Skip to content

Commit

Permalink
shader_recompiler: Use SRT flatbuf for bounds check size.
Browse files Browse the repository at this point in the history
  • Loading branch information
@squidbus
squidbus committed Feb 15, 2025
1 parent 9ceb577 commit 507722b
Show file tree
Hide file tree
Showing 6 changed files with 61 additions and 49 deletions.
9 changes: 8 additions & 1 deletion src/shader_recompiler/backend/spirv/emit_spirv_context_get_set.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -178,7 +178,14 @@ Id EmitReadConstBuffer(EmitContext& ctx, u32 handle, Id index) {
index = ctx.OpIAdd(ctx.U32[1], index, buffer.offset_dwords);
const auto [id, pointer_type] = buffer[BufferAlias::U32];
const Id ptr{ctx.OpAccessChain(pointer_type, id, ctx.u32_zero_value, index)};
return ctx.OpLoad(ctx.U32[1], ptr);
const Id result{ctx.OpLoad(ctx.U32[1], ptr)};

if (Sirit::ValidId(buffer.size_dwords)) {
const Id in_bounds = ctx.OpULessThan(ctx.U1[1], index, buffer.size_dwords);
return ctx.OpSelect(ctx.U32[1], in_bounds, result, ctx.u32_zero_value);
} else {
return result;
}
}

Id EmitReadStepRate(EmitContext& ctx, int rate_idx) {
Expand Down
80 changes: 45 additions & 35 deletions src/shader_recompiler/backend/spirv/spirv_emit_context.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -192,32 +192,49 @@ EmitContext::SpirvAttribute EmitContext::GetAttributeInfo(AmdGpu::NumberFormat f
UNREACHABLE_MSG("Invalid attribute type {}", fmt);
}

Id EmitContext::GetBufferSize(const u32 sharp_idx) {
const auto& srt_flatbuf = buffers.back();
ASSERT(srt_flatbuf.buffer_type == BufferType::ReadConstUbo);
const auto [id, pointer_type] = srt_flatbuf[BufferAlias::U32];

const auto rsrc1{
OpLoad(U32[1], OpAccessChain(pointer_type, id, u32_zero_value, ConstU32(sharp_idx + 1)))};
const auto rsrc2{
OpLoad(U32[1], OpAccessChain(pointer_type, id, u32_zero_value, ConstU32(sharp_idx + 2)))};

const auto stride{OpBitFieldUExtract(U32[1], rsrc1, ConstU32(16u), ConstU32(14u))};
const auto num_records{rsrc2};

const auto stride_zero{OpIEqual(U1[1], stride, u32_zero_value)};
const auto stride_size{OpIMul(U32[1], num_records, stride)};
return OpSelect(U32[1], stride_zero, num_records, stride_size);
}

void EmitContext::DefineBufferProperties() {
for (BufferDefinition& buffer : buffers) {
if (buffer.buffer_type != BufferType::Guest) {
continue;
}
const u32 binding = buffer.binding;
const u32 offset_half = PushData::BufOffsetIndex + (binding >> 4);
const u32 offset_comp = (binding & 0xf) >> 2;
const u32 offset_bit = (binding & 0x3) << 3;
const Id offset_ptr{OpAccessChain(TypePointer(spv::StorageClass::PushConstant, U32[1]),
push_data_block, ConstU32(offset_half),
ConstU32(offset_comp))};
const Id offset_value{OpLoad(U32[1], offset_ptr)};
buffer.offset =
OpBitFieldUExtract(U32[1], offset_value, ConstU32(offset_bit), ConstU32(8U));
const u32 half = PushData::BufOffsetIndex + (binding >> 4);
const u32 comp = (binding & 0xf) >> 2;
const u32 offset = (binding & 0x3) << 3;
const Id ptr{OpAccessChain(TypePointer(spv::StorageClass::PushConstant, U32[1]),
push_data_block, ConstU32(half), ConstU32(comp))};
const Id value{OpLoad(U32[1], ptr)};
buffer.offset = OpBitFieldUExtract(U32[1], value, ConstU32(offset), ConstU32(8U));
Name(buffer.offset, fmt::format("buf{}_off", binding));
buffer.offset_dwords = OpShiftRightLogical(U32[1], buffer.offset, ConstU32(2U));
Name(buffer.offset_dwords, fmt::format("buf{}_dword_off", binding));

