Index: clang/lib/CodeGen/CGBuiltin.cpp
===================================================================
--- clang/lib/CodeGen/CGBuiltin.cpp
+++ clang/lib/CodeGen/CGBuiltin.cpp
@@ -1060,7 +1060,10 @@
 
   llvm::InlineAsm *IA =
       llvm::InlineAsm::get(FTy, Asm, Constraints, /*hasSideEffects=*/true);
-  return CGF.Builder.CreateCall(IA, {Addr});
+  llvm::CallInst *CI = CGF.Builder.CreateCall(IA, {Addr});
+  CI->addParamAttr(0, Attribute::get(
+      CGF.getLLVMContext(), Attribute::ElementType, RetType));
+  return CI;
 }
 
 namespace {
Index: clang/lib/CodeGen/CGObjCMac.cpp
===================================================================
--- clang/lib/CodeGen/CGObjCMac.cpp
+++ clang/lib/CodeGen/CGObjCMac.cpp
@@ -4370,7 +4370,11 @@
 void FragileHazards::emitWriteHazard() {
   if (Locals.empty()) return;
 
-  CGF.EmitNounwindRuntimeCall(WriteHazard, Locals);
+  llvm::CallInst *Call = CGF.EmitNounwindRuntimeCall(WriteHazard, Locals);
+  for (auto Pair : llvm::enumerate(Locals))
+    Call->addParamAttr(Pair.index(), llvm::Attribute::get(
+        CGF.getLLVMContext(), llvm::Attribute::ElementType,
+        cast<llvm::AllocaInst>(Pair.value())->getAllocatedType()));
 }
 
 void FragileHazards::emitReadHazard(CGBuilderTy &Builder) {
@@ -4378,6 +4382,10 @@
   llvm::CallInst *call = Builder.CreateCall(ReadHazard, Locals);
   call->setDoesNotThrow();
   call->setCallingConv(CGF.getRuntimeCC());
+  for (auto Pair : llvm::enumerate(Locals))
+    call->addParamAttr(Pair.index(), llvm::Attribute::get(
+        Builder.getContext(), llvm::Attribute::ElementType,
+        cast<llvm::AllocaInst>(Pair.value())->getAllocatedType()));
 }
 
 /// Emit read hazards in all the protected blocks, i.e. all the blocks
Index: clang/lib/CodeGen/CGStmt.cpp
===================================================================
--- clang/lib/CodeGen/CGStmt.cpp
+++ clang/lib/CodeGen/CGStmt.cpp
@@ -2109,42 +2109,36 @@
   return (EarlyClobber ? "&{" : "{") + Register.str() + "}";
 }
 
-llvm::Value*
+std::pair<llvm::Value*, llvm::Type *>
 CodeGenFunction::EmitAsmInputLValue(const TargetInfo::ConstraintInfo &Info,
                                     LValue InputValue, QualType InputType,
                                     std::string &ConstraintStr,
                                     SourceLocation Loc) {
-  llvm::Value *Arg;
   if (Info.allowsRegister() || !Info.allowsMemory()) {
-    if (CodeGenFunction::hasScalarEvaluationKind(InputType)) {
-      Arg = EmitLoadOfLValue(InputValue, Loc).getScalarVal();
-    } else {
-      llvm::Type *Ty = ConvertType(InputType);
-      uint64_t Size = CGM.getDataLayout().getTypeSizeInBits(Ty);
-      if ((Size <= 64 && llvm::isPowerOf2_64(Size)) ||
-          getTargetHooks().isScalarizableAsmOperand(*this, Ty)) {
-        Ty = llvm::IntegerType::get(getLLVMContext(), Size);
-        Ty = llvm::PointerType::getUnqual(Ty);
-
-        Arg = Builder.CreateLoad(
-            Builder.CreateBitCast(InputValue.getAddress(*this), Ty));
-      } else {
-        Arg = InputValue.getPointer(*this);
-        ConstraintStr += '*';
-      }
+    if (CodeGenFunction::hasScalarEvaluationKind(InputType))
+      return {EmitLoadOfLValue(InputValue, Loc).getScalarVal(), nullptr};
+
+    llvm::Type *Ty = ConvertType(InputType);
+    uint64_t Size = CGM.getDataLayout().getTypeSizeInBits(Ty);
+    if ((Size <= 64 && llvm::isPowerOf2_64(Size)) ||
+        getTargetHooks().isScalarizableAsmOperand(*this, Ty)) {
+      Ty = llvm::IntegerType::get(getLLVMContext(), Size);
+      Ty = llvm::PointerType::getUnqual(Ty);
+
+      return {Builder.CreateLoad(
+          Builder.CreateBitCast(InputValue.getAddress(*this), Ty)), nullptr};
     }
-  } else {
-    Arg = InputValue.getPointer(*this);
-    ConstraintStr += '*';
   }
 
-  return Arg;
+  Address Addr = InputValue.getAddress(*this);
+  ConstraintStr += '*';
+  return {Addr.getPointer(), Addr.getElementType()};
 }
 
-llvm::Value* CodeGenFunction::EmitAsmInput(
-                                         const TargetInfo::ConstraintInfo &Info,
-                                           const Expr *InputExpr,
-                                           std::string &ConstraintStr) {
+std::pair<llvm::Value *, llvm::Type *>
+CodeGenFunction::EmitAsmInput(const TargetInfo::ConstraintInfo &Info,
+                              const Expr *InputExpr,
+                              std::string &ConstraintStr) {
   // If this can't be a register or memory, i.e., has to be a constant
   // (immediate or symbolic), try to emit it as such.
   if (!Info.allowsRegister() && !Info.allowsMemory()) {
@@ -2155,19 +2149,20 @@
       llvm::APSInt IntResult;
       if (EVResult.Val.toIntegralConstant(IntResult, InputExpr->getType(),
                                           getContext()))
-        return llvm::ConstantInt::get(getLLVMContext(), IntResult);
+        return {llvm::ConstantInt::get(getLLVMContext(), IntResult), nullptr};
     }
 
     Expr::EvalResult Result;
     if (InputExpr->EvaluateAsInt(Result, getContext()))
-      return llvm::ConstantInt::get(getLLVMContext(), Result.Val.getInt());
+      return {llvm::ConstantInt::get(getLLVMContext(), Result.Val.getInt()),
+              nullptr};
   }
 
   if (Info.allowsRegister() || !Info.allowsMemory())
     if (CodeGenFunction::hasScalarEvaluationKind(InputExpr->getType()))
-      return EmitScalarExpr(InputExpr);
+      return {EmitScalarExpr(InputExpr), nullptr};
   if (InputExpr->getStmtClass() == Expr::CXXThisExprClass)
-    return EmitScalarExpr(InputExpr);
+    return {EmitScalarExpr(InputExpr), nullptr};
   InputExpr = InputExpr->IgnoreParenNoopCasts(getContext());
   LValue Dest = EmitLValue(InputExpr);
   return EmitAsmInputLValue(Info, Dest, InputExpr->getType(), ConstraintStr,
@@ -2209,6 +2204,7 @@
                               bool HasUnwindClobber, bool ReadOnly,
                               bool ReadNone, bool NoMerge, const AsmStmt &S,
                               const std::vector<llvm::Type *> &ResultRegTypes,
+                              const std::vector<llvm::Type *> &ArgElemTypes,
                               CodeGenFunction &CGF,
                               std::vector<llvm::Value *> &RegResults) {
   if (!HasUnwindClobber)
@@ -2224,6 +2220,15 @@
       Result.addFnAttr(llvm::Attribute::ReadOnly);
   }
 
+  // Add elementtype attribute for indirect constraints.
+  for (auto Pair : llvm::enumerate(ArgElemTypes)) {
+    if (Pair.value()) {
+      auto Attr = llvm::Attribute::get(
+          CGF.getLLVMContext(), llvm::Attribute::ElementType, Pair.value());
+      Result.addParamAttr(Pair.index(), Attr);
+    }
+  }
+
   // Slap the source location of the inline asm into a !srcloc metadata on the
   // call.
   if (const auto *gccAsmStmt = dyn_cast<GCCAsmStmt>(&S))
@@ -2291,6 +2296,7 @@
   std::vector<llvm::Type *> ResultRegTypes;
   std::vector<llvm::Type *> ResultTruncRegTypes;
   std::vector<llvm::Type *> ArgTypes;
+  std::vector<llvm::Type *> ArgElemTypes;
   std::vector<llvm::Value*> Args;
   llvm::BitVector ResultTypeRequiresCast;
 
@@ -2298,6 +2304,7 @@
   std::string InOutConstraints;
   std::vector<llvm::Value*> InOutArgs;
   std::vector<llvm::Type*> InOutArgTypes;
+  std::vector<llvm::Type*> InOutArgElemTypes;
 
   // Keep track of out constraints for tied input operand.
   std::vector<std::string> OutputConstraints;
@@ -2399,21 +2406,19 @@
             std::max((uint64_t)LargestVectorWidth,
                      VT->getPrimitiveSizeInBits().getKnownMinSize());
     } else {
-      llvm::Type *DestAddrTy = Dest.getAddress(*this).getType();
-      llvm::Value *DestPtr = Dest.getPointer(*this);
+      Address DestAddr = Dest.getAddress(*this);
       // Matrix types in memory are represented by arrays, but accessed through
       // vector pointers, with the alignment specified on the access operation.
       // For inline assembly, update pointer arguments to use vector pointers.
       // Otherwise there will be a mis-match if the matrix is also an
       // input-argument which is represented as vector.
-      if (isa<MatrixType>(OutExpr->getType().getCanonicalType())) {
-        DestAddrTy = llvm::PointerType::get(
-            ConvertType(OutExpr->getType()),
-            cast<llvm::PointerType>(DestAddrTy)->getAddressSpace());
-        DestPtr = Builder.CreateBitCast(DestPtr, DestAddrTy);
-      }
-      ArgTypes.push_back(DestAddrTy);
-      Args.push_back(DestPtr);
+      if (isa<MatrixType>(OutExpr->getType().getCanonicalType()))
+        DestAddr = Builder.CreateElementBitCast(
+            DestAddr, ConvertType(OutExpr->getType()));
+
+      ArgTypes.push_back(DestAddr.getType());
+      ArgElemTypes.push_back(DestAddr.getElementType());
+      Args.push_back(DestAddr.getPointer());
       Constraints += "=*";
       Constraints += OutputConstraint;
       ReadOnly = ReadNone = false;
@@ -2423,9 +2428,11 @@
       InOutConstraints += ',';
 
       const Expr *InputExpr = S.getOutputExpr(i);
-      llvm::Value *Arg = EmitAsmInputLValue(Info, Dest, InputExpr->getType(),
-                                            InOutConstraints,
-                                            InputExpr->getExprLoc());
+      llvm::Value *Arg;
+      llvm::Type *ArgElemType;
+      std::tie(Arg, ArgElemType) = EmitAsmInputLValue(
+          Info, Dest, InputExpr->getType(), InOutConstraints,
+          InputExpr->getExprLoc());
 
       if (llvm::Type* AdjTy =
           getTargetHooks().adjustInlineAsmType(*this, OutputConstraint,
@@ -2444,6 +2451,7 @@
         InOutConstraints += OutputConstraint;
 
       InOutArgTypes.push_back(Arg->getType());
+      InOutArgElemTypes.push_back(ArgElemType);
       InOutArgs.push_back(Arg);
     }
   }
@@ -2483,7 +2491,9 @@
         getTarget(), CGM, S, false /* No EarlyClobber */);
 
     std::string ReplaceConstraint (InputConstraint);
-    llvm::Value *Arg = EmitAsmInput(Info, InputExpr, Constraints);
+    llvm::Value *Arg;
+    llvm::Type *ArgElemType;
+    std::tie(Arg, ArgElemType) = EmitAsmInput(Info, InputExpr, Constraints);
 
     // If this input argument is tied to a larger output result, extend the
     // input to be the same size as the output.  The LLVM backend wants to see
@@ -2528,6 +2538,7 @@
                    VT->getPrimitiveSizeInBits().getKnownMinSize());
 
