summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
-rw-r--r--CMakeLists.txt10
m---------externals/opus0
-rw-r--r--src/common/CMakeLists.txt6
-rw-r--r--src/common/memory_hook.cpp (renamed from src/core/memory_hook.cpp)6
-rw-r--r--src/common/memory_hook.h (renamed from src/core/memory_hook.h)4
-rw-r--r--src/common/page_table.cpp29
-rw-r--r--src/common/page_table.h80
-rw-r--r--src/common/uint128.cpp41
-rw-r--r--src/common/uint128.h14
-rw-r--r--src/core/CMakeLists.txt2
-rw-r--r--src/core/arm/dynarmic/arm_dynarmic.cpp5
-rw-r--r--src/core/arm/dynarmic/arm_dynarmic.h4
-rw-r--r--src/core/core_timing_util.cpp6
-rw-r--r--src/core/core_timing_util.h3
-rw-r--r--src/core/hle/ipc_helpers.h48
-rw-r--r--src/core/hle/kernel/client_port.cpp9
-rw-r--r--src/core/hle/kernel/process.cpp13
-rw-r--r--src/core/hle/kernel/process.h6
-rw-r--r--src/core/hle/kernel/scheduler.cpp8
-rw-r--r--src/core/hle/kernel/server_port.cpp4
-rw-r--r--src/core/hle/kernel/server_port.h35
-rw-r--r--src/core/hle/kernel/svc.cpp32
-rw-r--r--src/core/hle/kernel/thread.cpp19
-rw-r--r--src/core/hle/kernel/thread.h13
-rw-r--r--src/core/hle/kernel/vm_manager.cpp6
-rw-r--r--src/core/hle/kernel/vm_manager.h9
-rw-r--r--src/core/hle/result.h18
-rw-r--r--src/core/hle/service/nvdrv/devices/nvhost_as_gpu.cpp3
-rw-r--r--src/core/hle/service/service.cpp3
-rw-r--r--src/core/hle/service/sm/sm.h2
-rw-r--r--src/core/hle/service/vi/vi.cpp38
-rw-r--r--src/core/hle/service/vi/vi.h40
-rw-r--r--src/core/hle/service/vi/vi_m.cpp12
-rw-r--r--src/core/hle/service/vi/vi_m.h19
-rw-r--r--src/core/hle/service/vi/vi_s.cpp12
-rw-r--r--src/core/hle/service/vi/vi_s.h19
-rw-r--r--src/core/hle/service/vi/vi_u.cpp12
-rw-r--r--src/core/hle/service/vi/vi_u.h19
-rw-r--r--src/core/memory.cpp218
-rw-r--r--src/core/memory.h80
-rw-r--r--src/core/memory_setup.h19
-rw-r--r--src/input_common/CMakeLists.txt15
-rw-r--r--src/input_common/main.cpp23
-rw-r--r--src/input_common/main.h2
-rw-r--r--src/input_common/sdl/sdl.cpp636
-rw-r--r--src/input_common/sdl/sdl.h53
-rw-r--r--src/input_common/sdl/sdl_impl.cpp669
-rw-r--r--src/input_common/sdl/sdl_impl.h64
-rw-r--r--src/tests/core/arm/arm_test_common.cpp3
-rw-r--r--src/tests/core/arm/arm_test_common.h8
-rw-r--r--src/video_core/CMakeLists.txt2
-rw-r--r--src/video_core/dma_pusher.cpp7
-rw-r--r--src/video_core/engines/kepler_memory.cpp15
-rw-r--r--src/video_core/engines/maxwell_3d.cpp42
-rw-r--r--src/video_core/engines/maxwell_dma.cpp27
-rw-r--r--src/video_core/gpu.cpp13
-rw-r--r--src/video_core/gpu.h11
-rw-r--r--src/video_core/gpu_asynch.cpp6
-rw-r--r--src/video_core/gpu_asynch.h6
-rw-r--r--src/video_core/gpu_synch.cpp6
-rw-r--r--src/video_core/gpu_synch.h6
-rw-r--r--src/video_core/gpu_thread.cpp136
-rw-r--r--src/video_core/gpu_thread.h132
-rw-r--r--src/video_core/memory_manager.cpp55
-rw-r--r--src/video_core/memory_manager.h17
-rw-r--r--src/video_core/morton.cpp324
-rw-r--r--src/video_core/morton.h6
-rw-r--r--src/video_core/rasterizer_cache.h70
-rw-r--r--src/video_core/rasterizer_interface.h8
-rw-r--r--src/video_core/renderer_opengl/gl_buffer_cache.cpp26
-rw-r--r--src/video_core/renderer_opengl/gl_buffer_cache.h31
-rw-r--r--src/video_core/renderer_opengl/gl_global_cache.cpp43
-rw-r--r--src/video_core/renderer_opengl/gl_global_cache.h21
-rw-r--r--src/video_core/renderer_opengl/gl_primitive_assembler.cpp7
-rw-r--r--src/video_core/renderer_opengl/gl_rasterizer.cpp75
-rw-r--r--src/video_core/renderer_opengl/gl_rasterizer.h9
-rw-r--r--src/video_core/renderer_opengl/gl_rasterizer_cache.cpp69
-rw-r--r--src/video_core/renderer_opengl/gl_rasterizer_cache.h19
-rw-r--r--src/video_core/renderer_opengl/gl_shader_cache.cpp62
-rw-r--r--src/video_core/renderer_opengl/gl_shader_cache.h19
-rw-r--r--src/video_core/renderer_opengl/renderer_opengl.cpp11
-rw-r--r--src/video_core/renderer_vulkan/vk_buffer_cache.cpp27
-rw-r--r--src/video_core/renderer_vulkan/vk_buffer_cache.h31
-rw-r--r--src/video_core/renderer_vulkan/vk_sampler_cache.cpp81
-rw-r--r--src/video_core/renderer_vulkan/vk_sampler_cache.h56
-rw-r--r--src/video_core/textures/decoders.cpp32
-rw-r--r--src/video_core/textures/decoders.h13
-rw-r--r--src/video_core/textures/texture.h55
-rw-r--r--src/yuzu/bootmanager.cpp1
-rw-r--r--src/yuzu/configuration/config.cpp448
-rw-r--r--src/yuzu/configuration/config.h5
-rw-r--r--src/yuzu/debugger/graphics/graphics_surface.cpp5
-rw-r--r--src/yuzu_cmd/emu_window/emu_window_sdl2.cpp13
93 files changed, 2501 insertions, 1946 deletions
diff --git a/CMakeLists.txt b/CMakeLists.txt
index 32cfa8580..a4914f37d 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -163,12 +163,6 @@ else()
set(CMAKE_EXE_LINKER_FLAGS_RELEASE "/DEBUG /MANIFEST:NO /INCREMENTAL:NO /OPT:REF,ICF" CACHE STRING "" FORCE)
endif()
-# Fix GCC C++17 and Boost.ICL incompatibility (needed to build dynarmic)
-# See https://bugzilla.redhat.com/show_bug.cgi?id=1485641#c1
-if (CMAKE_COMPILER_IS_GNUCC)
- set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -fno-new-ttp-matching")
-endif()
-
# Set file offset size to 64 bits.
#
# On modern Unixes, this is typically already the case. The lone exception is
@@ -185,9 +179,9 @@ set_property(DIRECTORY APPEND PROPERTY
# System imported libraries
# ======================
-find_package(Boost 1.63.0 QUIET)
+find_package(Boost 1.66.0 QUIET)
if (NOT Boost_FOUND)
- message(STATUS "Boost 1.63.0 or newer not found, falling back to externals")
+ message(STATUS "Boost 1.66.0 or newer not found, falling back to externals")
set(BOOST_ROOT "${PROJECT_SOURCE_DIR}/externals/boost")
set(Boost_NO_SYSTEM_PATHS OFF)
diff --git a/externals/opus b/externals/opus
-Subproject b2871922a12abb49579512d604cabc471a59ad9
+Subproject 562f8ba555c4181e1b57e82e496e4a959b9c019
diff --git a/src/common/CMakeLists.txt b/src/common/CMakeLists.txt
index 3d30f0e3e..43ae8a9e7 100644
--- a/src/common/CMakeLists.txt
+++ b/src/common/CMakeLists.txt
@@ -92,10 +92,14 @@ add_library(common STATIC
logging/text_formatter.cpp
logging/text_formatter.h
math_util.h
+ memory_hook.cpp
+ memory_hook.h
microprofile.cpp
microprofile.h
microprofileui.h
misc.cpp
+ page_table.cpp
+ page_table.h
param_package.cpp
param_package.h
quaternion.h
@@ -114,6 +118,8 @@ add_library(common STATIC
threadsafe_queue.h
timer.cpp
timer.h
+ uint128.cpp
+ uint128.h
vector_math.h
web_result.h
)
diff --git a/src/core/memory_hook.cpp b/src/common/memory_hook.cpp
index c61c6c1fb..3986986d6 100644
--- a/src/core/memory_hook.cpp
+++ b/src/common/memory_hook.cpp
@@ -2,10 +2,10 @@
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
-#include "core/memory_hook.h"
+#include "common/memory_hook.h"
-namespace Memory {
+namespace Common {
MemoryHook::~MemoryHook() = default;
-} // namespace Memory
+} // namespace Common
diff --git a/src/core/memory_hook.h b/src/common/memory_hook.h
index 940777107..adaa4c2c5 100644
--- a/src/core/memory_hook.h
+++ b/src/common/memory_hook.h
@@ -9,7 +9,7 @@
#include "common/common_types.h"
-namespace Memory {
+namespace Common {
/**
* Memory hooks have two purposes:
@@ -44,4 +44,4 @@ public:
};
using MemoryHookPointer = std::shared_ptr<MemoryHook>;
-} // namespace Memory
+} // namespace Common
diff --git a/src/common/page_table.cpp b/src/common/page_table.cpp
new file mode 100644
index 000000000..8eba1c3f1
--- /dev/null
+++ b/src/common/page_table.cpp
@@ -0,0 +1,29 @@
+// Copyright 2019 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "common/page_table.h"
+
+namespace Common {
+
+PageTable::PageTable(std::size_t page_size_in_bits) : page_size_in_bits{page_size_in_bits} {}
+
+PageTable::~PageTable() = default;
+
+void PageTable::Resize(std::size_t address_space_width_in_bits) {
+ const std::size_t num_page_table_entries = 1ULL
+ << (address_space_width_in_bits - page_size_in_bits);
+
+ pointers.resize(num_page_table_entries);
+ attributes.resize(num_page_table_entries);
+
+ // The default is a 39-bit address space, which causes an initial 1GB allocation size. If the
+ // vector size is subsequently decreased (via resize), the vector might not automatically
+ // actually reallocate/resize its underlying allocation, which wastes up to ~800 MB for
+ // 36-bit titles. Call shrink_to_fit to reduce capacity to what's actually in use.
+
+ pointers.shrink_to_fit();
+ attributes.shrink_to_fit();
+}
+
+} // namespace Common
diff --git a/src/common/page_table.h b/src/common/page_table.h
new file mode 100644
index 000000000..8339f2890
--- /dev/null
+++ b/src/common/page_table.h
@@ -0,0 +1,80 @@
+// Copyright 2019 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <vector>
+#include <boost/icl/interval_map.hpp>
+#include "common/common_types.h"
+#include "common/memory_hook.h"
+
+namespace Common {
+
+enum class PageType : u8 {
+ /// Page is unmapped and should cause an access error.
+ Unmapped,
+ /// Page is mapped to regular memory. This is the only type you can get pointers to.
+ Memory,
+ /// Page is mapped to regular memory, but also needs to check for rasterizer cache flushing and
+ /// invalidation
+ RasterizerCachedMemory,
+ /// Page is mapped to a I/O region. Writing and reading to this page is handled by functions.
+ Special,
+};
+
+struct SpecialRegion {
+ enum class Type {
+ DebugHook,
+ IODevice,
+ } type;
+
+ MemoryHookPointer handler;
+
+ bool operator<(const SpecialRegion& other) const {
+ return std::tie(type, handler) < std::tie(other.type, other.handler);
+ }
+
+ bool operator==(const SpecialRegion& other) const {
+ return std::tie(type, handler) == std::tie(other.type, other.handler);
+ }
+};
+
+/**
+ * A (reasonably) fast way of allowing switchable and remappable process address spaces. It loosely
+ * mimics the way a real CPU page table works.
+ */
+struct PageTable {
+ explicit PageTable(std::size_t page_size_in_bits);
+ ~PageTable();
+
+ /**
+ * Resizes the page table to be able to accomodate enough pages within
+ * a given address space.
+ *
+ * @param address_space_width_in_bits The address size width in bits.
+ */
+ void Resize(std::size_t address_space_width_in_bits);
+
+ /**
+ * Vector of memory pointers backing each page. An entry can only be non-null if the
+ * corresponding entry in the `attributes` vector is of type `Memory`.
+ */
+ std::vector<u8*> pointers;
+
+ /**
+ * Contains MMIO handlers that back memory regions whose entries in the `attribute` vector is
+ * of type `Special`.
+ */
+ boost::icl::interval_map<VAddr, std::set<SpecialRegion>> special_regions;
+
+ /**
+ * Vector of fine grained page attributes. If it is set to any value other than `Memory`, then
+ * the corresponding entry in `pointers` MUST be set to null.
+ */
+ std::vector<PageType> attributes;
+
+ const std::size_t page_size_in_bits{};
+};
+
+} // namespace Common
diff --git a/src/common/uint128.cpp b/src/common/uint128.cpp
new file mode 100644
index 000000000..2238a52c5
--- /dev/null
+++ b/src/common/uint128.cpp
@@ -0,0 +1,41 @@
+#ifdef _MSC_VER
+#include <intrin.h>
+
+#pragma intrinsic(_umul128)
+#endif
+#include <cstring>
+#include "common/uint128.h"
+
+namespace Common {
+
+u128 Multiply64Into128(u64 a, u64 b) {
+ u128 result;
+#ifdef _MSC_VER
+ result[0] = _umul128(a, b, &result[1]);
+#else
+ unsigned __int128 tmp = a;
+ tmp *= b;
+ std::memcpy(&result, &tmp, sizeof(u128));
+#endif
+ return result;
+}
+
+std::pair<u64, u64> Divide128On32(u128 dividend, u32 divisor) {
+ u64 remainder = dividend[0] % divisor;
+ u64 accum = dividend[0] / divisor;
+ if (dividend[1] == 0)
+ return {accum, remainder};
+ // We ignore dividend[1] / divisor as that overflows
+ const u64 first_segment = (dividend[1] % divisor) << 32;
+ accum += (first_segment / divisor) << 32;
+ const u64 second_segment = (first_segment % divisor) << 32;
+ accum += (second_segment / divisor);
+ remainder += second_segment % divisor;
+ if (remainder >= divisor) {
+ accum++;
+ remainder -= divisor;
+ }
+ return {accum, remainder};
+}
+
+} // namespace Common
diff --git a/src/common/uint128.h b/src/common/uint128.h
new file mode 100644
index 000000000..52e6b46eb
--- /dev/null
+++ b/src/common/uint128.h
@@ -0,0 +1,14 @@
+
+#include <utility>
+#include "common/common_types.h"
+
+namespace Common {
+
+// This function multiplies 2 u64 values and produces a u128 value;
+u128 Multiply64Into128(u64 a, u64 b);
+
+// This function divides a u128 by a u32 value and produces two u64 values:
+// the result of division and the remainder
+std::pair<u64, u64> Divide128On32(u128 dividend, u32 divisor);
+
+} // namespace Common
diff --git a/src/core/CMakeLists.txt b/src/core/CMakeLists.txt
index 8ccb2d5f0..aee8bc27d 100644
--- a/src/core/CMakeLists.txt
+++ b/src/core/CMakeLists.txt
@@ -437,8 +437,6 @@ add_library(core STATIC
loader/xci.h
memory.cpp
memory.h
- memory_hook.cpp
- memory_hook.h
memory_setup.h
perf_stats.cpp
perf_stats.h
diff --git a/src/core/arm/dynarmic/arm_dynarmic.cpp b/src/core/arm/dynarmic/arm_dynarmic.cpp
index 9b7ca4030..4fdc12f11 100644
--- a/src/core/arm/dynarmic/arm_dynarmic.cpp
+++ b/src/core/arm/dynarmic/arm_dynarmic.cpp
@@ -12,6 +12,7 @@
#include "core/core.h"
#include "core/core_cpu.h"
#include "core/core_timing.h"
+#include "core/core_timing_util.h"
#include "core/gdbstub/gdbstub.h"
#include "core/hle/kernel/process.h"
#include "core/hle/kernel/svc.h"
@@ -119,7 +120,7 @@ public:
return std::max(parent.core_timing.GetDowncount(), 0);
}
u64 GetCNTPCT() override {
- return parent.core_timing.GetTicks();
+ return Timing::CpuCyclesToClockCycles(parent.core_timing.GetTicks());
}
ARM_Dynarmic& parent;
@@ -151,7 +152,7 @@ std::unique_ptr<Dynarmic::A64::Jit> ARM_Dynarmic::MakeJit() const {
config.tpidr_el0 = &cb->tpidr_el0;
config.dczid_el0 = 4;
config.ctr_el0 = 0x8444c004;
- config.cntfrq_el0 = 19200000; // Value from fusee.
+ config.cntfrq_el0 = Timing::CNTFREQ;
// Unpredictable instructions
config.define_unpredictable_behaviour = true;
diff --git a/src/core/arm/dynarmic/arm_dynarmic.h b/src/core/arm/dynarmic/arm_dynarmic.h
index 6cc458296..aada1e862 100644
--- a/src/core/arm/dynarmic/arm_dynarmic.h
+++ b/src/core/arm/dynarmic/arm_dynarmic.h
@@ -12,7 +12,7 @@
#include "core/arm/exclusive_monitor.h"
#include "core/arm/unicorn/arm_unicorn.h"
-namespace Memory {
+namespace Common {
struct PageTable;
}
@@ -70,7 +70,7 @@ private:
Timing::CoreTiming& core_timing;
DynarmicExclusiveMonitor& exclusive_monitor;
- Memory::PageTable* current_page_table = nullptr;
+ Common::PageTable* current_page_table = nullptr;
};
class DynarmicExclusiveMonitor final : public ExclusiveMonitor {
diff --git a/src/core/core_timing_util.cpp b/src/core/core_timing_util.cpp
index 88ff70233..7942f30d6 100644
--- a/src/core/core_timing_util.cpp
+++ b/src/core/core_timing_util.cpp
@@ -7,6 +7,7 @@
#include <cinttypes>
#include <limits>
#include "common/logging/log.h"
+#include "common/uint128.h"
namespace Core::Timing {
@@ -60,4 +61,9 @@ s64 nsToCycles(u64 ns) {
return (BASE_CLOCK_RATE * static_cast<s64>(ns)) / 1000000000;
}
+u64 CpuCyclesToClockCycles(u64 ticks) {
+ const u128 temporal = Common::Multiply64Into128(ticks, CNTFREQ);
+ return Common::Divide128On32(temporal, static_cast<u32>(BASE_CLOCK_RATE)).first;
+}
+
} // namespace Core::Timing
diff --git a/src/core/core_timing_util.h b/src/core/core_timing_util.h
index 513cfac1b..679aa3123 100644
--- a/src/core/core_timing_util.h
+++ b/src/core/core_timing_util.h
@@ -11,6 +11,7 @@ namespace Core::Timing {
// The below clock rate is based on Switch's clockspeed being widely known as 1.020GHz
// The exact value used is of course unverified.
constexpr u64 BASE_CLOCK_RATE = 1019215872; // Switch clock speed is 1020MHz un/docked
+constexpr u64 CNTFREQ = 19200000; // Value from fusee.
inline s64 msToCycles(int ms) {
// since ms is int there is no way to overflow
@@ -61,4 +62,6 @@ inline u64 cyclesToMs(s64 cycles) {
return cycles * 1000 / BASE_CLOCK_RATE;
}
+u64 CpuCyclesToClockCycles(u64 ticks);
+
} // namespace Core::Timing
diff --git a/src/core/hle/ipc_helpers.h b/src/core/hle/ipc_helpers.h
index 079283830..68406eb63 100644
--- a/src/core/hle/ipc_helpers.h
+++ b/src/core/hle/ipc_helpers.h
@@ -19,9 +19,12 @@
#include "core/hle/kernel/hle_ipc.h"
#include "core/hle/kernel/object.h"
#include "core/hle/kernel/server_session.h"
+#include "core/hle/result.h"
namespace IPC {
+constexpr ResultCode ERR_REMOTE_PROCESS_DEAD{ErrorModule::HIPC, 301};
+
class RequestHelperBase {
protected:
Kernel::HLERequestContext* context = nullptr;
@@ -272,6 +275,20 @@ inline void ResponseBuilder::Push(u64 value) {
}
template <>
+inline void ResponseBuilder::Push(float value) {
+ u32 integral;
+ std::memcpy(&integral, &value, sizeof(u32));
+ Push(integral);
+}
+
+template <>
+inline void ResponseBuilder::Push(double value) {
+ u64 integral;
+ std::memcpy(&integral, &value, sizeof(u64));
+ Push(integral);
+}
+
+template <>
inline void ResponseBuilder::Push(bool value) {
Push(static_cast<u8>(value));
}
@@ -362,6 +379,11 @@ inline u32 RequestParser::Pop() {
return cmdbuf[index++];
}
+template <>
+inline s32 RequestParser::Pop() {
+ return static_cast<s32>(Pop<u32>());
+}
+
template <typename T>
void RequestParser::PopRaw(T& value) {
std::memcpy(&value, cmdbuf + index, sizeof(T));
@@ -393,11 +415,37 @@ inline u64 RequestParser::Pop() {
}
template <>
+inline s8 RequestParser::Pop() {
+ return static_cast<s8>(Pop<u8>());
+}
+
+template <>
+inline s16 RequestParser::Pop() {
+ return static_cast<s16>(Pop<u16>());
+}
+
+template <>
inline s64 RequestParser::Pop() {
return static_cast<s64>(Pop<u64>());
}
template <>
+inline float RequestParser::Pop() {
+ const u32 value = Pop<u32>();
+ float real;
+ std::memcpy(&real, &value, sizeof(real));
+ return real;
+}
+
+template <>
+inline double RequestParser::Pop() {
+ const u64 value = Pop<u64>();
+ float real;
+ std::memcpy(&real, &value, sizeof(real));
+ return real;
+}
+
+template <>
inline bool RequestParser::Pop() {
return Pop<u8>() != 0;
}
diff --git a/src/core/hle/kernel/client_port.cpp b/src/core/hle/kernel/client_port.cpp
index d4c91d529..aa432658e 100644
--- a/src/core/hle/kernel/client_port.cpp
+++ b/src/core/hle/kernel/client_port.cpp
@@ -33,10 +33,11 @@ ResultVal<SharedPtr<ClientSession>> ClientPort::Connect() {
// Create a new session pair, let the created sessions inherit the parent port's HLE handler.
auto sessions = ServerSession::CreateSessionPair(kernel, server_port->GetName(), this);
- if (server_port->hle_handler)
- server_port->hle_handler->ClientConnected(std::get<SharedPtr<ServerSession>>(sessions));
- else
- server_port->pending_sessions.push_back(std::get<SharedPtr<ServerSession>>(sessions));
+ if (server_port->HasHLEHandler()) {
+ server_port->GetHLEHandler()->ClientConnected(std::get<SharedPtr<ServerSession>>(sessions));
+ } else {
+ server_port->AppendPendingSession(std::get<SharedPtr<ServerSession>>(sessions));
+ }
// Wake the threads waiting on the ServerPort
server_port->WakeupAllWaitingThreads();
diff --git a/src/core/hle/kernel/process.cpp b/src/core/hle/kernel/process.cpp
index 7e8ba978c..65c51003d 100644
--- a/src/core/hle/kernel/process.cpp
+++ b/src/core/hle/kernel/process.cpp
@@ -31,7 +31,7 @@ namespace {
*/
void SetupMainThread(Process& owner_process, KernelCore& kernel, VAddr entry_point, u32 priority) {
// Setup page table so we can write to memory
- SetCurrentPageTable(&owner_process.VMManager().page_table);
+ Memory::SetCurrentPageTable(&owner_process.VMManager().page_table);
// Initialize new "main" thread
const VAddr stack_top = owner_process.VMManager().GetTLSIORegionEndAddress();
@@ -133,7 +133,7 @@ void Process::PrepareForTermination() {
if (thread->GetOwnerProcess() != this)
continue;
- if (thread == GetCurrentThread())
+ if (thread == system.CurrentScheduler().GetCurrentThread())
continue;
// TODO(Subv): When are the other running/ready threads terminated?
@@ -145,7 +145,6 @@ void Process::PrepareForTermination() {
}
};
- const auto& system = Core::System::GetInstance();
stop_threads(system.Scheduler(0).GetThreadList());
stop_threads(system.Scheduler(1).GetThreadList());
stop_threads(system.Scheduler(2).GetThreadList());
@@ -228,13 +227,11 @@ void Process::LoadModule(CodeSet module_, VAddr base_addr) {
MapSegment(module_.DataSegment(), VMAPermission::ReadWrite, MemoryState::CodeMutable);
// Clear instruction cache in CPU JIT
- Core::System::GetInstance().ArmInterface(0).ClearInstructionCache();
- Core::System::GetInstance().ArmInterface(1).ClearInstructionCache();
- Core::System::GetInstance().ArmInterface(2).ClearInstructionCache();
- Core::System::GetInstance().ArmInterface(3).ClearInstructionCache();
+ system.InvalidateCpuInstructionCaches();
}
-Process::Process(Core::System& system) : WaitObject{system.Kernel()}, address_arbiter{system} {}
+Process::Process(Core::System& system)
+ : WaitObject{system.Kernel()}, address_arbiter{system}, system{system} {}
Process::~Process() = default;
void Process::Acquire(Thread* thread) {
diff --git a/src/core/hle/kernel/process.h b/src/core/hle/kernel/process.h
index 2a132c894..47ffd4ad3 100644
--- a/src/core/hle/kernel/process.h
+++ b/src/core/hle/kernel/process.h
@@ -266,7 +266,7 @@ public:
void FreeTLSSlot(VAddr tls_address);
private:
- explicit Process(Core::System& kernel);
+ explicit Process(Core::System& system);
~Process() override;
/// Checks if the specified thread should wait until this process is available.
@@ -330,6 +330,10 @@ private:
/// Random values for svcGetInfo RandomEntropy
std::array<u64, RANDOM_ENTROPY_SIZE> random_entropy;
+ /// System context
+ Core::System& system;
+
+ /// Name of this process
std::string name;
};
diff --git a/src/core/hle/kernel/scheduler.cpp b/src/core/hle/kernel/scheduler.cpp
index 5fccfd9f4..cc189cc64 100644
--- a/src/core/hle/kernel/scheduler.cpp
+++ b/src/core/hle/kernel/scheduler.cpp
@@ -96,7 +96,7 @@ void Scheduler::SwitchContext(Thread* new_thread) {
auto* const thread_owner_process = current_thread->GetOwnerProcess();
if (previous_process != thread_owner_process) {
system.Kernel().MakeCurrentProcess(thread_owner_process);
- SetCurrentPageTable(&thread_owner_process->VMManager().page_table);
+ Memory::SetCurrentPageTable(&thread_owner_process->VMManager().page_table);
}
cpu_core.LoadContext(new_thread->GetContext());
@@ -199,8 +199,7 @@ void Scheduler::YieldWithoutLoadBalancing(Thread* thread) {
ASSERT(thread->GetPriority() < THREADPRIO_COUNT);
// Yield this thread -- sleep for zero time and force reschedule to different thread
- WaitCurrentThread_Sleep();
- GetCurrentThread()->WakeAfterDelay(0);
+ GetCurrentThread()->Sleep(0);
}
void Scheduler::YieldWithLoadBalancing(Thread* thread) {
@@ -215,8 +214,7 @@ void Scheduler::YieldWithLoadBalancing(Thread* thread) {
ASSERT(priority < THREADPRIO_COUNT);
// Sleep for zero time to be able to force reschedule to different thread
- WaitCurrentThread_Sleep();
- GetCurrentThread()->WakeAfterDelay(0);
+ GetCurrentThread()->Sleep(0);
Thread* suggested_thread = nullptr;
diff --git a/src/core/hle/kernel/server_port.cpp b/src/core/hle/kernel/server_port.cpp
index d6ceeb2da..0e1515c89 100644
--- a/src/core/hle/kernel/server_port.cpp
+++ b/src/core/hle/kernel/server_port.cpp
@@ -26,6 +26,10 @@ ResultVal<SharedPtr<ServerSession>> ServerPort::Accept() {
return MakeResult(std::move(session));
}
+void ServerPort::AppendPendingSession(SharedPtr<ServerSession> pending_session) {
+ pending_sessions.push_back(std::move(pending_session));
+}
+
bool ServerPort::ShouldWait(Thread* thread) const {
// If there are no pending sessions, we wait until a new one is added.
return pending_sessions.empty();
diff --git a/src/core/hle/kernel/server_port.h b/src/core/hle/kernel/server_port.h
index e52f8245f..9bc667cf2 100644
--- a/src/core/hle/kernel/server_port.h
+++ b/src/core/hle/kernel/server_port.h
@@ -22,6 +22,8 @@ class SessionRequestHandler;
class ServerPort final : public WaitObject {
public:
+ using HLEHandler = std::shared_ptr<SessionRequestHandler>;
+
/**
* Creates a pair of ServerPort and an associated ClientPort.
*
@@ -51,22 +53,27 @@ public:
*/
ResultVal<SharedPtr<ServerSession>> Accept();
+ /// Whether or not this server port has an HLE handler available.
+ bool HasHLEHandler() const {
+ return hle_handler != nullptr;
+ }
+
+ /// Gets the HLE handler for this port.
+ HLEHandler GetHLEHandler() const {
+ return hle_handler;
+ }
+
/**
* Sets the HLE handler template for the port. ServerSessions crated by connecting to this port
* will inherit a reference to this handler.
*/
- void SetHleHandler(std::shared_ptr<SessionRequestHandler> hle_handler_) {
+ void SetHleHandler(HLEHandler hle_handler_) {
hle_handler = std::move(hle_handler_);
}
- std::string name; ///< Name of port (optional)
-
- /// ServerSessions waiting to be accepted by the port
- std::vector<SharedPtr<ServerSession>> pending_sessions;
-
- /// This session's HLE request handler template (optional)
- /// ServerSessions created from this port inherit a reference to this handler.
- std::shared_ptr<SessionRequestHandler> hle_handler;
+ /// Appends a ServerSession to the collection of ServerSessions
+ /// waiting to be accepted by this port.
+ void AppendPendingSession(SharedPtr<ServerSession> pending_session);
bool ShouldWait(Thread* thread) const override;
void Acquire(Thread* thread) override;
@@ -74,6 +81,16 @@ public:
private:
explicit ServerPort(KernelCore& kernel);
~ServerPort() override;
+
+ /// ServerSessions waiting to be accepted by the port
+ std::vector<SharedPtr<ServerSession>> pending_sessions;
+
+ /// This session's HLE request handler template (optional)
+ /// ServerSessions created from this port inherit a reference to this handler.
+ HLEHandler hle_handler;
+
+ /// Name of the port (optional)
+ std::string name;
};
} // namespace Kernel
diff --git a/src/core/hle/kernel/svc.cpp b/src/core/hle/kernel/svc.cpp
index 77d0e3d96..047fa0c19 100644
--- a/src/core/hle/kernel/svc.cpp
+++ b/src/core/hle/kernel/svc.cpp
@@ -1284,10 +1284,14 @@ static ResultCode StartThread(Handle thread_handle) {
/// Called when a thread exits
static void ExitThread() {
- LOG_TRACE(Kernel_SVC, "called, pc=0x{:08X}", Core::CurrentArmInterface().GetPC());
+ auto& system = Core::System::GetInstance();
- ExitCurrentThread();
- Core::System::GetInstance().PrepareReschedule();
+ LOG_TRACE(Kernel_SVC, "called, pc=0x{:08X}", system.CurrentArmInterface().GetPC());
+
+ auto* const current_thread = system.CurrentScheduler().GetCurrentThread();
+ current_thread->Stop();
+ system.CurrentScheduler().RemoveThread(current_thread);
+ system.PrepareReschedule();
}
/// Sleep the current thread
@@ -1300,32 +1304,32 @@ static void SleepThread(s64 nanoseconds) {
YieldAndWaitForLoadBalancing = -2,
};
+ auto& system = Core::System::GetInstance();
+ auto& scheduler = system.CurrentScheduler();
+ auto* const current_thread = scheduler.GetCurrentThread();
+
if (nanoseconds <= 0) {
- auto& scheduler{Core::System::GetInstance().CurrentScheduler()};
switch (static_cast<SleepType>(nanoseconds)) {
case SleepType::YieldWithoutLoadBalancing:
- scheduler.YieldWithoutLoadBalancing(GetCurrentThread());
+ scheduler.YieldWithoutLoadBalancing(current_thread);
break;
case SleepType::YieldWithLoadBalancing:
- scheduler.YieldWithLoadBalancing(GetCurrentThread());
+ scheduler.YieldWithLoadBalancing(current_thread);
break;
case SleepType::YieldAndWaitForLoadBalancing:
- scheduler.YieldAndWaitForLoadBalancing(GetCurrentThread());
+ scheduler.YieldAndWaitForLoadBalancing(current_thread);
break;
default:
UNREACHABLE_MSG("Unimplemented sleep yield type '{:016X}'!", nanoseconds);
}
} else {
- // Sleep current thread and check for next thread to schedule
- WaitCurrentThread_Sleep();
-
- // Create an event to wake the thread up after the specified nanosecond delay has passed
- GetCurrentThread()->WakeAfterDelay(nanoseconds);
+ current_thread->Sleep(nanoseconds);
}
// Reschedule all CPU cores
- for (std::size_t i = 0; i < Core::NUM_CPU_CORES; ++i)
- Core::System::GetInstance().CpuCore(i).PrepareReschedule();
+ for (std::size_t i = 0; i < Core::NUM_CPU_CORES; ++i) {
+ system.CpuCore(i).PrepareReschedule();
+ }
}
/// Wait process wide key atomic
diff --git a/src/core/hle/kernel/thread.cpp b/src/core/hle/kernel/thread.cpp
index 202997d20..d9ffebc3f 100644
--- a/src/core/hle/kernel/thread.cpp
+++ b/src/core/hle/kernel/thread.cpp
@@ -66,17 +66,6 @@ void Thread::Stop() {
owner_process->FreeTLSSlot(tls_address);
}
-void WaitCurrentThread_Sleep() {
- Thread* thread = GetCurrentThread();
- thread->SetStatus(ThreadStatus::WaitSleep);
-}
-
-void ExitCurrentThread() {
- Thread* thread = GetCurrentThread();
- thread->Stop();
- Core::System::GetInstance().CurrentScheduler().RemoveThread(thread);
-}
-
void Thread::WakeAfterDelay(s64 nanoseconds) {
// Don't schedule a wakeup if the thread wants to wait forever
if (nanoseconds == -1)
@@ -405,6 +394,14 @@ void Thread::SetActivity(ThreadActivity value) {
}
}
+void Thread::Sleep(s64 nanoseconds) {
+ // Sleep current thread and check for next thread to schedule
+ SetStatus(ThreadStatus::WaitSleep);
+
+ // Create an event to wake the thread up after the specified nanosecond delay has passed
+ WakeAfterDelay(nanoseconds);
+}
+
////////////////////////////////////////////////////////////////////////////////////////////////////
/**
diff --git a/src/core/hle/kernel/thread.h b/src/core/hle/kernel/thread.h
index 96d9982d8..faad5f391 100644
--- a/src/core/hle/kernel/thread.h
+++ b/src/core/hle/kernel/thread.h
@@ -383,6 +383,9 @@ public:
void SetActivity(ThreadActivity value);
+ /// Sleeps this thread for the given amount of nanoseconds.
+ void Sleep(s64 nanoseconds);
+
private:
explicit Thread(KernelCore& kernel);
~Thread() override;
@@ -466,14 +469,4 @@ private:
*/
Thread* GetCurrentThread();
-/**
- * Waits the current thread on a sleep
- */
-void WaitCurrentThread_Sleep();
-
-/**
- * Stops the current thread and removes it from the thread_list
- */
-void ExitCurrentThread();
-
} // namespace Kernel
diff --git a/src/core/hle/kernel/vm_manager.cpp b/src/core/hle/kernel/vm_manager.cpp
index 05c59af34..3def3e52c 100644
--- a/src/core/hle/kernel/vm_manager.cpp
+++ b/src/core/hle/kernel/vm_manager.cpp
@@ -7,13 +7,13 @@
#include <utility>
#include "common/assert.h"
#include "common/logging/log.h"
+#include "common/memory_hook.h"
#include "core/arm/arm_interface.h"
#include "core/core.h"
#include "core/file_sys/program_metadata.h"
#include "core/hle/kernel/errors.h"
#include "core/hle/kernel/vm_manager.h"
#include "core/memory.h"
-#include "core/memory_hook.h"
#include "core/memory_setup.h"
namespace Kernel {
@@ -177,7 +177,7 @@ ResultVal<VAddr> VMManager::FindFreeRegion(u64 size) const {
ResultVal<VMManager::VMAHandle> VMManager::MapMMIO(VAddr target, PAddr paddr, u64 size,
MemoryState state,
- Memory::MemoryHookPointer mmio_handler) {
+ Common::MemoryHookPointer mmio_handler) {
// This is the appropriately sized VMA that will turn into our allocation.
CASCADE_RESULT(VMAIter vma_handle, CarveVMA(target, size));
VirtualMemoryArea& final_vma = vma_handle->second;
@@ -624,7 +624,7 @@ void VMManager::ClearPageTable() {
std::fill(page_table.pointers.begin(), page_table.pointers.end(), nullptr);
page_table.special_regions.clear();
std::fill(page_table.attributes.begin(), page_table.attributes.end(),
- Memory::PageType::Unmapped);
+ Common::PageType::Unmapped);
}
VMManager::CheckResults VMManager::CheckRangeState(VAddr address, u64 size, MemoryState state_mask,
diff --git a/src/core/hle/kernel/vm_manager.h b/src/core/hle/kernel/vm_manager.h
index 88e0b3c02..b96980f8f 100644
--- a/src/core/hle/kernel/vm_manager.h
+++ b/src/core/hle/kernel/vm_manager.h
@@ -9,9 +9,10 @@
#include <tuple>
#include <vector>
#include "common/common_types.h"
+#include "common/memory_hook.h"
+#include "common/page_table.h"
#include "core/hle/result.h"
#include "core/memory.h"
-#include "core/memory_hook.h"
namespace FileSys {
enum class ProgramAddressSpaceType : u8;
@@ -290,7 +291,7 @@ struct VirtualMemoryArea {
// Settings for type = MMIO
/// Physical address of the register area this VMA maps to.
PAddr paddr = 0;
- Memory::MemoryHookPointer mmio_handler = nullptr;
+ Common::MemoryHookPointer mmio_handler = nullptr;
/// Tests if this area can be merged to the right with `next`.
bool CanBeMergedWith(const VirtualMemoryArea& next) const;
@@ -368,7 +369,7 @@ public:
* @param mmio_handler The handler that will implement read and write for this MMIO region.
*/
ResultVal<VMAHandle> MapMMIO(VAddr target, PAddr paddr, u64 size, MemoryState state,
- Memory::MemoryHookPointer mmio_handler);
+ Common::MemoryHookPointer mmio_handler);
/// Unmaps a range of addresses, splitting VMAs as necessary.
ResultCode UnmapRange(VAddr target, u64 size);
@@ -509,7 +510,7 @@ public:
/// Each VMManager has its own page table, which is set as the main one when the owning process
/// is scheduled.
- Memory::PageTable page_table;
+ Common::PageTable page_table{Memory::PAGE_BITS};
private:
using VMAIter = VMAMap::iterator;
diff --git a/src/core/hle/result.h b/src/core/hle/result.h
index 1ed144481..ab84f5ddc 100644
--- a/src/core/hle/result.h
+++ b/src/core/hle/result.h
@@ -13,14 +13,6 @@
// All the constants in this file come from http://switchbrew.org/index.php?title=Error_codes
/**
- * Detailed description of the error. Code 0 always means success.
- */
-enum class ErrorDescription : u32 {
- Success = 0,
- RemoteProcessDead = 301,
-};
-
-/**
* Identifies the module which caused the error. Error codes can be propagated through a call
* chain, meaning that this doesn't always correspond to the module where the API call made is
* contained.
@@ -120,7 +112,7 @@ enum class ErrorModule : u32 {
ShopN = 811,
};
-/// Encapsulates a CTR-OS error code, allowing it to be separated into its constituent fields.
