cast<VAddr>(copy_amount);
+ remaining_size -= copy_amount;
}
}
-}
-u8 Read8(const VAddr addr) {
- return Read<u8>(addr);
-}
+ void CopyBlock(VAddr dest_addr, VAddr src_addr, std::size_t size) {
+ return CopyBlock(*system.CurrentProcess(), dest_addr, src_addr, size);
+ }
-u16 Read16(const VAddr addr) {
- return Read<u16_le>(addr);
-}
+ void RasterizerMarkRegionCached(VAddr vaddr, u64 size, bool cached) {
+ if (vaddr == 0) {
+ return;
+ }
-u32 Read32(const VAddr addr) {
- return Read<u32_le>(addr);
-}
+ // Iterate over a contiguous CPU address space, which corresponds to the specified GPU
+ // address space, marking the region as un/cached. The region is marked un/cached at a
+ // granularity of CPU pages, hence why we iterate on a CPU page basis (note: GPU page size
+ // is different). This assumes the specified GPU address region is contiguous as well.
+
+ u64 num_pages = ((vaddr + size - 1) >> PAGE_BITS) - (vaddr >> PAGE_BITS) + 1;
+ for (unsigned i = 0; i < num_pages; ++i, vaddr += PAGE_SIZE) {
+ Common::PageType& page_type = current_page_table->attributes[vaddr >> PAGE_BITS];
+
+ if (cached) {
+ // Switch page type to cached if now cached
+ switch (page_type) {
+ case Common::PageType::Unmapped:
+ // It is not necessary for a process to have this region mapped into its address
+ // space, for example, a system module need not have a VRAM mapping.
+ break;
+ case Common::PageType::Memory:
+ page_type = Common::PageType::RasterizerCachedMemory;
+ current_page_table->pointers[vaddr >> PAGE_BITS] = nullptr;
+ break;
+ case Common::PageType::RasterizerCachedMemory:
+ // There can be more than one GPU region mapped per CPU region, so it's common
+ // that this area is already marked as cached.
+ break;
+ default:
+ UNREACHABLE();
+ }
+ } else {
+ // Switch page type to uncached if now uncached
+ switch (page_type) {
+ case Common::PageType::Unmapped:
+ // It is not necessary for a process to have this region mapped into its address
+ // space, for example, a system module need not have a VRAM mapping.
+ break;
+ case Common::PageType::Memory:
+ // There can be more than one GPU region mapped per CPU region, so it's common
+ // that this area is already unmarked as cached.
+ break;
+ case Common::PageType::RasterizerCachedMemory: {
+ u8* pointer = GetPointerFromVMA(vaddr & ~PAGE_MASK);
+ if (pointer == nullptr) {
+ // It's possible that this function has been called while updating the
+ // pagetable after unmapping a VMA. In that case the underlying VMA will no
+ // longer exist, and we should just leave the pagetable entry blank.
+ page_type = Common::PageType::Unmapped;
+ } else {
+ page_type = Common::PageType::Memory;
+ current_page_table->pointers[vaddr >> PAGE_BITS] = pointer;
+ }
+ break;
+ }
+ default:
+ UNREACHABLE();
+ }
+ }
+ }
+ }
-u64 Read64(const VAddr addr) {
- return Read<u64_le>(addr);
-}
+ /**
+ * Maps a region of pages as a specific type.
+ *
+ * @param page_table The page table to use to perform the mapping.
+ * @param base The base address to begin mapping at.
+ * @param size The total size of the range in bytes.
+ * @param memory The memory to map.
+ * @param type The page type to map the memory as.
+ */
+ void MapPages(Common::PageTable& page_table, VAddr base, u64 size, u8* memory,
+ Common::PageType type) {
+ LOG_DEBUG(HW_Memory, "Mapping {} onto {:016X}-{:016X}", fmt::ptr(memory), base * PAGE_SIZE,
+ (base + size) * PAGE_SIZE);
+
+ // During boot, current_page_table might not be set yet, in which case we need not flush
+ if (system.IsPoweredOn()) {
+ auto& gpu = system.GPU();
+ for (u64 i = 0; i < size; i++) {
+ const auto page = base + i;
+ if (page_table.attributes[page] == Common::PageType::RasterizerCachedMemory) {
+ gpu.FlushAndInvalidateRegion(page << PAGE_BITS, PAGE_SIZE);
+ }
+ }
+ }
-void ReadBlock(const Kernel::Process& process, const VAddr src_addr, void* dest_buffer,
- const std::size_t size) {
- const auto& page_table = process.VMManager().page_table;
-
- std::size_t remaining_size = size;
- std::size_t page_index = src_addr >> PAGE_BITS;
- std::size_t page_offset = src_addr & PAGE_MASK;
-
- while (remaining_size > 0) {
- const std::size_t copy_amount =
- std::min(static_cast<std::size_t>(PAGE_SIZE) - page_offset, remaining_size);
- const VAddr current_vaddr = static_cast<VAddr>((page_index << PAGE_BITS) + page_offset);
-
- switch (page_table.attributes[page_index]) {
- case Common::PageType::Unmapped: {
- LOG_ERROR(HW_Memory,
- "Unmapped ReadBlock @ 0x{:016X} (start address = 0x{:016X}, size = {})",
- current_vaddr, src_addr, size);
- std::memset(dest_buffer, 0, copy_amount);
- break;
+ const VAddr end = base + size;
+ ASSERT_MSG(end <= page_table.pointers.size(), "out of range mapping at {:016X}",
+ base + page_table.pointers.size());
+
+ std::fill(page_table.attributes.begin() + base, page_table.attributes.begin() + end, type);
+
+ if (memory == nullptr) {
+ std::fill(page_table.pointers.begin() + base, page_table.pointers.begin() + end,
+ memory);
+ } else {
+ while (base != end) {
+ page_table.pointers[base] = memory;
+
+ base += 1;
+ memory += PAGE_SIZE;
+ }
+ }
+ }
+
+ /**
+ * Reads a particular data type out of memory at the given virtual address.
+ *
+ * @param vaddr The virtual address to read the data type from.
+ *
+ * @tparam T The data type to read out of memory. This type *must* be
+ * trivially copyable, otherwise the behavior of this function
+ * is undefined.
+ *
+ * @returns The instance of T read from the specified virtual address.
+ */
+ template <typename T>
+ T Read(const VAddr vaddr) {
+ const u8* const page_pointer = current_page_table->pointers[vaddr >> PAGE_BITS];
+ if (page_pointer != nullptr) {
+ // NOTE: Avoid adding any extra logic to this fast-path block
+ T value;
+ std::memcpy(&value, &page_pointer[vaddr & PAGE_MASK], sizeof(T));
+ return value;
}
- case Common::PageType::Memory: {
- DEBUG_ASSERT(page_table.pointers[page_index]);
- const u8* src_ptr = page_table.pointers[page_index] + page_offset;
- std::memcpy(dest_buffer, src_ptr, copy_amount);
+ const Common::PageType type = current_page_table->attributes[vaddr >> PAGE_BITS];
+ switch (type) {
+ case Common::PageType::Unmapped:
+ LOG_ERROR(HW_Memory, "Unmapped Read{} @ 0x{:08X}", sizeof(T) * 8, vaddr);
+ return 0;
+ case Common::PageType::Memory:
+ ASSERT_MSG(false, "Mapped memory page without a pointer @ {:016X}", vaddr);
break;
+ case Common::PageType::RasterizerCachedMemory: {
+ const u8* const host_ptr = GetPointerFromVMA(vaddr);
+ system.GPU().FlushRegion(ToCacheAddr(host_ptr), sizeof(T));
+ T value;
+ std::memcpy(&value, host_ptr, sizeof(T));
+ return value;
+ }
+ default:
+ UNREACHABLE();
}
+ return {};
+ }
+
+ /**
+ * Writes a particular data type to memory at the given virtual address.
+ *
+ * @param vaddr The virtual address to write the data type to.
+ *
+ * @tparam T The data type to write to memory. This type *must* be
+ * trivially copyable, otherwise the behavior of this function
+ * is undefined.
+ *
+ * @returns The instance of T write to the specified virtual address.
+ */
+ template <typename T>
+ void Write(const VAddr vaddr, const T data) {
+ u8* const page_pointer = current_page_table->pointers[vaddr >> PAGE_BITS];
+ if (page_pointer != nullptr) {
+ // NOTE: Avoid adding any extra logic to this fast-path block
+ std::memcpy(&page_pointer[vaddr & PAGE_MASK], &data, sizeof(T));
+ return;
+ }
+
+ const Common::PageType type = current_page_table->attributes[vaddr >> PAGE_BITS];
+ switch (type) {
+ case Common::PageType::Unmapped:
+ LOG_ERROR(HW_Memory, "Unmapped Write{} 0x{:08X} @ 0x{:016X}", sizeof(data) * 8,
+ static_cast<u32>(data), vaddr);
+ return;
+ case Common::PageType::Memory:
+ ASSERT_MSG(false, "Mapped memory page without a pointer @ {:016X}", vaddr);
+ break;
case Common::PageType::RasterizerCachedMemory: {
- const auto& host_ptr{GetPointerFromVMA(process, current_vaddr)};
- Core::System::GetInstance().GPU().FlushRegion(ToCacheAddr(host_ptr), copy_amount);
- std::memcpy(dest_buffer, host_ptr, copy_amount);
+ u8* const host_ptr{GetPointerFromVMA(vaddr)};
+ system.GPU().InvalidateRegion(ToCacheAddr(host_ptr), sizeof(T));
+ std::memcpy(host_ptr, &data, sizeof(T));
break;
}
default:
UNREACHABLE();
}
-
- page_index++;
- page_offset = 0;
- dest_buffer = static_cast<u8*>(dest_buffer) + copy_amount;
- remaining_size -= copy_amount;
}
+
+ Common::PageTable* current_page_table = nullptr;
+ Core::System& system;
+};
+
+Memory::Memory(Core::System& system) : impl{std::make_unique<Impl>(system)} {}
+Memory::~Memory() = default;
+
+void Memory::SetCurrentPageTable(Kernel::Process& process) {
+ impl->SetCurrentPageTable(process);
}
-void ReadBlock(const VAddr src_addr, void* dest_buffer, const std::size_t size) {
- ReadBlock(*Core::System::GetInstance().CurrentProcess(), src_addr, dest_buffer, size);
+void Memory::MapMemoryRegion(Common::PageTable& page_table, VAddr base, u64 size, u8* target) {
+ impl->MapMemoryRegion(page_table, base, size, target);
}
-void Write8(const VAddr addr, const u8 data) {
- Write<u8>(addr, data);
+void Memory::MapIoRegion(Common::PageTable& page_table, VAddr base, u64 size,
+ Common::MemoryHookPointer mmio_handler) {
+ impl->MapIoRegion(page_table, base, size, std::move(mmio_handler));
}
-void Write16(const VAddr addr, const u16 data) {
- Write<u16_le>(addr, data);
+void Memory::UnmapRegion(Common::PageTable& page_table, VAddr base, u64 size) {
+ impl->UnmapRegion(page_table, base, size);
}
-void Write32(const VAddr addr, const u32 data) {
- Write<u32_le>(addr, data);
+void Memory::AddDebugHook(Common::PageTable& page_table, VAddr base, u64 size,
+ Common::MemoryHookPointer hook) {
+ impl->AddDebugHook(page_table, base, size, std::move(hook));
}
-void Write64(const VAddr addr, const u64 data) {
- Write<u64_le>(addr, data);
+void Memory::RemoveDebugHook(Common::PageTable& page_table, VAddr base, u64 size,
+ Common::MemoryHookPointer hook) {
+ impl->RemoveDebugHook(page_table, base, size, std::move(hook));
}
-void WriteBlock(const Kernel::Process& process, const VAddr dest_addr, const void* src_buffer,
- const std::size_t size) {
- const auto& page_table = process.VMManager().page_table;
- std::size_t remaining_size = size;
- std::size_t page_index = dest_addr >> PAGE_BITS;
- std::size_t page_offset = dest_addr & PAGE_MASK;
-
- while (remaining_size > 0) {
- const std::size_t copy_amount =
- std::min(static_cast<std::size_t>(PAGE_SIZE) - page_offset, remaining_size);
- const VAddr current_vaddr = static_cast<VAddr>((page_index << PAGE_BITS) + page_offset);
-
- switch (page_table.attributes[page_index]) {
- case Common::PageType::Unmapped: {
- LOG_ERROR(HW_Memory,
- "Unmapped WriteBlock @ 0x{:016X} (start address = 0x{:016X}, size = {})",
- current_vaddr, dest_addr, size);
- break;
- }
- case Common::PageType::Memory: {
- DEBUG_ASSERT(page_table.pointers[page_index]);
+bool Memory::IsValidVirtualAddress(const Kernel::Process& process, const VAddr vaddr) const {
+ return impl->IsValidVirtualAddress(process, vaddr);
+}
- u8* dest_ptr = page_table.pointers[page_index] + page_offset;
- std::memcpy(dest_ptr, src_buffer, copy_amount);
- break;
- }
- case Common::PageType::RasterizerCachedMemory: {
- const auto& host_ptr{GetPointerFromVMA(process, current_vaddr)};
- Core::System::GetInstance().GPU().InvalidateRegion(ToCacheAddr(host_ptr), copy_amount);
- std::memcpy(host_ptr, src_buffer, copy_amount);
- break;
- }
- default:
- UNREACHABLE();
- }
+bool Memory::IsValidVirtualAddress(const VAddr vaddr) const {
+ return impl->IsValidVirtualAddress(vaddr);
+}
- page_index++;
- page_offset = 0;
- src_buffer = static_cast<const u8*>(src_buffer) + copy_amount;
- remaining_size -= copy_amount;
- }
+u8* Memory::GetPointer(VAddr vaddr) {
+ return impl->GetPointer(vaddr);
}
-void WriteBlock(const VAddr dest_addr, const void* src_buffer, const std::size_t size) {
- WriteBlock(*Core::System::GetInstance().CurrentProcess(), dest_addr, src_buffer, size);
+const u8* Memory::GetPointer(VAddr vaddr) const {
+ return impl->GetPointer(vaddr);
}
-void ZeroBlock(const Kernel::Process& process, const VAddr dest_addr, const std::size_t size) {
- const auto& page_table = process.VMManager().page_table;
- std::size_t remaining_size = size;
- std::size_t page_index = dest_addr >> PAGE_BITS;
- std::size_t page_offset = dest_addr & PAGE_MASK;
-
- while (remaining_size > 0) {
- const std::size_t copy_amount =
- std::min(static_cast<std::size_t>(PAGE_SIZE) - page_offset, remaining_size);
- const VAddr current_vaddr = static_cast<VAddr>((page_index << PAGE_BITS) + page_offset);
-
- switch (page_table.attributes[page_index]) {
- case Common::PageType::Unmapped: {
- LOG_ERROR(HW_Memory,
- "Unmapped ZeroBlock @ 0x{:016X} (start address = 0x{:016X}, size = {})",
- current_vaddr, dest_addr, size);
- break;
- }
- case Common::PageType::Memory: {
- DEBUG_ASSERT(page_table.pointers[page_index]);
+u8 Memory::Read8(const VAddr addr) {
+ return impl->Read8(addr);
+}
- u8* dest_ptr = page_table.pointers[page_index] + page_offset;
- std::memset(dest_ptr, 0, copy_amount);
- break;
- }
- case Common::PageType::RasterizerCachedMemory: {
- const auto& host_ptr{GetPointerFromVMA(process, current_vaddr)};
- Core::System::GetInstance().GPU().InvalidateRegion(ToCacheAddr(host_ptr), copy_amount);
- std::memset(host_ptr, 0, copy_amount);
- break;
- }
- default:
- UNREACHABLE();
- }
+u16 Memory::Read16(const VAddr addr) {
+ return impl->Read16(addr);
+}
- page_index++;
- page_offset = 0;
- remaining_size -= copy_amount;
- }
+u32 Memory::Read32(const VAddr addr) {
+ return impl->Read32(addr);
}
-void CopyBlock(const Kernel::Process& process, VAddr dest_addr, VAddr src_addr,
- const std::size_t size) {
- const auto& page_table = process.VMManager().page_table;
- std::size_t remaining_size = size;
- std::size_t page_index = src_addr >> PAGE_BITS;
- std::size_t page_offset = src_addr & PAGE_MASK;
-
- while (remaining_size > 0) {
- const std::size_t copy_amount =
- std::min(static_cast<std::size_t>(PAGE_SIZE) - page_offset, remaining_size);
- const VAddr current_vaddr = static_cast<VAddr>((page_index << PAGE_BITS) + page_offset);
-
- switch (page_table.attributes[page_index]) {
- case Common::PageType::Unmapped: {
- LOG_ERROR(HW_Memory,
- "Unmapped CopyBlock @ 0x{:016X} (start address = 0x{:016X}, size = {})",
- current_vaddr, src_addr, size);
- ZeroBlock(process, dest_addr, copy_amount);
- break;
- }
- case Common::PageType::Memory: {
- DEBUG_ASSERT(page_table.pointers[page_index]);
- const u8* src_ptr = page_table.pointers[page_index] + page_offset;
- WriteBlock(process, dest_addr, src_ptr, copy_amount);
- break;
- }
- case Common::PageType::RasterizerCachedMemory: {
- const auto& host_ptr{GetPointerFromVMA(process, current_vaddr)};
- Core::System::GetInstance().GPU().FlushRegion(ToCacheAddr(host_ptr), copy_amount);
- WriteBlock(process, dest_addr, host_ptr, copy_amount);
- break;
- }
- default:
- UNREACHABLE();
- }
+u64 Memory::Read64(const VAddr addr) {
+ return impl->Read64(addr);
+}
- page_index++;
- page_offset = 0;
- dest_addr += static_cast<VAddr>(copy_amount);
- src_addr += static_cast<VAddr>(copy_amount);
- remaining_size -= copy_amount;
- }
+void Memory::Write8(VAddr addr, u8 data) {
+ impl->Write8(addr, data);
+}
+
+void Memory::Write16(VAddr addr, u16 data) {
+ impl->Write16(addr, data);
+}
+
+void Memory::Write32(VAddr addr, u32 data) {
+ impl->Write32(addr, data);
+}
+
+void Memory::Write64(VAddr addr, u64 data) {
+ impl->Write64(addr, data);
+}
+
+std::string Memory::ReadCString(VAddr vaddr, std::size_t max_length) {
+ return impl->ReadCString(vaddr, max_length);
+}
+
+void Memory::ReadBlock(const Kernel::Process& process, const VAddr src_addr, void* dest_buffer,
+ const std::size_t size) {
+ impl->ReadBlock(process, src_addr, dest_buffer, size);
+}
+
+void Memory::ReadBlock(const VAddr src_addr, void* dest_buffer, const std::size_t size) {
+ impl->ReadBlock(src_addr, dest_buffer, size);
+}
+
+void Memory::WriteBlock(const Kernel::Process& process, VAddr dest_addr, const void* src_buffer,
+ std::size_t size) {
+ impl->WriteBlock(process, dest_addr, src_buffer, size);
+}
+
+void Memory::WriteBlock(const VAddr dest_addr, const void* src_buffer, const std::size_t size) {
+ impl->WriteBlock(dest_addr, src_buffer, size);
}
-void CopyBlock(VAddr dest_addr, VAddr src_addr, std::size_t size) {
- CopyBlock(*Core::System::GetInstance().CurrentProcess(), dest_addr, src_addr, size);
+void Memory::ZeroBlock(const Kernel::Process& process, VAddr dest_addr, std::size_t size) {
+ impl->ZeroBlock(process, dest_addr, size);
+}
+
+void Memory::ZeroBlock(VAddr dest_addr, std::size_t size) {
+ impl->ZeroBlock(dest_addr, size);
+}
+
+void Memory::CopyBlock(const Kernel::Process& process, VAddr dest_addr, VAddr src_addr,
+ const std::size_t size) {
+ impl->CopyBlock(process, dest_addr, src_addr, size);
+}
+
+void Memory::CopyBlock(VAddr dest_addr, VAddr src_addr, std::size_t size) {
+ impl->CopyBlock(dest_addr, src_addr, size);
+}
+
+void Memory::RasterizerMarkRegionCached(VAddr vaddr, u64 size, bool cached) {
+ impl->RasterizerMarkRegionCached(vaddr, size, cached);
+}
+
+bool IsKernelVirtualAddress(const VAddr vaddr) {
+ return KERNEL_REGION_VADDR <= vaddr && vaddr < KERNEL_REGION_END;
}
} // namespace Memory
diff --git a/src/core/memory.h b/src/core/memory.h
index 09008e1dd..1428a6d60 100644
--- a/src/core/memory.h
+++ b/src/core/memory.h
@@ -5,8 +5,18 @@
#pragma once
#include <cstddef>
+#include <memory>
#include <string>
#include "common/common_types.h"
+#include "common/memory_hook.h"
+
+namespace Common {
+struct PageTable;
+}
+
+namespace Core {
+class System;
+}
namespace Kernel {
class Process;
@@ -36,41 +46,369 @@ enum : VAddr {
KERNEL_REGION_END = KERNEL_REGION_VADDR + KERNEL_REGION_SIZE,
};
-/// Changes the currently active page table to that of
-/// the given process instance.
-void SetCurrentPageTable(Kernel::Process& process);
+/// Central class that handles all memory operations and state.
+class Memory {
+public:
+ explicit Memory(Core::System& system);
+ ~Memory();
-/// Determines if the given VAddr is valid for the specified process.
-bool IsValidVirtualAddress(const Kernel::Process& process, VAddr vaddr);
-bool IsValidVirtualAddress(VAddr vaddr);
-/// Determines if the given VAddr is a kernel address
-bool IsKernelVirtualAddress(VAddr vaddr);
+ Memory(const Memory&) = delete;
+ Memory& operator=(const Memory&) = delete;
-u8 Read8(VAddr addr);
-u16 Read16(VAddr addr);
-u32 Read32(VAddr addr);
-u64 Read64(VAddr addr);
+ Memory(Memory&&) = default;
+ Memory& operator=(Memory&&) = default;
-void Write8(VAddr addr, u8 data);
-void Write16(VAddr addr, u16 data);
-void Write32(VAddr addr, u32 data);
-void Write64(VAddr addr, u64 data);
+ /**
+ * Changes the currently active page table to that of the given process instance.
+ *
+ * @param process The process to use the page table of.
+ */
+ void SetCurrentPageTable(Kernel::Process& process);
-void ReadBlock(const Kernel::Process& process, VAddr src_addr, void* dest_buffer, std::size_t size);
-void ReadBlock(VAddr src_addr, void* dest_buffer, std::size_t size);
-void WriteBlock(const Kernel::Process& process, VAddr dest_addr, const void* src_buffer,
- std::size_t size);
-void WriteBlock(VAddr dest_addr, const void* src_buffer, std::size_t size);
-void ZeroBlock(const Kernel::Process& process, VAddr dest_addr, std::size_t size);
-void CopyBlock(VAddr dest_addr, VAddr src_addr, std::size_t size);
+ /**
+ * Maps an allocated buffer onto a region of the emulated process address space.
+ *
+ * @param page_table The page table of the emulated process.
+ * @param base The address to start mapping at. Must be page-aligned.
+ * @param size The amount of bytes to map. Must be page-aligned.
+ * @param target Buffer with the memory backing the mapping. Must be of length at least
+ * `size`.
+ */
+ void MapMemoryRegion(Common::PageTable& page_table, VAddr base, u64 size, u8* target);
-u8* GetPointer(VAddr vaddr);
+ /**
+ * Maps a region of the emulated process address space as a IO region.
+ *
+ * @param page_table The page table of the emulated process.
+ * @param base The address to start mapping at. Must be page-aligned.
+ * @param size The amount of bytes to map. Must be page-aligned.
+ * @param mmio_handler The handler that backs the mapping.
+ */
+ void MapIoRegion(Common::PageTable& page_table, VAddr base, u64 size,
+ Common::MemoryHookPointer mmio_handler);
-std::string ReadCString(VAddr vaddr, std::size_t max_length);
+ /**
+ * Unmaps a region of the emulated process address space.
+ *
+ * @param page_table The page table of the emulated process.
+ * @param base The address to begin unmapping at.
+ * @param size The amount of bytes to unmap.
+ */
+ void UnmapRegion(Common::PageTable& page_table, VAddr base, u64 size);
-/**
- * Mark each page touching the region as cached.
- */
-void RasterizerMarkRegionCached(VAddr vaddr, u64 size, bool cached);
+ /**
+ * Adds a memory hook to intercept reads and writes to given region of memory.
+ *
+ * @param page_table The page table of the emulated process
+ * @param base The starting address to apply the hook to.
+ * @param size The size of the memory region to apply the hook to, in bytes.
+ * @param hook The hook to apply to the region of memory.
+ */
+ void AddDebugHook(Common::PageTable& page_table, VAddr base, u64 size,
+ Common::MemoryHookPointer hook);
+
+ /**
+ * Removes a memory hook from a given range of memory.
+ *
+ * @param page_table The page table of the emulated process.
+ * @param base The starting address to remove the hook from.
+ * @param size The size of the memory region to remove the hook from, in bytes.
+ * @param hook The hook to remove from the specified region of memory.
+ */
+ void RemoveDebugHook(Common::PageTable& page_table, VAddr base, u64 size,
+ Common::MemoryHookPointer hook);
+
+ /**
+ * Checks whether or not the supplied address is a valid virtual
+ * address for the given process.
+ *
+ * @param process The emulated process to check the address against.
+ * @param vaddr The virtual address to check the validity of.
+ *
+ * @returns True if the given virtual address is valid, false otherwise.
+ */
+ bool IsValidVirtualAddress(const Kernel::Process& process, VAddr vaddr) const;
+
+ /**
+ * Checks whether or not the supplied address is a valid virtual
+ * address for the current process.
+ *
+ * @param vaddr The virtual address to check the validity of.
+ *
+ * @returns True if the given virtual address is valid, false otherwise.
+ */
+ bool IsValidVirtualAddress(VAddr vaddr) const;
+
+ /**
+ * Gets a pointer to the given address.
+ *
+ * @param vaddr Virtual address to retrieve a pointer to.
+ *
+ * @returns The pointer to the given address, if the address is valid.
+ * If the address is not valid, nullptr will be returned.
+ */
+ u8* GetPointer(VAddr vaddr);
+
+ /**
+ * Gets a pointer to the given address.
+ *
+ * @param vaddr Virtual address to retrieve a pointer to.
+ *
+ * @returns The pointer to the given address, if the address is valid.
+ * If the address is not valid, nullptr will be returned.
+ */
+ const u8* GetPointer(VAddr vaddr) const;
+
+ /**
+ * Reads an 8-bit unsigned value from the current process' address space
+ * at the given virtual address.
+ *
+ * @param addr The virtual address to read the 8-bit value from.
+ *
+ * @returns the read 8-bit unsigned value.
+ */
+ u8 Read8(VAddr addr);
+
+ /**
+ * Reads a 16-bit unsigned value from the current process' address space
+ * at the given virtual address.
+ *
+ * @param addr The virtual address to read the 16-bit value from.
+ *
+ * @returns the read 16-bit unsigned value.
+ */
+ u16 Read16(VAddr addr);
+
+ /**
+ * Reads a 32-bit unsigned value from the current process' address space
+ * at the given virtual address.
+ *
+ * @param addr The virtual address to read the 32-bit value from.
+ *
+ * @returns the read 32-bit unsigned value.
+ */
+ u32 Read32(VAddr addr);
+
+ /**
+ * Reads a 64-bit unsigned value from the current process' address space
+ * at the given virtual address.
+ *
+ * @param addr The virtual address to read the 64-bit value from.
+ *
+ * @returns the read 64-bit value.
+ */
+ u64 Read64(VAddr addr);
+
+ /**
+ * Writes an 8-bit unsigned integer to the given virtual address in
+ * the current process' address space.
+ *
+ * @param addr The virtual address to write the 8-bit unsigned integer to.
+ * @param data The 8-bit unsigned integer to write to the given virtual address.
+ *
+ * @post The memory at the given virtual address contains the specified data value.
+ */
+ void Write8(VAddr addr, u8 data);
+
+ /**
+ * Writes a 16-bit unsigned integer to the given virtual address in
+ * the current process' address space.
+ *
+ * @param addr The virtual address to write the 16-bit unsigned integer to.
+ * @param data The 16-bit unsigned integer to write to the given virtual address.
+ *
+ * @post The memory range [addr, sizeof(data)) contains the given data value.
+ */
+ void Write16(VAddr addr, u16 data);
+
+ /**
+ * Writes a 32-bit unsigned integer to the given virtual address in
+ * the current process' address space.
+ *
+ * @param addr The virtual address to write the 32-bit unsigned integer to.
+ * @param data The 32-bit unsigned integer to write to the given virtual address.
+ *
+ * @post The memory range [addr, sizeof(data)) contains the given data value.
+ */
+ void Write32(VAddr addr, u32 data);
+
+ /**
+ * Writes a 64-bit unsigned integer to the given virtual address in
+ * the current process' address space.
+ *
+ * @param addr The virtual address to write the 64-bit unsigned integer to.
+ * @param data The 64-bit unsigned integer to write to the given virtual address.
+ *
+ * @post The memory range [addr, sizeof(data)) contains the given data value.
+ */
+ void Write64(VAddr addr, u64 data);
+
+ /**
+ * Reads a null-terminated string from the given virtual address.
+ * This function will continually read characters until either:
+ *
+ * - A null character ('\0') is reached.
+ * - max_length characters have been read.
+ *
+ * @note The final null-terminating character (if found) is not included
+ * in the returned string.
+ *
+ * @param vaddr The address to begin reading the string from.
+ * @param max_length The maximum length of the string to read in characters.
+ *
+ * @returns The read string.
+ */
+ std::string ReadCString(VAddr vaddr, std::size_t max_length);
+
+ /**
+ * Reads a contiguous block of bytes from a specified process' address space.
+ *
+ * @param process The process to read the data from.
+ * @param src_addr The virtual address to begin reading from.
+ * @param dest_buffer The buffer to place the read bytes into.
+ * @param size The amount of data to read, in bytes.
+ *
+ * @note If a size of 0 is specified, then this function reads nothing and
+ * no attempts to access memory are made at all.
+ *
+ * @pre dest_buffer must be at least size bytes in length, otherwise a
+ * buffer overrun will occur.
+ *
+ * @post The range [dest_buffer, size) contains the read bytes from the
+ * process' address space.
+ */
+ void ReadBlock(const Kernel::Process& process, VAddr src_addr, void* dest_buffer,
+ std::size_t size);
+
+ /**
+ * Reads a contiguous block of bytes from the current process' address space.
+ *
+ * @param src_addr The virtual address to begin reading from.
+ * @param dest_buffer The buffer to place the read bytes into.
+ * @param size The amount of data to read, in bytes.
+ *
+ * @note If a size of 0 is specified, then this function reads nothing and
+ * no attempts to access memory are made at all.
+ *
+ * @pre dest_buffer must be at least size bytes in length, otherwise a
+ * buffer overrun will occur.
+ *
+ * @post The range [dest_buffer, size) contains the read bytes from the
+ * current process' address space.
+ */
+ void ReadBlock(VAddr src_addr, void* dest_buffer, std::size_t size);
+
+ /**
+ * Writes a range of bytes into a given process' address space at the specified
+ * virtual address.
+ *
+ * @param process The process to write data into the address space of.
+ * @param dest_addr The destination virtual address to begin writing the data at.
+ * @param src_buffer The data to write into the process' address space.
+ * @param size The size of the data to write, in bytes.
+ *
+ * @post The address range [dest_addr, size) in the process' address space
+ * contains the data that was within src_buffer.
+ *
+ * @post If an attempt is made to write into an unmapped region of memory, the writes
+ * will be ignored and an error will be logged.
+ *
+ * @post If a write is performed into a region of memory that is considered cached
+ * rasterizer memory, will cause the currently active rasterizer to be notified
+ * and will mark that region as invalidated to caches that the active
+ * graphics backend may be maintaining over the course of execution.
+ */
+ void WriteBlock(const Kernel::Process& process, VAddr dest_addr, const void* src_buffer,
+ std::size_t size);
+
+ /**
+ * Writes a range of bytes into the current process' address space at the specified
+ * virtual address.
+ *
+ * @param dest_addr The destination virtual address to begin writing the data at.
+ * @param src_buffer The data to write into the current process' address space.
+ * @param size The size of the data to write, in bytes.
+ *
+ * @post The address range [dest_addr, size) in the current process' address space
+ * contains the data that was within src_buffer.
+ *
+ * @post If an attempt is made to write into an unmapped region of memory, the writes
+ * will be ignored and an error will be logged.
+ *
+ * @post If a write is performed into a region of memory that is considered cached
+ * rasterizer memory, will cause the currently active rasterizer to be notified
+ * and will mark that region as invalidated to caches that the active
+ * graphics backend may be maintaining over the course of execution.
+ */
+ void WriteBlock(VAddr dest_addr, const void* src_buffer, std::size_t size);
+
+ /**
+ * Fills the specified address range within a process' address space with zeroes.
+ *
+ * @param process The process that will have a portion of its memory zeroed out.
+ * @param dest_addr The starting virtual address of the range to zero out.
+ * @param size The size of the address range to zero out, in bytes.
+ *
+ * @post The range [dest_addr, size) within the process' address space is
+ * filled with zeroes.
+ */
+ void ZeroBlock(const Kernel::Process& process, VAddr dest_addr, std::size_t size);
+
+ /**
+ * Fills the specified address range within the current process' address space with zeroes.
+ *
+ * @param dest_addr The starting virtual address of the range to zero out.
+ * @param size The size of the address range to zero out, in bytes.
+ *
+ * @post The range [dest_addr, size) within the current process' address space is
+ * filled with zeroes.
+ */
+ void ZeroBlock(VAddr dest_addr, std::size_t size);
+
+ /**
+ * Copies data within a process' address space to another location within the
+ * same address space.
+ *
+ * @param process The process that will have data copied within its address space.
+ * @param dest_addr The destination virtual address to begin copying the data into.
+ * @param src_addr The source virtual address to begin copying the data from.
+ * @param size The size of the data to copy, in bytes.
+ *
+ * @post The range [dest_addr, size) within the process' address space contains the
+ * same data within the range [src_addr, size).
+ */
+ void CopyBlock(const Kernel::Process& process, VAddr dest_addr, VAddr src_addr,
+ std::size_t size);
+
+ /**
+ * Copies data within the current process' address space to another location within the
+ * same address space.
+ *
+ * @param dest_addr The destination virtual address to begin copying the data into.
+ * @param src_addr The source virtual address to begin copying the data from.
+ * @param size The size of the data to copy, in bytes.
+ *
+ * @post The range [dest_addr, size) within the current process' address space
+ * contains the same data within the range [src_addr, size).
+ */
+ void CopyBlock(VAddr dest_addr, VAddr src_addr, std::size_t size);
+
+ /**
+ * Marks each page within the specified address range as cached or uncached.
+ *
+ * @param vaddr The virtual address indicating the start of the address range.
+ * @param size The size of the address range in bytes.
+ * @param cached Whether or not any pages within the address range should be
+ * marked as cached or uncached.
+ */
+ void RasterizerMarkRegionCached(VAddr vaddr, u64 size, bool cached);
+
+private:
+ struct Impl;
+ std::unique_ptr<Impl> impl;
+};
+
+/// Determines if the given VAddr is a kernel address
+bool IsKernelVirtualAddress(VAddr vaddr);
} // namespace Memory
diff --git a/src/core/memory/cheat_engine.cpp b/src/core/memory/cheat_engine.cpp
index 10821d452..d1e6bed93 100644
--- a/src/core/memory/cheat_engine.cpp
+++ b/src/core/memory/cheat_engine.cpp
@@ -20,18 +20,17 @@ namespace Memory {
constexpr s64 CHEAT_ENGINE_TICKS = static_cast<s64>(Core::Timing::BASE_CLOCK_RATE / 12);
constexpr u32 KEYPAD_BITMASK = 0x3FFFFFF;
-StandardVmCallbacks::StandardVmCallbacks(const Core::System& system,
- const CheatProcessMetadata& metadata)
+StandardVmCallbacks::StandardVmCallbacks(Core::System& system, const CheatProcessMetadata& metadata)
: metadata(metadata), system(system) {}
StandardVmCallbacks::~StandardVmCallbacks() = default;
void StandardVmCallbacks::MemoryRead(VAddr address, void* data, u64 size) {
- ReadBlock(SanitizeAddress(address), data, size);
+ system.Memory().ReadBlock(SanitizeAddress(address), data, size);
}
void StandardVmCallbacks::MemoryWrite(VAddr address, const void* data, u64 size) {
- WriteBlock(SanitizeAddress(address), data, size);
+ system.Memory().WriteBlock(SanitizeAddress(address), data, size);
}
u64 StandardVmCallbacks::HidKeysDown() {
@@ -186,7 +185,7 @@ CheatEngine::~CheatEngine() {
}
void CheatEngine::Initialize() {
- event = core_timing.RegisterEvent(
+ event = Core::Timing::CreateEvent(
"CheatEngine::FrameCallback::" + Common::HexToString(metadata.main_nso_build_id),
[this](u64 userdata, s64 cycles_late) { FrameCallback(userdata, cycles_late); });
core_timing.ScheduleEvent(CHEAT_ENGINE_TICKS, event);
diff --git a/src/core/memory/cheat_engine.h b/src/core/memory/cheat_engine.h
index 0f012e9b5..3d6b2298a 100644
--- a/src/core/memory/cheat_engine.h
+++ b/src/core/memory/cheat_engine.h
@@ -5,6 +5,7 @@
#pragma once
#include <atomic>
+#include <memory>
#include <vector>
#include "common/common_types.h"
#include "core/memory/dmnt_cheat_types.h"
@@ -23,7 +24,7 @@ namespace Memory {
class StandardVmCallbacks : public DmntCheatVm::Callbacks {
public:
- StandardVmCallbacks(const Core::System& system, const CheatProcessMetadata& metadata);
+ StandardVmCallbacks(Core::System& system, const CheatProcessMetadata& metadata);
~StandardVmCallbacks() override;
void MemoryRead(VAddr address, void* data, u64 size) override;
@@ -36,7 +37,7 @@ private:
VAddr SanitizeAddress(VAddr address) const;
const CheatProcessMetadata& metadata;
- const Core::System& system;
+ Core::System& system;
};
// Intermediary class that parses a text file or other disk format for storing cheats into a
@@ -78,7 +79,7 @@ private:
std::vector<CheatEntry> cheats;
std::atomic_bool is_pending_reload{false};
- Core::Timing::EventType* event{};
+ std::shared_ptr<Core::Timing::EventType> event;
Core::Timing::CoreTiming& core_timing;
Core::System& system;
};
diff --git a/src/core/memory_setup.h b/src/core/memory_setup.h
deleted file mode 100644
index 5225ee8e2..000000000
--- a/src/core/memory_setup.h
+++ /dev/null
@@ -1,43 +0,0 @@
-// Copyright 2015 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include "common/common_types.h"
-#include "common/memory_hook.h"
-
-namespace Common {
-struct PageTable;
-}
-
-namespace Memory {
-
-/**
- * Maps an allocated buffer onto a region of the emulated process address space.
- *
- * @param page_table The page table of the emulated process.
- * @param base The address to start mapping at. Must be page-aligned.
- * @param size The amount of bytes to map. Must be page-aligned.
- * @param target Buffer with the memory backing the mapping. Must be of length at least `size`.
- */
-void MapMemoryRegion(Common::PageTable& page_table, VAddr base, u64 size, u8* target);
-
-/**
- * Maps a region of the emulated process address space as a IO region.
- * @param page_table The page table of the emulated process.
- * @param base The address to start mapping at. Must be page-aligned.
- * @param size The amount of bytes to map. Must be page-aligned.
- * @param mmio_handler The handler that backs the mapping.
- */
-void MapIoRegion(Common::PageTable& page_table, VAddr base, u64 size,
- Common::MemoryHookPointer mmio_handler);
-
-void UnmapRegion(Common::PageTable& page_table, VAddr base, u64 size);
-
-void AddDebugHook(Common::PageTable& page_table, VAddr base, u64 size,
- Common::MemoryHookPointer hook);
-void RemoveDebugHook(Common::PageTable& page_table, VAddr base, u64 size,
- Common::MemoryHookPointer hook);
-
-} // namespace Memory
diff --git a/src/core/perf_stats.cpp b/src/core/perf_stats.cpp
index d2c69d1a0..f1ae9d4df 100644
--- a/src/core/perf_stats.cpp
+++ b/src/core/perf_stats.cpp
@@ -81,7 +81,7 @@ double PerfStats::GetMeanFrametime() {
return 0;
}
const double sum = std::accumulate(perf_history.begin() + IgnoreFrames,
- perf_history.begin() + current_index, 0);
+ perf_history.begin() + current_index, 0.0);
return sum / (current_index - IgnoreFrames);
}
diff --git a/src/core/reporter.cpp b/src/core/reporter.cpp
index 6f4af77fd..f95eee3b1 100644
--- a/src/core/reporter.cpp
+++ b/src/core/reporter.cpp
@@ -147,7 +147,7 @@ json GetFullDataAuto(const std::string& timestamp, u64 title_id, Core::System& s
}
template <bool read_value, typename DescriptorType>
-json GetHLEBufferDescriptorData(const std::vector<DescriptorType>& buffer) {
+json GetHLEBufferDescriptorData(const std::vector<DescriptorType>& buffer, Memory::Memory& memory) {
auto buffer_out = json::array();
for (const auto& desc : buffer) {
auto entry = json{
@@ -157,7 +157,7 @@ json GetHLEBufferDescriptorData(const std::vector<DescriptorType>& buffer) {
if constexpr (read_value) {
std::vector<u8> data(desc.Size());
- Memory::ReadBlock(desc.Address(), data.data(), desc.Size());
+ memory.ReadBlock(desc.Address(), data.data(), desc.Size());
entry["data"] = Common::HexToString(data);
}
@@ -167,7 +167,7 @@ json GetHLEBufferDescriptorData(const std::vector<DescriptorType>& buffer) {
return buffer_out;
}
-json GetHLERequestContextData(Kernel::HLERequestContext& ctx) {
+json GetHLERequestContextData(Kernel::HLERequestContext& ctx, Memory::Memory& memory) {
json out;
auto cmd_buf = json::array();
@@ -177,10 +177,10 @@ json GetHLERequestContextData(Kernel::HLERequestContext& ctx) {
out["command_buffer"] = std::move(cmd_buf);
- out["buffer_descriptor_a"] = GetHLEBufferDescriptorData<true>(ctx.BufferDescriptorA());
- out["buffer_descriptor_b"] = GetHLEBufferDescriptorData<false>(ctx.BufferDescriptorB());
- out["buffer_descriptor_c"] = GetHLEBufferDescriptorData<false>(ctx.BufferDescriptorC());
- out["buffer_descriptor_x"] = GetHLEBufferDescriptorData<true>(ctx.BufferDescriptorX());
+ out["buffer_descriptor_a"] = GetHLEBufferDescriptorData<true>(ctx.BufferDescriptorA(), memory);
+ out["buffer_descriptor_b"] = GetHLEBufferDescriptorData<false>(ctx.BufferDescriptorB(), memory);
+ out["buffer_descriptor_c"] = GetHLEBufferDescriptorData<false>(ctx.BufferDescriptorC(), memory);
+ out["buffer_descriptor_x"] = GetHLEBufferDescriptorData<true>(ctx.BufferDescriptorX(), memory);
return out;
}
@@ -259,7 +259,7 @@ void Reporter::SaveUnimplementedFunctionReport(Kernel::HLERequestContext& ctx, u
const auto title_id = system.CurrentProcess()->GetTitleID();
auto out = GetFullDataAuto(timestamp, title_id, system);
- auto function_out = GetHLERequestContextData(ctx);
+ auto function_out = GetHLERequestContextData(ctx, system.Memory());
function_out["command_id"] = command_id;
function_out["function_name"] = name;
function_out["service_name"] = service_name;
diff --git a/src/core/telemetry_session.cpp b/src/core/telemetry_session.cpp
index 793d102d3..320e8ad73 100644
--- a/src/core/telemetry_session.cpp
+++ b/src/core/telemetry_session.cpp
@@ -165,24 +165,20 @@ void TelemetrySession::AddInitialInfo(Loader::AppLoader& app_loader) {
Telemetry::AppendOSInfo(field_collection);
// Log user configuration information
- AddField(Telemetry::FieldType::UserConfig, "Audio_SinkId", Settings::values.sink_id);
- AddField(Telemetry::FieldType::UserConfig, "Audio_EnableAudioStretching",
- Settings::values.enable_audio_stretching);
- AddField(Telemetry::FieldType::UserConfig, "Core_UseMultiCore",
- Settings::values.use_multi_core);
- AddField(Telemetry::FieldType::UserConfig, "Renderer_ResolutionFactor",
- Settings::values.resolution_factor);
- AddField(Telemetry::FieldType::UserConfig, "Renderer_UseFrameLimit",
- Settings::values.use_frame_limit);
- AddField(Telemetry::FieldType::UserConfig, "Renderer_FrameLimit", Settings::values.frame_limit);
- AddField(Telemetry::FieldType::UserConfig, "Renderer_UseDiskShaderCache",
- Settings::values.use_disk_shader_cache);
- AddField(Telemetry::FieldType::UserConfig, "Renderer_UseAccurateGpuEmulation",
+ constexpr auto field_type = Telemetry::FieldType::UserConfig;
+ AddField(field_type, "Audio_SinkId", Settings::values.sink_id);
+ AddField(field_type, "Audio_EnableAudioStretching", Settings::values.enable_audio_stretching);
+ AddField(field_type, "Core_UseMultiCore", Settings::values.use_multi_core);
+ AddField(field_type, "Renderer_Backend", "OpenGL");
+ AddField(field_type, "Renderer_ResolutionFactor", Settings::values.resolution_factor);
+ AddField(field_type, "Renderer_UseFrameLimit", Settings::values.use_frame_limit);
+ AddField(field_type, "Renderer_FrameLimit", Settings::values.frame_limit);
+ AddField(field_type, "Renderer_UseDiskShaderCache", Settings::values.use_disk_shader_cache);
+ AddField(field_type, "Renderer_UseAccurateGpuEmulation",
Settings::values.use_accurate_gpu_emulation);
- AddField(Telemetry::FieldType::UserConfig, "Renderer_UseAsynchronousGpuEmulation",
+ AddField(field_type, "Renderer_UseAsynchronousGpuEmulation",
Settings::values.use_asynchronous_gpu_emulation);
- AddField(Telemetry::FieldType::UserConfig, "System_UseDockedMode",
- Settings::values.use_docked_mode);
+ AddField(field_type, "System_UseDockedMode", Settings::values.use_docked_mode);
}
bool TelemetrySession::SubmitTestcase() {
diff --git a/src/core/tools/freezer.cpp b/src/core/tools/freezer.cpp
index 17f050068..55e0dbc49 100644
--- a/src/core/tools/freezer.cpp
+++ b/src/core/tools/freezer.cpp
@@ -11,40 +11,39 @@
#include "core/tools/freezer.h"
namespace Tools {
-
namespace {
constexpr s64 MEMORY_FREEZER_TICKS = static_cast<s64>(Core::Timing::BASE_CLOCK_RATE / 60);
-u64 MemoryReadWidth(u32 width, VAddr addr) {
+u64 MemoryReadWidth(Memory::Memory& memory, u32 width, VAddr addr) {
switch (width) {
case 1:
- return Memory::Read8(addr);
+ return memory.Read8(addr);
case 2:
- return Memory::Read16(addr);
+ return memory.Read16(addr);
case 4:
- return Memory::Read32(addr);
+ return memory.Read32(addr);
case 8:
- return Memory::Read64(addr);
+ return memory.Read64(addr);
default:
UNREACHABLE();
return 0;
}
}
-void MemoryWriteWidth(u32 width, VAddr addr, u64 value) {
+void MemoryWriteWidth(Memory::Memory& memory, u32 width, VAddr addr, u64 value) {
switch (width) {
case 1:
- Memory::Write8(addr, static_cast<u8>(value));
+ memory.Write8(addr, static_cast<u8>(value));
break;
case 2:
- Memory::Write16(addr, static_cast<u16>(value));
+ memory.Write16(addr, static_cast<u16>(value));
break;
case 4:
- Memory::Write32(addr, static_cast<u32>(value));
+ memory.Write32(addr, static_cast<u32>(value));
break;
case 8:
- Memory::Write64(addr, value);
+ memory.Write64(addr, value);
break;
default:
UNREACHABLE();
@@ -53,8 +52,9 @@ void MemoryWriteWidth(u32 width, VAddr addr, u64 value) {
} // Anonymous namespace
-Freezer::Freezer(Core::Timing::CoreTiming& core_timing) : core_timing(core_timing) {
- event = core_timing.RegisterEvent(
+Freezer::Freezer(Core::Timing::CoreTiming& core_timing_, Memory::Memory& memory_)
+ : core_timing{core_timing_}, memory{memory_} {
+ event = Core::Timing::CreateEvent(
"MemoryFreezer::FrameCallback",
[this](u64 userdata, s64 cycles_late) { FrameCallback(userdata, cycles_late); });
core_timing.ScheduleEvent(MEMORY_FREEZER_TICKS, event);
@@ -89,7 +89,7 @@ void Freezer::Clear() {
u64 Freezer::Freeze(VAddr address, u32 width) {
std::lock_guard lock{entries_mutex};
- const auto current_value = MemoryReadWidth(width, address);
+ const auto current_value = MemoryReadWidth(memory, width, address);
entries.push_back({address, width, current_value});
LOG_DEBUG(Common_Memory,
@@ -169,7 +169,7 @@ void Freezer::FrameCallback(u64 userdata, s64 cycles_late) {
LOG_DEBUG(Common_Memory,
"Enforcing memory freeze at address={:016X}, value={:016X}, width={:02X}",
entry.address, entry.value, entry.width);
- MemoryWriteWidth(entry.width, entry.address, entry.value);
+ MemoryWriteWidth(memory, entry.width, entry.address, entry.value);
}
core_timing.ScheduleEvent(MEMORY_FREEZER_TICKS - cycles_late, event);
@@ -181,7 +181,7 @@ void Freezer::FillEntryReads() {
LOG_DEBUG(Common_Memory, "Updating memory freeze entries to current values.");
for (auto& entry : entries) {
- entry.value = MemoryReadWidth(entry.width, entry.address);
+ entry.value = MemoryReadWidth(memory, entry.width, entry.address);
}
}
diff --git a/src/core/tools/freezer.h b/src/core/tools/freezer.h
index b58de5472..916339c6c 100644
--- a/src/core/tools/freezer.h
+++ b/src/core/tools/freezer.h
@@ -5,6 +5,7 @@
#pragma once
#include <atomic>
+#include <memory>
#include <mutex>
#include <optional>
#include <vector>
@@ -15,6 +16,10 @@ class CoreTiming;
struct EventType;
} // namespace Core::Timing
+namespace Memory {
+class Memory;
+}
+
namespace Tools {
/**
@@ -33,7 +38,7 @@ public:
u64 value;
};
- explicit Freezer(Core::Timing::CoreTiming& core_timing);
+ explicit Freezer(Core::Timing::CoreTiming& core_timing_, Memory::Memory& memory_);
~Freezer();
// Enables or disables the entire memory freezer.
@@ -75,8 +80,9 @@ private:
mutable std::mutex entries_mutex;
std::vector<Entry> entries;
- Core::Timing::EventType* event;
+ std::shared_ptr<Core::Timing::EventType> event;
Core::Timing::CoreTiming& core_timing;
+ Memory::Memory& memory;
};
} // namespace Tools
diff --git a/src/tests/core/arm/arm_test_common.cpp b/src/tests/core/arm/arm_test_common.cpp
index ac7ae3e52..17043346b 100644
--- a/src/tests/core/arm/arm_test_common.cpp
+++ b/src/tests/core/arm/arm_test_common.cpp
@@ -8,7 +8,6 @@
#include "core/core.h"
#include "core/hle/kernel/process.h"
#include "core/memory.h"
-#include "core/memory_setup.h"
#include "tests/core/arm/arm_test_common.h"
namespace ArmTests {
@@ -16,8 +15,9 @@ namespace ArmTests {
TestEnvironment::TestEnvironment(bool mutable_memory_)
: mutable_memory(mutable_memory_),
test_memory(std::make_shared<TestMemory>(this)), kernel{Core::System::GetInstance()} {
- auto process = Kernel::Process::Create(Core::System::GetInstance(), "",
- Kernel::Process::ProcessType::Userland);
+ auto& system = Core::System::GetInstance();
+
+ auto process = Kernel::Process::Create(system, "", Kernel::Process::ProcessType::Userland);
page_table = &process->VMManager().page_table;
std::fill(page_table->pointers.begin(), page_table->pointers.end(), nullptr);
@@ -25,15 +25,16 @@ TestEnvironment::TestEnvironment(bool mutable_memory_)
std::fill(page_table->attributes.begin(), page_table->attributes.end(),
Common::PageType::Unmapped);
- Memory::MapIoRegion(*page_table, 0x00000000, 0x80000000, test_memory);
- Memory::MapIoRegion(*page_table, 0x80000000, 0x80000000, test_memory);
+ system.Memory().MapIoRegion(*page_table, 0x00000000, 0x80000000, test_memory);
+ system.Memory().MapIoRegion(*page_table, 0x80000000, 0x80000000, test_memory);
kernel.MakeCurrentProcess(process.get());
}
TestEnvironment::~TestEnvironment() {
- Memory::UnmapRegion(*page_table, 0x80000000, 0x80000000);
- Memory::UnmapRegion(*page_table, 0x00000000, 0x80000000);
+ auto& system = Core::System::GetInstance();
+ system.Memory().UnmapRegion(*page_table, 0x80000000, 0x80000000);
+ system.Memory().UnmapRegion(*page_table, 0x00000000, 0x80000000);
}
void TestEnvironment::SetMemory64(VAddr vaddr, u64 value) {
diff --git a/src/tests/core/core_timing.cpp b/src/tests/core/core_timing.cpp
index 3443bf05e..1e3940801 100644
--- a/src/tests/core/core_timing.cpp
+++ b/src/tests/core/core_timing.cpp
@@ -7,7 +7,9 @@
#include <array>
#include <bitset>
#include <cstdlib>
+#include <memory>
#include <string>
+
#include "common/file_util.h"
#include "core/core.h"
#include "core/core_timing.h"
@@ -65,11 +67,16 @@ TEST_CASE("CoreTiming[BasicOrder]", "[core]") {
ScopeInit guard;
auto& core_timing = guard.core_timing;
- Core::Timing::EventType* cb_a = core_timing.RegisterEvent("callbackA", CallbackTemplate<0>);
- Core::Timing::EventType* cb_b = core_timing.RegisterEvent("callbackB", CallbackTemplate<1>);
- Core::Timing::EventType* cb_c = core_timing.RegisterEvent("callbackC", CallbackTemplate<2>);
- Core::Timing::EventType* cb_d = core_timing.RegisterEvent("callbackD", CallbackTemplate<3>);
- Core::Timing::EventType* cb_e = core_timing.RegisterEvent("callbackE", CallbackTemplate<4>);
+ std::shared_ptr<Core::Timing::EventType> cb_a =
+ Core::Timing::CreateEvent("callbackA", CallbackTemplate<0>);
+ std::shared_ptr<Core::Timing::EventType> cb_b =
+ Core::Timing::CreateEvent("callbackB", CallbackTemplate<1>);
+ std::shared_ptr<Core::Timing::EventType> cb_c =
+ Core::Timing::CreateEvent("callbackC", CallbackTemplate<2>);
+ std::shared_ptr<Core::Timing::EventType> cb_d =
+ Core::Timing::CreateEvent("callbackD", CallbackTemplate<3>);
+ std::shared_ptr<Core::Timing::EventType> cb_e =
+ Core::Timing::CreateEvent("callbackE", CallbackTemplate<4>);
// Enter slice 0
core_timing.ResetRun();
@@ -99,8 +106,8 @@ TEST_CASE("CoreTiming[FairSharing]", "[core]") {
ScopeInit guard;
auto& core_timing = guard.core_timing;
- Core::Timing::EventType* empty_callback =
- core_timing.RegisterEvent("empty_callback", EmptyCallback);
+ std::shared_ptr<Core::Timing::EventType> empty_callback =
+ Core::Timing::CreateEvent("empty_callback", EmptyCallback);
callbacks_done = 0;
u64 MAX_CALLBACKS = 10;
@@ -133,8 +140,10 @@ TEST_CASE("Core::Timing[PredictableLateness]", "[core]") {
ScopeInit guard;
auto& core_timing = guard.core_timing;
- Core::Timing::EventType* cb_a = core_timing.RegisterEvent("callbackA", CallbackTemplate<0>);
- Core::Timing::EventType* cb_b = core_timing.RegisterEvent("callbackB", CallbackTemplate<1>);
+ std::shared_ptr<Core::Timing::EventType> cb_a =
+ Core::Timing::CreateEvent("callbackA", CallbackTemplate<0>);
+ std::shared_ptr<Core::Timing::EventType> cb_b =
+ Core::Timing::CreateEvent("callbackB", CallbackTemplate<1>);
// Enter slice 0
core_timing.ResetRun();
@@ -145,60 +154,3 @@ TEST_CASE("Core::Timing[PredictableLateness]", "[core]") {
AdvanceAndCheck(core_timing, 0, 0, 10, -10); // (100 - 10)
AdvanceAndCheck(core_timing, 1, 1, 50, -50);
}
-
-namespace ChainSchedulingTest {
-static int reschedules = 0;
-
-static void RescheduleCallback(Core::Timing::CoreTiming& core_timing, u64 userdata,
- s64 cycles_late) {
- --reschedules;
- REQUIRE(reschedules >= 0);
- REQUIRE(lateness == cycles_late);
-
- if (reschedules > 0) {
- core_timing.ScheduleEvent(1000, reinterpret_cast<Core::Timing::EventType*>(userdata),
- userdata);
- }
-}
-} // namespace ChainSchedulingTest
-
-TEST_CASE("CoreTiming[ChainScheduling]", "[core]") {
- using namespace ChainSchedulingTest;
-
- ScopeInit guard;
- auto& core_timing = guard.core_timing;
-
- Core::Timing::EventType* cb_a = core_timing.RegisterEvent("callbackA", CallbackTemplate<0>);
- Core::Timing::EventType* cb_b = core_timing.RegisterEvent("callbackB", CallbackTemplate<1>);
- Core::Timing::EventType* cb_c = core_timing.RegisterEvent("callbackC", CallbackTemplate<2>);
- Core::Timing::EventType* cb_rs = core_timing.RegisterEvent(
- "callbackReschedule", [&core_timing](u64 userdata, s64 cycles_late) {
- RescheduleCallback(core_timing, userdata, cycles_late);
- });
-
- // Enter slice 0
- core_timing.ResetRun();
-
- core_timing.ScheduleEvent(800, cb_a, CB_IDS[0]);
- core_timing.ScheduleEvent(1000, cb_b, CB_IDS[1]);
- core_timing.ScheduleEvent(2200, cb_c, CB_IDS[2]);
- core_timing.ScheduleEvent(1000, cb_rs, reinterpret_cast<u64>(cb_rs));
- REQUIRE(800 == core_timing.GetDowncount());
-
- reschedules = 3;
- AdvanceAndCheck(core_timing, 0, 0); // cb_a
- AdvanceAndCheck(core_timing, 1, 1); // cb_b, cb_rs
- REQUIRE(2 == reschedules);
-
- core_timing.AddTicks(core_timing.GetDowncount());
- core_timing.Advance(); // cb_rs
- core_timing.SwitchContext(3);
- REQUIRE(1 == reschedules);
- REQUIRE(200 == core_timing.GetDowncount());
-
- AdvanceAndCheck(core_timing, 2, 3); // cb_c
-
- core_timing.AddTicks(core_timing.GetDowncount());
- core_timing.Advance(); // cb_rs
- REQUIRE(0 == reschedules);
-}
diff --git a/src/video_core/CMakeLists.txt b/src/video_core/CMakeLists.txt
index c911c6ec4..65d7b9f93 100644
--- a/src/video_core/CMakeLists.txt
+++ b/src/video_core/CMakeLists.txt
@@ -4,8 +4,6 @@ add_library(video_core STATIC
buffer_cache/map_interval.h
dma_pusher.cpp
dma_pusher.h
- debug_utils/debug_utils.cpp
- debug_utils/debug_utils.h
engines/const_buffer_engine_interface.h
engines/const_buffer_info.h
engines/engine_upload.cpp
@@ -22,6 +20,7 @@ add_library(video_core STATIC
engines/maxwell_dma.h
engines/shader_bytecode.h
engines/shader_header.h
+ engines/shader_type.h
gpu.cpp
gpu.h
gpu_asynch.cpp
@@ -127,6 +126,8 @@ add_library(video_core STATIC
shader/track.cpp
surface.cpp
surface.h
+ texture_cache/format_lookup_table.cpp
+ texture_cache/format_lookup_table.h
texture_cache/surface_base.cpp
texture_cache/surface_base.h
texture_cache/surface_params.cpp
@@ -148,12 +149,16 @@ add_library(video_core STATIC
if (ENABLE_VULKAN)
target_sources(video_core PRIVATE
renderer_vulkan/declarations.h
+ renderer_vulkan/fixed_pipeline_state.cpp
+ renderer_vulkan/fixed_pipeline_state.h
renderer_vulkan/maxwell_to_vk.cpp
renderer_vulkan/maxwell_to_vk.h
renderer_vulkan/vk_buffer_cache.cpp
renderer_vulkan/vk_buffer_cache.h
renderer_vulkan/vk_device.cpp
renderer_vulkan/vk_device.h
+ renderer_vulkan/vk_image.cpp
+ renderer_vulkan/vk_image.h
renderer_vulkan/vk_memory_manager.cpp
renderer_vulkan/vk_memory_manager.h
renderer_vulkan/vk_resource_manager.cpp
@@ -164,6 +169,8 @@ if (ENABLE_VULKAN)
renderer_vulkan/vk_scheduler.h
renderer_vulkan/vk_shader_decompiler.cpp
renderer_vulkan/vk_shader_decompiler.h
+ renderer_vulkan/vk_staging_buffer_pool.cpp
+ renderer_vulkan/vk_staging_buffer_pool.h
renderer_vulkan/vk_stream_buffer.cpp
renderer_vulkan/vk_stream_buffer.h
renderer_vulkan/vk_swapchain.cpp
@@ -180,3 +187,9 @@ target_link_libraries(video_core PRIVATE glad)
if (ENABLE_VULKAN)
target_link_libraries(video_core PRIVATE sirit)
endif()
+
+if (MSVC)
+ target_compile_options(video_core PRIVATE /we4267)
+else()
+ target_compile_options(video_core PRIVATE -Werror=conversion -Wno-error=sign-conversion)
+endif()
diff --git a/src/video_core/buffer_cache/buffer_cache.h b/src/video_core/buffer_cache/buffer_cache.h
index 63b3a8205..0510ed777 100644
--- a/src/video_core/buffer_cache/buffer_cache.h
+++ b/src/video_core/buffer_cache/buffer_cache.h
@@ -12,6 +12,10 @@
#include <utility>
#include <vector>
+#include <boost/icl/interval_map.hpp>
+#include <boost/icl/interval_set.hpp>
+#include <boost/range/iterator_range.hpp>
+
#include "common/alignment.h"
#include "common/common_types.h"
#include "core/core.h"
@@ -423,8 +427,8 @@ private:
VideoCore::RasterizerInterface& rasterizer;
Core::System& system;
- std::unique_ptr<StreamBuffer> stream_buffer;
+ std::unique_ptr<StreamBuffer> stream_buffer;
TBufferType stream_buffer_handle{};
bool invalidated = false;
@@ -436,18 +440,18 @@ private:
using IntervalSet = boost::icl::interval_set<CacheAddr>;
using IntervalCache = boost::icl::interval_map<CacheAddr, MapInterval>;
using IntervalType = typename IntervalCache::interval_type;
- IntervalCache mapped_addresses{};
+ IntervalCache mapped_addresses;
- static constexpr u64 write_page_bit{11};
- std::unordered_map<u64, u32> written_pages{};
+ static constexpr u64 write_page_bit = 11;
+ std::unordered_map<u64, u32> written_pages;
- static constexpr u64 block_page_bits{21};
- static constexpr u64 block_page_size{1 << block_page_bits};
- std::unordered_map<u64, TBuffer> blocks{};
+ static constexpr u64 block_page_bits = 21;
+ static constexpr u64 block_page_size = 1ULL << block_page_bits;
+ std::unordered_map<u64, TBuffer> blocks;
- std::list<TBuffer> pending_destruction{};
- u64 epoch{};
- u64 modified_ticks{};
+ std::list<TBuffer> pending_destruction;
+ u64 epoch = 0;
+ u64 modified_ticks = 0;
std::recursive_mutex mutex;
};
diff --git a/src/video_core/debug_utils/debug_utils.cpp b/src/video_core/debug_utils/debug_utils.cpp
deleted file mode 100644
index f0ef67535..000000000
--- a/src/video_core/debug_utils/debug_utils.cpp
+++ /dev/null
@@ -1,49 +0,0 @@
-// Copyright 2014 Citra Emulator Project
-// Licensed under GPLv2
-// Refer to the license.txt file included.
-
-#include <mutex>
-
-#include "video_core/debug_utils/debug_utils.h"
-
-namespace Tegra {
-
-void DebugContext::DoOnEvent(Event event, void* data) {
- {
- std::unique_lock lock{breakpoint_mutex};
-
- // TODO(Subv): Commit the rasterizer's caches so framebuffers, render targets, etc. will
- // show on debug widgets
-
- // TODO: Should stop the CPU thread here once we multithread emulation.
-
- active_breakpoint = event;
- at_breakpoint = true;
-
- // Tell all observers that we hit a breakpoint
- for (auto& breakpoint_observer : breakpoint_observers) {
- breakpoint_observer->OnMaxwellBreakPointHit(event, data);
- }
-
- // Wait until another thread tells us to Resume()
- resume_from_breakpoint.wait(lock, [&] { return !at_breakpoint; });
- }
-}
-
-void DebugContext::Resume() {
- {
- std::lock_guard lock{breakpoint_mutex};
-
- // Tell all observers that we are about to resume
- for (auto& breakpoint_observer : breakpoint_observers) {
- breakpoint_observer->OnMaxwellResume();
- }
-
- // Resume the waiting thread (i.e. OnEvent())
- at_breakpoint = false;
- }
-
- resume_from_breakpoint.notify_one();
-}
-
-} // namespace Tegra
diff --git a/src/video_core/debug_utils/debug_utils.h b/src/video_core/debug_utils/debug_utils.h
deleted file mode 100644
index ac3a2eb01..000000000
--- a/src/video_core/debug_utils/debug_utils.h
+++ /dev/null
@@ -1,157 +0,0 @@
-// Copyright 2014 Citra Emulator Project
-// Licensed under GPLv2
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include <array>
-#include <condition_variable>
-#include <list>
-#include <memory>
-#include <mutex>
-
-namespace Tegra {
-
-class DebugContext {
-public:
- enum class Event {
- FirstEvent = 0,
-
- MaxwellCommandLoaded = FirstEvent,
- MaxwellCommandProcessed,
- IncomingPrimitiveBatch,
- FinishedPrimitiveBatch,
-
- NumEvents
- };
-
- /**
- * Inherit from this class to be notified of events registered to some debug context.
- * Most importantly this is used for our debugger GUI.
- *
- * To implement event handling, override the OnMaxwellBreakPointHit and OnMaxwellResume methods.
- * @warning All BreakPointObservers need to be on the same thread to guarantee thread-safe state
- * access
- * @todo Evaluate an alternative interface, in which there is only one managing observer and
- * multiple child observers running (by design) on the same thread.
- */
- class BreakPointObserver {
- public:
- /// Constructs the object such that it observes events of the given DebugContext.
- explicit BreakPointObserver(std::shared_ptr<DebugContext> debug_context)
- : context_weak(debug_context) {
- std::unique_lock lock{debug_context->breakpoint_mutex};
- debug_context->breakpoint_observers.push_back(this);
- }
-
- virtual ~BreakPointObserver() {
- auto context = context_weak.lock();
- if (context) {
- {
- std::unique_lock lock{context->breakpoint_mutex};
- context->breakpoint_observers.remove(this);
- }
-
- // If we are the last observer to be destroyed, tell the debugger context that
- // it is free to continue. In particular, this is required for a proper yuzu
- // shutdown, when the emulation thread is waiting at a breakpoint.
- if (context->breakpoint_observers.empty())
- context->Resume();
- }
- }
-
- /**
- * Action to perform when a breakpoint was reached.
- * @param event Type of event which triggered the breakpoint
- * @param data Optional data pointer (if unused, this is a nullptr)
- * @note This function will perform nothing unless it is overridden in the child class.
- */
- virtual void OnMaxwellBreakPointHit(Event event, void* data) {}
-
- /**
- * Action to perform when emulation is resumed from a breakpoint.
- * @note This function will perform nothing unless it is overridden in the child class.
- */
- virtual void OnMaxwellResume() {}
-
- protected:
- /**
- * Weak context pointer. This need not be valid, so when requesting a shared_ptr via
- * context_weak.lock(), always compare the result against nullptr.
- */
- std::weak_ptr<DebugContext> context_weak;
- };
-
- /**
- * Simple structure defining a breakpoint state
- */
- struct BreakPoint {
- bool enabled = false;
- };
-
- /**
- * Static constructor used to create a shared_ptr of a DebugContext.
- */
- static std::shared_ptr<DebugContext> Construct() {
- return std::shared_ptr<DebugContext>(new DebugContext);
- }
-
- /**
- * Used by the emulation core when a given event has happened. If a breakpoint has been set
- * for this event, OnEvent calls the event handlers of the registered breakpoint observers.
- * The current thread then is halted until Resume() is called from another thread (or until
- * emulation is stopped).
- * @param event Event which has happened
- * @param data Optional data pointer (pass nullptr if unused). Needs to remain valid until
- * Resume() is called.
- */
- void OnEvent(Event event, void* data) {
- // This check is left in the header to allow the compiler to inline it.
- if (!breakpoints[(int)event].enabled)
- return;
- // For the rest of event handling, call a separate function.
- DoOnEvent(event, data);
- }
-
- void DoOnEvent(Event event, void* data);
-
- /**
- * Resume from the current breakpoint.
- * @warning Calling this from the same thread that OnEvent was called in will cause a deadlock.
- * Calling from any other thread is safe.
- */
- void Resume();
-
- /**
- * Delete all set breakpoints and resume emulation.
- */
- void ClearBreakpoints() {
- for (auto& bp : breakpoints) {
- bp.enabled = false;
- }
- Resume();
- }
-
- // TODO: Evaluate if access to these members should be hidden behind a public interface.
- std::array<BreakPoint, static_cast<int>(Event::NumEvents)> breakpoints;
- Event active_breakpoint{};
- bool at_breakpoint = false;
-
-private:
- /**
- * Private default constructor to make sure people always construct this through Construct()
- * instead.
- */
- DebugContext() = default;
-
- /// Mutex protecting current breakpoint state and the observer list.
- std::mutex breakpoint_mutex;
-
- /// Used by OnEvent to wait for resumption.
- std::condition_variable resume_from_breakpoint;
-
- /// List of registered observers
- std::list<BreakPointObserver*> breakpoint_observers;
-};
-
-} // namespace Tegra
diff --git a/src/video_core/engines/const_buffer_engine_interface.h b/src/video_core/engines/const_buffer_engine_interface.h
index ac27b6cbe..44b8b8d22 100644
--- a/src/video_core/engines/const_buffer_engine_interface.h
+++ b/src/video_core/engines/const_buffer_engine_interface.h
@@ -8,19 +8,11 @@
#include "common/bit_field.h"
#include "common/common_types.h"
#include "video_core/engines/shader_bytecode.h"
+#include "video_core/engines/shader_type.h"
#include "video_core/textures/texture.h"
namespace Tegra::Engines {
-enum class ShaderType : u32 {
- Vertex = 0,
- TesselationControl = 1,
- TesselationEval = 2,
- Geometry = 3,
- Fragment = 4,
- Compute = 5,
-};
-
struct SamplerDescriptor {
union {
BitField<0, 20, Tegra::Shader::TextureType> texture_type;
diff --git a/src/video_core/engines/kepler_compute.cpp b/src/video_core/engines/kepler_compute.cpp
index 3a39aeabe..110406f2f 100644
--- a/src/video_core/engines/kepler_compute.cpp
+++ b/src/video_core/engines/kepler_compute.cpp
@@ -8,6 +8,7 @@
#include "core/core.h"
#include "video_core/engines/kepler_compute.h"
#include "video_core/engines/maxwell_3d.h"
+#include "video_core/engines/shader_type.h"
#include "video_core/memory_manager.h"
#include "video_core/rasterizer_interface.h"
#include "video_core/renderer_base.h"
diff --git a/src/video_core/engines/kepler_compute.h b/src/video_core/engines/kepler_compute.h
index 5259d92bd..4ef3e0613 100644
--- a/src/video_core/engines/kepler_compute.h
+++ b/src/video_core/engines/kepler_compute.h
@@ -12,6 +12,7 @@
#include "common/common_types.h"
#include "video_core/engines/const_buffer_engine_interface.h"
#include "video_core/engines/engine_upload.h"
+#include "video_core/engines/shader_type.h"
#include "video_core/gpu.h"
#include "video_core/textures/texture.h"
@@ -140,7 +141,7 @@ public:
INSERT_PADDING_WORDS(0x3);
- BitField<0, 16, u32> shared_alloc;
+ BitField<0, 18, u32> shared_alloc;
BitField<16, 16, u32> block_dim_x;
union {
@@ -178,7 +179,12 @@ public:
BitField<24, 5, u32> gpr_alloc;
};
- INSERT_PADDING_WORDS(0x11);
+ union {
+ BitField<0, 20, u32> local_crs_alloc;
+ BitField<24, 5, u32> sass_version;
+ };
+
+ INSERT_PADDING_WORDS(0x10);
} launch_description{};
struct {
diff --git a/src/video_core/engines/maxwell_3d.cpp b/src/video_core/engines/maxwell_3d.cpp
index 2bed6cb38..1d1f780e7 100644
--- a/src/video_core/engines/maxwell_3d.cpp
+++ b/src/video_core/engines/maxwell_3d.cpp
@@ -7,8 +7,8 @@
#include "common/assert.h"
#include "core/core.h"
#include "core/core_timing.h"
-#include "video_core/debug_utils/debug_utils.h"
#include "video_core/engines/maxwell_3d.h"
+#include "video_core/engines/shader_type.h"
#include "video_core/memory_manager.h"
#include "video_core/rasterizer_interface.h"
#include "video_core/textures/texture.h"
@@ -87,11 +87,11 @@ void Maxwell3D::InitializeRegisterDefaults() {
color_mask.A.Assign(1);
}
- // Commercial games seem to assume this value is enabled and nouveau sets this value manually.
+ // NVN games expect these values to be enabled at boot
+ regs.rasterize_enable = 1;
regs.rt_separate_frag_data = 1;
-
- // Some games (like Super Mario Odyssey) assume that SRGB is enabled.
regs.framebuffer_srgb = 1;
+
mme_inline[MAXWELL3D_REG_INDEX(draw.vertex_end_gl)] = true;
mme_inline[MAXWELL3D_REG_INDEX(draw.vertex_begin_gl)] = true;
mme_inline[MAXWELL3D_REG_INDEX(vertex_buffer.count)] = true;
@@ -261,7 +261,8 @@ void Maxwell3D::CallMacroMethod(u32 method, std::size_t num_parameters, const u3
executing_macro = 0;
// Lookup the macro offset
- const u32 entry = ((method - MacroRegistersStart) >> 1) % macro_positions.size();
+ const u32 entry =
+ ((method - MacroRegistersStart) >> 1) % static_cast<u32>(macro_positions.size());
// Execute the current macro.
macro_interpreter.Execute(macro_positions[entry], num_parameters, parameters);
@@ -271,8 +272,6 @@ void Maxwell3D::CallMacroMethod(u32 method, std::size_t num_parameters, const u3
}
void Maxwell3D::CallMethod(const GPU::MethodCall& method_call) {
- auto debug_context = system.GetGPUDebugContext();
-
const u32 method = method_call.method;
if (method == cb_data_state.current) {
@@ -313,10 +312,6 @@ void Maxwell3D::CallMethod(const GPU::MethodCall& method_call) {
ASSERT_MSG(method < Regs::NUM_REGS,
"Invalid Maxwell3D register, increase the size of the Regs structure");
- if (debug_context) {
- debug_context->OnEvent(Tegra::DebugContext::Event::MaxwellCommandLoaded, nullptr);
- }
-
if (regs.reg_array[method] != method_call.argument) {
regs.reg_array[method] = method_call.argument;
const std::size_t dirty_reg = dirty_pointers[method];
@@ -367,24 +362,24 @@ void Maxwell3D::CallMethod(const GPU::MethodCall& method_call) {
StartCBData(method);
break;
}
- case MAXWELL3D_REG_INDEX(cb_bind[0].raw_config): {
- ProcessCBBind(Regs::ShaderStage::Vertex);
+ case MAXWELL3D_REG_INDEX(cb_bind[0]): {
+ ProcessCBBind(0);
break;
}
- case MAXWELL3D_REG_INDEX(cb_bind[1].raw_config): {
- ProcessCBBind(Regs::ShaderStage::TesselationControl);
+ case MAXWELL3D_REG_INDEX(cb_bind[1]): {
+ ProcessCBBind(1);
break;
}
- case MAXWELL3D_REG_INDEX(cb_bind[2].raw_config): {
- ProcessCBBind(Regs::ShaderStage::TesselationEval);
+ case MAXWELL3D_REG_INDEX(cb_bind[2]): {
+ ProcessCBBind(2);
break;
}
- case MAXWELL3D_REG_INDEX(cb_bind[3].raw_config): {
- ProcessCBBind(Regs::ShaderStage::Geometry);
+ case MAXWELL3D_REG_INDEX(cb_bind[3]): {
+ ProcessCBBind(3);
break;
}
- case MAXWELL3D_REG_INDEX(cb_bind[4].raw_config): {
- ProcessCBBind(Regs::ShaderStage::Fragment);
+ case MAXWELL3D_REG_INDEX(cb_bind[4]): {
+ ProcessCBBind(4);
break;
}
case MAXWELL3D_REG_INDEX(draw.vertex_end_gl): {
@@ -422,10 +417,6 @@ void Maxwell3D::CallMethod(const GPU::MethodCall& method_call) {
default:
break;
}
-
- if (debug_context) {
- debug_context->OnEvent(Tegra::DebugContext::Event::MaxwellCommandProcessed, nullptr);
- }
}
void Maxwell3D::StepInstance(const MMEDrawMode expected_mode, const u32 count) {
@@ -483,12 +474,6 @@ void Maxwell3D::FlushMMEInlineDraw() {
ASSERT_MSG(!(regs.index_array.count && regs.vertex_buffer.count), "Both indexed and direct?");
ASSERT(mme_draw.instance_count == mme_draw.gl_end_count);
- auto debug_context = system.GetGPUDebugContext();
-
- if (debug_context) {
- debug_context->OnEvent(Tegra::DebugContext::Event::IncomingPrimitiveBatch, nullptr);
- }
-
// Both instance configuration registers can not be set at the same time.
ASSERT_MSG(!regs.draw.instance_next || !regs.draw.instance_cont,
"Illegal combination of instancing parameters");
@@ -498,10 +483,6 @@ void Maxwell3D::FlushMMEInlineDraw() {
rasterizer.DrawMultiBatch(is_indexed);
}
- if (debug_context) {
- debug_context->OnEvent(Tegra::DebugContext::Event::FinishedPrimitiveBatch, nullptr);
- }
-
// TODO(bunnei): Below, we reset vertex count so that we can use these registers to determine if
// the game is trying to draw indexed or direct mode. This needs to be verified on HW still -
// it's possible that it is incorrect and that there is some other register used to specify the
@@ -648,12 +629,6 @@ void Maxwell3D::DrawArrays() {
regs.vertex_buffer.count);
ASSERT_MSG(!(regs.index_array.count && regs.vertex_buffer.count), "Both indexed and direct?");
- auto debug_context = system.GetGPUDebugContext();
-
- if (debug_context) {
- debug_context->OnEvent(Tegra::DebugContext::Event::IncomingPrimitiveBatch, nullptr);
- }
-
// Both instance configuration registers can not be set at the same time.
ASSERT_MSG(!regs.draw.instance_next || !regs.draw.instance_cont,
"Illegal combination of instancing parameters");
@@ -671,10 +646,6 @@ void Maxwell3D::DrawArrays() {
rasterizer.DrawBatch(is_indexed);
}
- if (debug_context) {
- debug_context->OnEvent(Tegra::DebugContext::Event::FinishedPrimitiveBatch, nullptr);
- }
-
// TODO(bunnei): Below, we reset vertex count so that we can use these registers to determine if
// the game is trying to draw indexed or direct mode. This needs to be verified on HW still -
// it's possible that it is incorrect and that there is some other register used to specify the
@@ -686,10 +657,10 @@ void Maxwell3D::DrawArrays() {
}
}
-void Maxwell3D::ProcessCBBind(Regs::ShaderStage stage) {
+void Maxwell3D::ProcessCBBind(std::size_t stage_index) {
// Bind the buffer currently in CB_ADDRESS to the specified index in the desired shader stage.
- auto& shader = state.shader_stages[static_cast<std::size_t>(stage)];
- auto& bind_data = regs.cb_bind[static_cast<std::size_t>(stage)];
+ auto& shader = state.shader_stages[stage_index];
+ auto& bind_data = regs.cb_bind[stage_index];
ASSERT(bind_data.index < Regs::MaxConstBuffers);
auto& buffer = shader.const_buffers[bind_data.index];
@@ -741,14 +712,6 @@ Texture::TICEntry Maxwell3D::GetTICEntry(u32 tic_index) const {
Texture::TICEntry tic_entry;
memory_manager.ReadBlockUnsafe(tic_address_gpu, &tic_entry, sizeof(Texture::TICEntry));
- [[maybe_unused]] const auto r_type{tic_entry.r_type.Value()};
- [[maybe_unused]] const auto g_type{tic_entry.g_type.Value()};
- [[maybe_unused]] const auto b_type{tic_entry.b_type.Value()};
- [[maybe_unused]] const auto a_type{tic_entry.a_type.Value()};
-
- // TODO(Subv): Different data types for separate components are not supported
- DEBUG_ASSERT(r_type == g_type && r_type == b_type && r_type == a_type);
-
return tic_entry;
}
@@ -764,9 +727,9 @@ Texture::FullTextureInfo Maxwell3D::GetTextureInfo(Texture::TextureHandle tex_ha
return Texture::FullTextureInfo{GetTICEntry(tex_handle.tic_id), GetTSCEntry(tex_handle.tsc_id)};
}
-Texture::FullTextureInfo Maxwell3D::GetStageTexture(Regs::ShaderStage stage,
- std::size_t offset) const {
- const auto& shader = state.shader_stages[static_cast<std::size_t>(stage)];
+Texture::FullTextureInfo Maxwell3D::GetStageTexture(ShaderType stage, std::size_t offset) const {
+ const auto stage_index = static_cast<std::size_t>(stage);
+ const auto& shader = state.shader_stages[stage_index];
const auto& tex_info_buffer = shader.const_buffers[regs.tex_cb_index];
ASSERT(tex_info_buffer.enabled && tex_info_buffer.address != 0);
diff --git a/src/video_core/engines/maxwell_3d.h b/src/video_core/engines/maxwell_3d.h
index 1aa7c274f..a35e7a195 100644
--- a/src/video_core/engines/maxwell_3d.h
+++ b/src/video_core/engines/maxwell_3d.h
@@ -18,6 +18,7 @@
#include "video_core/engines/const_buffer_engine_interface.h"
#include "video_core/engines/const_buffer_info.h"
#include "video_core/engines/engine_upload.h"
+#include "video_core/engines/shader_type.h"
#include "video_core/gpu.h"
#include "video_core/macro_interpreter.h"
#include "video_core/textures/texture.h"
@@ -62,7 +63,6 @@ public:
static constexpr std::size_t NumVertexArrays = 32;
static constexpr std::size_t NumVertexAttributes = 32;
static constexpr std::size_t NumVaryings = 31;
- static constexpr std::size_t NumTextureSamplers = 32;
static constexpr std::size_t NumImages = 8; // TODO(Rodrigo): Investigate this number
static constexpr std::size_t NumClipDistances = 8;
static constexpr std::size_t MaxShaderProgram = 6;
@@ -130,14 +130,6 @@ public:
Fragment = 5,
};
- enum class ShaderStage : u32 {
- Vertex = 0,
- TesselationControl = 1,
- TesselationEval = 2,
- Geometry = 3,
- Fragment = 4,
- };
-
struct VertexAttribute {
enum class Size : u32 {
Invalid = 0x0,
@@ -318,6 +310,11 @@ public:
}
};
+ enum class DepthMode : u32 {
+ MinusOneToOne = 0,
+ ZeroToOne = 1,
+ };
+
enum class PrimitiveTopology : u32 {
Points = 0x0,
Lines = 0x1,
@@ -499,6 +496,18 @@ public:
INSERT_UNION_PADDING_WORDS(1);
};
+ enum class TessellationPrimitive : u32 {
+ Isolines = 0,
+ Triangles = 1,
+ Quads = 2,
+ };
+
+ enum class TessellationSpacing : u32 {
+ Equal = 0,
+ FractionalOdd = 1,
+ FractionalEven = 2,
+ };
+
struct RenderTargetConfig {
u32 address_high;
u32 address_low;
@@ -636,7 +645,23 @@ public:
};
} sync_info;
- INSERT_UNION_PADDING_WORDS(0x11E);
+ INSERT_UNION_PADDING_WORDS(0x15);
+
+ union {
+ BitField<0, 2, TessellationPrimitive> prim;
+ BitField<4, 2, TessellationSpacing> spacing;
+ BitField<8, 1, u32> cw;
+ BitField<9, 1, u32> connected;
+ } tess_mode;
+
+ std::array<f32, 4> tess_level_outer;
+ std::array<f32, 2> tess_level_inner;
+
+ INSERT_UNION_PADDING_WORDS(0x10);
+
+ u32 rasterize_enable;
+
+ INSERT_UNION_PADDING_WORDS(0xF1);
u32 tfb_enabled;
@@ -655,7 +680,7 @@ public:
u32 count;
} vertex_buffer;
- INSERT_UNION_PADDING_WORDS(1);
+ DepthMode depth_mode;
float clear_color[4];
float clear_depth;
@@ -670,27 +695,31 @@ public:
u32 polygon_offset_line_enable;
u32 polygon_offset_fill_enable;
- INSERT_UNION_PADDING_WORDS(0xD);
+ u32 patch_vertices;
+
+ INSERT_UNION_PADDING_WORDS(0xC);
std::array<ScissorTest, NumViewports> scissor_test;
INSERT_UNION_PADDING_WORDS(0x15);
s32 stencil_back_func_ref;
- u32 stencil_back_mask;
u32 stencil_back_func_mask;
+ u32 stencil_back_mask;
INSERT_UNION_PADDING_WORDS(0xC);
u32 color_mask_common;
- INSERT_UNION_PADDING_WORDS(0x6);
-
- u32 rt_separate_frag_data;
+ INSERT_UNION_PADDING_WORDS(0x2);
f32 depth_bounds[2];
- INSERT_UNION_PADDING_WORDS(0xA);
+ INSERT_UNION_PADDING_WORDS(0x2);
+
+ u32 rt_separate_frag_data;
+
+ INSERT_UNION_PADDING_WORDS(0xC);
struct {
u32 address_high;
@@ -1007,7 +1036,12 @@ public:
BitField<4, 1, u32> depth_clamp_far;
} view_volume_clip_control;
- INSERT_UNION_PADDING_WORDS(0x21);
+ INSERT_UNION_PADDING_WORDS(0x1F);
+
+ u32 depth_bounds_enable;
+
+ INSERT_UNION_PADDING_WORDS(1);
+
struct {
u32 enable;
LogicOperation operation;
@@ -1254,7 +1288,7 @@ public:
Texture::FullTextureInfo GetTextureInfo(Texture::TextureHandle tex_handle) const;
/// Returns the texture information for a specific texture in a specific shader stage.
- Texture::FullTextureInfo GetStageTexture(Regs::ShaderStage stage, std::size_t offset) const;
+ Texture::FullTextureInfo GetStageTexture(ShaderType stage, std::size_t offset) const;
u32 AccessConstBuffer32(ShaderType stage, u64 const_buffer, u64 offset) const override;
@@ -1376,7 +1410,7 @@ private:
void FinishCBData();
/// Handles a write to the CB_BIND register.
- void ProcessCBBind(Regs::ShaderStage stage);
+ void ProcessCBBind(std::size_t stage_index);
/// Handles a write to the VERTEX_END_GL register, triggering a draw.
void DrawArrays();
@@ -1394,24 +1428,30 @@ ASSERT_REG_POSITION(upload, 0x60);
ASSERT_REG_POSITION(exec_upload, 0x6C);
ASSERT_REG_POSITION(data_upload, 0x6D);
ASSERT_REG_POSITION(sync_info, 0xB2);
+ASSERT_REG_POSITION(tess_mode, 0xC8);
+ASSERT_REG_POSITION(tess_level_outer, 0xC9);
+ASSERT_REG_POSITION(tess_level_inner, 0xCD);
+ASSERT_REG_POSITION(rasterize_enable, 0xDF);
ASSERT_REG_POSITION(tfb_enabled, 0x1D1);
ASSERT_REG_POSITION(rt, 0x200);
ASSERT_REG_POSITION(viewport_transform, 0x280);
ASSERT_REG_POSITION(viewports, 0x300);
ASSERT_REG_POSITION(vertex_buffer, 0x35D);
+ASSERT_REG_POSITION(depth_mode, 0x35F);
ASSERT_REG_POSITION(clear_color[0], 0x360);
ASSERT_REG_POSITION(clear_depth, 0x364);
ASSERT_REG_POSITION(clear_stencil, 0x368);
ASSERT_REG_POSITION(polygon_offset_point_enable, 0x370);
ASSERT_REG_POSITION(polygon_offset_line_enable, 0x371);
ASSERT_REG_POSITION(polygon_offset_fill_enable, 0x372);
+ASSERT_REG_POSITION(patch_vertices, 0x373);
ASSERT_REG_POSITION(scissor_test, 0x380);
ASSERT_REG_POSITION(stencil_back_func_ref, 0x3D5);
-ASSERT_REG_POSITION(stencil_back_mask, 0x3D6);
-ASSERT_REG_POSITION(stencil_back_func_mask, 0x3D7);
+ASSERT_REG_POSITION(stencil_back_func_mask, 0x3D6);
+ASSERT_REG_POSITION(stencil_back_mask, 0x3D7);
ASSERT_REG_POSITION(color_mask_common, 0x3E4);
ASSERT_REG_POSITION(rt_separate_frag_data, 0x3EB);
-ASSERT_REG_POSITION(depth_bounds, 0x3EC);
+ASSERT_REG_POSITION(depth_bounds, 0x3E7);
ASSERT_REG_POSITION(zeta, 0x3F8);
ASSERT_REG_POSITION(clear_flags, 0x43E);
ASSERT_REG_POSITION(vertex_attrib_format, 0x458);
@@ -1467,6 +1507,7 @@ ASSERT_REG_POSITION(cull, 0x646);
ASSERT_REG_POSITION(pixel_center_integer, 0x649);
ASSERT_REG_POSITION(viewport_transform_enabled, 0x64B);
ASSERT_REG_POSITION(view_volume_clip_control, 0x64F);
+ASSERT_REG_POSITION(depth_bounds_enable, 0x66F);
ASSERT_REG_POSITION(logic_op, 0x671);
ASSERT_REG_POSITION(clear_buffers, 0x674);
ASSERT_REG_POSITION(color_mask, 0x680);
diff --git a/src/video_core/engines/shader_bytecode.h b/src/video_core/engines/shader_bytecode.h
index 8f6bc76eb..57b57c647 100644
--- a/src/video_core/engines/shader_bytecode.h
+++ b/src/video_core/engines/shader_bytecode.h
@@ -98,10 +98,11 @@ union Attribute {
BitField<20, 10, u64> immediate;
BitField<22, 2, u64> element;
BitField<24, 6, Index> index;
+ BitField<31, 1, u64> patch;
BitField<47, 3, AttributeSize> size;
bool IsPhysical() const {
- return element == 0 && static_cast<u64>(index.Value()) == 0;
+ return patch == 0 && element == 0 && static_cast<u64>(index.Value()) == 0;
}
} fmt20;
@@ -383,6 +384,15 @@ enum class IsberdMode : u64 {
enum class IsberdShift : u64 { None = 0, U16 = 1, B32 = 2 };
+enum class MembarType : u64 {
+ CTA = 0,
+ GL = 1,
+ SYS = 2,
+ VC = 3,
+};
+
+enum class MembarUnknown : u64 { Default = 0, IVALLD = 1, IVALLT = 2, IVALLTD = 3 };
+
enum class HalfType : u64 {
H0_H1 = 0,
F32 = 1,
@@ -616,6 +626,14 @@ union Instruction {
} shfl;
union {
+ BitField<44, 1, u64> ftz;
+ BitField<39, 2, u64> tab5cb8_2;
+ BitField<38, 1, u64> ndv;
+ BitField<47, 1, u64> cc;
+ BitField<28, 8, u64> swizzle;
+ } fswzadd;
+
+ union {
BitField<8, 8, Register> gpr;
BitField<20, 24, s64> offset;
} gmem;
@@ -792,6 +810,12 @@ union Instruction {
} popc;
union {
+ BitField<41, 1, u64> sh;
+ BitField<40, 1, u64> invert;
+ BitField<48, 1, u64> is_signed;
+ } flo;
+
+ union {
BitField<39, 3, u64> pred;
BitField<42, 1, u64> neg_pred;
} sel;
@@ -1027,7 +1051,7 @@ union Instruction {
BitField<40, 1, R2pMode> mode;
BitField<41, 2, u64> byte;
BitField<20, 7, u64> immediate_mask;
- } r2p;
+ } p2r_r2p;
union {
BitField<39, 3, u64> pred39;
@@ -1215,7 +1239,7 @@ union Instruction {
BitField<35, 1, u64> ndv_flag;
BitField<49, 1, u64> nodep_flag;
BitField<50, 1, u64> dc_flag;
- BitField<54, 2, u64> info;
+ BitField<54, 2, u64> offset_mode;
BitField<56, 2, u64> component;
bool UsesMiscMode(TextureMiscMode mode) const {
@@ -1227,9 +1251,9 @@ union Instruction {
case TextureMiscMode::DC:
return dc_flag != 0;
case TextureMiscMode::AOFFI:
- return info == 1;
+ return offset_mode == 1;
case TextureMiscMode::PTP:
- return info == 2;
+ return offset_mode == 2;
default:
break;
}
@@ -1241,7 +1265,7 @@ union Instruction {
BitField<35, 1, u64> ndv_flag;
BitField<49, 1, u64> nodep_flag;
BitField<50, 1, u64> dc_flag;
- BitField<33, 2, u64> info;
+ BitField<33, 2, u64> offset_mode;
BitField<37, 2, u64> component;
bool UsesMiscMode(TextureMiscMode mode) const {
@@ -1253,9 +1277,9 @@ union Instruction {
case TextureMiscMode::DC:
return dc_flag != 0;
case TextureMiscMode::AOFFI:
- return info == 1;
+ return offset_mode == 1;
case TextureMiscMode::PTP:
- return info == 2;
+ return offset_mode == 2;
default:
break;
}
@@ -1268,6 +1292,7 @@ union Instruction {
BitField<50, 1, u64> dc_flag;
BitField<51, 1, u64> aoffi_flag;
BitField<52, 2, u64> component;
+ BitField<55, 1, u64> fp16_flag;
bool UsesMiscMode(TextureMiscMode mode) const {
switch (mode) {
@@ -1432,6 +1457,26 @@ union Instruction {
} tlds;
union {
+ BitField<28, 1, u64> is_array;
+ BitField<29, 2, TextureType> texture_type;
+ BitField<35, 1, u64> aoffi_flag;
+ BitField<49, 1, u64> nodep_flag;
+
+ bool UsesMiscMode(TextureMiscMode mode) const {
+ switch (mode) {
+ case TextureMiscMode::AOFFI:
+ return aoffi_flag != 0;
+ case TextureMiscMode::NODEP:
+ return nodep_flag != 0;
+ default:
+ break;
+ }
+ return false;
+ }
+
+ } txd;
+
+ union {
BitField<24, 2, StoreCacheManagement> cache_management;
BitField<33, 3, ImageType> image_type;
BitField<49, 2, OutOfBoundsStore> out_of_bounds_store;
@@ -1478,7 +1523,8 @@ union Instruction {
u32 value = static_cast<u32>(target);
// The branch offset is relative to the next instruction and is stored in bytes, so
// divide it by the size of an instruction and add 1 to it.
- return static_cast<s32>((value ^ mask) - mask) / sizeof(Instruction) + 1;
+ return static_cast<s32>((value ^ mask) - mask) / static_cast<s32>(sizeof(Instruction)) +
+ 1;
}
} bra;
@@ -1492,7 +1538,8 @@ union Instruction {
u32 value = static_cast<u32>(target);
// The branch offset is relative to the next instruction and is stored in bytes, so
// divide it by the size of an instruction and add 1 to it.
