rjx-mirror/Ryujinx.Graphics.Shader/Translation/Translator.cs
gdkchan cb171f6ebf Support shared color mask, implement more shader instructions
Support shared color masks (used by Nouveau and maybe the NVIDIA
driver).
Support draw buffers (also required by OpenGL).
Support viewport transform disable (disabled for now as it breaks some
games).
Fix instanced rendering draw being ignored for multi draw.
Fix IADD and IADD3 immediate shader encodings, that was not matching
some ops.
Implement FFMA32I shader instruction.
Implement IMAD shader instruction.
2020-01-09 02:13:00 +01:00

358 lines
No EOL
11 KiB
C#

using Ryujinx.Graphics.Shader.CodeGen.Glsl;
using Ryujinx.Graphics.Shader.Decoders;
using Ryujinx.Graphics.Shader.IntermediateRepresentation;
using Ryujinx.Graphics.Shader.StructuredIr;
using Ryujinx.Graphics.Shader.Translation.Optimizations;
using System;
using System.Collections.Generic;
using static Ryujinx.Graphics.Shader.IntermediateRepresentation.OperandHelper;
namespace Ryujinx.Graphics.Shader.Translation
{
public static class Translator
{
private const int HeaderSize = 0x50;
public static Span<byte> ExtractCode(Span<byte> code, bool compute, out int headerSize)
{
if (compute)
{
headerSize = 0;
}
else
{
headerSize = HeaderSize;
}
Block[] cfg = Decoder.Decode(code, (ulong)headerSize);
if (cfg == null)
{
// TODO: Error.
return code;
}
ulong endAddress = 0;
foreach (Block block in cfg)
{
if (endAddress < block.EndAddress)
{
endAddress = block.EndAddress;
}
}
return code.Slice(0, headerSize + (int)endAddress);
}
public static ShaderProgram Translate(Span<byte> code, ShaderCapabilities capabilities, TranslationFlags flags)
{
bool compute = (flags & TranslationFlags.Compute) != 0;
Operation[] ops = DecodeShader(code, capabilities, flags, out ShaderHeader header, out int size);
ShaderStage stage;
if (compute)
{
stage = ShaderStage.Compute;
}
else
{
stage = header.Stage;
}
int maxOutputVertexCount = 0;
OutputTopology outputTopology = OutputTopology.LineStrip;
if (!compute)
{
maxOutputVertexCount = header.MaxOutputVertexCount;
outputTopology = header.OutputTopology;
}
ShaderConfig config = new ShaderConfig(
stage,
capabilities,
flags,
maxOutputVertexCount,
outputTopology);
return Translate(ops, config, size);
}
public static ShaderProgram Translate(Span<byte> vpACode, Span<byte> vpBCode, ShaderCapabilities capabilities, TranslationFlags flags)
{
bool debugMode = (flags & TranslationFlags.DebugMode) != 0;
Operation[] vpAOps = DecodeShader(vpACode, capabilities, flags, out _, out _);
Operation[] vpBOps = DecodeShader(vpBCode, capabilities, flags, out ShaderHeader header, out int sizeB);
ShaderConfig config = new ShaderConfig(
header.Stage,
capabilities,
flags,
header.MaxOutputVertexCount,
header.OutputTopology);
return Translate(Combine(vpAOps, vpBOps), config, sizeB);
}
private static ShaderProgram Translate(Operation[] ops, ShaderConfig config, int size)
{
BasicBlock[] blocks = ControlFlowGraph.MakeCfg(ops);
if (blocks.Length > 0)
{
Dominance.FindDominators(blocks[0], blocks.Length);
Dominance.FindDominanceFrontiers(blocks);
Ssa.Rename(blocks);
Optimizer.RunPass(blocks, config);
Lowering.RunPass(blocks, config);
}
StructuredProgramInfo sInfo = StructuredProgram.MakeStructuredProgram(blocks, config);
GlslProgram program = GlslGenerator.Generate(sInfo, config);
ShaderProgramInfo spInfo = new ShaderProgramInfo(
program.CBufferDescriptors,
program.SBufferDescriptors,
program.TextureDescriptors,
program.ImageDescriptors,
sInfo.InterpolationQualifiers,
sInfo.UsesInstanceId);
string glslCode = program.Code;
return new ShaderProgram(spInfo, config.Stage, glslCode, size);
}
private static Operation[] DecodeShader(
Span<byte> code,
ShaderCapabilities capabilities,
TranslationFlags flags,
out ShaderHeader header,
out int size)
{
Block[] cfg;
EmitterContext context;
if ((flags & TranslationFlags.Compute) != 0)
{
header = null;
cfg = Decoder.Decode(code, 0);
context = new EmitterContext(ShaderStage.Compute, header, capabilities, flags);
}
else
{
header = new ShaderHeader(code);
cfg = Decoder.Decode(code, HeaderSize);
context = new EmitterContext(header.Stage, header, capabilities, flags);
}
if (cfg == null)
{
// TODO: Error.