+/// Encapsulates a Horizon OS error code, allowing it to be separated into its constituent fields.
union ResultCode {
u32 raw;
@@ -133,17 +125,9 @@ union ResultCode {
constexpr explicit ResultCode(u32 raw) : raw(raw) {}
- constexpr ResultCode(ErrorModule module, ErrorDescription description)
- : ResultCode(module, static_cast<u32>(description)) {}
-
constexpr ResultCode(ErrorModule module_, u32 description_)
: raw(module.FormatValue(module_) | description.FormatValue(description_)) {}
- constexpr ResultCode& operator=(const ResultCode& o) {
- raw = o.raw;
- return *this;
- }
-
constexpr bool IsSuccess() const {
return raw == 0;
}
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 a34b9e753..b031ebc66 100644
--- a/src/core/hle/service/nvdrv/devices/nvhost_as_gpu.cpp
+++ b/src/core/hle/service/nvdrv/devices/nvhost_as_gpu.cpp
@@ -10,6 +10,7 @@
#include "core/core.h"
#include "core/hle/service/nvdrv/devices/nvhost_as_gpu.h"
#include "core/hle/service/nvdrv/devices/nvmap.h"
+#include "core/memory.h"
#include "video_core/memory_manager.h"
#include "video_core/rasterizer_interface.h"
#include "video_core/renderer_base.h"
@@ -178,7 +179,7 @@ u32 nvhost_as_gpu::UnmapBuffer(const std::vector<u8>& input, std::vector<u8>& ou
auto& gpu = system_instance.GPU();
auto cpu_addr = gpu.MemoryManager().GpuToCpuAddress(params.offset);
ASSERT(cpu_addr);
- gpu.FlushAndInvalidateRegion(*cpu_addr, itr->second.size);
+ gpu.FlushAndInvalidateRegion(ToCacheAddr(Memory::GetPointer(*cpu_addr)), itr->second.size);
params.offset = gpu.MemoryManager().UnmapBuffer(params.offset, itr->second.size);
diff --git a/src/core/hle/service/service.cpp b/src/core/hle/service/service.cpp
index 576fd6407..00806b0ed 100644
--- a/src/core/hle/service/service.cpp
+++ b/src/core/hle/service/service.cpp
@@ -11,7 +11,6 @@
#include "core/hle/ipc.h"
#include "core/hle/ipc_helpers.h"
#include "core/hle/kernel/client_port.h"
-#include "core/hle/kernel/handle_table.h"
#include "core/hle/kernel/kernel.h"
#include "core/hle/kernel/process.h"
#include "core/hle/kernel/server_port.h"
@@ -168,7 +167,7 @@ ResultCode ServiceFrameworkBase::HandleSyncRequest(Kernel::HLERequestContext& co
case IPC::CommandType::Close: {
IPC::ResponseBuilder rb{context, 2};
rb.Push(RESULT_SUCCESS);
- return ResultCode(ErrorModule::HIPC, ErrorDescription::RemoteProcessDead);
+ return IPC::ERR_REMOTE_PROCESS_DEAD;
}
case IPC::CommandType::ControlWithContext:
case IPC::CommandType::Control: {
diff --git a/src/core/hle/service/sm/sm.h b/src/core/hle/service/sm/sm.h
index bef25433e..b9d6381b4 100644
--- a/src/core/hle/service/sm/sm.h
+++ b/src/core/hle/service/sm/sm.h
@@ -67,7 +67,7 @@ public:
if (port == nullptr) {
return nullptr;
}
- return std::static_pointer_cast<T>(port->hle_handler);
+ return std::static_pointer_cast<T>(port->GetHLEHandler());
}
void InvokeControlRequest(Kernel::HLERequestContext& context);
diff --git a/src/core/hle/service/vi/vi.cpp b/src/core/hle/service/vi/vi.cpp
index a975767bb..566cd6006 100644
--- a/src/core/hle/service/vi/vi.cpp
+++ b/src/core/hle/service/vi/vi.cpp
@@ -24,6 +24,7 @@
#include "core/hle/service/nvdrv/nvdrv.h"
#include "core/hle/service/nvflinger/buffer_queue.h"
#include "core/hle/service/nvflinger/nvflinger.h"
+#include "core/hle/service/service.h"
#include "core/hle/service/vi/vi.h"
#include "core/hle/service/vi/vi_m.h"
#include "core/hle/service/vi/vi_s.h"
@@ -33,6 +34,7 @@
namespace Service::VI {
constexpr ResultCode ERR_OPERATION_FAILED{ErrorModule::VI, 1};
+constexpr ResultCode ERR_PERMISSION_DENIED{ErrorModule::VI, 5};
constexpr ResultCode ERR_UNSUPPORTED{ErrorModule::VI, 6};
constexpr ResultCode ERR_NOT_FOUND{ErrorModule::VI, 7};
@@ -1203,26 +1205,40 @@ IApplicationDisplayService::IApplicationDisplayService(
RegisterHandlers(functions);
}
-Module::Interface::Interface(std::shared_ptr<Module> module, const char* name,
- std::shared_ptr<NVFlinger::NVFlinger> nv_flinger)
- : ServiceFramework(name), module(std::move(module)), nv_flinger(std::move(nv_flinger)) {}
+static bool IsValidServiceAccess(Permission permission, Policy policy) {
+ if (permission == Permission::User) {
+ return policy == Policy::User;
+ }
+
+ if (permission == Permission::System || permission == Permission::Manager) {
+ return policy == Policy::User || policy == Policy::Compositor;
+ }
-Module::Interface::~Interface() = default;
+ return false;
+}
-void Module::Interface::GetDisplayService(Kernel::HLERequestContext& ctx) {
- LOG_WARNING(Service_VI, "(STUBBED) called");
+void detail::GetDisplayServiceImpl(Kernel::HLERequestContext& ctx,
+ std::shared_ptr<NVFlinger::NVFlinger> nv_flinger,
+ Permission permission) {
+ IPC::RequestParser rp{ctx};
+ const auto policy = rp.PopEnum<Policy>();
+
+ if (!IsValidServiceAccess(permission, policy)) {
+ IPC::ResponseBuilder rb{ctx, 2};
+ rb.Push(ERR_PERMISSION_DENIED);
+ return;
+ }
IPC::ResponseBuilder rb{ctx, 2, 0, 1};
rb.Push(RESULT_SUCCESS);
- rb.PushIpcInterface<IApplicationDisplayService>(nv_flinger);
+ rb.PushIpcInterface<IApplicationDisplayService>(std::move(nv_flinger));
}
void InstallInterfaces(SM::ServiceManager& service_manager,
std::shared_ptr<NVFlinger::NVFlinger> nv_flinger) {
- auto module = std::make_shared<Module>();
- std::make_shared<VI_M>(module, nv_flinger)->InstallAsService(service_manager);
- std::make_shared<VI_S>(module, nv_flinger)->InstallAsService(service_manager);
- std::make_shared<VI_U>(module, nv_flinger)->InstallAsService(service_manager);
+ std::make_shared<VI_M>(nv_flinger)->InstallAsService(service_manager);
+ std::make_shared<VI_S>(nv_flinger)->InstallAsService(service_manager);
+ std::make_shared<VI_U>(nv_flinger)->InstallAsService(service_manager);
}
} // namespace Service::VI
diff --git a/src/core/hle/service/vi/vi.h b/src/core/hle/service/vi/vi.h
index e3963502a..6b66f8b81 100644
--- a/src/core/hle/service/vi/vi.h
+++ b/src/core/hle/service/vi/vi.h
@@ -4,12 +4,21 @@
#pragma once
-#include "core/hle/service/service.h"
+#include <memory>
+#include "common/common_types.h"
+
+namespace Kernel {
+class HLERequestContext;
+}
namespace Service::NVFlinger {
class NVFlinger;
}
+namespace Service::SM {
+class ServiceManager;
+}
+
namespace Service::VI {
enum class DisplayResolution : u32 {
@@ -19,22 +28,25 @@ enum class DisplayResolution : u32 {
UndockedHeight = 720,
};
-class Module final {
-public:
- class Interface : public ServiceFramework<Interface> {
- public:
- explicit Interface(std::shared_ptr<Module> module, const char* name,
- std::shared_ptr<NVFlinger::NVFlinger> nv_flinger);
- ~Interface() override;
-
- void GetDisplayService(Kernel::HLERequestContext& ctx);
+/// Permission level for a particular VI service instance
+enum class Permission {
+ User,
+ System,
+ Manager,
+};
- protected:
- std::shared_ptr<Module> module;
- std::shared_ptr<NVFlinger::NVFlinger> nv_flinger;
- };
+/// A policy type that may be requested via GetDisplayService and
+/// GetDisplayServiceWithProxyNameExchange
+enum class Policy {
+ User,
+ Compositor,
};
+namespace detail {
+void GetDisplayServiceImpl(Kernel::HLERequestContext& ctx,
+ std::shared_ptr<NVFlinger::NVFlinger> nv_flinger, Permission permission);
+} // namespace detail
+
/// Registers all VI services with the specified service manager.
void InstallInterfaces(SM::ServiceManager& service_manager,
std::shared_ptr<NVFlinger::NVFlinger> nv_flinger);
diff --git a/src/core/hle/service/vi/vi_m.cpp b/src/core/hle/service/vi/vi_m.cpp
index 207c06b16..06070087f 100644
--- a/src/core/hle/service/vi/vi_m.cpp
+++ b/src/core/hle/service/vi/vi_m.cpp
@@ -2,12 +2,14 @@
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
+#include "common/logging/log.h"
+#include "core/hle/service/vi/vi.h"
#include "core/hle/service/vi/vi_m.h"
namespace Service::VI {
-VI_M::VI_M(std::shared_ptr<Module> module, std::shared_ptr<NVFlinger::NVFlinger> nv_flinger)
- : Module::Interface(std::move(module), "vi:m", std::move(nv_flinger)) {
+VI_M::VI_M(std::shared_ptr<NVFlinger::NVFlinger> nv_flinger)
+ : ServiceFramework{"vi:m"}, nv_flinger{std::move(nv_flinger)} {
static const FunctionInfo functions[] = {
{2, &VI_M::GetDisplayService, "GetDisplayService"},
{3, nullptr, "GetDisplayServiceWithProxyNameExchange"},
@@ -17,4 +19,10 @@ VI_M::VI_M(std::shared_ptr<Module> module, std::shared_ptr<NVFlinger::NVFlinger>
VI_M::~VI_M() = default;
+void VI_M::GetDisplayService(Kernel::HLERequestContext& ctx) {
+ LOG_DEBUG(Service_VI, "called");
+
+ detail::GetDisplayServiceImpl(ctx, nv_flinger, Permission::Manager);
+}
+
} // namespace Service::VI
diff --git a/src/core/hle/service/vi/vi_m.h b/src/core/hle/service/vi/vi_m.h
index 487d58d50..290e06689 100644
--- a/src/core/hle/service/vi/vi_m.h
+++ b/src/core/hle/service/vi/vi_m.h
@@ -4,14 +4,27 @@
#pragma once
-#include "core/hle/service/vi/vi.h"
+#include "core/hle/service/service.h"
+
+namespace Kernel {
+class HLERequestContext;
+}
+
+namespace Service::NVFlinger {
+class NVFlinger;
+}
namespace Service::VI {
-class VI_M final : public Module::Interface {
+class VI_M final : public ServiceFramework<VI_M> {
public:
- explicit VI_M(std::shared_ptr<Module> module, std::shared_ptr<NVFlinger::NVFlinger> nv_flinger);
+ explicit VI_M(std::shared_ptr<NVFlinger::NVFlinger> nv_flinger);
~VI_M() override;
+
+private:
+ void GetDisplayService(Kernel::HLERequestContext& ctx);
+
+ std::shared_ptr<NVFlinger::NVFlinger> nv_flinger;
};
} // namespace Service::VI
diff --git a/src/core/hle/service/vi/vi_s.cpp b/src/core/hle/service/vi/vi_s.cpp
index 920e6a1f6..57c596cc4 100644
--- a/src/core/hle/service/vi/vi_s.cpp
+++ b/src/core/hle/service/vi/vi_s.cpp
@@ -2,12 +2,14 @@
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
+#include "common/logging/log.h"
+#include "core/hle/service/vi/vi.h"
#include "core/hle/service/vi/vi_s.h"
namespace Service::VI {
-VI_S::VI_S(std::shared_ptr<Module> module, std::shared_ptr<NVFlinger::NVFlinger> nv_flinger)
- : Module::Interface(std::move(module), "vi:s", std::move(nv_flinger)) {
+VI_S::VI_S(std::shared_ptr<NVFlinger::NVFlinger> nv_flinger)
+ : ServiceFramework{"vi:s"}, nv_flinger{std::move(nv_flinger)} {
static const FunctionInfo functions[] = {
{1, &VI_S::GetDisplayService, "GetDisplayService"},
{3, nullptr, "GetDisplayServiceWithProxyNameExchange"},
@@ -17,4 +19,10 @@ VI_S::VI_S(std::shared_ptr<Module> module, std::shared_ptr<NVFlinger::NVFlinger>
VI_S::~VI_S() = default;
+void VI_S::GetDisplayService(Kernel::HLERequestContext& ctx) {
+ LOG_DEBUG(Service_VI, "called");
+
+ detail::GetDisplayServiceImpl(ctx, nv_flinger, Permission::System);
+}
+
} // namespace Service::VI
diff --git a/src/core/hle/service/vi/vi_s.h b/src/core/hle/service/vi/vi_s.h
index bbc31148f..47804dc0b 100644
--- a/src/core/hle/service/vi/vi_s.h
+++ b/src/core/hle/service/vi/vi_s.h
@@ -4,14 +4,27 @@
#pragma once
-#include "core/hle/service/vi/vi.h"
+#include "core/hle/service/service.h"
+
+namespace Kernel {
+class HLERequestContext;
+}
+
+namespace Service::NVFlinger {
+class NVFlinger;
+}
namespace Service::VI {
-class VI_S final : public Module::Interface {
+class VI_S final : public ServiceFramework<VI_S> {
public:
- explicit VI_S(std::shared_ptr<Module> module, std::shared_ptr<NVFlinger::NVFlinger> nv_flinger);
+ explicit VI_S(std::shared_ptr<NVFlinger::NVFlinger> nv_flinger);
~VI_S() override;
+
+private:
+ void GetDisplayService(Kernel::HLERequestContext& ctx);
+
+ std::shared_ptr<NVFlinger::NVFlinger> nv_flinger;
};
} // namespace Service::VI
diff --git a/src/core/hle/service/vi/vi_u.cpp b/src/core/hle/service/vi/vi_u.cpp
index d81e410d6..9d5ceb608 100644
--- a/src/core/hle/service/vi/vi_u.cpp
+++ b/src/core/hle/service/vi/vi_u.cpp
@@ -2,12 +2,14 @@
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
+#include "common/logging/log.h"
+#include "core/hle/service/vi/vi.h"
#include "core/hle/service/vi/vi_u.h"
namespace Service::VI {
-VI_U::VI_U(std::shared_ptr<Module> module, std::shared_ptr<NVFlinger::NVFlinger> nv_flinger)
- : Module::Interface(std::move(module), "vi:u", std::move(nv_flinger)) {
+VI_U::VI_U(std::shared_ptr<NVFlinger::NVFlinger> nv_flinger)
+ : ServiceFramework{"vi:u"}, nv_flinger{std::move(nv_flinger)} {
static const FunctionInfo functions[] = {
{0, &VI_U::GetDisplayService, "GetDisplayService"},
};
@@ -16,4 +18,10 @@ VI_U::VI_U(std::shared_ptr<Module> module, std::shared_ptr<NVFlinger::NVFlinger>
VI_U::~VI_U() = default;
+void VI_U::GetDisplayService(Kernel::HLERequestContext& ctx) {
+ LOG_DEBUG(Service_VI, "called");
+
+ detail::GetDisplayServiceImpl(ctx, nv_flinger, Permission::User);
+}
+
} // namespace Service::VI
diff --git a/src/core/hle/service/vi/vi_u.h b/src/core/hle/service/vi/vi_u.h
index b92f28c92..19bdb73b0 100644
--- a/src/core/hle/service/vi/vi_u.h
+++ b/src/core/hle/service/vi/vi_u.h
@@ -4,14 +4,27 @@
#pragma once
-#include "core/hle/service/vi/vi.h"
+#include "core/hle/service/service.h"
+
+namespace Kernel {
+class HLERequestContext;
+}
+
+namespace Service::NVFlinger {
+class NVFlinger;
+}
namespace Service::VI {
-class VI_U final : public Module::Interface {
+class VI_U final : public ServiceFramework<VI_U> {
public:
- explicit VI_U(std::shared_ptr<Module> module, std::shared_ptr<NVFlinger::NVFlinger> nv_flinger);
+ explicit VI_U(std::shared_ptr<NVFlinger::NVFlinger> nv_flinger);
~VI_U() override;
+
+private:
+ void GetDisplayService(Kernel::HLERequestContext& ctx);
+
+ std::shared_ptr<NVFlinger::NVFlinger> nv_flinger;
};
} // namespace Service::VI
diff --git a/src/core/memory.cpp b/src/core/memory.cpp
index 6591c45d2..365ac82b4 100644
--- a/src/core/memory.cpp
+++ b/src/core/memory.cpp
@@ -10,6 +10,7 @@
#include "common/assert.h"
#include "common/common_types.h"
#include "common/logging/log.h"
+#include "common/page_table.h"
#include "common/swap.h"
#include "core/arm/arm_interface.h"
#include "core/core.h"
@@ -18,13 +19,14 @@
#include "core/hle/lock.h"
#include "core/memory.h"
#include "core/memory_setup.h"
+#include "video_core/gpu.h"
#include "video_core/renderer_base.h"
namespace Memory {
-static PageTable* current_page_table = nullptr;
+static Common::PageTable* current_page_table = nullptr;
-void SetCurrentPageTable(PageTable* page_table) {
+void SetCurrentPageTable(Common::PageTable* page_table) {
current_page_table = page_table;
auto& system = Core::System::GetInstance();
@@ -36,39 +38,20 @@ void SetCurrentPageTable(PageTable* page_table) {
}
}
-PageTable* GetCurrentPageTable() {
+Common::PageTable* GetCurrentPageTable() {
return current_page_table;
}
-PageTable::PageTable() = default;
-
-PageTable::PageTable(std::size_t address_space_width_in_bits) {
- Resize(address_space_width_in_bits);
-}
-
-PageTable::~PageTable() = default;
-
-void PageTable::Resize(std::size_t address_space_width_in_bits) {
- const std::size_t num_page_table_entries = 1ULL << (address_space_width_in_bits - PAGE_BITS);
-
- pointers.resize(num_page_table_entries);
- attributes.resize(num_page_table_entries);
-
- // The default is a 39-bit address space, which causes an initial 1GB allocation size. If the
- // vector size is subsequently decreased (via resize), the vector might not automatically
- // actually reallocate/resize its underlying allocation, which wastes up to ~800 MB for
- // 36-bit titles. Call shrink_to_fit to reduce capacity to what's actually in use.
-
- pointers.shrink_to_fit();
- attributes.shrink_to_fit();
-}
-
-static void MapPages(PageTable& page_table, VAddr base, u64 size, u8* memory, PageType type) {
+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);
- RasterizerFlushVirtualRegion(base << PAGE_BITS, size * PAGE_SIZE,
- FlushMode::FlushAndInvalidate);
+ // During boot, current_page_table might not be set yet, in which case we need not flush
+ if (current_page_table) {
+ Core::System::GetInstance().GPU().FlushAndInvalidateRegion(base << PAGE_BITS,
+ size * PAGE_SIZE);
+ }
VAddr end = base + size;
ASSERT_MSG(end <= page_table.pointers.size(), "out of range mapping at {:016X}",
@@ -88,41 +71,47 @@ static void MapPages(PageTable& page_table, VAddr base, u64 size, u8* memory, Pa
}
}
-void MapMemoryRegion(PageTable& page_table, VAddr base, u64 size, u8* target) {
+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, PageType::Memory);
+ MapPages(page_table, base / PAGE_SIZE, size / PAGE_SIZE, target, Common::PageType::Memory);
}
-void MapIoRegion(PageTable& page_table, VAddr base, u64 size, MemoryHookPointer mmio_handler) {
+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, PageType::Special);
+ MapPages(page_table, base / PAGE_SIZE, size / PAGE_SIZE, nullptr, Common::PageType::Special);
auto interval = boost::icl::discrete_interval<VAddr>::closed(base, base + size - 1);
- SpecialRegion region{SpecialRegion::Type::IODevice, std::move(mmio_handler)};
- page_table.special_regions.add(std::make_pair(interval, std::set<SpecialRegion>{region}));
+ 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}));
}
-void UnmapRegion(PageTable& page_table, VAddr base, u64 size) {
+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, PageType::Unmapped);
+ MapPages(page_table, base / PAGE_SIZE, size / PAGE_SIZE, nullptr, Common::PageType::Unmapped);
auto interval = boost::icl::discrete_interval<VAddr>::closed(base, base + size - 1);
page_table.special_regions.erase(interval);
}
-void AddDebugHook(PageTable& page_table, VAddr base, u64 size, MemoryHookPointer hook) {
+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);
- SpecialRegion region{SpecialRegion::Type::DebugHook, std::move(hook)};
- page_table.special_regions.add(std::make_pair(interval, std::set<SpecialRegion>{region}));
+ 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 RemoveDebugHook(PageTable& page_table, VAddr base, u64 size, MemoryHookPointer hook) {
+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);
- SpecialRegion region{SpecialRegion::Type::DebugHook, std::move(hook)};
- page_table.special_regions.subtract(std::make_pair(interval, std::set<SpecialRegion>{region}));
+ Common::SpecialRegion region{Common::SpecialRegion::Type::DebugHook, std::move(hook)};
+ page_table.special_regions.subtract(
+ std::make_pair(interval, std::set<Common::SpecialRegion>{region}));
}
/**
@@ -171,19 +160,19 @@ T Read(const VAddr vaddr) {
return value;
}
- PageType type = current_page_table->attributes[vaddr >> PAGE_BITS];
+ Common::PageType type = current_page_table->attributes[vaddr >> PAGE_BITS];
switch (type) {
- case PageType::Unmapped:
+ case Common::PageType::Unmapped:
LOG_ERROR(HW_Memory, "Unmapped Read{} @ 0x{:08X}", sizeof(T) * 8, vaddr);
return 0;
- case PageType::Memory:
+ case Common::PageType::Memory:
ASSERT_MSG(false, "Mapped memory page without a pointer @ {:016X}", vaddr);
break;
- case PageType::RasterizerCachedMemory: {
- RasterizerFlushVirtualRegion(vaddr, sizeof(T), FlushMode::Flush);
-
+ case Common::PageType::RasterizerCachedMemory: {
+ auto host_ptr{GetPointerFromVMA(vaddr)};
+ Core::System::GetInstance().GPU().FlushRegion(ToCacheAddr(host_ptr), sizeof(T));
T value;
- std::memcpy(&value, GetPointerFromVMA(vaddr), sizeof(T));
+ std::memcpy(&value, host_ptr, sizeof(T));
return value;
}
default:
@@ -201,18 +190,19 @@ void Write(const VAddr vaddr, const T data) {
return;
}
- PageType type = current_page_table->attributes[vaddr >> PAGE_BITS];
+ Common::PageType type = current_page_table->attributes[vaddr >> PAGE_BITS];
switch (type) {
- case PageType::Unmapped:
+ case Common::PageType::Unmapped:
LOG_ERROR(HW_Memory, "Unmapped Write{} 0x{:08X} @ 0x{:016X}", sizeof(data) * 8,
static_cast<u32>(data), vaddr);
return;
- case PageType::Memory:
+ case Common::PageType::Memory:
ASSERT_MSG(false, "Mapped memory page without a pointer @ {:016X}", vaddr);
break;
- case PageType::RasterizerCachedMemory: {
- RasterizerFlushVirtualRegion(vaddr, sizeof(T), FlushMode::Invalidate);
- std::memcpy(GetPointerFromVMA(vaddr), &data, sizeof(T));
+ 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:
@@ -227,10 +217,10 @@ bool IsValidVirtualAddress(const Kernel::Process& process, const VAddr vaddr) {
if (page_pointer)
return true;
- if (page_table.attributes[vaddr >> PAGE_BITS] == PageType::RasterizerCachedMemory)
+ if (page_table.attributes[vaddr >> PAGE_BITS] == Common::PageType::RasterizerCachedMemory)
return true;
- if (page_table.attributes[vaddr >> PAGE_BITS] != PageType::Special)
+ if (page_table.attributes[vaddr >> PAGE_BITS] != Common::PageType::Special)
return false;
return false;
@@ -250,7 +240,8 @@ u8* GetPointer(const VAddr vaddr) {
return page_pointer + (vaddr & PAGE_MASK);
}
- if (current_page_table->attributes[vaddr >> PAGE_BITS] == PageType::RasterizerCachedMemory) {
+ if (current_page_table->attributes[vaddr >> PAGE_BITS] ==
+ Common::PageType::RasterizerCachedMemory) {
return GetPointerFromVMA(vaddr);
}
@@ -284,20 +275,20 @@ void RasterizerMarkRegionCached(VAddr vaddr, u64 size, bool cached) {
u64 num_pages = ((vaddr + size - 1) >> PAGE_BITS) - (vaddr >> PAGE_BITS) + 1;
for (unsigned i = 0; i < num_pages; ++i, vaddr += PAGE_SIZE) {
- PageType& page_type = current_page_table->attributes[vaddr >> PAGE_BITS];
+ 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 PageType::Unmapped:
+ 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 PageType::Memory:
- page_type = PageType::RasterizerCachedMemory;
+ case Common::PageType::Memory:
+ page_type = Common::PageType::RasterizerCachedMemory;
current_page_table->pointers[vaddr >> PAGE_BITS] = nullptr;
break;
- case PageType::RasterizerCachedMemory:
+ 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;
@@ -307,23 +298,23 @@ void RasterizerMarkRegionCached(VAddr vaddr, u64 size, bool cached) {
} else {
// Switch page type to uncached if now uncached
switch (page_type) {
- case PageType::Unmapped:
+ 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 PageType::Memory:
+ 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 PageType::RasterizerCachedMemory: {
+ 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 = PageType::Unmapped;
+ page_type = Common::PageType::Unmapped;
} else {
- page_type = PageType::Memory;
+ page_type = Common::PageType::Memory;
current_page_table->pointers[vaddr >> PAGE_BITS] = pointer;
}
break;
@@ -335,47 +326,6 @@ void RasterizerMarkRegionCached(VAddr vaddr, u64 size, bool cached) {
}
}
-void RasterizerFlushVirtualRegion(VAddr start, u64 size, FlushMode mode) {
- auto& system_instance = Core::System::GetInstance();
-
- // Since pages are unmapped on shutdown after video core is shutdown, the renderer may be
- // null here
- if (!system_instance.IsPoweredOn()) {
- return;
- }
-
- const VAddr end = start + size;
-
- const auto CheckRegion = [&](VAddr region_start, VAddr region_end) {
- if (start >= region_end || end <= region_start) {
- // No overlap with region
- return;
- }
-
- const VAddr overlap_start = std::max(start, region_start);
- const VAddr overlap_end = std::min(end, region_end);
- const VAddr overlap_size = overlap_end - overlap_start;
-
- auto& gpu = system_instance.GPU();
- switch (mode) {
- case FlushMode::Flush:
- gpu.FlushRegion(overlap_start, overlap_size);
- break;
- case FlushMode::Invalidate:
- gpu.InvalidateRegion(overlap_start, overlap_size);
- break;
- case FlushMode::FlushAndInvalidate:
- gpu.FlushAndInvalidateRegion(overlap_start, overlap_size);
- break;
- }
- };
-
- const auto& vm_manager = Core::CurrentProcess()->VMManager();
-
- CheckRegion(vm_manager.GetCodeRegionBaseAddress(), vm_manager.GetCodeRegionEndAddress());
- CheckRegion(vm_manager.GetHeapRegionBaseAddress(), vm_manager.GetHeapRegionEndAddress());
-}
-
u8 Read8(const VAddr addr) {
return Read<u8>(addr);
}
@@ -406,24 +356,24 @@ void ReadBlock(const Kernel::Process& process, const VAddr src_addr, void* dest_
const VAddr current_vaddr = static_cast<VAddr>((page_index << PAGE_BITS) + page_offset);
switch (page_table.attributes[page_index]) {
- case PageType::Unmapped: {
+ 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 PageType::Memory: {
+ 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);
break;
}
- case PageType::RasterizerCachedMemory: {
- RasterizerFlushVirtualRegion(current_vaddr, static_cast<u32>(copy_amount),
- FlushMode::Flush);
- std::memcpy(dest_buffer, GetPointerFromVMA(process, current_vaddr), copy_amount);
+ 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);
break;
}
default:
@@ -470,23 +420,23 @@ void WriteBlock(const Kernel::Process& process, const VAddr dest_addr, const voi
const VAddr current_vaddr = static_cast<VAddr>((page_index << PAGE_BITS) + page_offset);
switch (page_table.attributes[page_index]) {
- case PageType::Unmapped: {
+ case Common::PageType::Unmapped: {
LOG_ERROR(HW_Memory,
"Unmapped WriteBlock @ 0x{:016X} (start address = 0x{:016X}, size = {})",
current_vaddr, dest_addr, size);
break;
}
- case PageType::Memory: {
+ case Common::PageType::Memory: {
DEBUG_ASSERT(page_table.pointers[page_index]);
u8* dest_ptr = page_table.pointers[page_index] + page_offset;
std::memcpy(dest_ptr, src_buffer, copy_amount);
break;
}
- case PageType::RasterizerCachedMemory: {
- RasterizerFlushVirtualRegion(current_vaddr, static_cast<u32>(copy_amount),
- FlushMode::Invalidate);
- std::memcpy(GetPointerFromVMA(process, current_vaddr), src_buffer, copy_amount);
+ 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:
@@ -516,23 +466,23 @@ void ZeroBlock(const Kernel::Process& process, const VAddr dest_addr, const std:
const VAddr current_vaddr = static_cast<VAddr>((page_index << PAGE_BITS) + page_offset);
switch (page_table.attributes[page_index]) {
- case PageType::Unmapped: {
+ case Common::PageType::Unmapped: {
LOG_ERROR(HW_Memory,
"Unmapped ZeroBlock @ 0x{:016X} (start address = 0x{:016X}, size = {})",
current_vaddr, dest_addr, size);
break;
}
- case PageType::Memory: {
+ 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 PageType::RasterizerCachedMemory: {
- RasterizerFlushVirtualRegion(current_vaddr, static_cast<u32>(copy_amount),
- FlushMode::Invalidate);
- std::memset(GetPointerFromVMA(process, current_vaddr), 0, copy_amount);
+ 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:
@@ -558,23 +508,23 @@ void CopyBlock(const Kernel::Process& process, VAddr dest_addr, VAddr src_addr,
const VAddr current_vaddr = static_cast<VAddr>((page_index << PAGE_BITS) + page_offset);
switch (page_table.attributes[page_index]) {
- case PageType::Unmapped: {
+ 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 PageType::Memory: {
+ 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 PageType::RasterizerCachedMemory: {
- RasterizerFlushVirtualRegion(current_vaddr, static_cast<u32>(copy_amount),
- FlushMode::Flush);
- WriteBlock(process, dest_addr, GetPointerFromVMA(process, current_vaddr), copy_amount);
+ 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:
diff --git a/src/core/memory.h b/src/core/memory.h
index 1acf5ce8c..3f60d868c 100644
--- a/src/core/memory.h
+++ b/src/core/memory.h
@@ -10,7 +10,10 @@
#include <vector>
#include <boost/icl/interval_map.hpp>
#include "common/common_types.h"
-#include "core/memory_hook.h"
+
+namespace Common {
+struct PageTable;
+}
namespace Kernel {
class Process;
@@ -26,71 +29,6 @@ constexpr std::size_t PAGE_BITS = 12;
constexpr u64 PAGE_SIZE = 1ULL << PAGE_BITS;
constexpr u64 PAGE_MASK = PAGE_SIZE - 1;
-enum class PageType : u8 {
- /// Page is unmapped and should cause an access error.
- Unmapped,
- /// Page is mapped to regular memory. This is the only type you can get pointers to.
- Memory,
- /// Page is mapped to regular memory, but also needs to check for rasterizer cache flushing and
- /// invalidation
- RasterizerCachedMemory,
- /// Page is mapped to a I/O region. Writing and reading to this page is handled by functions.
- Special,
-};
-
-struct SpecialRegion {
- enum class Type {
- DebugHook,
- IODevice,
- } type;
-
- MemoryHookPointer handler;
-
- bool operator<(const SpecialRegion& other) const {
- return std::tie(type, handler) < std::tie(other.type, other.handler);
- }
-
- bool operator==(const SpecialRegion& other) const {
- return std::tie(type, handler) == std::tie(other.type, other.handler);
- }
-};
-
-/**
- * A (reasonably) fast way of allowing switchable and remappable process address spaces. It loosely
- * mimics the way a real CPU page table works.
- */
-struct PageTable {
- explicit PageTable();
- explicit PageTable(std::size_t address_space_width_in_bits);
- ~PageTable();
-
- /**
- * Resizes the page table to be able to accomodate enough pages within
- * a given address space.
- *
- * @param address_space_width_in_bits The address size width in bits.
- */
- void Resize(std::size_t address_space_width_in_bits);
-
- /**
- * Vector of memory pointers backing each page. An entry can only be non-null if the
- * corresponding entry in the `attributes` vector is of type `Memory`.
- */
- std::vector<u8*> pointers;
-
- /**
- * Contains MMIO handlers that back memory regions whose entries in the `attribute` vector is
- * of type `Special`.
- */
- boost::icl::interval_map<VAddr, std::set<SpecialRegion>> special_regions;
-
- /**
- * Vector of fine grained page attributes. If it is set to any value other than `Memory`, then
- * the corresponding entry in `pointers` MUST be set to null.
- */
- std::vector<PageType> attributes;
-};
-
/// Virtual user-space memory regions
enum : VAddr {
/// Read-only page containing kernel and system configuration values.
@@ -116,8 +54,8 @@ enum : VAddr {
};
/// Currently active page table
-void SetCurrentPageTable(PageTable* page_table);
-PageTable* GetCurrentPageTable();
+void SetCurrentPageTable(Common::PageTable* page_table);
+Common::PageTable* GetCurrentPageTable();
/// Determines if the given VAddr is valid for the specified process.
bool IsValidVirtualAddress(const Kernel::Process& process, VAddr vaddr);
@@ -161,10 +99,4 @@ enum class FlushMode {
*/
void RasterizerMarkRegionCached(VAddr vaddr, u64 size, bool cached);
-/**
- * Flushes and invalidates any externally cached rasterizer resources touching the given virtual
- * address region.
- */
-void RasterizerFlushVirtualRegion(VAddr start, u64 size, FlushMode mode);
-
} // namespace Memory
diff --git a/src/core/memory_setup.h b/src/core/memory_setup.h
index 9a1a4f4be..5225ee8e2 100644
--- a/src/core/memory_setup.h
+++ b/src/core/memory_setup.h
@@ -5,7 +5,11 @@
#pragma once
#include "common/common_types.h"
-#include "core/memory_hook.h"
+#include "common/memory_hook.h"
+
+namespace Common {
+struct PageTable;
+}
namespace Memory {
@@ -17,7 +21,7 @@ namespace Memory {
* @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(PageTable& page_table, VAddr base, u64 size, u8* target);
+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.
@@ -26,11 +30,14 @@ void MapMemoryRegion(PageTable& page_table, VAddr base, u64 size, u8* target);
* @param size The amount of bytes to map. Must be page-aligned.
* @param mmio_handler The handler that backs the mapping.
*/
-void MapIoRegion(PageTable& page_table, VAddr base, u64 size, MemoryHookPointer mmio_handler);
+void MapIoRegion(Common::PageTable& page_table, VAddr base, u64 size,
+ Common::MemoryHookPointer mmio_handler);
-void UnmapRegion(PageTable& page_table, VAddr base, u64 size);
+void UnmapRegion(Common::PageTable& page_table, VAddr base, u64 size);
-void AddDebugHook(PageTable& page_table, VAddr base, u64 size, MemoryHookPointer hook);
-void RemoveDebugHook(PageTable& page_table, VAddr base, u64 size, MemoryHookPointer hook);
+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/input_common/CMakeLists.txt b/src/input_common/CMakeLists.txt
index 1c7db28c0..5b4e032bd 100644
--- a/src/input_common/CMakeLists.txt
+++ b/src/input_common/CMakeLists.txt
@@ -7,15 +7,18 @@ add_library(input_common STATIC
main.h
motion_emu.cpp
motion_emu.h
-
- $<$<BOOL:${SDL2_FOUND}>:sdl/sdl.cpp sdl/sdl.h>
+ sdl/sdl.cpp
+ sdl/sdl.h
)
-create_target_directory_groups(input_common)
-
-target_link_libraries(input_common PUBLIC core PRIVATE common)
-
if(SDL2_FOUND)
+ target_sources(input_common PRIVATE
+ sdl/sdl_impl.cpp
+ sdl/sdl_impl.h
+ )
target_link_libraries(input_common PRIVATE SDL2)
target_compile_definitions(input_common PRIVATE HAVE_SDL2)
endif()
+
+create_target_directory_groups(input_common)
+target_link_libraries(input_common PUBLIC core PRIVATE common)
diff --git a/src/input_common/main.cpp b/src/input_common/main.cpp
index 37f572853..8e66c1b15 100644
--- a/src/input_common/main.cpp
+++ b/src/input_common/main.cpp
@@ -17,10 +17,7 @@ namespace InputCommon {
static std::shared_ptr<Keyboard> keyboard;
static std::shared_ptr<MotionEmu> motion_emu;
-
-#ifdef HAVE_SDL2
-static std::thread poll_thread;
-#endif
+static std::unique_ptr<SDL::State> sdl;
void Init() {
keyboard = std::make_shared<Keyboard>();
@@ -30,15 +27,7 @@ void Init() {
motion_emu = std::make_shared<MotionEmu>();
Input::RegisterFactory<Input::MotionDevice>("motion_emu", motion_emu);
-#ifdef HAVE_SDL2
- SDL::Init();
-#endif
-}
-
-void StartJoystickEventHandler() {
-#ifdef HAVE_SDL2
- poll_thread = std::thread(SDL::PollLoop);
-#endif
+ sdl = SDL::Init();
}
void Shutdown() {
@@ -47,11 +36,7 @@ void Shutdown() {
Input::UnregisterFactory<Input::AnalogDevice>("analog_from_button");
Input::UnregisterFactory<Input::MotionDevice>("motion_emu");
motion_emu.reset();
-
-#ifdef HAVE_SDL2
- SDL::Shutdown();
- poll_thread.join();
-#endif
+ sdl.reset();
}
Keyboard* GetKeyboard() {
@@ -88,7 +73,7 @@ namespace Polling {
std::vector<std::unique_ptr<DevicePoller>> GetPollers(DeviceType type) {
#ifdef HAVE_SDL2
- return SDL::Polling::GetPollers(type);
+ return sdl->GetPollers(type);
#else
return {};
#endif
diff --git a/src/input_common/main.h b/src/input_common/main.h
index 9eb13106e..77a0ce90b 100644
--- a/src/input_common/main.h
+++ b/src/input_common/main.h
@@ -20,8 +20,6 @@ void Init();
/// Deregisters all built-in input device factories and shuts them down.
void Shutdown();
-void StartJoystickEventHandler();
-
class Keyboard;
/// Gets the keyboard button device factory.
diff --git a/src/input_common/sdl/sdl.cpp b/src/input_common/sdl/sdl.cpp
index faf3c1fa3..644db3448 100644
--- a/src/input_common/sdl/sdl.cpp
+++ b/src/input_common/sdl/sdl.cpp
@@ -1,631 +1,19 @@
-// Copyright 2017 Citra Emulator Project
+// Copyright 2018 Citra Emulator Project
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
-#include <algorithm>
-#include <atomic>
-#include <cmath>
-#include <functional>
-#include <iterator>
-#include <mutex>
-#include <string>
-#include <thread>
-#include <tuple>
-#include <unordered_map>
-#include <utility>
-#include <vector>
-#include <SDL.h>
-#include "common/assert.h"
-#include "common/logging/log.h"
-#include "common/math_util.h"
-#include "common/param_package.h"
-#include "common/threadsafe_queue.h"
-#include "input_common/main.h"
#include "input_common/sdl/sdl.h"
+#ifdef HAVE_SDL2
+#include "input_common/sdl/sdl_impl.h"
+#endif
-namespace InputCommon {
+namespace InputCommon::SDL {
-namespace SDL {
-
-class SDLJoystick;
-class SDLButtonFactory;
-class SDLAnalogFactory;
-
-/// Map of GUID of a list of corresponding virtual Joysticks
-static std::unordered_map<std::string, std::vector<std::shared_ptr<SDLJoystick>>> joystick_map;
-static std::mutex joystick_map_mutex;
-
-static std::shared_ptr<SDLButtonFactory> button_factory;
-static std::shared_ptr<SDLAnalogFactory> analog_factory;
-
-/// Used by the Pollers during config
-static std::atomic<bool> polling;
-static Common::SPSCQueue<SDL_Event> event_queue;
-
-static std::atomic<bool> initialized = false;
-
-static std::string GetGUID(SDL_Joystick* joystick) {
- SDL_JoystickGUID guid = SDL_JoystickGetGUID(joystick);
- char guid_str[33];
- SDL_JoystickGetGUIDString(guid, guid_str, sizeof(guid_str));
- return guid_str;
-}
-
-class SDLJoystick {
-public:
- SDLJoystick(std::string guid_, int port_, SDL_Joystick* joystick,
- decltype(&SDL_JoystickClose) deleter = &SDL_JoystickClose)
- : guid{std::move(guid_)}, port{port_}, sdl_joystick{joystick, deleter} {}
-
- void SetButton(int button, bool value) {
- std::lock_guard<std::mutex> lock(mutex);
- state.buttons[button] = value;
- }
-
- bool GetButton(int button) const {
- std::lock_guard<std::mutex> lock(mutex);
- return state.buttons.at(button);
- }
-
- void SetAxis(int axis, Sint16 value) {
- std::lock_guard<std::mutex> lock(mutex);
- state.axes[axis] = value;
- }
-
- float GetAxis(int axis) const {
- std::lock_guard<std::mutex> lock(mutex);
- return state.axes.at(axis) / 32767.0f;
- }
-
- std::tuple<float, float> GetAnalog(int axis_x, int axis_y) const {
- float x = GetAxis(axis_x);
- float y = GetAxis(axis_y);
- y = -y; // 3DS uses an y-axis inverse from SDL
-
- // Make sure the coordinates are in the unit circle,
- // otherwise normalize it.