- return static_cast<s32>((value ^ mask) - mask) / sizeof(Instruction) + 1;
+ return static_cast<s32>((value ^ mask) - mask) / static_cast<s32>(sizeof(Instruction)) +
+ 1;
}
} brx;
@@ -1509,6 +1556,11 @@ union Instruction {
} isberd;
union {
+ BitField<8, 2, MembarType> type;
+ BitField<0, 2, MembarUnknown> unknown;
+ } membar;
+
+ union {
BitField<48, 1, u64> signed_a;
BitField<38, 1, u64> is_byte_chunk_a;
BitField<36, 2, VideoType> type_a;
@@ -1590,6 +1642,7 @@ public:
DEPBAR,
VOTE,
SHFL,
+ FSWZADD,
BFE_C,
BFE_R,
BFE_IMM,
@@ -1621,6 +1674,8 @@ public:
TLD4S, // Texture Load 4 with scalar / non - vec4 source / destinations
TMML_B, // Texture Mip Map Level
TMML, // Texture Mip Map Level
+ TXD, // Texture Gradient/Load with Derivates
+ TXD_B, // Texture Gradient/Load with Derivates Bindless
SUST, // Surface Store
SULD, // Surface Load
SUATOM, // Surface Atomic Operation
@@ -1629,6 +1684,7 @@ public:
IPA,
OUT_R, // Emit vertex/primitive
ISBERD,
+ MEMBAR,
VMAD,
VSETP,
FFMA_IMM, // Fused Multiply and Add
@@ -1653,6 +1709,9 @@ public:
ISCADD_C, // Scale and Add
ISCADD_R,
ISCADD_IMM,
+ FLO_R,
+ FLO_C,
+ FLO_IMM,
LEA_R1,
LEA_R2,
LEA_RZ,
@@ -1716,6 +1775,10 @@ public:
SHR_C,
SHR_R,
SHR_IMM,
+ SHF_RIGHT_R,
+ SHF_RIGHT_IMM,
+ SHF_LEFT_R,
+ SHF_LEFT_IMM,
FMNMX_C,
FMNMX_R,
FMNMX_IMM,
@@ -1738,6 +1801,7 @@ public:
PSET,
CSETP,
R2P_IMM,
+ P2R_IMM,
XMAD_IMM,
XMAD_CR,
XMAD_RC,
@@ -1851,11 +1915,11 @@ private:
const std::size_t bit_position = opcode_bitsize - i - 1;
switch (bitstring[i]) {
case '0':
- mask |= 1 << bit_position;
+ mask |= static_cast<u16>(1U << bit_position);
break;
case '1':
- expect |= 1 << bit_position;
- mask |= 1 << bit_position;
+ expect |= static_cast<u16>(1U << bit_position);
+ mask |= static_cast<u16>(1U << bit_position);
break;
default:
// Ignore
@@ -1883,11 +1947,12 @@ private:
INST("111000100100----", Id::BRA, Type::Flow, "BRA"),
INST("111000100101----", Id::BRX, Type::Flow, "BRX"),
INST("1111000011111---", Id::SYNC, Type::Flow, "SYNC"),
- INST("111000110100---", Id::BRK, Type::Flow, "BRK"),
+ INST("111000110100----", Id::BRK, Type::Flow, "BRK"),
INST("111000110000----", Id::EXIT, Type::Flow, "EXIT"),
INST("1111000011110---", Id::DEPBAR, Type::Synch, "DEPBAR"),
INST("0101000011011---", Id::VOTE, Type::Warp, "VOTE"),
INST("1110111100010---", Id::SHFL, Type::Warp, "SHFL"),
+ INST("0101000011111---", Id::FSWZADD, Type::Warp, "FSWZADD"),
INST("1110111111011---", Id::LD_A, Type::Memory, "LD_A"),
INST("1110111101001---", Id::LD_S, Type::Memory, "LD_S"),
INST("1110111101000---", Id::LD_L, Type::Memory, "LD_L"),
@@ -1909,9 +1974,11 @@ private:
INST("1101-01---------", Id::TLDS, Type::Texture, "TLDS"),
INST("110010----111---", Id::TLD4, Type::Texture, "TLD4"),
INST("1101111011111---", Id::TLD4_B, Type::Texture, "TLD4_B"),
- INST("1101111100------", Id::TLD4S, Type::Texture, "TLD4S"),
+ INST("11011111-0------", Id::TLD4S, Type::Texture, "TLD4S"),
INST("110111110110----", Id::TMML_B, Type::Texture, "TMML_B"),
INST("1101111101011---", Id::TMML, Type::Texture, "TMML"),
+ INST("11011110011110--", Id::TXD_B, Type::Texture, "TXD_B"),
+ INST("11011110001110--", Id::TXD, Type::Texture, "TXD"),
INST("11101011001-----", Id::SUST, Type::Image, "SUST"),
INST("11101011000-----", Id::SULD, Type::Image, "SULD"),
INST("1110101000------", Id::SUATOM, Type::Image, "SUATOM_D"),
@@ -1919,6 +1986,7 @@ private:
INST("11100000--------", Id::IPA, Type::Trivial, "IPA"),
INST("1111101111100---", Id::OUT_R, Type::Trivial, "OUT_R"),
INST("1110111111010---", Id::ISBERD, Type::Trivial, "ISBERD"),
+ INST("1110111110011---", Id::MEMBAR, Type::Trivial, "MEMBAR"),
INST("01011111--------", Id::VMAD, Type::Video, "VMAD"),
INST("0101000011110---", Id::VSETP, Type::Video, "VSETP"),
INST("0011001-1-------", Id::FFMA_IMM, Type::Ffma, "FFMA_IMM"),
@@ -1953,6 +2021,9 @@ private:
INST("010110110100----", Id::ICMP_R, Type::ArithmeticInteger, "ICMP_R"),
INST("010010110100----", Id::ICMP_CR, Type::ArithmeticInteger, "ICMP_CR"),
INST("0011011-0100----", Id::ICMP_IMM, Type::ArithmeticInteger, "ICMP_IMM"),
+ INST("0101110000110---", Id::FLO_R, Type::ArithmeticInteger, "FLO_R"),
+ INST("0100110000110---", Id::FLO_C, Type::ArithmeticInteger, "FLO_C"),
+ INST("0011100-00110---", Id::FLO_IMM, Type::ArithmeticInteger, "FLO_IMM"),
INST("0101101111011---", Id::LEA_R2, Type::ArithmeticInteger, "LEA_R2"),
INST("0101101111010---", Id::LEA_R1, Type::ArithmeticInteger, "LEA_R1"),
INST("001101101101----", Id::LEA_IMM, Type::ArithmeticInteger, "LEA_IMM"),
@@ -2010,6 +2081,10 @@ private:
INST("0100110000101---", Id::SHR_C, Type::Shift, "SHR_C"),
INST("0101110000101---", Id::SHR_R, Type::Shift, "SHR_R"),
INST("0011100-00101---", Id::SHR_IMM, Type::Shift, "SHR_IMM"),
+ INST("0101110011111---", Id::SHF_RIGHT_R, Type::Shift, "SHF_RIGHT_R"),
+ INST("0011100-11111---", Id::SHF_RIGHT_IMM, Type::Shift, "SHF_RIGHT_IMM"),
+ INST("0101101111111---", Id::SHF_LEFT_R, Type::Shift, "SHF_LEFT_R"),
+ INST("0011011-11111---", Id::SHF_LEFT_IMM, Type::Shift, "SHF_LEFT_IMM"),
INST("0100110011100---", Id::I2I_C, Type::Conversion, "I2I_C"),
INST("0101110011100---", Id::I2I_R, Type::Conversion, "I2I_R"),
INST("0011101-11100---", Id::I2I_IMM, Type::Conversion, "I2I_IMM"),
@@ -2032,6 +2107,7 @@ private:
INST("0101000010010---", Id::PSETP, Type::PredicateSetPredicate, "PSETP"),
INST("010100001010----", Id::CSETP, Type::PredicateSetPredicate, "CSETP"),
INST("0011100-11110---", Id::R2P_IMM, Type::RegisterSetPredicate, "R2P_IMM"),
+ INST("0011100-11101---", Id::P2R_IMM, Type::RegisterSetPredicate, "P2R_IMM"),
INST("0011011-00------", Id::XMAD_IMM, Type::Xmad, "XMAD_IMM"),
INST("0100111---------", Id::XMAD_CR, Type::Xmad, "XMAD_CR"),
INST("010100010-------", Id::XMAD_RC, Type::Xmad, "XMAD_RC"),
diff --git a/src/video_core/engines/shader_type.h b/src/video_core/engines/shader_type.h
new file mode 100644
index 000000000..49ce5cde5
--- /dev/null
+++ b/src/video_core/engines/shader_type.h
@@ -0,0 +1,21 @@
+// Copyright 2019 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "common/common_types.h"
+
+namespace Tegra::Engines {
+
+enum class ShaderType : u32 {
+ Vertex = 0,
+ TesselationControl = 1,
+ TesselationEval = 2,
+ Geometry = 3,
+ Fragment = 4,
+ Compute = 5,
+};
+static constexpr std::size_t MaxShaderTypes = 6;
+
+} // namespace Tegra::Engines
diff --git a/src/video_core/gpu_thread.cpp b/src/video_core/gpu_thread.cpp
index 758a37f14..2cdf1aa7f 100644
--- a/src/video_core/gpu_thread.cpp
+++ b/src/video_core/gpu_thread.cpp
@@ -31,24 +31,22 @@ static void RunThread(VideoCore::RendererBase& renderer, Tegra::DmaPusher& dma_p
CommandDataContainer next;
while (state.is_running) {
- while (!state.queue.Empty()) {
- state.queue.Pop(next);
- if (const auto submit_list = std::get_if<SubmitListCommand>(&next.data)) {
- dma_pusher.Push(std::move(submit_list->entries));
- dma_pusher.DispatchCalls();
- } else if (const auto data = std::get_if<SwapBuffersCommand>(&next.data)) {
- renderer.SwapBuffers(data->framebuffer ? &*data->framebuffer : nullptr);
- } else if (const auto data = std::get_if<FlushRegionCommand>(&next.data)) {
- renderer.Rasterizer().FlushRegion(data->addr, data->size);
- } else if (const auto data = std::get_if<InvalidateRegionCommand>(&next.data)) {
- renderer.Rasterizer().InvalidateRegion(data->addr, data->size);
- } else if (std::holds_alternative<EndProcessingCommand>(next.data)) {
- return;
- } else {
- UNREACHABLE();
- }
- state.signaled_fence.store(next.fence);
+ next = state.queue.PopWait();
+ if (const auto submit_list = std::get_if<SubmitListCommand>(&next.data)) {
+ dma_pusher.Push(std::move(submit_list->entries));
+ dma_pusher.DispatchCalls();
+ } else if (const auto data = std::get_if<SwapBuffersCommand>(&next.data)) {
+ renderer.SwapBuffers(data->framebuffer ? &*data->framebuffer : nullptr);
+ } else if (const auto data = std::get_if<FlushRegionCommand>(&next.data)) {
+ renderer.Rasterizer().FlushRegion(data->addr, data->size);
+ } else if (const auto data = std::get_if<InvalidateRegionCommand>(&next.data)) {
+ renderer.Rasterizer().InvalidateRegion(data->addr, data->size);
+ } else if (std::holds_alternative<EndProcessingCommand>(next.data)) {
+ return;
+ } else {
+ UNREACHABLE();
}
+ state.signaled_fence.store(next.fence);
}
}
@@ -73,8 +71,7 @@ void ThreadManager::SubmitList(Tegra::CommandList&& entries) {
}
void ThreadManager::SwapBuffers(const Tegra::FramebufferConfig* framebuffer) {
- PushCommand(SwapBuffersCommand(framebuffer ? *framebuffer
- : std::optional<const Tegra::FramebufferConfig>{}));
+ PushCommand(SwapBuffersCommand(framebuffer ? std::make_optional(*framebuffer) : std::nullopt));
}
void ThreadManager::FlushRegion(CacheAddr addr, u64 size) {
diff --git a/src/video_core/memory_manager.cpp b/src/video_core/memory_manager.cpp
index bffae940c..11848fbce 100644
--- a/src/video_core/memory_manager.cpp
+++ b/src/video_core/memory_manager.cpp
@@ -52,7 +52,7 @@ GPUVAddr MemoryManager::MapBufferEx(VAddr cpu_addr, u64 size) {
const u64 aligned_size{Common::AlignUp(size, page_size)};
const GPUVAddr gpu_addr{FindFreeRegion(address_space_base, aligned_size)};
- MapBackingMemory(gpu_addr, Memory::GetPointer(cpu_addr), aligned_size, cpu_addr);
+ MapBackingMemory(gpu_addr, system.Memory().GetPointer(cpu_addr), aligned_size, cpu_addr);
ASSERT(system.CurrentProcess()
->VMManager()
.SetMemoryAttribute(cpu_addr, size, Kernel::MemoryAttribute::DeviceMapped,
@@ -67,7 +67,7 @@ GPUVAddr MemoryManager::MapBufferEx(VAddr cpu_addr, GPUVAddr gpu_addr, u64 size)
const u64 aligned_size{Common::AlignUp(size, page_size)};
- MapBackingMemory(gpu_addr, Memory::GetPointer(cpu_addr), aligned_size, cpu_addr);
+ MapBackingMemory(gpu_addr, system.Memory().GetPointer(cpu_addr), aligned_size, cpu_addr);
ASSERT(system.CurrentProcess()
->VMManager()
.SetMemoryAttribute(cpu_addr, size, Kernel::MemoryAttribute::DeviceMapped,
diff --git a/src/video_core/rasterizer_accelerated.cpp b/src/video_core/rasterizer_accelerated.cpp
index b230dcc18..d01db97da 100644
--- a/src/video_core/rasterizer_accelerated.cpp
+++ b/src/video_core/rasterizer_accelerated.cpp
@@ -5,6 +5,7 @@
#include <mutex>
#include <boost/icl/interval_map.hpp>
+#include <boost/range/iterator_range.hpp>
#include "common/assert.h"
#include "common/common_types.h"
@@ -22,7 +23,8 @@ constexpr auto RangeFromInterval(Map& map, const Interval& interval) {
} // Anonymous namespace
-RasterizerAccelerated::RasterizerAccelerated() = default;
+RasterizerAccelerated::RasterizerAccelerated(Memory::Memory& cpu_memory_)
+ : cpu_memory{cpu_memory_} {}
RasterizerAccelerated::~RasterizerAccelerated() = default;
@@ -47,9 +49,9 @@ void RasterizerAccelerated::UpdatePagesCachedCount(VAddr addr, u64 size, int del
const u64 interval_size = interval_end_addr - interval_start_addr;
if (delta > 0 && count == delta) {
- Memory::RasterizerMarkRegionCached(interval_start_addr, interval_size, true);
+ cpu_memory.RasterizerMarkRegionCached(interval_start_addr, interval_size, true);
} else if (delta < 0 && count == -delta) {
- Memory::RasterizerMarkRegionCached(interval_start_addr, interval_size, false);
+ cpu_memory.RasterizerMarkRegionCached(interval_start_addr, interval_size, false);
} else {
ASSERT(count >= 0);
}
diff --git a/src/video_core/rasterizer_accelerated.h b/src/video_core/rasterizer_accelerated.h
index 8f7e3547e..315798e7c 100644
--- a/src/video_core/rasterizer_accelerated.h
+++ b/src/video_core/rasterizer_accelerated.h
@@ -11,12 +11,16 @@
#include "common/common_types.h"
#include "video_core/rasterizer_interface.h"
+namespace Memory {
+class Memory;
+}
+
namespace VideoCore {
/// Implements the shared part in GPU accelerated rasterizers in RasterizerInterface.
class RasterizerAccelerated : public RasterizerInterface {
public:
- explicit RasterizerAccelerated();
+ explicit RasterizerAccelerated(Memory::Memory& cpu_memory_);
~RasterizerAccelerated() override;
void UpdatePagesCachedCount(VAddr addr, u64 size, int delta) override;
@@ -24,8 +28,9 @@ public:
private:
using CachedPageMap = boost::icl::interval_map<u64, int>;
CachedPageMap cached_pages;
-
std::mutex pages_mutex;
+
+ Memory::Memory& cpu_memory;
};
} // namespace VideoCore
diff --git a/src/video_core/renderer_opengl/gl_device.cpp b/src/video_core/renderer_opengl/gl_device.cpp
index c65b24c69..1a2e2a9f7 100644
--- a/src/video_core/renderer_opengl/gl_device.cpp
+++ b/src/video_core/renderer_opengl/gl_device.cpp
@@ -5,7 +5,10 @@
#include <algorithm>
#include <array>
#include <cstddef>
+#include <cstring>
+#include <optional>
#include <vector>
+
#include <glad/glad.h>
#include "common/logging/log.h"
@@ -17,6 +20,30 @@ namespace OpenGL {
namespace {
+// One uniform block is reserved for emulation purposes
+constexpr u32 ReservedUniformBlocks = 1;
+
+constexpr u32 NumStages = 5;
+
+constexpr std::array LimitUBOs = {GL_MAX_VERTEX_UNIFORM_BLOCKS, GL_MAX_TESS_CONTROL_UNIFORM_BLOCKS,
+ GL_MAX_TESS_EVALUATION_UNIFORM_BLOCKS,
+ GL_MAX_GEOMETRY_UNIFORM_BLOCKS, GL_MAX_FRAGMENT_UNIFORM_BLOCKS};
+
+constexpr std::array LimitSSBOs = {
+ GL_MAX_VERTEX_SHADER_STORAGE_BLOCKS, GL_MAX_TESS_CONTROL_SHADER_STORAGE_BLOCKS,
+ GL_MAX_TESS_EVALUATION_SHADER_STORAGE_BLOCKS, GL_MAX_GEOMETRY_SHADER_STORAGE_BLOCKS,
+ GL_MAX_FRAGMENT_SHADER_STORAGE_BLOCKS};
+
+constexpr std::array LimitSamplers = {
+ GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS, GL_MAX_TESS_CONTROL_TEXTURE_IMAGE_UNITS,
+ GL_MAX_TESS_EVALUATION_TEXTURE_IMAGE_UNITS, GL_MAX_GEOMETRY_TEXTURE_IMAGE_UNITS,
+ GL_MAX_TEXTURE_IMAGE_UNITS};
+
+constexpr std::array LimitImages = {GL_MAX_VERTEX_IMAGE_UNIFORMS,
+ GL_MAX_TESS_CONTROL_IMAGE_UNIFORMS,
+ GL_MAX_TESS_EVALUATION_IMAGE_UNIFORMS,
+ GL_MAX_GEOMETRY_IMAGE_UNIFORMS, GL_MAX_FRAGMENT_IMAGE_UNIFORMS};
+
template <typename T>
T GetInteger(GLenum pname) {
GLint temporary;
@@ -48,13 +75,73 @@ bool HasExtension(const std::vector<std::string_view>& images, std::string_view
return std::find(images.begin(), images.end(), extension) != images.end();
}
+u32 Extract(u32& base, u32& num, u32 amount, std::optional<GLenum> limit = {}) {
+ ASSERT(num >= amount);
+ if (limit) {
+ amount = std::min(amount, GetInteger<u32>(*limit));
+ }
+ num -= amount;
+ return std::exchange(base, base + amount);
+}
+
+std::array<Device::BaseBindings, Tegra::Engines::MaxShaderTypes> BuildBaseBindings() noexcept {
+ std::array<Device::BaseBindings, Tegra::Engines::MaxShaderTypes> bindings;
+
+ static std::array<std::size_t, 5> stage_swizzle = {0, 1, 2, 3, 4};
+ const u32 total_ubos = GetInteger<u32>(GL_MAX_UNIFORM_BUFFER_BINDINGS);
+ const u32 total_ssbos = GetInteger<u32>(GL_MAX_SHADER_STORAGE_BUFFER_BINDINGS);
+ const u32 total_samplers = GetInteger<u32>(GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS);
+
+ u32 num_ubos = total_ubos - ReservedUniformBlocks;
+ u32 num_ssbos = total_ssbos;
+ u32 num_samplers = total_samplers;
+
+ u32 base_ubo = ReservedUniformBlocks;
+ u32 base_ssbo = 0;
+ u32 base_samplers = 0;
+
+ for (std::size_t i = 0; i < NumStages; ++i) {
+ const std::size_t stage = stage_swizzle[i];
+ bindings[stage] = {
+ Extract(base_ubo, num_ubos, total_ubos / NumStages, LimitUBOs[stage]),
+ Extract(base_ssbo, num_ssbos, total_ssbos / NumStages, LimitSSBOs[stage]),
+ Extract(base_samplers, num_samplers, total_samplers / NumStages, LimitSamplers[stage])};
+ }
+
+ u32 num_images = GetInteger<u32>(GL_MAX_IMAGE_UNITS);
+ u32 base_images = 0;
+
+ // Reserve more image bindings on fragment and vertex stages.
+ bindings[4].image =
+ Extract(base_images, num_images, num_images / NumStages + 2, LimitImages[4]);
+ bindings[0].image =
+ Extract(base_images, num_images, num_images / NumStages + 1, LimitImages[0]);
+
+ // Reserve the other image bindings.
+ const u32 total_extracted_images = num_images / (NumStages - 2);
+ for (std::size_t i = 2; i < NumStages; ++i) {
+ const std::size_t stage = stage_swizzle[i];
+ bindings[stage].image =
+ Extract(base_images, num_images, total_extracted_images, LimitImages[stage]);
+ }
+
+ // Compute doesn't care about any of this.
+ bindings[5] = {0, 0, 0, 0};
+
+ return bindings;
+}
+
} // Anonymous namespace
-Device::Device() {
+Device::Device() : base_bindings{BuildBaseBindings()} {
const std::string_view vendor = reinterpret_cast<const char*>(glGetString(GL_VENDOR));
+ const auto renderer = reinterpret_cast<const char*>(glGetString(GL_RENDERER));
const std::vector extensions = GetExtensions();
const bool is_nvidia = vendor == "NVIDIA Corporation";
+ const bool is_amd = vendor == "ATI Technologies Inc.";
+ const bool is_intel = vendor == "Intel";
+ const bool is_intel_proprietary = is_intel && std::strstr(renderer, "Mesa") == nullptr;
uniform_buffer_alignment = GetInteger<std::size_t>(GL_UNIFORM_BUFFER_OFFSET_ALIGNMENT);
shader_storage_alignment = GetInteger<std::size_t>(GL_SHADER_STORAGE_BUFFER_OFFSET_ALIGNMENT);
@@ -62,11 +149,13 @@ Device::Device() {
max_varyings = GetInteger<u32>(GL_MAX_VARYING_VECTORS);
has_warp_intrinsics = GLAD_GL_NV_gpu_shader5 && GLAD_GL_NV_shader_thread_group &&
GLAD_GL_NV_shader_thread_shuffle;
+ has_shader_ballot = GLAD_GL_ARB_shader_ballot;
has_vertex_viewport_layer = GLAD_GL_ARB_shader_viewport_layer_array;
has_image_load_formatted = HasExtension(extensions, "GL_EXT_shader_image_load_formatted");
has_variable_aoffi = TestVariableAoffi();
- has_component_indexing_bug = TestComponentIndexingBug();
+ has_component_indexing_bug = is_amd;
has_precise_bug = TestPreciseBug();
+ has_broken_compute = is_intel_proprietary;
has_fast_buffer_sub_data = is_nvidia;
LOG_INFO(Render_OpenGL, "Renderer_VariableAOFFI: {}", has_variable_aoffi);
@@ -79,10 +168,12 @@ Device::Device(std::nullptr_t) {
max_vertex_attributes = 16;
max_varyings = 15;
has_warp_intrinsics = true;
+ has_shader_ballot = true;
has_vertex_viewport_layer = true;
has_image_load_formatted = true;
has_variable_aoffi = true;
has_component_indexing_bug = false;
+ has_broken_compute = false;
has_precise_bug = false;
}
@@ -97,52 +188,6 @@ void main() {
})");
}
-bool Device::TestComponentIndexingBug() {
- const GLchar* COMPONENT_TEST = R"(#version 430 core
-layout (std430, binding = 0) buffer OutputBuffer {
- uint output_value;
-};
-layout (std140, binding = 0) uniform InputBuffer {
- uvec4 input_value[4096];
-};
-layout (location = 0) uniform uint idx;
-void main() {
- output_value = input_value[idx >> 2][idx & 3];
-})";
- const GLuint shader{glCreateShaderProgramv(GL_VERTEX_SHADER, 1, &COMPONENT_TEST)};
- SCOPE_EXIT({ glDeleteProgram(shader); });
- glUseProgram(shader);
-
- OGLVertexArray vao;
- vao.Create();
- glBindVertexArray(vao.handle);
-
- constexpr std::array<GLuint, 8> values{0, 0, 0, 0, 0x1236327, 0x985482, 0x872753, 0x2378432};
- OGLBuffer ubo;
- ubo.Create();
- glNamedBufferData(ubo.handle, sizeof(values), values.data(), GL_STATIC_DRAW);
- glBindBufferBase(GL_UNIFORM_BUFFER, 0, ubo.handle);
-
- OGLBuffer ssbo;
- ssbo.Create();
- glNamedBufferStorage(ssbo.handle, sizeof(GLuint), nullptr, GL_CLIENT_STORAGE_BIT);
-
- for (GLuint index = 4; index < 8; ++index) {
- glInvalidateBufferData(ssbo.handle);
- glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 0, ssbo.handle);
-
- glProgramUniform1ui(shader, 0, index);
- glDrawArrays(GL_POINTS, 0, 1);
-
- GLuint result;
- glGetNamedBufferSubData(ssbo.handle, 0, sizeof(result), &result);
- if (result != values.at(index)) {
- return true;
- }
- }
- return false;
-}
-
bool Device::TestPreciseBug() {
return !TestProgram(R"(#version 430 core
in vec3 coords;
diff --git a/src/video_core/renderer_opengl/gl_device.h b/src/video_core/renderer_opengl/gl_device.h
index bf35bd0b6..d73b099d0 100644
--- a/src/video_core/renderer_opengl/gl_device.h
+++ b/src/video_core/renderer_opengl/gl_device.h
@@ -6,14 +6,32 @@
#include <cstddef>
#include "common/common_types.h"
+#include "video_core/engines/shader_type.h"
namespace OpenGL {
-class Device {
+static constexpr u32 EmulationUniformBlockBinding = 0;
+
+class Device final {
public:
+ struct BaseBindings final {
+ u32 uniform_buffer{};
+ u32 shader_storage_buffer{};
+ u32 sampler{};
+ u32 image{};
+ };
+
explicit Device();
explicit Device(std::nullptr_t);
+ const BaseBindings& GetBaseBindings(std::size_t stage_index) const noexcept {
+ return base_bindings[stage_index];
+ }
+
+ const BaseBindings& GetBaseBindings(Tegra::Engines::ShaderType shader_type) const noexcept {
+ return GetBaseBindings(static_cast<std::size_t>(shader_type));
+ }
+
std::size_t GetUniformBufferAlignment() const {
return uniform_buffer_alignment;
}
@@ -34,6 +52,10 @@ public:
return has_warp_intrinsics;
}
+ bool HasShaderBallot() const {
+ return has_shader_ballot;
+ }
+
bool HasVertexViewportLayer() const {
return has_vertex_viewport_layer;
}
@@ -54,25 +76,31 @@ public:
return has_precise_bug;
}
+ bool HasBrokenCompute() const {
+ return has_broken_compute;
+ }
+
bool HasFastBufferSubData() const {
return has_fast_buffer_sub_data;
}
private:
static bool TestVariableAoffi();
- static bool TestComponentIndexingBug();
static bool TestPreciseBug();
+ std::array<BaseBindings, Tegra::Engines::MaxShaderTypes> base_bindings;
std::size_t uniform_buffer_alignment{};
std::size_t shader_storage_alignment{};
u32 max_vertex_attributes{};
u32 max_varyings{};
bool has_warp_intrinsics{};
+ bool has_shader_ballot{};
bool has_vertex_viewport_layer{};
bool has_image_load_formatted{};
bool has_variable_aoffi{};
bool has_component_indexing_bug{};
bool has_precise_bug{};
+ bool has_broken_compute{};
bool has_fast_buffer_sub_data{};
};
diff --git a/src/video_core/renderer_opengl/gl_framebuffer_cache.cpp b/src/video_core/renderer_opengl/gl_framebuffer_cache.cpp
index a5d69d78d..874ed3c6e 100644
--- a/src/video_core/renderer_opengl/gl_framebuffer_cache.cpp
+++ b/src/video_core/renderer_opengl/gl_framebuffer_cache.cpp
@@ -3,9 +3,12 @@
// Refer to the license.txt file included.
#include <tuple>
+#include <unordered_map>
+#include <utility>
-#include "common/cityhash.h"
-#include "common/scope_exit.h"
+#include <glad/glad.h>
+
+#include "common/common_types.h"
#include "video_core/engines/maxwell_3d.h"
#include "video_core/renderer_opengl/gl_framebuffer_cache.h"
#include "video_core/renderer_opengl/gl_state.h"
@@ -13,6 +16,7 @@
namespace OpenGL {
using Maxwell = Tegra::Engines::Maxwell3D::Regs;
+using VideoCore::Surface::SurfaceType;
FramebufferCacheOpenGL::FramebufferCacheOpenGL() = default;
@@ -35,36 +39,49 @@ OGLFramebuffer FramebufferCacheOpenGL::CreateFramebuffer(const FramebufferCacheK
local_state.draw.draw_framebuffer = framebuffer.handle;
local_state.ApplyFramebufferState();
+ if (key.zeta) {
+ const bool stencil = key.zeta->GetSurfaceParams().type == SurfaceType::DepthStencil;
+ const GLenum attach_target = stencil ? GL_DEPTH_STENCIL_ATTACHMENT : GL_DEPTH_ATTACHMENT;
+ key.zeta->Attach(attach_target, GL_DRAW_FRAMEBUFFER);
+ }
+
+ std::size_t num_buffers = 0;
+ std::array<GLenum, Maxwell::NumRenderTargets> targets;
+
for (std::size_t index = 0; index < Maxwell::NumRenderTargets; ++index) {
- if (key.colors[index]) {
- key.colors[index]->Attach(GL_COLOR_ATTACHMENT0 + static_cast<GLenum>(index),
- GL_DRAW_FRAMEBUFFER);
+ if (!key.colors[index]) {
+ targets[index] = GL_NONE;
+ continue;
}
+ const GLenum attach_target = GL_COLOR_ATTACHMENT0 + static_cast<GLenum>(index);
+ key.colors[index]->Attach(attach_target, GL_DRAW_FRAMEBUFFER);
+
+ const u32 attachment = (key.color_attachments >> (BitsPerAttachment * index)) & 0b1111;
+ targets[index] = GL_COLOR_ATTACHMENT0 + attachment;
+ num_buffers = index + 1;
}
- if (key.colors_count) {
- glDrawBuffers(key.colors_count, key.color_attachments.data());
+
+ if (num_buffers > 0) {
+ glDrawBuffers(static_cast<GLsizei>(num_buffers), std::data(targets));
} else {
glDrawBuffer(GL_NONE);
}
- if (key.zeta) {
- key.zeta->Attach(key.stencil_enable ? GL_DEPTH_STENCIL_ATTACHMENT : GL_DEPTH_ATTACHMENT,
- GL_DRAW_FRAMEBUFFER);
- }
-
return framebuffer;
}
-std::size_t FramebufferCacheKey::Hash() const {
- static_assert(sizeof(*this) % sizeof(u64) == 0, "Unaligned struct");
- return static_cast<std::size_t>(
- Common::CityHash64(reinterpret_cast<const char*>(this), sizeof(*this)));
+std::size_t FramebufferCacheKey::Hash() const noexcept {
+ std::size_t hash = std::hash<View>{}(zeta);
+ for (const auto& color : colors) {
+ hash ^= std::hash<View>{}(color);
+ }
+ hash ^= static_cast<std::size_t>(color_attachments) << 16;
+ return hash;
}
-bool FramebufferCacheKey::operator==(const FramebufferCacheKey& rhs) const {
- return std::tie(stencil_enable, colors_count, color_attachments, colors, zeta) ==
- std::tie(rhs.stencil_enable, rhs.colors_count, rhs.color_attachments, rhs.colors,
- rhs.zeta);
+bool FramebufferCacheKey::operator==(const FramebufferCacheKey& rhs) const noexcept {
+ return std::tie(colors, zeta, color_attachments) ==
+ std::tie(rhs.colors, rhs.zeta, rhs.color_attachments);
}
} // namespace OpenGL
diff --git a/src/video_core/renderer_opengl/gl_framebuffer_cache.h b/src/video_core/renderer_opengl/gl_framebuffer_cache.h
index 424344c48..02ec80ae9 100644
--- a/src/video_core/renderer_opengl/gl_framebuffer_cache.h
+++ b/src/video_core/renderer_opengl/gl_framebuffer_cache.h
@@ -18,21 +18,24 @@
namespace OpenGL {
-struct alignas(sizeof(u64)) FramebufferCacheKey {
- bool stencil_enable = false;
- u16 colors_count = 0;
+constexpr std::size_t BitsPerAttachment = 4;
- std::array<GLenum, Tegra::Engines::Maxwell3D::Regs::NumRenderTargets> color_attachments{};
- std::array<View, Tegra::Engines::Maxwell3D::Regs::NumRenderTargets> colors;
+struct FramebufferCacheKey {
View zeta;
+ std::array<View, Tegra::Engines::Maxwell3D::Regs::NumRenderTargets> colors;
+ u32 color_attachments = 0;
- std::size_t Hash() const;
+ std::size_t Hash() const noexcept;
- bool operator==(const FramebufferCacheKey& rhs) const;
+ bool operator==(const FramebufferCacheKey& rhs) const noexcept;
- bool operator!=(const FramebufferCacheKey& rhs) const {
+ bool operator!=(const FramebufferCacheKey& rhs) const noexcept {
return !operator==(rhs);
}
+
+ void SetAttachment(std::size_t index, u32 attachment) {
+ color_attachments |= attachment << (BitsPerAttachment * index);
+ }
};
} // namespace OpenGL
diff --git a/src/video_core/renderer_opengl/gl_rasterizer.cpp b/src/video_core/renderer_opengl/gl_rasterizer.cpp
index e560d70d5..672051102 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer.cpp
+++ b/src/video_core/renderer_opengl/gl_rasterizer.cpp
@@ -19,9 +19,11 @@
#include "common/scope_exit.h"
#include "core/core.h"
#include "core/hle/kernel/process.h"
+#include "core/memory.h"
#include "core/settings.h"
#include "video_core/engines/kepler_compute.h"
#include "video_core/engines/maxwell_3d.h"
+#include "video_core/engines/shader_type.h"
#include "video_core/memory_manager.h"
#include "video_core/renderer_opengl/gl_rasterizer.h"
#include "video_core/renderer_opengl/gl_shader_cache.h"
@@ -49,8 +51,25 @@ MICROPROFILE_DEFINE(OpenGL_Blits, "OpenGL", "Blits", MP_RGB(128, 128, 192));
MICROPROFILE_DEFINE(OpenGL_CacheManagement, "OpenGL", "Cache Mgmt", MP_RGB(100, 255, 100));
MICROPROFILE_DEFINE(OpenGL_PrimitiveAssembly, "OpenGL", "Prim Asmbl", MP_RGB(255, 100, 100));
-static std::size_t GetConstBufferSize(const Tegra::Engines::ConstBufferInfo& buffer,
- const GLShader::ConstBufferEntry& entry) {
+namespace {
+
+template <typename Engine, typename Entry>
+Tegra::Texture::FullTextureInfo GetTextureInfo(const Engine& engine, const Entry& entry,
+ Tegra::Engines::ShaderType shader_type) {
+ if (entry.IsBindless()) {
+ const Tegra::Texture::TextureHandle tex_handle =
+ engine.AccessConstBuffer32(shader_type, entry.GetBuffer(), entry.GetOffset());
+ return engine.GetTextureInfo(tex_handle);
+ }
+ if constexpr (std::is_same_v<Engine, Tegra::Engines::Maxwell3D>) {
+ return engine.GetStageTexture(shader_type, entry.GetOffset());
+ } else {
+ return engine.GetTexture(entry.GetOffset());
+ }
+}
+
+std::size_t GetConstBufferSize(const Tegra::Engines::ConstBufferInfo& buffer,
+ const GLShader::ConstBufferEntry& entry) {
if (!entry.IsIndirect()) {
return entry.GetSize();
}
@@ -64,14 +83,16 @@ static std::size_t GetConstBufferSize(const Tegra::Engines::ConstBufferInfo& buf
return buffer.size;
}
+} // Anonymous namespace
+
RasterizerOpenGL::RasterizerOpenGL(Core::System& system, Core::Frontend::EmuWindow& emu_window,
ScreenInfo& info)
- : texture_cache{system, *this, device}, shader_cache{*this, system, emu_window, device},
- system{system}, screen_info{info}, buffer_cache{*this, system, device, STREAM_BUFFER_SIZE} {
+ : RasterizerAccelerated{system.Memory()}, texture_cache{system, *this, device},
+ shader_cache{*this, system, emu_window, device}, system{system}, screen_info{info},
+ buffer_cache{*this, system, device, STREAM_BUFFER_SIZE} {
shader_program_manager = std::make_unique<GLShader::ProgramManager>();
state.draw.shader_program = 0;
state.Apply();
- clear_framebuffer.Create();
LOG_DEBUG(Render_OpenGL, "Sync fixed function OpenGL state here");
CheckExtensions();
@@ -238,12 +259,11 @@ void RasterizerOpenGL::SetupShaders(GLenum primitive_mode) {
MICROPROFILE_SCOPE(OpenGL_Shader);
auto& gpu = system.GPU().Maxwell3D();
- BaseBindings base_bindings;
std::array<bool, Maxwell::NumClipDistances> clip_distances{};
for (std::size_t index = 0; index < Maxwell::MaxShaderProgram; ++index) {
const auto& shader_config = gpu.regs.shader_config[index];
- const Maxwell::ShaderProgram program{static_cast<Maxwell::ShaderProgram>(index)};
+ const auto program{static_cast<Maxwell::ShaderProgram>(index)};
// Skip stages that are not enabled
if (!gpu.regs.IsShaderConfigEnabled(index)) {
@@ -251,31 +271,34 @@ void RasterizerOpenGL::SetupShaders(GLenum primitive_mode) {
case Maxwell::ShaderProgram::Geometry:
shader_program_manager->UseTrivialGeometryShader();
break;
+ case Maxwell::ShaderProgram::Fragment:
+ shader_program_manager->UseTrivialFragmentShader();
+ break;
default:
break;
}
continue;
}
- const std::size_t stage{index == 0 ? 0 : index - 1}; // Stage indices are 0 - 5
-
- GLShader::MaxwellUniformData ubo{};
- ubo.SetFromRegs(gpu, stage);
- const auto [buffer, offset] =
- buffer_cache.UploadHostMemory(&ubo, sizeof(ubo), device.GetUniformBufferAlignment());
-
- // Bind the emulation info buffer
- bind_ubo_pushbuffer.Push(buffer, offset, static_cast<GLsizeiptr>(sizeof(ubo)));
+ // Currently this stages are not supported in the OpenGL backend.
+ // Todo(Blinkhawk): Port tesselation shaders from Vulkan to OpenGL
+ if (program == Maxwell::ShaderProgram::TesselationControl) {
+ continue;
+ } else if (program == Maxwell::ShaderProgram::TesselationEval) {
+ continue;
+ }
Shader shader{shader_cache.GetStageProgram(program)};
- const auto stage_enum = static_cast<Maxwell::ShaderStage>(stage);
- SetupDrawConstBuffers(stage_enum, shader);
- SetupDrawGlobalMemory(stage_enum, shader);
- const auto texture_buffer_usage{SetupDrawTextures(stage_enum, shader, base_bindings)};
+ // Stage indices are 0 - 5
+ const std::size_t stage = index == 0 ? 0 : index - 1;
+ SetupDrawConstBuffers(stage, shader);
+ SetupDrawGlobalMemory(stage, shader);
+ SetupDrawTextures(stage, shader);
+ SetupDrawImages(stage, shader);
- const ProgramVariant variant{base_bindings, primitive_mode, texture_buffer_usage};
- const auto [program_handle, next_bindings] = shader->GetProgramHandle(variant);
+ const ProgramVariant variant(primitive_mode);
+ const auto program_handle = shader->GetHandle(variant);
switch (program) {
case Maxwell::ShaderProgram::VertexA:
@@ -304,10 +327,8 @@ void RasterizerOpenGL::SetupShaders(GLenum primitive_mode) {
// When VertexA is enabled, we have dual vertex shaders
if (program == Maxwell::ShaderProgram::VertexA) {
// VertexB was combined with VertexA, so we skip the VertexB iteration
- index++;
+ ++index;
}
-
- base_bindings = next_bindings;
}
SyncClipEnabled(clip_distances);
@@ -362,78 +383,58 @@ void RasterizerOpenGL::ConfigureFramebuffers() {
UNIMPLEMENTED_IF(regs.rt_separate_frag_data == 0);
// Bind the framebuffer surfaces
- FramebufferCacheKey fbkey;
- for (std::size_t index = 0; index < Maxwell::NumRenderTargets; ++index) {
+ FramebufferCacheKey key;
+ const auto colors_count = static_cast<std::size_t>(regs.rt_control.count);
+ for (std::size_t index = 0; index < colors_count; ++index) {
View color_surface{texture_cache.GetColorBufferSurface(index, true)};
-
- if (color_surface) {
- // Assume that a surface will be written to if it is used as a framebuffer, even
- // if the shader doesn't actually write to it.
- texture_cache.MarkColorBufferInUse(index);
+ if (!color_surface) {
+ continue;
}
+ // Assume that a surface will be written to if it is used as a framebuffer, even
+ // if the shader doesn't actually write to it.
+ texture_cache.MarkColorBufferInUse(index);
- fbkey.color_attachments[index] = GL_COLOR_ATTACHMENT0 + regs.rt_control.GetMap(index);
- fbkey.colors[index] = std::move(color_surface);
+ key.SetAttachment(index, regs.rt_control.GetMap(index));
+ key.colors[index] = std::move(color_surface);
}
- fbkey.colors_count = regs.rt_control.count;
if (depth_surface) {
// Assume that a surface will be written to if it is used as a framebuffer, even if
// the shader doesn't actually write to it.
texture_cache.MarkDepthBufferInUse();
-
- fbkey.stencil_enable = depth_surface->GetSurfaceParams().type == SurfaceType::DepthStencil;
- fbkey.zeta = std::move(depth_surface);
+ key.zeta = std::move(depth_surface);
}
texture_cache.GuardRenderTargets(false);
- state.draw.draw_framebuffer = framebuffer_cache.GetFramebuffer(fbkey);
+ state.draw.draw_framebuffer = framebuffer_cache.GetFramebuffer(key);
SyncViewport(state);
}
void RasterizerOpenGL::ConfigureClearFramebuffer(OpenGLState& current_state, bool using_color_fb,
bool using_depth_fb, bool using_stencil_fb) {
+ using VideoCore::Surface::SurfaceType;
+
auto& gpu = system.GPU().Maxwell3D();
const auto& regs = gpu.regs;
texture_cache.GuardRenderTargets(true);
- View color_surface{};
+ View color_surface;
if (using_color_fb) {
color_surface = texture_cache.GetColorBufferSurface(regs.clear_buffers.RT, false);
}
- View depth_surface{};
+ View depth_surface;
if (using_depth_fb || using_stencil_fb) {
depth_surface = texture_cache.GetDepthBufferSurface(false);
}
texture_cache.GuardRenderTargets(false);
- current_state.draw.draw_framebuffer = clear_framebuffer.handle;
- current_state.ApplyFramebufferState();
-
- if (color_surface) {
- color_surface->Attach(GL_COLOR_ATTACHMENT0, GL_DRAW_FRAMEBUFFER);
- } else {
- glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, 0, 0);
- }
+ FramebufferCacheKey key;
+ key.colors[0] = color_surface;
+ key.zeta = depth_surface;
- if (depth_surface) {
- const auto& params = depth_surface->GetSurfaceParams();
- switch (params.type) {
- case VideoCore::Surface::SurfaceType::Depth:
- depth_surface->Attach(GL_DEPTH_ATTACHMENT, GL_DRAW_FRAMEBUFFER);
- glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_STENCIL_ATTACHMENT, GL_TEXTURE_2D, 0, 0);
- break;
- case VideoCore::Surface::SurfaceType::DepthStencil:
- depth_surface->Attach(GL_DEPTH_STENCIL_ATTACHMENT, GL_DRAW_FRAMEBUFFER);
- break;
- default:
- UNIMPLEMENTED();
- }
- } else {
- glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D, 0,
- 0);
- }
+ current_state.draw.draw_framebuffer = framebuffer_cache.GetFramebuffer(key);
+ current_state.ApplyFramebufferState();
}
void RasterizerOpenGL::Clear() {
@@ -516,6 +517,7 @@ void RasterizerOpenGL::Clear() {
ConfigureClearFramebuffer(clear_state, use_color, use_depth, use_stencil);
SyncViewport(clear_state);
+ SyncRasterizeEnable(clear_state);
if (regs.clear_flags.scissor) {
SyncScissorTest(clear_state);
}
@@ -543,6 +545,7 @@ void RasterizerOpenGL::Clear() {
void RasterizerOpenGL::DrawPrelude() {
auto& gpu = system.GPU().Maxwell3D();
+ SyncRasterizeEnable(state);
SyncColorMask();
SyncFragmentColorClampState();
SyncMultiSampleState();
@@ -592,8 +595,16 @@ void RasterizerOpenGL::DrawPrelude() {
index_buffer_offset = SetupIndexBuffer();
// Prepare packed bindings.
- bind_ubo_pushbuffer.Setup(0);
- bind_ssbo_pushbuffer.Setup(0);
+ bind_ubo_pushbuffer.Setup();
+ bind_ssbo_pushbuffer.Setup();
+
+ // Setup emulation uniform buffer.
+ GLShader::MaxwellUniformData ubo;
+ ubo.SetFromRegs(gpu);
+ const auto [buffer, offset] =
+ buffer_cache.UploadHostMemory(&ubo, sizeof(ubo), device.GetUniformBufferAlignment());
+ bind_ubo_pushbuffer.Push(EmulationUniformBlockBinding, buffer, offset,
+ static_cast<GLsizeiptr>(sizeof(ubo)));
// Setup shaders and their used resources.
texture_cache.GuardSamplers(true);
@@ -726,19 +737,21 @@ bool RasterizerOpenGL::DrawMultiBatch(bool is_indexed) {
}
void RasterizerOpenGL::DispatchCompute(GPUVAddr code_addr) {
- if (!GLAD_GL_ARB_compute_variable_group_size) {
- LOG_ERROR(Render_OpenGL, "Compute is currently not supported on this device due to the "
- "lack of GL_ARB_compute_variable_group_size");
+ if (device.HasBrokenCompute()) {
return;
}
+ buffer_cache.Acquire();
+
auto kernel = shader_cache.GetComputeKernel(code_addr);
- ProgramVariant variant;
- variant.texture_buffer_usage = SetupComputeTextures(kernel);
+ SetupComputeTextures(kernel);
SetupComputeImages(kernel);
- const auto [program, next_bindings] = kernel->GetProgramHandle(variant);
- state.draw.shader_program = program;
+ const auto& launch_desc = system.GPU().KeplerCompute().launch_description;
+ const ProgramVariant variant(launch_desc.block_dim_x, launch_desc.block_dim_y,
+ launch_desc.block_dim_z, launch_desc.shared_alloc,
+ launch_desc.local_pos_alloc);
+ state.draw.shader_program = kernel->GetHandle(variant);
state.draw.program_pipeline = 0;
const std::size_t buffer_size =
@@ -746,8 +759,8 @@ void RasterizerOpenGL::DispatchCompute(GPUVAddr code_addr) {
(Maxwell::MaxConstBufferSize + device.GetUniformBufferAlignment());
buffer_cache.Map(buffer_size);
- bind_ubo_pushbuffer.Setup(0);
- bind_ssbo_pushbuffer.Setup(0);
+ bind_ubo_pushbuffer.Setup();
+ bind_ssbo_pushbuffer.Setup();
SetupComputeConstBuffers(kernel);
SetupComputeGlobalMemory(kernel);
@@ -762,10 +775,7 @@ void RasterizerOpenGL::DispatchCompute(GPUVAddr code_addr) {
state.ApplyShaderProgram();
state.ApplyProgramPipeline();
- const auto& launch_desc = system.GPU().KeplerCompute().launch_description;
- glDispatchComputeGroupSizeARB(launch_desc.grid_dim_x, launch_desc.grid_dim_y,
- launch_desc.grid_dim_z, launch_desc.block_dim_x,
- launch_desc.block_dim_y, launch_desc.block_dim_z);
+ glDispatchCompute(launch_desc.grid_dim_x, launch_desc.grid_dim_y, launch_desc.grid_dim_z);
}
void RasterizerOpenGL::FlushAll() {}
@@ -821,7 +831,7 @@ bool RasterizerOpenGL::AccelerateDisplay(const Tegra::FramebufferConfig& config,
MICROPROFILE_SCOPE(OpenGL_CacheManagement);
const auto surface{
- texture_cache.TryFindFramebufferSurface(Memory::GetPointer(framebuffer_addr))};
+ texture_cache.TryFindFramebufferSurface(system.Memory().GetPointer(framebuffer_addr))};
if (!surface) {
return {};
}
@@ -834,7 +844,7 @@ bool RasterizerOpenGL::AccelerateDisplay(const Tegra::FramebufferConfig& config,
ASSERT_MSG(params.height == config.height, "Framebuffer height is different");
if (params.pixel_format != pixel_format) {
- LOG_WARNING(Render_OpenGL, "Framebuffer pixel_format is different");
+ LOG_DEBUG(Render_OpenGL, "Framebuffer pixel_format is different");
}
screen_info.display_texture = surface->GetTexture();
@@ -843,20 +853,23 @@ bool RasterizerOpenGL::AccelerateDisplay(const Tegra::FramebufferConfig& config,
return true;
}
-void RasterizerOpenGL::SetupDrawConstBuffers(Tegra::Engines::Maxwell3D::Regs::ShaderStage stage,
- const Shader& shader) {
+void RasterizerOpenGL::SetupDrawConstBuffers(std::size_t stage_index, const Shader& shader) {
MICROPROFILE_SCOPE(OpenGL_UBO);
const auto& stages = system.GPU().Maxwell3D().state.shader_stages;
- const auto& shader_stage = stages[static_cast<std::size_t>(stage)];
+ const auto& shader_stage = stages[stage_index];
+
+ u32 binding = device.GetBaseBindings(stage_index).uniform_buffer;
for (const auto& entry : shader->GetShaderEntries().const_buffers) {
const auto& buffer = shader_stage.const_buffers[entry.GetIndex()];
- SetupConstBuffer(buffer, entry);
+ SetupConstBuffer(binding++, buffer, entry);
}
}
void RasterizerOpenGL::SetupComputeConstBuffers(const Shader& kernel) {
MICROPROFILE_SCOPE(OpenGL_UBO);
const auto& launch_desc = system.GPU().KeplerCompute().launch_description;
+
+ u32 binding = 0;
for (const auto& entry : kernel->GetShaderEntries().const_buffers) {
const auto& config = launch_desc.const_buffer_config[entry.GetIndex()];
const std::bitset<8> mask = launch_desc.const_buffer_enable_mask.Value();
@@ -864,15 +877,16 @@ void RasterizerOpenGL::SetupComputeConstBuffers(const Shader& kernel) {
buffer.address = config.Address();
buffer.size = config.size;
buffer.enabled = mask[entry.GetIndex()];
- SetupConstBuffer(buffer, entry);
+ SetupConstBuffer(binding++, buffer, entry);
}
}
-void RasterizerOpenGL::SetupConstBuffer(const Tegra::Engines::ConstBufferInfo& buffer,
+void RasterizerOpenGL::SetupConstBuffer(u32 binding, const Tegra::Engines::ConstBufferInfo& buffer,
const GLShader::ConstBufferEntry& entry) {
if (!buffer.enabled) {
// Set values to zero to unbind buffers
- bind_ubo_pushbuffer.Push(buffer_cache.GetEmptyBuffer(sizeof(float)), 0, sizeof(float));
+ bind_ubo_pushbuffer.Push(binding, buffer_cache.GetEmptyBuffer(sizeof(float)), 0,
+ sizeof(float));
return;
}
@@ -883,19 +897,20 @@ void RasterizerOpenGL::SetupConstBuffer(const Tegra::Engines::ConstBufferInfo& b
const auto alignment = device.GetUniformBufferAlignment();
const auto [cbuf, offset] = buffer_cache.UploadMemory(buffer.address, size, alignment, false,
device.HasFastBufferSubData());
- bind_ubo_pushbuffer.Push(cbuf, offset, size);
+ bind_ubo_pushbuffer.Push(binding, cbuf, offset, size);
}
-void RasterizerOpenGL::SetupDrawGlobalMemory(Tegra::Engines::Maxwell3D::Regs::ShaderStage stage,
- const Shader& shader) {
+void RasterizerOpenGL::SetupDrawGlobalMemory(std::size_t stage_index, const Shader& shader) {
auto& gpu{system.GPU()};
auto& memory_manager{gpu.MemoryManager()};
- const auto cbufs{gpu.Maxwell3D().state.shader_stages[static_cast<std::size_t>(stage)]};
+ const auto cbufs{gpu.Maxwell3D().state.shader_stages[stage_index]};
+
+ u32 binding = device.GetBaseBindings(stage_index).shader_storage_buffer;
for (const auto& entry : shader->GetShaderEntries().global_memory_entries) {
const auto addr{cbufs.const_buffers[entry.GetCbufIndex()].address + entry.GetCbufOffset()};
const auto gpu_addr{memory_manager.Read<u64>(addr)};
const auto size{memory_manager.Read<u32>(addr + 8)};
- SetupGlobalMemory(entry, gpu_addr, size);
+ SetupGlobalMemory(binding++, entry, gpu_addr, size);
}
}
@@ -903,120 +918,82 @@ void RasterizerOpenGL::SetupComputeGlobalMemory(const Shader& kernel) {
auto& gpu{system.GPU()};
auto& memory_manager{gpu.MemoryManager()};
const auto cbufs{gpu.KeplerCompute().launch_description.const_buffer_config};
+
+ u32 binding = 0;
for (const auto& entry : kernel->GetShaderEntries().global_memory_entries) {
const auto addr{cbufs[entry.GetCbufIndex()].Address() + entry.GetCbufOffset()};
const auto gpu_addr{memory_manager.Read<u64>(addr)};
const auto size{memory_manager.Read<u32>(addr + 8)};
- SetupGlobalMemory(entry, gpu_addr, size);
+ SetupGlobalMemory(binding++, entry, gpu_addr, size);
}
}
-void RasterizerOpenGL::SetupGlobalMemory(const GLShader::GlobalMemoryEntry& entry,
+void RasterizerOpenGL::SetupGlobalMemory(u32 binding, const GLShader::GlobalMemoryEntry& entry,
GPUVAddr gpu_addr, std::size_t size) {
const auto alignment{device.GetShaderStorageBufferAlignment()};
const auto [ssbo, buffer_offset] =
buffer_cache.UploadMemory(gpu_addr, size, alignment, entry.IsWritten());
- bind_ssbo_pushbuffer.Push(ssbo, buffer_offset, static_cast<GLsizeiptr>(size));
+ bind_ssbo_pushbuffer.Push(binding, ssbo, buffer_offset, static_cast<GLsizeiptr>(size));
}
-TextureBufferUsage RasterizerOpenGL::SetupDrawTextures(Maxwell::ShaderStage stage,
- const Shader& shader,
- BaseBindings base_bindings) {
+void RasterizerOpenGL::SetupDrawTextures(std::size_t stage_index, const Shader& shader) {
MICROPROFILE_SCOPE(OpenGL_Texture);
- const auto& gpu = system.GPU();
- const auto& maxwell3d = gpu.Maxwell3D();
- const auto& entries = shader->GetShaderEntries().samplers;
-
- ASSERT_MSG(base_bindings.sampler + entries.size() <= std::size(state.textures),
- "Exceeded the number of active textures.");
-
- TextureBufferUsage texture_buffer_usage{0};
-
- for (u32 bindpoint = 0; bindpoint < entries.size(); ++bindpoint) {
- const auto& entry = entries[bindpoint];
- const auto texture = [&] {
- if (!entry.IsBindless()) {
- return maxwell3d.GetStageTexture(stage, entry.GetOffset());
- }
- const auto shader_type = static_cast<Tegra::Engines::ShaderType>(stage);
- const Tegra::Texture::TextureHandle tex_handle =
- maxwell3d.AccessConstBuffer32(shader_type, entry.GetBuffer(), entry.GetOffset());
- return maxwell3d.GetTextureInfo(tex_handle);
- }();
-
- if (SetupTexture(base_bindings.sampler + bindpoint, texture, entry)) {
- texture_buffer_usage.set(bindpoint);
- }
+ const auto& maxwell3d = system.GPU().Maxwell3D();
+ u32 binding = device.GetBaseBindings(stage_index).sampler;
+ for (const auto& entry : shader->GetShaderEntries().samplers) {
+ const auto shader_type = static_cast<Tegra::Engines::ShaderType>(stage_index);
+ const auto texture = GetTextureInfo(maxwell3d, entry, shader_type);
+ SetupTexture(binding++, texture, entry);
}
-
- return texture_buffer_usage;
}
-TextureBufferUsage RasterizerOpenGL::SetupComputeTextures(const Shader& kernel) {
+void RasterizerOpenGL::SetupComputeTextures(const Shader& kernel) {
MICROPROFILE_SCOPE(OpenGL_Texture);
const auto& compute = system.GPU().KeplerCompute();
- const auto& entries = kernel->GetShaderEntries().samplers;
-
- ASSERT_MSG(entries.size() <= std::size(state.textures),
- "Exceeded the number of active textures.");
-
- TextureBufferUsage texture_buffer_usage{0};
-
- for (u32 bindpoint = 0; bindpoint < entries.size(); ++bindpoint) {
- const auto& entry = entries[bindpoint];
- const auto texture = [&] {
- if (!entry.IsBindless()) {
- return compute.GetTexture(entry.GetOffset());
- }
- const Tegra::Texture::TextureHandle tex_handle = compute.AccessConstBuffer32(
- Tegra::Engines::ShaderType::Compute, entry.GetBuffer(), entry.GetOffset());
- return compute.GetTextureInfo(tex_handle);
- }();
-
- if (SetupTexture(bindpoint, texture, entry)) {
- texture_buffer_usage.set(bindpoint);
- }
+ u32 binding = 0;
+ for (const auto& entry : kernel->GetShaderEntries().samplers) {
+ const auto texture = GetTextureInfo(compute, entry, Tegra::Engines::ShaderType::Compute);
+ SetupTexture(binding++, texture, entry);
}
-
- return texture_buffer_usage;
}
-bool RasterizerOpenGL::SetupTexture(u32 binding, const Tegra::Texture::FullTextureInfo& texture,
+void RasterizerOpenGL::SetupTexture(u32 binding, const Tegra::Texture::FullTextureInfo& texture,
const GLShader::SamplerEntry& entry) {
- state.samplers[binding] = sampler_cache.GetSampler(texture.tsc);
-
const auto view = texture_cache.GetTextureSurface(texture.tic, entry);
if (!view) {
// Can occur when texture addr is null or its memory is unmapped/invalid
+ state.samplers[binding] = 0;
state.textures[binding] = 0;
- return false;
+ return;
}
state.textures[binding] = view->GetTexture();
if (view->GetSurfaceParams().IsBuffer()) {
- return true;
+ return;
}
+ state.samplers[binding] = sampler_cache.GetSampler(texture.tsc);
// Apply swizzle to textures that are not buffers.
view->ApplySwizzle(texture.tic.x_source, texture.tic.y_source, texture.tic.z_source,
texture.tic.w_source);
- return false;
+}
+
+void RasterizerOpenGL::SetupDrawImages(std::size_t stage_index, const Shader& shader) {
+ const auto& maxwell3d = system.GPU().Maxwell3D();
+ u32 binding = device.GetBaseBindings(stage_index).image;
+ for (const auto& entry : shader->GetShaderEntries().images) {
+ const auto shader_type = static_cast<Tegra::Engines::ShaderType>(stage_index);
+ const auto tic = GetTextureInfo(maxwell3d, entry, shader_type).tic;
+ SetupImage(binding++, tic, entry);
+ }
}
void RasterizerOpenGL::SetupComputeImages(const Shader& shader) {
const auto& compute = system.GPU().KeplerCompute();
- const auto& entries = shader->GetShaderEntries().images;
- for (u32 bindpoint = 0; bindpoint < entries.size(); ++bindpoint) {
- const auto& entry = entries[bindpoint];
- const auto tic = [&] {
- if (!entry.IsBindless()) {
- return compute.GetTexture(entry.GetOffset()).tic;
- }
- const Tegra::Texture::TextureHandle tex_handle = compute.AccessConstBuffer32(
- Tegra::Engines::ShaderType::Compute, entry.GetBuffer(), entry.GetOffset());
- return compute.GetTextureInfo(tex_handle).tic;
- }();
- SetupImage(bindpoint, tic, entry);
+ u32 binding = 0;
+ for (const auto& entry : shader->GetShaderEntries().images) {
+ const auto tic = GetTextureInfo(compute, entry, Tegra::Engines::ShaderType::Compute).tic;
+ SetupImage(binding++, tic, entry);
}
}
@@ -1055,6 +1032,19 @@ void RasterizerOpenGL::SyncViewport(OpenGLState& current_state) {
}
state.depth_clamp.far_plane = regs.view_volume_clip_control.depth_clamp_far != 0;
state.depth_clamp.near_plane = regs.view_volume_clip_control.depth_clamp_near != 0;
+
+ bool flip_y = false;
+ if (regs.viewport_transform[0].scale_y < 0.0) {
+ flip_y = !flip_y;
+ }
+ if (regs.screen_y_control.y_negate != 0) {
+ flip_y = !flip_y;
+ }
+ state.clip_control.origin = flip_y ? GL_UPPER_LEFT : GL_LOWER_LEFT;
+ state.clip_control.depth_mode =
+ regs.depth_mode == Tegra::Engines::Maxwell3D::Regs::DepthMode::ZeroToOne
+ ? GL_ZERO_TO_ONE
+ : GL_NEGATIVE_ONE_TO_ONE;
}
void RasterizerOpenGL::SyncClipEnabled(
@@ -1077,28 +1067,14 @@ void RasterizerOpenGL::SyncClipCoef() {
}
void RasterizerOpenGL::SyncCullMode() {
- auto& maxwell3d = system.GPU().Maxwell3D();
-
- const auto& regs = maxwell3d.regs;
+ const auto& regs = system.GPU().Maxwell3D().regs;
state.cull.enabled = regs.cull.enabled != 0;
if (state.cull.enabled) {
- state.cull.front_face = MaxwellToGL::FrontFace(regs.cull.front_face);
state.cull.mode = MaxwellToGL::CullFace(regs.cull.cull_face);
-
- const bool flip_triangles{regs.screen_y_control.triangle_rast_flip == 0 ||
- regs.viewport_transform[0].scale_y < 0.0f};
-
- // If the GPU is configured to flip the rasterized triangles, then we need to flip the
- // notion of front and back. Note: We flip the triangles when the value of the register is 0
- // because OpenGL already does it for us.
- if (flip_triangles) {
- if (state.cull.front_face == GL_CCW)
- state.cull.front_face = GL_CW;
- else if (state.cull.front_face == GL_CW)
- state.cull.front_face = GL_CCW;
- }
}
+
+ state.cull.front_face = MaxwellToGL::FrontFace(regs.cull.front_face);
}
void RasterizerOpenGL::SyncPrimitiveRestart() {
@@ -1162,6 +1138,11 @@ void RasterizerOpenGL::SyncStencilTestState() {
}
}
+void RasterizerOpenGL::SyncRasterizeEnable(OpenGLState& current_state) {
+ const auto& regs = system.GPU().Maxwell3D().regs;
+ current_state.rasterizer_discard = regs.rasterize_enable == 0;
+}
+
void RasterizerOpenGL::SyncColorMask() {
auto& maxwell3d = system.GPU().Maxwell3D();
if (!maxwell3d.dirty.color_mask) {
diff --git a/src/video_core/renderer_opengl/gl_rasterizer.h b/src/video_core/renderer_opengl/gl_rasterizer.h
index bd6fe5c3a..6a27cf497 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer.h
+++ b/src/video_core/renderer_opengl/gl_rasterizer.h
@@ -83,42 +83,41 @@ private:
bool using_depth_fb, bool using_stencil_fb);
/// Configures the current constbuffers to use for the draw command.
- void SetupDrawConstBuffers(Tegra::Engines::Maxwell3D::Regs::ShaderStage stage,
- const Shader& shader);
+ void SetupDrawConstBuffers(std::size_t stage_index, const Shader& shader);
/// Configures the current constbuffers to use for the kernel invocation.
void SetupComputeConstBuffers(const Shader& kernel);
/// Configures a constant buffer.
- void SetupConstBuffer(const Tegra::Engines::ConstBufferInfo& buffer,
+ void SetupConstBuffer(u32 binding, const Tegra::Engines::ConstBufferInfo& buffer,
const GLShader::ConstBufferEntry& entry);
/// Configures the current global memory entries to use for the draw command.
- void SetupDrawGlobalMemory(Tegra::Engines::Maxwell3D::Regs::ShaderStage stage,
- const Shader& shader);
+ void SetupDrawGlobalMemory(std::size_t stage_index, const Shader& shader);
/// Configures the current global memory entries to use for the kernel invocation.
void SetupComputeGlobalMemory(const Shader& kernel);
/// Configures a constant buffer.
- void SetupGlobalMemory(const GLShader::GlobalMemoryEntry& entry, GPUVAddr gpu_addr,
+ void SetupGlobalMemory(u32 binding, const GLShader::GlobalMemoryEntry& entry, GPUVAddr gpu_addr,
std::size_t size);
/// Syncs all the state, shaders, render targets and textures setting before a draw call.
void DrawPrelude();
- /// Configures the current textures to use for the draw command. Returns shaders texture buffer
- /// usage.
- TextureBufferUsage SetupDrawTextures(Tegra::Engines::Maxwell3D::Regs::ShaderStage stage,
- const Shader& shader, BaseBindings base_bindings);
+ /// Configures the current textures to use for the draw command.
+ void SetupDrawTextures(std::size_t stage_index, const Shader& shader);
- /// Configures the textures used in a compute shader. Returns texture buffer usage.
- TextureBufferUsage SetupComputeTextures(const Shader& kernel);
+ /// Configures the textures used in a compute shader.
+ void SetupComputeTextures(const Shader& kernel);
- /// Configures a texture. Returns true when the texture is a texture buffer.
- bool SetupTexture(u32 binding, const Tegra::Texture::FullTextureInfo& texture,
+ /// Configures a texture.
+ void SetupTexture(u32 binding, const Tegra::Texture::FullTextureInfo& texture,
const GLShader::SamplerEntry& entry);
+ /// Configures images in a graphics shader.
+ void SetupDrawImages(std::size_t stage_index, const Shader& shader);
+
/// Configures images in a compute shader.
void SetupComputeImages(const Shader& shader);
@@ -169,6 +168,9 @@ private:
/// Syncs the point state to match the guest state
void SyncPointState();
+ /// Syncs the rasterizer enable state to match the guest state
+ void SyncRasterizeEnable(OpenGLState& current_state);
+
/// Syncs Color Mask
void SyncColorMask();
@@ -224,8 +226,6 @@ private:
enum class AccelDraw { Disabled, Arrays, Indexed };
AccelDraw accelerate_draw = AccelDraw::Disabled;
-
- OGLFramebuffer clear_framebuffer;
};
} // namespace OpenGL
diff --git a/src/video_core/renderer_opengl/gl_shader_cache.cpp b/src/video_core/renderer_opengl/gl_shader_cache.cpp
index f1b89165d..de742d11c 100644
--- a/src/video_core/renderer_opengl/gl_shader_cache.cpp
+++ b/src/video_core/renderer_opengl/gl_shader_cache.cpp
@@ -8,12 +8,15 @@
#include <thread>
#include <unordered_set>
#include <boost/functional/hash.hpp>
+#include "common/alignment.h"
#include "common/assert.h"
+#include "common/logging/log.h"
#include "common/scope_exit.h"
#include "core/core.h"
#include "core/frontend/emu_window.h"
#include "video_core/engines/kepler_compute.h"
#include "video_core/engines/maxwell_3d.h"
+#include "video_core/engines/shader_type.h"
#include "video_core/memory_manager.h"
#include "video_core/renderer_opengl/gl_rasterizer.h"
#include "video_core/renderer_opengl/gl_shader_cache.h"
@@ -82,28 +85,26 @@ std::size_t CalculateProgramSize(const GLShader::ProgramCode& program) {
/// Gets the shader program code from memory for the specified address
ProgramCode GetShaderCode(Tegra::MemoryManager& memory_manager, const GPUVAddr gpu_addr,
const u8* host_ptr) {
- ProgramCode program_code(VideoCommon::Shader::MAX_PROGRAM_LENGTH);
+ ProgramCode code(VideoCommon::Shader::MAX_PROGRAM_LENGTH);
ASSERT_OR_EXECUTE(host_ptr != nullptr, {
- std::fill(program_code.begin(), program_code.end(), 0);
- return program_code;
+ std::fill(code.begin(), code.end(), 0);
+ return code;
});
- memory_manager.ReadBlockUnsafe(gpu_addr, program_code.data(),
- program_code.size() * sizeof(u64));
- program_code.resize(CalculateProgramSize(program_code));
- return program_code;
+ memory_manager.ReadBlockUnsafe(gpu_addr, code.data(), code.size() * sizeof(u64));
+ code.resize(CalculateProgramSize(code));
+ return code;
}
/// Gets the shader type from a Maxwell program type
-constexpr GLenum GetShaderType(ProgramType program_type) {
- switch (program_type) {
- case ProgramType::VertexA:
- case ProgramType::VertexB:
+constexpr GLenum GetGLShaderType(ShaderType shader_type) {
+ switch (shader_type) {
+ case ShaderType::Vertex:
return GL_VERTEX_SHADER;
- case ProgramType::Geometry:
+ case ShaderType::Geometry:
return GL_GEOMETRY_SHADER;
- case ProgramType::Fragment:
+ case ShaderType::Fragment:
return GL_FRAGMENT_SHADER;
- case ProgramType::Compute:
+ case ShaderType::Compute:
return GL_COMPUTE_SHADER;
default:
return GL_NONE;
@@ -111,52 +112,33 @@ constexpr GLenum GetShaderType(ProgramType program_type) {
}
/// Describes primitive behavior on geometry shaders
-constexpr std::tuple<const char*, const char*, u32> GetPrimitiveDescription(GLenum primitive_mode) {
+constexpr std::pair<const char*, u32> GetPrimitiveDescription(GLenum primitive_mode) {
switch (primitive_mode) {
case GL_POINTS:
- return {"points", "Points", 1};
+ return {"points", 1};
case GL_LINES:
case GL_LINE_STRIP:
- return {"lines", "Lines", 2};
+ return {"lines", 2};
case GL_LINES_ADJACENCY:
case GL_LINE_STRIP_ADJACENCY:
- return {"lines_adjacency", "LinesAdj", 4};
+ return {"lines_adjacency", 4};
case GL_TRIANGLES:
case GL_TRIANGLE_STRIP:
case GL_TRIANGLE_FAN:
- return {"triangles", "Triangles", 3};
+ return {"triangles", 3};
case GL_TRIANGLES_ADJACENCY:
case GL_TRIANGLE_STRIP_ADJACENCY:
- return {"triangles_adjacency", "TrianglesAdj", 6};
+ return {"triangles_adjacency", 6};
default:
- return {"points", "Invalid", 1};
+ return {"points", 1};
}
}
-ProgramType GetProgramType(Maxwell::ShaderProgram program) {
- switch (program) {
- case Maxwell::ShaderProgram::VertexA:
- return ProgramType::VertexA;
- case Maxwell::ShaderProgram::VertexB:
- return ProgramType::VertexB;
- case Maxwell::ShaderProgram::TesselationControl:
- return ProgramType::TessellationControl;
- case Maxwell::ShaderProgram::TesselationEval:
- return ProgramType::TessellationEval;
- case Maxwell::ShaderProgram::Geometry:
- return ProgramType::Geometry;
- case Maxwell::ShaderProgram::Fragment:
- return ProgramType::Fragment;
- }
- UNREACHABLE();
- return {};
-}
-
/// Hashes one (or two) program streams
-u64 GetUniqueIdentifier(ProgramType program_type, const ProgramCode& code,
+u64 GetUniqueIdentifier(ShaderType shader_type, bool is_a, const ProgramCode& code,
const ProgramCode& code_b) {
u64 unique_identifier = boost::hash_value(code);
- if (program_type == ProgramType::VertexA) {
+ if (is_a) {
// VertexA programs include two programs
boost::hash_combine(unique_identifier, boost::hash_value(code_b));
}
@@ -164,79 +146,74 @@ u64 GetUniqueIdentifier(ProgramType program_type, const ProgramCode& code,
}
/// Creates an unspecialized program from code streams
-std::string GenerateGLSL(const Device& device, ProgramType program_type, const ShaderIR& ir,
+std::string GenerateGLSL(const Device& device, ShaderType shader_type, const ShaderIR& ir,
const std::optional<ShaderIR>& ir_b) {
- switch (program_type) {
- case ProgramType::VertexA:
- case ProgramType::VertexB:
+ switch (shader_type) {
+ case ShaderType::Vertex:
return GLShader::GenerateVertexShader(device, ir, ir_b ? &*ir_b : nullptr);
- case ProgramType::Geometry:
+ case ShaderType::Geometry:
return GLShader::GenerateGeometryShader(device, ir);
- case ProgramType::Fragment:
+ case ShaderType::Fragment:
return GLShader::GenerateFragmentShader(device, ir);
- case ProgramType::Compute:
+ case ShaderType::Compute:
return GLShader::GenerateComputeShader(device, ir);
default:
- UNIMPLEMENTED_MSG("Unimplemented program_type={}", static_cast<u32>(program_type));
+ UNIMPLEMENTED_MSG("Unimplemented shader_type={}", static_cast<u32>(shader_type));
return {};
}
}
-constexpr const char* GetProgramTypeName(ProgramType program_type) {
- switch (program_type) {
- case ProgramType::VertexA:
- case ProgramType::VertexB:
+constexpr const char* GetShaderTypeName(ShaderType shader_type) {
+ switch (shader_type) {
+ case ShaderType::Vertex:
return "VS";
- case ProgramType::TessellationControl:
- return "TCS";
- case ProgramType::TessellationEval:
- return "TES";
- case ProgramType::Geometry:
+ case ShaderType::TesselationControl:
+ return "HS";
+ case ShaderType::TesselationEval:
+ return "DS";
+ case ShaderType::Geometry:
return "GS";
- case ProgramType::Fragment:
+ case ShaderType::Fragment:
return "FS";
- case ProgramType::Compute:
+ case ShaderType::Compute:
return "CS";
}
return "UNK";
}
-Tegra::Engines::ShaderType GetEnginesShaderType(ProgramType program_type) {
+constexpr ShaderType GetShaderType(Maxwell::ShaderProgram program_type) {
switch (program_type) {
- case ProgramType::VertexA:
- case ProgramType::VertexB:
- return Tegra::Engines::ShaderType::Vertex;
- case ProgramType::TessellationControl:
- return Tegra::Engines::ShaderType::TesselationControl;
- case ProgramType::TessellationEval:
- return Tegra::Engines::ShaderType::TesselationEval;
- case ProgramType::Geometry:
- return Tegra::Engines::ShaderType::Geometry;
- case ProgramType::Fragment:
- return Tegra::Engines::ShaderType::Fragment;
- case ProgramType::Compute:
- return Tegra::Engines::ShaderType::Compute;
- }
- UNREACHABLE();
+ case Maxwell::ShaderProgram::VertexA:
+ case Maxwell::ShaderProgram::VertexB:
+ return ShaderType::Vertex;
+ case Maxwell::ShaderProgram::TesselationControl:
+ return ShaderType::TesselationControl;
+ case Maxwell::ShaderProgram::TesselationEval:
+ return ShaderType::TesselationEval;
+ case Maxwell::ShaderProgram::Geometry:
+ return ShaderType::Geometry;
+ case Maxwell::ShaderProgram::Fragment:
+ return ShaderType::Fragment;
+ }
return {};
}
-std::string GetShaderId(u64 unique_identifier, ProgramType program_type) {
- return fmt::format("{}{:016X}", GetProgramTypeName(program_type), unique_identifier);
+std::string GetShaderId(u64 unique_identifier, ShaderType shader_type) {
+ return fmt::format("{}{:016X}", GetShaderTypeName(shader_type), unique_identifier);
}
-Tegra::Engines::ConstBufferEngineInterface& GetConstBufferEngineInterface(
- Core::System& system, ProgramType program_type) {
- if (program_type == ProgramType::Compute) {
+Tegra::Engines::ConstBufferEngineInterface& GetConstBufferEngineInterface(Core::System& system,
+ ShaderType shader_type) {
+ if (shader_type == ShaderType::Compute) {
return system.GPU().KeplerCompute();
} else {
return system.GPU().Maxwell3D();
}
}
-std::unique_ptr<ConstBufferLocker> MakeLocker(Core::System& system, ProgramType program_type) {
- return std::make_unique<ConstBufferLocker>(GetEnginesShaderType(program_type),
- GetConstBufferEngineInterface(system, program_type));
+std::unique_ptr<ConstBufferLocker> MakeLocker(Core::System& system, ShaderType shader_type) {
+ return std::make_unique<ConstBufferLocker>(shader_type,
+ GetConstBufferEngineInterface(system, shader_type));
}
void FillLocker(ConstBufferLocker& locker, const ShaderDiskCacheUsage& usage) {
@@ -253,88 +230,66 @@ void FillLocker(ConstBufferLocker& locker, const ShaderDiskCacheUsage& usage) {
}
}
-CachedProgram BuildShader(const Device& device, u64 unique_identifier, ProgramType program_type,
- const ProgramCode& program_code, const ProgramCode& program_code_b,
- const ProgramVariant& variant, ConstBufferLocker& locker,
+CachedProgram BuildShader(const Device& device, u64 unique_identifier, ShaderType shader_type,
+ const ProgramCode& code, const ProgramCode& code_b,
+ ConstBufferLocker& locker, const ProgramVariant& variant,
bool hint_retrievable = false) {
- LOG_INFO(Render_OpenGL, "called. {}", GetShaderId(unique_identifier, program_type));
+ LOG_INFO(Render_OpenGL, "called. {}", GetShaderId(unique_identifier, shader_type));
- const bool is_compute = program_type == ProgramType::Compute;
+ const bool is_compute = shader_type == ShaderType::Compute;
const u32 main_offset = is_compute ? KERNEL_MAIN_OFFSET : STAGE_MAIN_OFFSET;
- const ShaderIR ir(program_code, main_offset, COMPILER_SETTINGS, locker);
+ const ShaderIR ir(code, main_offset, COMPILER_SETTINGS, locker);
std::optional<ShaderIR> ir_b;
- if (!program_code_b.empty()) {
- ir_b.emplace(program_code_b, main_offset, COMPILER_SETTINGS, locker);
+ if (!code_b.empty()) {
+ ir_b.emplace(code_b, main_offset, COMPILER_SETTINGS, locker);
}
const auto entries = GLShader::GetEntries(ir);
- auto base_bindings{variant.base_bindings};
- const auto primitive_mode{variant.primitive_mode};
- const auto texture_buffer_usage{variant.texture_buffer_usage};
-
std::string source = fmt::format(R"(// {}
#version 430 core
#extension GL_ARB_separate_shader_objects : enable
-#extension GL_ARB_shader_viewport_layer_array : enable
-#extension GL_EXT_shader_image_load_formatted : enable
-#extension GL_NV_gpu_shader5 : enable
-#extension GL_NV_shader_thread_group : enable
-#extension GL_NV_shader_thread_shuffle : enable
)",
- GetShaderId(unique_identifier, program_type));
- if (is_compute) {
- source += "#extension GL_ARB_compute_variable_group_size : require\n";
+ GetShaderId(unique_identifier, shader_type));
+ if (device.HasShaderBallot()) {
+ source += "#extension GL_ARB_shader_ballot : require\n";
}
- source += '\n';
-
- if (!is_compute) {
- source += fmt::format("#define EMULATION_UBO_BINDING {}\n", base_bindings.cbuf++);
+ if (device.HasVertexViewportLayer()) {
+ source += "#extension GL_ARB_shader_viewport_layer_array : require\n";
}
-
- for (const auto& cbuf : entries.const_buffers) {
- source +=
- fmt::format("#define CBUF_BINDING_{} {}\n", cbuf.GetIndex(), base_bindings.cbuf++);
+ if (device.HasImageLoadFormatted()) {
+ source += "#extension GL_EXT_shader_image_load_formatted : require\n";
}
- for (const auto& gmem : entries.global_memory_entries) {
- source += fmt::format("#define GMEM_BINDING_{}_{} {}\n", gmem.GetCbufIndex(),
- gmem.GetCbufOffset(), base_bindings.gmem++);
+ if (device.HasWarpIntrinsics()) {
+ source += "#extension GL_NV_gpu_shader5 : require\n"
+ "#extension GL_NV_shader_thread_group : require\n"
+ "#extension GL_NV_shader_thread_shuffle : require\n";
}
- for (const auto& sampler : entries.samplers) {
- source += fmt::format("#define SAMPLER_BINDING_{} {}\n", sampler.GetIndex(),
- base_bindings.sampler++);
+
+ if (shader_type == ShaderType::Geometry) {
+ const auto [glsl_topology, max_vertices] = GetPrimitiveDescription(variant.primitive_mode);
+ source += fmt::format("#define MAX_VERTEX_INPUT {}\n", max_vertices);
+ source += fmt::format("layout ({}) in;\n", glsl_topology);
}
- for (const auto& image : entries.images) {
+ if (shader_type == ShaderType::Compute) {
+ if (variant.local_memory_size > 0) {
+ source += fmt::format("#define LOCAL_MEMORY_SIZE {}\n",
+ Common::AlignUp(variant.local_memory_size, 4) / 4);
+ }
source +=
- fmt::format("#define IMAGE_BINDING_{} {}\n", image.GetIndex(), base_bindings.image++);
- }
+ fmt::format("layout (local_size_x = {}, local_size_y = {}, local_size_z = {}) in;\n",
+ variant.block_x, variant.block_y, variant.block_z);
- // Transform 1D textures to texture samplers by declaring its preprocessor macros.
- for (std::size_t i = 0; i < texture_buffer_usage.size(); ++i) {
- if (!texture_buffer_usage.test(i)) {
- continue;
+ if (variant.shared_memory_size > 0) {
+ // shared_memory_size is described in number of words
+ source += fmt::format("shared uint smem[{}];\n", variant.shared_memory_size);
}
- source += fmt::format("#define SAMPLER_{}_IS_BUFFER\n", i);
- }
- if (texture_buffer_usage.any()) {
- source += '\n';
- }
-
- if (program_type == ProgramType::Geometry) {
- const auto [glsl_topology, debug_name, max_vertices] =
- GetPrimitiveDescription(primitive_mode);
-
- source += "layout (" + std::string(glsl_topology) + ") in;\n\n";
- source += "#define MAX_VERTEX_INPUT " + std::to_string(max_vertices) + '\n';
- }
- if (program_type == ProgramType::Compute) {
- source += "layout (local_size_variable) in;\n";
}
source += '\n';
- source += GenerateGLSL(device, program_type, ir, ir_b);
+ source += GenerateGLSL(device, shader_type, ir, ir_b);
OGLShader shader;
- shader.Create(source.c_str(), GetShaderType(program_type));
+ shader.Create(source.c_str(), GetGLShaderType(shader_type));
auto program = std::make_shared<OGLProgram>();
program->Create(true, hint_retrievable, shader.handle);
@@ -357,18 +312,16 @@ std::unordered_set<GLenum> GetSupportedFormats() {
} // Anonymous namespace
-CachedShader::CachedShader(const ShaderParameters& params, ProgramType program_type,
- GLShader::ShaderEntries entries, ProgramCode program_code,
- ProgramCode program_code_b)
- : RasterizerCacheObject{params.host_ptr}, system{params.system},
- disk_cache{params.disk_cache}, device{params.device}, cpu_addr{params.cpu_addr},
- unique_identifier{params.unique_identifier}, program_type{program_type}, entries{entries},
- program_code{std::move(program_code)}, program_code_b{std::move(program_code_b)} {
+CachedShader::CachedShader(const ShaderParameters& params, ShaderType shader_type,
+ GLShader::ShaderEntries entries, ProgramCode code, ProgramCode code_b)
+ : RasterizerCacheObject{params.host_ptr}, system{params.system}, disk_cache{params.disk_cache},
+ device{params.device}, cpu_addr{params.cpu_addr}, unique_identifier{params.unique_identifier},
+ shader_type{shader_type}, entries{entries}, code{std::move(code)}, code_b{std::move(code_b)} {
if (!params.precompiled_variants) {
return;
}
for (const auto& pair : *params.precompiled_variants) {
- auto locker = MakeLocker(system, program_type);
+ auto locker = MakeLocker(system, shader_type);
const auto& usage = pair->first;
FillLocker(*locker, usage);
@@ -389,92 +342,83 @@ CachedShader::CachedShader(const ShaderParameters& params, ProgramType program_t
}
Shader CachedShader::CreateStageFromMemory(const ShaderParameters& params,
- Maxwell::ShaderProgram program_type,
- ProgramCode program_code, ProgramCode program_code_b) {
- params.disk_cache.SaveRaw(ShaderDiskCacheRaw(
- params.unique_identifier, GetProgramType(program_type), program_code, program_code_b));
+ Maxwell::ShaderProgram program_type, ProgramCode code,
+ ProgramCode code_b) {
+ const auto shader_type = GetShaderType(program_type);
+ params.disk_cache.SaveRaw(
+ ShaderDiskCacheRaw(params.unique_identifier, shader_type, code, code_b));
- ConstBufferLocker locker(GetEnginesShaderType(GetProgramType(program_type)));
- const ShaderIR ir(program_code, STAGE_MAIN_OFFSET, COMPILER_SETTINGS, locker);
+ ConstBufferLocker locker(shader_type, params.system.GPU().Maxwell3D());
+ const ShaderIR ir(code, STAGE_MAIN_OFFSET, COMPILER_SETTINGS, locker);
// TODO(Rodrigo): Handle VertexA shaders
// std::optional<ShaderIR> ir_b;
- // if (!program_code_b.empty()) {
- // ir_b.emplace(program_code_b, STAGE_MAIN_OFFSET);
+ // if (!code_b.empty()) {
+ // ir_b.emplace(code_b, STAGE_MAIN_OFFSET);
// }
- return std::shared_ptr<CachedShader>(
- new CachedShader(params, GetProgramType(program_type), GLShader::GetEntries(ir),
- std::move(program_code), std::move(program_code_b)));
+ return std::shared_ptr<CachedShader>(new CachedShader(
+ params, shader_type, GLShader::GetEntries(ir), std::move(code), std::move(code_b)));
}
Shader CachedShader::CreateKernelFromMemory(const ShaderParameters& params, ProgramCode code) {
params.disk_cache.SaveRaw(
- ShaderDiskCacheRaw(params.unique_identifier, ProgramType::Compute, code));
+ ShaderDiskCacheRaw(params.unique_identifier, ShaderType::Compute, code));
- ConstBufferLocker locker(Tegra::Engines::ShaderType::Compute);
+ ConstBufferLocker locker(Tegra::Engines::ShaderType::Compute,
+ params.system.GPU().KeplerCompute());
const ShaderIR ir(code, KERNEL_MAIN_OFFSET, COMPILER_SETTINGS, locker);
return std::shared_ptr<CachedShader>(new CachedShader(
- params, ProgramType::Compute, GLShader::GetEntries(ir), std::move(code), {}));
+ params, ShaderType::Compute, GLShader::GetEntries(ir), std::move(code), {}));
}
Shader CachedShader::CreateFromCache(const ShaderParameters& params,
const UnspecializedShader& unspecialized) {
- return std::shared_ptr<CachedShader>(new CachedShader(params, unspecialized.program_type,
+ return std::shared_ptr<CachedShader>(new CachedShader(params, unspecialized.type,
unspecialized.entries, unspecialized.code,
unspecialized.code_b));
}
-std::tuple<GLuint, BaseBindings> CachedShader::GetProgramHandle(const ProgramVariant& variant) {
- UpdateVariant();
+GLuint CachedShader::GetHandle(const ProgramVariant& variant) {
+ EnsureValidLockerVariant();
- const auto [entry, is_cache_miss] = curr_variant->programs.try_emplace(variant);
+ const auto [entry, is_cache_miss] = curr_locker_variant->programs.try_emplace(variant);
auto& program = entry->second;
- if (is_cache_miss) {
- program = BuildShader(device, unique_identifier, program_type, program_code, program_code_b,
- variant, *curr_variant->locker);
- disk_cache.SaveUsage(GetUsage(variant, *curr_variant->locker));
-
- LabelGLObject(GL_PROGRAM, program->handle, cpu_addr);
+ if (!is_cache_miss) {
+ return program->handle;
}
- auto base_bindings = variant.base_bindings;
- base_bindings.cbuf += static_cast<u32>(entries.const_buffers.size());
- if (program_type != ProgramType::Compute) {
- base_bindings.cbuf += STAGE_RESERVED_UBOS;
- }
- base_bindings.gmem += static_cast<u32>(entries.global_memory_entries.size());
- base_bindings.sampler += static_cast<u32>(entries.samplers.size());
+ program = BuildShader(device, unique_identifier, shader_type, code, code_b,
+ *curr_locker_variant->locker, variant);
+ disk_cache.SaveUsage(GetUsage(variant, *curr_locker_variant->locker));
- return {program->handle, base_bindings};
+ LabelGLObject(GL_PROGRAM, program->handle, cpu_addr);
+ return program->handle;
}
-void CachedShader::UpdateVariant() {
- if (curr_variant && !curr_variant->locker->IsConsistent()) {
- curr_variant = nullptr;
+bool CachedShader::EnsureValidLockerVariant() {
+ const auto previous_variant = curr_locker_variant;
+ if (curr_locker_variant && !curr_locker_variant->locker->IsConsistent()) {
+ curr_locker_variant = nullptr;
}
- if (!curr_variant) {
+ if (!curr_locker_variant) {
for (auto& variant : locker_variants) {
if (variant->locker->IsConsistent()) {
- curr_variant = variant.get();
+ curr_locker_variant = variant.get();
}
}
}
- if (!curr_variant) {
+ if (!curr_locker_variant) {
auto& new_variant = locker_variants.emplace_back();
new_variant = std::make_unique<LockerVariant>();
- new_variant->locker = MakeLocker(system, program_type);
- curr_variant = new_variant.get();
+ new_variant->locker = MakeLocker(system, shader_type);
+ curr_locker_variant = new_variant.get();
}
+ return previous_variant == curr_locker_variant;
}
ShaderDiskCacheUsage CachedShader::GetUsage(const ProgramVariant& variant,
const ConstBufferLocker& locker) const {
- ShaderDiskCacheUsage usage;
- usage.unique_identifier = unique_identifier;
- usage.variant = variant;
- usage.keys = locker.GetKeys();
- usage.bound_samplers = locker.GetBoundSamplers();
- usage.bindless_samplers = locker.GetBindlessSamplers();
- return usage;
+ return ShaderDiskCacheUsage{unique_identifier, variant, locker.GetKeys(),
+ locker.GetBoundSamplers(), locker.GetBindlessSamplers()};
}
ShaderCacheOpenGL::ShaderCacheOpenGL(RasterizerOpenGL& rasterizer, Core::System& system,
@@ -533,11 +477,12 @@ void ShaderCacheOpenGL::LoadDiskCache(const std::atomic_bool& stop_loading,
}
}
if (!shader) {
- auto locker{MakeLocker(system, unspecialized.program_type)};
+ auto locker{MakeLocker(system, unspecialized.type)};
FillLocker(*locker, usage);
- shader = BuildShader(device, usage.unique_identifier, unspecialized.program_type,
- unspecialized.code, unspecialized.code_b, usage.variant,
- *locker, true);
+
+ shader = BuildShader(device, usage.unique_identifier, unspecialized.type,
+ unspecialized.code, unspecialized.code_b, *locker,
+ usage.variant, true);
}
std::scoped_lock lock{mutex};
@@ -640,7 +585,7 @@ bool ShaderCacheOpenGL::GenerateUnspecializedShaders(
const auto& raw{raws[i]};
const u64 unique_identifier{raw.GetUniqueIdentifier()};
const u64 calculated_hash{
- GetUniqueIdentifier(raw.GetProgramType(), raw.GetProgramCode(), raw.GetProgramCodeB())};
+ GetUniqueIdentifier(raw.GetType(), raw.HasProgramA(), raw.GetCode(), raw.GetCodeB())};
if (unique_identifier != calculated_hash) {
LOG_ERROR(Render_OpenGL,
"Invalid hash in entry={:016x} (obtained hash={:016x}) - "
@@ -651,9 +596,9 @@ bool ShaderCacheOpenGL::GenerateUnspecializedShaders(
}
const u32 main_offset =
- raw.GetProgramType() == ProgramType::Compute ? KERNEL_MAIN_OFFSET : STAGE_MAIN_OFFSET;
- ConstBufferLocker locker(GetEnginesShaderType(raw.GetProgramType()));
- const ShaderIR ir(raw.GetProgramCode(), main_offset, COMPILER_SETTINGS, locker);
+ raw.GetType() == ShaderType::Compute ? KERNEL_MAIN_OFFSET : STAGE_MAIN_OFFSET;
+ ConstBufferLocker locker(raw.GetType());
+ const ShaderIR ir(raw.GetCode(), main_offset, COMPILER_SETTINGS, locker);
// TODO(Rodrigo): Handle VertexA shaders
// std::optional<ShaderIR> ir_b;
// if (raw.HasProgramA()) {
@@ -662,9 +607,9 @@ bool ShaderCacheOpenGL::GenerateUnspecializedShaders(
UnspecializedShader unspecialized;
unspecialized.entries = GLShader::GetEntries(ir);
- unspecialized.program_type = raw.GetProgramType();
- unspecialized.code = raw.GetProgramCode();
- unspecialized.code_b = raw.GetProgramCodeB();
+ unspecialized.type = raw.GetType();
+ unspecialized.code = raw.GetCode();
+ unspecialized.code_b = raw.GetCodeB();
unspecialized_shaders.emplace(raw.GetUniqueIdentifier(), unspecialized);
if (callback) {
@@ -697,7 +642,8 @@ Shader ShaderCacheOpenGL::GetStageProgram(Maxwell::ShaderProgram program) {
code_b = GetShaderCode(memory_manager, address_b, memory_manager.GetPointer(address_b));
}
- const auto unique_identifier = GetUniqueIdentifier(GetProgramType(program), code, code_b);
+ const auto unique_identifier = GetUniqueIdentifier(
+ GetShaderType(program), program == Maxwell::ShaderProgram::VertexA, code, code_b);
const auto precompiled_variants = GetPrecompiledVariants(unique_identifier);
const auto cpu_addr{*memory_manager.GpuToCpuAddress(address)};
const ShaderParameters params{system, disk_cache, precompiled_variants, device,
@@ -725,7 +671,7 @@ Shader ShaderCacheOpenGL::GetComputeKernel(GPUVAddr code_addr) {
// No kernel found - create a new one
auto code{GetShaderCode(memory_manager, code_addr, host_ptr)};
- const auto unique_identifier{GetUniqueIdentifier(ProgramType::Compute, code, {})};
+ const auto unique_identifier{GetUniqueIdentifier(ShaderType::Compute, false, code, {})};
const auto precompiled_variants = GetPrecompiledVariants(unique_identifier);
const auto cpu_addr{*memory_manager.GpuToCpuAddress(code_addr)};
const ShaderParameters params{system, disk_cache, precompiled_variants, device,
diff --git a/src/video_core/renderer_opengl/gl_shader_cache.h b/src/video_core/renderer_opengl/gl_shader_cache.h
index 6bd7c9cf1..7b1470db3 100644
--- a/src/video_core/renderer_opengl/gl_shader_cache.h
+++ b/src/video_core/renderer_opengl/gl_shader_cache.h
@@ -17,6 +17,7 @@
#include <glad/glad.h>
#include "common/common_types.h"
+#include "video_core/engines/shader_type.h"
#include "video_core/rasterizer_cache.h"
#include "video_core/renderer_opengl/gl_resource_manager.h"
#include "video_core/renderer_opengl/gl_shader_decompiler.h"
@@ -47,7 +48,7 @@ using PrecompiledVariants = std::vector<PrecompiledPrograms::iterator>;
struct UnspecializedShader {
GLShader::ShaderEntries entries;
- ProgramType program_type;
+ Tegra::Engines::ShaderType type;
ProgramCode code;
ProgramCode code_b;
};
@@ -77,7 +78,7 @@ public:
}
std::size_t GetSizeInBytes() const override {
- return program_code.size() * sizeof(u64);
+ return code.size() * sizeof(u64);
}
/// Gets the shader entries for the shader
@@ -86,7 +87,7 @@ public:
}
/// Gets the GL program handle for the shader
- std::tuple<GLuint, BaseBindings> GetProgramHandle(const ProgramVariant& variant);
+ GLuint GetHandle(const ProgramVariant& variant);
private:
struct LockerVariant {
@@ -94,11 +95,11 @@ private:
std::unordered_map<ProgramVariant, CachedProgram> programs;
};
- explicit CachedShader(const ShaderParameters& params, ProgramType program_type,
+ explicit CachedShader(const ShaderParameters& params, Tegra::Engines::ShaderType shader_type,
GLShader::ShaderEntries entries, ProgramCode program_code,
ProgramCode program_code_b);
- void UpdateVariant();
+ bool EnsureValidLockerVariant();
ShaderDiskCacheUsage GetUsage(const ProgramVariant& variant,
const VideoCommon::Shader::ConstBufferLocker& locker) const;
@@ -110,14 +111,14 @@ private:
VAddr cpu_addr{};
u64 unique_identifier{};
- ProgramType program_type{};
+ Tegra::Engines::ShaderType shader_type{};
GLShader::ShaderEntries entries;
- ProgramCode program_code;
- ProgramCode program_code_b;
+ ProgramCode code;
+ ProgramCode code_b;
- LockerVariant* curr_variant = nullptr;
+ LockerVariant* curr_locker_variant = nullptr;
std::vector<std::unique_ptr<LockerVariant>> locker_variants;
};
diff --git a/src/video_core/renderer_opengl/gl_shader_decompiler.cpp b/src/video_core/renderer_opengl/gl_shader_decompiler.cpp
index 92ee8459e..f9f7a97b5 100644
--- a/src/video_core/renderer_opengl/gl_shader_decompiler.cpp
+++ b/src/video_core/renderer_opengl/gl_shader_decompiler.cpp
@@ -16,6 +16,7 @@
#include "common/common_types.h"
#include "common/logging/log.h"
#include "video_core/engines/maxwell_3d.h"
+#include "video_core/engines/shader_type.h"
#include "video_core/renderer_opengl/gl_device.h"
#include "video_core/renderer_opengl/gl_rasterizer.h"
#include "video_core/renderer_opengl/gl_shader_decompiler.h"
@@ -27,6 +28,7 @@ namespace OpenGL::GLShader {
namespace {
+using Tegra::Engines::ShaderType;
using Tegra::Shader::Attribute;
using Tegra::Shader::AttributeUse;
using Tegra::Shader::Header;
@@ -41,11 +43,15 @@ using namespace VideoCommon::Shader;
using Maxwell = Tegra::Engines::Maxwell3D::Regs;
using Operation = const OperationNode&;
+class ASTDecompiler;
+class ExprDecompiler;
+
enum class Type { Void, Bool, Bool2, Float, Int, Uint, HalfFloat };
-struct TextureAoffi {};
+struct TextureOffset {};
+struct TextureDerivates {};
using TextureArgument = std::pair<Type, Node>;
-using TextureIR = std::variant<TextureAoffi, TextureArgument>;
+using TextureIR = std::variant<TextureOffset, TextureDerivates, TextureArgument>;
constexpr u32 MAX_CONSTBUFFER_ELEMENTS =
static_cast<u32>(Maxwell::MaxConstBufferSize) / (4 * sizeof(float));
@@ -223,7 +229,7 @@ private:
Type type{};
};
-constexpr const char* GetTypeString(Type type) {
+const char* GetTypeString(Type type) {
switch (type) {
case Type::Bool:
return "bool";
@@ -243,7 +249,7 @@ constexpr const char* GetTypeString(Type type) {
}
}
-constexpr const char* GetImageTypeDeclaration(Tegra::Shader::ImageType image_type) {
+const char* GetImageTypeDeclaration(Tegra::Shader::ImageType image_type) {
switch (image_type) {
case Tegra::Shader::ImageType::Texture1D:
return "1D";
@@ -331,16 +337,13 @@ std::string FlowStackTopName(MetaStackClass stack) {
return fmt::format("{}_flow_stack_top", GetFlowStackPrefix(stack));
}
-constexpr bool IsVertexShader(ProgramType stage) {
- return stage == ProgramType::VertexA || stage == ProgramType::VertexB;
+[[deprecated]] constexpr bool IsVertexShader(ShaderType stage) {
+ return stage == ShaderType::Vertex;
}
-class ASTDecompiler;
-class ExprDecompiler;
-
class GLSLDecompiler final {
public:
- explicit GLSLDecompiler(const Device& device, const ShaderIR& ir, ProgramType stage,
+ explicit GLSLDecompiler(const Device& device, const ShaderIR& ir, ShaderType stage,
std::string suffix)
: device{device}, ir{ir}, stage{stage}, suffix{suffix}, header{ir.GetHeader()} {}
@@ -396,6 +399,7 @@ public:
DeclareConstantBuffers();
DeclareGlobalMemory();
DeclareSamplers();
+ DeclareImages();
DeclarePhysicalAttributeReader();
code.AddLine("void execute_{}() {{", suffix);
@@ -427,7 +431,7 @@ private:
}
void DeclareGeometry() {
- if (stage != ProgramType::Geometry) {
+ if (stage != ShaderType::Geometry) {
return;
}
@@ -510,10 +514,14 @@ private:
}
void DeclareLocalMemory() {
- // TODO(Rodrigo): Unstub kernel local memory size and pass it from a register at
- // specialization time.
