[HLSL] Add Load overload with status (llvm#166449)

This PR adds a Load method for resources, which takes an additional
parameter by reference, status. It fills the status parameter with a 1
or 0, depending on whether or not the resource access was mapped.
CheckAccessFullyMapped is also added as an intrinsic, and called in the
production of this status bit.
Only addresses DXIL for the below issue:
llvm#138910
Also only addresses the DXIL variant for the below issue:
llvm#99204

Author: bob80905

clang/include/clang/Basic/Builtins.td

@@ -4946,6 +4946,12 @@ def HLSLResourceGetPointer : LangBuiltin<"HLSL_LANG"> {
   let Prototype = "void(...)";
 }
 
+def HLSLResourceLoadWithStatus : LangBuiltin<"HLSL_LANG"> {
+  let Spellings = ["__builtin_hlsl_resource_load_with_status"];
+  let Attributes = [NoThrow];
+  let Prototype = "void(...)";
+}
+
 def HLSLResourceUninitializedHandle : LangBuiltin<"HLSL_LANG"> {
   let Spellings = ["__builtin_hlsl_resource_uninitializedhandle"];
   let Attributes = [NoThrow];

clang/lib/CodeGen/CGHLSLBuiltins.cpp

@@ -404,6 +404,49 @@ Value *CodeGenFunction::EmitHLSLBuiltinExpr(unsigned BuiltinID,
         RetTy, CGM.getHLSLRuntime().getCreateResourceGetPointerIntrinsic(),
         ArrayRef<Value *>{HandleOp, IndexOp});
   }
+  case Builtin::BI__builtin_hlsl_resource_load_with_status: {
+    Value *HandleOp = EmitScalarExpr(E->getArg(0));
+    Value *IndexOp = EmitScalarExpr(E->getArg(1));
+
+    // Get the *address* of the status argument to write to it by reference
+    LValue StatusLVal = EmitLValue(E->getArg(2));
+    Address StatusAddr = StatusLVal.getAddress();
+
+    QualType HandleTy = E->getArg(0)->getType();
+    const HLSLAttributedResourceType *RT =
+        HandleTy->getAs<HLSLAttributedResourceType>();
+    assert(CGM.getTarget().getTriple().getArch() == llvm::Triple::dxil &&
+           "Only DXIL currently implements load with status");
+
+    Intrinsic::ID IntrID = RT->getAttrs().RawBuffer
+                               ? llvm::Intrinsic::dx_resource_load_rawbuffer
+                               : llvm::Intrinsic::dx_resource_load_typedbuffer;
+
+    llvm::Type *DataTy = ConvertType(E->getType());
+    llvm::Type *RetTy = llvm::StructType::get(Builder.getContext(),
+                                              {DataTy, Builder.getInt1Ty()});
+
+    SmallVector<Value *, 3> Args;
+    Args.push_back(HandleOp);
+    Args.push_back(IndexOp);
+
+    if (RT->getAttrs().RawBuffer) {
+      Value *Offset = Builder.getInt32(0);
+      Args.push_back(Offset);
+    }
+
+    // The load intrinsics give us a (T value, i1 status) pair -
+    // shepherd these into the return value and out reference respectively.
+    Value *ResRet =
+        Builder.CreateIntrinsic(RetTy, IntrID, Args, {}, "ld.struct");
+    Value *LoadedValue = Builder.CreateExtractValue(ResRet, {0}, "ld.value");
+    Value *StatusBit = Builder.CreateExtractValue(ResRet, {1}, "ld.status");
+    Value *ExtendedStatus =
+        Builder.CreateZExt(StatusBit, Builder.getInt32Ty(), "ld.status.ext");
+    Builder.CreateStore(ExtendedStatus, StatusAddr);
+
+    return LoadedValue;
+  }
   case Builtin::BI__builtin_hlsl_resource_uninitializedhandle: {
     llvm::Type *HandleTy = CGM.getTypes().ConvertType(E->getType());
     return llvm::PoisonValue::get(HandleTy);

clang/lib/Headers/hlsl/hlsl_intrinsics.h

@@ -666,6 +666,10 @@ smoothstep(__detail::HLSL_FIXED_VECTOR<float, N> Min,
   return __detail::smoothstep_vec_impl(Min, Max, X);
 }
 
