[SystemZ] Add support for z13 and its vector facility

This patch adds support for the z13 architecture type.  For compatibility
with GCC, a pair of options -mvx / -mno-vx can be used to selectively
enable/disable use of the vector facility.

When the vector facility is present, we default to the new vector ABI.
This is characterized by two major differences:
- Vector types are passed/returned in vector registers
  (except for unnamed arguments of a variable-argument list function).
- Vector types are at most 8-byte aligned.

The reason for the choice of 8-byte vector alignment is that the hardware
is able to efficiently load vectors at 8-byte alignment, and the ABI only
guarantees 8-byte alignment of the stack pointer, so requiring any higher
alignment for vectors would require dynamic stack re-alignment code.

However, for compatibility with old code that may use vector types, when
*not* using the vector facility, the old alignment rules (vector types
are naturally aligned) remain in use.

These alignment rules are not only implemented at the C language level,
but also at the LLVM IR level.  This is done by selecting a different
DataLayout string depending on whether the vector ABI is in effect or not.

Based on a patch by Richard Sandiford.

git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@236531 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/include/clang/Driver/Options.td b/include/clang/Driver/Options.td
index 9c4cf8c656432a2290e7194f6b4c3e6b4085644d..cef5cc3452678e20a33fafaddab48b788425db7c 100644 (file)
--- a/include/clang/Driver/Options.td
+++ b/include/clang/Driver/Options.td
@@ -1310,6 +1310,9 @@ def fno_altivec : Flag<["-"], "fno-altivec">, Group<f_Group>, Flags<[CC1Option]>
 def maltivec : Flag<["-"], "maltivec">, Alias<faltivec>;
 def mno_altivec : Flag<["-"], "mno-altivec">, Alias<fno_altivec>;
 
+def mvx : Flag<["-"], "mvx">, Group<m_Group>;
+def mno_vx : Flag<["-"], "mno-vx">, Group<m_Group>;
+
 def mno_warn_nonportable_cfstrings : Flag<["-"], "mno-warn-nonportable-cfstrings">, Group<m_Group>;
 def mno_omit_leaf_frame_pointer : Flag<["-"], "mno-omit-leaf-frame-pointer">, Group<m_Group>;
 def momit_leaf_frame_pointer : Flag<["-"], "momit-leaf-frame-pointer">, Group<m_Group>,

diff --git a/lib/Basic/Targets.cpp b/lib/Basic/Targets.cpp
index 6d946236e28caff79a495710ef229a5973dd3633..91babd3d14eb99650de1ecd8aca4c31645823203 100644 (file)
--- a/lib/Basic/Targets.cpp
+++ b/lib/Basic/Targets.cpp
@@ -5531,10 +5531,11 @@ class SystemZTargetInfo : public TargetInfo {
   static const char *const GCCRegNames[];
   std::string CPU;
   bool HasTransactionalExecution;
+  bool HasVector;
 
 public:
   SystemZTargetInfo(const llvm::Triple &Triple)
-    : TargetInfo(Triple), CPU("z10"), HasTransactionalExecution(false) {
+    : TargetInfo(Triple), CPU("z10"), HasTransactionalExecution(false), HasVector(false) {
     IntMaxType = SignedLong;
     Int64Type = SignedLong;
     TLSSupported = true;
@@ -5587,6 +5588,7 @@ public:
       .Case("z10", true)
       .Case("z196", true)
       .Case("zEC12", true)
+      .Case("z13", true)
       .Default(false);
 
     return CPUKnown;
@@ -5594,6 +5596,10 @@ public:
   void getDefaultFeatures(llvm::StringMap<bool> &Features) const override {
     if (CPU == "zEC12")
       Features["transactional-execution"] = true;
+    if (CPU == "z13") {
+      Features["transactional-execution"] = true;
+      Features["vector"] = true;
+    }
   }
 
   bool handleTargetFeatures(std::vector<std::string> &Features,
@@ -5602,6 +5608,14 @@ public:
     for (unsigned i = 0, e = Features.size(); i != e; ++i) {
       if (Features[i] == "+transactional-execution")
         HasTransactionalExecution = true;
+      if (Features[i] == "+vector")
+        HasVector = true;
+    }
+    // If we use the vector ABI, vector types are 64-bit aligned.
+    if (HasVector) {
+      MaxVectorAlign = 64;
+      DescriptionString = "E-m:e-i1:8:16-i8:8:16-i64:64-f128:64"
+                          "-v128:64-a:8:16-n32:64";
     }
     return true;
   }
@@ -5610,8 +5624,15 @@ public:
     return llvm::StringSwitch<bool>(Feature)
         .Case("systemz", true)
         .Case("htm", HasTransactionalExecution)
+        .Case("vx", HasVector)
         .Default(false);
   }
+
+  StringRef getABI() const override {
+    if (HasVector)
+      return "vector";
+    return "";
+  }
 };
 
 const Builtin::Info SystemZTargetInfo::BuiltinInfo[] = {

diff --git a/lib/CodeGen/TargetInfo.cpp b/lib/CodeGen/TargetInfo.cpp
index 3a12304fb5379488689d6dc2bd6517f99013163b..9d2871649c832771605e9461075d53aa48f0b261 100644 (file)
--- a/lib/CodeGen/TargetInfo.cpp
+++ b/lib/CodeGen/TargetInfo.cpp
@@ -5146,12 +5146,17 @@ void NVPTXTargetCodeGenInfo::addNVVMMetadata(llvm::Function *F, StringRef Name,
 namespace {
 
 class SystemZABIInfo : public ABIInfo {
+  bool HasVector;
+
 public:
-  SystemZABIInfo(CodeGenTypes &CGT) : ABIInfo(CGT) {}
+  SystemZABIInfo(CodeGenTypes &CGT, bool HV)
+    : ABIInfo(CGT), HasVector(HV) {}
 
   bool isPromotableIntegerType(QualType Ty) const;
   bool isCompoundType(QualType Ty) const;
+  bool isVectorArgumentType(QualType Ty) const;
   bool isFPArgumentType(QualType Ty) const;
+  QualType GetSingleElementType(QualType Ty) const;
 
   ABIArgInfo classifyReturnType(QualType RetTy) const;
   ABIArgInfo classifyArgumentType(QualType ArgTy) const;
@@ -5169,8 +5174,8 @@ public:
 
 class SystemZTargetCodeGenInfo : public TargetCodeGenInfo {
 public:
-  SystemZTargetCodeGenInfo(CodeGenTypes &CGT)
-    : TargetCodeGenInfo(new SystemZABIInfo(CGT)) {}
+  SystemZTargetCodeGenInfo(CodeGenTypes &CGT, bool HasVector)
+    : TargetCodeGenInfo(new SystemZABIInfo(CGT, HasVector)) {}
 };
 
 }
@@ -5202,6 +5207,12 @@ bool SystemZABIInfo::isCompoundType(QualType Ty) const {
           isAggregateTypeForABI(Ty));
 }
 
+bool SystemZABIInfo::isVectorArgumentType(QualType Ty) const {
+  return (HasVector &&
+          Ty->isVectorType() &&
+          getContext().getTypeSize(Ty) <= 128);
+}
+
 bool SystemZABIInfo::isFPArgumentType(QualType Ty) const {
   if (const BuiltinType *BT = Ty->getAs<BuiltinType>())
     switch (BT->getKind()) {
@@ -5212,9 +5223,13 @@ bool SystemZABIInfo::isFPArgumentType(QualType Ty) const {
       return false;
     }
 
+  return false;
+}
+
+QualType SystemZABIInfo::GetSingleElementType(QualType Ty) const {
   if (const RecordType *RT = Ty->getAsStructureType()) {
     const RecordDecl *RD = RT->getDecl();
-    bool Found = false;
+    QualType Found;
 
     // If this is a C++ record, check the bases first.
     if (const CXXRecordDecl *CXXRD = dyn_cast<CXXRecordDecl>(RD))
@@ -5225,11 +5240,9 @@ bool SystemZABIInfo::isFPArgumentType(QualType Ty) const {
         if (isEmptyRecord(getContext(), Base, true))
           continue;
 
-        if (Found)
-          return false;
-        Found = isFPArgumentType(Base);
-        if (!Found)
-          return false;
+        if (!Found.isNull())
+          return Ty;
+        Found = GetSingleElementType(Base);
       }
 
     // Check the fields.
@@ -5242,20 +5255,19 @@ bool SystemZABIInfo::isFPArgumentType(QualType Ty) const {
         continue;
 
       // Unlike isSingleElementStruct(), arrays do not count.
-      // Nested isFPArgumentType structures still do though.
-      if (Found)
-        return false;
-      Found = isFPArgumentType(FD->getType());
-      if (!Found)
-        return false;
+      // Nested structures still do though.
+      if (!Found.isNull())
+        return Ty;
+      Found = GetSingleElementType(FD->getType());
     }
 
     // Unlike isSingleElementStruct(), trailing padding is allowed.
     // An 8-byte aligned struct s { float f; } is passed as a double.
-    return Found;
+    if (!Found.isNull())
+      return Found;
   }
 
-  return false;
+  return Ty;
 }
 
 llvm::Value *SystemZABIInfo::EmitVAArg(llvm::Value *VAListAddr, QualType Ty,
@@ -5268,14 +5280,16 @@ llvm::Value *SystemZABIInfo::EmitVAArg(llvm::Value *VAListAddr, QualType Ty,
   //   i8 *__reg_save_area;
   // };
 
