Mirrors/shadPS4
1
0
Fork 0
mirror of https://github.com/shadps4-emu/shadPS4.git synced 2025-06-12 05:33:15 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 6

shader_recompiler: Small instruction parsing refactor/bugfixes (#340)

Browse source 
* translator: Implemtn f32 to f16 convert

* shader_recompiler: Add bit instructions

* shader_recompiler: More data share instructions

* shader_recompiler: Remove exec contexts, fix S_MOV_B64

* shader_recompiler: Split instruction parsing into categories

* shader_recompiler: Better BFS search

* shader_recompiler: Constant propagation pass for cmp_class_f32

* shader_recompiler: Partial readfirstlane implementation

* shader_recompiler: Stub readlane/writelane only for non-compute

* hack: Fix swizzle on RDR

* Will properly fix this when merging this

* clang format

* address_space: Bump user area size to full

* shader_recompiler: V_INTERP_MOV_F32

* Should work the same as spirv will emit flat decoration on demand

* kernel: Add MAP_OP_MAP_FLEXIBLE

* image_view: Attempt to apply storage swizzle on format

* vk_scheduler: Barrier attachments on renderpass end

* clang format

* liverpool: cs state backup

* shader_recompiler: More instructions and formats

* vector_alu: Proper V_MBCNT_U32_B32

* shader_recompiler: Port some dark souls things

* file_system: Implement sceKernelRename

* more formats

* clang format

* resource_tracking_pass: Back to assert

* translate: Tracedata

* kernel: Remove tracy lock

* Solves random crashes in Dark Souls

* code: Review comments
This commit is contained in:
TheTurtle 2024-07-31 00:32:40 +03:00 • committed by GitHub
parent ac6dc20c3b
commit a7c9bfa5c5
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
66 changed files with 1349 additions and 904 deletions
Download patch file Download diff file Expand all files Collapse all files

25
src/video_core/texture_cache/image_view.cpp
View file

@ -47,6 +47,20 @@ vk::ComponentSwizzle ConvertComponentSwizzle(u32 dst_sel) {
}
}
bool IsIdentityMapping(u32 dst_sel, u32 num_components) {
return (num_components == 1 && dst_sel == 0b100) ||
(num_components == 2 && dst_sel == 0b101'100) ||
(num_components == 3 && dst_sel == 0b110'101'100) ||
(num_components == 4 && dst_sel == 0b111'110'101'100);
}
vk::Format TrySwizzleFormat(vk::Format format, u32 dst_sel) {
if (format == vk::Format::eR8G8B8A8Unorm && dst_sel == 0b111100101110) {
return vk::Format::eB8G8R8A8Unorm;
}
return format;
}
ImageViewInfo::ImageViewInfo(const AmdGpu::Image& image, bool is_storage) noexcept
: is_storage{is_storage} {
type = ConvertImageViewType(image.GetType());
@ -60,9 +74,16 @@ ImageViewInfo::ImageViewInfo(const AmdGpu::Image& image, bool is_storage) noexce
mapping.b = ConvertComponentSwizzle(image.dst_sel_z);
mapping.a = ConvertComponentSwizzle(image.dst_sel_w);
// Check for unfortunate case of storage images being swizzled
if (is_storage && (mapping != vk::ComponentMapping{})) {
LOG_ERROR(Render_Vulkan, "Storage image requires swizzling");
const u32 num_comps = AmdGpu::NumComponents(image.GetDataFmt());
const u32 dst_sel = image.DstSelect();
if (is_storage && !IsIdentityMapping(dst_sel, num_comps)) {
mapping = vk::ComponentMapping{};
if (auto new_format = TrySwizzleFormat(format, dst_sel); new_format != format) {
format = new_format;
return;
}
LOG_ERROR(Render_Vulkan, "Storage image (num_comps = {}) requires swizzling {}", num_comps,
image.DstSelectName());
}
}

2
src/video_core/texture_cache/image_view.h
View file

@ -35,6 +35,8 @@ struct ImageViewInfo {
struct Image;
constexpr Common::SlotId NULL_IMAGE_VIEW_ID{0};
struct ImageView {
explicit ImageView(const Vulkan::Instance& instance, const ImageViewInfo& info, Image& image,
ImageId image_id, std::optional<vk::ImageUsageFlags> usage_override = {});

15
src/video_core/texture_cache/texture_cache.cpp
View file

@ -142,14 +142,14 @@ ImageId TextureCache::FindImage(const ImageInfo& info, bool refresh_on_create) {
image_ids.push_back(image_id);
});
ASSERT_MSG(image_ids.size() <= 1, "Overlapping images not allowed!");
// ASSERT_MSG(image_ids.size() <= 1, "Overlapping images not allowed!");
ImageId image_id{};
if (image_ids.empty()) {
image_id = slot_images.insert(instance, scheduler, info);
RegisterImage(image_id);
} else {
image_id = image_ids[0];
image_id = image_ids[image_ids.size() > 1 ? 1 : 0];
}
Image& image = slot_images[image_id];
@ -183,12 +183,17 @@ ImageView& TextureCache::RegisterImageView(ImageId image_id, const ImageViewInfo
}
ImageView& TextureCache::FindTexture(const ImageInfo& info, const ImageViewInfo& view_info) {
if (info.guest_address == 0) [[unlikely]] {
return slot_image_views[NULL_IMAGE_VIEW_ID];
}
const ImageId image_id = FindImage(info);
Image& image = slot_images[image_id];
auto& usage = image.info.usage;
if (view_info.is_storage) {
image.Transit(vk::ImageLayout::eGeneral, vk::AccessFlagBits::eShaderWrite);
image.Transit(vk::ImageLayout::eGeneral,
vk::AccessFlagBits::eShaderRead | vk::AccessFlagBits::eShaderWrite);
usage.storage = true;
} else {
const auto new_layout = image.info.IsDepthStencil()
@ -206,7 +211,7 @@ ImageView& TextureCache::FindTexture(const ImageInfo& info, const ImageViewInfo&
view_info_tmp.range.extent.levels > image.info.resources.levels ||
view_info_tmp.range.extent.layers > image.info.resources.layers) {
LOG_ERROR(Render_Vulkan,
LOG_DEBUG(Render_Vulkan,
"Subresource range ({}~{},{}~{}) exceeds base image extents ({},{})",
view_info_tmp.range.base.level, view_info_tmp.range.extent.levels,
view_info_tmp.range.base.layer, view_info_tmp.range.extent.layers,
@ -341,7 +346,7 @@ void TextureCache::RefreshImage(Image& image) {
cmdbuf.copyBufferToImage(buffer, image.image, vk::ImageLayout::eTransferDstOptimal, image_copy);
image.Transit(vk::ImageLayout::eGeneral,
vk::AccessFlagBits::eShaderRead | vk::AccessFlagBits::eTransferRead);
vk::AccessFlagBits::eMemoryWrite | vk::AccessFlagBits::eMemoryRead);
}
vk::Sampler TextureCache::GetSampler(const AmdGpu::Sampler& sampler) {