- const u64 local_memory_size =
- stage == ProgramType::Compute ? 0x400 : header.GetLocalMemorySize();
+ if (stage == ShaderType::Compute) {
+ code.AddLine("#ifdef LOCAL_MEMORY_SIZE");
+ code.AddLine("uint {}[LOCAL_MEMORY_SIZE];", GetLocalMemory());
+ code.AddLine("#endif");
+ return;
+ }
+
+ const u64 local_memory_size = header.GetLocalMemorySize();
if (local_memory_size == 0) {
return;
}
@@ -522,13 +530,6 @@ private:
code.AddNewLine();
}
- void DeclareSharedMemory() {
- if (stage != ProgramType::Compute) {
- return;
- }
- code.AddLine("shared uint {}[];", GetSharedMemory());
- }
-
void DeclareInternalFlags() {
for (u32 flag = 0; flag < static_cast<u32>(InternalFlag::Amount); flag++) {
const auto flag_code = static_cast<InternalFlag>(flag);
@@ -578,12 +579,12 @@ private:
const u32 location{GetGenericAttributeIndex(index)};
std::string name{GetInputAttribute(index)};
- if (stage == ProgramType::Geometry) {
+ if (stage == ShaderType::Geometry) {
name = "gs_" + name + "[]";
}
std::string suffix;
- if (stage == ProgramType::Fragment) {
+ if (stage == ShaderType::Fragment) {
const auto input_mode{header.ps.GetAttributeUse(location)};
if (skip_unused && input_mode == AttributeUse::Unused) {
return;
@@ -595,7 +596,7 @@ private:
}
void DeclareOutputAttributes() {
- if (ir.HasPhysicalAttributes() && stage != ProgramType::Fragment) {
+ if (ir.HasPhysicalAttributes() && stage != ShaderType::Fragment) {
for (u32 i = 0; i < GetNumPhysicalVaryings(); ++i) {
DeclareOutputAttribute(ToGenericAttribute(i));
}
@@ -620,9 +621,9 @@ private:
}
void DeclareConstantBuffers() {
- for (const auto& entry : ir.GetConstantBuffers()) {
- const auto [index, size] = entry;
- code.AddLine("layout (std140, binding = CBUF_BINDING_{}) uniform {} {{", index,
+ u32 binding = device.GetBaseBindings(stage).uniform_buffer;
+ for (const auto& [index, cbuf] : ir.GetConstantBuffers()) {
+ code.AddLine("layout (std140, binding = {}) uniform {} {{", binding++,
GetConstBufferBlock(index));
code.AddLine(" uvec4 {}[{}];", GetConstBuffer(index), MAX_CONSTBUFFER_ELEMENTS);
code.AddLine("}};");
@@ -631,9 +632,8 @@ private:
}
void DeclareGlobalMemory() {
- for (const auto& gmem : ir.GetGlobalMemory()) {
- const auto& [base, usage] = gmem;
-
+ u32 binding = device.GetBaseBindings(stage).shader_storage_buffer;
+ for (const auto& [base, usage] : ir.GetGlobalMemory()) {
// Since we don't know how the shader will use the shader, hint the driver to disable as
// much optimizations as possible
std::string qualifier = "coherent volatile";
@@ -643,8 +643,8 @@ private:
qualifier += " writeonly";
}
- code.AddLine("layout (std430, binding = GMEM_BINDING_{}_{}) {} buffer {} {{",
- base.cbuf_index, base.cbuf_offset, qualifier, GetGlobalMemoryBlock(base));
+ code.AddLine("layout (std430, binding = {}) {} buffer {} {{", binding++, qualifier,
+ GetGlobalMemoryBlock(base));
code.AddLine(" uint {}[];", GetGlobalMemory(base));
code.AddLine("}};");
code.AddNewLine();
@@ -652,15 +652,17 @@ private:
}
void DeclareSamplers() {
- const auto& samplers = ir.GetSamplers();
- for (const auto& sampler : samplers) {
- const std::string name{GetSampler(sampler)};
- const std::string description{"layout (binding = SAMPLER_BINDING_" +
- std::to_string(sampler.GetIndex()) + ") uniform"};
+ u32 binding = device.GetBaseBindings(stage).sampler;
+ for (const auto& sampler : ir.GetSamplers()) {
+ const std::string name = GetSampler(sampler);
+ const std::string description = fmt::format("layout (binding = {}) uniform", binding++);
+
std::string sampler_type = [&]() {
+ if (sampler.IsBuffer()) {
+ return "samplerBuffer";
+ }
switch (sampler.GetType()) {
case Tegra::Shader::TextureType::Texture1D:
- // Special cased, read below.
return "sampler1D";
case Tegra::Shader::TextureType::Texture2D:
return "sampler2D";
@@ -680,21 +682,9 @@ private:
sampler_type += "Shadow";
}
- if (sampler.GetType() == Tegra::Shader::TextureType::Texture1D) {
- // 1D textures can be aliased to texture buffers, hide the declarations behind a
- // preprocessor flag and use one or the other from the GPU state. This has to be
- // done because shaders don't have enough information to determine the texture type.
- EmitIfdefIsBuffer(sampler);
- code.AddLine("{} samplerBuffer {};", description, name);
- code.AddLine("#else");
- code.AddLine("{} {} {};", description, sampler_type, name);
- code.AddLine("#endif");
- } else {
- // The other texture types (2D, 3D and cubes) don't have this issue.
- code.AddLine("{} {} {};", description, sampler_type, name);
- }
+ code.AddLine("{} {} {};", description, sampler_type, name);
}
- if (!samplers.empty()) {
+ if (!ir.GetSamplers().empty()) {
code.AddNewLine();
}
}
@@ -717,7 +707,7 @@ private:
constexpr u32 element_stride = 4;
const u32 address{generic_base + index * generic_stride + element * element_stride};
- const bool declared = stage != ProgramType::Fragment ||
+ const bool declared = stage != ShaderType::Fragment ||
header.ps.GetAttributeUse(index) != AttributeUse::Unused;
const std::string value =
declared ? ReadAttribute(attribute, element).AsFloat() : "0.0f";
@@ -734,8 +724,8 @@ private:
}
void DeclareImages() {
- const auto& images{ir.GetImages()};
- for (const auto& image : images) {
+ u32 binding = device.GetBaseBindings(stage).image;
+ for (const auto& image : ir.GetImages()) {
std::string qualifier = "coherent volatile";
if (image.IsRead() && !image.IsWritten()) {
qualifier += " readonly";
@@ -745,10 +735,10 @@ private:
const char* format = image.IsAtomic() ? "r32ui, " : "";
const char* type_declaration = GetImageTypeDeclaration(image.GetType());
- code.AddLine("layout ({}binding = IMAGE_BINDING_{}) {} uniform uimage{} {};", format,
- image.GetIndex(), qualifier, type_declaration, GetImage(image));
+ code.AddLine("layout ({}binding = {}) {} uniform uimage{} {};", format, binding++,
+ qualifier, type_declaration, GetImage(image));
}
- if (!images.empty()) {
+ if (!ir.GetImages().empty()) {
code.AddNewLine();
}
}
@@ -761,6 +751,9 @@ private:
Expression Visit(const Node& node) {
if (const auto operation = std::get_if<OperationNode>(&*node)) {
+ if (const auto amend_index = operation->GetAmendIndex()) {
+ Visit(ir.GetAmendNode(*amend_index)).CheckVoid();
+ }
const auto operation_index = static_cast<std::size_t>(operation->GetCode());
if (operation_index >= operation_decompilers.size()) {
UNREACHABLE_MSG("Out of bounds operation: {}", operation_index);
@@ -809,7 +802,7 @@ private:
}
if (const auto abuf = std::get_if<AbufNode>(&*node)) {
- UNIMPLEMENTED_IF_MSG(abuf->IsPhysicalBuffer() && stage == ProgramType::Geometry,
+ UNIMPLEMENTED_IF_MSG(abuf->IsPhysicalBuffer() && stage == ShaderType::Geometry,
"Physical attributes in geometry shaders are not implemented");
if (abuf->IsPhysicalBuffer()) {
return {fmt::format("ReadPhysicalAttribute({})",
@@ -868,18 +861,13 @@ private:
}
if (const auto lmem = std::get_if<LmemNode>(&*node)) {
- if (stage == ProgramType::Compute) {
- LOG_WARNING(Render_OpenGL, "Local memory is stubbed on compute shaders");
- }
return {
fmt::format("{}[{} >> 2]", GetLocalMemory(), Visit(lmem->GetAddress()).AsUint()),
Type::Uint};
}
if (const auto smem = std::get_if<SmemNode>(&*node)) {
- return {
- fmt::format("{}[{} >> 2]", GetSharedMemory(), Visit(smem->GetAddress()).AsUint()),
- Type::Uint};
+ return {fmt::format("smem[{} >> 2]", Visit(smem->GetAddress()).AsUint()), Type::Uint};
}
if (const auto internal_flag = std::get_if<InternalFlagNode>(&*node)) {
@@ -887,6 +875,9 @@ private:
}
if (const auto conditional = std::get_if<ConditionalNode>(&*node)) {
+ if (const auto amend_index = conditional->GetAmendIndex()) {
+ Visit(ir.GetAmendNode(*amend_index)).CheckVoid();
+ }
// It's invalid to call conditional on nested nodes, use an operation instead
code.AddLine("if ({}) {{", Visit(conditional->GetCondition()).AsBool());
++code.scope;
@@ -909,7 +900,7 @@ private:
Expression ReadAttribute(Attribute::Index attribute, u32 element, const Node& buffer = {}) {
const auto GeometryPass = [&](std::string_view name) {
- if (stage == ProgramType::Geometry && buffer) {
+ if (stage == ShaderType::Geometry && buffer) {
// TODO(Rodrigo): Guard geometry inputs against out of bound reads. Some games
// set an 0x80000000 index for those and the shader fails to build. Find out why
// this happens and what's its intent.
@@ -921,11 +912,11 @@ private:
switch (attribute) {
case Attribute::Index::Position:
switch (stage) {
- case ProgramType::Geometry:
+ case ShaderType::Geometry:
return {fmt::format("gl_in[{}].gl_Position{}", Visit(buffer).AsUint(),
GetSwizzle(element)),
Type::Float};
- case ProgramType::Fragment:
+ case ShaderType::Fragment:
return {element == 3 ? "1.0f" : ("gl_FragCoord"s + GetSwizzle(element)),
Type::Float};
default:
@@ -959,7 +950,7 @@ private:
return {"0", Type::Int};
case Attribute::Index::FrontFacing:
// TODO(Subv): Find out what the values are for the other elements.
- ASSERT(stage == ProgramType::Fragment);
+ ASSERT(stage == ShaderType::Fragment);
switch (element) {
case 3:
return {"(gl_FrontFacing ? -1 : 0)", Type::Int};
@@ -985,7 +976,7 @@ private:
// be found in fragment shaders, so we disable precise there. There are vertex shaders that
// also fail to build but nobody seems to care about those.
// Note: Only bugged drivers will skip precise.
- const bool disable_precise = device.HasPreciseBug() && stage == ProgramType::Fragment;
+ const bool disable_precise = device.HasPreciseBug() && stage == ShaderType::Fragment;
std::string temporary = code.GenerateTemporary();
code.AddLine("{}{} {} = {};", disable_precise ? "" : "precise ", GetTypeString(type),
@@ -1092,7 +1083,7 @@ private:
}
std::string GenerateTexture(Operation operation, const std::string& function_suffix,
- const std::vector<TextureIR>& extras) {
+ const std::vector<TextureIR>& extras, bool separate_dc = false) {
constexpr std::array coord_constructors = {"float", "vec2", "vec3", "vec4"};
const auto meta = std::get_if<MetaTexture>(&operation.GetMeta());
@@ -1105,9 +1096,12 @@ private:
std::string expr = "texture" + function_suffix;
if (!meta->aoffi.empty()) {
expr += "Offset";
+ } else if (!meta->ptp.empty()) {
+ expr += "Offsets";
}
expr += '(' + GetSampler(meta->sampler) + ", ";
- expr += coord_constructors.at(count + (has_array ? 1 : 0) + (has_shadow ? 1 : 0) - 1);
+ expr += coord_constructors.at(count + (has_array ? 1 : 0) +
+ (has_shadow && !separate_dc ? 1 : 0) - 1);
expr += '(';
for (std::size_t i = 0; i < count; ++i) {
expr += Visit(operation[i]).AsFloat();
@@ -1120,15 +1114,26 @@ private:
expr += ", float(" + Visit(meta->array).AsInt() + ')';
}
if (has_shadow) {
- expr += ", " + Visit(meta->depth_compare).AsFloat();
+ if (separate_dc) {
+ expr += "), " + Visit(meta->depth_compare).AsFloat();
+ } else {
+ expr += ", " + Visit(meta->depth_compare).AsFloat() + ')';
+ }
+ } else {
+ expr += ')';
}
- expr += ')';
for (const auto& variant : extras) {
if (const auto argument = std::get_if<TextureArgument>(&variant)) {
expr += GenerateTextureArgument(*argument);
- } else if (std::holds_alternative<TextureAoffi>(variant)) {
- expr += GenerateTextureAoffi(meta->aoffi);
+ } else if (std::holds_alternative<TextureOffset>(variant)) {
+ if (!meta->aoffi.empty()) {
+ expr += GenerateTextureAoffi(meta->aoffi);
+ } else if (!meta->ptp.empty()) {
+ expr += GenerateTexturePtp(meta->ptp);
+ }
+ } else if (std::holds_alternative<TextureDerivates>(variant)) {
+ expr += GenerateTextureDerivates(meta->derivates);
} else {
UNREACHABLE();
}
@@ -1167,6 +1172,20 @@ private:
return expr;
}
+ std::string ReadTextureOffset(const Node& value) {
+ if (const auto immediate = std::get_if<ImmediateNode>(&*value)) {
+ // Inline the string as an immediate integer in GLSL (AOFFI arguments are required
+ // to be constant by the standard).
+ return std::to_string(static_cast<s32>(immediate->GetValue()));
+ } else if (device.HasVariableAoffi()) {
+ // Avoid using variable AOFFI on unsupported devices.
+ return Visit(value).AsInt();
+ } else {
+ // Insert 0 on devices not supporting variable AOFFI.
+ return "0";
+ }
+ }
+
std::string GenerateTextureAoffi(const std::vector<Node>& aoffi) {
if (aoffi.empty()) {
return {};
@@ -1177,18 +1196,7 @@ private:
expr += '(';
for (std::size_t index = 0; index < aoffi.size(); ++index) {
- const auto operand{aoffi.at(index)};
- if (const auto immediate = std::get_if<ImmediateNode>(&*operand)) {
- // Inline the string as an immediate integer in GLSL (AOFFI arguments are required
- // to be constant by the standard).
- expr += std::to_string(static_cast<s32>(immediate->GetValue()));
- } else if (device.HasVariableAoffi()) {
- // Avoid using variable AOFFI on unsupported devices.
- expr += Visit(operand).AsInt();
- } else {
- // Insert 0 on devices not supporting variable AOFFI.
- expr += '0';
- }
+ expr += ReadTextureOffset(aoffi.at(index));
if (index + 1 < aoffi.size()) {
expr += ", ";
}
@@ -1198,6 +1206,50 @@ private:
return expr;
}
+ std::string GenerateTexturePtp(const std::vector<Node>& ptp) {
+ static constexpr std::size_t num_vectors = 4;
+ ASSERT(ptp.size() == num_vectors * 2);
+
+ std::string expr = ", ivec2[](";
+ for (std::size_t vector = 0; vector < num_vectors; ++vector) {
+ const bool has_next = vector + 1 < num_vectors;
+ expr += fmt::format("ivec2({}, {}){}", ReadTextureOffset(ptp.at(vector * 2)),
+ ReadTextureOffset(ptp.at(vector * 2 + 1)), has_next ? ", " : "");
+ }
+ expr += ')';
+ return expr;
+ }
+
+ std::string GenerateTextureDerivates(const std::vector<Node>& derivates) {
+ if (derivates.empty()) {
+ return {};
+ }
+ constexpr std::array coord_constructors = {"float", "vec2", "vec3"};
+ std::string expr = ", ";
+ const std::size_t components = derivates.size() / 2;
+ std::string dx = coord_constructors.at(components - 1);
+ std::string dy = coord_constructors.at(components - 1);
+ dx += '(';
+ dy += '(';
+
+ for (std::size_t index = 0; index < components; ++index) {
+ const auto operand_x{derivates.at(index * 2)};
+ const auto operand_y{derivates.at(index * 2 + 1)};
+ dx += Visit(operand_x).AsFloat();
+ dy += Visit(operand_y).AsFloat();
+
+ if (index + 1 < components) {
+ dx += ", ";
+ dy += ", ";
+ }
+ }
+ dx += ')';
+ dy += ')';
+ expr += dx + ", " + dy;
+
+ return expr;
+ }
+
std::string BuildIntegerCoordinates(Operation operation) {
constexpr std::array constructors{"int(", "ivec2(", "ivec3(", "ivec4("};
const std::size_t coords_count{operation.GetOperandsCount()};
@@ -1247,17 +1299,12 @@ private:
}
target = std::move(*output);
} else if (const auto lmem = std::get_if<LmemNode>(&*dest)) {
- if (stage == ProgramType::Compute) {
- LOG_WARNING(Render_OpenGL, "Local memory is stubbed on compute shaders");
- }
target = {
fmt::format("{}[{} >> 2]", GetLocalMemory(), Visit(lmem->GetAddress()).AsUint()),
Type::Uint};
} else if (const auto smem = std::get_if<SmemNode>(&*dest)) {
- ASSERT(stage == ProgramType::Compute);
- target = {
- fmt::format("{}[{} >> 2]", GetSharedMemory(), Visit(smem->GetAddress()).AsUint()),
- Type::Uint};
+ ASSERT(stage == ShaderType::Compute);
+ target = {fmt::format("smem[{} >> 2]", Visit(smem->GetAddress()).AsUint()), Type::Uint};
} else if (const auto gmem = std::get_if<GmemNode>(&*dest)) {
const std::string real = Visit(gmem->GetRealAddress()).AsUint();
const std::string base = Visit(gmem->GetBaseAddress()).AsUint();
@@ -1379,6 +1426,26 @@ private:
return GenerateUnary(operation, "float", Type::Float, type);
}
+ Expression FSwizzleAdd(Operation operation) {
+ const std::string op_a = VisitOperand(operation, 0).AsFloat();
+ const std::string op_b = VisitOperand(operation, 1).AsFloat();
+
+ if (!device.HasShaderBallot()) {
+ LOG_ERROR(Render_OpenGL, "Shader ballot is unavailable but required by the shader");
+ return {fmt::format("{} + {}", op_a, op_b), Type::Float};
+ }
+
+ const std::string instr_mask = VisitOperand(operation, 2).AsUint();
+ const std::string mask = code.GenerateTemporary();
+ code.AddLine("uint {} = ({} >> ((gl_SubGroupInvocationARB & 3) << 1)) & 3;", mask,
+ instr_mask);
+
+ const std::string modifier_a = fmt::format("fswzadd_modifiers_a[{}]", mask);
+ const std::string modifier_b = fmt::format("fswzadd_modifiers_b[{}]", mask);
+ return {fmt::format("(({} * {}) + ({} * {}))", op_a, modifier_a, op_b, modifier_b),
+ Type::Float};
+ }
+
Expression ICastFloat(Operation operation) {
return GenerateUnary(operation, "int", Type::Int, Type::Float);
}
@@ -1452,6 +1519,11 @@ private:
return GenerateUnary(operation, "bitCount", type, type);
}
+ template <Type type>
+ Expression BitMSB(Operation operation) {
+ return GenerateUnary(operation, "findMSB", type, type);
+ }
+
Expression HNegate(Operation operation) {
const auto GetNegate = [&](std::size_t index) {
return VisitOperand(operation, index).AsBool() + " ? -1 : 1";
@@ -1471,7 +1543,8 @@ private:
}
Expression HCastFloat(Operation operation) {
- return {fmt::format("vec2({})", VisitOperand(operation, 0).AsFloat()), Type::HalfFloat};
+ return {fmt::format("vec2({}, 0.0f)", VisitOperand(operation, 0).AsFloat()),
+ Type::HalfFloat};
}
Expression HUnpack(Operation operation) {
@@ -1645,7 +1718,7 @@ private:
ASSERT(meta);
std::string expr = GenerateTexture(
- operation, "", {TextureAoffi{}, TextureArgument{Type::Float, meta->bias}});
+ operation, "", {TextureOffset{}, TextureArgument{Type::Float, meta->bias}});
if (meta->sampler.IsShadow()) {
expr = "vec4(" + expr + ')';
}
@@ -1657,7 +1730,7 @@ private:
ASSERT(meta);
std::string expr = GenerateTexture(
- operation, "Lod", {TextureArgument{Type::Float, meta->lod}, TextureAoffi{}});
+ operation, "Lod", {TextureArgument{Type::Float, meta->lod}, TextureOffset{}});
if (meta->sampler.IsShadow()) {
expr = "vec4(" + expr + ')';
}
@@ -1665,13 +1738,18 @@ private:
}
Expression TextureGather(Operation operation) {
- const auto meta = std::get_if<MetaTexture>(&operation.GetMeta());
- ASSERT(meta);
+ const auto& meta = std::get<MetaTexture>(operation.GetMeta());
- const auto type = meta->sampler.IsShadow() ? Type::Float : Type::Int;
- return {GenerateTexture(operation, "Gather",
- {TextureArgument{type, meta->component}, TextureAoffi{}}) +
- GetSwizzle(meta->element),
+ const auto type = meta.sampler.IsShadow() ? Type::Float : Type::Int;
+ const bool separate_dc = meta.sampler.IsShadow();
+
+ std::vector<TextureIR> ir;
+ if (meta.sampler.IsShadow()) {
+ ir = {TextureOffset{}};
+ } else {
+ ir = {TextureOffset{}, TextureArgument{type, meta.component}};
+ }
+ return {GenerateTexture(operation, "Gather", ir, separate_dc) + GetSwizzle(meta.element),
Type::Float};
}
@@ -1729,27 +1807,23 @@ private:
expr += ", ";
}
- // Store a copy of the expression without the lod to be used with texture buffers
- std::string expr_buffer = expr;
-
- if (meta->lod) {
+ if (meta->lod && !meta->sampler.IsBuffer()) {
expr += ", ";
expr += Visit(meta->lod).AsInt();
}
expr += ')';
expr += GetSwizzle(meta->element);
- expr_buffer += ')';
- expr_buffer += GetSwizzle(meta->element);
+ return {std::move(expr), Type::Float};
+ }
- const std::string tmp{code.GenerateTemporary()};
- EmitIfdefIsBuffer(meta->sampler);
- code.AddLine("float {} = {};", tmp, expr_buffer);
- code.AddLine("#else");
- code.AddLine("float {} = {};", tmp, expr);
- code.AddLine("#endif");
+ Expression TextureGradient(Operation operation) {
+ const auto meta = std::get_if<MetaTexture>(&operation.GetMeta());
+ ASSERT(meta);
- return {tmp, Type::Float};
+ std::string expr =
+ GenerateTexture(operation, "Grad", {TextureDerivates{}, TextureOffset{}});
+ return {std::move(expr) + GetSwizzle(meta->element), Type::Float};
}
Expression ImageLoad(Operation operation) {
@@ -1817,7 +1891,7 @@ private:
}
void PreExit() {
- if (stage != ProgramType::Fragment) {
+ if (stage != ShaderType::Fragment) {
return;
}
const auto& used_registers = ir.GetRegisters();
@@ -1870,27 +1944,25 @@ private:
}
Expression EmitVertex(Operation operation) {
- ASSERT_MSG(stage == ProgramType::Geometry,
+ ASSERT_MSG(stage == ShaderType::Geometry,
"EmitVertex is expected to be used in a geometry shader.");
-
- // If a geometry shader is attached, it will always flip (it's the last stage before
- // fragment). For more info about flipping, refer to gl_shader_gen.cpp.
- code.AddLine("gl_Position.xy *= viewport_flip.xy;");
code.AddLine("EmitVertex();");
return {};
}
Expression EndPrimitive(Operation operation) {
- ASSERT_MSG(stage == ProgramType::Geometry,
+ ASSERT_MSG(stage == ShaderType::Geometry,
"EndPrimitive is expected to be used in a geometry shader.");
-
code.AddLine("EndPrimitive();");
return {};
}
+ Expression InvocationId(Operation operation) {
+ return {"gl_InvocationID", Type::Int};
+ }
+
Expression YNegate(Operation operation) {
- // Config pack's third value is Y_NEGATE's state.
- return {"config_pack[2]", Type::Uint};
+ return {"y_direction", Type::Float};
}
template <u32 element>
@@ -1942,34 +2014,29 @@ private:
return Vote(operation, "allThreadsEqualNV");
}
- template <const std::string_view& func>
- Expression Shuffle(Operation operation) {
- const std::string value = VisitOperand(operation, 0).AsFloat();
- if (!device.HasWarpIntrinsics()) {
- LOG_ERROR(Render_OpenGL, "Nvidia shuffle intrinsics are required by this shader");
- // On a "single-thread" device we are either on the same thread or out of bounds. Both
- // cases return the passed value.
- return {value, Type::Float};
+ Expression ThreadId(Operation operation) {
+ if (!device.HasShaderBallot()) {
+ LOG_ERROR(Render_OpenGL, "Shader ballot is unavailable but required by the shader");
+ return {"0U", Type::Uint};
}
-
- const std::string index = VisitOperand(operation, 1).AsUint();
- const std::string width = VisitOperand(operation, 2).AsUint();
- return {fmt::format("{}({}, {}, {})", func, value, index, width), Type::Float};
+ return {"gl_SubGroupInvocationARB", Type::Uint};
}
- template <const std::string_view& func>
- Expression InRangeShuffle(Operation operation) {
- const std::string index = VisitOperand(operation, 0).AsUint();
- const std::string width = VisitOperand(operation, 1).AsUint();
- if (!device.HasWarpIntrinsics()) {
- // On a "single-thread" device we are only in bounds when the requested index is 0.
- return {fmt::format("({} == 0U)", index), Type::Bool};
+ Expression ShuffleIndexed(Operation operation) {
+ std::string value = VisitOperand(operation, 0).AsFloat();
+
+ if (!device.HasShaderBallot()) {
+ LOG_ERROR(Render_OpenGL, "Shader ballot is unavailable but required by the shader");
+ return {std::move(value), Type::Float};
}
- const std::string in_range = code.GenerateTemporary();
- code.AddLine("bool {};", in_range);
- code.AddLine("{}(0U, {}, {}, {});", func, index, width, in_range);
- return {in_range, Type::Bool};
+ const std::string index = VisitOperand(operation, 1).AsUint();
+ return {fmt::format("readInvocationARB({}, {})", value, index), Type::Float};
+ }
+
+ Expression MemoryBarrierGL(Operation) {
+ code.AddLine("memoryBarrier();");
+ return {};
}
struct Func final {
@@ -1981,11 +2048,6 @@ private:
static constexpr std::string_view Or = "Or";
static constexpr std::string_view Xor = "Xor";
static constexpr std::string_view Exchange = "Exchange";
-
- static constexpr std::string_view ShuffleIndexed = "shuffleNV";
- static constexpr std::string_view ShuffleUp = "shuffleUpNV";
- static constexpr std::string_view ShuffleDown = "shuffleDownNV";
- static constexpr std::string_view ShuffleButterfly = "shuffleXorNV";
};
static constexpr std::array operation_decompilers = {
@@ -2016,6 +2078,7 @@ private:
&GLSLDecompiler::FTrunc,
&GLSLDecompiler::FCastInteger<Type::Int>,
&GLSLDecompiler::FCastInteger<Type::Uint>,
+ &GLSLDecompiler::FSwizzleAdd,
&GLSLDecompiler::Add<Type::Int>,
&GLSLDecompiler::Mul<Type::Int>,
@@ -2037,6 +2100,7 @@ private:
&GLSLDecompiler::BitfieldInsert<Type::Int>,
&GLSLDecompiler::BitfieldExtract<Type::Int>,
&GLSLDecompiler::BitCount<Type::Int>,
+ &GLSLDecompiler::BitMSB<Type::Int>,
&GLSLDecompiler::Add<Type::Uint>,
&GLSLDecompiler::Mul<Type::Uint>,
@@ -2055,6 +2119,7 @@ private:
&GLSLDecompiler::BitfieldInsert<Type::Uint>,
&GLSLDecompiler::BitfieldExtract<Type::Uint>,
&GLSLDecompiler::BitCount<Type::Uint>,
+ &GLSLDecompiler::BitMSB<Type::Uint>,
&GLSLDecompiler::Add<Type::HalfFloat>,
&GLSLDecompiler::Mul<Type::HalfFloat>,
@@ -2118,6 +2183,7 @@ private:
&GLSLDecompiler::TextureQueryDimensions,
&GLSLDecompiler::TextureQueryLod,
&GLSLDecompiler::TexelFetch,
+ &GLSLDecompiler::TextureGradient,
&GLSLDecompiler::ImageLoad,
&GLSLDecompiler::ImageStore,
@@ -2138,6 +2204,7 @@ private:
&GLSLDecompiler::EmitVertex,
&GLSLDecompiler::EndPrimitive,
+ &GLSLDecompiler::InvocationId,
&GLSLDecompiler::YNegate,
&GLSLDecompiler::LocalInvocationId<0>,
&GLSLDecompiler::LocalInvocationId<1>,
@@ -2151,15 +2218,10 @@ private:
&GLSLDecompiler::VoteAny,
&GLSLDecompiler::VoteEqual,
- &GLSLDecompiler::Shuffle<Func::ShuffleIndexed>,
- &GLSLDecompiler::Shuffle<Func::ShuffleUp>,
- &GLSLDecompiler::Shuffle<Func::ShuffleDown>,
- &GLSLDecompiler::Shuffle<Func::ShuffleButterfly>,
+ &GLSLDecompiler::ThreadId,
+ &GLSLDecompiler::ShuffleIndexed,
- &GLSLDecompiler::InRangeShuffle<Func::ShuffleIndexed>,
- &GLSLDecompiler::InRangeShuffle<Func::ShuffleUp>,
- &GLSLDecompiler::InRangeShuffle<Func::ShuffleDown>,
- &GLSLDecompiler::InRangeShuffle<Func::ShuffleButterfly>,
+ &GLSLDecompiler::MemoryBarrierGL,
};
static_assert(operation_decompilers.size() == static_cast<std::size_t>(OperationCode::Amount));
@@ -2200,10 +2262,6 @@ private:
return "lmem_" + suffix;
}
- std::string GetSharedMemory() const {
- return fmt::format("smem_{}", suffix);
- }
-
std::string GetInternalFlag(InternalFlag flag) const {
constexpr std::array InternalFlagNames = {"zero_flag", "sign_flag", "carry_flag",
"overflow_flag"};
@@ -2221,10 +2279,6 @@ private:
return GetDeclarationWithSuffix(static_cast<u32>(image.GetIndex()), "image");
}
- void EmitIfdefIsBuffer(const Sampler& sampler) {
- code.AddLine("#ifdef SAMPLER_{}_IS_BUFFER", sampler.GetIndex());
- }
-
std::string GetDeclarationWithSuffix(u32 index, std::string_view name) const {
return fmt::format("{}_{}_{}", name, index, suffix);
}
@@ -2243,7 +2297,7 @@ private:
const Device& device;
const ShaderIR& ir;
- const ProgramType stage;
+ const ShaderType stage;
const std::string suffix;
const Header header;
@@ -2492,10 +2546,13 @@ bvec2 HalfFloatNanComparison(bvec2 comparison, vec2 pair1, vec2 pair2) {
bvec2 is_nan2 = isnan(pair2);
return bvec2(comparison.x || is_nan1.x || is_nan2.x, comparison.y || is_nan1.y || is_nan2.y);
}
+
+const float fswzadd_modifiers_a[] = float[4](-1.0f, 1.0f, -1.0f, 0.0f );
+const float fswzadd_modifiers_b[] = float[4](-1.0f, -1.0f, 1.0f, -1.0f );
)";
}
-std::string Decompile(const Device& device, const ShaderIR& ir, ProgramType stage,
+std::string Decompile(const Device& device, const ShaderIR& ir, ShaderType stage,
const std::string& suffix) {
GLSLDecompiler decompiler(device, ir, stage, suffix);
decompiler.Decompile();
diff --git a/src/video_core/renderer_opengl/gl_shader_decompiler.h b/src/video_core/renderer_opengl/gl_shader_decompiler.h
index b1e75e6cc..7876f48d6 100644
--- a/src/video_core/renderer_opengl/gl_shader_decompiler.h
+++ b/src/video_core/renderer_opengl/gl_shader_decompiler.h
@@ -10,6 +10,7 @@
#include <vector>
#include "common/common_types.h"
#include "video_core/engines/maxwell_3d.h"
+#include "video_core/engines/shader_type.h"
#include "video_core/shader/shader_ir.h"
namespace VideoCommon::Shader {
@@ -17,20 +18,8 @@ class ShaderIR;
}
namespace OpenGL {
-
class Device;
-
-enum class ProgramType : u32 {
- VertexA = 0,
- VertexB = 1,
- TessellationControl = 2,
- TessellationEval = 3,
- Geometry = 4,
- Fragment = 5,
- Compute = 6
-};
-
-} // namespace OpenGL
+}
namespace OpenGL::GLShader {
@@ -94,6 +83,6 @@ ShaderEntries GetEntries(const VideoCommon::Shader::ShaderIR& ir);
std::string GetCommonDeclarations();
std::string Decompile(const Device& device, const VideoCommon::Shader::ShaderIR& ir,
- ProgramType stage, const std::string& suffix);
+ Tegra::Engines::ShaderType stage, const std::string& suffix);
} // namespace OpenGL::GLShader
diff --git a/src/video_core/renderer_opengl/gl_shader_disk_cache.cpp b/src/video_core/renderer_opengl/gl_shader_disk_cache.cpp
index 184a565e6..cf874a09a 100644
--- a/src/video_core/renderer_opengl/gl_shader_disk_cache.cpp
+++ b/src/video_core/renderer_opengl/gl_shader_disk_cache.cpp
@@ -3,6 +3,7 @@
// Refer to the license.txt file included.
#include <cstring>
+
#include <fmt/format.h>
#include "common/assert.h"
@@ -12,50 +13,50 @@
#include "common/logging/log.h"
#include "common/scm_rev.h"
#include "common/zstd_compression.h"
-
#include "core/core.h"
#include "core/hle/kernel/process.h"
#include "core/settings.h"
-
+#include "video_core/engines/shader_type.h"
#include "video_core/renderer_opengl/gl_shader_cache.h"
#include "video_core/renderer_opengl/gl_shader_disk_cache.h"
namespace OpenGL {
+using Tegra::Engines::ShaderType;
using VideoCommon::Shader::BindlessSamplerMap;
using VideoCommon::Shader::BoundSamplerMap;
using VideoCommon::Shader::KeyMap;
namespace {
+using ShaderCacheVersionHash = std::array<u8, 64>;
+
+enum class TransferableEntryKind : u32 {
+ Raw,
+ Usage,
+};
+
struct ConstBufferKey {
- u32 cbuf;
- u32 offset;
- u32 value;
+ u32 cbuf{};
+ u32 offset{};
+ u32 value{};
};
struct BoundSamplerKey {
- u32 offset;
- Tegra::Engines::SamplerDescriptor sampler;
+ u32 offset{};
+ Tegra::Engines::SamplerDescriptor sampler{};
};
struct BindlessSamplerKey {
- u32 cbuf;
- u32 offset;
- Tegra::Engines::SamplerDescriptor sampler;
-};
-
-using ShaderCacheVersionHash = std::array<u8, 64>;
-
-enum class TransferableEntryKind : u32 {
- Raw,
- Usage,
+ u32 cbuf{};
+ u32 offset{};
+ Tegra::Engines::SamplerDescriptor sampler{};
};
-constexpr u32 NativeVersion = 5;
+constexpr u32 NativeVersion = 11;
// Making sure sizes doesn't change by accident
-static_assert(sizeof(BaseBindings) == 16);
+static_assert(sizeof(ProgramVariant) == 20);
ShaderCacheVersionHash GetShaderCacheVersionHash() {
ShaderCacheVersionHash hash{};
@@ -66,10 +67,10 @@ ShaderCacheVersionHash GetShaderCacheVersionHash() {
} // Anonymous namespace
-ShaderDiskCacheRaw::ShaderDiskCacheRaw(u64 unique_identifier, ProgramType program_type,
- ProgramCode program_code, ProgramCode program_code_b)
- : unique_identifier{unique_identifier}, program_type{program_type},
- program_code{std::move(program_code)}, program_code_b{std::move(program_code_b)} {}
+ShaderDiskCacheRaw::ShaderDiskCacheRaw(u64 unique_identifier, ShaderType type, ProgramCode code,
+ ProgramCode code_b)
+ : unique_identifier{unique_identifier}, type{type}, code{std::move(code)}, code_b{std::move(
+ code_b)} {}
ShaderDiskCacheRaw::ShaderDiskCacheRaw() = default;
@@ -77,42 +78,39 @@ ShaderDiskCacheRaw::~ShaderDiskCacheRaw() = default;
bool ShaderDiskCacheRaw::Load(FileUtil::IOFile& file) {
if (file.ReadBytes(&unique_identifier, sizeof(u64)) != sizeof(u64) ||
- file.ReadBytes(&program_type, sizeof(u32)) != sizeof(u32)) {
+ file.ReadBytes(&type, sizeof(u32)) != sizeof(u32)) {
return false;
}
- u32 program_code_size{};
- u32 program_code_size_b{};
- if (file.ReadBytes(&program_code_size, sizeof(u32)) != sizeof(u32) ||
- file.ReadBytes(&program_code_size_b, sizeof(u32)) != sizeof(u32)) {
+ u32 code_size{};
+ u32 code_size_b{};
+ if (file.ReadBytes(&code_size, sizeof(u32)) != sizeof(u32) ||
+ file.ReadBytes(&code_size_b, sizeof(u32)) != sizeof(u32)) {
return false;
}
- program_code.resize(program_code_size);
- program_code_b.resize(program_code_size_b);
+ code.resize(code_size);
+ code_b.resize(code_size_b);
- if (file.ReadArray(program_code.data(), program_code_size) != program_code_size)
+ if (file.ReadArray(code.data(), code_size) != code_size)
return false;
- if (HasProgramA() &&
- file.ReadArray(program_code_b.data(), program_code_size_b) != program_code_size_b) {
+ if (HasProgramA() && file.ReadArray(code_b.data(), code_size_b) != code_size_b) {
return false;
}
return true;
}
bool ShaderDiskCacheRaw::Save(FileUtil::IOFile& file) const {
- if (file.WriteObject(unique_identifier) != 1 ||
- file.WriteObject(static_cast<u32>(program_type)) != 1 ||
- file.WriteObject(static_cast<u32>(program_code.size())) != 1 ||
- file.WriteObject(static_cast<u32>(program_code_b.size())) != 1) {
+ if (file.WriteObject(unique_identifier) != 1 || file.WriteObject(static_cast<u32>(type)) != 1 ||
+ file.WriteObject(static_cast<u32>(code.size())) != 1 ||
+ file.WriteObject(static_cast<u32>(code_b.size())) != 1) {
return false;
}
- if (file.WriteArray(program_code.data(), program_code.size()) != program_code.size())
+ if (file.WriteArray(code.data(), code.size()) != code.size())
return false;
- if (HasProgramA() &&
- file.WriteArray(program_code_b.data(), program_code_b.size()) != program_code_b.size()) {
+ if (HasProgramA() && file.WriteArray(code_b.data(), code_b.size()) != code_b.size()) {
return false;
}
return true;
diff --git a/src/video_core/renderer_opengl/gl_shader_disk_cache.h b/src/video_core/renderer_opengl/gl_shader_disk_cache.h
index db23ada93..69a2fbdda 100644
--- a/src/video_core/renderer_opengl/gl_shader_disk_cache.h
+++ b/src/video_core/renderer_opengl/gl_shader_disk_cache.h
@@ -4,7 +4,6 @@
#pragma once
-#include <bitset>
#include <optional>
#include <string>
#include <tuple>
@@ -19,6 +18,7 @@
#include "common/assert.h"
#include "common/common_types.h"
#include "core/file_sys/vfs_vector.h"
+#include "video_core/engines/shader_type.h"
#include "video_core/renderer_opengl/gl_shader_gen.h"
#include "video_core/shader/const_buffer_locker.h"
@@ -37,42 +37,42 @@ struct ShaderDiskCacheDump;
using ProgramCode = std::vector<u64>;
using ShaderDumpsMap = std::unordered_map<ShaderDiskCacheUsage, ShaderDiskCacheDump>;
-using TextureBufferUsage = std::bitset<64>;
-
-/// Allocated bindings used by an OpenGL shader program
-struct BaseBindings {
- u32 cbuf{};
- u32 gmem{};
- u32 sampler{};
- u32 image{};
-
- bool operator==(const BaseBindings& rhs) const {
- return std::tie(cbuf, gmem, sampler, image) ==
- std::tie(rhs.cbuf, rhs.gmem, rhs.sampler, rhs.image);
- }
- bool operator!=(const BaseBindings& rhs) const {
- return !operator==(rhs);
- }
-};
-static_assert(std::is_trivially_copyable_v<BaseBindings>);
+/// Describes the different variants a program can be compiled with.
+struct ProgramVariant final {
+ ProgramVariant() = default;
+
+ /// Graphics constructor.
+ explicit constexpr ProgramVariant(GLenum primitive_mode) noexcept
+ : primitive_mode{primitive_mode} {}
+
+ /// Compute constructor.
+ explicit constexpr ProgramVariant(u32 block_x, u32 block_y, u32 block_z, u32 shared_memory_size,
+ u32 local_memory_size) noexcept
+ : block_x{block_x}, block_y{static_cast<u16>(block_y)}, block_z{static_cast<u16>(block_z)},
+ shared_memory_size{shared_memory_size}, local_memory_size{local_memory_size} {}
-/// Describes the different variants a single program can be compiled.
-struct ProgramVariant {
- BaseBindings base_bindings;
+ // Graphics specific parameters.
GLenum primitive_mode{};
- TextureBufferUsage texture_buffer_usage{};
- bool operator==(const ProgramVariant& rhs) const {
- return std::tie(base_bindings, primitive_mode, texture_buffer_usage) ==
- std::tie(rhs.base_bindings, rhs.primitive_mode, rhs.texture_buffer_usage);
+ // Compute specific parameters.
+ u32 block_x{};
+ u16 block_y{};
+ u16 block_z{};
+ u32 shared_memory_size{};
+ u32 local_memory_size{};
+
+ bool operator==(const ProgramVariant& rhs) const noexcept {
+ return std::tie(primitive_mode, block_x, block_y, block_z, shared_memory_size,
+ local_memory_size) == std::tie(rhs.primitive_mode, rhs.block_x, rhs.block_y,
+ rhs.block_z, rhs.shared_memory_size,
+ rhs.local_memory_size);
}
- bool operator!=(const ProgramVariant& rhs) const {
+ bool operator!=(const ProgramVariant& rhs) const noexcept {
return !operator==(rhs);
}
};
-
static_assert(std::is_trivially_copyable_v<ProgramVariant>);
/// Describes how a shader is used.
@@ -99,21 +99,14 @@ struct ShaderDiskCacheUsage {
namespace std {
template <>
-struct hash<OpenGL::BaseBindings> {
- std::size_t operator()(const OpenGL::BaseBindings& bindings) const noexcept {
- return static_cast<std::size_t>(bindings.cbuf) ^
- (static_cast<std::size_t>(bindings.gmem) << 8) ^
- (static_cast<std::size_t>(bindings.sampler) << 16) ^
- (static_cast<std::size_t>(bindings.image) << 24);
- }
-};
-
-template <>
struct hash<OpenGL::ProgramVariant> {
std::size_t operator()(const OpenGL::ProgramVariant& variant) const noexcept {
- return std::hash<OpenGL::BaseBindings>()(variant.base_bindings) ^
- std::hash<OpenGL::TextureBufferUsage>()(variant.texture_buffer_usage) ^
- (static_cast<std::size_t>(variant.primitive_mode) << 6);
+ return (static_cast<std::size_t>(variant.primitive_mode) << 6) ^
+ static_cast<std::size_t>(variant.block_x) ^
+ (static_cast<std::size_t>(variant.block_y) << 32) ^
+ (static_cast<std::size_t>(variant.block_z) << 48) ^
+ (static_cast<std::size_t>(variant.shared_memory_size) << 16) ^
+ (static_cast<std::size_t>(variant.local_memory_size) << 36);
}
};
@@ -121,7 +114,7 @@ template <>
struct hash<OpenGL::ShaderDiskCacheUsage> {
std::size_t operator()(const OpenGL::ShaderDiskCacheUsage& usage) const noexcept {
return static_cast<std::size_t>(usage.unique_identifier) ^
- std::hash<OpenGL::ProgramVariant>()(usage.variant);
+ std::hash<OpenGL::ProgramVariant>{}(usage.variant);
}
};
@@ -132,8 +125,8 @@ namespace OpenGL {
/// Describes a shader how it's used by the guest GPU
class ShaderDiskCacheRaw {
public:
- explicit ShaderDiskCacheRaw(u64 unique_identifier, ProgramType program_type,
- ProgramCode program_code, ProgramCode program_code_b = {});
+ explicit ShaderDiskCacheRaw(u64 unique_identifier, Tegra::Engines::ShaderType type,
+ ProgramCode code, ProgramCode code_b = {});
ShaderDiskCacheRaw();
~ShaderDiskCacheRaw();
@@ -146,27 +139,26 @@ public:
}
bool HasProgramA() const {
- return program_type == ProgramType::VertexA;
+ return !code.empty() && !code_b.empty();
}
- ProgramType GetProgramType() const {
- return program_type;
+ Tegra::Engines::ShaderType GetType() const {
+ return type;
}
- const ProgramCode& GetProgramCode() const {
- return program_code;
+ const ProgramCode& GetCode() const {
+ return code;
}
- const ProgramCode& GetProgramCodeB() const {
- return program_code_b;
+ const ProgramCode& GetCodeB() const {
+ return code_b;
}
private:
u64 unique_identifier{};
- ProgramType program_type{};
-
- ProgramCode program_code;
- ProgramCode program_code_b;
+ Tegra::Engines::ShaderType type{};
+ ProgramCode code;
+ ProgramCode code_b;
};
/// Contains an OpenGL dumped binary program
diff --git a/src/video_core/renderer_opengl/gl_shader_gen.cpp b/src/video_core/renderer_opengl/gl_shader_gen.cpp
index 0e22eede9..34946fb47 100644
--- a/src/video_core/renderer_opengl/gl_shader_gen.cpp
+++ b/src/video_core/renderer_opengl/gl_shader_gen.cpp
@@ -2,8 +2,13 @@
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
+#include <string>
+
#include <fmt/format.h>
+
#include "video_core/engines/maxwell_3d.h"
+#include "video_core/engines/shader_type.h"
+#include "video_core/renderer_opengl/gl_device.h"
#include "video_core/renderer_opengl/gl_shader_decompiler.h"
#include "video_core/renderer_opengl/gl_shader_gen.h"
#include "video_core/shader/shader_ir.h"
@@ -11,6 +16,7 @@
namespace OpenGL::GLShader {
using Tegra::Engines::Maxwell3D;
+using Tegra::Engines::ShaderType;
using VideoCommon::Shader::CompileDepth;
using VideoCommon::Shader::CompilerSettings;
using VideoCommon::Shader::ProgramCode;
@@ -18,53 +24,40 @@ using VideoCommon::Shader::ShaderIR;
std::string GenerateVertexShader(const Device& device, const ShaderIR& ir, const ShaderIR* ir_b) {
std::string out = GetCommonDeclarations();
- out += R"(
-layout (std140, binding = EMULATION_UBO_BINDING) uniform vs_config {
- vec4 viewport_flip;
- uvec4 config_pack; // instance_id, flip_stage, y_direction, padding
-};
-
-)";
- const auto stage = ir_b ? ProgramType::VertexA : ProgramType::VertexB;
- out += Decompile(device, ir, stage, "vertex");
+ out += fmt::format(R"(
+layout (std140, binding = {}) uniform vs_config {{
+ float y_direction;
+}};
+
+)",
+ EmulationUniformBlockBinding);
+ out += Decompile(device, ir, ShaderType::Vertex, "vertex");
if (ir_b) {
- out += Decompile(device, *ir_b, ProgramType::VertexB, "vertex_b");
+ out += Decompile(device, *ir_b, ShaderType::Vertex, "vertex_b");
}
out += R"(
void main() {
+ gl_Position = vec4(0.0f, 0.0f, 0.0f, 1.0f);
execute_vertex();
)";
-
if (ir_b) {
out += " execute_vertex_b();";
}
-
- out += R"(
-
- // Set Position Y direction
- gl_Position.y *= utof(config_pack[2]);
- // Check if the flip stage is VertexB
- // Config pack's second value is flip_stage
- if (config_pack[1] == 1) {
- // Viewport can be flipped, which is unsupported by glViewport
- gl_Position.xy *= viewport_flip.xy;
- }
-}
-)";
+ out += "}\n";
return out;
}
std::string GenerateGeometryShader(const Device& device, const ShaderIR& ir) {
std::string out = GetCommonDeclarations();
- out += R"(
-layout (std140, binding = EMULATION_UBO_BINDING) uniform gs_config {
- vec4 viewport_flip;
- uvec4 config_pack; // instance_id, flip_stage, y_direction, padding
-};
+ out += fmt::format(R"(
+layout (std140, binding = {}) uniform gs_config {{
+ float y_direction;
+}};
-)";
- out += Decompile(device, ir, ProgramType::Geometry, "geometry");
+)",
+ EmulationUniformBlockBinding);
+ out += Decompile(device, ir, ShaderType::Geometry, "geometry");
out += R"(
void main() {
@@ -76,7 +69,7 @@ void main() {
std::string GenerateFragmentShader(const Device& device, const ShaderIR& ir) {
std::string out = GetCommonDeclarations();
- out += R"(
+ out += fmt::format(R"(
layout (location = 0) out vec4 FragColor0;
layout (location = 1) out vec4 FragColor1;
layout (location = 2) out vec4 FragColor2;
@@ -86,13 +79,13 @@ layout (location = 5) out vec4 FragColor5;
layout (location = 6) out vec4 FragColor6;
layout (location = 7) out vec4 FragColor7;
-layout (std140, binding = EMULATION_UBO_BINDING) uniform fs_config {
- vec4 viewport_flip;
- uvec4 config_pack; // instance_id, flip_stage, y_direction, padding
-};
+layout (std140, binding = {}) uniform fs_config {{
+ float y_direction;
+}};
-)";
- out += Decompile(device, ir, ProgramType::Fragment, "fragment");
+)",
+ EmulationUniformBlockBinding);
+ out += Decompile(device, ir, ShaderType::Fragment, "fragment");
out += R"(
void main() {
@@ -104,7 +97,7 @@ void main() {
std::string GenerateComputeShader(const Device& device, const ShaderIR& ir) {
std::string out = GetCommonDeclarations();
- out += Decompile(device, ir, ProgramType::Compute, "compute");
+ out += Decompile(device, ir, ShaderType::Compute, "compute");
out += R"(
void main() {
execute_compute();
diff --git a/src/video_core/renderer_opengl/gl_shader_manager.cpp b/src/video_core/renderer_opengl/gl_shader_manager.cpp
index b05f90f20..75d3fac04 100644
--- a/src/video_core/renderer_opengl/gl_shader_manager.cpp
+++ b/src/video_core/renderer_opengl/gl_shader_manager.cpp
@@ -40,27 +40,11 @@ void ProgramManager::UpdatePipeline() {
old_state = current_state;
}
-void MaxwellUniformData::SetFromRegs(const Maxwell3D& maxwell, std::size_t shader_stage) {
+void MaxwellUniformData::SetFromRegs(const Maxwell3D& maxwell) {
const auto& regs = maxwell.regs;
- const auto& state = maxwell.state;
-
- // TODO(bunnei): Support more than one viewport
- viewport_flip[0] = regs.viewport_transform[0].scale_x < 0.0 ? -1.0f : 1.0f;
- viewport_flip[1] = regs.viewport_transform[0].scale_y < 0.0 ? -1.0f : 1.0f;
-
- instance_id = state.current_instance;
-
- // Assign in which stage the position has to be flipped
- // (the last stage before the fragment shader).
- constexpr u32 geometry_index = static_cast<u32>(Maxwell3D::Regs::ShaderProgram::Geometry);
- if (maxwell.regs.shader_config[geometry_index].enable) {
- flip_stage = geometry_index;
- } else {
- flip_stage = static_cast<u32>(Maxwell3D::Regs::ShaderProgram::VertexB);
- }
// Y_NEGATE controls what value S2R returns for the Y_DIRECTION system value.
- y_direction = regs.screen_y_control.y_negate == 0 ? 1.f : -1.f;
+ y_direction = regs.screen_y_control.y_negate == 0 ? 1.0f : -1.0f;
}
} // namespace OpenGL::GLShader
diff --git a/src/video_core/renderer_opengl/gl_shader_manager.h b/src/video_core/renderer_opengl/gl_shader_manager.h
index 6961e702a..478c165ce 100644
--- a/src/video_core/renderer_opengl/gl_shader_manager.h
+++ b/src/video_core/renderer_opengl/gl_shader_manager.h
@@ -18,17 +18,12 @@ namespace OpenGL::GLShader {
/// @note Always keep a vec4 at the end. The GL spec is not clear whether the alignment at
/// the end of a uniform block is included in UNIFORM_BLOCK_DATA_SIZE or not.
/// Not following that rule will cause problems on some AMD drivers.
-struct MaxwellUniformData {
- void SetFromRegs(const Tegra::Engines::Maxwell3D& maxwell, std::size_t shader_stage);
-
- alignas(16) GLvec4 viewport_flip;
- struct alignas(16) {
- GLuint instance_id;
- GLuint flip_stage;
- GLfloat y_direction;
- };
+struct alignas(16) MaxwellUniformData {
+ void SetFromRegs(const Tegra::Engines::Maxwell3D& maxwell);
+
+ GLfloat y_direction;
};
-static_assert(sizeof(MaxwellUniformData) == 32, "MaxwellUniformData structure size is incorrect");
+static_assert(sizeof(MaxwellUniformData) == 16, "MaxwellUniformData structure size is incorrect");
static_assert(sizeof(MaxwellUniformData) < 16384,
"MaxwellUniformData structure must be less than 16kb as per the OpenGL spec");
@@ -55,6 +50,10 @@ public:
current_state.geometry_shader = 0;
}
+ void UseTrivialFragmentShader() {
+ current_state.fragment_shader = 0;
+ }
+
private:
struct PipelineState {
bool operator==(const PipelineState& rhs) const {
diff --git a/src/video_core/renderer_opengl/gl_state.cpp b/src/video_core/renderer_opengl/gl_state.cpp
index f25148362..df2e2395a 100644
--- a/src/video_core/renderer_opengl/gl_state.cpp
+++ b/src/video_core/renderer_opengl/gl_state.cpp
@@ -182,6 +182,10 @@ void OpenGLState::ApplyCulling() {
}
}
+void OpenGLState::ApplyRasterizerDiscard() {
+ Enable(GL_RASTERIZER_DISCARD, cur_state.rasterizer_discard, rasterizer_discard);
+}
+
void OpenGLState::ApplyColorMask() {
if (!dirty.color_mask) {
return;
@@ -410,15 +414,32 @@ void OpenGLState::ApplyAlphaTest() {
}
}
+void OpenGLState::ApplyClipControl() {
+ if (UpdateTie(std::tie(cur_state.clip_control.origin, cur_state.clip_control.depth_mode),
+ std::tie(clip_control.origin, clip_control.depth_mode))) {
+ glClipControl(clip_control.origin, clip_control.depth_mode);
+ }
+}
+
void OpenGLState::ApplyTextures() {
- if (const auto update = UpdateArray(cur_state.textures, textures)) {
- glBindTextures(update->first, update->second, textures.data() + update->first);
+ const std::size_t size = std::size(textures);
+ for (std::size_t i = 0; i < size; ++i) {
+ if (UpdateValue(cur_state.textures[i], textures[i])) {
+ // BindTextureUnit doesn't support binding null textures, skip those binds.
+ // TODO(Rodrigo): Stop using null textures
+ if (textures[i] != 0) {
+ glBindTextureUnit(static_cast<GLuint>(i), textures[i]);
+ }
+ }
}
}
void OpenGLState::ApplySamplers() {
- if (const auto update = UpdateArray(cur_state.samplers, samplers)) {
- glBindSamplers(update->first, update->second, samplers.data() + update->first);
+ const std::size_t size = std::size(samplers);
+ for (std::size_t i = 0; i < size; ++i) {
+ if (UpdateValue(cur_state.samplers[i], samplers[i])) {
+ glBindSampler(static_cast<GLuint>(i), samplers[i]);
+ }
}
}
@@ -438,6 +459,7 @@ void OpenGLState::Apply() {
ApplyPointSize();
ApplyFragmentColorClamp();
ApplyMultisample();
+ ApplyRasterizerDiscard();
ApplyColorMask();
ApplyDepthClamp();
ApplyViewport();
@@ -453,6 +475,7 @@ void OpenGLState::Apply() {
ApplyImages();
ApplyPolygonOffset();
ApplyAlphaTest();
+ ApplyClipControl();
}
void OpenGLState::EmulateViewportWithScissor() {
diff --git a/src/video_core/renderer_opengl/gl_state.h b/src/video_core/renderer_opengl/gl_state.h
index cca25206b..fb180f302 100644
--- a/src/video_core/renderer_opengl/gl_state.h
+++ b/src/video_core/renderer_opengl/gl_state.h
@@ -48,6 +48,8 @@ public:
GLuint index = 0;
} primitive_restart; // GL_PRIMITIVE_RESTART
+ bool rasterizer_discard = false; // GL_RASTERIZER_DISCARD
+
struct ColorMask {
GLboolean red_enabled = GL_TRUE;
GLboolean green_enabled = GL_TRUE;
@@ -56,6 +58,7 @@ public:
};
std::array<ColorMask, Tegra::Engines::Maxwell3D::Regs::NumRenderTargets>
color_mask; // GL_COLOR_WRITEMASK
+
struct {
bool test_enabled = false; // GL_STENCIL_TEST
struct {
@@ -96,9 +99,11 @@ public:
GLenum operation = GL_COPY;
} logic_op;
- std::array<GLuint, Tegra::Engines::Maxwell3D::Regs::NumTextureSamplers> textures = {};
- std::array<GLuint, Tegra::Engines::Maxwell3D::Regs::NumTextureSamplers> samplers = {};
- std::array<GLuint, Tegra::Engines::Maxwell3D::Regs::NumImages> images = {};
+ static constexpr std::size_t NumSamplers = 32 * 5;
+ static constexpr std::size_t NumImages = 8 * 5;
+ std::array<GLuint, NumSamplers> textures = {};
+ std::array<GLuint, NumSamplers> samplers = {};
+ std::array<GLuint, NumImages> images = {};
struct {
GLuint read_framebuffer = 0; // GL_READ_FRAMEBUFFER_BINDING
@@ -146,6 +151,11 @@ public:
std::array<bool, 8> clip_distance = {}; // GL_CLIP_DISTANCE
+ struct {
+ GLenum origin = GL_LOWER_LEFT;
+ GLenum depth_mode = GL_NEGATIVE_ONE_TO_ONE;
+ } clip_control;
+
OpenGLState();
/// Get the currently active OpenGL state
@@ -167,6 +177,7 @@ public:
void ApplyMultisample();
void ApplySRgb();
void ApplyCulling();
+ void ApplyRasterizerDiscard();
void ApplyColorMask();
void ApplyDepth();
void ApplyPrimitiveRestart();
@@ -182,6 +193,7 @@ public:
void ApplyDepthClamp();
void ApplyPolygonOffset();
void ApplyAlphaTest();
+ void ApplyClipControl();
/// Resets any references to the given resource
OpenGLState& UnbindTexture(GLuint handle);
diff --git a/src/video_core/renderer_opengl/gl_texture_cache.cpp b/src/video_core/renderer_opengl/gl_texture_cache.cpp
index 55b3e58b2..b790b0ef4 100644
--- a/src/video_core/renderer_opengl/gl_texture_cache.cpp
+++ b/src/video_core/renderer_opengl/gl_texture_cache.cpp
@@ -23,7 +23,6 @@ namespace OpenGL {
using Tegra::Texture::SwizzleSource;
using VideoCore::MortonSwizzleMode;
-using VideoCore::Surface::ComponentType;
using VideoCore::Surface::PixelFormat;
using VideoCore::Surface::SurfaceCompression;
using VideoCore::Surface::SurfaceTarget;
@@ -40,114 +39,95 @@ struct FormatTuple {
GLint internal_format;
GLenum format;
GLenum type;
- ComponentType component_type;
bool compressed;
};
constexpr std::array<FormatTuple, VideoCore::Surface::MaxPixelFormat> tex_format_tuples = {{
- {GL_RGBA8, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8_REV, ComponentType::UNorm, false}, // ABGR8U
- {GL_RGBA8, GL_RGBA, GL_BYTE, ComponentType::SNorm, false}, // ABGR8S
- {GL_RGBA8UI, GL_RGBA_INTEGER, GL_UNSIGNED_BYTE, ComponentType::UInt, false}, // ABGR8UI
- {GL_RGB565, GL_RGB, GL_UNSIGNED_SHORT_5_6_5_REV, ComponentType::UNorm, false}, // B5G6R5U
- {GL_RGB10_A2, GL_RGBA, GL_UNSIGNED_INT_2_10_10_10_REV, ComponentType::UNorm,
- false}, // A2B10G10R10U
- {GL_RGB5_A1, GL_RGBA, GL_UNSIGNED_SHORT_1_5_5_5_REV, ComponentType::UNorm, false}, // A1B5G5R5U
- {GL_R8, GL_RED, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // R8U
- {GL_R8UI, GL_RED_INTEGER, GL_UNSIGNED_BYTE, ComponentType::UInt, false}, // R8UI
- {GL_RGBA16F, GL_RGBA, GL_HALF_FLOAT, ComponentType::Float, false}, // RGBA16F
- {GL_RGBA16, GL_RGBA, GL_UNSIGNED_SHORT, ComponentType::UNorm, false}, // RGBA16U
- {GL_RGBA16UI, GL_RGBA_INTEGER, GL_UNSIGNED_SHORT, ComponentType::UInt, false}, // RGBA16UI
- {GL_R11F_G11F_B10F, GL_RGB, GL_UNSIGNED_INT_10F_11F_11F_REV, ComponentType::Float,
- false}, // R11FG11FB10F
- {GL_RGBA32UI, GL_RGBA_INTEGER, GL_UNSIGNED_INT, ComponentType::UInt, false}, // RGBA32UI
- {GL_COMPRESSED_RGBA_S3TC_DXT1_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, ComponentType::UNorm,
- true}, // DXT1
- {GL_COMPRESSED_RGBA_S3TC_DXT3_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, ComponentType::UNorm,
- true}, // DXT23
- {GL_COMPRESSED_RGBA_S3TC_DXT5_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, ComponentType::UNorm,
- true}, // DXT45
- {GL_COMPRESSED_RED_RGTC1, GL_RED, GL_UNSIGNED_INT_8_8_8_8, ComponentType::UNorm, true}, // DXN1
- {GL_COMPRESSED_RG_RGTC2, GL_RG, GL_UNSIGNED_INT_8_8_8_8, ComponentType::UNorm,
- true}, // DXN2UNORM
- {GL_COMPRESSED_SIGNED_RG_RGTC2, GL_RG, GL_INT, ComponentType::SNorm, true}, // DXN2SNORM
- {GL_COMPRESSED_RGBA_BPTC_UNORM, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, ComponentType::UNorm,
- true}, // BC7U
- {GL_COMPRESSED_RGB_BPTC_UNSIGNED_FLOAT, GL_RGB, GL_UNSIGNED_INT_8_8_8_8, ComponentType::Float,
- true}, // BC6H_UF16
- {GL_COMPRESSED_RGB_BPTC_SIGNED_FLOAT, GL_RGB, GL_UNSIGNED_INT_8_8_8_8, ComponentType::Float,
- true}, // BC6H_SF16
- {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_4X4
- {GL_RGBA8, GL_BGRA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // BGRA8
- {GL_RGBA32F, GL_RGBA, GL_FLOAT, ComponentType::Float, false}, // RGBA32F
- {GL_RG32F, GL_RG, GL_FLOAT, ComponentType::Float, false}, // RG32F
- {GL_R32F, GL_RED, GL_FLOAT, ComponentType::Float, false}, // R32F
- {GL_R16F, GL_RED, GL_HALF_FLOAT, ComponentType::Float, false}, // R16F
- {GL_R16, GL_RED, GL_UNSIGNED_SHORT, ComponentType::UNorm, false}, // R16U
- {GL_R16_SNORM, GL_RED, GL_SHORT, ComponentType::SNorm, false}, // R16S
- {GL_R16UI, GL_RED_INTEGER, GL_UNSIGNED_SHORT, ComponentType::UInt, false}, // R16UI
- {GL_R16I, GL_RED_INTEGER, GL_SHORT, ComponentType::SInt, false}, // R16I
- {GL_RG16, GL_RG, GL_UNSIGNED_SHORT, ComponentType::UNorm, false}, // RG16
- {GL_RG16F, GL_RG, GL_HALF_FLOAT, ComponentType::Float, false}, // RG16F
- {GL_RG16UI, GL_RG_INTEGER, GL_UNSIGNED_SHORT, ComponentType::UInt, false}, // RG16UI
- {GL_RG16I, GL_RG_INTEGER, GL_SHORT, ComponentType::SInt, false}, // RG16I
- {GL_RG16_SNORM, GL_RG, GL_SHORT, ComponentType::SNorm, false}, // RG16S
- {GL_RGB32F, GL_RGB, GL_FLOAT, ComponentType::Float, false}, // RGB32F
- {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8_REV, ComponentType::UNorm,
- false}, // RGBA8_SRGB
- {GL_RG8, GL_RG, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // RG8U
- {GL_RG8, GL_RG, GL_BYTE, ComponentType::SNorm, false}, // RG8S
- {GL_RG32UI, GL_RG_INTEGER, GL_UNSIGNED_INT, ComponentType::UInt, false}, // RG32UI
- {GL_RGB16F, GL_RGBA16, GL_HALF_FLOAT, ComponentType::Float, false}, // RGBX16F
- {GL_R32UI, GL_RED_INTEGER, GL_UNSIGNED_INT, ComponentType::UInt, false}, // R32UI
- {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_8X8
- {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_8X5
- {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_5X4
- {GL_SRGB8_ALPHA8, GL_BGRA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // BGRA8
+ {GL_RGBA8, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8_REV, false}, // ABGR8U
+ {GL_RGBA8, GL_RGBA, GL_BYTE, false}, // ABGR8S
+ {GL_RGBA8UI, GL_RGBA_INTEGER, GL_UNSIGNED_BYTE, false}, // ABGR8UI
+ {GL_RGB565, GL_RGB, GL_UNSIGNED_SHORT_5_6_5_REV, false}, // B5G6R5U
+ {GL_RGB10_A2, GL_RGBA, GL_UNSIGNED_INT_2_10_10_10_REV, false}, // A2B10G10R10U
+ {GL_RGB5_A1, GL_RGBA, GL_UNSIGNED_SHORT_1_5_5_5_REV, false}, // A1B5G5R5U
+ {GL_R8, GL_RED, GL_UNSIGNED_BYTE, false}, // R8U
+ {GL_R8UI, GL_RED_INTEGER, GL_UNSIGNED_BYTE, false}, // R8UI
+ {GL_RGBA16F, GL_RGBA, GL_HALF_FLOAT, false}, // RGBA16F
+ {GL_RGBA16, GL_RGBA, GL_UNSIGNED_SHORT, false}, // RGBA16U
+ {GL_RGBA16UI, GL_RGBA_INTEGER, GL_UNSIGNED_SHORT, false}, // RGBA16UI
+ {GL_R11F_G11F_B10F, GL_RGB, GL_UNSIGNED_INT_10F_11F_11F_REV, false}, // R11FG11FB10F
+ {GL_RGBA32UI, GL_RGBA_INTEGER, GL_UNSIGNED_INT, false}, // RGBA32UI
+ {GL_COMPRESSED_RGBA_S3TC_DXT1_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, true}, // DXT1
+ {GL_COMPRESSED_RGBA_S3TC_DXT3_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, true}, // DXT23
+ {GL_COMPRESSED_RGBA_S3TC_DXT5_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, true}, // DXT45
+ {GL_COMPRESSED_RED_RGTC1, GL_RED, GL_UNSIGNED_INT_8_8_8_8, true}, // DXN1
+ {GL_COMPRESSED_RG_RGTC2, GL_RG, GL_UNSIGNED_INT_8_8_8_8, true}, // DXN2UNORM
+ {GL_COMPRESSED_SIGNED_RG_RGTC2, GL_RG, GL_INT, true}, // DXN2SNORM
+ {GL_COMPRESSED_RGBA_BPTC_UNORM, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, true}, // BC7U
+ {GL_COMPRESSED_RGB_BPTC_UNSIGNED_FLOAT, GL_RGB, GL_UNSIGNED_INT_8_8_8_8, true}, // BC6H_UF16
+ {GL_COMPRESSED_RGB_BPTC_SIGNED_FLOAT, GL_RGB, GL_UNSIGNED_INT_8_8_8_8, true}, // BC6H_SF16
+ {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, false}, // ASTC_2D_4X4
+ {GL_RGBA8, GL_BGRA, GL_UNSIGNED_BYTE, false}, // BGRA8
+ {GL_RGBA32F, GL_RGBA, GL_FLOAT, false}, // RGBA32F
+ {GL_RG32F, GL_RG, GL_FLOAT, false}, // RG32F
+ {GL_R32F, GL_RED, GL_FLOAT, false}, // R32F
+ {GL_R16F, GL_RED, GL_HALF_FLOAT, false}, // R16F
+ {GL_R16, GL_RED, GL_UNSIGNED_SHORT, false}, // R16U
+ {GL_R16_SNORM, GL_RED, GL_SHORT, false}, // R16S
+ {GL_R16UI, GL_RED_INTEGER, GL_UNSIGNED_SHORT, false}, // R16UI
+ {GL_R16I, GL_RED_INTEGER, GL_SHORT, false}, // R16I
+ {GL_RG16, GL_RG, GL_UNSIGNED_SHORT, false}, // RG16
+ {GL_RG16F, GL_RG, GL_HALF_FLOAT, false}, // RG16F
+ {GL_RG16UI, GL_RG_INTEGER, GL_UNSIGNED_SHORT, false}, // RG16UI
+ {GL_RG16I, GL_RG_INTEGER, GL_SHORT, false}, // RG16I
+ {GL_RG16_SNORM, GL_RG, GL_SHORT, false}, // RG16S
+ {GL_RGB32F, GL_RGB, GL_FLOAT, false}, // RGB32F
+ {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8_REV, false}, // RGBA8_SRGB
+ {GL_RG8, GL_RG, GL_UNSIGNED_BYTE, false}, // RG8U
+ {GL_RG8, GL_RG, GL_BYTE, false}, // RG8S
+ {GL_RG32UI, GL_RG_INTEGER, GL_UNSIGNED_INT, false}, // RG32UI
+ {GL_RGB16F, GL_RGBA16, GL_HALF_FLOAT, false}, // RGBX16F
+ {GL_R32UI, GL_RED_INTEGER, GL_UNSIGNED_INT, false}, // R32UI
+ {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, false}, // ASTC_2D_8X8
+ {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, false}, // ASTC_2D_8X5
+ {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, false}, // ASTC_2D_5X4
+ {GL_SRGB8_ALPHA8, GL_BGRA, GL_UNSIGNED_BYTE, false}, // BGRA8
// Compressed sRGB formats
- {GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT1_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, ComponentType::UNorm,
- true}, // DXT1_SRGB
- {GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT3_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, ComponentType::UNorm,
- true}, // DXT23_SRGB
- {GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT5_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, ComponentType::UNorm,
- true}, // DXT45_SRGB
- {GL_COMPRESSED_SRGB_ALPHA_BPTC_UNORM, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, ComponentType::UNorm,
- true}, // BC7U_SRGB
- {GL_RGBA4, GL_RGBA, GL_UNSIGNED_SHORT_4_4_4_4_REV, ComponentType::UNorm, false}, // R4G4B4A4U
- {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_4X4_SRGB
- {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_8X8_SRGB
- {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_8X5_SRGB
- {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_5X4_SRGB
- {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_5X5
- {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_5X5_SRGB
- {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_10X8
- {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_10X8_SRGB
- {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_6X6
- {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_6X6_SRGB
- {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_10X10
- {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_10X10_SRGB
- {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_12X12
- {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_12X12_SRGB
- {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_8X6
- {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_8X6_SRGB
- {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_6X5
- {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_6X5_SRGB
- {GL_RGB9_E5, GL_RGB, GL_UNSIGNED_INT_5_9_9_9_REV, ComponentType::Float, false}, // E5B9G9R9F
+ {GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT1_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, true}, // DXT1_SRGB
+ {GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT3_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, true}, // DXT23_SRGB
+ {GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT5_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, true}, // DXT45_SRGB
+ {GL_COMPRESSED_SRGB_ALPHA_BPTC_UNORM, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, true}, // BC7U_SRGB
+ {GL_RGBA4, GL_RGBA, GL_UNSIGNED_SHORT_4_4_4_4_REV, false}, // R4G4B4A4U
+ {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, false}, // ASTC_2D_4X4_SRGB
+ {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, false}, // ASTC_2D_8X8_SRGB
+ {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, false}, // ASTC_2D_8X5_SRGB
+ {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, false}, // ASTC_2D_5X4_SRGB
+ {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, false}, // ASTC_2D_5X5
+ {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, false}, // ASTC_2D_5X5_SRGB
+ {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, false}, // ASTC_2D_10X8
+ {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, false}, // ASTC_2D_10X8_SRGB
+ {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, false}, // ASTC_2D_6X6
+ {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, false}, // ASTC_2D_6X6_SRGB
+ {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, false}, // ASTC_2D_10X10
+ {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, false}, // ASTC_2D_10X10_SRGB
+ {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, false}, // ASTC_2D_12X12
+ {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, false}, // ASTC_2D_12X12_SRGB
+ {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, false}, // ASTC_2D_8X6
+ {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, false}, // ASTC_2D_8X6_SRGB
+ {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, false}, // ASTC_2D_6X5
+ {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, false}, // ASTC_2D_6X5_SRGB
+ {GL_RGB9_E5, GL_RGB, GL_UNSIGNED_INT_5_9_9_9_REV, false}, // E5B9G9R9F
// Depth formats
- {GL_DEPTH_COMPONENT32F, GL_DEPTH_COMPONENT, GL_FLOAT, ComponentType::Float, false}, // Z32F
- {GL_DEPTH_COMPONENT16, GL_DEPTH_COMPONENT, GL_UNSIGNED_SHORT, ComponentType::UNorm,
- false}, // Z16
+ {GL_DEPTH_COMPONENT32F, GL_DEPTH_COMPONENT, GL_FLOAT, false}, // Z32F
+ {GL_DEPTH_COMPONENT16, GL_DEPTH_COMPONENT, GL_UNSIGNED_SHORT, false}, // Z16
// DepthStencil formats
- {GL_DEPTH24_STENCIL8, GL_DEPTH_STENCIL, GL_UNSIGNED_INT_24_8, ComponentType::UNorm,
- false}, // Z24S8
- {GL_DEPTH24_STENCIL8, GL_DEPTH_STENCIL, GL_UNSIGNED_INT_24_8, ComponentType::UNorm,
- false}, // S8Z24
- {GL_DEPTH32F_STENCIL8, GL_DEPTH_STENCIL, GL_FLOAT_32_UNSIGNED_INT_24_8_REV,
- ComponentType::Float, false}, // Z32FS8
+ {GL_DEPTH24_STENCIL8, GL_DEPTH_STENCIL, GL_UNSIGNED_INT_24_8, false}, // Z24S8
+ {GL_DEPTH24_STENCIL8, GL_DEPTH_STENCIL, GL_UNSIGNED_INT_24_8, false}, // S8Z24
+ {GL_DEPTH32F_STENCIL8, GL_DEPTH_STENCIL, GL_FLOAT_32_UNSIGNED_INT_24_8_REV, false}, // Z32FS8
}};
-const FormatTuple& GetFormatTuple(PixelFormat pixel_format, ComponentType component_type) {
+const FormatTuple& GetFormatTuple(PixelFormat pixel_format) {
ASSERT(static_cast<std::size_t>(pixel_format) < tex_format_tuples.size());
const auto& format{tex_format_tuples[static_cast<std::size_t>(pixel_format)]};
return format;
@@ -249,7 +229,7 @@ OGLTexture CreateTexture(const SurfaceParams& params, GLenum target, GLenum inte
CachedSurface::CachedSurface(const GPUVAddr gpu_addr, const SurfaceParams& params)
: VideoCommon::SurfaceBase<View>(gpu_addr, params) {
- const auto& tuple{GetFormatTuple(params.pixel_format, params.component_type)};
+ const auto& tuple{GetFormatTuple(params.pixel_format)};
internal_format = tuple.internal_format;
format = tuple.format;
type = tuple.type;
@@ -451,8 +431,7 @@ OGLTextureView CachedSurfaceView::CreateTextureView() const {
texture_view.Create();
const GLuint handle{texture_view.handle};
- const FormatTuple& tuple{
- GetFormatTuple(owner_params.pixel_format, owner_params.component_type)};
+ const FormatTuple& tuple{GetFormatTuple(owner_params.pixel_format)};
glTextureView(handle, target, surface.texture.handle, tuple.internal_format, params.base_level,
params.num_levels, params.base_layer, params.num_layers);
@@ -509,6 +488,7 @@ void TextureCacheOpenGL::ImageBlit(View& src_view, View& dst_view,
OpenGLState state;
state.draw.read_framebuffer = src_framebuffer.handle;
state.draw.draw_framebuffer = dst_framebuffer.handle;
+ state.framebuffer_srgb.enabled = dst_params.srgb_conversion;
state.AllDirty();
state.Apply();
@@ -562,8 +542,8 @@ void TextureCacheOpenGL::BufferCopy(Surface& src_surface, Surface& dst_surface)
const auto& dst_params = dst_surface->GetSurfaceParams();
UNIMPLEMENTED_IF(src_params.num_levels > 1 || dst_params.num_levels > 1);
- const auto source_format = GetFormatTuple(src_params.pixel_format, src_params.component_type);
- const auto dest_format = GetFormatTuple(dst_params.pixel_format, dst_params.component_type);
+ const auto source_format = GetFormatTuple(src_params.pixel_format);
+ const auto dest_format = GetFormatTuple(dst_params.pixel_format);
const std::size_t source_size = src_surface->GetHostSizeInBytes();
const std::size_t dest_size = dst_surface->GetHostSizeInBytes();
diff --git a/src/video_core/renderer_opengl/maxwell_to_gl.h b/src/video_core/renderer_opengl/maxwell_to_gl.h
index 9ed738171..ea4f35663 100644
--- a/src/video_core/renderer_opengl/maxwell_to_gl.h
+++ b/src/video_core/renderer_opengl/maxwell_to_gl.h
@@ -120,6 +120,8 @@ inline GLenum PrimitiveTopology(Maxwell::PrimitiveTopology topology) {
return GL_POINTS;
case Maxwell::PrimitiveTopology::Lines:
return GL_LINES;
+ case Maxwell::PrimitiveTopology::LineLoop:
+ return GL_LINE_LOOP;
case Maxwell::PrimitiveTopology::LineStrip:
return GL_LINE_STRIP;
case Maxwell::PrimitiveTopology::Triangles:
@@ -130,11 +132,23 @@ inline GLenum PrimitiveTopology(Maxwell::PrimitiveTopology topology) {
return GL_TRIANGLE_FAN;
case Maxwell::PrimitiveTopology::Quads:
return GL_QUADS;
- default:
- LOG_CRITICAL(Render_OpenGL, "Unimplemented topology={}", static_cast<u32>(topology));
- UNREACHABLE();
- return {};
+ case Maxwell::PrimitiveTopology::QuadStrip:
+ return GL_QUAD_STRIP;
+ case Maxwell::PrimitiveTopology::Polygon:
+ return GL_POLYGON;
+ case Maxwell::PrimitiveTopology::LinesAdjacency:
+ return GL_LINES_ADJACENCY;
+ case Maxwell::PrimitiveTopology::LineStripAdjacency:
+ return GL_LINE_STRIP_ADJACENCY;
+ case Maxwell::PrimitiveTopology::TrianglesAdjacency:
+ return GL_TRIANGLES_ADJACENCY;
+ case Maxwell::PrimitiveTopology::TriangleStripAdjacency:
+ return GL_TRIANGLE_STRIP_ADJACENCY;
+ case Maxwell::PrimitiveTopology::Patches:
+ return GL_PATCHES;
}
+ UNREACHABLE_MSG("Invalid topology={}", static_cast<int>(topology));
+ return GL_POINTS;
}
inline GLenum TextureFilterMode(Tegra::Texture::TextureFilter filter_mode,
diff --git a/src/video_core/renderer_opengl/renderer_opengl.cpp b/src/video_core/renderer_opengl/renderer_opengl.cpp
index 4bbd17b12..bba16afaf 100644
--- a/src/video_core/renderer_opengl/renderer_opengl.cpp
+++ b/src/video_core/renderer_opengl/renderer_opengl.cpp
@@ -24,19 +24,21 @@
namespace OpenGL {
-static const char vertex_shader[] = R"(
-#version 150 core
+namespace {
-in vec2 vert_position;
-in vec2 vert_tex_coord;
-out vec2 frag_tex_coord;
+constexpr char vertex_shader[] = R"(
+#version 430 core
+
+layout (location = 0) in vec2 vert_position;
+layout (location = 1) in vec2 vert_tex_coord;
+layout (location = 0) out vec2 frag_tex_coord;
// This is a truncated 3x3 matrix for 2D transformations:
// The upper-left 2x2 submatrix performs scaling/rotation/mirroring.
// The third column performs translation.
// The third row could be used for projection, which we don't need in 2D. It hence is assumed to
// implicitly be [0, 0, 1]
-uniform mat3x2 modelview_matrix;
+layout (location = 0) uniform mat3x2 modelview_matrix;
void main() {
// Multiply input position by the rotscale part of the matrix and then manually translate by
@@ -47,34 +49,29 @@ void main() {
}
)";
-static const char fragment_shader[] = R"(
-#version 150 core
+constexpr char fragment_shader[] = R"(
+#version 430 core
-in vec2 frag_tex_coord;
-out vec4 color;
+layout (location = 0) in vec2 frag_tex_coord;
+layout (location = 0) out vec4 color;
-uniform sampler2D color_texture;
+layout (binding = 0) uniform sampler2D color_texture;
void main() {
- // Swap RGBA -> ABGR so we don't have to do this on the CPU. This needs to change if we have to
- // support more framebuffer pixel formats.
color = texture(color_texture, frag_tex_coord);
}
)";
-/**
- * Vertex structure that the drawn screen rectangles are composed of.
- */
+constexpr GLint PositionLocation = 0;
+constexpr GLint TexCoordLocation = 1;
+constexpr GLint ModelViewMatrixLocation = 0;
+
struct ScreenRectVertex {
- ScreenRectVertex(GLfloat x, GLfloat y, GLfloat u, GLfloat v) {
- position[0] = x;
- position[1] = y;
- tex_coord[0] = u;
- tex_coord[1] = v;
- }
+ constexpr ScreenRectVertex(GLfloat x, GLfloat y, GLfloat u, GLfloat v)
+ : position{{x, y}}, tex_coord{{u, v}} {}
- GLfloat position[2];
- GLfloat tex_coord[2];
+ std::array<GLfloat, 2> position;
+ std::array<GLfloat, 2> tex_coord;
};
/**
@@ -84,18 +81,82 @@ struct ScreenRectVertex {
* The projection part of the matrix is trivial, hence these operations are represented
* by a 3x2 matrix.
*/
-static std::array<GLfloat, 3 * 2> MakeOrthographicMatrix(const float width, const float height) {
+std::array<GLfloat, 3 * 2> MakeOrthographicMatrix(float width, float height) {
std::array<GLfloat, 3 * 2> matrix; // Laid out in column-major order
// clang-format off
- matrix[0] = 2.f / width; matrix[2] = 0.f; matrix[4] = -1.f;
- matrix[1] = 0.f; matrix[3] = -2.f / height; matrix[5] = 1.f;
+ matrix[0] = 2.f / width; matrix[2] = 0.f; matrix[4] = -1.f;
+ matrix[1] = 0.f; matrix[3] = -2.f / height; matrix[5] = 1.f;
// Last matrix row is implicitly assumed to be [0, 0, 1].
// clang-format on
return matrix;
}
+const char* GetSource(GLenum source) {
+ switch (source) {
+ case GL_DEBUG_SOURCE_API:
+ return "API";
+ case GL_DEBUG_SOURCE_WINDOW_SYSTEM:
+ return "WINDOW_SYSTEM";
+ case GL_DEBUG_SOURCE_SHADER_COMPILER:
+ return "SHADER_COMPILER";
+ case GL_DEBUG_SOURCE_THIRD_PARTY:
+ return "THIRD_PARTY";
+ case GL_DEBUG_SOURCE_APPLICATION:
+ return "APPLICATION";
+ case GL_DEBUG_SOURCE_OTHER:
+ return "OTHER";
+ default:
+ UNREACHABLE();
+ return "Unknown source";
+ }
+}
+
+const char* GetType(GLenum type) {
+ switch (type) {
+ case GL_DEBUG_TYPE_ERROR:
+ return "ERROR";
+ case GL_DEBUG_TYPE_DEPRECATED_BEHAVIOR:
+ return "DEPRECATED_BEHAVIOR";
+ case GL_DEBUG_TYPE_UNDEFINED_BEHAVIOR:
+ return "UNDEFINED_BEHAVIOR";
+ case GL_DEBUG_TYPE_PORTABILITY:
+ return "PORTABILITY";
+ case GL_DEBUG_TYPE_PERFORMANCE:
+ return "PERFORMANCE";
+ case GL_DEBUG_TYPE_OTHER:
+ return "OTHER";
+ case GL_DEBUG_TYPE_MARKER:
+ return "MARKER";
+ default:
+ UNREACHABLE();
+ return "Unknown type";
+ }
+}
+
+void APIENTRY DebugHandler(GLenum source, GLenum type, GLuint id, GLenum severity, GLsizei length,
+ const GLchar* message, const void* user_param) {
+ const char format[] = "{} {} {}: {}";
+ const char* const str_source = GetSource(source);
+ const char* const str_type = GetType(type);
+
+ switch (severity) {
+ case GL_DEBUG_SEVERITY_HIGH:
+ LOG_CRITICAL(Render_OpenGL, format, str_source, str_type, id, message);
+ break;
+ case GL_DEBUG_SEVERITY_MEDIUM:
+ LOG_WARNING(Render_OpenGL, format, str_source, str_type, id, message);
+ break;
+ case GL_DEBUG_SEVERITY_NOTIFICATION:
+ case GL_DEBUG_SEVERITY_LOW:
+ LOG_DEBUG(Render_OpenGL, format, str_source, str_type, id, message);
+ break;
+ }
+}
+
+} // Anonymous namespace
+
RendererOpenGL::RendererOpenGL(Core::Frontend::EmuWindow& emu_window, Core::System& system)
: VideoCore::RendererBase{emu_window}, emu_window{emu_window}, system{system} {}
@@ -138,9 +199,6 @@ void RendererOpenGL::SwapBuffers(const Tegra::FramebufferConfig* framebuffer) {
prev_state.Apply();
}
-/**
- * Loads framebuffer from emulated memory into the active OpenGL texture.
- */
void RendererOpenGL::LoadFBToScreenInfo(const Tegra::FramebufferConfig& framebuffer) {
// Framebuffer orientation handling
framebuffer_transform_flags = framebuffer.transform_flags;
@@ -158,7 +216,7 @@ void RendererOpenGL::LoadFBToScreenInfo(const Tegra::FramebufferConfig& framebuf
VideoCore::Surface::PixelFormatFromGPUPixelFormat(framebuffer.pixel_format)};
const u32 bytes_per_pixel{VideoCore::Surface::GetBytesPerPixel(pixel_format)};
const u64 size_in_bytes{framebuffer.stride * framebuffer.height * bytes_per_pixel};
- const auto host_ptr{Memory::GetPointer(framebuffer_addr)};
+ u8* const host_ptr{system.Memory().GetPointer(framebuffer_addr)};
rasterizer->FlushRegion(ToCacheAddr(host_ptr), size_in_bytes);
// TODO(Rodrigo): Read this from HLE
@@ -181,19 +239,12 @@ void RendererOpenGL::LoadFBToScreenInfo(const Tegra::FramebufferConfig& framebuf
glPixelStorei(GL_UNPACK_ROW_LENGTH, 0);
}
-/**
- * Fills active OpenGL texture with the given RGB color. Since the color is solid, the texture can
- * be 1x1 but will stretch across whatever it's rendered on.
- */
void RendererOpenGL::LoadColorToActiveGLTexture(u8 color_r, u8 color_g, u8 color_b, u8 color_a,
const TextureInfo& texture) {
const u8 framebuffer_data[4] = {color_a, color_b, color_g, color_r};
glClearTexImage(texture.resource.handle, 0, GL_RGBA, GL_UNSIGNED_BYTE, framebuffer_data);
}
-/**
- * Initializes the OpenGL state and creates persistent objects.
- */
void RendererOpenGL::InitOpenGLObjects() {
glClearColor(Settings::values.bg_red, Settings::values.bg_green, Settings::values.bg_blue,
0.0f);
@@ -203,10 +254,6 @@ void RendererOpenGL::InitOpenGLObjects() {
state.draw.shader_program = shader.handle;
state.AllDirty();
state.Apply();
- uniform_modelview_matrix = glGetUniformLocation(shader.handle, "modelview_matrix");
- uniform_color_texture = glGetUniformLocation(shader.handle, "color_texture");
- attrib_position = glGetAttribLocation(shader.handle, "vert_position");
- attrib_tex_coord = glGetAttribLocation(shader.handle, "vert_tex_coord");
// Generate VBO handle for drawing
vertex_buffer.Create();
@@ -217,14 +264,14 @@ void RendererOpenGL::InitOpenGLObjects() {
// Attach vertex data to VAO
glNamedBufferData(vertex_buffer.handle, sizeof(ScreenRectVertex) * 4, nullptr, GL_STREAM_DRAW);
- glVertexArrayAttribFormat(vertex_array.handle, attrib_position, 2, GL_FLOAT, GL_FALSE,
+ glVertexArrayAttribFormat(vertex_array.handle, PositionLocation, 2, GL_FLOAT, GL_FALSE,
offsetof(ScreenRectVertex, position));
- glVertexArrayAttribFormat(vertex_array.handle, attrib_tex_coord, 2, GL_FLOAT, GL_FALSE,
+ glVertexArrayAttribFormat(vertex_array.handle, TexCoordLocation, 2, GL_FLOAT, GL_FALSE,
offsetof(ScreenRectVertex, tex_coord));
- glVertexArrayAttribBinding(vertex_array.handle, attrib_position, 0);
- glVertexArrayAttribBinding(vertex_array.handle, attrib_tex_coord, 0);
- glEnableVertexArrayAttrib(vertex_array.handle, attrib_position);
- glEnableVertexArrayAttrib(vertex_array.handle, attrib_tex_coord);
+ glVertexArrayAttribBinding(vertex_array.handle, PositionLocation, 0);
+ glVertexArrayAttribBinding(vertex_array.handle, TexCoordLocation, 0);
+ glEnableVertexArrayAttrib(vertex_array.handle, PositionLocation);
+ glEnableVertexArrayAttrib(vertex_array.handle, TexCoordLocation);
glVertexArrayVertexBuffer(vertex_array.handle, 0, vertex_buffer.handle, 0,
sizeof(ScreenRectVertex));
@@ -323,24 +370,26 @@ void RendererOpenGL::DrawScreenTriangles(const ScreenInfo& screen_info, float x,
// (e.g. handheld mode) on a 1920x1080 framebuffer.
f32 scale_u = 1.f, scale_v = 1.f;
if (framebuffer_crop_rect.GetWidth() > 0) {
- scale_u = static_cast<f32>(framebuffer_crop_rect.GetWidth()) / screen_info.texture.width;
+ scale_u = static_cast<f32>(framebuffer_crop_rect.GetWidth()) /
+ static_cast<f32>(screen_info.texture.width);
}
if (framebuffer_crop_rect.GetHeight() > 0) {
- scale_v = static_cast<f32>(framebuffer_crop_rect.GetHeight()) / screen_info.texture.height;
+ scale_v = static_cast<f32>(framebuffer_crop_rect.GetHeight()) /
+ static_cast<f32>(screen_info.texture.height);
}
- std::array<ScreenRectVertex, 4> vertices = {{
+ const std::array vertices = {
ScreenRectVertex(x, y, texcoords.top * scale_u, left * scale_v),
ScreenRectVertex(x + w, y, texcoords.bottom * scale_u, left * scale_v),
ScreenRectVertex(x, y + h, texcoords.top * scale_u, right * scale_v),
ScreenRectVertex(x + w, y + h, texcoords.bottom * scale_u, right * scale_v),
- }};
+ };
state.textures[0] = screen_info.display_texture;
state.framebuffer_srgb.enabled = screen_info.display_srgb;
state.AllDirty();
state.Apply();
- glNamedBufferSubData(vertex_buffer.handle, 0, sizeof(vertices), vertices.data());
+ glNamedBufferSubData(vertex_buffer.handle, 0, sizeof(vertices), std::data(vertices));
glDrawArrays(GL_TRIANGLE_STRIP, 0, 4);
// Restore default state
state.framebuffer_srgb.enabled = false;
@@ -349,9 +398,6 @@ void RendererOpenGL::DrawScreenTriangles(const ScreenInfo& screen_info, float x,
state.Apply();
}
-/**
- * Draws the emulated screens to the emulator window.