-  // Every argument occupies 8 bytes and is passed by preference in either
-  // GPRs or FPRs.
+  // Every non-vector argument occupies 8 bytes and is passed by preference
+  // in either GPRs or FPRs.  Vector arguments occupy 8 or 16 bytes and are
+  // always passed on the stack.
   Ty = CGF.getContext().getCanonicalType(Ty);
   llvm::Type *ArgTy = CGF.ConvertTypeForMem(Ty);
   llvm::Type *APTy = llvm::PointerType::getUnqual(ArgTy);
   ABIArgInfo AI = classifyArgumentType(Ty);
   bool IsIndirect = AI.isIndirect();
   bool InFPRs = false;
+  bool IsVector = false;
   unsigned UnpaddedBitSize;
   if (IsIndirect) {
     APTy = llvm::PointerType::getUnqual(APTy);
@@ -5284,14 +5298,38 @@ llvm::Value *SystemZABIInfo::EmitVAArg(llvm::Value *VAListAddr, QualType Ty,
     if (AI.getCoerceToType())
       ArgTy = AI.getCoerceToType();
     InFPRs = ArgTy->isFloatTy() || ArgTy->isDoubleTy();
+    IsVector = ArgTy->isVectorTy();
     UnpaddedBitSize = getContext().getTypeSize(Ty);
   }
-  unsigned PaddedBitSize = 64;
+  unsigned PaddedBitSize = (IsVector && UnpaddedBitSize > 64) ? 128 : 64;
   assert((UnpaddedBitSize <= PaddedBitSize) && "Invalid argument size.");
 
   unsigned PaddedSize = PaddedBitSize / 8;
   unsigned Padding = (PaddedBitSize - UnpaddedBitSize) / 8;
 
+  llvm::Type *IndexTy = CGF.Int64Ty;
+  llvm::Value *PaddedSizeV = llvm::ConstantInt::get(IndexTy, PaddedSize);
+
+  if (IsVector) {
+    // Work out the address of a vector argument on the stack.
+    // Vector arguments are always passed in the high bits of a
+    // single (8 byte) or double (16 byte) stack slot.
+    llvm::Value *OverflowArgAreaPtr =
+      CGF.Builder.CreateStructGEP(nullptr, VAListAddr, 2,
+                                  "overflow_arg_area_ptr");
+    llvm::Value *OverflowArgArea =
+      CGF.Builder.CreateLoad(OverflowArgAreaPtr, "overflow_arg_area");
+    llvm::Value *MemAddr =
+      CGF.Builder.CreateBitCast(OverflowArgArea, APTy, "mem_addr");
+
+    // Update overflow_arg_area_ptr pointer
+    llvm::Value *NewOverflowArgArea =
+      CGF.Builder.CreateGEP(OverflowArgArea, PaddedSizeV, "overflow_arg_area");
+    CGF.Builder.CreateStore(NewOverflowArgArea, OverflowArgAreaPtr);
+
+    return MemAddr;
+  }
+
   unsigned MaxRegs, RegCountField, RegSaveIndex, RegPadding;
   if (InFPRs) {
     MaxRegs = 4; // Maximum of 4 FPR arguments
@@ -5308,7 +5346,6 @@ llvm::Value *SystemZABIInfo::EmitVAArg(llvm::Value *VAListAddr, QualType Ty,
   llvm::Value *RegCountPtr = CGF.Builder.CreateStructGEP(
       nullptr, VAListAddr, RegCountField, "reg_count_ptr");
   llvm::Value *RegCount = CGF.Builder.CreateLoad(RegCountPtr, "reg_count");
-  llvm::Type *IndexTy = RegCount->getType();
   llvm::Value *MaxRegsV = llvm::ConstantInt::get(IndexTy, MaxRegs);
   llvm::Value *InRegs = CGF.Builder.CreateICmpULT(RegCount, MaxRegsV,
                                                  "fits_in_regs");
@@ -5322,7 +5359,6 @@ llvm::Value *SystemZABIInfo::EmitVAArg(llvm::Value *VAListAddr, QualType Ty,
   CGF.EmitBlock(InRegBlock);
 
   // Work out the address of an argument register.
-  llvm::Value *PaddedSizeV = llvm::ConstantInt::get(IndexTy, PaddedSize);
   llvm::Value *ScaledRegCount =
     CGF.Builder.CreateMul(RegCount, PaddedSizeV, "scaled_reg_count");
   llvm::Value *RegBase =
@@ -5380,6 +5416,8 @@ llvm::Value *SystemZABIInfo::EmitVAArg(llvm::Value *VAListAddr, QualType Ty,
 ABIArgInfo SystemZABIInfo::classifyReturnType(QualType RetTy) const {
   if (RetTy->isVoidType())
     return ABIArgInfo::getIgnore();
+  if (isVectorArgumentType(RetTy))
+    return ABIArgInfo::getDirect();
   if (isCompoundType(RetTy) || getContext().getTypeSize(RetTy) > 64)
     return ABIArgInfo::getIndirect(0);
   return (isPromotableIntegerType(RetTy) ?
@@ -5395,8 +5433,16 @@ ABIArgInfo SystemZABIInfo::classifyArgumentType(QualType Ty) const {
   if (isPromotableIntegerType(Ty))
     return ABIArgInfo::getExtend();
 
-  // Values that are not 1, 2, 4 or 8 bytes in size are passed indirectly.
+  // Handle vector types and vector-like structure types.  Note that
+  // as opposed to float-like structure types, we do not allow any
+  // padding for vector-like structures, so verify the sizes match.
   uint64_t Size = getContext().getTypeSize(Ty);
+  QualType SingleElementTy = GetSingleElementType(Ty);
+  if (isVectorArgumentType(SingleElementTy) &&
+      getContext().getTypeSize(SingleElementTy) == Size)
+    return ABIArgInfo::getDirect(CGT.ConvertType(SingleElementTy));
+
+  // Values that are not 1, 2, 4 or 8 bytes in size are passed indirectly.
   if (Size != 8 && Size != 16 && Size != 32 && Size != 64)
     return ABIArgInfo::getIndirect(0, /*ByVal=*/false);
 
@@ -5410,7 +5456,7 @@ ABIArgInfo SystemZABIInfo::classifyArgumentType(QualType Ty) const {
 
     // The structure is passed as an unextended integer, a float, or a double.
     llvm::Type *PassTy;
-    if (isFPArgumentType(Ty)) {
+    if (isFPArgumentType(SingleElementTy)) {
       assert(Size == 32 || Size == 64);
       if (Size == 32)
         PassTy = llvm::Type::getFloatTy(getVMContext());
@@ -7067,8 +7113,11 @@ const TargetCodeGenInfo &CodeGenModule::getTargetCodeGenInfo() {
   case llvm::Triple::msp430:
     return *(TheTargetCodeGenInfo = new MSP430TargetCodeGenInfo(Types));
 
-  case llvm::Triple::systemz:
-    return *(TheTargetCodeGenInfo = new SystemZTargetCodeGenInfo(Types));
+  case llvm::Triple::systemz: {
+    bool HasVector = getTarget().getABI() == "vector";
+    return *(TheTargetCodeGenInfo = new SystemZTargetCodeGenInfo(Types,
+                                                                 HasVector));
+  }
 
   case llvm::Triple::tce:
     return *(TheTargetCodeGenInfo = new TCETargetCodeGenInfo(Types));

diff --git a/lib/Driver/Tools.cpp b/lib/Driver/Tools.cpp
index 27e6427d0d4fa4ba4bb00c42fccfde47c8884b3f..3f4c317133234622f845c2c9103887c04e2e9571 100644 (file)
--- a/lib/Driver/Tools.cpp
+++ b/lib/Driver/Tools.cpp
@@ -1442,6 +1442,14 @@ static void getSystemZTargetFeatures(const ArgList &Args,
     else
       Features.push_back("-transactional-execution");
   }
+  // -m(no-)vx overrides use of the vector facility.
+  if (Arg *A = Args.getLastArg(options::OPT_mvx,
+                               options::OPT_mno_vx)) {
+    if (A->getOption().matches(options::OPT_mvx))
+      Features.push_back("+vector");
+    else
+      Features.push_back("-vector");
+  }
 }
 
 static const char *getX86TargetCPU(const ArgList &Args,

diff --git a/test/CodeGen/systemz-abi-vector.c b/test/CodeGen/systemz-abi-vector.c
index 06c5f3f8bd6634b32f64b1bef4c850cbc02a3dbf..1d1f302e632453e673d185da58fd257a899a369d 100644 (file)
--- a/test/CodeGen/systemz-abi-vector.c
+++ b/test/CodeGen/systemz-abi-vector.c
@@ -1,4 +1,9 @@
-// RUN: %clang_cc1 -triple s390x-linux-gnu -emit-llvm -o - %s | FileCheck %s
+// RUN: %clang_cc1 -triple s390x-linux-gnu \
+// RUN:   -emit-llvm -o - %s | FileCheck %s
+// RUN: %clang_cc1 -triple s390x-linux-gnu -target-feature +vector \
+// RUN:   -emit-llvm -o - %s | FileCheck --check-prefix=CHECK-VECTOR %s
+// RUN: %clang_cc1 -triple s390x-linux-gnu -target-cpu z13 \
+// RUN:   -emit-llvm -o - %s | FileCheck --check-prefix=CHECK-VECTOR %s
 
 // Vector types
 
@@ -30,71 +35,153 @@ typedef __attribute__((vector_size(16))) long double v1f128;
 
 typedef __attribute__((vector_size(32))) char v32i8;
 
