[ARM] Use AEABI aligned function variants

[john.brawn]

Summary

AEABI defines aligned variants of memcpy etc. that can be faster than the default version due to not having to do alignment checks. When emitting target code for these functions make use of these aligned variants if possible. Also convert memset to memclr if possible.

[john.brawn]

It may be worth noting that with glibc these functions may be no
better and in some cases slower than the standard ISO C functions.

Trunk glibc aliases aeabi_memcpy4 to aeabi_memcpy (and similarly
with the other functions) so it should be no slower as LLVM already
translates memcpy to __aeabi_memcpy for AEABI targets through the
generic Libcall lowering mechanism.

John

[john.brawn]

New patch changes the decision of whether to do this to check what getLibcallName says the default function name is instead of duplicating the logic in ARMTargetLowering in case that logic changes.

[john.brawn]

New patch that adjusts the tests to cope with changes since the last patch.

[asl]

Otherwise LGTM. Renato, what do you think?

lib/Target/ARM/ARMSelectionDAGInfo.cpp

36

Minor nit: put brace on the preceding line

43

Should we start with cheaper ABI check here and bail out early for non-AAPCS?

[rengolin]

LGTM, Thanks!

[john.brawn]

lib/Target/ARM/ARMSelectionDAGInfo.cpp

36

Will do.

43

I'm not sure what you mean by this?

[john.brawn]

Move curly brace position.

[john.brawn]

Ping.

[john.brawn]

Ping x 2.

[jroelofs]

lib/Target/ARM/ARMSelectionDAGInfo.cpp

78

No need for '{'s on single-statement if/else-if/else blocks.

104

If it's already MVT::i32, don't generate the zext?

[john.brawn]

Remove unnecessary curly braces, only zext to i32 if < i32.

[jroelofs]

lib/Target/ARM/ARMSelectionDAGInfo.cpp

42

I think @asl is suggesting:

if (!Subtarget.isAAPCS_ABI() ||
    std::strncmp(TLI->getLibcallName(LC), "__aeabi", 7) != 0)

here.

[john.brawn]

lib/Target/ARM/ARMSelectionDAGInfo.cpp

42

Ah, I see. Seems somewhat redundant to me (we would only get an aeabi function if the subtarget is already aapcs), and it's not like this is a time-critical function where avoiding a strncmp may be a good idea. I'll leave it as-is.

lib/Target/ARM/ARMSelectionDAGInfo.h

public:
   SDValue EmitTargetCodeForMemcpy(SelectionDAG &DAG, SDLoc dl,
                                   SDValue Chain,
                                   SDValue Dst, SDValue Src,
                                   SDValue Size, unsigned Align,
                                   bool isVolatile, bool AlwaysInline,
                                   MachinePointerInfo DstPtrInfo,
                                   MachinePointerInfo SrcPtrInfo) const override;
 
+  SDValue EmitTargetCodeForMemmove(SelectionDAG &DAG, SDLoc dl,
+                                   SDValue Chain,
+                                   SDValue Dst, SDValue Src,
+                                   SDValue Size, unsigned Align, bool isVolatile,
+                                   MachinePointerInfo DstPtrInfo,
+                                   MachinePointerInfo SrcPtrInfo) const override;
+
   // Adjust parameters for memset, see RTABI section 4.3.4
   SDValue EmitTargetCodeForMemset(SelectionDAG &DAG, SDLoc dl,
                                   SDValue Chain,
                                   SDValue Op1, SDValue Op2,
                                   SDValue Op3, unsigned Align,
                                   bool isVolatile,
                                   MachinePointerInfo DstPtrInfo) const override;
+
+  SDValue EmitSpecializedLibcall(SelectionDAG &DAG, SDLoc dl,
+                                 SDValue Chain,
+                                 SDValue Dst, SDValue Src,
+                                 SDValue Size, unsigned Align,
+                                 RTLIB::Libcall LC) const;
 };
 
 }
 
 #endif

lib/Target/ARM/ARMSelectionDAGInfo.cpp

 #define DEBUG_TYPE "arm-selectiondag-info"
 