- */
void RendererOpenGL::DrawScreen(const Layout::FramebufferLayout& layout) {
if (renderer_settings.set_background_color) {
// Update background color before drawing
@@ -365,21 +411,17 @@ void RendererOpenGL::DrawScreen(const Layout::FramebufferLayout& layout) {
glClear(GL_COLOR_BUFFER_BIT);
// Set projection matrix
- std::array<GLfloat, 3 * 2> ortho_matrix =
- MakeOrthographicMatrix((float)layout.width, (float)layout.height);
- glUniformMatrix3x2fv(uniform_modelview_matrix, 1, GL_FALSE, ortho_matrix.data());
-
- // Bind texture in Texture Unit 0
- glActiveTexture(GL_TEXTURE0);
- glUniform1i(uniform_color_texture, 0);
+ const std::array ortho_matrix =
+ MakeOrthographicMatrix(static_cast<float>(layout.width), static_cast<float>(layout.height));
+ glUniformMatrix3x2fv(ModelViewMatrixLocation, 1, GL_FALSE, ortho_matrix.data());
- DrawScreenTriangles(screen_info, (float)screen.left, (float)screen.top,
- (float)screen.GetWidth(), (float)screen.GetHeight());
+ DrawScreenTriangles(screen_info, static_cast<float>(screen.left),
+ static_cast<float>(screen.top), static_cast<float>(screen.GetWidth()),
+ static_cast<float>(screen.GetHeight()));
m_current_frame++;
}
-/// Updates the framerate
void RendererOpenGL::UpdateFramerate() {}
void RendererOpenGL::CaptureScreenshot() {
@@ -416,63 +458,6 @@ void RendererOpenGL::CaptureScreenshot() {
renderer_settings.screenshot_requested = false;
}
-static const char* GetSource(GLenum source) {
-#define RET(s) \
- case GL_DEBUG_SOURCE_##s: \
- return #s
- switch (source) {
- RET(API);
- RET(WINDOW_SYSTEM);
- RET(SHADER_COMPILER);
- RET(THIRD_PARTY);
- RET(APPLICATION);
- RET(OTHER);
- default:
- UNREACHABLE();
- return "Unknown source";
- }
-#undef RET
-}
-
-static const char* GetType(GLenum type) {
-#define RET(t) \
- case GL_DEBUG_TYPE_##t: \
- return #t
- switch (type) {
- RET(ERROR);
- RET(DEPRECATED_BEHAVIOR);
- RET(UNDEFINED_BEHAVIOR);
- RET(PORTABILITY);
- RET(PERFORMANCE);
- RET(OTHER);
- RET(MARKER);
- default:
- UNREACHABLE();
- return "Unknown type";
- }
-#undef RET
-}
-
-static void APIENTRY DebugHandler(GLenum source, GLenum type, GLuint id, GLenum severity,
- GLsizei length, const GLchar* message, const void* user_param) {
- const char format[] = "{} {} {}: {}";
- const char* const str_source = GetSource(source);
- const char* const str_type = GetType(type);
-
- switch (severity) {
- case GL_DEBUG_SEVERITY_HIGH:
- LOG_CRITICAL(Render_OpenGL, format, str_source, str_type, id, message);
- break;
- case GL_DEBUG_SEVERITY_MEDIUM:
- LOG_WARNING(Render_OpenGL, format, str_source, str_type, id, message);
- break;
- case GL_DEBUG_SEVERITY_NOTIFICATION:
- case GL_DEBUG_SEVERITY_LOW:
- LOG_DEBUG(Render_OpenGL, format, str_source, str_type, id, message);
- break;
- }
-}
-
bool RendererOpenGL::Init() {
Core::Frontend::ScopeAcquireWindowContext acquire_context{render_window};
@@ -493,7 +478,6 @@ bool RendererOpenGL::Init() {
return true;
}
-/// Shutdown the renderer
void RendererOpenGL::ShutDown() {}
} // namespace OpenGL
diff --git a/src/video_core/renderer_opengl/renderer_opengl.h b/src/video_core/renderer_opengl/renderer_opengl.h
index cf26628ca..b56328a7f 100644
--- a/src/video_core/renderer_opengl/renderer_opengl.h
+++ b/src/video_core/renderer_opengl/renderer_opengl.h
@@ -59,21 +59,31 @@ public:
void ShutDown() override;
private:
+ /// Initializes the OpenGL state and creates persistent objects.
void InitOpenGLObjects();
+
void AddTelemetryFields();
+
void CreateRasterizer();
void ConfigureFramebufferTexture(TextureInfo& texture,
const Tegra::FramebufferConfig& framebuffer);
+
+ /// Draws the emulated screens to the emulator window.
void DrawScreen(const Layout::FramebufferLayout& layout);
+
void DrawScreenTriangles(const ScreenInfo& screen_info, float x, float y, float w, float h);
+
+ /// Updates the framerate.
void UpdateFramerate();
void CaptureScreenshot();
- // Loads framebuffer from emulated memory into the display information structure
+ /// Loads framebuffer from emulated memory into the active OpenGL texture.
void LoadFBToScreenInfo(const Tegra::FramebufferConfig& framebuffer);
- // Fills active OpenGL texture with the given RGBA color.
+
+ /// Fills active OpenGL texture with the given RGB color.Since the color is solid, the texture
+ /// can be 1x1 but will stretch across whatever it's rendered on.
void LoadColorToActiveGLTexture(u8 color_r, u8 color_g, u8 color_b, u8 color_a,
const TextureInfo& texture);
@@ -94,14 +104,6 @@ private:
/// OpenGL framebuffer data
std::vector<u8> gl_framebuffer_data;
- // Shader uniform location indices
- GLuint uniform_modelview_matrix;
- GLuint uniform_color_texture;
-
- // Shader attribute input indices
- GLuint attrib_position;
- GLuint attrib_tex_coord;
-
/// Used for transforming the framebuffer orientation
Tegra::FramebufferConfig::TransformFlags framebuffer_transform_flags;
Common::Rectangle<int> framebuffer_crop_rect;
diff --git a/src/video_core/renderer_opengl/utils.cpp b/src/video_core/renderer_opengl/utils.cpp
index c504a2c1a..9770dda1c 100644
--- a/src/video_core/renderer_opengl/utils.cpp
+++ b/src/video_core/renderer_opengl/utils.cpp
@@ -3,7 +3,10 @@
// Refer to the license.txt file included.
#include <string>
+#include <vector>
+
#include <fmt/format.h>
+
#include <glad/glad.h>
#include "common/assert.h"
@@ -48,34 +51,19 @@ BindBuffersRangePushBuffer::BindBuffersRangePushBuffer(GLenum target) : target{t
BindBuffersRangePushBuffer::~BindBuffersRangePushBuffer() = default;
-void BindBuffersRangePushBuffer::Setup(GLuint first_) {
- first = first_;
- buffer_pointers.clear();
- offsets.clear();
- sizes.clear();
+void BindBuffersRangePushBuffer::Setup() {
+ entries.clear();
}
-void BindBuffersRangePushBuffer::Push(const GLuint* buffer, GLintptr offset, GLsizeiptr size) {
- buffer_pointers.push_back(buffer);
- offsets.push_back(offset);
- sizes.push_back(size);
+void BindBuffersRangePushBuffer::Push(GLuint binding, const GLuint* buffer, GLintptr offset,
+ GLsizeiptr size) {
+ entries.push_back(Entry{binding, buffer, offset, size});
}
void BindBuffersRangePushBuffer::Bind() {
- // Ensure sizes are valid.
- const std::size_t count{buffer_pointers.size()};
- DEBUG_ASSERT(count == offsets.size() && count == sizes.size());
- if (count == 0) {
- return;
+ for (const Entry& entry : entries) {
+ glBindBufferRange(target, entry.binding, *entry.buffer, entry.offset, entry.size);
}
-
- // Dereference buffers.
- buffers.resize(count);
- std::transform(buffer_pointers.begin(), buffer_pointers.end(), buffers.begin(),
- [](const GLuint* pointer) { return *pointer; });
-
- glBindBuffersRange(target, first, static_cast<GLsizei>(count), buffers.data(), offsets.data(),
- sizes.data());
}
void LabelGLObject(GLenum identifier, GLuint handle, VAddr addr, std::string_view extra_info) {
diff --git a/src/video_core/renderer_opengl/utils.h b/src/video_core/renderer_opengl/utils.h
index 6c2b45546..d56153fe7 100644
--- a/src/video_core/renderer_opengl/utils.h
+++ b/src/video_core/renderer_opengl/utils.h
@@ -43,20 +43,22 @@ public:
explicit BindBuffersRangePushBuffer(GLenum target);
~BindBuffersRangePushBuffer();
- void Setup(GLuint first_);
+ void Setup();
- void Push(const GLuint* buffer, GLintptr offset, GLsizeiptr size);
+ void Push(GLuint binding, const GLuint* buffer, GLintptr offset, GLsizeiptr size);
void Bind();
private:
- GLenum target{};
- GLuint first{};
- std::vector<const GLuint*> buffer_pointers;
+ struct Entry {
+ GLuint binding;
+ const GLuint* buffer;
+ GLintptr offset;
+ GLsizeiptr size;
+ };
- std::vector<GLuint> buffers;
- std::vector<GLintptr> offsets;
- std::vector<GLsizeiptr> sizes;
+ GLenum target;
+ std::vector<Entry> entries;
};
void LabelGLObject(GLenum identifier, GLuint handle, VAddr addr, std::string_view extra_info = {});
diff --git a/src/video_core/renderer_vulkan/declarations.h b/src/video_core/renderer_vulkan/declarations.h
index ba25b5bc7..323bf6b39 100644
--- a/src/video_core/renderer_vulkan/declarations.h
+++ b/src/video_core/renderer_vulkan/declarations.h
@@ -4,6 +4,17 @@
#pragma once
+namespace vk {
+class DispatchLoaderDynamic;
+}
+
+namespace Vulkan {
+constexpr vk::DispatchLoaderDynamic* dont_use_me_dld = nullptr;
+}
+
+#define VULKAN_HPP_DEFAULT_DISPATCHER (*::Vulkan::dont_use_me_dld)
+#define VULKAN_HPP_ENABLE_DYNAMIC_LOADER_TOOL 0
+#define VULKAN_HPP_DISPATCH_LOADER_DYNAMIC 1
#include <vulkan/vulkan.hpp>
namespace Vulkan {
@@ -41,5 +52,7 @@ using UniqueSemaphore = UniqueHandle<vk::Semaphore>;
using UniqueShaderModule = UniqueHandle<vk::ShaderModule>;
using UniqueSwapchainKHR = UniqueHandle<vk::SwapchainKHR>;
using UniqueValidationCacheEXT = UniqueHandle<vk::ValidationCacheEXT>;
+using UniqueDebugReportCallbackEXT = UniqueHandle<vk::DebugReportCallbackEXT>;
+using UniqueDebugUtilsMessengerEXT = UniqueHandle<vk::DebugUtilsMessengerEXT>;
} // namespace Vulkan
diff --git a/src/video_core/renderer_vulkan/fixed_pipeline_state.cpp b/src/video_core/renderer_vulkan/fixed_pipeline_state.cpp
new file mode 100644
index 000000000..5a490f6ef
--- /dev/null
+++ b/src/video_core/renderer_vulkan/fixed_pipeline_state.cpp
@@ -0,0 +1,296 @@
+// Copyright 2019 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <tuple>
+
+#include <boost/functional/hash.hpp>
+
+#include "common/common_types.h"
+#include "video_core/renderer_vulkan/fixed_pipeline_state.h"
+
+namespace Vulkan {
+
+namespace {
+
+constexpr FixedPipelineState::DepthStencil GetDepthStencilState(const Maxwell& regs) {
+ const FixedPipelineState::StencilFace front_stencil(
+ regs.stencil_front_op_fail, regs.stencil_front_op_zfail, regs.stencil_front_op_zpass,
+ regs.stencil_front_func_func);
+ const FixedPipelineState::StencilFace back_stencil =
+ regs.stencil_two_side_enable
+ ? FixedPipelineState::StencilFace(regs.stencil_back_op_fail, regs.stencil_back_op_zfail,
+ regs.stencil_back_op_zpass,
+ regs.stencil_back_func_func)
+ : front_stencil;
+ return FixedPipelineState::DepthStencil(
+ regs.depth_test_enable == 1, regs.depth_write_enabled == 1, regs.depth_bounds_enable == 1,
+ regs.stencil_enable == 1, regs.depth_test_func, front_stencil, back_stencil);
+}
+
+constexpr FixedPipelineState::InputAssembly GetInputAssemblyState(const Maxwell& regs) {
+ return FixedPipelineState::InputAssembly(
+ regs.draw.topology, regs.primitive_restart.enabled,
+ regs.draw.topology == Maxwell::PrimitiveTopology::Points ? regs.point_size : 0.0f);
+}
+
+constexpr FixedPipelineState::BlendingAttachment GetBlendingAttachmentState(
+ const Maxwell& regs, std::size_t render_target) {
+ const auto& mask = regs.color_mask[regs.color_mask_common ? 0 : render_target];
+ const std::array components = {mask.R != 0, mask.G != 0, mask.B != 0, mask.A != 0};
+
+ const FixedPipelineState::BlendingAttachment default_blending(
+ false, Maxwell::Blend::Equation::Add, Maxwell::Blend::Factor::One,
+ Maxwell::Blend::Factor::Zero, Maxwell::Blend::Equation::Add, Maxwell::Blend::Factor::One,
+ Maxwell::Blend::Factor::Zero, components);
+ if (render_target >= regs.rt_control.count) {
+ return default_blending;
+ }
+
+ if (!regs.independent_blend_enable) {
+ const auto& src = regs.blend;
+ if (!src.enable[render_target]) {
+ return default_blending;
+ }
+ return FixedPipelineState::BlendingAttachment(
+ true, src.equation_rgb, src.factor_source_rgb, src.factor_dest_rgb, src.equation_a,
+ src.factor_source_a, src.factor_dest_a, components);
+ }
+
+ if (!regs.blend.enable[render_target]) {
+ return default_blending;
+ }
+ const auto& src = regs.independent_blend[render_target];
+ return FixedPipelineState::BlendingAttachment(
+ true, src.equation_rgb, src.factor_source_rgb, src.factor_dest_rgb, src.equation_a,
+ src.factor_source_a, src.factor_dest_a, components);
+}
+
+constexpr FixedPipelineState::ColorBlending GetColorBlendingState(const Maxwell& regs) {
+ return FixedPipelineState::ColorBlending(
+ {regs.blend_color.r, regs.blend_color.g, regs.blend_color.b, regs.blend_color.a},
+ regs.rt_control.count,
+ {GetBlendingAttachmentState(regs, 0), GetBlendingAttachmentState(regs, 1),
+ GetBlendingAttachmentState(regs, 2), GetBlendingAttachmentState(regs, 3),
+ GetBlendingAttachmentState(regs, 4), GetBlendingAttachmentState(regs, 5),
+ GetBlendingAttachmentState(regs, 6), GetBlendingAttachmentState(regs, 7)});
+}
+
+constexpr FixedPipelineState::Tessellation GetTessellationState(const Maxwell& regs) {
+ return FixedPipelineState::Tessellation(regs.patch_vertices, regs.tess_mode.prim,
+ regs.tess_mode.spacing, regs.tess_mode.cw != 0);
+}
+
+constexpr std::size_t Point = 0;
+constexpr std::size_t Line = 1;
+constexpr std::size_t Polygon = 2;
+constexpr std::array PolygonOffsetEnableLUT = {
+ Point, // Points
+ Line, // Lines
+ Line, // LineLoop
+ Line, // LineStrip
+ Polygon, // Triangles
+ Polygon, // TriangleStrip
+ Polygon, // TriangleFan
+ Polygon, // Quads
+ Polygon, // QuadStrip
+ Polygon, // Polygon
+ Line, // LinesAdjacency
+ Line, // LineStripAdjacency
+ Polygon, // TrianglesAdjacency
+ Polygon, // TriangleStripAdjacency
+ Polygon, // Patches
+};
+
+constexpr FixedPipelineState::Rasterizer GetRasterizerState(const Maxwell& regs) {
+ const std::array enabled_lut = {regs.polygon_offset_point_enable,
+ regs.polygon_offset_line_enable,
+ regs.polygon_offset_fill_enable};
+ const auto topology = static_cast<std::size_t>(regs.draw.topology.Value());
+ const bool depth_bias_enabled = enabled_lut[PolygonOffsetEnableLUT[topology]];
+
+ Maxwell::Cull::FrontFace front_face = regs.cull.front_face;
+ if (regs.screen_y_control.triangle_rast_flip != 0 &&
+ regs.viewport_transform[0].scale_y > 0.0f) {
+ if (front_face == Maxwell::Cull::FrontFace::CounterClockWise)
+ front_face = Maxwell::Cull::FrontFace::ClockWise;
+ else if (front_face == Maxwell::Cull::FrontFace::ClockWise)
+ front_face = Maxwell::Cull::FrontFace::CounterClockWise;
+ }
+
+ const bool gl_ndc = regs.depth_mode == Maxwell::DepthMode::MinusOneToOne;
+ return FixedPipelineState::Rasterizer(regs.cull.enabled, depth_bias_enabled, gl_ndc,
+ regs.cull.cull_face, front_face);
+}
+
+} // Anonymous namespace
+
+std::size_t FixedPipelineState::VertexBinding::Hash() const noexcept {
+ return (index << stride) ^ divisor;
+}
+
+bool FixedPipelineState::VertexBinding::operator==(const VertexBinding& rhs) const noexcept {
+ return std::tie(index, stride, divisor) == std::tie(rhs.index, rhs.stride, rhs.divisor);
+}
+
+std::size_t FixedPipelineState::VertexAttribute::Hash() const noexcept {
+ return static_cast<std::size_t>(index) ^ (static_cast<std::size_t>(buffer) << 13) ^
+ (static_cast<std::size_t>(type) << 22) ^ (static_cast<std::size_t>(size) << 31) ^
+ (static_cast<std::size_t>(offset) << 36);
+}
+
+bool FixedPipelineState::VertexAttribute::operator==(const VertexAttribute& rhs) const noexcept {
+ return std::tie(index, buffer, type, size, offset) ==
+ std::tie(rhs.index, rhs.buffer, rhs.type, rhs.size, rhs.offset);
+}
+
+std::size_t FixedPipelineState::StencilFace::Hash() const noexcept {
+ return static_cast<std::size_t>(action_stencil_fail) ^
+ (static_cast<std::size_t>(action_depth_fail) << 4) ^
+ (static_cast<std::size_t>(action_depth_fail) << 20) ^
+ (static_cast<std::size_t>(action_depth_pass) << 36);
+}
+
+bool FixedPipelineState::StencilFace::operator==(const StencilFace& rhs) const noexcept {
+ return std::tie(action_stencil_fail, action_depth_fail, action_depth_pass, test_func) ==
+ std::tie(rhs.action_stencil_fail, rhs.action_depth_fail, rhs.action_depth_pass,
+ rhs.test_func);
+}
+
+std::size_t FixedPipelineState::BlendingAttachment::Hash() const noexcept {
+ return static_cast<std::size_t>(enable) ^ (static_cast<std::size_t>(rgb_equation) << 5) ^
+ (static_cast<std::size_t>(src_rgb_func) << 10) ^
+ (static_cast<std::size_t>(dst_rgb_func) << 15) ^
+ (static_cast<std::size_t>(a_equation) << 20) ^
+ (static_cast<std::size_t>(src_a_func) << 25) ^
+ (static_cast<std::size_t>(dst_a_func) << 30) ^
+ (static_cast<std::size_t>(components[0]) << 35) ^
+ (static_cast<std::size_t>(components[1]) << 36) ^
+ (static_cast<std::size_t>(components[2]) << 37) ^
+ (static_cast<std::size_t>(components[3]) << 38);
+}
+
+bool FixedPipelineState::BlendingAttachment::operator==(const BlendingAttachment& rhs) const
+ noexcept {
+ return std::tie(enable, rgb_equation, src_rgb_func, dst_rgb_func, a_equation, src_a_func,
+ dst_a_func, components) ==
+ std::tie(rhs.enable, rhs.rgb_equation, rhs.src_rgb_func, rhs.dst_rgb_func,
+ rhs.a_equation, rhs.src_a_func, rhs.dst_a_func, rhs.components);
+}
+
+std::size_t FixedPipelineState::VertexInput::Hash() const noexcept {
+ std::size_t hash = num_bindings ^ (num_attributes << 32);
+ for (std::size_t i = 0; i < num_bindings; ++i) {
+ boost::hash_combine(hash, bindings[i].Hash());
+ }
+ for (std::size_t i = 0; i < num_attributes; ++i) {
+ boost::hash_combine(hash, attributes[i].Hash());
+ }
+ return hash;
+}
+
+bool FixedPipelineState::VertexInput::operator==(const VertexInput& rhs) const noexcept {
+ return std::equal(bindings.begin(), bindings.begin() + num_bindings, rhs.bindings.begin(),
+ rhs.bindings.begin() + rhs.num_bindings) &&
+ std::equal(attributes.begin(), attributes.begin() + num_attributes,
+ rhs.attributes.begin(), rhs.attributes.begin() + rhs.num_attributes);
+}
+
+std::size_t FixedPipelineState::InputAssembly::Hash() const noexcept {
+ std::size_t point_size_int = 0;
+ std::memcpy(&point_size_int, &point_size, sizeof(point_size));
+ return (static_cast<std::size_t>(topology) << 24) ^ (point_size_int << 32) ^
+ static_cast<std::size_t>(primitive_restart_enable);
+}
+
+bool FixedPipelineState::InputAssembly::operator==(const InputAssembly& rhs) const noexcept {
+ return std::tie(topology, primitive_restart_enable, point_size) ==
+ std::tie(rhs.topology, rhs.primitive_restart_enable, rhs.point_size);
+}
+
+std::size_t FixedPipelineState::Tessellation::Hash() const noexcept {
+ return static_cast<std::size_t>(patch_control_points) ^
+ (static_cast<std::size_t>(primitive) << 6) ^ (static_cast<std::size_t>(spacing) << 8) ^
+ (static_cast<std::size_t>(clockwise) << 10);
+}
+
+bool FixedPipelineState::Tessellation::operator==(const Tessellation& rhs) const noexcept {
+ return std::tie(patch_control_points, primitive, spacing, clockwise) ==
+ std::tie(rhs.patch_control_points, rhs.primitive, rhs.spacing, rhs.clockwise);
+}
+
+std::size_t FixedPipelineState::Rasterizer::Hash() const noexcept {
+ return static_cast<std::size_t>(cull_enable) ^
+ (static_cast<std::size_t>(depth_bias_enable) << 1) ^
+ (static_cast<std::size_t>(ndc_minus_one_to_one) << 2) ^
+ (static_cast<std::size_t>(cull_face) << 24) ^
+ (static_cast<std::size_t>(front_face) << 48);
+}
+
+bool FixedPipelineState::Rasterizer::operator==(const Rasterizer& rhs) const noexcept {
+ return std::tie(cull_enable, depth_bias_enable, ndc_minus_one_to_one, cull_face, front_face) ==
+ std::tie(rhs.cull_enable, rhs.depth_bias_enable, rhs.ndc_minus_one_to_one, rhs.cull_face,
+ rhs.front_face);
+}
+
+std::size_t FixedPipelineState::DepthStencil::Hash() const noexcept {
+ std::size_t hash = static_cast<std::size_t>(depth_test_enable) ^
+ (static_cast<std::size_t>(depth_write_enable) << 1) ^
+ (static_cast<std::size_t>(depth_bounds_enable) << 2) ^
+ (static_cast<std::size_t>(stencil_enable) << 3) ^
+ (static_cast<std::size_t>(depth_test_function) << 4);
+ boost::hash_combine(hash, front_stencil.Hash());
+ boost::hash_combine(hash, back_stencil.Hash());
+ return hash;
+}
+
+bool FixedPipelineState::DepthStencil::operator==(const DepthStencil& rhs) const noexcept {
+ return std::tie(depth_test_enable, depth_write_enable, depth_bounds_enable, depth_test_function,
+ stencil_enable, front_stencil, back_stencil) ==
+ std::tie(rhs.depth_test_enable, rhs.depth_write_enable, rhs.depth_bounds_enable,
+ rhs.depth_test_function, rhs.stencil_enable, rhs.front_stencil,
+ rhs.back_stencil);
+}
+
+std::size_t FixedPipelineState::ColorBlending::Hash() const noexcept {
+ std::size_t hash = attachments_count << 13;
+ for (std::size_t rt = 0; rt < static_cast<std::size_t>(attachments_count); ++rt) {
+ boost::hash_combine(hash, attachments[rt].Hash());
+ }
+ return hash;
+}
+
+bool FixedPipelineState::ColorBlending::operator==(const ColorBlending& rhs) const noexcept {
+ return std::equal(attachments.begin(), attachments.begin() + attachments_count,
+ rhs.attachments.begin(), rhs.attachments.begin() + rhs.attachments_count);
+}
+
+std::size_t FixedPipelineState::Hash() const noexcept {
+ std::size_t hash = 0;
+ boost::hash_combine(hash, vertex_input.Hash());
+ boost::hash_combine(hash, input_assembly.Hash());
+ boost::hash_combine(hash, tessellation.Hash());
+ boost::hash_combine(hash, rasterizer.Hash());
+ boost::hash_combine(hash, depth_stencil.Hash());
+ boost::hash_combine(hash, color_blending.Hash());
+ return hash;
+}
+
+bool FixedPipelineState::operator==(const FixedPipelineState& rhs) const noexcept {
+ return std::tie(vertex_input, input_assembly, tessellation, rasterizer, depth_stencil,
+ color_blending) == std::tie(rhs.vertex_input, rhs.input_assembly,
+ rhs.tessellation, rhs.rasterizer, rhs.depth_stencil,
+ rhs.color_blending);
+}
+
+FixedPipelineState GetFixedPipelineState(const Maxwell& regs) {
+ FixedPipelineState fixed_state;
+ fixed_state.input_assembly = GetInputAssemblyState(regs);
+ fixed_state.tessellation = GetTessellationState(regs);
+ fixed_state.rasterizer = GetRasterizerState(regs);
+ fixed_state.depth_stencil = GetDepthStencilState(regs);
+ fixed_state.color_blending = GetColorBlendingState(regs);
+ return fixed_state;
+}
+
+} // namespace Vulkan
diff --git a/src/video_core/renderer_vulkan/fixed_pipeline_state.h b/src/video_core/renderer_vulkan/fixed_pipeline_state.h
new file mode 100644
index 000000000..04152c0d4
--- /dev/null
+++ b/src/video_core/renderer_vulkan/fixed_pipeline_state.h
@@ -0,0 +1,282 @@
+// Copyright 2019 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <array>
+#include <type_traits>
+
+#include "common/common_types.h"
+
+#include "video_core/engines/maxwell_3d.h"
+#include "video_core/surface.h"
+
+namespace Vulkan {
+
+using Maxwell = Tegra::Engines::Maxwell3D::Regs;
+
+// TODO(Rodrigo): Optimize this structure.
+
+struct FixedPipelineState {
+ using PixelFormat = VideoCore::Surface::PixelFormat;
+
+ struct VertexBinding {
+ constexpr VertexBinding(u32 index, u32 stride, u32 divisor)
+ : index{index}, stride{stride}, divisor{divisor} {}
+ VertexBinding() = default;
+
+ u32 index;
+ u32 stride;
+ u32 divisor;
+
+ std::size_t Hash() const noexcept;
+
+ bool operator==(const VertexBinding& rhs) const noexcept;
+
+ bool operator!=(const VertexBinding& rhs) const noexcept {
+ return !operator==(rhs);
+ }
+ };
+
+ struct VertexAttribute {
+ constexpr VertexAttribute(u32 index, u32 buffer, Maxwell::VertexAttribute::Type type,
+ Maxwell::VertexAttribute::Size size, u32 offset)
+ : index{index}, buffer{buffer}, type{type}, size{size}, offset{offset} {}
+ VertexAttribute() = default;
+
+ u32 index;
+ u32 buffer;
+ Maxwell::VertexAttribute::Type type;
+ Maxwell::VertexAttribute::Size size;
+ u32 offset;
+
+ std::size_t Hash() const noexcept;
+
+ bool operator==(const VertexAttribute& rhs) const noexcept;
+
+ bool operator!=(const VertexAttribute& rhs) const noexcept {
+ return !operator==(rhs);
+ }
+ };
+
+ struct StencilFace {
+ constexpr StencilFace(Maxwell::StencilOp action_stencil_fail,
+ Maxwell::StencilOp action_depth_fail,
+ Maxwell::StencilOp action_depth_pass, Maxwell::ComparisonOp test_func)
+ : action_stencil_fail{action_stencil_fail}, action_depth_fail{action_depth_fail},
+ action_depth_pass{action_depth_pass}, test_func{test_func} {}
+ StencilFace() = default;
+
+ Maxwell::StencilOp action_stencil_fail;
+ Maxwell::StencilOp action_depth_fail;
+ Maxwell::StencilOp action_depth_pass;
+ Maxwell::ComparisonOp test_func;
+
+ std::size_t Hash() const noexcept;
+
+ bool operator==(const StencilFace& rhs) const noexcept;
+
+ bool operator!=(const StencilFace& rhs) const noexcept {
+ return !operator==(rhs);
+ }
+ };
+
+ struct BlendingAttachment {
+ constexpr BlendingAttachment(bool enable, Maxwell::Blend::Equation rgb_equation,
+ Maxwell::Blend::Factor src_rgb_func,
+ Maxwell::Blend::Factor dst_rgb_func,
+ Maxwell::Blend::Equation a_equation,
+ Maxwell::Blend::Factor src_a_func,
+ Maxwell::Blend::Factor dst_a_func,
+ std::array<bool, 4> components)
+ : enable{enable}, rgb_equation{rgb_equation}, src_rgb_func{src_rgb_func},
+ dst_rgb_func{dst_rgb_func}, a_equation{a_equation}, src_a_func{src_a_func},
+ dst_a_func{dst_a_func}, components{components} {}
+ BlendingAttachment() = default;
+
+ bool enable;
+ Maxwell::Blend::Equation rgb_equation;
+ Maxwell::Blend::Factor src_rgb_func;
+ Maxwell::Blend::Factor dst_rgb_func;
+ Maxwell::Blend::Equation a_equation;
+ Maxwell::Blend::Factor src_a_func;
+ Maxwell::Blend::Factor dst_a_func;
+ std::array<bool, 4> components;
+
+ std::size_t Hash() const noexcept;
+
+ bool operator==(const BlendingAttachment& rhs) const noexcept;
+
+ bool operator!=(const BlendingAttachment& rhs) const noexcept {
+ return !operator==(rhs);
+ }
+ };
+
+ struct VertexInput {
+ std::size_t num_bindings = 0;
+ std::size_t num_attributes = 0;
+ std::array<VertexBinding, Maxwell::NumVertexArrays> bindings;
+ std::array<VertexAttribute, Maxwell::NumVertexAttributes> attributes;
+
+ std::size_t Hash() const noexcept;
+
+ bool operator==(const VertexInput& rhs) const noexcept;
+
+ bool operator!=(const VertexInput& rhs) const noexcept {
+ return !operator==(rhs);
+ }
+ };
+
+ struct InputAssembly {
+ constexpr InputAssembly(Maxwell::PrimitiveTopology topology, bool primitive_restart_enable,
+ float point_size)
+ : topology{topology}, primitive_restart_enable{primitive_restart_enable},
+ point_size{point_size} {}
+ InputAssembly() = default;
+
+ Maxwell::PrimitiveTopology topology;
+ bool primitive_restart_enable;
+ float point_size;
+
+ std::size_t Hash() const noexcept;
+
+ bool operator==(const InputAssembly& rhs) const noexcept;
+
+ bool operator!=(const InputAssembly& rhs) const noexcept {
+ return !operator==(rhs);
+ }
+ };
+
+ struct Tessellation {
+ constexpr Tessellation(u32 patch_control_points, Maxwell::TessellationPrimitive primitive,
+ Maxwell::TessellationSpacing spacing, bool clockwise)
+ : patch_control_points{patch_control_points}, primitive{primitive}, spacing{spacing},
+ clockwise{clockwise} {}
+ Tessellation() = default;
+
+ u32 patch_control_points;
+ Maxwell::TessellationPrimitive primitive;
+ Maxwell::TessellationSpacing spacing;
+ bool clockwise;
+
+ std::size_t Hash() const noexcept;
+
+ bool operator==(const Tessellation& rhs) const noexcept;
+
+ bool operator!=(const Tessellation& rhs) const noexcept {
+ return !operator==(rhs);
+ }
+ };
+
+ struct Rasterizer {
+ constexpr Rasterizer(bool cull_enable, bool depth_bias_enable, bool ndc_minus_one_to_one,
+ Maxwell::Cull::CullFace cull_face, Maxwell::Cull::FrontFace front_face)
+ : cull_enable{cull_enable}, depth_bias_enable{depth_bias_enable},
+ ndc_minus_one_to_one{ndc_minus_one_to_one}, cull_face{cull_face}, front_face{
+ front_face} {}
+ Rasterizer() = default;
+
+ bool cull_enable;
+ bool depth_bias_enable;
+ bool ndc_minus_one_to_one;
+ Maxwell::Cull::CullFace cull_face;
+ Maxwell::Cull::FrontFace front_face;
+
+ std::size_t Hash() const noexcept;
+
+ bool operator==(const Rasterizer& rhs) const noexcept;
+
+ bool operator!=(const Rasterizer& rhs) const noexcept {
+ return !operator==(rhs);
+ }
+ };
+
+ struct DepthStencil {
+ constexpr DepthStencil(bool depth_test_enable, bool depth_write_enable,
+ bool depth_bounds_enable, bool stencil_enable,
+ Maxwell::ComparisonOp depth_test_function, StencilFace front_stencil,
+ StencilFace back_stencil)
+ : depth_test_enable{depth_test_enable}, depth_write_enable{depth_write_enable},
+ depth_bounds_enable{depth_bounds_enable}, stencil_enable{stencil_enable},
+ depth_test_function{depth_test_function}, front_stencil{front_stencil},
+ back_stencil{back_stencil} {}
+ DepthStencil() = default;
+
+ bool depth_test_enable;
+ bool depth_write_enable;
+ bool depth_bounds_enable;
+ bool stencil_enable;
+ Maxwell::ComparisonOp depth_test_function;
+ StencilFace front_stencil;
+ StencilFace back_stencil;
+
+ std::size_t Hash() const noexcept;
+
+ bool operator==(const DepthStencil& rhs) const noexcept;
+
+ bool operator!=(const DepthStencil& rhs) const noexcept {
+ return !operator==(rhs);
+ }
+ };
+
+ struct ColorBlending {
+ constexpr ColorBlending(
+ std::array<float, 4> blend_constants, std::size_t attachments_count,
+ std::array<BlendingAttachment, Maxwell::NumRenderTargets> attachments)
+ : attachments_count{attachments_count}, attachments{attachments} {}
+ ColorBlending() = default;
+
+ std::size_t attachments_count;
+ std::array<BlendingAttachment, Maxwell::NumRenderTargets> attachments;
+
+ std::size_t Hash() const noexcept;
+
+ bool operator==(const ColorBlending& rhs) const noexcept;
+
+ bool operator!=(const ColorBlending& rhs) const noexcept {
+ return !operator==(rhs);
+ }
+ };
+
+ std::size_t Hash() const noexcept;
+
+ bool operator==(const FixedPipelineState& rhs) const noexcept;
+
+ bool operator!=(const FixedPipelineState& rhs) const noexcept {
+ return !operator==(rhs);
+ }
+
+ VertexInput vertex_input;
+ InputAssembly input_assembly;
+ Tessellation tessellation;
+ Rasterizer rasterizer;
+ DepthStencil depth_stencil;
+ ColorBlending color_blending;
+};
+static_assert(std::is_trivially_copyable_v<FixedPipelineState::VertexBinding>);
+static_assert(std::is_trivially_copyable_v<FixedPipelineState::VertexAttribute>);
+static_assert(std::is_trivially_copyable_v<FixedPipelineState::StencilFace>);
+static_assert(std::is_trivially_copyable_v<FixedPipelineState::BlendingAttachment>);
+static_assert(std::is_trivially_copyable_v<FixedPipelineState::VertexInput>);
+static_assert(std::is_trivially_copyable_v<FixedPipelineState::InputAssembly>);
+static_assert(std::is_trivially_copyable_v<FixedPipelineState::Tessellation>);
+static_assert(std::is_trivially_copyable_v<FixedPipelineState::Rasterizer>);
+static_assert(std::is_trivially_copyable_v<FixedPipelineState::DepthStencil>);
+static_assert(std::is_trivially_copyable_v<FixedPipelineState::ColorBlending>);
+static_assert(std::is_trivially_copyable_v<FixedPipelineState>);
+
+FixedPipelineState GetFixedPipelineState(const Maxwell& regs);
+
+} // namespace Vulkan
+
+namespace std {
+
+template <>
+struct hash<Vulkan::FixedPipelineState> {
+ std::size_t operator()(const Vulkan::FixedPipelineState& k) const noexcept {
+ return k.Hash();
+ }
+};
+
+} // namespace std
diff --git a/src/video_core/renderer_vulkan/maxwell_to_vk.cpp b/src/video_core/renderer_vulkan/maxwell_to_vk.cpp
index 3c5acda3e..000e3616d 100644
--- a/src/video_core/renderer_vulkan/maxwell_to_vk.cpp
+++ b/src/video_core/renderer_vulkan/maxwell_to_vk.cpp
@@ -13,6 +13,8 @@
namespace Vulkan::MaxwellToVK {
+using Maxwell = Tegra::Engines::Maxwell3D::Regs;
+
namespace Sampler {
vk::Filter Filter(Tegra::Texture::TextureFilter filter) {
@@ -42,7 +44,8 @@ vk::SamplerMipmapMode MipmapMode(Tegra::Texture::TextureMipmapFilter mipmap_filt
return {};
}
-vk::SamplerAddressMode WrapMode(Tegra::Texture::WrapMode wrap_mode) {
+vk::SamplerAddressMode WrapMode(Tegra::Texture::WrapMode wrap_mode,
+ Tegra::Texture::TextureFilter filter) {
switch (wrap_mode) {
case Tegra::Texture::WrapMode::Wrap:
return vk::SamplerAddressMode::eRepeat;
@@ -53,10 +56,15 @@ vk::SamplerAddressMode WrapMode(Tegra::Texture::WrapMode wrap_mode) {
case Tegra::Texture::WrapMode::Border:
return vk::SamplerAddressMode::eClampToBorder;
case Tegra::Texture::WrapMode::Clamp:
- // TODO(Rodrigo): GL_CLAMP was removed as of OpenGL 3.1, to implement GL_CLAMP, we can use
- // eClampToBorder to get the border color of the texture, and then sample the edge to
- // manually mix them. However the shader part of this is not yet implemented.
- return vk::SamplerAddressMode::eClampToBorder;
+ // TODO(Rodrigo): Emulate GL_CLAMP properly
+ switch (filter) {
+ case Tegra::Texture::TextureFilter::Nearest:
+ return vk::SamplerAddressMode::eClampToEdge;
+ case Tegra::Texture::TextureFilter::Linear:
+ return vk::SamplerAddressMode::eClampToBorder;
+ }
+ UNREACHABLE();
+ return vk::SamplerAddressMode::eClampToEdge;
case Tegra::Texture::WrapMode::MirrorOnceClampToEdge:
return vk::SamplerAddressMode::eMirrorClampToEdge;
case Tegra::Texture::WrapMode::MirrorOnceBorder:
@@ -94,128 +102,161 @@ vk::CompareOp DepthCompareFunction(Tegra::Texture::DepthCompareFunc depth_compar
} // namespace Sampler
-struct FormatTuple {
- vk::Format format; ///< Vulkan format
- ComponentType component_type; ///< Abstracted component type
- bool attachable; ///< True when this format can be used as an attachment
-};
+namespace {
-static constexpr std::array<FormatTuple, VideoCore::Surface::MaxPixelFormat> tex_format_tuples = {{
- {vk::Format::eA8B8G8R8UnormPack32, ComponentType::UNorm, true}, // ABGR8U
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // ABGR8S
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // ABGR8UI
- {vk::Format::eB5G6R5UnormPack16, ComponentType::UNorm, false}, // B5G6R5U
- {vk::Format::eA2B10G10R10UnormPack32, ComponentType::UNorm, true}, // A2B10G10R10U
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // A1B5G5R5U
- {vk::Format::eR8Unorm, ComponentType::UNorm, true}, // R8U
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // R8UI
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // RGBA16F
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // RGBA16U
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // RGBA16UI
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // R11FG11FB10F
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // RGBA32UI
- {vk::Format::eBc1RgbaUnormBlock, ComponentType::UNorm, false}, // DXT1
- {vk::Format::eBc2UnormBlock, ComponentType::UNorm, false}, // DXT23
- {vk::Format::eBc3UnormBlock, ComponentType::UNorm, false}, // DXT45
- {vk::Format::eBc4UnormBlock, ComponentType::UNorm, false}, // DXN1
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // DXN2UNORM
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // DXN2SNORM
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // BC7U
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // BC6H_UF16
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // BC6H_SF16
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // ASTC_2D_4X4
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // BGRA8
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // RGBA32F
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // RG32F
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // R32F
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // R16F
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // R16U
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // R16S
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // R16UI
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // R16I
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // RG16
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // RG16F
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // RG16UI
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // RG16I
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // RG16S
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // RGB32F
- {vk::Format::eA8B8G8R8SrgbPack32, ComponentType::UNorm, true}, // RGBA8_SRGB
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // RG8U
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // RG8S
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // RG32UI
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // RGBX16F
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // R32UI
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // ASTC_2D_8X8
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // ASTC_2D_8X5
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // ASTC_2D_5X4
-
- // Compressed sRGB formats
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // BGRA8_SRGB
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // DXT1_SRGB
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // DXT23_SRGB
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // DXT45_SRGB
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // BC7U_SRGB
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // ASTC_2D_4X4_SRGB
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // ASTC_2D_8X8_SRGB
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // ASTC_2D_8X5_SRGB
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // ASTC_2D_5X4_SRGB
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // ASTC_2D_5X5
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // ASTC_2D_5X5_SRGB
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // ASTC_2D_10X8
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // ASTC_2D_10X8_SRGB
+enum : u32 { Attachable = 1, Storage = 2 };
+
+struct FormatTuple {
+ vk::Format format; ///< Vulkan format
+ int usage; ///< Describes image format usage
+} constexpr tex_format_tuples[] = {
+ {vk::Format::eA8B8G8R8UnormPack32, Attachable | Storage}, // ABGR8U
+ {vk::Format::eA8B8G8R8SnormPack32, Attachable | Storage}, // ABGR8S
+ {vk::Format::eA8B8G8R8UintPack32, Attachable | Storage}, // ABGR8UI
+ {vk::Format::eB5G6R5UnormPack16, {}}, // B5G6R5U
+ {vk::Format::eA2B10G10R10UnormPack32, Attachable | Storage}, // A2B10G10R10U
+ {vk::Format::eA1R5G5B5UnormPack16, Attachable | Storage}, // A1B5G5R5U (flipped with swizzle)
+ {vk::Format::eR8Unorm, Attachable | Storage}, // R8U
+ {vk::Format::eR8Uint, Attachable | Storage}, // R8UI
+ {vk::Format::eR16G16B16A16Sfloat, Attachable | Storage}, // RGBA16F
+ {vk::Format::eR16G16B16A16Unorm, Attachable | Storage}, // RGBA16U
+ {vk::Format::eR16G16B16A16Uint, Attachable | Storage}, // RGBA16UI
+ {vk::Format::eB10G11R11UfloatPack32, Attachable | Storage}, // R11FG11FB10F
+ {vk::Format::eR32G32B32A32Uint, Attachable | Storage}, // RGBA32UI
+ {vk::Format::eBc1RgbaUnormBlock, {}}, // DXT1
+ {vk::Format::eBc2UnormBlock, {}}, // DXT23
+ {vk::Format::eBc3UnormBlock, {}}, // DXT45
+ {vk::Format::eBc4UnormBlock, {}}, // DXN1
+ {vk::Format::eBc5UnormBlock, {}}, // DXN2UNORM
+ {vk::Format::eBc5SnormBlock, {}}, // DXN2SNORM
+ {vk::Format::eBc7UnormBlock, {}}, // BC7U
+ {vk::Format::eBc6HUfloatBlock, {}}, // BC6H_UF16
+ {vk::Format::eBc6HSfloatBlock, {}}, // BC6H_SF16
+ {vk::Format::eAstc4x4UnormBlock, {}}, // ASTC_2D_4X4
+ {vk::Format::eB8G8R8A8Unorm, {}}, // BGRA8
+ {vk::Format::eR32G32B32A32Sfloat, Attachable | Storage}, // RGBA32F
+ {vk::Format::eR32G32Sfloat, Attachable | Storage}, // RG32F
+ {vk::Format::eR32Sfloat, Attachable | Storage}, // R32F
+ {vk::Format::eR16Sfloat, Attachable | Storage}, // R16F
+ {vk::Format::eR16Unorm, Attachable | Storage}, // R16U
+ {vk::Format::eUndefined, {}}, // R16S
+ {vk::Format::eUndefined, {}}, // R16UI
+ {vk::Format::eUndefined, {}}, // R16I
+ {vk::Format::eR16G16Unorm, Attachable | Storage}, // RG16
+ {vk::Format::eR16G16Sfloat, Attachable | Storage}, // RG16F
+ {vk::Format::eUndefined, {}}, // RG16UI
+ {vk::Format::eUndefined, {}}, // RG16I
+ {vk::Format::eR16G16Snorm, Attachable | Storage}, // RG16S
+ {vk::Format::eUndefined, {}}, // RGB32F
+ {vk::Format::eR8G8B8A8Srgb, Attachable}, // RGBA8_SRGB
+ {vk::Format::eR8G8Unorm, Attachable | Storage}, // RG8U
+ {vk::Format::eR8G8Snorm, Attachable | Storage}, // RG8S
+ {vk::Format::eR32G32Uint, Attachable | Storage}, // RG32UI
+ {vk::Format::eUndefined, {}}, // RGBX16F
+ {vk::Format::eR32Uint, Attachable | Storage}, // R32UI
+ {vk::Format::eAstc8x8UnormBlock, {}}, // ASTC_2D_8X8
+ {vk::Format::eUndefined, {}}, // ASTC_2D_8X5
+ {vk::Format::eUndefined, {}}, // ASTC_2D_5X4
+ {vk::Format::eUndefined, {}}, // BGRA8_SRGB
+ {vk::Format::eBc1RgbaSrgbBlock, {}}, // DXT1_SRGB
+ {vk::Format::eUndefined, {}}, // DXT23_SRGB
+ {vk::Format::eBc3SrgbBlock, {}}, // DXT45_SRGB
+ {vk::Format::eBc7SrgbBlock, {}}, // BC7U_SRGB
+ {vk::Format::eR4G4B4A4UnormPack16, Attachable}, // R4G4B4A4U
+ {vk::Format::eAstc4x4SrgbBlock, {}}, // ASTC_2D_4X4_SRGB
+ {vk::Format::eAstc8x8SrgbBlock, {}}, // ASTC_2D_8X8_SRGB
+ {vk::Format::eAstc8x5SrgbBlock, {}}, // ASTC_2D_8X5_SRGB
+ {vk::Format::eAstc5x4SrgbBlock, {}}, // ASTC_2D_5X4_SRGB
+ {vk::Format::eAstc5x5UnormBlock, {}}, // ASTC_2D_5X5
+ {vk::Format::eAstc5x5SrgbBlock, {}}, // ASTC_2D_5X5_SRGB
+ {vk::Format::eAstc10x8UnormBlock, {}}, // ASTC_2D_10X8
+ {vk::Format::eAstc10x8SrgbBlock, {}}, // ASTC_2D_10X8_SRGB
+ {vk::Format::eAstc6x6UnormBlock, {}}, // ASTC_2D_6X6
+ {vk::Format::eAstc6x6SrgbBlock, {}}, // ASTC_2D_6X6_SRGB
+ {vk::Format::eAstc10x10UnormBlock, {}}, // ASTC_2D_10X10
+ {vk::Format::eAstc10x10SrgbBlock, {}}, // ASTC_2D_10X10_SRGB
+ {vk::Format::eAstc12x12UnormBlock, {}}, // ASTC_2D_12X12
+ {vk::Format::eAstc12x12SrgbBlock, {}}, // ASTC_2D_12X12_SRGB
+ {vk::Format::eAstc8x6UnormBlock, {}}, // ASTC_2D_8X6
+ {vk::Format::eAstc8x6SrgbBlock, {}}, // ASTC_2D_8X6_SRGB
+ {vk::Format::eAstc6x5UnormBlock, {}}, // ASTC_2D_6X5
+ {vk::Format::eAstc6x5SrgbBlock, {}}, // ASTC_2D_6X5_SRGB
+ {vk::Format::eE5B9G9R9UfloatPack32, {}}, // E5B9G9R9F
// Depth formats
- {vk::Format::eD32Sfloat, ComponentType::Float, true}, // Z32F
- {vk::Format::eD16Unorm, ComponentType::UNorm, true}, // Z16
+ {vk::Format::eD32Sfloat, Attachable}, // Z32F
+ {vk::Format::eD16Unorm, Attachable}, // Z16
// DepthStencil formats
- {vk::Format::eD24UnormS8Uint, ComponentType::UNorm, true}, // Z24S8
- {vk::Format::eD24UnormS8Uint, ComponentType::UNorm, true}, // S8Z24 (emulated)
- {vk::Format::eUndefined, ComponentType::Invalid, false}, // Z32FS8
-}};
+ {vk::Format::eD24UnormS8Uint, Attachable}, // Z24S8
+ {vk::Format::eD24UnormS8Uint, Attachable}, // S8Z24 (emulated)
+ {vk::Format::eD32SfloatS8Uint, Attachable}, // Z32FS8
+};
+static_assert(std::size(tex_format_tuples) == VideoCore::Surface::MaxPixelFormat);
-static constexpr bool IsZetaFormat(PixelFormat pixel_format) {
+constexpr bool IsZetaFormat(PixelFormat pixel_format) {
return pixel_format >= PixelFormat::MaxColorFormat &&
pixel_format < PixelFormat::MaxDepthStencilFormat;
}
-std::pair<vk::Format, bool> SurfaceFormat(const VKDevice& device, FormatType format_type,
- PixelFormat pixel_format, ComponentType component_type) {
- ASSERT(static_cast<std::size_t>(pixel_format) < tex_format_tuples.size());
-
- const auto tuple = tex_format_tuples[static_cast<u32>(pixel_format)];
- UNIMPLEMENTED_IF_MSG(tuple.format == vk::Format::eUndefined,
- "Unimplemented texture format with pixel format={} and component type={}",
- static_cast<u32>(pixel_format), static_cast<u32>(component_type));
- ASSERT_MSG(component_type == tuple.component_type, "Component type mismatch");
-
- auto usage = vk::FormatFeatureFlagBits::eSampledImage |
- vk::FormatFeatureFlagBits::eTransferDst | vk::FormatFeatureFlagBits::eTransferSrc;
- if (tuple.attachable) {
- usage |= IsZetaFormat(pixel_format) ? vk::FormatFeatureFlagBits::eDepthStencilAttachment
- : vk::FormatFeatureFlagBits::eColorAttachment;
+} // Anonymous namespace
+
+FormatInfo SurfaceFormat(const VKDevice& device, FormatType format_type, PixelFormat pixel_format) {
+ ASSERT(static_cast<std::size_t>(pixel_format) < std::size(tex_format_tuples));
+
+ auto tuple = tex_format_tuples[static_cast<std::size_t>(pixel_format)];
+ if (tuple.format == vk::Format::eUndefined) {
+ UNIMPLEMENTED_MSG("Unimplemented texture format with pixel format={}",
+ static_cast<u32>(pixel_format));
+ return {vk::Format::eA8B8G8R8UnormPack32, true, true};
+ }
+
+ // Use ABGR8 on hardware that doesn't support ASTC natively
+ if (!device.IsOptimalAstcSupported() && VideoCore::Surface::IsPixelFormatASTC(pixel_format)) {
+ tuple.format = VideoCore::Surface::IsPixelFormatSRGB(pixel_format)
+ ? vk::Format::eA8B8G8R8SrgbPack32
+ : vk::Format::eA8B8G8R8UnormPack32;
}
- return {device.GetSupportedFormat(tuple.format, usage, format_type), tuple.attachable};
+ const bool attachable = tuple.usage & Attachable;
+ const bool storage = tuple.usage & Storage;
+
+ vk::FormatFeatureFlags usage;
+ if (format_type == FormatType::Buffer) {
+ usage = vk::FormatFeatureFlagBits::eStorageTexelBuffer |
+ vk::FormatFeatureFlagBits::eUniformTexelBuffer;
+ } else {
+ usage = vk::FormatFeatureFlagBits::eSampledImage | vk::FormatFeatureFlagBits::eTransferDst |
+ vk::FormatFeatureFlagBits::eTransferSrc;
+ if (attachable) {
+ usage |= IsZetaFormat(pixel_format) ? vk::FormatFeatureFlagBits::eDepthStencilAttachment
+ : vk::FormatFeatureFlagBits::eColorAttachment;
+ }
+ if (storage) {
+ usage |= vk::FormatFeatureFlagBits::eStorageImage;
+ }
+ }
+ return {device.GetSupportedFormat(tuple.format, usage, format_type), attachable, storage};
}
-vk::ShaderStageFlagBits ShaderStage(Maxwell::ShaderStage stage) {
+vk::ShaderStageFlagBits ShaderStage(Tegra::Engines::ShaderType stage) {
switch (stage) {
- case Maxwell::ShaderStage::Vertex:
+ case Tegra::Engines::ShaderType::Vertex:
return vk::ShaderStageFlagBits::eVertex;
- case Maxwell::ShaderStage::TesselationControl:
+ case Tegra::Engines::ShaderType::TesselationControl:
return vk::ShaderStageFlagBits::eTessellationControl;
- case Maxwell::ShaderStage::TesselationEval:
+ case Tegra::Engines::ShaderType::TesselationEval:
return vk::ShaderStageFlagBits::eTessellationEvaluation;
- case Maxwell::ShaderStage::Geometry:
+ case Tegra::Engines::ShaderType::Geometry:
return vk::ShaderStageFlagBits::eGeometry;
- case Maxwell::ShaderStage::Fragment:
+ case Tegra::Engines::ShaderType::Fragment:
return vk::ShaderStageFlagBits::eFragment;
}
UNIMPLEMENTED_MSG("Unimplemented shader stage={}", static_cast<u32>(stage));
return {};
}
-vk::PrimitiveTopology PrimitiveTopology(Maxwell::PrimitiveTopology topology) {
+vk::PrimitiveTopology PrimitiveTopology([[maybe_unused]] const VKDevice& device,
+ Maxwell::PrimitiveTopology topology) {
switch (topology) {
case Maxwell::PrimitiveTopology::Points:
return vk::PrimitiveTopology::ePointList;
@@ -227,6 +268,13 @@ vk::PrimitiveTopology PrimitiveTopology(Maxwell::PrimitiveTopology topology) {
return vk::PrimitiveTopology::eTriangleList;
case Maxwell::PrimitiveTopology::TriangleStrip:
return vk::PrimitiveTopology::eTriangleStrip;
+ case Maxwell::PrimitiveTopology::TriangleFan:
+ return vk::PrimitiveTopology::eTriangleFan;
+ case Maxwell::PrimitiveTopology::Quads:
+ // TODO(Rodrigo): Use VK_PRIMITIVE_TOPOLOGY_QUAD_LIST_EXT whenever it releases
+ return vk::PrimitiveTopology::eTriangleList;
+ case Maxwell::PrimitiveTopology::Patches:
+ return vk::PrimitiveTopology::ePatchList;
default:
UNIMPLEMENTED_MSG("Unimplemented topology={}", static_cast<u32>(topology));
return {};
@@ -236,37 +284,111 @@ vk::PrimitiveTopology PrimitiveTopology(Maxwell::PrimitiveTopology topology) {
vk::Format VertexFormat(Maxwell::VertexAttribute::Type type, Maxwell::VertexAttribute::Size size) {
switch (type) {
case Maxwell::VertexAttribute::Type::SignedNorm:
+ switch (size) {
+ case Maxwell::VertexAttribute::Size::Size_8:
+ return vk::Format::eR8Snorm;
+ case Maxwell::VertexAttribute::Size::Size_8_8:
+ return vk::Format::eR8G8Snorm;
+ case Maxwell::VertexAttribute::Size::Size_8_8_8:
+ return vk::Format::eR8G8B8Snorm;
+ case Maxwell::VertexAttribute::Size::Size_8_8_8_8:
+ return vk::Format::eR8G8B8A8Snorm;
+ case Maxwell::VertexAttribute::Size::Size_16:
+ return vk::Format::eR16Snorm;
+ case Maxwell::VertexAttribute::Size::Size_16_16:
+ return vk::Format::eR16G16Snorm;
+ case Maxwell::VertexAttribute::Size::Size_16_16_16:
+ return vk::Format::eR16G16B16Snorm;
+ case Maxwell::VertexAttribute::Size::Size_16_16_16_16:
+ return vk::Format::eR16G16B16A16Snorm;
+ case Maxwell::VertexAttribute::Size::Size_10_10_10_2:
+ return vk::Format::eA2B10G10R10SnormPack32;
+ default:
+ break;
+ }
break;
case Maxwell::VertexAttribute::Type::UnsignedNorm:
switch (size) {
+ case Maxwell::VertexAttribute::Size::Size_8:
+ return vk::Format::eR8Unorm;
+ case Maxwell::VertexAttribute::Size::Size_8_8:
+ return vk::Format::eR8G8Unorm;
+ case Maxwell::VertexAttribute::Size::Size_8_8_8:
+ return vk::Format::eR8G8B8Unorm;
case Maxwell::VertexAttribute::Size::Size_8_8_8_8:
return vk::Format::eR8G8B8A8Unorm;
+ case Maxwell::VertexAttribute::Size::Size_16:
+ return vk::Format::eR16Unorm;
+ case Maxwell::VertexAttribute::Size::Size_16_16:
+ return vk::Format::eR16G16Unorm;
+ case Maxwell::VertexAttribute::Size::Size_16_16_16:
+ return vk::Format::eR16G16B16Unorm;
+ case Maxwell::VertexAttribute::Size::Size_16_16_16_16:
+ return vk::Format::eR16G16B16A16Unorm;
default:
break;
}
break;
case Maxwell::VertexAttribute::Type::SignedInt:
- break;
+ switch (size) {
+ case Maxwell::VertexAttribute::Size::Size_16_16_16_16:
+ return vk::Format::eR16G16B16A16Sint;
+ case Maxwell::VertexAttribute::Size::Size_8:
+ return vk::Format::eR8Sint;
+ case Maxwell::VertexAttribute::Size::Size_8_8:
+ return vk::Format::eR8G8Sint;
+ case Maxwell::VertexAttribute::Size::Size_8_8_8:
+ return vk::Format::eR8G8B8Sint;
+ case Maxwell::VertexAttribute::Size::Size_8_8_8_8:
+ return vk::Format::eR8G8B8A8Sint;
+ case Maxwell::VertexAttribute::Size::Size_32:
+ return vk::Format::eR32Sint;
+ default:
+ break;
+ }
case Maxwell::VertexAttribute::Type::UnsignedInt:
switch (size) {
+ case Maxwell::VertexAttribute::Size::Size_8:
+ return vk::Format::eR8Uint;
+ case Maxwell::VertexAttribute::Size::Size_8_8:
+ return vk::Format::eR8G8Uint;
+ case Maxwell::VertexAttribute::Size::Size_8_8_8:
+ return vk::Format::eR8G8B8Uint;
+ case Maxwell::VertexAttribute::Size::Size_8_8_8_8:
+ return vk::Format::eR8G8B8A8Uint;
case Maxwell::VertexAttribute::Size::Size_32:
return vk::Format::eR32Uint;
default:
break;
}
case Maxwell::VertexAttribute::Type::UnsignedScaled:
+ switch (size) {
+ case Maxwell::VertexAttribute::Size::Size_8_8:
+ return vk::Format::eR8G8Uscaled;
+ default:
+ break;
+ }
+ break;
case Maxwell::VertexAttribute::Type::SignedScaled:
break;
case Maxwell::VertexAttribute::Type::Float:
switch (size) {
- case Maxwell::VertexAttribute::Size::Size_32_32_32_32:
- return vk::Format::eR32G32B32A32Sfloat;
- case Maxwell::VertexAttribute::Size::Size_32_32_32:
- return vk::Format::eR32G32B32Sfloat;
- case Maxwell::VertexAttribute::Size::Size_32_32:
- return vk::Format::eR32G32Sfloat;
case Maxwell::VertexAttribute::Size::Size_32:
return vk::Format::eR32Sfloat;
+ case Maxwell::VertexAttribute::Size::Size_32_32:
+ return vk::Format::eR32G32Sfloat;
+ case Maxwell::VertexAttribute::Size::Size_32_32_32:
+ return vk::Format::eR32G32B32Sfloat;
+ case Maxwell::VertexAttribute::Size::Size_32_32_32_32:
+ return vk::Format::eR32G32B32A32Sfloat;
+ case Maxwell::VertexAttribute::Size::Size_16:
+ return vk::Format::eR16Sfloat;
+ case Maxwell::VertexAttribute::Size::Size_16_16:
+ return vk::Format::eR16G16Sfloat;
+ case Maxwell::VertexAttribute::Size::Size_16_16_16:
+ return vk::Format::eR16G16B16Sfloat;
+ case Maxwell::VertexAttribute::Size::Size_16_16_16_16:
+ return vk::Format::eR16G16B16A16Sfloat;
default:
break;
}
@@ -308,11 +430,14 @@ vk::CompareOp ComparisonOp(Maxwell::ComparisonOp comparison) {
return {};
}
-vk::IndexType IndexFormat(Maxwell::IndexFormat index_format) {
+vk::IndexType IndexFormat(const VKDevice& device, Maxwell::IndexFormat index_format) {
switch (index_format) {
case Maxwell::IndexFormat::UnsignedByte:
- UNIMPLEMENTED_MSG("Vulkan does not support native u8 index format");
- return vk::IndexType::eUint16;
+ if (!device.IsExtIndexTypeUint8Supported()) {
+ UNIMPLEMENTED_MSG("Native uint8 indices are not supported on this device");
+ return vk::IndexType::eUint16;
+ }
+ return vk::IndexType::eUint8EXT;
case Maxwell::IndexFormat::UnsignedShort:
return vk::IndexType::eUint16;
case Maxwell::IndexFormat::UnsignedInt:
diff --git a/src/video_core/renderer_vulkan/maxwell_to_vk.h b/src/video_core/renderer_vulkan/maxwell_to_vk.h
index 4cadc0721..1534b738b 100644
--- a/src/video_core/renderer_vulkan/maxwell_to_vk.h
+++ b/src/video_core/renderer_vulkan/maxwell_to_vk.h
@@ -4,7 +4,6 @@
#pragma once
-#include <utility>
#include "common/common_types.h"
#include "video_core/engines/maxwell_3d.h"
#include "video_core/renderer_vulkan/declarations.h"
@@ -16,7 +15,6 @@ namespace Vulkan::MaxwellToVK {
using Maxwell = Tegra::Engines::Maxwell3D::Regs;
using PixelFormat = VideoCore::Surface::PixelFormat;
-using ComponentType = VideoCore::Surface::ComponentType;
namespace Sampler {
@@ -24,24 +22,31 @@ vk::Filter Filter(Tegra::Texture::TextureFilter filter);
vk::SamplerMipmapMode MipmapMode(Tegra::Texture::TextureMipmapFilter mipmap_filter);
-vk::SamplerAddressMode WrapMode(Tegra::Texture::WrapMode wrap_mode);
+vk::SamplerAddressMode WrapMode(Tegra::Texture::WrapMode wrap_mode,
+ Tegra::Texture::TextureFilter filter);
vk::CompareOp DepthCompareFunction(Tegra::Texture::DepthCompareFunc depth_compare_func);
} // namespace Sampler
-std::pair<vk::Format, bool> SurfaceFormat(const VKDevice& device, FormatType format_type,
- PixelFormat pixel_format, ComponentType component_type);
+struct FormatInfo {
+ vk::Format format;
+ bool attachable;
+ bool storage;
+};
-vk::ShaderStageFlagBits ShaderStage(Maxwell::ShaderStage stage);
+FormatInfo SurfaceFormat(const VKDevice& device, FormatType format_type, PixelFormat pixel_format);
-vk::PrimitiveTopology PrimitiveTopology(Maxwell::PrimitiveTopology topology);
+vk::ShaderStageFlagBits ShaderStage(Tegra::Engines::ShaderType stage);
+
+vk::PrimitiveTopology PrimitiveTopology(const VKDevice& device,
+ Maxwell::PrimitiveTopology topology);
vk::Format VertexFormat(Maxwell::VertexAttribute::Type type, Maxwell::VertexAttribute::Size size);
vk::CompareOp ComparisonOp(Maxwell::ComparisonOp comparison);
-vk::IndexType IndexFormat(Maxwell::IndexFormat index_format);
+vk::IndexType IndexFormat(const VKDevice& device, Maxwell::IndexFormat index_format);
vk::StencilOp StencilOp(Maxwell::StencilOp stencil_op);
diff --git a/src/video_core/renderer_vulkan/shaders/blit.frag b/src/video_core/renderer_vulkan/shaders/blit.frag
new file mode 100644
index 000000000..a06ecd24a
--- /dev/null
+++ b/src/video_core/renderer_vulkan/shaders/blit.frag
@@ -0,0 +1,24 @@
+// Copyright 2019 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+/*
+ * Build instructions:
+ * $ glslangValidator -V $THIS_FILE -o output.spv
+ * $ spirv-opt -O --strip-debug output.spv -o optimized.spv
+ * $ xxd -i optimized.spv
+ *
+ * Then copy that bytecode to the C++ file
+ */
+
+#version 460 core
+
+layout (location = 0) in vec2 frag_tex_coord;
+
+layout (location = 0) out vec4 color;
+
+layout (binding = 1) uniform sampler2D color_texture;
+
+void main() {
+ color = texture(color_texture, frag_tex_coord);
+}
diff --git a/src/video_core/renderer_vulkan/shaders/blit.vert b/src/video_core/renderer_vulkan/shaders/blit.vert
new file mode 100644
index 000000000..c64d9235a
--- /dev/null
+++ b/src/video_core/renderer_vulkan/shaders/blit.vert
@@ -0,0 +1,28 @@
+// Copyright 2019 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+/*
+ * Build instructions:
+ * $ glslangValidator -V $THIS_FILE -o output.spv
+ * $ spirv-opt -O --strip-debug output.spv -o optimized.spv
+ * $ xxd -i optimized.spv
+ *
+ * Then copy that bytecode to the C++ file
+ */
+
+#version 460 core
+
+layout (location = 0) in vec2 vert_position;
+layout (location = 1) in vec2 vert_tex_coord;
+
+layout (location = 0) out vec2 frag_tex_coord;
+
+layout (set = 0, binding = 0) uniform MatrixBlock {
+ mat4 modelview_matrix;
+};
+
+void main() {
+ gl_Position = modelview_matrix * vec4(vert_position, 0.0, 1.0);
+ frag_tex_coord = vert_tex_coord;
+}
diff --git a/src/video_core/renderer_vulkan/shaders/quad_array.comp b/src/video_core/renderer_vulkan/shaders/quad_array.comp
new file mode 100644
index 000000000..5a5703308
--- /dev/null
+++ b/src/video_core/renderer_vulkan/shaders/quad_array.comp
@@ -0,0 +1,37 @@
+// Copyright 2019 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+/*
+ * Build instructions:
+ * $ glslangValidator -V $THIS_FILE -o output.spv
+ * $ spirv-opt -O --strip-debug output.spv -o optimized.spv
+ * $ xxd -i optimized.spv
+ *
+ * Then copy that bytecode to the C++ file
+ */
+
+#version 460 core
+
+layout (local_size_x = 1024) in;
+
+layout (std430, set = 0, binding = 0) buffer OutputBuffer {
+ uint output_indexes[];
+};
+
+layout (push_constant) uniform PushConstants {
+ uint first;
+};
+
+void main() {
+ uint primitive = gl_GlobalInvocationID.x;
+ if (primitive * 6 >= output_indexes.length()) {
+ return;
+ }
+
+ const uint quad_map[6] = uint[](0, 1, 2, 0, 2, 3);
+ for (uint vertex = 0; vertex < 6; ++vertex) {
+ uint index = first + primitive * 4 + quad_map[vertex];
+ output_indexes[primitive * 6 + vertex] = index;
+ }
+}
diff --git a/src/video_core/renderer_vulkan/shaders/uint8.comp b/src/video_core/renderer_vulkan/shaders/uint8.comp
new file mode 100644
index 000000000..a320f3ae0
--- /dev/null
+++ b/src/video_core/renderer_vulkan/shaders/uint8.comp
@@ -0,0 +1,33 @@
+// Copyright 2019 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+/*
+ * Build instructions:
+ * $ glslangValidator -V $THIS_FILE -o output.spv
+ * $ spirv-opt -O --strip-debug output.spv -o optimized.spv
+ * $ xxd -i optimized.spv
+ *
+ * Then copy that bytecode to the C++ file
+ */
+
+#version 460 core
+#extension GL_EXT_shader_16bit_storage : require
+#extension GL_EXT_shader_8bit_storage : require
+
+layout (local_size_x = 1024) in;
+
+layout (std430, set = 0, binding = 0) readonly buffer InputBuffer {
+ uint8_t input_indexes[];
+};
+
+layout (std430, set = 0, binding = 1) writeonly buffer OutputBuffer {
+ uint16_t output_indexes[];
+};
+
+void main() {
+ uint id = gl_GlobalInvocationID.x;
+ if (id < input_indexes.length()) {
+ output_indexes[id] = uint16_t(input_indexes[id]);
+ }
+}
diff --git a/src/video_core/renderer_vulkan/vk_buffer_cache.cpp b/src/video_core/renderer_vulkan/vk_buffer_cache.cpp
index d2e9f4031..46da81aaa 100644
--- a/src/video_core/renderer_vulkan/vk_buffer_cache.cpp
+++ b/src/video_core/renderer_vulkan/vk_buffer_cache.cpp
@@ -24,9 +24,11 @@ CachedBufferEntry::CachedBufferEntry(VAddr cpu_addr, std::size_t size, u64 offse
alignment{alignment} {}
VKBufferCache::VKBufferCache(Tegra::MemoryManager& tegra_memory_manager,
+ Memory::Memory& cpu_memory_,
VideoCore::RasterizerInterface& rasterizer, const VKDevice& device,
VKMemoryManager& memory_manager, VKScheduler& scheduler, u64 size)
- : RasterizerCache{rasterizer}, tegra_memory_manager{tegra_memory_manager} {
+ : RasterizerCache{rasterizer}, tegra_memory_manager{tegra_memory_manager}, cpu_memory{
+ cpu_memory_} {
const auto usage = vk::BufferUsageFlagBits::eVertexBuffer |
vk::BufferUsageFlagBits::eIndexBuffer |
vk::BufferUsageFlagBits::eUniformBuffer;
@@ -48,9 +50,9 @@ u64 VKBufferCache::UploadMemory(GPUVAddr gpu_addr, std::size_t size, u64 alignme
// TODO: Figure out which size is the best for given games.
cache &= size >= 2048;
- const auto& host_ptr{Memory::GetPointer(*cpu_addr)};
+ u8* const host_ptr{cpu_memory.GetPointer(*cpu_addr)};
if (cache) {
- auto entry = TryGet(host_ptr);
+ const auto entry = TryGet(host_ptr);
if (entry) {
if (entry->GetSize() >= size && entry->GetAlignment() == alignment) {
return entry->GetOffset();
@@ -62,7 +64,7 @@ u64 VKBufferCache::UploadMemory(GPUVAddr gpu_addr, std::size_t size, u64 alignme
AlignBuffer(alignment);
const u64 uploaded_offset = buffer_offset;
- if (!host_ptr) {
+ if (host_ptr == nullptr) {
return uploaded_offset;
}
diff --git a/src/video_core/renderer_vulkan/vk_buffer_cache.h b/src/video_core/renderer_vulkan/vk_buffer_cache.h
index 49f13bcdc..daa8ccf66 100644
--- a/src/video_core/renderer_vulkan/vk_buffer_cache.h
+++ b/src/video_core/renderer_vulkan/vk_buffer_cache.h
@@ -13,6 +13,10 @@
#include "video_core/renderer_vulkan/declarations.h"
#include "video_core/renderer_vulkan/vk_scheduler.h"
+namespace Memory {
+class Memory;
+}
+
namespace Tegra {
class MemoryManager;
}
@@ -58,7 +62,7 @@ private:
class VKBufferCache final : public RasterizerCache<std::shared_ptr<CachedBufferEntry>> {
public:
- explicit VKBufferCache(Tegra::MemoryManager& tegra_memory_manager,
+ explicit VKBufferCache(Tegra::MemoryManager& tegra_memory_manager, Memory::Memory& cpu_memory_,
VideoCore::RasterizerInterface& rasterizer, const VKDevice& device,
VKMemoryManager& memory_manager, VKScheduler& scheduler, u64 size);
~VKBufferCache();
@@ -92,6 +96,7 @@ private:
void AlignBuffer(std::size_t alignment);
Tegra::MemoryManager& tegra_memory_manager;
+ Memory::Memory& cpu_memory;
std::unique_ptr<VKStreamBuffer> stream_buffer;
vk::Buffer buffer_handle;
diff --git a/src/video_core/renderer_vulkan/vk_device.cpp b/src/video_core/renderer_vulkan/vk_device.cpp
index 897cbb4e8..939eebe83 100644
--- a/src/video_core/renderer_vulkan/vk_device.cpp
+++ b/src/video_core/renderer_vulkan/vk_device.cpp
@@ -3,11 +3,15 @@
// Refer to the license.txt file included.
#include <bitset>
+#include <chrono>
+#include <cstdlib>
#include <optional>
#include <set>
#include <string_view>
+#include <thread>
#include <vector>
#include "common/assert.h"
+#include "core/settings.h"
#include "video_core/renderer_vulkan/declarations.h"
#include "video_core/renderer_vulkan/vk_device.h"
@@ -15,6 +19,15 @@ namespace Vulkan {
namespace {
+namespace Alternatives {
+
+constexpr std::array Depth24UnormS8Uint = {vk::Format::eD32SfloatS8Uint,
+ vk::Format::eD16UnormS8Uint, vk::Format{}};
+constexpr std::array Depth16UnormS8Uint = {vk::Format::eD24UnormS8Uint,
+ vk::Format::eD32SfloatS8Uint, vk::Format{}};
+
+} // namespace Alternatives
+
template <typename T>
void SetNext(void**& next, T& data) {
*next = &data;
@@ -22,7 +35,7 @@ void SetNext(void**& next, T& data) {
}
template <typename T>
-T GetFeatures(vk::PhysicalDevice physical, vk::DispatchLoaderDynamic dldi) {
+T GetFeatures(vk::PhysicalDevice physical, const vk::DispatchLoaderDynamic& dldi) {
vk::PhysicalDeviceFeatures2 features;
T extension_features;
features.pNext = &extension_features;
@@ -30,17 +43,14 @@ T GetFeatures(vk::PhysicalDevice physical, vk::DispatchLoaderDynamic dldi) {
return extension_features;
}
-} // Anonymous namespace
-
-namespace Alternatives {
-
-constexpr std::array Depth24UnormS8Uint = {vk::Format::eD32SfloatS8Uint,
- vk::Format::eD16UnormS8Uint, vk::Format{}};
-constexpr std::array Depth16UnormS8Uint = {vk::Format::eD24UnormS8Uint,
- vk::Format::eD32SfloatS8Uint, vk::Format{}};
-constexpr std::array Astc = {vk::Format::eA8B8G8R8UnormPack32, vk::Format{}};
-
-} // namespace Alternatives
+template <typename T>
+T GetProperties(vk::PhysicalDevice physical, const vk::DispatchLoaderDynamic& dldi) {
+ vk::PhysicalDeviceProperties2 properties;
+ T extension_properties;
+ properties.pNext = &extension_properties;
+ physical.getProperties2(&properties, dldi);
+ return extension_properties;
+}
constexpr const vk::Format* GetFormatAlternatives(vk::Format format) {
switch (format) {
@@ -53,8 +63,7 @@ constexpr const vk::Format* GetFormatAlternatives(vk::Format format) {
}
}
-constexpr vk::FormatFeatureFlags GetFormatFeatures(vk::FormatProperties properties,
- FormatType format_type) {
+vk::FormatFeatureFlags GetFormatFeatures(vk::FormatProperties properties, FormatType format_type) {
switch (format_type) {
case FormatType::Linear:
return properties.linearTilingFeatures;
@@ -67,11 +76,13 @@ constexpr vk::FormatFeatureFlags GetFormatFeatures(vk::FormatProperties properti
}
}
+} // Anonymous namespace
+
VKDevice::VKDevice(const vk::DispatchLoaderDynamic& dldi, vk::PhysicalDevice physical,
vk::SurfaceKHR surface)
- : physical{physical}, format_properties{GetFormatProperties(dldi, physical)} {
+ : physical{physical}, properties{physical.getProperties(dldi)},
+ format_properties{GetFormatProperties(dldi, physical)} {
SetupFamilies(dldi, surface);
- SetupProperties(dldi);
SetupFeatures(dldi);
}
@@ -89,12 +100,22 @@ bool VKDevice::Create(const vk::DispatchLoaderDynamic& dldi, vk::Instance instan
features.depthClamp = true;
features.samplerAnisotropy = true;
features.largePoints = true;
+ features.multiViewport = true;
+ features.depthBiasClamp = true;
+ features.geometryShader = true;
+ features.tessellationShader = true;
+ features.fragmentStoresAndAtomics = true;
+ features.shaderImageGatherExtended = true;
+ features.shaderStorageImageWriteWithoutFormat = true;
features.textureCompressionASTC_LDR = is_optimal_astc_supported;
- vk::PhysicalDeviceVertexAttributeDivisorFeaturesEXT vertex_divisor;
- vertex_divisor.vertexAttributeInstanceRateDivisor = true;
- vertex_divisor.vertexAttributeInstanceRateZeroDivisor = true;
- SetNext(next, vertex_divisor);
+ vk::PhysicalDevice16BitStorageFeaturesKHR bit16_storage;
+ bit16_storage.uniformAndStorageBuffer16BitAccess = true;
+ SetNext(next, bit16_storage);
+
+ vk::PhysicalDevice8BitStorageFeaturesKHR bit8_storage;
+ bit8_storage.uniformAndStorageBuffer8BitAccess = true;
+ SetNext(next, bit8_storage);
vk::PhysicalDeviceFloat16Int8FeaturesKHR float16_int8;
if (is_float16_supported) {
@@ -120,6 +141,10 @@ bool VKDevice::Create(const vk::DispatchLoaderDynamic& dldi, vk::Instance instan
LOG_INFO(Render_Vulkan, "Device doesn't support uint8 indexes");
}
+ if (!ext_depth_range_unrestricted) {
+ LOG_INFO(Render_Vulkan, "Device doesn't support depth range unrestricted");
+ }
+
vk::DeviceCreateInfo device_ci({}, static_cast<u32>(queue_cis.size()), queue_cis.data(), 0,
nullptr, static_cast<u32>(extensions.size()), extensions.data(),
nullptr);
@@ -135,16 +160,7 @@ bool VKDevice::Create(const vk::DispatchLoaderDynamic& dldi, vk::Instance instan
logical = UniqueDevice(
dummy_logical, vk::ObjectDestroy<vk::NoParent, vk::DispatchLoaderDynamic>(nullptr, dld));
- if (khr_driver_properties) {
- vk::PhysicalDeviceDriverPropertiesKHR driver;
- vk::PhysicalDeviceProperties2 properties;
- properties.pNext = &driver;
- physical.getProperties2(&properties, dld);
- driver_id = driver.driverID;
- LOG_INFO(Render_Vulkan, "Driver: {} {}", driver.driverName, driver.driverInfo);
- } else {
- LOG_INFO(Render_Vulkan, "Driver: Unknown");
- }
+ CollectTelemetryParameters();
graphics_queue = logical->getQueue(graphics_family, 0, dld);
present_queue = logical->getQueue(present_family, 0, dld);
@@ -188,8 +204,36 @@ vk::Format VKDevice::GetSupportedFormat(vk::Format wanted_format,
return wanted_format;
}
+void VKDevice::ReportLoss() const {
+ LOG_CRITICAL(Render_Vulkan, "Device loss occured!");
+
+ // Wait some time to let the log flush
+ std::this_thread::sleep_for(std::chrono::seconds{1});
+
+ if (!nv_device_diagnostic_checkpoints) {
+ return;
+ }
+
+ [[maybe_unused]] const std::vector data = graphics_queue.getCheckpointDataNV(dld);
+ // Catch here in debug builds (or with optimizations disabled) the last graphics pipeline to be
+ // executed. It can be done on a debugger by evaluating the expression:
+ // *(VKGraphicsPipeline*)data[0]
+}
+
bool VKDevice::IsOptimalAstcSupported(const vk::PhysicalDeviceFeatures& features,
const vk::DispatchLoaderDynamic& dldi) const {
+ // Disable for now to avoid converting ASTC twice.
+ return false;
+ static constexpr std::array astc_formats = {
+ vk::Format::eAstc4x4SrgbBlock, vk::Format::eAstc8x8SrgbBlock,
+ vk::Format::eAstc8x5SrgbBlock, vk::Format::eAstc5x4SrgbBlock,
+ vk::Format::eAstc5x5UnormBlock, vk::Format::eAstc5x5SrgbBlock,
+ vk::Format::eAstc10x8UnormBlock, vk::Format::eAstc10x8SrgbBlock,
+ vk::Format::eAstc6x6UnormBlock, vk::Format::eAstc6x6SrgbBlock,
+ vk::Format::eAstc10x10UnormBlock, vk::Format::eAstc10x10SrgbBlock,
+ vk::Format::eAstc12x12UnormBlock, vk::Format::eAstc12x12SrgbBlock,
+ vk::Format::eAstc8x6UnormBlock, vk::Format::eAstc8x6SrgbBlock,
+ vk::Format::eAstc6x5UnormBlock, vk::Format::eAstc6x5SrgbBlock};
if (!features.textureCompressionASTC_LDR) {
return false;
}
@@ -197,12 +241,6 @@ bool VKDevice::IsOptimalAstcSupported(const vk::PhysicalDeviceFeatures& features
vk::FormatFeatureFlagBits::eSampledImage | vk::FormatFeatureFlagBits::eBlitSrc |
vk::FormatFeatureFlagBits::eBlitDst | vk::FormatFeatureFlagBits::eTransferSrc |
vk::FormatFeatureFlagBits::eTransferDst};
- constexpr std::array astc_formats = {
- vk::Format::eAstc4x4UnormBlock, vk::Format::eAstc4x4SrgbBlock,
- vk::Format::eAstc8x8SrgbBlock, vk::Format::eAstc8x6SrgbBlock,
- vk::Format::eAstc5x4SrgbBlock, vk::Format::eAstc5x5UnormBlock,
- vk::Format::eAstc5x5SrgbBlock, vk::Format::eAstc10x8UnormBlock,
- vk::Format::eAstc10x8SrgbBlock};
for (const auto format : astc_formats) {
const auto format_properties{physical.getFormatProperties(format, dldi)};
if (!(format_properties.optimalTilingFeatures & format_feature_usage)) {
@@ -225,11 +263,17 @@ bool VKDevice::IsFormatSupported(vk::Format wanted_format, vk::FormatFeatureFlag
bool VKDevice::IsSuitable(const vk::DispatchLoaderDynamic& dldi, vk::PhysicalDevice physical,
vk::SurfaceKHR surface) {
- LOG_INFO(Render_Vulkan, "{}", physical.getProperties(dldi).deviceName);
bool is_suitable = true;
- constexpr std::array required_extensions = {VK_KHR_SWAPCHAIN_EXTENSION_NAME,
- VK_EXT_VERTEX_ATTRIBUTE_DIVISOR_EXTENSION_NAME};
+ constexpr std::array required_extensions = {
+ VK_KHR_SWAPCHAIN_EXTENSION_NAME,
+ VK_KHR_16BIT_STORAGE_EXTENSION_NAME,
+ VK_KHR_8BIT_STORAGE_EXTENSION_NAME,
+ VK_KHR_DRIVER_PROPERTIES_EXTENSION_NAME,
+ VK_EXT_VERTEX_ATTRIBUTE_DIVISOR_EXTENSION_NAME,
+ VK_EXT_SHADER_SUBGROUP_BALLOT_EXTENSION_NAME,
+ VK_EXT_SHADER_SUBGROUP_VOTE_EXTENSION_NAME,
+ };
std::bitset<required_extensions.size()> available_extensions{};
for (const auto& prop : physical.enumerateDeviceExtensionProperties(nullptr, dldi)) {
@@ -246,7 +290,7 @@ bool VKDevice::IsSuitable(const vk::DispatchLoaderDynamic& dldi, vk::PhysicalDev
if (available_extensions[i]) {
continue;
}
- LOG_INFO(Render_Vulkan, "Missing required extension: {}", required_extensions[i]);
+ LOG_ERROR(Render_Vulkan, "Missing required extension: {}", required_extensions[i]);
is_suitable = false;
}
}
@@ -263,7 +307,7 @@ bool VKDevice::IsSuitable(const vk::DispatchLoaderDynamic& dldi, vk::PhysicalDev
has_present |= physical.getSurfaceSupportKHR(i, surface, dldi) != 0;
}
if (!has_graphics || !has_present) {
- LOG_INFO(Render_Vulkan, "Device lacks a graphics and present queue");
+ LOG_ERROR(Render_Vulkan, "Device lacks a graphics and present queue");
is_suitable = false;
}
@@ -273,8 +317,15 @@ bool VKDevice::IsSuitable(const vk::DispatchLoaderDynamic& dldi, vk::PhysicalDev
constexpr u32 required_ubo_size = 65536;
if (limits.maxUniformBufferRange < required_ubo_size) {
- LOG_INFO(Render_Vulkan, "Device UBO size {} is too small, {} is required)",
- limits.maxUniformBufferRange, required_ubo_size);
+ LOG_ERROR(Render_Vulkan, "Device UBO size {} is too small, {} is required",
+ limits.maxUniformBufferRange, required_ubo_size);
+ is_suitable = false;
+ }
+
+ constexpr u32 required_num_viewports = 16;
+ if (limits.maxViewports < required_num_viewports) {
+ LOG_INFO(Render_Vulkan, "Device number of viewports {} is too small, {} is required",
+ limits.maxViewports, required_num_viewports);
is_suitable = false;
}
@@ -285,24 +336,32 @@ bool VKDevice::IsSuitable(const vk::DispatchLoaderDynamic& dldi, vk::PhysicalDev
std::make_pair(features.depthClamp, "depthClamp"),
std::make_pair(features.samplerAnisotropy, "samplerAnisotropy"),
std::make_pair(features.largePoints, "largePoints"),
+ std::make_pair(features.multiViewport, "multiViewport"),
+ std::make_pair(features.depthBiasClamp, "depthBiasClamp"),
+ std::make_pair(features.geometryShader, "geometryShader"),
+ std::make_pair(features.tessellationShader, "tessellationShader"),
+ std::make_pair(features.fragmentStoresAndAtomics, "fragmentStoresAndAtomics"),
+ std::make_pair(features.shaderImageGatherExtended, "shaderImageGatherExtended"),
+ std::make_pair(features.shaderStorageImageWriteWithoutFormat,
+ "shaderStorageImageWriteWithoutFormat"),
};
for (const auto& [supported, name] : feature_report) {
if (supported) {
continue;
}
- LOG_INFO(Render_Vulkan, "Missing required feature: {}", name);
+ LOG_ERROR(Render_Vulkan, "Missing required feature: {}", name);
is_suitable = false;
}
+ if (!is_suitable) {
+ LOG_ERROR(Render_Vulkan, "{} is not suitable", properties.deviceName);
+ }
+
return is_suitable;
}
std::vector<const char*> VKDevice::LoadExtensions(const vk::DispatchLoaderDynamic& dldi) {
std::vector<const char*> extensions;
- extensions.reserve(7);
- extensions.push_back(VK_KHR_SWAPCHAIN_EXTENSION_NAME);
- extensions.push_back(VK_EXT_VERTEX_ATTRIBUTE_DIVISOR_EXTENSION_NAME);
-
const auto Test = [&](const vk::ExtensionProperties& extension,
std::optional<std::reference_wrapper<bool>> status, const char* name,
bool push) {
@@ -317,13 +376,32 @@ std::vector<const char*> VKDevice::LoadExtensions(const vk::DispatchLoaderDynami
}
};
+ extensions.reserve(13);
+ extensions.push_back(VK_KHR_SWAPCHAIN_EXTENSION_NAME);
+ extensions.push_back(VK_KHR_16BIT_STORAGE_EXTENSION_NAME);
+ extensions.push_back(VK_KHR_8BIT_STORAGE_EXTENSION_NAME);
+ extensions.push_back(VK_KHR_DRIVER_PROPERTIES_EXTENSION_NAME);
+ extensions.push_back(VK_EXT_VERTEX_ATTRIBUTE_DIVISOR_EXTENSION_NAME);
+ extensions.push_back(VK_EXT_SHADER_SUBGROUP_BALLOT_EXTENSION_NAME);
+ extensions.push_back(VK_EXT_SHADER_SUBGROUP_VOTE_EXTENSION_NAME);
+
+ [[maybe_unused]] const bool nsight =
+ std::getenv("NVTX_INJECTION64_PATH") || std::getenv("NSIGHT_LAUNCHED");
bool khr_shader_float16_int8{};
+ bool ext_subgroup_size_control{};
for (const auto& extension : physical.enumerateDeviceExtensionProperties(nullptr, dldi)) {
Test(extension, khr_uniform_buffer_standard_layout,
VK_KHR_UNIFORM_BUFFER_STANDARD_LAYOUT_EXTENSION_NAME, true);
- Test(extension, ext_index_type_uint8, VK_EXT_INDEX_TYPE_UINT8_EXTENSION_NAME, true);
- Test(extension, khr_driver_properties, VK_KHR_DRIVER_PROPERTIES_EXTENSION_NAME, true);
Test(extension, khr_shader_float16_int8, VK_KHR_SHADER_FLOAT16_INT8_EXTENSION_NAME, false);
+ Test(extension, ext_depth_range_unrestricted,
+ VK_EXT_DEPTH_RANGE_UNRESTRICTED_EXTENSION_NAME, true);
+ Test(extension, ext_index_type_uint8, VK_EXT_INDEX_TYPE_UINT8_EXTENSION_NAME, true);
+ Test(extension, ext_shader_viewport_index_layer,
+ VK_EXT_SHADER_VIEWPORT_INDEX_LAYER_EXTENSION_NAME, true);
+ Test(extension, ext_subgroup_size_control, VK_EXT_SUBGROUP_SIZE_CONTROL_EXTENSION_NAME,
+ false);
+ Test(extension, nv_device_diagnostic_checkpoints,
+ VK_NV_DEVICE_DIAGNOSTIC_CHECKPOINTS_EXTENSION_NAME, true);
}
if (khr_shader_float16_int8) {
@@ -332,6 +410,23 @@ std::vector<const char*> VKDevice::LoadExtensions(const vk::DispatchLoaderDynami
extensions.push_back(VK_KHR_SHADER_FLOAT16_INT8_EXTENSION_NAME);
}
+ if (ext_subgroup_size_control) {
+ const auto features =
+ GetFeatures<vk::PhysicalDeviceSubgroupSizeControlFeaturesEXT>(physical, dldi);
+ const auto properties =
+ GetProperties<vk::PhysicalDeviceSubgroupSizeControlPropertiesEXT>(physical, dldi);
+
+ is_warp_potentially_bigger = properties.maxSubgroupSize > GuestWarpSize;
+
+ if (features.subgroupSizeControl && properties.minSubgroupSize <= GuestWarpSize &&
+ properties.maxSubgroupSize >= GuestWarpSize) {
+ extensions.push_back(VK_EXT_SUBGROUP_SIZE_CONTROL_EXTENSION_NAME);
+ guest_warp_stages = properties.requiredSubgroupSizeStages;
+ }
+ } else {
+ is_warp_potentially_bigger = true;
+ }
+
return extensions;
}
@@ -358,19 +453,23 @@ void VKDevice::SetupFamilies(const vk::DispatchLoaderDynamic& dldi, vk::SurfaceK
present_family = *present_family_;
}
-void VKDevice::SetupProperties(const vk::DispatchLoaderDynamic& dldi) {
- const auto props = physical.getProperties(dldi);
- device_type = props.deviceType;
- uniform_buffer_alignment = static_cast<u64>(props.limits.minUniformBufferOffsetAlignment);
- storage_buffer_alignment = static_cast<u64>(props.limits.minStorageBufferOffsetAlignment);
- max_storage_buffer_range = static_cast<u64>(props.limits.maxStorageBufferRange);
-}
-
void VKDevice::SetupFeatures(const vk::DispatchLoaderDynamic& dldi) {
const auto supported_features{physical.getFeatures(dldi)};
is_optimal_astc_supported = IsOptimalAstcSupported(supported_features, dldi);
}
+void VKDevice::CollectTelemetryParameters() {
+ const auto driver = GetProperties<vk::PhysicalDeviceDriverPropertiesKHR>(physical, dld);
+ driver_id = driver.driverID;
+ vendor_name = driver.driverName;
+
+ const auto extensions = physical.enumerateDeviceExtensionProperties(nullptr, dld);
+ reported_extensions.reserve(std::size(extensions));
+ for (const auto& extension : extensions) {
+ reported_extensions.push_back(extension.extensionName);
+ }
+}
+
std::vector<vk::DeviceQueueCreateInfo> VKDevice::GetDeviceQueueCreateInfos() const {
static const float QUEUE_PRIORITY = 1.0f;
@@ -385,50 +484,71 @@ std::vector<vk::DeviceQueueCreateInfo> VKDevice::GetDeviceQueueCreateInfos() con
std::unordered_map<vk::Format, vk::FormatProperties> VKDevice::GetFormatProperties(
const vk::DispatchLoaderDynamic& dldi, vk::PhysicalDevice physical) {
- constexpr std::array formats{vk::Format::eA8B8G8R8UnormPack32,
- vk::Format::eA8B8G8R8SnormPack32,
- vk::Format::eA8B8G8R8SrgbPack32,
- vk::Format::eB5G6R5UnormPack16,
- vk::Format::eA2B10G10R10UnormPack32,
- vk::Format::eR32G32B32A32Sfloat,
- vk::Format::eR16G16B16A16Uint,
- vk::Format::eR16G16Unorm,
- vk::Format::eR16G16Snorm,
- vk::Format::eR16G16Sfloat,
- vk::Format::eR16Unorm,
- vk::Format::eR8G8B8A8Srgb,
- vk::Format::eR8G8Unorm,
- vk::Format::eR8G8Snorm,
- vk::Format::eR8Unorm,
- vk::Format::eB10G11R11UfloatPack32,
- vk::Format::eR32Sfloat,
- vk::Format::eR16Sfloat,
- vk::Format::eR16G16B16A16Sfloat,
- vk::Format::eB8G8R8A8Unorm,
- vk::Format::eD32Sfloat,
- vk::Format::eD16Unorm,
- vk::Format::eD16UnormS8Uint,
- vk::Format::eD24UnormS8Uint,
- vk::Format::eD32SfloatS8Uint,
- vk::Format::eBc1RgbaUnormBlock,
- vk::Format::eBc2UnormBlock,
- vk::Format::eBc3UnormBlock,
- vk::Format::eBc4UnormBlock,
- vk::Format::eBc5UnormBlock,
- vk::Format::eBc5SnormBlock,
- vk::Format::eBc7UnormBlock,
- vk::Format::eBc1RgbaSrgbBlock,
- vk::Format::eBc3SrgbBlock,
- vk::Format::eBc7SrgbBlock,
- vk::Format::eAstc4x4UnormBlock,
- vk::Format::eAstc4x4SrgbBlock,
- vk::Format::eAstc8x8SrgbBlock,
- vk::Format::eAstc8x6SrgbBlock,
- vk::Format::eAstc5x4SrgbBlock,
- vk::Format::eAstc5x5UnormBlock,
- vk::Format::eAstc5x5SrgbBlock,
- vk::Format::eAstc10x8UnormBlock,
- vk::Format::eAstc10x8SrgbBlock};
+ static constexpr std::array formats{vk::Format::eA8B8G8R8UnormPack32,
+ vk::Format::eA8B8G8R8UintPack32,
+ vk::Format::eA8B8G8R8SnormPack32,
+ vk::Format::eA8B8G8R8SrgbPack32,
+ vk::Format::eB5G6R5UnormPack16,
+ vk::Format::eA2B10G10R10UnormPack32,
+ vk::Format::eA1R5G5B5UnormPack16,
+ vk::Format::eR32G32B32A32Sfloat,
+ vk::Format::eR32G32B32A32Uint,
+ vk::Format::eR32G32Sfloat,
+ vk::Format::eR32G32Uint,
+ vk::Format::eR16G16B16A16Uint,
+ vk::Format::eR16G16B16A16Unorm,
+ vk::Format::eR16G16Unorm,
+ vk::Format::eR16G16Snorm,
+ vk::Format::eR16G16Sfloat,
+ vk::Format::eR16Unorm,
+ vk::Format::eR8G8B8A8Srgb,
+ vk::Format::eR8G8Unorm,
+ vk::Format::eR8G8Snorm,
+ vk::Format::eR8Unorm,
+ vk::Format::eR8Uint,
+ vk::Format::eB10G11R11UfloatPack32,
+ vk::Format::eR32Sfloat,
+ vk::Format::eR32Uint,
+ vk::Format::eR16Sfloat,
+ vk::Format::eR16G16B16A16Sfloat,
+ vk::Format::eB8G8R8A8Unorm,
+ vk::Format::eR4G4B4A4UnormPack16,
+ vk::Format::eD32Sfloat,
+ vk::Format::eD16Unorm,
+ vk::Format::eD16UnormS8Uint,
+ vk::Format::eD24UnormS8Uint,
+ vk::Format::eD32SfloatS8Uint,
+ vk::Format::eBc1RgbaUnormBlock,
+ vk::Format::eBc2UnormBlock,
+ vk::Format::eBc3UnormBlock,
+ vk::Format::eBc4UnormBlock,
+ vk::Format::eBc5UnormBlock,
+ vk::Format::eBc5SnormBlock,
+ vk::Format::eBc7UnormBlock,
+ vk::Format::eBc6HUfloatBlock,
+ vk::Format::eBc6HSfloatBlock,
+ vk::Format::eBc1RgbaSrgbBlock,
+ vk::Format::eBc3SrgbBlock,
+ vk::Format::eBc7SrgbBlock,
+ vk::Format::eAstc4x4SrgbBlock,
+ vk::Format::eAstc8x8SrgbBlock,
+ vk::Format::eAstc8x5SrgbBlock,
+ vk::Format::eAstc5x4SrgbBlock,
+ vk::Format::eAstc5x5UnormBlock,
+ vk::Format::eAstc5x5SrgbBlock,
+ vk::Format::eAstc10x8UnormBlock,
+ vk::Format::eAstc10x8SrgbBlock,
+ vk::Format::eAstc6x6UnormBlock,
+ vk::Format::eAstc6x6SrgbBlock,
+ vk::Format::eAstc10x10UnormBlock,
+ vk::Format::eAstc10x10SrgbBlock,
+ vk::Format::eAstc12x12UnormBlock,
+ vk::Format::eAstc12x12SrgbBlock,
+ vk::Format::eAstc8x6UnormBlock,
+ vk::Format::eAstc8x6SrgbBlock,
+ vk::Format::eAstc6x5UnormBlock,
+ vk::Format::eAstc6x5SrgbBlock,
+ vk::Format::eE5B9G9R9UfloatPack32};
std::unordered_map<vk::Format, vk::FormatProperties> format_properties;
for (const auto format : formats) {
format_properties.emplace(format, physical.getFormatProperties(format, dldi));
diff --git a/src/video_core/renderer_vulkan/vk_device.h b/src/video_core/renderer_vulkan/vk_device.h
index 010d4c3d6..72603f9f6 100644
--- a/src/video_core/renderer_vulkan/vk_device.h
+++ b/src/video_core/renderer_vulkan/vk_device.h
@@ -4,6 +4,8 @@
#pragma once
+#include <string>
+#include <string_view>
#include <unordered_map>
#include <vector>
#include "common/common_types.h"
@@ -14,6 +16,9 @@ namespace Vulkan {
/// Format usage descriptor.
enum class FormatType { Linear, Optimal, Buffer };
+/// Subgroup size of the guest emulated hardware (Nvidia has 32 threads per subgroup).