- float r = x * x + y * y;
- if (r > 1.0f) {
- r = std::sqrt(r);
- x /= r;
- y /= r;
- }
-
- return std::make_tuple(x, y);
- }
-
- void SetHat(int hat, Uint8 direction) {
- std::lock_guard<std::mutex> lock(mutex);
- state.hats[hat] = direction;
- }
-
- bool GetHatDirection(int hat, Uint8 direction) const {
- std::lock_guard<std::mutex> lock(mutex);
- return (state.hats.at(hat) & direction) != 0;
- }
- /**
- * The guid of the joystick
- */
- const std::string& GetGUID() const {
- return guid;
- }
-
- /**
- * The number of joystick from the same type that were connected before this joystick
- */
- int GetPort() const {
- return port;
- }
-
- SDL_Joystick* GetSDLJoystick() const {
- return sdl_joystick.get();
- }
-
- void SetSDLJoystick(SDL_Joystick* joystick,
- decltype(&SDL_JoystickClose) deleter = &SDL_JoystickClose) {
- sdl_joystick =
- std::unique_ptr<SDL_Joystick, decltype(&SDL_JoystickClose)>(joystick, deleter);
- }
-
-private:
- struct State {
- std::unordered_map<int, bool> buttons;
- std::unordered_map<int, Sint16> axes;
- std::unordered_map<int, Uint8> hats;
- } state;
- std::string guid;
- int port;
- std::unique_ptr<SDL_Joystick, decltype(&SDL_JoystickClose)> sdl_joystick;
- mutable std::mutex mutex;
-};
-
-/**
- * Get the nth joystick with the corresponding GUID
- */
-static std::shared_ptr<SDLJoystick> GetSDLJoystickByGUID(const std::string& guid, int port) {
- std::lock_guard<std::mutex> lock(joystick_map_mutex);
- const auto it = joystick_map.find(guid);
- if (it != joystick_map.end()) {
- while (it->second.size() <= port) {
- auto joystick = std::make_shared<SDLJoystick>(guid, it->second.size(), nullptr,
- [](SDL_Joystick*) {});
- it->second.emplace_back(std::move(joystick));
- }
- return it->second[port];
- }
- auto joystick = std::make_shared<SDLJoystick>(guid, 0, nullptr, [](SDL_Joystick*) {});
- return joystick_map[guid].emplace_back(std::move(joystick));
-}
-
-/**
- * Check how many identical joysticks (by guid) were connected before the one with sdl_id and so tie
- * it to a SDLJoystick with the same guid and that port
- */
-static std::shared_ptr<SDLJoystick> GetSDLJoystickBySDLID(SDL_JoystickID sdl_id) {
- std::lock_guard<std::mutex> lock(joystick_map_mutex);
- auto sdl_joystick = SDL_JoystickFromInstanceID(sdl_id);
- const std::string guid = GetGUID(sdl_joystick);
- auto map_it = joystick_map.find(guid);
- if (map_it != joystick_map.end()) {
- auto vec_it = std::find_if(map_it->second.begin(), map_it->second.end(),
- [&sdl_joystick](const std::shared_ptr<SDLJoystick>& joystick) {
- return sdl_joystick == joystick->GetSDLJoystick();
- });
- if (vec_it != map_it->second.end()) {
- // This is the common case: There is already an existing SDL_Joystick maped to a
- // SDLJoystick. return the SDLJoystick
- return *vec_it;
- }
- // Search for a SDLJoystick without a mapped SDL_Joystick...
- auto nullptr_it = std::find_if(map_it->second.begin(), map_it->second.end(),
- [](const std::shared_ptr<SDLJoystick>& joystick) {
- return !joystick->GetSDLJoystick();
- });
- if (nullptr_it != map_it->second.end()) {
- // ... and map it
- (*nullptr_it)->SetSDLJoystick(sdl_joystick);
- return *nullptr_it;
- }
- // There is no SDLJoystick without a mapped SDL_Joystick
- // Create a new SDLJoystick
- auto joystick = std::make_shared<SDLJoystick>(guid, map_it->second.size(), sdl_joystick);
- return map_it->second.emplace_back(std::move(joystick));
- }
- auto joystick = std::make_shared<SDLJoystick>(guid, 0, sdl_joystick);
- return joystick_map[guid].emplace_back(std::move(joystick));
-}
-
-void InitJoystick(int joystick_index) {
- std::lock_guard<std::mutex> lock(joystick_map_mutex);
- SDL_Joystick* sdl_joystick = SDL_JoystickOpen(joystick_index);
- if (!sdl_joystick) {
- LOG_ERROR(Input, "failed to open joystick {}", joystick_index);
- return;
- }
- std::string guid = GetGUID(sdl_joystick);
- if (joystick_map.find(guid) == joystick_map.end()) {
- auto joystick = std::make_shared<SDLJoystick>(guid, 0, sdl_joystick);
- joystick_map[guid].emplace_back(std::move(joystick));
- return;
- }
- auto& joystick_guid_list = joystick_map[guid];
- const auto it = std::find_if(
- joystick_guid_list.begin(), joystick_guid_list.end(),
- [](const std::shared_ptr<SDLJoystick>& joystick) { return !joystick->GetSDLJoystick(); });
- if (it != joystick_guid_list.end()) {
- (*it)->SetSDLJoystick(sdl_joystick);
- return;
- }
- auto joystick = std::make_shared<SDLJoystick>(guid, joystick_guid_list.size(), sdl_joystick);
- joystick_guid_list.emplace_back(std::move(joystick));
-}
-
-void CloseJoystick(SDL_Joystick* sdl_joystick) {
- std::lock_guard<std::mutex> lock(joystick_map_mutex);
- std::string guid = GetGUID(sdl_joystick);
- // This call to guid is save since the joystick is guranteed to be in that map
- auto& joystick_guid_list = joystick_map[guid];
- const auto joystick_it =
- std::find_if(joystick_guid_list.begin(), joystick_guid_list.end(),
- [&sdl_joystick](const std::shared_ptr<SDLJoystick>& joystick) {
- return joystick->GetSDLJoystick() == sdl_joystick;
- });
- (*joystick_it)->SetSDLJoystick(nullptr, [](SDL_Joystick*) {});
-}
-
-void HandleGameControllerEvent(const SDL_Event& event) {
- switch (event.type) {
- case SDL_JOYBUTTONUP: {
- auto joystick = GetSDLJoystickBySDLID(event.jbutton.which);
- if (joystick) {
- joystick->SetButton(event.jbutton.button, false);
- }
- break;
- }
- case SDL_JOYBUTTONDOWN: {
- auto joystick = GetSDLJoystickBySDLID(event.jbutton.which);
- if (joystick) {
- joystick->SetButton(event.jbutton.button, true);
- }
- break;
- }
- case SDL_JOYHATMOTION: {
- auto joystick = GetSDLJoystickBySDLID(event.jhat.which);
- if (joystick) {
- joystick->SetHat(event.jhat.hat, event.jhat.value);
- }
- break;
- }
- case SDL_JOYAXISMOTION: {
- auto joystick = GetSDLJoystickBySDLID(event.jaxis.which);
- if (joystick) {
- joystick->SetAxis(event.jaxis.axis, event.jaxis.value);
- }
- break;
- }
- case SDL_JOYDEVICEREMOVED:
- LOG_DEBUG(Input, "Controller removed with Instance_ID {}", event.jdevice.which);
- CloseJoystick(SDL_JoystickFromInstanceID(event.jdevice.which));
- break;
- case SDL_JOYDEVICEADDED:
- LOG_DEBUG(Input, "Controller connected with device index {}", event.jdevice.which);
- InitJoystick(event.jdevice.which);
- break;
- }
-}
-
-void CloseSDLJoysticks() {
- std::lock_guard<std::mutex> lock(joystick_map_mutex);
- joystick_map.clear();
-}
-
-void PollLoop() {
- if (SDL_Init(SDL_INIT_JOYSTICK) < 0) {
- LOG_CRITICAL(Input, "SDL_Init(SDL_INIT_JOYSTICK) failed with: {}", SDL_GetError());
- return;
- }
-
- SDL_Event event;
- while (initialized) {
- // Wait for 10 ms or until an event happens
- if (SDL_WaitEventTimeout(&event, 10)) {
- // Don't handle the event if we are configuring
- if (polling) {
- event_queue.Push(event);
- } else {
- HandleGameControllerEvent(event);
- }
- }
- }
- CloseSDLJoysticks();
- SDL_QuitSubSystem(SDL_INIT_JOYSTICK);
-}
-
-class SDLButton final : public Input::ButtonDevice {
-public:
- explicit SDLButton(std::shared_ptr<SDLJoystick> joystick_, int button_)
- : joystick(std::move(joystick_)), button(button_) {}
-
- bool GetStatus() const override {
- return joystick->GetButton(button);
- }
-
-private:
- std::shared_ptr<SDLJoystick> joystick;
- int button;
-};
-
-class SDLDirectionButton final : public Input::ButtonDevice {
-public:
- explicit SDLDirectionButton(std::shared_ptr<SDLJoystick> joystick_, int hat_, Uint8 direction_)
- : joystick(std::move(joystick_)), hat(hat_), direction(direction_) {}
-
- bool GetStatus() const override {
- return joystick->GetHatDirection(hat, direction);
- }
-
-private:
- std::shared_ptr<SDLJoystick> joystick;
- int hat;
- Uint8 direction;
-};
-
-class SDLAxisButton final : public Input::ButtonDevice {
-public:
- explicit SDLAxisButton(std::shared_ptr<SDLJoystick> joystick_, int axis_, float threshold_,
- bool trigger_if_greater_)
- : joystick(std::move(joystick_)), axis(axis_), threshold(threshold_),
- trigger_if_greater(trigger_if_greater_) {}
-
- bool GetStatus() const override {
- float axis_value = joystick->GetAxis(axis);
- if (trigger_if_greater)
- return axis_value > threshold;
- return axis_value < threshold;
- }
-
-private:
- std::shared_ptr<SDLJoystick> joystick;
- int axis;
- float threshold;
- bool trigger_if_greater;
-};
-
-class SDLAnalog final : public Input::AnalogDevice {
-public:
- SDLAnalog(std::shared_ptr<SDLJoystick> joystick_, int axis_x_, int axis_y_)
- : joystick(std::move(joystick_)), axis_x(axis_x_), axis_y(axis_y_) {}
-
- std::tuple<float, float> GetStatus() const override {
- return joystick->GetAnalog(axis_x, axis_y);
- }
-
-private:
- std::shared_ptr<SDLJoystick> joystick;
- int axis_x;
- int axis_y;
-};
-
-/// A button device factory that creates button devices from SDL joystick
-class SDLButtonFactory final : public Input::Factory<Input::ButtonDevice> {
-public:
- /**
- * Creates a button device from a joystick button
- * @param params contains parameters for creating the device:
- * - "guid": the guid of the joystick to bind
- * - "port": the nth joystick of the same type to bind
- * - "button"(optional): the index of the button to bind
- * - "hat"(optional): the index of the hat to bind as direction buttons
- * - "axis"(optional): the index of the axis to bind
- * - "direction"(only used for hat): the direction name of the hat to bind. Can be "up",
- * "down", "left" or "right"
- * - "threshold"(only used for axis): a float value in (-1.0, 1.0) which the button is
- * triggered if the axis value crosses
- * - "direction"(only used for axis): "+" means the button is triggered when the axis
- * value is greater than the threshold; "-" means the button is triggered when the axis
- * value is smaller than the threshold
- */
- std::unique_ptr<Input::ButtonDevice> Create(const Common::ParamPackage& params) override {
- const std::string guid = params.Get("guid", "0");
- const int port = params.Get("port", 0);
-
- auto joystick = GetSDLJoystickByGUID(guid, port);
-
- if (params.Has("hat")) {
- const int hat = params.Get("hat", 0);
- const std::string direction_name = params.Get("direction", "");
- Uint8 direction;
- if (direction_name == "up") {
- direction = SDL_HAT_UP;
- } else if (direction_name == "down") {
- direction = SDL_HAT_DOWN;
- } else if (direction_name == "left") {
- direction = SDL_HAT_LEFT;
- } else if (direction_name == "right") {
- direction = SDL_HAT_RIGHT;
- } else {
- direction = 0;
- }
- // This is necessary so accessing GetHat with hat won't crash
- joystick->SetHat(hat, SDL_HAT_CENTERED);
- return std::make_unique<SDLDirectionButton>(joystick, hat, direction);
- }
-
- if (params.Has("axis")) {
- const int axis = params.Get("axis", 0);
- const float threshold = params.Get("threshold", 0.5f);
- const std::string direction_name = params.Get("direction", "");
- bool trigger_if_greater;
- if (direction_name == "+") {
- trigger_if_greater = true;
- } else if (direction_name == "-") {
- trigger_if_greater = false;
- } else {
- trigger_if_greater = true;
- LOG_ERROR(Input, "Unknown direction '{}'", direction_name);
- }
- // This is necessary so accessing GetAxis with axis won't crash
- joystick->SetAxis(axis, 0);
- return std::make_unique<SDLAxisButton>(joystick, axis, threshold, trigger_if_greater);
- }
-
- const int button = params.Get("button", 0);
- // This is necessary so accessing GetButton with button won't crash
- joystick->SetButton(button, false);
- return std::make_unique<SDLButton>(joystick, button);
- }
-};
-
-/// An analog device factory that creates analog devices from SDL joystick
-class SDLAnalogFactory final : public Input::Factory<Input::AnalogDevice> {
-public:
- /**
- * Creates analog device from joystick axes
- * @param params contains parameters for creating the device:
- * - "guid": the guid of the joystick to bind
- * - "port": the nth joystick of the same type
- * - "axis_x": the index of the axis to be bind as x-axis
- * - "axis_y": the index of the axis to be bind as y-axis
- */
- std::unique_ptr<Input::AnalogDevice> Create(const Common::ParamPackage& params) override {
- const std::string guid = params.Get("guid", "0");
- const int port = params.Get("port", 0);
- const int axis_x = params.Get("axis_x", 0);
- const int axis_y = params.Get("axis_y", 1);
-
- auto joystick = GetSDLJoystickByGUID(guid, port);
-
- // This is necessary so accessing GetAxis with axis_x and axis_y won't crash
- joystick->SetAxis(axis_x, 0);
- joystick->SetAxis(axis_y, 0);
- return std::make_unique<SDLAnalog>(joystick, axis_x, axis_y);
- }
-};
-
-void Init() {
- using namespace Input;
- RegisterFactory<ButtonDevice>("sdl", std::make_shared<SDLButtonFactory>());
- RegisterFactory<AnalogDevice>("sdl", std::make_shared<SDLAnalogFactory>());
- polling = false;
- initialized = true;
-}
-
-void Shutdown() {
- if (initialized) {
- using namespace Input;
- UnregisterFactory<ButtonDevice>("sdl");
- UnregisterFactory<AnalogDevice>("sdl");
- initialized = false;
- }
-}
-
-Common::ParamPackage SDLEventToButtonParamPackage(const SDL_Event& event) {
- Common::ParamPackage params({{"engine", "sdl"}});
- switch (event.type) {
- case SDL_JOYAXISMOTION: {
- auto joystick = GetSDLJoystickBySDLID(event.jaxis.which);
- params.Set("port", joystick->GetPort());
- params.Set("guid", joystick->GetGUID());
- params.Set("axis", event.jaxis.axis);
- if (event.jaxis.value > 0) {
- params.Set("direction", "+");
- params.Set("threshold", "0.5");
- } else {
- params.Set("direction", "-");
- params.Set("threshold", "-0.5");
- }
- break;
- }
- case SDL_JOYBUTTONUP: {
- auto joystick = GetSDLJoystickBySDLID(event.jbutton.which);
- params.Set("port", joystick->GetPort());
- params.Set("guid", joystick->GetGUID());
- params.Set("button", event.jbutton.button);
- break;
- }
- case SDL_JOYHATMOTION: {
- auto joystick = GetSDLJoystickBySDLID(event.jhat.which);
- params.Set("port", joystick->GetPort());
- params.Set("guid", joystick->GetGUID());
- params.Set("hat", event.jhat.hat);
- switch (event.jhat.value) {
- case SDL_HAT_UP:
- params.Set("direction", "up");
- break;
- case SDL_HAT_DOWN:
- params.Set("direction", "down");
- break;
- case SDL_HAT_LEFT:
- params.Set("direction", "left");
- break;
- case SDL_HAT_RIGHT:
- params.Set("direction", "right");
- break;
- default:
- return {};
- }
- break;
- }
- }
- return params;
-}
-
-namespace Polling {
-
-class SDLPoller : public InputCommon::Polling::DevicePoller {
-public:
- void Start() override {
- event_queue.Clear();
- polling = true;
- }
-
- void Stop() override {
- polling = false;
- }
-};
-
-class SDLButtonPoller final : public SDLPoller {
-public:
- Common::ParamPackage GetNextInput() override {
- SDL_Event event;
- while (event_queue.Pop(event)) {
- switch (event.type) {
- case SDL_JOYAXISMOTION:
- if (std::abs(event.jaxis.value / 32767.0) < 0.5) {
- break;
- }
- case SDL_JOYBUTTONUP:
- case SDL_JOYHATMOTION:
- return SDLEventToButtonParamPackage(event);
- }
- }
- return {};
- }
-};
-
-class SDLAnalogPoller final : public SDLPoller {
-public:
- void Start() override {
- SDLPoller::Start();
-
- // Reset stored axes
- analog_xaxis = -1;
- analog_yaxis = -1;
- analog_axes_joystick = -1;
- }
-
- Common::ParamPackage GetNextInput() override {
- SDL_Event event;
- while (event_queue.Pop(event)) {
- if (event.type != SDL_JOYAXISMOTION || std::abs(event.jaxis.value / 32767.0) < 0.5) {
- continue;
- }
- // An analog device needs two axes, so we need to store the axis for later and wait for
- // a second SDL event. The axes also must be from the same joystick.
- int axis = event.jaxis.axis;
- if (analog_xaxis == -1) {
- analog_xaxis = axis;
- analog_axes_joystick = event.jaxis.which;
- } else if (analog_yaxis == -1 && analog_xaxis != axis &&
- analog_axes_joystick == event.jaxis.which) {
- analog_yaxis = axis;
- }
- }
- Common::ParamPackage params;
- if (analog_xaxis != -1 && analog_yaxis != -1) {
- auto joystick = GetSDLJoystickBySDLID(event.jaxis.which);
- params.Set("engine", "sdl");
- params.Set("port", joystick->GetPort());
- params.Set("guid", joystick->GetGUID());
- params.Set("axis_x", analog_xaxis);
- params.Set("axis_y", analog_yaxis);
- analog_xaxis = -1;
- analog_yaxis = -1;
- analog_axes_joystick = -1;
- return params;
- }
- return params;
- }
-
-private:
- int analog_xaxis = -1;
- int analog_yaxis = -1;
- SDL_JoystickID analog_axes_joystick = -1;
-};
-
-std::vector<std::unique_ptr<InputCommon::Polling::DevicePoller>> GetPollers(
- InputCommon::Polling::DeviceType type) {
- std::vector<std::unique_ptr<InputCommon::Polling::DevicePoller>> pollers;
- switch (type) {
- case InputCommon::Polling::DeviceType::Analog:
- pollers.push_back(std::make_unique<SDLAnalogPoller>());
- break;
- case InputCommon::Polling::DeviceType::Button:
- pollers.push_back(std::make_unique<SDLButtonPoller>());
- break;
- }
- return pollers;
+std::unique_ptr<State> Init() {
+#ifdef HAVE_SDL2
+ return std::make_unique<SDLState>();
+#else
+ return std::make_unique<NullState>();
+#endif
}
-} // namespace Polling
-} // namespace SDL
-} // namespace InputCommon
+} // namespace InputCommon::SDL
diff --git a/src/input_common/sdl/sdl.h b/src/input_common/sdl/sdl.h
index 0206860d3..02a8d2e2c 100644
--- a/src/input_common/sdl/sdl.h
+++ b/src/input_common/sdl/sdl.h
@@ -1,4 +1,4 @@
-// Copyright 2017 Citra Emulator Project
+// Copyright 2018 Citra Emulator Project
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
@@ -7,45 +7,36 @@
#include <memory>
#include <vector>
#include "core/frontend/input.h"
+#include "input_common/main.h"
union SDL_Event;
+
namespace Common {
class ParamPackage;
-}
-namespace InputCommon {
-namespace Polling {
+} // namespace Common
+
+namespace InputCommon::Polling {
class DevicePoller;
enum class DeviceType;
-} // namespace Polling
-} // namespace InputCommon
-
-namespace InputCommon {
-namespace SDL {
-
-/// Initializes and registers SDL device factories
-void Init();
-
-/// Unresisters SDL device factories and shut them down.
-void Shutdown();
-
-/// Needs to be called before SDL_QuitSubSystem.
-void CloseSDLJoysticks();
+} // namespace InputCommon::Polling
-/// Handle SDL_Events for joysticks from SDL_PollEvent
-void HandleGameControllerEvent(const SDL_Event& event);
+namespace InputCommon::SDL {
-/// A Loop that calls HandleGameControllerEvent until Shutdown is called
-void PollLoop();
+class State {
+public:
+ /// Unresisters SDL device factories and shut them down.
+ virtual ~State() = default;
-/// Creates a ParamPackage from an SDL_Event that can directly be used to create a ButtonDevice
-Common::ParamPackage SDLEventToButtonParamPackage(const SDL_Event& event);
+ virtual std::vector<std::unique_ptr<InputCommon::Polling::DevicePoller>> GetPollers(
+ InputCommon::Polling::DeviceType type) = 0;
+};
-namespace Polling {
+class NullState : public State {
+public:
+ std::vector<std::unique_ptr<InputCommon::Polling::DevicePoller>> GetPollers(
+ InputCommon::Polling::DeviceType type) override {}
+};
-/// Get all DevicePoller that use the SDL backend for a specific device type
-std::vector<std::unique_ptr<InputCommon::Polling::DevicePoller>> GetPollers(
- InputCommon::Polling::DeviceType type);
+std::unique_ptr<State> Init();
-} // namespace Polling
-} // namespace SDL
-} // namespace InputCommon
+} // namespace InputCommon::SDL
diff --git a/src/input_common/sdl/sdl_impl.cpp b/src/input_common/sdl/sdl_impl.cpp
new file mode 100644
index 000000000..934339d3b
--- /dev/null
+++ b/src/input_common/sdl/sdl_impl.cpp
@@ -0,0 +1,669 @@
+// Copyright 2018 Citra Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <algorithm>
+#include <atomic>
+#include <cmath>
+#include <functional>
+#include <iterator>
+#include <mutex>
+#include <string>
+#include <thread>
+#include <tuple>
+#include <unordered_map>
+#include <utility>
+#include <vector>
+#include <SDL.h>
+#include "common/assert.h"
+#include "common/logging/log.h"
+#include "common/math_util.h"
+#include "common/param_package.h"
+#include "common/threadsafe_queue.h"
+#include "core/frontend/input.h"
+#include "input_common/sdl/sdl_impl.h"
+
+namespace InputCommon {
+
+namespace SDL {
+
+static std::string GetGUID(SDL_Joystick* joystick) {
+ SDL_JoystickGUID guid = SDL_JoystickGetGUID(joystick);
+ char guid_str[33];
+ SDL_JoystickGetGUIDString(guid, guid_str, sizeof(guid_str));
+ return guid_str;
+}
+
+/// Creates a ParamPackage from an SDL_Event that can directly be used to create a ButtonDevice
+static Common::ParamPackage SDLEventToButtonParamPackage(SDLState& state, const SDL_Event& event);
+
+static int SDLEventWatcher(void* userdata, SDL_Event* event) {
+ SDLState* sdl_state = reinterpret_cast<SDLState*>(userdata);
+ // Don't handle the event if we are configuring
+ if (sdl_state->polling) {
+ sdl_state->event_queue.Push(*event);
+ } else {
+ sdl_state->HandleGameControllerEvent(*event);
+ }
+ return 0;
+}
+
+class SDLJoystick {
+public:
+ SDLJoystick(std::string guid_, int port_, SDL_Joystick* joystick,
+ decltype(&SDL_JoystickClose) deleter = &SDL_JoystickClose)
+ : guid{std::move(guid_)}, port{port_}, sdl_joystick{joystick, deleter} {}
+
+ void SetButton(int button, bool value) {
+ std::lock_guard<std::mutex> lock(mutex);
+ state.buttons[button] = value;
+ }
+
+ bool GetButton(int button) const {
+ std::lock_guard<std::mutex> lock(mutex);
+ return state.buttons.at(button);
+ }
+
+ void SetAxis(int axis, Sint16 value) {
+ std::lock_guard<std::mutex> lock(mutex);
+ state.axes[axis] = value;
+ }
+
+ float GetAxis(int axis) const {
+ std::lock_guard<std::mutex> lock(mutex);
+ return state.axes.at(axis) / 32767.0f;
+ }
+
+ std::tuple<float, float> GetAnalog(int axis_x, int axis_y) const {
+ float x = GetAxis(axis_x);
+ float y = GetAxis(axis_y);
+ y = -y; // 3DS uses an y-axis inverse from SDL
+
+ // Make sure the coordinates are in the unit circle,
+ // otherwise normalize it.
+ float r = x * x + y * y;
+ if (r > 1.0f) {
+ r = std::sqrt(r);
+ x /= r;
+ y /= r;
+ }
+
+ return std::make_tuple(x, y);
+ }
+
+ void SetHat(int hat, Uint8 direction) {
+ std::lock_guard<std::mutex> lock(mutex);
+ state.hats[hat] = direction;
+ }
+
+ bool GetHatDirection(int hat, Uint8 direction) const {
+ std::lock_guard<std::mutex> lock(mutex);
+ return (state.hats.at(hat) & direction) != 0;
+ }
+ /**
+ * The guid of the joystick
+ */
+ const std::string& GetGUID() const {
+ return guid;
+ }
+
+ /**
+ * The number of joystick from the same type that were connected before this joystick
+ */
+ int GetPort() const {
+ return port;
+ }
+
+ SDL_Joystick* GetSDLJoystick() const {
+ return sdl_joystick.get();
+ }
+
+ void SetSDLJoystick(SDL_Joystick* joystick,
+ decltype(&SDL_JoystickClose) deleter = &SDL_JoystickClose) {
+ sdl_joystick =
+ std::unique_ptr<SDL_Joystick, decltype(&SDL_JoystickClose)>(joystick, deleter);
+ }
+
+private:
+ struct State {
+ std::unordered_map<int, bool> buttons;
+ std::unordered_map<int, Sint16> axes;
+ std::unordered_map<int, Uint8> hats;
+ } state;
+ std::string guid;
+ int port;
+ std::unique_ptr<SDL_Joystick, decltype(&SDL_JoystickClose)> sdl_joystick;
+ mutable std::mutex mutex;
+};
+
+/**
+ * Get the nth joystick with the corresponding GUID
+ */
+std::shared_ptr<SDLJoystick> SDLState::GetSDLJoystickByGUID(const std::string& guid, int port) {
+ std::lock_guard<std::mutex> lock(joystick_map_mutex);
+ const auto it = joystick_map.find(guid);
+ if (it != joystick_map.end()) {
+ while (it->second.size() <= port) {
+ auto joystick = std::make_shared<SDLJoystick>(guid, it->second.size(), nullptr,
+ [](SDL_Joystick*) {});
+ it->second.emplace_back(std::move(joystick));
+ }
+ return it->second[port];
+ }
+ auto joystick = std::make_shared<SDLJoystick>(guid, 0, nullptr, [](SDL_Joystick*) {});
+ return joystick_map[guid].emplace_back(std::move(joystick));
+}
+
+/**
+ * Check how many identical joysticks (by guid) were connected before the one with sdl_id and so tie
+ * it to a SDLJoystick with the same guid and that port
+ */
+std::shared_ptr<SDLJoystick> SDLState::GetSDLJoystickBySDLID(SDL_JoystickID sdl_id) {
+ auto sdl_joystick = SDL_JoystickFromInstanceID(sdl_id);
+ const std::string guid = GetGUID(sdl_joystick);
+ std::lock_guard<std::mutex> lock(joystick_map_mutex);
+ auto map_it = joystick_map.find(guid);
+ if (map_it != joystick_map.end()) {
+ auto vec_it = std::find_if(map_it->second.begin(), map_it->second.end(),
+ [&sdl_joystick](const std::shared_ptr<SDLJoystick>& joystick) {
+ return sdl_joystick == joystick->GetSDLJoystick();
+ });
+ if (vec_it != map_it->second.end()) {
+ // This is the common case: There is already an existing SDL_Joystick maped to a
+ // SDLJoystick. return the SDLJoystick
+ return *vec_it;
+ }
+ // Search for a SDLJoystick without a mapped SDL_Joystick...
+ auto nullptr_it = std::find_if(map_it->second.begin(), map_it->second.end(),
+ [](const std::shared_ptr<SDLJoystick>& joystick) {
+ return !joystick->GetSDLJoystick();
+ });
+ if (nullptr_it != map_it->second.end()) {
+ // ... and map it
+ (*nullptr_it)->SetSDLJoystick(sdl_joystick);
+ return *nullptr_it;
+ }
+ // There is no SDLJoystick without a mapped SDL_Joystick
+ // Create a new SDLJoystick
+ auto joystick = std::make_shared<SDLJoystick>(guid, map_it->second.size(), sdl_joystick);
+ return map_it->second.emplace_back(std::move(joystick));
+ }
+ auto joystick = std::make_shared<SDLJoystick>(guid, 0, sdl_joystick);
+ return joystick_map[guid].emplace_back(std::move(joystick));
+}
+
+void SDLState::InitJoystick(int joystick_index) {
+ SDL_Joystick* sdl_joystick = SDL_JoystickOpen(joystick_index);
+ if (!sdl_joystick) {
+ LOG_ERROR(Input, "failed to open joystick {}", joystick_index);
+ return;
+ }
+ std::string guid = GetGUID(sdl_joystick);
+ std::lock_guard<std::mutex> lock(joystick_map_mutex);
+ if (joystick_map.find(guid) == joystick_map.end()) {
+ auto joystick = std::make_shared<SDLJoystick>(guid, 0, sdl_joystick);
+ joystick_map[guid].emplace_back(std::move(joystick));
+ return;
+ }
+ auto& joystick_guid_list = joystick_map[guid];
+ const auto it = std::find_if(
+ joystick_guid_list.begin(), joystick_guid_list.end(),
+ [](const std::shared_ptr<SDLJoystick>& joystick) { return !joystick->GetSDLJoystick(); });
+ if (it != joystick_guid_list.end()) {
+ (*it)->SetSDLJoystick(sdl_joystick);
+ return;
+ }
+ auto joystick = std::make_shared<SDLJoystick>(guid, joystick_guid_list.size(), sdl_joystick);
+ joystick_guid_list.emplace_back(std::move(joystick));
+}
+
+void SDLState::CloseJoystick(SDL_Joystick* sdl_joystick) {
+ std::string guid = GetGUID(sdl_joystick);
+ std::shared_ptr<SDLJoystick> joystick;
+ {
+ std::lock_guard<std::mutex> lock(joystick_map_mutex);
+ // This call to guid is safe since the joystick is guaranteed to be in the map
+ auto& joystick_guid_list = joystick_map[guid];
+ const auto joystick_it =
+ std::find_if(joystick_guid_list.begin(), joystick_guid_list.end(),
+ [&sdl_joystick](const std::shared_ptr<SDLJoystick>& joystick) {
+ return joystick->GetSDLJoystick() == sdl_joystick;
+ });
+ joystick = *joystick_it;
+ }
+ // Destruct SDL_Joystick outside the lock guard because SDL can internally call event calback
+ // which locks the mutex again
+ joystick->SetSDLJoystick(nullptr, [](SDL_Joystick*) {});
+}
+
+void SDLState::HandleGameControllerEvent(const SDL_Event& event) {
+ switch (event.type) {
+ case SDL_JOYBUTTONUP: {
+ if (auto joystick = GetSDLJoystickBySDLID(event.jbutton.which)) {
+ joystick->SetButton(event.jbutton.button, false);
+ }
+ break;
+ }
+ case SDL_JOYBUTTONDOWN: {
+ if (auto joystick = GetSDLJoystickBySDLID(event.jbutton.which)) {
+ joystick->SetButton(event.jbutton.button, true);
+ }
+ break;
+ }
+ case SDL_JOYHATMOTION: {
+ if (auto joystick = GetSDLJoystickBySDLID(event.jhat.which)) {
+ joystick->SetHat(event.jhat.hat, event.jhat.value);
+ }
+ break;
+ }
+ case SDL_JOYAXISMOTION: {
+ if (auto joystick = GetSDLJoystickBySDLID(event.jaxis.which)) {
+ joystick->SetAxis(event.jaxis.axis, event.jaxis.value);
+ }
+ break;
+ }
+ case SDL_JOYDEVICEREMOVED:
+ LOG_DEBUG(Input, "Controller removed with Instance_ID {}", event.jdevice.which);
+ CloseJoystick(SDL_JoystickFromInstanceID(event.jdevice.which));
+ break;
+ case SDL_JOYDEVICEADDED:
+ LOG_DEBUG(Input, "Controller connected with device index {}", event.jdevice.which);
+ InitJoystick(event.jdevice.which);
+ break;
+ }
+}
+
+void SDLState::CloseJoysticks() {
+ std::lock_guard<std::mutex> lock(joystick_map_mutex);
+ joystick_map.clear();
+}
+
+class SDLButton final : public Input::ButtonDevice {
+public:
+ explicit SDLButton(std::shared_ptr<SDLJoystick> joystick_, int button_)
+ : joystick(std::move(joystick_)), button(button_) {}
+
+ bool GetStatus() const override {
+ return joystick->GetButton(button);
+ }
+
+private:
+ std::shared_ptr<SDLJoystick> joystick;
+ int button;
+};
+
+class SDLDirectionButton final : public Input::ButtonDevice {
+public:
+ explicit SDLDirectionButton(std::shared_ptr<SDLJoystick> joystick_, int hat_, Uint8 direction_)
+ : joystick(std::move(joystick_)), hat(hat_), direction(direction_) {}
+
+ bool GetStatus() const override {
+ return joystick->GetHatDirection(hat, direction);
+ }
+
+private:
+ std::shared_ptr<SDLJoystick> joystick;
+ int hat;
+ Uint8 direction;
+};
+
+class SDLAxisButton final : public Input::ButtonDevice {
+public:
+ explicit SDLAxisButton(std::shared_ptr<SDLJoystick> joystick_, int axis_, float threshold_,
+ bool trigger_if_greater_)
+ : joystick(std::move(joystick_)), axis(axis_), threshold(threshold_),
+ trigger_if_greater(trigger_if_greater_) {}
+
+ bool GetStatus() const override {
+ float axis_value = joystick->GetAxis(axis);
+ if (trigger_if_greater)
+ return axis_value > threshold;
+ return axis_value < threshold;
+ }
+
+private:
+ std::shared_ptr<SDLJoystick> joystick;
+ int axis;
+ float threshold;
+ bool trigger_if_greater;
+};
+
+class SDLAnalog final : public Input::AnalogDevice {
+public:
+ SDLAnalog(std::shared_ptr<SDLJoystick> joystick_, int axis_x_, int axis_y_, float deadzone_)
+ : joystick(std::move(joystick_)), axis_x(axis_x_), axis_y(axis_y_), deadzone(deadzone_) {}
+
+ std::tuple<float, float> GetStatus() const override {
+ const auto [x, y] = joystick->GetAnalog(axis_x, axis_y);
+ const float r = std::sqrt((x * x) + (y * y));
+ if (r > deadzone) {
+ return std::make_tuple(x / r * (r - deadzone) / (1 - deadzone),
+ y / r * (r - deadzone) / (1 - deadzone));
+ }
+ return std::make_tuple<float, float>(0.0f, 0.0f);
+ }
+
+private:
+ std::shared_ptr<SDLJoystick> joystick;
+ const int axis_x;
+ const int axis_y;
+ const float deadzone;
+};
+
+/// A button device factory that creates button devices from SDL joystick
+class SDLButtonFactory final : public Input::Factory<Input::ButtonDevice> {
+public:
+ explicit SDLButtonFactory(SDLState& state_) : state(state_) {}
+
+ /**
+ * Creates a button device from a joystick button
+ * @param params contains parameters for creating the device:
+ * - "guid": the guid of the joystick to bind
+ * - "port": the nth joystick of the same type to bind
+ * - "button"(optional): the index of the button to bind
+ * - "hat"(optional): the index of the hat to bind as direction buttons
+ * - "axis"(optional): the index of the axis to bind
+ * - "direction"(only used for hat): the direction name of the hat to bind. Can be "up",
+ * "down", "left" or "right"
+ * - "threshold"(only used for axis): a float value in (-1.0, 1.0) which the button is
+ * triggered if the axis value crosses
+ * - "direction"(only used for axis): "+" means the button is triggered when the axis
+ * value is greater than the threshold; "-" means the button is triggered when the axis
+ * value is smaller than the threshold
+ */
+ std::unique_ptr<Input::ButtonDevice> Create(const Common::ParamPackage& params) override {
+ const std::string guid = params.Get("guid", "0");
+ const int port = params.Get("port", 0);
+
+ auto joystick = state.GetSDLJoystickByGUID(guid, port);
+
+ if (params.Has("hat")) {
+ const int hat = params.Get("hat", 0);
+ const std::string direction_name = params.Get("direction", "");
+ Uint8 direction;
+ if (direction_name == "up") {
+ direction = SDL_HAT_UP;
+ } else if (direction_name == "down") {
+ direction = SDL_HAT_DOWN;
+ } else if (direction_name == "left") {
+ direction = SDL_HAT_LEFT;
+ } else if (direction_name == "right") {
+ direction = SDL_HAT_RIGHT;
+ } else {
+ direction = 0;
+ }
+ // This is necessary so accessing GetHat with hat won't crash
+ joystick->SetHat(hat, SDL_HAT_CENTERED);
+ return std::make_unique<SDLDirectionButton>(joystick, hat, direction);
+ }
+
+ if (params.Has("axis")) {
+ const int axis = params.Get("axis", 0);
+ const float threshold = params.Get("threshold", 0.5f);
+ const std::string direction_name = params.Get("direction", "");
+ bool trigger_if_greater;
+ if (direction_name == "+") {
+ trigger_if_greater = true;
+ } else if (direction_name == "-") {
+ trigger_if_greater = false;
+ } else {
+ trigger_if_greater = true;
+ LOG_ERROR(Input, "Unknown direction {}", direction_name);
+ }
+ // This is necessary so accessing GetAxis with axis won't crash
+ joystick->SetAxis(axis, 0);
+ return std::make_unique<SDLAxisButton>(joystick, axis, threshold, trigger_if_greater);
+ }
+
+ const int button = params.Get("button", 0);
+ // This is necessary so accessing GetButton with button won't crash
+ joystick->SetButton(button, false);
+ return std::make_unique<SDLButton>(joystick, button);
+ }
+
+private:
+ SDLState& state;
+};
+
+/// An analog device factory that creates analog devices from SDL joystick
+class SDLAnalogFactory final : public Input::Factory<Input::AnalogDevice> {
+public:
+ explicit SDLAnalogFactory(SDLState& state_) : state(state_) {}
+ /**
+ * Creates analog device from joystick axes
+ * @param params contains parameters for creating the device:
+ * - "guid": the guid of the joystick to bind
+ * - "port": the nth joystick of the same type
+ * - "axis_x": the index of the axis to be bind as x-axis
+ * - "axis_y": the index of the axis to be bind as y-axis
+ */
+ std::unique_ptr<Input::AnalogDevice> Create(const Common::ParamPackage& params) override {
+ const std::string guid = params.Get("guid", "0");
+ const int port = params.Get("port", 0);
+ const int axis_x = params.Get("axis_x", 0);
+ const int axis_y = params.Get("axis_y", 1);
+ float deadzone = std::clamp(params.Get("deadzone", 0.0f), 0.0f, .99f);
+
+ auto joystick = state.GetSDLJoystickByGUID(guid, port);
+
+ // This is necessary so accessing GetAxis with axis_x and axis_y won't crash
+ joystick->SetAxis(axis_x, 0);
+ joystick->SetAxis(axis_y, 0);
+ return std::make_unique<SDLAnalog>(joystick, axis_x, axis_y, deadzone);
+ }
+
+private:
+ SDLState& state;
+};
+
+SDLState::SDLState() {
+ using namespace Input;
+ RegisterFactory<ButtonDevice>("sdl", std::make_shared<SDLButtonFactory>(*this));
+ RegisterFactory<AnalogDevice>("sdl", std::make_shared<SDLAnalogFactory>(*this));
+
+ // If the frontend is going to manage the event loop, then we dont start one here
+ start_thread = !SDL_WasInit(SDL_INIT_JOYSTICK);
+ if (start_thread && SDL_Init(SDL_INIT_JOYSTICK) < 0) {
+ LOG_CRITICAL(Input, "SDL_Init(SDL_INIT_JOYSTICK) failed with: {}", SDL_GetError());
+ return;
+ }
+ if (SDL_SetHint(SDL_HINT_JOYSTICK_ALLOW_BACKGROUND_EVENTS, "1") == SDL_FALSE) {
+ LOG_ERROR(Input, "Failed to set Hint for background events", SDL_GetError());
+ }
+
+ SDL_AddEventWatch(&SDLEventWatcher, this);
+
+ initialized = true;
+ if (start_thread) {
+ poll_thread = std::thread([&] {
+ using namespace std::chrono_literals;
+ SDL_Event event;
+ while (initialized) {
+ SDL_PumpEvents();
+ std::this_thread::sleep_for(std::chrono::duration(10ms));
+ }
+ });
+ }
+ // Because the events for joystick connection happens before we have our event watcher added, we
+ // can just open all the joysticks right here
+ for (int i = 0; i < SDL_NumJoysticks(); ++i) {
+ InitJoystick(i);
+ }
+}
+
+SDLState::~SDLState() {
+ using namespace Input;
+ UnregisterFactory<ButtonDevice>("sdl");
+ UnregisterFactory<AnalogDevice>("sdl");
+
+ CloseJoysticks();
+ SDL_DelEventWatch(&SDLEventWatcher, this);
+
+ initialized = false;
+ if (start_thread) {
+ poll_thread.join();
+ SDL_QuitSubSystem(SDL_INIT_JOYSTICK);
+ }
+}
+
+Common::ParamPackage SDLEventToButtonParamPackage(SDLState& state, const SDL_Event& event) {
+ Common::ParamPackage params({{"engine", "sdl"}});
+
+ switch (event.type) {
+ case SDL_JOYAXISMOTION: {
+ auto joystick = state.GetSDLJoystickBySDLID(event.jaxis.which);
+ params.Set("port", joystick->GetPort());
+ params.Set("guid", joystick->GetGUID());
+ params.Set("axis", event.jaxis.axis);
+ if (event.jaxis.value > 0) {
+ params.Set("direction", "+");
+ params.Set("threshold", "0.5");
+ } else {
+ params.Set("direction", "-");
+ params.Set("threshold", "-0.5");
+ }
+ break;
+ }
+ case SDL_JOYBUTTONUP: {
+ auto joystick = state.GetSDLJoystickBySDLID(event.jbutton.which);
+ params.Set("port", joystick->GetPort());
+ params.Set("guid", joystick->GetGUID());
+ params.Set("button", event.jbutton.button);
+ break;
+ }
+ case SDL_JOYHATMOTION: {
+ auto joystick = state.GetSDLJoystickBySDLID(event.jhat.which);
+ params.Set("port", joystick->GetPort());
+ params.Set("guid", joystick->GetGUID());
+ params.Set("hat", event.jhat.hat);
+ switch (event.jhat.value) {
+ case SDL_HAT_UP:
+ params.Set("direction", "up");
+ break;
+ case SDL_HAT_DOWN:
+ params.Set("direction", "down");
+ break;
+ case SDL_HAT_LEFT:
+ params.Set("direction", "left");
+ break;
+ case SDL_HAT_RIGHT:
+ params.Set("direction", "right");
+ break;
+ default:
+ return {};
+ }
+ break;
+ }
+ }
+ return params;
+}
+
+namespace Polling {
+
+class SDLPoller : public InputCommon::Polling::DevicePoller {
+public:
+ explicit SDLPoller(SDLState& state_) : state(state_) {}
+
+ void Start() override {
+ state.event_queue.Clear();
+ state.polling = true;
+ }
+
+ void Stop() override {
+ state.polling = false;
+ }
+
+protected:
+ SDLState& state;
+};
+
+class SDLButtonPoller final : public SDLPoller {
+public:
+ explicit SDLButtonPoller(SDLState& state_) : SDLPoller(state_) {}
+
+ Common::ParamPackage GetNextInput() override {
+ SDL_Event event;
+ while (state.event_queue.Pop(event)) {
+ switch (event.type) {
+ case SDL_JOYAXISMOTION:
+ if (std::abs(event.jaxis.value / 32767.0) < 0.5) {
+ break;
+ }
+ case SDL_JOYBUTTONUP:
+ case SDL_JOYHATMOTION:
+ return SDLEventToButtonParamPackage(state, event);
+ }
+ }
+ return {};
+ }
+};
+
+class SDLAnalogPoller final : public SDLPoller {
+public:
+ explicit SDLAnalogPoller(SDLState& state_) : SDLPoller(state_) {}
+
+ void Start() override {
+ SDLPoller::Start();
+
+ // Reset stored axes
+ analog_xaxis = -1;
+ analog_yaxis = -1;
+ analog_axes_joystick = -1;
+ }
+
+ Common::ParamPackage GetNextInput() override {
+ SDL_Event event;
+ while (state.event_queue.Pop(event)) {
+ if (event.type != SDL_JOYAXISMOTION || std::abs(event.jaxis.value / 32767.0) < 0.5) {
+ continue;
+ }
+ // An analog device needs two axes, so we need to store the axis for later and wait for
+ // a second SDL event. The axes also must be from the same joystick.