 ARMSelectionDAGInfo::ARMSelectionDAGInfo(const DataLayout &DL)
     : TargetSelectionDAGInfo(&DL) {}
 
 ARMSelectionDAGInfo::~ARMSelectionDAGInfo() {
 }
 
+// Emit, if possible, a specialized version of the given Libcall. Typically this
+// means selecting the appropriately aligned version, but we also convert memset
+// of 0 into memclr.
+SDValue ARMSelectionDAGInfo::
+EmitSpecializedLibcall(SelectionDAG &DAG, SDLoc dl,
+                       SDValue Chain,
+                       SDValue Dst, SDValue Src,
+                       SDValue Size, unsigned Align,
+                       RTLIB::Libcall LC) const {
+  const ARMSubtarget &Subtarget =
+      DAG.getMachineFunction().getSubtarget<ARMSubtarget>();
+  const ARMTargetLowering *TLI = Subtarget.getTargetLowering();
+
+  // Only use a specialized AEABI function if the default version of this
+  // Libcall is an AEABI function.
+  if (std::strncmp(TLI->getLibcallName(LC), "__aeabi", 7) != 0)
+    return SDValue();
+
+  // Translate RTLIB::Libcall to AEABILibcall. We only do this in order to be
+  // able to translate memset to memclr and use the value to index the function
+  // name array.
+  enum {
+    AEABI_MEMCPY = 0,
+    AEABI_MEMMOVE,
+    AEABI_MEMSET,
+    AEABI_MEMCLR
+  } AEABILibcall;
+  switch (LC) {
+  case RTLIB::MEMCPY:
+    AEABILibcall = AEABI_MEMCPY;
+    break;
+  case RTLIB::MEMMOVE:
+    AEABILibcall = AEABI_MEMMOVE;
+    break;
+  case RTLIB::MEMSET: 
+    AEABILibcall = AEABI_MEMSET;
+    if (ConstantSDNode *ConstantSrc = dyn_cast<ConstantSDNode>(Src))
+      if (ConstantSrc->getZExtValue() == 0)
+        AEABILibcall = AEABI_MEMCLR;
+    break;
+  default:
+    return SDValue();
+  }
+
+  // Choose the most-aligned libcall variant that we can
+  enum {
+    ALIGN1 = 0,
+    ALIGN4,
+    ALIGN8
+  } AlignVariant;
+  if ((Align & 7) == 0)
+    AlignVariant = ALIGN8;
+  else if ((Align & 3) == 0)
+    AlignVariant = ALIGN4;
+  else
+    AlignVariant = ALIGN1;
+
+  TargetLowering::ArgListTy Args;
+  TargetLowering::ArgListEntry Entry;
+  Entry.Ty = TLI->getDataLayout()->getIntPtrType(*DAG.getContext());
+  Entry.Node = Dst;
+  Args.push_back(Entry);
+  if (AEABILibcall == AEABI_MEMCLR) {
+    Entry.Node = Size;
+    Args.push_back(Entry);
+  } else if (AEABILibcall == AEABI_MEMSET) {
+    // Adjust parameters for memset, EABI uses format (ptr, size, value),
+    // GNU library uses (ptr, value, size)
+    // See RTABI section 4.3.4
+    Entry.Node = Size;
+    Args.push_back(Entry);
+
+    // Extend or truncate the argument to be an i32 value for the call.
+    if (Src.getValueType().bitsGT(MVT::i32))