     ArgTypes.push_back(Arg->getType());
+    ArgElemTypes.push_back(ArgElemType);
     Args.push_back(Arg);
     Constraints += InputConstraint;
   }
@@ -2546,6 +2557,7 @@
             llvm::BlockAddress::get(CurFn, Dest.getBlock());
         Args.push_back(BA);
         ArgTypes.push_back(BA->getType());
+        ArgElemTypes.push_back(nullptr);
         if (!Constraints.empty())
           Constraints += ',';
         Constraints += 'X';
@@ -2557,6 +2569,7 @@
   // Append the "input" part of inout constraints last.
   for (unsigned i = 0, e = InOutArgs.size(); i != e; i++) {
     ArgTypes.push_back(InOutArgTypes[i]);
+    ArgElemTypes.push_back(InOutArgElemTypes[i]);
     Args.push_back(InOutArgs[i]);
   }
   Constraints += InOutConstraints;
@@ -2647,18 +2660,18 @@
     EmitBlock(Fallthrough);
     UpdateAsmCallInst(cast<llvm::CallBase>(*Result), HasSideEffect, false,
                       ReadOnly, ReadNone, InNoMergeAttributedStmt, S,
-                      ResultRegTypes, *this, RegResults);
+                      ResultRegTypes, ArgElemTypes, *this, RegResults);
   } else if (HasUnwindClobber) {
     llvm::CallBase *Result = EmitCallOrInvoke(IA, Args, "");
     UpdateAsmCallInst(*Result, HasSideEffect, true, ReadOnly, ReadNone,
-                      InNoMergeAttributedStmt, S, ResultRegTypes, *this,
-                      RegResults);
+                      InNoMergeAttributedStmt, S, ResultRegTypes, ArgElemTypes,
+                      *this, RegResults);
   } else {
     llvm::CallInst *Result =
         Builder.CreateCall(IA, Args, getBundlesForFunclet(IA));
     UpdateAsmCallInst(cast<llvm::CallBase>(*Result), HasSideEffect, false,
                       ReadOnly, ReadNone, InNoMergeAttributedStmt, S,
-                      ResultRegTypes, *this, RegResults);
+                      ResultRegTypes, ArgElemTypes, *this, RegResults);
   }
 
   assert(RegResults.size() == ResultRegTypes.size());
Index: clang/lib/CodeGen/CodeGenFunction.h
===================================================================
--- clang/lib/CodeGen/CodeGenFunction.h
+++ clang/lib/CodeGen/CodeGenFunction.h
@@ -4677,13 +4677,14 @@
                         SmallVectorImpl<llvm::Value *> &IRCallArgs,
                         unsigned &IRCallArgPos);
 
-  llvm::Value* EmitAsmInput(const TargetInfo::ConstraintInfo &Info,
-                            const Expr *InputExpr, std::string &ConstraintStr);
-
-  llvm::Value* EmitAsmInputLValue(const TargetInfo::ConstraintInfo &Info,
-                                  LValue InputValue, QualType InputType,
-                                  std::string &ConstraintStr,
-                                  SourceLocation Loc);
+  std::pair<llvm::Value *, llvm::Type *>
+  EmitAsmInput(const TargetInfo::ConstraintInfo &Info, const Expr *InputExpr,
+               std::string &ConstraintStr);
+
+  std::pair<llvm::Value *, llvm::Type *>
+  EmitAsmInputLValue(const TargetInfo::ConstraintInfo &Info, LValue InputValue,
+                     QualType InputType, std::string &ConstraintStr,
+                     SourceLocation Loc);
 
   /// Attempts to statically evaluate the object size of E. If that
   /// fails, emits code to figure the size of E out for us. This is
Index: clang/test/CodeGen/RISCV/riscv-inline-asm.c
===================================================================
--- clang/test/CodeGen/RISCV/riscv-inline-asm.c
+++ clang/test/CodeGen/RISCV/riscv-inline-asm.c
@@ -41,7 +41,7 @@
 
 void test_A(int *p) {
 // CHECK-LABEL: define{{.*}} void @test_A(i32* %p)
-// CHECK: call void asm sideeffect "", "*A"(i32* %p)
+// CHECK: call void asm sideeffect "", "*A"(i32* elementtype(i32) %p)
   asm volatile("" :: "A"(*p));
 }
 
Index: clang/test/CodeGen/SystemZ/systemz-inline-asm.c
===================================================================
--- clang/test/CodeGen/SystemZ/systemz-inline-asm.c
+++ clang/test/CodeGen/SystemZ/systemz-inline-asm.c
@@ -6,31 +6,31 @@
 void test_store_m(unsigned int i) {
   asm("st %1, %0" : "=m" (gi) : "r" (i));
 // CHECK-LABEL: define{{.*}} void @test_store_m(i32 zeroext %i)
-// CHECK: call void asm "st $1, $0", "=*m,r"(i32* nonnull @gi, i32 %i)
+// CHECK: call void asm "st $1, $0", "=*m,r"(i32* nonnull elementtype(i32) @gi, i32 %i)
 }
 
 void test_store_Q(unsigned int i) {
   asm("st %1, %0" : "=Q" (gi) : "r" (i));
 // CHECK-LABEL: define{{.*}} void @test_store_Q(i32 zeroext %i)
-// CHECK: call void asm "st $1, $0", "=*Q,r"(i32* nonnull @gi, i32 %i)
+// CHECK: call void asm "st $1, $0", "=*Q,r"(i32* nonnull elementtype(i32) @gi, i32 %i)
 }
 
 void test_store_R(unsigned int i) {
   asm("st %1, %0" : "=R" (gi) : "r" (i));
 // CHECK-LABEL: define{{.*}} void @test_store_R(i32 zeroext %i)
-// CHECK: call void asm "st $1, $0", "=*R,r"(i32* nonnull @gi, i32 %i)
+// CHECK: call void asm "st $1, $0", "=*R,r"(i32* nonnull elementtype(i32) @gi, i32 %i)
 }
 
 void test_store_S(unsigned int i) {
   asm("st %1, %0" : "=S" (gi) : "r" (i));
 // CHECK-LABEL: define{{.*}} void @test_store_S(i32 zeroext %i)
-// CHECK: call void asm "st $1, $0", "=*S,r"(i32* nonnull @gi, i32 %i)
+// CHECK: call void asm "st $1, $0", "=*S,r"(i32* nonnull elementtype(i32) @gi, i32 %i)
 }
 
 void test_store_T(unsigned int i) {
   asm("st %1, %0" : "=T" (gi) : "r" (i));
 // CHECK-LABEL: define{{.*}} void @test_store_T(i32 zeroext %i)
-// CHECK: call void asm "st $1, $0", "=*T,r"(i32* nonnull @gi, i32 %i)
+// CHECK: call void asm "st $1, $0", "=*T,r"(i32* nonnull elementtype(i32) @gi, i32 %i)
 }
 
 int test_load_m() {
@@ -38,7 +38,7 @@
   asm("l %0, %1" : "=r" (i) : "m" (gi));
   return i;
 // CHECK-LABEL: define{{.*}} signext i32 @test_load_m()
-// CHECK: call i32 asm "l $0, $1", "=r,*m"(i32* nonnull @gi)
+// CHECK: call i32 asm "l $0, $1", "=r,*m"(i32* nonnull elementtype(i32) @gi)
 }
 
 int test_load_Q() {
@@ -46,7 +46,7 @@
   asm("l %0, %1" : "=r" (i) : "Q" (gi));
   return i;
 // CHECK-LABEL: define{{.*}} signext i32 @test_load_Q()
-// CHECK: call i32 asm "l $0, $1", "=r,*Q"(i32* nonnull @gi)
+// CHECK: call i32 asm "l $0, $1", "=r,*Q"(i32* nonnull elementtype(i32) @gi)
 }
 
 int test_load_R() {
@@ -54,7 +54,7 @@
   asm("l %0, %1" : "=r" (i) : "R" (gi));
   return i;
 // CHECK-LABEL: define{{.*}} signext i32 @test_load_R()
-// CHECK: call i32 asm "l $0, $1", "=r,*R"(i32* nonnull @gi)
+// CHECK: call i32 asm "l $0, $1", "=r,*R"(i32* nonnull elementtype(i32) @gi)
 }
 
 int test_load_S() {
@@ -62,7 +62,7 @@
   asm("l %0, %1" : "=r" (i) : "S" (gi));
   return i;
 // CHECK-LABEL: define{{.*}} signext i32 @test_load_S()
-// CHECK: call i32 asm "l $0, $1", "=r,*S"(i32* nonnull @gi)
+// CHECK: call i32 asm "l $0, $1", "=r,*S"(i32* nonnull elementtype(i32) @gi)
 }
 
 int test_load_T() {
@@ -70,13 +70,13 @@
   asm("l %0, %1" : "=r" (i) : "T" (gi));
   return i;
 // CHECK-LABEL: define{{.*}} signext i32 @test_load_T()
-// CHECK: call i32 asm "l $0, $1", "=r,*T"(i32* nonnull @gi)
+// CHECK: call i32 asm "l $0, $1", "=r,*T"(i32* nonnull elementtype(i32) @gi)
 }
 
 void test_mI(unsigned char *c) {
   asm volatile("cli %0, %1" :: "Q" (*c), "I" (100));
 // CHECK-LABEL: define{{.*}} void @test_mI(i8* %c)
-// CHECK: call void asm sideeffect "cli $0, $1", "*Q,I"(i8* %c, i32 100)
+// CHECK: call void asm sideeffect "cli $0, $1", "*Q,I"(i8* elementtype(i8) %c, i32 100)
 }
 
 unsigned int test_dJa(unsigned int i, unsigned int j) {
Index: clang/test/CodeGen/aarch64-inline-asm.c
===================================================================
--- clang/test/CodeGen/aarch64-inline-asm.c
+++ clang/test/CodeGen/aarch64-inline-asm.c
@@ -17,7 +17,7 @@
 
     asm("ldr %0, %1" : "=r"(var32) : "m"(var));
     asm("ldr %0, [%1]" : "=r"(var64) : "r"(&var));
-// CHECK: call i32 asm "ldr $0, $1", "=r,*m"(i64* @var)
+// CHECK: call i32 asm "ldr $0, $1", "=r,*m"(i64* elementtype(i64) @var)
 // CHECK: call i64 asm "ldr $0, [$1]", "=r,r"(i64* @var)
 }
 
@@ -52,7 +52,7 @@
 void test_constraint_Q(void) {
     int val;
     asm("ldxr %0, %1" : "=r"(val) : "Q"(var));
-// CHECK: call i32 asm "ldxr $0, $1", "=r,*Q"(i64* @var)
+// CHECK: call i32 asm "ldxr $0, $1", "=r,*Q"(i64* elementtype(i64) @var)
 }
 
 void test_gcc_registers(void) {
Index: clang/test/CodeGen/asm-inout.c
===================================================================
--- clang/test/CodeGen/asm-inout.c
+++ clang/test/CodeGen/asm-inout.c
@@ -5,7 +5,7 @@
 // CHECK: @test1
 void test1() {
   // CHECK: [[REGCALLRESULT:%[a-zA-Z0-9\.]+]] = call i32* @foo()
-  // CHECK: call void asm "foobar", "=*m,*m,~{dirflag},~{fpsr},~{flags}"(i32* [[REGCALLRESULT]], i32* [[REGCALLRESULT]])
+  // CHECK: call void asm "foobar", "=*m,*m,~{dirflag},~{fpsr},~{flags}"(i32* elementtype(i32) [[REGCALLRESULT]], i32* elementtype(i32) [[REGCALLRESULT]])
   asm ("foobar" : "+m"(*foo()));
 }
 
