yuzu-emu
/
yuzu
Archived
1
0
Fork 0
Code

gl_shader_decompiler: Abstract VMAD into a video subset

This commit is contained in:
ReinUsesLisp 2018-10-18 02:38:58 -03:00
parent 496d155d7b
commit 5dfb43531c
2 changed files with 82 additions and 75 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
src/video_core/engines/shader_bytecode.h
View File

@ -214,7 +214,7 @@ enum class IMinMaxExchange : u64 {
XHi = 3,
};
enum class VmadType : u64 {
enum class VideoType : u64 {
Size16_Low = 0,
Size16_High = 1,
Size32 = 2,
@ -782,6 +782,14 @@ union Instruction {
BitField<45, 2, PredOperation> op;
} psetp;
union {
BitField<43, 4, PredCondition> cond;
BitField<45, 2, PredOperation> op;
BitField<3, 3, u64> pred3;
BitField<0, 3, u64> pred0;
BitField<39, 3, u64> pred39;
} vsetp;
union {
BitField<12, 3, u64> pred12;
BitField<15, 1, u64> neg_pred12;
@ -1154,15 +1162,17 @@ union Instruction {
union {
BitField<48, 1, u64> signed_a;
BitField<38, 1, u64> is_byte_chunk_a;
BitField<36, 2, VmadType> type_a;
BitField<36, 2, VideoType> type_a;
BitField<36, 2, u64> byte_height_a;
BitField<49, 1, u64> signed_b;
BitField<50, 1, u64> use_register_b;
BitField<30, 1, u64> is_byte_chunk_b;
BitField<28, 2, VmadType> type_b;
BitField<28, 2, VideoType> type_b;
BitField<28, 2, u64> byte_height_b;
} video;
union {
BitField<51, 2, VmadShr> shr;
BitField<55, 1, u64> saturate; // Saturates the result (a * b + c)
BitField<47, 1, u64> cc;

