shader/memory: Move unaligned load/store to functions

This commit is contained in:
ReinUsesLisp 2020-01-25 02:18:14 -03:00
parent 96638f57c9
commit 531f25a037

View File

@ -62,6 +62,20 @@ u32 GetMemorySize(Tegra::Shader::UniformType uniform_type) {
} }
} }
Node ExtractUnaligned(Node value, Node address, u32 mask, u32 size) {
Node offset = Operation(OperationCode::UBitwiseAnd, address, Immediate(mask));
offset = Operation(OperationCode::ULogicalShiftLeft, std::move(offset), Immediate(3));
return Operation(OperationCode::UBitfieldExtract, std::move(value), std::move(offset),
Immediate(size));
}
Node InsertUnaligned(Node dest, Node value, Node address, u32 mask, u32 size) {
Node offset = Operation(OperationCode::UBitwiseAnd, std::move(address), Immediate(mask));
offset = Operation(OperationCode::ULogicalShiftLeft, std::move(offset), Immediate(3));
return Operation(OperationCode::UBitfieldInsert, std::move(dest), std::move(value),
std::move(offset), Immediate(size));
}
Node Sign16Extend(Node value) { Node Sign16Extend(Node value) {
Node sign = Operation(OperationCode::UBitwiseAnd, value, Immediate(1U << 15)); Node sign = Operation(OperationCode::UBitwiseAnd, value, Immediate(1U << 15));
Node is_sign = Operation(OperationCode::LogicalUEqual, std::move(sign), Immediate(1U << 15)); Node is_sign = Operation(OperationCode::LogicalUEqual, std::move(sign), Immediate(1U << 15));
@ -144,19 +158,23 @@ u32 ShaderIR::DecodeMemory(NodeBlock& bb, u32 pc) {
LOG_DEBUG(HW_GPU, "LD_L cache management mode: {}", static_cast<u64>(instr.ld_l.unknown)); LOG_DEBUG(HW_GPU, "LD_L cache management mode: {}", static_cast<u64>(instr.ld_l.unknown));
[[fallthrough]]; [[fallthrough]];
case OpCode::Id::LD_S: { case OpCode::Id::LD_S: {
const auto GetMemory = [&](s32 offset) { const auto GetAddress = [&](s32 offset) {
ASSERT(offset % 4 == 0); ASSERT(offset % 4 == 0);
const Node immediate_offset = Immediate(static_cast<s32>(instr.smem_imm) + offset); const Node immediate_offset = Immediate(static_cast<s32>(instr.smem_imm) + offset);
const Node address = return Operation(OperationCode::IAdd, GetRegister(instr.gpr8), immediate_offset);
Operation(OperationCode::IAdd, GetRegister(instr.gpr8), immediate_offset); };
return opcode->get().GetId() == OpCode::Id::LD_S ? GetSharedMemory(address) const auto GetMemory = [&](s32 offset) {
: GetLocalMemory(address); return opcode->get().GetId() == OpCode::Id::LD_S ? GetSharedMemory(GetAddress(offset))
: GetLocalMemory(GetAddress(offset));
}; };
switch (instr.ldst_sl.type.Value()) { switch (instr.ldst_sl.type.Value()) {
case StoreType::Signed16: case StoreType::Signed16: {
Node address = GetAddress(0);
SetRegister(bb, instr.gpr0, Sign16Extend(GetMemory(0))); SetRegister(bb, instr.gpr0, Sign16Extend(GetMemory(0)));
break; break;
}
case StoreType::Bits32: case StoreType::Bits32:
case StoreType::Bits64: case StoreType::Bits64:
case StoreType::Bits128: { case StoreType::Bits128: {
@ -223,12 +241,7 @@ u32 ShaderIR::DecodeMemory(NodeBlock& bb, u32 pc) {
// To handle unaligned loads get the bytes used to dereference global memory and extract // To handle unaligned loads get the bytes used to dereference global memory and extract
// those bytes from the loaded u32. // those bytes from the loaded u32.
if (IsUnaligned(type)) { if (IsUnaligned(type)) {
Node mask = Immediate(GetUnalignedMask(type)); gmem = ExtractUnaligned(gmem, real_address, GetUnalignedMask(type), size);
Node offset = Operation(OperationCode::UBitwiseAnd, real_address, std::move(mask));
offset = Operation(OperationCode::ULogicalShiftLeft, offset, Immediate(3));
gmem = Operation(OperationCode::UBitfieldExtract, std::move(gmem),
std::move(offset), Immediate(size));
} }
SetTemporary(bb, i, gmem); SetTemporary(bb, i, gmem);
@ -334,12 +347,8 @@ u32 ShaderIR::DecodeMemory(NodeBlock& bb, u32 pc) {
Node value = GetRegister(instr.gpr0.Value() + i); Node value = GetRegister(instr.gpr0.Value() + i);
if (IsUnaligned(type)) { if (IsUnaligned(type)) {
Node mask = Immediate(GetUnalignedMask(type)); const u32 mask = GetUnalignedMask(type);
Node offset = Operation(OperationCode::UBitwiseAnd, real_address, std::move(mask)); value = InsertUnaligned(gmem, std::move(value), real_address, mask, size);
offset = Operation(OperationCode::ULogicalShiftLeft, offset, Immediate(3));
value = Operation(OperationCode::UBitfieldInsert, gmem, std::move(value), offset,
Immediate(size));
} }
bb.push_back(Operation(OperationCode::Assign, gmem, value)); bb.push_back(Operation(OperationCode::Assign, gmem, value));