[DirectX] Implement UseNativeLowPrecision shader flag analysis (#134288)

Fixes #112267 

Implement the shader flag analysis to set the UseNativeLowPrecision DXIL
module flag.

The flag is only able to be set when the command-line flag
`-enable-16bit-types` is passed to clang-dxc, or equivalently
`-fnative-half-type` is passed to clang.
When the command-line flag is passed, a module metadata flag called
"dx.nativelowprec" is set to 1.
The DXILShaderFlags shader flags analysis checks that the module
metadata flag "dx.nativelowprec" is set to 1 and the DXIL Version is 1.2
or greater before setting the UseNativeLowPrecision DXIL module flag.

Author: Icohedron

clang/lib/CodeGen/CGHLSLRuntime.cpp

@@ -282,11 +282,19 @@ void CGHLSLRuntime::addHLSLBufferLayoutType(const RecordType *StructType,
 
 void CGHLSLRuntime::finishCodeGen() {
   auto &TargetOpts = CGM.getTarget().getTargetOpts();
+  auto &LangOpts = CGM.getLangOpts();
   llvm::Module &M = CGM.getModule();
   Triple T(M.getTargetTriple());
   if (T.getArch() == Triple::ArchType::dxil)
     addDxilValVersion(TargetOpts.DxilValidatorVersion, M);
 
+  // NativeHalfType corresponds to the -fnative-half-type clang option which is
+  // aliased by clang-dxc's -enable-16bit-types option. This option is used to
+  // set the UseNativeLowPrecision DXIL module flag in the DirectX backend
+  if (LangOpts.NativeHalfType)
+    M.setModuleFlag(llvm::Module::ModFlagBehavior::Error, "dx.nativelowprec",
+                    1);
+
   generateGlobalCtorDtorCalls();
 }
 

clang/test/CodeGenHLSL/enable-16bit-types.hlsl

@@ -0,0 +1,7 @@
+// RUN: %clang_dxc -enable-16bit-types -T lib_6_3 -HV 202x -Vd -Xclang -emit-llvm %s | FileCheck %s --check-prefix=FLAG
+// RUN: %clang_dxc -T lib_6_3 -HV 202x -Vd -Xclang -emit-llvm %s | FileCheck %s --check-prefix=NOFLAG
+
+// FLAG-DAG: ![[NLP:.*]] = !{i32 1, !"dx.nativelowprec", i32 1}
+// FLAG-DAG: !llvm.module.flags = !{{{.*}}![[NLP]]{{.*}}}
+
+// NOFLAG-NOT: dx.nativelowprec

llvm/lib/Target/DirectX/DXILShaderFlags.cpp

@@ -188,6 +188,13 @@ void ModuleShaderFlags::initialize(Module &M, DXILResourceTypeMap &DRTM,
         continue;
       }
 
+      // Set UseNativeLowPrecision using dx.nativelowprec module metadata
+      if (auto *NativeLowPrec = mdconst::extract_or_null<ConstantInt>(
+              M.getModuleFlag("dx.nativelowprec")))
+        if (MMDI.DXILVersion >= VersionTuple(1, 2) &&
+            NativeLowPrec->getValue() != 0)
+          SCCSF.UseNativeLowPrecision = true;
+
       ComputedShaderFlags CSF;
       for (const auto &BB : *F)
         for (const auto &I : BB)

llvm/test/CodeGen/DirectX/ShaderFlags/low-precision.ll

@@ -1,5 +1,4 @@
 ; RUN: opt -S --passes="print-dx-shader-flags" 2>&1 %s | FileCheck %s
-; RUN: llc %s --filetype=obj -o - | obj2yaml | FileCheck %s --check-prefix=DXC
 
 target triple = "dxil-pc-shadermodel6.7-library"
 
@@ -13,26 +12,19 @@ target triple = "dxil-pc-shadermodel6.7-library"
 ;CHECK-NEXT: ; Shader Flags for Module Functions
 
 ;CHECK-LABEL: ; Function add_i16 : 0x00000020
-define i16 @add_i16(i16 %a, i16 %b) #0 {
+define i16 @add_i16(i16 %a, i16 %b) {
   %sum = add i16 %a, %b
   ret i16 %sum
 }
 
 ;CHECK-LABEL: ; Function add_i32 : 0x00000000
-define i32 @add_i32(i32 %a, i32 %b) #0 {
+define i32 @add_i32(i32 %a, i32 %b) {
   %sum = add i32 %a, %b
   ret i32 %sum
 }
 
 ;CHECK-LABEL: ; Function add_half : 0x00000020
-define half @add_half(half %a, half %b) #0 {
+define half @add_half(half %a, half %b) {
   %sum = fadd half %a, %b
   ret half %sum
 }
-
-attributes #0 = { convergent norecurse nounwind "hlsl.export"}
-
-; DXC: - Name:            SFI0
-; DXC-NEXT:     Size:            8
-; DXC-NOT:     Flags:
-; DXC: ...

llvm/test/CodeGen/DirectX/ShaderFlags/use-native-low-precision.ll

@@ -0,0 +1,37 @@
+; RUN: opt -S --passes="print-dx-shader-flags" 2>&1 %s | FileCheck %s
+
+target triple = "dxil-pc-shadermodel6.7-library"
+
+;CHECK: ; Combined Shader Flags for Module
+;CHECK-NEXT: ; Shader Flags Value: 0x00800020
+;CHECK-NEXT: ;
+;CHECK-NEXT: ; Note: shader requires additional functionality:
+;CHECK-NEXT: ; Note: extra DXIL module flags:
+;CHECK-NEXT: ;       D3D11_1_SB_GLOBAL_FLAG_ENABLE_MINIMUM_PRECISION
+;CHECK-NEXT: ;       Native 16bit types enabled
+;CHECK-NEXT: ;
+;CHECK-NEXT: ; Shader Flags for Module Functions
+
+;CHECK-LABEL: ; Function add_i16 : 0x00800020
+define i16 @add_i16(i16 %a, i16 %b) {
+  %sum = add i16 %a, %b
+  ret i16 %sum
+}
+
+; NOTE: The flag for native low precision (0x80000) is set for every function
+; in the module regardless of whether or not the function uses low precision
+; data types (flag 0x20). This matches the behavior in DXC
+;CHECK-LABEL: ; Function add_i32 : 0x00800000
+define i32 @add_i32(i32 %a, i32 %b) {
+  %sum = add i32 %a, %b
+  ret i32 %sum
+}
+
+;CHECK-LABEL: ; Function add_half : 0x00800020
+define half @add_half(half %a, half %b) {
+  %sum = fadd half %a, %b
+  ret half %sum
+}
+
+!llvm.module.flags = !{!0}
+!0 = !{i32 1, !"dx.nativelowprec", i32 1}