+unsigned int align = __alignof__ (v16i8);
+// CHECK: @align = global i32 16
+// CHECK-VECTOR: @align = global i32 8
+
 v1i8 pass_v1i8(v1i8 arg) { return arg; }
 // CHECK-LABEL: define void @pass_v1i8(<1 x i8>* noalias sret %{{.*}}, <1 x i8>*)
+// CHECK-VECTOR-LABEL: define <1 x i8> @pass_v1i8(<1 x i8> %{{.*}})
 
 v2i8 pass_v2i8(v2i8 arg) { return arg; }
 // CHECK-LABEL: define void @pass_v2i8(<2 x i8>* noalias sret %{{.*}}, <2 x i8>*)
+// CHECK-VECTOR-LABEL: define <2 x i8> @pass_v2i8(<2 x i8> %{{.*}})
 
 v4i8 pass_v4i8(v4i8 arg) { return arg; }
 // CHECK-LABEL: define void @pass_v4i8(<4 x i8>* noalias sret %{{.*}}, <4 x i8>*)
+// CHECK-VECTOR-LABEL: define <4 x i8> @pass_v4i8(<4 x i8> %{{.*}})
 
 v8i8 pass_v8i8(v8i8 arg) { return arg; }
 // CHECK-LABEL: define void @pass_v8i8(<8 x i8>* noalias sret %{{.*}}, <8 x i8>*)
+// CHECK-VECTOR-LABEL: define <8 x i8> @pass_v8i8(<8 x i8> %{{.*}})
 
 v16i8 pass_v16i8(v16i8 arg) { return arg; }
 // CHECK-LABEL: define void @pass_v16i8(<16 x i8>* noalias sret %{{.*}}, <16 x i8>*)
+// CHECK-VECTOR-LABEL: define <16 x i8> @pass_v16i8(<16 x i8> %{{.*}})
 
 v32i8 pass_v32i8(v32i8 arg) { return arg; }
 // CHECK-LABEL: define void @pass_v32i8(<32 x i8>* noalias sret %{{.*}}, <32 x i8>*)
+// CHECK-VECTOR-LABEL: define void @pass_v32i8(<32 x i8>* noalias sret %{{.*}}, <32 x i8>*)
 
 v1i16 pass_v1i16(v1i16 arg) { return arg; }
 // CHECK-LABEL: define void @pass_v1i16(<1 x i16>* noalias sret %{{.*}}, <1 x i16>*)
+// CHECK-VECTOR-LABEL: define <1 x i16> @pass_v1i16(<1 x i16> %{{.*}})
 
 v2i16 pass_v2i16(v2i16 arg) { return arg; }
 // CHECK-LABEL: define void @pass_v2i16(<2 x i16>* noalias sret %{{.*}}, <2 x i16>*)
+// CHECK-VECTOR-LABEL: define <2 x i16> @pass_v2i16(<2 x i16> %{{.*}})
 
 v4i16 pass_v4i16(v4i16 arg) { return arg; }
 // CHECK-LABEL: define void @pass_v4i16(<4 x i16>* noalias sret %{{.*}}, <4 x i16>*)
+// CHECK-VECTOR-LABEL: define <4 x i16> @pass_v4i16(<4 x i16> %{{.*}})
 
 v8i16 pass_v8i16(v8i16 arg) { return arg; }
 // CHECK-LABEL: define void @pass_v8i16(<8 x i16>* noalias sret %{{.*}}, <8 x i16>*)
+// CHECK-VECTOR-LABEL: define <8 x i16> @pass_v8i16(<8 x i16> %{{.*}})
 
 v1i32 pass_v1i32(v1i32 arg) { return arg; }
 // CHECK-LABEL: define void @pass_v1i32(<1 x i32>* noalias sret %{{.*}}, <1 x i32>*)
+// CHECK-VECTOR-LABEL: define <1 x i32> @pass_v1i32(<1 x i32> %{{.*}})
 
 v2i32 pass_v2i32(v2i32 arg) { return arg; }
 // CHECK-LABEL: define void @pass_v2i32(<2 x i32>* noalias sret %{{.*}}, <2 x i32>*)
+// CHECK-VECTOR-LABEL: define <2 x i32> @pass_v2i32(<2 x i32> %{{.*}})
 
 v4i32 pass_v4i32(v4i32 arg) { return arg; }
 // CHECK-LABEL: define void @pass_v4i32(<4 x i32>* noalias sret %{{.*}}, <4 x i32>*)
+// CHECK-VECTOR-LABEL: define <4 x i32> @pass_v4i32(<4 x i32> %{{.*}})
 
 v1i64 pass_v1i64(v1i64 arg) { return arg; }
 // CHECK-LABEL: define void @pass_v1i64(<1 x i64>* noalias sret %{{.*}}, <1 x i64>*)
+// CHECK-VECTOR-LABEL: define <1 x i64> @pass_v1i64(<1 x i64> %{{.*}})
 
 v2i64 pass_v2i64(v2i64 arg) { return arg; }
 // CHECK-LABEL: define void @pass_v2i64(<2 x i64>* noalias sret %{{.*}}, <2 x i64>*)
+// CHECK-VECTOR-LABEL: define <2 x i64> @pass_v2i64(<2 x i64> %{{.*}})
 
 v1i128 pass_v1i128(v1i128 arg) { return arg; }
 // CHECK-LABEL: define void @pass_v1i128(<1 x i128>* noalias sret %{{.*}}, <1 x i128>*)
+// CHECK-VECTOR-LABEL: define <1 x i128> @pass_v1i128(<1 x i128> %{{.*}})
 
 v1f32 pass_v1f32(v1f32 arg) { return arg; }
 // CHECK-LABEL: define void @pass_v1f32(<1 x float>* noalias sret %{{.*}}, <1 x float>*)
+// CHECK-VECTOR-LABEL: define <1 x float> @pass_v1f32(<1 x float> %{{.*}})
 
 v2f32 pass_v2f32(v2f32 arg) { return arg; }
 // CHECK-LABEL: define void @pass_v2f32(<2 x float>* noalias sret %{{.*}}, <2 x float>*)
+// CHECK-VECTOR-LABEL: define <2 x float> @pass_v2f32(<2 x float> %{{.*}})
 
 v4f32 pass_v4f32(v4f32 arg) { return arg; }
 // CHECK-LABEL: define void @pass_v4f32(<4 x float>* noalias sret %{{.*}}, <4 x float>*)
+// CHECK-VECTOR-LABEL: define <4 x float> @pass_v4f32(<4 x float> %{{.*}})
 
 v1f64 pass_v1f64(v1f64 arg) { return arg; }
 // CHECK-LABEL: define void @pass_v1f64(<1 x double>* noalias sret %{{.*}}, <1 x double>*)
+// CHECK-VECTOR-LABEL: define <1 x double> @pass_v1f64(<1 x double> %{{.*}})
 
 v2f64 pass_v2f64(v2f64 arg) { return arg; }
 // CHECK-LABEL: define void @pass_v2f64(<2 x double>* noalias sret %{{.*}}, <2 x double>*)
+// CHECK-VECTOR-LABEL: define <2 x double> @pass_v2f64(<2 x double> %{{.*}})
 
 v1f128 pass_v1f128(v1f128 arg) { return arg; }
 // CHECK-LABEL: define void @pass_v1f128(<1 x fp128>* noalias sret %{{.*}}, <1 x fp128>*)
+// CHECK-VECTOR-LABEL: define <1 x fp128> @pass_v1f128(<1 x fp128> %{{.*}})
+
+
+// Vector-like aggregate types
+
+struct agg_v1i8 { v1i8 a; };
+struct agg_v1i8 pass_agg_v1i8(struct agg_v1i8 arg) { return arg; }
+// CHECK-LABEL: define void @pass_agg_v1i8(%struct.agg_v1i8* noalias sret %{{.*}}, i8 %{{.*}})
+// CHECK-VECTOR-LABEL: define void @pass_agg_v1i8(%struct.agg_v1i8* noalias sret %{{.*}}, <1 x i8> %{{.*}})
+
+struct agg_v2i8 { v2i8 a; };
+struct agg_v2i8 pass_agg_v2i8(struct agg_v2i8 arg) { return arg; }
+// CHECK-LABEL: define void @pass_agg_v2i8(%struct.agg_v2i8* noalias sret %{{.*}}, i16 %{{.*}})
+// CHECK-VECTOR-LABEL: define void @pass_agg_v2i8(%struct.agg_v2i8* noalias sret %{{.*}}, <2 x i8> %{{.*}})
+
+struct agg_v4i8 { v4i8 a; };
+struct agg_v4i8 pass_agg_v4i8(struct agg_v4i8 arg) { return arg; }
+// CHECK-LABEL: define void @pass_agg_v4i8(%struct.agg_v4i8* noalias sret %{{.*}}, i32 %{{.*}})
+// CHECK-VECTOR-LABEL: define void @pass_agg_v4i8(%struct.agg_v4i8* noalias sret %{{.*}}, <4 x i8> %{{.*}})
+
+struct agg_v8i8 { v8i8 a; };
+struct agg_v8i8 pass_agg_v8i8(struct agg_v8i8 arg) { return arg; }
+// CHECK-LABEL: define void @pass_agg_v8i8(%struct.agg_v8i8* noalias sret %{{.*}}, i64 %{{.*}})
+// CHECK-VECTOR-LABEL: define void @pass_agg_v8i8(%struct.agg_v8i8* noalias sret %{{.*}}, <8 x i8> %{{.*}})
+
+struct agg_v16i8 { v16i8 a; };
+struct agg_v16i8 pass_agg_v16i8(struct agg_v16i8 arg) { return arg; }
+// CHECK-LABEL: define void @pass_agg_v16i8(%struct.agg_v16i8* noalias sret %{{.*}}, %struct.agg_v16i8* %{{.*}})
+// CHECK-VECTOR-LABEL: define void @pass_agg_v16i8(%struct.agg_v16i8* noalias sret %{{.*}}, <16 x i8> %{{.*}})
+
+struct agg_v32i8 { v32i8 a; };
+struct agg_v32i8 pass_agg_v32i8(struct agg_v32i8 arg) { return arg; }
+// CHECK-LABEL: define void @pass_agg_v32i8(%struct.agg_v32i8* noalias sret %{{.*}}, %struct.agg_v32i8* %{{.*}})
+// CHECK-VECTOR-LABEL: define void @pass_agg_v32i8(%struct.agg_v32i8* noalias sret %{{.*}}, %struct.agg_v32i8* %{{.*}})
+
+
+// Verify that the following are *not* vector-like aggregate types
+
+struct agg_novector1 { v4i8 a; v4i8 b; };
+struct agg_novector1 pass_agg_novector1(struct agg_novector1 arg) { return arg; }
+// CHECK-LABEL: define void @pass_agg_novector1(%struct.agg_novector1* noalias sret %{{.*}}, i64 %{{.*}})
+// CHECK-VECTOR-LABEL: define void @pass_agg_novector1(%struct.agg_novector1* noalias sret %{{.*}}, i64 %{{.*}})
+
+struct agg_novector2 { v4i8 a; float b; };
+struct agg_novector2 pass_agg_novector2(struct agg_novector2 arg) { return arg; }
+// CHECK-LABEL: define void @pass_agg_novector2(%struct.agg_novector2* noalias sret %{{.*}}, i64 %{{.*}})
+// CHECK-VECTOR-LABEL: define void @pass_agg_novector2(%struct.agg_novector2* noalias sret %{{.*}}, i64 %{{.*}})
+
+struct agg_novector3 { v4i8 a; int : 0; };
+struct agg_novector3 pass_agg_novector3(struct agg_novector3 arg) { return arg; }
+// CHECK-LABEL: define void @pass_agg_novector3(%struct.agg_novector3* noalias sret %{{.*}}, i32 %{{.*}})
+// CHECK-VECTOR-LABEL: define void @pass_agg_novector3(%struct.agg_novector3* noalias sret %{{.*}}, i32 %{{.*}})
+
+struct agg_novector4 { v4i8 a __attribute__((aligned (8))); };
+struct agg_novector4 pass_agg_novector4(struct agg_novector4 arg) { return arg; }
+// CHECK-LABEL: define void @pass_agg_novector4(%struct.agg_novector4* noalias sret %{{.*}}, i64 %{{.*}})
+// CHECK-VECTOR-LABEL: define void @pass_agg_novector4(%struct.agg_novector4* noalias sret %{{.*}}, i64 %{{.*}})
 