Index: clang/test/CodeGen/asm.c
===================================================================
--- clang/test/CodeGen/asm.c
+++ clang/test/CodeGen/asm.c
@@ -246,7 +246,7 @@
                :
                : "m"(t29_var));
   // CHECK: @t29
-  // CHECK: call void asm sideeffect "movl %eax, $0", "*m,~{dirflag},~{fpsr},~{flags}"([1 x i32]* @t29_var)
+  // CHECK: call void asm sideeffect "movl %eax, $0", "*m,~{dirflag},~{fpsr},~{flags}"([1 x i32]* elementtype([1 x i32]) @t29_var)
 }
 
 void t30(int len) {
Index: clang/test/CodeGen/builtins-ppc-xlcompat-LoadReseve-StoreCond-64bit-only.c
===================================================================
--- clang/test/CodeGen/builtins-ppc-xlcompat-LoadReseve-StoreCond-64bit-only.c
+++ clang/test/CodeGen/builtins-ppc-xlcompat-LoadReseve-StoreCond-64bit-only.c
@@ -10,7 +10,7 @@
 
 long test_ldarx(volatile long* a) {
   // CHECK64-LABEL: @test_ldarx
-  // CHECK64: %0 = tail call i64 asm sideeffect "ldarx $0, ${1:y}", "=r,*Z,~{memory}"(i64* %a)
+  // CHECK64: %0 = tail call i64 asm sideeffect "ldarx $0, ${1:y}", "=r,*Z,~{memory}"(i64* elementtype(i64) %a)
   // CHECK32-ERROR: error: this builtin is only available on 64-bit targets
   return __ldarx(a);
 }
Index: clang/test/CodeGen/builtins-ppc-xlcompat-LoadReseve-StoreCond.c
===================================================================
--- clang/test/CodeGen/builtins-ppc-xlcompat-LoadReseve-StoreCond.c
+++ clang/test/CodeGen/builtins-ppc-xlcompat-LoadReseve-StoreCond.c
@@ -13,20 +13,20 @@
 
 int test_lwarx(volatile int* a) {
   // CHECK-LABEL: @test_lwarx
-  // CHECK: %0 = tail call i32 asm sideeffect "lwarx $0, ${1:y}", "=r,*Z,~{memory}"(i32* %a)
+  // CHECK: %0 = tail call i32 asm sideeffect "lwarx $0, ${1:y}", "=r,*Z,~{memory}"(i32* elementtype(i32) %a)
   return __lwarx(a);
 }
 
 short test_lharx(volatile short *a) {
   // CHECK-LABEL: @test_lharx
-  // CHECK: %0 = tail call i16 asm sideeffect "lharx $0, ${1:y}", "=r,*Z,~{memory}"(i16* %a)
+  // CHECK: %0 = tail call i16 asm sideeffect "lharx $0, ${1:y}", "=r,*Z,~{memory}"(i16* elementtype(i16) %a)
   // CHECK-NON-PWR8-ERR:  error: this builtin is only valid on POWER8 or later CPUs
   return __lharx(a);
 }
 
 char test_lbarx(volatile char *a) {
   // CHECK-LABEL: @test_lbarx
-  // CHECK: %0 = tail call i8 asm sideeffect "lbarx $0, ${1:y}", "=r,*Z,~{memory}"(i8* %a)
+  // CHECK: %0 = tail call i8 asm sideeffect "lbarx $0, ${1:y}", "=r,*Z,~{memory}"(i8* elementtype(i8) %a)
   // CHECK-NON-PWR8-ERR:  error: this builtin is only valid on POWER8 or later CPUs
   return __lbarx(a);
 }
@@ -50,14 +50,14 @@
 // Extra test cases that previously caused error during usage.
 int test_lharx_intret(volatile short *a) {
   // CHECK-LABEL: @test_lharx_intret
-  // CHECK: %0 = tail call i16 asm sideeffect "lharx $0, ${1:y}", "=r,*Z,~{memory}"(i16* %a)
+  // CHECK: %0 = tail call i16 asm sideeffect "lharx $0, ${1:y}", "=r,*Z,~{memory}"(i16* elementtype(i16) %a)
   // CHECK-NON-PWR8-ERR:  error: this builtin is only valid on POWER8 or later CPUs
   return __lharx(a);
 }
 
 int test_lbarx_intret(volatile char *a) {
   // CHECK-LABEL: @test_lbarx_intret
-  // CHECK: %0 = tail call i8 asm sideeffect "lbarx $0, ${1:y}", "=r,*Z,~{memory}"(i8* %a)
+  // CHECK: %0 = tail call i8 asm sideeffect "lbarx $0, ${1:y}", "=r,*Z,~{memory}"(i8* elementtype(i8) %a)
   // CHECK-NON-PWR8-ERR:  error: this builtin is only valid on POWER8 or later CPUs
   return __lbarx(a);
 }
Index: clang/test/CodeGen/inline-asm-x86-flag-output.c
===================================================================
--- clang/test/CodeGen/inline-asm-x86-flag-output.c
+++ clang/test/CodeGen/inline-asm-x86-flag-output.c
@@ -2,7 +2,7 @@
 
 int test_cca(long nr, volatile long *addr) {
   //CHECK-LABEL: @test_cca
-  //CHECK: = tail call i32 asm "cmp $2,$1", "={@cca},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* %addr, i64 %nr)
+  //CHECK: = tail call i32 asm "cmp $2,$1", "={@cca},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* elementtype(i64) %addr, i64 %nr)
   int x;
   asm("cmp %2,%1"
       : "=@cca"(x), "=m"(*(volatile long *)(addr))
@@ -15,7 +15,7 @@
 
 int test_ccae(long nr, volatile long *addr) {
   //CHECK-LABEL: @test_ccae
-  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccae},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* %addr, i64 %nr)
+  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccae},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* elementtype(i64) %addr, i64 %nr)
   int x;
   asm("cmp %2,%1"
       : "=@ccae"(x), "=m"(*(volatile long *)(addr))
@@ -28,7 +28,7 @@
 
 int test_ccb(long nr, volatile long *addr) {
   //CHECK-LABEL: @test_ccb
-  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccb},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* %addr, i64 %nr)
+  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccb},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* elementtype(i64) %addr, i64 %nr)
   int x;
   asm("cmp %2,%1"
       : "=@ccb"(x), "=m"(*(volatile long *)(addr))
@@ -41,7 +41,7 @@
 
 int test_ccbe(long nr, volatile long *addr) {
   //CHECK-LABEL: @test_ccbe
-  //CHECK: tail call i32 asm "cmp $2,$1", "={@ccbe},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* %addr, i64 %nr)
+  //CHECK: tail call i32 asm "cmp $2,$1", "={@ccbe},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* elementtype(i64) %addr, i64 %nr)
   int x;
   asm("cmp %2,%1"
       : "=@ccbe"(x), "=m"(*(volatile long *)(addr))
@@ -54,7 +54,7 @@
 
 int test_ccc(long nr, volatile long *addr) {
   //CHECK-LABEL: @test_ccc
-  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccc},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* %addr, i64 %nr)
+  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccc},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* elementtype(i64) %addr, i64 %nr)
   int x;
   asm("cmp %2,%1"
       : "=@ccc"(x), "=m"(*(volatile long *)(addr))
@@ -67,7 +67,7 @@
 
 int test_cce(long nr, volatile long *addr) {
   //CHECK-LABEL: @test_cce
-  //CHECK: = tail call i32 asm "cmp $2,$1", "={@cce},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* %addr, i64 %nr)
+  //CHECK: = tail call i32 asm "cmp $2,$1", "={@cce},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* elementtype(i64) %addr, i64 %nr)
   int x;
   asm("cmp %2,%1"
       : "=@cce"(x), "=m"(*(volatile long *)(addr))
@@ -80,7 +80,7 @@
 
 int test_ccz(long nr, volatile long *addr) {
   //CHECK-LABEL: @test_ccz
-  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccz},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* %addr, i64 %nr)
+  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccz},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* elementtype(i64) %addr, i64 %nr)
   int x;
   asm("cmp %2,%1"
       : "=@ccz"(x), "=m"(*(volatile long *)(addr))
@@ -93,7 +93,7 @@
 
 int test_ccg(long nr, volatile long *addr) {
   //CHECK-LABEL: @test_ccg
-  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccg},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* %addr, i64 %nr)
+  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccg},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* elementtype(i64) %addr, i64 %nr)
   int x;
   asm("cmp %2,%1"
       : "=@ccg"(x), "=m"(*(volatile long *)(addr))
@@ -106,7 +106,7 @@
 
 int test_ccge(long nr, volatile long *addr) {
   //CHECK-LABEL: @test_ccge
-  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccge},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* %addr, i64 %nr)
+  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccge},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* elementtype(i64) %addr, i64 %nr)
   int x;
   asm("cmp %2,%1"
       : "=@ccge"(x), "=m"(*(volatile long *)(addr))
@@ -119,7 +119,7 @@
 
 int test_ccl(long nr, volatile long *addr) {
   //CHECK-LABEL: @test_ccl
-  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccl},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* %addr, i64 %nr)
+  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccl},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* elementtype(i64) %addr, i64 %nr)
   int x;
   asm("cmp %2,%1"
       : "=@ccl"(x), "=m"(*(volatile long *)(addr))
@@ -132,7 +132,7 @@
 
 int test_ccle(long nr, volatile long *addr) {
   //CHECK-LABEL: @test_ccle
-  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccle},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* %addr, i64 %nr)
+  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccle},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* elementtype(i64) %addr, i64 %nr)
   int x;
   asm("cmp %2,%1"
       : "=@ccle"(x), "=m"(*(volatile long *)(addr))
@@ -145,7 +145,7 @@
 
 int test_ccna(long nr, volatile long *addr) {
   //CHECK-LABEL: @test_ccna
-  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccna},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* %addr, i64 %nr)
+  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccna},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* elementtype(i64) %addr, i64 %nr)
   int x;
   asm("cmp %2,%1"
       : "=@ccna"(x), "=m"(*(volatile long *)(addr))
@@ -158,7 +158,7 @@
 
 int test_ccnae(long nr, volatile long *addr) {
   //CHECK-LABEL: @test_ccnae
-  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccnae},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* %addr, i64 %nr)
+  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccnae},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* elementtype(i64) %addr, i64 %nr)
   int x;
   asm("cmp %2,%1"
       : "=@ccnae"(x), "=m"(*(volatile long *)(addr))
@@ -171,7 +171,7 @@
 
 int test_ccnb(long nr, volatile long *addr) {
   //CHECK-LABEL: @test_ccnb
-  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccnb},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* %addr, i64 %nr)
+  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccnb},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* elementtype(i64) %addr, i64 %nr)
   int x;
   asm("cmp %2,%1"
       : "=@ccnb"(x), "=m"(*(volatile long *)(addr))
@@ -184,7 +184,7 @@
 
 int test_ccnbe(long nr, volatile long *addr) {
   //CHECK-LABEL: @test_ccnbe
-  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccnbe},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* %addr, i64 %nr)
+  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccnbe},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* elementtype(i64) %addr, i64 %nr)
   int x;
   asm("cmp %2,%1"
       : "=@ccnbe"(x), "=m"(*(volatile long *)(addr))
@@ -197,7 +197,7 @@
 
 int test_ccnc(long nr, volatile long *addr) {
   //CHECK-LABEL: @test_ccnc
-  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccnc},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* %addr, i64 %nr)
+  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccnc},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* elementtype(i64) %addr, i64 %nr)
   int x;
   asm("cmp %2,%1"
       : "=@ccnc"(x), "=m"(*(volatile long *)(addr))
@@ -210,7 +210,7 @@
 
 int test_ccne(long nr, volatile long *addr) {
   //CHECK-LABEL: @test_ccne
-  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccne},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* %addr, i64 %nr)
+  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccne},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* elementtype(i64) %addr, i64 %nr)
   int x;
   asm("cmp %2,%1"
       : "=@ccne"(x), "=m"(*(volatile long *)(addr))
@@ -223,7 +223,7 @@
 
 int test_ccnz(long nr, volatile long *addr) {
   //CHECK-LABEL: @test_ccnz
-  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccnz},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* %addr, i64 %nr)
+  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccnz},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* elementtype(i64) %addr, i64 %nr)
   int x;
   asm("cmp %2,%1"
       : "=@ccnz"(x), "=m"(*(volatile long *)(addr))
@@ -236,7 +236,7 @@
 
 int test_ccng(long nr, volatile long *addr) {
   //CHECK-LABEL: @test_ccng
-  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccng},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* %addr, i64 %nr)
+  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccng},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* elementtype(i64) %addr, i64 %nr)
   int x;
   asm("cmp %2,%1"
       : "=@ccng"(x), "=m"(*(volatile long *)(addr))
@@ -249,7 +249,7 @@
 
 int test_ccnge(long nr, volatile long *addr) {
   //CHECK-LABEL: @test_ccnge
-  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccnge},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* %addr, i64 %nr)
+  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccnge},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* elementtype(i64) %addr, i64 %nr)
   int x;
   asm("cmp %2,%1"
       : "=@ccnge"(x), "=m"(*(volatile long *)(addr))
@@ -262,7 +262,7 @@
 
 int test_ccnl(long nr, volatile long *addr) {
   //CHECK-LABEL: @test_ccnl
-  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccnl},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* %addr, i64 %nr)
+  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccnl},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* elementtype(i64) %addr, i64 %nr)
   int x;
   asm("cmp %2,%1"
       : "=@ccnl"(x), "=m"(*(volatile long *)(addr))
@@ -275,7 +275,7 @@
 
 int test_ccnle(long nr, volatile long *addr) {
   //CHECK-LABEL: @test_ccnle
-  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccnle},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* %addr, i64 %nr)
+  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccnle},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* elementtype(i64) %addr, i64 %nr)
   int x;
   asm("cmp %2,%1"
       : "=@ccnle"(x), "=m"(*(volatile long *)(addr))
@@ -288,7 +288,7 @@
 
 int test_ccno(long nr, volatile long *addr) {
   //CHECK-LABEL: @test_ccno
-  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccno},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* %addr, i64 %nr)
+  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccno},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* elementtype(i64) %addr, i64 %nr)
   int x;
   asm("cmp %2,%1"
       : "=@ccno"(x), "=m"(*(volatile long *)(addr))
@@ -301,7 +301,7 @@
 