+      Src = DAG.getNode(ISD::TRUNCATE, dl, MVT::i32, Src);
+    else if (Src.getValueType().bitsLT(MVT::i32))
+      Src = DAG.getNode(ISD::ZERO_EXTEND, dl, MVT::i32, Src);
+
+    Entry.Node = Src; 
+    Entry.Ty = Type::getInt32Ty(*DAG.getContext());
+    Entry.isSExt = false;
+    Args.push_back(Entry);
+  } else {
+    Entry.Node = Src;
+    Args.push_back(Entry);
+    
+    Entry.Node = Size;
+    Args.push_back(Entry);
+  }
+
+  char const *FunctionNames[4][3] = {
+    { "__aeabi_memcpy",  "__aeabi_memcpy4",  "__aeabi_memcpy8"  },
+    { "__aeabi_memmove", "__aeabi_memmove4", "__aeabi_memmove8" },
+    { "__aeabi_memset",  "__aeabi_memset4",  "__aeabi_memset8"  },
+    { "__aeabi_memclr",  "__aeabi_memclr4",  "__aeabi_memclr8"  }
+  };
+  TargetLowering::CallLoweringInfo CLI(DAG);
+  CLI.setDebugLoc(dl).setChain(Chain)
+    .setCallee(TLI->getLibcallCallingConv(LC),
+               Type::getVoidTy(*DAG.getContext()),
+               DAG.getExternalSymbol(FunctionNames[AEABILibcall][AlignVariant],
+                                     TLI->getPointerTy()), std::move(Args), 0)
+    .setDiscardResult();
+  std::pair<SDValue,SDValue> CallResult = TLI->LowerCallTo(CLI);
+  
+  return CallResult.second;
+}
+
 SDValue
 ARMSelectionDAGInfo::EmitTargetCodeForMemcpy(SelectionDAG &DAG, SDLoc dl,
                                              SDValue Chain,
                                              SDValue Dst, SDValue Src,
                                              SDValue Size, unsigned Align,
                                              bool isVolatile, bool AlwaysInline,
                                              MachinePointerInfo DstPtrInfo,
                                           MachinePointerInfo SrcPtrInfo) const {
   const ARMSubtarget &Subtarget =
       DAG.getMachineFunction().getSubtarget<ARMSubtarget>();
   // Do repeated 4-byte loads and stores. To be improved.
   // This requires 4-byte alignment.
   if ((Align & 3) != 0)
     return SDValue();
   // This requires the copy size to be a constant, preferably
   // within a subtarget-specific limit.
   ConstantSDNode *ConstantSize = dyn_cast<ConstantSDNode>(Size);
   if (!ConstantSize)
-    return SDValue();
+    return EmitSpecializedLibcall(DAG, dl, Chain, Dst, Src, Size, Align,
+                                  RTLIB::MEMCPY);
   uint64_t SizeVal = ConstantSize->getZExtValue();
   if (!AlwaysInline && SizeVal > Subtarget.getMaxInlineSizeThreshold())
-    return SDValue();
+    return EmitSpecializedLibcall(DAG, dl, Chain, Dst, Src, Size, Align,
+                                  RTLIB::MEMCPY);
 