 
 // Accessing variable argument lists
@@ -122,6 +209,14 @@ v1i8 va_v1i8(__builtin_va_list l) { return __builtin_va_arg(l, v1i8); }
 // CHECK: [[VA_ARG_ADDR:%[^ ]+]] = phi <1 x i8>** [ [[REG_ADDR]], %{{.*}} ], [ [[MEM_ADDR]], %{{.*}} ]
 // CHECK: [[INDIRECT_ARG:%[^ ]+]] = load <1 x i8>*, <1 x i8>** [[VA_ARG_ADDR]]
 // CHECK: ret void
+// CHECK-VECTOR-LABEL: define <1 x i8> @va_v1i8(%struct.__va_list_tag* %{{.*}})
+// CHECK-VECTOR: [[OVERFLOW_ARG_AREA_PTR:%[^ ]+]] = getelementptr inbounds %struct.__va_list_tag, %struct.__va_list_tag* %{{.*}}, i32 0, i32 2
+// CHECK-VECTOR: [[OVERFLOW_ARG_AREA:%[^ ]+]] = load i8*, i8** [[OVERFLOW_ARG_AREA_PTR]]
+// CHECK-VECTOR: [[MEM_ADDR:%[^ ]+]] = bitcast i8* [[OVERFLOW_ARG_AREA]] to <1 x i8>*
+// CHECK-VECTOR: [[OVERFLOW_ARG_AREA1:%[^ ]+]] = getelementptr i8, i8* [[OVERFLOW_ARG_AREA]], i64 8
+// CHECK-VECTOR: store i8* [[OVERFLOW_ARG_AREA1]], i8** [[OVERFLOW_ARG_AREA_PTR]]
+// CHECK-VECTOR: [[RET:%[^ ]+]] = load <1 x i8>, <1 x i8>* [[MEM_ADDR]]
+// CHECK-VECTOR: ret <1 x i8> [[RET]]
 
 v2i8 va_v2i8(__builtin_va_list l) { return __builtin_va_arg(l, v2i8); }
 // CHECK-LABEL: define void @va_v2i8(<2 x i8>* noalias sret %{{.*}}, %struct.__va_list_tag* %{{.*}})
@@ -146,6 +241,14 @@ v2i8 va_v2i8(__builtin_va_list l) { return __builtin_va_arg(l, v2i8); }
 // CHECK: [[VA_ARG_ADDR:%[^ ]+]] = phi <2 x i8>** [ [[REG_ADDR]], %{{.*}} ], [ [[MEM_ADDR]], %{{.*}} ]
 // CHECK: [[INDIRECT_ARG:%[^ ]+]] = load <2 x i8>*, <2 x i8>** [[VA_ARG_ADDR]]
 // CHECK: ret void
+// CHECK-VECTOR-LABEL: define <2 x i8> @va_v2i8(%struct.__va_list_tag* %{{.*}})
+// CHECK-VECTOR: [[OVERFLOW_ARG_AREA_PTR:%[^ ]+]] = getelementptr inbounds %struct.__va_list_tag, %struct.__va_list_tag* %{{.*}}, i32 0, i32 2
+// CHECK-VECTOR: [[OVERFLOW_ARG_AREA:%[^ ]+]] = load i8*, i8** [[OVERFLOW_ARG_AREA_PTR]]
+// CHECK-VECTOR: [[MEM_ADDR:%[^ ]+]] = bitcast i8* [[OVERFLOW_ARG_AREA]] to <2 x i8>*
+// CHECK-VECTOR: [[OVERFLOW_ARG_AREA1:%[^ ]+]] = getelementptr i8, i8* [[OVERFLOW_ARG_AREA]], i64 8
+// CHECK-VECTOR: store i8* [[OVERFLOW_ARG_AREA1]], i8** [[OVERFLOW_ARG_AREA_PTR]]
+// CHECK-VECTOR: [[RET:%[^ ]+]] = load <2 x i8>, <2 x i8>* [[MEM_ADDR]]
+// CHECK-VECTOR: ret <2 x i8> [[RET]]
 
 v4i8 va_v4i8(__builtin_va_list l) { return __builtin_va_arg(l, v4i8); }
 // CHECK-LABEL: define void @va_v4i8(<4 x i8>* noalias sret %{{.*}}, %struct.__va_list_tag* %{{.*}})
@@ -170,6 +273,14 @@ v4i8 va_v4i8(__builtin_va_list l) { return __builtin_va_arg(l, v4i8); }
 // CHECK: [[VA_ARG_ADDR:%[^ ]+]] = phi <4 x i8>** [ [[REG_ADDR]], %{{.*}} ], [ [[MEM_ADDR]], %{{.*}} ]
 // CHECK: [[INDIRECT_ARG:%[^ ]+]] = load <4 x i8>*, <4 x i8>** [[VA_ARG_ADDR]]
 // CHECK: ret void
+// CHECK-VECTOR-LABEL: define <4 x i8> @va_v4i8(%struct.__va_list_tag* %{{.*}})
+// CHECK-VECTOR: [[OVERFLOW_ARG_AREA_PTR:%[^ ]+]] = getelementptr inbounds %struct.__va_list_tag, %struct.__va_list_tag* %{{.*}}, i32 0, i32 2
+// CHECK-VECTOR: [[OVERFLOW_ARG_AREA:%[^ ]+]] = load i8*, i8** [[OVERFLOW_ARG_AREA_PTR]]
+// CHECK-VECTOR: [[MEM_ADDR:%[^ ]+]] = bitcast i8* [[OVERFLOW_ARG_AREA]] to <4 x i8>*
+// CHECK-VECTOR: [[OVERFLOW_ARG_AREA1:%[^ ]+]] = getelementptr i8, i8* [[OVERFLOW_ARG_AREA]], i64 8
+// CHECK-VECTOR: store i8* [[OVERFLOW_ARG_AREA1]], i8** [[OVERFLOW_ARG_AREA_PTR]]
+// CHECK-VECTOR: [[RET:%[^ ]+]] = load <4 x i8>, <4 x i8>* [[MEM_ADDR]]
+// CHECK-VECTOR: ret <4 x i8> [[RET]]
 
 v8i8 va_v8i8(__builtin_va_list l) { return __builtin_va_arg(l, v8i8); }
 // CHECK-LABEL: define void @va_v8i8(<8 x i8>* noalias sret %{{.*}}, %struct.__va_list_tag* %{{.*}})
@@ -194,6 +305,14 @@ v8i8 va_v8i8(__builtin_va_list l) { return __builtin_va_arg(l, v8i8); }
 // CHECK: [[VA_ARG_ADDR:%[^ ]+]] = phi <8 x i8>** [ [[REG_ADDR]], %{{.*}} ], [ [[MEM_ADDR]], %{{.*}} ]
 // CHECK: [[INDIRECT_ARG:%[^ ]+]] = load <8 x i8>*, <8 x i8>** [[VA_ARG_ADDR]]
 // CHECK: ret void
+// CHECK-VECTOR-LABEL: define <8 x i8> @va_v8i8(%struct.__va_list_tag* %{{.*}})
+// CHECK-VECTOR: [[OVERFLOW_ARG_AREA_PTR:%[^ ]+]] = getelementptr inbounds %struct.__va_list_tag, %struct.__va_list_tag* %{{.*}}, i32 0, i32 2
+// CHECK-VECTOR: [[OVERFLOW_ARG_AREA:%[^ ]+]] = load i8*, i8** [[OVERFLOW_ARG_AREA_PTR]]
+// CHECK-VECTOR: [[MEM_ADDR:%[^ ]+]] = bitcast i8* [[OVERFLOW_ARG_AREA]] to <8 x i8>*
+// CHECK-VECTOR: [[OVERFLOW_ARG_AREA1:%[^ ]+]] = getelementptr i8, i8* [[OVERFLOW_ARG_AREA]], i64 8
+// CHECK-VECTOR: store i8* [[OVERFLOW_ARG_AREA1]], i8** [[OVERFLOW_ARG_AREA_PTR]]
+// CHECK-VECTOR: [[RET:%[^ ]+]] = load <8 x i8>, <8 x i8>* [[MEM_ADDR]]
+// CHECK-VECTOR: ret <8 x i8> [[RET]]
 