 int test_ccnp(long nr, volatile long *addr) {
   //CHECK-LABEL: @test_ccnp
-  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccnp},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* %addr, i64 %nr)
+  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccnp},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* elementtype(i64) %addr, i64 %nr)
   int x;
   asm("cmp %2,%1"
       : "=@ccnp"(x), "=m"(*(volatile long *)(addr))
@@ -314,7 +314,7 @@
 
 int test_ccns(long nr, volatile long *addr) {
   //CHECK-LABEL: @test_ccns
-  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccns},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* %addr, i64 %nr)
+  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccns},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* elementtype(i64) %addr, i64 %nr)
   int x;
   asm("cmp %2,%1"
       : "=@ccns"(x), "=m"(*(volatile long *)(addr))
@@ -327,7 +327,7 @@
 
 int test_cco(long nr, volatile long *addr) {
   //CHECK-LABEL: @test_cco
-  //CHECK: = tail call i32 asm "cmp $2,$1", "={@cco},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* %addr, i64 %nr)
+  //CHECK: = tail call i32 asm "cmp $2,$1", "={@cco},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* elementtype(i64) %addr, i64 %nr)
   int x;
   asm("cmp %2,%1"
       : "=@cco"(x), "=m"(*(volatile long *)(addr))
@@ -340,7 +340,7 @@
 
 int test_ccp(long nr, volatile long *addr) {
   //CHECK-LABEL: @test_ccp
-  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccp},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* %addr, i64 %nr)
+  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccp},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* elementtype(i64) %addr, i64 %nr)
   int x;
   asm("cmp %2,%1"
       : "=@ccp"(x), "=m"(*(volatile long *)(addr))
@@ -353,7 +353,7 @@
 
 int test_ccs(long nr, volatile long *addr) {
   //CHECK-LABEL: @test_ccs
-  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccs},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* %addr, i64 %nr)
+  //CHECK: = tail call i32 asm "cmp $2,$1", "={@ccs},=*m,r,~{cc},~{dirflag},~{fpsr},~{flags}"(i64* elementtype(i64) %addr, i64 %nr)
   int x;
   asm("cmp %2,%1"
       : "=@ccs"(x), "=m"(*(volatile long *)(addr))
Index: clang/test/CodeGen/matrix-type.c
===================================================================
--- clang/test/CodeGen/matrix-type.c
+++ clang/test/CodeGen/matrix-type.c
@@ -169,7 +169,7 @@
   // CHECK-NEXT:    [[PTR1:%.+]] = bitcast [16 x double]* [[ALLOCA]] to <16 x double>*
   // CHECK-NEXT:    [[PTR2:%.+]] = bitcast [16 x double]* [[ALLOCA]] to <16 x double>*
   // CHECK-NEXT:    [[VAL:%.+]] = load <16 x double>, <16 x double>* [[PTR2]], align 8
-  // CHECK-NEXT:    call void asm sideeffect "", "=*r|m,0,~{memory},~{dirflag},~{fpsr},~{flags}"(<16 x double>* [[PTR1]], <16 x double> [[VAL]])
+  // CHECK-NEXT:    call void asm sideeffect "", "=*r|m,0,~{memory},~{dirflag},~{fpsr},~{flags}"(<16 x double>* elementtype(<16 x double>) [[PTR1]], <16 x double> [[VAL]])
   // CHECK-NEXT:    ret void
 
   dx4x4_t m;
Index: clang/test/CodeGen/mips-constraints-mem.c
===================================================================
--- clang/test/CodeGen/mips-constraints-mem.c
+++ clang/test/CodeGen/mips-constraints-mem.c
@@ -9,7 +9,7 @@
  // 'R': An address that can be used in a non-macro load or stor'
  // This test will result in the higher and lower nibbles being
  // switched due to the lwl/lwr instruction pairs.
- // CHECK:   %{{[0-9]+}} = call i32 asm sideeffect  "lwl $0, 1 + $1\0A\09lwr $0, 2 + $1\0A\09", "=r,*R,~{$1}"(i32* %{{[0-9,a-f]+}}) #1,
+ // CHECK:   %{{[0-9]+}} = call i32 asm sideeffect  "lwl $0, 1 + $1\0A\09lwr $0, 2 + $1\0A\09", "=r,*R,~{$1}"(i32* elementtype(i32) %{{[0-9,a-f]+}}) #1,
 
   int c = 0xffbbccdd;
 
Index: clang/test/CodeGen/mips-inline-asm-modifiers.c
===================================================================
--- clang/test/CodeGen/mips-inline-asm-modifiers.c
+++ clang/test/CodeGen/mips-inline-asm-modifiers.c
@@ -7,8 +7,8 @@
 
 typedef int v4i32 __attribute__((vector_size(16)));
 
-  // CHECK: %{{[0-9]+}} = call i32 asm ".set noreorder;\0Alw    $0,$1;\0A.set reorder;\0A", "=r,*m,~{$1}"(i32* getelementptr inbounds ([8 x i32], [8 x i32]* @b, i32 {{[0-9]+}}, i32 {{[0-9]+}})) #2,
-  // CHECK: %{{[0-9]+}} = call i32 asm "lw    $0,${1:D};\0A", "=r,*m,~{$1}"(i32* getelementptr inbounds ([8 x i32], [8 x i32]* @b, i32 {{[0-9]+}}, i32 {{[0-9]+}})) #2,
+  // CHECK: %{{[0-9]+}} = call i32 asm ".set noreorder;\0Alw    $0,$1;\0A.set reorder;\0A", "=r,*m,~{$1}"(i32* elementtype(i32) getelementptr inbounds ([8 x i32], [8 x i32]* @b, i32 {{[0-9]+}}, i32 {{[0-9]+}})) #2,
+  // CHECK: %{{[0-9]+}} = call i32 asm "lw    $0,${1:D};\0A", "=r,*m,~{$1}"(i32* elementtype(i32) getelementptr inbounds ([8 x i32], [8 x i32]* @b, i32 {{[0-9]+}}, i32 {{[0-9]+}})) #2,
   // CHECK: %{{[0-9]+}} = call <4 x i32> asm "ldi.w ${0:w},1", "=f,~{$1}"
 int b[8] = {0,1,2,3,4,5,6,7};
 int  main()
Index: clang/test/CodeGen/mips-inline-asm.c
===================================================================
--- clang/test/CodeGen/mips-inline-asm.c
+++ clang/test/CodeGen/mips-inline-asm.c
@@ -5,17 +5,17 @@
 
 void m () {
   asm("lw $1, %0" :: "m"(data));
-  // CHECK: call void asm sideeffect "lw $$1, $0", "*m,~{$1}"(i32* @data)
+  // CHECK: call void asm sideeffect "lw $$1, $0", "*m,~{$1}"(i32* elementtype(i32) @data)
 }
 
 void ZC () {
   asm("ll $1, %0" :: "ZC"(data));
-  // CHECK: call void asm sideeffect "ll $$1, $0", "*^ZC,~{$1}"(i32* @data)
+  // CHECK: call void asm sideeffect "ll $$1, $0", "*^ZC,~{$1}"(i32* elementtype(i32) @data)
 }
 
 void R () {
   asm("lw $1, %0" :: "R"(data));
-  // CHECK: call void asm sideeffect "lw $$1, $0", "*R,~{$1}"(i32* @data)
+  // CHECK: call void asm sideeffect "lw $$1, $0", "*R,~{$1}"(i32* elementtype(i32) @data)
 }
 
 int additionalClobberedRegisters () {
Index: clang/test/CodeGen/mozilla-ms-inline-asm.c
===================================================================
--- clang/test/CodeGen/mozilla-ms-inline-asm.c
+++ clang/test/CodeGen/mozilla-ms-inline-asm.c
@@ -38,7 +38,7 @@
 // CHECK-SAME: pop ebp
 // CHECK-SAME: ret
 // CHECK: "=*m,*m,*m,*m,*m,~{eax},~{ebp},~{ecx},~{edx},~{flags},~{esp},~{dirflag},~{fpsr},~{flags}"
-// CHECK: (i8** %8, i32* %7, void (...)* bitcast (void ()* @invoke_copy_to_stack to void (...)*), i8** %5, i32* %6)
+// CHECK: (i8** elementtype(i8*) %8, i32* elementtype(i32) %7, void (...)* elementtype(void (...)) bitcast (void ()* @invoke_copy_to_stack to void (...)*), i8** elementtype(i8*) %5, i32* elementtype(i32) %6)
 // CHECK: ret void
     __asm {
         mov     edx,paramCount
Index: clang/test/CodeGen/ms-inline-asm-64.c
===================================================================
--- clang/test/CodeGen/ms-inline-asm-64.c
+++ clang/test/CodeGen/ms-inline-asm-64.c
@@ -36,7 +36,7 @@
 // CHECK-SAME: lea ebx, $0
 // CHECK-SAME: mov eax, [ebx]
 // CHECK-SAME: mov [ebx + $$4], ecx
-// CHECK-SAME: "*m,~{eax},~{ebx},~{dirflag},~{fpsr},~{flags}"(%struct.t3_type* %{{.*}})
+// CHECK-SAME: "*m,~{eax},~{ebx},~{dirflag},~{fpsr},~{flags}"(%struct.t3_type* elementtype(%struct.t3_type) %{{.*}})
 }
 
 int t4() {
@@ -56,7 +56,7 @@
 // CHECK-SAME: lea ebx, $0
 // CHECK-SAME: mov eax, [ebx]
 // CHECK-SAME: mov [ebx + $$4], ecx
-// CHECK-SAME: "*m,~{eax},~{ebx},~{dirflag},~{fpsr},~{flags}"(%struct.t3_type* %{{.*}})
+// CHECK-SAME: "*m,~{eax},~{ebx},~{dirflag},~{fpsr},~{flags}"(%struct.t3_type* elementtype(%struct.t3_type) %{{.*}})
 }
 
 void bar() {}
@@ -70,5 +70,5 @@
   // CHECK: call void asm sideeffect inteldialect
   // CHECK-SAME: call qword ptr ${0:P}
   // CHECK-SAME: jmp qword ptr ${1:P}
-  // CHECK-SAME: "*m,*m,~{dirflag},~{fpsr},~{flags}"(void (...)* bitcast (void ()* @bar to void (...)*), void (...)* bitcast (void ()* @bar to void (...)*))
+  // CHECK-SAME: "*m,*m,~{dirflag},~{fpsr},~{flags}"(void (...)* elementtype(void (...)) bitcast (void ()* @bar to void (...)*), void (...)* elementtype(void (...)) bitcast (void ()* @bar to void (...)*))
 }
Index: clang/test/CodeGen/ms-inline-asm-static-variable.c
===================================================================
--- clang/test/CodeGen/ms-inline-asm-static-variable.c
+++ clang/test/CodeGen/ms-inline-asm-static-variable.c
@@ -5,6 +5,6 @@
 static int arr[10];
 void t1() {
   // CHECK: @arr = internal global [10 x i32]
-  // CHECK: call void asm sideeffect inteldialect "mov dword ptr arr[edx * $$4],edx", "=*m,{{.*}}([10 x i32]* @arr)
+  // CHECK: call void asm sideeffect inteldialect "mov dword ptr arr[edx * $$4],edx", "=*m,{{.*}}([10 x i32]* elementtype([10 x i32]) @arr)
   __asm mov  dword ptr arr[edx*4],edx
 }
Index: clang/test/CodeGen/ms-inline-asm.c
===================================================================
--- clang/test/CodeGen/ms-inline-asm.c
+++ clang/test/CodeGen/ms-inline-asm.c
@@ -114,7 +114,7 @@
 // CHECK: call i32 asm sideeffect inteldialect
 // CHECK-SAME: mov eax, $2
 // CHECK-SAME: mov $0, eax
-// CHECK-SAME: "=*m,=&{eax},*m,~{dirflag},~{fpsr},~{flags}"(i32* %{{.*}}, i32* %{{.*}})
+// CHECK-SAME: "=*m,=&{eax},*m,~{dirflag},~{fpsr},~{flags}"(i32* elementtype(i32) %{{.*}}, i32* elementtype(i32) %{{.*}})
 // CHECK: [[RET:%[a-zA-Z0-9]+]] = load i32, i32* [[J]], align 4
 // CHECK: ret i32 [[RET]]
 }
@@ -140,7 +140,7 @@
 // CHECK-SAME: mov $0, eax
 // CHECK-SAME: mov eax, $4
 // CHECK-SAME: mov $1, eax
-// CHECK-SAME: "=*m,=*m,=&{eax},*m,*m,~{dirflag},~{fpsr},~{flags}"(i32* %{{.*}}, i32* %{{.*}}, i32* %{{.*}}, i32* %{{.*}})
+// CHECK-SAME: "=*m,=*m,=&{eax},*m,*m,~{dirflag},~{fpsr},~{flags}"(i32* elementtype(i32) %{{.*}}, i32* elementtype(i32) %{{.*}}, i32* elementtype(i32) %{{.*}}, i32* elementtype(i32) %{{.*}})
 }
 