   unsigned BytesLeft = SizeVal & 3;
   unsigned NumMemOps = SizeVal >> 2;
   unsigned EmittedNumMemOps = 0;
   EVT VT = MVT::i32;
   unsigned VTSize = 4;
   unsigned i = 0;
   // Emit a maximum of 4 loads in Thumb1 since we have fewer registers
while (BytesLeft) {
     ++i;
     DstOff += VTSize;
     BytesLeft -= VTSize;
   }
   return DAG.getNode(ISD::TokenFactor, dl, MVT::Other,
                      makeArrayRef(TFOps, i));
 }
 
-// Adjust parameters for memset, EABI uses format (ptr, size, value),
-// GNU library uses (ptr, value, size)
-// See RTABI section 4.3.4
+
+SDValue ARMSelectionDAGInfo::
+EmitTargetCodeForMemmove(SelectionDAG &DAG, SDLoc dl,
+                         SDValue Chain,
+                         SDValue Dst, SDValue Src,
+                         SDValue Size, unsigned Align,
+                         bool isVolatile,
+                         MachinePointerInfo DstPtrInfo,
+                         MachinePointerInfo SrcPtrInfo) const {
+  return EmitSpecializedLibcall(DAG, dl, Chain, Dst, Src, Size, Align,
+                                RTLIB::MEMMOVE);
+}
+
+
 SDValue ARMSelectionDAGInfo::
 EmitTargetCodeForMemset(SelectionDAG &DAG, SDLoc dl,
                         SDValue Chain, SDValue Dst,
                         SDValue Src, SDValue Size,
                         unsigned Align, bool isVolatile,
                         MachinePointerInfo DstPtrInfo) const {
-  const ARMSubtarget &Subtarget =
-      DAG.getMachineFunction().getSubtarget<ARMSubtarget>();
-  // Use default for non-AAPCS (or MachO) subtargets
-  if (!Subtarget.isAAPCS_ABI() || Subtarget.isTargetMachO() ||
-      Subtarget.isTargetWindows())
-    return SDValue();
-
-  const ARMTargetLowering &TLI = *Subtarget.getTargetLowering();
-  TargetLowering::ArgListTy Args;
-  TargetLowering::ArgListEntry Entry;
-
-  // First argument: data pointer
-  Type *IntPtrTy = TLI.getDataLayout()->getIntPtrType(*DAG.getContext());
-  Entry.Node = Dst;
-  Entry.Ty = IntPtrTy;
-  Args.push_back(Entry);
-
-  // Second argument: buffer size
-  Entry.Node = Size;
-  Entry.Ty = IntPtrTy;
-  Entry.isSExt = false;
-  Args.push_back(Entry);
-
-  // Extend or truncate the argument to be an i32 value for the call.
-  if (Src.getValueType().bitsGT(MVT::i32))
-    Src = DAG.getNode(ISD::TRUNCATE, dl, MVT::i32, Src);
-  else
-    Src = DAG.getNode(ISD::ZERO_EXTEND, dl, MVT::i32, Src);
-
-  // Third argument: value to fill
-  Entry.Node = Src;
-  Entry.Ty = Type::getInt32Ty(*DAG.getContext());
-  Entry.isSExt = true;
-  Args.push_back(Entry);
-
-  // Emit __eabi_memset call
-  TargetLowering::CallLoweringInfo CLI(DAG);
-  CLI.setDebugLoc(dl).setChain(Chain)
-    .setCallee(TLI.getLibcallCallingConv(RTLIB::MEMSET),
-               Type::getVoidTy(*DAG.getContext()),
-               DAG.getExternalSymbol(TLI.getLibcallName(RTLIB::MEMSET),
-                                     TLI.getPointerTy()), std::move(Args), 0)
-    .setDiscardResult();
-
-  std::pair<SDValue,SDValue> CallResult = TLI.LowerCallTo(CLI);
-  return CallResult.second;
+  return EmitSpecializedLibcall(DAG, dl, Chain, Dst, Src, Size, Align,
+                                RTLIB::MEMSET);
 }

test/CodeGen/ARM/memfunc.ll

entry:
   call void @llvm.memmove.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 500, i32 0, i1 false)
 
   ; CHECK-IOS: memcpy
   ; CHECK-DARWIN: memcpy
   ; CHECK-EABI: __aeabi_memcpy
   call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 500, i32 0, i1 false)
 