if (!profile.supports_robust_buffer_access) {
const u32 size_field = PushData::BufSizesIndex + (binding >> 2);
const u32 size_comp = binding & 0x3;
const Id size_ptr{OpAccessChain(TypePointer(spv::StorageClass::PushConstant, U32[1]),
push_data_block, ConstU32(size_field),
ConstU32(size_comp))};
buffer.size = OpLoad(U32[1], size_ptr);
// Only need to load size if performing bounds checks and the buffer is both guest and not
// inline.
if (!profile.supports_robust_buffer_access && buffer.buffer_type == BufferType::Guest) {
if (buffer.desc.sharp_idx == std::numeric_limits<u32>::max()) {
buffer.size = ConstU32(buffer.desc.inline_cbuf.GetSize());
} else {
buffer.size = GetBufferSize(buffer.desc.sharp_idx);
}
Name(buffer.size, fmt::format("buf{}_size", binding));
buffer.size_shorts = OpShiftRightLogical(U32[1], buffer.size, ConstU32(1U));
Name(buffer.size_shorts, fmt::format("buf{}_short_size", binding));
Expand Down Expand Up @@ -606,8 +623,7 @@ void EmitContext::DefineOutputs() {
void EmitContext::DefinePushDataBlock() {
// Create push constants block for instance steps rates
const Id struct_type{Name(TypeStruct(U32[1], U32[1], F32[1], F32[1], F32[1], F32[1], U32[4],
U32[4], U32[4], U32[4], U32[4], U32[4], U32[4], U32[4],
U32[4], U32[4], U32[4], U32[4], U32[4], U32[4]),
U32[4], U32[4], U32[4], U32[4], U32[4]),
"AuxData")};
Decorate(struct_type, spv::Decoration::Block);
MemberName(struct_type, PushData::Step0Index, "sr0");
Expand All @@ -622,14 +638,6 @@ void EmitContext::DefinePushDataBlock() {
MemberName(struct_type, PushData::UdRegsIndex + 3, "ud_regs3");
MemberName(struct_type, PushData::BufOffsetIndex + 0, "buf_offsets0");
MemberName(struct_type, PushData::BufOffsetIndex + 1, "buf_offsets1");
MemberName(struct_type, PushData::BufSizesIndex + 0, "buf_sizes0");
MemberName(struct_type, PushData::BufSizesIndex + 1, "buf_sizes1");
MemberName(struct_type, PushData::BufSizesIndex + 2, "buf_sizes2");
MemberName(struct_type, PushData::BufSizesIndex + 3, "buf_sizes3");
MemberName(struct_type, PushData::BufSizesIndex + 4, "buf_sizes4");
MemberName(struct_type, PushData::BufSizesIndex + 5, "buf_sizes5");
MemberName(struct_type, PushData::BufSizesIndex + 6, "buf_sizes6");
MemberName(struct_type, PushData::BufSizesIndex + 7, "buf_sizes7");
MemberDecorate(struct_type, PushData::Step0Index, spv::Decoration::Offset, 0U);
MemberDecorate(struct_type, PushData::Step1Index, spv::Decoration::Offset, 4U);
MemberDecorate(struct_type, PushData::XOffsetIndex, spv::Decoration::Offset, 8U);
Expand All @@ -642,14 +650,6 @@ void EmitContext::DefinePushDataBlock() {
MemberDecorate(struct_type, PushData::UdRegsIndex + 3, spv::Decoration::Offset, 72U);
MemberDecorate(struct_type, PushData::BufOffsetIndex + 0, spv::Decoration::Offset, 88U);
MemberDecorate(struct_type, PushData::BufOffsetIndex + 1, spv::Decoration::Offset, 104U);
MemberDecorate(struct_type, PushData::BufSizesIndex + 0, spv::Decoration::Offset, 120U);
MemberDecorate(struct_type, PushData::BufSizesIndex + 1, spv::Decoration::Offset, 136U);
MemberDecorate(struct_type, PushData::BufSizesIndex + 2, spv::Decoration::Offset, 152U);
MemberDecorate(struct_type, PushData::BufSizesIndex + 3, spv::Decoration::Offset, 168U);
MemberDecorate(struct_type, PushData::BufSizesIndex + 4, spv::Decoration::Offset, 184U);
MemberDecorate(struct_type, PushData::BufSizesIndex + 5, spv::Decoration::Offset, 200U);
MemberDecorate(struct_type, PushData::BufSizesIndex + 6, spv::Decoration::Offset, 216U);
MemberDecorate(struct_type, PushData::BufSizesIndex + 7, spv::Decoration::Offset, 232U);
push_data_block = DefineVar(struct_type, spv::StorageClass::PushConstant);
Name(push_data_block, "push_data");
interfaces.push_back(push_data_block);
Expand Down Expand Up @@ -694,18 +694,28 @@ EmitContext::BufferSpv EmitContext::DefineBuffer(bool is_storage, bool is_writte
break;
default:
Name(id, fmt::format("{}_{}", is_storage ? "ssbo" : "ubo", binding.buffer));
break;
}
interfaces.push_back(id);
return {id, pointer_type};
};

void EmitContext::DefineBuffers() {
if (!profile.supports_robust_buffer_access && !info.has_readconst) {
// In case ReadConstUbo has not already been bound by IR and is needed
// to query buffer sizes, bind it now.
info.buffers.push_back({
.used_types = IR::Type::U32,
.inline_cbuf = AmdGpu::Buffer::Null(),
.buffer_type = BufferType::ReadConstUbo,
});
}
for (const auto& desc : info.buffers) {
const auto buf_sharp = desc.GetSharp(info);
const bool is_storage = desc.IsStorage(buf_sharp, profile);