 void t13() {
@@ -152,7 +152,7 @@
 // CHECK: call void asm sideeffect inteldialect
 // CHECK-SAME: movzx eax, byte ptr $0
 // CHECK-SAME: movzx eax, word ptr $1
-// CHECK-SAME: "*m,*m,~{eax},~{dirflag},~{fpsr},~{flags}"(i8* %{{.*}}i, i16* %{{.*}}j)
+// CHECK-SAME: "*m,*m,~{eax},~{dirflag},~{fpsr},~{flags}"(i8* elementtype(i8) %{{.*}}i, i16* elementtype(i16) %{{.*}}j)
 }
 
 void t13_brac() {
@@ -164,7 +164,7 @@
 // CHECK: call void asm sideeffect inteldialect
 // CHECK-SAME: movzx eax, byte ptr $0
 // CHECK-SAME: movzx eax, word ptr $1
-// CHECK-SAME: "*m,*m,~{eax},~{dirflag},~{fpsr},~{flags}"(i8* %{{.*}}i, i16* %{{.*}}j)
+// CHECK-SAME: "*m,*m,~{eax},~{dirflag},~{fpsr},~{flags}"(i8* elementtype(i8) %{{.*}}i, i16* elementtype(i16) %{{.*}}j)
 }
 
 void t14() {
@@ -177,7 +177,7 @@
     .endif
   }
 // CHECK: t14
-// CHECK: call void asm sideeffect inteldialect ".if 1\0A\09mov eax, $0\0A\09.else\0A\09mov ebx, j\0A\09.endif", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"(i32* %{{.*}})
+// CHECK: call void asm sideeffect inteldialect ".if 1\0A\09mov eax, $0\0A\09.else\0A\09mov ebx, j\0A\09.endif", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"(i32* elementtype(i32) %{{.*}})
 }
 
 int gvar = 10;
@@ -196,7 +196,7 @@
 // CHECK: mov eax, $4 + $$1
   __asm mov eax, 1+offset gvar+1 ; eax = 2 + address of gvar
 // CHECK: mov eax, $5 + $$2
-// CHECK: "*m,r,i,i,i,i,~{eax},~{dirflag},~{fpsr},~{flags}"(i32* %{{.*}}, i32* %{{.*}}, i32* @{{.*}}, i32* @{{.*}}, i32* @{{.*}}, i32* @{{.*}})
+// CHECK: "*m,r,i,i,i,i,~{eax},~{dirflag},~{fpsr},~{flags}"(i32* elementtype(i32) %{{.*}}, i32* %{{.*}}, i32* @{{.*}}, i32* @{{.*}}, i32* @{{.*}}, i32* @{{.*}})
 }
 
 void t16() {
@@ -312,7 +312,7 @@
 void t24() {
   __asm call t24_helper
 // CHECK: t24
-// CHECK: call void asm sideeffect inteldialect "call dword ptr ${0:P}", "*m,~{dirflag},~{fpsr},~{flags}"(void ()* @t24_helper)
+// CHECK: call void asm sideeffect inteldialect "call dword ptr ${0:P}", "*m,~{dirflag},~{fpsr},~{flags}"(void ()* elementtype(void ()) @t24_helper)
 }
 
 void t25() {
@@ -376,7 +376,7 @@
 // CHECK: mov dword ptr $1, $$8
   __asm mov otype, TYPE arr
 // CHECK: mov dword ptr $2, $$4
-// CHECK: "=*m,=*m,=*m,~{dirflag},~{fpsr},~{flags}"(i32* %{{.*}}, i32* %{{.*}}, i32* %{{.*}})
+// CHECK: "=*m,=*m,=*m,~{dirflag},~{fpsr},~{flags}"(i32* elementtype(i32) %{{.*}}, i32* elementtype(i32) %{{.*}}, i32* elementtype(i32) %{{.*}})
 }
 
 int results[2] = {13, 37};
@@ -389,7 +389,7 @@
   __asm mov res, edi
 // CHECK: mov $0, edi
   return res;
-// CHECK: "=*m,={eax},*m,~{edi},~{dirflag},~{fpsr},~{flags}"(i32** %{{.*}}, [2 x i32]* @{{.*}})
+// CHECK: "=*m,={eax},*m,~{edi},~{dirflag},~{fpsr},~{flags}"(i32** elementtype(i32*) %{{.*}}, [2 x i32]* elementtype([2 x i32]) @{{.*}})
 }
 
 void t31() {
@@ -412,7 +412,7 @@
 // CHECK: mov ax, word ptr $2
   __asm mov al, byte ptr i
 // CHECK: mov al, byte ptr $3
-// CHECK: "*m,*m,*m,*m,~{al},~{ax},~{eax},~{dirflag},~{fpsr},~{flags}"(i32* %{{.*}}, i32* %{{.*}}, i32* %{{.*}}, i32* %{{.*}})
+// CHECK: "*m,*m,*m,*m,~{al},~{ax},~{eax},~{dirflag},~{fpsr},~{flags}"(i32* elementtype(i32) %{{.*}}, i32* elementtype(i32) %{{.*}}, i32* elementtype(i32) %{{.*}}, i32* elementtype(i32) %{{.*}})
 }
 
 void t33() {
@@ -426,7 +426,7 @@
 // CHECK: mov ax, word ptr $2
   __asm mov al, byte ptr [i]
 // CHECK: mov al, byte ptr $3
-// CHECK: "*m,*m,*m,*m,~{al},~{ax},~{eax},~{dirflag},~{fpsr},~{flags}"(i32* %{{.*}}, i32* %{{.*}}, i32* %{{.*}}, i32* %{{.*}})
+// CHECK: "*m,*m,*m,*m,~{al},~{ax},~{eax},~{dirflag},~{fpsr},~{flags}"(i32* elementtype(i32) %{{.*}}, i32* elementtype(i32) %{{.*}}, i32* elementtype(i32) %{{.*}}, i32* elementtype(i32) %{{.*}})
 }
 
 void t34() {
@@ -452,31 +452,31 @@
   int arr[4];
   // Work around PR20368: These should be single line blocks
   __asm { mov eax, 4[arr] }
-// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$4]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* %{{.*}})
+// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$4]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* elementtype([4 x i32]) %{{.*}})
   __asm { mov eax, 4[arr + 4] }
-// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$8]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* %{{.*}})
+// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$8]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* elementtype([4 x i32]) %{{.*}})
   __asm { mov eax, 8[arr + 4 + 32*2 - 4] }
-// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$72]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* %{{.*}})
+// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$72]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* elementtype([4 x i32]) %{{.*}})
   __asm { mov eax, 12[4 + arr] }
-// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$16]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* %{{.*}})
+// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$16]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* elementtype([4 x i32]) %{{.*}})
   __asm { mov eax, 4[4 + arr + 4] }
-// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$12]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* %{{.*}})
+// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$12]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* elementtype([4 x i32]) %{{.*}})
   __asm { mov eax, 4[64 + arr + (2*32)] }
-// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$132]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* %{{.*}})
+// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$132]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* elementtype([4 x i32]) %{{.*}})
   __asm { mov eax, 4[64 + arr - 2*32] }
-// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$4]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* %{{.*}})
+// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$4]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* elementtype([4 x i32]) %{{.*}})
   __asm { mov eax, [arr + 4] }
-// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$4]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* %{{.*}})
+// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$4]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* elementtype([4 x i32]) %{{.*}})
   __asm { mov eax, [arr + 4 + 32*2 - 4] }
-// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$64]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* %{{.*}})
+// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$64]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* elementtype([4 x i32]) %{{.*}})
   __asm { mov eax, [4 + arr] }
-// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$4]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* %{{.*}})
+// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$4]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* elementtype([4 x i32]) %{{.*}})
   __asm { mov eax, [4 + arr + 4] }
-// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$8]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* %{{.*}})
+// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$8]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* elementtype([4 x i32]) %{{.*}})
   __asm { mov eax, [64 + arr + (2*32)] }
-// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$128]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* %{{.*}})
+// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$128]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* elementtype([4 x i32]) %{{.*}})
   __asm { mov eax, [64 + arr - 2*32] }
-// CHECK: call void asm sideeffect inteldialect "mov eax, $0", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* %{{.*}})
+// CHECK: call void asm sideeffect inteldialect "mov eax, $0", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* elementtype([4 x i32]) %{{.*}})
 }
 
 void t37() {
@@ -507,21 +507,21 @@
   int arr[4];
   // Work around PR20368: These should be single line blocks
   __asm { mov eax, 4+4[arr] }
-// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$8]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* %{{.*}})
+// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$8]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* elementtype([4 x i32]) %{{.*}})
   __asm { mov eax, (4+4)[arr + 4] }
-// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$12]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* %{{.*}})
+// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$12]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* elementtype([4 x i32]) %{{.*}})
   __asm { mov eax, 8*2[arr + 4 + 32*2 - 4] }
-// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$80]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* %{{.*}})
+// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$80]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* elementtype([4 x i32]) %{{.*}})
   __asm { mov eax, 12+20[4 + arr] }
-// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$36]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* %{{.*}})
+// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$36]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* elementtype([4 x i32]) %{{.*}})
   __asm { mov eax, 4*16+4[4 + arr + 4] }
-// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$76]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* %{{.*}})
+// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$76]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* elementtype([4 x i32]) %{{.*}})
   __asm { mov eax, 4*4[64 + arr + (2*32)] }
-// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$144]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* %{{.*}})
+// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$144]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* elementtype([4 x i32]) %{{.*}})
   __asm { mov eax, 4*(4-2)[64 + arr - 2*32] }
-// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$8]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* %{{.*}})
+// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$8]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* elementtype([4 x i32]) %{{.*}})
   __asm { mov eax, 32*(4-2)[arr - 2*32] }
-// CHECK: call void asm sideeffect inteldialect "mov eax, $0", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* %{{.*}})
+// CHECK: call void asm sideeffect inteldialect "mov eax, $0", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"([4 x i32]* elementtype([4 x i32]) %{{.*}})
 }
 
 void cpuid() {
@@ -577,7 +577,7 @@
 // CHECK: fld dword ptr $1
   __asm fistp i
 // CHECK: fistp dword ptr $0
-// CHECK: "=*m,*m,~{dirflag},~{fpsr},~{flags}"(i32* %{{.*}}, float* %{{.*}})
+// CHECK: "=*m,*m,~{dirflag},~{fpsr},~{flags}"(i32* elementtype(i32) %{{.*}}, float* elementtype(float) %{{.*}})
 }
 
 void t41(unsigned short a) {
@@ -602,7 +602,7 @@
   int flags;
   __asm mov flags, eax
 // CHECK: mov $0, eax
-// CHECK: "=*m,~{dirflag},~{fpsr},~{flags}"(i32* %flags)
+// CHECK: "=*m,~{dirflag},~{fpsr},~{flags}"(i32* elementtype(i32) %flags)
 }
 
 void t42b() {
@@ -610,7 +610,7 @@
   int mxcsr;
   __asm mov mxcsr, eax
 // CHECK: mov $0, eax
-// CHECK: "=*m,~{dirflag},~{fpsr},~{flags}"(i32* %mxcsr)
+// CHECK: "=*m,~{dirflag},~{fpsr},~{flags}"(i32* elementtype(i32) %mxcsr)
 }
 