+ int axis = event.jaxis.axis;
+ if (analog_xaxis == -1) {
+ analog_xaxis = axis;
+ analog_axes_joystick = event.jaxis.which;
+ } else if (analog_yaxis == -1 && analog_xaxis != axis &&
+ analog_axes_joystick == event.jaxis.which) {
+ analog_yaxis = axis;
+ }
+ }
+ Common::ParamPackage params;
+ if (analog_xaxis != -1 && analog_yaxis != -1) {
+ auto joystick = state.GetSDLJoystickBySDLID(event.jaxis.which);
+ params.Set("engine", "sdl");
+ params.Set("port", joystick->GetPort());
+ params.Set("guid", joystick->GetGUID());
+ params.Set("axis_x", analog_xaxis);
+ params.Set("axis_y", analog_yaxis);
+ analog_xaxis = -1;
+ analog_yaxis = -1;
+ analog_axes_joystick = -1;
+ return params;
+ }
+ return params;
+ }
+
+private:
+ int analog_xaxis = -1;
+ int analog_yaxis = -1;
+ SDL_JoystickID analog_axes_joystick = -1;
+};
+} // namespace Polling
+
+std::vector<std::unique_ptr<InputCommon::Polling::DevicePoller>> SDLState::GetPollers(
+ InputCommon::Polling::DeviceType type) {
+ std::vector<std::unique_ptr<InputCommon::Polling::DevicePoller>> pollers;
+ switch (type) {
+ case InputCommon::Polling::DeviceType::Analog:
+ pollers.emplace_back(std::make_unique<Polling::SDLAnalogPoller>(*this));
+ break;
+ case InputCommon::Polling::DeviceType::Button:
+ pollers.emplace_back(std::make_unique<Polling::SDLButtonPoller>(*this));
+ break;
+ return pollers;
+ }
+}
+
+} // namespace SDL
+} // namespace InputCommon
diff --git a/src/input_common/sdl/sdl_impl.h b/src/input_common/sdl/sdl_impl.h
new file mode 100644
index 000000000..fec82fbe6
--- /dev/null
+++ b/src/input_common/sdl/sdl_impl.h
@@ -0,0 +1,64 @@
+// Copyright 2018 Citra Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <atomic>
+#include <memory>
+#include <thread>
+#include "common/threadsafe_queue.h"
+#include "input_common/sdl/sdl.h"
+
+union SDL_Event;
+using SDL_Joystick = struct _SDL_Joystick;
+using SDL_JoystickID = s32;
+
+namespace InputCommon::SDL {
+
+class SDLJoystick;
+class SDLButtonFactory;
+class SDLAnalogFactory;
+
+class SDLState : public State {
+public:
+ /// Initializes and registers SDL device factories
+ SDLState();
+
+ /// Unresisters SDL device factories and shut them down.
+ ~SDLState() override;
+
+ /// Handle SDL_Events for joysticks from SDL_PollEvent
+ void HandleGameControllerEvent(const SDL_Event& event);
+
+ std::shared_ptr<SDLJoystick> GetSDLJoystickBySDLID(SDL_JoystickID sdl_id);
+ std::shared_ptr<SDLJoystick> GetSDLJoystickByGUID(const std::string& guid, int port);
+
+ /// Get all DevicePoller that use the SDL backend for a specific device type
+ std::vector<std::unique_ptr<InputCommon::Polling::DevicePoller>> GetPollers(
+ InputCommon::Polling::DeviceType type) override;
+
+ /// Used by the Pollers during config
+ std::atomic<bool> polling = false;
+ Common::SPSCQueue<SDL_Event> event_queue;
+
+private:
+ void InitJoystick(int joystick_index);
+ void CloseJoystick(SDL_Joystick* sdl_joystick);
+
+ /// Needs to be called before SDL_QuitSubSystem.
+ void CloseJoysticks();
+
+ /// Map of GUID of a list of corresponding virtual Joysticks
+ std::unordered_map<std::string, std::vector<std::shared_ptr<SDLJoystick>>> joystick_map;
+ std::mutex joystick_map_mutex;
+
+ std::shared_ptr<SDLButtonFactory> button_factory;
+ std::shared_ptr<SDLAnalogFactory> analog_factory;
+
+ bool start_thread = false;
+ std::atomic<bool> initialized = false;
+
+ std::thread poll_thread;
+};
+} // namespace InputCommon::SDL
diff --git a/src/tests/core/arm/arm_test_common.cpp b/src/tests/core/arm/arm_test_common.cpp
index 6fe56833d..3e1a735c3 100644
--- a/src/tests/core/arm/arm_test_common.cpp
+++ b/src/tests/core/arm/arm_test_common.cpp
@@ -4,6 +4,7 @@
#include <algorithm>
+#include "common/page_table.h"
#include "core/core.h"
#include "core/hle/kernel/process.h"
#include "core/memory.h"
@@ -22,7 +23,7 @@ TestEnvironment::TestEnvironment(bool mutable_memory_)
std::fill(page_table->pointers.begin(), page_table->pointers.end(), nullptr);
page_table->special_regions.clear();
std::fill(page_table->attributes.begin(), page_table->attributes.end(),
- Memory::PageType::Unmapped);
+ Common::PageType::Unmapped);
Memory::MapIoRegion(*page_table, 0x00000000, 0x80000000, test_memory);
Memory::MapIoRegion(*page_table, 0x80000000, 0x80000000, test_memory);
diff --git a/src/tests/core/arm/arm_test_common.h b/src/tests/core/arm/arm_test_common.h
index 0b7539601..d145dbfcc 100644
--- a/src/tests/core/arm/arm_test_common.h
+++ b/src/tests/core/arm/arm_test_common.h
@@ -9,10 +9,10 @@
#include <vector>
#include "common/common_types.h"
+#include "common/memory_hook.h"
#include "core/hle/kernel/kernel.h"
-#include "core/memory_hook.h"
-namespace Memory {
+namespace Common {
struct PageTable;
}
@@ -58,7 +58,7 @@ public:
private:
friend struct TestMemory;
- struct TestMemory final : Memory::MemoryHook {
+ struct TestMemory final : Common::MemoryHook {
explicit TestMemory(TestEnvironment* env_) : env(env_) {}
TestEnvironment* env;
@@ -86,7 +86,7 @@ private:
bool mutable_memory;
std::shared_ptr<TestMemory> test_memory;
std::vector<WriteRecord> write_records;
- Memory::PageTable* page_table = nullptr;
+ Common::PageTable* page_table = nullptr;
Kernel::KernelCore kernel;
};
diff --git a/src/video_core/CMakeLists.txt b/src/video_core/CMakeLists.txt
index 0c3038c52..14b76680f 100644
--- a/src/video_core/CMakeLists.txt
+++ b/src/video_core/CMakeLists.txt
@@ -123,6 +123,8 @@ if (ENABLE_VULKAN)
renderer_vulkan/vk_memory_manager.h
renderer_vulkan/vk_resource_manager.cpp
renderer_vulkan/vk_resource_manager.h
+ renderer_vulkan/vk_sampler_cache.cpp
+ renderer_vulkan/vk_sampler_cache.h
renderer_vulkan/vk_scheduler.cpp
renderer_vulkan/vk_scheduler.h
renderer_vulkan/vk_stream_buffer.cpp
diff --git a/src/video_core/dma_pusher.cpp b/src/video_core/dma_pusher.cpp
index bff1a37ff..8b1bea1ae 100644
--- a/src/video_core/dma_pusher.cpp
+++ b/src/video_core/dma_pusher.cpp
@@ -55,12 +55,9 @@ bool DmaPusher::Step() {
}
// Push buffer non-empty, read a word
- const auto address = gpu.MemoryManager().GpuToCpuAddress(dma_get);
- ASSERT_MSG(address, "Invalid GPU address");
-
command_headers.resize(command_list_header.size);
-
- Memory::ReadBlock(*address, command_headers.data(), command_list_header.size * sizeof(u32));
+ gpu.MemoryManager().ReadBlock(dma_get, command_headers.data(),
+ command_list_header.size * sizeof(u32));
for (const CommandHeader& command_header : command_headers) {
diff --git a/src/video_core/engines/kepler_memory.cpp b/src/video_core/engines/kepler_memory.cpp
index aae2a4019..0931b9626 100644
--- a/src/video_core/engines/kepler_memory.cpp
+++ b/src/video_core/engines/kepler_memory.cpp
@@ -9,6 +9,7 @@
#include "video_core/engines/kepler_memory.h"
#include "video_core/engines/maxwell_3d.h"
#include "video_core/rasterizer_interface.h"
+#include "video_core/renderer_base.h"
namespace Tegra::Engines {
@@ -40,17 +41,13 @@ void KeplerMemory::ProcessData(u32 data) {
ASSERT_MSG(regs.exec.linear, "Non-linear uploads are not supported");
ASSERT(regs.dest.x == 0 && regs.dest.y == 0 && regs.dest.z == 0);
- const GPUVAddr address = regs.dest.Address();
- const auto dest_address =
- memory_manager.GpuToCpuAddress(address + state.write_offset * sizeof(u32));
- ASSERT_MSG(dest_address, "Invalid GPU address");
-
// We have to invalidate the destination region to evict any outdated surfaces from the cache.
- // We do this before actually writing the new data because the destination address might contain
- // a dirty surface that will have to be written back to memory.
- Core::System::GetInstance().GPU().InvalidateRegion(*dest_address, sizeof(u32));
+ // We do this before actually writing the new data because the destination address might
+ // contain a dirty surface that will have to be written back to memory.
+ const GPUVAddr address{regs.dest.Address() + state.write_offset * sizeof(u32)};
+ rasterizer.InvalidateRegion(ToCacheAddr(memory_manager.GetPointer(address)), sizeof(u32));
+ memory_manager.Write32(address, data);
- Memory::Write32(*dest_address, data);
system.GPU().Maxwell3D().dirty_flags.OnMemoryWrite();
state.write_offset++;
diff --git a/src/video_core/engines/maxwell_3d.cpp b/src/video_core/engines/maxwell_3d.cpp
index 144e7fa82..c5d5be4ef 100644
--- a/src/video_core/engines/maxwell_3d.cpp
+++ b/src/video_core/engines/maxwell_3d.cpp
@@ -270,11 +270,9 @@ void Maxwell3D::ProcessMacroBind(u32 data) {
}
void Maxwell3D::ProcessQueryGet() {
- GPUVAddr sequence_address = regs.query.QueryAddress();
+ const GPUVAddr sequence_address{regs.query.QueryAddress()};
// Since the sequence address is given as a GPU VAddr, we have to convert it to an application
// VAddr before writing.
- const auto address = memory_manager.GpuToCpuAddress(sequence_address);
- ASSERT_MSG(address, "Invalid GPU address");
// TODO(Subv): Support the other query units.
ASSERT_MSG(regs.query.query_get.unit == Regs::QueryUnit::Crop,
@@ -309,7 +307,7 @@ void Maxwell3D::ProcessQueryGet() {
// Write the current query sequence to the sequence address.
// TODO(Subv): Find out what happens if you use a long query type but mark it as a short
// query.
- Memory::Write32(*address, sequence);
+ memory_manager.Write32(sequence_address, sequence);
} else {
// Write the 128-bit result structure in long mode. Note: We emulate an infinitely fast
// GPU, this command may actually take a while to complete in real hardware due to GPU
@@ -318,7 +316,7 @@ void Maxwell3D::ProcessQueryGet() {
query_result.value = result;
// TODO(Subv): Generate a real GPU timestamp and write it here instead of CoreTiming
query_result.timestamp = system.CoreTiming().GetTicks();
- Memory::WriteBlock(*address, &query_result, sizeof(query_result));
+ memory_manager.WriteBlock(sequence_address, &query_result, sizeof(query_result));
}
dirty_flags.OnMemoryWrite();
break;
@@ -393,10 +391,12 @@ void Maxwell3D::ProcessCBData(u32 value) {
// Don't allow writing past the end of the buffer.
ASSERT(regs.const_buffer.cb_pos + sizeof(u32) <= regs.const_buffer.cb_size);
- const auto address = memory_manager.GpuToCpuAddress(buffer_address + regs.const_buffer.cb_pos);
- ASSERT_MSG(address, "Invalid GPU address");
+ const GPUVAddr address{buffer_address + regs.const_buffer.cb_pos};
+
+ u8* ptr{memory_manager.GetPointer(address)};
+ rasterizer.InvalidateRegion(ToCacheAddr(ptr), sizeof(u32));
+ memory_manager.Write32(address, value);
- Memory::Write32(*address, value);
dirty_flags.OnMemoryWrite();
// Increment the current buffer position.
@@ -404,14 +404,10 @@ void Maxwell3D::ProcessCBData(u32 value) {
}
Texture::TICEntry Maxwell3D::GetTICEntry(u32 tic_index) const {
- const GPUVAddr tic_base_address = regs.tic.TICAddress();
-
- const GPUVAddr tic_address_gpu = tic_base_address + tic_index * sizeof(Texture::TICEntry);
- const auto tic_address_cpu = memory_manager.GpuToCpuAddress(tic_address_gpu);
- ASSERT_MSG(tic_address_cpu, "Invalid GPU address");
+ const GPUVAddr tic_address_gpu{regs.tic.TICAddress() + tic_index * sizeof(Texture::TICEntry)};
Texture::TICEntry tic_entry;
- Memory::ReadBlock(*tic_address_cpu, &tic_entry, sizeof(Texture::TICEntry));
+ memory_manager.ReadBlock(tic_address_gpu, &tic_entry, sizeof(Texture::TICEntry));
ASSERT_MSG(tic_entry.header_version == Texture::TICHeaderVersion::BlockLinear ||
tic_entry.header_version == Texture::TICHeaderVersion::Pitch,
@@ -429,14 +425,10 @@ Texture::TICEntry Maxwell3D::GetTICEntry(u32 tic_index) const {
}
Texture::TSCEntry Maxwell3D::GetTSCEntry(u32 tsc_index) const {
- const GPUVAddr tsc_base_address = regs.tsc.TSCAddress();
-
- const GPUVAddr tsc_address_gpu = tsc_base_address + tsc_index * sizeof(Texture::TSCEntry);
- const auto tsc_address_cpu = memory_manager.GpuToCpuAddress(tsc_address_gpu);
- ASSERT_MSG(tsc_address_cpu, "Invalid GPU address");
+ const GPUVAddr tsc_address_gpu{regs.tsc.TSCAddress() + tsc_index * sizeof(Texture::TSCEntry)};
Texture::TSCEntry tsc_entry;
- Memory::ReadBlock(*tsc_address_cpu, &tsc_entry, sizeof(Texture::TSCEntry));
+ memory_manager.ReadBlock(tsc_address_gpu, &tsc_entry, sizeof(Texture::TSCEntry));
return tsc_entry;
}
@@ -455,10 +447,7 @@ std::vector<Texture::FullTextureInfo> Maxwell3D::GetStageTextures(Regs::ShaderSt
for (GPUVAddr current_texture = tex_info_buffer.address + TextureInfoOffset;
current_texture < tex_info_buffer_end; current_texture += sizeof(Texture::TextureHandle)) {
- const auto address = memory_manager.GpuToCpuAddress(current_texture);
- ASSERT_MSG(address, "Invalid GPU address");
-
- const Texture::TextureHandle tex_handle{Memory::Read32(*address)};
+ const Texture::TextureHandle tex_handle{memory_manager.Read32(current_texture)};
Texture::FullTextureInfo tex_info{};
// TODO(Subv): Use the shader to determine which textures are actually accessed.
@@ -493,10 +482,7 @@ Texture::FullTextureInfo Maxwell3D::GetStageTexture(Regs::ShaderStage stage,
ASSERT(tex_info_address < tex_info_buffer.address + tex_info_buffer.size);
- const auto tex_address_cpu = memory_manager.GpuToCpuAddress(tex_info_address);
- ASSERT_MSG(tex_address_cpu, "Invalid GPU address");
-
- const Texture::TextureHandle tex_handle{Memory::Read32(*tex_address_cpu)};
+ const Texture::TextureHandle tex_handle{memory_manager.Read32(tex_info_address)};
Texture::FullTextureInfo tex_info{};
tex_info.index = static_cast<u32>(offset);
diff --git a/src/video_core/engines/maxwell_dma.cpp b/src/video_core/engines/maxwell_dma.cpp
index 9dfea5999..a0ded4c25 100644
--- a/src/video_core/engines/maxwell_dma.cpp
+++ b/src/video_core/engines/maxwell_dma.cpp
@@ -9,6 +9,7 @@
#include "video_core/engines/maxwell_3d.h"
#include "video_core/engines/maxwell_dma.h"
#include "video_core/rasterizer_interface.h"
+#include "video_core/renderer_base.h"
#include "video_core/textures/decoders.h"
namespace Tegra::Engines {
@@ -42,11 +43,6 @@ void MaxwellDMA::HandleCopy() {
const GPUVAddr source = regs.src_address.Address();
const GPUVAddr dest = regs.dst_address.Address();
- const auto source_cpu = memory_manager.GpuToCpuAddress(source);
- const auto dest_cpu = memory_manager.GpuToCpuAddress(dest);
- ASSERT_MSG(source_cpu, "Invalid source GPU address");
- ASSERT_MSG(dest_cpu, "Invalid destination GPU address");
-
// TODO(Subv): Perform more research and implement all features of this engine.
ASSERT(regs.exec.enable_swizzle == 0);
ASSERT(regs.exec.query_mode == Regs::QueryMode::None);
@@ -69,7 +65,7 @@ void MaxwellDMA::HandleCopy() {
// buffer of length `x_count`, otherwise we copy a 2D image of dimensions (x_count,
// y_count).
if (!regs.exec.enable_2d) {
- Memory::CopyBlock(*dest_cpu, *source_cpu, regs.x_count);
+ memory_manager.CopyBlock(dest, source, regs.x_count);
return;
}
@@ -78,9 +74,9 @@ void MaxwellDMA::HandleCopy() {
// rectangle. There is no need to manually flush/invalidate the regions because
// CopyBlock does that for us.
for (u32 line = 0; line < regs.y_count; ++line) {
- const VAddr source_line = *source_cpu + line * regs.src_pitch;
- const VAddr dest_line = *dest_cpu + line * regs.dst_pitch;
- Memory::CopyBlock(dest_line, source_line, regs.x_count);
+ const GPUVAddr source_line = source + line * regs.src_pitch;
+ const GPUVAddr dest_line = dest + line * regs.dst_pitch;
+ memory_manager.CopyBlock(dest_line, source_line, regs.x_count);
}
return;
}
@@ -89,15 +85,18 @@ void MaxwellDMA::HandleCopy() {
const std::size_t copy_size = regs.x_count * regs.y_count;
+ auto source_ptr{memory_manager.GetPointer(source)};
+ auto dst_ptr{memory_manager.GetPointer(dest)};
+
const auto FlushAndInvalidate = [&](u32 src_size, u64 dst_size) {
// TODO(Subv): For now, manually flush the regions until we implement GPU-accelerated
// copying.
- Core::System::GetInstance().GPU().FlushRegion(*source_cpu, src_size);
+ rasterizer.FlushRegion(ToCacheAddr(source_ptr), src_size);
// We have to invalidate the destination region to evict any outdated surfaces from the
// cache. We do this before actually writing the new data because the destination address
// might contain a dirty surface that will have to be written back to memory.
- Core::System::GetInstance().GPU().InvalidateRegion(*dest_cpu, dst_size);
+ rasterizer.InvalidateRegion(ToCacheAddr(dst_ptr), dst_size);
};
if (regs.exec.is_dst_linear && !regs.exec.is_src_linear) {
@@ -110,8 +109,8 @@ void MaxwellDMA::HandleCopy() {
copy_size * src_bytes_per_pixel);
Texture::UnswizzleSubrect(regs.x_count, regs.y_count, regs.dst_pitch,
- regs.src_params.size_x, src_bytes_per_pixel, *source_cpu,
- *dest_cpu, regs.src_params.BlockHeight(), regs.src_params.pos_x,
+ regs.src_params.size_x, src_bytes_per_pixel, source_ptr, dst_ptr,
+ regs.src_params.BlockHeight(), regs.src_params.pos_x,
regs.src_params.pos_y);
} else {
ASSERT(regs.dst_params.size_z == 1);
@@ -124,7 +123,7 @@ void MaxwellDMA::HandleCopy() {
// If the input is linear and the output is tiled, swizzle the input and copy it over.
Texture::SwizzleSubrect(regs.x_count, regs.y_count, regs.src_pitch, regs.dst_params.size_x,
- src_bpp, *dest_cpu, *source_cpu, regs.dst_params.BlockHeight());
+ src_bpp, dst_ptr, source_ptr, regs.dst_params.BlockHeight());
}
}
diff --git a/src/video_core/gpu.cpp b/src/video_core/gpu.cpp
index 08abf8ac9..66c690494 100644
--- a/src/video_core/gpu.cpp
+++ b/src/video_core/gpu.cpp
@@ -274,7 +274,6 @@ void GPU::ProcessSemaphoreTriggerMethod() {
const auto op =
static_cast<GpuSemaphoreOperation>(regs.semaphore_trigger & semaphoreOperationMask);
if (op == GpuSemaphoreOperation::WriteLong) {
- auto address = memory_manager->GpuToCpuAddress(regs.smaphore_address.SmaphoreAddress());
struct Block {
u32 sequence;
u32 zeros = 0;
@@ -286,11 +285,9 @@ void GPU::ProcessSemaphoreTriggerMethod() {
// TODO(Kmather73): Generate a real GPU timestamp and write it here instead of
// CoreTiming
block.timestamp = Core::System::GetInstance().CoreTiming().GetTicks();
- Memory::WriteBlock(*address, &block, sizeof(block));
+ memory_manager->WriteBlock(regs.smaphore_address.SmaphoreAddress(), &block, sizeof(block));
} else {
- const auto address =
- memory_manager->GpuToCpuAddress(regs.smaphore_address.SmaphoreAddress());
- const u32 word = Memory::Read32(*address);
+ const u32 word{memory_manager->Read32(regs.smaphore_address.SmaphoreAddress())};
if ((op == GpuSemaphoreOperation::AcquireEqual && word == regs.semaphore_sequence) ||
(op == GpuSemaphoreOperation::AcquireGequal &&
static_cast<s32>(word - regs.semaphore_sequence) > 0) ||
@@ -317,13 +314,11 @@ void GPU::ProcessSemaphoreTriggerMethod() {
}
void GPU::ProcessSemaphoreRelease() {
- const auto address = memory_manager->GpuToCpuAddress(regs.smaphore_address.SmaphoreAddress());
- Memory::Write32(*address, regs.semaphore_release);
+ memory_manager->Write32(regs.smaphore_address.SmaphoreAddress(), regs.semaphore_release);
}
void GPU::ProcessSemaphoreAcquire() {
- const auto address = memory_manager->GpuToCpuAddress(regs.smaphore_address.SmaphoreAddress());
- const u32 word = Memory::Read32(*address);
+ const u32 word = memory_manager->Read32(regs.smaphore_address.SmaphoreAddress());
const auto value = regs.semaphore_acquire;
if (word != value) {
regs.acquire_active = true;
diff --git a/src/video_core/gpu.h b/src/video_core/gpu.h
index 56a203275..a14b95c30 100644
--- a/src/video_core/gpu.h
+++ b/src/video_core/gpu.h
@@ -11,6 +11,11 @@
#include "video_core/dma_pusher.h"
#include "video_core/memory_manager.h"
+using CacheAddr = std::uintptr_t;
+inline CacheAddr ToCacheAddr(const void* host_ptr) {
+ return reinterpret_cast<CacheAddr>(host_ptr);
+}
+
namespace Core {
class System;
}
@@ -209,13 +214,13 @@ public:
std::optional<std::reference_wrapper<const Tegra::FramebufferConfig>> framebuffer) = 0;
/// Notify rasterizer that any caches of the specified region should be flushed to Switch memory
- virtual void FlushRegion(VAddr addr, u64 size) = 0;
+ virtual void FlushRegion(CacheAddr addr, u64 size) = 0;
/// Notify rasterizer that any caches of the specified region should be invalidated
- virtual void InvalidateRegion(VAddr addr, u64 size) = 0;
+ virtual void InvalidateRegion(CacheAddr addr, u64 size) = 0;
/// Notify rasterizer that any caches of the specified region should be flushed and invalidated
- virtual void FlushAndInvalidateRegion(VAddr addr, u64 size) = 0;
+ virtual void FlushAndInvalidateRegion(CacheAddr addr, u64 size) = 0;
private:
void ProcessBindMethod(const MethodCall& method_call);
diff --git a/src/video_core/gpu_asynch.cpp b/src/video_core/gpu_asynch.cpp
index ad0a747e3..8b355cf7b 100644
--- a/src/video_core/gpu_asynch.cpp
+++ b/src/video_core/gpu_asynch.cpp
@@ -22,15 +22,15 @@ void GPUAsynch::SwapBuffers(
gpu_thread.SwapBuffers(std::move(framebuffer));
}
-void GPUAsynch::FlushRegion(VAddr addr, u64 size) {
+void GPUAsynch::FlushRegion(CacheAddr addr, u64 size) {
gpu_thread.FlushRegion(addr, size);
}
-void GPUAsynch::InvalidateRegion(VAddr addr, u64 size) {
+void GPUAsynch::InvalidateRegion(CacheAddr addr, u64 size) {
gpu_thread.InvalidateRegion(addr, size);
}
-void GPUAsynch::FlushAndInvalidateRegion(VAddr addr, u64 size) {
+void GPUAsynch::FlushAndInvalidateRegion(CacheAddr addr, u64 size) {
gpu_thread.FlushAndInvalidateRegion(addr, size);
}
diff --git a/src/video_core/gpu_asynch.h b/src/video_core/gpu_asynch.h
index e6a807aba..1dcc61a6c 100644
--- a/src/video_core/gpu_asynch.h
+++ b/src/video_core/gpu_asynch.h
@@ -26,9 +26,9 @@ public:
void PushGPUEntries(Tegra::CommandList&& entries) override;
void SwapBuffers(
std::optional<std::reference_wrapper<const Tegra::FramebufferConfig>> framebuffer) override;
- void FlushRegion(VAddr addr, u64 size) override;
- void InvalidateRegion(VAddr addr, u64 size) override;
- void FlushAndInvalidateRegion(VAddr addr, u64 size) override;
+ void FlushRegion(CacheAddr addr, u64 size) override;
+ void InvalidateRegion(CacheAddr addr, u64 size) override;
+ void FlushAndInvalidateRegion(CacheAddr addr, u64 size) override;
private:
GPUThread::ThreadManager gpu_thread;
diff --git a/src/video_core/gpu_synch.cpp b/src/video_core/gpu_synch.cpp
index 4c00b96c7..2cfc900ed 100644
--- a/src/video_core/gpu_synch.cpp
+++ b/src/video_core/gpu_synch.cpp
@@ -22,15 +22,15 @@ void GPUSynch::SwapBuffers(
renderer.SwapBuffers(std::move(framebuffer));
}
-void GPUSynch::FlushRegion(VAddr addr, u64 size) {
+void GPUSynch::FlushRegion(CacheAddr addr, u64 size) {
renderer.Rasterizer().FlushRegion(addr, size);
}
-void GPUSynch::InvalidateRegion(VAddr addr, u64 size) {
+void GPUSynch::InvalidateRegion(CacheAddr addr, u64 size) {
renderer.Rasterizer().InvalidateRegion(addr, size);
}
-void GPUSynch::FlushAndInvalidateRegion(VAddr addr, u64 size) {
+void GPUSynch::FlushAndInvalidateRegion(CacheAddr addr, u64 size) {
renderer.Rasterizer().FlushAndInvalidateRegion(addr, size);
}
diff --git a/src/video_core/gpu_synch.h b/src/video_core/gpu_synch.h
index 7d5a241ff..766b5631c 100644
--- a/src/video_core/gpu_synch.h
+++ b/src/video_core/gpu_synch.h
@@ -21,9 +21,9 @@ public:
void PushGPUEntries(Tegra::CommandList&& entries) override;
void SwapBuffers(
std::optional<std::reference_wrapper<const Tegra::FramebufferConfig>> framebuffer) override;
- void FlushRegion(VAddr addr, u64 size) override;
- void InvalidateRegion(VAddr addr, u64 size) override;
- void FlushAndInvalidateRegion(VAddr addr, u64 size) override;
+ void FlushRegion(CacheAddr addr, u64 size) override;
+ void InvalidateRegion(CacheAddr addr, u64 size) override;
+ void FlushAndInvalidateRegion(CacheAddr addr, u64 size) override;
};
} // namespace VideoCommon
diff --git a/src/video_core/gpu_thread.cpp b/src/video_core/gpu_thread.cpp
index c5bdd2a17..086b2f625 100644
--- a/src/video_core/gpu_thread.cpp
+++ b/src/video_core/gpu_thread.cpp
@@ -5,7 +5,6 @@
#include "common/assert.h"
#include "common/microprofile.h"
#include "core/frontend/scope_acquire_window_context.h"
-#include "core/settings.h"
#include "video_core/dma_pusher.h"
#include "video_core/gpu.h"
#include "video_core/gpu_thread.h"
@@ -13,38 +12,13 @@
namespace VideoCommon::GPUThread {
-/// Executes a single GPU thread command
-static void ExecuteCommand(CommandData* command, VideoCore::RendererBase& renderer,
- Tegra::DmaPusher& dma_pusher) {
- if (const auto submit_list = std::get_if<SubmitListCommand>(command)) {
- dma_pusher.Push(std::move(submit_list->entries));
- dma_pusher.DispatchCalls();
- } else if (const auto data = std::get_if<SwapBuffersCommand>(command)) {
- renderer.SwapBuffers(data->framebuffer);
- } else if (const auto data = std::get_if<FlushRegionCommand>(command)) {
- renderer.Rasterizer().FlushRegion(data->addr, data->size);
- } else if (const auto data = std::get_if<InvalidateRegionCommand>(command)) {
- renderer.Rasterizer().InvalidateRegion(data->addr, data->size);
- } else if (const auto data = std::get_if<FlushAndInvalidateRegionCommand>(command)) {
- renderer.Rasterizer().FlushAndInvalidateRegion(data->addr, data->size);
- } else {
- UNREACHABLE();
- }
-}
-
/// Runs the GPU thread
static void RunThread(VideoCore::RendererBase& renderer, Tegra::DmaPusher& dma_pusher,
SynchState& state) {
-
MicroProfileOnThreadCreate("GpuThread");
- auto WaitForWakeup = [&]() {
- std::unique_lock<std::mutex> lock{state.signal_mutex};
- state.signal_condition.wait(lock, [&] { return !state.is_idle || !state.is_running; });
- };
-
// Wait for first GPU command before acquiring the window context
- WaitForWakeup();
+ state.WaitForCommands();
// If emulation was stopped during disk shader loading, abort before trying to acquire context
if (!state.is_running) {
@@ -53,100 +27,72 @@ static void RunThread(VideoCore::RendererBase& renderer, Tegra::DmaPusher& dma_p
Core::Frontend::ScopeAcquireWindowContext acquire_context{renderer.GetRenderWindow()};
+ CommandDataContainer next;
while (state.is_running) {
- if (!state.is_running) {
- return;
- }
-
- {
- // Thread has been woken up, so make the previous write queue the next read queue
- std::lock_guard<std::mutex> lock{state.signal_mutex};
- std::swap(state.push_queue, state.pop_queue);
- }
-
- // Execute all of the GPU commands
- while (!state.pop_queue->empty()) {
- ExecuteCommand(&state.pop_queue->front(), renderer, dma_pusher);
- state.pop_queue->pop();
+ state.WaitForCommands();
+ 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)) {
+ state.DecrementFramesCounter();
+ renderer.SwapBuffers(std::move(data->framebuffer));
+ } 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 (const auto data = std::get_if<EndProcessingCommand>(&next.data)) {
+ return;
+ } else {
+ UNREACHABLE();
+ }
}
-
- state.UpdateIdleState();
-
- // Signal that the GPU thread has finished processing commands
- if (state.is_idle) {
- state.idle_condition.notify_one();
- }
-
- // Wait for CPU thread to send more GPU commands
- WaitForWakeup();
}
}
ThreadManager::ThreadManager(VideoCore::RendererBase& renderer, Tegra::DmaPusher& dma_pusher)
: renderer{renderer}, dma_pusher{dma_pusher}, thread{RunThread, std::ref(renderer),
- std::ref(dma_pusher), std::ref(state)},
- thread_id{thread.get_id()} {}
+ std::ref(dma_pusher), std::ref(state)} {}
ThreadManager::~ThreadManager() {
- {
- // Notify GPU thread that a shutdown is pending
- std::lock_guard<std::mutex> lock{state.signal_mutex};
- state.is_running = false;
- }
-
- state.signal_condition.notify_one();
+ // Notify GPU thread that a shutdown is pending
+ PushCommand(EndProcessingCommand());
thread.join();
}
void ThreadManager::SubmitList(Tegra::CommandList&& entries) {
- if (entries.empty()) {
- return;
- }
-
- PushCommand(SubmitListCommand(std::move(entries)), false, false);
+ PushCommand(SubmitListCommand(std::move(entries)));
}
void ThreadManager::SwapBuffers(
std::optional<std::reference_wrapper<const Tegra::FramebufferConfig>> framebuffer) {
- PushCommand(SwapBuffersCommand(std::move(framebuffer)), true, false);
+ state.IncrementFramesCounter();
+ PushCommand(SwapBuffersCommand(std::move(framebuffer)));
+ state.WaitForFrames();
}
-void ThreadManager::FlushRegion(VAddr addr, u64 size) {
- // Block the CPU when using accurate emulation
- PushCommand(FlushRegionCommand(addr, size), Settings::values.use_accurate_gpu_emulation, false);
+void ThreadManager::FlushRegion(CacheAddr addr, u64 size) {
+ PushCommand(FlushRegionCommand(addr, size));
}
-void ThreadManager::InvalidateRegion(VAddr addr, u64 size) {
- PushCommand(InvalidateRegionCommand(addr, size), true, true);
+void ThreadManager::InvalidateRegion(CacheAddr addr, u64 size) {
+ if (state.queue.Empty()) {
+ // It's quicker to invalidate a single region on the CPU if the queue is already empty
+ renderer.Rasterizer().InvalidateRegion(addr, size);
+ } else {
+ PushCommand(InvalidateRegionCommand(addr, size));
+ }
}
-void ThreadManager::FlushAndInvalidateRegion(VAddr addr, u64 size) {
+void ThreadManager::FlushAndInvalidateRegion(CacheAddr addr, u64 size) {
+ // Skip flush on asynch mode, as FlushAndInvalidateRegion is not used for anything too important
InvalidateRegion(addr, size);
}
-void ThreadManager::PushCommand(CommandData&& command_data, bool wait_for_idle, bool allow_on_cpu) {
- {
- std::lock_guard<std::mutex> lock{state.signal_mutex};
-
- if ((allow_on_cpu && state.is_idle) || IsGpuThread()) {
- // Execute the command synchronously on the current thread
- ExecuteCommand(&command_data, renderer, dma_pusher);
- return;
- }
-
- // Push the command to the GPU thread
- state.UpdateIdleState();
- state.push_queue->emplace(command_data);
- }
-
- // Signal the GPU thread that commands are pending
- state.signal_condition.notify_one();
-
- if (wait_for_idle) {
- // Wait for the GPU to be idle (all commands to be executed)
- std::unique_lock<std::mutex> lock{state.idle_mutex};
- state.idle_condition.wait(lock, [this] { return static_cast<bool>(state.is_idle); });
- }
+void ThreadManager::PushCommand(CommandData&& command_data) {
+ state.queue.Push(CommandDataContainer(std::move(command_data)));
+ state.SignalCommands();
}
} // namespace VideoCommon::GPUThread
diff --git a/src/video_core/gpu_thread.h b/src/video_core/gpu_thread.h
index edb148b14..8cd7db1c6 100644
--- a/src/video_core/gpu_thread.h
+++ b/src/video_core/gpu_thread.h
@@ -13,6 +13,9 @@
#include <thread>
#include <variant>
+#include "common/threadsafe_queue.h"
+#include "video_core/gpu.h"
+
namespace Tegra {
struct FramebufferConfig;
class DmaPusher;
@@ -24,6 +27,9 @@ class RendererBase;
namespace VideoCommon::GPUThread {
+/// Command to signal to the GPU thread that processing has ended
+struct EndProcessingCommand final {};
+
/// Command to signal to the GPU thread that a command list is ready for processing
struct SubmitListCommand final {
explicit SubmitListCommand(Tegra::CommandList&& entries) : entries{std::move(entries)} {}
@@ -36,59 +42,110 @@ struct SwapBuffersCommand final {
explicit SwapBuffersCommand(std::optional<const Tegra::FramebufferConfig> framebuffer)
: framebuffer{std::move(framebuffer)} {}
- std::optional<const Tegra::FramebufferConfig> framebuffer;
+ std::optional<Tegra::FramebufferConfig> framebuffer;
};
/// Command to signal to the GPU thread to flush a region
struct FlushRegionCommand final {
- explicit constexpr FlushRegionCommand(VAddr addr, u64 size) : addr{addr}, size{size} {}
+ explicit constexpr FlushRegionCommand(CacheAddr addr, u64 size) : addr{addr}, size{size} {}
- const VAddr addr;
- const u64 size;
+ CacheAddr addr;
+ u64 size;
};
/// Command to signal to the GPU thread to invalidate a region
struct InvalidateRegionCommand final {
- explicit constexpr InvalidateRegionCommand(VAddr addr, u64 size) : addr{addr}, size{size} {}
+ explicit constexpr InvalidateRegionCommand(CacheAddr addr, u64 size) : addr{addr}, size{size} {}
- const VAddr addr;
- const u64 size;
+ CacheAddr addr;
+ u64 size;
};
/// Command to signal to the GPU thread to flush and invalidate a region
struct FlushAndInvalidateRegionCommand final {
- explicit constexpr FlushAndInvalidateRegionCommand(VAddr addr, u64 size)
+ explicit constexpr FlushAndInvalidateRegionCommand(CacheAddr addr, u64 size)
: addr{addr}, size{size} {}
- const VAddr addr;
- const u64 size;
+ CacheAddr addr;
+ u64 size;
};
-using CommandData = std::variant<SubmitListCommand, SwapBuffersCommand, FlushRegionCommand,
- InvalidateRegionCommand, FlushAndInvalidateRegionCommand>;
+using CommandData =
+ std::variant<EndProcessingCommand, SubmitListCommand, SwapBuffersCommand, FlushRegionCommand,
+ InvalidateRegionCommand, FlushAndInvalidateRegionCommand>;
+
+struct CommandDataContainer {
+ CommandDataContainer() = default;
+
+ CommandDataContainer(CommandData&& data) : data{std::move(data)} {}
+
+ CommandDataContainer& operator=(const CommandDataContainer& t) {
+ data = std::move(t.data);
+ return *this;
+ }
+
+ CommandData data;
+};
/// Struct used to synchronize the GPU thread
struct SynchState final {
- std::atomic<bool> is_running{true};
- std::atomic<bool> is_idle{true};
- std::condition_variable signal_condition;
- std::mutex signal_mutex;
- std::condition_variable idle_condition;
- std::mutex idle_mutex;
-
- // We use two queues for sending commands to the GPU thread, one for writing (push_queue) to and
- // one for reading from (pop_queue). These are swapped whenever the current pop_queue becomes
- // empty. This allows for efficient thread-safe access, as it does not require any copies.