// Define aliases depending on the shader usage.
auto& spv_buffer = buffers.emplace_back(binding.buffer++, desc.buffer_type);
auto& spv_buffer = buffers.emplace_back(binding.buffer++, desc.buffer_type, desc);
if (True(desc.used_types & IR::Type::U32)) {
spv_buffer[BufferAlias::U32] =
DefineBuffer(is_storage, desc.is_written, 2, desc.buffer_type, U32[1]);
Expand Down
3 changes: 3 additions & 0 deletions src/shader_recompiler/backend/spirv/spirv_emit_context.h
View file
Original file line number Diff line number Diff line change
Expand Up @@ -246,6 +246,7 @@ class EmitContext final : public Sirit::Module {
struct BufferDefinition {
u32 binding;
BufferType buffer_type;
const BufferResource& desc;
Id offset;
Id offset_dwords;
Id size;
Expand Down Expand Up @@ -310,6 +311,8 @@ class EmitContext final : public Sirit::Module {

Id DefineFloat32ToUfloatM5(u32 mantissa_bits, std::string_view name);
Id DefineUfloatM5ToFloat32(u32 mantissa_bits, std::string_view name);

Id GetBufferSize(u32 sharp_idx);
};

} // namespace Shader::Backend::SPIRV
10 changes: 3 additions & 7 deletions src/shader_recompiler/info.h
View file
Original file line number Diff line number Diff line change
Expand Up @@ -106,7 +106,6 @@ struct PushData {
static constexpr u32 YScaleIndex = 5;
static constexpr u32 UdRegsIndex = 6;
static constexpr u32 BufOffsetIndex = UdRegsIndex + NumUserDataRegs / 4;
static constexpr u32 BufSizesIndex = BufOffsetIndex + NumBuffers / sizeof(u32) / 4;

u32 step0;
u32 step1;
Expand All @@ -116,17 +115,14 @@ struct PushData {
float yscale;
std::array<u32, NumUserDataRegs> ud_regs;
std::array<u8, NumBuffers> buf_offsets;
std::array<u32, NumBuffers> buf_sizes;

void AddBuffer(u32 binding, u32 offset, u32 size) {
void AddOffset(u32 binding, u32 offset) {
ASSERT(offset < 256 && binding < buf_offsets.size());
buf_offsets[binding] = offset;
buf_sizes[binding] = size;
}
};
static_assert(offsetof(PushData, buf_sizes) <= 128,
"PushData size without buf_sizes is greater than guaranteed by Vulkan spec");
static_assert(sizeof(PushData) <= 256, "PushData size is greater than guaranteed by most GPUs");
static_assert(sizeof(PushData) <= 128,
"PushData size is greater than minimum size guaranteed by Vulkan spec");

/**
* Contains general information generated by the shader recompiler for an input program.
Expand Down
6 changes: 1 addition & 5 deletions src/video_core/renderer_vulkan/vk_pipeline_common.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -38,11 +38,7 @@ void Pipeline::BindResources(DescriptorWrites& set_writes, const BufferBarriers&
}

const auto stage_flags = IsCompute() ? vk::ShaderStageFlagBits::eCompute : gp_stage_flags;
// If not emulating buffer bounds checks, buffer sizes are not needed.
const auto push_constants_size = instance.IsRobustBufferAccess2Supported()
? offsetof(Shader::PushData, buf_sizes)
: sizeof(Shader::PushData);
cmdbuf.pushConstants(*pipeline_layout, stage_flags, 0u, push_constants_size, &push_data);
cmdbuf.pushConstants(*pipeline_layout, stage_flags, 0u, sizeof(push_data), &push_data);

// Bind descriptor set.
if (set_writes.empty()) {
Expand Down
2 changes: 1 addition & 1 deletion src/video_core/renderer_vulkan/vk_rasterizer.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -548,7 +548,7 @@ void Rasterizer::BindBuffers(const Shader::Info& stage, Shader::Backend::Binding
const u32 offset_aligned = Common::AlignDown(offset, alignment);
const u32 adjust = offset - offset_aligned;
ASSERT(adjust % 4 == 0);
push_data.AddBuffer(binding.buffer, adjust, vsharp.GetSize());
push_data.AddOffset(binding.buffer, adjust);
buffer_infos.emplace_back(vk_buffer->Handle(), offset_aligned,
vsharp.GetSize() + adjust);
if (auto barrier =
Expand Down

0 comments on commit 507722b

Please sign in to comment.