 v16i8 va_v16i8(__builtin_va_list l) { return __builtin_va_arg(l, v16i8); }
 // CHECK-LABEL: define void @va_v16i8(<16 x i8>* noalias sret %{{.*}}, %struct.__va_list_tag* %{{.*}})
@@ -218,6 +337,14 @@ v16i8 va_v16i8(__builtin_va_list l) { return __builtin_va_arg(l, v16i8); }
 // CHECK: [[VA_ARG_ADDR:%[^ ]+]] = phi <16 x i8>** [ [[REG_ADDR]], %{{.*}} ], [ [[MEM_ADDR]], %{{.*}} ]
 // CHECK: [[INDIRECT_ARG:%[^ ]+]] = load <16 x i8>*, <16 x i8>** [[VA_ARG_ADDR]]
 // CHECK: ret void
+// CHECK-VECTOR-LABEL: define <16 x i8> @va_v16i8(%struct.__va_list_tag* %{{.*}})
+// CHECK-VECTOR: [[OVERFLOW_ARG_AREA_PTR:%[^ ]+]] = getelementptr inbounds %struct.__va_list_tag, %struct.__va_list_tag* %{{.*}}, i32 0, i32 2
+// CHECK-VECTOR: [[OVERFLOW_ARG_AREA:%[^ ]+]] = load i8*, i8** [[OVERFLOW_ARG_AREA_PTR]]
+// CHECK-VECTOR: [[MEM_ADDR:%[^ ]+]] = bitcast i8* [[OVERFLOW_ARG_AREA]] to <16 x i8>*
+// CHECK-VECTOR: [[OVERFLOW_ARG_AREA1:%[^ ]+]] = getelementptr i8, i8* [[OVERFLOW_ARG_AREA]], i64 16
+// CHECK-VECTOR: store i8* [[OVERFLOW_ARG_AREA1]], i8** [[OVERFLOW_ARG_AREA_PTR]]
+// CHECK-VECTOR: [[RET:%[^ ]+]] = load <16 x i8>, <16 x i8>* [[MEM_ADDR]]
+// CHECK-VECTOR: ret <16 x i8> [[RET]]
 