+const u32 GuestWarpSize = 32;
+
/// Handles data specific to a physical device.
class VKDevice final {
public:
@@ -34,6 +39,9 @@ public:
vk::Format GetSupportedFormat(vk::Format wanted_format, vk::FormatFeatureFlags wanted_usage,
FormatType format_type) const;
+ /// Reports a device loss.
+ void ReportLoss() const;
+
/// Returns the dispatch loader with direct function pointers of the device.
const vk::DispatchLoaderDynamic& GetDispatchLoader() const {
return dld;
@@ -71,7 +79,22 @@ public:
/// Returns true if the device is integrated with the host CPU.
bool IsIntegrated() const {
- return device_type == vk::PhysicalDeviceType::eIntegratedGpu;
+ return properties.deviceType == vk::PhysicalDeviceType::eIntegratedGpu;
+ }
+
+ /// Returns the current Vulkan API version provided in Vulkan-formatted version numbers.
+ u32 GetApiVersion() const {
+ return properties.apiVersion;
+ }
+
+ /// Returns the current driver version provided in Vulkan-formatted version numbers.
+ u32 GetDriverVersion() const {
+ return properties.driverVersion;
+ }
+
+ /// Returns the device name.
+ std::string_view GetModelName() const {
+ return properties.deviceName;
}
/// Returns the driver ID.
@@ -80,18 +103,23 @@ public:
}
/// Returns uniform buffer alignment requeriment.
- u64 GetUniformBufferAlignment() const {
- return uniform_buffer_alignment;
+ vk::DeviceSize GetUniformBufferAlignment() const {
+ return properties.limits.minUniformBufferOffsetAlignment;
}
/// Returns storage alignment requeriment.
- u64 GetStorageBufferAlignment() const {
- return storage_buffer_alignment;
+ vk::DeviceSize GetStorageBufferAlignment() const {
+ return properties.limits.minStorageBufferOffsetAlignment;
}
/// Returns the maximum range for storage buffers.
- u64 GetMaxStorageBufferRange() const {
- return max_storage_buffer_range;
+ vk::DeviceSize GetMaxStorageBufferRange() const {
+ return properties.limits.maxStorageBufferRange;
+ }
+
+ /// Returns the maximum size for push constants.
+ vk::DeviceSize GetMaxPushConstantsSize() const {
+ return properties.limits.maxPushConstantsSize;
}
/// Returns true if ASTC is natively supported.
@@ -104,6 +132,16 @@ public:
return is_float16_supported;
}
+ /// Returns true if the device warp size can potentially be bigger than guest's warp size.
+ bool IsWarpSizePotentiallyBiggerThanGuest() const {
+ return is_warp_potentially_bigger;
+ }
+
+ /// Returns true if the device can be forced to use the guest warp size.
+ bool IsGuestWarpSizeSupported(vk::ShaderStageFlagBits stage) const {
+ return (guest_warp_stages & stage) != vk::ShaderStageFlags{};
+ }
+
/// Returns true if the device supports VK_EXT_scalar_block_layout.
bool IsKhrUniformBufferStandardLayoutSupported() const {
return khr_uniform_buffer_standard_layout;
@@ -114,6 +152,31 @@ public:
return ext_index_type_uint8;
}
+ /// Returns true if the device supports VK_EXT_depth_range_unrestricted.
+ bool IsExtDepthRangeUnrestrictedSupported() const {
+ return ext_depth_range_unrestricted;
+ }
+
+ /// Returns true if the device supports VK_EXT_shader_viewport_index_layer.
+ bool IsExtShaderViewportIndexLayerSupported() const {
+ return ext_shader_viewport_index_layer;
+ }
+
+ /// Returns true if the device supports VK_NV_device_diagnostic_checkpoints.
+ bool IsNvDeviceDiagnosticCheckpoints() const {
+ return nv_device_diagnostic_checkpoints;
+ }
+
+ /// Returns the vendor name reported from Vulkan.
+ std::string_view GetVendorName() const {
+ return vendor_name;
+ }
+
+ /// Returns the list of available extensions.
+ const std::vector<std::string>& GetAvailableExtensions() const {
+ return reported_extensions;
+ }
+
/// Checks if the physical device is suitable.
static bool IsSuitable(const vk::DispatchLoaderDynamic& dldi, vk::PhysicalDevice physical,
vk::SurfaceKHR surface);
@@ -125,12 +188,12 @@ private:
/// Sets up queue families.
void SetupFamilies(const vk::DispatchLoaderDynamic& dldi, vk::SurfaceKHR surface);
- /// Sets up device properties.
- void SetupProperties(const vk::DispatchLoaderDynamic& dldi);
-
/// Sets up device features.
void SetupFeatures(const vk::DispatchLoaderDynamic& dldi);
+ /// Collects telemetry information from the device.
+ void CollectTelemetryParameters();
+
/// Returns a list of queue initialization descriptors.
std::vector<vk::DeviceQueueCreateInfo> GetDeviceQueueCreateInfos() const;
@@ -148,23 +211,29 @@ private:
const vk::PhysicalDevice physical; ///< Physical device.
vk::DispatchLoaderDynamic dld; ///< Device function pointers.
+ vk::PhysicalDeviceProperties properties; ///< Device properties.
UniqueDevice logical; ///< Logical device.
vk::Queue graphics_queue; ///< Main graphics queue.
vk::Queue present_queue; ///< Main present queue.
u32 graphics_family{}; ///< Main graphics queue family index.
u32 present_family{}; ///< Main present queue family index.
- vk::PhysicalDeviceType device_type; ///< Physical device type.
vk::DriverIdKHR driver_id{}; ///< Driver ID.
- u64 uniform_buffer_alignment{}; ///< Uniform buffer alignment requeriment.
- u64 storage_buffer_alignment{}; ///< Storage buffer alignment requeriment.
- u64 max_storage_buffer_range{}; ///< Max storage buffer size.
+ vk::ShaderStageFlags guest_warp_stages{}; ///< Stages where the guest warp size can be forced.
bool is_optimal_astc_supported{}; ///< Support for native ASTC.
bool is_float16_supported{}; ///< Support for float16 arithmetics.
+ bool is_warp_potentially_bigger{}; ///< Host warp size can be bigger than guest.
bool khr_uniform_buffer_standard_layout{}; ///< Support for std430 on UBOs.
bool ext_index_type_uint8{}; ///< Support for VK_EXT_index_type_uint8.
- bool khr_driver_properties{}; ///< Support for VK_KHR_driver_properties.
- std::unordered_map<vk::Format, vk::FormatProperties>
- format_properties; ///< Format properties dictionary.
+ bool ext_depth_range_unrestricted{}; ///< Support for VK_EXT_depth_range_unrestricted.
+ bool ext_shader_viewport_index_layer{}; ///< Support for VK_EXT_shader_viewport_index_layer.
+ bool nv_device_diagnostic_checkpoints{}; ///< Support for VK_NV_device_diagnostic_checkpoints.
+
+ // Telemetry parameters
+ std::string vendor_name; ///< Device's driver name.
+ std::vector<std::string> reported_extensions; ///< Reported Vulkan extensions.
+
+ /// Format properties dictionary.
+ std::unordered_map<vk::Format, vk::FormatProperties> format_properties;
};
} // namespace Vulkan
diff --git a/src/video_core/renderer_vulkan/vk_image.cpp b/src/video_core/renderer_vulkan/vk_image.cpp
new file mode 100644
index 000000000..4bcbef959
--- /dev/null
+++ b/src/video_core/renderer_vulkan/vk_image.cpp
@@ -0,0 +1,106 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <memory>
+#include <vector>
+
+#include "common/assert.h"
+#include "video_core/renderer_vulkan/declarations.h"
+#include "video_core/renderer_vulkan/vk_device.h"
+#include "video_core/renderer_vulkan/vk_image.h"
+#include "video_core/renderer_vulkan/vk_scheduler.h"
+
+namespace Vulkan {
+
+VKImage::VKImage(const VKDevice& device, VKScheduler& scheduler,
+ const vk::ImageCreateInfo& image_ci, vk::ImageAspectFlags aspect_mask)
+ : device{device}, scheduler{scheduler}, format{image_ci.format}, aspect_mask{aspect_mask},
+ image_num_layers{image_ci.arrayLayers}, image_num_levels{image_ci.mipLevels} {
+ UNIMPLEMENTED_IF_MSG(image_ci.queueFamilyIndexCount != 0,
+ "Queue family tracking is not implemented");
+
+ const auto dev = device.GetLogical();
+ image = dev.createImageUnique(image_ci, nullptr, device.GetDispatchLoader());
+
+ const u32 num_ranges = image_num_layers * image_num_levels;
+ barriers.resize(num_ranges);
+ subrange_states.resize(num_ranges, {{}, image_ci.initialLayout});
+}
+
+VKImage::~VKImage() = default;
+
+void VKImage::Transition(u32 base_layer, u32 num_layers, u32 base_level, u32 num_levels,
+ vk::PipelineStageFlags new_stage_mask, vk::AccessFlags new_access,
+ vk::ImageLayout new_layout) {
+ if (!HasChanged(base_layer, num_layers, base_level, num_levels, new_access, new_layout)) {
+ return;
+ }
+
+ std::size_t cursor = 0;
+ for (u32 layer_it = 0; layer_it < num_layers; ++layer_it) {
+ for (u32 level_it = 0; level_it < num_levels; ++level_it, ++cursor) {
+ const u32 layer = base_layer + layer_it;
+ const u32 level = base_level + level_it;
+ auto& state = GetSubrangeState(layer, level);
+ barriers[cursor] = vk::ImageMemoryBarrier(
+ state.access, new_access, state.layout, new_layout, VK_QUEUE_FAMILY_IGNORED,
+ VK_QUEUE_FAMILY_IGNORED, *image, {aspect_mask, level, 1, layer, 1});
+ state.access = new_access;
+ state.layout = new_layout;
+ }
+ }
+
+ scheduler.RequestOutsideRenderPassOperationContext();
+
+ scheduler.Record([barriers = barriers, cursor](auto cmdbuf, auto& dld) {
+ // TODO(Rodrigo): Implement a way to use the latest stage across subresources.
+ constexpr auto stage_stub = vk::PipelineStageFlagBits::eAllCommands;
+ cmdbuf.pipelineBarrier(stage_stub, stage_stub, {}, 0, nullptr, 0, nullptr,
+ static_cast<u32>(cursor), barriers.data(), dld);
+ });
+}
+
+bool VKImage::HasChanged(u32 base_layer, u32 num_layers, u32 base_level, u32 num_levels,
+ vk::AccessFlags new_access, vk::ImageLayout new_layout) noexcept {
+ const bool is_full_range = base_layer == 0 && num_layers == image_num_layers &&
+ base_level == 0 && num_levels == image_num_levels;
+ if (!is_full_range) {
+ state_diverged = true;
+ }
+
+ if (!state_diverged) {
+ auto& state = GetSubrangeState(0, 0);
+ if (state.access != new_access || state.layout != new_layout) {
+ return true;
+ }
+ }
+
+ for (u32 layer_it = 0; layer_it < num_layers; ++layer_it) {
+ for (u32 level_it = 0; level_it < num_levels; ++level_it) {
+ const u32 layer = base_layer + layer_it;
+ const u32 level = base_level + level_it;
+ auto& state = GetSubrangeState(layer, level);
+ if (state.access != new_access || state.layout != new_layout) {
+ return true;
+ }
+ }
+ }
+ return false;
+}
+
+void VKImage::CreatePresentView() {
+ // Image type has to be 2D to be presented.
+ const vk::ImageViewCreateInfo image_view_ci({}, *image, vk::ImageViewType::e2D, format, {},
+ {aspect_mask, 0, 1, 0, 1});
+ const auto dev = device.GetLogical();
+ const auto& dld = device.GetDispatchLoader();
+ present_view = dev.createImageViewUnique(image_view_ci, nullptr, dld);
+}
+
+VKImage::SubrangeState& VKImage::GetSubrangeState(u32 layer, u32 level) noexcept {
+ return subrange_states[static_cast<std::size_t>(layer * image_num_levels) +
+ static_cast<std::size_t>(level)];
+}
+
+} // namespace Vulkan \ No newline at end of file
diff --git a/src/video_core/renderer_vulkan/vk_image.h b/src/video_core/renderer_vulkan/vk_image.h
new file mode 100644
index 000000000..b78242512
--- /dev/null
+++ b/src/video_core/renderer_vulkan/vk_image.h
@@ -0,0 +1,84 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <memory>
+#include <vector>
+
+#include "common/common_types.h"
+#include "video_core/renderer_vulkan/declarations.h"
+
+namespace Vulkan {
+
+class VKDevice;
+class VKScheduler;
+
+class VKImage {
+public:
+ explicit VKImage(const VKDevice& device, VKScheduler& scheduler,
+ const vk::ImageCreateInfo& image_ci, vk::ImageAspectFlags aspect_mask);
+ ~VKImage();
+
+ /// Records in the passed command buffer an image transition and updates the state of the image.
+ void Transition(u32 base_layer, u32 num_layers, u32 base_level, u32 num_levels,
+ vk::PipelineStageFlags new_stage_mask, vk::AccessFlags new_access,
+ vk::ImageLayout new_layout);
+
+ /// Returns a view compatible with presentation, the image has to be 2D.
+ vk::ImageView GetPresentView() {
+ if (!present_view) {
+ CreatePresentView();
+ }
+ return *present_view;
+ }
+
+ /// Returns the Vulkan image handler.
+ vk::Image GetHandle() const {
+ return *image;
+ }
+
+ /// Returns the Vulkan format for this image.
+ vk::Format GetFormat() const {
+ return format;
+ }
+
+ /// Returns the Vulkan aspect mask.
+ vk::ImageAspectFlags GetAspectMask() const {
+ return aspect_mask;
+ }
+
+private:
+ struct SubrangeState final {
+ vk::AccessFlags access{}; ///< Current access bits.
+ vk::ImageLayout layout = vk::ImageLayout::eUndefined; ///< Current image layout.
+ };
+
+ bool HasChanged(u32 base_layer, u32 num_layers, u32 base_level, u32 num_levels,
+ vk::AccessFlags new_access, vk::ImageLayout new_layout) noexcept;
+
+ /// Creates a presentation view.
+ void CreatePresentView();
+
+ /// Returns the subrange state for a layer and layer.
+ SubrangeState& GetSubrangeState(u32 layer, u32 level) noexcept;
+
+ const VKDevice& device; ///< Device handler.
+ VKScheduler& scheduler; ///< Device scheduler.
+
+ const vk::Format format; ///< Vulkan format.
+ const vk::ImageAspectFlags aspect_mask; ///< Vulkan aspect mask.
+ const u32 image_num_layers; ///< Number of layers.
+ const u32 image_num_levels; ///< Number of mipmap levels.
+
+ UniqueImage image; ///< Image handle.
+ UniqueImageView present_view; ///< Image view compatible with presentation.
+
+ std::vector<vk::ImageMemoryBarrier> barriers; ///< Pool of barriers.
+ std::vector<SubrangeState> subrange_states; ///< Current subrange state.
+
+ bool state_diverged = false; ///< True when subresources mismatch in layout.
+};
+
+} // namespace Vulkan
diff --git a/src/video_core/renderer_vulkan/vk_resource_manager.cpp b/src/video_core/renderer_vulkan/vk_resource_manager.cpp
index 13c46e5b8..525b4bb46 100644
--- a/src/video_core/renderer_vulkan/vk_resource_manager.cpp
+++ b/src/video_core/renderer_vulkan/vk_resource_manager.cpp
@@ -72,12 +72,22 @@ VKFence::VKFence(const VKDevice& device, UniqueFence handle)
VKFence::~VKFence() = default;
void VKFence::Wait() {
+ static constexpr u64 timeout = std::numeric_limits<u64>::max();
const auto dev = device.GetLogical();
const auto& dld = device.GetDispatchLoader();
- dev.waitForFences({*handle}, true, std::numeric_limits<u64>::max(), dld);
+ switch (const auto result = dev.waitForFences(1, &*handle, true, timeout, dld)) {
+ case vk::Result::eSuccess:
+ return;
+ case vk::Result::eErrorDeviceLost:
+ device.ReportLoss();
+ [[fallthrough]];
+ default:
+ vk::throwResultException(result, "vk::waitForFences");
+ }
}
void VKFence::Release() {
+ ASSERT(is_owned);
is_owned = false;
}
@@ -133,8 +143,32 @@ void VKFence::Unprotect(VKResource* resource) {
protected_resources.erase(it);
}
+void VKFence::RedirectProtection(VKResource* old_resource, VKResource* new_resource) noexcept {
+ std::replace(std::begin(protected_resources), std::end(protected_resources), old_resource,
+ new_resource);
+}
+
VKFenceWatch::VKFenceWatch() = default;
+VKFenceWatch::VKFenceWatch(VKFence& initial_fence) {
+ Watch(initial_fence);
+}
+
+VKFenceWatch::VKFenceWatch(VKFenceWatch&& rhs) noexcept {
+ fence = std::exchange(rhs.fence, nullptr);
+ if (fence) {
+ fence->RedirectProtection(&rhs, this);
+ }
+}
+
+VKFenceWatch& VKFenceWatch::operator=(VKFenceWatch&& rhs) noexcept {
+ fence = std::exchange(rhs.fence, nullptr);
+ if (fence) {
+ fence->RedirectProtection(&rhs, this);
+ }
+ return *this;
+}
+
VKFenceWatch::~VKFenceWatch() {
if (fence) {
fence->Unprotect(this);
diff --git a/src/video_core/renderer_vulkan/vk_resource_manager.h b/src/video_core/renderer_vulkan/vk_resource_manager.h
index 08ee86fa6..d4cbc95a5 100644
--- a/src/video_core/renderer_vulkan/vk_resource_manager.h
+++ b/src/video_core/renderer_vulkan/vk_resource_manager.h
@@ -65,6 +65,9 @@ public:
/// Removes protection for a resource.
void Unprotect(VKResource* resource);
+ /// Redirects one protected resource to a new address.
+ void RedirectProtection(VKResource* old_resource, VKResource* new_resource) noexcept;
+
/// Retreives the fence.
operator vk::Fence() const {
return *handle;
@@ -97,8 +100,13 @@ private:
class VKFenceWatch final : public VKResource {
public:
explicit VKFenceWatch();
+ VKFenceWatch(VKFence& initial_fence);
+ VKFenceWatch(VKFenceWatch&&) noexcept;
+ VKFenceWatch(const VKFenceWatch&) = delete;
~VKFenceWatch() override;
+ VKFenceWatch& operator=(VKFenceWatch&&) noexcept;
+
/// Waits for the fence to be released.
void Wait();
@@ -116,6 +124,14 @@ public:
void OnFenceRemoval(VKFence* signaling_fence) override;
+ /**
+ * Do not use it paired with Watch. Use TryWatch instead.
+ * Returns true when the watch is free.
+ */
+ bool IsUsed() const {
+ return fence != nullptr;
+ }
+
private:
VKFence* fence{}; ///< Fence watching this resource. nullptr when the watch is free.
};
diff --git a/src/video_core/renderer_vulkan/vk_sampler_cache.cpp b/src/video_core/renderer_vulkan/vk_sampler_cache.cpp
index 801826d3d..1ce583f75 100644
--- a/src/video_core/renderer_vulkan/vk_sampler_cache.cpp
+++ b/src/video_core/renderer_vulkan/vk_sampler_cache.cpp
@@ -46,9 +46,10 @@ UniqueSampler VKSamplerCache::CreateSampler(const Tegra::Texture::TSCEntry& tsc)
{}, MaxwellToVK::Sampler::Filter(tsc.mag_filter),
MaxwellToVK::Sampler::Filter(tsc.min_filter),
MaxwellToVK::Sampler::MipmapMode(tsc.mipmap_filter),
- MaxwellToVK::Sampler::WrapMode(tsc.wrap_u), MaxwellToVK::Sampler::WrapMode(tsc.wrap_v),
- MaxwellToVK::Sampler::WrapMode(tsc.wrap_p), tsc.GetLodBias(), has_anisotropy,
- max_anisotropy, tsc.depth_compare_enabled,
+ MaxwellToVK::Sampler::WrapMode(tsc.wrap_u, tsc.mag_filter),
+ MaxwellToVK::Sampler::WrapMode(tsc.wrap_v, tsc.mag_filter),
+ MaxwellToVK::Sampler::WrapMode(tsc.wrap_p, tsc.mag_filter), tsc.GetLodBias(),
+ has_anisotropy, max_anisotropy, tsc.depth_compare_enabled,
MaxwellToVK::Sampler::DepthCompareFunction(tsc.depth_compare_func), tsc.GetMinLod(),
tsc.GetMaxLod(), vk_border_color.value_or(vk::BorderColor::eFloatTransparentBlack),
unnormalized_coords);
diff --git a/src/video_core/renderer_vulkan/vk_scheduler.cpp b/src/video_core/renderer_vulkan/vk_scheduler.cpp
index 0f8116458..d66133ad1 100644
--- a/src/video_core/renderer_vulkan/vk_scheduler.cpp
+++ b/src/video_core/renderer_vulkan/vk_scheduler.cpp
@@ -3,7 +3,7 @@
// Refer to the license.txt file included.
#include "common/assert.h"
-#include "common/logging/log.h"
+#include "common/microprofile.h"
#include "video_core/renderer_vulkan/declarations.h"
#include "video_core/renderer_vulkan/vk_device.h"
#include "video_core/renderer_vulkan/vk_resource_manager.h"
@@ -11,46 +11,172 @@
namespace Vulkan {
+MICROPROFILE_DECLARE(Vulkan_WaitForWorker);
+
+void VKScheduler::CommandChunk::ExecuteAll(vk::CommandBuffer cmdbuf,
+ const vk::DispatchLoaderDynamic& dld) {
+ auto command = first;
+ while (command != nullptr) {
+ auto next = command->GetNext();
+ command->Execute(cmdbuf, dld);
+ command->~Command();
+ command = next;
+ }
+
+ command_offset = 0;
+ first = nullptr;
+ last = nullptr;
+}
+
VKScheduler::VKScheduler(const VKDevice& device, VKResourceManager& resource_manager)
- : device{device}, resource_manager{resource_manager} {
- next_fence = &resource_manager.CommitFence();
+ : device{device}, resource_manager{resource_manager}, next_fence{
+ &resource_manager.CommitFence()} {
+ AcquireNewChunk();
AllocateNewContext();
+ worker_thread = std::thread(&VKScheduler::WorkerThread, this);
}
-VKScheduler::~VKScheduler() = default;
+VKScheduler::~VKScheduler() {
+ quit = true;
+ cv.notify_all();
+ worker_thread.join();
+}
void VKScheduler::Flush(bool release_fence, vk::Semaphore semaphore) {
SubmitExecution(semaphore);
- if (release_fence)
+ if (release_fence) {
current_fence->Release();
+ }
AllocateNewContext();
}
void VKScheduler::Finish(bool release_fence, vk::Semaphore semaphore) {
SubmitExecution(semaphore);
current_fence->Wait();
- if (release_fence)
+ if (release_fence) {
current_fence->Release();
+ }
AllocateNewContext();
}
+void VKScheduler::WaitWorker() {
+ MICROPROFILE_SCOPE(Vulkan_WaitForWorker);
+ DispatchWork();
+
+ bool finished = false;
+ do {
+ cv.notify_all();
+ std::unique_lock lock{mutex};
+ finished = chunk_queue.Empty();
+ } while (!finished);
+}
+
+void VKScheduler::DispatchWork() {
+ if (chunk->Empty()) {
+ return;
+ }
+ chunk_queue.Push(std::move(chunk));
+ cv.notify_all();
+ AcquireNewChunk();
+}
+
+void VKScheduler::RequestRenderpass(const vk::RenderPassBeginInfo& renderpass_bi) {
+ if (state.renderpass && renderpass_bi == *state.renderpass) {
+ return;
+ }
+ const bool end_renderpass = state.renderpass.has_value();
+ state.renderpass = renderpass_bi;
+ Record([renderpass_bi, end_renderpass](auto cmdbuf, auto& dld) {
+ if (end_renderpass) {
+ cmdbuf.endRenderPass(dld);
+ }
+ cmdbuf.beginRenderPass(renderpass_bi, vk::SubpassContents::eInline, dld);
+ });
+}
+
+void VKScheduler::RequestOutsideRenderPassOperationContext() {
+ EndRenderPass();
+}
+
+void VKScheduler::BindGraphicsPipeline(vk::Pipeline pipeline) {
+ if (state.graphics_pipeline == pipeline) {
+ return;
+ }
+ state.graphics_pipeline = pipeline;
+ Record([pipeline](auto cmdbuf, auto& dld) {
+ cmdbuf.bindPipeline(vk::PipelineBindPoint::eGraphics, pipeline, dld);
+ });
+}
+
+void VKScheduler::WorkerThread() {
+ std::unique_lock lock{mutex};
+ do {
+ cv.wait(lock, [this] { return !chunk_queue.Empty() || quit; });
+ if (quit) {
+ continue;
+ }
+ auto extracted_chunk = std::move(chunk_queue.Front());
+ chunk_queue.Pop();
+ extracted_chunk->ExecuteAll(current_cmdbuf, device.GetDispatchLoader());
+ chunk_reserve.Push(std::move(extracted_chunk));
+ } while (!quit);
+}
+
void VKScheduler::SubmitExecution(vk::Semaphore semaphore) {
+ EndPendingOperations();
+ InvalidateState();
+ WaitWorker();
+
+ std::unique_lock lock{mutex};
+
+ const auto queue = device.GetGraphicsQueue();
const auto& dld = device.GetDispatchLoader();
current_cmdbuf.end(dld);
- const auto queue = device.GetGraphicsQueue();
- const vk::SubmitInfo submit_info(0, nullptr, nullptr, 1, &current_cmdbuf, semaphore ? 1u : 0u,
+ const vk::SubmitInfo submit_info(0, nullptr, nullptr, 1, &current_cmdbuf, semaphore ? 1U : 0U,
&semaphore);
- queue.submit({submit_info}, *current_fence, dld);
+ queue.submit({submit_info}, static_cast<vk::Fence>(*current_fence), dld);
}
void VKScheduler::AllocateNewContext() {
+ std::unique_lock lock{mutex};
current_fence = next_fence;
- current_cmdbuf = resource_manager.CommitCommandBuffer(*current_fence);
next_fence = &resource_manager.CommitFence();
- const auto& dld = device.GetDispatchLoader();
- current_cmdbuf.begin({vk::CommandBufferUsageFlagBits::eOneTimeSubmit}, dld);
+ current_cmdbuf = resource_manager.CommitCommandBuffer(*current_fence);
+ current_cmdbuf.begin({vk::CommandBufferUsageFlagBits::eOneTimeSubmit},
+ device.GetDispatchLoader());
+}
+
+void VKScheduler::InvalidateState() {
+ state.graphics_pipeline = nullptr;
+ state.viewports = false;
+ state.scissors = false;
+ state.depth_bias = false;
+ state.blend_constants = false;
+ state.depth_bounds = false;
+ state.stencil_values = false;
+}
+
+void VKScheduler::EndPendingOperations() {
+ EndRenderPass();
+}
+
+void VKScheduler::EndRenderPass() {
+ if (!state.renderpass) {
+ return;
+ }
+ state.renderpass = std::nullopt;
+ Record([](auto cmdbuf, auto& dld) { cmdbuf.endRenderPass(dld); });
+}
+
+void VKScheduler::AcquireNewChunk() {
+ if (chunk_reserve.Empty()) {
+ chunk = std::make_unique<CommandChunk>();
+ return;
+ }
+ chunk = std::move(chunk_reserve.Front());
+ chunk_reserve.Pop();
}
} // namespace Vulkan
diff --git a/src/video_core/renderer_vulkan/vk_scheduler.h b/src/video_core/renderer_vulkan/vk_scheduler.h
index 0e5b49c7f..bcdffbba0 100644
--- a/src/video_core/renderer_vulkan/vk_scheduler.h
+++ b/src/video_core/renderer_vulkan/vk_scheduler.h
@@ -4,7 +4,14 @@
#pragma once
+#include <condition_variable>
+#include <memory>
+#include <optional>
+#include <stack>
+#include <thread>
+#include <utility>
#include "common/common_types.h"
+#include "common/threadsafe_queue.h"
#include "video_core/renderer_vulkan/declarations.h"
namespace Vulkan {
@@ -30,56 +37,197 @@ private:
VKFence* const& fence;
};
-class VKCommandBufferView {
+/// The scheduler abstracts command buffer and fence management with an interface that's able to do
+/// OpenGL-like operations on Vulkan command buffers.
+class VKScheduler {
public:
- VKCommandBufferView() = default;
- VKCommandBufferView(const vk::CommandBuffer& cmdbuf) : cmdbuf{cmdbuf} {}
+ explicit VKScheduler(const VKDevice& device, VKResourceManager& resource_manager);
+ ~VKScheduler();
+
+ /// Sends the current execution context to the GPU.
+ void Flush(bool release_fence = true, vk::Semaphore semaphore = nullptr);
+
+ /// Sends the current execution context to the GPU and waits for it to complete.
+ void Finish(bool release_fence = true, vk::Semaphore semaphore = nullptr);
+
+ /// Waits for the worker thread to finish executing everything. After this function returns it's
+ /// safe to touch worker resources.
+ void WaitWorker();
+
+ /// Sends currently recorded work to the worker thread.
+ void DispatchWork();
+
+ /// Requests to begin a renderpass.
+ void RequestRenderpass(const vk::RenderPassBeginInfo& renderpass_bi);
+
+ /// Requests the current executino context to be able to execute operations only allowed outside
+ /// of a renderpass.
+ void RequestOutsideRenderPassOperationContext();
+
+ /// Binds a pipeline to the current execution context.
+ void BindGraphicsPipeline(vk::Pipeline pipeline);
- const vk::CommandBuffer* operator->() const noexcept {
- return &cmdbuf;
+ /// Returns true when viewports have been set in the current command buffer.
+ bool TouchViewports() {
+ return std::exchange(state.viewports, true);
}
- operator vk::CommandBuffer() const noexcept {
- return cmdbuf;
+ /// Returns true when scissors have been set in the current command buffer.
+ bool TouchScissors() {
+ return std::exchange(state.scissors, true);
}
-private:
- const vk::CommandBuffer& cmdbuf;
-};
+ /// Returns true when depth bias have been set in the current command buffer.
+ bool TouchDepthBias() {
+ return std::exchange(state.depth_bias, true);
+ }
-/// The scheduler abstracts command buffer and fence management with an interface that's able to do
-/// OpenGL-like operations on Vulkan command buffers.
-class VKScheduler {
-public:
- explicit VKScheduler(const VKDevice& device, VKResourceManager& resource_manager);
- ~VKScheduler();
+ /// Returns true when blend constants have been set in the current command buffer.
+ bool TouchBlendConstants() {
+ return std::exchange(state.blend_constants, true);
+ }
+
+ /// Returns true when depth bounds have been set in the current command buffer.
+ bool TouchDepthBounds() {
+ return std::exchange(state.depth_bounds, true);
+ }
+
+ /// Returns true when stencil values have been set in the current command buffer.
+ bool TouchStencilValues() {
+ return std::exchange(state.stencil_values, true);
+ }
+
+ /// Send work to a separate thread.
+ template <typename T>
+ void Record(T&& command) {
+ if (chunk->Record(command)) {
+ return;
+ }
+ DispatchWork();
+ (void)chunk->Record(command);
+ }
/// Gets a reference to the current fence.
VKFenceView GetFence() const {
return current_fence;
}
- /// Gets a reference to the current command buffer.
- VKCommandBufferView GetCommandBuffer() const {
- return current_cmdbuf;
- }
+private:
+ class Command {
+ public:
+ virtual ~Command() = default;
- /// Sends the current execution context to the GPU.
- void Flush(bool release_fence = true, vk::Semaphore semaphore = nullptr);
+ virtual void Execute(vk::CommandBuffer cmdbuf,
+ const vk::DispatchLoaderDynamic& dld) const = 0;
- /// Sends the current execution context to the GPU and waits for it to complete.
- void Finish(bool release_fence = true, vk::Semaphore semaphore = nullptr);
+ Command* GetNext() const {
+ return next;
+ }
+
+ void SetNext(Command* next_) {
+ next = next_;
+ }
+
+ private:
+ Command* next = nullptr;
+ };
+
+ template <typename T>
+ class TypedCommand final : public Command {
+ public:
+ explicit TypedCommand(T&& command) : command{std::move(command)} {}
+ ~TypedCommand() override = default;
+
+ TypedCommand(TypedCommand&&) = delete;
+ TypedCommand& operator=(TypedCommand&&) = delete;
+
+ void Execute(vk::CommandBuffer cmdbuf,
+ const vk::DispatchLoaderDynamic& dld) const override {
+ command(cmdbuf, dld);
+ }
+
+ private:
+ T command;
+ };
+
+ class CommandChunk final {
+ public:
+ void ExecuteAll(vk::CommandBuffer cmdbuf, const vk::DispatchLoaderDynamic& dld);
+
+ template <typename T>
+ bool Record(T& command) {
+ using FuncType = TypedCommand<T>;
+ static_assert(sizeof(FuncType) < sizeof(data), "Lambda is too large");
+
+ if (command_offset > sizeof(data) - sizeof(FuncType)) {
+ return false;
+ }
+
+ Command* current_last = last;
+
+ last = new (data.data() + command_offset) FuncType(std::move(command));
+
+ if (current_last) {
+ current_last->SetNext(last);
+ } else {
+ first = last;
+ }
+
+ command_offset += sizeof(FuncType);
+ return true;
+ }
+
+ bool Empty() const {
+ return command_offset == 0;
+ }
+
+ private:
+ Command* first = nullptr;
+ Command* last = nullptr;
+
+ std::size_t command_offset = 0;
+ std::array<u8, 0x8000> data{};
+ };
+
+ void WorkerThread();
-private:
void SubmitExecution(vk::Semaphore semaphore);
void AllocateNewContext();
+ void InvalidateState();
+
+ void EndPendingOperations();
+
+ void EndRenderPass();
+
+ void AcquireNewChunk();
+
const VKDevice& device;
VKResourceManager& resource_manager;
vk::CommandBuffer current_cmdbuf;
VKFence* current_fence = nullptr;
VKFence* next_fence = nullptr;
+
+ struct State {
+ std::optional<vk::RenderPassBeginInfo> renderpass;
+ vk::Pipeline graphics_pipeline;
+ bool viewports = false;
+ bool scissors = false;
+ bool depth_bias = false;
+ bool blend_constants = false;
+ bool depth_bounds = false;
+ bool stencil_values = false;
+ } state;
+
+ std::unique_ptr<CommandChunk> chunk;
+ std::thread worker_thread;
+
+ Common::SPSCQueue<std::unique_ptr<CommandChunk>> chunk_queue;
+ Common::SPSCQueue<std::unique_ptr<CommandChunk>> chunk_reserve;
+ std::mutex mutex;
+ std::condition_variable cv;
+ bool quit = false;
};
} // namespace Vulkan
diff --git a/src/video_core/renderer_vulkan/vk_shader_decompiler.cpp b/src/video_core/renderer_vulkan/vk_shader_decompiler.cpp
index 42cf068b6..8fe852ce8 100644
--- a/src/video_core/renderer_vulkan/vk_shader_decompiler.cpp
+++ b/src/video_core/renderer_vulkan/vk_shader_decompiler.cpp
@@ -3,8 +3,10 @@
// Refer to the license.txt file included.
#include <functional>
+#include <limits>
#include <map>
-#include <set>
+#include <type_traits>
+#include <utility>
#include <fmt/format.h>
@@ -17,39 +19,80 @@
#include "video_core/engines/maxwell_3d.h"
#include "video_core/engines/shader_bytecode.h"
#include "video_core/engines/shader_header.h"
+#include "video_core/engines/shader_type.h"
#include "video_core/renderer_vulkan/vk_device.h"
#include "video_core/renderer_vulkan/vk_shader_decompiler.h"
#include "video_core/shader/node.h"
#include "video_core/shader/shader_ir.h"
-namespace Vulkan::VKShader {
+namespace Vulkan {
+
+namespace {
using Sirit::Id;
+using Tegra::Engines::ShaderType;
using Tegra::Shader::Attribute;
using Tegra::Shader::AttributeUse;
using Tegra::Shader::Register;
using namespace VideoCommon::Shader;
using Maxwell = Tegra::Engines::Maxwell3D::Regs;
-using ShaderStage = Tegra::Engines::Maxwell3D::Regs::ShaderStage;
using Operation = const OperationNode&;
+class ASTDecompiler;
+class ExprDecompiler;
+
// TODO(Rodrigo): Use rasterizer's value
-constexpr u32 MAX_CONSTBUFFER_FLOATS = 0x4000;
-constexpr u32 MAX_CONSTBUFFER_ELEMENTS = MAX_CONSTBUFFER_FLOATS / 4;
-constexpr u32 STAGE_BINDING_STRIDE = 0x100;
+constexpr u32 MaxConstBufferFloats = 0x4000;
+constexpr u32 MaxConstBufferElements = MaxConstBufferFloats / 4;
-enum class Type { Bool, Bool2, Float, Int, Uint, HalfFloat };
+constexpr u32 NumInputPatches = 32; // This value seems to be the standard
+
+enum class Type { Void, Bool, Bool2, Float, Int, Uint, HalfFloat };
+
+class Expression final {
+public:
+ Expression(Id id, Type type) : id{id}, type{type} {
+ ASSERT(type != Type::Void);
+ }
+ Expression() : type{Type::Void} {}
-struct SamplerImage {
- Id image_type;
- Id sampled_image_type;
- Id sampler;
+ Id id{};
+ Type type{};
};
+static_assert(std::is_standard_layout_v<Expression>);
-namespace {
+struct TexelBuffer {
+ Id image_type{};
+ Id image{};
+};
+
+struct SampledImage {
+ Id image_type{};
+ Id sampled_image_type{};
+ Id sampler{};
+};
+
+struct StorageImage {
+ Id image_type{};
+ Id image{};
+};
+
+struct AttributeType {
+ Type type;
+ Id scalar;
+ Id vector;
+};
+
+struct VertexIndices {
+ std::optional<u32> position;
+ std::optional<u32> viewport;
+ std::optional<u32> point_size;
+ std::optional<u32> clip_distances;
+};
spv::Dim GetSamplerDim(const Sampler& sampler) {
+ ASSERT(!sampler.IsBuffer());
switch (sampler.GetType()) {
case Tegra::Shader::TextureType::Texture1D:
return spv::Dim::Dim1D;
@@ -65,6 +108,138 @@ spv::Dim GetSamplerDim(const Sampler& sampler) {
}
}
+std::pair<spv::Dim, bool> GetImageDim(const Image& image) {
+ switch (image.GetType()) {
+ case Tegra::Shader::ImageType::Texture1D:
+ return {spv::Dim::Dim1D, false};
+ case Tegra::Shader::ImageType::TextureBuffer:
+ return {spv::Dim::Buffer, false};
+ case Tegra::Shader::ImageType::Texture1DArray:
+ return {spv::Dim::Dim1D, true};
+ case Tegra::Shader::ImageType::Texture2D:
+ return {spv::Dim::Dim2D, false};
+ case Tegra::Shader::ImageType::Texture2DArray:
+ return {spv::Dim::Dim2D, true};
+ case Tegra::Shader::ImageType::Texture3D:
+ return {spv::Dim::Dim3D, false};
+ default:
+ UNIMPLEMENTED_MSG("Unimplemented image type={}", static_cast<u32>(image.GetType()));
+ return {spv::Dim::Dim2D, false};
+ }
+}
+
+/// Returns the number of vertices present in a primitive topology.
+u32 GetNumPrimitiveTopologyVertices(Maxwell::PrimitiveTopology primitive_topology) {
+ switch (primitive_topology) {
+ case Maxwell::PrimitiveTopology::Points:
+ return 1;
+ case Maxwell::PrimitiveTopology::Lines:
+ case Maxwell::PrimitiveTopology::LineLoop:
+ case Maxwell::PrimitiveTopology::LineStrip:
+ return 2;
+ case Maxwell::PrimitiveTopology::Triangles:
+ case Maxwell::PrimitiveTopology::TriangleStrip:
+ case Maxwell::PrimitiveTopology::TriangleFan:
+ return 3;
+ case Maxwell::PrimitiveTopology::LinesAdjacency:
+ case Maxwell::PrimitiveTopology::LineStripAdjacency:
+ return 4;
+ case Maxwell::PrimitiveTopology::TrianglesAdjacency:
+ case Maxwell::PrimitiveTopology::TriangleStripAdjacency:
+ return 6;
+ case Maxwell::PrimitiveTopology::Quads:
+ UNIMPLEMENTED_MSG("Quads");
+ return 3;
+ case Maxwell::PrimitiveTopology::QuadStrip:
+ UNIMPLEMENTED_MSG("QuadStrip");
+ return 3;
+ case Maxwell::PrimitiveTopology::Polygon:
+ UNIMPLEMENTED_MSG("Polygon");
+ return 3;
+ case Maxwell::PrimitiveTopology::Patches:
+ UNIMPLEMENTED_MSG("Patches");
+ return 3;
+ default:
+ UNREACHABLE();
+ return 3;
+ }
+}
+
+spv::ExecutionMode GetExecutionMode(Maxwell::TessellationPrimitive primitive) {
+ switch (primitive) {
+ case Maxwell::TessellationPrimitive::Isolines:
+ return spv::ExecutionMode::Isolines;
+ case Maxwell::TessellationPrimitive::Triangles:
+ return spv::ExecutionMode::Triangles;
+ case Maxwell::TessellationPrimitive::Quads:
+ return spv::ExecutionMode::Quads;
+ }
+ UNREACHABLE();
+ return spv::ExecutionMode::Triangles;
+}
+
+spv::ExecutionMode GetExecutionMode(Maxwell::TessellationSpacing spacing) {
+ switch (spacing) {
+ case Maxwell::TessellationSpacing::Equal:
+ return spv::ExecutionMode::SpacingEqual;
+ case Maxwell::TessellationSpacing::FractionalOdd:
+ return spv::ExecutionMode::SpacingFractionalOdd;
+ case Maxwell::TessellationSpacing::FractionalEven:
+ return spv::ExecutionMode::SpacingFractionalEven;
+ }
+ UNREACHABLE();
+ return spv::ExecutionMode::SpacingEqual;
+}
+
+spv::ExecutionMode GetExecutionMode(Maxwell::PrimitiveTopology input_topology) {
+ switch (input_topology) {
+ case Maxwell::PrimitiveTopology::Points:
+ return spv::ExecutionMode::InputPoints;
+ case Maxwell::PrimitiveTopology::Lines:
+ case Maxwell::PrimitiveTopology::LineLoop:
+ case Maxwell::PrimitiveTopology::LineStrip:
+ return spv::ExecutionMode::InputLines;
+ case Maxwell::PrimitiveTopology::Triangles:
+ case Maxwell::PrimitiveTopology::TriangleStrip:
+ case Maxwell::PrimitiveTopology::TriangleFan:
+ return spv::ExecutionMode::Triangles;
+ case Maxwell::PrimitiveTopology::LinesAdjacency:
+ case Maxwell::PrimitiveTopology::LineStripAdjacency:
+ return spv::ExecutionMode::InputLinesAdjacency;
+ case Maxwell::PrimitiveTopology::TrianglesAdjacency:
+ case Maxwell::PrimitiveTopology::TriangleStripAdjacency:
+ return spv::ExecutionMode::InputTrianglesAdjacency;
+ case Maxwell::PrimitiveTopology::Quads:
+ UNIMPLEMENTED_MSG("Quads");
+ return spv::ExecutionMode::Triangles;
+ case Maxwell::PrimitiveTopology::QuadStrip:
+ UNIMPLEMENTED_MSG("QuadStrip");
+ return spv::ExecutionMode::Triangles;
+ case Maxwell::PrimitiveTopology::Polygon:
+ UNIMPLEMENTED_MSG("Polygon");
+ return spv::ExecutionMode::Triangles;
+ case Maxwell::PrimitiveTopology::Patches:
+ UNIMPLEMENTED_MSG("Patches");
+ return spv::ExecutionMode::Triangles;
+ }
+ UNREACHABLE();
+ return spv::ExecutionMode::Triangles;
+}
+
+spv::ExecutionMode GetExecutionMode(Tegra::Shader::OutputTopology output_topology) {
+ switch (output_topology) {
+ case Tegra::Shader::OutputTopology::PointList:
+ return spv::ExecutionMode::OutputPoints;
+ case Tegra::Shader::OutputTopology::LineStrip:
+ return spv::ExecutionMode::OutputLineStrip;
+ case Tegra::Shader::OutputTopology::TriangleStrip:
+ return spv::ExecutionMode::OutputTriangleStrip;
+ default:
+ UNREACHABLE();
+ return spv::ExecutionMode::OutputPoints;
+ }
+}
+
/// Returns true if an attribute index is one of the 32 generic attributes
constexpr bool IsGenericAttribute(Attribute::Index attribute) {
return attribute >= Attribute::Index::Attribute_0 &&
@@ -72,7 +247,7 @@ constexpr bool IsGenericAttribute(Attribute::Index attribute) {
}
/// Returns the location of a generic attribute
-constexpr u32 GetGenericAttributeLocation(Attribute::Index attribute) {
+u32 GetGenericAttributeLocation(Attribute::Index attribute) {
ASSERT(IsGenericAttribute(attribute));
return static_cast<u32>(attribute) - static_cast<u32>(Attribute::Index::Attribute_0);
}
@@ -86,20 +261,146 @@ bool IsPrecise(Operation operand) {
return false;
}
-} // namespace
-
-class ASTDecompiler;
-class ExprDecompiler;
-
-class SPIRVDecompiler : public Sirit::Module {
+class SPIRVDecompiler final : public Sirit::Module {
public:
- explicit SPIRVDecompiler(const VKDevice& device, const ShaderIR& ir, ShaderStage stage)
- : Module(0x00010300), device{device}, ir{ir}, stage{stage}, header{ir.GetHeader()} {
+ explicit SPIRVDecompiler(const VKDevice& device, const ShaderIR& ir, ShaderType stage,
+ const Specialization& specialization)
+ : Module(0x00010300), device{device}, ir{ir}, stage{stage}, header{ir.GetHeader()},
+ specialization{specialization} {
AddCapability(spv::Capability::Shader);
+ AddCapability(spv::Capability::UniformAndStorageBuffer16BitAccess);
+ AddCapability(spv::Capability::ImageQuery);
+ AddCapability(spv::Capability::Image1D);
+ AddCapability(spv::Capability::ImageBuffer);
+ AddCapability(spv::Capability::ImageGatherExtended);
+ AddCapability(spv::Capability::SampledBuffer);
+ AddCapability(spv::Capability::StorageImageWriteWithoutFormat);
+ AddCapability(spv::Capability::SubgroupBallotKHR);
+ AddCapability(spv::Capability::SubgroupVoteKHR);
+ AddExtension("SPV_KHR_shader_ballot");
+ AddExtension("SPV_KHR_subgroup_vote");
AddExtension("SPV_KHR_storage_buffer_storage_class");
AddExtension("SPV_KHR_variable_pointers");
+
+ if (ir.UsesViewportIndex()) {
+ AddCapability(spv::Capability::MultiViewport);
+ if (device.IsExtShaderViewportIndexLayerSupported()) {
+ AddExtension("SPV_EXT_shader_viewport_index_layer");
+ AddCapability(spv::Capability::ShaderViewportIndexLayerEXT);
+ }
+ }
+
+ if (device.IsFloat16Supported()) {
+ AddCapability(spv::Capability::Float16);
+ }
+ t_scalar_half = Name(TypeFloat(device.IsFloat16Supported() ? 16 : 32), "scalar_half");
+ t_half = Name(TypeVector(t_scalar_half, 2), "half");
+
+ const Id main = Decompile();
+
+ switch (stage) {
+ case ShaderType::Vertex:
+ AddEntryPoint(spv::ExecutionModel::Vertex, main, "main", interfaces);
+ break;
+ case ShaderType::TesselationControl:
+ AddCapability(spv::Capability::Tessellation);
+ AddEntryPoint(spv::ExecutionModel::TessellationControl, main, "main", interfaces);
+ AddExecutionMode(main, spv::ExecutionMode::OutputVertices,
+ header.common2.threads_per_input_primitive);
+ break;
+ case ShaderType::TesselationEval:
+ AddCapability(spv::Capability::Tessellation);
+ AddEntryPoint(spv::ExecutionModel::TessellationEvaluation, main, "main", interfaces);
+ AddExecutionMode(main, GetExecutionMode(specialization.tessellation.primitive));
+ AddExecutionMode(main, GetExecutionMode(specialization.tessellation.spacing));
+ AddExecutionMode(main, specialization.tessellation.clockwise
+ ? spv::ExecutionMode::VertexOrderCw
+ : spv::ExecutionMode::VertexOrderCcw);
+ break;
+ case ShaderType::Geometry:
+ AddCapability(spv::Capability::Geometry);
+ AddEntryPoint(spv::ExecutionModel::Geometry, main, "main", interfaces);
+ AddExecutionMode(main, GetExecutionMode(specialization.primitive_topology));
+ AddExecutionMode(main, GetExecutionMode(header.common3.output_topology));
+ AddExecutionMode(main, spv::ExecutionMode::OutputVertices,
+ header.common4.max_output_vertices);
+ // TODO(Rodrigo): Where can we get this info from?
+ AddExecutionMode(main, spv::ExecutionMode::Invocations, 1U);
+ break;
+ case ShaderType::Fragment:
+ AddEntryPoint(spv::ExecutionModel::Fragment, main, "main", interfaces);
+ AddExecutionMode(main, spv::ExecutionMode::OriginUpperLeft);
+ if (header.ps.omap.depth) {
+ AddExecutionMode(main, spv::ExecutionMode::DepthReplacing);
+ }
+ break;
+ case ShaderType::Compute:
+ const auto workgroup_size = specialization.workgroup_size;
+ AddExecutionMode(main, spv::ExecutionMode::LocalSize, workgroup_size[0],
+ workgroup_size[1], workgroup_size[2]);
+ AddEntryPoint(spv::ExecutionModel::GLCompute, main, "main", interfaces);
+ break;
+ }
}
+private:
+ Id Decompile() {
+ DeclareCommon();
+ DeclareVertex();
+ DeclareTessControl();
+ DeclareTessEval();
+ DeclareGeometry();
+ DeclareFragment();
+ DeclareCompute();
+ DeclareRegisters();
+ DeclarePredicates();
+ DeclareLocalMemory();
+ DeclareSharedMemory();
+ DeclareInternalFlags();
+ DeclareInputAttributes();
+ DeclareOutputAttributes();
+
+ u32 binding = specialization.base_binding;
+ binding = DeclareConstantBuffers(binding);
+ binding = DeclareGlobalBuffers(binding);
+ binding = DeclareTexelBuffers(binding);
+ binding = DeclareSamplers(binding);
+ binding = DeclareImages(binding);
+
+ const Id main = OpFunction(t_void, {}, TypeFunction(t_void));
+ AddLabel();
+
+ if (ir.IsDecompiled()) {
+ DeclareFlowVariables();
+ DecompileAST();
+ } else {
+ AllocateLabels();
+ DecompileBranchMode();
+ }
+
+ OpReturn();
+ OpFunctionEnd();
+
+ return main;
+ }
+
+ void DefinePrologue() {
+ if (stage == ShaderType::Vertex) {
+ // Clear Position to avoid reading trash on the Z conversion.
+ const auto position_index = out_indices.position.value();
+ const Id position = AccessElement(t_out_float4, out_vertex, position_index);
+ OpStore(position, v_varying_default);
+
+ if (specialization.point_size) {
+ const u32 point_size_index = out_indices.point_size.value();
+ const Id out_point_size = AccessElement(t_out_float, out_vertex, point_size_index);
+ OpStore(out_point_size, Constant(t_float, *specialization.point_size));
+ }
+ }
+ }
+
+ void DecompileAST();
+
void DecompileBranchMode() {
const u32 first_address = ir.GetBasicBlocks().begin()->first;
const Id loop_label = OpLabel("loop");
@@ -110,14 +411,15 @@ public:
std::vector<Sirit::Literal> literals;
std::vector<Id> branch_labels;
- for (const auto& pair : labels) {
- const auto [literal, label] = pair;
+ for (const auto& [literal, label] : labels) {
literals.push_back(literal);
branch_labels.push_back(label);
}
- jmp_to = Emit(OpVariable(TypePointer(spv::StorageClass::Function, t_uint),
- spv::StorageClass::Function, Constant(t_uint, first_address)));
+ jmp_to = OpVariable(TypePointer(spv::StorageClass::Function, t_uint),
+ spv::StorageClass::Function, Constant(t_uint, first_address));
+ AddLocalVariable(jmp_to);
+
std::tie(ssy_flow_stack, ssy_flow_stack_top) = CreateFlowStack();
std::tie(pbk_flow_stack, pbk_flow_stack_top) = CreateFlowStack();
@@ -127,154 +429,121 @@ public:
Name(pbk_flow_stack, "pbk_flow_stack");
Name(pbk_flow_stack_top, "pbk_flow_stack_top");
- Emit(OpBranch(loop_label));
- Emit(loop_label);
- Emit(OpLoopMerge(merge_label, continue_label, spv::LoopControlMask::Unroll));
- Emit(OpBranch(dummy_label));
+ DefinePrologue();
- Emit(dummy_label);
+ OpBranch(loop_label);
+ AddLabel(loop_label);
+ OpLoopMerge(merge_label, continue_label, spv::LoopControlMask::MaskNone);
+ OpBranch(dummy_label);
+
+ AddLabel(dummy_label);
const Id default_branch = OpLabel();
- const Id jmp_to_load = Emit(OpLoad(t_uint, jmp_to));
- Emit(OpSelectionMerge(jump_label, spv::SelectionControlMask::MaskNone));
- Emit(OpSwitch(jmp_to_load, default_branch, literals, branch_labels));
+ const Id jmp_to_load = OpLoad(t_uint, jmp_to);
+ OpSelectionMerge(jump_label, spv::SelectionControlMask::MaskNone);
+ OpSwitch(jmp_to_load, default_branch, literals, branch_labels);
- Emit(default_branch);
- Emit(OpReturn());
+ AddLabel(default_branch);
+ OpReturn();
- for (const auto& pair : ir.GetBasicBlocks()) {
- const auto& [address, bb] = pair;
- Emit(labels.at(address));
+ for (const auto& [address, bb] : ir.GetBasicBlocks()) {
+ AddLabel(labels.at(address));
VisitBasicBlock(bb);
const auto next_it = labels.lower_bound(address + 1);
const Id next_label = next_it != labels.end() ? next_it->second : default_branch;
- Emit(OpBranch(next_label));
+ OpBranch(next_label);
}
- Emit(jump_label);
- Emit(OpBranch(continue_label));
- Emit(continue_label);
- Emit(OpBranch(loop_label));
- Emit(merge_label);
+ AddLabel(jump_label);
+ OpBranch(continue_label);
+ AddLabel(continue_label);
+ OpBranch(loop_label);
+ AddLabel(merge_label);
}
- void DecompileAST();
+private:
+ friend class ASTDecompiler;
+ friend class ExprDecompiler;
- void Decompile() {
- const bool is_fully_decompiled = ir.IsDecompiled();
- AllocateBindings();
- if (!is_fully_decompiled) {
- AllocateLabels();
- }
+ static constexpr auto INTERNAL_FLAGS_COUNT = static_cast<std::size_t>(InternalFlag::Amount);
- DeclareVertex();
- DeclareGeometry();
- DeclareFragment();
- DeclareRegisters();
- DeclarePredicates();
- if (is_fully_decompiled) {
- DeclareFlowVariables();
+ void AllocateLabels() {
+ for (const auto& pair : ir.GetBasicBlocks()) {
+ const u32 address = pair.first;
+ labels.emplace(address, OpLabel(fmt::format("label_0x{:x}", address)));
}
- DeclareLocalMemory();
- DeclareInternalFlags();
- DeclareInputAttributes();
- DeclareOutputAttributes();
- DeclareConstantBuffers();
- DeclareGlobalBuffers();
- DeclareSamplers();
+ }
- execute_function =
- Emit(OpFunction(t_void, spv::FunctionControlMask::Inline, TypeFunction(t_void)));
- Emit(OpLabel());
+ void DeclareCommon() {
+ thread_id =
+ DeclareInputBuiltIn(spv::BuiltIn::SubgroupLocalInvocationId, t_in_uint, "thread_id");
+ }
- if (is_fully_decompiled) {
- DecompileAST();
- } else {
- DecompileBranchMode();
+ void DeclareVertex() {
+ if (stage != ShaderType::Vertex) {
+ return;
}
+ Id out_vertex_struct;
+ std::tie(out_vertex_struct, out_indices) = DeclareVertexStruct();
+ const Id vertex_ptr = TypePointer(spv::StorageClass::Output, out_vertex_struct);
+ out_vertex = OpVariable(vertex_ptr, spv::StorageClass::Output);
+ interfaces.push_back(AddGlobalVariable(Name(out_vertex, "out_vertex")));
- Emit(OpReturn());
- Emit(OpFunctionEnd());
+ // Declare input attributes
+ vertex_index = DeclareInputBuiltIn(spv::BuiltIn::VertexIndex, t_in_uint, "vertex_index");
+ instance_index =
+ DeclareInputBuiltIn(spv::BuiltIn::InstanceIndex, t_in_uint, "instance_index");
}
- ShaderEntries GetShaderEntries() const {
- ShaderEntries entries;
- entries.const_buffers_base_binding = const_buffers_base_binding;
- entries.global_buffers_base_binding = global_buffers_base_binding;
- entries.samplers_base_binding = samplers_base_binding;
- for (const auto& cbuf : ir.GetConstantBuffers()) {
- entries.const_buffers.emplace_back(cbuf.second, cbuf.first);
- }
- for (const auto& gmem_pair : ir.GetGlobalMemory()) {
- const auto& [base, usage] = gmem_pair;
- entries.global_buffers.emplace_back(base.cbuf_index, base.cbuf_offset);
- }
- for (const auto& sampler : ir.GetSamplers()) {
- entries.samplers.emplace_back(sampler);
- }
- for (const auto& attribute : ir.GetInputAttributes()) {
- if (IsGenericAttribute(attribute)) {
- entries.attributes.insert(GetGenericAttributeLocation(attribute));
- }
+ void DeclareTessControl() {
+ if (stage != ShaderType::TesselationControl) {
+ return;
}
- entries.clip_distances = ir.GetClipDistances();
- entries.shader_length = ir.GetLength();
- entries.entry_function = execute_function;
- entries.interfaces = interfaces;
- return entries;
- }
-
-private:
- friend class ASTDecompiler;
- friend class ExprDecompiler;
+ DeclareInputVertexArray(NumInputPatches);
+ DeclareOutputVertexArray(header.common2.threads_per_input_primitive);
- static constexpr auto INTERNAL_FLAGS_COUNT = static_cast<std::size_t>(InternalFlag::Amount);
+ tess_level_outer = DeclareBuiltIn(
+ spv::BuiltIn::TessLevelOuter, spv::StorageClass::Output,
+ TypePointer(spv::StorageClass::Output, TypeArray(t_float, Constant(t_uint, 4U))),
+ "tess_level_outer");
+ Decorate(tess_level_outer, spv::Decoration::Patch);
- void AllocateBindings() {
- const u32 binding_base = static_cast<u32>(stage) * STAGE_BINDING_STRIDE;
- u32 binding_iterator = binding_base;
+ tess_level_inner = DeclareBuiltIn(
+ spv::BuiltIn::TessLevelInner, spv::StorageClass::Output,
+ TypePointer(spv::StorageClass::Output, TypeArray(t_float, Constant(t_uint, 2U))),
+ "tess_level_inner");
+ Decorate(tess_level_inner, spv::Decoration::Patch);
- const auto Allocate = [&binding_iterator](std::size_t count) {
- const u32 current_binding = binding_iterator;
- binding_iterator += static_cast<u32>(count);
- return current_binding;
- };
- const_buffers_base_binding = Allocate(ir.GetConstantBuffers().size());
- global_buffers_base_binding = Allocate(ir.GetGlobalMemory().size());
- samplers_base_binding = Allocate(ir.GetSamplers().size());
-
- ASSERT_MSG(binding_iterator - binding_base < STAGE_BINDING_STRIDE,
- "Stage binding stride is too small");
+ invocation_id = DeclareInputBuiltIn(spv::BuiltIn::InvocationId, t_in_int, "invocation_id");
}
- void AllocateLabels() {
- for (const auto& pair : ir.GetBasicBlocks()) {
- const u32 address = pair.first;
- labels.emplace(address, OpLabel(fmt::format("label_0x{:x}", address)));
- }
- }
-
- void DeclareVertex() {
- if (stage != ShaderStage::Vertex)
+ void DeclareTessEval() {
+ if (stage != ShaderType::TesselationEval) {
return;
+ }
+ DeclareInputVertexArray(NumInputPatches);
+ DeclareOutputVertex();
- DeclareVertexRedeclarations();
+ tess_coord = DeclareInputBuiltIn(spv::BuiltIn::TessCoord, t_in_float3, "tess_coord");
}
void DeclareGeometry() {
- if (stage != ShaderStage::Geometry)
+ if (stage != ShaderType::Geometry) {
return;
-
- UNIMPLEMENTED();
+ }
+ const u32 num_input = GetNumPrimitiveTopologyVertices(specialization.primitive_topology);
+ DeclareInputVertexArray(num_input);
+ DeclareOutputVertex();
}
void DeclareFragment() {
- if (stage != ShaderStage::Fragment)
+ if (stage != ShaderType::Fragment) {
return;
+ }
for (u32 rt = 0; rt < static_cast<u32>(frag_colors.size()); ++rt) {
- if (!IsRenderTargetUsed(rt)) {
+ if (!specialization.enabled_rendertargets[rt]) {
continue;
}
@@ -295,10 +564,19 @@ private:
interfaces.push_back(frag_depth);
}
- frag_coord = DeclareBuiltIn(spv::BuiltIn::FragCoord, spv::StorageClass::Input, t_in_float4,
- "frag_coord");
- front_facing = DeclareBuiltIn(spv::BuiltIn::FrontFacing, spv::StorageClass::Input,
- t_in_bool, "front_facing");
+ frag_coord = DeclareInputBuiltIn(spv::BuiltIn::FragCoord, t_in_float4, "frag_coord");
+ front_facing = DeclareInputBuiltIn(spv::BuiltIn::FrontFacing, t_in_bool, "front_facing");
+ point_coord = DeclareInputBuiltIn(spv::BuiltIn::PointCoord, t_in_float2, "point_coord");
+ }
+
+ void DeclareCompute() {
+ if (stage != ShaderType::Compute) {
+ return;
+ }
+
+ workgroup_id = DeclareInputBuiltIn(spv::BuiltIn::WorkgroupId, t_in_uint3, "workgroup_id");
+ local_invocation_id =
+ DeclareInputBuiltIn(spv::BuiltIn::LocalInvocationId, t_in_uint3, "local_invocation_id");
}
void DeclareRegisters() {
@@ -326,21 +604,44 @@ private:
}
void DeclareLocalMemory() {
- if (const u64 local_memory_size = header.GetLocalMemorySize(); local_memory_size > 0) {
- const auto element_count = static_cast<u32>(Common::AlignUp(local_memory_size, 4) / 4);
- const Id type_array = TypeArray(t_float, Constant(t_uint, element_count));
- const Id type_pointer = TypePointer(spv::StorageClass::Private, type_array);
- Name(type_pointer, "LocalMemory");
+ // TODO(Rodrigo): Unstub kernel local memory size and pass it from a register at
+ // specialization time.
+ const u64 lmem_size = stage == ShaderType::Compute ? 0x400 : header.GetLocalMemorySize();
+ if (lmem_size == 0) {
+ return;
+ }
+ const auto element_count = static_cast<u32>(Common::AlignUp(lmem_size, 4) / 4);
+ const Id type_array = TypeArray(t_float, Constant(t_uint, element_count));
+ const Id type_pointer = TypePointer(spv::StorageClass::Private, type_array);
+ Name(type_pointer, "LocalMemory");
+
+ local_memory =
+ OpVariable(type_pointer, spv::StorageClass::Private, ConstantNull(type_array));
+ AddGlobalVariable(Name(local_memory, "local_memory"));
+ }
- local_memory =
- OpVariable(type_pointer, spv::StorageClass::Private, ConstantNull(type_array));
- AddGlobalVariable(Name(local_memory, "local_memory"));
+ void DeclareSharedMemory() {
+ if (stage != ShaderType::Compute) {
+ return;
+ }
+ t_smem_uint = TypePointer(spv::StorageClass::Workgroup, t_uint);
+
+ const u32 smem_size = specialization.shared_memory_size;
+ if (smem_size == 0) {
+ // Avoid declaring an empty array.
+ return;
}
+ const auto element_count = static_cast<u32>(Common::AlignUp(smem_size, 4) / 4);
+ const Id type_array = TypeArray(t_uint, Constant(t_uint, element_count));
+ const Id type_pointer = TypePointer(spv::StorageClass::Workgroup, type_array);
+ Name(type_pointer, "SharedMemory");
+
+ shared_memory = OpVariable(type_pointer, spv::StorageClass::Workgroup);
+ AddGlobalVariable(Name(shared_memory, "shared_memory"));
}
void DeclareInternalFlags() {
- constexpr std::array<const char*, INTERNAL_FLAGS_COUNT> names = {"zero", "sign", "carry",
- "overflow"};
+ constexpr std::array names = {"zero", "sign", "carry", "overflow"};
for (std::size_t flag = 0; flag < INTERNAL_FLAGS_COUNT; ++flag) {
const auto flag_code = static_cast<InternalFlag>(flag);
const Id id = OpVariable(t_prv_bool, spv::StorageClass::Private, v_false);
@@ -348,23 +649,59 @@ private:
}
}
+ void DeclareInputVertexArray(u32 length) {
+ constexpr auto storage = spv::StorageClass::Input;
+ std::tie(in_indices, in_vertex) = DeclareVertexArray(storage, "in_indices", length);
+ }
+
+ void DeclareOutputVertexArray(u32 length) {
+ constexpr auto storage = spv::StorageClass::Output;
+ std::tie(out_indices, out_vertex) = DeclareVertexArray(storage, "out_indices", length);
+ }
+
+ std::tuple<VertexIndices, Id> DeclareVertexArray(spv::StorageClass storage_class,
+ std::string name, u32 length) {
+ const auto [struct_id, indices] = DeclareVertexStruct();
+ const Id vertex_array = TypeArray(struct_id, Constant(t_uint, length));
+ const Id vertex_ptr = TypePointer(storage_class, vertex_array);
+ const Id vertex = OpVariable(vertex_ptr, storage_class);
+ AddGlobalVariable(Name(vertex, std::move(name)));
+ interfaces.push_back(vertex);
+ return {indices, vertex};
+ }
+
+ void DeclareOutputVertex() {
+ Id out_vertex_struct;
+ std::tie(out_vertex_struct, out_indices) = DeclareVertexStruct();
+ const Id out_vertex_ptr = TypePointer(spv::StorageClass::Output, out_vertex_struct);
+ out_vertex = OpVariable(out_vertex_ptr, spv::StorageClass::Output);
+ interfaces.push_back(AddGlobalVariable(Name(out_vertex, "out_vertex")));
+ }
+
void DeclareInputAttributes() {
for (const auto index : ir.GetInputAttributes()) {
if (!IsGenericAttribute(index)) {
continue;
}
- UNIMPLEMENTED_IF(stage == ShaderStage::Geometry);
-
const u32 location = GetGenericAttributeLocation(index);
- const Id id = OpVariable(t_in_float4, spv::StorageClass::Input);
- Name(AddGlobalVariable(id), fmt::format("in_attr{}", location));
+ const auto type_descriptor = GetAttributeType(location);
+ Id type;
+ if (IsInputAttributeArray()) {
+ type = GetTypeVectorDefinitionLut(type_descriptor.type).at(3);
+ type = TypeArray(type, Constant(t_uint, GetNumInputVertices()));
+ type = TypePointer(spv::StorageClass::Input, type);
+ } else {
+ type = type_descriptor.vector;
+ }
+ const Id id = OpVariable(type, spv::StorageClass::Input);
+ AddGlobalVariable(Name(id, fmt::format("in_attr{}", location)));
input_attributes.emplace(index, id);
interfaces.push_back(id);
Decorate(id, spv::Decoration::Location, location);
- if (stage != ShaderStage::Fragment) {
+ if (stage != ShaderType::Fragment) {
continue;
}
switch (header.ps.GetAttributeUse(location)) {
@@ -388,8 +725,21 @@ private:
if (!IsGenericAttribute(index)) {
continue;
}
- const auto location = GetGenericAttributeLocation(index);
- const Id id = OpVariable(t_out_float4, spv::StorageClass::Output);
+ const u32 location = GetGenericAttributeLocation(index);
+ Id type = t_float4;
+ Id varying_default = v_varying_default;
+ if (IsOutputAttributeArray()) {
+ const u32 num = GetNumOutputVertices();
+ type = TypeArray(type, Constant(t_uint, num));
+ if (device.GetDriverID() != vk::DriverIdKHR::eIntelProprietaryWindows) {
+ // Intel's proprietary driver fails to setup defaults for arrayed output
+ // attributes.
+ varying_default = ConstantComposite(type, std::vector(num, varying_default));
+ }
+ }
+ type = TypePointer(spv::StorageClass::Output, type);
+
+ const Id id = OpVariable(type, spv::StorageClass::Output, varying_default);
Name(AddGlobalVariable(id), fmt::format("out_attr{}", location));
output_attributes.emplace(index, id);
interfaces.push_back(id);
@@ -398,10 +748,8 @@ private:
}
}
- void DeclareConstantBuffers() {
- u32 binding = const_buffers_base_binding;
- for (const auto& entry : ir.GetConstantBuffers()) {
- const auto [index, size] = entry;
+ u32 DeclareConstantBuffers(u32 binding) {
+ for (const auto& [index, size] : ir.GetConstantBuffers()) {
const Id type = device.IsKhrUniformBufferStandardLayoutSupported() ? t_cbuf_scalar_ubo
: t_cbuf_std140_ubo;
const Id id = OpVariable(type, spv::StorageClass::Uniform);
@@ -411,12 +759,11 @@ private:
Decorate(id, spv::Decoration::DescriptorSet, DESCRIPTOR_SET);
constant_buffers.emplace(index, id);
}
+ return binding;
}
- void DeclareGlobalBuffers() {
- u32 binding = global_buffers_base_binding;
- for (const auto& entry : ir.GetGlobalMemory()) {
- const auto [base, usage] = entry;
+ u32 DeclareGlobalBuffers(u32 binding) {
+ for (const auto& [base, usage] : ir.GetGlobalMemory()) {
const Id id = OpVariable(t_gmem_ssbo, spv::StorageClass::StorageBuffer);
AddGlobalVariable(
Name(id, fmt::format("gmem_{}_{}", base.cbuf_index, base.cbuf_offset)));
@@ -425,108 +772,213 @@ private:
Decorate(id, spv::Decoration::DescriptorSet, DESCRIPTOR_SET);
global_buffers.emplace(base, id);
}
+ return binding;
}
- void DeclareSamplers() {
- u32 binding = samplers_base_binding;
+ u32 DeclareTexelBuffers(u32 binding) {
for (const auto& sampler : ir.GetSamplers()) {
+ if (!sampler.IsBuffer()) {
+ continue;
+ }
+ ASSERT(!sampler.IsArray());
+ ASSERT(!sampler.IsShadow());
+
+ constexpr auto dim = spv::Dim::Buffer;
+ constexpr int depth = 0;
+ constexpr int arrayed = 0;
+ constexpr bool ms = false;
+ constexpr int sampled = 1;
+ constexpr auto format = spv::ImageFormat::Unknown;
+ const Id image_type = TypeImage(t_float, dim, depth, arrayed, ms, sampled, format);
+ const Id pointer_type = TypePointer(spv::StorageClass::UniformConstant, image_type);
+ const Id id = OpVariable(pointer_type, spv::StorageClass::UniformConstant);
+ AddGlobalVariable(Name(id, fmt::format("sampler_{}", sampler.GetIndex())));
+ Decorate(id, spv::Decoration::Binding, binding++);
+ Decorate(id, spv::Decoration::DescriptorSet, DESCRIPTOR_SET);
+
+ texel_buffers.emplace(sampler.GetIndex(), TexelBuffer{image_type, id});
+ }
+ return binding;
+ }
+
+ u32 DeclareSamplers(u32 binding) {
+ for (const auto& sampler : ir.GetSamplers()) {
+ if (sampler.IsBuffer()) {
+ continue;
+ }
const auto dim = GetSamplerDim(sampler);
const int depth = sampler.IsShadow() ? 1 : 0;
const int arrayed = sampler.IsArray() ? 1 : 0;
- // TODO(Rodrigo): Sampled 1 indicates that the image will be used with a sampler. When
- // SULD and SUST instructions are implemented, replace this value.
- const int sampled = 1;
- const Id image_type =
- TypeImage(t_float, dim, depth, arrayed, false, sampled, spv::ImageFormat::Unknown);
+ constexpr bool ms = false;
+ constexpr int sampled = 1;
+ constexpr auto format = spv::ImageFormat::Unknown;
+ const Id image_type = TypeImage(t_float, dim, depth, arrayed, ms, sampled, format);
const Id sampled_image_type = TypeSampledImage(image_type);
const Id pointer_type =
TypePointer(spv::StorageClass::UniformConstant, sampled_image_type);
const Id id = OpVariable(pointer_type, spv::StorageClass::UniformConstant);
AddGlobalVariable(Name(id, fmt::format("sampler_{}", sampler.GetIndex())));
+ Decorate(id, spv::Decoration::Binding, binding++);
+ Decorate(id, spv::Decoration::DescriptorSet, DESCRIPTOR_SET);
- sampler_images.insert(
- {static_cast<u32>(sampler.GetIndex()), {image_type, sampled_image_type, id}});
+ sampled_images.emplace(sampler.GetIndex(),
+ SampledImage{image_type, sampled_image_type, id});
+ }
+ return binding;
+ }
+
+ u32 DeclareImages(u32 binding) {
+ for (const auto& image : ir.GetImages()) {
+ const auto [dim, arrayed] = GetImageDim(image);
+ constexpr int depth = 0;
+ constexpr bool ms = false;
+ constexpr int sampled = 2; // This won't be accessed with a sampler
+ constexpr auto format = spv::ImageFormat::Unknown;
+ const Id image_type = TypeImage(t_uint, dim, depth, arrayed, ms, sampled, format, {});
+ const Id pointer_type = TypePointer(spv::StorageClass::UniformConstant, image_type);
+ const Id id = OpVariable(pointer_type, spv::StorageClass::UniformConstant);
+ AddGlobalVariable(Name(id, fmt::format("image_{}", image.GetIndex())));
Decorate(id, spv::Decoration::Binding, binding++);
Decorate(id, spv::Decoration::DescriptorSet, DESCRIPTOR_SET);
+ if (image.IsRead() && !image.IsWritten()) {
+ Decorate(id, spv::Decoration::NonWritable);
+ } else if (image.IsWritten() && !image.IsRead()) {
+ Decorate(id, spv::Decoration::NonReadable);
+ }
+
+ images.emplace(static_cast<u32>(image.GetIndex()), StorageImage{image_type, id});
}
+ return binding;
}
- void DeclareVertexRedeclarations() {
- vertex_index = DeclareBuiltIn(spv::BuiltIn::VertexIndex, spv::StorageClass::Input,
- t_in_uint, "vertex_index");
- instance_index = DeclareBuiltIn(spv::BuiltIn::InstanceIndex, spv::StorageClass::Input,
- t_in_uint, "instance_index");
+ bool IsInputAttributeArray() const {
+ return stage == ShaderType::TesselationControl || stage == ShaderType::TesselationEval ||
+ stage == ShaderType::Geometry;
+ }
- bool is_clip_distances_declared = false;
- for (const auto index : ir.GetOutputAttributes()) {
- if (index == Attribute::Index::ClipDistances0123 ||
- index == Attribute::Index::ClipDistances4567) {
- is_clip_distances_declared = true;
- }
+ bool IsOutputAttributeArray() const {
+ return stage == ShaderType::TesselationControl;
+ }
+
+ u32 GetNumInputVertices() const {
+ switch (stage) {
+ case ShaderType::Geometry:
+ return GetNumPrimitiveTopologyVertices(specialization.primitive_topology);
+ case ShaderType::TesselationControl:
+ case ShaderType::TesselationEval:
+ return NumInputPatches;
+ default:
+ UNREACHABLE();
+ return 1;
}
+ }
- std::vector<Id> members;
- members.push_back(t_float4);
- if (ir.UsesPointSize()) {
- members.push_back(t_float);
- }
- if (is_clip_distances_declared) {
- members.push_back(TypeArray(t_float, Constant(t_uint, 8)));
- }
-
- const Id gl_per_vertex_struct = Name(TypeStruct(members), "PerVertex");
- Decorate(gl_per_vertex_struct, spv::Decoration::Block);
-
- u32 declaration_index = 0;
- const auto MemberDecorateBuiltIn = [&](spv::BuiltIn builtin, std::string name,
- bool condition) {
- if (!condition)
- return u32{};
- MemberName(gl_per_vertex_struct, declaration_index, name);
- MemberDecorate(gl_per_vertex_struct, declaration_index, spv::Decoration::BuiltIn,
- static_cast<u32>(builtin));
- return declaration_index++;
+ u32 GetNumOutputVertices() const {
+ switch (stage) {
+ case ShaderType::TesselationControl:
+ return header.common2.threads_per_input_primitive;
+ default:
+ UNREACHABLE();
+ return 1;
+ }
+ }
+
+ std::tuple<Id, VertexIndices> DeclareVertexStruct() {
+ struct BuiltIn {
+ Id type;
+ spv::BuiltIn builtin;
+ const char* name;
+ };
+ std::vector<BuiltIn> members;
+ members.reserve(4);
+
+ const auto AddBuiltIn = [&](Id type, spv::BuiltIn builtin, const char* name) {
+ const auto index = static_cast<u32>(members.size());
+ members.push_back(BuiltIn{type, builtin, name});
+ return index;
};
- position_index = MemberDecorateBuiltIn(spv::BuiltIn::Position, "position", true);
- point_size_index =
- MemberDecorateBuiltIn(spv::BuiltIn::PointSize, "point_size", ir.UsesPointSize());
- clip_distances_index = MemberDecorateBuiltIn(spv::BuiltIn::ClipDistance, "clip_distances",
- is_clip_distances_declared);
+ VertexIndices indices;
+ indices.position = AddBuiltIn(t_float4, spv::BuiltIn::Position, "position");
- const Id type_pointer = TypePointer(spv::StorageClass::Output, gl_per_vertex_struct);
- per_vertex = OpVariable(type_pointer, spv::StorageClass::Output);
- AddGlobalVariable(Name(per_vertex, "per_vertex"));
- interfaces.push_back(per_vertex);
+ if (ir.UsesViewportIndex()) {
+ if (stage != ShaderType::Vertex || device.IsExtShaderViewportIndexLayerSupported()) {
+ indices.viewport = AddBuiltIn(t_int, spv::BuiltIn::ViewportIndex, "viewport_index");
+ } else {
+ LOG_ERROR(Render_Vulkan,
+ "Shader requires ViewportIndex but it's not supported on this "
+ "stage with this device.");
+ }
+ }
+
+ if (ir.UsesPointSize() || specialization.point_size) {
+ indices.point_size = AddBuiltIn(t_float, spv::BuiltIn::PointSize, "point_size");
+ }
+
+ const auto& output_attributes = ir.GetOutputAttributes();
+ const bool declare_clip_distances =
+ std::any_of(output_attributes.begin(), output_attributes.end(), [](const auto& index) {
+ return index == Attribute::Index::ClipDistances0123 ||
+ index == Attribute::Index::ClipDistances4567;
+ });
+ if (declare_clip_distances) {
+ indices.clip_distances = AddBuiltIn(TypeArray(t_float, Constant(t_uint, 8)),
+ spv::BuiltIn::ClipDistance, "clip_distances");
+ }
+
+ std::vector<Id> member_types;
+ member_types.reserve(members.size());
+ for (std::size_t i = 0; i < members.size(); ++i) {
+ member_types.push_back(members[i].type);
+ }
+ const Id per_vertex_struct = Name(TypeStruct(member_types), "PerVertex");
+ Decorate(per_vertex_struct, spv::Decoration::Block);
+
+ for (std::size_t index = 0; index < members.size(); ++index) {
+ const auto& member = members[index];
+ MemberName(per_vertex_struct, static_cast<u32>(index), member.name);
+ MemberDecorate(per_vertex_struct, static_cast<u32>(index), spv::Decoration::BuiltIn,
+ static_cast<u32>(member.builtin));
+ }
+
+ return {per_vertex_struct, indices};
}
void VisitBasicBlock(const NodeBlock& bb) {
for (const auto& node : bb) {
- static_cast<void>(Visit(node));
+ [[maybe_unused]] const Type type = Visit(node).type;
+ ASSERT(type == Type::Void);
}
}
- Id Visit(const Node& node) {
+ Expression Visit(const Node& node) {
if (const auto operation = std::get_if<OperationNode>(&*node)) {
+ if (const auto amend_index = operation->GetAmendIndex()) {
+ [[maybe_unused]] const Type type = Visit(ir.GetAmendNode(*amend_index)).type;
+ ASSERT(type == Type::Void);
+ }
const auto operation_index = static_cast<std::size_t>(operation->GetCode());
const auto decompiler = operation_decompilers[operation_index];
if (decompiler == nullptr) {
UNREACHABLE_MSG("Operation decompiler {} not defined", operation_index);
}
return (this->*decompiler)(*operation);
+ }
- } else if (const auto gpr = std::get_if<GprNode>(&*node)) {
+ if (const auto gpr = std::get_if<GprNode>(&*node)) {
const u32 index = gpr->GetIndex();
if (index == Register::ZeroIndex) {
- return Constant(t_float, 0.0f);
+ return {v_float_zero, Type::Float};
}
- return Emit(OpLoad(t_float, registers.at(index)));
+ return {OpLoad(t_float, registers.at(index)), Type::Float};
+ }
- } else if (const auto immediate = std::get_if<ImmediateNode>(&*node)) {
- return BitcastTo<Type::Float>(Constant(t_uint, immediate->GetValue()));
+ if (const auto immediate = std::get_if<ImmediateNode>(&*node)) {
+ return {Constant(t_uint, immediate->GetValue()), Type::Uint};
+ }
- } else if (const auto predicate = std::get_if<PredicateNode>(&*node)) {
+ if (const auto predicate = std::get_if<PredicateNode>(&*node)) {
const auto value = [&]() -> Id {
switch (const auto index = predicate->GetIndex(); index) {
case Tegra::Shader::Pred::UnusedIndex:
@@ -534,74 +986,108 @@ private:
case Tegra::Shader::Pred::NeverExecute:
return v_false;
default:
- return Emit(OpLoad(t_bool, predicates.at(index)));
+ return OpLoad(t_bool, predicates.at(index));
}
}();
if (predicate->IsNegated()) {
- return Emit(OpLogicalNot(t_bool, value));
+ return {OpLogicalNot(t_bool, value), Type::Bool};
}
- return value;
+ return {value, Type::Bool};
+ }
- } else if (const auto abuf = std::get_if<AbufNode>(&*node)) {
+ if (const auto abuf = std::get_if<AbufNode>(&*node)) {
const auto attribute = abuf->GetIndex();
- const auto element = abuf->GetElement();
+ const u32 element = abuf->GetElement();
+ const auto& buffer = abuf->GetBuffer();
+
+ const auto ArrayPass = [&](Id pointer_type, Id composite, std::vector<u32> indices) {
+ std::vector<Id> members;
+ members.reserve(std::size(indices) + 1);
+
+ if (buffer && IsInputAttributeArray()) {
+ members.push_back(AsUint(Visit(buffer)));
+ }
+ for (const u32 index : indices) {
+ members.push_back(Constant(t_uint, index));
+ }
+ return OpAccessChain(pointer_type, composite, members);
+ };
switch (attribute) {
- case Attribute::Index::Position:
- if (stage != ShaderStage::Fragment) {
- UNIMPLEMENTED();
- break;
- } else {
+ case Attribute::Index::Position: {
+ if (stage == ShaderType::Fragment) {
if (element == 3) {
- return Constant(t_float, 1.0f);
+ return {Constant(t_float, 1.0f), Type::Float};
}
- return Emit(OpLoad(t_float, AccessElement(t_in_float, frag_coord, element)));
+ return {OpLoad(t_float, AccessElement(t_in_float, frag_coord, element)),
+ Type::Float};
}
+ const std::vector elements = {in_indices.position.value(), element};
+ return {OpLoad(t_float, ArrayPass(t_in_float, in_vertex, elements)), Type::Float};
+ }
+ case Attribute::Index::PointCoord: {
+ switch (element) {
+ case 0:
+ case 1:
+ return {OpCompositeExtract(t_float, OpLoad(t_float2, point_coord), element),
+ Type::Float};
+ }
+ UNIMPLEMENTED_MSG("Unimplemented point coord element={}", element);
+ return {v_float_zero, Type::Float};
+ }
case Attribute::Index::TessCoordInstanceIDVertexID:
// TODO(Subv): Find out what the values are for the first two elements when inside a
// vertex shader, and what's the value of the fourth element when inside a Tess Eval
// shader.
- ASSERT(stage == ShaderStage::Vertex);
switch (element) {
+ case 0:
+ case 1:
+ return {OpLoad(t_float, AccessElement(t_in_float, tess_coord, element)),
+ Type::Float};
case 2:
- return BitcastFrom<Type::Uint>(Emit(OpLoad(t_uint, instance_index)));
+ return {OpLoad(t_uint, instance_index), Type::Uint};
case 3:
- return BitcastFrom<Type::Uint>(Emit(OpLoad(t_uint, vertex_index)));
+ return {OpLoad(t_uint, vertex_index), Type::Uint};
}
UNIMPLEMENTED_MSG("Unmanaged TessCoordInstanceIDVertexID element={}", element);
- return Constant(t_float, 0);
+ return {Constant(t_uint, 0U), Type::Uint};
case Attribute::Index::FrontFacing:
// TODO(Subv): Find out what the values are for the other elements.