 void t43() {
@@ -618,31 +618,31 @@
   C strct;
 // Work around PR20368: These should be single line blocks
  __asm { mov eax, 4[strct.c1] }
-// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$4]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"(i32* %{{.*}})
+// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$4]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"(i32* elementtype(i32) %{{.*}})
   __asm { mov eax, 4[strct.c3 + 4] }
-// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$8]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"(i32* %{{.*}})
+// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$8]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"(i32* elementtype(i32) %{{.*}})
   __asm { mov eax, 8[strct.c2.a + 4 + 32*2 - 4] }
-// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$72]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"(i32* %{{.*}})
+// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$72]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"(i32* elementtype(i32) %{{.*}})
   __asm { mov eax, 12[4 + strct.c2.b] }
-// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$16]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"(i32* %{{.*}})
+// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$16]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"(i32* elementtype(i32) %{{.*}})
   __asm { mov eax, 4[4 + strct.c4.b2.b + 4] }
-// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$12]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"(i32* %{{.*}})
+// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$12]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"(i32* elementtype(i32) %{{.*}})
   __asm { mov eax, 4[64 + strct.c1 + (2*32)] }
-// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$132]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"(i32* %{{.*}})
+// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$132]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"(i32* elementtype(i32) %{{.*}})
   __asm { mov eax, 4[64 + strct.c2.a - 2*32] }
-// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$4]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"(i32* %{{.*}})
+// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$4]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"(i32* elementtype(i32) %{{.*}})
   __asm { mov eax, [strct.c4.b1 + 4] }
-// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$4]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"(i32* %{{.*}})
+// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$4]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"(i32* elementtype(i32) %{{.*}})
   __asm { mov eax, [strct.c4.b2.a + 4 + 32*2 - 4] }
-// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$64]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"(i32* %{{.*}})
+// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$64]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"(i32* elementtype(i32) %{{.*}})
   __asm { mov eax, [4 + strct.c1] }
-// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$4]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"(i32* %{{.*}})
+// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$4]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"(i32* elementtype(i32) %{{.*}})
   __asm { mov eax, [4 + strct.c2.b + 4] }
-// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$8]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"(i32* %{{.*}})
+// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$8]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"(i32* elementtype(i32) %{{.*}})
   __asm { mov eax, [64 + strct.c3 + (2*32)] }
-// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$128]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"(i32* %{{.*}})
+// CHECK: call void asm sideeffect inteldialect "mov eax, $0[$$128]", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"(i32* elementtype(i32) %{{.*}})
   __asm { mov eax, [64 + strct.c4.b2.b - 2*32] }
-// CHECK: call void asm sideeffect inteldialect "mov eax, $0", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"(i32* %{{.*}})
+// CHECK: call void asm sideeffect inteldialect "mov eax, $0", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"(i32* elementtype(i32) %{{.*}})
 }
 
 void t44() {
@@ -684,7 +684,7 @@
 void call_clobber() {
   __asm call t41
   // CHECK-LABEL: define{{.*}} void @call_clobber
-  // CHECK: call void asm sideeffect inteldialect "call dword ptr ${0:P}", "*m,~{dirflag},~{fpsr},~{flags}"(void (i16)* @t41)
+  // CHECK: call void asm sideeffect inteldialect "call dword ptr ${0:P}", "*m,~{dirflag},~{fpsr},~{flags}"(void (i16)* elementtype(void (i16)) @t41)
 }
 
 void xgetbv() {
Index: clang/test/CodeGen/ms-inline-asm.cpp
===================================================================
--- clang/test/CodeGen/ms-inline-asm.cpp
+++ clang/test/CodeGen/ms-inline-asm.cpp
@@ -23,7 +23,7 @@
 // CHECK-SAME: mov eax, $2
 // CHECK-SAME: mov eax, dword ptr $3
 // CHECK-SAME: mov eax, dword ptr $4
-// CHECK-SAME: "*m,*m,*m,*m,*m,~{eax},~{dirflag},~{fpsr},~{flags}"(i32** @_ZN3Foo3ptrE, i32** @_ZN3Foo3Bar3ptrE, i32** @_ZN3Foo3ptrE, i32** @_ZN3Foo3ptrE, i32** @_ZN3Foo3ptrE)
+// CHECK-SAME: "*m,*m,*m,*m,*m,~{eax},~{dirflag},~{fpsr},~{flags}"(i32** elementtype(i32*) @_ZN3Foo3ptrE, i32** elementtype(i32*) @_ZN3Foo3Bar3ptrE, i32** elementtype(i32*) @_ZN3Foo3ptrE, i32** elementtype(i32*) @_ZN3Foo3ptrE, i32** elementtype(i32*) @_ZN3Foo3ptrE)
   __asm mov eax, Foo ::ptr
   __asm mov eax, Foo :: Bar :: ptr
   __asm mov eax, [Foo:: ptr]
@@ -92,7 +92,7 @@
 // CHECK: call void asm sideeffect inteldialect
 // CHECK-SAME: mov eax, $1
 // CHECK-SAME: mov $0, eax
-// CHECK-SAME: "=*m,*m,~{eax},~{dirflag},~{fpsr},~{flags}"(i32* @_ZN2T41yE, i32* {{.*}})
+// CHECK-SAME: "=*m,*m,~{eax},~{dirflag},~{fpsr},~{flags}"(i32* elementtype(i32) @_ZN2T41yE, i32* elementtype(i32) {{.*}})
 }
 
 template <class T> struct T5 {
@@ -111,7 +111,7 @@
   // CHECK-SAME: push $0
   // CHECK-SAME: call dword ptr ${2:P}
   // CHECK-SAME: mov $1, eax
-  // CHECK-SAME: "=*m,=*m,*m,~{esp},~{dirflag},~{fpsr},~{flags}"(i32* %y, i32* %x, i32 (float)* @_ZN2T5IiE6createIfEEiT_)
+  // CHECK-SAME: "=*m,=*m,*m,~{esp},~{dirflag},~{fpsr},~{flags}"(i32* elementtype(i32) %y, i32* elementtype(i32) %x, i32 (float)* elementtype(i32 (float)) @_ZN2T5IiE6createIfEEiT_)
 }
 
 // Just verify this doesn't emit an error.
Index: clang/test/CodeGen/ms-intrinsics.c
===================================================================
--- clang/test/CodeGen/ms-intrinsics.c
+++ clang/test/CodeGen/ms-intrinsics.c
@@ -626,48 +626,48 @@
 #if defined(__i386__) || defined(__x86_64__)
 long test_InterlockedExchange_HLEAcquire(long volatile *Target, long Value) {
 // CHECK-INTEL: define{{.*}} i32 @test_InterlockedExchange_HLEAcquire(i32*{{[a-z_ ]*}}%Target, i32{{[a-z_ ]*}}%Value)
-// CHECK-INTEL: call i32 asm sideeffect ".byte 0xf2 ; lock ; xchg $($0, $1$|$1, $0$)", "=r,=*m,0,*m,~{memory},~{dirflag},~{fpsr},~{flags}"(i32* %Target, i32 %Value, i32* %Target)
+// CHECK-INTEL: call i32 asm sideeffect ".byte 0xf2 ; lock ; xchg $($0, $1$|$1, $0$)", "=r,=*m,0,*m,~{memory},~{dirflag},~{fpsr},~{flags}"(i32* elementtype(i32) %Target, i32 %Value, i32* elementtype(i32) %Target)
   return _InterlockedExchange_HLEAcquire(Target, Value);
 }
 long test_InterlockedExchange_HLERelease(long volatile *Target, long Value) {
 // CHECK-INTEL: define{{.*}} i32 @test_InterlockedExchange_HLERelease(i32*{{[a-z_ ]*}}%Target, i32{{[a-z_ ]*}}%Value)
-// CHECK-INTEL: call i32 asm sideeffect ".byte 0xf3 ; lock ; xchg $($0, $1$|$1, $0$)", "=r,=*m,0,*m,~{memory},~{dirflag},~{fpsr},~{flags}"(i32* %Target, i32 %Value, i32* %Target)
+// CHECK-INTEL: call i32 asm sideeffect ".byte 0xf3 ; lock ; xchg $($0, $1$|$1, $0$)", "=r,=*m,0,*m,~{memory},~{dirflag},~{fpsr},~{flags}"(i32* elementtype(i32) %Target, i32 %Value, i32* elementtype(i32) %Target)
   return _InterlockedExchange_HLERelease(Target, Value);
 }
 long test_InterlockedCompareExchange_HLEAcquire(long volatile *Destination,
                                                 long Exchange, long Comparand) {
 // CHECK-INTEL: define{{.*}} i32 @test_InterlockedCompareExchange_HLEAcquire(i32*{{[a-z_ ]*}}%Destination, i32{{[a-z_ ]*}}%Exchange, i32{{[a-z_ ]*}}%Comparand)
-// CHECK-INTEL: call i32 asm sideeffect ".byte 0xf2 ; lock ; cmpxchg $($2, $1$|$1, $2$)", "={ax},=*m,r,0,*m,~{memory},~{dirflag},~{fpsr},~{flags}"(i32* %Destination, i32 %Exchange, i32 %Comparand, i32* %Destination)
+// CHECK-INTEL: call i32 asm sideeffect ".byte 0xf2 ; lock ; cmpxchg $($2, $1$|$1, $2$)", "={ax},=*m,r,0,*m,~{memory},~{dirflag},~{fpsr},~{flags}"(i32* elementtype(i32) %Destination, i32 %Exchange, i32 %Comparand, i32* elementtype(i32) %Destination)
   return _InterlockedCompareExchange_HLEAcquire(Destination, Exchange, Comparand);
 }
 long test_InterlockedCompareExchange_HLERelease(long volatile *Destination,
                                             long Exchange, long Comparand) {
 // CHECK-INTEL: define{{.*}} i32 @test_InterlockedCompareExchange_HLERelease(i32*{{[a-z_ ]*}}%Destination, i32{{[a-z_ ]*}}%Exchange, i32{{[a-z_ ]*}}%Comparand)
-// CHECK-INTEL: call i32 asm sideeffect ".byte 0xf3 ; lock ; cmpxchg $($2, $1$|$1, $2$)", "={ax},=*m,r,0,*m,~{memory},~{dirflag},~{fpsr},~{flags}"(i32* %Destination, i32 %Exchange, i32 %Comparand, i32* %Destination)
+// CHECK-INTEL: call i32 asm sideeffect ".byte 0xf3 ; lock ; cmpxchg $($2, $1$|$1, $2$)", "={ax},=*m,r,0,*m,~{memory},~{dirflag},~{fpsr},~{flags}"(i32* elementtype(i32) %Destination, i32 %Exchange, i32 %Comparand, i32* elementtype(i32) %Destination)
   return _InterlockedCompareExchange_HLERelease(Destination, Exchange, Comparand);
 }
 #endif
 #if defined(__x86_64__)
 __int64 test_InterlockedExchange64_HLEAcquire(__int64 volatile *Target, __int64 Value) {
 // CHECK-X64: define{{.*}} i64 @test_InterlockedExchange64_HLEAcquire(i64*{{[a-z_ ]*}}%Target, i64{{[a-z_ ]*}}%Value)
-// CHECK-X64: call i64 asm sideeffect ".byte 0xf2 ; lock ; xchg $($0, $1$|$1, $0$)", "=r,=*m,0,*m,~{memory},~{dirflag},~{fpsr},~{flags}"(i64* %Target, i64 %Value, i64* %Target)
+// CHECK-X64: call i64 asm sideeffect ".byte 0xf2 ; lock ; xchg $($0, $1$|$1, $0$)", "=r,=*m,0,*m,~{memory},~{dirflag},~{fpsr},~{flags}"(i64* elementtype(i64) %Target, i64 %Value, i64* elementtype(i64) %Target)
   return _InterlockedExchange64_HLEAcquire(Target, Value);
 }
 __int64 test_InterlockedExchange64_HLERelease(__int64 volatile *Target, __int64 Value) {
 // CHECK-X64: define{{.*}} i64 @test_InterlockedExchange64_HLERelease(i64*{{[a-z_ ]*}}%Target, i64{{[a-z_ ]*}}%Value)
-// CHECK-X64: call i64 asm sideeffect ".byte 0xf3 ; lock ; xchg $($0, $1$|$1, $0$)", "=r,=*m,0,*m,~{memory},~{dirflag},~{fpsr},~{flags}"(i64* %Target, i64 %Value, i64* %Target)
+// CHECK-X64: call i64 asm sideeffect ".byte 0xf3 ; lock ; xchg $($0, $1$|$1, $0$)", "=r,=*m,0,*m,~{memory},~{dirflag},~{fpsr},~{flags}"(i64* elementtype(i64) %Target, i64 %Value, i64* elementtype(i64) %Target)
   return _InterlockedExchange64_HLERelease(Target, Value);
 }
 __int64 test_InterlockedCompareExchange64_HLEAcquire(__int64 volatile *Destination,
                                                      __int64 Exchange, __int64 Comparand) {
 // CHECK-X64: define{{.*}} i64 @test_InterlockedCompareExchange64_HLEAcquire(i64*{{[a-z_ ]*}}%Destination, i64{{[a-z_ ]*}}%Exchange, i64{{[a-z_ ]*}}%Comparand)
-// CHECK-X64: call i64 asm sideeffect ".byte 0xf2 ; lock ; cmpxchg $($2, $1$|$1, $2$)", "={ax},=*m,r,0,*m,~{memory},~{dirflag},~{fpsr},~{flags}"(i64* %Destination, i64 %Exchange, i64 %Comparand, i64* %Destination)
+// CHECK-X64: call i64 asm sideeffect ".byte 0xf2 ; lock ; cmpxchg $($2, $1$|$1, $2$)", "={ax},=*m,r,0,*m,~{memory},~{dirflag},~{fpsr},~{flags}"(i64* elementtype(i64) %Destination, i64 %Exchange, i64 %Comparand, i64* elementtype(i64) %Destination)
   return _InterlockedCompareExchange64_HLEAcquire(Destination, Exchange, Comparand);
 }
 __int64 test_InterlockedCompareExchange64_HLERelease(__int64 volatile *Destination,
                                                      __int64 Exchange, __int64 Comparand) {
 // CHECK-X64: define{{.*}} i64 @test_InterlockedCompareExchange64_HLERelease(i64*{{[a-z_ ]*}}%Destination, i64{{[a-z_ ]*}}%Exchange, i64{{[a-z_ ]*}}%Comparand)
-// CHECK-X64: call i64 asm sideeffect ".byte 0xf3 ; lock ; cmpxchg $($2, $1$|$1, $2$)", "={ax},=*m,r,0,*m,~{memory},~{dirflag},~{fpsr},~{flags}"(i64* %Destination, i64 %Exchange, i64 %Comparand, i64* %Destination)
+// CHECK-X64: call i64 asm sideeffect ".byte 0xf3 ; lock ; cmpxchg $($2, $1$|$1, $2$)", "={ax},=*m,r,0,*m,~{memory},~{dirflag},~{fpsr},~{flags}"(i64* elementtype(i64) %Destination, i64 %Exchange, i64 %Comparand, i64* elementtype(i64) %Destination)
   return _InterlockedCompareExchange64_HLERelease(Destination, Exchange, Comparand);
 }
 #endif
Index: clang/test/CodeGen/mult-alt-generic.c
===================================================================
--- clang/test/CodeGen/mult-alt-generic.c
+++ clang/test/CodeGen/mult-alt-generic.c
@@ -17,7 +17,7 @@
 // CHECK: @single_m
 void single_m()
 {
-  // CHECK: call void asm "foo $1,$0", "=*m,*m[[CLOBBERS:[a-zA-Z0-9@%{},~_$ ]*\"]](i32* {{[a-zA-Z0-9@%]+}}, i32* {{[a-zA-Z0-9@%]+}})
+  // CHECK: call void asm "foo $1,$0", "=*m,*m[[CLOBBERS:[a-zA-Z0-9@%{},~_$ ]*\"]](i32* elementtype(i32) {{[a-zA-Z0-9@%]+}}, i32* elementtype(i32) {{[a-zA-Z0-9@%]+}})
   asm("foo %1,%0" : "=m" (mout0) : "m" (min1));
 }
 
