[CodeGen] Avoid pointer element type access for blocks

Pass the block struct type down to the TargetInfo hooks.

clang/lib/CodeGen/CGBlocks.cpp

@@ -1105,7 +1105,7 @@ llvm::Value *CodeGenFunction::EmitBlockLiteral(const CGBlockInfo &blockInfo) {
 
   if (IsOpenCL) {
     CGM.getOpenCLRuntime().recordBlockInfo(blockInfo.BlockExpression, InvokeFn,
-                                           result);
+                                           result, blockInfo.StructureType);
   }
 
   return result;
@@ -1401,7 +1401,8 @@ static llvm::Constant *buildGlobalBlock(CodeGenModule &CGM,
   if (CGM.getContext().getLangOpts().OpenCL)
     CGM.getOpenCLRuntime().recordBlockInfo(
         blockInfo.BlockExpression,
-        cast<llvm::Function>(blockFn->stripPointerCasts()), Result);
+        cast<llvm::Function>(blockFn->stripPointerCasts()), Result,
+        literal->getValueType());
   return Result;
 }
 

clang/lib/CodeGen/CGOpenCLRuntime.cpp

@@ -148,13 +148,14 @@ static const BlockExpr *getBlockExpr(const Expr *E) {
 /// corresponding block expression.
 void CGOpenCLRuntime::recordBlockInfo(const BlockExpr *E,
                                       llvm::Function *InvokeF,
-                                      llvm::Value *Block) {
+                                      llvm::Value *Block, llvm::Type *BlockTy) {
   assert(EnqueuedBlockMap.find(E) == EnqueuedBlockMap.end() &&
          "Block expression emitted twice");
   assert(isa<llvm::Function>(InvokeF) && "Invalid invoke function");
   assert(Block->getType()->isPointerTy() && "Invalid block literal type");
   EnqueuedBlockMap[E].InvokeFunc = InvokeF;
   EnqueuedBlockMap[E].BlockArg = Block;
+  EnqueuedBlockMap[E].BlockTy = BlockTy;
   EnqueuedBlockMap[E].Kernel = nullptr;
 }
 
@@ -179,8 +180,7 @@ CGOpenCLRuntime::emitOpenCLEnqueuedBlock(CodeGenFunction &CGF, const Expr *E) {
   }
 
   auto *F = CGF.getTargetHooks().createEnqueuedBlockKernel(
-      CGF, EnqueuedBlockMap[Block].InvokeFunc,
-      EnqueuedBlockMap[Block].BlockArg->stripPointerCasts());
+      CGF, EnqueuedBlockMap[Block].InvokeFunc, EnqueuedBlockMap[Block].BlockTy);
 
   // The common part of the post-processing of the kernel goes here.
   F->addFnAttr(llvm::Attribute::NoUnwind);

clang/lib/CodeGen/CGOpenCLRuntime.h

@@ -46,6 +46,7 @@ protected:
     llvm::Function *InvokeFunc; /// Block invoke function.
     llvm::Function *Kernel;     /// Enqueued block kernel.
     llvm::Value *BlockArg;      /// The first argument to enqueued block kernel.
+    llvm::Type *BlockTy;        /// Type of the block argument.
   };
   /// Maps block expression to block information.
   llvm::DenseMap<const Expr *, EnqueuedBlockInfo> EnqueuedBlockMap;
@@ -93,7 +94,7 @@ public:
   /// \param InvokeF invoke function emitted for the block expression.
   /// \param Block block literal emitted for the block expression.
   void recordBlockInfo(const BlockExpr *E, llvm::Function *InvokeF,
-                       llvm::Value *Block);
+                       llvm::Value *Block, llvm::Type *BlockTy);
 
   /// \return LLVM block invoke function emitted for an expression derived from
   /// the block expression.

clang/lib/CodeGen/TargetInfo.cpp

@@ -9214,7 +9214,7 @@ public:
   llvm::Function *
   createEnqueuedBlockKernel(CodeGenFunction &CGF,
                             llvm::Function *BlockInvokeFunc,
-                            llvm::Value *BlockLiteral) const override;
+                            llvm::Type *BlockTy) const override;
   bool shouldEmitStaticExternCAliases() const override;
   void setCUDAKernelCallingConvention(const FunctionType *&FT) const override;
 };
@@ -11461,7 +11461,7 @@ const TargetCodeGenInfo &CodeGenModule::getTargetCodeGenInfo() {
 llvm::Function *
 TargetCodeGenInfo::createEnqueuedBlockKernel(CodeGenFunction &CGF,
                                              llvm::Function *Invoke,
-                                             llvm::Value *BlockLiteral) const {
+                                             llvm::Type *BlockTy) const {
   auto *InvokeFT = Invoke->getFunctionType();
   llvm::SmallVector<llvm::Type *, 2> ArgTys;
   for (auto &P : InvokeFT->params())
@@ -11495,11 +11495,10 @@ TargetCodeGenInfo::createEnqueuedBlockKernel(CodeGenFunction &CGF,
 /// has "enqueued-block" function attribute and kernel argument metadata.
 llvm::Function *AMDGPUTargetCodeGenInfo::createEnqueuedBlockKernel(
     CodeGenFunction &CGF, llvm::Function *Invoke,
-    llvm::Value *BlockLiteral) const {
+    llvm::Type *BlockTy) const {
   auto &Builder = CGF.Builder;
   auto &C = CGF.getLLVMContext();
 
-  auto *BlockTy = BlockLiteral->getType()->getPointerElementType();
   auto *InvokeFT = Invoke->getFunctionType();
   llvm::SmallVector<llvm::Type *, 2> ArgTys;
   llvm::SmallVector<llvm::Metadata *, 8> AddressQuals;

clang/lib/CodeGen/TargetInfo.h

@@ -329,7 +329,7 @@ public:
   virtual llvm::Function *
   createEnqueuedBlockKernel(CodeGenFunction &CGF,
                             llvm::Function *BlockInvokeFunc,
-                            llvm::Value *BlockLiteral) const;
+                            llvm::Type *BlockTy) const;
 
   /// \return true if the target supports alias from the unmangled name to the
   /// mangled name of functions declared within an extern "C" region and marked