- ASSERT(stage == ShaderStage::Fragment);
+ ASSERT(stage == ShaderType::Fragment);
if (element == 3) {
- const Id is_front_facing = Emit(OpLoad(t_bool, front_facing));
- const Id true_value =
- BitcastTo<Type::Float>(Constant(t_int, static_cast<s32>(-1)));
- const Id false_value = BitcastTo<Type::Float>(Constant(t_int, 0));
- return Emit(OpSelect(t_float, is_front_facing, true_value, false_value));
+ const Id is_front_facing = OpLoad(t_bool, front_facing);
+ const Id true_value = Constant(t_int, static_cast<s32>(-1));
+ const Id false_value = Constant(t_int, 0);
+ return {OpSelect(t_int, is_front_facing, true_value, false_value), Type::Int};
}
UNIMPLEMENTED_MSG("Unmanaged FrontFacing element={}", element);
- return Constant(t_float, 0.0f);
+ return {v_float_zero, Type::Float};
default:
if (IsGenericAttribute(attribute)) {
- const Id pointer =
- AccessElement(t_in_float, input_attributes.at(attribute), element);
- return Emit(OpLoad(t_float, pointer));
+ const u32 location = GetGenericAttributeLocation(attribute);
+ const auto type_descriptor = GetAttributeType(location);
+ const Type type = type_descriptor.type;
+ const Id attribute_id = input_attributes.at(attribute);
+ const std::vector elements = {element};
+ const Id pointer = ArrayPass(type_descriptor.scalar, attribute_id, elements);
+ return {OpLoad(GetTypeDefinition(type), pointer), type};
}
break;
}
UNIMPLEMENTED_MSG("Unhandled input attribute: {}", static_cast<u32>(attribute));
+ return {v_float_zero, Type::Float};
+ }
- } else if (const auto cbuf = std::get_if<CbufNode>(&*node)) {
+ if (const auto cbuf = std::get_if<CbufNode>(&*node)) {
const Node& offset = cbuf->GetOffset();
const Id buffer_id = constant_buffers.at(cbuf->GetIndex());
Id pointer{};
if (device.IsKhrUniformBufferStandardLayoutSupported()) {
- const Id buffer_offset = Emit(OpShiftRightLogical(
- t_uint, BitcastTo<Type::Uint>(Visit(offset)), Constant(t_uint, 2u)));
- pointer = Emit(
- OpAccessChain(t_cbuf_float, buffer_id, Constant(t_uint, 0u), buffer_offset));
+ const Id buffer_offset =
+ OpShiftRightLogical(t_uint, AsUint(Visit(offset)), Constant(t_uint, 2U));
+ pointer =
+ OpAccessChain(t_cbuf_float, buffer_id, Constant(t_uint, 0U), buffer_offset);
} else {
Id buffer_index{};
Id buffer_element{};
@@ -613,53 +1099,80 @@ private:
buffer_element = Constant(t_uint, (offset_imm / 4) % 4);
} else if (std::holds_alternative<OperationNode>(*offset)) {
// Indirect access
- const Id offset_id = BitcastTo<Type::Uint>(Visit(offset));
- const Id unsafe_offset = Emit(OpUDiv(t_uint, offset_id, Constant(t_uint, 4)));
- const Id final_offset = Emit(OpUMod(
- t_uint, unsafe_offset, Constant(t_uint, MAX_CONSTBUFFER_ELEMENTS - 1)));
- buffer_index = Emit(OpUDiv(t_uint, final_offset, Constant(t_uint, 4)));
- buffer_element = Emit(OpUMod(t_uint, final_offset, Constant(t_uint, 4)));
+ const Id offset_id = AsUint(Visit(offset));
+ const Id unsafe_offset = OpUDiv(t_uint, offset_id, Constant(t_uint, 4));
+ const Id final_offset =
+ OpUMod(t_uint, unsafe_offset, Constant(t_uint, MaxConstBufferElements - 1));
+ buffer_index = OpUDiv(t_uint, final_offset, Constant(t_uint, 4));
+ buffer_element = OpUMod(t_uint, final_offset, Constant(t_uint, 4));
} else {
UNREACHABLE_MSG("Unmanaged offset node type");
}
- pointer = Emit(OpAccessChain(t_cbuf_float, buffer_id, Constant(t_uint, 0),
- buffer_index, buffer_element));
+ pointer = OpAccessChain(t_cbuf_float, buffer_id, Constant(t_uint, 0), buffer_index,
+ buffer_element);
}
- return Emit(OpLoad(t_float, pointer));
+ return {OpLoad(t_float, pointer), Type::Float};
+ }
- } else if (const auto gmem = std::get_if<GmemNode>(&*node)) {
+ if (const auto gmem = std::get_if<GmemNode>(&*node)) {
const Id gmem_buffer = global_buffers.at(gmem->GetDescriptor());
- const Id real = BitcastTo<Type::Uint>(Visit(gmem->GetRealAddress()));
- const Id base = BitcastTo<Type::Uint>(Visit(gmem->GetBaseAddress()));
+ const Id real = AsUint(Visit(gmem->GetRealAddress()));
+ const Id base = AsUint(Visit(gmem->GetBaseAddress()));
+
+ Id offset = OpISub(t_uint, real, base);
+ offset = OpUDiv(t_uint, offset, Constant(t_uint, 4U));
+ return {OpLoad(t_float,
+ OpAccessChain(t_gmem_float, gmem_buffer, Constant(t_uint, 0U), offset)),
+ Type::Float};
+ }
+
+ if (const auto lmem = std::get_if<LmemNode>(&*node)) {
+ Id address = AsUint(Visit(lmem->GetAddress()));
+ address = OpShiftRightLogical(t_uint, address, Constant(t_uint, 2U));
+ const Id pointer = OpAccessChain(t_prv_float, local_memory, address);
+ return {OpLoad(t_float, pointer), Type::Float};
+ }
+
+ if (const auto smem = std::get_if<SmemNode>(&*node)) {
+ Id address = AsUint(Visit(smem->GetAddress()));
+ address = OpShiftRightLogical(t_uint, address, Constant(t_uint, 2U));
+ const Id pointer = OpAccessChain(t_smem_uint, shared_memory, address);
+ return {OpLoad(t_uint, pointer), Type::Uint};
+ }
- Id offset = Emit(OpISub(t_uint, real, base));
- offset = Emit(OpUDiv(t_uint, offset, Constant(t_uint, 4u)));
- return Emit(OpLoad(t_float, Emit(OpAccessChain(t_gmem_float, gmem_buffer,
- Constant(t_uint, 0u), offset))));
+ if (const auto internal_flag = std::get_if<InternalFlagNode>(&*node)) {
+ const Id flag = internal_flags.at(static_cast<std::size_t>(internal_flag->GetFlag()));
+ return {OpLoad(t_bool, flag), Type::Bool};
+ }
- } else if (const auto conditional = std::get_if<ConditionalNode>(&*node)) {
+ if (const auto conditional = std::get_if<ConditionalNode>(&*node)) {
+ if (const auto amend_index = conditional->GetAmendIndex()) {
+ [[maybe_unused]] const Type type = Visit(ir.GetAmendNode(*amend_index)).type;
+ ASSERT(type == Type::Void);
+ }
// It's invalid to call conditional on nested nodes, use an operation instead
const Id true_label = OpLabel();
const Id skip_label = OpLabel();
- const Id condition = Visit(conditional->GetCondition());
- Emit(OpSelectionMerge(skip_label, spv::SelectionControlMask::MaskNone));
- Emit(OpBranchConditional(condition, true_label, skip_label));
- Emit(true_label);
+ const Id condition = AsBool(Visit(conditional->GetCondition()));
+ OpSelectionMerge(skip_label, spv::SelectionControlMask::MaskNone);
+ OpBranchConditional(condition, true_label, skip_label);
+ AddLabel(true_label);
- ++conditional_nest_count;
+ conditional_branch_set = true;
+ inside_branch = false;
VisitBasicBlock(conditional->GetCode());
- --conditional_nest_count;
-
- if (inside_branch == 0) {
- Emit(OpBranch(skip_label));
+ conditional_branch_set = false;
+ if (!inside_branch) {
+ OpBranch(skip_label);
} else {
- inside_branch--;
+ inside_branch = false;
}
- Emit(skip_label);
+ AddLabel(skip_label);
return {};
+ }
- } else if (const auto comment = std::get_if<CommentNode>(&*node)) {
- Name(Emit(OpUndef(t_void)), comment->GetText());
+ if (const auto comment = std::get_if<CommentNode>(&*node)) {
+ Name(OpUndef(t_void), comment->GetText());
return {};
}
@@ -668,94 +1181,126 @@ private:
}
template <Id (Module::*func)(Id, Id), Type result_type, Type type_a = result_type>
- Id Unary(Operation operation) {
+ Expression Unary(Operation operation) {
const Id type_def = GetTypeDefinition(result_type);
- const Id op_a = VisitOperand<type_a>(operation, 0);
+ const Id op_a = As(Visit(operation[0]), type_a);
- const Id value = BitcastFrom<result_type>(Emit((this->*func)(type_def, op_a)));
+ const Id value = (this->*func)(type_def, op_a);
if (IsPrecise(operation)) {
Decorate(value, spv::Decoration::NoContraction);
}
- return value;
+ return {value, result_type};
}
template <Id (Module::*func)(Id, Id, Id), Type result_type, Type type_a = result_type,
Type type_b = type_a>
- Id Binary(Operation operation) {
+ Expression Binary(Operation operation) {
const Id type_def = GetTypeDefinition(result_type);
- const Id op_a = VisitOperand<type_a>(operation, 0);
- const Id op_b = VisitOperand<type_b>(operation, 1);
+ const Id op_a = As(Visit(operation[0]), type_a);
+ const Id op_b = As(Visit(operation[1]), type_b);
- const Id value = BitcastFrom<result_type>(Emit((this->*func)(type_def, op_a, op_b)));
+ const Id value = (this->*func)(type_def, op_a, op_b);
if (IsPrecise(operation)) {
Decorate(value, spv::Decoration::NoContraction);
}
- return value;
+ return {value, result_type};
}
template <Id (Module::*func)(Id, Id, Id, Id), Type result_type, Type type_a = result_type,
Type type_b = type_a, Type type_c = type_b>
- Id Ternary(Operation operation) {
+ Expression Ternary(Operation operation) {
const Id type_def = GetTypeDefinition(result_type);
- const Id op_a = VisitOperand<type_a>(operation, 0);
- const Id op_b = VisitOperand<type_b>(operation, 1);
- const Id op_c = VisitOperand<type_c>(operation, 2);
+ const Id op_a = As(Visit(operation[0]), type_a);
+ const Id op_b = As(Visit(operation[1]), type_b);
+ const Id op_c = As(Visit(operation[2]), type_c);
- const Id value = BitcastFrom<result_type>(Emit((this->*func)(type_def, op_a, op_b, op_c)));
+ const Id value = (this->*func)(type_def, op_a, op_b, op_c);
if (IsPrecise(operation)) {
Decorate(value, spv::Decoration::NoContraction);
}
- return value;
+ return {value, result_type};
}
template <Id (Module::*func)(Id, Id, Id, Id, Id), Type result_type, Type type_a = result_type,
Type type_b = type_a, Type type_c = type_b, Type type_d = type_c>
- Id Quaternary(Operation operation) {
+ Expression Quaternary(Operation operation) {
const Id type_def = GetTypeDefinition(result_type);
- const Id op_a = VisitOperand<type_a>(operation, 0);
- const Id op_b = VisitOperand<type_b>(operation, 1);
- const Id op_c = VisitOperand<type_c>(operation, 2);
- const Id op_d = VisitOperand<type_d>(operation, 3);
+ const Id op_a = As(Visit(operation[0]), type_a);
+ const Id op_b = As(Visit(operation[1]), type_b);
+ const Id op_c = As(Visit(operation[2]), type_c);
+ const Id op_d = As(Visit(operation[3]), type_d);
- const Id value =
- BitcastFrom<result_type>(Emit((this->*func)(type_def, op_a, op_b, op_c, op_d)));
+ const Id value = (this->*func)(type_def, op_a, op_b, op_c, op_d);
if (IsPrecise(operation)) {
Decorate(value, spv::Decoration::NoContraction);
}
- return value;
+ return {value, result_type};
}
- Id Assign(Operation operation) {
+ Expression Assign(Operation operation) {
const Node& dest = operation[0];
const Node& src = operation[1];
- Id target{};
+ Expression target{};
if (const auto gpr = std::get_if<GprNode>(&*dest)) {
if (gpr->GetIndex() == Register::ZeroIndex) {
// Writing to Register::ZeroIndex is a no op
return {};
}
- target = registers.at(gpr->GetIndex());
+ target = {registers.at(gpr->GetIndex()), Type::Float};
} else if (const auto abuf = std::get_if<AbufNode>(&*dest)) {
- target = [&]() -> Id {
+ const auto& buffer = abuf->GetBuffer();
+ const auto ArrayPass = [&](Id pointer_type, Id composite, std::vector<u32> indices) {
+ std::vector<Id> members;
+ members.reserve(std::size(indices) + 1);
+
+ if (buffer && IsOutputAttributeArray()) {
+ members.push_back(AsUint(Visit(buffer)));
+ }
+ for (const u32 index : indices) {
+ members.push_back(Constant(t_uint, index));
+ }
+ return OpAccessChain(pointer_type, composite, members);
+ };
+
+ target = [&]() -> Expression {
+ const u32 element = abuf->GetElement();
switch (const auto attribute = abuf->GetIndex(); attribute) {
- case Attribute::Index::Position:
- return AccessElement(t_out_float, per_vertex, position_index,
- abuf->GetElement());
+ case Attribute::Index::Position: {
+ const u32 index = out_indices.position.value();
+ return {ArrayPass(t_out_float, out_vertex, {index, element}), Type::Float};
+ }
case Attribute::Index::LayerViewportPointSize:
- UNIMPLEMENTED_IF(abuf->GetElement() != 3);
- return AccessElement(t_out_float, per_vertex, point_size_index);
- case Attribute::Index::ClipDistances0123:
- return AccessElement(t_out_float, per_vertex, clip_distances_index,
- abuf->GetElement());
- case Attribute::Index::ClipDistances4567:
- return AccessElement(t_out_float, per_vertex, clip_distances_index,
- abuf->GetElement() + 4);
+ switch (element) {
+ case 2: {
+ if (!out_indices.viewport) {
+ return {};
+ }
+ const u32 index = out_indices.viewport.value();
+ return {AccessElement(t_out_int, out_vertex, index), Type::Int};
+ }
+ case 3: {
+ const auto index = out_indices.point_size.value();
+ return {AccessElement(t_out_float, out_vertex, index), Type::Float};
+ }
+ default:
+ UNIMPLEMENTED_MSG("LayerViewportPoint element={}", abuf->GetElement());
+ return {};
+ }
+ case Attribute::Index::ClipDistances0123: {
+ const u32 index = out_indices.clip_distances.value();
+ return {AccessElement(t_out_float, out_vertex, index, element), Type::Float};
+ }
+ case Attribute::Index::ClipDistances4567: {
+ const u32 index = out_indices.clip_distances.value();
+ return {AccessElement(t_out_float, out_vertex, index, element + 4),
+ Type::Float};
+ }
default:
if (IsGenericAttribute(attribute)) {
- return AccessElement(t_out_float, output_attributes.at(attribute),
- abuf->GetElement());
+ const Id composite = output_attributes.at(attribute);
+ return {ArrayPass(t_out_float, composite, {element}), Type::Float};
}
UNIMPLEMENTED_MSG("Unhandled output attribute: {}",
static_cast<u32>(attribute));
@@ -763,67 +1308,154 @@ private:
}
}();
+ } else if (const auto patch = std::get_if<PatchNode>(&*dest)) {
+ target = [&]() -> Expression {
+ const u32 offset = patch->GetOffset();
+ switch (offset) {
+ case 0:
+ case 1:
+ case 2:
+ case 3:
+ return {AccessElement(t_out_float, tess_level_outer, offset % 4), Type::Float};
+ case 4:
+ case 5:
+ return {AccessElement(t_out_float, tess_level_inner, offset % 4), Type::Float};
+ }
+ UNIMPLEMENTED_MSG("Unhandled patch output offset: {}", offset);
+ return {};
+ }();
+
} else if (const auto lmem = std::get_if<LmemNode>(&*dest)) {
- Id address = BitcastTo<Type::Uint>(Visit(lmem->GetAddress()));
- address = Emit(OpUDiv(t_uint, address, Constant(t_uint, 4)));
- target = Emit(OpAccessChain(t_prv_float, local_memory, {address}));
+ Id address = AsUint(Visit(lmem->GetAddress()));
+ address = OpUDiv(t_uint, address, Constant(t_uint, 4));
+ target = {OpAccessChain(t_prv_float, local_memory, address), Type::Float};
+
+ } else if (const auto smem = std::get_if<SmemNode>(&*dest)) {
+ ASSERT(stage == ShaderType::Compute);
+ Id address = AsUint(Visit(smem->GetAddress()));
+ address = OpShiftRightLogical(t_uint, address, Constant(t_uint, 2U));
+ target = {OpAccessChain(t_smem_uint, shared_memory, address), Type::Uint};
+
+ } else if (const auto gmem = std::get_if<GmemNode>(&*dest)) {
+ const Id real = AsUint(Visit(gmem->GetRealAddress()));
+ const Id base = AsUint(Visit(gmem->GetBaseAddress()));
+ const Id diff = OpISub(t_uint, real, base);
+ const Id offset = OpShiftRightLogical(t_uint, diff, Constant(t_uint, 2));
+
+ const Id gmem_buffer = global_buffers.at(gmem->GetDescriptor());
+ target = {OpAccessChain(t_gmem_float, gmem_buffer, Constant(t_uint, 0), offset),
+ Type::Float};
+
+ } else {
+ UNIMPLEMENTED();
}
- Emit(OpStore(target, Visit(src)));
+ OpStore(target.id, As(Visit(src), target.type));
return {};
}
- Id FCastHalf0(Operation operation) {
- UNIMPLEMENTED();
- return {};
+ template <u32 offset>
+ Expression FCastHalf(Operation operation) {
+ const Id value = AsHalfFloat(Visit(operation[0]));
+ return {GetFloatFromHalfScalar(OpCompositeExtract(t_scalar_half, value, offset)),
+ Type::Float};
}
- Id FCastHalf1(Operation operation) {
- UNIMPLEMENTED();
- return {};
- }
+ Expression FSwizzleAdd(Operation operation) {
+ const Id minus = Constant(t_float, -1.0f);
+ const Id plus = v_float_one;
+ const Id zero = v_float_zero;
+ const Id lut_a = ConstantComposite(t_float4, minus, plus, minus, zero);
+ const Id lut_b = ConstantComposite(t_float4, minus, minus, plus, minus);
- Id HNegate(Operation operation) {
- UNIMPLEMENTED();
- return {};
+ Id mask = OpLoad(t_uint, thread_id);
+ mask = OpBitwiseAnd(t_uint, mask, Constant(t_uint, 3));
+ mask = OpShiftLeftLogical(t_uint, mask, Constant(t_uint, 1));
+ mask = OpShiftRightLogical(t_uint, AsUint(Visit(operation[2])), mask);
+ mask = OpBitwiseAnd(t_uint, mask, Constant(t_uint, 3));
+
+ const Id modifier_a = OpVectorExtractDynamic(t_float, lut_a, mask);
+ const Id modifier_b = OpVectorExtractDynamic(t_float, lut_b, mask);
+
+ const Id op_a = OpFMul(t_float, AsFloat(Visit(operation[0])), modifier_a);
+ const Id op_b = OpFMul(t_float, AsFloat(Visit(operation[1])), modifier_b);
+ return {OpFAdd(t_float, op_a, op_b), Type::Float};
}
- Id HClamp(Operation operation) {
- UNIMPLEMENTED();
- return {};
+ Expression HNegate(Operation operation) {
+ const bool is_f16 = device.IsFloat16Supported();
+ const Id minus_one = Constant(t_scalar_half, is_f16 ? 0xbc00 : 0xbf800000);
+ const Id one = Constant(t_scalar_half, is_f16 ? 0x3c00 : 0x3f800000);
+ const auto GetNegate = [&](std::size_t index) {
+ return OpSelect(t_scalar_half, AsBool(Visit(operation[index])), minus_one, one);
+ };
+ const Id negation = OpCompositeConstruct(t_half, GetNegate(1), GetNegate(2));
+ return {OpFMul(t_half, AsHalfFloat(Visit(operation[0])), negation), Type::HalfFloat};
}
- Id HCastFloat(Operation operation) {
- UNIMPLEMENTED();
- return {};
+ Expression HClamp(Operation operation) {
+ const auto Pack = [&](std::size_t index) {
+ const Id scalar = GetHalfScalarFromFloat(AsFloat(Visit(operation[index])));
+ return OpCompositeConstruct(t_half, scalar, scalar);
+ };
+ const Id value = AsHalfFloat(Visit(operation[0]));
+ const Id min = Pack(1);
+ const Id max = Pack(2);
+
+ const Id clamped = OpFClamp(t_half, value, min, max);
+ if (IsPrecise(operation)) {
+ Decorate(clamped, spv::Decoration::NoContraction);
+ }
+ return {clamped, Type::HalfFloat};
}
- Id HUnpack(Operation operation) {
- UNIMPLEMENTED();
- return {};
+ Expression HCastFloat(Operation operation) {
+ const Id value = GetHalfScalarFromFloat(AsFloat(Visit(operation[0])));
+ return {OpCompositeConstruct(t_half, value, Constant(t_scalar_half, 0)), Type::HalfFloat};
}
- Id HMergeF32(Operation operation) {
- UNIMPLEMENTED();
- return {};
+ Expression HUnpack(Operation operation) {
+ Expression operand = Visit(operation[0]);
+ const auto type = std::get<Tegra::Shader::HalfType>(operation.GetMeta());
+ if (type == Tegra::Shader::HalfType::H0_H1) {
+ return operand;
+ }
+ const auto value = [&] {
+ switch (std::get<Tegra::Shader::HalfType>(operation.GetMeta())) {
+ case Tegra::Shader::HalfType::F32:
+ return GetHalfScalarFromFloat(AsFloat(operand));
+ case Tegra::Shader::HalfType::H0_H0:
+ return OpCompositeExtract(t_scalar_half, AsHalfFloat(operand), 0);
+ case Tegra::Shader::HalfType::H1_H1:
+ return OpCompositeExtract(t_scalar_half, AsHalfFloat(operand), 1);
+ default:
+ UNREACHABLE();
+ return ConstantNull(t_half);
+ }
+ }();
+ return {OpCompositeConstruct(t_half, value, value), Type::HalfFloat};
}
- Id HMergeH0(Operation operation) {
- UNIMPLEMENTED();
- return {};
+ Expression HMergeF32(Operation operation) {
+ const Id value = AsHalfFloat(Visit(operation[0]));
+ return {GetFloatFromHalfScalar(OpCompositeExtract(t_scalar_half, value, 0)), Type::Float};
}
- Id HMergeH1(Operation operation) {
- UNIMPLEMENTED();
- return {};
+ template <u32 offset>
+ Expression HMergeHN(Operation operation) {
+ const Id target = AsHalfFloat(Visit(operation[0]));
+ const Id source = AsHalfFloat(Visit(operation[1]));
+ const Id object = OpCompositeExtract(t_scalar_half, source, offset);
+ return {OpCompositeInsert(t_half, object, target, offset), Type::HalfFloat};
}
- Id HPack2(Operation operation) {
- UNIMPLEMENTED();
- return {};
+ Expression HPack2(Operation operation) {
+ const Id low = GetHalfScalarFromFloat(AsFloat(Visit(operation[0])));
+ const Id high = GetHalfScalarFromFloat(AsFloat(Visit(operation[1])));
+ return {OpCompositeConstruct(t_half, low, high), Type::HalfFloat};
}
- Id LogicalAssign(Operation operation) {
+ Expression LogicalAssign(Operation operation) {
const Node& dest = operation[0];
const Node& src = operation[1];
@@ -844,106 +1476,198 @@ private:
target = internal_flags.at(static_cast<u32>(flag->GetFlag()));
}
- Emit(OpStore(target, Visit(src)));
+ OpStore(target, AsBool(Visit(src)));
return {};
}
- Id LogicalPick2(Operation operation) {
- UNIMPLEMENTED();
- return {};
+ Id GetTextureSampler(Operation operation) {
+ const auto& meta = std::get<MetaTexture>(operation.GetMeta());
+ ASSERT(!meta.sampler.IsBuffer());
+
+ const auto& entry = sampled_images.at(meta.sampler.GetIndex());
+ return OpLoad(entry.sampled_image_type, entry.sampler);
}
- Id LogicalAnd2(Operation operation) {
- UNIMPLEMENTED();
- return {};
+ Id GetTextureImage(Operation operation) {
+ const auto& meta = std::get<MetaTexture>(operation.GetMeta());
+ const u32 index = meta.sampler.GetIndex();
+ if (meta.sampler.IsBuffer()) {
+ const auto& entry = texel_buffers.at(index);
+ return OpLoad(entry.image_type, entry.image);
+ } else {
+ const auto& entry = sampled_images.at(index);
+ return OpImage(entry.image_type, GetTextureSampler(operation));
+ }
}
- Id GetTextureSampler(Operation operation) {
- const auto meta = std::get_if<MetaTexture>(&operation.GetMeta());
- const auto entry = sampler_images.at(static_cast<u32>(meta->sampler.GetIndex()));
- return Emit(OpLoad(entry.sampled_image_type, entry.sampler));
+ Id GetImage(Operation operation) {
+ const auto& meta = std::get<MetaImage>(operation.GetMeta());
+ const auto entry = images.at(meta.image.GetIndex());
+ return OpLoad(entry.image_type, entry.image);
}
- Id GetTextureImage(Operation operation) {
- const auto meta = std::get_if<MetaTexture>(&operation.GetMeta());
- const auto entry = sampler_images.at(static_cast<u32>(meta->sampler.GetIndex()));
- return Emit(OpImage(entry.image_type, GetTextureSampler(operation)));
+ Id AssembleVector(const std::vector<Id>& coords, Type type) {
+ const Id coords_type = GetTypeVectorDefinitionLut(type).at(coords.size() - 1);
+ return coords.size() == 1 ? coords[0] : OpCompositeConstruct(coords_type, coords);
}
- Id GetTextureCoordinates(Operation operation) {
- const auto meta = std::get_if<MetaTexture>(&operation.GetMeta());
+ Id GetCoordinates(Operation operation, Type type) {
std::vector<Id> coords;
for (std::size_t i = 0; i < operation.GetOperandsCount(); ++i) {
- coords.push_back(Visit(operation[i]));
+ coords.push_back(As(Visit(operation[i]), type));
}
- if (meta->sampler.IsArray()) {
- const Id array_integer = BitcastTo<Type::Int>(Visit(meta->array));
- coords.push_back(Emit(OpConvertSToF(t_float, array_integer)));
+ if (const auto meta = std::get_if<MetaTexture>(&operation.GetMeta())) {
+ // Add array coordinate for textures
+ if (meta->sampler.IsArray()) {
+ Id array = AsInt(Visit(meta->array));
+ if (type == Type::Float) {
+ array = OpConvertSToF(t_float, array);
+ }
+ coords.push_back(array);
+ }
}
- if (meta->sampler.IsShadow()) {
- coords.push_back(Visit(meta->depth_compare));
+ return AssembleVector(coords, type);
+ }
+
+ Id GetOffsetCoordinates(Operation operation) {
+ const auto& meta = std::get<MetaTexture>(operation.GetMeta());
+ std::vector<Id> coords;
+ coords.reserve(meta.aoffi.size());
+ for (const auto& coord : meta.aoffi) {
+ coords.push_back(AsInt(Visit(coord)));
}
+ return AssembleVector(coords, Type::Int);
+ }
- const std::array<Id, 4> t_float_lut = {nullptr, t_float2, t_float3, t_float4};
- return coords.size() == 1
- ? coords[0]
- : Emit(OpCompositeConstruct(t_float_lut.at(coords.size() - 1), coords));
+ std::pair<Id, Id> GetDerivatives(Operation operation) {
+ const auto& meta = std::get<MetaTexture>(operation.GetMeta());
+ const auto& derivatives = meta.derivates;
+ ASSERT(derivatives.size() % 2 == 0);
+
+ const std::size_t components = derivatives.size() / 2;
+ std::vector<Id> dx, dy;
+ dx.reserve(components);
+ dy.reserve(components);
+ for (std::size_t index = 0; index < components; ++index) {
+ dx.push_back(AsFloat(Visit(derivatives.at(index * 2 + 0))));
+ dy.push_back(AsFloat(Visit(derivatives.at(index * 2 + 1))));
+ }
+ return {AssembleVector(dx, Type::Float), AssembleVector(dy, Type::Float)};
}
- Id GetTextureElement(Operation operation, Id sample_value) {
- const auto meta = std::get_if<MetaTexture>(&operation.GetMeta());
- ASSERT(meta);
- return Emit(OpCompositeExtract(t_float, sample_value, meta->element));
+ Expression GetTextureElement(Operation operation, Id sample_value, Type type) {
+ const auto& meta = std::get<MetaTexture>(operation.GetMeta());
+ const auto type_def = GetTypeDefinition(type);
+ return {OpCompositeExtract(type_def, sample_value, meta.element), type};
}
- Id Texture(Operation operation) {
- const Id texture = Emit(OpImageSampleImplicitLod(t_float4, GetTextureSampler(operation),
- GetTextureCoordinates(operation)));
- return GetTextureElement(operation, texture);
+ Expression Texture(Operation operation) {
+ const auto& meta = std::get<MetaTexture>(operation.GetMeta());
+
+ const bool can_implicit = stage == ShaderType::Fragment;
+ const Id sampler = GetTextureSampler(operation);
+ const Id coords = GetCoordinates(operation, Type::Float);
+
+ std::vector<Id> operands;
+ spv::ImageOperandsMask mask{};
+ if (meta.bias) {
+ mask = mask | spv::ImageOperandsMask::Bias;
+ operands.push_back(AsFloat(Visit(meta.bias)));
+ }
+
+ if (!can_implicit) {
+ mask = mask | spv::ImageOperandsMask::Lod;
+ operands.push_back(v_float_zero);
+ }
+
+ if (!meta.aoffi.empty()) {
+ mask = mask | spv::ImageOperandsMask::Offset;
+ operands.push_back(GetOffsetCoordinates(operation));
+ }
+
+ if (meta.depth_compare) {
+ // Depth sampling
+ UNIMPLEMENTED_IF(meta.bias);
+ const Id dref = AsFloat(Visit(meta.depth_compare));
+ if (can_implicit) {
+ return {
+ OpImageSampleDrefImplicitLod(t_float, sampler, coords, dref, mask, operands),
+ Type::Float};
+ } else {
+ return {
+ OpImageSampleDrefExplicitLod(t_float, sampler, coords, dref, mask, operands),
+ Type::Float};
+ }
+ }
+
+ Id texture;
+ if (can_implicit) {
+ texture = OpImageSampleImplicitLod(t_float4, sampler, coords, mask, operands);
+ } else {
+ texture = OpImageSampleExplicitLod(t_float4, sampler, coords, mask, operands);
+ }
+ return GetTextureElement(operation, texture, Type::Float);
}
- Id TextureLod(Operation operation) {
- const auto meta = std::get_if<MetaTexture>(&operation.GetMeta());
- const Id texture = Emit(OpImageSampleExplicitLod(
- t_float4, GetTextureSampler(operation), GetTextureCoordinates(operation),
- spv::ImageOperandsMask::Lod, Visit(meta->lod)));
- return GetTextureElement(operation, texture);
+ Expression TextureLod(Operation operation) {
+ const auto& meta = std::get<MetaTexture>(operation.GetMeta());
+
+ const Id sampler = GetTextureSampler(operation);
+ const Id coords = GetCoordinates(operation, Type::Float);
+ const Id lod = AsFloat(Visit(meta.lod));
+
+ spv::ImageOperandsMask mask = spv::ImageOperandsMask::Lod;
+ std::vector<Id> operands{lod};
+
+ if (!meta.aoffi.empty()) {
+ mask = mask | spv::ImageOperandsMask::Offset;
+ operands.push_back(GetOffsetCoordinates(operation));
+ }
+
+ if (meta.sampler.IsShadow()) {
+ const Id dref = AsFloat(Visit(meta.depth_compare));
+ return {OpImageSampleDrefExplicitLod(t_float, sampler, coords, dref, mask, operands),
+ Type::Float};
+ }
+ const Id texture = OpImageSampleExplicitLod(t_float4, sampler, coords, mask, operands);
+ return GetTextureElement(operation, texture, Type::Float);
}
- Id TextureGather(Operation operation) {
- const auto meta = std::get_if<MetaTexture>(&operation.GetMeta());
- const auto coords = GetTextureCoordinates(operation);
+ Expression TextureGather(Operation operation) {
+ const auto& meta = std::get<MetaTexture>(operation.GetMeta());
+ UNIMPLEMENTED_IF(!meta.aoffi.empty());
- Id texture;
- if (meta->sampler.IsShadow()) {
- texture = Emit(OpImageDrefGather(t_float4, GetTextureSampler(operation), coords,
- Visit(meta->component)));
+ const Id coords = GetCoordinates(operation, Type::Float);
+ Id texture{};
+ if (meta.sampler.IsShadow()) {
+ texture = OpImageDrefGather(t_float4, GetTextureSampler(operation), coords,
+ AsFloat(Visit(meta.depth_compare)));
} else {
u32 component_value = 0;
- if (meta->component) {
- const auto component = std::get_if<ImmediateNode>(&*meta->component);
+ if (meta.component) {
+ const auto component = std::get_if<ImmediateNode>(&*meta.component);
ASSERT_MSG(component, "Component is not an immediate value");
component_value = component->GetValue();
}
- texture = Emit(OpImageGather(t_float4, GetTextureSampler(operation), coords,
- Constant(t_uint, component_value)));
+ texture = OpImageGather(t_float4, GetTextureSampler(operation), coords,
+ Constant(t_uint, component_value));
}
-
- return GetTextureElement(operation, texture);
+ return GetTextureElement(operation, texture, Type::Float);
}
- Id TextureQueryDimensions(Operation operation) {
- const auto meta = std::get_if<MetaTexture>(&operation.GetMeta());
- const auto image_id = GetTextureImage(operation);
- AddCapability(spv::Capability::ImageQuery);
+ Expression TextureQueryDimensions(Operation operation) {
+ const auto& meta = std::get<MetaTexture>(operation.GetMeta());
+ UNIMPLEMENTED_IF(!meta.aoffi.empty());
+ UNIMPLEMENTED_IF(meta.depth_compare);
- if (meta->element == 3) {
- return BitcastTo<Type::Float>(Emit(OpImageQueryLevels(t_int, image_id)));
+ const auto image_id = GetTextureImage(operation);
+ if (meta.element == 3) {
+ return {OpImageQueryLevels(t_int, image_id), Type::Int};
}
- const Id lod = VisitOperand<Type::Uint>(operation, 0);
+ const Id lod = AsUint(Visit(operation[0]));
const std::size_t coords_count = [&]() {
- switch (const auto type = meta->sampler.GetType(); type) {
+ switch (const auto type = meta.sampler.GetType(); type) {
case Tegra::Shader::TextureType::Texture1D:
return 1;
case Tegra::Shader::TextureType::Texture2D:
@@ -957,136 +1681,190 @@ private:
}
}();
- if (meta->element >= coords_count) {
- return Constant(t_float, 0.0f);
+ if (meta.element >= coords_count) {
+ return {v_float_zero, Type::Float};
}
const std::array<Id, 3> types = {t_int, t_int2, t_int3};
- const Id sizes = Emit(OpImageQuerySizeLod(types.at(coords_count - 1), image_id, lod));
- const Id size = Emit(OpCompositeExtract(t_int, sizes, meta->element));
- return BitcastTo<Type::Float>(size);
+ const Id sizes = OpImageQuerySizeLod(types.at(coords_count - 1), image_id, lod);
+ const Id size = OpCompositeExtract(t_int, sizes, meta.element);
+ return {size, Type::Int};
+ }
+
+ Expression TextureQueryLod(Operation operation) {
+ const auto& meta = std::get<MetaTexture>(operation.GetMeta());
+ UNIMPLEMENTED_IF(!meta.aoffi.empty());
+ UNIMPLEMENTED_IF(meta.depth_compare);
+
+ if (meta.element >= 2) {
+ UNREACHABLE_MSG("Invalid element");
+ return {v_float_zero, Type::Float};
+ }
+ const auto sampler_id = GetTextureSampler(operation);
+
+ const Id multiplier = Constant(t_float, 256.0f);
+ const Id multipliers = ConstantComposite(t_float2, multiplier, multiplier);
+
+ const Id coords = GetCoordinates(operation, Type::Float);
+ Id size = OpImageQueryLod(t_float2, sampler_id, coords);
+ size = OpFMul(t_float2, size, multipliers);
+ size = OpConvertFToS(t_int2, size);
+ return GetTextureElement(operation, size, Type::Int);
+ }
+
+ Expression TexelFetch(Operation operation) {
+ const auto& meta = std::get<MetaTexture>(operation.GetMeta());
+ UNIMPLEMENTED_IF(meta.depth_compare);
+
+ const Id image = GetTextureImage(operation);
+ const Id coords = GetCoordinates(operation, Type::Int);
+ Id fetch;
+ if (meta.lod && !meta.sampler.IsBuffer()) {
+ fetch = OpImageFetch(t_float4, image, coords, spv::ImageOperandsMask::Lod,
+ AsInt(Visit(meta.lod)));
+ } else {
+ fetch = OpImageFetch(t_float4, image, coords);
+ }
+ return GetTextureElement(operation, fetch, Type::Float);
}
- Id TextureQueryLod(Operation operation) {
+ Expression TextureGradient(Operation operation) {
+ const auto& meta = std::get<MetaTexture>(operation.GetMeta());
+ UNIMPLEMENTED_IF(!meta.aoffi.empty());
+
+ const Id sampler = GetTextureSampler(operation);
+ const Id coords = GetCoordinates(operation, Type::Float);
+ const auto [dx, dy] = GetDerivatives(operation);
+ const std::vector grad = {dx, dy};
+
+ static constexpr auto mask = spv::ImageOperandsMask::Grad;
+ const Id texture = OpImageSampleExplicitLod(t_float4, sampler, coords, mask, grad);
+ return GetTextureElement(operation, texture, Type::Float);
+ }
+
+ Expression ImageLoad(Operation operation) {
UNIMPLEMENTED();
return {};
}
- Id TexelFetch(Operation operation) {
- UNIMPLEMENTED();
+ Expression ImageStore(Operation operation) {
+ const auto meta{std::get<MetaImage>(operation.GetMeta())};
+ std::vector<Id> colors;
+ for (const auto& value : meta.values) {
+ colors.push_back(AsUint(Visit(value)));
+ }
+
+ const Id coords = GetCoordinates(operation, Type::Int);
+ const Id texel = OpCompositeConstruct(t_uint4, colors);
+
+ OpImageWrite(GetImage(operation), coords, texel, {});
return {};
}
- Id ImageLoad(Operation operation) {
+ Expression AtomicImageAdd(Operation operation) {
UNIMPLEMENTED();
return {};
}
- Id ImageStore(Operation operation) {
+ Expression AtomicImageMin(Operation operation) {
UNIMPLEMENTED();
return {};
}
- Id AtomicImageAdd(Operation operation) {
+ Expression AtomicImageMax(Operation operation) {
UNIMPLEMENTED();
return {};
}
- Id AtomicImageAnd(Operation operation) {
+ Expression AtomicImageAnd(Operation operation) {
UNIMPLEMENTED();
return {};
}
- Id AtomicImageOr(Operation operation) {
+ Expression AtomicImageOr(Operation operation) {
UNIMPLEMENTED();
return {};
}
- Id AtomicImageXor(Operation operation) {
+ Expression AtomicImageXor(Operation operation) {
UNIMPLEMENTED();
return {};
}
- Id AtomicImageExchange(Operation operation) {
+ Expression AtomicImageExchange(Operation operation) {
UNIMPLEMENTED();
return {};
}
- Id Branch(Operation operation) {
- const auto target = std::get_if<ImmediateNode>(&*operation[0]);
- UNIMPLEMENTED_IF(!target);
-
- Emit(OpStore(jmp_to, Constant(t_uint, target->GetValue())));
- Emit(OpBranch(continue_label));
- inside_branch = conditional_nest_count;
- if (conditional_nest_count == 0) {
- Emit(OpLabel());
+ Expression Branch(Operation operation) {
+ const auto& target = std::get<ImmediateNode>(*operation[0]);
+ OpStore(jmp_to, Constant(t_uint, target.GetValue()));
+ OpBranch(continue_label);
+ inside_branch = true;
+ if (!conditional_branch_set) {
+ AddLabel();
}
return {};
}
- Id BranchIndirect(Operation operation) {
- const Id op_a = VisitOperand<Type::Uint>(operation, 0);
+ Expression BranchIndirect(Operation operation) {
+ const Id op_a = AsUint(Visit(operation[0]));
- Emit(OpStore(jmp_to, op_a));
- Emit(OpBranch(continue_label));
- inside_branch = conditional_nest_count;
- if (conditional_nest_count == 0) {
- Emit(OpLabel());
+ OpStore(jmp_to, op_a);
+ OpBranch(continue_label);
+ inside_branch = true;
+ if (!conditional_branch_set) {
+ AddLabel();
}
return {};
}
- Id PushFlowStack(Operation operation) {
- const auto target = std::get_if<ImmediateNode>(&*operation[0]);
- ASSERT(target);
-
+ Expression PushFlowStack(Operation operation) {
+ const auto& target = std::get<ImmediateNode>(*operation[0]);
const auto [flow_stack, flow_stack_top] = GetFlowStack(operation);
- const Id current = Emit(OpLoad(t_uint, flow_stack_top));
- const Id next = Emit(OpIAdd(t_uint, current, Constant(t_uint, 1)));
- const Id access = Emit(OpAccessChain(t_func_uint, flow_stack, current));
+ const Id current = OpLoad(t_uint, flow_stack_top);
+ const Id next = OpIAdd(t_uint, current, Constant(t_uint, 1));
+ const Id access = OpAccessChain(t_func_uint, flow_stack, current);
- Emit(OpStore(access, Constant(t_uint, target->GetValue())));
- Emit(OpStore(flow_stack_top, next));
+ OpStore(access, Constant(t_uint, target.GetValue()));
+ OpStore(flow_stack_top, next);
return {};
}
- Id PopFlowStack(Operation operation) {
+ Expression PopFlowStack(Operation operation) {
const auto [flow_stack, flow_stack_top] = GetFlowStack(operation);
- const Id current = Emit(OpLoad(t_uint, flow_stack_top));
- const Id previous = Emit(OpISub(t_uint, current, Constant(t_uint, 1)));
- const Id access = Emit(OpAccessChain(t_func_uint, flow_stack, previous));
- const Id target = Emit(OpLoad(t_uint, access));
-
- Emit(OpStore(flow_stack_top, previous));
- Emit(OpStore(jmp_to, target));
- Emit(OpBranch(continue_label));
- inside_branch = conditional_nest_count;
- if (conditional_nest_count == 0) {
- Emit(OpLabel());
+ const Id current = OpLoad(t_uint, flow_stack_top);
+ const Id previous = OpISub(t_uint, current, Constant(t_uint, 1));
+ const Id access = OpAccessChain(t_func_uint, flow_stack, previous);
+ const Id target = OpLoad(t_uint, access);
+
+ OpStore(flow_stack_top, previous);
+ OpStore(jmp_to, target);
+ OpBranch(continue_label);
+ inside_branch = true;
+ if (!conditional_branch_set) {
+ AddLabel();
}
return {};
}
- Id PreExit() {
- switch (stage) {
- case ShaderStage::Vertex: {
- // TODO(Rodrigo): We should use VK_EXT_depth_range_unrestricted instead, but it doesn't
- // seem to be working on Nvidia's drivers and Intel (mesa and blob) doesn't support it.
- const Id z_pointer = AccessElement(t_out_float, per_vertex, position_index, 2u);
- Id depth = Emit(OpLoad(t_float, z_pointer));
- depth = Emit(OpFAdd(t_float, depth, Constant(t_float, 1.0f)));
- depth = Emit(OpFMul(t_float, depth, Constant(t_float, 0.5f)));
- Emit(OpStore(z_pointer, depth));
- break;
+ void PreExit() {
+ if (stage == ShaderType::Vertex && specialization.ndc_minus_one_to_one) {
+ const u32 position_index = out_indices.position.value();
+ const Id z_pointer = AccessElement(t_out_float, out_vertex, position_index, 2U);
+ const Id w_pointer = AccessElement(t_out_float, out_vertex, position_index, 3U);
+ Id depth = OpLoad(t_float, z_pointer);
+ depth = OpFAdd(t_float, depth, OpLoad(t_float, w_pointer));
+ depth = OpFMul(t_float, depth, Constant(t_float, 0.5f));
+ OpStore(z_pointer, depth);
}
- case ShaderStage::Fragment: {
+ if (stage == ShaderType::Fragment) {
const auto SafeGetRegister = [&](u32 reg) {
// TODO(Rodrigo): Replace with contains once C++20 releases
if (const auto it = registers.find(reg); it != registers.end()) {
- return Emit(OpLoad(t_float, it->second));
+ return OpLoad(t_float, it->second);
}
- return Constant(t_float, 0.0f);
+ return v_float_zero;
};
UNIMPLEMENTED_IF_MSG(header.ps.omap.sample_mask != 0,
@@ -1098,159 +1876,145 @@ private:
// rendertargets/components are skipped in the register assignment.
u32 current_reg = 0;
for (u32 rt = 0; rt < Maxwell::NumRenderTargets; ++rt) {
+ if (!specialization.enabled_rendertargets[rt]) {
+ // Skip rendertargets that are not enabled
+ continue;
+ }
// TODO(Subv): Figure out how dual-source blending is configured in the Switch.
for (u32 component = 0; component < 4; ++component) {
+ const Id pointer = AccessElement(t_out_float, frag_colors.at(rt), component);
if (header.ps.IsColorComponentOutputEnabled(rt, component)) {
- Emit(OpStore(AccessElement(t_out_float, frag_colors.at(rt), component),
- SafeGetRegister(current_reg)));
+ OpStore(pointer, SafeGetRegister(current_reg));
++current_reg;
+ } else {
+ OpStore(pointer, component == 3 ? v_float_one : v_float_zero);
}
}
}
if (header.ps.omap.depth) {
// The depth output is always 2 registers after the last color output, and
// current_reg already contains one past the last color register.
- Emit(OpStore(frag_depth, SafeGetRegister(current_reg + 1)));
+ OpStore(frag_depth, SafeGetRegister(current_reg + 1));
}
- break;
- }
}
-
- return {};
}
- Id Exit(Operation operation) {
+ Expression Exit(Operation operation) {
PreExit();
- inside_branch = conditional_nest_count;
- if (conditional_nest_count > 0) {
- Emit(OpReturn());
+ inside_branch = true;
+ if (conditional_branch_set) {
+ OpReturn();
} else {
const Id dummy = OpLabel();
- Emit(OpBranch(dummy));
- Emit(dummy);
- Emit(OpReturn());
- Emit(OpLabel());
+ OpBranch(dummy);
+ AddLabel(dummy);
+ OpReturn();
+ AddLabel();
}
return {};
}
- Id Discard(Operation operation) {
- inside_branch = conditional_nest_count;
- if (conditional_nest_count > 0) {
- Emit(OpKill());
+ Expression Discard(Operation operation) {
+ inside_branch = true;
+ if (conditional_branch_set) {
+ OpKill();
} else {
const Id dummy = OpLabel();
- Emit(OpBranch(dummy));
- Emit(dummy);
- Emit(OpKill());
- Emit(OpLabel());
+ OpBranch(dummy);
+ AddLabel(dummy);
+ OpKill();
+ AddLabel();
}
return {};
}
- Id EmitVertex(Operation operation) {
- UNIMPLEMENTED();
+ Expression EmitVertex(Operation) {
+ OpEmitVertex();
return {};
}
- Id EndPrimitive(Operation operation) {
- UNIMPLEMENTED();
+ Expression EndPrimitive(Operation operation) {
+ OpEndPrimitive();
return {};
}
- Id YNegate(Operation operation) {
- UNIMPLEMENTED();
- return {};
+ Expression InvocationId(Operation) {
+ return {OpLoad(t_int, invocation_id), Type::Int};
}
- template <u32 element>
- Id LocalInvocationId(Operation) {
- UNIMPLEMENTED();
- return {};
+ Expression YNegate(Operation) {
+ LOG_WARNING(Render_Vulkan, "(STUBBED)");
+ return {Constant(t_float, 1.0f), Type::Float};
}
template <u32 element>
- Id WorkGroupId(Operation) {
- UNIMPLEMENTED();
- return {};
- }
-
- Id BallotThread(Operation) {
- UNIMPLEMENTED();
- return {};
+ Expression LocalInvocationId(Operation) {
+ const Id id = OpLoad(t_uint3, local_invocation_id);
+ return {OpCompositeExtract(t_uint, id, element), Type::Uint};
}
- Id VoteAll(Operation) {
- UNIMPLEMENTED();
- return {};
- }
-
- Id VoteAny(Operation) {
- UNIMPLEMENTED();
- return {};
+ template <u32 element>
+ Expression WorkGroupId(Operation operation) {
+ const Id id = OpLoad(t_uint3, workgroup_id);
+ return {OpCompositeExtract(t_uint, id, element), Type::Uint};
}
- Id VoteEqual(Operation) {
- UNIMPLEMENTED();
- return {};
- }
+ Expression BallotThread(Operation operation) {
+ const Id predicate = AsBool(Visit(operation[0]));
+ const Id ballot = OpSubgroupBallotKHR(t_uint4, predicate);
- Id ShuffleIndexed(Operation) {
- UNIMPLEMENTED();
- return {};
- }
+ if (!device.IsWarpSizePotentiallyBiggerThanGuest()) {
+ // Guest-like devices can just return the first index.
+ return {OpCompositeExtract(t_uint, ballot, 0U), Type::Uint};
+ }
- Id ShuffleUp(Operation) {
- UNIMPLEMENTED();
- return {};
+ // The others will have to return what is local to the current thread.
+ // For instance a device with a warp size of 64 will return the upper uint when the current
+ // thread is 38.
+ const Id tid = OpLoad(t_uint, thread_id);
+ const Id thread_index = OpShiftRightLogical(t_uint, tid, Constant(t_uint, 5));
+ return {OpVectorExtractDynamic(t_uint, ballot, thread_index), Type::Uint};
}
- Id ShuffleDown(Operation) {
- UNIMPLEMENTED();
- return {};
+ template <Id (Module::*func)(Id, Id)>
+ Expression Vote(Operation operation) {
+ // TODO(Rodrigo): Handle devices with different warp sizes
+ const Id predicate = AsBool(Visit(operation[0]));
+ return {(this->*func)(t_bool, predicate), Type::Bool};
}
- Id ShuffleButterfly(Operation) {
- UNIMPLEMENTED();
- return {};
+ Expression ThreadId(Operation) {
+ return {OpLoad(t_uint, thread_id), Type::Uint};
}
- Id InRangeShuffleIndexed(Operation) {
- UNIMPLEMENTED();
- return {};
- }
-
- Id InRangeShuffleUp(Operation) {
- UNIMPLEMENTED();
- return {};
+ Expression ShuffleIndexed(Operation operation) {
+ const Id value = AsFloat(Visit(operation[0]));
+ const Id index = AsUint(Visit(operation[1]));
+ return {OpSubgroupReadInvocationKHR(t_float, value, index), Type::Float};
}
- Id InRangeShuffleDown(Operation) {
- UNIMPLEMENTED();
- return {};
- }
+ Expression MemoryBarrierGL(Operation) {
+ const auto scope = spv::Scope::Device;
+ const auto semantics =
+ spv::MemorySemanticsMask::AcquireRelease | spv::MemorySemanticsMask::UniformMemory |
+ spv::MemorySemanticsMask::WorkgroupMemory |
+ spv::MemorySemanticsMask::AtomicCounterMemory | spv::MemorySemanticsMask::ImageMemory;
- Id InRangeShuffleButterfly(Operation) {
- UNIMPLEMENTED();
+ OpMemoryBarrier(Constant(t_uint, static_cast<u32>(scope)),
+ Constant(t_uint, static_cast<u32>(semantics)));
return {};
}
- Id DeclareBuiltIn(spv::BuiltIn builtin, spv::StorageClass storage, Id type,
- const std::string& name) {
+ Id DeclareBuiltIn(spv::BuiltIn builtin, spv::StorageClass storage, Id type, std::string name) {
const Id id = OpVariable(type, storage);
Decorate(id, spv::Decoration::BuiltIn, static_cast<u32>(builtin));
- AddGlobalVariable(Name(id, name));
+ AddGlobalVariable(Name(id, std::move(name)));
interfaces.push_back(id);
return id;
}
- bool IsRenderTargetUsed(u32 rt) const {
- for (u32 component = 0; component < 4; ++component) {
- if (header.ps.IsColorComponentOutputEnabled(rt, component)) {
- return true;
- }
- }
- return false;
+ Id DeclareInputBuiltIn(spv::BuiltIn builtin, Id type, std::string name) {
+ return DeclareBuiltIn(builtin, spv::StorageClass::Input, type, std::move(name));
}
template <typename... Args>
@@ -1261,66 +2025,148 @@ private:
members.push_back(Constant(t_uint, element));
}
- return Emit(OpAccessChain(pointer_type, composite, members));
+ return OpAccessChain(pointer_type, composite, members);
}
- template <Type type>
- Id VisitOperand(Operation operation, std::size_t operand_index) {
- const Id value = Visit(operation[operand_index]);
-
- switch (type) {
+ Id As(Expression expr, Type wanted_type) {
+ switch (wanted_type) {
case Type::Bool:
+ return AsBool(expr);
case Type::Bool2:
+ return AsBool2(expr);
case Type::Float:
- return value;
+ return AsFloat(expr);
case Type::Int:
- return Emit(OpBitcast(t_int, value));
+ return AsInt(expr);
case Type::Uint:
- return Emit(OpBitcast(t_uint, value));
+ return AsUint(expr);
case Type::HalfFloat:
- UNIMPLEMENTED();
+ return AsHalfFloat(expr);
+ default:
+ UNREACHABLE();
+ return expr.id;
}
- UNREACHABLE();
- return value;
}
- template <Type type>
- Id BitcastFrom(Id value) {
- switch (type) {
- case Type::Bool:
- case Type::Bool2:
+ Id AsBool(Expression expr) {
+ ASSERT(expr.type == Type::Bool);
+ return expr.id;
+ }
+
+ Id AsBool2(Expression expr) {
+ ASSERT(expr.type == Type::Bool2);
+ return expr.id;
+ }
+
+ Id AsFloat(Expression expr) {
+ switch (expr.type) {
case Type::Float:
- return value;
+ return expr.id;
case Type::Int:
case Type::Uint:
- return Emit(OpBitcast(t_float, value));
+ return OpBitcast(t_float, expr.id);
case Type::HalfFloat:
- UNIMPLEMENTED();
+ if (device.IsFloat16Supported()) {
+ return OpBitcast(t_float, expr.id);
+ }
+ return OpBitcast(t_float, OpPackHalf2x16(t_uint, expr.id));
+ default:
+ UNREACHABLE();
+ return expr.id;
}
- UNREACHABLE();
- return value;
}
- template <Type type>
- Id BitcastTo(Id value) {
- switch (type) {
- case Type::Bool:
- case Type::Bool2:
+ Id AsInt(Expression expr) {
+ switch (expr.type) {
+ case Type::Int:
+ return expr.id;
+ case Type::Float:
+ case Type::Uint:
+ return OpBitcast(t_int, expr.id);
+ case Type::HalfFloat:
+ if (device.IsFloat16Supported()) {
+ return OpBitcast(t_int, expr.id);
+ }
+ return OpPackHalf2x16(t_int, expr.id);
+ default:
UNREACHABLE();
+ return expr.id;
+ }
+ }
+
+ Id AsUint(Expression expr) {
+ switch (expr.type) {
+ case Type::Uint:
+ return expr.id;
case Type::Float:
- return Emit(OpBitcast(t_float, value));
case Type::Int:
- return Emit(OpBitcast(t_int, value));
- case Type::Uint:
- return Emit(OpBitcast(t_uint, value));
+ return OpBitcast(t_uint, expr.id);
case Type::HalfFloat:
- UNIMPLEMENTED();
+ if (device.IsFloat16Supported()) {
+ return OpBitcast(t_uint, expr.id);
+ }
+ return OpPackHalf2x16(t_uint, expr.id);
+ default:
+ UNREACHABLE();
+ return expr.id;
+ }
+ }
+
+ Id AsHalfFloat(Expression expr) {
+ switch (expr.type) {
+ case Type::HalfFloat:
+ return expr.id;
+ case Type::Float:
+ case Type::Int:
+ case Type::Uint:
+ if (device.IsFloat16Supported()) {
+ return OpBitcast(t_half, expr.id);
+ }
+ return OpUnpackHalf2x16(t_half, AsUint(expr));
+ default:
+ UNREACHABLE();
+ return expr.id;
+ }
+ }
+
+ Id GetHalfScalarFromFloat(Id value) {
+ if (device.IsFloat16Supported()) {
+ return OpFConvert(t_scalar_half, value);
+ }
+ return value;
+ }
+
+ Id GetFloatFromHalfScalar(Id value) {
+ if (device.IsFloat16Supported()) {
+ return OpFConvert(t_float, value);
}
- UNREACHABLE();
return value;
}
- Id GetTypeDefinition(Type type) {
+ AttributeType GetAttributeType(u32 location) const {
+ if (stage != ShaderType::Vertex) {
+ return {Type::Float, t_in_float, t_in_float4};
+ }
+ switch (specialization.attribute_types.at(location)) {
+ case Maxwell::VertexAttribute::Type::SignedNorm:
+ case Maxwell::VertexAttribute::Type::UnsignedNorm:
+ case Maxwell::VertexAttribute::Type::Float:
+ return {Type::Float, t_in_float, t_in_float4};
+ case Maxwell::VertexAttribute::Type::SignedInt:
+ return {Type::Int, t_in_int, t_in_int4};
+ case Maxwell::VertexAttribute::Type::UnsignedInt:
+ return {Type::Uint, t_in_uint, t_in_uint4};
+ case Maxwell::VertexAttribute::Type::UnsignedScaled:
+ case Maxwell::VertexAttribute::Type::SignedScaled:
+ UNIMPLEMENTED();
+ return {Type::Float, t_in_float, t_in_float4};
+ default:
+ UNREACHABLE();
+ return {Type::Float, t_in_float, t_in_float4};
+ }
+ }
+
+ Id GetTypeDefinition(Type type) const {
switch (type) {
case Type::Bool:
return t_bool;
@@ -1333,10 +2179,25 @@ private:
case Type::Uint:
return t_uint;
case Type::HalfFloat:
+ return t_half;
+ default:
+ UNREACHABLE();
+ return {};
+ }
+ }
+
+ std::array<Id, 4> GetTypeVectorDefinitionLut(Type type) const {
+ switch (type) {
+ case Type::Float:
+ return {nullptr, t_float2, t_float3, t_float4};
+ case Type::Int:
+ return {nullptr, t_int2, t_int3, t_int4};
+ case Type::Uint:
+ return {nullptr, t_uint2, t_uint3, t_uint4};
+ default:
UNIMPLEMENTED();
+ return {};
}
- UNREACHABLE();
- return {};
}
std::tuple<Id, Id> CreateFlowStack() {
@@ -1346,9 +2207,11 @@ private:
constexpr auto storage_class = spv::StorageClass::Function;
const Id flow_stack_type = TypeArray(t_uint, Constant(t_uint, FLOW_STACK_SIZE));
- const Id stack = Emit(OpVariable(TypePointer(storage_class, flow_stack_type), storage_class,
- ConstantNull(flow_stack_type)));
- const Id top = Emit(OpVariable(t_func_uint, storage_class, Constant(t_uint, 0)));
+ const Id stack = OpVariable(TypePointer(storage_class, flow_stack_type), storage_class,
+ ConstantNull(flow_stack_type));
+ const Id top = OpVariable(t_func_uint, storage_class, Constant(t_uint, 0));
+ AddLocalVariable(stack);
+ AddLocalVariable(top);
return std::tie(stack, top);
}
@@ -1377,8 +2240,8 @@ private:
&SPIRVDecompiler::Unary<&Module::OpFNegate, Type::Float>,
&SPIRVDecompiler::Unary<&Module::OpFAbs, Type::Float>,
&SPIRVDecompiler::Ternary<&Module::OpFClamp, Type::Float>,
- &SPIRVDecompiler::FCastHalf0,
- &SPIRVDecompiler::FCastHalf1,
+ &SPIRVDecompiler::FCastHalf<0>,
+ &SPIRVDecompiler::FCastHalf<1>,
&SPIRVDecompiler::Binary<&Module::OpFMin, Type::Float>,
&SPIRVDecompiler::Binary<&Module::OpFMax, Type::Float>,
&SPIRVDecompiler::Unary<&Module::OpCos, Type::Float>,
@@ -1393,6 +2256,7 @@ private:
&SPIRVDecompiler::Unary<&Module::OpTrunc, Type::Float>,
&SPIRVDecompiler::Unary<&Module::OpConvertSToF, Type::Float, Type::Int>,
&SPIRVDecompiler::Unary<&Module::OpConvertUToF, Type::Float, Type::Uint>,
+ &SPIRVDecompiler::FSwizzleAdd,
&SPIRVDecompiler::Binary<&Module::OpIAdd, Type::Int>,
&SPIRVDecompiler::Binary<&Module::OpIMul, Type::Int>,
@@ -1414,6 +2278,7 @@ private:
&SPIRVDecompiler::Quaternary<&Module::OpBitFieldInsert, Type::Int>,
&SPIRVDecompiler::Ternary<&Module::OpBitFieldSExtract, Type::Int>,
&SPIRVDecompiler::Unary<&Module::OpBitCount, Type::Int>,
+ &SPIRVDecompiler::Unary<&Module::OpFindSMsb, Type::Int>,
&SPIRVDecompiler::Binary<&Module::OpIAdd, Type::Uint>,
&SPIRVDecompiler::Binary<&Module::OpIMul, Type::Uint>,
@@ -1424,7 +2289,7 @@ private:
&SPIRVDecompiler::Unary<&Module::OpBitcast, Type::Uint, Type::Int>,
&SPIRVDecompiler::Binary<&Module::OpShiftLeftLogical, Type::Uint>,
&SPIRVDecompiler::Binary<&Module::OpShiftRightLogical, Type::Uint>,
- &SPIRVDecompiler::Binary<&Module::OpShiftRightArithmetic, Type::Uint>,
+ &SPIRVDecompiler::Binary<&Module::OpShiftRightLogical, Type::Uint>,
&SPIRVDecompiler::Binary<&Module::OpBitwiseAnd, Type::Uint>,
&SPIRVDecompiler::Binary<&Module::OpBitwiseOr, Type::Uint>,
&SPIRVDecompiler::Binary<&Module::OpBitwiseXor, Type::Uint>,
@@ -1432,6 +2297,7 @@ private:
&SPIRVDecompiler::Quaternary<&Module::OpBitFieldInsert, Type::Uint>,
&SPIRVDecompiler::Ternary<&Module::OpBitFieldUExtract, Type::Uint>,
&SPIRVDecompiler::Unary<&Module::OpBitCount, Type::Uint>,
+ &SPIRVDecompiler::Unary<&Module::OpFindUMsb, Type::Uint>,
&SPIRVDecompiler::Binary<&Module::OpFAdd, Type::HalfFloat>,
&SPIRVDecompiler::Binary<&Module::OpFMul, Type::HalfFloat>,
@@ -1442,8 +2308,8 @@ private:
&SPIRVDecompiler::HCastFloat,
&SPIRVDecompiler::HUnpack,
&SPIRVDecompiler::HMergeF32,
- &SPIRVDecompiler::HMergeH0,
- &SPIRVDecompiler::HMergeH1,
+ &SPIRVDecompiler::HMergeHN<0>,
+ &SPIRVDecompiler::HMergeHN<1>,
&SPIRVDecompiler::HPack2,
&SPIRVDecompiler::LogicalAssign,
@@ -1451,8 +2317,9 @@ private:
&SPIRVDecompiler::Binary<&Module::OpLogicalOr, Type::Bool>,
&SPIRVDecompiler::Binary<&Module::OpLogicalNotEqual, Type::Bool>,
&SPIRVDecompiler::Unary<&Module::OpLogicalNot, Type::Bool>,
- &SPIRVDecompiler::LogicalPick2,
- &SPIRVDecompiler::LogicalAnd2,
+ &SPIRVDecompiler::Binary<&Module::OpVectorExtractDynamic, Type::Bool, Type::Bool2,
+ Type::Uint>,
+ &SPIRVDecompiler::Unary<&Module::OpAll, Type::Bool, Type::Bool2>,
&SPIRVDecompiler::Binary<&Module::OpFOrdLessThan, Type::Bool, Type::Float>,
&SPIRVDecompiler::Binary<&Module::OpFOrdEqual, Type::Bool, Type::Float>,
@@ -1460,7 +2327,7 @@ private:
&SPIRVDecompiler::Binary<&Module::OpFOrdGreaterThan, Type::Bool, Type::Float>,
&SPIRVDecompiler::Binary<&Module::OpFOrdNotEqual, Type::Bool, Type::Float>,
&SPIRVDecompiler::Binary<&Module::OpFOrdGreaterThanEqual, Type::Bool, Type::Float>,
- &SPIRVDecompiler::Unary<&Module::OpIsNan, Type::Bool>,
+ &SPIRVDecompiler::Unary<&Module::OpIsNan, Type::Bool, Type::Float>,
&SPIRVDecompiler::Binary<&Module::OpSLessThan, Type::Bool, Type::Int>,
&SPIRVDecompiler::Binary<&Module::OpIEqual, Type::Bool, Type::Int>,
@@ -1476,19 +2343,19 @@ private:
&SPIRVDecompiler::Binary<&Module::OpINotEqual, Type::Bool, Type::Uint>,
&SPIRVDecompiler::Binary<&Module::OpUGreaterThanEqual, Type::Bool, Type::Uint>,
- &SPIRVDecompiler::Binary<&Module::OpFOrdLessThan, Type::Bool, Type::HalfFloat>,
- &SPIRVDecompiler::Binary<&Module::OpFOrdEqual, Type::Bool, Type::HalfFloat>,
- &SPIRVDecompiler::Binary<&Module::OpFOrdLessThanEqual, Type::Bool, Type::HalfFloat>,
- &SPIRVDecompiler::Binary<&Module::OpFOrdGreaterThan, Type::Bool, Type::HalfFloat>,
- &SPIRVDecompiler::Binary<&Module::OpFOrdNotEqual, Type::Bool, Type::HalfFloat>,
- &SPIRVDecompiler::Binary<&Module::OpFOrdGreaterThanEqual, Type::Bool, Type::HalfFloat>,
+ &SPIRVDecompiler::Binary<&Module::OpFOrdLessThan, Type::Bool2, Type::HalfFloat>,
+ &SPIRVDecompiler::Binary<&Module::OpFOrdEqual, Type::Bool2, Type::HalfFloat>,
+ &SPIRVDecompiler::Binary<&Module::OpFOrdLessThanEqual, Type::Bool2, Type::HalfFloat>,
+ &SPIRVDecompiler::Binary<&Module::OpFOrdGreaterThan, Type::Bool2, Type::HalfFloat>,
+ &SPIRVDecompiler::Binary<&Module::OpFOrdNotEqual, Type::Bool2, Type::HalfFloat>,
+ &SPIRVDecompiler::Binary<&Module::OpFOrdGreaterThanEqual, Type::Bool2, Type::HalfFloat>,
// TODO(Rodrigo): Should these use the OpFUnord* variants?
- &SPIRVDecompiler::Binary<&Module::OpFOrdLessThan, Type::Bool, Type::HalfFloat>,
- &SPIRVDecompiler::Binary<&Module::OpFOrdEqual, Type::Bool, Type::HalfFloat>,
- &SPIRVDecompiler::Binary<&Module::OpFOrdLessThanEqual, Type::Bool, Type::HalfFloat>,
- &SPIRVDecompiler::Binary<&Module::OpFOrdGreaterThan, Type::Bool, Type::HalfFloat>,
- &SPIRVDecompiler::Binary<&Module::OpFOrdNotEqual, Type::Bool, Type::HalfFloat>,
- &SPIRVDecompiler::Binary<&Module::OpFOrdGreaterThanEqual, Type::Bool, Type::HalfFloat>,
+ &SPIRVDecompiler::Binary<&Module::OpFOrdLessThan, Type::Bool2, Type::HalfFloat>,
+ &SPIRVDecompiler::Binary<&Module::OpFOrdEqual, Type::Bool2, Type::HalfFloat>,
+ &SPIRVDecompiler::Binary<&Module::OpFOrdLessThanEqual, Type::Bool2, Type::HalfFloat>,
+ &SPIRVDecompiler::Binary<&Module::OpFOrdGreaterThan, Type::Bool2, Type::HalfFloat>,
+ &SPIRVDecompiler::Binary<&Module::OpFOrdNotEqual, Type::Bool2, Type::HalfFloat>,
+ &SPIRVDecompiler::Binary<&Module::OpFOrdGreaterThanEqual, Type::Bool2, Type::HalfFloat>,
&SPIRVDecompiler::Texture,
&SPIRVDecompiler::TextureLod,
@@ -1496,6 +2363,7 @@ private:
&SPIRVDecompiler::TextureQueryDimensions,
&SPIRVDecompiler::TextureQueryLod,
&SPIRVDecompiler::TexelFetch,
+ &SPIRVDecompiler::TextureGradient,
&SPIRVDecompiler::ImageLoad,
&SPIRVDecompiler::ImageStore,
@@ -1515,6 +2383,7 @@ private:
&SPIRVDecompiler::EmitVertex,
&SPIRVDecompiler::EndPrimitive,
+ &SPIRVDecompiler::InvocationId,
&SPIRVDecompiler::YNegate,
&SPIRVDecompiler::LocalInvocationId<0>,
&SPIRVDecompiler::LocalInvocationId<1>,
@@ -1524,28 +2393,22 @@ private:
&SPIRVDecompiler::WorkGroupId<2>,
&SPIRVDecompiler::BallotThread,
- &SPIRVDecompiler::VoteAll,
- &SPIRVDecompiler::VoteAny,
- &SPIRVDecompiler::VoteEqual,
+ &SPIRVDecompiler::Vote<&Module::OpSubgroupAllKHR>,
+ &SPIRVDecompiler::Vote<&Module::OpSubgroupAnyKHR>,
+ &SPIRVDecompiler::Vote<&Module::OpSubgroupAllEqualKHR>,
+ &SPIRVDecompiler::ThreadId,
&SPIRVDecompiler::ShuffleIndexed,
- &SPIRVDecompiler::ShuffleUp,
- &SPIRVDecompiler::ShuffleDown,
- &SPIRVDecompiler::ShuffleButterfly,
-
- &SPIRVDecompiler::InRangeShuffleIndexed,
- &SPIRVDecompiler::InRangeShuffleUp,
- &SPIRVDecompiler::InRangeShuffleDown,
- &SPIRVDecompiler::InRangeShuffleButterfly,
+
+ &SPIRVDecompiler::MemoryBarrierGL,
};
static_assert(operation_decompilers.size() == static_cast<std::size_t>(OperationCode::Amount));
const VKDevice& device;
const ShaderIR& ir;
- const ShaderStage stage;
+ const ShaderType stage;
const Tegra::Shader::Header header;
- u64 conditional_nest_count{};
- u64 inside_branch{};
+ const Specialization& specialization;
const Id t_void = Name(TypeVoid(), "void");
@@ -1573,20 +2436,28 @@ private:
const Id t_func_uint = Name(TypePointer(spv::StorageClass::Function, t_uint), "func_uint");
const Id t_in_bool = Name(TypePointer(spv::StorageClass::Input, t_bool), "in_bool");
+ const Id t_in_int = Name(TypePointer(spv::StorageClass::Input, t_int), "in_int");
+ const Id t_in_int4 = Name(TypePointer(spv::StorageClass::Input, t_int4), "in_int4");
const Id t_in_uint = Name(TypePointer(spv::StorageClass::Input, t_uint), "in_uint");
+ const Id t_in_uint3 = Name(TypePointer(spv::StorageClass::Input, t_uint3), "in_uint3");
+ const Id t_in_uint4 = Name(TypePointer(spv::StorageClass::Input, t_uint4), "in_uint4");
const Id t_in_float = Name(TypePointer(spv::StorageClass::Input, t_float), "in_float");
+ const Id t_in_float2 = Name(TypePointer(spv::StorageClass::Input, t_float2), "in_float2");
+ const Id t_in_float3 = Name(TypePointer(spv::StorageClass::Input, t_float3), "in_float3");
const Id t_in_float4 = Name(TypePointer(spv::StorageClass::Input, t_float4), "in_float4");
+ const Id t_out_int = Name(TypePointer(spv::StorageClass::Output, t_int), "out_int");
+
const Id t_out_float = Name(TypePointer(spv::StorageClass::Output, t_float), "out_float");
const Id t_out_float4 = Name(TypePointer(spv::StorageClass::Output, t_float4), "out_float4");
const Id t_cbuf_float = TypePointer(spv::StorageClass::Uniform, t_float);
const Id t_cbuf_std140 = Decorate(
- Name(TypeArray(t_float4, Constant(t_uint, MAX_CONSTBUFFER_ELEMENTS)), "CbufStd140Array"),
- spv::Decoration::ArrayStride, 16u);
+ Name(TypeArray(t_float4, Constant(t_uint, MaxConstBufferElements)), "CbufStd140Array"),
+ spv::Decoration::ArrayStride, 16U);
const Id t_cbuf_scalar = Decorate(
- Name(TypeArray(t_float, Constant(t_uint, MAX_CONSTBUFFER_FLOATS)), "CbufScalarArray"),
- spv::Decoration::ArrayStride, 4u);
+ Name(TypeArray(t_float, Constant(t_uint, MaxConstBufferFloats)), "CbufScalarArray"),
+ spv::Decoration::ArrayStride, 4U);
const Id t_cbuf_std140_struct = MemberDecorate(
Decorate(TypeStruct(t_cbuf_std140), spv::Decoration::Block), 0, spv::Decoration::Offset, 0);
const Id t_cbuf_scalar_struct = MemberDecorate(
@@ -1594,28 +2465,43 @@ private:
const Id t_cbuf_std140_ubo = TypePointer(spv::StorageClass::Uniform, t_cbuf_std140_struct);
const Id t_cbuf_scalar_ubo = TypePointer(spv::StorageClass::Uniform, t_cbuf_scalar_struct);
+ Id t_smem_uint{};
+
const Id t_gmem_float = TypePointer(spv::StorageClass::StorageBuffer, t_float);
const Id t_gmem_array =
- Name(Decorate(TypeRuntimeArray(t_float), spv::Decoration::ArrayStride, 4u), "GmemArray");
+ Name(Decorate(TypeRuntimeArray(t_float), spv::Decoration::ArrayStride, 4U), "GmemArray");
const Id t_gmem_struct = MemberDecorate(
Decorate(TypeStruct(t_gmem_array), spv::Decoration::Block), 0, spv::Decoration::Offset, 0);
const Id t_gmem_ssbo = TypePointer(spv::StorageClass::StorageBuffer, t_gmem_struct);
const Id v_float_zero = Constant(t_float, 0.0f);
+ const Id v_float_one = Constant(t_float, 1.0f);
+
+ // Nvidia uses these defaults for varyings (e.g. position and generic attributes)
+ const Id v_varying_default =
+ ConstantComposite(t_float4, v_float_zero, v_float_zero, v_float_zero, v_float_one);
+
const Id v_true = ConstantTrue(t_bool);
const Id v_false = ConstantFalse(t_bool);
- Id per_vertex{};
+ Id t_scalar_half{};
+ Id t_half{};
+
+ Id out_vertex{};
+ Id in_vertex{};
std::map<u32, Id> registers;
std::map<Tegra::Shader::Pred, Id> predicates;
std::map<u32, Id> flow_variables;
Id local_memory{};
+ Id shared_memory{};
std::array<Id, INTERNAL_FLAGS_COUNT> internal_flags{};
std::map<Attribute::Index, Id> input_attributes;
std::map<Attribute::Index, Id> output_attributes;
std::map<u32, Id> constant_buffers;
std::map<GlobalMemoryBase, Id> global_buffers;
- std::map<u32, SamplerImage> sampler_images;
+ std::map<u32, TexelBuffer> texel_buffers;
+ std::map<u32, SampledImage> sampled_images;
+ std::map<u32, StorageImage> images;
Id instance_index{};
Id vertex_index{};
@@ -1623,18 +2509,20 @@ private:
Id frag_depth{};
Id frag_coord{};
Id front_facing{};
-
- u32 position_index{};
- u32 point_size_index{};
- u32 clip_distances_index{};
+ Id point_coord{};
+ Id tess_level_outer{};
+ Id tess_level_inner{};
+ Id tess_coord{};
+ Id invocation_id{};
+ Id workgroup_id{};
+ Id local_invocation_id{};
+ Id thread_id{};
+
+ VertexIndices in_indices;
+ VertexIndices out_indices;
std::vector<Id> interfaces;
- u32 const_buffers_base_binding{};
- u32 global_buffers_base_binding{};
- u32 samplers_base_binding{};
-
- Id execute_function{};
Id jmp_to{};
Id ssy_flow_stack_top{};
Id pbk_flow_stack_top{};
@@ -1642,6 +2530,9 @@ private:
Id pbk_flow_stack{};
Id continue_label{};
std::map<u32, Id> labels;
+
+ bool conditional_branch_set{};
+ bool inside_branch{};
};
class ExprDecompiler {
@@ -1652,52 +2543,33 @@ public:
const Id type_def = decomp.GetTypeDefinition(Type::Bool);
const Id op1 = Visit(expr.operand1);
const Id op2 = Visit(expr.operand2);
- return decomp.Emit(decomp.OpLogicalAnd(type_def, op1, op2));
+ return decomp.OpLogicalAnd(type_def, op1, op2);
}
Id operator()(const ExprOr& expr) {
const Id type_def = decomp.GetTypeDefinition(Type::Bool);
const Id op1 = Visit(expr.operand1);
const Id op2 = Visit(expr.operand2);
- return decomp.Emit(decomp.OpLogicalOr(type_def, op1, op2));
+ return decomp.OpLogicalOr(type_def, op1, op2);
}
Id operator()(const ExprNot& expr) {
const Id type_def = decomp.GetTypeDefinition(Type::Bool);
const Id op1 = Visit(expr.operand1);
- return decomp.Emit(decomp.OpLogicalNot(type_def, op1));
+ return decomp.OpLogicalNot(type_def, op1);
}
Id operator()(const ExprPredicate& expr) {
const auto pred = static_cast<Tegra::Shader::Pred>(expr.predicate);
- return decomp.Emit(decomp.OpLoad(decomp.t_bool, decomp.predicates.at(pred)));
+ return decomp.OpLoad(decomp.t_bool, decomp.predicates.at(pred));
}
Id operator()(const ExprCondCode& expr) {
- const Node cc = decomp.ir.GetConditionCode(expr.cc);
- Id target;
-
- if (const auto pred = std::get_if<PredicateNode>(&*cc)) {
- const auto index = pred->GetIndex();
- switch (index) {
- case Tegra::Shader::Pred::NeverExecute:
- target = decomp.v_false;
- break;
- case Tegra::Shader::Pred::UnusedIndex:
- target = decomp.v_true;
- break;
- default:
- target = decomp.predicates.at(index);
- break;
- }
- } else if (const auto flag = std::get_if<InternalFlagNode>(&*cc)) {
- target = decomp.internal_flags.at(static_cast<u32>(flag->GetFlag()));
- }
- return decomp.Emit(decomp.OpLoad(decomp.t_bool, target));
+ return decomp.AsBool(decomp.Visit(decomp.ir.GetConditionCode(expr.cc)));
}
Id operator()(const ExprVar& expr) {
- return decomp.Emit(decomp.OpLoad(decomp.t_bool, decomp.flow_variables.at(expr.var_index)));
+ return decomp.OpLoad(decomp.t_bool, decomp.flow_variables.at(expr.var_index));
}
Id operator()(const ExprBoolean& expr) {
@@ -1706,9 +2578,9 @@ public:
Id operator()(const ExprGprEqual& expr) {
const Id target = decomp.Constant(decomp.t_uint, expr.value);
- const Id gpr = decomp.BitcastTo<Type::Uint>(
- decomp.Emit(decomp.OpLoad(decomp.t_float, decomp.registers.at(expr.gpr))));
- return decomp.Emit(decomp.OpLogicalEqual(decomp.t_uint, gpr, target));
+ Id gpr = decomp.OpLoad(decomp.t_float, decomp.registers.at(expr.gpr));
+ gpr = decomp.OpBitcast(decomp.t_uint, gpr);
+ return decomp.OpIEqual(decomp.t_bool, gpr, target);
}
Id Visit(const Expr& node) {
@@ -1736,16 +2608,16 @@ public:
const Id condition = expr_parser.Visit(ast.condition);
const Id then_label = decomp.OpLabel();
const Id endif_label = decomp.OpLabel();
- decomp.Emit(decomp.OpSelectionMerge(endif_label, spv::SelectionControlMask::MaskNone));
- decomp.Emit(decomp.OpBranchConditional(condition, then_label, endif_label));
- decomp.Emit(then_label);
+ decomp.OpSelectionMerge(endif_label, spv::SelectionControlMask::MaskNone);
+ decomp.OpBranchConditional(condition, then_label, endif_label);
+ decomp.AddLabel(then_label);
ASTNode current = ast.nodes.GetFirst();
while (current) {
Visit(current);
current = current->GetNext();
}
- decomp.Emit(decomp.OpBranch(endif_label));
- decomp.Emit(endif_label);
+ decomp.OpBranch(endif_label);
+ decomp.AddLabel(endif_label);
}
void operator()([[maybe_unused]] const ASTIfElse& ast) {
@@ -1763,7 +2635,7 @@ public:
void operator()(const ASTVarSet& ast) {
ExprDecompiler expr_parser{decomp};
const Id condition = expr_parser.Visit(ast.condition);
- decomp.Emit(decomp.OpStore(decomp.flow_variables.at(ast.index), condition));
+ decomp.OpStore(decomp.flow_variables.at(ast.index), condition);
}
void operator()([[maybe_unused]] const ASTLabel& ast) {
@@ -1778,23 +2650,24 @@ public:
const Id loop_label = decomp.OpLabel();
const Id endloop_label = decomp.OpLabel();
const Id loop_start_block = decomp.OpLabel();
- const Id loop_end_block = decomp.OpLabel();
+ const Id loop_continue_block = decomp.OpLabel();
current_loop_exit = endloop_label;
- decomp.Emit(decomp.OpBranch(loop_label));
- decomp.Emit(loop_label);
- decomp.Emit(
- decomp.OpLoopMerge(endloop_label, loop_end_block, spv::LoopControlMask::MaskNone));
- decomp.Emit(decomp.OpBranch(loop_start_block));
- decomp.Emit(loop_start_block);
+ decomp.OpBranch(loop_label);
+ decomp.AddLabel(loop_label);
+ decomp.OpLoopMerge(endloop_label, loop_continue_block, spv::LoopControlMask::MaskNone);
+ decomp.OpBranch(loop_start_block);
+ decomp.AddLabel(loop_start_block);
ASTNode current = ast.nodes.GetFirst();
while (current) {
Visit(current);
current = current->GetNext();
}
+ decomp.OpBranch(loop_continue_block);
+ decomp.AddLabel(loop_continue_block);
ExprDecompiler expr_parser{decomp};
const Id condition = expr_parser.Visit(ast.condition);
- decomp.Emit(decomp.OpBranchConditional(condition, loop_label, endloop_label));
- decomp.Emit(endloop_label);
+ decomp.OpBranchConditional(condition, loop_label, endloop_label);
+ decomp.AddLabel(endloop_label);
}
void operator()(const ASTReturn& ast) {
@@ -1803,27 +2676,27 @@ public:
const Id condition = expr_parser.Visit(ast.condition);
const Id then_label = decomp.OpLabel();
const Id endif_label = decomp.OpLabel();
- decomp.Emit(decomp.OpSelectionMerge(endif_label, spv::SelectionControlMask::MaskNone));
- decomp.Emit(decomp.OpBranchConditional(condition, then_label, endif_label));
- decomp.Emit(then_label);
+ decomp.OpSelectionMerge(endif_label, spv::SelectionControlMask::MaskNone);
+ decomp.OpBranchConditional(condition, then_label, endif_label);
+ decomp.AddLabel(then_label);
if (ast.kills) {
- decomp.Emit(decomp.OpKill());
+ decomp.OpKill();
} else {
decomp.PreExit();
- decomp.Emit(decomp.OpReturn());
+ decomp.OpReturn();
}
- decomp.Emit(endif_label);
+ decomp.AddLabel(endif_label);
} else {
const Id next_block = decomp.OpLabel();
- decomp.Emit(decomp.OpBranch(next_block));
- decomp.Emit(next_block);
+ decomp.OpBranch(next_block);
+ decomp.AddLabel(next_block);
if (ast.kills) {
- decomp.Emit(decomp.OpKill());
+ decomp.OpKill();
} else {
decomp.PreExit();
- decomp.Emit(decomp.OpReturn());
+ decomp.OpReturn();
}
- decomp.Emit(decomp.OpLabel());
+ decomp.AddLabel(decomp.OpLabel());
}
}
@@ -1833,17 +2706,17 @@ public:
const Id condition = expr_parser.Visit(ast.condition);
const Id then_label = decomp.OpLabel();
const Id endif_label = decomp.OpLabel();
- decomp.Emit(decomp.OpSelectionMerge(endif_label, spv::SelectionControlMask::MaskNone));
- decomp.Emit(decomp.OpBranchConditional(condition, then_label, endif_label));
- decomp.Emit(then_label);
- decomp.Emit(decomp.OpBranch(current_loop_exit));
- decomp.Emit(endif_label);
+ decomp.OpSelectionMerge(endif_label, spv::SelectionControlMask::MaskNone);
+ decomp.OpBranchConditional(condition, then_label, endif_label);
+ decomp.AddLabel(then_label);
+ decomp.OpBranch(current_loop_exit);
+ decomp.AddLabel(endif_label);
} else {
const Id next_block = decomp.OpLabel();
- decomp.Emit(decomp.OpBranch(next_block));
- decomp.Emit(next_block);
- decomp.Emit(decomp.OpBranch(current_loop_exit));
- decomp.Emit(decomp.OpLabel());
+ decomp.OpBranch(next_block);
+ decomp.AddLabel(next_block);
+ decomp.OpBranch(current_loop_exit);
+ decomp.AddLabel(decomp.OpLabel());
}
}
@@ -1864,20 +2737,51 @@ void SPIRVDecompiler::DecompileAST() {
flow_variables.emplace(i, AddGlobalVariable(id));
}
+ DefinePrologue();
+
const ASTNode program = ir.GetASTProgram();
ASTDecompiler decompiler{*this};
decompiler.Visit(program);
const Id next_block = OpLabel();
- Emit(OpBranch(next_block));
- Emit(next_block);
+ OpBranch(next_block);
+ AddLabel(next_block);
+}
+
+} // Anonymous namespace
+
+ShaderEntries GenerateShaderEntries(const VideoCommon::Shader::ShaderIR& ir) {
+ ShaderEntries entries;
+ for (const auto& cbuf : ir.GetConstantBuffers()) {
+ entries.const_buffers.emplace_back(cbuf.second, cbuf.first);
+ }
+ for (const auto& [base, usage] : ir.GetGlobalMemory()) {
+ entries.global_buffers.emplace_back(base.cbuf_index, base.cbuf_offset, usage.is_written);
+ }
+ for (const auto& sampler : ir.GetSamplers()) {
+ if (sampler.IsBuffer()) {
+ entries.texel_buffers.emplace_back(sampler);
+ } else {
+ entries.samplers.emplace_back(sampler);
+ }
+ }
+ for (const auto& image : ir.GetImages()) {
+ entries.images.emplace_back(image);
+ }
+ for (const auto& attribute : ir.GetInputAttributes()) {
+ if (IsGenericAttribute(attribute)) {
+ entries.attributes.insert(GetGenericAttributeLocation(attribute));
+ }
+ }
+ entries.clip_distances = ir.GetClipDistances();
+ entries.shader_length = ir.GetLength();
+ entries.uses_warps = ir.UsesWarps();
+ return entries;
}
-DecompilerResult Decompile(const VKDevice& device, const VideoCommon::Shader::ShaderIR& ir,
- Maxwell::ShaderStage stage) {
- auto decompiler = std::make_unique<SPIRVDecompiler>(device, ir, stage);
- decompiler->Decompile();
- return {std::move(decompiler), decompiler->GetShaderEntries()};
+std::vector<u32> Decompile(const VKDevice& device, const VideoCommon::Shader::ShaderIR& ir,
+ ShaderType stage, const Specialization& specialization) {
+ return SPIRVDecompiler(device, ir, stage, specialization).Assemble();
}
-} // namespace Vulkan::VKShader
+} // namespace Vulkan
diff --git a/src/video_core/renderer_vulkan/vk_shader_decompiler.h b/src/video_core/renderer_vulkan/vk_shader_decompiler.h
index f90541cc1..10794be1c 100644
--- a/src/video_core/renderer_vulkan/vk_shader_decompiler.h
+++ b/src/video_core/renderer_vulkan/vk_shader_decompiler.h
@@ -5,29 +5,28 @@
#pragma once
#include <array>
+#include <bitset>
#include <memory>
#include <set>
+#include <type_traits>
#include <utility>
#include <vector>
-#include <sirit/sirit.h>
-
#include "common/common_types.h"
#include "video_core/engines/maxwell_3d.h"
+#include "video_core/engines/shader_type.h"
#include "video_core/shader/shader_ir.h"
-namespace VideoCommon::Shader {
-class ShaderIR;
-}
-
namespace Vulkan {
class VKDevice;
}
-namespace Vulkan::VKShader {
+namespace Vulkan {
using Maxwell = Tegra::Engines::Maxwell3D::Regs;
+using TexelBufferEntry = VideoCommon::Shader::Sampler;
using SamplerEntry = VideoCommon::Shader::Sampler;
+using ImageEntry = VideoCommon::Shader::Image;
constexpr u32 DESCRIPTOR_SET = 0;
@@ -46,39 +45,78 @@ private:
class GlobalBufferEntry {
public:
- explicit GlobalBufferEntry(u32 cbuf_index, u32 cbuf_offset)
- : cbuf_index{cbuf_index}, cbuf_offset{cbuf_offset} {}
+ constexpr explicit GlobalBufferEntry(u32 cbuf_index, u32 cbuf_offset, bool is_written)
+ : cbuf_index{cbuf_index}, cbuf_offset{cbuf_offset}, is_written{is_written} {}
- u32 GetCbufIndex() const {
+ constexpr u32 GetCbufIndex() const {
return cbuf_index;
}
- u32 GetCbufOffset() const {
+ constexpr u32 GetCbufOffset() const {
return cbuf_offset;
}
+ constexpr bool IsWritten() const {
+ return is_written;
+ }
+
private:
u32 cbuf_index{};
u32 cbuf_offset{};
+ bool is_written{};
};
struct ShaderEntries {
- u32 const_buffers_base_binding{};
- u32 global_buffers_base_binding{};
- u32 samplers_base_binding{};
+ u32 NumBindings() const {
+ return static_cast<u32>(const_buffers.size() + global_buffers.size() +
+ texel_buffers.size() + samplers.size() + images.size());
+ }
+
std::vector<ConstBufferEntry> const_buffers;
std::vector<GlobalBufferEntry> global_buffers;
+ std::vector<TexelBufferEntry> texel_buffers;
std::vector<SamplerEntry> samplers;
+ std::vector<ImageEntry> images;
std::set<u32> attributes;
std::array<bool, Maxwell::NumClipDistances> clip_distances{};
std::size_t shader_length{};
- Sirit::Id entry_function{};
- std::vector<Sirit::Id> interfaces;
+ bool uses_warps{};
+};
+
+struct Specialization final {
+ u32 base_binding{};
+
+ // Compute specific
+ std::array<u32, 3> workgroup_size{};
+ u32 shared_memory_size{};
+
+ // Graphics specific
+ Maxwell::PrimitiveTopology primitive_topology{};
+ std::optional<float> point_size{};
+ std::array<Maxwell::VertexAttribute::Type, Maxwell::NumVertexAttributes> attribute_types{};
+ bool ndc_minus_one_to_one{};
+
+ // Tessellation specific
+ struct {
+ Maxwell::TessellationPrimitive primitive{};
+ Maxwell::TessellationSpacing spacing{};
+ bool clockwise{};
+ } tessellation;
+
+ // Fragment specific
+ std::bitset<8> enabled_rendertargets;
+};
+// Old gcc versions don't consider this trivially copyable.
+// static_assert(std::is_trivially_copyable_v<Specialization>);
+
+struct SPIRVShader {
+ std::vector<u32> code;
+ ShaderEntries entries;
};
-using DecompilerResult = std::pair<std::unique_ptr<Sirit::Module>, ShaderEntries>;
+ShaderEntries GenerateShaderEntries(const VideoCommon::Shader::ShaderIR& ir);
-DecompilerResult Decompile(const VKDevice& device, const VideoCommon::Shader::ShaderIR& ir,
- Maxwell::ShaderStage stage);
+std::vector<u32> Decompile(const VKDevice& device, const VideoCommon::Shader::ShaderIR& ir,
+ Tegra::Engines::ShaderType stage, const Specialization& specialization);
-} // namespace Vulkan::VKShader
+} // namespace Vulkan
diff --git a/src/video_core/renderer_vulkan/vk_staging_buffer_pool.cpp b/src/video_core/renderer_vulkan/vk_staging_buffer_pool.cpp
new file mode 100644
index 000000000..171d78afc
--- /dev/null
+++ b/src/video_core/renderer_vulkan/vk_staging_buffer_pool.cpp
@@ -0,0 +1,127 @@
+// Copyright 2019 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <algorithm>
+#include <unordered_map>
+#include <utility>
+#include <vector>
+
+#include "common/bit_util.h"
+#include "common/common_types.h"
+#include "video_core/renderer_vulkan/vk_device.h"
+#include "video_core/renderer_vulkan/vk_resource_manager.h"
+#include "video_core/renderer_vulkan/vk_scheduler.h"
+#include "video_core/renderer_vulkan/vk_staging_buffer_pool.h"
+
+namespace Vulkan {
+
+VKStagingBufferPool::StagingBuffer::StagingBuffer(std::unique_ptr<VKBuffer> buffer, VKFence& fence,
+ u64 last_epoch)
+ : buffer{std::move(buffer)}, watch{fence}, last_epoch{last_epoch} {}
+
+VKStagingBufferPool::StagingBuffer::StagingBuffer(StagingBuffer&& rhs) noexcept {
+ buffer = std::move(rhs.buffer);
+ watch = std::move(rhs.watch);
+ last_epoch = rhs.last_epoch;
+}
+
+VKStagingBufferPool::StagingBuffer::~StagingBuffer() = default;
+
+VKStagingBufferPool::StagingBuffer& VKStagingBufferPool::StagingBuffer::operator=(
+ StagingBuffer&& rhs) noexcept {
+ buffer = std::move(rhs.buffer);
+ watch = std::move(rhs.watch);
+ last_epoch = rhs.last_epoch;
+ return *this;
+}
+
+VKStagingBufferPool::VKStagingBufferPool(const VKDevice& device, VKMemoryManager& memory_manager,
+ VKScheduler& scheduler)
+ : device{device}, memory_manager{memory_manager}, scheduler{scheduler},
+ is_device_integrated{device.IsIntegrated()} {}
+
+VKStagingBufferPool::~VKStagingBufferPool() = default;
+
+VKBuffer& VKStagingBufferPool::GetUnusedBuffer(std::size_t size, bool host_visible) {
+ if (const auto buffer = TryGetReservedBuffer(size, host_visible)) {
+ return *buffer;
+ }
+ return CreateStagingBuffer(size, host_visible);
+}
+
+void VKStagingBufferPool::TickFrame() {
+ ++epoch;
+ current_delete_level = (current_delete_level + 1) % NumLevels;
+
+ ReleaseCache(true);
+ if (!is_device_integrated) {
+ ReleaseCache(false);
+ }
+}
+
+VKBuffer* VKStagingBufferPool::TryGetReservedBuffer(std::size_t size, bool host_visible) {
+ for (auto& entry : GetCache(host_visible)[Common::Log2Ceil64(size)].entries) {
+ if (entry.watch.TryWatch(scheduler.GetFence())) {
+ entry.last_epoch = epoch;
+ return &*entry.buffer;
+ }
+ }
+ return nullptr;
+}
+
+VKBuffer& VKStagingBufferPool::CreateStagingBuffer(std::size_t size, bool host_visible) {
+ const auto usage =
+ vk::BufferUsageFlagBits::eTransferSrc | vk::BufferUsageFlagBits::eTransferDst |
+ vk::BufferUsageFlagBits::eStorageBuffer | vk::BufferUsageFlagBits::eIndexBuffer;
+ const u32 log2 = Common::Log2Ceil64(size);
+ const vk::BufferCreateInfo buffer_ci({}, 1ULL << log2, usage, vk::SharingMode::eExclusive, 0,
+ nullptr);
+ const auto dev = device.GetLogical();
+ auto buffer = std::make_unique<VKBuffer>();
+ buffer->handle = dev.createBufferUnique(buffer_ci, nullptr, device.GetDispatchLoader());
+ buffer->commit = memory_manager.Commit(*buffer->handle, host_visible);
+
+ auto& entries = GetCache(host_visible)[log2].entries;
+ return *entries.emplace_back(std::move(buffer), scheduler.GetFence(), epoch).buffer;
+}
+
+VKStagingBufferPool::StagingBuffersCache& VKStagingBufferPool::GetCache(bool host_visible) {
+ return is_device_integrated || host_visible ? host_staging_buffers : device_staging_buffers;
+}
+
+void VKStagingBufferPool::ReleaseCache(bool host_visible) {
+ auto& cache = GetCache(host_visible);
+ const u64 size = ReleaseLevel(cache, current_delete_level);
+ if (size == 0) {
+ return;
+ }
+}
+
+u64 VKStagingBufferPool::ReleaseLevel(StagingBuffersCache& cache, std::size_t log2) {
+ static constexpr u64 epochs_to_destroy = 180;
+ static constexpr std::size_t deletions_per_tick = 16;
+
+ auto& staging = cache[log2];
+ auto& entries = staging.entries;
+ const std::size_t old_size = entries.size();
+
+ const auto is_deleteable = [this](const auto& entry) {
+ return entry.last_epoch + epochs_to_destroy < epoch && !entry.watch.IsUsed();
+ };
+ const std::size_t begin_offset = staging.delete_index;
+ const std::size_t end_offset = std::min(begin_offset + deletions_per_tick, old_size);
+ const auto begin = std::begin(entries) + begin_offset;
+ const auto end = std::begin(entries) + end_offset;
+ entries.erase(std::remove_if(begin, end, is_deleteable), end);
+
+ const std::size_t new_size = entries.size();
+ staging.delete_index += deletions_per_tick;
+ if (staging.delete_index >= new_size) {
+ staging.delete_index = 0;
+ }
+
+ return (1ULL << log2) * (old_size - new_size);
+}
+
+} // namespace Vulkan
diff --git a/src/video_core/renderer_vulkan/vk_staging_buffer_pool.h b/src/video_core/renderer_vulkan/vk_staging_buffer_pool.h
new file mode 100644
index 000000000..02310375f
--- /dev/null
+++ b/src/video_core/renderer_vulkan/vk_staging_buffer_pool.h
@@ -0,0 +1,83 @@
+// Copyright 2019 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <climits>
+#include <unordered_map>
+#include <utility>
+#include <vector>
+
+#include "common/common_types.h"
+
+#include "video_core/renderer_vulkan/declarations.h"
+#include "video_core/renderer_vulkan/vk_memory_manager.h"
+
+namespace Vulkan {
+
+class VKDevice;
+class VKFenceWatch;
+class VKScheduler;
+
+struct VKBuffer final {
+ UniqueBuffer handle;
+ VKMemoryCommit commit;
+};
+
+class VKStagingBufferPool final {
+public:
+ explicit VKStagingBufferPool(const VKDevice& device, VKMemoryManager& memory_manager,
+ VKScheduler& scheduler);
+ ~VKStagingBufferPool();
+
+ VKBuffer& GetUnusedBuffer(std::size_t size, bool host_visible);
+
+ void TickFrame();
+
+private:
+ struct StagingBuffer final {
+ explicit StagingBuffer(std::unique_ptr<VKBuffer> buffer, VKFence& fence, u64 last_epoch);
+ StagingBuffer(StagingBuffer&& rhs) noexcept;
+ StagingBuffer(const StagingBuffer&) = delete;
+ ~StagingBuffer();
+
+ StagingBuffer& operator=(StagingBuffer&& rhs) noexcept;
+
+ std::unique_ptr<VKBuffer> buffer;
+ VKFenceWatch watch;
+ u64 last_epoch = 0;
+ };
+
+ struct StagingBuffers final {
+ std::vector<StagingBuffer> entries;
+ std::size_t delete_index = 0;
+ };
+
+ static constexpr std::size_t NumLevels = sizeof(std::size_t) * CHAR_BIT;
+ using StagingBuffersCache = std::array<StagingBuffers, NumLevels>;
+
+ VKBuffer* TryGetReservedBuffer(std::size_t size, bool host_visible);
+
+ VKBuffer& CreateStagingBuffer(std::size_t size, bool host_visible);
+
+ StagingBuffersCache& GetCache(bool host_visible);
+
+ void ReleaseCache(bool host_visible);
+
+ u64 ReleaseLevel(StagingBuffersCache& cache, std::size_t log2);
+
+ const VKDevice& device;
+ VKMemoryManager& memory_manager;
+ VKScheduler& scheduler;
+ const bool is_device_integrated;
+
+ StagingBuffersCache host_staging_buffers;
+ StagingBuffersCache device_staging_buffers;
+
+ u64 epoch = 0;
+
+ std::size_t current_delete_level = 0;
+};
+
+} // namespace Vulkan
diff --git a/src/video_core/renderer_vulkan/vk_swapchain.cpp b/src/video_core/renderer_vulkan/vk_swapchain.cpp
index 08279e562..ebc68f030 100644
--- a/src/video_core/renderer_vulkan/vk_swapchain.cpp
+++ b/src/video_core/renderer_vulkan/vk_swapchain.cpp
@@ -19,12 +19,18 @@
namespace Vulkan {
namespace {
-vk::SurfaceFormatKHR ChooseSwapSurfaceFormat(const std::vector<vk::SurfaceFormatKHR>& formats) {
+
+vk::SurfaceFormatKHR ChooseSwapSurfaceFormat(const std::vector<vk::SurfaceFormatKHR>& formats,
+ bool srgb) {
if (formats.size() == 1 && formats[0].format == vk::Format::eUndefined) {
- return {vk::Format::eB8G8R8A8Unorm, vk::ColorSpaceKHR::eSrgbNonlinear};
+ vk::SurfaceFormatKHR format;
+ format.format = vk::Format::eB8G8R8A8Unorm;
+ format.colorSpace = vk::ColorSpaceKHR::eSrgbNonlinear;
+ return format;
}
- const auto& found = std::find_if(formats.begin(), formats.end(), [](const auto& format) {
- return format.format == vk::Format::eB8G8R8A8Unorm &&
+ const auto& found = std::find_if(formats.begin(), formats.end(), [srgb](const auto& format) {
+ const auto request_format = srgb ? vk::Format::eB8G8R8A8Srgb : vk::Format::eB8G8R8A8Unorm;
+ return format.format == request_format &&
format.colorSpace == vk::ColorSpaceKHR::eSrgbNonlinear;
});
return found != formats.end() ? *found : formats[0];
@@ -51,28 +57,26 @@ vk::Extent2D ChooseSwapExtent(const vk::SurfaceCapabilitiesKHR& capabilities, u3
std::min(capabilities.maxImageExtent.height, extent.height));
return extent;
}
-} // namespace
+
+} // Anonymous namespace
VKSwapchain::VKSwapchain(vk::SurfaceKHR surface, const VKDevice& device)
: surface{surface}, device{device} {}
VKSwapchain::~VKSwapchain() = default;
-void VKSwapchain::Create(u32 width, u32 height) {
- const auto dev = device.GetLogical();
+void VKSwapchain::Create(u32 width, u32 height, bool srgb) {
const auto& dld = device.GetDispatchLoader();
const auto physical_device = device.GetPhysical();
-
- const vk::SurfaceCapabilitiesKHR capabilities{
- physical_device.getSurfaceCapabilitiesKHR(surface, dld)};
+ const auto capabilities{physical_device.getSurfaceCapabilitiesKHR(surface, dld)};
if (capabilities.maxImageExtent.width == 0 || capabilities.maxImageExtent.height == 0) {
return;
}
- dev.waitIdle(dld);
+ device.GetLogical().waitIdle(dld);
Destroy();
- CreateSwapchain(capabilities, width, height);
+ CreateSwapchain(capabilities, width, height, srgb);
CreateSemaphores();
CreateImageViews();
@@ -107,7 +111,7 @@ bool VKSwapchain::Present(vk::Semaphore render_semaphore, VKFence& fence) {
break;
case vk::Result::eErrorOutOfDateKHR:
if (current_width > 0 && current_height > 0) {
- Create(current_width, current_height);
+ Create(current_width, current_height, current_srgb);
recreated = true;
}
break;
@@ -129,23 +133,19 @@ bool VKSwapchain::HasFramebufferChanged(const Layout::FramebufferLayout& framebu
}
void VKSwapchain::CreateSwapchain(const vk::SurfaceCapabilitiesKHR& capabilities, u32 width,
- u32 height) {
- const auto dev{device.GetLogical()};
+ u32 height, bool srgb) {
const auto& dld{device.GetDispatchLoader()};
const auto physical_device{device.GetPhysical()};
+ const auto formats{physical_device.getSurfaceFormatsKHR(surface, dld)};
+ const auto present_modes{physical_device.getSurfacePresentModesKHR(surface, dld)};
- const std::vector<vk::SurfaceFormatKHR> formats{
- physical_device.getSurfaceFormatsKHR(surface, dld)};
-
- const std::vector<vk::PresentModeKHR> present_modes{
- physical_device.getSurfacePresentModesKHR(surface, dld)};
-
- const vk::SurfaceFormatKHR surface_format{ChooseSwapSurfaceFormat(formats)};
+ const vk::SurfaceFormatKHR surface_format{ChooseSwapSurfaceFormat(formats, srgb)};
const vk::PresentModeKHR present_mode{ChooseSwapPresentMode(present_modes)};
extent = ChooseSwapExtent(capabilities, width, height);
current_width = extent.width;
current_height = extent.height;
+ current_srgb = srgb;
u32 requested_image_count{capabilities.minImageCount + 1};
if (capabilities.maxImageCount > 0 && requested_image_count > capabilities.maxImageCount) {
@@ -169,6 +169,7 @@ void VKSwapchain::CreateSwapchain(const vk::SurfaceCapabilitiesKHR& capabilities
swapchain_ci.imageSharingMode = vk::SharingMode::eExclusive;
}
+ const auto dev{device.GetLogical()};
swapchain = dev.createSwapchainKHRUnique(swapchain_ci, nullptr, dld);
images = dev.getSwapchainImagesKHR(*swapchain, dld);
diff --git a/src/video_core/renderer_vulkan/vk_swapchain.h b/src/video_core/renderer_vulkan/vk_swapchain.h
index 2ad84f185..a1e7938d2 100644
--- a/src/video_core/renderer_vulkan/vk_swapchain.h
+++ b/src/video_core/renderer_vulkan/vk_swapchain.h
@@ -24,7 +24,7 @@ public:
~VKSwapchain();
/// Creates (or recreates) the swapchain with a given size.
