[HLSL] Fix resrouce wrapper declaration

s-perron · s-perron · commit a65ccd00e320 · 2025-02-18T10:31:34.000-05:00
The resource wrapper should have internal linkage because it cotains a handle to the global resource, and it not the actual global. Makeing this changed exposed that we were zeroinitializing the resouce, which is a problem. The handle cannot be zeroinitialized. Fixes #122767.
diff --git a/clang/lib/CodeGen/CGHLSLRuntime.cpp b/clang/lib/CodeGen/CGHLSLRuntime.cpp
@@ -528,11 +528,6 @@ void CGHLSLRuntime::generateGlobalCtorDtorCalls() {
   }
 }
 
-// Returns true if the type is an HLSL resource class
-static bool isResourceRecordType(const clang::Type *Ty) {
-  return HLSLAttributedResourceType::findHandleTypeOnResource(Ty) != nullptr;
-}
-
 static void createResourceInitFn(CodeGenModule &CGM, const VarDecl *VD,
                                  llvm::GlobalVariable *GV, unsigned Slot,
                                  unsigned Space) {
@@ -599,7 +594,7 @@ void CGHLSLRuntime::handleGlobalVarDefinition(const VarDecl *VD,
     // on?
     return;
 
-  if (!isResourceRecordType(VD->getType().getTypePtr()))
+  if (!isResource(VD))
     // FIXME: Only simple declarations of resources are supported for now.
     // Arrays of resources or resources in user defined classes are
     // not implemented yet.
@@ -623,3 +618,8 @@ llvm::Instruction *CGHLSLRuntime::getConvergenceToken(BasicBlock &BB) {
   llvm_unreachable("Convergence token should have been emitted.");
   return nullptr;
 }
+
+bool CGHLSLRuntime::isResource(const VarDecl *D) {
+  const clang::Type *ty = D->getType().getTypePtr();
+  return HLSLAttributedResourceType::findHandleTypeOnResource(ty) != nullptr;
+}
diff --git a/clang/lib/CodeGen/CGHLSLRuntime.h b/clang/lib/CodeGen/CGHLSLRuntime.h
@@ -135,6 +135,8 @@ class CGHLSLRuntime {
     llvm::StructType *LayoutStruct = nullptr;
   };
 
+  bool isResource(const VarDecl *D);
+
 protected:
   CodeGenModule &CGM;
 
diff --git a/clang/lib/CodeGen/CodeGenModule.cpp b/clang/lib/CodeGen/CodeGenModule.cpp
@@ -5589,7 +5589,10 @@ void CodeGenModule::EmitGlobalVarDefinition(const VarDecl *D,
       if (D->getType()->isReferenceType())
         T = D->getType();
 
-      if (getLangOpts().CPlusPlus) {
+      if (getLangOpts().HLSL && getHLSLRuntime().isResource(D)) {
+        Init = llvm::UndefValue::get(getTypes().ConvertType(ASTTy));
+        NeedsGlobalCtor = true;
+      } else if (getLangOpts().CPlusPlus) {
         Init = EmitNullConstant(T);
         if (!IsDefinitionAvailableExternally)
           NeedsGlobalCtor = true;
@@ -5735,6 +5738,11 @@ void CodeGenModule::EmitGlobalVarDefinition(const VarDecl *D,
       !D->hasAttr<ConstInitAttr>())
     Linkage = llvm::GlobalValue::InternalLinkage;
 
+  // For HLSL resources, the GV is an internal wrapper containing a handle to
+  // external resource.
+  if (getLangOpts().HLSL && getHLSLRuntime().isResource(D))
+    Linkage = llvm::GlobalValue::InternalLinkage;
+
   GV->setLinkage(Linkage);
   if (D->hasAttr<DLLImportAttr>())
     GV->setDLLStorageClass(llvm::GlobalVariable::DLLImportStorageClass);
diff --git a/clang/test/CodeGenHLSL/builtins/ByteAddressBuffers-constructors.hlsl b/clang/test/CodeGenHLSL/builtins/ByteAddressBuffers-constructors.hlsl
@@ -11,9 +11,9 @@ RasterizerOrderedByteAddressBuffer Buffer2: register(u3, space4);
 // CHECK: "class.hlsl::RWByteAddressBuffer" = type { target("dx.RawBuffer", i8, 1, 0) }
 // CHECK: "class.hlsl::RasterizerOrderedByteAddressBuffer" = type { target("dx.RawBuffer", i8, 1, 1) }
 
-// CHECK: @Buffer0 = global %"class.hlsl::ByteAddressBuffer" zeroinitializer, align 4
-// CHECK: @Buffer1 = global %"class.hlsl::RWByteAddressBuffer" zeroinitializer, align 4
-// CHECK: @Buffer2 = global %"class.hlsl::RasterizerOrderedByteAddressBuffer" zeroinitializer, align 4
+// CHECK: @Buffer0 = internal global %"class.hlsl::ByteAddressBuffer" undef, align 4
+// CHECK: @Buffer1 = internal global %"class.hlsl::RWByteAddressBuffer" undef, align 4
+// CHECK: @Buffer2 = internal global %"class.hlsl::RasterizerOrderedByteAddressBuffer" undef, align 4
 
