[DirectX backend] set target triple to "dxil-ms-dx"

[python3kgae]

Summary

Set target triple to "dxil-ms-dx" for DXIL at the end of DXILTranslateMetadata.

[beanz]

So... this is actually going to be terrible, but we can't do it this way. We rely on the module having a sane triple through MC emission.

What we need to do is change the triple right before DXIL bitcode generation, then change it back so that it is correct through MC.

[python3kgae]

Only set "dxil-ms-dx" TargetTriple for the output bitcode.

[beanz]

llvm/lib/Target/DirectX/DXILTranslateMetadata.cpp

54 ↗

(On Diff #451921)

This is a bit confusingly named. Probably worth calling this SMVer instead and explaining in a comment that the DXIL and SM minor versions are the same, and the major versions are offset.

75 ↗

(On Diff #451921)

nit: Rather than static const constexpr. The later is guaranteed to be compile-time resolved, the former isn't.

85 ↗

(On Diff #451921)

nit: https://llvm.org/docs/CodingStandards.html#use-auto-type-deduction-to-make-code-more-readable

86 ↗

(On Diff #451921)

This one really requires you to know the API to know that getMinor() returns an Optional rather than a value.

Other places in LLVM use a pattern more like:

unsigned Minor = V.getMinor().value_or(0);

Which would also be more clear.

llvm/lib/Target/DirectX/DXILWriter/DXILBitcodeWriter.h

78 ↗

(On Diff #451921)

I know we had talked about where to mutate the triple and I thought WriteDXILToFile made sense, but seeing the code now that we lose the const-ness here I'm rethinking.

We only really need the triple to be altered when we're embedding DXIL in the module for emission of a container. I don't even think the WriteDXILPass is ever used (unless it were called manually).

It might make more sense to mutate the triple in EmbedDXILPass::runOnModule to preserve the const-ness of the writing function. We can also in a subsequent patch delete the WriteDXILPass.

[python3kgae]

Move triple update into EmbedDXILPass.

[python3kgae]

Rebase and fix test fail.

[python3kgae]

Rebase

[python3kgae]

Rebase and only set target triple to "dxil-ms-dx"

[beanz]

Please add a check line to the embed-dxil.ll test to verify that the triple is restored after embedding the DXIL global, otherwise LGTM.

[python3kgae]

Make sure triple is restored.

llvm/lib/Target/DirectX/DXILWriter/DXILWriterPass.cpp

EmbedDXILPass() : ModulePass(ID) {
     initializeEmbedDXILPassPass(*PassRegistry::getPassRegistry());
   }
 
   StringRef getPassName() const override { return "DXIL Embedder"; }
 
   bool runOnModule(Module &M) override {
     std::string Data;
     llvm::raw_string_ostream OS(Data);
+
+    const std::string OriginalTriple = M.getTargetTriple();
+    // Set to DXIL triple when write to bitcode.
+    // Only the output bitcode need to be DXIL triple.
+    M.setTargetTriple("dxil-ms-dx");
+
     WriteDXILToFile(M, OS);
 
+    // Recover triple.
+    M.setTargetTriple(OriginalTriple);
+
     Constant *ModuleConstant =
         ConstantDataArray::get(M.getContext(), arrayRefFromStringRef(Data));
     auto *GV = new llvm::GlobalVariable(M, ModuleConstant->getType(), true,
                                         GlobalValue::PrivateLinkage,
                                         ModuleConstant, "dx.dxil");
     GV->setSection("DXIL");
     GV->setAlignment(Align(4));
     appendToCompilerUsed(M, {GV});

llvm/test/CodeGen/DirectX/embed-dxil.ll

 ; RUN: llc %s --filetype=asm -o - | FileCheck %s
 ; RUN: opt %s -dxil-embed -S -o - | FileCheck %s
 ; RUN: llc %s --filetype=obj -o - | obj2yaml | FileCheck %s --check-prefix=DXC
 target triple = "dxil-unknown-shadermodel6.5-library"
 
+; Make sure triple is restored after updated to dxil.
+; CHECK:target triple = "dxil-unknown-shadermodel6.5-library"
+
 define i32 @add(i32 %a, i32 %b) {
   %sum = add i32 %a, %b
   ret i32 %sum
 }
 
 ; CHECK: @dx.dxil = private constant [[BC_TYPE:\[[0-9]+ x i8\]]] c"BC\C0\DE{{[^"]+}}", section "DXIL", align 4
 
 ; The dxil global should be the first here because we generate it before the

llvm/test/tools/dxil-dis/attribute-filter.ll

 ; RUN: llc %s --filetype=obj -o - | dxil-dis -o - | FileCheck %s
 
-; CHECK: target triple = "dxil-unknown-shadermodel6.7-library"
+; CHECK: target triple = "dxil-ms-dx"
 target triple = "dxil-unknown-shadermodel6.7-library"
 
 ; CHECK: Function Attrs: nounwind readnone
 ; Function Attrs: norecurse nounwind readnone willreturn
 define float @fma(float %0, float %1, float %2) #0 {
   %4 = fmul float %0, %1
   %5 = fadd float %4, %2
   ret float %5
 }
 
 ; CHECK: attributes #0 = { nounwind readnone "disable-tail-calls"="false" }
 attributes #0 = { norecurse nounwind readnone willreturn "disable-tail-calls"="false" }

llvm/test/tools/dxil-dis/triple.ll

+; RUN: llc --filetype=obj %s -o - | dxil-dis  -o - | FileCheck %s
+target triple = "dxil-unknown-shadermodel6.7-library"
+
+; Make sure triple updated to dxil.
+; CHECK:target triple = "dxil-ms-dx"