+inline bool CheckAccessFullyMapped(uint Status) {
+  return static_cast<bool>(Status);
+}
+
 //===----------------------------------------------------------------------===//
 // fwidth builtin
 //===----------------------------------------------------------------------===//

clang/lib/Sema/HLSLBuiltinTypeDeclBuilder.cpp

@@ -202,7 +202,7 @@ struct BuiltinTypeMethodBuilder {
   BuiltinTypeMethodBuilder &declareLocalVar(LocalVar &Var);
   template <typename... Ts>
   BuiltinTypeMethodBuilder &callBuiltin(StringRef BuiltinName,
-                                        QualType ReturnType, Ts... ArgSpecs);
+                                        QualType ReturnType, Ts &&...ArgSpecs);
   template <typename TLHS, typename TRHS>
   BuiltinTypeMethodBuilder &assign(TLHS LHS, TRHS RHS);
   template <typename T> BuiltinTypeMethodBuilder &dereference(T Ptr);
@@ -572,7 +572,7 @@ BuiltinTypeMethodBuilder &BuiltinTypeMethodBuilder::returnThis() {
 template <typename... Ts>
 BuiltinTypeMethodBuilder &
 BuiltinTypeMethodBuilder::callBuiltin(StringRef BuiltinName,
-                                      QualType ReturnType, Ts... ArgSpecs) {
+                                      QualType ReturnType, Ts &&...ArgSpecs) {
   ensureCompleteDecl();
 
   std::array<Expr *, sizeof...(ArgSpecs)> Args{
@@ -1140,6 +1140,7 @@ BuiltinTypeDeclBuilder &BuiltinTypeDeclBuilder::addLoadMethods() {
   DeclarationName Load(&II);
   // TODO: We also need versions with status for CheckAccessFullyMapped.
   addHandleAccessFunction(Load, /*IsConst=*/false, /*IsRef=*/false);
+  addLoadWithStatusFunction(Load, /*IsConst=*/false);
 
   return *this;
 }
@@ -1232,6 +1233,22 @@ BuiltinTypeDeclBuilder &BuiltinTypeDeclBuilder::addDecrementCounterMethod() {
       .finalize();
 }
 
+BuiltinTypeDeclBuilder &
+BuiltinTypeDeclBuilder::addLoadWithStatusFunction(DeclarationName &Name,
+                                                  bool IsConst) {
+  assert(!Record->isCompleteDefinition() && "record is already complete");
+  ASTContext &AST = SemaRef.getASTContext();
+  using PH = BuiltinTypeMethodBuilder::PlaceHolder;
+
+  QualType ReturnTy = getHandleElementType();
+  return BuiltinTypeMethodBuilder(*this, Name, ReturnTy, IsConst)
+      .addParam("Index", AST.UnsignedIntTy)
+      .addParam("Status", AST.UnsignedIntTy, HLSLParamModifierAttr::Keyword_out)
+      .callBuiltin("__builtin_hlsl_resource_load_with_status", ReturnTy,
+                   PH::Handle, PH::_0, PH::_1)
+      .finalize();
+}
+
 BuiltinTypeDeclBuilder &
 BuiltinTypeDeclBuilder::addHandleAccessFunction(DeclarationName &Name,
                                                 bool IsConst, bool IsRef) {

clang/lib/Sema/HLSLBuiltinTypeDeclBuilder.h

@@ -91,6 +91,8 @@ class BuiltinTypeDeclBuilder {
   BuiltinTypeDeclBuilder &addDecrementCounterMethod();
   BuiltinTypeDeclBuilder &addHandleAccessFunction(DeclarationName &Name,
                                                   bool IsConst, bool IsRef);
+  BuiltinTypeDeclBuilder &addLoadWithStatusFunction(DeclarationName &Name,
+                                                    bool IsConst);
   BuiltinTypeDeclBuilder &addAppendMethod();
   BuiltinTypeDeclBuilder &addConsumeMethod();
 