   ; EABI memset swaps arguments
+  ; CHECK-IOS: mov r1, #1
+  ; CHECK-IOS: memset
+  ; CHECK-DARWIN: movs r1, #1
+  ; CHECK-DARWIN: memset
+  ; CHECK-EABI: mov r2, #1
+  ; CHECK-EABI: __aeabi_memset
+  call void @llvm.memset.p0i8.i32(i8* %dest, i8 1, i32 500, i32 0, i1 false)
+
+  ; EABI uses memclr if value set to 0
   ; CHECK-IOS: mov r1, #0
   ; CHECK-IOS: memset
   ; CHECK-DARWIN: movs r1, #0
   ; CHECK-DARWIN: memset
-  ; CHECK-EABI: mov r2, #0
-  ; CHECK-EABI: __aeabi_memset
+  ; CHECK-EABI: __aeabi_memclr
   call void @llvm.memset.p0i8.i32(i8* %dest, i8 0, i32 500, i32 0, i1 false)
+  
+  ; EABI uses aligned function variants if possible
+
+  ; CHECK-IOS: memmove
+  ; CHECK-DARWIN: memmove
+  ; CHECK-EABI: __aeabi_memmove4
+  call void @llvm.memmove.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 500, i32 4, i1 false)
+
+  ; CHECK-IOS: memcpy
+  ; CHECK-DARWIN: memcpy
+  ; CHECK-EABI: __aeabi_memcpy4
+  call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 500, i32 4, i1 false)
+
+  ; CHECK-IOS: memset
+  ; CHECK-DARWIN: memset
+  ; CHECK-EABI: __aeabi_memset4
+  call void @llvm.memset.p0i8.i32(i8* %dest, i8 1, i32 500, i32 4, i1 false)
+
+  ; CHECK-IOS: memset
+  ; CHECK-DARWIN: memset
+  ; CHECK-EABI: __aeabi_memclr4
+  call void @llvm.memset.p0i8.i32(i8* %dest, i8 0, i32 500, i32 4, i1 false)
+
+  ; CHECK-IOS: memmove
+  ; CHECK-DARWIN: memmove
+  ; CHECK-EABI: __aeabi_memmove8
+  call void @llvm.memmove.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 500, i32 8, i1 false)
+
+  ; CHECK-IOS: memcpy
+  ; CHECK-DARWIN: memcpy
+  ; CHECK-EABI: __aeabi_memcpy8
+  call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 500, i32 8, i1 false)
+
+  ; CHECK-IOS: memset
+  ; CHECK-DARWIN: memset
+  ; CHECK-EABI: __aeabi_memset8
+  call void @llvm.memset.p0i8.i32(i8* %dest, i8 1, i32 500, i32 8, i1 false)
+
+  ; CHECK-IOS: memset
+  ; CHECK-DARWIN: memset
+  ; CHECK-EABI: __aeabi_memclr8
+  call void @llvm.memset.p0i8.i32(i8* %dest, i8 0, i32 500, i32 8, i1 false)
+
   unreachable
 }
 
 ; Check that alloca arguments to memory intrinsics are automatically aligned if at least 8 bytes in size
 define void @f2(i8* %dest, i32 %n) {
 entry:
   ; CHECK-LABEL: f2
 
entry:
   ; CHECK-IOS: memcpy
   ; CHECK-DARWIN: memcpy
   ; CHECK-EABI: __aeabi_memcpy
   %arr1 = alloca [9 x i8], align 1
   %1 = bitcast [9 x i8]* %arr1 to i8*
   call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %1, i32 %n, i32 0, i1 false)
 
   ; CHECK-IOS: mov r0, sp
-  ; CHECK-IOS: mov r1, #0
+  ; CHECK-IOS: mov r1, #1
   ; CHECK-IOS: memset
-  ; CHECK-DARINW: add r0, sp, #4
-  ; CHECK-DARWIN: movs r1, #0
+  ; CHECK-DARWIN: add r0, sp, #4
+  ; CHECK-DARWIN: movs r1, #1
   ; CHECK-DARWIN: memset
   ; CHECK-EABI: add r0, sp, #4
-  ; CHECK-EABI: mov r2, #0
+  ; CHECK-EABI: mov r2, #1
   ; CHECK-EABI: __aeabi_memset
   %arr2 = alloca [9 x i8], align 1
   %2 = bitcast [9 x i8]* %arr2 to i8*
-  call void @llvm.memset.p0i8.i32(i8* %2, i8 0, i32 %n, i32 0, i1 false)
+  call void @llvm.memset.p0i8.i32(i8* %2, i8 1, i32 %n, i32 0, i1 false)
 
   unreachable
 }
 
 ; Check that alloca arguments are not aligned if less than 8 bytes in size
 define void @f3(i8* %dest, i32 %n) {
 entry:
   ; CHECK-LABEL: f3
entry:
   ; CHECK-IOS: memcpy
   ; CHECK-DARWIN: memcpy
   ; CHECK-EABI: __aeabi_memcpy
   %arr1 = alloca [7 x i8], align 1
   %1 = bitcast [7 x i8]* %arr1 to i8*
   call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %1, i32 %n, i32 0, i1 false)
 