-
- using CommandQueue = std::queue<CommandData>;
- std::array<CommandQueue, 2> command_queues;
- CommandQueue* push_queue{&command_queues[0]};
- CommandQueue* pop_queue{&command_queues[1]};
-
- void UpdateIdleState() {
- std::lock_guard<std::mutex> lock{idle_mutex};
- is_idle = command_queues[0].empty() && command_queues[1].empty();
+ std::atomic_bool is_running{true};
+ std::atomic_int queued_frame_count{};
+ std::mutex frames_mutex;
+ std::mutex commands_mutex;
+ std::condition_variable commands_condition;
+ std::condition_variable frames_condition;
+
+ void IncrementFramesCounter() {
+ std::lock_guard<std::mutex> lock{frames_mutex};
+ ++queued_frame_count;
+ }
+
+ void DecrementFramesCounter() {
+ {
+ std::lock_guard<std::mutex> lock{frames_mutex};
+ --queued_frame_count;
+
+ if (queued_frame_count) {
+ return;
+ }
+ }
+ frames_condition.notify_one();
}
+
+ void WaitForFrames() {
+ {
+ std::lock_guard<std::mutex> lock{frames_mutex};
+ if (!queued_frame_count) {
+ return;
+ }
+ }
+
+ // Wait for the GPU to be idle (all commands to be executed)
+ {
+ std::unique_lock<std::mutex> lock{frames_mutex};
+ frames_condition.wait(lock, [this] { return !queued_frame_count; });
+ }
+ }
+
+ void SignalCommands() {
+ {
+ std::unique_lock<std::mutex> lock{commands_mutex};
+ if (queue.Empty()) {
+ return;
+ }
+ }
+
+ commands_condition.notify_one();
+ }
+
+ void WaitForCommands() {
+ std::unique_lock<std::mutex> lock{commands_mutex};
+ commands_condition.wait(lock, [this] { return !queue.Empty(); });
+ }
+
+ using CommandQueue = Common::SPSCQueue<CommandDataContainer>;
+ CommandQueue queue;
};
/// Class used to manage the GPU thread
@@ -105,22 +162,17 @@ public:
std::optional<std::reference_wrapper<const Tegra::FramebufferConfig>> framebuffer);
/// Notify rasterizer that any caches of the specified region should be flushed to Switch memory
- void FlushRegion(VAddr addr, u64 size);
+ void FlushRegion(CacheAddr addr, u64 size);
/// Notify rasterizer that any caches of the specified region should be invalidated
- void InvalidateRegion(VAddr addr, u64 size);
+ void InvalidateRegion(CacheAddr addr, u64 size);
/// Notify rasterizer that any caches of the specified region should be flushed and invalidated
- void FlushAndInvalidateRegion(VAddr addr, u64 size);
+ void FlushAndInvalidateRegion(CacheAddr addr, u64 size);
private:
/// Pushes a command to be executed by the GPU thread
- void PushCommand(CommandData&& command_data, bool wait_for_idle, bool allow_on_cpu);
-
- /// Returns true if this is called by the GPU thread
- bool IsGpuThread() const {
- return std::this_thread::get_id() == thread_id;
- }
+ void PushCommand(CommandData&& command_data);
private:
SynchState state;
diff --git a/src/video_core/memory_manager.cpp b/src/video_core/memory_manager.cpp
index 54abe5298..8e8f36f28 100644
--- a/src/video_core/memory_manager.cpp
+++ b/src/video_core/memory_manager.cpp
@@ -5,6 +5,7 @@
#include "common/alignment.h"
#include "common/assert.h"
#include "common/logging/log.h"
+#include "core/memory.h"
#include "video_core/memory_manager.h"
namespace Tegra {
@@ -162,15 +163,51 @@ std::optional<VAddr> MemoryManager::GpuToCpuAddress(GPUVAddr gpu_addr) {
return base_addr + (gpu_addr & PAGE_MASK);
}
-std::vector<GPUVAddr> MemoryManager::CpuToGpuAddress(VAddr cpu_addr) const {
- std::vector<GPUVAddr> results;
- for (const auto& region : mapped_regions) {
- if (cpu_addr >= region.cpu_addr && cpu_addr < (region.cpu_addr + region.size)) {
- const u64 offset{cpu_addr - region.cpu_addr};
- results.push_back(region.gpu_addr + offset);
- }
- }
- return results;
+u8 MemoryManager::Read8(GPUVAddr addr) {
+ return Memory::Read8(*GpuToCpuAddress(addr));
+}
+
+u16 MemoryManager::Read16(GPUVAddr addr) {
+ return Memory::Read16(*GpuToCpuAddress(addr));
+}
+
+u32 MemoryManager::Read32(GPUVAddr addr) {
+ return Memory::Read32(*GpuToCpuAddress(addr));
+}
+
+u64 MemoryManager::Read64(GPUVAddr addr) {
+ return Memory::Read64(*GpuToCpuAddress(addr));
+}
+
+void MemoryManager::Write8(GPUVAddr addr, u8 data) {
+ Memory::Write8(*GpuToCpuAddress(addr), data);
+}
+
+void MemoryManager::Write16(GPUVAddr addr, u16 data) {
+ Memory::Write16(*GpuToCpuAddress(addr), data);
+}
+
+void MemoryManager::Write32(GPUVAddr addr, u32 data) {
+ Memory::Write32(*GpuToCpuAddress(addr), data);
+}
+
+void MemoryManager::Write64(GPUVAddr addr, u64 data) {
+ Memory::Write64(*GpuToCpuAddress(addr), data);
+}
+
+u8* MemoryManager::GetPointer(GPUVAddr addr) {
+ return Memory::GetPointer(*GpuToCpuAddress(addr));
+}
+
+void MemoryManager::ReadBlock(GPUVAddr src_addr, void* dest_buffer, std::size_t size) {
+ std::memcpy(dest_buffer, GetPointer(src_addr), size);
+}
+void MemoryManager::WriteBlock(GPUVAddr dest_addr, const void* src_buffer, std::size_t size) {
+ std::memcpy(GetPointer(dest_addr), src_buffer, size);
+}
+
+void MemoryManager::CopyBlock(GPUVAddr dest_addr, GPUVAddr src_addr, std::size_t size) {
+ std::memcpy(GetPointer(dest_addr), GetPointer(src_addr), size);
}
VAddr& MemoryManager::PageSlot(GPUVAddr gpu_addr) {
diff --git a/src/video_core/memory_manager.h b/src/video_core/memory_manager.h
index fb03497ca..425e2f31c 100644
--- a/src/video_core/memory_manager.h
+++ b/src/video_core/memory_manager.h
@@ -27,12 +27,27 @@ public:
GPUVAddr UnmapBuffer(GPUVAddr gpu_addr, u64 size);
GPUVAddr GetRegionEnd(GPUVAddr region_start) const;
std::optional<VAddr> GpuToCpuAddress(GPUVAddr gpu_addr);
- std::vector<GPUVAddr> CpuToGpuAddress(VAddr cpu_addr) const;
static constexpr u64 PAGE_BITS = 16;
static constexpr u64 PAGE_SIZE = 1 << PAGE_BITS;
static constexpr u64 PAGE_MASK = PAGE_SIZE - 1;
+ u8 Read8(GPUVAddr addr);
+ u16 Read16(GPUVAddr addr);
+ u32 Read32(GPUVAddr addr);
+ u64 Read64(GPUVAddr addr);
+
+ void Write8(GPUVAddr addr, u8 data);
+ void Write16(GPUVAddr addr, u16 data);
+ void Write32(GPUVAddr addr, u32 data);
+ void Write64(GPUVAddr addr, u64 data);
+
+ u8* GetPointer(GPUVAddr vaddr);
+
+ void ReadBlock(GPUVAddr src_addr, void* dest_buffer, std::size_t size);
+ void WriteBlock(GPUVAddr dest_addr, const void* src_buffer, std::size_t size);
+ void CopyBlock(VAddr dest_addr, VAddr src_addr, std::size_t size);
+
private:
enum class PageStatus : u64 {
Unmapped = 0xFFFFFFFFFFFFFFFFULL,
diff --git a/src/video_core/morton.cpp b/src/video_core/morton.cpp
index b68f4fb13..3e91cbc83 100644
--- a/src/video_core/morton.cpp
+++ b/src/video_core/morton.cpp
@@ -6,7 +6,6 @@
#include <cstring>
#include "common/assert.h"
#include "common/common_types.h"
-#include "core/memory.h"
#include "video_core/morton.h"
#include "video_core/surface.h"
#include "video_core/textures/decoders.h"
@@ -16,12 +15,12 @@ namespace VideoCore {
using Surface::GetBytesPerPixel;
using Surface::PixelFormat;
-using MortonCopyFn = void (*)(u32, u32, u32, u32, u32, u32, u8*, std::size_t, VAddr);
+using MortonCopyFn = void (*)(u32, u32, u32, u32, u32, u32, u8*, u8*);
using ConversionArray = std::array<MortonCopyFn, Surface::MaxPixelFormat>;
template <bool morton_to_linear, PixelFormat format>
static void MortonCopy(u32 stride, u32 block_height, u32 height, u32 block_depth, u32 depth,
- u32 tile_width_spacing, u8* buffer, std::size_t buffer_size, VAddr addr) {
+ u32 tile_width_spacing, u8* buffer, u8* addr) {
constexpr u32 bytes_per_pixel = GetBytesPerPixel(format);
// With the BCn formats (DXT and DXN), each 4x4 tile is swizzled instead of just individual
@@ -34,150 +33,146 @@ static void MortonCopy(u32 stride, u32 block_height, u32 height, u32 block_depth
stride, height, depth, block_height, block_depth,
tile_width_spacing);
} else {
- Tegra::Texture::CopySwizzledData(
- (stride + tile_size_x - 1) / tile_size_x, (height + tile_size_y - 1) / tile_size_y,
- depth, bytes_per_pixel, bytes_per_pixel, Memory::GetPointer(addr), buffer, false,
- block_height, block_depth, tile_width_spacing);
+ Tegra::Texture::CopySwizzledData((stride + tile_size_x - 1) / tile_size_x,
+ (height + tile_size_y - 1) / tile_size_y, depth,
+ bytes_per_pixel, bytes_per_pixel, addr, buffer, false,
+ block_height, block_depth, tile_width_spacing);
}
}
static constexpr ConversionArray morton_to_linear_fns = {
- // clang-format off
- MortonCopy<true, PixelFormat::ABGR8U>,
- MortonCopy<true, PixelFormat::ABGR8S>,
- MortonCopy<true, PixelFormat::ABGR8UI>,
- MortonCopy<true, PixelFormat::B5G6R5U>,
- MortonCopy<true, PixelFormat::A2B10G10R10U>,
- MortonCopy<true, PixelFormat::A1B5G5R5U>,
- MortonCopy<true, PixelFormat::R8U>,
- MortonCopy<true, PixelFormat::R8UI>,
- MortonCopy<true, PixelFormat::RGBA16F>,
- MortonCopy<true, PixelFormat::RGBA16U>,
- MortonCopy<true, PixelFormat::RGBA16UI>,
- MortonCopy<true, PixelFormat::R11FG11FB10F>,
- MortonCopy<true, PixelFormat::RGBA32UI>,
- MortonCopy<true, PixelFormat::DXT1>,
- MortonCopy<true, PixelFormat::DXT23>,
- MortonCopy<true, PixelFormat::DXT45>,
- MortonCopy<true, PixelFormat::DXN1>,
- MortonCopy<true, PixelFormat::DXN2UNORM>,
- MortonCopy<true, PixelFormat::DXN2SNORM>,
- MortonCopy<true, PixelFormat::BC7U>,
- MortonCopy<true, PixelFormat::BC6H_UF16>,
- MortonCopy<true, PixelFormat::BC6H_SF16>,
- MortonCopy<true, PixelFormat::ASTC_2D_4X4>,
- MortonCopy<true, PixelFormat::BGRA8>,
- MortonCopy<true, PixelFormat::RGBA32F>,
- MortonCopy<true, PixelFormat::RG32F>,
- MortonCopy<true, PixelFormat::R32F>,
- MortonCopy<true, PixelFormat::R16F>,
- MortonCopy<true, PixelFormat::R16U>,
- MortonCopy<true, PixelFormat::R16S>,
- MortonCopy<true, PixelFormat::R16UI>,
- MortonCopy<true, PixelFormat::R16I>,
- MortonCopy<true, PixelFormat::RG16>,
- MortonCopy<true, PixelFormat::RG16F>,
- MortonCopy<true, PixelFormat::RG16UI>,
- MortonCopy<true, PixelFormat::RG16I>,
- MortonCopy<true, PixelFormat::RG16S>,
- MortonCopy<true, PixelFormat::RGB32F>,
- MortonCopy<true, PixelFormat::RGBA8_SRGB>,
- MortonCopy<true, PixelFormat::RG8U>,
- MortonCopy<true, PixelFormat::RG8S>,
- MortonCopy<true, PixelFormat::RG32UI>,
- MortonCopy<true, PixelFormat::R32UI>,
- MortonCopy<true, PixelFormat::ASTC_2D_8X8>,
- MortonCopy<true, PixelFormat::ASTC_2D_8X5>,
- MortonCopy<true, PixelFormat::ASTC_2D_5X4>,
- MortonCopy<true, PixelFormat::BGRA8_SRGB>,
- MortonCopy<true, PixelFormat::DXT1_SRGB>,
- MortonCopy<true, PixelFormat::DXT23_SRGB>,
- MortonCopy<true, PixelFormat::DXT45_SRGB>,
- MortonCopy<true, PixelFormat::BC7U_SRGB>,
- MortonCopy<true, PixelFormat::ASTC_2D_4X4_SRGB>,
- MortonCopy<true, PixelFormat::ASTC_2D_8X8_SRGB>,
- MortonCopy<true, PixelFormat::ASTC_2D_8X5_SRGB>,
- MortonCopy<true, PixelFormat::ASTC_2D_5X4_SRGB>,
- MortonCopy<true, PixelFormat::ASTC_2D_5X5>,
- MortonCopy<true, PixelFormat::ASTC_2D_5X5_SRGB>,
- MortonCopy<true, PixelFormat::ASTC_2D_10X8>,
- MortonCopy<true, PixelFormat::ASTC_2D_10X8_SRGB>,
- MortonCopy<true, PixelFormat::Z32F>,
- MortonCopy<true, PixelFormat::Z16>,
- MortonCopy<true, PixelFormat::Z24S8>,
- MortonCopy<true, PixelFormat::S8Z24>,
- MortonCopy<true, PixelFormat::Z32FS8>,
- // clang-format on
+ MortonCopy<true, PixelFormat::ABGR8U>,
+ MortonCopy<true, PixelFormat::ABGR8S>,
+ MortonCopy<true, PixelFormat::ABGR8UI>,
+ MortonCopy<true, PixelFormat::B5G6R5U>,
+ MortonCopy<true, PixelFormat::A2B10G10R10U>,
+ MortonCopy<true, PixelFormat::A1B5G5R5U>,
+ MortonCopy<true, PixelFormat::R8U>,
+ MortonCopy<true, PixelFormat::R8UI>,
+ MortonCopy<true, PixelFormat::RGBA16F>,
+ MortonCopy<true, PixelFormat::RGBA16U>,
+ MortonCopy<true, PixelFormat::RGBA16UI>,
+ MortonCopy<true, PixelFormat::R11FG11FB10F>,
+ MortonCopy<true, PixelFormat::RGBA32UI>,
+ MortonCopy<true, PixelFormat::DXT1>,
+ MortonCopy<true, PixelFormat::DXT23>,
+ MortonCopy<true, PixelFormat::DXT45>,
+ MortonCopy<true, PixelFormat::DXN1>,
+ MortonCopy<true, PixelFormat::DXN2UNORM>,
+ MortonCopy<true, PixelFormat::DXN2SNORM>,
+ MortonCopy<true, PixelFormat::BC7U>,
+ MortonCopy<true, PixelFormat::BC6H_UF16>,
+ MortonCopy<true, PixelFormat::BC6H_SF16>,
+ MortonCopy<true, PixelFormat::ASTC_2D_4X4>,
+ MortonCopy<true, PixelFormat::BGRA8>,
+ MortonCopy<true, PixelFormat::RGBA32F>,
+ MortonCopy<true, PixelFormat::RG32F>,
+ MortonCopy<true, PixelFormat::R32F>,
+ MortonCopy<true, PixelFormat::R16F>,
+ MortonCopy<true, PixelFormat::R16U>,
+ MortonCopy<true, PixelFormat::R16S>,
+ MortonCopy<true, PixelFormat::R16UI>,
+ MortonCopy<true, PixelFormat::R16I>,
+ MortonCopy<true, PixelFormat::RG16>,
+ MortonCopy<true, PixelFormat::RG16F>,
+ MortonCopy<true, PixelFormat::RG16UI>,
+ MortonCopy<true, PixelFormat::RG16I>,
+ MortonCopy<true, PixelFormat::RG16S>,
+ MortonCopy<true, PixelFormat::RGB32F>,
+ MortonCopy<true, PixelFormat::RGBA8_SRGB>,
+ MortonCopy<true, PixelFormat::RG8U>,
+ MortonCopy<true, PixelFormat::RG8S>,
+ MortonCopy<true, PixelFormat::RG32UI>,
+ MortonCopy<true, PixelFormat::R32UI>,
+ MortonCopy<true, PixelFormat::ASTC_2D_8X8>,
+ MortonCopy<true, PixelFormat::ASTC_2D_8X5>,
+ MortonCopy<true, PixelFormat::ASTC_2D_5X4>,
+ MortonCopy<true, PixelFormat::BGRA8_SRGB>,
+ MortonCopy<true, PixelFormat::DXT1_SRGB>,
+ MortonCopy<true, PixelFormat::DXT23_SRGB>,
+ MortonCopy<true, PixelFormat::DXT45_SRGB>,
+ MortonCopy<true, PixelFormat::BC7U_SRGB>,
+ MortonCopy<true, PixelFormat::ASTC_2D_4X4_SRGB>,
+ MortonCopy<true, PixelFormat::ASTC_2D_8X8_SRGB>,
+ MortonCopy<true, PixelFormat::ASTC_2D_8X5_SRGB>,
+ MortonCopy<true, PixelFormat::ASTC_2D_5X4_SRGB>,
+ MortonCopy<true, PixelFormat::ASTC_2D_5X5>,
+ MortonCopy<true, PixelFormat::ASTC_2D_5X5_SRGB>,
+ MortonCopy<true, PixelFormat::ASTC_2D_10X8>,
+ MortonCopy<true, PixelFormat::ASTC_2D_10X8_SRGB>,
+ MortonCopy<true, PixelFormat::Z32F>,
+ MortonCopy<true, PixelFormat::Z16>,
+ MortonCopy<true, PixelFormat::Z24S8>,
+ MortonCopy<true, PixelFormat::S8Z24>,
+ MortonCopy<true, PixelFormat::Z32FS8>,
};
static constexpr ConversionArray linear_to_morton_fns = {
- // clang-format off
- MortonCopy<false, PixelFormat::ABGR8U>,
- MortonCopy<false, PixelFormat::ABGR8S>,
- MortonCopy<false, PixelFormat::ABGR8UI>,
- MortonCopy<false, PixelFormat::B5G6R5U>,
- MortonCopy<false, PixelFormat::A2B10G10R10U>,
- MortonCopy<false, PixelFormat::A1B5G5R5U>,
- MortonCopy<false, PixelFormat::R8U>,
- MortonCopy<false, PixelFormat::R8UI>,
- MortonCopy<false, PixelFormat::RGBA16F>,
- MortonCopy<false, PixelFormat::RGBA16U>,
- MortonCopy<false, PixelFormat::RGBA16UI>,
- MortonCopy<false, PixelFormat::R11FG11FB10F>,
- MortonCopy<false, PixelFormat::RGBA32UI>,
- MortonCopy<false, PixelFormat::DXT1>,
- MortonCopy<false, PixelFormat::DXT23>,
- MortonCopy<false, PixelFormat::DXT45>,
- MortonCopy<false, PixelFormat::DXN1>,
- MortonCopy<false, PixelFormat::DXN2UNORM>,
- MortonCopy<false, PixelFormat::DXN2SNORM>,
- MortonCopy<false, PixelFormat::BC7U>,
- MortonCopy<false, PixelFormat::BC6H_UF16>,
- MortonCopy<false, PixelFormat::BC6H_SF16>,
- // TODO(Subv): Swizzling ASTC formats are not supported
- nullptr,
- MortonCopy<false, PixelFormat::BGRA8>,
- MortonCopy<false, PixelFormat::RGBA32F>,
- MortonCopy<false, PixelFormat::RG32F>,
- MortonCopy<false, PixelFormat::R32F>,
- MortonCopy<false, PixelFormat::R16F>,
- MortonCopy<false, PixelFormat::R16U>,
- MortonCopy<false, PixelFormat::R16S>,
- MortonCopy<false, PixelFormat::R16UI>,
- MortonCopy<false, PixelFormat::R16I>,
- MortonCopy<false, PixelFormat::RG16>,
- MortonCopy<false, PixelFormat::RG16F>,
- MortonCopy<false, PixelFormat::RG16UI>,
- MortonCopy<false, PixelFormat::RG16I>,
- MortonCopy<false, PixelFormat::RG16S>,
- MortonCopy<false, PixelFormat::RGB32F>,
- MortonCopy<false, PixelFormat::RGBA8_SRGB>,
- MortonCopy<false, PixelFormat::RG8U>,
- MortonCopy<false, PixelFormat::RG8S>,
- MortonCopy<false, PixelFormat::RG32UI>,
- MortonCopy<false, PixelFormat::R32UI>,
- nullptr,
- nullptr,
- nullptr,
- MortonCopy<false, PixelFormat::BGRA8_SRGB>,
- MortonCopy<false, PixelFormat::DXT1_SRGB>,
- MortonCopy<false, PixelFormat::DXT23_SRGB>,
- MortonCopy<false, PixelFormat::DXT45_SRGB>,
- MortonCopy<false, PixelFormat::BC7U_SRGB>,
- nullptr,
- nullptr,
- nullptr,
- nullptr,
- nullptr,
- nullptr,
- nullptr,
- nullptr,
- MortonCopy<false, PixelFormat::Z32F>,
- MortonCopy<false, PixelFormat::Z16>,
- MortonCopy<false, PixelFormat::Z24S8>,
- MortonCopy<false, PixelFormat::S8Z24>,
- MortonCopy<false, PixelFormat::Z32FS8>,
- // clang-format on
+ MortonCopy<false, PixelFormat::ABGR8U>,
+ MortonCopy<false, PixelFormat::ABGR8S>,
+ MortonCopy<false, PixelFormat::ABGR8UI>,
+ MortonCopy<false, PixelFormat::B5G6R5U>,
+ MortonCopy<false, PixelFormat::A2B10G10R10U>,
+ MortonCopy<false, PixelFormat::A1B5G5R5U>,
+ MortonCopy<false, PixelFormat::R8U>,
+ MortonCopy<false, PixelFormat::R8UI>,
+ MortonCopy<false, PixelFormat::RGBA16F>,
+ MortonCopy<false, PixelFormat::RGBA16U>,
+ MortonCopy<false, PixelFormat::RGBA16UI>,
+ MortonCopy<false, PixelFormat::R11FG11FB10F>,
+ MortonCopy<false, PixelFormat::RGBA32UI>,
+ MortonCopy<false, PixelFormat::DXT1>,
+ MortonCopy<false, PixelFormat::DXT23>,
+ MortonCopy<false, PixelFormat::DXT45>,
+ MortonCopy<false, PixelFormat::DXN1>,
+ MortonCopy<false, PixelFormat::DXN2UNORM>,
+ MortonCopy<false, PixelFormat::DXN2SNORM>,
+ MortonCopy<false, PixelFormat::BC7U>,
+ MortonCopy<false, PixelFormat::BC6H_UF16>,
+ MortonCopy<false, PixelFormat::BC6H_SF16>,
+ // TODO(Subv): Swizzling ASTC formats are not supported
+ nullptr,
+ MortonCopy<false, PixelFormat::BGRA8>,
+ MortonCopy<false, PixelFormat::RGBA32F>,
+ MortonCopy<false, PixelFormat::RG32F>,
+ MortonCopy<false, PixelFormat::R32F>,
+ MortonCopy<false, PixelFormat::R16F>,
+ MortonCopy<false, PixelFormat::R16U>,
+ MortonCopy<false, PixelFormat::R16S>,
+ MortonCopy<false, PixelFormat::R16UI>,
+ MortonCopy<false, PixelFormat::R16I>,
+ MortonCopy<false, PixelFormat::RG16>,
+ MortonCopy<false, PixelFormat::RG16F>,
+ MortonCopy<false, PixelFormat::RG16UI>,
+ MortonCopy<false, PixelFormat::RG16I>,
+ MortonCopy<false, PixelFormat::RG16S>,
+ MortonCopy<false, PixelFormat::RGB32F>,
+ MortonCopy<false, PixelFormat::RGBA8_SRGB>,
+ MortonCopy<false, PixelFormat::RG8U>,
+ MortonCopy<false, PixelFormat::RG8S>,
+ MortonCopy<false, PixelFormat::RG32UI>,
+ MortonCopy<false, PixelFormat::R32UI>,
+ nullptr,
+ nullptr,
+ nullptr,
+ MortonCopy<false, PixelFormat::BGRA8_SRGB>,
+ MortonCopy<false, PixelFormat::DXT1_SRGB>,
+ MortonCopy<false, PixelFormat::DXT23_SRGB>,
+ MortonCopy<false, PixelFormat::DXT45_SRGB>,
+ MortonCopy<false, PixelFormat::BC7U_SRGB>,
+ nullptr,
+ nullptr,
+ nullptr,
+ nullptr,
+ nullptr,
+ nullptr,
+ nullptr,
+ nullptr,
+ MortonCopy<false, PixelFormat::Z32F>,
+ MortonCopy<false, PixelFormat::Z16>,
+ MortonCopy<false, PixelFormat::Z24S8>,
+ MortonCopy<false, PixelFormat::S8Z24>,
+ MortonCopy<false, PixelFormat::Z32FS8>,
};
static MortonCopyFn GetSwizzleFunction(MortonSwizzleMode mode, Surface::PixelFormat format) {
@@ -191,45 +186,6 @@ static MortonCopyFn GetSwizzleFunction(MortonSwizzleMode mode, Surface::PixelFor
return morton_to_linear_fns[static_cast<std::size_t>(format)];
}
-/// 8x8 Z-Order coordinate from 2D coordinates
-static u32 MortonInterleave(u32 x, u32 y) {
- static const u32 xlut[] = {0x00, 0x01, 0x04, 0x05, 0x10, 0x11, 0x14, 0x15};
- static const u32 ylut[] = {0x00, 0x02, 0x08, 0x0a, 0x20, 0x22, 0x28, 0x2a};
- return xlut[x % 8] + ylut[y % 8];
-}
-
-/// Calculates the offset of the position of the pixel in Morton order
-static u32 GetMortonOffset(u32 x, u32 y, u32 bytes_per_pixel) {
- // Images are split into 8x8 tiles. Each tile is composed of four 4x4 subtiles each
- // of which is composed of four 2x2 subtiles each of which is composed of four texels.
- // Each structure is embedded into the next-bigger one in a diagonal pattern, e.g.
- // texels are laid out in a 2x2 subtile like this:
- // 2 3
- // 0 1
- //
- // The full 8x8 tile has the texels arranged like this:
- //
- // 42 43 46 47 58 59 62 63
- // 40 41 44 45 56 57 60 61
- // 34 35 38 39 50 51 54 55
- // 32 33 36 37 48 49 52 53
- // 10 11 14 15 26 27 30 31
- // 08 09 12 13 24 25 28 29
- // 02 03 06 07 18 19 22 23
- // 00 01 04 05 16 17 20 21
- //
- // This pattern is what's called Z-order curve, or Morton order.
-
- const unsigned int block_height = 8;
- const unsigned int coarse_x = x & ~7;
-
- u32 i = MortonInterleave(x, y);
-
- const unsigned int offset = coarse_x * block_height;
-
- return (i + offset) * bytes_per_pixel;
-}
-
static u32 MortonInterleave128(u32 x, u32 y) {
// 128x128 Z-Order coordinate from 2D coordinates
static constexpr u32 xlut[] = {
@@ -325,14 +281,14 @@ static u32 GetMortonOffset128(u32 x, u32 y, u32 bytes_per_pixel) {
void MortonSwizzle(MortonSwizzleMode mode, Surface::PixelFormat format, u32 stride,
u32 block_height, u32 height, u32 block_depth, u32 depth, u32 tile_width_spacing,
- u8* buffer, std::size_t buffer_size, VAddr addr) {
-
+ u8* buffer, u8* addr) {
GetSwizzleFunction(mode, format)(stride, block_height, height, block_depth, depth,
- tile_width_spacing, buffer, buffer_size, addr);
+ tile_width_spacing, buffer, addr);
}
-void MortonCopyPixels128(u32 width, u32 height, u32 bytes_per_pixel, u32 linear_bytes_per_pixel,
- u8* morton_data, u8* linear_data, bool morton_to_linear) {
+void MortonCopyPixels128(MortonSwizzleMode mode, u32 width, u32 height, u32 bytes_per_pixel,
+ u32 linear_bytes_per_pixel, u8* morton_data, u8* linear_data) {
+ const bool morton_to_linear = mode == MortonSwizzleMode::MortonToLinear;
u8* data_ptrs[2];
for (u32 y = 0; y < height; ++y) {
for (u32 x = 0; x < width; ++x) {
diff --git a/src/video_core/morton.h b/src/video_core/morton.h
index 065f59ce3..ee5b45555 100644
--- a/src/video_core/morton.h
+++ b/src/video_core/morton.h
@@ -13,9 +13,9 @@ enum class MortonSwizzleMode { MortonToLinear, LinearToMorton };
void MortonSwizzle(MortonSwizzleMode mode, VideoCore::Surface::PixelFormat format, u32 stride,
u32 block_height, u32 height, u32 block_depth, u32 depth, u32 tile_width_spacing,
- u8* buffer, std::size_t buffer_size, VAddr addr);
+ u8* buffer, u8* addr);
-void MortonCopyPixels128(u32 width, u32 height, u32 bytes_per_pixel, u32 linear_bytes_per_pixel,
- u8* morton_data, u8* linear_data, bool morton_to_linear);
+void MortonCopyPixels128(MortonSwizzleMode mode, u32 width, u32 height, u32 bytes_per_pixel,
+ u32 linear_bytes_per_pixel, u8* morton_data, u8* linear_data);
} // namespace VideoCore
diff --git a/src/video_core/rasterizer_cache.h b/src/video_core/rasterizer_cache.h
index a7bcf26fb..ecd9986a0 100644
--- a/src/video_core/rasterizer_cache.h
+++ b/src/video_core/rasterizer_cache.h
@@ -4,6 +4,7 @@
#pragma once
+#include <mutex>
#include <set>
#include <unordered_map>
@@ -12,14 +13,26 @@
#include "common/common_types.h"
#include "core/settings.h"
+#include "video_core/gpu.h"
#include "video_core/rasterizer_interface.h"
class RasterizerCacheObject {
public:
+ explicit RasterizerCacheObject(const u8* host_ptr)
+ : host_ptr{host_ptr}, cache_addr{ToCacheAddr(host_ptr)} {}
+
virtual ~RasterizerCacheObject();
+ CacheAddr GetCacheAddr() const {
+ return cache_addr;
+ }
+
+ const u8* GetHostPtr() const {
+ return host_ptr;
+ }
+
/// Gets the address of the shader in guest memory, required for cache management
- virtual VAddr GetAddr() const = 0;
+ virtual VAddr GetCpuAddr() const = 0;
/// Gets the size of the shader in guest memory, required for cache management
virtual std::size_t GetSizeInBytes() const = 0;
@@ -58,6 +71,8 @@ private:
bool is_registered{}; ///< Whether the object is currently registered with the cache
bool is_dirty{}; ///< Whether the object is dirty (out of sync with guest memory)
u64 last_modified_ticks{}; ///< When the object was last modified, used for in-order flushing
+ CacheAddr cache_addr{}; ///< Cache address memory, unique from emulated virtual address space
+ const u8* host_ptr{}; ///< Pointer to the memory backing this cached region
};
template <class T>
@@ -68,7 +83,9 @@ public:
explicit RasterizerCache(VideoCore::RasterizerInterface& rasterizer) : rasterizer{rasterizer} {}
/// Write any cached resources overlapping the specified region back to memory
- void FlushRegion(Tegra::GPUVAddr addr, size_t size) {
+ void FlushRegion(CacheAddr addr, std::size_t size) {
+ std::lock_guard<std::recursive_mutex> lock{mutex};
+
const auto& objects{GetSortedObjectsFromRegion(addr, size)};
for (auto& object : objects) {
FlushObject(object);
@@ -76,7 +93,9 @@ public:
}
/// Mark the specified region as being invalidated
- void InvalidateRegion(VAddr addr, u64 size) {
+ void InvalidateRegion(CacheAddr addr, u64 size) {
+ std::lock_guard<std::recursive_mutex> lock{mutex};
+
const auto& objects{GetSortedObjectsFromRegion(addr, size)};
for (auto& object : objects) {
if (!object->IsRegistered()) {
@@ -89,48 +108,60 @@ public:
/// Invalidates everything in the cache
void InvalidateAll() {
+ std::lock_guard<std::recursive_mutex> lock{mutex};
+
while (interval_cache.begin() != interval_cache.end()) {
Unregister(*interval_cache.begin()->second.begin());
}
}
protected:
- /// Tries to get an object from the cache with the specified address
- T TryGet(VAddr addr) const {
+ /// Tries to get an object from the cache with the specified cache address
+ T TryGet(CacheAddr addr) const {
const auto iter = map_cache.find(addr);
if (iter != map_cache.end())
return iter->second;
return nullptr;
}
+ T TryGet(const void* addr) const {
+ const auto iter = map_cache.find(ToCacheAddr(addr));
+ if (iter != map_cache.end())
+ return iter->second;
+ return nullptr;
+ }
+
/// Register an object into the cache
void Register(const T& object) {
+ std::lock_guard<std::recursive_mutex> lock{mutex};
+
object->SetIsRegistered(true);
interval_cache.add({GetInterval(object), ObjectSet{object}});
- map_cache.insert({object->GetAddr(), object});
- rasterizer.UpdatePagesCachedCount(object->GetAddr(), object->GetSizeInBytes(), 1);
+ map_cache.insert({object->GetCacheAddr(), object});
+ rasterizer.UpdatePagesCachedCount(object->GetCpuAddr(), object->GetSizeInBytes(), 1);
}
/// Unregisters an object from the cache
void Unregister(const T& object) {
- object->SetIsRegistered(false);
- rasterizer.UpdatePagesCachedCount(object->GetAddr(), object->GetSizeInBytes(), -1);
- // Only flush if use_accurate_gpu_emulation is enabled, as it incurs a performance hit
- if (Settings::values.use_accurate_gpu_emulation) {
- FlushObject(object);
- }
+ std::lock_guard<std::recursive_mutex> lock{mutex};
+ object->SetIsRegistered(false);
+ rasterizer.UpdatePagesCachedCount(object->GetCpuAddr(), object->GetSizeInBytes(), -1);
interval_cache.subtract({GetInterval(object), ObjectSet{object}});
- map_cache.erase(object->GetAddr());
+ map_cache.erase(object->GetCacheAddr());
}
/// Returns a ticks counter used for tracking when cached objects were last modified
u64 GetModifiedTicks() {
+ std::lock_guard<std::recursive_mutex> lock{mutex};
+
return ++modified_ticks;
}
/// Flushes the specified object, updating appropriate cache state as needed
void FlushObject(const T& object) {
+ std::lock_guard<std::recursive_mutex> lock{mutex};
+
if (!object->IsDirty()) {
return;
}
@@ -140,7 +171,7 @@ protected:
private:
/// Returns a list of cached objects from the specified memory region, ordered by access time
- std::vector<T> GetSortedObjectsFromRegion(VAddr addr, u64 size) {
+ std::vector<T> GetSortedObjectsFromRegion(CacheAddr addr, u64 size) {
if (size == 0) {
return {};
}
@@ -164,17 +195,18 @@ private:
}
using ObjectSet = std::set<T>;
- using ObjectCache = std::unordered_map<VAddr, T>;
- using IntervalCache = boost::icl::interval_map<VAddr, ObjectSet>;
+ using ObjectCache = std::unordered_map<CacheAddr, T>;
+ using IntervalCache = boost::icl::interval_map<CacheAddr, ObjectSet>;
using ObjectInterval = typename IntervalCache::interval_type;
static auto GetInterval(const T& object) {
- return ObjectInterval::right_open(object->GetAddr(),
- object->GetAddr() + object->GetSizeInBytes());
+ return ObjectInterval::right_open(object->GetCacheAddr(),
+ object->GetCacheAddr() + object->GetSizeInBytes());
}
ObjectCache map_cache;
IntervalCache interval_cache; ///< Cache of objects
u64 modified_ticks{}; ///< Counter of cache state ticks, used for in-order flushing
VideoCore::RasterizerInterface& rasterizer;
+ std::recursive_mutex mutex;
};
diff --git a/src/video_core/rasterizer_interface.h b/src/video_core/rasterizer_interface.h
index 6a1dc9cf6..76e292e87 100644
--- a/src/video_core/rasterizer_interface.h
+++ b/src/video_core/rasterizer_interface.h
@@ -35,14 +35,14 @@ public:
virtual void FlushAll() = 0;
/// Notify rasterizer that any caches of the specified region should be flushed to Switch memory
- virtual void FlushRegion(VAddr addr, u64 size) = 0;
+ virtual void FlushRegion(CacheAddr addr, u64 size) = 0;
/// Notify rasterizer that any caches of the specified region should be invalidated
- virtual void InvalidateRegion(VAddr addr, u64 size) = 0;
+ virtual void InvalidateRegion(CacheAddr addr, u64 size) = 0;
/// Notify rasterizer that any caches of the specified region should be flushed to Switch memory
/// and invalidated
- virtual void FlushAndInvalidateRegion(VAddr addr, u64 size) = 0;
+ virtual void FlushAndInvalidateRegion(CacheAddr addr, u64 size) = 0;
/// Attempt to use a faster method to perform a surface copy
virtual bool AccelerateSurfaceCopy(const Tegra::Engines::Fermi2D::Regs::Surface& src,
@@ -63,7 +63,7 @@ public:
}
/// Increase/decrease the number of object in pages touching the specified region
- virtual void UpdatePagesCachedCount(Tegra::GPUVAddr addr, u64 size, int delta) {}
+ virtual void UpdatePagesCachedCount(VAddr addr, u64 size, int delta) {}
/// Initialize disk cached resources for the game being emulated
virtual void LoadDiskResources(const std::atomic_bool& stop_loading = false,
diff --git a/src/video_core/renderer_opengl/gl_buffer_cache.cpp b/src/video_core/renderer_opengl/gl_buffer_cache.cpp
index b3062e5ba..5048ed6ce 100644
--- a/src/video_core/renderer_opengl/gl_buffer_cache.cpp
+++ b/src/video_core/renderer_opengl/gl_buffer_cache.cpp
@@ -13,24 +13,28 @@
namespace OpenGL {
+CachedBufferEntry::CachedBufferEntry(VAddr cpu_addr, std::size_t size, GLintptr offset,
+ std::size_t alignment, u8* host_ptr)
+ : cpu_addr{cpu_addr}, size{size}, offset{offset}, alignment{alignment}, RasterizerCacheObject{
+ host_ptr} {}
+
OGLBufferCache::OGLBufferCache(RasterizerOpenGL& rasterizer, std::size_t size)
: RasterizerCache{rasterizer}, stream_buffer(size, true) {}
GLintptr OGLBufferCache::UploadMemory(Tegra::GPUVAddr gpu_addr, std::size_t size,
std::size_t alignment, bool cache) {
auto& memory_manager = Core::System::GetInstance().GPU().MemoryManager();
- const auto cpu_addr{memory_manager.GpuToCpuAddress(gpu_addr)};
- ASSERT_MSG(cpu_addr, "Invalid GPU address");
// Cache management is a big overhead, so only cache entries with a given size.
