summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorFeng Chen <VonChenPlus@gmail.com>2021-07-08 20:06:09 +0200
committerGitHub <noreply@github.com>2021-07-08 20:06:09 +0200
commitc7ad195fd3eeea380465be48264f4e69165178c7 (patch)
tree77a348320c12feaca4248d090c6d686de7bde39a
parentMerge pull request #6564 from Kelebek1/Audio (diff)
downloadyuzu-c7ad195fd3eeea380465be48264f4e69165178c7.tar
yuzu-c7ad195fd3eeea380465be48264f4e69165178c7.tar.gz
yuzu-c7ad195fd3eeea380465be48264f4e69165178c7.tar.bz2
yuzu-c7ad195fd3eeea380465be48264f4e69165178c7.tar.lz
yuzu-c7ad195fd3eeea380465be48264f4e69165178c7.tar.xz
yuzu-c7ad195fd3eeea380465be48264f4e69165178c7.tar.zst
yuzu-c7ad195fd3eeea380465be48264f4e69165178c7.zip
-rw-r--r--src/video_core/engines/fermi_2d.cpp22
-rw-r--r--src/video_core/texture_cache/texture_cache.h71
2 files changed, 35 insertions, 58 deletions
diff --git a/src/video_core/engines/fermi_2d.cpp b/src/video_core/engines/fermi_2d.cpp
index 0f640fdae..f26530ede 100644
--- a/src/video_core/engines/fermi_2d.cpp
+++ b/src/video_core/engines/fermi_2d.cpp
@@ -7,6 +7,10 @@
#include "video_core/engines/fermi_2d.h"
#include "video_core/memory_manager.h"
#include "video_core/rasterizer_interface.h"
+#include "video_core/surface.h"
+
+using VideoCore::Surface::BytesPerBlock;
+using VideoCore::Surface::PixelFormatFromRenderTargetFormat;
namespace Tegra::Engines {
@@ -49,7 +53,7 @@ void Fermi2D::Blit() {
UNIMPLEMENTED_IF_MSG(regs.clip_enable != 0, "Clipped blit enabled");
const auto& args = regs.pixels_from_memory;
- const Config config{
+ Config config{
.operation = regs.operation,
.filter = args.sample_mode.filter,
.dst_x0 = args.dst_x0,
@@ -61,7 +65,21 @@ void Fermi2D::Blit() {
.src_x1 = static_cast<s32>((args.du_dx * args.dst_width + args.src_x0) >> 32),
.src_y1 = static_cast<s32>((args.dv_dy * args.dst_height + args.src_y0) >> 32),
};
- if (!rasterizer->AccelerateSurfaceCopy(regs.src, regs.dst, config)) {
+ Surface src = regs.src;
+ const auto bytes_per_pixel = BytesPerBlock(PixelFormatFromRenderTargetFormat(src.format));
+ const auto need_align_to_pitch =
+ src.linear == Tegra::Engines::Fermi2D::MemoryLayout::Pitch &&
+ static_cast<s32>(src.width) == config.src_x1 &&
+ config.src_x1 > static_cast<s32>(src.pitch / bytes_per_pixel) && config.src_x0 > 0;
+ if (need_align_to_pitch) {
+ auto address = src.Address() + config.src_x0 * bytes_per_pixel;
+ src.addr_upper = static_cast<u32>(address >> 32);
+ src.addr_lower = static_cast<u32>(address);
+ src.width -= config.src_x0;
+ config.src_x1 -= config.src_x0;
+ config.src_x0 = 0;
+ }
+ if (!rasterizer->AccelerateSurfaceCopy(src, regs.dst, config)) {
UNIMPLEMENTED();
}
}
diff --git a/src/video_core/texture_cache/texture_cache.h b/src/video_core/texture_cache/texture_cache.h
index e3542301e..01de2d498 100644
--- a/src/video_core/texture_cache/texture_cache.h
+++ b/src/video_core/texture_cache/texture_cache.h
@@ -159,9 +159,7 @@ public:
/// Blit an image with the given parameters
void BlitImage(const Tegra::Engines::Fermi2D::Surface& dst,
const Tegra::Engines::Fermi2D::Surface& src,
- const Tegra::Engines::Fermi2D::Config& copy,
- std::optional<Region2D> src_region_override = {},
- std::optional<Region2D> dst_region_override = {});
+ const Tegra::Engines::Fermi2D::Config& copy);
/// Invalidate the contents of the color buffer index
/// These contents become unspecified, the cache can assume aggressive optimizations.