@@ -150,7 +150,7 @@
 // CHECK: @multi_m
 void multi_m()
 {
-  // CHECK: call void asm "foo $1,$0", "=*m|r,m|r[[CLOBBERS]](i32* {{[a-zA-Z0-9@%]+}}, i32 {{[a-zA-Z0-9@%]+}})
+  // CHECK: call void asm "foo $1,$0", "=*m|r,m|r[[CLOBBERS]](i32* elementtype(i32) {{[a-zA-Z0-9@%]+}}, i32 {{[a-zA-Z0-9@%]+}})
   asm("foo %1,%0" : "=m,r" (mout0) : "m,r" (min1));
 }
 
Index: clang/test/CodeGen/mult-alt-x86.c
===================================================================
--- clang/test/CodeGen/mult-alt-x86.c
+++ clang/test/CodeGen/mult-alt-x86.c
@@ -123,144 +123,144 @@
 // CHECK: @single_I
 void single_I()
 {
-  // CHECK: asm "foo $1,$0", "=*m,I[[CLOBBERS]](i32* @mout0, i32 1)
+  // CHECK: asm "foo $1,$0", "=*m,I[[CLOBBERS]](i32* elementtype(i32) @mout0, i32 1)
   asm("foo %1,%0" : "=m" (mout0) : "I" (1));
 }
 
 // CHECK: @single_J
 void single_J()
 {
-  // CHECK: asm "foo $1,$0", "=*m,J[[CLOBBERS]](i32* @mout0, i32 1)
+  // CHECK: asm "foo $1,$0", "=*m,J[[CLOBBERS]](i32* elementtype(i32) @mout0, i32 1)
   asm("foo %1,%0" : "=m" (mout0) : "J" (1));
 }
 
 // CHECK: @single_K
 void single_K()
 {
-  // CHECK: asm "foo $1,$0", "=*m,K[[CLOBBERS]](i32* @mout0, i32 1)
+  // CHECK: asm "foo $1,$0", "=*m,K[[CLOBBERS]](i32* elementtype(i32) @mout0, i32 1)
   asm("foo %1,%0" : "=m" (mout0) : "K" (1));
 }
 
 // CHECK: @single_L
 void single_L()
 {
-  // CHECK: asm "foo $1,$0", "=*m,L[[CLOBBERS]](i32* @mout0, i32 255)
+  // CHECK: asm "foo $1,$0", "=*m,L[[CLOBBERS]](i32* elementtype(i32) @mout0, i32 255)
   asm("foo %1,%0" : "=m" (mout0) : "L" (0xff));
-  // CHECK: asm "foo $1,$0", "=*m,L[[CLOBBERS]](i32* @mout0, i32 65535)
+  // CHECK: asm "foo $1,$0", "=*m,L[[CLOBBERS]](i32* elementtype(i32) @mout0, i32 65535)
   asm("foo %1,%0" : "=m" (mout0) : "L" (0xffff));
-  // CHECK: asm "foo $1,$0", "=*m,L[[CLOBBERS]](i32* @mout0, i32 -1)
+  // CHECK: asm "foo $1,$0", "=*m,L[[CLOBBERS]](i32* elementtype(i32) @mout0, i32 -1)
   asm("foo %1,%0" : "=m" (mout0) : "L" (0xffffffff));
 }
 
 // CHECK: @single_M
 void single_M()
 {
-  // CHECK: asm "foo $1,$0", "=*m,M[[CLOBBERS]](i32* @mout0, i32 1)
+  // CHECK: asm "foo $1,$0", "=*m,M[[CLOBBERS]](i32* elementtype(i32) @mout0, i32 1)
   asm("foo %1,%0" : "=m" (mout0) : "M" (1));
 }
 
 // CHECK: @single_N
 void single_N()
 {
-  // CHECK: asm "foo $1,$0", "=*m,N[[CLOBBERS]](i32* @mout0, i32 1)
+  // CHECK: asm "foo $1,$0", "=*m,N[[CLOBBERS]](i32* elementtype(i32) @mout0, i32 1)
   asm("foo %1,%0" : "=m" (mout0) : "N" (1));
 }
 
 // CHECK: @single_G
 void single_G()
 {
-  // CHECK: asm "foo $1,$0", "=*m,G[[CLOBBERS]](i32* @mout0, double {{1.[0]+e[+]*[0]+}})
+  // CHECK: asm "foo $1,$0", "=*m,G[[CLOBBERS]](i32* elementtype(i32) @mout0, double {{1.[0]+e[+]*[0]+}})
   asm("foo %1,%0" : "=m" (mout0) : "G" (1.0));
 }
 
 // CHECK: @single_C
 void single_C()
 {
-  // CHECK: asm "foo $1,$0", "=*m,C[[CLOBBERS]](i32* @mout0, double {{1.[0]+e[+]*[0]+}})
+  // CHECK: asm "foo $1,$0", "=*m,C[[CLOBBERS]](i32* elementtype(i32) @mout0, double {{1.[0]+e[+]*[0]+}})
   asm("foo %1,%0" : "=m" (mout0) : "C" (1.0));
 }
 
 // CHECK: @single_e
 void single_e()
 {
-  // CHECK: asm "foo $1,$0", "=*m,e[[CLOBBERS]](i32* @mout0, i32 1)
+  // CHECK: asm "foo $1,$0", "=*m,e[[CLOBBERS]](i32* elementtype(i32) @mout0, i32 1)
   asm("foo %1,%0" : "=m" (mout0) : "e" (1));
 }
 
 // CHECK: @single_Z
 void single_Z()
 {
-  // CHECK: asm "foo $1,$0", "=*m,Z[[CLOBBERS]](i32* @mout0, i32 1)
+  // CHECK: asm "foo $1,$0", "=*m,Z[[CLOBBERS]](i32* elementtype(i32) @mout0, i32 1)
   asm("foo %1,%0" : "=m" (mout0) : "Z" (1));
 }
 
 // CHECK: @multi_R
 void multi_R()
 {
-  // CHECK: asm "foo $1,$0", "=*r|R|m,r|R|m[[CLOBBERS]](i32* @mout0, i32 {{[a-zA-Z0-9@%]+}})
+  // CHECK: asm "foo $1,$0", "=*r|R|m,r|R|m[[CLOBBERS]](i32* elementtype(i32) @mout0, i32 {{[a-zA-Z0-9@%]+}})
   asm("foo %1,%0" : "=r,R,m" (mout0) : "r,R,m" (min1));
 }
 
 // CHECK: @multi_q
 void multi_q()
 {
-  // CHECK: asm "foo $1,$0", "=*r|q|m,r|q|m[[CLOBBERS]](i32* @mout0, i32 {{[a-zA-Z0-9@%]+}})
+  // CHECK: asm "foo $1,$0", "=*r|q|m,r|q|m[[CLOBBERS]](i32* elementtype(i32) @mout0, i32 {{[a-zA-Z0-9@%]+}})
   asm("foo %1,%0" : "=r,q,m" (mout0) : "r,q,m" (min1));
 }
 
 // CHECK: @multi_Q
 void multi_Q()
 {
-  // CHECK: asm "foo $1,$0", "=*r|Q|m,r|Q|m[[CLOBBERS]](i32* @mout0, i32 {{[a-zA-Z0-9@%]+}})
+  // CHECK: asm "foo $1,$0", "=*r|Q|m,r|Q|m[[CLOBBERS]](i32* elementtype(i32) @mout0, i32 {{[a-zA-Z0-9@%]+}})
   asm("foo %1,%0" : "=r,Q,m" (mout0) : "r,Q,m" (min1));
 }
 
 // CHECK: @multi_a
 void multi_a()
 {
-  // CHECK: asm "foo $1,$0", "=*r|{ax}|m,r|{ax}|m[[CLOBBERS]](i32* @mout0, i32 {{[a-zA-Z0-9@%]+}})
+  // CHECK: asm "foo $1,$0", "=*r|{ax}|m,r|{ax}|m[[CLOBBERS]](i32* elementtype(i32) @mout0, i32 {{[a-zA-Z0-9@%]+}})
   asm("foo %1,%0" : "=r,a,m" (mout0) : "r,a,m" (min1));
 }
 
 // CHECK: @multi_b
 void multi_b()
 {
-  // CHECK: asm "foo $1,$0", "=*r|{bx}|m,r|{bx}|m[[CLOBBERS]](i32* @mout0, i32 {{[a-zA-Z0-9@%]+}})
+  // CHECK: asm "foo $1,$0", "=*r|{bx}|m,r|{bx}|m[[CLOBBERS]](i32* elementtype(i32) @mout0, i32 {{[a-zA-Z0-9@%]+}})
   asm("foo %1,%0" : "=r,b,m" (mout0) : "r,b,m" (min1));
 }
 
 // CHECK: @multi_c
 void multi_c()
 {
-  // CHECK: asm "foo $1,$0", "=*r|{cx}|m,r|{cx}|m[[CLOBBERS]](i32* @mout0, i32 {{[a-zA-Z0-9@%]+}})
+  // CHECK: asm "foo $1,$0", "=*r|{cx}|m,r|{cx}|m[[CLOBBERS]](i32* elementtype(i32) @mout0, i32 {{[a-zA-Z0-9@%]+}})
   asm("foo %1,%0" : "=r,c,m" (mout0) : "r,c,m" (min1));
 }
 
 // CHECK: @multi_d
 void multi_d()
 {
-  // CHECK: asm "foo $1,$0", "=*r|{dx}|m,r|{dx}|m[[CLOBBERS]](i32* @mout0, i32 {{[a-zA-Z0-9@%]+}})
+  // CHECK: asm "foo $1,$0", "=*r|{dx}|m,r|{dx}|m[[CLOBBERS]](i32* elementtype(i32) @mout0, i32 {{[a-zA-Z0-9@%]+}})
   asm("foo %1,%0" : "=r,d,m" (mout0) : "r,d,m" (min1));
 }
 
 // CHECK: @multi_S
 void multi_S()
 {
-  // CHECK: asm "foo $1,$0", "=*r|{si}|m,r|{si}|m[[CLOBBERS]](i32* @mout0, i32 {{[a-zA-Z0-9@%]+}})
+  // CHECK: asm "foo $1,$0", "=*r|{si}|m,r|{si}|m[[CLOBBERS]](i32* elementtype(i32) @mout0, i32 {{[a-zA-Z0-9@%]+}})
   asm("foo %1,%0" : "=r,S,m" (mout0) : "r,S,m" (min1));
 }
 