 v32i8 va_v32i8(__builtin_va_list l) { return __builtin_va_arg(l, v32i8); }
 // CHECK-LABEL: define void @va_v32i8(<32 x i8>* noalias sret %{{.*}}, %struct.__va_list_tag* %{{.*}})
@@ -242,4 +369,222 @@ v32i8 va_v32i8(__builtin_va_list l) { return __builtin_va_arg(l, v32i8); }
 // CHECK: [[VA_ARG_ADDR:%[^ ]+]] = phi <32 x i8>** [ [[REG_ADDR]], %{{.*}} ], [ [[MEM_ADDR]], %{{.*}} ]
 // CHECK: [[INDIRECT_ARG:%[^ ]+]] = load <32 x i8>*, <32 x i8>** [[VA_ARG_ADDR]]
 // CHECK: ret void
-
+// CHECK-VECTOR-LABEL: define void @va_v32i8(<32 x i8>* noalias sret %{{.*}}, %struct.__va_list_tag* %{{.*}})
+// CHECK-VECTOR: [[REG_COUNT_PTR:%[^ ]+]] = getelementptr inbounds %struct.__va_list_tag, %struct.__va_list_tag* %{{.*}}, i32 0, i32 0
+// CHECK-VECTOR: [[REG_COUNT:%[^ ]+]] = load i64, i64* [[REG_COUNT_PTR]]
+// CHECK-VECTOR: [[FITS_IN_REGS:%[^ ]+]] = icmp ult i64 [[REG_COUNT]], 5
+// CHECK-VECTOR: br i1 [[FITS_IN_REGS]],
+// CHECK-VECTOR: [[SCALED_REG_COUNT:%[^ ]+]] = mul i64 [[REG_COUNT]], 8
+// CHECK-VECTOR: [[REG_OFFSET:%[^ ]+]] = add i64 [[SCALED_REG_COUNT]], 16
+// CHECK-VECTOR: [[REG_SAVE_AREA_PTR:%[^ ]+]] = getelementptr inbounds %struct.__va_list_tag, %struct.__va_list_tag* %{{.*}}, i32 0, i32 3
+// CHECK-VECTOR: [[REG_SAVE_AREA:%[^ ]+]] = load i8*, i8** [[REG_SAVE_AREA_PTR:[^ ]+]]
+// CHECK-VECTOR: [[RAW_REG_ADDR:%[^ ]+]] = getelementptr i8, i8* [[REG_SAVE_AREA]], i64 [[REG_OFFSET]]
+// CHECK-VECTOR: [[REG_ADDR:%[^ ]+]] = bitcast i8* [[RAW_REG_ADDR]] to <32 x i8>**
+// CHECK-VECTOR: [[REG_COUNT1:%[^ ]+]] = add i64 [[REG_COUNT]], 1
+// CHECK-VECTOR: store i64 [[REG_COUNT1]], i64* [[REG_COUNT_PTR]]
+// CHECK-VECTOR: [[OVERFLOW_ARG_AREA_PTR:%[^ ]+]] = getelementptr inbounds %struct.__va_list_tag, %struct.__va_list_tag* %{{.*}}, i32 0, i32 2
+// CHECK-VECTOR: [[OVERFLOW_ARG_AREA:%[^ ]+]] = load i8*, i8** [[OVERFLOW_ARG_AREA_PTR]]
+// CHECK-VECTOR: [[RAW_MEM_ADDR:%[^ ]+]] = getelementptr i8, i8* [[OVERFLOW_ARG_AREA]], i64 0
+// CHECK-VECTOR: [[MEM_ADDR:%[^ ]+]] = bitcast i8* [[RAW_MEM_ADDR]] to <32 x i8>**
+// CHECK-VECTOR: [[OVERFLOW_ARG_AREA2:%[^ ]+]] = getelementptr i8, i8* [[OVERFLOW_ARG_AREA]], i64 8
+// CHECK-VECTOR: store i8* [[OVERFLOW_ARG_AREA2]], i8** [[OVERFLOW_ARG_AREA_PTR]]
+// CHECK-VECTOR: [[VA_ARG_ADDR:%[^ ]+]] = phi <32 x i8>** [ [[REG_ADDR]], %{{.*}} ], [ [[MEM_ADDR]], %{{.*}} ]
+// CHECK-VECTOR: [[INDIRECT_ARG:%[^ ]+]] = load <32 x i8>*, <32 x i8>** [[VA_ARG_ADDR]]
+// CHECK-VECTOR: ret void
+
+struct agg_v1i8 va_agg_v1i8(__builtin_va_list l) { return __builtin_va_arg(l, struct agg_v1i8); }
+// CHECK-LABEL: define void @va_agg_v1i8(%struct.agg_v1i8* noalias sret %{{.*}}, %struct.__va_list_tag* %{{.*}})
+// CHECK: [[REG_COUNT_PTR:%[^ ]+]] = getelementptr inbounds %struct.__va_list_tag, %struct.__va_list_tag* %{{.*}}, i32 0, i32 0
+// CHECK: [[REG_COUNT:%[^ ]+]] = load i64, i64* [[REG_COUNT_PTR]]
+// CHECK: [[FITS_IN_REGS:%[^ ]+]] = icmp ult i64 [[REG_COUNT]], 5
+// CHECK: br i1 [[FITS_IN_REGS]],
+// CHECK: [[SCALED_REG_COUNT:%[^ ]+]] = mul i64 [[REG_COUNT]], 8
+// CHECK: [[REG_OFFSET:%[^ ]+]] = add i64 [[SCALED_REG_COUNT]], 23
+// CHECK: [[REG_SAVE_AREA_PTR:%[^ ]+]] = getelementptr inbounds %struct.__va_list_tag, %struct.__va_list_tag* %{{.*}}, i32 0, i32 3
+// CHECK: [[REG_SAVE_AREA:%[^ ]+]] = load i8*, i8** [[REG_SAVE_AREA_PTR:[^ ]+]]
+// CHECK: [[RAW_REG_ADDR:%[^ ]+]] = getelementptr i8, i8* [[REG_SAVE_AREA]], i64 [[REG_OFFSET]]
+// CHECK: [[REG_ADDR:%[^ ]+]] = bitcast i8* [[RAW_REG_ADDR]] to %struct.agg_v1i8*
+// CHECK: [[REG_COUNT1:%[^ ]+]] = add i64 [[REG_COUNT]], 1
+// CHECK: store i64 [[REG_COUNT1]], i64* [[REG_COUNT_PTR]]
+// CHECK: [[OVERFLOW_ARG_AREA_PTR:%[^ ]+]] = getelementptr inbounds %struct.__va_list_tag, %struct.__va_list_tag* %{{.*}}, i32 0, i32 2
+// CHECK: [[OVERFLOW_ARG_AREA:%[^ ]+]] = load i8*, i8** [[OVERFLOW_ARG_AREA_PTR]]
+// CHECK: [[RAW_MEM_ADDR:%[^ ]+]] = getelementptr i8, i8* [[OVERFLOW_ARG_AREA]], i64 7
+// CHECK: [[MEM_ADDR:%[^ ]+]] = bitcast i8* [[RAW_MEM_ADDR]] to %struct.agg_v1i8*
+// CHECK: [[OVERFLOW_ARG_AREA2:%[^ ]+]] = getelementptr i8, i8* [[OVERFLOW_ARG_AREA]], i64 8
+// CHECK: store i8* [[OVERFLOW_ARG_AREA2]], i8** [[OVERFLOW_ARG_AREA_PTR]]
+// CHECK: [[VA_ARG_ADDR:%[^ ]+]] = phi %struct.agg_v1i8* [ [[REG_ADDR]], %{{.*}} ], [ [[MEM_ADDR]], %{{.*}} ]
+// CHECK: ret void
+// CHECK-VECTOR-LABEL: define void @va_agg_v1i8(%struct.agg_v1i8* noalias sret %{{.*}}, %struct.__va_list_tag* %{{.*}})
+// CHECK-VECTOR: [[OVERFLOW_ARG_AREA_PTR:%[^ ]+]] = getelementptr inbounds %struct.__va_list_tag, %struct.__va_list_tag* %{{.*}}, i32 0, i32 2
+// CHECK-VECTOR: [[OVERFLOW_ARG_AREA:%[^ ]+]] = load i8*, i8** [[OVERFLOW_ARG_AREA_PTR]]
+// CHECK-VECTOR: [[MEM_ADDR:%[^ ]+]] = bitcast i8* [[OVERFLOW_ARG_AREA]] to %struct.agg_v1i8*
+// CHECK-VECTOR: [[OVERFLOW_ARG_AREA1:%[^ ]+]] = getelementptr i8, i8* [[OVERFLOW_ARG_AREA]], i64 8
+// CHECK-VECTOR: store i8* [[OVERFLOW_ARG_AREA1]], i8** [[OVERFLOW_ARG_AREA_PTR]]
+// CHECK-VECTOR: ret void
+
+struct agg_v2i8 va_agg_v2i8(__builtin_va_list l) { return __builtin_va_arg(l, struct agg_v2i8); }
+// CHECK-LABEL: define void @va_agg_v2i8(%struct.agg_v2i8* noalias sret %{{.*}}, %struct.__va_list_tag* %{{.*}})
+// CHECK: [[REG_COUNT_PTR:%[^ ]+]] = getelementptr inbounds %struct.__va_list_tag, %struct.__va_list_tag* %{{.*}}, i32 0, i32 0
+// CHECK: [[REG_COUNT:%[^ ]+]] = load i64, i64* [[REG_COUNT_PTR]]
+// CHECK: [[FITS_IN_REGS:%[^ ]+]] = icmp ult i64 [[REG_COUNT]], 5
+// CHECK: br i1 [[FITS_IN_REGS]],
+// CHECK: [[SCALED_REG_COUNT:%[^ ]+]] = mul i64 [[REG_COUNT]], 8
+// CHECK: [[REG_OFFSET:%[^ ]+]] = add i64 [[SCALED_REG_COUNT]], 22
+// CHECK: [[REG_SAVE_AREA_PTR:%[^ ]+]] = getelementptr inbounds %struct.__va_list_tag, %struct.__va_list_tag* %{{.*}}, i32 0, i32 3
+// CHECK: [[REG_SAVE_AREA:%[^ ]+]] = load i8*, i8** [[REG_SAVE_AREA_PTR:[^ ]+]]
+// CHECK: [[RAW_REG_ADDR:%[^ ]+]] = getelementptr i8, i8* [[REG_SAVE_AREA]], i64 [[REG_OFFSET]]
+// CHECK: [[REG_ADDR:%[^ ]+]] = bitcast i8* [[RAW_REG_ADDR]] to %struct.agg_v2i8*
+// CHECK: [[REG_COUNT1:%[^ ]+]] = add i64 [[REG_COUNT]], 1
+// CHECK: store i64 [[REG_COUNT1]], i64* [[REG_COUNT_PTR]]
+// CHECK: [[OVERFLOW_ARG_AREA_PTR:%[^ ]+]] = getelementptr inbounds %struct.__va_list_tag, %struct.__va_list_tag* %{{.*}}, i32 0, i32 2
+// CHECK: [[OVERFLOW_ARG_AREA:%[^ ]+]] = load i8*, i8** [[OVERFLOW_ARG_AREA_PTR]]
+// CHECK: [[RAW_MEM_ADDR:%[^ ]+]] = getelementptr i8, i8* [[OVERFLOW_ARG_AREA]], i64 6
+// CHECK: [[MEM_ADDR:%[^ ]+]] = bitcast i8* [[RAW_MEM_ADDR]] to %struct.agg_v2i8*
+// CHECK: [[OVERFLOW_ARG_AREA2:%[^ ]+]] = getelementptr i8, i8* [[OVERFLOW_ARG_AREA]], i64 8
+// CHECK: store i8* [[OVERFLOW_ARG_AREA2]], i8** [[OVERFLOW_ARG_AREA_PTR]]
+// CHECK: [[VA_ARG_ADDR:%[^ ]+]] = phi %struct.agg_v2i8* [ [[REG_ADDR]], %{{.*}} ], [ [[MEM_ADDR]], %{{.*}} ]
+// CHECK: ret void
+// CHECK-VECTOR-LABEL: define void @va_agg_v2i8(%struct.agg_v2i8* noalias sret %{{.*}}, %struct.__va_list_tag* %{{.*}})
+// CHECK-VECTOR: [[OVERFLOW_ARG_AREA_PTR:%[^ ]+]] = getelementptr inbounds %struct.__va_list_tag, %struct.__va_list_tag* %{{.*}}, i32 0, i32 2
+// CHECK-VECTOR: [[OVERFLOW_ARG_AREA:%[^ ]+]] = load i8*, i8** [[OVERFLOW_ARG_AREA_PTR]]
+// CHECK-VECTOR: [[MEM_ADDR:%[^ ]+]] = bitcast i8* [[OVERFLOW_ARG_AREA]] to %struct.agg_v2i8*
+// CHECK-VECTOR: [[OVERFLOW_ARG_AREA1:%[^ ]+]] = getelementptr i8, i8* [[OVERFLOW_ARG_AREA]], i64 8
+// CHECK-VECTOR: store i8* [[OVERFLOW_ARG_AREA1]], i8** [[OVERFLOW_ARG_AREA_PTR]]
+// CHECK-VECTOR: ret void
+
+struct agg_v4i8 va_agg_v4i8(__builtin_va_list l) { return __builtin_va_arg(l, struct agg_v4i8); }
+// CHECK-LABEL: define void @va_agg_v4i8(%struct.agg_v4i8* noalias sret %{{.*}}, %struct.__va_list_tag* %{{.*}})
+// CHECK: [[REG_COUNT_PTR:%[^ ]+]] = getelementptr inbounds %struct.__va_list_tag, %struct.__va_list_tag* %{{.*}}, i32 0, i32 0
+// CHECK: [[REG_COUNT:%[^ ]+]] = load i64, i64* [[REG_COUNT_PTR]]
+// CHECK: [[FITS_IN_REGS:%[^ ]+]] = icmp ult i64 [[REG_COUNT]], 5
+// CHECK: br i1 [[FITS_IN_REGS]],
+// CHECK: [[SCALED_REG_COUNT:%[^ ]+]] = mul i64 [[REG_COUNT]], 8
+// CHECK: [[REG_OFFSET:%[^ ]+]] = add i64 [[SCALED_REG_COUNT]], 20
+// CHECK: [[REG_SAVE_AREA_PTR:%[^ ]+]] = getelementptr inbounds %struct.__va_list_tag, %struct.__va_list_tag* %{{.*}}, i32 0, i32 3
+// CHECK: [[REG_SAVE_AREA:%[^ ]+]] = load i8*, i8** [[REG_SAVE_AREA_PTR:[^ ]+]]
+// CHECK: [[RAW_REG_ADDR:%[^ ]+]] = getelementptr i8, i8* [[REG_SAVE_AREA]], i64 [[REG_OFFSET]]
+// CHECK: [[REG_ADDR:%[^ ]+]] = bitcast i8* [[RAW_REG_ADDR]] to %struct.agg_v4i8*
+// CHECK: [[REG_COUNT1:%[^ ]+]] = add i64 [[REG_COUNT]], 1
+// CHECK: store i64 [[REG_COUNT1]], i64* [[REG_COUNT_PTR]]
+// CHECK: [[OVERFLOW_ARG_AREA_PTR:%[^ ]+]] = getelementptr inbounds %struct.__va_list_tag, %struct.__va_list_tag* %{{.*}}, i32 0, i32 2
+// CHECK: [[OVERFLOW_ARG_AREA:%[^ ]+]] = load i8*, i8** [[OVERFLOW_ARG_AREA_PTR]]
+// CHECK: [[RAW_MEM_ADDR:%[^ ]+]] = getelementptr i8, i8* [[OVERFLOW_ARG_AREA]], i64 4
+// CHECK: [[MEM_ADDR:%[^ ]+]] = bitcast i8* [[RAW_MEM_ADDR]] to %struct.agg_v4i8*
+// CHECK: [[OVERFLOW_ARG_AREA2:%[^ ]+]] = getelementptr i8, i8* [[OVERFLOW_ARG_AREA]], i64 8
+// CHECK: store i8* [[OVERFLOW_ARG_AREA2]], i8** [[OVERFLOW_ARG_AREA_PTR]]
+// CHECK: [[VA_ARG_ADDR:%[^ ]+]] = phi %struct.agg_v4i8* [ [[REG_ADDR]], %{{.*}} ], [ [[MEM_ADDR]], %{{.*}} ]
+// CHECK: ret void
+// CHECK-VECTOR-LABEL: define void @va_agg_v4i8(%struct.agg_v4i8* noalias sret %{{.*}}, %struct.__va_list_tag* %{{.*}})
+// CHECK-VECTOR: [[OVERFLOW_ARG_AREA_PTR:%[^ ]+]] = getelementptr inbounds %struct.__va_list_tag, %struct.__va_list_tag* %{{.*}}, i32 0, i32 2