@@ -760,9 +758,7 @@ void TextureCache<P>::UnmapGPUMemory(GPUVAddr gpu_addr, size_t size) {
template <class P>
void TextureCache<P>::BlitImage(const Tegra::Engines::Fermi2D::Surface& dst,
const Tegra::Engines::Fermi2D::Surface& src,
- const Tegra::Engines::Fermi2D::Config& copy,
- std::optional<Region2D> src_override,
- std::optional<Region2D> dst_override) {
+ const Tegra::Engines::Fermi2D::Config& copy) {
const BlitImages images = GetBlitImages(dst, src);
const ImageId dst_id = images.dst_id;
const ImageId src_id = images.src_id;
@@ -773,47 +769,25 @@ void TextureCache<P>::BlitImage(const Tegra::Engines::Fermi2D::Surface& dst,
const ImageBase& src_image = slot_images[src_id];
// TODO: Deduplicate
- const std::optional dst_base = dst_image.TryFindBase(dst.Address());
- const SubresourceRange dst_range{.base = dst_base.value(), .extent = {1, 1}};
- const ImageViewInfo dst_view_info(ImageViewType::e2D, images.dst_format, dst_range);
- const auto [dst_framebuffer_id, dst_view_id] = RenderTargetFromImage(dst_id, dst_view_info);
- const auto [src_samples_x, src_samples_y] = SamplesLog2(src_image.info.num_samples);
-
- // out of bounds texture blit checking
- const bool use_override = src_override.has_value();
- const s32 src_x0 = copy.src_x0 >> src_samples_x;
- s32 src_x1 = use_override ? src_override->end.x : copy.src_x1 >> src_samples_x;
- const s32 src_y0 = copy.src_y0 >> src_samples_y;
- const s32 src_y1 = copy.src_y1 >> src_samples_y;
-
- const auto src_width = static_cast<s32>(src_image.info.size.width);
- const bool width_oob = src_x1 > src_width;
- const auto width_diff = width_oob ? src_x1 - src_width : 0;
- if (width_oob) {
- src_x1 = src_width;
- }
-
- const Region2D src_dimensions{
- Offset2D{.x = src_x0, .y = src_y0},
- Offset2D{.x = src_x1, .y = src_y1},
- };
- const auto src_region = use_override ? *src_override : src_dimensions;
-
const std::optional src_base = src_image.TryFindBase(src.Address());
const SubresourceRange src_range{.base = src_base.value(), .extent = {1, 1}};
const ImageViewInfo src_view_info(ImageViewType::e2D, images.src_format, src_range);
const auto [src_framebuffer_id, src_view_id] = RenderTargetFromImage(src_id, src_view_info);
- const auto [dst_samples_x, dst_samples_y] = SamplesLog2(dst_image.info.num_samples);
+ const auto [src_samples_x, src_samples_y] = SamplesLog2(src_image.info.num_samples);
+ const Region2D src_region{
+ Offset2D{.x = copy.src_x0 >> src_samples_x, .y = copy.src_y0 >> src_samples_y},
+ Offset2D{.x = copy.src_x1 >> src_samples_x, .y = copy.src_y1 >> src_samples_y},
+ };
- const s32 dst_x0 = copy.dst_x0 >> dst_samples_x;
- const s32 dst_x1 = copy.dst_x1 >> dst_samples_x;
- const s32 dst_y0 = copy.dst_y0 >> dst_samples_y;
- const s32 dst_y1 = copy.dst_y1 >> dst_samples_y;
- const Region2D dst_dimensions{
- Offset2D{.x = dst_x0, .y = dst_y0},
- Offset2D{.x = dst_x1 - width_diff, .y = dst_y1},
+ const std::optional dst_base = dst_image.TryFindBase(dst.Address());
+ const SubresourceRange dst_range{.base = dst_base.value(), .extent = {1, 1}};
+ const ImageViewInfo dst_view_info(ImageViewType::e2D, images.dst_format, dst_range);
+ const auto [dst_framebuffer_id, dst_view_id] = RenderTargetFromImage(dst_id, dst_view_info);
+ const auto [dst_samples_x, dst_samples_y] = SamplesLog2(dst_image.info.num_samples);
+ const Region2D dst_region{
+ Offset2D{.x = copy.dst_x0 >> dst_samples_x, .y = copy.dst_y0 >> dst_samples_y},
+ Offset2D{.x = copy.dst_x1 >> dst_samples_x, .y = copy.dst_y1 >> dst_samples_y},
};
- const auto dst_region = use_override ? *dst_override : dst_dimensions;
// Always call this after src_framebuffer_id was queried, as the address might be invalidated.
Framebuffer* const dst_framebuffer = &slot_framebuffers[dst_framebuffer_id];
@@ -830,21 +804,6 @@ void TextureCache<P>::BlitImage(const Tegra::Engines::Fermi2D::Surface& dst,
runtime.BlitImage(dst_framebuffer, dst_view, src_view, dst_region, src_region, copy.filter,
copy.operation);
}
-
- if (width_oob) {
- // Continue copy of the oob region of the texture on the next row
- auto oob_src = src;
- oob_src.height++;
- const Region2D src_region_override{
- Offset2D{.x = 0, .y = src_y0 + 1},
- Offset2D{.x = width_diff, .y = src_y1 + 1},
- };
- const Region2D dst_region_override{
- Offset2D{.x = dst_x1 - width_diff, .y = dst_y0},
- Offset2D{.x = dst_x1, .y = dst_y1},
- };
- BlitImage(dst, oob_src, copy, src_region_override, dst_region_override);
- }
}
template <class P>