clang/lib/Sema/SemaHLSL.cpp

@@ -3057,6 +3057,24 @@ bool SemaHLSL::CheckBuiltinFunctionCall(unsigned BuiltinID, CallExpr *TheCall) {
 
     break;
   }
+  case Builtin::BI__builtin_hlsl_resource_load_with_status: {
+    if (SemaRef.checkArgCount(TheCall, 3) ||
+        CheckResourceHandle(&SemaRef, TheCall, 0) ||
+        CheckArgTypeMatches(&SemaRef, TheCall->getArg(1),
+                            SemaRef.getASTContext().UnsignedIntTy) ||
+        CheckArgTypeMatches(&SemaRef, TheCall->getArg(2),
+                            SemaRef.getASTContext().UnsignedIntTy) ||
+        CheckModifiableLValue(&SemaRef, TheCall, 2))
+      return true;
+
+    auto *ResourceTy =
+        TheCall->getArg(0)->getType()->castAs<HLSLAttributedResourceType>();
+    QualType ReturnType = ResourceTy->getContainedType();
+    TheCall->setType(ReturnType);
+
+    break;
+  }
+
   case Builtin::BI__builtin_hlsl_resource_uninitializedhandle: {
     assert(TheCall->getNumArgs() == 1 && "expected 1 arg");
     // Update return type to be the attributed resource type from arg0.

clang/test/AST/HLSL/StructuredBuffers-AST.hlsl

@@ -326,6 +326,26 @@ RESOURCE<float> Buffer;
 // CHECK-LOAD-NEXT: DeclRefExpr {{.*}} 'unsigned int' ParmVar {{.*}} 'Index' 'unsigned int'
 // CHECK-LOAD-NEXT: AlwaysInlineAttr {{.*}} Implicit always_inline
 
+// Load with status method
+
+// CHECK-LOAD: CXXMethodDecl {{.*}} Load 'element_type (unsigned int, out unsigned int)'
+// CHECK-LOAD-NEXT: ParmVarDecl {{.*}} Index 'unsigned int'
+// CHECK-LOAD-NEXT: ParmVarDecl {{.*}} Status 'unsigned int &__restrict'
+// CHECK-LOAD-NEXT: HLSLParamModifierAttr {{.*}} out
+// CHECK-LOAD-NEXT: CompoundStmt
+// CHECK-LOAD-NEXT: ReturnStmt
+// CHECK-LOAD-NEXT: CallExpr {{.*}} 'element_type'
+// CHECK-LOAD-NEXT: ImplicitCastExpr {{.*}} <BuiltinFnToFnPtr>
+// CHECK-LOAD-NEXT: DeclRefExpr {{.*}} '<builtin fn type>' Function {{.*}} '__builtin_hlsl_resource_load_with_status' 'void (...) noexcept'
+// CHECK-LOAD-NEXT: MemberExpr {{.*}} '__hlsl_resource_t
+// CHECK-LOAD-UAV-SAME{LITERAL}: [[hlsl::resource_class(UAV)]]
+// CHECK-LOAD-SRV-SAME{LITERAL}: [[hlsl::resource_class(SRV)]]
+// CHECK-LOAD-SAME{LITERAL}: [[hlsl::contained_type(element_type)]]
+// CHECK-LOAD-NEXT: CXXThisExpr {{.*}} 'hlsl::[[RESOURCE]]<element_type>' lvalue implicit this
+// CHECK-LOAD-NEXT: DeclRefExpr {{.*}} 'unsigned int' ParmVar {{.*}} 'Index' 'unsigned int'
+// CHECK-LOAD-NEXT: DeclRefExpr {{.*}} 'unsigned int' ParmVar {{.*}} 'Status' 'unsigned int &__restrict'
+// CHECK-LOAD-NEXT: AlwaysInlineAttr {{.*}} Implicit always_inline
+
 // IncrementCounter method
 