- void Create(u32 width, u32 height);
+ void Create(u32 width, u32 height, bool srgb);
/// Acquires the next image in the swapchain, waits as needed.
void AcquireNextImage();
@@ -60,8 +60,13 @@ public:
return image_format;
}
+ bool GetSrgbState() const {
+ return current_srgb;
+ }
+
private:
- void CreateSwapchain(const vk::SurfaceCapabilitiesKHR& capabilities, u32 width, u32 height);
+ void CreateSwapchain(const vk::SurfaceCapabilitiesKHR& capabilities, u32 width, u32 height,
+ bool srgb);
void CreateSemaphores();
void CreateImageViews();
@@ -87,6 +92,7 @@ private:
u32 current_width{};
u32 current_height{};
+ bool current_srgb{};
};
} // namespace Vulkan
diff --git a/src/video_core/shader/const_buffer_locker.cpp b/src/video_core/shader/const_buffer_locker.cpp
index fe467608e..a4a0319eb 100644
--- a/src/video_core/shader/const_buffer_locker.cpp
+++ b/src/video_core/shader/const_buffer_locker.cpp
@@ -2,13 +2,12 @@
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
-#pragma once
-
#include <algorithm>
-#include <memory>
-#include "common/assert.h"
+#include <tuple>
+
#include "common/common_types.h"
#include "video_core/engines/maxwell_3d.h"
+#include "video_core/engines/shader_type.h"
#include "video_core/shader/const_buffer_locker.h"
namespace VideoCommon::Shader {
@@ -103,8 +102,8 @@ bool ConstBufferLocker::IsConsistent() const {
}
bool ConstBufferLocker::HasEqualKeys(const ConstBufferLocker& rhs) const {
- return keys == rhs.keys && bound_samplers == rhs.bound_samplers &&
- bindless_samplers == rhs.bindless_samplers;
+ return std::tie(keys, bound_samplers, bindless_samplers) ==
+ std::tie(rhs.keys, rhs.bound_samplers, rhs.bindless_samplers);
}
} // namespace VideoCommon::Shader
diff --git a/src/video_core/shader/const_buffer_locker.h b/src/video_core/shader/const_buffer_locker.h
index 600e2f3c3..d32e2d657 100644
--- a/src/video_core/shader/const_buffer_locker.h
+++ b/src/video_core/shader/const_buffer_locker.h
@@ -4,10 +4,12 @@
#pragma once
+#include <optional>
#include <unordered_map>
#include "common/common_types.h"
#include "common/hash.h"
#include "video_core/engines/const_buffer_engine_interface.h"
+#include "video_core/engines/shader_type.h"
namespace VideoCommon::Shader {
@@ -20,7 +22,7 @@ using BindlessSamplerMap =
* The ConstBufferLocker is a class use to interface the 3D and compute engines with the shader
* compiler. with it, the shader can obtain required data from GPU state and store it for disk
* shader compilation.
- **/
+ */
class ConstBufferLocker {
public:
explicit ConstBufferLocker(Tegra::Engines::ShaderType shader_stage);
diff --git a/src/video_core/shader/control_flow.cpp b/src/video_core/shader/control_flow.cpp
index d47c63d9f..b427ac873 100644
--- a/src/video_core/shader/control_flow.cpp
+++ b/src/video_core/shader/control_flow.cpp
@@ -16,7 +16,9 @@
#include "video_core/shader/shader_ir.h"
namespace VideoCommon::Shader {
+
namespace {
+
using Tegra::Shader::Instruction;
using Tegra::Shader::OpCode;
@@ -68,15 +70,15 @@ struct CFGRebuildState {
const ProgramCode& program_code;
ConstBufferLocker& locker;
u32 start{};
- std::vector<BlockInfo> block_info{};
- std::list<u32> inspect_queries{};
- std::list<Query> queries{};
- std::unordered_map<u32, u32> registered{};
- std::set<u32> labels{};
- std::map<u32, u32> ssy_labels{};
- std::map<u32, u32> pbk_labels{};
- std::unordered_map<u32, BlockStack> stacks{};
- ASTManager* manager;
+ std::vector<BlockInfo> block_info;
+ std::list<u32> inspect_queries;
+ std::list<Query> queries;
+ std::unordered_map<u32, u32> registered;
+ std::set<u32> labels;
+ std::map<u32, u32> ssy_labels;
+ std::map<u32, u32> pbk_labels;
+ std::unordered_map<u32, BlockStack> stacks;
+ ASTManager* manager{};
};
enum class BlockCollision : u32 { None, Found, Inside };
@@ -109,7 +111,7 @@ BlockInfo& CreateBlockInfo(CFGRebuildState& state, u32 start, u32 end) {
}
Pred GetPredicate(u32 index, bool negated) {
- return static_cast<Pred>(index + (negated ? 8 : 0));
+ return static_cast<Pred>(static_cast<u64>(index) + (negated ? 8ULL : 0ULL));
}
/**
@@ -136,15 +138,13 @@ struct BranchIndirectInfo {
s32 relative_position{};
};
-std::optional<BranchIndirectInfo> TrackBranchIndirectInfo(const CFGRebuildState& state,
- u32 start_address, u32 current_position) {
- const u32 shader_start = state.start;
- u32 pos = current_position;
- BranchIndirectInfo result{};
- u64 track_register = 0;
+struct BufferInfo {
+ u32 index;
+ u32 offset;
+};
- // Step 0 Get BRX Info
- const Instruction instr = {state.program_code[pos]};
+std::optional<std::pair<s32, u64>> GetBRXInfo(const CFGRebuildState& state, u32& pos) {
+ const Instruction instr = state.program_code[pos];
const auto opcode = OpCode::Decode(instr);
if (opcode->get().GetId() != OpCode::Id::BRX) {
return std::nullopt;
@@ -152,86 +152,94 @@ std::optional<BranchIndirectInfo> TrackBranchIndirectInfo(const CFGRebuildState&
if (instr.brx.constant_buffer != 0) {
return std::nullopt;
}
- track_register = instr.gpr8.Value();
- result.relative_position = instr.brx.GetBranchExtend();
- pos--;
- bool found_track = false;
+ --pos;
+ return std::make_pair(instr.brx.GetBranchExtend(), instr.gpr8.Value());
+}
- // Step 1 Track LDC
- while (pos >= shader_start) {
- if (IsSchedInstruction(pos, shader_start)) {
- pos--;
+template <typename Result, typename TestCallable, typename PackCallable>
+// requires std::predicate<TestCallable, Instruction, const OpCode::Matcher&>
+// requires std::invocable<PackCallable, Instruction, const OpCode::Matcher&>
+std::optional<Result> TrackInstruction(const CFGRebuildState& state, u32& pos, TestCallable test,
+ PackCallable pack) {
+ for (; pos >= state.start; --pos) {
+ if (IsSchedInstruction(pos, state.start)) {
continue;
}
- const Instruction instr = {state.program_code[pos]};
+ const Instruction instr = state.program_code[pos];
const auto opcode = OpCode::Decode(instr);
- if (opcode->get().GetId() == OpCode::Id::LD_C) {
- if (instr.gpr0.Value() == track_register &&
- instr.ld_c.type.Value() == Tegra::Shader::UniformType::Single) {
- result.buffer = instr.cbuf36.index.Value();
- result.offset = static_cast<u32>(instr.cbuf36.GetOffset());
- track_register = instr.gpr8.Value();
- pos--;
- found_track = true;
- break;
- }
+ if (!opcode) {
+ continue;
+ }
+ if (test(instr, opcode->get())) {
+ --pos;
+ return std::make_optional(pack(instr, opcode->get()));
}
- pos--;
}
+ return std::nullopt;
+}
- if (!found_track) {
- return std::nullopt;
- }
- found_track = false;
+std::optional<std::pair<BufferInfo, u64>> TrackLDC(const CFGRebuildState& state, u32& pos,
+ u64 brx_tracked_register) {
+ return TrackInstruction<std::pair<BufferInfo, u64>>(
+ state, pos,
+ [brx_tracked_register](auto instr, const auto& opcode) {
+ return opcode.GetId() == OpCode::Id::LD_C &&
+ instr.gpr0.Value() == brx_tracked_register &&
+ instr.ld_c.type.Value() == Tegra::Shader::UniformType::Single;
+ },
+ [](auto instr, const auto& opcode) {
+ const BufferInfo info = {static_cast<u32>(instr.cbuf36.index.Value()),
+ static_cast<u32>(instr.cbuf36.GetOffset())};
+ return std::make_pair(info, instr.gpr8.Value());
+ });
+}
- // Step 2 Track SHL
- while (pos >= shader_start) {
- if (IsSchedInstruction(pos, shader_start)) {
- pos--;
- continue;
- }
- const Instruction instr = state.program_code[pos];
- const auto opcode = OpCode::Decode(instr);
- if (opcode->get().GetId() == OpCode::Id::SHL_IMM) {
- if (instr.gpr0.Value() == track_register) {
- track_register = instr.gpr8.Value();
- pos--;
- found_track = true;
- break;
- }
- }
- pos--;
+std::optional<u64> TrackSHLRegister(const CFGRebuildState& state, u32& pos,
+ u64 ldc_tracked_register) {
+ return TrackInstruction<u64>(state, pos,
+ [ldc_tracked_register](auto instr, const auto& opcode) {
+ return opcode.GetId() == OpCode::Id::SHL_IMM &&
+ instr.gpr0.Value() == ldc_tracked_register;
+ },
+ [](auto instr, const auto&) { return instr.gpr8.Value(); });
+}
+
+std::optional<u32> TrackIMNMXValue(const CFGRebuildState& state, u32& pos,
+ u64 shl_tracked_register) {
+ return TrackInstruction<u32>(state, pos,
+ [shl_tracked_register](auto instr, const auto& opcode) {
+ return opcode.GetId() == OpCode::Id::IMNMX_IMM &&
+ instr.gpr0.Value() == shl_tracked_register;
+ },
+ [](auto instr, const auto&) {
+ return static_cast<u32>(instr.alu.GetSignedImm20_20() + 1);
+ });
+}
+
+std::optional<BranchIndirectInfo> TrackBranchIndirectInfo(const CFGRebuildState& state, u32 pos) {
+ const auto brx_info = GetBRXInfo(state, pos);
+ if (!brx_info) {
+ return std::nullopt;
}
+ const auto [relative_position, brx_tracked_register] = *brx_info;
- if (!found_track) {
+ const auto ldc_info = TrackLDC(state, pos, brx_tracked_register);
+ if (!ldc_info) {
return std::nullopt;
}
- found_track = false;
+ const auto [buffer_info, ldc_tracked_register] = *ldc_info;
- // Step 3 Track IMNMX
- while (pos >= shader_start) {
- if (IsSchedInstruction(pos, shader_start)) {
- pos--;
- continue;
- }
- const Instruction instr = state.program_code[pos];
- const auto opcode = OpCode::Decode(instr);
- if (opcode->get().GetId() == OpCode::Id::IMNMX_IMM) {
- if (instr.gpr0.Value() == track_register) {
- track_register = instr.gpr8.Value();
- result.entries = instr.alu.GetSignedImm20_20() + 1;
- pos--;
- found_track = true;
- break;
- }
- }
- pos--;
+ const auto shl_tracked_register = TrackSHLRegister(state, pos, ldc_tracked_register);
+ if (!shl_tracked_register) {
+ return std::nullopt;
}
- if (!found_track) {
+ const auto entries = TrackIMNMXValue(state, pos, *shl_tracked_register);
+ if (!entries) {
return std::nullopt;
}
- return result;
+
+ return BranchIndirectInfo{buffer_info.index, buffer_info.offset, *entries, relative_position};
}
std::pair<ParseResult, ParseInfo> ParseCode(CFGRebuildState& state, u32 address) {
@@ -420,30 +428,30 @@ std::pair<ParseResult, ParseInfo> ParseCode(CFGRebuildState& state, u32 address)
break;
}
case OpCode::Id::BRX: {
- auto tmp = TrackBranchIndirectInfo(state, address, offset);
- if (tmp) {
- auto result = *tmp;
- std::vector<CaseBranch> branches{};
- s32 pc_target = offset + result.relative_position;
- for (u32 i = 0; i < result.entries; i++) {
- auto k = state.locker.ObtainKey(result.buffer, result.offset + i * 4);
- if (!k) {
- return {ParseResult::AbnormalFlow, parse_info};
- }
- u32 value = *k;
- u32 target = static_cast<u32>((value >> 3) + pc_target);
- insert_label(state, target);
- branches.emplace_back(value, target);
- }
- parse_info.end_address = offset;
- parse_info.branch_info = MakeBranchInfo<MultiBranch>(
- static_cast<u32>(instr.gpr8.Value()), std::move(branches));
-
- return {ParseResult::ControlCaught, parse_info};
- } else {
+ const auto tmp = TrackBranchIndirectInfo(state, offset);
+ if (!tmp) {
LOG_WARNING(HW_GPU, "BRX Track Unsuccesful");
+ return {ParseResult::AbnormalFlow, parse_info};
}
- return {ParseResult::AbnormalFlow, parse_info};
+
+ const auto result = *tmp;
+ const s32 pc_target = offset + result.relative_position;
+ std::vector<CaseBranch> branches;
+ for (u32 i = 0; i < result.entries; i++) {
+ auto key = state.locker.ObtainKey(result.buffer, result.offset + i * 4);
+ if (!key) {
+ return {ParseResult::AbnormalFlow, parse_info};
+ }
+ u32 value = *key;
+ u32 target = static_cast<u32>((value >> 3) + pc_target);
+ insert_label(state, target);
+ branches.emplace_back(value, target);
+ }
+ parse_info.end_address = offset;
+ parse_info.branch_info = MakeBranchInfo<MultiBranch>(
+ static_cast<u32>(instr.gpr8.Value()), std::move(branches));
+
+ return {ParseResult::ControlCaught, parse_info};
}
default:
break;
diff --git a/src/video_core/shader/decode.cpp b/src/video_core/shader/decode.cpp
index 21fb9cb83..22c3e5120 100644
--- a/src/video_core/shader/decode.cpp
+++ b/src/video_core/shader/decode.cpp
@@ -154,10 +154,10 @@ void ShaderIR::Decode() {
LOG_CRITICAL(HW_GPU, "Unknown decompilation mode!");
[[fallthrough]];
case CompileDepth::BruteForce: {
+ const auto shader_end = static_cast<u32>(program_code.size());
coverage_begin = main_offset;
- const std::size_t shader_end = program_code.size();
coverage_end = shader_end;
- for (u32 label = main_offset; label < shader_end; label++) {
+ for (u32 label = main_offset; label < shader_end; ++label) {
basic_blocks.insert({label, DecodeRange(label, label + 1)});
}
break;
diff --git a/src/video_core/shader/decode/arithmetic.cpp b/src/video_core/shader/decode/arithmetic.cpp
index 1473c282a..fcedd2af6 100644
--- a/src/video_core/shader/decode/arithmetic.cpp
+++ b/src/video_core/shader/decode/arithmetic.cpp
@@ -43,12 +43,12 @@ u32 ShaderIR::DecodeArithmetic(NodeBlock& bb, u32 pc) {
case OpCode::Id::FMUL_IMM: {
// FMUL does not have 'abs' bits and only the second operand has a 'neg' bit.
if (instr.fmul.tab5cb8_2 != 0) {
- LOG_WARNING(HW_GPU, "FMUL tab5cb8_2({}) is not implemented",
- instr.fmul.tab5cb8_2.Value());
+ LOG_DEBUG(HW_GPU, "FMUL tab5cb8_2({}) is not implemented",
+ instr.fmul.tab5cb8_2.Value());
}
if (instr.fmul.tab5c68_0 != 1) {
- LOG_WARNING(HW_GPU, "FMUL tab5cb8_0({}) is not implemented",
- instr.fmul.tab5c68_0.Value());
+ LOG_DEBUG(HW_GPU, "FMUL tab5cb8_0({}) is not implemented",
+ instr.fmul.tab5c68_0.Value());
}
op_b = GetOperandAbsNegFloat(op_b, false, instr.fmul.negate_b);
@@ -144,10 +144,11 @@ u32 ShaderIR::DecodeArithmetic(NodeBlock& bb, u32 pc) {
case OpCode::Id::RRO_C:
case OpCode::Id::RRO_R:
case OpCode::Id::RRO_IMM: {
+ LOG_DEBUG(HW_GPU, "(STUBBED) RRO used");
+
// Currently RRO is only implemented as a register move.
op_b = GetOperandAbsNegFloat(op_b, instr.alu.abs_b, instr.alu.negate_b);
SetRegister(bb, instr.gpr0, op_b);
- LOG_WARNING(HW_GPU, "RRO instruction is incomplete");
break;
}
default:
diff --git a/src/video_core/shader/decode/arithmetic_half.cpp b/src/video_core/shader/decode/arithmetic_half.cpp
index b06cbe441..ee7d9a29d 100644
--- a/src/video_core/shader/decode/arithmetic_half.cpp
+++ b/src/video_core/shader/decode/arithmetic_half.cpp
@@ -21,8 +21,8 @@ u32 ShaderIR::DecodeArithmeticHalf(NodeBlock& bb, u32 pc) {
if (opcode->get().GetId() == OpCode::Id::HADD2_C ||
opcode->get().GetId() == OpCode::Id::HADD2_R) {
- if (instr.alu_half.ftz != 0) {
- LOG_WARNING(HW_GPU, "{} FTZ not implemented", opcode->get().GetName());
+ if (instr.alu_half.ftz == 0) {
+ LOG_DEBUG(HW_GPU, "{} without FTZ is not implemented", opcode->get().GetName());
}
}
diff --git a/src/video_core/shader/decode/arithmetic_half_immediate.cpp b/src/video_core/shader/decode/arithmetic_half_immediate.cpp
index 6466fc011..d179b9873 100644
--- a/src/video_core/shader/decode/arithmetic_half_immediate.cpp
+++ b/src/video_core/shader/decode/arithmetic_half_immediate.cpp
@@ -19,12 +19,12 @@ u32 ShaderIR::DecodeArithmeticHalfImmediate(NodeBlock& bb, u32 pc) {
const auto opcode = OpCode::Decode(instr);
if (opcode->get().GetId() == OpCode::Id::HADD2_IMM) {
- if (instr.alu_half_imm.ftz != 0) {
- LOG_WARNING(HW_GPU, "{} FTZ not implemented", opcode->get().GetName());
+ if (instr.alu_half_imm.ftz == 0) {
+ LOG_DEBUG(HW_GPU, "{} without FTZ is not implemented", opcode->get().GetName());
}
} else {
- if (instr.alu_half_imm.precision != Tegra::Shader::HalfPrecision::None) {
- LOG_WARNING(HW_GPU, "{} FTZ not implemented", opcode->get().GetName());
+ if (instr.alu_half_imm.precision != Tegra::Shader::HalfPrecision::FTZ) {
+ LOG_DEBUG(HW_GPU, "{} without FTZ is not implemented", opcode->get().GetName());
}
}
diff --git a/src/video_core/shader/decode/arithmetic_integer.cpp b/src/video_core/shader/decode/arithmetic_integer.cpp
index a33d242e9..371fae127 100644
--- a/src/video_core/shader/decode/arithmetic_integer.cpp
+++ b/src/video_core/shader/decode/arithmetic_integer.cpp
@@ -130,6 +130,25 @@ u32 ShaderIR::DecodeArithmeticInteger(NodeBlock& bb, u32 pc) {
SetRegister(bb, instr.gpr0, value);
break;
}
+ case OpCode::Id::FLO_R:
+ case OpCode::Id::FLO_C:
+ case OpCode::Id::FLO_IMM: {
+ Node value;
+ if (instr.flo.invert) {
+ op_b = Operation(OperationCode::IBitwiseNot, NO_PRECISE, std::move(op_b));
+ }
+ if (instr.flo.is_signed) {
+ value = Operation(OperationCode::IBitMSB, NO_PRECISE, std::move(op_b));
+ } else {
+ value = Operation(OperationCode::UBitMSB, NO_PRECISE, std::move(op_b));
+ }
+ if (instr.flo.sh) {
+ value =
+ Operation(OperationCode::UBitwiseXor, NO_PRECISE, std::move(value), Immediate(31));
+ }
+ SetRegister(bb, instr.gpr0, std::move(value));
+ break;
+ }
case OpCode::Id::SEL_C:
case OpCode::Id::SEL_R:
case OpCode::Id::SEL_IMM: {
diff --git a/src/video_core/shader/decode/conversion.cpp b/src/video_core/shader/decode/conversion.cpp
index 32facd6ba..0eeb75559 100644
--- a/src/video_core/shader/decode/conversion.cpp
+++ b/src/video_core/shader/decode/conversion.cpp
@@ -63,12 +63,11 @@ u32 ShaderIR::DecodeConversion(NodeBlock& bb, u32 pc) {
case OpCode::Id::I2F_R:
case OpCode::Id::I2F_C:
case OpCode::Id::I2F_IMM: {
- UNIMPLEMENTED_IF(instr.conversion.int_src.selector != 0);
UNIMPLEMENTED_IF(instr.conversion.dst_size == Register::Size::Long);
UNIMPLEMENTED_IF_MSG(instr.generates_cc,
"Condition codes generation in I2F is not implemented");
- Node value = [&]() {
+ Node value = [&] {
switch (opcode->get().GetId()) {
case OpCode::Id::I2F_R:
return GetRegister(instr.gpr20);
@@ -81,7 +80,19 @@ u32 ShaderIR::DecodeConversion(NodeBlock& bb, u32 pc) {
return Immediate(0);
}
}();
+
const bool input_signed = instr.conversion.is_input_signed;
+
+ if (instr.conversion.src_size == Register::Size::Byte) {
+ const u32 offset = static_cast<u32>(instr.conversion.int_src.selector) * 8;
+ if (offset > 0) {
+ value = SignedOperation(OperationCode::ILogicalShiftRight, input_signed,
+ std::move(value), Immediate(offset));
+ }
+ } else {
+ UNIMPLEMENTED_IF(instr.conversion.int_src.selector != 0);
+ }
+
value = ConvertIntegerSize(value, instr.conversion.src_size, input_signed);
value = GetOperandAbsNegInteger(value, instr.conversion.abs_a, false, input_signed);
value = SignedOperation(OperationCode::FCastInteger, input_signed, PRECISE, value);
diff --git a/src/video_core/shader/decode/ffma.cpp b/src/video_core/shader/decode/ffma.cpp
index ca2f39e8d..5973588d6 100644
--- a/src/video_core/shader/decode/ffma.cpp
+++ b/src/video_core/shader/decode/ffma.cpp
@@ -19,10 +19,10 @@ u32 ShaderIR::DecodeFfma(NodeBlock& bb, u32 pc) {
UNIMPLEMENTED_IF_MSG(instr.ffma.cc != 0, "FFMA cc not implemented");
if (instr.ffma.tab5980_0 != 1) {
- LOG_WARNING(HW_GPU, "FFMA tab5980_0({}) not implemented", instr.ffma.tab5980_0.Value());
+ LOG_DEBUG(HW_GPU, "FFMA tab5980_0({}) not implemented", instr.ffma.tab5980_0.Value());
}
if (instr.ffma.tab5980_1 != 0) {
- LOG_WARNING(HW_GPU, "FFMA tab5980_1({}) not implemented", instr.ffma.tab5980_1.Value());
+ LOG_DEBUG(HW_GPU, "FFMA tab5980_1({}) not implemented", instr.ffma.tab5980_1.Value());
}
const Node op_a = GetRegister(instr.gpr8);
diff --git a/src/video_core/shader/decode/half_set.cpp b/src/video_core/shader/decode/half_set.cpp
index 48ca7a4af..848e46874 100644
--- a/src/video_core/shader/decode/half_set.cpp
+++ b/src/video_core/shader/decode/half_set.cpp
@@ -20,8 +20,8 @@ u32 ShaderIR::DecodeHalfSet(NodeBlock& bb, u32 pc) {
const Instruction instr = {program_code[pc]};
const auto opcode = OpCode::Decode(instr);
- if (instr.hset2.ftz != 0) {
- LOG_WARNING(HW_GPU, "{} FTZ not implemented", opcode->get().GetName());
+ if (instr.hset2.ftz == 0) {
+ LOG_DEBUG(HW_GPU, "{} without FTZ is not implemented", opcode->get().GetName());
}
Node op_a = UnpackHalfFloat(GetRegister(instr.gpr8), instr.hset2.type_a);
diff --git a/src/video_core/shader/decode/half_set_predicate.cpp b/src/video_core/shader/decode/half_set_predicate.cpp
index fec8f2dbe..310655619 100644
--- a/src/video_core/shader/decode/half_set_predicate.cpp
+++ b/src/video_core/shader/decode/half_set_predicate.cpp
@@ -19,7 +19,9 @@ u32 ShaderIR::DecodeHalfSetPredicate(NodeBlock& bb, u32 pc) {
const Instruction instr = {program_code[pc]};
const auto opcode = OpCode::Decode(instr);
- LOG_DEBUG(HW_GPU, "ftz={}", static_cast<u32>(instr.hsetp2.ftz));
+ if (instr.hsetp2.ftz != 0) {
+ LOG_DEBUG(HW_GPU, "{} without FTZ is not implemented", opcode->get().GetName());
+ }
Node op_a = UnpackHalfFloat(GetRegister(instr.gpr8), instr.hsetp2.type_a);
op_a = GetOperandAbsNegHalf(op_a, instr.hsetp2.abs_a, instr.hsetp2.negate_a);
diff --git a/src/video_core/shader/decode/memory.cpp b/src/video_core/shader/decode/memory.cpp
index 335d78146..c934d0719 100644
--- a/src/video_core/shader/decode/memory.cpp
+++ b/src/video_core/shader/decode/memory.cpp
@@ -21,8 +21,10 @@ using Tegra::Shader::OpCode;
using Tegra::Shader::Register;
namespace {
-u32 GetUniformTypeElementsCount(Tegra::Shader::UniformType uniform_type) {
+
+u32 GetLdgMemorySize(Tegra::Shader::UniformType uniform_type) {
switch (uniform_type) {
+ case Tegra::Shader::UniformType::UnsignedByte:
case Tegra::Shader::UniformType::Single:
return 1;
case Tegra::Shader::UniformType::Double:
@@ -35,6 +37,22 @@ u32 GetUniformTypeElementsCount(Tegra::Shader::UniformType uniform_type) {
return 1;
}
}
+
+u32 GetStgMemorySize(Tegra::Shader::UniformType uniform_type) {
+ switch (uniform_type) {
+ case Tegra::Shader::UniformType::Single:
+ return 1;
+ case Tegra::Shader::UniformType::Double:
+ return 2;
+ case Tegra::Shader::UniformType::Quad:
+ case Tegra::Shader::UniformType::UnsignedQuad:
+ return 4;
+ default:
+ UNIMPLEMENTED_MSG("Unimplemented size={}!", static_cast<u32>(uniform_type));
+ return 1;
+ }
+}
+
} // Anonymous namespace
u32 ShaderIR::DecodeMemory(NodeBlock& bb, u32 pc) {
@@ -168,7 +186,7 @@ u32 ShaderIR::DecodeMemory(NodeBlock& bb, u32 pc) {
const auto [real_address_base, base_address, descriptor] =
TrackGlobalMemory(bb, instr, false);
- const u32 count = GetUniformTypeElementsCount(type);
+ const u32 count = GetLdgMemorySize(type);
if (!real_address_base || !base_address) {
// Tracking failed, load zeroes.
for (u32 i = 0; i < count; ++i) {
@@ -179,12 +197,22 @@ u32 ShaderIR::DecodeMemory(NodeBlock& bb, u32 pc) {
for (u32 i = 0; i < count; ++i) {
const Node it_offset = Immediate(i * 4);
- const Node real_address =
- Operation(OperationCode::UAdd, NO_PRECISE, real_address_base, it_offset);
- const Node gmem = MakeNode<GmemNode>(real_address, base_address, descriptor);
+ const Node real_address = Operation(OperationCode::UAdd, real_address_base, it_offset);
+ Node gmem = MakeNode<GmemNode>(real_address, base_address, descriptor);
+
+ if (type == Tegra::Shader::UniformType::UnsignedByte) {
+ // To handle unaligned loads get the byte used to dereferenced global memory
+ // and extract that byte from the loaded uint32.
+ Node byte = Operation(OperationCode::UBitwiseAnd, real_address, Immediate(3));
+ byte = Operation(OperationCode::ULogicalShiftLeft, std::move(byte), Immediate(3));
+
+ gmem = Operation(OperationCode::UBitfieldExtract, std::move(gmem), std::move(byte),
+ Immediate(8));
+ }
SetTemporary(bb, i, gmem);
}
+
for (u32 i = 0; i < count; ++i) {
SetRegister(bb, instr.gpr0.Value() + i, GetTemporary(i));
}
@@ -196,28 +224,28 @@ u32 ShaderIR::DecodeMemory(NodeBlock& bb, u32 pc) {
UNIMPLEMENTED_IF_MSG((instr.attribute.fmt20.immediate.Value() % sizeof(u32)) != 0,
"Unaligned attribute loads are not supported");
- u64 next_element = instr.attribute.fmt20.element;
- auto next_index = static_cast<u64>(instr.attribute.fmt20.index.Value());
+ u64 element = instr.attribute.fmt20.element;
+ auto index = static_cast<u64>(instr.attribute.fmt20.index.Value());
- const auto StoreNextElement = [&](u32 reg_offset) {
- const auto dest = GetOutputAttribute(static_cast<Attribute::Index>(next_index),
- next_element, GetRegister(instr.gpr39));
+ const u32 num_words = static_cast<u32>(instr.attribute.fmt20.size.Value()) + 1;
+ for (u32 reg_offset = 0; reg_offset < num_words; ++reg_offset) {
+ Node dest;
+ if (instr.attribute.fmt20.patch) {
+ const u32 offset = static_cast<u32>(index) * 4 + static_cast<u32>(element);
+ dest = MakeNode<PatchNode>(offset);
+ } else {
+ dest = GetOutputAttribute(static_cast<Attribute::Index>(index), element,
+ GetRegister(instr.gpr39));
+ }
const auto src = GetRegister(instr.gpr0.Value() + reg_offset);
bb.push_back(Operation(OperationCode::Assign, dest, src));
- // Load the next attribute element into the following register. If the element
- // to load goes beyond the vec4 size, load the first element of the next
- // attribute.
- next_element = (next_element + 1) % 4;
- next_index = next_index + (next_element == 0 ? 1 : 0);
- };
-
- const u32 num_words = static_cast<u32>(instr.attribute.fmt20.size.Value()) + 1;
- for (u32 reg_offset = 0; reg_offset < num_words; ++reg_offset) {
- StoreNextElement(reg_offset);
+ // Load the next attribute element into the following register. If the element to load
+ // goes beyond the vec4 size, load the first element of the next attribute.
+ element = (element + 1) % 4;
+ index = index + (element == 0 ? 1 : 0);
}
-
break;
}
case OpCode::Id::ST_L:
@@ -274,7 +302,7 @@ u32 ShaderIR::DecodeMemory(NodeBlock& bb, u32 pc) {
break;
}
- const u32 count = GetUniformTypeElementsCount(type);
+ const u32 count = GetStgMemorySize(type);
for (u32 i = 0; i < count; ++i) {
const Node it_offset = Immediate(i * 4);
const Node real_address = Operation(OperationCode::UAdd, real_address_base, it_offset);
diff --git a/src/video_core/shader/decode/other.cpp b/src/video_core/shader/decode/other.cpp
index 116b95f76..7321698b2 100644
--- a/src/video_core/shader/decode/other.cpp
+++ b/src/video_core/shader/decode/other.cpp
@@ -69,6 +69,8 @@ u32 ShaderIR::DecodeOther(NodeBlock& bb, u32 pc) {
case OpCode::Id::MOV_SYS: {
const Node value = [this, instr] {
switch (instr.sys20) {
+ case SystemVariable::InvocationId:
+ return Operation(OperationCode::InvocationId);
case SystemVariable::Ydirection:
return Operation(OperationCode::YNegate);
case SystemVariable::InvocationInfo:
@@ -255,8 +257,14 @@ u32 ShaderIR::DecodeOther(NodeBlock& bb, u32 pc) {
SetRegister(bb, instr.gpr0, GetRegister(instr.gpr8));
break;
}
+ case OpCode::Id::MEMBAR: {
+ UNIMPLEMENTED_IF(instr.membar.type != Tegra::Shader::MembarType::GL);
+ UNIMPLEMENTED_IF(instr.membar.unknown != Tegra::Shader::MembarUnknown::Default);
+ bb.push_back(Operation(OperationCode::MemoryBarrierGL));
+ break;
+ }
case OpCode::Id::DEPBAR: {
- LOG_WARNING(HW_GPU, "DEPBAR instruction is stubbed");
+ LOG_DEBUG(HW_GPU, "DEPBAR instruction is stubbed");
break;
}
default:
diff --git a/src/video_core/shader/decode/register_set_predicate.cpp b/src/video_core/shader/decode/register_set_predicate.cpp
index e6c9d287e..8d54cce34 100644
--- a/src/video_core/shader/decode/register_set_predicate.cpp
+++ b/src/video_core/shader/decode/register_set_predicate.cpp
@@ -13,37 +13,65 @@ namespace VideoCommon::Shader {
using Tegra::Shader::Instruction;
using Tegra::Shader::OpCode;
+namespace {
+constexpr u64 NUM_PROGRAMMABLE_PREDICATES = 7;
+}
+
u32 ShaderIR::DecodeRegisterSetPredicate(NodeBlock& bb, u32 pc) {
const Instruction instr = {program_code[pc]};
const auto opcode = OpCode::Decode(instr);
- UNIMPLEMENTED_IF(instr.r2p.mode != Tegra::Shader::R2pMode::Pr);
+ UNIMPLEMENTED_IF(instr.p2r_r2p.mode != Tegra::Shader::R2pMode::Pr);
- const Node apply_mask = [&]() {
+ const Node apply_mask = [&] {
switch (opcode->get().GetId()) {
case OpCode::Id::R2P_IMM:
- return Immediate(static_cast<u32>(instr.r2p.immediate_mask));
+ case OpCode::Id::P2R_IMM:
+ return Immediate(static_cast<u32>(instr.p2r_r2p.immediate_mask));
default:
UNREACHABLE();
- return Immediate(static_cast<u32>(instr.r2p.immediate_mask));
+ return Immediate(0);
}
}();
- const Node mask = GetRegister(instr.gpr8);
- const auto offset = static_cast<u32>(instr.r2p.byte) * 8;
- constexpr u32 programmable_preds = 7;
- for (u64 pred = 0; pred < programmable_preds; ++pred) {
- const auto shift = static_cast<u32>(pred);
+ const auto offset = static_cast<u32>(instr.p2r_r2p.byte) * 8;
+
+ switch (opcode->get().GetId()) {
+ case OpCode::Id::R2P_IMM: {
+ const Node mask = GetRegister(instr.gpr8);
- const Node apply_compare = BitfieldExtract(apply_mask, shift, 1);
- const Node condition =
- Operation(OperationCode::LogicalUNotEqual, apply_compare, Immediate(0));
+ for (u64 pred = 0; pred < NUM_PROGRAMMABLE_PREDICATES; ++pred) {
+ const auto shift = static_cast<u32>(pred);
- const Node value_compare = BitfieldExtract(mask, offset + shift, 1);
- const Node value = Operation(OperationCode::LogicalUNotEqual, value_compare, Immediate(0));
+ const Node apply_compare = BitfieldExtract(apply_mask, shift, 1);
+ const Node condition =
+ Operation(OperationCode::LogicalUNotEqual, apply_compare, Immediate(0));
- const Node code = Operation(OperationCode::LogicalAssign, GetPredicate(pred), value);
- bb.push_back(Conditional(condition, {code}));
+ const Node value_compare = BitfieldExtract(mask, offset + shift, 1);
+ const Node value =
+ Operation(OperationCode::LogicalUNotEqual, value_compare, Immediate(0));
+
+ const Node code = Operation(OperationCode::LogicalAssign, GetPredicate(pred), value);
+ bb.push_back(Conditional(condition, {code}));
+ }
+ break;
+ }
+ case OpCode::Id::P2R_IMM: {
+ Node value = Immediate(0);
+ for (u64 pred = 0; pred < NUM_PROGRAMMABLE_PREDICATES; ++pred) {
+ Node bit = Operation(OperationCode::Select, GetPredicate(pred), Immediate(1U << pred),
+ Immediate(0));
+ value = Operation(OperationCode::UBitwiseOr, std::move(value), std::move(bit));
+ }
+ value = Operation(OperationCode::UBitwiseAnd, std::move(value), apply_mask);
+ value = BitfieldInsert(GetRegister(instr.gpr8), std::move(value), offset, 8);
+
+ SetRegister(bb, instr.gpr0, std::move(value));
+ break;
+ }
+ default:
+ UNIMPLEMENTED_MSG("Unhandled P2R/R2R instruction: {}", opcode->get().GetName());
+ break;
}
return pc;
diff --git a/src/video_core/shader/decode/texture.cpp b/src/video_core/shader/decode/texture.cpp
index ca690b58b..4b14cdf58 100644
--- a/src/video_core/shader/decode/texture.cpp
+++ b/src/video_core/shader/decode/texture.cpp
@@ -44,10 +44,6 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
bool is_bindless = false;
switch (opcode->get().GetId()) {
case OpCode::Id::TEX: {
- if (instr.tex.UsesMiscMode(TextureMiscMode::NODEP)) {
- LOG_WARNING(HW_GPU, "TEX.NODEP implementation is incomplete");
- }
-
const TextureType texture_type{instr.tex.texture_type};
const bool is_array = instr.tex.array != 0;
const bool is_aoffi = instr.tex.UsesMiscMode(TextureMiscMode::AOFFI);
@@ -62,10 +58,6 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
UNIMPLEMENTED_IF_MSG(instr.tex.UsesMiscMode(TextureMiscMode::AOFFI),
"AOFFI is not implemented");
- if (instr.tex.UsesMiscMode(TextureMiscMode::NODEP)) {
- LOG_WARNING(HW_GPU, "TEX.NODEP implementation is incomplete");
- }
-
const TextureType texture_type{instr.tex_b.texture_type};
const bool is_array = instr.tex_b.array != 0;
const bool is_aoffi = instr.tex.UsesMiscMode(TextureMiscMode::AOFFI);
@@ -82,10 +74,6 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
const bool depth_compare = instr.texs.UsesMiscMode(TextureMiscMode::DC);
const auto process_mode = instr.texs.GetTextureProcessMode();
- if (instr.texs.UsesMiscMode(TextureMiscMode::NODEP)) {
- LOG_WARNING(HW_GPU, "TEXS.NODEP implementation is incomplete");
- }
-
const Node4 components =
GetTexsCode(instr, texture_type, process_mode, depth_compare, is_array);
@@ -101,78 +89,142 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
[[fallthrough]];
}
case OpCode::Id::TLD4: {
- ASSERT(instr.tld4.array == 0);
UNIMPLEMENTED_IF_MSG(instr.tld4.UsesMiscMode(TextureMiscMode::NDV),
"NDV is not implemented");
- UNIMPLEMENTED_IF_MSG(instr.tld4.UsesMiscMode(TextureMiscMode::PTP),
- "PTP is not implemented");
-
- if (instr.tld4.UsesMiscMode(TextureMiscMode::NODEP)) {
- LOG_WARNING(HW_GPU, "TLD4.NODEP implementation is incomplete");
- }
-
const auto texture_type = instr.tld4.texture_type.Value();
const bool depth_compare = is_bindless ? instr.tld4_b.UsesMiscMode(TextureMiscMode::DC)
: instr.tld4.UsesMiscMode(TextureMiscMode::DC);
const bool is_array = instr.tld4.array != 0;
const bool is_aoffi = is_bindless ? instr.tld4_b.UsesMiscMode(TextureMiscMode::AOFFI)
: instr.tld4.UsesMiscMode(TextureMiscMode::AOFFI);
- WriteTexInstructionFloat(
- bb, instr,
- GetTld4Code(instr, texture_type, depth_compare, is_array, is_aoffi, is_bindless));
+ const bool is_ptp = is_bindless ? instr.tld4_b.UsesMiscMode(TextureMiscMode::PTP)
+ : instr.tld4.UsesMiscMode(TextureMiscMode::PTP);
+ WriteTexInstructionFloat(bb, instr,
+ GetTld4Code(instr, texture_type, depth_compare, is_array, is_aoffi,
+ is_ptp, is_bindless));
break;
}
case OpCode::Id::TLD4S: {
- UNIMPLEMENTED_IF_MSG(instr.tld4s.UsesMiscMode(TextureMiscMode::AOFFI),
- "AOFFI is not implemented");
- if (instr.tld4s.UsesMiscMode(TextureMiscMode::NODEP)) {
- LOG_WARNING(HW_GPU, "TLD4S.NODEP implementation is incomplete");
- }
-
- const bool depth_compare = instr.tld4s.UsesMiscMode(TextureMiscMode::DC);
+ constexpr std::size_t num_coords = 2;
+ const bool is_aoffi = instr.tld4s.UsesMiscMode(TextureMiscMode::AOFFI);
+ const bool is_depth_compare = instr.tld4s.UsesMiscMode(TextureMiscMode::DC);
const Node op_a = GetRegister(instr.gpr8);
const Node op_b = GetRegister(instr.gpr20);
// TODO(Subv): Figure out how the sampler type is encoded in the TLD4S instruction.
std::vector<Node> coords;
- if (depth_compare) {
+ std::vector<Node> aoffi;
+ Node depth_compare;
+ if (is_depth_compare) {
// Note: TLD4S coordinate encoding works just like TEXS's
const Node op_y = GetRegister(instr.gpr8.Value() + 1);
coords.push_back(op_a);
coords.push_back(op_y);
- coords.push_back(op_b);
+ if (is_aoffi) {
+ aoffi = GetAoffiCoordinates(op_b, num_coords, true);
+ depth_compare = GetRegister(instr.gpr20.Value() + 1);
+ } else {
+ depth_compare = op_b;
+ }
} else {
+ // There's no depth compare
coords.push_back(op_a);
- coords.push_back(op_b);
+ if (is_aoffi) {
+ coords.push_back(GetRegister(instr.gpr8.Value() + 1));
+ aoffi = GetAoffiCoordinates(op_b, num_coords, true);
+ } else {
+ coords.push_back(op_b);
+ }
}
const Node component = Immediate(static_cast<u32>(instr.tld4s.component));
- const auto& sampler =
- GetSampler(instr.sampler, {{TextureType::Texture2D, false, depth_compare}});
+ const SamplerInfo info{TextureType::Texture2D, false, is_depth_compare};
+ const Sampler& sampler = *GetSampler(instr.sampler, info);
Node4 values;
for (u32 element = 0; element < values.size(); ++element) {
auto coords_copy = coords;
- MetaTexture meta{sampler, {}, {}, {}, {}, {}, component, element};
+ MetaTexture meta{sampler, {}, depth_compare, aoffi, {}, {}, {}, {}, component, element};
values[element] = Operation(OperationCode::TextureGather, meta, std::move(coords_copy));
}
- WriteTexsInstructionFloat(bb, instr, values, true);
+ if (instr.tld4s.fp16_flag) {
+ WriteTexsInstructionHalfFloat(bb, instr, values, true);
+ } else {
+ WriteTexsInstructionFloat(bb, instr, values, true);
+ }
break;
}
- case OpCode::Id::TXQ_B:
+ case OpCode::Id::TXD_B:
is_bindless = true;
[[fallthrough]];
- case OpCode::Id::TXQ: {
- if (instr.txq.UsesMiscMode(TextureMiscMode::NODEP)) {
- LOG_WARNING(HW_GPU, "TXQ.NODEP implementation is incomplete");
+ case OpCode::Id::TXD: {
+ UNIMPLEMENTED_IF_MSG(instr.txd.UsesMiscMode(TextureMiscMode::AOFFI),
+ "AOFFI is not implemented");
+ UNIMPLEMENTED_IF_MSG(instr.txd.is_array != 0, "TXD Array is not implemented");
+
+ u64 base_reg = instr.gpr8.Value();
+ const auto derivate_reg = instr.gpr20.Value();
+ const auto texture_type = instr.txd.texture_type.Value();
+ const auto coord_count = GetCoordCount(texture_type);
+
+ const Sampler* sampler = is_bindless
+ ? GetBindlessSampler(base_reg, {{texture_type, false, false}})
+ : GetSampler(instr.sampler, {{texture_type, false, false}});
+ Node4 values;
+ if (sampler == nullptr) {
+ for (u32 element = 0; element < values.size(); ++element) {
+ values[element] = Immediate(0);
+ }
+ WriteTexInstructionFloat(bb, instr, values);
+ break;
+ }
+ if (is_bindless) {
+ base_reg++;
}
+ std::vector<Node> coords;
+ std::vector<Node> derivates;
+ for (std::size_t i = 0; i < coord_count; ++i) {
+ coords.push_back(GetRegister(base_reg + i));
+ const std::size_t derivate = i * 2;
+ derivates.push_back(GetRegister(derivate_reg + derivate));
+ derivates.push_back(GetRegister(derivate_reg + derivate + 1));
+ }
+
+ for (u32 element = 0; element < values.size(); ++element) {
+ MetaTexture meta{*sampler, {}, {}, {}, {}, derivates, {}, {}, {}, element};
+ values[element] = Operation(OperationCode::TextureGradient, std::move(meta), coords);
+ }
+
+ WriteTexInstructionFloat(bb, instr, values);
+
+ break;
+ }
+ case OpCode::Id::TXQ_B:
+ is_bindless = true;
+ [[fallthrough]];
+ case OpCode::Id::TXQ: {
// TODO: The new commits on the texture refactor, change the way samplers work.
// Sadly, not all texture instructions specify the type of texture their sampler
// uses. This must be fixed at a later instance.
- const auto& sampler =
- is_bindless ? GetBindlessSampler(instr.gpr8, {}) : GetSampler(instr.sampler, {});
+ const Sampler* sampler =
+ is_bindless ? GetBindlessSampler(instr.gpr8) : GetSampler(instr.sampler);
+
+ if (sampler == nullptr) {
+ u32 indexer = 0;
+ for (u32 element = 0; element < 4; ++element) {
+ if (!instr.txq.IsComponentEnabled(element)) {
+ continue;
+ }
+ const Node value = Immediate(0);
+ SetTemporary(bb, indexer++, value);
+ }
+ for (u32 i = 0; i < indexer; ++i) {
+ SetRegister(bb, instr.gpr0.Value() + i, GetTemporary(i));
+ }
+ break;
+ }
u32 indexer = 0;
switch (instr.txq.query_type) {
@@ -181,7 +233,7 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
if (!instr.txq.IsComponentEnabled(element)) {
continue;
}
- MetaTexture meta{sampler, {}, {}, {}, {}, {}, {}, element};
+ MetaTexture meta{*sampler, {}, {}, {}, {}, {}, {}, {}, {}, element};
const Node value =
Operation(OperationCode::TextureQueryDimensions, meta,
GetRegister(instr.gpr8.Value() + (is_bindless ? 1 : 0)));
@@ -205,15 +257,25 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
UNIMPLEMENTED_IF_MSG(instr.tmml.UsesMiscMode(Tegra::Shader::TextureMiscMode::NDV),
"NDV is not implemented");
- if (instr.tmml.UsesMiscMode(TextureMiscMode::NODEP)) {
- LOG_WARNING(HW_GPU, "TMML.NODEP implementation is incomplete");
- }
-
auto texture_type = instr.tmml.texture_type.Value();
const bool is_array = instr.tmml.array != 0;
- const auto& sampler =
- is_bindless ? GetBindlessSampler(instr.gpr20, {{texture_type, is_array, false}})
- : GetSampler(instr.sampler, {{texture_type, is_array, false}});
+ const Sampler* sampler =
+ is_bindless ? GetBindlessSampler(instr.gpr20) : GetSampler(instr.sampler);
+
+ if (sampler == nullptr) {
+ u32 indexer = 0;
+ for (u32 element = 0; element < 2; ++element) {
+ if (!instr.tmml.IsComponentEnabled(element)) {
+ continue;
+ }
+ const Node value = Immediate(0);
+ SetTemporary(bb, indexer++, value);
+ }
+ for (u32 i = 0; i < indexer; ++i) {
+ SetRegister(bb, instr.gpr0.Value() + i, GetTemporary(i));
+ }
+ break;
+ }
std::vector<Node> coords;
@@ -240,7 +302,7 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
continue;
}
auto params = coords;
- MetaTexture meta{sampler, {}, {}, {}, {}, {}, {}, element};
+ MetaTexture meta{*sampler, {}, {}, {}, {}, {}, {}, {}, {}, element};
const Node value = Operation(OperationCode::TextureQueryLod, meta, std::move(params));
SetTemporary(bb, indexer++, value);
}
@@ -254,10 +316,6 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
UNIMPLEMENTED_IF_MSG(instr.tld.ms, "MS is not implemented");
UNIMPLEMENTED_IF_MSG(instr.tld.cl, "CL is not implemented");
- if (instr.tld.nodep_flag) {
- LOG_WARNING(HW_GPU, "TLD.NODEP implementation is incomplete");
- }
-
WriteTexInstructionFloat(bb, instr, GetTldCode(instr));
break;
}
@@ -269,10 +327,6 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
"AOFFI is not implemented");
UNIMPLEMENTED_IF_MSG(instr.tlds.UsesMiscMode(TextureMiscMode::MZ), "MZ is not implemented");
- if (instr.tlds.UsesMiscMode(TextureMiscMode::NODEP)) {
- LOG_WARNING(HW_GPU, "TLDS.NODEP implementation is incomplete");
- }
-
const Node4 components = GetTldsCode(instr, texture_type, is_array);
if (instr.tlds.fp32_flag) {
@@ -289,68 +343,54 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
return pc;
}
-const Sampler& ShaderIR::GetSampler(const Tegra::Shader::Sampler& sampler,
- std::optional<SamplerInfo> sampler_info) {
- const auto offset = static_cast<u32>(sampler.index.Value());
-
- TextureType type;
- bool is_array;
- bool is_shadow;
+ShaderIR::SamplerInfo ShaderIR::GetSamplerInfo(std::optional<SamplerInfo> sampler_info, u32 offset,
+ std::optional<u32> buffer) {
if (sampler_info) {
- type = sampler_info->type;
- is_array = sampler_info->is_array;
- is_shadow = sampler_info->is_shadow;
- } else if (const auto sampler = locker.ObtainBoundSampler(offset)) {
- type = sampler->texture_type.Value();
- is_array = sampler->is_array.Value() != 0;
- is_shadow = sampler->is_shadow.Value() != 0;
- } else {
+ return *sampler_info;
+ }
+ const auto sampler =
+ buffer ? locker.ObtainBindlessSampler(*buffer, offset) : locker.ObtainBoundSampler(offset);
+ if (!sampler) {
LOG_WARNING(HW_GPU, "Unknown sampler info");
- type = TextureType::Texture2D;
- is_array = false;
- is_shadow = false;
+ return SamplerInfo{TextureType::Texture2D, false, false, false};
}
+ return SamplerInfo{sampler->texture_type, sampler->is_array != 0, sampler->is_shadow != 0,
+ sampler->is_buffer != 0};
+}
+
+const Sampler* ShaderIR::GetSampler(const Tegra::Shader::Sampler& sampler,
+ std::optional<SamplerInfo> sampler_info) {
+ const auto offset = static_cast<u32>(sampler.index.Value());
+ const auto info = GetSamplerInfo(sampler_info, offset);
// If this sampler has already been used, return the existing mapping.
const auto it =
std::find_if(used_samplers.begin(), used_samplers.end(),
[offset](const Sampler& entry) { return entry.GetOffset() == offset; });
if (it != used_samplers.end()) {
- ASSERT(!it->IsBindless() && it->GetType() == type && it->IsArray() == is_array &&
- it->IsShadow() == is_shadow);
- return *it;
+ ASSERT(!it->IsBindless() && it->GetType() == info.type && it->IsArray() == info.is_array &&
+ it->IsShadow() == info.is_shadow && it->IsBuffer() == info.is_buffer);
+ return &*it;
}
// Otherwise create a new mapping for this sampler
const auto next_index = static_cast<u32>(used_samplers.size());
- return used_samplers.emplace_back(Sampler(next_index, offset, type, is_array, is_shadow));
+ return &used_samplers.emplace_back(next_index, offset, info.type, info.is_array, info.is_shadow,
+ info.is_buffer);
}
-const Sampler& ShaderIR::GetBindlessSampler(const Tegra::Shader::Register& reg,
+const Sampler* ShaderIR::GetBindlessSampler(Tegra::Shader::Register reg,
std::optional<SamplerInfo> sampler_info) {
const Node sampler_register = GetRegister(reg);
const auto [base_sampler, buffer, offset] =
TrackCbuf(sampler_register, global_code, static_cast<s64>(global_code.size()));
ASSERT(base_sampler != nullptr);
-
- TextureType type;
- bool is_array;
- bool is_shadow;
- if (sampler_info) {
- type = sampler_info->type;
- is_array = sampler_info->is_array;
- is_shadow = sampler_info->is_shadow;
- } else if (const auto sampler = locker.ObtainBindlessSampler(buffer, offset)) {
- type = sampler->texture_type.Value();
- is_array = sampler->is_array.Value() != 0;
- is_shadow = sampler->is_shadow.Value() != 0;
- } else {
- LOG_WARNING(HW_GPU, "Unknown sampler info");
- type = TextureType::Texture2D;
- is_array = false;
- is_shadow = false;
+ if (base_sampler == nullptr) {
+ return nullptr;
}
+ const auto info = GetSamplerInfo(sampler_info, offset, buffer);
+
// If this sampler has already been used, return the existing mapping.
const auto it =
std::find_if(used_samplers.begin(), used_samplers.end(),
@@ -358,15 +398,15 @@ const Sampler& ShaderIR::GetBindlessSampler(const Tegra::Shader::Register& reg,
return entry.GetBuffer() == buffer && entry.GetOffset() == offset;
});
if (it != used_samplers.end()) {
- ASSERT(it->IsBindless() && it->GetType() == type && it->IsArray() == is_array &&
- it->IsShadow() == is_shadow);
- return *it;
+ ASSERT(it->IsBindless() && it->GetType() == info.type && it->IsArray() == info.is_array &&
+ it->IsShadow() == info.is_shadow);
+ return &*it;
}
// Otherwise create a new mapping for this sampler
const auto next_index = static_cast<u32>(used_samplers.size());
- return used_samplers.emplace_back(
- Sampler(next_index, offset, buffer, type, is_array, is_shadow));
+ return &used_samplers.emplace_back(next_index, offset, buffer, info.type, info.is_array,
+ info.is_shadow, info.is_buffer);
}
void ShaderIR::WriteTexInstructionFloat(NodeBlock& bb, Instruction instr, const Node4& components) {
@@ -409,14 +449,14 @@ void ShaderIR::WriteTexsInstructionFloat(NodeBlock& bb, Instruction instr, const
}
void ShaderIR::WriteTexsInstructionHalfFloat(NodeBlock& bb, Instruction instr,
- const Node4& components) {
+ const Node4& components, bool ignore_mask) {
// TEXS.F16 destionation registers are packed in two registers in pairs (just like any half
// float instruction).
Node4 values;
u32 dest_elem = 0;
for (u32 component = 0; component < 4; ++component) {
- if (!instr.texs.IsComponentEnabled(component))
+ if (!instr.texs.IsComponentEnabled(component) && !ignore_mask)
continue;
values[dest_elem++] = components[component];
}
@@ -451,17 +491,23 @@ Node4 ShaderIR::GetTextureCode(Instruction instr, TextureType texture_type,
(texture_type == TextureType::TextureCube && is_array && is_shadow),
"This method is not supported.");
- const auto& sampler =
- is_bindless ? GetBindlessSampler(*bindless_reg, {{texture_type, is_array, is_shadow}})
- : GetSampler(instr.sampler, {{texture_type, is_array, is_shadow}});
+ const SamplerInfo info{texture_type, is_array, is_shadow, false};
+ const Sampler* sampler =
+ is_bindless ? GetBindlessSampler(*bindless_reg, info) : GetSampler(instr.sampler, info);
+ Node4 values;
+ if (sampler == nullptr) {
+ for (u32 element = 0; element < values.size(); ++element) {
+ values[element] = Immediate(0);
+ }
+ return values;
+ }
const bool lod_needed = process_mode == TextureProcessMode::LZ ||
process_mode == TextureProcessMode::LL ||
process_mode == TextureProcessMode::LLA;
- // LOD selection (either via bias or explicit textureLod) not
- // supported in GL for sampler2DArrayShadow and
- // samplerCubeArrayShadow.
+ // LOD selection (either via bias or explicit textureLod) not supported in GL for
+ // sampler2DArrayShadow and samplerCubeArrayShadow.
const bool gl_lod_supported =
!((texture_type == Tegra::Shader::TextureType::Texture2D && is_array && is_shadow) ||
(texture_type == Tegra::Shader::TextureType::TextureCube && is_array && is_shadow));
@@ -471,8 +517,8 @@ Node4 ShaderIR::GetTextureCode(Instruction instr, TextureType texture_type,
UNIMPLEMENTED_IF(process_mode != TextureProcessMode::None && !gl_lod_supported);
- Node bias = {};
- Node lod = {};
+ Node bias;
+ Node lod;
if (process_mode != TextureProcessMode::None && gl_lod_supported) {
switch (process_mode) {
case TextureProcessMode::LZ:
@@ -493,10 +539,9 @@ Node4 ShaderIR::GetTextureCode(Instruction instr, TextureType texture_type,
}
}
- Node4 values;
for (u32 element = 0; element < values.size(); ++element) {
auto copy_coords = coords;
- MetaTexture meta{sampler, array, depth_compare, aoffi, bias, lod, {}, element};
+ MetaTexture meta{*sampler, array, depth_compare, aoffi, {}, {}, bias, lod, {}, element};
values[element] = Operation(read_method, meta, std::move(copy_coords));
}
@@ -593,7 +638,9 @@ Node4 ShaderIR::GetTexsCode(Instruction instr, TextureType texture_type,
}
Node4 ShaderIR::GetTld4Code(Instruction instr, TextureType texture_type, bool depth_compare,
- bool is_array, bool is_aoffi, bool is_bindless) {
+ bool is_array, bool is_aoffi, bool is_ptp, bool is_bindless) {
+ ASSERT_MSG(!(is_aoffi && is_ptp), "AOFFI and PTP can't be enabled at the same time");
+
const std::size_t coord_count = GetCoordCount(texture_type);
// If enabled arrays index is always stored in the gpr8 field
@@ -608,17 +655,26 @@ Node4 ShaderIR::GetTld4Code(Instruction instr, TextureType texture_type, bool de
u64 parameter_register = instr.gpr20.Value();
- const auto& sampler =
- is_bindless
- ? GetBindlessSampler(parameter_register++, {{texture_type, is_array, depth_compare}})
- : GetSampler(instr.sampler, {{texture_type, is_array, depth_compare}});
+ const SamplerInfo info{texture_type, is_array, depth_compare, false};
+ const Sampler* sampler = is_bindless ? GetBindlessSampler(parameter_register++, info)
+ : GetSampler(instr.sampler, info);
+ Node4 values;
+ if (sampler == nullptr) {
+ for (u32 element = 0; element < values.size(); ++element) {
+ values[element] = Immediate(0);
+ }
+ return values;
+ }
- std::vector<Node> aoffi;
+ std::vector<Node> aoffi, ptp;
if (is_aoffi) {
aoffi = GetAoffiCoordinates(GetRegister(parameter_register++), coord_count, true);
+ } else if (is_ptp) {
+ ptp = GetPtpCoordinates(
+ {GetRegister(parameter_register++), GetRegister(parameter_register++)});
}
- Node dc{};
+ Node dc;
if (depth_compare) {
dc = GetRegister(parameter_register++);
}
@@ -626,11 +682,10 @@ Node4 ShaderIR::GetTld4Code(Instruction instr, TextureType texture_type, bool de
const Node component = is_bindless ? Immediate(static_cast<u32>(instr.tld4_b.component))
: Immediate(static_cast<u32>(instr.tld4.component));
- Node4 values;
for (u32 element = 0; element < values.size(); ++element) {
auto coords_copy = coords;
- MetaTexture meta{sampler, GetRegister(array_register), dc, aoffi, {}, {}, component,
- element};
+ MetaTexture meta{
+ *sampler, GetRegister(array_register), dc, aoffi, ptp, {}, {}, {}, component, element};
values[element] = Operation(OperationCode::TextureGather, meta, std::move(coords_copy));
}
@@ -658,12 +713,12 @@ Node4 ShaderIR::GetTldCode(Tegra::Shader::Instruction instr) {
// const Node aoffi_register{is_aoffi ? GetRegister(gpr20_cursor++) : nullptr};
// const Node multisample{is_multisample ? GetRegister(gpr20_cursor++) : nullptr};
- const auto& sampler = GetSampler(instr.sampler, {{texture_type, is_array, false}});
+ const auto& sampler = *GetSampler(instr.sampler);
Node4 values;
for (u32 element = 0; element < values.size(); ++element) {
auto coords_copy = coords;
- MetaTexture meta{sampler, array_register, {}, {}, {}, lod, {}, element};
+ MetaTexture meta{sampler, array_register, {}, {}, {}, {}, {}, lod, {}, element};
values[element] = Operation(OperationCode::TexelFetch, meta, std::move(coords_copy));
}
@@ -671,6 +726,8 @@ Node4 ShaderIR::GetTldCode(Tegra::Shader::Instruction instr) {
}
Node4 ShaderIR::GetTldsCode(Instruction instr, TextureType texture_type, bool is_array) {
+ const Sampler& sampler = *GetSampler(instr.sampler);
+
const std::size_t type_coord_count = GetCoordCount(texture_type);
const bool lod_enabled = instr.tlds.GetTextureProcessMode() == TextureProcessMode::LL;
@@ -694,12 +751,24 @@ Node4 ShaderIR::GetTldsCode(Instruction instr, TextureType texture_type, bool is
// When lod is used always is in gpr20
const Node lod = lod_enabled ? GetRegister(instr.gpr20) : Immediate(0);
- const auto& sampler = GetSampler(instr.sampler, {{texture_type, is_array, false}});
+ // Fill empty entries from the guest sampler
+ const std::size_t entry_coord_count = GetCoordCount(sampler.GetType());
+ if (type_coord_count != entry_coord_count) {
+ LOG_WARNING(HW_GPU, "Bound and built texture types mismatch");
+
+ // When the size is higher we insert zeroes
+ for (std::size_t i = type_coord_count; i < entry_coord_count; ++i) {
+ coords.push_back(GetRegister(Register::ZeroIndex));
+ }
+
+ // Then we ensure the size matches the number of entries (dropping unused values)
+ coords.resize(entry_coord_count);
+ }
Node4 values;
for (u32 element = 0; element < values.size(); ++element) {
auto coords_copy = coords;
- MetaTexture meta{sampler, array, {}, {}, {}, lod, {}, element};
+ MetaTexture meta{sampler, array, {}, {}, {}, {}, {}, lod, {}, element};
values[element] = Operation(OperationCode::TexelFetch, meta, std::move(coords_copy));
}
return values;
@@ -764,4 +833,38 @@ std::vector<Node> ShaderIR::GetAoffiCoordinates(Node aoffi_reg, std::size_t coor
return aoffi;
}
+std::vector<Node> ShaderIR::GetPtpCoordinates(std::array<Node, 2> ptp_regs) {
+ static constexpr u32 num_entries = 8;
+
+ std::vector<Node> ptp;
+ ptp.reserve(num_entries);
+
+ const auto global_size = static_cast<s64>(global_code.size());
+ const std::optional low = TrackImmediate(ptp_regs[0], global_code, global_size);
+ const std::optional high = TrackImmediate(ptp_regs[1], global_code, global_size);
+ if (!low || !high) {
+ for (u32 entry = 0; entry < num_entries; ++entry) {
+ const u32 reg = entry / 4;
+ const u32 offset = entry % 4;
+ const Node value = BitfieldExtract(ptp_regs[reg], offset * 8, 6);
+ const Node condition =
+ Operation(OperationCode::LogicalIGreaterEqual, value, Immediate(32));
+ const Node negative = Operation(OperationCode::IAdd, value, Immediate(-64));
+ ptp.push_back(Operation(OperationCode::Select, condition, negative, value));
+ }
+ return ptp;
+ }
+
+ const u64 immediate = (static_cast<u64>(*high) << 32) | static_cast<u64>(*low);
+ for (u32 entry = 0; entry < num_entries; ++entry) {
+ s32 value = (immediate >> (entry * 8)) & 0b111111;
+ if (value >= 32) {
+ value -= 64;
+ }
+ ptp.push_back(Immediate(value));
+ }
+
+ return ptp;
+}
+
} // namespace VideoCommon::Shader
diff --git a/src/video_core/shader/decode/warp.cpp b/src/video_core/shader/decode/warp.cpp
index fa8a250cc..11b77f795 100644
--- a/src/video_core/shader/decode/warp.cpp
+++ b/src/video_core/shader/decode/warp.cpp
@@ -17,6 +17,7 @@ using Tegra::Shader::ShuffleOperation;
using Tegra::Shader::VoteOperation;
namespace {
+
OperationCode GetOperationCode(VoteOperation vote_op) {
switch (vote_op) {
case VoteOperation::All:
@@ -30,12 +31,16 @@ OperationCode GetOperationCode(VoteOperation vote_op) {
return OperationCode::VoteAll;
}
}
+
} // Anonymous namespace
u32 ShaderIR::DecodeWarp(NodeBlock& bb, u32 pc) {
const Instruction instr = {program_code[pc]};
const auto opcode = OpCode::Decode(instr);
+ // Signal the backend that this shader uses warp instructions.
+ uses_warps = true;
+
switch (opcode->get().GetId()) {
case OpCode::Id::VOTE: {
const Node value = GetPredicate(instr.vote.value, instr.vote.negate_value != 0);
@@ -46,50 +51,59 @@ u32 ShaderIR::DecodeWarp(NodeBlock& bb, u32 pc) {
break;
}
case OpCode::Id::SHFL: {
- Node width = [this, instr] {
- Node mask = instr.shfl.is_mask_imm ? Immediate(static_cast<u32>(instr.shfl.mask_imm))
- : GetRegister(instr.gpr39);
-
- // Convert the obscure SHFL mask back into GL_NV_shader_thread_shuffle's width. This has
- // been done reversing Nvidia's math. It won't work on all cases due to SHFL having
- // different parameters that don't properly map to GLSL's interface, but it should work
- // for cases emitted by Nvidia's compiler.
- if (instr.shfl.operation == ShuffleOperation::Up) {
- return Operation(
- OperationCode::ILogicalShiftRight,
- Operation(OperationCode::IAdd, std::move(mask), Immediate(-0x2000)),
- Immediate(8));
- } else {
- return Operation(OperationCode::ILogicalShiftRight,
- Operation(OperationCode::IAdd, Immediate(0x201F),
- Operation(OperationCode::INegate, std::move(mask))),
- Immediate(8));
- }
- }();
+ Node mask = instr.shfl.is_mask_imm ? Immediate(static_cast<u32>(instr.shfl.mask_imm))
+ : GetRegister(instr.gpr39);
+ Node index = instr.shfl.is_index_imm ? Immediate(static_cast<u32>(instr.shfl.index_imm))
+ : GetRegister(instr.gpr20);
- const auto [operation, in_range] = [instr]() -> std::pair<OperationCode, OperationCode> {
+ Node thread_id = Operation(OperationCode::ThreadId);
+ Node clamp = Operation(OperationCode::IBitwiseAnd, mask, Immediate(0x1FU));
+ Node seg_mask = BitfieldExtract(mask, 8, 16);
+
+ Node neg_seg_mask = Operation(OperationCode::IBitwiseNot, seg_mask);
+ Node min_thread_id = Operation(OperationCode::IBitwiseAnd, thread_id, seg_mask);
+ Node max_thread_id = Operation(OperationCode::IBitwiseOr, min_thread_id,
+ Operation(OperationCode::IBitwiseAnd, clamp, neg_seg_mask));
+
+ Node src_thread_id = [instr, index, neg_seg_mask, min_thread_id, thread_id] {
switch (instr.shfl.operation) {
case ShuffleOperation::Idx:
- return {OperationCode::ShuffleIndexed, OperationCode::InRangeShuffleIndexed};
- case ShuffleOperation::Up:
- return {OperationCode::ShuffleUp, OperationCode::InRangeShuffleUp};
+ return Operation(OperationCode::IBitwiseOr,
+ Operation(OperationCode::IBitwiseAnd, index, neg_seg_mask),
+ min_thread_id);
case ShuffleOperation::Down:
- return {OperationCode::ShuffleDown, OperationCode::InRangeShuffleDown};
+ return Operation(OperationCode::IAdd, thread_id, index);
+ case ShuffleOperation::Up:
+ return Operation(OperationCode::IAdd, thread_id,
+ Operation(OperationCode::INegate, index));
case ShuffleOperation::Bfly:
- return {OperationCode::ShuffleButterfly, OperationCode::InRangeShuffleButterfly};
+ return Operation(OperationCode::IBitwiseXor, thread_id, index);
}
- UNREACHABLE_MSG("Invalid SHFL operation: {}",
- static_cast<u64>(instr.shfl.operation.Value()));
- return {};
+ UNREACHABLE();
+ return Immediate(0U);
}();
- // Setting the predicate before the register is intentional to avoid overwriting.
- Node index = instr.shfl.is_index_imm ? Immediate(static_cast<u32>(instr.shfl.index_imm))
- : GetRegister(instr.gpr20);
- SetPredicate(bb, instr.shfl.pred48, Operation(in_range, index, width));
+ Node in_bounds = [instr, src_thread_id, min_thread_id, max_thread_id] {
+ if (instr.shfl.operation == ShuffleOperation::Up) {
+ return Operation(OperationCode::LogicalIGreaterEqual, src_thread_id, min_thread_id);
+ } else {
+ return Operation(OperationCode::LogicalILessEqual, src_thread_id, max_thread_id);
+ }
+ }();
+
+ SetPredicate(bb, instr.shfl.pred48, in_bounds);
SetRegister(
bb, instr.gpr0,
- Operation(operation, GetRegister(instr.gpr8), std::move(index), std::move(width)));
+ Operation(OperationCode::ShuffleIndexed, GetRegister(instr.gpr8), src_thread_id));
+ break;
+ }
+ case OpCode::Id::FSWZADD: {
+ UNIMPLEMENTED_IF(instr.fswzadd.ndv);
+
+ Node op_a = GetRegister(instr.gpr8);
+ Node op_b = GetRegister(instr.gpr20);
+ Node mask = Immediate(static_cast<u32>(instr.fswzadd.swizzle));
+ SetRegister(bb, instr.gpr0, Operation(OperationCode::FSwizzleAdd, op_a, op_b, mask));
break;
}
default:
diff --git a/src/video_core/shader/node.h b/src/video_core/shader/node.h
index 4300d9ff4..4e155542a 100644
--- a/src/video_core/shader/node.h
+++ b/src/video_core/shader/node.h
@@ -47,6 +47,7 @@ enum class OperationCode {
FTrunc, /// (MetaArithmetic, float a) -> float
FCastInteger, /// (MetaArithmetic, int a) -> float
FCastUInteger, /// (MetaArithmetic, uint a) -> float
+ FSwizzleAdd, /// (float a, float b, uint mask) -> float
IAdd, /// (MetaArithmetic, int a, int b) -> int
IMul, /// (MetaArithmetic, int a, int b) -> int
@@ -67,6 +68,7 @@ enum class OperationCode {
IBitfieldInsert, /// (MetaArithmetic, int base, int insert, int offset, int bits) -> int
IBitfieldExtract, /// (MetaArithmetic, int value, int offset, int offset) -> int
IBitCount, /// (MetaArithmetic, int) -> int
+ IBitMSB, /// (MetaArithmetic, int) -> int
UAdd, /// (MetaArithmetic, uint a, uint b) -> uint
UMul, /// (MetaArithmetic, uint a, uint b) -> uint
@@ -85,6 +87,7 @@ enum class OperationCode {
UBitfieldInsert, /// (MetaArithmetic, uint base, uint insert, int offset, int bits) -> uint
UBitfieldExtract, /// (MetaArithmetic, uint value, int offset, int offset) -> uint
UBitCount, /// (MetaArithmetic, uint) -> uint
+ UBitMSB, /// (MetaArithmetic, uint) -> uint
HAdd, /// (MetaArithmetic, f16vec2 a, f16vec2 b) -> f16vec2
HMul, /// (MetaArithmetic, f16vec2 a, f16vec2 b) -> f16vec2
@@ -148,6 +151,7 @@ enum class OperationCode {
TextureQueryDimensions, /// (MetaTexture, float a) -> float4
TextureQueryLod, /// (MetaTexture, float[N] coords) -> float4
TexelFetch, /// (MetaTexture, int[N], int) -> float4
+ TextureGradient, /// (MetaTexture, float[N] coords, float[N*2] derivates) -> float4
ImageLoad, /// (MetaImage, int[N] coords) -> void
ImageStore, /// (MetaImage, int[N] coords) -> void
@@ -168,6 +172,7 @@ enum class OperationCode {
EmitVertex, /// () -> void
EndPrimitive, /// () -> void
+ InvocationId, /// () -> int
YNegate, /// () -> float
LocalInvocationIdX, /// () -> uint
LocalInvocationIdY, /// () -> uint
@@ -181,15 +186,10 @@ enum class OperationCode {
VoteAny, /// (bool) -> bool
VoteEqual, /// (bool) -> bool
- ShuffleIndexed, /// (uint value, uint index, uint width) -> uint
- ShuffleUp, /// (uint value, uint index, uint width) -> uint
- ShuffleDown, /// (uint value, uint index, uint width) -> uint
- ShuffleButterfly, /// (uint value, uint index, uint width) -> uint
+ ThreadId, /// () -> uint
+ ShuffleIndexed, /// (uint value, uint index) -> uint
- InRangeShuffleIndexed, /// (uint index, uint width) -> bool
- InRangeShuffleUp, /// (uint index, uint width) -> bool
- InRangeShuffleDown, /// (uint index, uint width) -> bool
- InRangeShuffleButterfly, /// (uint index, uint width) -> bool
+ MemoryBarrierGL, /// () -> void
Amount,
};
@@ -216,13 +216,14 @@ class PredicateNode;
class AbufNode;
class CbufNode;
class LmemNode;
+class PatchNode;
class SmemNode;
class GmemNode;
class CommentNode;
-using NodeData =
- std::variant<OperationNode, ConditionalNode, GprNode, ImmediateNode, InternalFlagNode,
- PredicateNode, AbufNode, CbufNode, LmemNode, SmemNode, GmemNode, CommentNode>;
+using NodeData = std::variant<OperationNode, ConditionalNode, GprNode, ImmediateNode,
+ InternalFlagNode, PredicateNode, AbufNode, PatchNode, CbufNode,
+ LmemNode, SmemNode, GmemNode, CommentNode>;
using Node = std::shared_ptr<NodeData>;
using Node4 = std::array<Node, 4>;
using NodeBlock = std::vector<Node>;
@@ -231,14 +232,15 @@ class Sampler {
public:
/// This constructor is for bound samplers
constexpr explicit Sampler(u32 index, u32 offset, Tegra::Shader::TextureType type,
- bool is_array, bool is_shadow)
- : index{index}, offset{offset}, type{type}, is_array{is_array}, is_shadow{is_shadow} {}
+ bool is_array, bool is_shadow, bool is_buffer)
+ : index{index}, offset{offset}, type{type}, is_array{is_array}, is_shadow{is_shadow},
+ is_buffer{is_buffer} {}
/// This constructor is for bindless samplers
constexpr explicit Sampler(u32 index, u32 offset, u32 buffer, Tegra::Shader::TextureType type,
- bool is_array, bool is_shadow)
+ bool is_array, bool is_shadow, bool is_buffer)
: index{index}, offset{offset}, buffer{buffer}, type{type}, is_array{is_array},
- is_shadow{is_shadow}, is_bindless{true} {}
+ is_shadow{is_shadow}, is_buffer{is_buffer}, is_bindless{true} {}
constexpr u32 GetIndex() const {
return index;
@@ -264,6 +266,10 @@ public:
return is_shadow;
}
+ constexpr bool IsBuffer() const {
+ return is_buffer;
+ }
+
constexpr bool IsBindless() const {
return is_bindless;
}
@@ -276,6 +282,7 @@ private:
Tegra::Shader::TextureType type{}; ///< The type used to sample this texture (Texture2D, etc)
bool is_array{}; ///< Whether the texture is being sampled as an array texture or not.
bool is_shadow{}; ///< Whether the texture is being sampled as a depth texture or not.
+ bool is_buffer{}; ///< Whether the texture is a texture buffer without sampler.
bool is_bindless{}; ///< Whether this sampler belongs to a bindless texture or not.
};
@@ -367,6 +374,8 @@ struct MetaTexture {
Node array;
Node depth_compare;
std::vector<Node> aoffi;
+ std::vector<Node> ptp;
+ std::vector<Node> derivates;
Node bias;
Node lod;
Node component{};
@@ -383,8 +392,30 @@ struct MetaImage {
using Meta =
std::variant<MetaArithmetic, MetaTexture, MetaImage, MetaStackClass, Tegra::Shader::HalfType>;
+class AmendNode {
+public:
+ std::optional<std::size_t> GetAmendIndex() const {
+ if (amend_index == amend_null_index) {
+ return std::nullopt;
+ }
+ return {amend_index};
+ }
+
+ void SetAmendIndex(std::size_t index) {
+ amend_index = index;
+ }
+
+ void ClearAmend() {
+ amend_index = amend_null_index;
+ }
+
+private:
+ static constexpr std::size_t amend_null_index = 0xFFFFFFFFFFFFFFFFULL;
+ std::size_t amend_index{amend_null_index};
+};
+
/// Holds any kind of operation that can be done in the IR
-class OperationNode final {
+class OperationNode final : public AmendNode {
public:
explicit OperationNode(OperationCode code) : OperationNode(code, Meta{}) {}
@@ -424,7 +455,7 @@ private:
};
/// Encloses inside any kind of node that returns a boolean conditionally-executed code
-class ConditionalNode final {
+class ConditionalNode final : public AmendNode {
public:
explicit ConditionalNode(Node condition, std::vector<Node>&& code)
: condition{std::move(condition)}, code{std::move(code)} {}
@@ -538,6 +569,19 @@ private:
u32 element{};
};
+/// Patch memory (used to communicate tessellation stages).
+class PatchNode final {
+public:
+ explicit PatchNode(u32 offset) : offset{offset} {}
+
+ u32 GetOffset() const {
+ return offset;
+ }
+
+private:
+ u32 offset{};
+};
+
/// Constant buffer node, usually mapped to uniform buffers in GLSL
class CbufNode final {
public:
diff --git a/src/video_core/shader/shader_ir.cpp b/src/video_core/shader/shader_ir.cpp
index 1d9825c76..31eecb3f4 100644
--- a/src/video_core/shader/shader_ir.cpp
+++ b/src/video_core/shader/shader_ir.cpp
@@ -446,4 +446,10 @@ Node ShaderIR::BitfieldInsert(Node base, Node insert, u32 offset, u32 bits) {
Immediate(bits));
}
+std::size_t ShaderIR::DeclareAmend(Node new_amend) {
+ const std::size_t id = amend_code.size();
+ amend_code.push_back(new_amend);
+ return id;
+}
+
} // namespace VideoCommon::Shader
diff --git a/src/video_core/shader/shader_ir.h b/src/video_core/shader/shader_ir.h
index 26c8fde22..aacd0a0da 100644
--- a/src/video_core/shader/shader_ir.h
+++ b/src/video_core/shader/shader_ir.h
@@ -49,7 +49,7 @@ public:
}
u32 GetSize() const {
- return max_offset + sizeof(float);
+ return max_offset + static_cast<u32>(sizeof(float));
}
u32 GetMaxOffset() const {
@@ -137,6 +137,10 @@ public:
return uses_vertex_id;
}
+ bool UsesWarps() const {
+ return uses_warps;
+ }
+
bool HasPhysicalAttributes() const {
return uses_physical_attributes;
}
@@ -165,13 +169,17 @@ public:
return program_manager.GetVariables();
}
- u32 ConvertAddressToNvidiaSpace(const u32 address) const {
- return (address - main_offset) * sizeof(Tegra::Shader::Instruction);
+ u32 ConvertAddressToNvidiaSpace(u32 address) const {
+ return (address - main_offset) * static_cast<u32>(sizeof(Tegra::Shader::Instruction));
}
/// Returns a condition code evaluated from internal flags
Node GetConditionCode(Tegra::Shader::ConditionCode cc) const;
+ const Node& GetAmendNode(std::size_t index) const {
+ return amend_code[index];
+ }
+
private:
friend class ASTDecoder;
@@ -179,6 +187,7 @@ private:
Tegra::Shader::TextureType type;
bool is_array;
bool is_shadow;
+ bool is_buffer;
};
void Decode();
@@ -303,13 +312,17 @@ private:
/// Returns a predicate combiner operation
OperationCode GetPredicateCombiner(Tegra::Shader::PredOperation operation);
+ /// Queries the missing sampler info from the execution context.
+ SamplerInfo GetSamplerInfo(std::optional<SamplerInfo> sampler_info, u32 offset,
+ std::optional<u32> buffer = std::nullopt);
+
/// Accesses a texture sampler
- const Sampler& GetSampler(const Tegra::Shader::Sampler& sampler,
- std::optional<SamplerInfo> sampler_info);
+ const Sampler* GetSampler(const Tegra::Shader::Sampler& sampler,
+ std::optional<SamplerInfo> sampler_info = std::nullopt);
- // Accesses a texture sampler for a bindless texture.
- const Sampler& GetBindlessSampler(const Tegra::Shader::Register& reg,
- std::optional<SamplerInfo> sampler_info);
+ /// Accesses a texture sampler for a bindless texture.
+ const Sampler* GetBindlessSampler(Tegra::Shader::Register reg,
+ std::optional<SamplerInfo> sampler_info = std::nullopt);
/// Accesses an image.
Image& GetImage(Tegra::Shader::Image image, Tegra::Shader::ImageType type);
@@ -329,7 +342,7 @@ private:
void WriteTexsInstructionFloat(NodeBlock& bb, Tegra::Shader::Instruction instr,
const Node4& components, bool ignore_mask = false);
void WriteTexsInstructionHalfFloat(NodeBlock& bb, Tegra::Shader::Instruction instr,
- const Node4& components);
+ const Node4& components, bool ignore_mask = false);
Node4 GetTexCode(Tegra::Shader::Instruction instr, Tegra::Shader::TextureType texture_type,
Tegra::Shader::TextureProcessMode process_mode, bool depth_compare,
@@ -341,7 +354,8 @@ private:
bool is_array);
Node4 GetTld4Code(Tegra::Shader::Instruction instr, Tegra::Shader::TextureType texture_type,
- bool depth_compare, bool is_array, bool is_aoffi, bool is_bindless);
+ bool depth_compare, bool is_array, bool is_aoffi, bool is_ptp,
+ bool is_bindless);
Node4 GetTldCode(Tegra::Shader::Instruction instr);
@@ -354,6 +368,8 @@ private:
std::vector<Node> GetAoffiCoordinates(Node aoffi_reg, std::size_t coord_count, bool is_tld4);
+ std::vector<Node> GetPtpCoordinates(std::array<Node, 2> ptp_regs);
+
Node4 GetTextureCode(Tegra::Shader::Instruction instr, Tegra::Shader::TextureType texture_type,
Tegra::Shader::TextureProcessMode process_mode, std::vector<Node> coords,
Node array, Node depth_compare, u32 bias_offset, std::vector<Node> aoffi,
@@ -380,6 +396,9 @@ private:
Tegra::Shader::Instruction instr,
bool is_write);
+ /// Register new amending code and obtain the reference id.
+ std::size_t DeclareAmend(Node new_amend);
+
const ProgramCode& program_code;
const u32 main_offset;
const CompilerSettings settings;
@@ -394,6 +413,7 @@ private:
std::map<u32, NodeBlock> basic_blocks;
NodeBlock global_code;
ASTManager program_manager{true, true};
+ std::vector<Node> amend_code;
std::set<u32> used_registers;
std::set<Tegra::Shader::Pred> used_predicates;
@@ -410,6 +430,7 @@ private:
bool uses_physical_attributes{}; // Shader uses AL2P or physical attribute read/writes
bool uses_instance_id{};
bool uses_vertex_id{};
+ bool uses_warps{};
Tegra::Shader::Header header;
};
diff --git a/src/video_core/shader/track.cpp b/src/video_core/shader/track.cpp
index 55f5949e4..165c79330 100644
--- a/src/video_core/shader/track.cpp
+++ b/src/video_core/shader/track.cpp
@@ -7,6 +7,7 @@
#include <variant>
#include "common/common_types.h"
+#include "video_core/shader/node.h"
#include "video_core/shader/shader_ir.h"
namespace VideoCommon::Shader {
diff --git a/src/video_core/surface.cpp b/src/video_core/surface.cpp
index 621136b6e..1655ccf16 100644
--- a/src/video_core/surface.cpp
+++ b/src/video_core/surface.cpp
@@ -168,309 +168,6 @@ PixelFormat PixelFormatFromRenderTargetFormat(Tegra::RenderTargetFormat format)
}
}
-PixelFormat PixelFormatFromTextureFormat(Tegra::Texture::TextureFormat format,
- Tegra::Texture::ComponentType component_type,
- bool is_srgb) {
- // TODO(Subv): Properly implement this
- switch (format) {
- case Tegra::Texture::TextureFormat::A8R8G8B8:
- if (is_srgb) {
- return PixelFormat::RGBA8_SRGB;
- }
- switch (component_type) {
- case Tegra::Texture::ComponentType::UNORM:
- return PixelFormat::ABGR8U;
- case Tegra::Texture::ComponentType::SNORM:
- return PixelFormat::ABGR8S;
- case Tegra::Texture::ComponentType::UINT:
- return PixelFormat::ABGR8UI;
- default:
- break;
- }
- break;
- case Tegra::Texture::TextureFormat::B5G6R5:
- switch (component_type) {
- case Tegra::Texture::ComponentType::UNORM:
- return PixelFormat::B5G6R5U;
- default:
- break;
- }
- break;
- case Tegra::Texture::TextureFormat::A2B10G10R10:
- switch (component_type) {
- case Tegra::Texture::ComponentType::UNORM:
- return PixelFormat::A2B10G10R10U;
- default:
- break;
- }
- break;
- case Tegra::Texture::TextureFormat::A1B5G5R5:
- switch (component_type) {
- case Tegra::Texture::ComponentType::UNORM:
- return PixelFormat::A1B5G5R5U;
- default:
- break;
- }
- break;
- case Tegra::Texture::TextureFormat::A4B4G4R4:
- switch (component_type) {
- case Tegra::Texture::ComponentType::UNORM:
- return PixelFormat::R4G4B4A4U;
- default:
- break;
- }
- break;
- case Tegra::Texture::TextureFormat::R8:
- switch (component_type) {
- case Tegra::Texture::ComponentType::UNORM:
- return PixelFormat::R8U;
- case Tegra::Texture::ComponentType::UINT:
- return PixelFormat::R8UI;
- default:
- break;
- }
- break;
- case Tegra::Texture::TextureFormat::G8R8:
- // TextureFormat::G8R8 is actually ordered red then green, as such we can use
- // PixelFormat::RG8U and PixelFormat::RG8S. This was tested with The Legend of Zelda: Breath
- // of the Wild, which uses this format to render the hearts on the UI.