 // CHECK: @multi_D
 void multi_D()
 {
-  // CHECK: asm "foo $1,$0", "=*r|{di}|m,r|{di}|m[[CLOBBERS]](i32* @mout0, i32 {{[a-zA-Z0-9@%]+}})
+  // CHECK: asm "foo $1,$0", "=*r|{di}|m,r|{di}|m[[CLOBBERS]](i32* elementtype(i32) @mout0, i32 {{[a-zA-Z0-9@%]+}})
   asm("foo %1,%0" : "=r,D,m" (mout0) : "r,D,m" (min1));
 }
 
 // CHECK: @multi_A
 void multi_A()
 {
-  // CHECK: asm "foo $1,$0", "=*r|A|m,r|A|m[[CLOBBERS]](i32* @mout0, i32 {{[a-zA-Z0-9@%]+}})
+  // CHECK: asm "foo $1,$0", "=*r|A|m,r|A|m[[CLOBBERS]](i32* elementtype(i32) @mout0, i32 {{[a-zA-Z0-9@%]+}})
   asm("foo %1,%0" : "=r,A,m" (mout0) : "r,A,m" (min1));
 }
 
@@ -285,14 +285,14 @@
 // CHECK: @multi_y
 void multi_y()
 {
-  // CHECK: asm "foo $1,$0", "=*r|y|m,r|y|m[[CLOBBERS]](double* @dout0, double {{[a-zA-Z0-9@%]+}})
+  // CHECK: asm "foo $1,$0", "=*r|y|m,r|y|m[[CLOBBERS]](double* elementtype(double) @dout0, double {{[a-zA-Z0-9@%]+}})
   asm("foo %1,%0" : "=r,y,m" (dout0) : "r,y,m" (din1));
 }
 
 // CHECK: @multi_x
 void multi_x()
 {
-  // CHECK: asm "foo $1,$0", "=*r|x|m,r|x|m[[CLOBBERS]](double* @dout0, double {{[a-zA-Z0-9@%]+}})
+  // CHECK: asm "foo $1,$0", "=*r|x|m,r|x|m[[CLOBBERS]](double* elementtype(double) @dout0, double {{[a-zA-Z0-9@%]+}})
   asm("foo %1,%0" : "=r,x,m" (dout0) : "r,x,m" (din1));
 }
 
@@ -310,69 +310,69 @@
 // CHECK: @multi_I
 void multi_I()
 {
-  // CHECK: asm "foo $1,$0", "=*r|m|m,r|I|m[[CLOBBERS]](i32* @mout0, i32 1)
+  // CHECK: asm "foo $1,$0", "=*r|m|m,r|I|m[[CLOBBERS]](i32* elementtype(i32) @mout0, i32 1)
   asm("foo %1,%0" : "=r,m,m" (mout0) : "r,I,m" (1));
 }
 
 // CHECK: @multi_J
 void multi_J()
 {
-  // CHECK: asm "foo $1,$0", "=*r|m|m,r|J|m[[CLOBBERS]](i32* @mout0, i32 1)
+  // CHECK: asm "foo $1,$0", "=*r|m|m,r|J|m[[CLOBBERS]](i32* elementtype(i32) @mout0, i32 1)
   asm("foo %1,%0" : "=r,m,m" (mout0) : "r,J,m" (1));
 }
 
 // CHECK: @multi_K
 void multi_K()
 {
-  // CHECK: asm "foo $1,$0", "=*r|m|m,r|K|m[[CLOBBERS]](i32* @mout0, i32 1)
+  // CHECK: asm "foo $1,$0", "=*r|m|m,r|K|m[[CLOBBERS]](i32* elementtype(i32) @mout0, i32 1)
   asm("foo %1,%0" : "=r,m,m" (mout0) : "r,K,m" (1));
 }
 
 // CHECK: @multi_L
 void multi_L()
 {
-  // CHECK: asm "foo $1,$0", "=*r|m|m,r|L|m[[CLOBBERS]](i32* @mout0, i32 1)
+  // CHECK: asm "foo $1,$0", "=*r|m|m,r|L|m[[CLOBBERS]](i32* elementtype(i32) @mout0, i32 1)
   asm("foo %1,%0" : "=r,m,m" (mout0) : "r,L,m" (1));
 }
 
 // CHECK: @multi_M
 void multi_M()
 {
-  // CHECK: asm "foo $1,$0", "=*r|m|m,r|M|m[[CLOBBERS]](i32* @mout0, i32 1)
+  // CHECK: asm "foo $1,$0", "=*r|m|m,r|M|m[[CLOBBERS]](i32* elementtype(i32) @mout0, i32 1)
   asm("foo %1,%0" : "=r,m,m" (mout0) : "r,M,m" (1));
 }
 
 // CHECK: @multi_N
 void multi_N()
 {
-  // CHECK: asm "foo $1,$0", "=*r|m|m,r|N|m[[CLOBBERS]](i32* @mout0, i32 1)
+  // CHECK: asm "foo $1,$0", "=*r|m|m,r|N|m[[CLOBBERS]](i32* elementtype(i32) @mout0, i32 1)
   asm("foo %1,%0" : "=r,m,m" (mout0) : "r,N,m" (1));
 }
 
 // CHECK: @multi_G
 void multi_G()
 {
-  // CHECK: asm "foo $1,$0", "=*r|m|m,r|G|m[[CLOBBERS]](i32* @mout0, double {{1.[0]+e[+]*[0]+}})
+  // CHECK: asm "foo $1,$0", "=*r|m|m,r|G|m[[CLOBBERS]](i32* elementtype(i32) @mout0, double {{1.[0]+e[+]*[0]+}})
   asm("foo %1,%0" : "=r,m,m" (mout0) : "r,G,m" (1.0));
 }
 
 // CHECK: @multi_C
 void multi_C()
 {
-  // CHECK: asm "foo $1,$0", "=*r|m|m,r|C|m[[CLOBBERS]](i32* @mout0, double {{1.[0]+e[+]*[0]+}})
+  // CHECK: asm "foo $1,$0", "=*r|m|m,r|C|m[[CLOBBERS]](i32* elementtype(i32) @mout0, double {{1.[0]+e[+]*[0]+}})
   asm("foo %1,%0" : "=r,m,m" (mout0) : "r,C,m" (1.0));
 }
 
 // CHECK: @multi_e
 void multi_e()
 {
-  // CHECK: asm "foo $1,$0", "=*r|m|m,r|e|m[[CLOBBERS]](i32* @mout0, i32 1)
+  // CHECK: asm "foo $1,$0", "=*r|m|m,r|e|m[[CLOBBERS]](i32* elementtype(i32) @mout0, i32 1)
   asm("foo %1,%0" : "=r,m,m" (mout0) : "r,e,m" (1));
 }
 
 // CHECK: @multi_Z
 void multi_Z()
 {
-  // CHECK: asm "foo $1,$0", "=*r|m|m,r|Z|m[[CLOBBERS]](i32* @mout0, i32 1)
+  // CHECK: asm "foo $1,$0", "=*r|m|m,r|Z|m[[CLOBBERS]](i32* elementtype(i32) @mout0, i32 1)
   asm("foo %1,%0" : "=r,m,m" (mout0) : "r,Z,m" (1));
 }
Index: clang/test/CodeGen/ppc64-inline-asm.c
===================================================================
--- clang/test/CodeGen/ppc64-inline-asm.c
+++ clang/test/CodeGen/ppc64-inline-asm.c
@@ -41,12 +41,12 @@
 void testZ(void *addr) {
   asm volatile ("dcbz %y0\n" :: "Z"(*(unsigned char *)addr) : "memory");
 // CHECK-LABEL: void @testZ(i8* %addr)
-// CHECK: call void asm sideeffect "dcbz ${0:y}\0A", "*Z,~{memory}"(i8* %addr)
+// CHECK: call void asm sideeffect "dcbz ${0:y}\0A", "*Z,~{memory}"(i8* elementtype(i8) %addr)
 }
 
 void testZwOff(void *addr, long long off) {
   asm volatile ("dcbz %y0\n" :: "Z"(*(unsigned char *)(addr + off)) : "memory");
 // CHECK-LABEL: void @testZwOff(i8* %addr, i64 %off)
 // CHECK: %[[VAL:[^ ]+]] = getelementptr i8, i8* %addr, i64 %off
-// CHECK: call void asm sideeffect "dcbz ${0:y}\0A", "*Z,~{memory}"(i8* %[[VAL]])
+// CHECK: call void asm sideeffect "dcbz ${0:y}\0A", "*Z,~{memory}"(i8* elementtype(i8) %[[VAL]])
 }
Index: clang/test/CodeGenCXX/ms-inline-asm-fields.cpp
===================================================================
--- clang/test/CodeGenCXX/ms-inline-asm-fields.cpp
+++ clang/test/CodeGenCXX/ms-inline-asm-fields.cpp
@@ -24,7 +24,7 @@
 
 extern "C" int test_namespace_global() {
 // CHECK: define{{.*}} i32 @test_namespace_global()
-// CHECK: call i32 asm sideeffect inteldialect "mov eax, $1", "{{.*}}"(i32* getelementptr inbounds (%struct.A, %struct.A* @_ZN4asdf8a_globalE, i32 0, i32 2, i32 1))
+// CHECK: call i32 asm sideeffect inteldialect "mov eax, $1", "{{.*}}"(i32* elementtype(i32) getelementptr inbounds (%struct.A, %struct.A* @_ZN4asdf8a_globalE, i32 0, i32 2, i32 1))
 // CHECK: ret i32
   __asm mov eax, asdf::a_global.a3.b2
 }
@@ -53,4 +53,4 @@
 // CHECK: %[[P:.*]] = alloca %"struct.make_storage_type<false>::type", align 4
 // CHECK: %[[B:.*]] = getelementptr inbounds %"struct.make_storage_type<false>::type", %"struct.make_storage_type<false>::type"* %[[P]], i32 0, i32 0
 // CHECK: %[[X:.*]] = getelementptr inbounds %"struct.make_storage_type<false>::type::B", %"struct.make_storage_type<false>::type::B"* %[[B]], i32 0, i32 1
-// CHECK: call void asm sideeffect inteldialect "mov edx, dword ptr $0", "*m,~{edx},~{dirflag},~{fpsr},~{flags}"(i32* %[[X]])
+// CHECK: call void asm sideeffect inteldialect "mov edx, dword ptr $0", "*m,~{edx},~{dirflag},~{fpsr},~{flags}"(i32* elementtype(i32) %[[X]])
Index: clang/test/CodeGenObjC/exceptions.m
===================================================================
--- clang/test/CodeGenObjC/exceptions.m
+++ clang/test/CodeGenObjC/exceptions.m
@@ -54,14 +54,14 @@
   // CHECK-NEXT:   br i1 [[CAUGHT]]
   @try {
     // Landing pad.  Note that we elide the re-enter.
-    // CHECK:      call void asm sideeffect "", "=*m,=*m"(i32* nonnull [[X]]
+    // CHECK:      call void asm sideeffect "", "=*m,=*m"(i32* nonnull elementtype(i32) [[X]]
     // CHECK-NEXT: call i8* @objc_exception_extract
     // CHECK-NEXT: [[T1:%.*]] = load i32, i32* [[X]]
     // CHECK-NEXT: [[T2:%.*]] = add nsw i32 [[T1]], -1
 
     // CHECK: store i32 6, i32* [[X]]
     x++;
-    // CHECK-NEXT: call void asm sideeffect "", "*m,*m"(i32* nonnull [[X]]
+    // CHECK-NEXT: call void asm sideeffect "", "*m,*m"(i32* nonnull elementtype(i32) [[X]]
     // CHECK-NEXT: call void @foo()
     // CHECK-NEXT: call void @objc_exception_try_exit
     // CHECK-NEXT: [[T:%.*]] = load i32, i32* [[X]]
Index: clang/test/CodeGenObjC/synchronized.m
===================================================================
--- clang/test/CodeGenObjC/synchronized.m
+++ clang/test/CodeGenObjC/synchronized.m
@@ -32,7 +32,7 @@
   // CHECK:      call i32 @_setjmp
   @synchronized(a) {
     // This is unreachable, but the optimizers can't know that.
-    // CHECK: call void asm sideeffect "", "=*m,=*m,=*m"(i8** nonnull [[A]], i8** nonnull [[SYNC]]
+    // CHECK: call void asm sideeffect "", "=*m,=*m,=*m"(i8** nonnull elementtype(i8*) [[A]], i8** nonnull elementtype(i8*) [[SYNC]]
     // CHECK: call i32 @objc_sync_exit
     // CHECK: call i8* @objc_exception_extract
     // CHECK: call void @objc_exception_throw