Index: clang/lib/CodeGen/TargetInfo.cpp
===================================================================
--- clang/lib/CodeGen/TargetInfo.cpp
+++ clang/lib/CodeGen/TargetInfo.cpp
@@ -295,17 +295,13 @@
 static llvm::Value *emitRoundPointerUpToAlignment(CodeGenFunction &CGF,
                                                   llvm::Value *Ptr,
                                                   CharUnits Align) {
-  llvm::Value *PtrAsInt = Ptr;
   // OverflowArgArea = (OverflowArgArea + Align - 1) & -Align;
-  PtrAsInt = CGF.Builder.CreatePtrToInt(PtrAsInt, CGF.IntPtrTy);
-  PtrAsInt = CGF.Builder.CreateAdd(PtrAsInt,
-        llvm::ConstantInt::get(CGF.IntPtrTy, Align.getQuantity() - 1));
-  PtrAsInt = CGF.Builder.CreateAnd(PtrAsInt,
-           llvm::ConstantInt::get(CGF.IntPtrTy, -Align.getQuantity()));
-  PtrAsInt = CGF.Builder.CreateIntToPtr(PtrAsInt,
-                                        Ptr->getType(),
-                                        Ptr->getName() + ".aligned");
-  return PtrAsInt;
+  llvm::Value *RoundUp = CGF.Builder.CreateConstInBoundsGEP1_32(
+      CGF.Builder.getInt8Ty(), Ptr, Align.getQuantity() - 1);
+  return CGF.Builder.CreateIntrinsic(
+      llvm::Intrinsic::ptrmask, {CGF.AllocaInt8PtrTy, CGF.IntPtrTy},
+      {RoundUp, llvm::ConstantInt::get(CGF.IntPtrTy, -Align.getQuantity())},
+      nullptr, Ptr->getName() + ".aligned");
 }
 
 /// Emit va_arg for a platform using the common void* representation,
Index: clang/test/CodeGen/PowerPC/ppc-varargs-struct.c
===================================================================
--- clang/test/CodeGen/PowerPC/ppc-varargs-struct.c
+++ clang/test/CodeGen/PowerPC/ppc-varargs-struct.c
@@ -37,10 +37,8 @@
 // CHECK-PPC-NEXT:  store i8 8, ptr [[GPRPTR]], align 4
 // CHECK-PPC-NEXT:  [[OVERFLOW_AREA_P:%[0-9]+]] = getelementptr inbounds %struct.__va_list_tag, ptr [[ARRAYDECAY]], i32 0, i32 3
 // CHECK-PPC-NEXT:  [[OVERFLOW_AREA:%.+]] = load ptr, ptr [[OVERFLOW_AREA_P]], align 4
-// CHECK-PPC-NEXT:  %{{[0-9]+}} =  ptrtoint ptr %argp.cur to i32
-// CHECK-PPC-NEXT:  %{{[0-9]+}} = add i32 %{{[0-9]+}}, 7
-// CHECK-PPC-NEXT:  %{{[0-9]+}} = and i32 %{{[0-9]+}}, -8
-// CHECK-PPC-NEXT:  %argp.cur.aligned = inttoptr i32 %{{[0-9]+}} to ptr
+// CHECK-PPC-NEXT: [[GEP_ALIGN:%[0-9]+]] = getelementptr inbounds i8, ptr %argp.cur, i32 7
+// CHECK-PPC-NEXT: %argp.cur.aligned = call ptr @llvm.ptrmask.p0.i32(ptr [[GEP_ALIGN]], i32 -8)
 // CHECK-PPC-NEXT:  [[NEW_OVERFLOW_AREA:%[0-9]+]] = getelementptr inbounds i8, ptr %argp.cur.aligned, i32 4
 // CHECK-PPC-NEXT:  store ptr [[NEW_OVERFLOW_AREA:%[0-9]+]], ptr [[OVERFLOW_AREA_P]], align 4
 // CHECK-PPC-NEXT:  br label %[[CONT]]
@@ -51,7 +49,7 @@
 // CHECK-PPC-NEXT:  call void @llvm.memcpy.p0.p0.i32(ptr align 8 %t, ptr align 8 [[AGGR]], i32 16, i1 false)
 
   int v = va_arg (ap, int);
-  
+
 // CHECK: getelementptr inbounds i8, ptr %{{[a-z.0-9]*}}, i64 4
 // CHECK-PPC:       [[ARRAYDECAY:%[a-z0-9]+]] = getelementptr inbounds [1 x %struct.__va_list_tag], ptr %ap, i32 0, i32 0
 // CHECK-PPC-NEXT:  [[GPRPTR:%.+]] = getelementptr inbounds %struct.__va_list_tag, ptr [[ARRAYDECAY]], i32 0, i32 0
Index: clang/test/CodeGen/x86_32-align-linux.c
===================================================================
--- clang/test/CodeGen/x86_32-align-linux.c
+++ clang/test/CodeGen/x86_32-align-linux.c
@@ -9,10 +9,8 @@
 
 // CHECK-LABEL: define dso_local void @testm128
 // CHECK-LABEL: %argp.cur = load ptr, ptr %args, align 4
-// CHECK-NEXT:  %0 = ptrtoint ptr %argp.cur to i32
-// CHECK-NEXT:  %1 = add i32 %0, 15
-// CHECK-NEXT:  %2 = and i32 %1, -16
-// CHECK-NEXT:  %argp.cur.aligned = inttoptr i32 %2 to ptr
+// CHECK-NEXT: %0 = getelementptr inbounds i8, ptr %argp.cur, i32 15
+// CHECK-NEXT: %argp.cur.aligned = call ptr @llvm.ptrmask.p0.i32(ptr %0, i32 -16)
 void testm128(int argCount, ...) {
   __m128 res;
   __builtin_va_list args;
@@ -23,10 +21,8 @@
 
 // CHECK-LABEL: define dso_local void @testm256
 // CHECK-LABEL: %argp.cur = load ptr, ptr %args, align 4
-// CHECK-NEXT:  %0 = ptrtoint ptr %argp.cur to i32
-// CHECK-NEXT:  %1 = add i32 %0, 31
-// CHECK-NEXT:  %2 = and i32 %1, -32
-// CHECK-NEXT:  %argp.cur.aligned = inttoptr i32 %2 to ptr
+// CHECK-NEXT: %0 = getelementptr inbounds i8, ptr %argp.cur, i32 31
+// CHECK-NEXT: %argp.cur.aligned = call ptr @llvm.ptrmask.p0.i32(ptr %0, i32 -32)
 void testm256(int argCount, ...) {
   __m256 res;
   __builtin_va_list args;
@@ -37,10 +33,8 @@
 
 // CHECK-LABEL: define dso_local void @testm512
 // CHECK-LABEL: %argp.cur = load ptr, ptr %args, align 4
-// CHECK-NEXT:  %0 = ptrtoint ptr %argp.cur to i32
-// CHECK-NEXT:  %1 = add i32 %0, 63
-// CHECK-NEXT:  %2 = and i32 %1, -64
-// CHECK-NEXT:  %argp.cur.aligned = inttoptr i32 %2 to ptr
+// CHECK-NEXT: %0 = getelementptr inbounds i8, ptr %argp.cur, i32 63
+// CHECK-NEXT: %argp.cur.aligned = call ptr @llvm.ptrmask.p0.i32(ptr %0, i32 -64)
 void testm512(int argCount, ...) {
   __m512 res;
   __builtin_va_list args;