// TODO: Figure out which size is the best for given games.
cache &= size >= 2048;
+ const auto& host_ptr{memory_manager.GetPointer(gpu_addr)};
if (cache) {
- auto entry = TryGet(*cpu_addr);
+ auto entry = TryGet(host_ptr);
if (entry) {
- if (entry->size >= size && entry->alignment == alignment) {
- return entry->offset;
+ if (entry->GetSize() >= size && entry->GetAlignment() == alignment) {
+ return entry->GetOffset();
}
Unregister(entry);
}
@@ -39,17 +43,17 @@ GLintptr OGLBufferCache::UploadMemory(Tegra::GPUVAddr gpu_addr, std::size_t size
AlignBuffer(alignment);
const GLintptr uploaded_offset = buffer_offset;
- Memory::ReadBlock(*cpu_addr, buffer_ptr, size);
+ if (!host_ptr) {
+ return uploaded_offset;
+ }
+ std::memcpy(buffer_ptr, host_ptr, size);
buffer_ptr += size;
buffer_offset += size;
if (cache) {
- auto entry = std::make_shared<CachedBufferEntry>();
- entry->offset = uploaded_offset;
- entry->size = size;
- entry->alignment = alignment;
- entry->addr = *cpu_addr;
+ auto entry = std::make_shared<CachedBufferEntry>(
+ *memory_manager.GpuToCpuAddress(gpu_addr), size, uploaded_offset, alignment, host_ptr);
Register(entry);
}
diff --git a/src/video_core/renderer_opengl/gl_buffer_cache.h b/src/video_core/renderer_opengl/gl_buffer_cache.h
index c11acfb79..1de1f84ae 100644
--- a/src/video_core/renderer_opengl/gl_buffer_cache.h
+++ b/src/video_core/renderer_opengl/gl_buffer_cache.h
@@ -17,22 +17,39 @@ namespace OpenGL {
class RasterizerOpenGL;
-struct CachedBufferEntry final : public RasterizerCacheObject {
- VAddr GetAddr() const override {
- return addr;
+class CachedBufferEntry final : public RasterizerCacheObject {
+public:
+ explicit CachedBufferEntry(VAddr cpu_addr, std::size_t size, GLintptr offset,
+ std::size_t alignment, u8* host_ptr);
+
+ VAddr GetCpuAddr() const override {
+ return cpu_addr;
}
std::size_t GetSizeInBytes() const override {
return size;
}
+ std::size_t GetSize() const {
+ return size;
+ }
+
+ GLintptr GetOffset() const {
+ return offset;
+ }
+
+ std::size_t GetAlignment() const {
+ return alignment;
+ }
+
// We do not have to flush this cache as things in it are never modified by us.
void Flush() override {}
- VAddr addr;
- std::size_t size;
- GLintptr offset;
- std::size_t alignment;
+private:
+ VAddr cpu_addr{};
+ std::size_t size{};
+ GLintptr offset{};
+ std::size_t alignment{};
};
class OGLBufferCache final : public RasterizerCache<std::shared_ptr<CachedBufferEntry>> {
diff --git a/src/video_core/renderer_opengl/gl_global_cache.cpp b/src/video_core/renderer_opengl/gl_global_cache.cpp
index c7f32feaa..c8dbcacbd 100644
--- a/src/video_core/renderer_opengl/gl_global_cache.cpp
+++ b/src/video_core/renderer_opengl/gl_global_cache.cpp
@@ -7,7 +7,6 @@
#include "common/assert.h"
#include "common/logging/log.h"
#include "core/core.h"
-#include "core/memory.h"
#include "video_core/renderer_opengl/gl_global_cache.h"
#include "video_core/renderer_opengl/gl_rasterizer.h"
#include "video_core/renderer_opengl/gl_shader_decompiler.h"
@@ -15,12 +14,13 @@
namespace OpenGL {
-CachedGlobalRegion::CachedGlobalRegion(VAddr addr, u32 size) : addr{addr}, size{size} {
+CachedGlobalRegion::CachedGlobalRegion(VAddr cpu_addr, u32 size, u8* host_ptr)
+ : cpu_addr{cpu_addr}, size{size}, RasterizerCacheObject{host_ptr} {
buffer.Create();
// Bind and unbind the buffer so it gets allocated by the driver
glBindBuffer(GL_SHADER_STORAGE_BUFFER, buffer.handle);
glBindBuffer(GL_SHADER_STORAGE_BUFFER, 0);
- LabelGLObject(GL_BUFFER, buffer.handle, addr, "GlobalMemory");
+ LabelGLObject(GL_BUFFER, buffer.handle, cpu_addr, "GlobalMemory");
}
void CachedGlobalRegion::Reload(u32 size_) {
@@ -35,10 +35,10 @@ void CachedGlobalRegion::Reload(u32 size_) {
// TODO(Rodrigo): Get rid of Memory::GetPointer with a staging buffer
glBindBuffer(GL_SHADER_STORAGE_BUFFER, buffer.handle);
- glBufferData(GL_SHADER_STORAGE_BUFFER, size, Memory::GetPointer(addr), GL_DYNAMIC_DRAW);
+ glBufferData(GL_SHADER_STORAGE_BUFFER, size, GetHostPtr(), GL_DYNAMIC_DRAW);
}
-GlobalRegion GlobalRegionCacheOpenGL::TryGetReservedGlobalRegion(VAddr addr, u32 size) const {
+GlobalRegion GlobalRegionCacheOpenGL::TryGetReservedGlobalRegion(CacheAddr addr, u32 size) const {
const auto search{reserve.find(addr)};
if (search == reserve.end()) {
return {};
@@ -46,19 +46,22 @@ GlobalRegion GlobalRegionCacheOpenGL::TryGetReservedGlobalRegion(VAddr addr, u32
return search->second;
}
-GlobalRegion GlobalRegionCacheOpenGL::GetUncachedGlobalRegion(VAddr addr, u32 size) {
- GlobalRegion region{TryGetReservedGlobalRegion(addr, size)};
+GlobalRegion GlobalRegionCacheOpenGL::GetUncachedGlobalRegion(Tegra::GPUVAddr addr, u32 size,
+ u8* host_ptr) {
+ GlobalRegion region{TryGetReservedGlobalRegion(ToCacheAddr(host_ptr), size)};
if (!region) {
// No reserved surface available, create a new one and reserve it
- region = std::make_shared<CachedGlobalRegion>(addr, size);
+ auto& memory_manager{Core::System::GetInstance().GPU().MemoryManager()};
+ const auto cpu_addr = *memory_manager.GpuToCpuAddress(addr);
+ region = std::make_shared<CachedGlobalRegion>(cpu_addr, size, host_ptr);
ReserveGlobalRegion(region);
}
region->Reload(size);
return region;
}
-void GlobalRegionCacheOpenGL::ReserveGlobalRegion(const GlobalRegion& region) {
- reserve[region->GetAddr()] = region;
+void GlobalRegionCacheOpenGL::ReserveGlobalRegion(GlobalRegion region) {
+ reserve.insert_or_assign(region->GetCacheAddr(), std::move(region));
}
GlobalRegionCacheOpenGL::GlobalRegionCacheOpenGL(RasterizerOpenGL& rasterizer)
@@ -69,22 +72,20 @@ GlobalRegion GlobalRegionCacheOpenGL::GetGlobalRegion(
Tegra::Engines::Maxwell3D::Regs::ShaderStage stage) {
auto& gpu{Core::System::GetInstance().GPU()};
- const auto cbufs = gpu.Maxwell3D().state.shader_stages[static_cast<u64>(stage)];
- const auto cbuf_addr = gpu.MemoryManager().GpuToCpuAddress(
- cbufs.const_buffers[global_region.GetCbufIndex()].address + global_region.GetCbufOffset());
- ASSERT(cbuf_addr);
-
- const auto actual_addr_gpu = Memory::Read64(*cbuf_addr);
- const auto size = Memory::Read32(*cbuf_addr + 8);
- const auto actual_addr = gpu.MemoryManager().GpuToCpuAddress(actual_addr_gpu);
- ASSERT(actual_addr);
+ auto& memory_manager{gpu.MemoryManager()};
+ const auto cbufs{gpu.Maxwell3D().state.shader_stages[static_cast<u64>(stage)]};
+ const auto addr{cbufs.const_buffers[global_region.GetCbufIndex()].address +
+ global_region.GetCbufOffset()};
+ const auto actual_addr{memory_manager.Read64(addr)};
+ const auto size{memory_manager.Read32(addr + 8)};
// Look up global region in the cache based on address
- GlobalRegion region = TryGet(*actual_addr);
+ const auto& host_ptr{memory_manager.GetPointer(actual_addr)};
+ GlobalRegion region{TryGet(host_ptr)};
if (!region) {
// No global region found - create a new one
- region = GetUncachedGlobalRegion(*actual_addr, size);
+ region = GetUncachedGlobalRegion(actual_addr, size, host_ptr);
Register(region);
}
diff --git a/src/video_core/renderer_opengl/gl_global_cache.h b/src/video_core/renderer_opengl/gl_global_cache.h
index 37830bb7c..a840491f7 100644
--- a/src/video_core/renderer_opengl/gl_global_cache.h
+++ b/src/video_core/renderer_opengl/gl_global_cache.h
@@ -27,15 +27,13 @@ using GlobalRegion = std::shared_ptr<CachedGlobalRegion>;
class CachedGlobalRegion final : public RasterizerCacheObject {
public:
- explicit CachedGlobalRegion(VAddr addr, u32 size);
+ explicit CachedGlobalRegion(VAddr cpu_addr, u32 size, u8* host_ptr);
- /// Gets the address of the shader in guest memory, required for cache management
- VAddr GetAddr() const {
- return addr;
+ VAddr GetCpuAddr() const override {
+ return cpu_addr;
}
- /// Gets the size of the shader in guest memory, required for cache management
- std::size_t GetSizeInBytes() const {
+ std::size_t GetSizeInBytes() const override {
return size;
}
@@ -53,9 +51,8 @@ public:
}
private:
- VAddr addr{};
+ VAddr cpu_addr{};
u32 size{};
-
OGLBuffer buffer;
};
@@ -68,11 +65,11 @@ public:
Tegra::Engines::Maxwell3D::Regs::ShaderStage stage);
private:
- GlobalRegion TryGetReservedGlobalRegion(VAddr addr, u32 size) const;
- GlobalRegion GetUncachedGlobalRegion(VAddr addr, u32 size);
- void ReserveGlobalRegion(const GlobalRegion& region);
+ GlobalRegion TryGetReservedGlobalRegion(CacheAddr addr, u32 size) const;
+ GlobalRegion GetUncachedGlobalRegion(Tegra::GPUVAddr addr, u32 size, u8* host_ptr);
+ void ReserveGlobalRegion(GlobalRegion region);
- std::unordered_map<VAddr, GlobalRegion> reserve;
+ std::unordered_map<CacheAddr, GlobalRegion> reserve;
};
} // namespace OpenGL
diff --git a/src/video_core/renderer_opengl/gl_primitive_assembler.cpp b/src/video_core/renderer_opengl/gl_primitive_assembler.cpp
index 77d5cedd2..75d816795 100644
--- a/src/video_core/renderer_opengl/gl_primitive_assembler.cpp
+++ b/src/video_core/renderer_opengl/gl_primitive_assembler.cpp
@@ -46,10 +46,7 @@ GLintptr PrimitiveAssembler::MakeQuadIndexed(Tegra::GPUVAddr gpu_addr, std::size
auto [dst_pointer, index_offset] = buffer_cache.ReserveMemory(map_size);
auto& memory_manager = Core::System::GetInstance().GPU().MemoryManager();
- const auto cpu_addr{memory_manager.GpuToCpuAddress(gpu_addr)};
- ASSERT_MSG(cpu_addr, "Invalid GPU address");
-
- const u8* source{Memory::GetPointer(*cpu_addr)};
+ const u8* source{memory_manager.GetPointer(gpu_addr)};
for (u32 primitive = 0; primitive < count / 4; ++primitive) {
for (std::size_t i = 0; i < TRIANGLES_PER_QUAD; ++i) {
@@ -64,4 +61,4 @@ GLintptr PrimitiveAssembler::MakeQuadIndexed(Tegra::GPUVAddr gpu_addr, std::size
return index_offset;
}
-} // namespace OpenGL \ No newline at end of file
+} // namespace OpenGL
diff --git a/src/video_core/renderer_opengl/gl_rasterizer.cpp b/src/video_core/renderer_opengl/gl_rasterizer.cpp
index 824863561..198c54872 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer.cpp
+++ b/src/video_core/renderer_opengl/gl_rasterizer.cpp
@@ -102,8 +102,9 @@ struct FramebufferCacheKey {
RasterizerOpenGL::RasterizerOpenGL(Core::Frontend::EmuWindow& window, Core::System& system,
ScreenInfo& info)
- : res_cache{*this}, shader_cache{*this, system}, global_cache{*this}, emu_window{window},
- screen_info{info}, buffer_cache(*this, STREAM_BUFFER_SIZE) {
+ : res_cache{*this}, shader_cache{*this, system}, global_cache{*this},
+ emu_window{window}, system{system}, screen_info{info},
+ buffer_cache(*this, STREAM_BUFFER_SIZE) {
// Create sampler objects
for (std::size_t i = 0; i < texture_samplers.size(); ++i) {
texture_samplers[i].Create();
@@ -138,7 +139,7 @@ void RasterizerOpenGL::CheckExtensions() {
}
GLuint RasterizerOpenGL::SetupVertexFormat() {
- auto& gpu = Core::System::GetInstance().GPU().Maxwell3D();
+ auto& gpu = system.GPU().Maxwell3D();
const auto& regs = gpu.regs;
if (!gpu.dirty_flags.vertex_attrib_format) {
@@ -207,7 +208,7 @@ GLuint RasterizerOpenGL::SetupVertexFormat() {
}
void RasterizerOpenGL::SetupVertexBuffer(GLuint vao) {
- auto& gpu = Core::System::GetInstance().GPU().Maxwell3D();
+ auto& gpu = system.GPU().Maxwell3D();
const auto& regs = gpu.regs;
if (gpu.dirty_flags.vertex_array.none())
@@ -248,7 +249,7 @@ void RasterizerOpenGL::SetupVertexBuffer(GLuint vao) {
}
DrawParameters RasterizerOpenGL::SetupDraw() {
- const auto& gpu = Core::System::GetInstance().GPU().Maxwell3D();
+ const auto& gpu = system.GPU().Maxwell3D();
const auto& regs = gpu.regs;
const bool is_indexed = accelerate_draw == AccelDraw::Indexed;
@@ -297,7 +298,7 @@ DrawParameters RasterizerOpenGL::SetupDraw() {
void RasterizerOpenGL::SetupShaders(GLenum primitive_mode) {
MICROPROFILE_SCOPE(OpenGL_Shader);
- auto& gpu = Core::System::GetInstance().GPU().Maxwell3D();
+ auto& gpu = system.GPU().Maxwell3D();
BaseBindings base_bindings;
std::array<bool, Maxwell::NumClipDistances> clip_distances{};
@@ -413,7 +414,7 @@ void RasterizerOpenGL::SetupCachedFramebuffer(const FramebufferCacheKey& fbkey,
}
std::size_t RasterizerOpenGL::CalculateVertexArraysSize() const {
- const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
+ const auto& regs = system.GPU().Maxwell3D().regs;
std::size_t size = 0;
for (u32 index = 0; index < Maxwell::NumVertexArrays; ++index) {
@@ -431,7 +432,7 @@ std::size_t RasterizerOpenGL::CalculateVertexArraysSize() const {
}
std::size_t RasterizerOpenGL::CalculateIndexBufferSize() const {
- const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
+ const auto& regs = system.GPU().Maxwell3D().regs;
return static_cast<std::size_t>(regs.index_array.count) *
static_cast<std::size_t>(regs.index_array.FormatSizeInBytes());
@@ -448,7 +449,7 @@ static constexpr auto RangeFromInterval(Map& map, const Interval& interval) {
return boost::make_iterator_range(map.equal_range(interval));
}
-void RasterizerOpenGL::UpdatePagesCachedCount(Tegra::GPUVAddr addr, u64 size, int delta) {
+void RasterizerOpenGL::UpdatePagesCachedCount(VAddr addr, u64 size, int delta) {
const u64 page_start{addr >> Memory::PAGE_BITS};
const u64 page_end{(addr + size + Memory::PAGE_SIZE - 1) >> Memory::PAGE_BITS};
@@ -487,7 +488,7 @@ std::pair<bool, bool> RasterizerOpenGL::ConfigureFramebuffers(
OpenGLState& current_state, bool using_color_fb, bool using_depth_fb, bool preserve_contents,
std::optional<std::size_t> single_color_target) {
MICROPROFILE_SCOPE(OpenGL_Framebuffer);
- auto& gpu = Core::System::GetInstance().GPU().Maxwell3D();
+ auto& gpu = system.GPU().Maxwell3D();
const auto& regs = gpu.regs;
const FramebufferConfigState fb_config_state{using_color_fb, using_depth_fb, preserve_contents,
@@ -581,7 +582,7 @@ void RasterizerOpenGL::Clear() {
const auto prev_state{state};
SCOPE_EXIT({ prev_state.Apply(); });
- const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
+ const auto& regs = system.GPU().Maxwell3D().regs;
bool use_color{};
bool use_depth{};
bool use_stencil{};
@@ -672,7 +673,7 @@ void RasterizerOpenGL::DrawArrays() {
return;
MICROPROFILE_SCOPE(OpenGL_Drawing);
- auto& gpu = Core::System::GetInstance().GPU().Maxwell3D();
+ auto& gpu = system.GPU().Maxwell3D();
const auto& regs = gpu.regs;
ConfigureFramebuffers(state);
@@ -746,20 +747,26 @@ void RasterizerOpenGL::DrawArrays() {
void RasterizerOpenGL::FlushAll() {}
-void RasterizerOpenGL::FlushRegion(VAddr addr, u64 size) {
+void RasterizerOpenGL::FlushRegion(CacheAddr addr, u64 size) {
MICROPROFILE_SCOPE(OpenGL_CacheManagement);
+ if (!addr || !size) {
+ return;
+ }
res_cache.FlushRegion(addr, size);
}
-void RasterizerOpenGL::InvalidateRegion(VAddr addr, u64 size) {
+void RasterizerOpenGL::InvalidateRegion(CacheAddr addr, u64 size) {
MICROPROFILE_SCOPE(OpenGL_CacheManagement);
+ if (!addr || !size) {
+ return;
+ }
res_cache.InvalidateRegion(addr, size);
shader_cache.InvalidateRegion(addr, size);
global_cache.InvalidateRegion(addr, size);
buffer_cache.InvalidateRegion(addr, size);
}
-void RasterizerOpenGL::FlushAndInvalidateRegion(VAddr addr, u64 size) {
+void RasterizerOpenGL::FlushAndInvalidateRegion(CacheAddr addr, u64 size) {
FlushRegion(addr, size);
InvalidateRegion(addr, size);
}
@@ -781,7 +788,7 @@ bool RasterizerOpenGL::AccelerateDisplay(const Tegra::FramebufferConfig& config,
MICROPROFILE_SCOPE(OpenGL_CacheManagement);
- const auto& surface{res_cache.TryFindFramebufferSurface(framebuffer_addr)};
+ const auto& surface{res_cache.TryFindFramebufferSurface(Memory::GetPointer(framebuffer_addr))};
if (!surface) {
return {};
}
@@ -892,7 +899,7 @@ void RasterizerOpenGL::SetupConstBuffers(Tegra::Engines::Maxwell3D::Regs::Shader
const Shader& shader, GLuint program_handle,
BaseBindings base_bindings) {
MICROPROFILE_SCOPE(OpenGL_UBO);
- const auto& gpu = Core::System::GetInstance().GPU();
+ const auto& gpu = system.GPU();
const auto& maxwell3d = gpu.Maxwell3D();
const auto& shader_stage = maxwell3d.state.shader_stages[static_cast<std::size_t>(stage)];
const auto& entries = shader->GetShaderEntries().const_buffers;
@@ -971,7 +978,7 @@ void RasterizerOpenGL::SetupGlobalRegions(Tegra::Engines::Maxwell3D::Regs::Shade
void RasterizerOpenGL::SetupTextures(Maxwell::ShaderStage stage, const Shader& shader,
GLuint program_handle, BaseBindings base_bindings) {
MICROPROFILE_SCOPE(OpenGL_Texture);
- const auto& gpu = Core::System::GetInstance().GPU();
+ const auto& gpu = system.GPU();
const auto& maxwell3d = gpu.Maxwell3D();
const auto& entries = shader->GetShaderEntries().samplers;
@@ -998,7 +1005,7 @@ void RasterizerOpenGL::SetupTextures(Maxwell::ShaderStage stage, const Shader& s
}
void RasterizerOpenGL::SyncViewport(OpenGLState& current_state) {
- const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
+ const auto& regs = system.GPU().Maxwell3D().regs;
const bool geometry_shaders_enabled =
regs.IsShaderConfigEnabled(static_cast<size_t>(Maxwell::ShaderProgram::Geometry));
const std::size_t viewport_count =
@@ -1021,7 +1028,7 @@ void RasterizerOpenGL::SyncViewport(OpenGLState& current_state) {
void RasterizerOpenGL::SyncClipEnabled(
const std::array<bool, Maxwell::Regs::NumClipDistances>& clip_mask) {
- const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
+ const auto& regs = system.GPU().Maxwell3D().regs;
const std::array<bool, Maxwell::Regs::NumClipDistances> reg_state{
regs.clip_distance_enabled.c0 != 0, regs.clip_distance_enabled.c1 != 0,
regs.clip_distance_enabled.c2 != 0, regs.clip_distance_enabled.c3 != 0,
@@ -1038,7 +1045,7 @@ void RasterizerOpenGL::SyncClipCoef() {
}
void RasterizerOpenGL::SyncCullMode() {
- const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
+ const auto& regs = system.GPU().Maxwell3D().regs;
state.cull.enabled = regs.cull.enabled != 0;
@@ -1062,14 +1069,14 @@ void RasterizerOpenGL::SyncCullMode() {
}
void RasterizerOpenGL::SyncPrimitiveRestart() {
- const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
+ const auto& regs = system.GPU().Maxwell3D().regs;
state.primitive_restart.enabled = regs.primitive_restart.enabled;
state.primitive_restart.index = regs.primitive_restart.index;
}
void RasterizerOpenGL::SyncDepthTestState() {
- const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
+ const auto& regs = system.GPU().Maxwell3D().regs;
state.depth.test_enabled = regs.depth_test_enable != 0;
state.depth.write_mask = regs.depth_write_enabled ? GL_TRUE : GL_FALSE;
@@ -1081,7 +1088,7 @@ void RasterizerOpenGL::SyncDepthTestState() {
}
void RasterizerOpenGL::SyncStencilTestState() {
- const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
+ const auto& regs = system.GPU().Maxwell3D().regs;
state.stencil.test_enabled = regs.stencil_enable != 0;
if (!regs.stencil_enable) {
@@ -1115,7 +1122,7 @@ void RasterizerOpenGL::SyncStencilTestState() {
}
void RasterizerOpenGL::SyncColorMask() {
- const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
+ const auto& regs = system.GPU().Maxwell3D().regs;
const std::size_t count =
regs.independent_blend_enable ? Tegra::Engines::Maxwell3D::Regs::NumRenderTargets : 1;
for (std::size_t i = 0; i < count; i++) {
@@ -1129,18 +1136,18 @@ void RasterizerOpenGL::SyncColorMask() {
}
void RasterizerOpenGL::SyncMultiSampleState() {
- const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
+ const auto& regs = system.GPU().Maxwell3D().regs;
state.multisample_control.alpha_to_coverage = regs.multisample_control.alpha_to_coverage != 0;
state.multisample_control.alpha_to_one = regs.multisample_control.alpha_to_one != 0;
}
void RasterizerOpenGL::SyncFragmentColorClampState() {
- const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
+ const auto& regs = system.GPU().Maxwell3D().regs;
state.fragment_color_clamp.enabled = regs.frag_color_clamp != 0;
}
void RasterizerOpenGL::SyncBlendState() {
- const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
+ const auto& regs = system.GPU().Maxwell3D().regs;
state.blend_color.red = regs.blend_color.r;
state.blend_color.green = regs.blend_color.g;
@@ -1182,7 +1189,7 @@ void RasterizerOpenGL::SyncBlendState() {
}
void RasterizerOpenGL::SyncLogicOpState() {
- const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
+ const auto& regs = system.GPU().Maxwell3D().regs;
state.logic_op.enabled = regs.logic_op.enable != 0;
@@ -1196,7 +1203,7 @@ void RasterizerOpenGL::SyncLogicOpState() {
}
void RasterizerOpenGL::SyncScissorTest(OpenGLState& current_state) {
- const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
+ const auto& regs = system.GPU().Maxwell3D().regs;
const bool geometry_shaders_enabled =
regs.IsShaderConfigEnabled(static_cast<size_t>(Maxwell::ShaderProgram::Geometry));
const std::size_t viewport_count =
@@ -1218,17 +1225,17 @@ void RasterizerOpenGL::SyncScissorTest(OpenGLState& current_state) {
}
void RasterizerOpenGL::SyncTransformFeedback() {
- const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
+ const auto& regs = system.GPU().Maxwell3D().regs;
UNIMPLEMENTED_IF_MSG(regs.tfb_enabled != 0, "Transform feedbacks are not implemented");
}
void RasterizerOpenGL::SyncPointState() {
- const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
+ const auto& regs = system.GPU().Maxwell3D().regs;
state.point.size = regs.point_size;
}
void RasterizerOpenGL::SyncPolygonOffset() {
- const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
+ const auto& regs = system.GPU().Maxwell3D().regs;
state.polygon_offset.fill_enable = regs.polygon_offset_fill_enable != 0;
state.polygon_offset.line_enable = regs.polygon_offset_line_enable != 0;
state.polygon_offset.point_enable = regs.polygon_offset_point_enable != 0;
@@ -1238,7 +1245,7 @@ void RasterizerOpenGL::SyncPolygonOffset() {
}
void RasterizerOpenGL::CheckAlphaTests() {
- const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
+ const auto& regs = system.GPU().Maxwell3D().regs;
UNIMPLEMENTED_IF_MSG(regs.alpha_test_enabled != 0 && regs.rt_control.count > 1,
"Alpha Testing is enabled with more than one rendertarget");
}
diff --git a/src/video_core/renderer_opengl/gl_rasterizer.h b/src/video_core/renderer_opengl/gl_rasterizer.h
index 7e63f8008..30f3e8acb 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer.h
+++ b/src/video_core/renderer_opengl/gl_rasterizer.h
@@ -57,9 +57,9 @@ public:
void DrawArrays() override;
void Clear() override;
void FlushAll() override;
- void FlushRegion(VAddr addr, u64 size) override;
- void InvalidateRegion(VAddr addr, u64 size) override;
- void FlushAndInvalidateRegion(VAddr addr, u64 size) override;
+ void FlushRegion(CacheAddr addr, u64 size) override;
+ void InvalidateRegion(CacheAddr addr, u64 size) override;
+ void FlushAndInvalidateRegion(CacheAddr addr, u64 size) override;
bool AccelerateSurfaceCopy(const Tegra::Engines::Fermi2D::Regs::Surface& src,
const Tegra::Engines::Fermi2D::Regs::Surface& dst,
const Common::Rectangle<u32>& src_rect,
@@ -67,7 +67,7 @@ public:
bool AccelerateDisplay(const Tegra::FramebufferConfig& config, VAddr framebuffer_addr,
u32 pixel_stride) override;
bool AccelerateDrawBatch(bool is_indexed) override;
- void UpdatePagesCachedCount(Tegra::GPUVAddr addr, u64 size, int delta) override;
+ void UpdatePagesCachedCount(VAddr addr, u64 size, int delta) override;
void LoadDiskResources(const std::atomic_bool& stop_loading,
const VideoCore::DiskResourceLoadCallback& callback) override;
@@ -215,6 +215,7 @@ private:
GlobalRegionCacheOpenGL global_cache;
Core::Frontend::EmuWindow& emu_window;
+ Core::System& system;
ScreenInfo& screen_info;
diff --git a/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp b/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp
index e9eb6e921..57329cd61 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp
+++ b/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp
@@ -57,10 +57,9 @@ static void ApplyTextureDefaults(GLuint texture, u32 max_mip_level) {
void SurfaceParams::InitCacheParameters(Tegra::GPUVAddr gpu_addr_) {
auto& memory_manager{Core::System::GetInstance().GPU().MemoryManager()};
- const auto cpu_addr{memory_manager.GpuToCpuAddress(gpu_addr_)};
- addr = cpu_addr ? *cpu_addr : 0;
gpu_addr = gpu_addr_;
+ host_ptr = memory_manager.GetPointer(gpu_addr_);
size_in_bytes = SizeInBytesRaw();
if (IsPixelFormatASTC(pixel_format)) {
@@ -446,7 +445,7 @@ void SwizzleFunc(const MortonSwizzleMode& mode, const SurfaceParams& params,
MortonSwizzle(mode, params.pixel_format, params.MipWidth(mip_level),
params.MipBlockHeight(mip_level), params.MipHeight(mip_level),
params.MipBlockDepth(mip_level), 1, params.tile_width_spacing,
- gl_buffer.data() + offset_gl, gl_size, params.addr + offset);
+ gl_buffer.data() + offset_gl, params.host_ptr + offset);
offset += layer_size;
offset_gl += gl_size;
}
@@ -455,7 +454,7 @@ void SwizzleFunc(const MortonSwizzleMode& mode, const SurfaceParams& params,
MortonSwizzle(mode, params.pixel_format, params.MipWidth(mip_level),
params.MipBlockHeight(mip_level), params.MipHeight(mip_level),
params.MipBlockDepth(mip_level), depth, params.tile_width_spacing,
- gl_buffer.data(), gl_buffer.size(), params.addr + offset);
+ gl_buffer.data(), params.host_ptr + offset);
}
}
@@ -513,9 +512,9 @@ void RasterizerCacheOpenGL::CopySurface(const Surface& src_surface, const Surfac
"reinterpretation but the texture is tiled.");
}
const std::size_t remaining_size = dst_params.size_in_bytes - src_params.size_in_bytes;
-
+ auto& memory_manager{Core::System::GetInstance().GPU().MemoryManager()};
glBufferSubData(GL_PIXEL_PACK_BUFFER, src_params.size_in_bytes, remaining_size,
- Memory::GetPointer(dst_params.addr + src_params.size_in_bytes));
+ memory_manager.GetPointer(dst_params.gpu_addr + src_params.size_in_bytes));
}
glBindBuffer(GL_PIXEL_PACK_BUFFER, 0);
@@ -563,8 +562,8 @@ void RasterizerCacheOpenGL::CopySurface(const Surface& src_surface, const Surfac
}
CachedSurface::CachedSurface(const SurfaceParams& params)
- : params(params), gl_target(SurfaceTargetToGL(params.target)),
- cached_size_in_bytes(params.size_in_bytes) {
+ : params{params}, gl_target{SurfaceTargetToGL(params.target)},
+ cached_size_in_bytes{params.size_in_bytes}, RasterizerCacheObject{params.host_ptr} {
texture.Create(gl_target);
// TODO(Rodrigo): Using params.GetRect() returns a different size than using its Mip*(0)
@@ -603,7 +602,7 @@ CachedSurface::CachedSurface(const SurfaceParams& params)
ApplyTextureDefaults(texture.handle, params.max_mip_level);
- OpenGL::LabelGLObject(GL_TEXTURE, texture.handle, params.addr, params.IdentityString());
+ OpenGL::LabelGLObject(GL_TEXTURE, texture.handle, params.gpu_addr, params.IdentityString());
// Clamp size to mapped GPU memory region
// TODO(bunnei): Super Mario Odyssey maps a 0x40000 byte region and then uses it for a 0x80000
@@ -616,6 +615,8 @@ CachedSurface::CachedSurface(const SurfaceParams& params)
LOG_ERROR(HW_GPU, "Surface size {} exceeds region size {}", params.size_in_bytes, max_size);
cached_size_in_bytes = max_size;
}
+
+ cpu_addr = *memory_manager.GpuToCpuAddress(params.gpu_addr);
}
MICROPROFILE_DEFINE(OpenGL_SurfaceLoad, "OpenGL", "Surface Load", MP_RGB(128, 192, 64));
@@ -633,10 +634,9 @@ void CachedSurface::LoadGLBuffer() {
const u32 bpp = params.GetFormatBpp() / 8;
const u32 copy_size = params.width * bpp;
if (params.pitch == copy_size) {
- std::memcpy(gl_buffer[0].data(), Memory::GetPointer(params.addr),
- params.size_in_bytes_gl);
+ std::memcpy(gl_buffer[0].data(), params.host_ptr, params.size_in_bytes_gl);
} else {
- const u8* start = Memory::GetPointer(params.addr);
+ const u8* start{params.host_ptr};
u8* write_to = gl_buffer[0].data();
for (u32 h = params.height; h > 0; h--) {
std::memcpy(write_to, start, copy_size);
@@ -680,8 +680,6 @@ void CachedSurface::FlushGLBuffer() {
glPixelStorei(GL_PACK_ROW_LENGTH, 0);
Tegra::Texture::ConvertFromHostToGuest(gl_buffer[0].data(), params.pixel_format, params.width,
params.height, params.depth, true, true);
- const u8* const texture_src_data = Memory::GetPointer(params.addr);
- ASSERT(texture_src_data);
if (params.is_tiled) {
ASSERT_MSG(params.block_width == 1, "Block width is defined as {} on texture type {}",
params.block_width, static_cast<u32>(params.target));
@@ -691,9 +689,9 @@ void CachedSurface::FlushGLBuffer() {
const u32 bpp = params.GetFormatBpp() / 8;
const u32 copy_size = params.width * bpp;
if (params.pitch == copy_size) {
- std::memcpy(Memory::GetPointer(params.addr), gl_buffer[0].data(), GetSizeInBytes());
+ std::memcpy(params.host_ptr, gl_buffer[0].data(), GetSizeInBytes());
} else {
- u8* start = Memory::GetPointer(params.addr);
+ u8* start{params.host_ptr};
const u8* read_to = gl_buffer[0].data();
for (u32 h = params.height; h > 0; h--) {
std::memcpy(start, read_to, copy_size);
@@ -927,12 +925,12 @@ void RasterizerCacheOpenGL::LoadSurface(const Surface& surface) {
}
Surface RasterizerCacheOpenGL::GetSurface(const SurfaceParams& params, bool preserve_contents) {
- if (params.addr == 0 || params.height * params.width == 0) {
+ if (params.gpu_addr == 0 || params.height * params.width == 0) {
return {};
}
// Look up surface in the cache based on address
- Surface surface{TryGet(params.addr)};
+ Surface surface{TryGet(params.host_ptr)};
if (surface) {
if (surface->GetSurfaceParams().IsCompatibleSurface(params)) {
// Use the cached surface as-is unless it's not synced with memory
@@ -981,14 +979,16 @@ void RasterizerCacheOpenGL::FastLayeredCopySurface(const Surface& src_surface,
const Surface& dst_surface) {
const auto& init_params{src_surface->GetSurfaceParams()};
const auto& dst_params{dst_surface->GetSurfaceParams()};
- VAddr address = init_params.addr;
- const std::size_t layer_size = dst_params.LayerMemorySize();
+ auto& memory_manager{Core::System::GetInstance().GPU().MemoryManager()};
+ Tegra::GPUVAddr address{init_params.gpu_addr};
+ const std::size_t layer_size{dst_params.LayerMemorySize()};
for (u32 layer = 0; layer < dst_params.depth; layer++) {
for (u32 mipmap = 0; mipmap < dst_params.max_mip_level; mipmap++) {
- const VAddr sub_address = address + dst_params.GetMipmapLevelOffset(mipmap);
- const Surface& copy = TryGet(sub_address);
- if (!copy)
+ const Tegra::GPUVAddr sub_address{address + dst_params.GetMipmapLevelOffset(mipmap)};
+ const Surface& copy{TryGet(memory_manager.GetPointer(sub_address))};
+ if (!copy) {
continue;
+ }
const auto& src_params{copy->GetSurfaceParams()};
const u32 width{std::min(src_params.width, dst_params.MipWidth(mipmap))};
const u32 height{std::min(src_params.height, dst_params.MipHeight(mipmap))};
@@ -1163,7 +1163,8 @@ void RasterizerCacheOpenGL::AccurateCopySurface(const Surface& src_surface,
const auto& dst_params{dst_surface->GetSurfaceParams()};
// Flush enough memory for both the source and destination surface
- FlushRegion(src_params.addr, std::max(src_params.MemorySize(), dst_params.MemorySize()));
+ FlushRegion(ToCacheAddr(src_params.host_ptr),
+ std::max(src_params.MemorySize(), dst_params.MemorySize()));
LoadSurface(dst_surface);
}
@@ -1215,8 +1216,8 @@ Surface RasterizerCacheOpenGL::RecreateSurface(const Surface& old_surface,
return new_surface;
}
-Surface RasterizerCacheOpenGL::TryFindFramebufferSurface(VAddr addr) const {
- return TryGet(addr);
+Surface RasterizerCacheOpenGL::TryFindFramebufferSurface(const u8* host_ptr) const {
+ return TryGet(host_ptr);
}
void RasterizerCacheOpenGL::ReserveSurface(const Surface& surface) {
@@ -1243,9 +1244,10 @@ static std::optional<u32> TryFindBestMipMap(std::size_t memory, const SurfacePar
return {};
}
-static std::optional<u32> TryFindBestLayer(VAddr addr, const SurfaceParams params, u32 mipmap) {
- const std::size_t size = params.LayerMemorySize();
- VAddr start = params.addr + params.GetMipmapLevelOffset(mipmap);
+static std::optional<u32> TryFindBestLayer(Tegra::GPUVAddr addr, const SurfaceParams params,
+ u32 mipmap) {
+ const std::size_t size{params.LayerMemorySize()};
+ Tegra::GPUVAddr start{params.gpu_addr + params.GetMipmapLevelOffset(mipmap)};
for (u32 i = 0; i < params.depth; i++) {
if (start == addr) {
return {i};
@@ -1267,7 +1269,7 @@ static bool LayerFitReinterpretSurface(RasterizerCacheOpenGL& cache, const Surfa
src_params.height == dst_params.MipHeight(*level) &&
src_params.block_height >= dst_params.MipBlockHeight(*level)) {
const std::optional<u32> slot =
- TryFindBestLayer(render_surface->GetAddr(), dst_params, *level);
+ TryFindBestLayer(render_surface->GetSurfaceParams().gpu_addr, dst_params, *level);
if (slot.has_value()) {
glCopyImageSubData(render_surface->Texture().handle,
SurfaceTargetToGL(src_params.target), 0, 0, 0, 0,
@@ -1283,8 +1285,8 @@ static bool LayerFitReinterpretSurface(RasterizerCacheOpenGL& cache, const Surfa
}
static bool IsReinterpretInvalid(const Surface render_surface, const Surface blitted_surface) {
- const VAddr bound1 = blitted_surface->GetAddr() + blitted_surface->GetMemorySize();
- const VAddr bound2 = render_surface->GetAddr() + render_surface->GetMemorySize();
+ const VAddr bound1 = blitted_surface->GetCpuAddr() + blitted_surface->GetMemorySize();
+ const VAddr bound2 = render_surface->GetCpuAddr() + render_surface->GetMemorySize();
if (bound2 > bound1)
return true;
const auto& dst_params = blitted_surface->GetSurfaceParams();
@@ -1327,7 +1329,8 @@ void RasterizerCacheOpenGL::SignalPreDrawCall() {
void RasterizerCacheOpenGL::SignalPostDrawCall() {
for (u32 i = 0; i < Maxwell::NumRenderTargets; i++) {
if (current_color_buffers[i] != nullptr) {
- Surface intersect = CollideOnReinterpretedSurface(current_color_buffers[i]->GetAddr());
+ Surface intersect =
+ CollideOnReinterpretedSurface(current_color_buffers[i]->GetCacheAddr());
if (intersect != nullptr) {
PartialReinterpretSurface(current_color_buffers[i], intersect);
texception = true;
diff --git a/src/video_core/renderer_opengl/gl_rasterizer_cache.h b/src/video_core/renderer_opengl/gl_rasterizer_cache.h
index 9cf6f50be..9366f47f2 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer_cache.h
+++ b/src/video_core/renderer_opengl/gl_rasterizer_cache.h
@@ -296,7 +296,7 @@ struct SurfaceParams {
bool is_array;
bool srgb_conversion;
// Parameters used for caching
- VAddr addr;
+ u8* host_ptr;
Tegra::GPUVAddr gpu_addr;
std::size_t size_in_bytes;
std::size_t size_in_bytes_gl;
@@ -345,10 +345,10 @@ class RasterizerOpenGL;
class CachedSurface final : public RasterizerCacheObject {
public:
- CachedSurface(const SurfaceParams& params);
+ explicit CachedSurface(const SurfaceParams& params);
- VAddr GetAddr() const override {
- return params.addr;
+ VAddr GetCpuAddr() const override {
+ return cpu_addr;
}
std::size_t GetSizeInBytes() const override {
@@ -432,6 +432,7 @@ private:
std::size_t memory_size;
bool reinterpreted = false;
bool must_reload = false;
+ VAddr cpu_addr{};
};
class RasterizerCacheOpenGL final : public RasterizerCache<Surface> {
@@ -449,7 +450,7 @@ public:
Surface GetColorBufferSurface(std::size_t index, bool preserve_contents);
/// Tries to find a framebuffer using on the provided CPU address
- Surface TryFindFramebufferSurface(VAddr addr) const;
+ Surface TryFindFramebufferSurface(const u8* host_ptr) const;
/// Copies the contents of one surface to another
void FermiCopySurface(const Tegra::Engines::Fermi2D::Regs::Surface& src_config,
@@ -506,12 +507,12 @@ private:
std::array<Surface, Maxwell::NumRenderTargets> current_color_buffers;
Surface last_depth_buffer;
- using SurfaceIntervalCache = boost::icl::interval_map<VAddr, Surface>;
+ using SurfaceIntervalCache = boost::icl::interval_map<CacheAddr, Surface>;
using SurfaceInterval = typename SurfaceIntervalCache::interval_type;
static auto GetReinterpretInterval(const Surface& object) {
- return SurfaceInterval::right_open(object->GetAddr() + 1,
- object->GetAddr() + object->GetMemorySize() - 1);
+ return SurfaceInterval::right_open(object->GetCacheAddr() + 1,
+ object->GetCacheAddr() + object->GetMemorySize() - 1);
}
// Reinterpreted surfaces are very fragil as the game may keep rendering into them.