+// CHECK-VECTOR: [[OVERFLOW_ARG_AREA:%[^ ]+]] = load i8*, i8** [[OVERFLOW_ARG_AREA_PTR]]
+// CHECK-VECTOR: [[MEM_ADDR:%[^ ]+]] = bitcast i8* [[OVERFLOW_ARG_AREA]] to %struct.agg_v4i8*
+// CHECK-VECTOR: [[OVERFLOW_ARG_AREA1:%[^ ]+]] = getelementptr i8, i8* [[OVERFLOW_ARG_AREA]], i64 8
+// CHECK-VECTOR: store i8* [[OVERFLOW_ARG_AREA1]], i8** [[OVERFLOW_ARG_AREA_PTR]]
+// CHECK-VECTOR: ret void
+
+struct agg_v8i8 va_agg_v8i8(__builtin_va_list l) { return __builtin_va_arg(l, struct agg_v8i8); }
+// CHECK-LABEL: define void @va_agg_v8i8(%struct.agg_v8i8* noalias sret %{{.*}}, %struct.__va_list_tag* %{{.*}})
+// CHECK: [[REG_COUNT_PTR:%[^ ]+]] = getelementptr inbounds %struct.__va_list_tag, %struct.__va_list_tag* %{{.*}}, i32 0, i32 0
+// CHECK: [[REG_COUNT:%[^ ]+]] = load i64, i64* [[REG_COUNT_PTR]]
+// CHECK: [[FITS_IN_REGS:%[^ ]+]] = icmp ult i64 [[REG_COUNT]], 5
+// CHECK: br i1 [[FITS_IN_REGS]],
+// CHECK: [[SCALED_REG_COUNT:%[^ ]+]] = mul i64 [[REG_COUNT]], 8
+// CHECK: [[REG_OFFSET:%[^ ]+]] = add i64 [[SCALED_REG_COUNT]], 16
+// CHECK: [[REG_SAVE_AREA_PTR:%[^ ]+]] = getelementptr inbounds %struct.__va_list_tag, %struct.__va_list_tag* %{{.*}}, i32 0, i32 3
+// CHECK: [[REG_SAVE_AREA:%[^ ]+]] = load i8*, i8** [[REG_SAVE_AREA_PTR:[^ ]+]]
+// CHECK: [[RAW_REG_ADDR:%[^ ]+]] = getelementptr i8, i8* [[REG_SAVE_AREA]], i64 [[REG_OFFSET]]
+// CHECK: [[REG_ADDR:%[^ ]+]] = bitcast i8* [[RAW_REG_ADDR]] to %struct.agg_v8i8*
+// CHECK: [[REG_COUNT1:%[^ ]+]] = add i64 [[REG_COUNT]], 1
+// CHECK: store i64 [[REG_COUNT1]], i64* [[REG_COUNT_PTR]]
+// CHECK: [[OVERFLOW_ARG_AREA_PTR:%[^ ]+]] = getelementptr inbounds %struct.__va_list_tag, %struct.__va_list_tag* %{{.*}}, i32 0, i32 2
+// CHECK: [[OVERFLOW_ARG_AREA:%[^ ]+]] = load i8*, i8** [[OVERFLOW_ARG_AREA_PTR]]
+// CHECK: [[RAW_MEM_ADDR:%[^ ]+]] = getelementptr i8, i8* [[OVERFLOW_ARG_AREA]], i64 0
+// CHECK: [[MEM_ADDR:%[^ ]+]] = bitcast i8* [[RAW_MEM_ADDR]] to %struct.agg_v8i8*
+// CHECK: [[OVERFLOW_ARG_AREA2:%[^ ]+]] = getelementptr i8, i8* [[OVERFLOW_ARG_AREA]], i64 8
+// CHECK: store i8* [[OVERFLOW_ARG_AREA2]], i8** [[OVERFLOW_ARG_AREA_PTR]]
+// CHECK: [[VA_ARG_ADDR:%[^ ]+]] = phi %struct.agg_v8i8* [ [[REG_ADDR]], %{{.*}} ], [ [[MEM_ADDR]], %{{.*}} ]
+// CHECK: ret void
+// CHECK-VECTOR-LABEL: define void @va_agg_v8i8(%struct.agg_v8i8* noalias sret %{{.*}}, %struct.__va_list_tag* %{{.*}})
+// CHECK-VECTOR: [[OVERFLOW_ARG_AREA_PTR:%[^ ]+]] = getelementptr inbounds %struct.__va_list_tag, %struct.__va_list_tag* %{{.*}}, i32 0, i32 2
+// CHECK-VECTOR: [[OVERFLOW_ARG_AREA:%[^ ]+]] = load i8*, i8** [[OVERFLOW_ARG_AREA_PTR]]
+// CHECK-VECTOR: [[MEM_ADDR:%[^ ]+]] = bitcast i8* [[OVERFLOW_ARG_AREA]] to %struct.agg_v8i8*
+// CHECK-VECTOR: [[OVERFLOW_ARG_AREA1:%[^ ]+]] = getelementptr i8, i8* [[OVERFLOW_ARG_AREA]], i64 8
+// CHECK-VECTOR: store i8* [[OVERFLOW_ARG_AREA1]], i8** [[OVERFLOW_ARG_AREA_PTR]]
+// CHECK-VECTOR: ret void
+
+struct agg_v16i8 va_agg_v16i8(__builtin_va_list l) { return __builtin_va_arg(l, struct agg_v16i8); }
+// CHECK-LABEL: define void @va_agg_v16i8(%struct.agg_v16i8* noalias sret %{{.*}}, %struct.__va_list_tag* %{{.*}})
+// CHECK: [[REG_COUNT_PTR:%[^ ]+]] = getelementptr inbounds %struct.__va_list_tag, %struct.__va_list_tag* %{{.*}}, i32 0, i32 0
+// CHECK: [[REG_COUNT:%[^ ]+]] = load i64, i64* [[REG_COUNT_PTR]]
+// CHECK: [[FITS_IN_REGS:%[^ ]+]] = icmp ult i64 [[REG_COUNT]], 5
+// CHECK: br i1 [[FITS_IN_REGS]],
+// CHECK: [[SCALED_REG_COUNT:%[^ ]+]] = mul i64 [[REG_COUNT]], 8
+// CHECK: [[REG_OFFSET:%[^ ]+]] = add i64 [[SCALED_REG_COUNT]], 16
+// CHECK: [[REG_SAVE_AREA_PTR:%[^ ]+]] = getelementptr inbounds %struct.__va_list_tag, %struct.__va_list_tag* %{{.*}}, i32 0, i32 3
+// CHECK: [[REG_SAVE_AREA:%[^ ]+]] = load i8*, i8** [[REG_SAVE_AREA_PTR:[^ ]+]]
+// CHECK: [[RAW_REG_ADDR:%[^ ]+]] = getelementptr i8, i8* [[REG_SAVE_AREA]], i64 [[REG_OFFSET]]
+// CHECK: [[REG_ADDR:%[^ ]+]] = bitcast i8* [[RAW_REG_ADDR]] to %struct.agg_v16i8**
+// CHECK: [[REG_COUNT1:%[^ ]+]] = add i64 [[REG_COUNT]], 1
+// CHECK: store i64 [[REG_COUNT1]], i64* [[REG_COUNT_PTR]]
+// CHECK: [[OVERFLOW_ARG_AREA_PTR:%[^ ]+]] = getelementptr inbounds %struct.__va_list_tag, %struct.__va_list_tag* %{{.*}}, i32 0, i32 2
+// CHECK: [[OVERFLOW_ARG_AREA:%[^ ]+]] = load i8*, i8** [[OVERFLOW_ARG_AREA_PTR]]
+// CHECK: [[RAW_MEM_ADDR:%[^ ]+]] = getelementptr i8, i8* [[OVERFLOW_ARG_AREA]], i64 0
+// CHECK: [[MEM_ADDR:%[^ ]+]] = bitcast i8* [[RAW_MEM_ADDR]] to %struct.agg_v16i8**
+// CHECK: [[OVERFLOW_ARG_AREA2:%[^ ]+]] = getelementptr i8, i8* [[OVERFLOW_ARG_AREA]], i64 8
+// CHECK: store i8* [[OVERFLOW_ARG_AREA2]], i8** [[OVERFLOW_ARG_AREA_PTR]]
+// CHECK: [[VA_ARG_ADDR:%[^ ]+]] = phi %struct.agg_v16i8** [ [[REG_ADDR]], %{{.*}} ], [ [[MEM_ADDR]], %{{.*}} ]
+// CHECK: [[INDIRECT_ARG:%[^ ]+]] = load %struct.agg_v16i8*, %struct.agg_v16i8** [[VA_ARG_ADDR]]
+// CHECK: ret void
+// CHECK-VECTOR-LABEL: define void @va_agg_v16i8(%struct.agg_v16i8* noalias sret %{{.*}}, %struct.__va_list_tag* %{{.*}})
+// CHECK-VECTOR: [[OVERFLOW_ARG_AREA_PTR:%[^ ]+]] = getelementptr inbounds %struct.__va_list_tag, %struct.__va_list_tag* %{{.*}}, i32 0, i32 2
+// CHECK-VECTOR: [[OVERFLOW_ARG_AREA:%[^ ]+]] = load i8*, i8** [[OVERFLOW_ARG_AREA_PTR]]
+// CHECK-VECTOR: [[MEM_ADDR:%[^ ]+]] = bitcast i8* [[OVERFLOW_ARG_AREA]] to %struct.agg_v16i8*
+// CHECK-VECTOR: [[OVERFLOW_ARG_AREA1:%[^ ]+]] = getelementptr i8, i8* [[OVERFLOW_ARG_AREA]], i64 16
+// CHECK-VECTOR: store i8* [[OVERFLOW_ARG_AREA1]], i8** [[OVERFLOW_ARG_AREA_PTR]]
+// CHECK-VECTOR: ret void
+
+struct agg_v32i8 va_agg_v32i8(__builtin_va_list l) { return __builtin_va_arg(l, struct agg_v32i8); }
+// CHECK-LABEL: define void @va_agg_v32i8(%struct.agg_v32i8* noalias sret %{{.*}}, %struct.__va_list_tag* %{{.*}})
+// CHECK: [[REG_COUNT_PTR:%[^ ]+]] = getelementptr inbounds %struct.__va_list_tag, %struct.__va_list_tag* %{{.*}}, i32 0, i32 0
+// CHECK: [[REG_COUNT:%[^ ]+]] = load i64, i64* [[REG_COUNT_PTR]]
+// CHECK: [[FITS_IN_REGS:%[^ ]+]] = icmp ult i64 [[REG_COUNT]], 5
+// CHECK: br i1 [[FITS_IN_REGS]],
+// CHECK: [[SCALED_REG_COUNT:%[^ ]+]] = mul i64 [[REG_COUNT]], 8
+// CHECK: [[REG_OFFSET:%[^ ]+]] = add i64 [[SCALED_REG_COUNT]], 16
+// CHECK: [[REG_SAVE_AREA_PTR:%[^ ]+]] = getelementptr inbounds %struct.__va_list_tag, %struct.__va_list_tag* %{{.*}}, i32 0, i32 3
+// CHECK: [[REG_SAVE_AREA:%[^ ]+]] = load i8*, i8** [[REG_SAVE_AREA_PTR:[^ ]+]]
+// CHECK: [[RAW_REG_ADDR:%[^ ]+]] = getelementptr i8, i8* [[REG_SAVE_AREA]], i64 [[REG_OFFSET]]
+// CHECK: [[REG_ADDR:%[^ ]+]] = bitcast i8* [[RAW_REG_ADDR]] to %struct.agg_v32i8**
+// CHECK: [[REG_COUNT1:%[^ ]+]] = add i64 [[REG_COUNT]], 1
+// CHECK: store i64 [[REG_COUNT1]], i64* [[REG_COUNT_PTR]]
+// CHECK: [[OVERFLOW_ARG_AREA_PTR:%[^ ]+]] = getelementptr inbounds %struct.__va_list_tag, %struct.__va_list_tag* %{{.*}}, i32 0, i32 2
+// CHECK: [[OVERFLOW_ARG_AREA:%[^ ]+]] = load i8*, i8** [[OVERFLOW_ARG_AREA_PTR]]
+// CHECK: [[RAW_MEM_ADDR:%[^ ]+]] = getelementptr i8, i8* [[OVERFLOW_ARG_AREA]], i64 0
+// CHECK: [[MEM_ADDR:%[^ ]+]] = bitcast i8* [[RAW_MEM_ADDR]] to %struct.agg_v32i8**
+// CHECK: [[OVERFLOW_ARG_AREA2:%[^ ]+]] = getelementptr i8, i8* [[OVERFLOW_ARG_AREA]], i64 8
+// CHECK: store i8* [[OVERFLOW_ARG_AREA2]], i8** [[OVERFLOW_ARG_AREA_PTR]]
+// CHECK: [[VA_ARG_ADDR:%[^ ]+]] = phi %struct.agg_v32i8** [ [[REG_ADDR]], %{{.*}} ], [ [[MEM_ADDR]], %{{.*}} ]
+// CHECK: [[INDIRECT_ARG:%[^ ]+]] = load %struct.agg_v32i8*, %struct.agg_v32i8** [[VA_ARG_ADDR]]
+// CHECK: ret void
+// CHECK-VECTOR-LABEL: define void @va_agg_v32i8(%struct.agg_v32i8* noalias sret %{{.*}}, %struct.__va_list_tag* %{{.*}})
+// CHECK-VECTOR: [[REG_COUNT_PTR:%[^ ]+]] = getelementptr inbounds %struct.__va_list_tag, %struct.__va_list_tag* %{{.*}}, i32 0, i32 0
+// CHECK-VECTOR: [[REG_COUNT:%[^ ]+]] = load i64, i64* [[REG_COUNT_PTR]]
+// CHECK-VECTOR: [[FITS_IN_REGS:%[^ ]+]] = icmp ult i64 [[REG_COUNT]], 5
+// CHECK-VECTOR: br i1 [[FITS_IN_REGS]],
+// CHECK-VECTOR: [[SCALED_REG_COUNT:%[^ ]+]] = mul i64 [[REG_COUNT]], 8
+// CHECK-VECTOR: [[REG_OFFSET:%[^ ]+]] = add i64 [[SCALED_REG_COUNT]], 16
+// CHECK-VECTOR: [[REG_SAVE_AREA_PTR:%[^ ]+]] = getelementptr inbounds %struct.__va_list_tag, %struct.__va_list_tag* %{{.*}}, i32 0, i32 3
+// CHECK-VECTOR: [[REG_SAVE_AREA:%[^ ]+]] = load i8*, i8** [[REG_SAVE_AREA_PTR:[^ ]+]]
+// CHECK-VECTOR: [[RAW_REG_ADDR:%[^ ]+]] = getelementptr i8, i8* [[REG_SAVE_AREA]], i64 [[REG_OFFSET]]
+// CHECK-VECTOR: [[REG_ADDR:%[^ ]+]] = bitcast i8* [[RAW_REG_ADDR]] to %struct.agg_v32i8**
+// CHECK-VECTOR: [[REG_COUNT1:%[^ ]+]] = add i64 [[REG_COUNT]], 1
+// CHECK-VECTOR: store i64 [[REG_COUNT1]], i64* [[REG_COUNT_PTR]]
+// CHECK-VECTOR: [[OVERFLOW_ARG_AREA_PTR:%[^ ]+]] = getelementptr inbounds %struct.__va_list_tag, %struct.__va_list_tag* %{{.*}}, i32 0, i32 2
+// CHECK-VECTOR: [[OVERFLOW_ARG_AREA:%[^ ]+]] = load i8*, i8** [[OVERFLOW_ARG_AREA_PTR]]
+// CHECK-VECTOR: [[RAW_MEM_ADDR:%[^ ]+]] = getelementptr i8, i8* [[OVERFLOW_ARG_AREA]], i64 0
+// CHECK-VECTOR: [[MEM_ADDR:%[^ ]+]] = bitcast i8* [[RAW_MEM_ADDR]] to %struct.agg_v32i8**
+// CHECK-VECTOR: [[OVERFLOW_ARG_AREA2:%[^ ]+]] = getelementptr i8, i8* [[OVERFLOW_ARG_AREA]], i64 8
+// CHECK-VECTOR: store i8* [[OVERFLOW_ARG_AREA2]], i8** [[OVERFLOW_ARG_AREA_PTR]]
+// CHECK-VECTOR: [[VA_ARG_ADDR:%[^ ]+]] = phi %struct.agg_v32i8** [ [[REG_ADDR]], %{{.*}} ], [ [[MEM_ADDR]], %{{.*}} ]
+// CHECK-VECTOR: [[INDIRECT_ARG:%[^ ]+]] = load %struct.agg_v32i8*, %struct.agg_v32i8** [[VA_ARG_ADDR]]
+// CHECK-VECTOR: ret void