 // CHECK-COUNTER: CXXMethodDecl {{.*}} IncrementCounter 'unsigned int ()'

clang/test/AST/HLSL/TypedBuffers-AST.hlsl

@@ -214,6 +214,25 @@ RESOURCE<float> Buffer;
 // CHECK-NEXT: DeclRefExpr {{.*}} 'unsigned int' ParmVar {{.*}}  'Index' 'unsigned int'
 // CHECK-NEXT: AlwaysInlineAttr {{.*}} Implicit always_inline
 
+// Load with status method
+// CHECK: CXXMethodDecl {{.*}} Load 'element_type (unsigned int, out unsigned int)'
+// CHECK-NEXT: ParmVarDecl {{.*}} Index 'unsigned int'
+// CHECK-NEXT: ParmVarDecl {{.*}} Status 'unsigned int &__restrict'
+// CHECK-NEXT: HLSLParamModifierAttr {{.*}} out
+// CHECK-NEXT: CompoundStmt
+// CHECK-NEXT: ReturnStmt
+// CHECK-NEXT: CallExpr {{.*}} 'element_type'
+// CHECK-NEXT: ImplicitCastExpr {{.*}} 'void (*)(...) noexcept' <BuiltinFnToFnPtr>
+// CHECK-NEXT: DeclRefExpr {{.*}} '<builtin fn type>' Function {{.*}} '__builtin_hlsl_resource_load_with_status' 'void (...) noexcept'
+// CHECK-NEXT: MemberExpr {{.*}} '__hlsl_resource_t
+// CHECK-UAV-SAME{LITERAL}: [[hlsl::resource_class(UAV)]]
+// CHECK-SRV-SAME{LITERAL}: [[hlsl::resource_class(SRV)]]
+// CHECK-SAME{LITERAL}: [[hlsl::contained_type(element_type)]]
+// CHECK-NEXT: CXXThisExpr {{.*}} 'hlsl::[[RESOURCE]]<element_type>' lvalue implicit this
+// CHECK-NEXT: DeclRefExpr {{.*}} 'unsigned int' ParmVar {{.*}} 'Index' 'unsigned int'
+// CHECK-NEXT: DeclRefExpr {{.*}} 'unsigned int' ParmVar {{.*}} 'Status' 'unsigned int &__restrict'
+// CHECK-NEXT: AlwaysInlineAttr {{.*}} Implicit always_inline
+
 // GetDimensions method
 
 // CHECK-NEXT: CXXMethodDecl {{.*}} GetDimensions 'void (out unsigned int)'

clang/test/CodeGenHLSL/resources/StructuredBuffers-methods-lib.hlsl

@@ -104,6 +104,44 @@ export float TestLoad() {
 // CHECK-NEXT: %[[VAL:.*]] = load float, ptr %[[PTR]]
 // CHECK-NEXT: ret float %[[VAL]]
 