- switch (component_type) {
- case Tegra::Texture::ComponentType::UNORM:
- return PixelFormat::RG8U;
- case Tegra::Texture::ComponentType::SNORM:
- return PixelFormat::RG8S;
- default:
- break;
- }
- break;
- case Tegra::Texture::TextureFormat::R16_G16_B16_A16:
- switch (component_type) {
- case Tegra::Texture::ComponentType::UNORM:
- return PixelFormat::RGBA16U;
- case Tegra::Texture::ComponentType::FLOAT:
- return PixelFormat::RGBA16F;
- default:
- break;
- }
- break;
- case Tegra::Texture::TextureFormat::BF10GF11RF11:
- switch (component_type) {
- case Tegra::Texture::ComponentType::FLOAT:
- return PixelFormat::R11FG11FB10F;
- default:
- break;
- }
- break;
- case Tegra::Texture::TextureFormat::R32_G32_B32_A32:
- switch (component_type) {
- case Tegra::Texture::ComponentType::FLOAT:
- return PixelFormat::RGBA32F;
- case Tegra::Texture::ComponentType::UINT:
- return PixelFormat::RGBA32UI;
- default:
- break;
- }
- break;
- case Tegra::Texture::TextureFormat::R32_G32:
- switch (component_type) {
- case Tegra::Texture::ComponentType::FLOAT:
- return PixelFormat::RG32F;
- case Tegra::Texture::ComponentType::UINT:
- return PixelFormat::RG32UI;
- default:
- break;
- }
- break;
- case Tegra::Texture::TextureFormat::R32_G32_B32:
- switch (component_type) {
- case Tegra::Texture::ComponentType::FLOAT:
- return PixelFormat::RGB32F;
- default:
- break;
- }
- break;
- case Tegra::Texture::TextureFormat::R16:
- switch (component_type) {
- case Tegra::Texture::ComponentType::FLOAT:
- return PixelFormat::R16F;
- case Tegra::Texture::ComponentType::UNORM:
- return PixelFormat::R16U;
- case Tegra::Texture::ComponentType::SNORM:
- return PixelFormat::R16S;
- case Tegra::Texture::ComponentType::UINT:
- return PixelFormat::R16UI;
- case Tegra::Texture::ComponentType::SINT:
- return PixelFormat::R16I;
- default:
- break;
- }
- break;
- case Tegra::Texture::TextureFormat::R32:
- switch (component_type) {
- case Tegra::Texture::ComponentType::FLOAT:
- return PixelFormat::R32F;
- case Tegra::Texture::ComponentType::UINT:
- return PixelFormat::R32UI;
- default:
- break;
- }
- break;
- case Tegra::Texture::TextureFormat::E5B9G9R9_SHAREDEXP:
- switch (component_type) {
- case Tegra::Texture::ComponentType::FLOAT:
- return PixelFormat::E5B9G9R9F;
- default:
- break;
- }
- break;
- case Tegra::Texture::TextureFormat::ZF32:
- return PixelFormat::Z32F;
- case Tegra::Texture::TextureFormat::Z16:
- return PixelFormat::Z16;
- case Tegra::Texture::TextureFormat::S8Z24:
- return PixelFormat::S8Z24;
- case Tegra::Texture::TextureFormat::ZF32_X24S8:
- return PixelFormat::Z32FS8;
- case Tegra::Texture::TextureFormat::DXT1:
- return is_srgb ? PixelFormat::DXT1_SRGB : PixelFormat::DXT1;
- case Tegra::Texture::TextureFormat::DXT23:
- return is_srgb ? PixelFormat::DXT23_SRGB : PixelFormat::DXT23;
- case Tegra::Texture::TextureFormat::DXT45:
- return is_srgb ? PixelFormat::DXT45_SRGB : PixelFormat::DXT45;
- case Tegra::Texture::TextureFormat::DXN1:
- return PixelFormat::DXN1;
- case Tegra::Texture::TextureFormat::DXN2:
- switch (component_type) {
- case Tegra::Texture::ComponentType::UNORM:
- return PixelFormat::DXN2UNORM;
- case Tegra::Texture::ComponentType::SNORM:
- return PixelFormat::DXN2SNORM;
- default:
- break;
- }
- break;
- case Tegra::Texture::TextureFormat::BC7U:
- return is_srgb ? PixelFormat::BC7U_SRGB : PixelFormat::BC7U;
- case Tegra::Texture::TextureFormat::BC6H_UF16:
- return PixelFormat::BC6H_UF16;
- case Tegra::Texture::TextureFormat::BC6H_SF16:
- return PixelFormat::BC6H_SF16;
- case Tegra::Texture::TextureFormat::ASTC_2D_4X4:
- return is_srgb ? PixelFormat::ASTC_2D_4X4_SRGB : PixelFormat::ASTC_2D_4X4;
- case Tegra::Texture::TextureFormat::ASTC_2D_5X4:
- return is_srgb ? PixelFormat::ASTC_2D_5X4_SRGB : PixelFormat::ASTC_2D_5X4;
- case Tegra::Texture::TextureFormat::ASTC_2D_5X5:
- return is_srgb ? PixelFormat::ASTC_2D_5X5_SRGB : PixelFormat::ASTC_2D_5X5;
- case Tegra::Texture::TextureFormat::ASTC_2D_8X8:
- return is_srgb ? PixelFormat::ASTC_2D_8X8_SRGB : PixelFormat::ASTC_2D_8X8;
- case Tegra::Texture::TextureFormat::ASTC_2D_8X5:
- return is_srgb ? PixelFormat::ASTC_2D_8X5_SRGB : PixelFormat::ASTC_2D_8X5;
- case Tegra::Texture::TextureFormat::ASTC_2D_10X8:
- return is_srgb ? PixelFormat::ASTC_2D_10X8_SRGB : PixelFormat::ASTC_2D_10X8;
- case Tegra::Texture::TextureFormat::ASTC_2D_6X6:
- return is_srgb ? PixelFormat::ASTC_2D_6X6_SRGB : PixelFormat::ASTC_2D_6X6;
- case Tegra::Texture::TextureFormat::ASTC_2D_10X10:
- return is_srgb ? PixelFormat::ASTC_2D_10X10_SRGB : PixelFormat::ASTC_2D_10X10;
- case Tegra::Texture::TextureFormat::ASTC_2D_12X12:
- return is_srgb ? PixelFormat::ASTC_2D_12X12_SRGB : PixelFormat::ASTC_2D_12X12;
- case Tegra::Texture::TextureFormat::ASTC_2D_8X6:
- return is_srgb ? PixelFormat::ASTC_2D_8X6_SRGB : PixelFormat::ASTC_2D_8X6;
- case Tegra::Texture::TextureFormat::ASTC_2D_6X5:
- return is_srgb ? PixelFormat::ASTC_2D_6X5_SRGB : PixelFormat::ASTC_2D_6X5;
- case Tegra::Texture::TextureFormat::R16_G16:
- switch (component_type) {
- case Tegra::Texture::ComponentType::FLOAT:
- return PixelFormat::RG16F;
- case Tegra::Texture::ComponentType::UNORM:
- return PixelFormat::RG16;
- case Tegra::Texture::ComponentType::SNORM:
- return PixelFormat::RG16S;
- case Tegra::Texture::ComponentType::UINT:
- return PixelFormat::RG16UI;
- case Tegra::Texture::ComponentType::SINT:
- return PixelFormat::RG16I;
- default:
- break;
- }
- break;
- default:
- break;
- }
- LOG_CRITICAL(HW_GPU, "Unimplemented format={}, component_type={}", static_cast<u32>(format),
- static_cast<u32>(component_type));
- UNREACHABLE();
- return PixelFormat::ABGR8U;
-}
-
-ComponentType ComponentTypeFromTexture(Tegra::Texture::ComponentType type) {
- // TODO(Subv): Implement more component types
- switch (type) {
- case Tegra::Texture::ComponentType::UNORM:
- return ComponentType::UNorm;
- case Tegra::Texture::ComponentType::FLOAT:
- return ComponentType::Float;
- case Tegra::Texture::ComponentType::SNORM:
- return ComponentType::SNorm;
- case Tegra::Texture::ComponentType::UINT:
- return ComponentType::UInt;
- case Tegra::Texture::ComponentType::SINT:
- return ComponentType::SInt;
- default:
- LOG_CRITICAL(HW_GPU, "Unimplemented component type={}", static_cast<u32>(type));
- UNREACHABLE();
- return ComponentType::UNorm;
- }
-}
-
-ComponentType ComponentTypeFromRenderTarget(Tegra::RenderTargetFormat format) {
- // TODO(Subv): Implement more render targets
- switch (format) {
- case Tegra::RenderTargetFormat::RGBA8_UNORM:
- case Tegra::RenderTargetFormat::RGBA8_SRGB:
- case Tegra::RenderTargetFormat::BGRA8_UNORM:
- case Tegra::RenderTargetFormat::BGRA8_SRGB:
- case Tegra::RenderTargetFormat::RGB10_A2_UNORM:
- case Tegra::RenderTargetFormat::R8_UNORM:
- case Tegra::RenderTargetFormat::RG16_UNORM:
- case Tegra::RenderTargetFormat::R16_UNORM:
- case Tegra::RenderTargetFormat::B5G6R5_UNORM:
- case Tegra::RenderTargetFormat::BGR5A1_UNORM:
- case Tegra::RenderTargetFormat::RG8_UNORM:
- case Tegra::RenderTargetFormat::RGBA16_UNORM:
- return ComponentType::UNorm;
- case Tegra::RenderTargetFormat::RGBA8_SNORM:
- case Tegra::RenderTargetFormat::RG16_SNORM:
- case Tegra::RenderTargetFormat::R16_SNORM:
- case Tegra::RenderTargetFormat::RG8_SNORM:
- return ComponentType::SNorm;
- case Tegra::RenderTargetFormat::RGBA16_FLOAT:
- case Tegra::RenderTargetFormat::RGBX16_FLOAT:
- case Tegra::RenderTargetFormat::R11G11B10_FLOAT:
- case Tegra::RenderTargetFormat::RGBA32_FLOAT:
- case Tegra::RenderTargetFormat::RG32_FLOAT:
- case Tegra::RenderTargetFormat::RG16_FLOAT:
- case Tegra::RenderTargetFormat::R16_FLOAT:
- case Tegra::RenderTargetFormat::R32_FLOAT:
- return ComponentType::Float;
- case Tegra::RenderTargetFormat::RGBA32_UINT:
- case Tegra::RenderTargetFormat::RGBA16_UINT:
- case Tegra::RenderTargetFormat::RG16_UINT:
- case Tegra::RenderTargetFormat::R8_UINT:
- case Tegra::RenderTargetFormat::R16_UINT:
- case Tegra::RenderTargetFormat::RG32_UINT:
- case Tegra::RenderTargetFormat::R32_UINT:
- case Tegra::RenderTargetFormat::RGBA8_UINT:
- return ComponentType::UInt;
- case Tegra::RenderTargetFormat::RG16_SINT:
- case Tegra::RenderTargetFormat::R16_SINT:
- return ComponentType::SInt;
- default:
- LOG_CRITICAL(HW_GPU, "Unimplemented format={}", static_cast<u32>(format));
- UNREACHABLE();
- return ComponentType::UNorm;
- }
-}
-
PixelFormat PixelFormatFromGPUPixelFormat(Tegra::FramebufferConfig::PixelFormat format) {
switch (format) {
case Tegra::FramebufferConfig::PixelFormat::ABGR8:
@@ -485,22 +182,6 @@ PixelFormat PixelFormatFromGPUPixelFormat(Tegra::FramebufferConfig::PixelFormat
}
}
-ComponentType ComponentTypeFromDepthFormat(Tegra::DepthFormat format) {
- switch (format) {
- case Tegra::DepthFormat::Z16_UNORM:
- case Tegra::DepthFormat::S8_Z24_UNORM:
- case Tegra::DepthFormat::Z24_S8_UNORM:
- return ComponentType::UNorm;
- case Tegra::DepthFormat::Z32_FLOAT:
- case Tegra::DepthFormat::Z32_S8_X24_FLOAT:
- return ComponentType::Float;
- default:
- LOG_CRITICAL(HW_GPU, "Unimplemented format={}", static_cast<u32>(format));
- UNREACHABLE();
- return ComponentType::UNorm;
- }
-}
-
SurfaceType GetFormatType(PixelFormat pixel_format) {
if (static_cast<std::size_t>(pixel_format) <
static_cast<std::size_t>(PixelFormat::MaxColorFormat)) {
diff --git a/src/video_core/surface.h b/src/video_core/surface.h
index d3bcd38c5..0d17a93ed 100644
--- a/src/video_core/surface.h
+++ b/src/video_core/surface.h
@@ -106,18 +106,8 @@ enum class PixelFormat {
Max = MaxDepthStencilFormat,
Invalid = 255,
};
-
static constexpr std::size_t MaxPixelFormat = static_cast<std::size_t>(PixelFormat::Max);
-enum class ComponentType {
- Invalid = 0,
- SNorm = 1,
- UNorm = 2,
- SInt = 3,
- UInt = 4,
- Float = 5,
-};
-
enum class SurfaceType {
ColorTexture = 0,
Depth = 1,
@@ -609,18 +599,8 @@ PixelFormat PixelFormatFromDepthFormat(Tegra::DepthFormat format);
PixelFormat PixelFormatFromRenderTargetFormat(Tegra::RenderTargetFormat format);
-PixelFormat PixelFormatFromTextureFormat(Tegra::Texture::TextureFormat format,
- Tegra::Texture::ComponentType component_type,
- bool is_srgb);
-
-ComponentType ComponentTypeFromTexture(Tegra::Texture::ComponentType type);
-
-ComponentType ComponentTypeFromRenderTarget(Tegra::RenderTargetFormat format);
-
PixelFormat PixelFormatFromGPUPixelFormat(Tegra::FramebufferConfig::PixelFormat format);
-ComponentType ComponentTypeFromDepthFormat(Tegra::DepthFormat format);
-
SurfaceType GetFormatType(PixelFormat pixel_format);
bool IsPixelFormatASTC(PixelFormat format);
diff --git a/src/video_core/texture_cache/format_lookup_table.cpp b/src/video_core/texture_cache/format_lookup_table.cpp
new file mode 100644
index 000000000..271e67533
--- /dev/null
+++ b/src/video_core/texture_cache/format_lookup_table.cpp
@@ -0,0 +1,208 @@
+// Copyright 2019 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <array>
+#include "common/common_types.h"
+#include "common/logging/log.h"
+#include "video_core/texture_cache/format_lookup_table.h"
+
+namespace VideoCommon {
+
+using Tegra::Texture::ComponentType;
+using Tegra::Texture::TextureFormat;
+using VideoCore::Surface::PixelFormat;
+
+namespace {
+
+constexpr auto SNORM = ComponentType::SNORM;
+constexpr auto UNORM = ComponentType::UNORM;
+constexpr auto SINT = ComponentType::SINT;
+constexpr auto UINT = ComponentType::UINT;
+constexpr auto SNORM_FORCE_FP16 = ComponentType::SNORM_FORCE_FP16;
+constexpr auto UNORM_FORCE_FP16 = ComponentType::UNORM_FORCE_FP16;
+constexpr auto FLOAT = ComponentType::FLOAT;
+constexpr bool C = false; // Normal color
+constexpr bool S = true; // Srgb
+
+struct Table {
+ constexpr Table(TextureFormat texture_format, bool is_srgb, ComponentType red_component,
+ ComponentType green_component, ComponentType blue_component,
+ ComponentType alpha_component, PixelFormat pixel_format)
+ : texture_format{texture_format}, pixel_format{pixel_format}, red_component{red_component},
+ green_component{green_component}, blue_component{blue_component},
+ alpha_component{alpha_component}, is_srgb{is_srgb} {}
+
+ TextureFormat texture_format;
+ PixelFormat pixel_format;
+ ComponentType red_component;
+ ComponentType green_component;
+ ComponentType blue_component;
+ ComponentType alpha_component;
+ bool is_srgb;
+};
+constexpr std::array<Table, 74> DefinitionTable = {{
+ {TextureFormat::A8R8G8B8, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::ABGR8U},
+ {TextureFormat::A8R8G8B8, C, SNORM, SNORM, SNORM, SNORM, PixelFormat::ABGR8S},
+ {TextureFormat::A8R8G8B8, C, UINT, UINT, UINT, UINT, PixelFormat::ABGR8UI},
+ {TextureFormat::A8R8G8B8, S, UNORM, UNORM, UNORM, UNORM, PixelFormat::RGBA8_SRGB},
+
+ {TextureFormat::B5G6R5, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::B5G6R5U},
+
+ {TextureFormat::A2B10G10R10, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::A2B10G10R10U},
+
+ {TextureFormat::A1B5G5R5, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::A1B5G5R5U},
+
+ {TextureFormat::A4B4G4R4, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::R4G4B4A4U},
+
+ {TextureFormat::R8, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::R8U},
+ {TextureFormat::R8, C, UINT, UINT, UINT, UINT, PixelFormat::R8UI},
+
+ {TextureFormat::G8R8, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::RG8U},
+ {TextureFormat::G8R8, C, SNORM, SNORM, SNORM, SNORM, PixelFormat::RG8S},
+
+ {TextureFormat::R16_G16_B16_A16, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::RGBA16U},
+ {TextureFormat::R16_G16_B16_A16, C, FLOAT, FLOAT, FLOAT, FLOAT, PixelFormat::RGBA16F},
+ {TextureFormat::R16_G16_B16_A16, C, UINT, UINT, UINT, UINT, PixelFormat::RGBA16UI},
+
+ {TextureFormat::R16_G16, C, FLOAT, FLOAT, FLOAT, FLOAT, PixelFormat::RG16F},
+ {TextureFormat::R16_G16, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::RG16},
+ {TextureFormat::R16_G16, C, SNORM, SNORM, SNORM, SNORM, PixelFormat::RG16S},
+ {TextureFormat::R16_G16, C, UINT, UINT, UINT, UINT, PixelFormat::RG16UI},
+ {TextureFormat::R16_G16, C, SINT, SINT, SINT, SINT, PixelFormat::RG16I},
+
+ {TextureFormat::R16, C, FLOAT, FLOAT, FLOAT, FLOAT, PixelFormat::R16F},
+ {TextureFormat::R16, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::R16U},
+ {TextureFormat::R16, C, SNORM, SNORM, SNORM, SNORM, PixelFormat::R16S},
+ {TextureFormat::R16, C, UINT, UINT, UINT, UINT, PixelFormat::R16UI},
+ {TextureFormat::R16, C, SINT, SINT, SINT, SINT, PixelFormat::R16I},
+
+ {TextureFormat::BF10GF11RF11, C, FLOAT, FLOAT, FLOAT, FLOAT, PixelFormat::R11FG11FB10F},
+
+ {TextureFormat::R32_G32_B32_A32, C, FLOAT, FLOAT, FLOAT, FLOAT, PixelFormat::RGBA32F},
+ {TextureFormat::R32_G32_B32_A32, C, UINT, UINT, UINT, UINT, PixelFormat::RGBA32UI},
+
+ {TextureFormat::R32_G32_B32, C, FLOAT, FLOAT, FLOAT, FLOAT, PixelFormat::RGB32F},
+
+ {TextureFormat::R32_G32, C, FLOAT, FLOAT, FLOAT, FLOAT, PixelFormat::RG32F},
+ {TextureFormat::R32_G32, C, UINT, UINT, UINT, UINT, PixelFormat::RG32UI},
+
+ {TextureFormat::R32, C, FLOAT, FLOAT, FLOAT, FLOAT, PixelFormat::R32F},
+ {TextureFormat::R32, C, UINT, UINT, UINT, UINT, PixelFormat::R32UI},
+
+ {TextureFormat::E5B9G9R9_SHAREDEXP, C, FLOAT, FLOAT, FLOAT, FLOAT, PixelFormat::E5B9G9R9F},
+
+ {TextureFormat::ZF32, C, FLOAT, FLOAT, FLOAT, FLOAT, PixelFormat::Z32F},
+ {TextureFormat::Z16, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::Z16},
+ {TextureFormat::S8Z24, C, UINT, UNORM, UNORM, UNORM, PixelFormat::S8Z24},
+ {TextureFormat::ZF32_X24S8, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::Z32FS8},
+
+ {TextureFormat::DXT1, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::DXT1},
+ {TextureFormat::DXT1, S, UNORM, UNORM, UNORM, UNORM, PixelFormat::DXT1_SRGB},
+
+ {TextureFormat::DXT23, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::DXT23},
+ {TextureFormat::DXT23, S, UNORM, UNORM, UNORM, UNORM, PixelFormat::DXT23_SRGB},
+
+ {TextureFormat::DXT45, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::DXT45},
+ {TextureFormat::DXT45, S, UNORM, UNORM, UNORM, UNORM, PixelFormat::DXT45_SRGB},
+
+ // TODO: Use a different pixel format for SNORM
+ {TextureFormat::DXN1, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::DXN1},
+ {TextureFormat::DXN1, C, SNORM, SNORM, SNORM, SNORM, PixelFormat::DXN1},
+
+ {TextureFormat::DXN2, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::DXN2UNORM},
+ {TextureFormat::DXN2, C, SNORM, SNORM, SNORM, SNORM, PixelFormat::DXN2SNORM},
+
+ {TextureFormat::BC7U, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::BC7U},
+ {TextureFormat::BC7U, S, UNORM, UNORM, UNORM, UNORM, PixelFormat::BC7U_SRGB},
+
+ {TextureFormat::BC6H_SF16, C, FLOAT, FLOAT, FLOAT, FLOAT, PixelFormat::BC6H_SF16},
+ {TextureFormat::BC6H_UF16, C, FLOAT, FLOAT, FLOAT, FLOAT, PixelFormat::BC6H_UF16},
+
+ {TextureFormat::ASTC_2D_4X4, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_4X4},
+ {TextureFormat::ASTC_2D_4X4, S, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_4X4_SRGB},
+
+ {TextureFormat::ASTC_2D_5X4, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_5X4},
+ {TextureFormat::ASTC_2D_5X4, S, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_5X4_SRGB},
+
+ {TextureFormat::ASTC_2D_5X5, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_5X5},
+ {TextureFormat::ASTC_2D_5X5, S, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_5X5_SRGB},
+
+ {TextureFormat::ASTC_2D_8X8, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_8X8},
+ {TextureFormat::ASTC_2D_8X8, S, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_8X8_SRGB},
+
+ {TextureFormat::ASTC_2D_8X5, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_8X5},
+ {TextureFormat::ASTC_2D_8X5, S, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_8X5_SRGB},
+
+ {TextureFormat::ASTC_2D_10X8, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_10X8},
+ {TextureFormat::ASTC_2D_10X8, S, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_10X8_SRGB},
+
+ {TextureFormat::ASTC_2D_6X6, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_6X6},
+ {TextureFormat::ASTC_2D_6X6, S, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_6X6_SRGB},
+
+ {TextureFormat::ASTC_2D_10X10, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_10X10},
+ {TextureFormat::ASTC_2D_10X10, S, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_10X10_SRGB},
+
+ {TextureFormat::ASTC_2D_12X12, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_12X12},
+ {TextureFormat::ASTC_2D_12X12, S, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_12X12_SRGB},
+
+ {TextureFormat::ASTC_2D_8X6, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_8X6},
+ {TextureFormat::ASTC_2D_8X6, S, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_8X6_SRGB},
+
+ {TextureFormat::ASTC_2D_6X5, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_6X5},
+ {TextureFormat::ASTC_2D_6X5, S, UNORM, UNORM, UNORM, UNORM, PixelFormat::ASTC_2D_6X5_SRGB},
+}};
+
+} // Anonymous namespace
+
+FormatLookupTable::FormatLookupTable() {
+ table.fill(static_cast<u8>(PixelFormat::Invalid));
+
+ for (const auto& entry : DefinitionTable) {
+ table[CalculateIndex(entry.texture_format, entry.is_srgb != 0, entry.red_component,
+ entry.green_component, entry.blue_component, entry.alpha_component)] =
+ static_cast<u8>(entry.pixel_format);
+ }
+}
+
+PixelFormat FormatLookupTable::GetPixelFormat(TextureFormat format, bool is_srgb,
+ ComponentType red_component,
+ ComponentType green_component,
+ ComponentType blue_component,
+ ComponentType alpha_component) const noexcept {
+ const auto pixel_format = static_cast<PixelFormat>(table[CalculateIndex(
+ format, is_srgb, red_component, green_component, blue_component, alpha_component)]);
+ // [[likely]]
+ if (pixel_format != PixelFormat::Invalid) {
+ return pixel_format;
+ }
+ UNIMPLEMENTED_MSG("texture format={} srgb={} components={{{} {} {} {}}}",
+ static_cast<int>(format), is_srgb, static_cast<int>(red_component),
+ static_cast<int>(green_component), static_cast<int>(blue_component),
+ static_cast<int>(alpha_component));
+ return PixelFormat::ABGR8U;
+}
+
+void FormatLookupTable::Set(TextureFormat format, bool is_srgb, ComponentType red_component,
+ ComponentType green_component, ComponentType blue_component,
+ ComponentType alpha_component, PixelFormat pixel_format) {}
+
+std::size_t FormatLookupTable::CalculateIndex(TextureFormat format, bool is_srgb,
+ ComponentType red_component,
+ ComponentType green_component,
+ ComponentType blue_component,
+ ComponentType alpha_component) noexcept {
+ const auto format_index = static_cast<std::size_t>(format);
+ const auto red_index = static_cast<std::size_t>(red_component);
+ const auto green_index = static_cast<std::size_t>(red_component);
+ const auto blue_index = static_cast<std::size_t>(red_component);
+ const auto alpha_index = static_cast<std::size_t>(red_component);
+ const std::size_t srgb_index = is_srgb ? 1 : 0;
+
+ return format_index * PerFormat +
+ srgb_index * PerComponent * PerComponent * PerComponent * PerComponent +
+ alpha_index * PerComponent * PerComponent * PerComponent +
+ blue_index * PerComponent * PerComponent + green_index * PerComponent + red_index;
+}
+
+} // namespace VideoCommon
diff --git a/src/video_core/texture_cache/format_lookup_table.h b/src/video_core/texture_cache/format_lookup_table.h
new file mode 100644
index 000000000..aa77e0a5a
--- /dev/null
+++ b/src/video_core/texture_cache/format_lookup_table.h
@@ -0,0 +1,51 @@
+// Copyright 2019 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <array>
+#include <limits>
+#include "video_core/surface.h"
+#include "video_core/textures/texture.h"
+
+namespace VideoCommon {
+
+class FormatLookupTable {
+public:
+ explicit FormatLookupTable();
+
+ VideoCore::Surface::PixelFormat GetPixelFormat(
+ Tegra::Texture::TextureFormat format, bool is_srgb,
+ Tegra::Texture::ComponentType red_component, Tegra::Texture::ComponentType green_component,
+ Tegra::Texture::ComponentType blue_component,
+ Tegra::Texture::ComponentType alpha_component) const noexcept;
+
+private:
+ static_assert(VideoCore::Surface::MaxPixelFormat <= std::numeric_limits<u8>::max());
+
+ static constexpr std::size_t NumTextureFormats = 128;
+
+ static constexpr std::size_t PerComponent = 8;
+ static constexpr std::size_t PerComponents2 = PerComponent * PerComponent;
+ static constexpr std::size_t PerComponents3 = PerComponents2 * PerComponent;
+ static constexpr std::size_t PerComponents4 = PerComponents3 * PerComponent;
+ static constexpr std::size_t PerFormat = PerComponents4 * 2;
+
+ static std::size_t CalculateIndex(Tegra::Texture::TextureFormat format, bool is_srgb,
+ Tegra::Texture::ComponentType red_component,
+ Tegra::Texture::ComponentType green_component,
+ Tegra::Texture::ComponentType blue_component,
+ Tegra::Texture::ComponentType alpha_component) noexcept;
+
+ void Set(Tegra::Texture::TextureFormat format, bool is_srgb,
+ Tegra::Texture::ComponentType red_component,
+ Tegra::Texture::ComponentType green_component,
+ Tegra::Texture::ComponentType blue_component,
+ Tegra::Texture::ComponentType alpha_component,
+ VideoCore::Surface::PixelFormat pixel_format);
+
+ std::array<u8, NumTextureFormats * PerFormat> table;
+};
+
+} // namespace VideoCommon
diff --git a/src/video_core/texture_cache/surface_base.h b/src/video_core/texture_cache/surface_base.h
index 1bed82898..5f79bb0aa 100644
--- a/src/video_core/texture_cache/surface_base.h
+++ b/src/video_core/texture_cache/surface_base.h
@@ -254,16 +254,14 @@ public:
if (!layer_mipmap) {
return {};
}
- const u32 end_layer{layer_mipmap->first};
- const u32 end_mipmap{layer_mipmap->second};
+ const auto [end_layer, end_mipmap] = *layer_mipmap;
if (layer != end_layer) {
if (mipmap == 0 && end_mipmap == 0) {
- return GetView(ViewParams(view_params.target, layer, end_layer - layer + 1, 0, 1));
+ return GetView(ViewParams(view_params.target, layer, end_layer - layer, 0, 1));
}
return {};
} else {
- return GetView(
- ViewParams(view_params.target, layer, 1, mipmap, end_mipmap - mipmap + 1));
+ return GetView(ViewParams(view_params.target, layer, 1, mipmap, end_mipmap - mipmap));
}
}
@@ -278,8 +276,7 @@ public:
if (!layer_mipmap) {
return {};
}
- const u32 layer{layer_mipmap->first};
- const u32 mipmap{layer_mipmap->second};
+ const auto [layer, mipmap] = *layer_mipmap;
if (GetMipmapSize(mipmap) != candidate_size) {
return EmplaceIrregularView(view_params, view_addr, candidate_size, mipmap, layer);
}
diff --git a/src/video_core/texture_cache/surface_params.cpp b/src/video_core/texture_cache/surface_params.cpp
index 1e4d3fb79..38b3a4ba8 100644
--- a/src/video_core/texture_cache/surface_params.cpp
+++ b/src/video_core/texture_cache/surface_params.cpp
@@ -2,24 +2,23 @@
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
-#include <map>
+#include <algorithm>
+#include <string>
+#include <tuple>
#include "common/alignment.h"
#include "common/bit_util.h"
#include "core/core.h"
#include "video_core/engines/shader_bytecode.h"
#include "video_core/surface.h"
+#include "video_core/texture_cache/format_lookup_table.h"
#include "video_core/texture_cache/surface_params.h"
namespace VideoCommon {
-using VideoCore::Surface::ComponentTypeFromDepthFormat;
-using VideoCore::Surface::ComponentTypeFromRenderTarget;
-using VideoCore::Surface::ComponentTypeFromTexture;
using VideoCore::Surface::PixelFormat;
using VideoCore::Surface::PixelFormatFromDepthFormat;
using VideoCore::Surface::PixelFormatFromRenderTargetFormat;
-using VideoCore::Surface::PixelFormatFromTextureFormat;
using VideoCore::Surface::SurfaceTarget;
using VideoCore::Surface::SurfaceTargetFromTextureType;
using VideoCore::Surface::SurfaceType;
@@ -69,7 +68,8 @@ constexpr u32 GetMipmapSize(bool uncompressed, u32 mip_size, u32 tile) {
} // Anonymous namespace
-SurfaceParams SurfaceParams::CreateForTexture(const Tegra::Texture::TICEntry& tic,
+SurfaceParams SurfaceParams::CreateForTexture(const FormatLookupTable& lookup_table,
+ const Tegra::Texture::TICEntry& tic,
const VideoCommon::Shader::Sampler& entry) {
SurfaceParams params;
params.is_tiled = tic.IsTiled();
@@ -78,8 +78,8 @@ SurfaceParams SurfaceParams::CreateForTexture(const Tegra::Texture::TICEntry& ti
params.block_height = params.is_tiled ? tic.BlockHeight() : 0,
params.block_depth = params.is_tiled ? tic.BlockDepth() : 0,
params.tile_width_spacing = params.is_tiled ? (1 << tic.tile_width_spacing.Value()) : 1;
- params.pixel_format =
- PixelFormatFromTextureFormat(tic.format, tic.r_type.Value(), params.srgb_conversion);
+ params.pixel_format = lookup_table.GetPixelFormat(
+ tic.format, params.srgb_conversion, tic.r_type, tic.g_type, tic.b_type, tic.a_type);
params.type = GetFormatType(params.pixel_format);
if (entry.IsShadow() && params.type == SurfaceType::ColorTexture) {
switch (params.pixel_format) {
@@ -99,7 +99,6 @@ SurfaceParams SurfaceParams::CreateForTexture(const Tegra::Texture::TICEntry& ti
}
params.type = GetFormatType(params.pixel_format);
}
- params.component_type = ComponentTypeFromTexture(tic.r_type.Value());
params.type = GetFormatType(params.pixel_format);
// TODO: on 1DBuffer we should use the tic info.
if (tic.IsBuffer()) {
@@ -128,7 +127,8 @@ SurfaceParams SurfaceParams::CreateForTexture(const Tegra::Texture::TICEntry& ti
return params;
}
-SurfaceParams SurfaceParams::CreateForImage(const Tegra::Texture::TICEntry& tic,
+SurfaceParams SurfaceParams::CreateForImage(const FormatLookupTable& lookup_table,
+ const Tegra::Texture::TICEntry& tic,
const VideoCommon::Shader::Image& entry) {
SurfaceParams params;
params.is_tiled = tic.IsTiled();
@@ -137,10 +137,9 @@ SurfaceParams SurfaceParams::CreateForImage(const Tegra::Texture::TICEntry& tic,
params.block_height = params.is_tiled ? tic.BlockHeight() : 0,
params.block_depth = params.is_tiled ? tic.BlockDepth() : 0,
params.tile_width_spacing = params.is_tiled ? (1 << tic.tile_width_spacing.Value()) : 1;
- params.pixel_format =
- PixelFormatFromTextureFormat(tic.format, tic.r_type.Value(), params.srgb_conversion);
+ params.pixel_format = lookup_table.GetPixelFormat(
+ tic.format, params.srgb_conversion, tic.r_type, tic.g_type, tic.b_type, tic.a_type);
params.type = GetFormatType(params.pixel_format);
- params.component_type = ComponentTypeFromTexture(tic.r_type.Value());
params.type = GetFormatType(params.pixel_format);
params.target = ImageTypeToSurfaceTarget(entry.GetType());
// TODO: on 1DBuffer we should use the tic info.
@@ -181,7 +180,6 @@ SurfaceParams SurfaceParams::CreateForDepthBuffer(
params.block_depth = std::min(block_depth, 5U);
params.tile_width_spacing = 1;
params.pixel_format = PixelFormatFromDepthFormat(format);
- params.component_type = ComponentTypeFromDepthFormat(format);
params.type = GetFormatType(params.pixel_format);
params.width = zeta_width;
params.height = zeta_height;
@@ -206,7 +204,6 @@ SurfaceParams SurfaceParams::CreateForFramebuffer(Core::System& system, std::siz
params.block_depth = config.memory_layout.block_depth;
params.tile_width_spacing = 1;
params.pixel_format = PixelFormatFromRenderTargetFormat(config.format);
- params.component_type = ComponentTypeFromRenderTarget(config.format);
params.type = GetFormatType(params.pixel_format);
if (params.is_tiled) {
params.pitch = 0;
@@ -236,7 +233,6 @@ SurfaceParams SurfaceParams::CreateForFermiCopySurface(
params.block_depth = params.is_tiled ? std::min(config.BlockDepth(), 5U) : 0,
params.tile_width_spacing = 1;
params.pixel_format = PixelFormatFromRenderTargetFormat(config.format);
- params.component_type = ComponentTypeFromRenderTarget(config.format);
params.type = GetFormatType(params.pixel_format);
params.width = config.width;
params.height = config.height;
@@ -250,6 +246,16 @@ SurfaceParams SurfaceParams::CreateForFermiCopySurface(
return params;
}
+VideoCore::Surface::SurfaceTarget SurfaceParams::ExpectedTarget(
+ const VideoCommon::Shader::Sampler& entry) {
+ return TextureTypeToSurfaceTarget(entry.GetType(), entry.IsArray());
+}
+
+VideoCore::Surface::SurfaceTarget SurfaceParams::ExpectedTarget(
+ const VideoCommon::Shader::Image& entry) {
+ return ImageTypeToSurfaceTarget(entry.GetType());
+}
+
bool SurfaceParams::IsLayered() const {
switch (target) {
case SurfaceTarget::Texture1DArray:
@@ -355,10 +361,10 @@ std::size_t SurfaceParams::GetInnerMipmapMemorySize(u32 level, bool as_host_size
bool SurfaceParams::operator==(const SurfaceParams& rhs) const {
return std::tie(is_tiled, block_width, block_height, block_depth, tile_width_spacing, width,
- height, depth, pitch, num_levels, pixel_format, component_type, type, target) ==
+ height, depth, pitch, num_levels, pixel_format, type, target) ==
std::tie(rhs.is_tiled, rhs.block_width, rhs.block_height, rhs.block_depth,
rhs.tile_width_spacing, rhs.width, rhs.height, rhs.depth, rhs.pitch,
- rhs.num_levels, rhs.pixel_format, rhs.component_type, rhs.type, rhs.target);
+ rhs.num_levels, rhs.pixel_format, rhs.type, rhs.target);
}
std::string SurfaceParams::TargetName() const {
@@ -386,4 +392,42 @@ std::string SurfaceParams::TargetName() const {
}
}
+u32 SurfaceParams::GetBlockSize() const {
+ const u32 x = 64U << block_width;
+ const u32 y = 8U << block_height;
+ const u32 z = 1U << block_depth;
+ return x * y * z;
+}
+
+std::pair<u32, u32> SurfaceParams::GetBlockXY() const {
+ const u32 x_pixels = 64U / GetBytesPerPixel();
+ const u32 x = x_pixels << block_width;
+ const u32 y = 8U << block_height;
+ return {x, y};
+}
+
+std::tuple<u32, u32, u32> SurfaceParams::GetBlockOffsetXYZ(u32 offset) const {
+ const auto div_ceil = [](const u32 x, const u32 y) { return ((x + y - 1) / y); };
+ const u32 block_size = GetBlockSize();
+ const u32 block_index = offset / block_size;
+ const u32 gob_offset = offset % block_size;
+ const u32 gob_index = gob_offset / static_cast<u32>(Tegra::Texture::GetGOBSize());
+ const u32 x_gob_pixels = 64U / GetBytesPerPixel();
+ const u32 x_block_pixels = x_gob_pixels << block_width;
+ const u32 y_block_pixels = 8U << block_height;
+ const u32 z_block_pixels = 1U << block_depth;
+ const u32 x_blocks = div_ceil(width, x_block_pixels);
+ const u32 y_blocks = div_ceil(height, y_block_pixels);
+ const u32 z_blocks = div_ceil(depth, z_block_pixels);
+ const u32 base_x = block_index % x_blocks;
+ const u32 base_y = (block_index / x_blocks) % y_blocks;
+ const u32 base_z = (block_index / (x_blocks * y_blocks)) % z_blocks;
+ u32 x = base_x * x_block_pixels;
+ u32 y = base_y * y_block_pixels;
+ u32 z = base_z * z_block_pixels;
+ z += gob_index >> block_height;
+ y += (gob_index * 8U) % y_block_pixels;
+ return {x, y, z};
+}
+
} // namespace VideoCommon
diff --git a/src/video_core/texture_cache/surface_params.h b/src/video_core/texture_cache/surface_params.h
index c58e7f8a4..992b5c022 100644
--- a/src/video_core/texture_cache/surface_params.h
+++ b/src/video_core/texture_cache/surface_params.h
@@ -4,6 +4,8 @@
#pragma once
+#include <utility>
+
#include "common/alignment.h"
#include "common/bit_util.h"
#include "common/cityhash.h"
@@ -16,16 +18,20 @@
namespace VideoCommon {
+class FormatLookupTable;
+
using VideoCore::Surface::SurfaceCompression;
class SurfaceParams {
public:
/// Creates SurfaceCachedParams from a texture configuration.
- static SurfaceParams CreateForTexture(const Tegra::Texture::TICEntry& tic,
+ static SurfaceParams CreateForTexture(const FormatLookupTable& lookup_table,
+ const Tegra::Texture::TICEntry& tic,
const VideoCommon::Shader::Sampler& entry);
/// Creates SurfaceCachedParams from an image configuration.
- static SurfaceParams CreateForImage(const Tegra::Texture::TICEntry& tic,
+ static SurfaceParams CreateForImage(const FormatLookupTable& lookup_table,
+ const Tegra::Texture::TICEntry& tic,
const VideoCommon::Shader::Image& entry);
/// Creates SurfaceCachedParams for a depth buffer configuration.
@@ -41,6 +47,14 @@ public:
static SurfaceParams CreateForFermiCopySurface(
const Tegra::Engines::Fermi2D::Regs::Surface& config);
+ /// Obtains the texture target from a shader's sampler entry.
+ static VideoCore::Surface::SurfaceTarget ExpectedTarget(
+ const VideoCommon::Shader::Sampler& entry);
+
+ /// Obtains the texture target from a shader's sampler entry.
+ static VideoCore::Surface::SurfaceTarget ExpectedTarget(
+ const VideoCommon::Shader::Image& entry);
+
std::size_t Hash() const {
return static_cast<std::size_t>(
Common::CityHash64(reinterpret_cast<const char*>(this), sizeof(*this)));
@@ -124,6 +138,15 @@ public:
std::size_t GetConvertedMipmapSize(u32 level) const;
+ /// Get this texture Tegra Block size in guest memory layout
+ u32 GetBlockSize() const;
+
+ /// Get X, Y coordinates max sizes of a single block.
+ std::pair<u32, u32> GetBlockXY() const;
+
+ /// Get the offset in x, y, z coordinates from a memory offset
+ std::tuple<u32, u32, u32> GetBlockOffsetXYZ(u32 offset) const;
+
/// Returns the size of a layer in bytes in guest memory.
std::size_t GetGuestLayerSize() const {
return GetLayerSize(false, false);
@@ -248,7 +271,6 @@ public:
u32 num_levels;
u32 emulated_levels;
VideoCore::Surface::PixelFormat pixel_format;
- VideoCore::Surface::ComponentType component_type;
VideoCore::Surface::SurfaceType type;
VideoCore::Surface::SurfaceTarget target;
@@ -258,7 +280,8 @@ private:
/// Returns the size of all mipmap levels and aligns as needed.
std::size_t GetInnerMemorySize(bool as_host_size, bool layer_only, bool uncompressed) const {
- return GetLayerSize(as_host_size, uncompressed) * (layer_only ? 1U : depth);
+ return GetLayerSize(as_host_size, uncompressed) *
+ (layer_only ? 1U : (is_layered ? depth : 1U));
}
/// Returns the size of a layer
diff --git a/src/video_core/texture_cache/texture_cache.h b/src/video_core/texture_cache/texture_cache.h
index 6a92b22d3..f4c015635 100644
--- a/src/video_core/texture_cache/texture_cache.h
+++ b/src/video_core/texture_cache/texture_cache.h
@@ -29,6 +29,7 @@
#include "video_core/rasterizer_interface.h"
#include "video_core/surface.h"
#include "video_core/texture_cache/copy_params.h"
+#include "video_core/texture_cache/format_lookup_table.h"
#include "video_core/texture_cache/surface_base.h"
#include "video_core/texture_cache/surface_params.h"
#include "video_core/texture_cache/surface_view.h"
@@ -94,10 +95,16 @@ public:
std::lock_guard lock{mutex};
const auto gpu_addr{tic.Address()};
if (!gpu_addr) {
- return {};
+ return GetNullSurface(SurfaceParams::ExpectedTarget(entry));
}
- const auto params{SurfaceParams::CreateForTexture(tic, entry)};
- const auto [surface, view] = GetSurface(gpu_addr, params, true, false);
+
+ const auto host_ptr{system.GPU().MemoryManager().GetPointer(gpu_addr)};
+ const auto cache_addr{ToCacheAddr(host_ptr)};
+ if (!cache_addr) {
+ return GetNullSurface(SurfaceParams::ExpectedTarget(entry));
+ }
+ const auto params{SurfaceParams::CreateForTexture(format_lookup_table, tic, entry)};
+ const auto [surface, view] = GetSurface(gpu_addr, cache_addr, params, true, false);
if (guard_samplers) {
sampled_textures.push_back(surface);
}
@@ -109,10 +116,15 @@ public:
std::lock_guard lock{mutex};
const auto gpu_addr{tic.Address()};
if (!gpu_addr) {
- return {};
+ return GetNullSurface(SurfaceParams::ExpectedTarget(entry));
+ }
+ const auto host_ptr{system.GPU().MemoryManager().GetPointer(gpu_addr)};
+ const auto cache_addr{ToCacheAddr(host_ptr)};
+ if (!cache_addr) {
+ return GetNullSurface(SurfaceParams::ExpectedTarget(entry));
}
- const auto params{SurfaceParams::CreateForImage(tic, entry)};
- const auto [surface, view] = GetSurface(gpu_addr, params, true, false);
+ const auto params{SurfaceParams::CreateForImage(format_lookup_table, tic, entry)};
+ const auto [surface, view] = GetSurface(gpu_addr, cache_addr, params, true, false);
if (guard_samplers) {
sampled_textures.push_back(surface);
}
@@ -142,11 +154,17 @@ public:
SetEmptyDepthBuffer();
return {};
}
+ const auto host_ptr{system.GPU().MemoryManager().GetPointer(gpu_addr)};
+ const auto cache_addr{ToCacheAddr(host_ptr)};
+ if (!cache_addr) {
+ SetEmptyDepthBuffer();
+ return {};
+ }
const auto depth_params{SurfaceParams::CreateForDepthBuffer(
system, regs.zeta_width, regs.zeta_height, regs.zeta.format,
regs.zeta.memory_layout.block_width, regs.zeta.memory_layout.block_height,
regs.zeta.memory_layout.block_depth, regs.zeta.memory_layout.type)};
- auto surface_view = GetSurface(gpu_addr, depth_params, preserve_contents, true);
+ auto surface_view = GetSurface(gpu_addr, cache_addr, depth_params, preserve_contents, true);
if (depth_buffer.target)
depth_buffer.target->MarkAsRenderTarget(false, NO_RT);
depth_buffer.target = surface_view.first;
@@ -179,8 +197,16 @@ public:
return {};
}
- auto surface_view = GetSurface(gpu_addr, SurfaceParams::CreateForFramebuffer(system, index),
- preserve_contents, true);
+ const auto host_ptr{system.GPU().MemoryManager().GetPointer(gpu_addr)};
+ const auto cache_addr{ToCacheAddr(host_ptr)};
+ if (!cache_addr) {
+ SetEmptyColorBuffer(index);
+ return {};
+ }
+
+ auto surface_view =
+ GetSurface(gpu_addr, cache_addr, SurfaceParams::CreateForFramebuffer(system, index),
+ preserve_contents, true);
if (render_targets[index].target)
render_targets[index].target->MarkAsRenderTarget(false, NO_RT);
render_targets[index].target = surface_view.first;
@@ -229,8 +255,14 @@ public:
const GPUVAddr src_gpu_addr = src_config.Address();
const GPUVAddr dst_gpu_addr = dst_config.Address();
DeduceBestBlit(src_params, dst_params, src_gpu_addr, dst_gpu_addr);
- std::pair<TSurface, TView> dst_surface = GetSurface(dst_gpu_addr, dst_params, true, false);
- std::pair<TSurface, TView> src_surface = GetSurface(src_gpu_addr, src_params, true, false);
+ const auto dst_host_ptr{system.GPU().MemoryManager().GetPointer(dst_gpu_addr)};
+ const auto dst_cache_addr{ToCacheAddr(dst_host_ptr)};
+ const auto src_host_ptr{system.GPU().MemoryManager().GetPointer(src_gpu_addr)};
+ const auto src_cache_addr{ToCacheAddr(src_host_ptr)};
+ std::pair<TSurface, TView> dst_surface =
+ GetSurface(dst_gpu_addr, dst_cache_addr, dst_params, true, false);
+ std::pair<TSurface, TView> src_surface =
+ GetSurface(src_gpu_addr, src_cache_addr, src_params, true, false);
ImageBlit(src_surface.second, dst_surface.second, copy_config);
dst_surface.first->MarkAsModified(true, Tick());
}
@@ -346,13 +378,6 @@ protected:
return new_surface;
}
- std::pair<TSurface, TView> GetFermiSurface(
- const Tegra::Engines::Fermi2D::Regs::Surface& config) {
- SurfaceParams params = SurfaceParams::CreateForFermiCopySurface(config);
- const GPUVAddr gpu_addr = config.Address();
- return GetSurface(gpu_addr, params, true, false);
- }
-
Core::System& system;
private:
@@ -485,15 +510,13 @@ private:
GetSiblingFormat(cr_params.pixel_format) == params.pixel_format) {
SurfaceParams new_params = params;
new_params.pixel_format = cr_params.pixel_format;
- new_params.component_type = cr_params.component_type;
new_params.type = cr_params.type;
new_surface = GetUncachedSurface(gpu_addr, new_params);
} else {
new_surface = GetUncachedSurface(gpu_addr, params);
}
const auto& final_params = new_surface->GetSurfaceParams();
- if (cr_params.type != final_params.type ||
- (cr_params.component_type != final_params.component_type)) {
+ if (cr_params.type != final_params.type) {
BufferCopy(current_surface, new_surface);
} else {
std::vector<CopyParams> bricks = current_surface->BreakDown(final_params);
@@ -593,6 +616,86 @@ private:
}
/**
+ * Takes care of managing 3D textures and its slices. Does HLE methods for reconstructing the 3D
+ * textures within the GPU if possible. Falls back to LLE when it isn't possible to use any of
+ * the HLE methods.
+ *
+ * @param overlaps The overlapping surfaces registered in the cache.
+ * @param params The parameters on the new surface.
+ * @param gpu_addr The starting address of the new surface.
+ * @param cache_addr The starting address of the new surface on physical memory.
+ * @param preserve_contents Indicates that the new surface should be loaded from memory or
+ * left blank.
+ */
+ std::optional<std::pair<TSurface, TView>> Manage3DSurfaces(std::vector<TSurface>& overlaps,
+ const SurfaceParams& params,
+ const GPUVAddr gpu_addr,
+ const CacheAddr cache_addr,
+ bool preserve_contents) {
+ if (params.target == SurfaceTarget::Texture3D) {
+ bool failed = false;
+ if (params.num_levels > 1) {
+ // We can't handle mipmaps in 3D textures yet, better fallback to LLE approach
+ return std::nullopt;
+ }
+ TSurface new_surface = GetUncachedSurface(gpu_addr, params);
+ bool modified = false;
+ for (auto& surface : overlaps) {
+ const SurfaceParams& src_params = surface->GetSurfaceParams();
+ if (src_params.target != SurfaceTarget::Texture2D) {
+ failed = true;
+ break;
+ }
+ if (src_params.height != params.height) {
+ failed = true;
+ break;
+ }
+ if (src_params.block_depth != params.block_depth ||
+ src_params.block_height != params.block_height) {
+ failed = true;
+ break;
+ }
+ const u32 offset = static_cast<u32>(surface->GetCacheAddr() - cache_addr);
+ const auto [x, y, z] = params.GetBlockOffsetXYZ(offset);
+ modified |= surface->IsModified();
+ const CopyParams copy_params(0, 0, 0, 0, 0, z, 0, 0, params.width, params.height,
+ 1);
+ ImageCopy(surface, new_surface, copy_params);
+ }
+ if (failed) {
+ return std::nullopt;
+ }
+ for (const auto& surface : overlaps) {
+ Unregister(surface);
+ }
+ new_surface->MarkAsModified(modified, Tick());
+ Register(new_surface);
+ auto view = new_surface->GetMainView();
+ return {{std::move(new_surface), view}};
+ } else {
+ for (const auto& surface : overlaps) {
+ if (!surface->MatchTarget(params.target)) {
+ if (overlaps.size() == 1 && surface->GetCacheAddr() == cache_addr) {
+ if (Settings::values.use_accurate_gpu_emulation) {
+ return std::nullopt;
+ }
+ Unregister(surface);
+ return InitializeSurface(gpu_addr, params, preserve_contents);
+ }
+ return std::nullopt;
+ }
+ if (surface->GetCacheAddr() != cache_addr) {
+ continue;
+ }
+ if (surface->MatchesStructure(params) == MatchStructureResult::FullMatch) {
+ return {{surface, surface->GetMainView()}};
+ }
+ }
+ return InitializeSurface(gpu_addr, params, preserve_contents);
+ }
+ }
+
+ /**
* Gets the starting address and parameters of a candidate surface and tries
* to find a matching surface within the cache. This is done in 3 big steps:
*
@@ -615,22 +718,9 @@ private:
* left blank.
* @param is_render Whether or not the surface is a render target.
**/
- std::pair<TSurface, TView> GetSurface(const GPUVAddr gpu_addr, const SurfaceParams& params,
- bool preserve_contents, bool is_render) {
- const auto host_ptr{system.GPU().MemoryManager().GetPointer(gpu_addr)};
- const auto cache_addr{ToCacheAddr(host_ptr)};
-
- // Step 0: guarantee a valid surface
- if (!cache_addr) {
- // Return a null surface if it's invalid
- SurfaceParams new_params = params;
- new_params.width = 1;
- new_params.height = 1;
- new_params.depth = 1;
- new_params.block_height = 0;
- new_params.block_depth = 0;
- return InitializeSurface(gpu_addr, new_params, false);
- }
+ std::pair<TSurface, TView> GetSurface(const GPUVAddr gpu_addr, const CacheAddr cache_addr,
+ const SurfaceParams& params, bool preserve_contents,
+ bool is_render) {
// Step 1
// Check Level 1 Cache for a fast structural match. If candidate surface
@@ -677,6 +767,15 @@ private:
}
}
+ // Check if it's a 3D texture
+ if (params.block_depth > 0) {
+ auto surface =
+ Manage3DSurfaces(overlaps, params, gpu_addr, cache_addr, preserve_contents);
+ if (surface) {
+ return *surface;
+ }
+ }
+
// Split cases between 1 overlap or many.
if (overlaps.size() == 1) {
TSurface current_surface = overlaps[0];
@@ -795,6 +894,41 @@ private:
}
/**
+ * Gets a null surface based on a target texture.
+ * @param target The target of the null surface.
+ */
+ TView GetNullSurface(SurfaceTarget target) {
+ const u32 i_target = static_cast<u32>(target);
+ if (const auto it = invalid_cache.find(i_target); it != invalid_cache.end()) {
+ return it->second->GetMainView();
+ }
+ SurfaceParams params{};
+ params.target = target;
+ params.is_tiled = false;
+ params.srgb_conversion = false;
+ params.is_layered = false;
+ params.block_width = 0;
+ params.block_height = 0;
+ params.block_depth = 0;
+ params.tile_width_spacing = 1;
+ params.width = 1;
+ params.height = 1;
+ params.depth = 1;
+ params.pitch = 4;
+ params.num_levels = 1;
+ params.emulated_levels = 1;
+ params.pixel_format = VideoCore::Surface::PixelFormat::RGBA16F;
+ params.type = VideoCore::Surface::SurfaceType::ColorTexture;
+ auto surface = CreateSurface(0ULL, params);
+ invalid_memory.clear();
+ invalid_memory.resize(surface->GetHostSizeInBytes(), 0U);
+ surface->UploadTexture(invalid_memory);
+ surface->MarkAsModified(false, Tick());
+ invalid_cache.emplace(i_target, surface);
+ return surface->GetMainView();
+ }
+
+ /**
* Gets the a source and destination starting address and parameters,
* and tries to deduce if they are supposed to be depth textures. If so, their
* parameters are modified and fixed into so.
@@ -835,12 +969,11 @@ private:
}
}
- const auto inherit_format = ([](SurfaceParams& to, TSurface from) {
+ const auto inherit_format = [](SurfaceParams& to, TSurface from) {
const SurfaceParams& params = from->GetSurfaceParams();
to.pixel_format = params.pixel_format;
- to.component_type = params.component_type;
to.type = params.type;
- });
+ };
// Now we got the cases where one or both is Depth and the other is not known
if (!incomplete_src) {
inherit_format(src_params, deduced_src.surface);
@@ -956,6 +1089,8 @@ private:
VideoCore::RasterizerInterface& rasterizer;
+ FormatLookupTable format_lookup_table;
+
u64 ticks{};
// Guards the cache for protection conflicts.
@@ -991,6 +1126,11 @@ private:
std::vector<TSurface> sampled_textures;
+ /// This cache stores null surfaces in order to be used as a placeholder
+ /// for invalid texture calls.
+ std::unordered_map<u32, TSurface> invalid_cache;
+ std::vector<u8> invalid_memory;
+
StagingCache staging_cache;
std::recursive_mutex mutex;
};
diff --git a/src/video_core/textures/astc.cpp b/src/video_core/textures/astc.cpp
index 58b608a36..33bd31865 100644
--- a/src/video_core/textures/astc.cpp
+++ b/src/video_core/textures/astc.cpp
@@ -92,11 +92,11 @@ private:
const unsigned int mask = 1 << m_NextBit++;
// clear the bit
- *m_CurByte &= ~mask;
+ *m_CurByte &= static_cast<unsigned char>(~mask);
// Write the bit, if necessary
if (b)
- *m_CurByte |= mask;
+ *m_CurByte |= static_cast<unsigned char>(mask);
// Next byte?
if (m_NextBit >= 8) {
@@ -137,7 +137,7 @@ public:
}
uint64_t mask = (1 << (end - start + 1)) - 1;
- return (m_Bits >> start) & mask;
+ return (m_Bits >> start) & static_cast<IntType>(mask);
}
private:
@@ -656,7 +656,7 @@ static IntType Replicate(const IntType& val, uint32_t numBits, uint32_t toBit) {
return 0;
if (toBit == 0)
return 0;
- IntType v = val & ((1 << numBits) - 1);
+ IntType v = val & static_cast<IntType>((1 << numBits) - 1);
IntType res = v;
uint32_t reslen = numBits;
while (reslen < toBit) {
@@ -666,8 +666,8 @@ static IntType Replicate(const IntType& val, uint32_t numBits, uint32_t toBit) {
comp = numBits - newshift;
numBits = newshift;
}
- res <<= numBits;
- res |= v >> comp;
+ res = static_cast<IntType>(res << numBits);
+ res = static_cast<IntType>(res | (v >> comp));
reslen += numBits;
}
return res;
@@ -714,7 +714,7 @@ public:
// Do nothing
return val;
} else if (oldDepth == 0 && newDepth != 0) {
- return (1 << newDepth) - 1;
+ return static_cast<ChannelType>((1 << newDepth) - 1);
} else if (newDepth > oldDepth) {
return Replicate(val, oldDepth, newDepth);
} else {
@@ -722,10 +722,11 @@ public:
if (newDepth == 0) {
return 0xFF;
} else {
- uint8_t bitsWasted = oldDepth - newDepth;
+ uint8_t bitsWasted = static_cast<uint8_t>(oldDepth - newDepth);
uint16_t v = static_cast<uint16_t>(val);
- v = (v + (1 << (bitsWasted - 1))) >> bitsWasted;
- v = ::std::min<uint16_t>(::std::max<uint16_t>(0, v), (1 << newDepth) - 1);
+ v = static_cast<uint16_t>((v + (1 << (bitsWasted - 1))) >> bitsWasted);
+ v = ::std::min<uint16_t>(::std::max<uint16_t>(0, v),
+ static_cast<uint16_t>((1 << newDepth) - 1));
return static_cast<uint8_t>(v);
}
}
@@ -1191,18 +1192,18 @@ static uint32_t SelectPartition(int32_t seed, int32_t x, int32_t y, int32_t z,
uint8_t seed11 = static_cast<uint8_t>((rnum >> 26) & 0xF);
uint8_t seed12 = static_cast<uint8_t>(((rnum >> 30) | (rnum << 2)) & 0xF);
- seed1 *= seed1;
- seed2 *= seed2;
- seed3 *= seed3;
- seed4 *= seed4;
- seed5 *= seed5;
- seed6 *= seed6;
- seed7 *= seed7;
- seed8 *= seed8;
- seed9 *= seed9;
- seed10 *= seed10;
- seed11 *= seed11;
- seed12 *= seed12;
+ seed1 = static_cast<uint8_t>(seed1 * seed1);
+ seed2 = static_cast<uint8_t>(seed2 * seed2);
+ seed3 = static_cast<uint8_t>(seed3 * seed3);
+ seed4 = static_cast<uint8_t>(seed4 * seed4);
+ seed5 = static_cast<uint8_t>(seed5 * seed5);
+ seed6 = static_cast<uint8_t>(seed6 * seed6);
+ seed7 = static_cast<uint8_t>(seed7 * seed7);
+ seed8 = static_cast<uint8_t>(seed8 * seed8);
+ seed9 = static_cast<uint8_t>(seed9 * seed9);
+ seed10 = static_cast<uint8_t>(seed10 * seed10);
+ seed11 = static_cast<uint8_t>(seed11 * seed11);
+ seed12 = static_cast<uint8_t>(seed12 * seed12);
int32_t sh1, sh2, sh3;
if (seed & 1) {
@@ -1214,18 +1215,18 @@ static uint32_t SelectPartition(int32_t seed, int32_t x, int32_t y, int32_t z,
}
sh3 = (seed & 0x10) ? sh1 : sh2;
- seed1 >>= sh1;
- seed2 >>= sh2;
- seed3 >>= sh1;
- seed4 >>= sh2;
- seed5 >>= sh1;
- seed6 >>= sh2;
- seed7 >>= sh1;
- seed8 >>= sh2;
- seed9 >>= sh3;
- seed10 >>= sh3;
- seed11 >>= sh3;
- seed12 >>= sh3;
+ seed1 = static_cast<uint8_t>(seed1 >> sh1);
+ seed2 = static_cast<uint8_t>(seed2 >> sh2);
+ seed3 = static_cast<uint8_t>(seed3 >> sh1);
+ seed4 = static_cast<uint8_t>(seed4 >> sh2);
+ seed5 = static_cast<uint8_t>(seed5 >> sh1);
+ seed6 = static_cast<uint8_t>(seed6 >> sh2);
+ seed7 = static_cast<uint8_t>(seed7 >> sh1);
+ seed8 = static_cast<uint8_t>(seed8 >> sh2);
+ seed9 = static_cast<uint8_t>(seed9 >> sh3);
+ seed10 = static_cast<uint8_t>(seed10 >> sh3);
+ seed11 = static_cast<uint8_t>(seed11 >> sh3);
+ seed12 = static_cast<uint8_t>(seed12 >> sh3);
int32_t a = seed1 * x + seed2 * y + seed11 * z + (rnum >> 14);
int32_t b = seed3 * x + seed4 * y + seed12 * z + (rnum >> 10);
@@ -1558,7 +1559,9 @@ static void DecompressBlock(const uint8_t inBuf[16], const uint32_t blockWidth,
// Make sure that higher non-texel bits are set to zero
const uint32_t clearByteStart = (weightParams.GetPackedBitSize() >> 3) + 1;
- texelWeightData[clearByteStart - 1] &= (1 << (weightParams.GetPackedBitSize() % 8)) - 1;
+ texelWeightData[clearByteStart - 1] =
+ texelWeightData[clearByteStart - 1] &
+ static_cast<uint8_t>((1 << (weightParams.GetPackedBitSize() % 8)) - 1);
memset(texelWeightData + clearByteStart, 0, 16 - clearByteStart);
std::vector<IntegerEncodedValue> texelWeightValues;
diff --git a/src/video_core/textures/decoders.h b/src/video_core/textures/decoders.h
index f1e3952bc..e5eac3f3b 100644
--- a/src/video_core/textures/decoders.h
+++ b/src/video_core/textures/decoders.h
@@ -12,6 +12,10 @@ namespace Tegra::Texture {
// GOBSize constant. Calculated by 64 bytes in x multiplied by 8 y coords, represents
// an small rect of (64/bytes_per_pixel)X8.
+inline std::size_t GetGOBSize() {
+ return 512;
+}
+
inline std::size_t GetGOBSizeShift() {
return 9;
}
diff --git a/src/video_core/textures/texture.h b/src/video_core/textures/texture.h
index 27c8ce975..8e82c6748 100644
--- a/src/video_core/textures/texture.h
+++ b/src/video_core/textures/texture.h
@@ -342,13 +342,14 @@ struct TSCEntry {
float GetLodBias() const {
// Sign extend the 13-bit value.
constexpr u32 mask = 1U << (13 - 1);
- return static_cast<s32>((mip_lod_bias ^ mask) - mask) / 256.0f;
+ return static_cast<float>(static_cast<s32>((mip_lod_bias ^ mask) - mask)) / 256.0f;
}
std::array<float, 4> GetBorderColor() const {
if (srgb_conversion) {
- return {srgb_border_color_r / 255.0f, srgb_border_color_g / 255.0f,
- srgb_border_color_b / 255.0f, border_color[3]};
+ return {static_cast<float>(srgb_border_color_r) / 255.0f,
+ static_cast<float>(srgb_border_color_g) / 255.0f,
+ static_cast<float>(srgb_border_color_b) / 255.0f, border_color[3]};
}
return border_color;
}
diff --git a/src/video_core/video_core.cpp b/src/video_core/video_core.cpp
index 60cda0ca3..8e947394c 100644
--- a/src/video_core/video_core.cpp
+++ b/src/video_core/video_core.cpp
@@ -28,7 +28,7 @@ std::unique_ptr<Tegra::GPU> CreateGPU(Core::System& system) {
u16 GetResolutionScaleFactor(const RendererBase& renderer) {
return static_cast<u16>(
- Settings::values.resolution_factor
+ Settings::values.resolution_factor != 0
? Settings::values.resolution_factor
: renderer.GetRenderWindow().GetFramebufferLayout().GetScalingRatio());
}
diff --git a/src/web_service/web_backend.cpp b/src/web_service/web_backend.cpp
index dc149d2ed..6683f459f 100644
--- a/src/web_service/web_backend.cpp
+++ b/src/web_service/web_backend.cpp
@@ -2,10 +2,12 @@
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
+#include <array>
#include <cstdlib>
#include <mutex>
#include <string>
#include <LUrlParser.h>
+#include <fmt/format.h>
#include <httplib.h>
#include "common/common_types.h"
#include "common/logging/log.h"
@@ -16,10 +18,10 @@ namespace WebService {
constexpr std::array<const char, 1> API_VERSION{'1'};
-constexpr u32 HTTP_PORT = 80;
-constexpr u32 HTTPS_PORT = 443;
+constexpr int HTTP_PORT = 80;
+constexpr int HTTPS_PORT = 443;
-constexpr u32 TIMEOUT_SECONDS = 30;
+constexpr std::size_t TIMEOUT_SECONDS = 30;
struct Client::Impl {
Impl(std::string host, std::string username, std::string token)
@@ -31,8 +33,9 @@ struct Client::Impl {
}
/// A generic function handles POST, GET and DELETE request together
- Common::WebResult GenericJson(const std::string& method, const std::string& path,
- const std::string& data, bool allow_anonymous) {
+ Common::WebResult GenericRequest(const std::string& method, const std::string& path,
+ const std::string& data, bool allow_anonymous,
+ const std::string& accept) {
if (jwt.empty()) {
UpdateJWT();
}
@@ -43,11 +46,11 @@ struct Client::Impl {
"Credentials needed"};
}
- auto result = GenericJson(method, path, data, jwt);
+ auto result = GenericRequest(method, path, data, accept, jwt);
if (result.result_string == "401") {
// Try again with new JWT
UpdateJWT();
- result = GenericJson(method, path, data, jwt);
+ result = GenericRequest(method, path, data, accept, jwt);
}
return result;
@@ -56,12 +59,13 @@ struct Client::Impl {
/**
* A generic function with explicit authentication method specified
* JWT is used if the jwt parameter is not empty
- * username + token is used if jwt is empty but username and token are not empty
- * anonymous if all of jwt, username and token are empty
+ * username + token is used if jwt is empty but username and token are
+ * not empty anonymous if all of jwt, username and token are empty
*/
- Common::WebResult GenericJson(const std::string& method, const std::string& path,
- const std::string& data, const std::string& jwt = "",
- const std::string& username = "", const std::string& token = "") {
+ Common::WebResult GenericRequest(const std::string& method, const std::string& path,
+ const std::string& data, const std::string& accept,
+ const std::string& jwt = "", const std::string& username = "",
+ const std::string& token = "") {
if (cli == nullptr) {
auto parsedUrl = LUrlParser::clParseURL::ParseURL(host);
int port;
@@ -132,8 +136,7 @@ struct Client::Impl {
return Common::WebResult{Common::WebResult::Code::WrongContent, ""};
}
- if (content_type->second.find("application/json") == std::string::npos &&
- content_type->second.find("text/html; charset=utf-8") == std::string::npos) {
+ if (content_type->second.find(accept) == std::string::npos) {
LOG_ERROR(WebService, "{} to {} returned wrong content: {}", method, host + path,
content_type->second);
return Common::WebResult{Common::WebResult::Code::WrongContent, "Wrong content"};
@@ -147,7 +150,7 @@ struct Client::Impl {
return;
}
- auto result = GenericJson("POST", "/jwt/internal", "", "", username, token);
+ auto result = GenericRequest("POST", "/jwt/internal", "", "text/html", "", username, token);
if (result.result_code != Common::WebResult::Code::Success) {
LOG_ERROR(WebService, "UpdateJWT failed");
} else {
@@ -180,16 +183,29 @@ Client::~Client() = default;
Common::WebResult Client::PostJson(const std::string& path, const std::string& data,
bool allow_anonymous) {
- return impl->GenericJson("POST", path, data, allow_anonymous);
+ return impl->GenericRequest("POST", path, data, allow_anonymous, "application/json");
}
Common::WebResult Client::GetJson(const std::string& path, bool allow_anonymous) {
- return impl->GenericJson("GET", path, "", allow_anonymous);
+ return impl->GenericRequest("GET", path, "", allow_anonymous, "application/json");
}
Common::WebResult Client::DeleteJson(const std::string& path, const std::string& data,
bool allow_anonymous) {
- return impl->GenericJson("DELETE", path, data, allow_anonymous);
+ return impl->GenericRequest("DELETE", path, data, allow_anonymous, "application/json");
+}
+
+Common::WebResult Client::GetPlain(const std::string& path, bool allow_anonymous) {
+ return impl->GenericRequest("GET", path, "", allow_anonymous, "text/plain");
+}
+
+Common::WebResult Client::GetImage(const std::string& path, bool allow_anonymous) {
+ return impl->GenericRequest("GET", path, "", allow_anonymous, "image/png");
+}
+
+Common::WebResult Client::GetExternalJWT(const std::string& audience) {
+ return impl->GenericRequest("POST", fmt::format("/jwt/external/{}", audience), "", false,
+ "text/html");
}
} // namespace WebService
diff --git a/src/web_service/web_backend.h b/src/web_service/web_backend.h
index c637e09df..04121f17e 100644
--- a/src/web_service/web_backend.h
+++ b/src/web_service/web_backend.h
@@ -46,6 +46,29 @@ public:
Common::WebResult DeleteJson(const std::string& path, const std::string& data,
bool allow_anonymous);
+ /**
+ * Gets a plain string from the specified path.
+ * @param path the URL segment after the host address.
+ * @param allow_anonymous If true, allow anonymous unauthenticated requests.
+ * @return the result of the request.
+ */
+ Common::WebResult GetPlain(const std::string& path, bool allow_anonymous);
+
+ /**
+ * Gets an PNG image from the specified path.
+ * @param path the URL segment after the host address.
+ * @param allow_anonymous If true, allow anonymous unauthenticated requests.
+ * @return the result of the request.
+ */
+ Common::WebResult GetImage(const std::string& path, bool allow_anonymous);
+
+ /**
+ * Requests an external JWT for the specific audience provided.
+ * @param audience the audience of the JWT requested.
+ * @return the result of the request.
+ */
+ Common::WebResult GetExternalJWT(const std::string& audience);
+
private:
struct Impl;
std::unique_ptr<Impl> impl;
diff --git a/src/yuzu/CMakeLists.txt b/src/yuzu/CMakeLists.txt
index ff1c1d985..11ae1e66e 100644
--- a/src/yuzu/CMakeLists.txt
+++ b/src/yuzu/CMakeLists.txt
@@ -78,11 +78,6 @@ add_executable(yuzu
configuration/configure_web.cpp
configuration/configure_web.h
configuration/configure_web.ui
- debugger/graphics/graphics_breakpoint_observer.cpp
- debugger/graphics/graphics_breakpoint_observer.h
- debugger/graphics/graphics_breakpoints.cpp
- debugger/graphics/graphics_breakpoints.h
- debugger/graphics/graphics_breakpoints_p.h
debugger/console.cpp
debugger/console.h
debugger/profiler.cpp
diff --git a/src/yuzu/configuration/configure_web.cpp b/src/yuzu/configuration/configure_web.cpp
index 336b062b3..8637f5b3c 100644
--- a/src/yuzu/configuration/configure_web.cpp
+++ b/src/yuzu/configuration/configure_web.cpp
@@ -11,6 +11,31 @@
#include "yuzu/configuration/configure_web.h"
#include "yuzu/uisettings.h"
+static constexpr char token_delimiter{':'};
+
+static std::string GenerateDisplayToken(const std::string& username, const std::string& token) {
+ if (username.empty() || token.empty()) {
+ return {};
+ }
+
+ const std::string unencoded_display_token{username + token_delimiter + token};
+ QByteArray b{unencoded_display_token.c_str()};
+ QByteArray b64 = b.toBase64();
+ return b64.toStdString();
+}
+
+static std::string UsernameFromDisplayToken(const std::string& display_token) {
+ const std::string unencoded_display_token{
+ QByteArray::fromBase64(display_token.c_str()).toStdString()};
+ return unencoded_display_token.substr(0, unencoded_display_token.find(token_delimiter));
+}
+
+static std::string TokenFromDisplayToken(const std::string& display_token) {
+ const std::string unencoded_display_token{
+ QByteArray::fromBase64(display_token.c_str()).toStdString()};
+ return unencoded_display_token.substr(unencoded_display_token.find(token_delimiter) + 1);
+}
+
ConfigureWeb::ConfigureWeb(QWidget* parent)
: QWidget(parent), ui(std::make_unique<Ui::ConfigureWeb>()) {
ui->setupUi(this);
@@ -63,13 +88,18 @@ void ConfigureWeb::SetConfiguration() {
ui->web_signup_link->setOpenExternalLinks(true);
ui->web_token_info_link->setOpenExternalLinks(true);
+ if (Settings::values.yuzu_username.empty()) {
+ ui->username->setText(tr("Unspecified"));
+ } else {
+ ui->username->setText(QString::fromStdString(Settings::values.yuzu_username));
+ }
+
ui->toggle_telemetry->setChecked(Settings::values.enable_telemetry);
- ui->edit_username->setText(QString::fromStdString(Settings::values.yuzu_username));
- ui->edit_token->setText(QString::fromStdString(Settings::values.yuzu_token));
+ ui->edit_token->setText(QString::fromStdString(
+ GenerateDisplayToken(Settings::values.yuzu_username, Settings::values.yuzu_token)));
// Connect after setting the values, to avoid calling OnLoginChanged now
connect(ui->edit_token, &QLineEdit::textChanged, this, &ConfigureWeb::OnLoginChanged);
- connect(ui->edit_username, &QLineEdit::textChanged, this, &ConfigureWeb::OnLoginChanged);
user_verified = true;
@@ -80,12 +110,13 @@ void ConfigureWeb::ApplyConfiguration() {
Settings::values.enable_telemetry = ui->toggle_telemetry->isChecked();
UISettings::values.enable_discord_presence = ui->toggle_discordrpc->isChecked();
if (user_verified) {
- Settings::values.yuzu_username = ui->edit_username->text().toStdString();
- Settings::values.yuzu_token = ui->edit_token->text().toStdString();
+ Settings::values.yuzu_username =
+ UsernameFromDisplayToken(ui->edit_token->text().toStdString());
+ Settings::values.yuzu_token = TokenFromDisplayToken(ui->edit_token->text().toStdString());
} else {
- QMessageBox::warning(this, tr("Username and token not verified"),
- tr("Username and token were not verified. The changes to your "
- "username and/or token have not been saved."));
+ QMessageBox::warning(
+ this, tr("Token not verified"),
+ tr("Token was not verified. The change to your token has not been saved."));
}
}
@@ -96,17 +127,15 @@ void ConfigureWeb::RefreshTelemetryID() {
}
void ConfigureWeb::OnLoginChanged() {
- if (ui->edit_username->text().isEmpty() && ui->edit_token->text().isEmpty()) {
+ if (ui->edit_token->text().isEmpty()) {
user_verified = true;
const QPixmap pixmap = QIcon::fromTheme(QStringLiteral("checked")).pixmap(16);
- ui->label_username_verified->setPixmap(pixmap);
ui->label_token_verified->setPixmap(pixmap);
} else {
user_verified = false;
const QPixmap pixmap = QIcon::fromTheme(QStringLiteral("failed")).pixmap(16);
- ui->label_username_verified->setPixmap(pixmap);
ui->label_token_verified->setPixmap(pixmap);
}
}
@@ -114,10 +143,11 @@ void ConfigureWeb::OnLoginChanged() {
void ConfigureWeb::VerifyLogin() {
ui->button_verify_login->setDisabled(true);
ui->button_verify_login->setText(tr("Verifying..."));
- verify_watcher.setFuture(QtConcurrent::run([username = ui->edit_username->text().toStdString(),
- token = ui->edit_token->text().toStdString()] {
- return Core::VerifyLogin(username, token);
- }));
+ verify_watcher.setFuture(QtConcurrent::run(
+ [username = UsernameFromDisplayToken(ui->edit_token->text().toStdString()),
+ token = TokenFromDisplayToken(ui->edit_token->text().toStdString())] {
+ return Core::VerifyLogin(username, token);
+ }));
}
void ConfigureWeb::OnLoginVerified() {
@@ -127,16 +157,15 @@ void ConfigureWeb::OnLoginVerified() {
user_verified = true;
const QPixmap pixmap = QIcon::fromTheme(QStringLiteral("checked")).pixmap(16);
- ui->label_username_verified->setPixmap(pixmap);
ui->label_token_verified->setPixmap(pixmap);
+ ui->username->setText(
+ QString::fromStdString(UsernameFromDisplayToken(ui->edit_token->text().toStdString())));
} else {
const QPixmap pixmap = QIcon::fromTheme(QStringLiteral("failed")).pixmap(16);
- ui->label_username_verified->setPixmap(pixmap);
ui->label_token_verified->setPixmap(pixmap);
-
- QMessageBox::critical(
- this, tr("Verification failed"),
- tr("Verification failed. Check that you have entered your username and token "
- "correctly, and that your internet connection is working."));
+ ui->username->setText(tr("Unspecified"));
+ QMessageBox::critical(this, tr("Verification failed"),
+ tr("Verification failed. Check that you have entered your token "
+ "correctly, and that your internet connection is working."));
}
}
diff --git a/src/yuzu/configuration/configure_web.ui b/src/yuzu/configuration/configure_web.ui
index 2f4b9dd73..8c07d1165 100644
--- a/src/yuzu/configuration/configure_web.ui
+++ b/src/yuzu/configuration/configure_web.ui
@@ -55,11 +55,7 @@
</widget>
</item>
<item row="0" column="1" colspan="3">
- <widget class="QLineEdit" name="edit_username">
- <property name="maxLength">
- <number>36</number>
- </property>
- </widget>
+ <widget class="QLabel" name="username" />
</item>
<item row="1" column="0">
<widget class="QLabel" name="label_token">
@@ -79,14 +75,10 @@
</property>
</widget>
</item>
- <item row="0" column="4">
- <widget class="QLabel" name="label_username_verified">
- </widget>
- </item>
<item row="1" column="1" colspan="3">
<widget class="QLineEdit" name="edit_token">
<property name="maxLength">
- <number>36</number>
+ <number>80</number>
</property>
<property name="echoMode">
<enum>QLineEdit::Password</enum>
diff --git a/src/yuzu/debugger/graphics/graphics_breakpoint_observer.cpp b/src/yuzu/debugger/graphics/graphics_breakpoint_observer.cpp
deleted file mode 100644
index 5f459ccfb..000000000
--- a/src/yuzu/debugger/graphics/graphics_breakpoint_observer.cpp
+++ /dev/null
@@ -1,27 +0,0 @@
-// Copyright 2014 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#include <QMetaType>
-#include "yuzu/debugger/graphics/graphics_breakpoint_observer.h"
-
-BreakPointObserverDock::BreakPointObserverDock(std::shared_ptr<Tegra::DebugContext> debug_context,
- const QString& title, QWidget* parent)
- : QDockWidget(title, parent), BreakPointObserver(debug_context) {
- qRegisterMetaType<Tegra::DebugContext::Event>("Tegra::DebugContext::Event");
-
- connect(this, &BreakPointObserverDock::Resumed, this, &BreakPointObserverDock::OnResumed);
-
- // NOTE: This signal is emitted from a non-GUI thread, but connect() takes
- // care of delaying its handling to the GUI thread.
- connect(this, &BreakPointObserverDock::BreakPointHit, this,
- &BreakPointObserverDock::OnBreakPointHit, Qt::BlockingQueuedConnection);
-}
-
-void BreakPointObserverDock::OnMaxwellBreakPointHit(Tegra::DebugContext::Event event, void* data) {
- emit BreakPointHit(event, data);
-}
-
-void BreakPointObserverDock::OnMaxwellResume() {
- emit Resumed();
-}
diff --git a/src/yuzu/debugger/graphics/graphics_breakpoint_observer.h b/src/yuzu/debugger/graphics/graphics_breakpoint_observer.h
deleted file mode 100644
index ab32f0115..000000000
--- a/src/yuzu/debugger/graphics/graphics_breakpoint_observer.h
+++ /dev/null
@@ -1,33 +0,0 @@
-// Copyright 2014 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include <QDockWidget>
-#include "video_core/debug_utils/debug_utils.h"
-
-/**
- * Utility class which forwards calls to OnMaxwellBreakPointHit and OnMaxwellResume to public slots.
- * This is because the Maxwell breakpoint callbacks are called from a non-GUI thread, while
- * the widget usually wants to perform reactions in the GUI thread.
- */
-class BreakPointObserverDock : public QDockWidget,
- protected Tegra::DebugContext::BreakPointObserver {
- Q_OBJECT
-
-public:
- BreakPointObserverDock(std::shared_ptr<Tegra::DebugContext> debug_context, const QString& title,
- QWidget* parent = nullptr);
-
- void OnMaxwellBreakPointHit(Tegra::DebugContext::Event event, void* data) override;
- void OnMaxwellResume() override;
-
-signals:
- void Resumed();
- void BreakPointHit(Tegra::DebugContext::Event event, void* data);
-
-private:
- virtual void OnBreakPointHit(Tegra::DebugContext::Event event, void* data) = 0;
- virtual void OnResumed() = 0;
-};
diff --git a/src/yuzu/debugger/graphics/graphics_breakpoints.cpp b/src/yuzu/debugger/graphics/graphics_breakpoints.cpp
deleted file mode 100644
index 1c80082a4..000000000
--- a/src/yuzu/debugger/graphics/graphics_breakpoints.cpp
+++ /dev/null
@@ -1,221 +0,0 @@
-// Copyright 2014 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#include <QLabel>
-#include <QMetaType>
-#include <QPushButton>
-#include <QTreeView>
-#include <QVBoxLayout>
-#include "common/assert.h"
-#include "yuzu/debugger/graphics/graphics_breakpoints.h"
-#include "yuzu/debugger/graphics/graphics_breakpoints_p.h"
-
-BreakPointModel::BreakPointModel(std::shared_ptr<Tegra::DebugContext> debug_context,
- QObject* parent)
- : QAbstractListModel(parent), context_weak(debug_context),
- at_breakpoint(debug_context->at_breakpoint),
- active_breakpoint(debug_context->active_breakpoint) {}
-
-int BreakPointModel::columnCount(const QModelIndex& parent) const {
- return 1;
-}
-
-int BreakPointModel::rowCount(const QModelIndex& parent) const {
- return static_cast<int>(Tegra::DebugContext::Event::NumEvents);
-}
-
-QVariant BreakPointModel::data(const QModelIndex& index, int role) const {
- const auto event = static_cast<Tegra::DebugContext::Event>(index.row());
-
- switch (role) {
- case Qt::DisplayRole: {
- if (index.column() == 0) {
- return DebugContextEventToString(event);
- }
- break;
- }
-
- case Qt::CheckStateRole: {
- if (index.column() == 0)
- return data(index, Role_IsEnabled).toBool() ? Qt::Checked : Qt::Unchecked;
- break;
- }
-
- case Qt::BackgroundRole: {
- if (at_breakpoint && index.row() == static_cast<int>(active_breakpoint)) {
- return QBrush(QColor(0xE0, 0xE0, 0x10));
- }
- break;
- }
-
- case Role_IsEnabled: {
- auto context = context_weak.lock();
- return context && context->breakpoints[(int)event].enabled;
- }
-
- default:
- break;
- }
- return QVariant();
-}
-
-Qt::ItemFlags BreakPointModel::flags(const QModelIndex& index) const {
- if (!index.isValid())
- return 0;
-
- Qt::ItemFlags flags = Qt::ItemIsEnabled;
- if (index.column() == 0)
- flags |= Qt::ItemIsUserCheckable;
- return flags;
-}
-
-bool BreakPointModel::setData(const QModelIndex& index, const QVariant& value, int role) {
- const auto event = static_cast<Tegra::DebugContext::Event>(index.row());
-
- switch (role) {
- case Qt::CheckStateRole: {
- if (index.column() != 0)
- return false;
-
- auto context = context_weak.lock();
- if (!context)
- return false;
-
- context->breakpoints[(int)event].enabled = value == Qt::Checked;
- QModelIndex changed_index = createIndex(index.row(), 0);
- emit dataChanged(changed_index, changed_index);
- return true;
- }
- }
-
- return false;
-}
-
-void BreakPointModel::OnBreakPointHit(Tegra::DebugContext::Event event) {
- auto context = context_weak.lock();
- if (!context)
- return;
-
- active_breakpoint = context->active_breakpoint;
- at_breakpoint = context->at_breakpoint;
- emit dataChanged(createIndex(static_cast<int>(event), 0),
- createIndex(static_cast<int>(event), 0));
-}
-
-void BreakPointModel::OnResumed() {
- auto context = context_weak.lock();
- if (!context)
- return;
-
- at_breakpoint = context->at_breakpoint;
- emit dataChanged(createIndex(static_cast<int>(active_breakpoint), 0),
- createIndex(static_cast<int>(active_breakpoint), 0));
- active_breakpoint = context->active_breakpoint;
-}
-
-QString BreakPointModel::DebugContextEventToString(Tegra::DebugContext::Event event) {
- switch (event) {
- case Tegra::DebugContext::Event::MaxwellCommandLoaded:
- return tr("Maxwell command loaded");
- case Tegra::DebugContext::Event::MaxwellCommandProcessed:
- return tr("Maxwell command processed");
- case Tegra::DebugContext::Event::IncomingPrimitiveBatch:
- return tr("Incoming primitive batch");
- case Tegra::DebugContext::Event::FinishedPrimitiveBatch:
- return tr("Finished primitive batch");
- case Tegra::DebugContext::Event::NumEvents:
- break;
- }
-
- return tr("Unknown debug context event");
-}
-
-GraphicsBreakPointsWidget::GraphicsBreakPointsWidget(
- std::shared_ptr<Tegra::DebugContext> debug_context, QWidget* parent)
- : QDockWidget(tr("Maxwell Breakpoints"), parent), Tegra::DebugContext::BreakPointObserver(
- debug_context) {
- setObjectName(QStringLiteral("TegraBreakPointsWidget"));
-
- status_text = new QLabel(tr("Emulation running"));
- resume_button = new QPushButton(tr("Resume"));
- resume_button->setEnabled(false);
-
- breakpoint_model = new BreakPointModel(debug_context, this);
- breakpoint_list = new QTreeView;
- breakpoint_list->setRootIsDecorated(false);
- breakpoint_list->setHeaderHidden(true);
- breakpoint_list->setModel(breakpoint_model);
-
- qRegisterMetaType<Tegra::DebugContext::Event>("Tegra::DebugContext::Event");
-
- connect(breakpoint_list, &QTreeView::doubleClicked, this,
- &GraphicsBreakPointsWidget::OnItemDoubleClicked);
-
- connect(resume_button, &QPushButton::clicked, this,
- &GraphicsBreakPointsWidget::OnResumeRequested);
-
- connect(this, &GraphicsBreakPointsWidget::BreakPointHit, this,
- &GraphicsBreakPointsWidget::OnBreakPointHit, Qt::BlockingQueuedConnection);
- connect(this, &GraphicsBreakPointsWidget::Resumed, this, &GraphicsBreakPointsWidget::OnResumed);
-
- connect(this, &GraphicsBreakPointsWidget::BreakPointHit, breakpoint_model,
- &BreakPointModel::OnBreakPointHit, Qt::BlockingQueuedConnection);
- connect(this, &GraphicsBreakPointsWidget::Resumed, breakpoint_model,
- &BreakPointModel::OnResumed);
-
- connect(this, &GraphicsBreakPointsWidget::BreakPointsChanged,
- [this](const QModelIndex& top_left, const QModelIndex& bottom_right) {
- breakpoint_model->dataChanged(top_left, bottom_right);
- });
-
- QWidget* main_widget = new QWidget;
- auto main_layout = new QVBoxLayout;
- {
- auto sub_layout = new QHBoxLayout;
- sub_layout->addWidget(status_text);
- sub_layout->addWidget(resume_button);
- main_layout->addLayout(sub_layout);
- }
- main_layout->addWidget(breakpoint_list);
- main_widget->setLayout(main_layout);
-
- setWidget(main_widget);
-}
-
-void GraphicsBreakPointsWidget::OnMaxwellBreakPointHit(Event event, void* data) {
- // Process in GUI thread
- emit BreakPointHit(event, data);
-}
-
-void GraphicsBreakPointsWidget::OnBreakPointHit(Tegra::DebugContext::Event event, void* data) {
- status_text->setText(tr("Emulation halted at breakpoint"));
- resume_button->setEnabled(true);
-}
-
-void GraphicsBreakPointsWidget::OnMaxwellResume() {
- // Process in GUI thread
- emit Resumed();
-}
-
-void GraphicsBreakPointsWidget::OnResumed() {
- status_text->setText(tr("Emulation running"));
- resume_button->setEnabled(false);
-}
-
-void GraphicsBreakPointsWidget::OnResumeRequested() {
- if (auto context = context_weak.lock())
- context->Resume();
-}
-
-void GraphicsBreakPointsWidget::OnItemDoubleClicked(const QModelIndex& index) {
- if (!index.isValid())
- return;
-
- QModelIndex check_index = breakpoint_list->model()->index(index.row(), 0);
- QVariant enabled = breakpoint_list->model()->data(check_index, Qt::CheckStateRole);
- QVariant new_state = Qt::Unchecked;
- if (enabled == Qt::Unchecked)
- new_state = Qt::Checked;
- breakpoint_list->model()->setData(check_index, new_state, Qt::CheckStateRole);
-}
diff --git a/src/yuzu/debugger/graphics/graphics_breakpoints.h b/src/yuzu/debugger/graphics/graphics_breakpoints.h
deleted file mode 100644
index a920a2ae5..000000000
--- a/src/yuzu/debugger/graphics/graphics_breakpoints.h
+++ /dev/null
@@ -1,45 +0,0 @@
-// Copyright 2014 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include <memory>
-#include <QDockWidget>
-#include "video_core/debug_utils/debug_utils.h"
-
-class QLabel;
-class QPushButton;
-class QTreeView;
-
-class BreakPointModel;
-
-class GraphicsBreakPointsWidget : public QDockWidget, Tegra::DebugContext::BreakPointObserver {
- Q_OBJECT
-
- using Event = Tegra::DebugContext::Event;
-
-public:
- explicit GraphicsBreakPointsWidget(std::shared_ptr<Tegra::DebugContext> debug_context,
- QWidget* parent = nullptr);
-
- void OnMaxwellBreakPointHit(Tegra::DebugContext::Event event, void* data) override;
- void OnMaxwellResume() override;
-
-signals:
- void Resumed();
- void BreakPointHit(Tegra::DebugContext::Event event, void* data);
- void BreakPointsChanged(const QModelIndex& topLeft, const QModelIndex& bottomRight);
-
-private:
- void OnBreakPointHit(Tegra::DebugContext::Event event, void* data);
- void OnItemDoubleClicked(const QModelIndex&);
- void OnResumeRequested();
- void OnResumed();
-
- QLabel* status_text;
- QPushButton* resume_button;
-
- BreakPointModel* breakpoint_model;
- QTreeView* breakpoint_list;
-};
diff --git a/src/yuzu/debugger/graphics/graphics_breakpoints_p.h b/src/yuzu/debugger/graphics/graphics_breakpoints_p.h
deleted file mode 100644
index fb488e38f..000000000
--- a/src/yuzu/debugger/graphics/graphics_breakpoints_p.h
+++ /dev/null
@@ -1,37 +0,0 @@
-// Copyright 2014 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include <memory>
-#include <QAbstractListModel>
-#include "video_core/debug_utils/debug_utils.h"
-
-class BreakPointModel : public QAbstractListModel {
- Q_OBJECT
-
-public:
- enum {
- Role_IsEnabled = Qt::UserRole,
- };
-
- BreakPointModel(std::shared_ptr<Tegra::DebugContext> context, QObject* parent);
-
- int columnCount(const QModelIndex& parent = QModelIndex()) const override;
- int rowCount(const QModelIndex& parent = QModelIndex()) const override;
- QVariant data(const QModelIndex& index, int role = Qt::DisplayRole) const override;
- Qt::ItemFlags flags(const QModelIndex& index) const override;
-
- bool setData(const QModelIndex& index, const QVariant& value, int role = Qt::EditRole) override;
-
- void OnBreakPointHit(Tegra::DebugContext::Event event);
- void OnResumed();
-
-private:
- static QString DebugContextEventToString(Tegra::DebugContext::Event event);
-
- std::weak_ptr<Tegra::DebugContext> context_weak;
- bool at_breakpoint;
- Tegra::DebugContext::Event active_breakpoint;
-};
diff --git a/src/yuzu/debugger/wait_tree.cpp b/src/yuzu/debugger/wait_tree.cpp
index 188f798c0..727bd8a94 100644
--- a/src/yuzu/debugger/wait_tree.cpp
+++ b/src/yuzu/debugger/wait_tree.cpp
@@ -57,7 +57,7 @@ std::size_t WaitTreeItem::Row() const {
std::vector<std::unique_ptr<WaitTreeThread>> WaitTreeItem::MakeThreadItemList() {
std::vector<std::unique_ptr<WaitTreeThread>> item_list;
std::size_t row = 0;
- auto add_threads = [&](const std::vector<Kernel::SharedPtr<Kernel::Thread>>& threads) {
+ auto add_threads = [&](const std::vector<std::shared_ptr<Kernel::Thread>>& threads) {
for (std::size_t i = 0; i < threads.size(); ++i) {
item_list.push_back(std::make_unique<WaitTreeThread>(*threads[i]));
item_list.back()->row = row;
@@ -80,7 +80,7 @@ QString WaitTreeText::GetText() const {
WaitTreeMutexInfo::WaitTreeMutexInfo(VAddr mutex_address, const Kernel::HandleTable& handle_table)
: mutex_address(mutex_address) {
- mutex_value = Memory::Read32(mutex_address);
+ mutex_value = Core::System::GetInstance().Memory().Read32(mutex_address);
owner_handle = static_cast<Kernel::Handle>(mutex_value & Kernel::Mutex::MutexOwnerMask);
owner = handle_table.Get<Kernel::Thread>(owner_handle);
}
@@ -115,10 +115,11 @@ std::vector<std::unique_ptr<WaitTreeItem>> WaitTreeCallstack::GetChildren() cons
std::vector<std::unique_ptr<WaitTreeItem>> list;
constexpr std::size_t BaseRegister = 29;
+ auto& memory = Core::System::GetInstance().Memory();
u64 base_pointer = thread.GetContext().cpu_registers[BaseRegister];
while (base_pointer != 0) {
- const u64 lr = Memory::Read64(base_pointer + sizeof(u64));
+ const u64 lr = memory.Read64(base_pointer + sizeof(u64));
if (lr == 0) {
break;
}
@@ -126,7 +127,7 @@ std::vector<std::unique_ptr<WaitTreeItem>> WaitTreeCallstack::GetChildren() cons
list.push_back(std::make_unique<WaitTreeText>(
tr("0x%1").arg(lr - sizeof(u32), 16, 16, QLatin1Char{'0'})));
- base_pointer = Memory::Read64(base_pointer);
+ base_pointer = memory.Read64(base_pointer);
}
return list;
@@ -172,8 +173,8 @@ std::vector<std::unique_ptr<WaitTreeItem>> WaitTreeWaitObject::GetChildren() con
return list;
}
-WaitTreeObjectList::WaitTreeObjectList(
- const std::vector<Kernel::SharedPtr<Kernel::WaitObject>>& list, bool w_all)
+WaitTreeObjectList::WaitTreeObjectList(const std::vector<std::shared_ptr<Kernel::WaitObject>>& list,
+ bool w_all)
: object_list(list), wait_all(w_all) {}
WaitTreeObjectList::~WaitTreeObjectList() = default;
@@ -325,7 +326,7 @@ std::vector<std::unique_ptr<WaitTreeItem>> WaitTreeThread::GetChildren() const {
WaitTreeEvent::WaitTreeEvent(const Kernel::ReadableEvent& object) : WaitTreeWaitObject(object) {}
WaitTreeEvent::~WaitTreeEvent() = default;
-WaitTreeThreadList::WaitTreeThreadList(const std::vector<Kernel::SharedPtr<Kernel::Thread>>& list)
+WaitTreeThreadList::WaitTreeThreadList(const std::vector<std::shared_ptr<Kernel::Thread>>& list)
: thread_list(list) {}
WaitTreeThreadList::~WaitTreeThreadList() = default;
diff --git a/src/yuzu/debugger/wait_tree.h b/src/yuzu/debugger/wait_tree.h
index f2b13be24..631274a5f 100644
--- a/src/yuzu/debugger/wait_tree.h
+++ b/src/yuzu/debugger/wait_tree.h
@@ -83,7 +83,7 @@ private:
VAddr mutex_address;
u32 mutex_value;
Kernel::Handle owner_handle;
- Kernel::SharedPtr<Kernel::Thread> owner;
+ std::shared_ptr<Kernel::Thread> owner;
};
class WaitTreeCallstack : public WaitTreeExpandableItem {
@@ -116,15 +116,14 @@ protected:
class WaitTreeObjectList : public WaitTreeExpandableItem {
Q_OBJECT
public:
- WaitTreeObjectList(const std::vector<Kernel::SharedPtr<Kernel::WaitObject>>& list,
- bool wait_all);
+ WaitTreeObjectList(const std::vector<std::shared_ptr<Kernel::WaitObject>>& list, bool wait_all);
~WaitTreeObjectList() override;
QString GetText() const override;
std::vector<std::unique_ptr<WaitTreeItem>> GetChildren() const override;
private:
- const std::vector<Kernel::SharedPtr<Kernel::WaitObject>>& object_list;
+ const std::vector<std::shared_ptr<Kernel::WaitObject>>& object_list;
bool wait_all;
};
@@ -149,14 +148,14 @@ public:
class WaitTreeThreadList : public WaitTreeExpandableItem {
Q_OBJECT
public:
- explicit WaitTreeThreadList(const std::vector<Kernel::SharedPtr<Kernel::Thread>>& list);
+ explicit WaitTreeThreadList(const std::vector<std::shared_ptr<Kernel::Thread>>& list);
~WaitTreeThreadList() override;
QString GetText() const override;
std::vector<std::unique_ptr<WaitTreeItem>> GetChildren() const override;
private:
- const std::vector<Kernel::SharedPtr<Kernel::Thread>>& thread_list;
+ const std::vector<std::shared_ptr<Kernel::Thread>>& thread_list;
};
class WaitTreeModel : public QAbstractItemModel {
diff --git a/src/yuzu/main.cpp b/src/yuzu/main.cpp
index 160613ee1..b21fbf826 100644
--- a/src/yuzu/main.cpp
+++ b/src/yuzu/main.cpp
@@ -93,7 +93,6 @@ static FileSys::VirtualFile VfsDirectoryCreateFileWrapper(const FileSys::Virtual
#include "core/perf_stats.h"
#include "core/settings.h"
#include "core/telemetry_session.h"
-#include "video_core/debug_utils/debug_utils.h"
#include "yuzu/about_dialog.h"
#include "yuzu/bootmanager.h"
#include "yuzu/compatdb.h"
@@ -101,7 +100,6 @@ static FileSys::VirtualFile VfsDirectoryCreateFileWrapper(const FileSys::Virtual
#include "yuzu/configuration/config.h"
#include "yuzu/configuration/configure_dialog.h"
#include "yuzu/debugger/console.h"
-#include "yuzu/debugger/graphics/graphics_breakpoints.h"
#include "yuzu/debugger/profiler.h"
#include "yuzu/debugger/wait_tree.h"
#include "yuzu/discord.h"
@@ -187,8 +185,6 @@ GMainWindow::GMainWindow()
provider(std::make_unique<FileSys::ManualContentProvider>()) {
InitializeLogging();
- debug_context = Tegra::DebugContext::Construct();
-
setAcceptDrops(true);
ui.setupUi(this);
statusBar()->hide();
@@ -495,11 +491,6 @@ void GMainWindow::InitializeDebugWidgets() {
debug_menu->addAction(microProfileDialog->toggleViewAction());
#endif
- graphicsBreakpointsWidget = new GraphicsBreakPointsWidget(debug_context, this);
- addDockWidget(Qt::RightDockWidgetArea, graphicsBreakpointsWidget);
- graphicsBreakpointsWidget->hide();
- debug_menu->addAction(graphicsBreakpointsWidget->toggleViewAction());
-
waitTreeWidget = new WaitTreeWidget(this);
addDockWidget(Qt::LeftDockWidgetArea, waitTreeWidget);
waitTreeWidget->hide();
@@ -817,6 +808,9 @@ QStringList GMainWindow::GetUnsupportedGLExtensions() {
if (!GLAD_GL_ARB_multi_bind) {
unsupported_ext.append(QStringLiteral("ARB_multi_bind"));
}
+ if (!GLAD_GL_ARB_clip_control) {
+ unsupported_ext.append(QStringLiteral("ARB_clip_control"));
+ }
// Extensions required to support some texture formats.
if (!GLAD_GL_EXT_texture_compression_s3tc) {
@@ -866,8 +860,6 @@ bool GMainWindow::LoadROM(const QString& filename) {
Core::System& system{Core::System::GetInstance()};
system.SetFilesystem(vfs);
- system.SetGPUDebugContext(debug_context);
-
system.SetAppletFrontendSet({
nullptr, // Parental Controls
std::make_unique<QtErrorDisplay>(*this), //
diff --git a/src/yuzu/main.h b/src/yuzu/main.h
index 7f46bea2b..a56f9a981 100644
--- a/src/yuzu/main.h
+++ b/src/yuzu/main.h
@@ -22,7 +22,6 @@ class Config;
class EmuThread;
class GameList;
class GImageInfo;
-class GraphicsBreakPointsWidget;
class GRenderWindow;
class LoadingScreen;
class MicroProfileDialog;
@@ -42,10 +41,6 @@ class ManualContentProvider;
class VfsFilesystem;
} // namespace FileSys
-namespace Tegra {
-class DebugContext;
-}
-
enum class EmulatedDirectoryTarget {
NAND,
SDMC,
@@ -223,8 +218,6 @@ private:
Ui::MainWindow ui;
- std::shared_ptr<Tegra::DebugContext> debug_context;
-
GRenderWindow* render_window;
GameList* game_list;
LoadingScreen* loading_screen;
@@ -255,7 +248,6 @@ private:
// Debugger panes
ProfilerWidget* profilerWidget;
MicroProfileDialog* microProfileDialog;
- GraphicsBreakPointsWidget* graphicsBreakpointsWidget;
WaitTreeWidget* waitTreeWidget;
QAction* actions_recent_files[max_recent_files_item];
diff --git a/src/yuzu/main.ui b/src/yuzu/main.ui
index a1ce3c0c3..21f422500 100644
--- a/src/yuzu/main.ui
+++ b/src/yuzu/main.ui
@@ -140,11 +140,6 @@
<string>Load Folder...</string>
</property>
</action>
- <action name="action_Load_Symbol_Map">
- <property name="text">
- <string>Load Symbol Map...</string>
- </property>
- </action>
<action name="action_Exit">
<property name="text">
<string>E&amp;xit</string>
@@ -221,14 +216,6 @@
<string>Show Status Bar</string>
</property>
</action>
- <action name="action_Select_Game_List_Root">
- <property name="text">
- <string>Select Game Directory...</string>
- </property>
- <property name="toolTip">
- <string>Selects a folder to display in the game list</string>
- </property>
- </action>
<action name="action_Select_NAND_Directory">
<property name="text">
<string>Select NAND Directory...</string>
diff --git a/src/yuzu_cmd/emu_window/emu_window_sdl2_gl.cpp b/src/yuzu_cmd/emu_window/emu_window_sdl2_gl.cpp
index f91b071bf..6fde694a2 100644
--- a/src/yuzu_cmd/emu_window/emu_window_sdl2_gl.cpp
+++ b/src/yuzu_cmd/emu_window/emu_window_sdl2_gl.cpp
@@ -50,7 +50,7 @@ private:
};
bool EmuWindow_SDL2_GL::SupportsRequiredGLExtensions() {
- std::vector<std::string> unsupported_ext;
+ std::vector<std::string_view> unsupported_ext;
if (!GLAD_GL_ARB_buffer_storage)
unsupported_ext.push_back("ARB_buffer_storage");
@@ -62,6 +62,8 @@ bool EmuWindow_SDL2_GL::SupportsRequiredGLExtensions() {
unsupported_ext.push_back("ARB_texture_mirror_clamp_to_edge");
if (!GLAD_GL_ARB_multi_bind)
unsupported_ext.push_back("ARB_multi_bind");
+ if (!GLAD_GL_ARB_clip_control)
+ unsupported_ext.push_back("ARB_clip_control");
// Extensions required to support some texture formats.
if (!GLAD_GL_EXT_texture_compression_s3tc)
@@ -71,8 +73,8 @@ bool EmuWindow_SDL2_GL::SupportsRequiredGLExtensions() {
if (!GLAD_GL_ARB_depth_buffer_float)
unsupported_ext.push_back("ARB_depth_buffer_float");
- for (const std::string& ext : unsupported_ext)
- LOG_CRITICAL(Frontend, "Unsupported GL extension: {}", ext);
+ for (const auto& extension : unsupported_ext)
+ LOG_CRITICAL(Frontend, "Unsupported GL extension: {}", extension);
return unsupported_ext.empty();
}