   ; CHECK: {{add(.w)? r0, sp, #3}}
-  ; CHECK-IOS: mov r1, #0
+  ; CHECK-IOS: mov r1, #1
   ; CHECK-IOS: memset
-  ; CHECK-DARWIN: movs r1, #0
+  ; CHECK-DARWIN: movs r1, #1
   ; CHECK-DARWIN: memset
-  ; CHECK-EABI: mov r2, #0
+  ; CHECK-EABI: mov r2, #1
   ; CHECK-EABI: __aeabi_memset
   %arr2 = alloca [7 x i8], align 1
   %2 = bitcast [7 x i8]* %arr2 to i8*
-  call void @llvm.memset.p0i8.i32(i8* %2, i8 0, i32 %n, i32 0, i1 false)
+  call void @llvm.memset.p0i8.i32(i8* %2, i8 1, i32 %n, i32 0, i1 false)
 
   unreachable
 }
 
 ; Check that alloca arguments are not aligned if size+offset is less than 8 bytes
 define void @f4(i8* %dest, i32 %n) {
 entry:
   ; CHECK-LABEL: f4
entry:
   ; CHECK-IOS: memcpy
   ; CHECK-DARWIN: memcpy
   ; CHECK-EABI: __aeabi_memcpy
   %arr1 = alloca [9 x i8], align 1
   %1 = getelementptr inbounds [9 x i8], [9 x i8]* %arr1, i32 0, i32 4
   call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %1, i32 %n, i32 0, i1 false)
 
   ; CHECK: {{add(.w)? r., sp, #(1|5)}}
-  ; CHECK-IOS: mov r1, #0
+  ; CHECK-IOS: mov r1, #1
   ; CHECK-IOS: memset
-  ; CHECK-DARWIN: movs r1, #0
+  ; CHECK-DARWIN: movs r1, #1
   ; CHECK-DARWIN: memset
-  ; CHECK-EABI: mov r2, #0
+  ; CHECK-EABI: mov r2, #1
   ; CHECK-EABI: __aeabi_memset
   %arr2 = alloca [9 x i8], align 1
   %2 = getelementptr inbounds [9 x i8], [9 x i8]* %arr2, i32 0, i32 4
-  call void @llvm.memset.p0i8.i32(i8* %2, i8 0, i32 %n, i32 0, i1 false)
+  call void @llvm.memset.p0i8.i32(i8* %2, i8 1, i32 %n, i32 0, i1 false)
 
   unreachable
 }
 
 ; Check that alloca arguments are not aligned if the offset is not a multiple of 4
 define void @f5(i8* %dest, i32 %n) {
 entry:
   ; CHECK-LABEL: f5
entry:
   ; CHECK-IOS: memcpy
   ; CHECK-DARWIN: memcpy
   ; CHECK-EABI: __aeabi_memcpy
   %arr1 = alloca [13 x i8], align 1
   %1 = getelementptr inbounds [13 x i8], [13 x i8]* %arr1, i32 0, i32 1
   call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %1, i32 %n, i32 0, i1 false)
 
   ; CHECK: {{add(.w)? r., sp, #(1|5)}}
-  ; CHECK-IOS: mov r1, #0
+  ; CHECK-IOS: mov r1, #1
   ; CHECK-IOS: memset
-  ; CHECK-DARWIN: movs r1, #0
+  ; CHECK-DARWIN: movs r1, #1
   ; CHECK-DARWIN: memset
-  ; CHECK-EABI: mov r2, #0
+  ; CHECK-EABI: mov r2, #1
   ; CHECK-EABI: __aeabi_memset
   %arr2 = alloca [13 x i8], align 1
   %2 = getelementptr inbounds [13 x i8], [13 x i8]* %arr2, i32 0, i32 1
-  call void @llvm.memset.p0i8.i32(i8* %2, i8 0, i32 %n, i32 0, i1 false)
+  call void @llvm.memset.p0i8.i32(i8* %2, i8 1, i32 %n, i32 0, i1 false)
 
   unreachable
 }
 
 ; Check that alloca arguments are not aligned if the offset is unknown
 define void @f6(i8* %dest, i32 %n, i32 %i) {
 entry:
   ; CHECK-LABEL: f6
entry:
   ; CHECK-IOS: memcpy
   ; CHECK-DARWIN: memcpy
   ; CHECK-EABI: __aeabi_memcpy
   %arr1 = alloca [13 x i8], align 1
   %1 = getelementptr inbounds [13 x i8], [13 x i8]* %arr1, i32 0, i32 %i
   call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %1, i32 %n, i32 0, i1 false)
 