+export float TestLoadWithStatus() {
+    uint s1;
+    uint s2;
+    float ret = RWSB1.Load(1, s1) + SB1.Load(2, s2);
+    ret += float(s1 + s2);
+    return ret;
+}
+
+// CHECK: define noundef nofpclass(nan inf) float @TestLoadWithStatus()()
+// CHECK: call {{.*}} float @hlsl::RWStructuredBuffer<float>::Load(unsigned int, unsigned int&)(ptr {{.*}} @RWSB1, i32 noundef 1, ptr {{.*}} %tmp)
+// CHECK: call {{.*}} float @hlsl::StructuredBuffer<float>::Load(unsigned int, unsigned int&)(ptr {{.*}} @SB1, i32 noundef 2, ptr {{.*}} %tmp1)
+// CHECK: add
+// CHECK: ret float
+
+// CHECK: define {{.*}} float @hlsl::RWStructuredBuffer<float>::Load(unsigned int, unsigned int&)(ptr {{.*}} %this, i32 noundef %Index, ptr {{.*}} %Status)
+// CHECK: %__handle = getelementptr inbounds nuw %"class.hlsl::RWStructuredBuffer", ptr %{{.*}}, i32 0, i32 0
+// DXIL-NEXT: %[[HANDLE:.*]] = load target("dx.RawBuffer", float, 1, 0), ptr %__handle
+// CHECK-NEXT: %[[INDEX:.*]] = load i32, ptr %Index.addr
+// CHECK-NEXT: %[[LOADED_STATUS_ADDR:.*]] = load ptr, ptr %Status.addr
+// DXIL-NEXT: %[[STRUCT:.*]] = call { float, i1 } @llvm.dx.resource.load.rawbuffer.f32.tdx.RawBuffer_f32_1_0t(target("dx.RawBuffer", float, 1, 0) %[[HANDLE]], i32 %[[INDEX]], i32 0)
+// CHECK-NEXT: %[[VALUE:.*]] = extractvalue { float, i1 } %[[STRUCT]], 0
+// CHECK-NEXT: %[[STATUS_TEMP:.*]] = extractvalue { float, i1 } %[[STRUCT]], 1
+// CHECK-NEXT: %[[STATUS_EXT:.*]] = zext i1 %[[STATUS_TEMP]] to i32
+// CHECK-NEXT: store i32 %[[STATUS_EXT]], ptr %[[LOADED_STATUS_ADDR]], align 4
+// CHECK-NEXT: ret float %[[VALUE]]
+
+// CHECK: define {{.*}} float @hlsl::StructuredBuffer<float>::Load(unsigned int, unsigned int&)(ptr {{.*}} %this, i32 noundef %Index, ptr {{.*}} %Status)
+// CHECK: %__handle = getelementptr inbounds nuw %"class.hlsl::StructuredBuffer", ptr %{{.*}}, i32 0, i32 0
+// DXIL-NEXT: %[[HANDLE:.*]] = load target("dx.RawBuffer", float, 0, 0), ptr %__handle
+// CHECK-NEXT: %[[INDEX:.*]] = load i32, ptr %Index.addr
+// CHECK-NEXT: %[[LOADED_STATUS_ADDR:.*]] = load ptr, ptr %Status.addr, align 4
+// DXIL-NEXT: %[[STRUCT:.*]] = call { float, i1 } @llvm.dx.resource.load.rawbuffer.f32.tdx.RawBuffer_f32_0_0t(target("dx.RawBuffer", float, 0, 0) %[[HANDLE]], i32 %[[INDEX]], i32 0)
+// CHECK-NEXT: %[[VALUE:.*]] = extractvalue { float, i1 } %[[STRUCT]], 0
+// CHECK-NEXT: %[[STATUS_TEMP:.*]] = extractvalue { float, i1 } %[[STRUCT]], 1
+// CHECK-NEXT: %[[STATUS_EXT:.*]] = zext i1 %[[STATUS_TEMP]] to i32
+// CHECK-NEXT: store i32 %[[STATUS_EXT]], ptr %[[LOADED_STATUS_ADDR]], align 4
+// CHECK-NEXT: ret float %[[VALUE]]
+
 export uint TestGetDimensions() {
     uint dim1, dim2, dim3, stride1, stride2, stride3;
     SB1.GetDimensions(dim1, stride1);

clang/test/CodeGenHLSL/resources/StructuredBuffers-methods-ps.hlsl

@@ -65,6 +65,43 @@ export float TestLoad() {
 // CHECK-NEXT: %[[VAL:.*]] = load <2 x i32>, ptr %[[BUFPTR]]
 // CHECK-NEXT: ret <2 x i32> %[[VAL]]
 