size = 0;
return new Operation[0];
}
ulong maxEndAddress = 0;
for (int blkIndex = 0; blkIndex < cfg.Length; blkIndex++)
{
Block block = cfg[blkIndex];
if (maxEndAddress < block.EndAddress)
{
maxEndAddress = block.EndAddress;
}
context.CurrBlock = block;
context.MarkLabel(context.GetLabel(block.Address));
for (int opIndex = 0; opIndex < block.OpCodes.Count; opIndex++)
{
OpCode op = block.OpCodes[opIndex];
if ((flags & TranslationFlags.DebugMode) != 0)
{
string instName;
if (op.Emitter != null)
{
instName = op.Emitter.Method.Name;
}
else
{
instName = "???";
}
string dbgComment = $"0x{op.Address:X6}: 0x{op.RawOpCode:X16} {instName}";
context.Add(new CommentNode(dbgComment));
}
if (op.NeverExecute)
{
continue;
}
Operand predSkipLbl = null;
bool skipPredicateCheck = op is OpCodeBranch opBranch && !opBranch.PushTarget;
if (op is OpCodeBranchPop opBranchPop)
{
// If the instruction is a SYNC or BRK instruction with only one
// possible target address, then the instruction is basically
// just a simple branch, we can generate code similar to branch
// instructions, with the condition check on the branch itself.
skipPredicateCheck = opBranchPop.Targets.Count < 2;
}
if (!(op.Predicate.IsPT || skipPredicateCheck))
{
Operand label;
if (opIndex == block.OpCodes.Count - 1 && block.Next != null)
{
label = context.GetLabel(block.Next.Address);
}
else
{
label = Label();
predSkipLbl = label;
}
Operand pred = Register(op.Predicate);
if (op.InvertPredicate)
{
context.BranchIfTrue(label, pred);
}
else
{
context.BranchIfFalse(label, pred);
}
}
context.CurrOp = op;
if (op.Emitter != null)
{
op.Emitter(context);
}
if (predSkipLbl != null)
{
context.MarkLabel(predSkipLbl);
}
}
}
size = (int)maxEndAddress + (((flags & TranslationFlags.Compute) != 0) ? 0 : HeaderSize);
return context.GetOperations();
}
private static Operation[] Combine(Operation[] a, Operation[] b)
{
// Here we combine two shaders.
// For shader A:
// - All user attribute stores on shader A are turned into copies to a
// temporary variable. It's assumed that shader B will consume them.
// - All return instructions are turned into branch instructions, the
// branch target being the start of the shader B code.
// For shader B:
// - All user attribute loads on shader B are turned into copies from a
// temporary variable, as long that attribute is written by shader A.
List<Operation> output = new List<Operation>(a.Length + b.Length);
Operand[] temps = new Operand[AttributeConsts.UserAttributesCount * 4];
Operand lblB = Label();
for (int index = 0; index < a.Length; index++)
{
Operation operation = a[index];
if (IsUserAttribute(operation.Dest))
{
int tIndex = (operation.Dest.Value - AttributeConsts.UserAttributeBase) / 4;
Operand temp = temps[tIndex];
if (temp == null)
{
temp = Local();
temps[tIndex] = temp;
}
operation.Dest = temp;
}
if (operation.Inst == Instruction.Return)
{
output.Add(new Operation(Instruction.Branch, lblB));
}
else
{
output.Add(operation);
}
}
output.Add(new Operation(Instruction.MarkLabel, lblB));
for (int index = 0; index < b.Length; index++)
{
Operation operation = b[index];
for (int srcIndex = 0; srcIndex < operation.SourcesCount; srcIndex++)
{
Operand src = operation.GetSource(srcIndex);
if (IsUserAttribute(src))
{
Operand temp = temps[(src.Value - AttributeConsts.UserAttributeBase) / 4];
if (temp != null)
{
operation.SetSource(srcIndex, temp);
}
}
}
output.Add(operation);
}
return output.ToArray();
}
private static bool IsUserAttribute(Operand operand)
{
return operand != null &&
operand.Type == OperandType.Attribute &&
operand.Value >= AttributeConsts.UserAttributeBase &&
operand.Value < AttributeConsts.UserAttributeEnd;
}
}
}