 // CHECK; define internal void @_init_resource_Buffer0()
 // CHECK-DXIL: %Buffer0_h = call target("dx.RawBuffer", i8, 0, 0) @llvm.dx.resource.handlefrombinding.tdx.RawBuffer_i8_0_0t(i32 0, i32 0, i32 1, i32 0, i1 false)
diff --git a/clang/test/CodeGenHLSL/builtins/RWBuffer-constructor-opt.hlsl b/clang/test/CodeGenHLSL/builtins/RWBuffer-constructor-opt.hlsl
@@ -1,21 +1,13 @@
-// RUN: %clang_cc1 -triple dxil-pc-shadermodel6.3-library -x hlsl -emit-llvm -O3 -o - %s | FileCheck %s --check-prefixes=CHECK,CHECK-DXIL
-// RUN: %clang_cc1 -triple spirv-vulkan-compute -x hlsl -emit-llvm -O3 -o - %s | FileCheck %s --check-prefixes=CHECK,CHECK-SPIRV
+// RUN: %clang_cc1 -triple dxil-pc-shadermodel6.3-library -x hlsl -emit-llvm -O3 -o - %s | FileCheck %s
+// RUN: %clang_cc1 -triple spirv-vulkan-compute -x hlsl -emit-llvm -O3 -o - %s | FileCheck %s
 
-// CHECK-SPIRV: %"class.hlsl::RWBuffer" = type { target("spirv.Image", float, 5, 2, 0, 0, 2, 0) }
-// CHECK-DXIL:  %"class.hlsl::RWBuffer" = type { target("dx.TypedBuffer", float, 1, 0, 0) }
+// All referenced to an unused resource should be removed by optimizations.
 RWBuffer<float> Buf : register(u5, space3);
 
 [shader("compute")]
 [numthreads(1, 1, 1)]
 void main() {
 // CHECK: define void @main()
 // CHECK-NEXT: entry:
-
-// CHECK-SPIRV-NEXT: %[[HANDLE:.*]] = tail call target("spirv.Image", float, 5, 2, 0, 0, 2, 0) @llvm.spv.resource.handlefrombinding.tspirv.Image_f32_5_2_0_0_2_0t(i32 3, i32 5, i32 1, i32 0, i1 false)
-// CHECK-SPIRV-NEXT: store target("spirv.Image", float, 5, 2, 0, 0, 2, 0) %[[HANDLE:.*]], ptr @Buf, align 8
-
-// CHECK-DXIL-NEXT: %[[HANDLE:.*]] = tail call target("dx.TypedBuffer", float, 1, 0, 0) @llvm.dx.resource.handlefrombinding.tdx.TypedBuffer_f32_1_0_0t(i32 3, i32 5, i32 1, i32 0, i1 false)
-// CHECK-DXIL-NEXT: store target("dx.TypedBuffer", float, 1, 0, 0) %[[HANDLE]], ptr @Buf, align 4
-
 // CHECK-NEXT: ret void
 }
diff --git a/clang/test/CodeGenHLSL/builtins/RWBuffer-constructor.hlsl b/clang/test/CodeGenHLSL/builtins/RWBuffer-constructor.hlsl
@@ -7,7 +7,7 @@
 RWBuffer<float> Buf : register(u5, space3);
 
 // CHECK: %"class.hlsl::RWBuffer" = type { target("dx.TypedBuffer", float, 1, 0, 0) }
-// CHECK: @Buf = global %"class.hlsl::RWBuffer" zeroinitializer, align 4
+// CHECK: @Buf = internal global %"class.hlsl::RWBuffer" undef, align 4
 
 // CHECK: define internal void @_init_resource_Buf()
 // CHECK-DXIL: %Buf_h = call target("dx.TypedBuffer", float, 1, 0, 0) @llvm.dx.resource.handlefrombinding.tdx.TypedBuffer_f32_1_0_0t(i32 3, i32 5, i32 1, i32 0, i1 false)
diff --git a/clang/test/CodeGenHLSL/builtins/StructuredBuffers-constructors.hlsl b/clang/test/CodeGenHLSL/builtins/StructuredBuffers-constructors.hlsl
@@ -15,11 +15,11 @@ RasterizerOrderedStructuredBuffer<float> Buf5 : register(u1, space2);
 // CHECK: %"class.hlsl::ConsumeStructuredBuffer" = type { target("dx.RawBuffer", float, 1, 0) }
 // CHECK: %"class.hlsl::RasterizerOrderedStructuredBuffer" = type { target("dx.RawBuffer", float, 1, 1) }
 
-// CHECK: @Buf = global %"class.hlsl::StructuredBuffer" zeroinitializer, align 4
-// CHECK: @Buf2 = global %"class.hlsl::RWStructuredBuffer" zeroinitializer, align 4
-// CHECK: @Buf3 = global %"class.hlsl::AppendStructuredBuffer" zeroinitializer, align 4
-// CHECK: @Buf4 = global %"class.hlsl::ConsumeStructuredBuffer" zeroinitializer, align 4
-// CHECK: @Buf5 = global %"class.hlsl::RasterizerOrderedStructuredBuffer" zeroinitializer, align 4
+// CHECK: @Buf = internal global %"class.hlsl::StructuredBuffer" undef, align 4
+// CHECK: @Buf2 = internal global %"class.hlsl::RWStructuredBuffer" undef, align 4
+// CHECK: @Buf3 = internal global %"class.hlsl::AppendStructuredBuffer" undef, align 4
+// CHECK: @Buf4 = internal global %"class.hlsl::ConsumeStructuredBuffer" undef, align 4
+// CHECK: @Buf5 = internal global %"class.hlsl::RasterizerOrderedStructuredBuffer" undef, align 4
 
 // CHECK: define internal void @_init_resource_Buf()
 // CHECK-DXIL: %Buf_h = call target("dx.RawBuffer", float, 0, 0) @llvm.dx.resource.handlefrombinding.tdx.RawBuffer_f32_0_0t(i32 0, i32 10, i32 1, i32 0, i1 false)