141
src/video_core/renderer_opengl/gl_shader_decompiler.cpp
View File

@ -1291,6 +1291,63 @@ private:
}
}
/// Unpacks a video instruction operand (e.g. VMAD).
std::string GetVideoOperand(const std::string& op, bool is_chunk, bool is_signed,
Tegra::Shader::VideoType type, u64 byte_height) {
const std::string value = [&]() {
if (!is_chunk) {
const auto offset = static_cast<u32>(byte_height * 8);
return "((" + op + " >> " + std::to_string(offset) + ") & 0xff)";
}
const std::string zero = "0";
switch (type) {
case Tegra::Shader::VideoType::Size16_Low:
return '(' + op + " & 0xffff)";
case Tegra::Shader::VideoType::Size16_High:
return '(' + op + " >> 16)";
case Tegra::Shader::VideoType::Size32:
// TODO(Rodrigo): From my hardware tests it becomes a bit "mad" when
// this type is used (1 * 1 + 0 == 0x5b800000). Until a better
// explanation is found: assert.
UNIMPLEMENTED();
return zero;
case Tegra::Shader::VideoType::Invalid:
UNREACHABLE_MSG("Invalid instruction encoding");
return zero;
default:
UNREACHABLE();
return zero;
}
}();
if (is_signed) {
return "int(" + value + ')';
}
return value;
};
/// Gets the A operand for a video instruction.
std::string GetVideoOperandA(Instruction instr) {
return GetVideoOperand(regs.GetRegisterAsInteger(instr.gpr8, 0, false),
instr.video.is_byte_chunk_a != 0, instr.video.signed_a,
instr.video.type_a, instr.video.byte_height_a);
}
/// Gets the B operand for a video instruction.
std::string GetVideoOperandB(Instruction instr) {
if (instr.video.use_register_b) {
return GetVideoOperand(regs.GetRegisterAsInteger(instr.gpr20, 0, false),
instr.video.is_byte_chunk_b != 0, instr.video.signed_b,
instr.video.type_b, instr.video.byte_height_b);
} else {
return '(' +
std::to_string(instr.video.signed_b ? static_cast<s16>(instr.alu.GetImm20_16())
: instr.alu.GetImm20_16()) +
')';
}
}
/**
* Compiles a single instruction from Tegra to GLSL.
* @param offset the offset of the Tegra shader instruction.
@ -3284,82 +3341,22 @@ private:
break;
}
case OpCode::Id::VMAD: {
const bool signed_a = instr.vmad.signed_a == 1;
const bool signed_b = instr.vmad.signed_b == 1;
const bool result_signed = signed_a || signed_b;
boost::optional<std::string> forced_result;
auto Unpack = [&](const std::string& op, bool is_chunk, bool is_signed,
Tegra::Shader::VmadType type, u64 byte_height) {
const std::string value = [&]() {
if (!is_chunk) {
const auto shift = static_cast<u32>(byte_height * 8);
return "((" + op + " >> " + std::to_string(shift) + ") & 0xff)";
}
const std::string zero = "0";
switch (type) {
case Tegra::Shader::VmadType::Size16_Low:
return '(' + op + " & 0xffff)";
case Tegra::Shader::VmadType::Size16_High:
return '(' + op + " >> 16)";
case Tegra::Shader::VmadType::Size32:
// TODO(Rodrigo): From my hardware tests it becomes a bit "mad" when
// this type is used (1 * 1 + 0 == 0x5b800000). Until a better
// explanation is found: assert.
UNREACHABLE_MSG("Unimplemented");
return zero;
case Tegra::Shader::VmadType::Invalid:
// Note(Rodrigo): This flag is invalid according to nvdisasm. From my
// testing (even though it's invalid) this makes the whole instruction
// assign zero to target register.
forced_result = boost::make_optional(zero);
return zero;
default:
UNREACHABLE();
return zero;
}
}();
if (is_signed) {
return "int(" + value + ')';
}
return value;
};
const std::string op_a = Unpack(regs.GetRegisterAsInteger(instr.gpr8, 0, false),
instr.vmad.is_byte_chunk_a != 0, signed_a,
instr.vmad.type_a, instr.vmad.byte_height_a);
std::string op_b;
if (instr.vmad.use_register_b) {
op_b = Unpack(regs.GetRegisterAsInteger(instr.gpr20, 0, false),
instr.vmad.is_byte_chunk_b != 0, signed_b, instr.vmad.type_b,
instr.vmad.byte_height_b);
} else {
op_b = '(' +
std::to_string(signed_b ? static_cast<s16>(instr.alu.GetImm20_16())
: instr.alu.GetImm20_16()) +
')';
}
const bool result_signed = instr.video.signed_a == 1 || instr.video.signed_b == 1;
const std::string op_a = GetVideoOperandA(instr);
const std::string op_b = GetVideoOperandB(instr);
const std::string op_c = regs.GetRegisterAsInteger(instr.gpr39, 0, result_signed);
std::string result;
if (forced_result) {
result = *forced_result;
} else {
result = '(' + op_a + " * " + op_b + " + " + op_c + ')';
std::string result = '(' + op_a + " * " + op_b + " + " + op_c + ')';
switch (instr.vmad.shr) {
case Tegra::Shader::VmadShr::Shr7:
result = '(' + result + " >> 7)";
break;
case Tegra::Shader::VmadShr::Shr15:
result = '(' + result + " >> 15)";
break;
}
switch (instr.vmad.shr) {
case Tegra::Shader::VmadShr::Shr7:
result = '(' + result + " >> 7)";
break;
case Tegra::Shader::VmadShr::Shr15:
result = '(' + result + " >> 15)";
break;
}
regs.SetRegisterToInteger(instr.gpr0, result_signed, 1, result, 1, 1,
instr.vmad.saturate == 1, 0, Register::Size::Word,
instr.vmad.cc);