+export float TestLoadWithStatus() {
+    uint status;
+    uint status2;
+    float val = ROSB1.Load(10, status).x + ROSB2.Load(20, status2).x;
+    return val + float(status + status2);
+}
+
+// CHECK: define {{.*}} float @TestLoadWithStatus()()
+// CHECK: call {{.*}} float @hlsl::RasterizerOrderedStructuredBuffer<float>::Load(unsigned int, unsigned int&)(ptr {{.*}} @ROSB1, i32 noundef 10, ptr {{.*}} %tmp)
+// CHECK: call {{.*}} <2 x i32> @hlsl::RasterizerOrderedStructuredBuffer<int vector[2]>::Load(unsigned int, unsigned int&)(ptr {{.*}} @ROSB2, i32 noundef 20, ptr {{.*}} %tmp2)
+// CHECK: ret
+
+// CHECK: define {{.*}} float @hlsl::RasterizerOrderedStructuredBuffer<float>::Load(unsigned int, unsigned int&)(ptr {{.*}} %this, i32 noundef %Index, ptr {{.*}} %Status)
+// CHECK: %__handle = getelementptr inbounds nuw %"class.hlsl::RasterizerOrderedStructuredBuffer", ptr {{.*}}, i32 0, i32 0
+// CHECK-NEXT: %[[HANDLE:.*]] = load target("dx.RawBuffer", float, 1, 1), ptr %__handle
+// CHECK-NEXT: %[[INDEX:.*]] = load i32, ptr %Index.addr
+// CHECK-NEXT: %[[LOADED_STATUS_ADDR:.*]] = load ptr, ptr %Status.addr, align 4
+// DXIL-NEXT: %[[STRUCT:.*]] = call { float, i1 } @llvm.dx.resource.load.rawbuffer.f32.tdx.RawBuffer_f32_1_1t(target("dx.RawBuffer", float, 1, 1) %[[HANDLE]], i32 %[[INDEX]], i32 0)
+// CHECK-NEXT: %[[VALUE:.*]] = extractvalue { float, i1 } %[[STRUCT]], 0
+// CHECK-NEXT: %[[STATUS_TEMP:.*]] = extractvalue { float, i1 } %[[STRUCT]], 1
+// CHECK-NEXT: %[[STATUS_EXT:.*]] = zext i1 %[[STATUS_TEMP]] to i32
+// CHECK-NEXT: store i32 %[[STATUS_EXT]], ptr %[[LOADED_STATUS_ADDR]], align 4
+// CHECK-NEXT: ret float %[[VALUE]]
+
+// CHECK: define {{.*}} <2 x i32> @hlsl::RasterizerOrderedStructuredBuffer<int vector[2]>::Load(unsigned int, unsigned int&)(ptr {{.*}} %this, i32 noundef %Index, ptr {{.*}} %Status)
+// CHECK: %__handle = getelementptr inbounds nuw %"class.hlsl::RasterizerOrderedStructuredBuffer.0", ptr {{.*}}, i32 0, i32 0
+// CHECK-NEXT: %[[HANDLE:.*]] = load target("dx.RawBuffer", <2 x i32>, 1, 1), ptr %__handle
+// CHECK-NEXT: %[[INDEX:.*]] = load i32, ptr %Index.addr
+// CHECK-NEXT: %[[LOADED_STATUS_ADDR:.*]] = load ptr, ptr %Status.addr, align 4
+// DXIL-NEXT: %[[STRUCT:.*]] = call { <2 x i32>, i1 } @llvm.dx.resource.load.rawbuffer.v2i32.tdx.RawBuffer_v2i32_1_1t(target("dx.RawBuffer", <2 x i32>, 1, 1) %[[HANDLE]], i32 %[[INDEX]], i32 0)
+// CHECK-NEXT: %[[VALUE:.*]] = extractvalue { <2 x i32>, i1 } %[[STRUCT]], 0
+// CHECK-NEXT: %[[STATUS_TEMP:.*]] = extractvalue { <2 x i32>, i1 } %[[STRUCT]], 1
+// CHECK-NEXT: %[[STATUS_EXT:.*]] = zext i1 %[[STATUS_TEMP]] to i32
+// CHECK-NEXT: store i32 %[[STATUS_EXT]], ptr %[[LOADED_STATUS_ADDR]], align 4
+// CHECK-NEXT: ret <2 x i32> %[[VALUE]]
+
+
 export uint TestGetDimensions() {
     uint dim1, dim2, stride1, stride2;
     ROSB1.GetDimensions(dim1, stride1);