@@ -523,7 +524,7 @@ private:
reinterpret_surface->MarkReinterpreted();
}
- Surface CollideOnReinterpretedSurface(VAddr addr) const {
+ Surface CollideOnReinterpretedSurface(CacheAddr addr) const {
const SurfaceInterval interval{addr};
for (auto& pair :
boost::make_iterator_range(reinterpreted_surfaces.equal_range(interval))) {
diff --git a/src/video_core/renderer_opengl/gl_shader_cache.cpp b/src/video_core/renderer_opengl/gl_shader_cache.cpp
index 4883e4f62..1ed740877 100644
--- a/src/video_core/renderer_opengl/gl_shader_cache.cpp
+++ b/src/video_core/renderer_opengl/gl_shader_cache.cpp
@@ -32,19 +32,16 @@ struct UnspecializedShader {
namespace {
/// Gets the address for the specified shader stage program
-VAddr GetShaderAddress(Maxwell::ShaderProgram program) {
- const auto& gpu = Core::System::GetInstance().GPU().Maxwell3D();
- const auto& shader_config = gpu.regs.shader_config[static_cast<std::size_t>(program)];
- const auto address = gpu.memory_manager.GpuToCpuAddress(gpu.regs.code_address.CodeAddress() +
- shader_config.offset);
- ASSERT_MSG(address, "Invalid GPU address");
- return *address;
+Tegra::GPUVAddr GetShaderAddress(Maxwell::ShaderProgram program) {
+ const auto& gpu{Core::System::GetInstance().GPU().Maxwell3D()};
+ const auto& shader_config{gpu.regs.shader_config[static_cast<std::size_t>(program)]};
+ return gpu.regs.code_address.CodeAddress() + shader_config.offset;
}
/// Gets the shader program code from memory for the specified address
-ProgramCode GetShaderCode(VAddr addr) {
+ProgramCode GetShaderCode(const u8* host_ptr) {
ProgramCode program_code(VideoCommon::Shader::MAX_PROGRAM_LENGTH);
- Memory::ReadBlock(addr, program_code.data(), program_code.size() * sizeof(u64));
+ std::memcpy(program_code.data(), host_ptr, program_code.size() * sizeof(u64));
return program_code;
}
@@ -214,12 +211,13 @@ std::set<GLenum> GetSupportedFormats() {
} // namespace
-CachedShader::CachedShader(VAddr addr, u64 unique_identifier, Maxwell::ShaderProgram program_type,
- ShaderDiskCacheOpenGL& disk_cache,
+CachedShader::CachedShader(VAddr cpu_addr, u64 unique_identifier,
+ Maxwell::ShaderProgram program_type, ShaderDiskCacheOpenGL& disk_cache,
const PrecompiledPrograms& precompiled_programs,
- ProgramCode&& program_code, ProgramCode&& program_code_b)
- : addr{addr}, unique_identifier{unique_identifier}, program_type{program_type},
- disk_cache{disk_cache}, precompiled_programs{precompiled_programs} {
+ ProgramCode&& program_code, ProgramCode&& program_code_b, u8* host_ptr)
+ : host_ptr{host_ptr}, cpu_addr{cpu_addr}, unique_identifier{unique_identifier},
+ program_type{program_type}, disk_cache{disk_cache},
+ precompiled_programs{precompiled_programs}, RasterizerCacheObject{host_ptr} {
const std::size_t code_size = CalculateProgramSize(program_code);
const std::size_t code_size_b =
@@ -243,12 +241,13 @@ CachedShader::CachedShader(VAddr addr, u64 unique_identifier, Maxwell::ShaderPro
disk_cache.SaveRaw(raw);
}
-CachedShader::CachedShader(VAddr addr, u64 unique_identifier, Maxwell::ShaderProgram program_type,
- ShaderDiskCacheOpenGL& disk_cache,
+CachedShader::CachedShader(VAddr cpu_addr, u64 unique_identifier,
+ Maxwell::ShaderProgram program_type, ShaderDiskCacheOpenGL& disk_cache,
const PrecompiledPrograms& precompiled_programs,
- GLShader::ProgramResult result)
- : addr{addr}, unique_identifier{unique_identifier}, program_type{program_type},
- disk_cache{disk_cache}, precompiled_programs{precompiled_programs} {
+ GLShader::ProgramResult result, u8* host_ptr)
+ : cpu_addr{cpu_addr}, unique_identifier{unique_identifier}, program_type{program_type},
+ disk_cache{disk_cache}, precompiled_programs{precompiled_programs}, RasterizerCacheObject{
+ host_ptr} {
code = std::move(result.first);
entries = result.second;
@@ -271,7 +270,7 @@ std::tuple<GLuint, BaseBindings> CachedShader::GetProgramHandle(GLenum primitive
disk_cache.SaveUsage(GetUsage(primitive_mode, base_bindings));
}
- LabelGLObject(GL_PROGRAM, program->handle, addr);
+ LabelGLObject(GL_PROGRAM, program->handle, cpu_addr);
}
handle = program->handle;
@@ -323,7 +322,7 @@ GLuint CachedShader::LazyGeometryProgram(CachedProgram& target_program, BaseBind
disk_cache.SaveUsage(GetUsage(primitive_mode, base_bindings));
}
- LabelGLObject(GL_PROGRAM, target_program->handle, addr, debug_name);
+ LabelGLObject(GL_PROGRAM, target_program->handle, cpu_addr, debug_name);
return target_program->handle;
};
@@ -486,29 +485,32 @@ Shader ShaderCacheOpenGL::GetStageProgram(Maxwell::ShaderProgram program) {
return last_shaders[static_cast<u32>(program)];
}
- const VAddr program_addr{GetShaderAddress(program)};
+ auto& memory_manager{Core::System::GetInstance().GPU().MemoryManager()};
+ const Tegra::GPUVAddr program_addr{GetShaderAddress(program)};
// Look up shader in the cache based on address
- Shader shader{TryGet(program_addr)};
+ const auto& host_ptr{memory_manager.GetPointer(program_addr)};
+ Shader shader{TryGet(host_ptr)};
if (!shader) {
// No shader found - create a new one
- ProgramCode program_code = GetShaderCode(program_addr);
+ ProgramCode program_code{GetShaderCode(host_ptr)};
ProgramCode program_code_b;
if (program == Maxwell::ShaderProgram::VertexA) {
- program_code_b = GetShaderCode(GetShaderAddress(Maxwell::ShaderProgram::VertexB));
+ program_code_b = GetShaderCode(
+ memory_manager.GetPointer(GetShaderAddress(Maxwell::ShaderProgram::VertexB)));
}
const u64 unique_identifier = GetUniqueIdentifier(program, program_code, program_code_b);
-
+ const VAddr cpu_addr{*memory_manager.GpuToCpuAddress(program_addr)};
const auto found = precompiled_shaders.find(unique_identifier);
if (found != precompiled_shaders.end()) {
shader =
- std::make_shared<CachedShader>(program_addr, unique_identifier, program, disk_cache,
- precompiled_programs, found->second);
+ std::make_shared<CachedShader>(cpu_addr, unique_identifier, program, disk_cache,
+ precompiled_programs, found->second, host_ptr);
} else {
shader = std::make_shared<CachedShader>(
- program_addr, unique_identifier, program, disk_cache, precompiled_programs,
- std::move(program_code), std::move(program_code_b));
+ cpu_addr, unique_identifier, program, disk_cache, precompiled_programs,
+ std::move(program_code), std::move(program_code_b), host_ptr);
}
Register(shader);
}
diff --git a/src/video_core/renderer_opengl/gl_shader_cache.h b/src/video_core/renderer_opengl/gl_shader_cache.h
index 97eed192f..fd1c85115 100644
--- a/src/video_core/renderer_opengl/gl_shader_cache.h
+++ b/src/video_core/renderer_opengl/gl_shader_cache.h
@@ -39,18 +39,18 @@ using PrecompiledShaders = std::unordered_map<u64, GLShader::ProgramResult>;
class CachedShader final : public RasterizerCacheObject {
public:
- explicit CachedShader(VAddr addr, u64 unique_identifier, Maxwell::ShaderProgram program_type,
- ShaderDiskCacheOpenGL& disk_cache,
+ explicit CachedShader(VAddr cpu_addr, u64 unique_identifier,
+ Maxwell::ShaderProgram program_type, ShaderDiskCacheOpenGL& disk_cache,
const PrecompiledPrograms& precompiled_programs,
- ProgramCode&& program_code, ProgramCode&& program_code_b);
+ ProgramCode&& program_code, ProgramCode&& program_code_b, u8* host_ptr);
- explicit CachedShader(VAddr addr, u64 unique_identifier, Maxwell::ShaderProgram program_type,
- ShaderDiskCacheOpenGL& disk_cache,
+ explicit CachedShader(VAddr cpu_addr, u64 unique_identifier,
+ Maxwell::ShaderProgram program_type, ShaderDiskCacheOpenGL& disk_cache,
const PrecompiledPrograms& precompiled_programs,
- GLShader::ProgramResult result);
+ GLShader::ProgramResult result, u8* host_ptr);
- VAddr GetAddr() const override {
- return addr;
+ VAddr GetCpuAddr() const override {
+ return cpu_addr;
}
std::size_t GetSizeInBytes() const override {
@@ -91,7 +91,8 @@ private:
ShaderDiskCacheUsage GetUsage(GLenum primitive_mode, BaseBindings base_bindings) const;
- VAddr addr{};
+ u8* host_ptr{};
+ VAddr cpu_addr{};
u64 unique_identifier{};
Maxwell::ShaderProgram program_type{};
ShaderDiskCacheOpenGL& disk_cache;
diff --git a/src/video_core/renderer_opengl/renderer_opengl.cpp b/src/video_core/renderer_opengl/renderer_opengl.cpp
index 8b510b6ae..5e3d862c6 100644
--- a/src/video_core/renderer_opengl/renderer_opengl.cpp
+++ b/src/video_core/renderer_opengl/renderer_opengl.cpp
@@ -164,12 +164,13 @@ void RendererOpenGL::LoadFBToScreenInfo(const Tegra::FramebufferConfig& framebuf
// Reset the screen info's display texture to its own permanent texture
screen_info.display_texture = screen_info.texture.resource.handle;
- Memory::RasterizerFlushVirtualRegion(framebuffer_addr, size_in_bytes,
- Memory::FlushMode::Flush);
+ rasterizer->FlushRegion(ToCacheAddr(Memory::GetPointer(framebuffer_addr)), size_in_bytes);
- VideoCore::MortonCopyPixels128(framebuffer.width, framebuffer.height, bytes_per_pixel, 4,
- Memory::GetPointer(framebuffer_addr),
- gl_framebuffer_data.data(), true);
+ constexpr u32 linear_bpp = 4;
+ VideoCore::MortonCopyPixels128(VideoCore::MortonSwizzleMode::MortonToLinear,
+ framebuffer.width, framebuffer.height, bytes_per_pixel,
+ linear_bpp, Memory::GetPointer(framebuffer_addr),
+ gl_framebuffer_data.data());
glPixelStorei(GL_UNPACK_ROW_LENGTH, static_cast<GLint>(framebuffer.stride));
diff --git a/src/video_core/renderer_vulkan/vk_buffer_cache.cpp b/src/video_core/renderer_vulkan/vk_buffer_cache.cpp
index 4a33a6c84..95eab3fec 100644
--- a/src/video_core/renderer_vulkan/vk_buffer_cache.cpp
+++ b/src/video_core/renderer_vulkan/vk_buffer_cache.cpp
@@ -17,6 +17,11 @@
namespace Vulkan {
+CachedBufferEntry::CachedBufferEntry(VAddr cpu_addr, std::size_t size, u64 offset,
+ std::size_t alignment, u8* host_ptr)
+ : cpu_addr{cpu_addr}, size{size}, offset{offset}, alignment{alignment}, RasterizerCacheObject{
+ host_ptr} {}
+
VKBufferCache::VKBufferCache(Tegra::MemoryManager& tegra_memory_manager,
VideoCore::RasterizerInterface& rasterizer, const VKDevice& device,
VKMemoryManager& memory_manager, VKScheduler& scheduler, u64 size)
@@ -37,16 +42,18 @@ VKBufferCache::~VKBufferCache() = default;
u64 VKBufferCache::UploadMemory(Tegra::GPUVAddr gpu_addr, std::size_t size, u64 alignment,
bool cache) {
const auto cpu_addr{tegra_memory_manager.GpuToCpuAddress(gpu_addr)};
- ASSERT(cpu_addr);
+ ASSERT_MSG(cpu_addr, "Invalid GPU address");
// Cache management is a big overhead, so only cache entries with a given size.
// TODO: Figure out which size is the best for given games.
cache &= size >= 2048;
+ const auto& host_ptr{Memory::GetPointer(*cpu_addr)};
if (cache) {
- if (auto entry = TryGet(*cpu_addr); entry) {
- if (entry->size >= size && entry->alignment == alignment) {
- return entry->offset;
+ auto entry = TryGet(host_ptr);
+ if (entry) {
+ if (entry->GetSize() >= size && entry->GetAlignment() == alignment) {
+ return entry->GetOffset();
}
Unregister(entry);
}
@@ -55,17 +62,17 @@ u64 VKBufferCache::UploadMemory(Tegra::GPUVAddr gpu_addr, std::size_t size, u64
AlignBuffer(alignment);
const u64 uploaded_offset = buffer_offset;
- Memory::ReadBlock(*cpu_addr, buffer_ptr, size);
+ if (!host_ptr) {
+ return uploaded_offset;
+ }
+ std::memcpy(buffer_ptr, host_ptr, size);
buffer_ptr += size;
buffer_offset += size;
if (cache) {
- auto entry = std::make_shared<CachedBufferEntry>();
- entry->offset = uploaded_offset;
- entry->size = size;
- entry->alignment = alignment;
- entry->addr = *cpu_addr;
+ auto entry = std::make_shared<CachedBufferEntry>(*cpu_addr, size, uploaded_offset,
+ alignment, host_ptr);
Register(entry);
}
diff --git a/src/video_core/renderer_vulkan/vk_buffer_cache.h b/src/video_core/renderer_vulkan/vk_buffer_cache.h
index d8e916f31..8b415744b 100644
--- a/src/video_core/renderer_vulkan/vk_buffer_cache.h
+++ b/src/video_core/renderer_vulkan/vk_buffer_cache.h
@@ -24,22 +24,39 @@ class VKFence;
class VKMemoryManager;
class VKStreamBuffer;
-struct CachedBufferEntry final : public RasterizerCacheObject {
- VAddr GetAddr() const override {
- return addr;
+class CachedBufferEntry final : public RasterizerCacheObject {
+public:
+ explicit CachedBufferEntry(VAddr cpu_addr, std::size_t size, u64 offset, std::size_t alignment,
+ u8* host_ptr);
+
+ VAddr GetCpuAddr() const override {
+ return cpu_addr;
}
std::size_t GetSizeInBytes() const override {
return size;
}
+ std::size_t GetSize() const {
+ return size;
+ }
+
+ u64 GetOffset() const {
+ return offset;
+ }
+
+ std::size_t GetAlignment() const {
+ return alignment;
+ }
+
// We do not have to flush this cache as things in it are never modified by us.
void Flush() override {}
- VAddr addr;
- std::size_t size;
- u64 offset;
- std::size_t alignment;
+private:
+ VAddr cpu_addr{};
+ std::size_t size{};
+ u64 offset{};
+ std::size_t alignment{};
};
class VKBufferCache final : public RasterizerCache<std::shared_ptr<CachedBufferEntry>> {
diff --git a/src/video_core/renderer_vulkan/vk_sampler_cache.cpp b/src/video_core/renderer_vulkan/vk_sampler_cache.cpp
new file mode 100644
index 000000000..ed3178f09
--- /dev/null
+++ b/src/video_core/renderer_vulkan/vk_sampler_cache.cpp
@@ -0,0 +1,81 @@
+// Copyright 2019 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <cstring>
+#include <optional>
+#include <unordered_map>
+
+#include "common/assert.h"
+#include "common/cityhash.h"
+#include "video_core/renderer_vulkan/declarations.h"
+#include "video_core/renderer_vulkan/maxwell_to_vk.h"
+#include "video_core/renderer_vulkan/vk_sampler_cache.h"
+#include "video_core/textures/texture.h"
+
+namespace Vulkan {
+
+static std::optional<vk::BorderColor> TryConvertBorderColor(std::array<float, 4> color) {
+ // TODO(Rodrigo): Manage integer border colors
+ if (color == std::array<float, 4>{0, 0, 0, 0}) {
+ return vk::BorderColor::eFloatTransparentBlack;
+ } else if (color == std::array<float, 4>{0, 0, 0, 1}) {
+ return vk::BorderColor::eFloatOpaqueBlack;
+ } else if (color == std::array<float, 4>{1, 1, 1, 1}) {
+ return vk::BorderColor::eFloatOpaqueWhite;
+ } else {
+ return {};
+ }
+}
+
+std::size_t SamplerCacheKey::Hash() const {
+ static_assert(sizeof(raw) % sizeof(u64) == 0);
+ return static_cast<std::size_t>(
+ Common::CityHash64(reinterpret_cast<const char*>(raw.data()), sizeof(raw) / sizeof(u64)));
+}
+
+bool SamplerCacheKey::operator==(const SamplerCacheKey& rhs) const {
+ return raw == rhs.raw;
+}
+
+VKSamplerCache::VKSamplerCache(const VKDevice& device) : device{device} {}
+
+VKSamplerCache::~VKSamplerCache() = default;
+
+vk::Sampler VKSamplerCache::GetSampler(const Tegra::Texture::TSCEntry& tsc) {
+ const auto [entry, is_cache_miss] = cache.try_emplace(SamplerCacheKey{tsc});
+ auto& sampler = entry->second;
+ if (is_cache_miss) {
+ sampler = CreateSampler(tsc);
+ }
+ return *sampler;
+}
+
+UniqueSampler VKSamplerCache::CreateSampler(const Tegra::Texture::TSCEntry& tsc) {
+ const float max_anisotropy = tsc.GetMaxAnisotropy();
+ const bool has_anisotropy = max_anisotropy > 1.0f;
+
+ const auto border_color = tsc.GetBorderColor();
+ const auto vk_border_color = TryConvertBorderColor(border_color);
+ UNIMPLEMENTED_IF_MSG(!vk_border_color, "Unimplemented border color {} {} {} {}",
+ border_color[0], border_color[1], border_color[2], border_color[3]);
+
+ constexpr bool unnormalized_coords = false;
+
+ const vk::SamplerCreateInfo sampler_ci(
+ {}, 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::DepthCompareFunction(tsc.depth_compare_func), tsc.GetMinLod(),
+ tsc.GetMaxLod(), vk_border_color.value_or(vk::BorderColor::eFloatTransparentBlack),
+ unnormalized_coords);
+
+ const auto& dld = device.GetDispatchLoader();
+ const auto dev = device.GetLogical();
+ return dev.createSamplerUnique(sampler_ci, nullptr, dld);
+}
+
+} // namespace Vulkan
diff --git a/src/video_core/renderer_vulkan/vk_sampler_cache.h b/src/video_core/renderer_vulkan/vk_sampler_cache.h
new file mode 100644
index 000000000..c6394dc87
--- /dev/null
+++ b/src/video_core/renderer_vulkan/vk_sampler_cache.h
@@ -0,0 +1,56 @@
+// Copyright 2019 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <unordered_map>
+
+#include "common/common_types.h"
+#include "video_core/renderer_vulkan/declarations.h"
+#include "video_core/textures/texture.h"
+
+namespace Vulkan {
+
+class VKDevice;
+
+struct SamplerCacheKey final : public Tegra::Texture::TSCEntry {
+ std::size_t Hash() const;
+
+ bool operator==(const SamplerCacheKey& rhs) const;
+
+ bool operator!=(const SamplerCacheKey& rhs) const {
+ return !operator==(rhs);
+ }
+};
+
+} // namespace Vulkan
+
+namespace std {
+
+template <>
+struct hash<Vulkan::SamplerCacheKey> {
+ std::size_t operator()(const Vulkan::SamplerCacheKey& k) const noexcept {
+ return k.Hash();
+ }
+};
+
+} // namespace std
+
+namespace Vulkan {
+
+class VKSamplerCache {
+public:
+ explicit VKSamplerCache(const VKDevice& device);
+ ~VKSamplerCache();
+
+ vk::Sampler GetSampler(const Tegra::Texture::TSCEntry& tsc);
+
+private:
+ UniqueSampler CreateSampler(const Tegra::Texture::TSCEntry& tsc);
+
+ const VKDevice& device;
+ std::unordered_map<SamplerCacheKey, UniqueSampler> cache;
+};
+
+} // namespace Vulkan
diff --git a/src/video_core/textures/decoders.cpp b/src/video_core/textures/decoders.cpp
index cad7340f5..995d0e068 100644
--- a/src/video_core/textures/decoders.cpp
+++ b/src/video_core/textures/decoders.cpp
@@ -6,7 +6,6 @@
#include <cstring>
#include "common/alignment.h"
#include "common/assert.h"
-#include "core/memory.h"
#include "video_core/gpu.h"
#include "video_core/textures/decoders.h"
#include "video_core/textures/texture.h"
@@ -230,18 +229,18 @@ u32 BytesPerPixel(TextureFormat format) {
}
}
-void UnswizzleTexture(u8* const unswizzled_data, VAddr address, u32 tile_size_x, u32 tile_size_y,
+void UnswizzleTexture(u8* const unswizzled_data, u8* address, u32 tile_size_x, u32 tile_size_y,
u32 bytes_per_pixel, u32 width, u32 height, u32 depth, u32 block_height,
u32 block_depth, u32 width_spacing) {
CopySwizzledData((width + tile_size_x - 1) / tile_size_x,
(height + tile_size_y - 1) / tile_size_y, depth, bytes_per_pixel,
- bytes_per_pixel, Memory::GetPointer(address), unswizzled_data, true,
- block_height, block_depth, width_spacing);
+ bytes_per_pixel, address, unswizzled_data, true, block_height, block_depth,
+ width_spacing);
}
-std::vector<u8> UnswizzleTexture(VAddr address, u32 tile_size_x, u32 tile_size_y,
- u32 bytes_per_pixel, u32 width, u32 height, u32 depth,
- u32 block_height, u32 block_depth, u32 width_spacing) {
+std::vector<u8> UnswizzleTexture(u8* address, u32 tile_size_x, u32 tile_size_y, u32 bytes_per_pixel,
+ u32 width, u32 height, u32 depth, u32 block_height,
+ u32 block_depth, u32 width_spacing) {
std::vector<u8> unswizzled_data(width * height * depth * bytes_per_pixel);
UnswizzleTexture(unswizzled_data.data(), address, tile_size_x, tile_size_y, bytes_per_pixel,
width, height, depth, block_height, block_depth, width_spacing);
@@ -249,8 +248,7 @@ std::vector<u8> UnswizzleTexture(VAddr address, u32 tile_size_x, u32 tile_size_y
}
void SwizzleSubrect(u32 subrect_width, u32 subrect_height, u32 source_pitch, u32 swizzled_width,
- u32 bytes_per_pixel, VAddr swizzled_data, VAddr unswizzled_data,
- u32 block_height) {
+ u32 bytes_per_pixel, u8* swizzled_data, u8* unswizzled_data, u32 block_height) {
const u32 image_width_in_gobs{(swizzled_width * bytes_per_pixel + (gob_size_x - 1)) /
gob_size_x};
for (u32 line = 0; line < subrect_height; ++line) {
@@ -262,17 +260,17 @@ void SwizzleSubrect(u32 subrect_width, u32 subrect_height, u32 source_pitch, u32
const u32 gob_address =
gob_address_y + (x * bytes_per_pixel / gob_size_x) * gob_size * block_height;
const u32 swizzled_offset = gob_address + table[(x * bytes_per_pixel) % gob_size_x];
- const VAddr source_line = unswizzled_data + line * source_pitch + x * bytes_per_pixel;
- const VAddr dest_addr = swizzled_data + swizzled_offset;
+ u8* source_line = unswizzled_data + line * source_pitch + x * bytes_per_pixel;
+ u8* dest_addr = swizzled_data + swizzled_offset;
- Memory::CopyBlock(dest_addr, source_line, bytes_per_pixel);
+ std::memcpy(dest_addr, source_line, bytes_per_pixel);
}
}
}
void UnswizzleSubrect(u32 subrect_width, u32 subrect_height, u32 dest_pitch, u32 swizzled_width,
- u32 bytes_per_pixel, VAddr swizzled_data, VAddr unswizzled_data,
- u32 block_height, u32 offset_x, u32 offset_y) {
+ u32 bytes_per_pixel, u8* swizzled_data, u8* unswizzled_data, u32 block_height,
+ u32 offset_x, u32 offset_y) {
for (u32 line = 0; line < subrect_height; ++line) {
const u32 y2 = line + offset_y;
const u32 gob_address_y = (y2 / (gob_size_y * block_height)) * gob_size * block_height +
@@ -282,10 +280,10 @@ void UnswizzleSubrect(u32 subrect_width, u32 subrect_height, u32 dest_pitch, u32
const u32 x2 = (x + offset_x) * bytes_per_pixel;
const u32 gob_address = gob_address_y + (x2 / gob_size_x) * gob_size * block_height;
const u32 swizzled_offset = gob_address + table[x2 % gob_size_x];
- const VAddr dest_line = unswizzled_data + line * dest_pitch + x * bytes_per_pixel;
- const VAddr source_addr = swizzled_data + swizzled_offset;
+ u8* dest_line = unswizzled_data + line * dest_pitch + x * bytes_per_pixel;
+ u8* source_addr = swizzled_data + swizzled_offset;
- Memory::CopyBlock(dest_line, source_addr, bytes_per_pixel);
+ std::memcpy(dest_line, source_addr, bytes_per_pixel);
}
}
}
diff --git a/src/video_core/textures/decoders.h b/src/video_core/textures/decoders.h
index 65df86890..e078fa274 100644
--- a/src/video_core/textures/decoders.h
+++ b/src/video_core/textures/decoders.h
@@ -17,14 +17,14 @@ inline std::size_t GetGOBSize() {
}
/// Unswizzles a swizzled texture without changing its format.
-void UnswizzleTexture(u8* unswizzled_data, VAddr address, u32 tile_size_x, u32 tile_size_y,
+void UnswizzleTexture(u8* unswizzled_data, u8* address, u32 tile_size_x, u32 tile_size_y,
u32 bytes_per_pixel, u32 width, u32 height, u32 depth,
u32 block_height = TICEntry::DefaultBlockHeight,
u32 block_depth = TICEntry::DefaultBlockHeight, u32 width_spacing = 0);
/// Unswizzles a swizzled texture without changing its format.
-std::vector<u8> UnswizzleTexture(VAddr address, u32 tile_size_x, u32 tile_size_y,
- u32 bytes_per_pixel, u32 width, u32 height, u32 depth,
+std::vector<u8> UnswizzleTexture(u8* address, u32 tile_size_x, u32 tile_size_y, u32 bytes_per_pixel,
+ u32 width, u32 height, u32 depth,
u32 block_height = TICEntry::DefaultBlockHeight,
u32 block_depth = TICEntry::DefaultBlockHeight,
u32 width_spacing = 0);
@@ -44,12 +44,11 @@ std::size_t CalculateSize(bool tiled, u32 bytes_per_pixel, u32 width, u32 height
/// Copies an untiled subrectangle into a tiled surface.
void SwizzleSubrect(u32 subrect_width, u32 subrect_height, u32 source_pitch, u32 swizzled_width,
- u32 bytes_per_pixel, VAddr swizzled_data, VAddr unswizzled_data,
- u32 block_height);
+ u32 bytes_per_pixel, u8* swizzled_data, u8* unswizzled_data, u32 block_height);
/// Copies a tiled subrectangle into a linear surface.
void UnswizzleSubrect(u32 subrect_width, u32 subrect_height, u32 dest_pitch, u32 swizzled_width,
- u32 bytes_per_pixel, VAddr swizzled_data, VAddr unswizzled_data,
- u32 block_height, u32 offset_x, u32 offset_y);
+ u32 bytes_per_pixel, u8* swizzled_data, u8* unswizzled_data, u32 block_height,
+ u32 offset_x, u32 offset_y);
} // namespace Tegra::Texture
diff --git a/src/video_core/textures/texture.h b/src/video_core/textures/texture.h
index 8c278c0e2..93ecc6e31 100644
--- a/src/video_core/textures/texture.h
+++ b/src/video_core/textures/texture.h
@@ -283,31 +283,36 @@ enum class TextureMipmapFilter : u32 {
struct TSCEntry {
union {
- BitField<0, 3, WrapMode> wrap_u;
- BitField<3, 3, WrapMode> wrap_v;
- BitField<6, 3, WrapMode> wrap_p;
- BitField<9, 1, u32> depth_compare_enabled;
- BitField<10, 3, DepthCompareFunc> depth_compare_func;
- BitField<13, 1, u32> srgb_conversion;
- BitField<20, 3, u32> max_anisotropy;
+ struct {
+ union {
+ BitField<0, 3, WrapMode> wrap_u;
+ BitField<3, 3, WrapMode> wrap_v;
+ BitField<6, 3, WrapMode> wrap_p;
+ BitField<9, 1, u32> depth_compare_enabled;
+ BitField<10, 3, DepthCompareFunc> depth_compare_func;
+ BitField<13, 1, u32> srgb_conversion;
+ BitField<20, 3, u32> max_anisotropy;
+ };
+ union {
+ BitField<0, 2, TextureFilter> mag_filter;
+ BitField<4, 2, TextureFilter> min_filter;
+ BitField<6, 2, TextureMipmapFilter> mipmap_filter;
+ BitField<9, 1, u32> cubemap_interface_filtering;
+ BitField<12, 13, u32> mip_lod_bias;
+ };
+ union {
+ BitField<0, 12, u32> min_lod_clamp;
+ BitField<12, 12, u32> max_lod_clamp;
+ BitField<24, 8, u32> srgb_border_color_r;
+ };
+ union {
+ BitField<12, 8, u32> srgb_border_color_g;
+ BitField<20, 8, u32> srgb_border_color_b;
+ };
+ std::array<f32, 4> border_color;
+ };
+ std::array<u8, 0x20> raw;
};
- union {
- BitField<0, 2, TextureFilter> mag_filter;
- BitField<4, 2, TextureFilter> min_filter;
- BitField<6, 2, TextureMipmapFilter> mipmap_filter;
- BitField<9, 1, u32> cubemap_interface_filtering;
- BitField<12, 13, u32> mip_lod_bias;
- };
- union {
- BitField<0, 12, u32> min_lod_clamp;
- BitField<12, 12, u32> max_lod_clamp;
- BitField<24, 8, u32> srgb_border_color_r;
- };
- union {
- BitField<12, 8, u32> srgb_border_color_g;
- BitField<20, 8, u32> srgb_border_color_b;
- };
- std::array<f32, 4> border_color;
float GetMaxAnisotropy() const {
return static_cast<float>(1U << max_anisotropy);
@@ -324,7 +329,7 @@ struct TSCEntry {
float GetLodBias() const {
// Sign extend the 13-bit value.