diff --git a/test/CodeGen/systemz-abi.c b/test/CodeGen/systemz-abi.c
index 5599c79e90056d3468188d539dcbdbf49a26f4a5..375d02a35889aba3d64c622ca1e5ecec1c5c619c 100644 (file)
--- a/test/CodeGen/systemz-abi.c
+++ b/test/CodeGen/systemz-abi.c
@@ -1,4 +1,9 @@
-// RUN: %clang_cc1 -triple s390x-linux-gnu -emit-llvm -o - %s | FileCheck %s
+// RUN: %clang_cc1 -triple s390x-linux-gnu \
+// RUN:   -emit-llvm -o - %s | FileCheck %s
+// RUN: %clang_cc1 -triple s390x-linux-gnu -target-feature +vector \
+// RUN:   -emit-llvm -o - %s | FileCheck %s
+// RUN: %clang_cc1 -triple s390x-linux-gnu -target-cpu z13 \
+// RUN:   -emit-llvm -o - %s | FileCheck %s
 
 // Scalar types
 

diff --git a/test/CodeGen/target-data.c b/test/CodeGen/target-data.c
index c3e3852bccd7524a45bbf27c5c97f70a3f0b730a..d601158af63a7152b4853889dc7766a590ebc6ab 100644 (file)
--- a/test/CodeGen/target-data.c
+++ b/test/CodeGen/target-data.c
@@ -155,6 +155,10 @@
 // RUN: FileCheck %s -check-prefix=SYSTEMZ
 // SYSTEMZ: target datalayout = "E-m:e-i1:8:16-i8:8:16-i64:64-f128:64-a:8:16-n32:64"
 
+// RUN: %clang_cc1 -triple s390x-unknown -target-cpu z13 -o - -emit-llvm %s | \
+// RUN: FileCheck %s -check-prefix=SYSTEMZ-VECTOR
+// SYSTEMZ-VECTOR: target datalayout = "E-m:e-i1:8:16-i8:8:16-i64:64-f128:64-v128:64-a:8:16-n32:64"
+
 // RUN: %clang_cc1 -triple msp430-unknown -o - -emit-llvm %s | \
 // RUN: FileCheck %s -check-prefix=MSP430
 // MSP430: target datalayout = "e-m:e-p:16:16-i32:16:32-a:16-n8:16"

diff --git a/test/Driver/systemz-features.cpp b/test/Driver/systemz-features.cpp
index 25937d1cdce8ccbda0246292eec7f13721a1dabc..be1818a032355cd641693aa7954799439a125bd8 100644 (file)
--- a/test/Driver/systemz-features.cpp
+++ b/test/Driver/systemz-features.cpp
@@ -2,6 +2,8 @@
 // RUN: %clang -target s390x-unknown-linux-gnu %s -### -o %t.o 2>&1 | FileCheck -check-prefix=CHECK-DEFAULT %s
 // CHECK-DEFAULT-NOT: "-target-feature" "+transactional-execution"
 // CHECK-DEFAULT-NOT: "-target-feature" "-transactional-execution"
+// CHECK-DEFAULT-NOT: "-target-feature" "+vector"
+// CHECK-DEFAULT-NOT: "-target-feature" "-vector"
 
 // RUN: %clang -target s390x-unknown-linux-gnu %s -mhtm -### -o %t.o 2>&1 | FileCheck -check-prefix=CHECK-HTM %s
 // RUN: %clang -target s390x-unknown-linux-gnu %s -mno-htm -mhtm -### -o %t.o 2>&1 | FileCheck -check-prefix=CHECK-HTM %s
@@ -13,3 +15,12 @@
 // CHECK-NOHTM: "-target-feature" "-transactional-execution"
 // CHECK-NOHTM-NOT: "-target-feature" "+transactional-execution"
 
+// RUN: %clang -target s390x-unknown-linux-gnu %s -mvx -### -o %t.o 2>&1 | FileCheck -check-prefix=CHECK-VX %s
+// RUN: %clang -target s390x-unknown-linux-gnu %s -mno-vx -mvx -### -o %t.o 2>&1 | FileCheck -check-prefix=CHECK-VX %s
+// CHECK-VX: "-target-feature" "+vector"
+// CHECK-VX-NOT: "-target-feature" "-vector"
+//
+// RUN: %clang -target s390x-unknown-linux-gnu %s -mno-vx -### -o %t.o 2>&1 | FileCheck -check-prefix=CHECK-NOVX %s
+// RUN: %clang -target s390x-unknown-linux-gnu %s -mvx -mno-vx -### -o %t.o 2>&1 | FileCheck -check-prefix=CHECK-NOVX %s
+// CHECK-NOVX: "-target-feature" "-vector"
+// CHECK-NOVX-NOT: "-target-feature" "+vector"