constexpr u32 mask = 1U << (13 - 1);
- return static_cast<float>((mip_lod_bias ^ mask) - mask) / 256.0f;
+ return static_cast<s32>((mip_lod_bias ^ mask) - mask) / 256.0f;
}
std::array<float, 4> GetBorderColor() const {
diff --git a/src/yuzu/bootmanager.cpp b/src/yuzu/bootmanager.cpp
index 3b070bfbb..d2c97b1f8 100644
--- a/src/yuzu/bootmanager.cpp
+++ b/src/yuzu/bootmanager.cpp
@@ -123,7 +123,6 @@ GRenderWindow::GRenderWindow(QWidget* parent, EmuThread* emu_thread)
setAttribute(Qt::WA_AcceptTouchEvents);
InputCommon::Init();
- InputCommon::StartJoystickEventHandler();
connect(this, &GRenderWindow::FirstFrameDisplayed, static_cast<GMainWindow*>(parent),
&GMainWindow::OnLoadComplete);
}
diff --git a/src/yuzu/configuration/config.cpp b/src/yuzu/configuration/config.cpp
index 74dc6bb28..4650f96a3 100644
--- a/src/yuzu/configuration/config.cpp
+++ b/src/yuzu/configuration/config.cpp
@@ -209,7 +209,7 @@ void Config::ReadPlayerValues() {
for (std::size_t p = 0; p < Settings::values.players.size(); ++p) {
auto& player = Settings::values.players[p];
- player.connected = qt_config->value(QString("player_%1_connected").arg(p), false).toBool();
+ player.connected = ReadSetting(QString("player_%1_connected").arg(p), false).toBool();
player.type = static_cast<Settings::ControllerType>(
qt_config
@@ -269,7 +269,7 @@ void Config::ReadPlayerValues() {
}
void Config::ReadDebugValues() {
- Settings::values.debug_pad_enabled = qt_config->value("debug_pad_enabled", false).toBool();
+ Settings::values.debug_pad_enabled = ReadSetting("debug_pad_enabled", false).toBool();
for (int i = 0; i < Settings::NativeButton::NumButtons; ++i) {
std::string default_param = InputCommon::GenerateKeyboardParam(default_buttons[i]);
Settings::values.debug_pad_buttons[i] =
@@ -298,7 +298,7 @@ void Config::ReadDebugValues() {
}
void Config::ReadKeyboardValues() {
- Settings::values.keyboard_enabled = qt_config->value("keyboard_enabled", false).toBool();
+ Settings::values.keyboard_enabled = ReadSetting("keyboard_enabled", false).toBool();
std::transform(default_keyboard_keys.begin(), default_keyboard_keys.end(),
Settings::values.keyboard_keys.begin(), InputCommon::GenerateKeyboardParam);
@@ -311,7 +311,7 @@ void Config::ReadKeyboardValues() {
}
void Config::ReadMouseValues() {
- Settings::values.mouse_enabled = qt_config->value("mouse_enabled", false).toBool();
+ Settings::values.mouse_enabled = ReadSetting("mouse_enabled", false).toBool();
for (int i = 0; i < Settings::NativeMouseButton::NumMouseButtons; ++i) {
std::string default_param = InputCommon::GenerateKeyboardParam(default_mouse_buttons[i]);
@@ -327,16 +327,14 @@ void Config::ReadMouseValues() {
}
void Config::ReadTouchscreenValues() {
- Settings::values.touchscreen.enabled = qt_config->value("touchscreen_enabled", true).toBool();
+ Settings::values.touchscreen.enabled = ReadSetting("touchscreen_enabled", true).toBool();
Settings::values.touchscreen.device =
- qt_config->value("touchscreen_device", "engine:emu_window").toString().toStdString();
-
- Settings::values.touchscreen.finger = qt_config->value("touchscreen_finger", 0).toUInt();
- Settings::values.touchscreen.rotation_angle = qt_config->value("touchscreen_angle", 0).toUInt();
- Settings::values.touchscreen.diameter_x =
- qt_config->value("touchscreen_diameter_x", 15).toUInt();
- Settings::values.touchscreen.diameter_y =
- qt_config->value("touchscreen_diameter_y", 15).toUInt();
+ ReadSetting("touchscreen_device", "engine:emu_window").toString().toStdString();
+
+ Settings::values.touchscreen.finger = ReadSetting("touchscreen_finger", 0).toUInt();
+ Settings::values.touchscreen.rotation_angle = ReadSetting("touchscreen_angle", 0).toUInt();
+ Settings::values.touchscreen.diameter_x = ReadSetting("touchscreen_diameter_x", 15).toUInt();
+ Settings::values.touchscreen.diameter_y = ReadSetting("touchscreen_diameter_y", 15).toUInt();
qt_config->endGroup();
}
@@ -357,42 +355,41 @@ void Config::ReadValues() {
ReadTouchscreenValues();
Settings::values.motion_device =
- qt_config->value("motion_device", "engine:motion_emu,update_period:100,sensitivity:0.01")
+ ReadSetting("motion_device", "engine:motion_emu,update_period:100,sensitivity:0.01")
.toString()
.toStdString();
qt_config->beginGroup("Core");
- Settings::values.use_cpu_jit = qt_config->value("use_cpu_jit", true).toBool();
- Settings::values.use_multi_core = qt_config->value("use_multi_core", false).toBool();
+ Settings::values.use_cpu_jit = ReadSetting("use_cpu_jit", true).toBool();
+ Settings::values.use_multi_core = ReadSetting("use_multi_core", false).toBool();
qt_config->endGroup();
qt_config->beginGroup("Renderer");
- Settings::values.resolution_factor = qt_config->value("resolution_factor", 1.0).toFloat();
- Settings::values.use_frame_limit = qt_config->value("use_frame_limit", true).toBool();
- Settings::values.frame_limit = qt_config->value("frame_limit", 100).toInt();
- Settings::values.use_disk_shader_cache =
- qt_config->value("use_disk_shader_cache", false).toBool();
+ Settings::values.resolution_factor = ReadSetting("resolution_factor", 1.0).toFloat();
+ Settings::values.use_frame_limit = ReadSetting("use_frame_limit", true).toBool();
+ Settings::values.frame_limit = ReadSetting("frame_limit", 100).toInt();
+ Settings::values.use_disk_shader_cache = ReadSetting("use_disk_shader_cache", true).toBool();
Settings::values.use_accurate_gpu_emulation =
- qt_config->value("use_accurate_gpu_emulation", false).toBool();
+ ReadSetting("use_accurate_gpu_emulation", false).toBool();
Settings::values.use_asynchronous_gpu_emulation =
- qt_config->value("use_asynchronous_gpu_emulation", false).toBool();
+ ReadSetting("use_asynchronous_gpu_emulation", false).toBool();
- Settings::values.bg_red = qt_config->value("bg_red", 0.0).toFloat();
- Settings::values.bg_green = qt_config->value("bg_green", 0.0).toFloat();
- Settings::values.bg_blue = qt_config->value("bg_blue", 0.0).toFloat();
+ Settings::values.bg_red = ReadSetting("bg_red", 0.0).toFloat();
+ Settings::values.bg_green = ReadSetting("bg_green", 0.0).toFloat();
+ Settings::values.bg_blue = ReadSetting("bg_blue", 0.0).toFloat();
qt_config->endGroup();
qt_config->beginGroup("Audio");
- Settings::values.sink_id = qt_config->value("output_engine", "auto").toString().toStdString();
+ Settings::values.sink_id = ReadSetting("output_engine", "auto").toString().toStdString();
Settings::values.enable_audio_stretching =
- qt_config->value("enable_audio_stretching", true).toBool();
+ ReadSetting("enable_audio_stretching", true).toBool();
Settings::values.audio_device_id =
- qt_config->value("output_device", "auto").toString().toStdString();
- Settings::values.volume = qt_config->value("volume", 1).toFloat();
+ ReadSetting("output_device", "auto").toString().toStdString();
+ Settings::values.volume = ReadSetting("volume", 1).toFloat();
qt_config->endGroup();
qt_config->beginGroup("Data Storage");
- Settings::values.use_virtual_sd = qt_config->value("use_virtual_sd", true).toBool();
+ Settings::values.use_virtual_sd = ReadSetting("use_virtual_sd", true).toBool();
FileUtil::GetUserPath(
FileUtil::UserPath::NANDDir,
qt_config
@@ -410,30 +407,30 @@ void Config::ReadValues() {
qt_config->endGroup();
qt_config->beginGroup("Core");
- Settings::values.use_cpu_jit = qt_config->value("use_cpu_jit", true).toBool();
- Settings::values.use_multi_core = qt_config->value("use_multi_core", false).toBool();
+ Settings::values.use_cpu_jit = ReadSetting("use_cpu_jit", true).toBool();
+ Settings::values.use_multi_core = ReadSetting("use_multi_core", false).toBool();
qt_config->endGroup();
qt_config->beginGroup("System");
- Settings::values.use_docked_mode = qt_config->value("use_docked_mode", false).toBool();
- Settings::values.enable_nfc = qt_config->value("enable_nfc", true).toBool();
+ Settings::values.use_docked_mode = ReadSetting("use_docked_mode", false).toBool();
+ Settings::values.enable_nfc = ReadSetting("enable_nfc", true).toBool();
- Settings::values.current_user = std::clamp<int>(qt_config->value("current_user", 0).toInt(), 0,
- Service::Account::MAX_USERS - 1);
+ Settings::values.current_user =
+ std::clamp<int>(ReadSetting("current_user", 0).toInt(), 0, Service::Account::MAX_USERS - 1);
- Settings::values.language_index = qt_config->value("language_index", 1).toInt();
+ Settings::values.language_index = ReadSetting("language_index", 1).toInt();
- const auto rng_seed_enabled = qt_config->value("rng_seed_enabled", false).toBool();
+ const auto rng_seed_enabled = ReadSetting("rng_seed_enabled", false).toBool();
if (rng_seed_enabled) {
- Settings::values.rng_seed = qt_config->value("rng_seed", 0).toULongLong();
+ Settings::values.rng_seed = ReadSetting("rng_seed", 0).toULongLong();
} else {
Settings::values.rng_seed = std::nullopt;
}
- const auto custom_rtc_enabled = qt_config->value("custom_rtc_enabled", false).toBool();
+ const auto custom_rtc_enabled = ReadSetting("custom_rtc_enabled", false).toBool();
if (custom_rtc_enabled) {
Settings::values.custom_rtc =
- std::chrono::seconds(qt_config->value("custom_rtc", 0).toULongLong());
+ std::chrono::seconds(ReadSetting("custom_rtc", 0).toULongLong());
} else {
Settings::values.custom_rtc = std::nullopt;
}
@@ -441,35 +438,35 @@ void Config::ReadValues() {
qt_config->endGroup();
qt_config->beginGroup("Miscellaneous");
- Settings::values.log_filter = qt_config->value("log_filter", "*:Info").toString().toStdString();
- Settings::values.use_dev_keys = qt_config->value("use_dev_keys", false).toBool();
+ Settings::values.log_filter = ReadSetting("log_filter", "*:Info").toString().toStdString();
+ Settings::values.use_dev_keys = ReadSetting("use_dev_keys", false).toBool();
qt_config->endGroup();
qt_config->beginGroup("Debugging");
- Settings::values.use_gdbstub = qt_config->value("use_gdbstub", false).toBool();
- Settings::values.gdbstub_port = qt_config->value("gdbstub_port", 24689).toInt();
- Settings::values.program_args = qt_config->value("program_args", "").toString().toStdString();
- Settings::values.dump_exefs = qt_config->value("dump_exefs", false).toBool();
- Settings::values.dump_nso = qt_config->value("dump_nso", false).toBool();
+ Settings::values.use_gdbstub = ReadSetting("use_gdbstub", false).toBool();
+ Settings::values.gdbstub_port = ReadSetting("gdbstub_port", 24689).toInt();
+ Settings::values.program_args = ReadSetting("program_args", "").toString().toStdString();
+ Settings::values.dump_exefs = ReadSetting("dump_exefs", false).toBool();
+ Settings::values.dump_nso = ReadSetting("dump_nso", false).toBool();
qt_config->endGroup();
qt_config->beginGroup("WebService");
- Settings::values.enable_telemetry = qt_config->value("enable_telemetry", true).toBool();
+ Settings::values.enable_telemetry = ReadSetting("enable_telemetry", true).toBool();
Settings::values.web_api_url =
- qt_config->value("web_api_url", "https://api.yuzu-emu.org").toString().toStdString();
- Settings::values.yuzu_username = qt_config->value("yuzu_username").toString().toStdString();
- Settings::values.yuzu_token = qt_config->value("yuzu_token").toString().toStdString();
+ ReadSetting("web_api_url", "https://api.yuzu-emu.org").toString().toStdString();
+ Settings::values.yuzu_username = ReadSetting("yuzu_username").toString().toStdString();
+ Settings::values.yuzu_token = ReadSetting("yuzu_token").toString().toStdString();
qt_config->endGroup();
const auto size = qt_config->beginReadArray("DisabledAddOns");
for (int i = 0; i < size; ++i) {
qt_config->setArrayIndex(i);
- const auto title_id = qt_config->value("title_id", 0).toULongLong();
+ const auto title_id = ReadSetting("title_id", 0).toULongLong();
std::vector<std::string> out;
const auto d_size = qt_config->beginReadArray("disabled");
for (int j = 0; j < d_size; ++j) {
qt_config->setArrayIndex(j);
- out.push_back(qt_config->value("d", "").toString().toStdString());
+ out.push_back(ReadSetting("d", "").toString().toStdString());
}
qt_config->endArray();
Settings::values.disabled_addons.insert_or_assign(title_id, out);
@@ -477,41 +474,38 @@ void Config::ReadValues() {
qt_config->endArray();
qt_config->beginGroup("UI");
- UISettings::values.theme = qt_config->value("theme", UISettings::themes[0].second).toString();
+ UISettings::values.theme = ReadSetting("theme", UISettings::themes[0].second).toString();
UISettings::values.enable_discord_presence =
- qt_config->value("enable_discord_presence", true).toBool();
+ ReadSetting("enable_discord_presence", true).toBool();
UISettings::values.screenshot_resolution_factor =
- static_cast<u16>(qt_config->value("screenshot_resolution_factor", 0).toUInt());
- UISettings::values.select_user_on_boot =
- qt_config->value("select_user_on_boot", false).toBool();
+ static_cast<u16>(ReadSetting("screenshot_resolution_factor", 0).toUInt());
+ UISettings::values.select_user_on_boot = ReadSetting("select_user_on_boot", false).toBool();
qt_config->beginGroup("UIGameList");
- UISettings::values.show_unknown = qt_config->value("show_unknown", true).toBool();
- UISettings::values.show_add_ons = qt_config->value("show_add_ons", true).toBool();
- UISettings::values.icon_size = qt_config->value("icon_size", 64).toUInt();
- UISettings::values.row_1_text_id = qt_config->value("row_1_text_id", 3).toUInt();
- UISettings::values.row_2_text_id = qt_config->value("row_2_text_id", 2).toUInt();
+ UISettings::values.show_unknown = ReadSetting("show_unknown", true).toBool();
+ UISettings::values.show_add_ons = ReadSetting("show_add_ons", true).toBool();
+ UISettings::values.icon_size = ReadSetting("icon_size", 64).toUInt();
+ UISettings::values.row_1_text_id = ReadSetting("row_1_text_id", 3).toUInt();
+ UISettings::values.row_2_text_id = ReadSetting("row_2_text_id", 2).toUInt();
qt_config->endGroup();
qt_config->beginGroup("UILayout");
- UISettings::values.geometry = qt_config->value("geometry").toByteArray();
- UISettings::values.state = qt_config->value("state").toByteArray();
- UISettings::values.renderwindow_geometry =
- qt_config->value("geometryRenderWindow").toByteArray();
- UISettings::values.gamelist_header_state =
- qt_config->value("gameListHeaderState").toByteArray();
+ UISettings::values.geometry = ReadSetting("geometry").toByteArray();
+ UISettings::values.state = ReadSetting("state").toByteArray();
+ UISettings::values.renderwindow_geometry = ReadSetting("geometryRenderWindow").toByteArray();
+ UISettings::values.gamelist_header_state = ReadSetting("gameListHeaderState").toByteArray();
UISettings::values.microprofile_geometry =
- qt_config->value("microProfileDialogGeometry").toByteArray();
+ ReadSetting("microProfileDialogGeometry").toByteArray();
UISettings::values.microprofile_visible =
- qt_config->value("microProfileDialogVisible", false).toBool();
+ ReadSetting("microProfileDialogVisible", false).toBool();
qt_config->endGroup();
qt_config->beginGroup("Paths");
- UISettings::values.roms_path = qt_config->value("romsPath").toString();
- UISettings::values.symbols_path = qt_config->value("symbolsPath").toString();
- UISettings::values.gamedir = qt_config->value("gameListRootDir", ".").toString();
- UISettings::values.gamedir_deepscan = qt_config->value("gameListDeepScan", false).toBool();
- UISettings::values.recent_files = qt_config->value("recentFiles").toStringList();
+ UISettings::values.roms_path = ReadSetting("romsPath").toString();
+ UISettings::values.symbols_path = ReadSetting("symbolsPath").toString();
+ UISettings::values.gamedir = ReadSetting("gameListRootDir", ".").toString();
+ UISettings::values.gamedir_deepscan = ReadSetting("gameListDeepScan", false).toBool();
+ UISettings::values.recent_files = ReadSetting("recentFiles").toStringList();
qt_config->endGroup();
qt_config->beginGroup("Shortcuts");
@@ -524,8 +518,8 @@ void Config::ReadValues() {
qt_config->beginGroup(hotkey);
UISettings::values.shortcuts.emplace_back(UISettings::Shortcut(
group + "/" + hotkey,
- UISettings::ContextualShortcut(qt_config->value("KeySeq").toString(),
- qt_config->value("Context").toInt())));
+ UISettings::ContextualShortcut(ReadSetting("KeySeq").toString(),
+ ReadSetting("Context").toInt())));
qt_config->endGroup();
}
@@ -533,16 +527,16 @@ void Config::ReadValues() {
}
qt_config->endGroup();
- UISettings::values.single_window_mode = qt_config->value("singleWindowMode", true).toBool();
- UISettings::values.fullscreen = qt_config->value("fullscreen", false).toBool();
- UISettings::values.display_titlebar = qt_config->value("displayTitleBars", true).toBool();
- UISettings::values.show_filter_bar = qt_config->value("showFilterBar", true).toBool();
- UISettings::values.show_status_bar = qt_config->value("showStatusBar", true).toBool();
- UISettings::values.confirm_before_closing = qt_config->value("confirmClose", true).toBool();
- UISettings::values.first_start = qt_config->value("firstStart", true).toBool();
- UISettings::values.callout_flags = qt_config->value("calloutFlags", 0).toUInt();
- UISettings::values.show_console = qt_config->value("showConsole", false).toBool();
- UISettings::values.profile_index = qt_config->value("profileIndex", 0).toUInt();
+ UISettings::values.single_window_mode = ReadSetting("singleWindowMode", true).toBool();
+ UISettings::values.fullscreen = ReadSetting("fullscreen", false).toBool();
+ UISettings::values.display_titlebar = ReadSetting("displayTitleBars", true).toBool();
+ UISettings::values.show_filter_bar = ReadSetting("showFilterBar", true).toBool();
+ UISettings::values.show_status_bar = ReadSetting("showStatusBar", true).toBool();
+ UISettings::values.confirm_before_closing = ReadSetting("confirmClose", true).toBool();
+ UISettings::values.first_start = ReadSetting("firstStart", true).toBool();
+ UISettings::values.callout_flags = ReadSetting("calloutFlags", 0).toUInt();
+ UISettings::values.show_console = ReadSetting("showConsole", false).toBool();
+ UISettings::values.profile_index = ReadSetting("profileIndex", 0).toUInt();
ApplyDefaultProfileIfInputInvalid();
@@ -553,62 +547,79 @@ void Config::SavePlayerValues() {
for (std::size_t p = 0; p < Settings::values.players.size(); ++p) {
const auto& player = Settings::values.players[p];
- qt_config->setValue(QString("player_%1_connected").arg(p), player.connected);
- qt_config->setValue(QString("player_%1_type").arg(p), static_cast<u8>(player.type));
+ WriteSetting(QString("player_%1_connected").arg(p), player.connected, false);
+ WriteSetting(QString("player_%1_type").arg(p), static_cast<u8>(player.type),
+ static_cast<u8>(Settings::ControllerType::DualJoycon));
- qt_config->setValue(QString("player_%1_body_color_left").arg(p), player.body_color_left);
- qt_config->setValue(QString("player_%1_body_color_right").arg(p), player.body_color_right);
- qt_config->setValue(QString("player_%1_button_color_left").arg(p),
- player.button_color_left);
- qt_config->setValue(QString("player_%1_button_color_right").arg(p),
- player.button_color_right);
+ WriteSetting(QString("player_%1_body_color_left").arg(p), player.body_color_left,
+ Settings::JOYCON_BODY_NEON_BLUE);
+ WriteSetting(QString("player_%1_body_color_right").arg(p), player.body_color_right,
+ Settings::JOYCON_BODY_NEON_RED);
+ WriteSetting(QString("player_%1_button_color_left").arg(p), player.button_color_left,
+ Settings::JOYCON_BUTTONS_NEON_BLUE);
+ WriteSetting(QString("player_%1_button_color_right").arg(p), player.button_color_right,
+ Settings::JOYCON_BUTTONS_NEON_RED);
for (int i = 0; i < Settings::NativeButton::NumButtons; ++i) {
- qt_config->setValue(QString("player_%1_").arg(p) +
- QString::fromStdString(Settings::NativeButton::mapping[i]),
- QString::fromStdString(player.buttons[i]));
+ std::string default_param = InputCommon::GenerateKeyboardParam(default_buttons[i]);
+ WriteSetting(QString("player_%1_").arg(p) +
+ QString::fromStdString(Settings::NativeButton::mapping[i]),
+ QString::fromStdString(player.buttons[i]),
+ QString::fromStdString(default_param));
}
for (int i = 0; i < Settings::NativeAnalog::NumAnalogs; ++i) {
- qt_config->setValue(QString("player_%1_").arg(p) +
- QString::fromStdString(Settings::NativeAnalog::mapping[i]),
- QString::fromStdString(player.analogs[i]));
+ std::string default_param = InputCommon::GenerateAnalogParamFromKeys(
+ default_analogs[i][0], default_analogs[i][1], default_analogs[i][2],
+ default_analogs[i][3], default_analogs[i][4], 0.5f);
+ WriteSetting(QString("player_%1_").arg(p) +
+ QString::fromStdString(Settings::NativeAnalog::mapping[i]),
+ QString::fromStdString(player.analogs[i]),
+ QString::fromStdString(default_param));
}
}
}
void Config::SaveDebugValues() {
- qt_config->setValue("debug_pad_enabled", Settings::values.debug_pad_enabled);
+ WriteSetting("debug_pad_enabled", Settings::values.debug_pad_enabled, false);
for (int i = 0; i < Settings::NativeButton::NumButtons; ++i) {
- qt_config->setValue(QString("debug_pad_") +
- QString::fromStdString(Settings::NativeButton::mapping[i]),
- QString::fromStdString(Settings::values.debug_pad_buttons[i]));
+ std::string default_param = InputCommon::GenerateKeyboardParam(default_buttons[i]);
+ WriteSetting(QString("debug_pad_") +
+ QString::fromStdString(Settings::NativeButton::mapping[i]),
+ QString::fromStdString(Settings::values.debug_pad_buttons[i]),
+ QString::fromStdString(default_param));
}
for (int i = 0; i < Settings::NativeAnalog::NumAnalogs; ++i) {
- qt_config->setValue(QString("debug_pad_") +
- QString::fromStdString(Settings::NativeAnalog::mapping[i]),
- QString::fromStdString(Settings::values.debug_pad_analogs[i]));
+ std::string default_param = InputCommon::GenerateAnalogParamFromKeys(
+ default_analogs[i][0], default_analogs[i][1], default_analogs[i][2],
+ default_analogs[i][3], default_analogs[i][4], 0.5f);
+ WriteSetting(QString("debug_pad_") +
+ QString::fromStdString(Settings::NativeAnalog::mapping[i]),
+ QString::fromStdString(Settings::values.debug_pad_analogs[i]),
+ QString::fromStdString(default_param));
}
}
void Config::SaveMouseValues() {
- qt_config->setValue("mouse_enabled", Settings::values.mouse_enabled);
+ WriteSetting("mouse_enabled", Settings::values.mouse_enabled, false);
for (int i = 0; i < Settings::NativeMouseButton::NumMouseButtons; ++i) {
- qt_config->setValue(QString("mouse_") +
- QString::fromStdString(Settings::NativeMouseButton::mapping[i]),
- QString::fromStdString(Settings::values.mouse_buttons[i]));
+ std::string default_param = InputCommon::GenerateKeyboardParam(default_mouse_buttons[i]);
+ WriteSetting(QString("mouse_") +
+ QString::fromStdString(Settings::NativeMouseButton::mapping[i]),
+ QString::fromStdString(Settings::values.mouse_buttons[i]),
+ QString::fromStdString(default_param));
}
}
void Config::SaveTouchscreenValues() {
- qt_config->setValue("touchscreen_enabled", Settings::values.touchscreen.enabled);
- qt_config->setValue("touchscreen_device",
- QString::fromStdString(Settings::values.touchscreen.device));
-
- qt_config->setValue("touchscreen_finger", Settings::values.touchscreen.finger);
- qt_config->setValue("touchscreen_angle", Settings::values.touchscreen.rotation_angle);
- qt_config->setValue("touchscreen_diameter_x", Settings::values.touchscreen.diameter_x);
- qt_config->setValue("touchscreen_diameter_y", Settings::values.touchscreen.diameter_y);
+ WriteSetting("touchscreen_enabled", Settings::values.touchscreen.enabled, true);
+ WriteSetting("touchscreen_device", QString::fromStdString(Settings::values.touchscreen.device),
+ "engine:emu_window");
+
+ WriteSetting("touchscreen_finger", Settings::values.touchscreen.finger, 0);
+ WriteSetting("touchscreen_angle", Settings::values.touchscreen.rotation_angle, 0);
+ WriteSetting("touchscreen_diameter_x", Settings::values.touchscreen.diameter_x, 15);
+ WriteSetting("touchscreen_diameter_y", Settings::values.touchscreen.diameter_y, 15);
}
void Config::SaveValues() {
@@ -619,91 +630,96 @@ void Config::SaveValues() {
SaveMouseValues();
SaveTouchscreenValues();
- qt_config->setValue("motion_device", QString::fromStdString(Settings::values.motion_device));
- qt_config->setValue("keyboard_enabled", Settings::values.keyboard_enabled);
+ WriteSetting("motion_device", QString::fromStdString(Settings::values.motion_device),
+ "engine:motion_emu,update_period:100,sensitivity:0.01");
+ WriteSetting("keyboard_enabled", Settings::values.keyboard_enabled, false);
qt_config->endGroup();
qt_config->beginGroup("Core");
- qt_config->setValue("use_cpu_jit", Settings::values.use_cpu_jit);
- qt_config->setValue("use_multi_core", Settings::values.use_multi_core);
+ WriteSetting("use_cpu_jit", Settings::values.use_cpu_jit, true);
+ WriteSetting("use_multi_core", Settings::values.use_multi_core, false);
qt_config->endGroup();
qt_config->beginGroup("Renderer");
- qt_config->setValue("resolution_factor", (double)Settings::values.resolution_factor);
- qt_config->setValue("use_frame_limit", Settings::values.use_frame_limit);
- qt_config->setValue("frame_limit", Settings::values.frame_limit);
- qt_config->setValue("use_disk_shader_cache", Settings::values.use_disk_shader_cache);
- qt_config->setValue("use_accurate_gpu_emulation", Settings::values.use_accurate_gpu_emulation);
- qt_config->setValue("use_asynchronous_gpu_emulation",
- Settings::values.use_asynchronous_gpu_emulation);
+ WriteSetting("resolution_factor", (double)Settings::values.resolution_factor, 1.0);
+ WriteSetting("use_frame_limit", Settings::values.use_frame_limit, true);
+ WriteSetting("frame_limit", Settings::values.frame_limit, 100);
+ WriteSetting("use_disk_shader_cache", Settings::values.use_disk_shader_cache, true);
+ WriteSetting("use_accurate_gpu_emulation", Settings::values.use_accurate_gpu_emulation, false);
+ WriteSetting("use_asynchronous_gpu_emulation", Settings::values.use_asynchronous_gpu_emulation,
+ false);
// Cast to double because Qt's written float values are not human-readable
- qt_config->setValue("bg_red", (double)Settings::values.bg_red);
- qt_config->setValue("bg_green", (double)Settings::values.bg_green);
- qt_config->setValue("bg_blue", (double)Settings::values.bg_blue);
+ WriteSetting("bg_red", (double)Settings::values.bg_red, 0.0);
+ WriteSetting("bg_green", (double)Settings::values.bg_green, 0.0);
+ WriteSetting("bg_blue", (double)Settings::values.bg_blue, 0.0);
qt_config->endGroup();
qt_config->beginGroup("Audio");
- qt_config->setValue("output_engine", QString::fromStdString(Settings::values.sink_id));
- qt_config->setValue("enable_audio_stretching", Settings::values.enable_audio_stretching);
- qt_config->setValue("output_device", QString::fromStdString(Settings::values.audio_device_id));
- qt_config->setValue("volume", Settings::values.volume);
+ WriteSetting("output_engine", QString::fromStdString(Settings::values.sink_id), "auto");
+ WriteSetting("enable_audio_stretching", Settings::values.enable_audio_stretching, true);
+ WriteSetting("output_device", QString::fromStdString(Settings::values.audio_device_id), "auto");
+ WriteSetting("volume", Settings::values.volume, 1.0f);
qt_config->endGroup();
qt_config->beginGroup("Data Storage");
- qt_config->setValue("use_virtual_sd", Settings::values.use_virtual_sd);
- qt_config->setValue("nand_directory",
- QString::fromStdString(FileUtil::GetUserPath(FileUtil::UserPath::NANDDir)));
- qt_config->setValue("sdmc_directory",
- QString::fromStdString(FileUtil::GetUserPath(FileUtil::UserPath::SDMCDir)));
+ WriteSetting("use_virtual_sd", Settings::values.use_virtual_sd, true);
+ WriteSetting("nand_directory",
+ QString::fromStdString(FileUtil::GetUserPath(FileUtil::UserPath::NANDDir)),
+ QString::fromStdString(FileUtil::GetUserPath(FileUtil::UserPath::NANDDir)));
+ WriteSetting("sdmc_directory",
+ QString::fromStdString(FileUtil::GetUserPath(FileUtil::UserPath::SDMCDir)),
+ QString::fromStdString(FileUtil::GetUserPath(FileUtil::UserPath::SDMCDir)));
qt_config->endGroup();
qt_config->beginGroup("System");
- qt_config->setValue("use_docked_mode", Settings::values.use_docked_mode);
- qt_config->setValue("enable_nfc", Settings::values.enable_nfc);
- qt_config->setValue("current_user", Settings::values.current_user);
- qt_config->setValue("language_index", Settings::values.language_index);
+ WriteSetting("use_docked_mode", Settings::values.use_docked_mode, false);
+ WriteSetting("enable_nfc", Settings::values.enable_nfc, true);
+ WriteSetting("current_user", Settings::values.current_user, 0);
+ WriteSetting("language_index", Settings::values.language_index, 1);
- qt_config->setValue("rng_seed_enabled", Settings::values.rng_seed.has_value());
- qt_config->setValue("rng_seed", Settings::values.rng_seed.value_or(0));
+ WriteSetting("rng_seed_enabled", Settings::values.rng_seed.has_value(), false);
+ WriteSetting("rng_seed", Settings::values.rng_seed.value_or(0), 0);
- qt_config->setValue("custom_rtc_enabled", Settings::values.custom_rtc.has_value());
- qt_config->setValue("custom_rtc",
- QVariant::fromValue<long long>(
- Settings::values.custom_rtc.value_or(std::chrono::seconds{}).count()));
+ WriteSetting("custom_rtc_enabled", Settings::values.custom_rtc.has_value(), false);
+ WriteSetting("custom_rtc",
+ QVariant::fromValue<long long>(
+ Settings::values.custom_rtc.value_or(std::chrono::seconds{}).count()),
+ 0);
qt_config->endGroup();
qt_config->beginGroup("Miscellaneous");
- qt_config->setValue("log_filter", QString::fromStdString(Settings::values.log_filter));
- qt_config->setValue("use_dev_keys", Settings::values.use_dev_keys);
+ WriteSetting("log_filter", QString::fromStdString(Settings::values.log_filter), "*:Info");
+ WriteSetting("use_dev_keys", Settings::values.use_dev_keys, false);
qt_config->endGroup();
qt_config->beginGroup("Debugging");
- qt_config->setValue("use_gdbstub", Settings::values.use_gdbstub);
- qt_config->setValue("gdbstub_port", Settings::values.gdbstub_port);
- qt_config->setValue("program_args", QString::fromStdString(Settings::values.program_args));
- qt_config->setValue("dump_exefs", Settings::values.dump_exefs);
- qt_config->setValue("dump_nso", Settings::values.dump_nso);
+ WriteSetting("use_gdbstub", Settings::values.use_gdbstub, false);
+ WriteSetting("gdbstub_port", Settings::values.gdbstub_port, 24689);
+ WriteSetting("program_args", QString::fromStdString(Settings::values.program_args), "");
+ WriteSetting("dump_exefs", Settings::values.dump_exefs, false);
+ WriteSetting("dump_nso", Settings::values.dump_nso, false);
qt_config->endGroup();
qt_config->beginGroup("WebService");
- qt_config->setValue("enable_telemetry", Settings::values.enable_telemetry);
- qt_config->setValue("web_api_url", QString::fromStdString(Settings::values.web_api_url));
- qt_config->setValue("yuzu_username", QString::fromStdString(Settings::values.yuzu_username));
- qt_config->setValue("yuzu_token", QString::fromStdString(Settings::values.yuzu_token));
+ WriteSetting("enable_telemetry", Settings::values.enable_telemetry, true);
+ WriteSetting("web_api_url", QString::fromStdString(Settings::values.web_api_url),
+ "https://api.yuzu-emu.org");
+ WriteSetting("yuzu_username", QString::fromStdString(Settings::values.yuzu_username));
+ WriteSetting("yuzu_token", QString::fromStdString(Settings::values.yuzu_token));
qt_config->endGroup();
qt_config->beginWriteArray("DisabledAddOns");
int i = 0;
for (const auto& elem : Settings::values.disabled_addons) {
qt_config->setArrayIndex(i);
- qt_config->setValue("title_id", QVariant::fromValue<u64>(elem.first));
+ WriteSetting("title_id", QVariant::fromValue<u64>(elem.first), 0);
qt_config->beginWriteArray("disabled");
for (std::size_t j = 0; j < elem.second.size(); ++j) {
qt_config->setArrayIndex(static_cast<int>(j));
- qt_config->setValue("d", QString::fromStdString(elem.second[j]));
+ WriteSetting("d", QString::fromStdString(elem.second[j]), "");
}
qt_config->endArray();
++i;
@@ -711,60 +727,86 @@ void Config::SaveValues() {
qt_config->endArray();
qt_config->beginGroup("UI");
- qt_config->setValue("theme", UISettings::values.theme);
- qt_config->setValue("enable_discord_presence", UISettings::values.enable_discord_presence);
- qt_config->setValue("screenshot_resolution_factor",
- UISettings::values.screenshot_resolution_factor);
- qt_config->setValue("select_user_on_boot", UISettings::values.select_user_on_boot);
+ WriteSetting("theme", UISettings::values.theme, UISettings::themes[0].second);
+ WriteSetting("enable_discord_presence", UISettings::values.enable_discord_presence, true);
+ WriteSetting("screenshot_resolution_factor", UISettings::values.screenshot_resolution_factor,
+ 0);
+ WriteSetting("select_user_on_boot", UISettings::values.select_user_on_boot, false);
qt_config->beginGroup("UIGameList");
- qt_config->setValue("show_unknown", UISettings::values.show_unknown);
- qt_config->setValue("show_add_ons", UISettings::values.show_add_ons);
- qt_config->setValue("icon_size", UISettings::values.icon_size);
- qt_config->setValue("row_1_text_id", UISettings::values.row_1_text_id);
- qt_config->setValue("row_2_text_id", UISettings::values.row_2_text_id);
+ WriteSetting("show_unknown", UISettings::values.show_unknown, true);
+ WriteSetting("show_add_ons", UISettings::values.show_add_ons, true);
+ WriteSetting("icon_size", UISettings::values.icon_size, 64);
+ WriteSetting("row_1_text_id", UISettings::values.row_1_text_id, 3);
+ WriteSetting("row_2_text_id", UISettings::values.row_2_text_id, 2);
qt_config->endGroup();
qt_config->beginGroup("UILayout");
- qt_config->setValue("geometry", UISettings::values.geometry);
- qt_config->setValue("state", UISettings::values.state);
- qt_config->setValue("geometryRenderWindow", UISettings::values.renderwindow_geometry);
- qt_config->setValue("gameListHeaderState", UISettings::values.gamelist_header_state);
- qt_config->setValue("microProfileDialogGeometry", UISettings::values.microprofile_geometry);
- qt_config->setValue("microProfileDialogVisible", UISettings::values.microprofile_visible);
+ WriteSetting("geometry", UISettings::values.geometry);
+ WriteSetting("state", UISettings::values.state);
+ WriteSetting("geometryRenderWindow", UISettings::values.renderwindow_geometry);
+ WriteSetting("gameListHeaderState", UISettings::values.gamelist_header_state);
+ WriteSetting("microProfileDialogGeometry", UISettings::values.microprofile_geometry);
+ WriteSetting("microProfileDialogVisible", UISettings::values.microprofile_visible, false);
qt_config->endGroup();
qt_config->beginGroup("Paths");
- qt_config->setValue("romsPath", UISettings::values.roms_path);
- qt_config->setValue("symbolsPath", UISettings::values.symbols_path);
- qt_config->setValue("screenshotPath", UISettings::values.screenshot_path);
- qt_config->setValue("gameListRootDir", UISettings::values.gamedir);
- qt_config->setValue("gameListDeepScan", UISettings::values.gamedir_deepscan);
- qt_config->setValue("recentFiles", UISettings::values.recent_files);
+ WriteSetting("romsPath", UISettings::values.roms_path);
+ WriteSetting("symbolsPath", UISettings::values.symbols_path);
+ WriteSetting("screenshotPath", UISettings::values.screenshot_path);
+ WriteSetting("gameListRootDir", UISettings::values.gamedir, ".");
+ WriteSetting("gameListDeepScan", UISettings::values.gamedir_deepscan, false);
+ WriteSetting("recentFiles", UISettings::values.recent_files);
qt_config->endGroup();
qt_config->beginGroup("Shortcuts");
for (auto shortcut : UISettings::values.shortcuts) {
- qt_config->setValue(shortcut.first + "/KeySeq", shortcut.second.first);
- qt_config->setValue(shortcut.first + "/Context", shortcut.second.second);
+ WriteSetting(shortcut.first + "/KeySeq", shortcut.second.first);
+ WriteSetting(shortcut.first + "/Context", shortcut.second.second);
}
qt_config->endGroup();
- qt_config->setValue("singleWindowMode", UISettings::values.single_window_mode);
- qt_config->setValue("fullscreen", UISettings::values.fullscreen);
- qt_config->setValue("displayTitleBars", UISettings::values.display_titlebar);
- qt_config->setValue("showFilterBar", UISettings::values.show_filter_bar);
- qt_config->setValue("showStatusBar", UISettings::values.show_status_bar);
- qt_config->setValue("confirmClose", UISettings::values.confirm_before_closing);
- qt_config->setValue("firstStart", UISettings::values.first_start);
- qt_config->setValue("calloutFlags", UISettings::values.callout_flags);
- qt_config->setValue("showConsole", UISettings::values.show_console);
- qt_config->setValue("profileIndex", UISettings::values.profile_index);
+ WriteSetting("singleWindowMode", UISettings::values.single_window_mode, true);
+ WriteSetting("fullscreen", UISettings::values.fullscreen, false);
+ WriteSetting("displayTitleBars", UISettings::values.display_titlebar, true);
+ WriteSetting("showFilterBar", UISettings::values.show_filter_bar, true);
+ WriteSetting("showStatusBar", UISettings::values.show_status_bar, true);
+ WriteSetting("confirmClose", UISettings::values.confirm_before_closing, true);
+ WriteSetting("firstStart", UISettings::values.first_start, true);
+ WriteSetting("calloutFlags", UISettings::values.callout_flags, 0);
+ WriteSetting("showConsole", UISettings::values.show_console, false);
+ WriteSetting("profileIndex", UISettings::values.profile_index, 0);
qt_config->endGroup();
}
+QVariant Config::ReadSetting(const QString& name) const {
+ return qt_config->value(name);
+}
+
+QVariant Config::ReadSetting(const QString& name, const QVariant& default_value) const {
+ QVariant result;
+ if (qt_config->value(name + "/default", false).toBool()) {
+ result = default_value;
+ } else {
+ result = qt_config->value(name, default_value);
+ }
+ return result;
+}
+
+void Config::WriteSetting(const QString& name, const QVariant& value) {
+ qt_config->setValue(name, value);
+}
+
+void Config::WriteSetting(const QString& name, const QVariant& value,
+ const QVariant& default_value) {
+ qt_config->setValue(name + "/default", value == default_value);
+ qt_config->setValue(name, value);
+}
+
void Config::Reload() {
ReadValues();
+ // To apply default value changes
+ SaveValues();
Settings::Apply();
}
diff --git a/src/yuzu/configuration/config.h b/src/yuzu/configuration/config.h
index e73ad19bb..f4185db18 100644
--- a/src/yuzu/configuration/config.h
+++ b/src/yuzu/configuration/config.h
@@ -42,6 +42,11 @@ private:
void SaveMouseValues();
void SaveTouchscreenValues();
+ QVariant ReadSetting(const QString& name) const;
+ QVariant ReadSetting(const QString& name, const QVariant& default_value) const;
+ void WriteSetting(const QString& name, const QVariant& value);
+ void WriteSetting(const QString& name, const QVariant& value, const QVariant& default_value);
+
std::unique_ptr<QSettings> qt_config;
std::string qt_config_loc;
};
diff --git a/src/yuzu/debugger/graphics/graphics_surface.cpp b/src/yuzu/debugger/graphics/graphics_surface.cpp
index 71683da8e..29f01dfb2 100644
--- a/src/yuzu/debugger/graphics/graphics_surface.cpp
+++ b/src/yuzu/debugger/graphics/graphics_surface.cpp
@@ -383,13 +383,12 @@ void GraphicsSurfaceWidget::OnUpdate() {
// TODO: Implement a good way to visualize alpha components!
QImage decoded_image(surface_width, surface_height, QImage::Format_ARGB32);
- std::optional<VAddr> address = gpu.MemoryManager().GpuToCpuAddress(surface_address);
// TODO(bunnei): Will not work with BCn formats that swizzle 4x4 tiles.
// Needs to be fixed if we plan to use this feature more, otherwise we may remove it.
auto unswizzled_data = Tegra::Texture::UnswizzleTexture(
- *address, 1, 1, Tegra::Texture::BytesPerPixel(surface_format), surface_width,
- surface_height, 1U);
+ gpu.MemoryManager().GetPointer(surface_address), 1, 1,
+ Tegra::Texture::BytesPerPixel(surface_format), surface_width, surface_height, 1U);
auto texture_data = Tegra::Texture::DecodeTexture(unswizzled_data, surface_format,
surface_width, surface_height);
diff --git a/src/yuzu_cmd/emu_window/emu_window_sdl2.cpp b/src/yuzu_cmd/emu_window/emu_window_sdl2.cpp
index 7df8eff53..de7a26e14 100644
--- a/src/yuzu_cmd/emu_window/emu_window_sdl2.cpp
+++ b/src/yuzu_cmd/emu_window/emu_window_sdl2.cpp
@@ -135,16 +135,16 @@ bool EmuWindow_SDL2::SupportsRequiredGLExtensions() {
}
EmuWindow_SDL2::EmuWindow_SDL2(bool fullscreen) {
- InputCommon::Init();
-
- SDL_SetMainReady();
-
// Initialize the window
if (SDL_Init(SDL_INIT_VIDEO | SDL_INIT_JOYSTICK) < 0) {
LOG_CRITICAL(Frontend, "Failed to initialize SDL2! Exiting...");
exit(1);
}
+ InputCommon::Init();
+
+ SDL_SetMainReady();
+
SDL_GL_SetAttribute(SDL_GL_CONTEXT_MAJOR_VERSION, 4);
SDL_GL_SetAttribute(SDL_GL_CONTEXT_MINOR_VERSION, 3);
SDL_GL_SetAttribute(SDL_GL_CONTEXT_PROFILE_MASK, SDL_GL_CONTEXT_PROFILE_CORE);
@@ -201,11 +201,9 @@ EmuWindow_SDL2::EmuWindow_SDL2(bool fullscreen) {
}
EmuWindow_SDL2::~EmuWindow_SDL2() {
- InputCommon::SDL::CloseSDLJoysticks();
+ InputCommon::Shutdown();
SDL_GL_DeleteContext(gl_context);
SDL_Quit();
-
- InputCommon::Shutdown();
}
void EmuWindow_SDL2::SwapBuffers() {
@@ -262,7 +260,6 @@ void EmuWindow_SDL2::PollEvents() {
is_open = false;
break;
default:
- InputCommon::SDL::HandleGameControllerEvent(event);
break;
}
}