   ; CHECK: {{add(.w)? r., sp, #(1|5)}}
-  ; CHECK-IOS: mov r1, #0
+  ; CHECK-IOS: mov r1, #1
   ; CHECK-IOS: memset
-  ; CHECK-DARWIN: movs r1, #0
+  ; CHECK-DARWIN: movs r1, #1
   ; CHECK-DARWIN: memset
-  ; CHECK-EABI: mov r2, #0
+  ; CHECK-EABI: mov r2, #1
   ; CHECK-EABI: __aeabi_memset
   %arr2 = alloca [13 x i8], align 1
   %2 = getelementptr inbounds [13 x i8], [13 x i8]* %arr2, i32 0, i32 %i
-  call void @llvm.memset.p0i8.i32(i8* %2, i8 0, i32 %n, i32 0, i1 false)
+  call void @llvm.memset.p0i8.i32(i8* %2, i8 1, i32 %n, i32 0, i1 false)
 
   unreachable
 }
 
 ; Check that alloca arguments are not aligned if the GEP is not inbounds
 define void @f7(i8* %dest, i32 %n) {
 entry:
   ; CHECK-LABEL: f7
entry:
   ; CHECK-IOS: memcpy
   ; CHECK-DARWIN: memcpy
   ; CHECK-EABI: __aeabi_memcpy
   %arr1 = alloca [13 x i8], align 1
   %1 = getelementptr [13 x i8], [13 x i8]* %arr1, i32 0, i32 4
   call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %1, i32 %n, i32 0, i1 false)
 
   ; CHECK: {{add(.w)? r., sp, #(1|5)}}
-  ; CHECK-IOS: mov r1, #0
+  ; CHECK-IOS: mov r1, #1
   ; CHECK-IOS: memset
-  ; CHECK-DARWIN: movs r1, #0
+  ; CHECK-DARWIN: movs r1, #1
   ; CHECK-DARWIN: memset
-  ; CHECK-EABI: mov r2, #0
+  ; CHECK-EABI: mov r2, #1
   ; CHECK-EABI: __aeabi_memset
   %arr2 = alloca [13 x i8], align 1
   %2 = getelementptr [13 x i8], [13 x i8]* %arr2, i32 0, i32 4
-  call void @llvm.memset.p0i8.i32(i8* %2, i8 0, i32 %n, i32 0, i1 false)
+  call void @llvm.memset.p0i8.i32(i8* %2, i8 1, i32 %n, i32 0, i1 false)
 
   unreachable
 }
 
 ; Check that alloca arguments are not aligned when the offset is past the end of the allocation
 define void @f8(i8* %dest, i32 %n) {
 entry:
   ; CHECK-LABEL: f8
entry:
   ; CHECK-IOS: memcpy
   ; CHECK-DARWIN: memcpy
   ; CHECK-EABI: __aeabi_memcpy
   %arr1 = alloca [13 x i8], align 1
   %1 = getelementptr inbounds [13 x i8], [13 x i8]* %arr1, i32 0, i32 16
   call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %1, i32 %n, i32 0, i1 false)
 
   ; CHECK: {{add(.w)? r., sp, #(1|5)}}
-  ; CHECK-IOS: mov r1, #0
+  ; CHECK-IOS: mov r1, #1
   ; CHECK-IOS: memset
-  ; CHECK-DARWIN: movs r1, #0
+  ; CHECK-DARWIN: movs r1, #1
   ; CHECK-DARWIN: memset
-  ; CHECK-EABI: mov r2, #0
+  ; CHECK-EABI: mov r2, #1
   ; CHECK-EABI: __aeabi_memset
   %arr2 = alloca [13 x i8], align 1
   %2 = getelementptr inbounds [13 x i8], [13 x i8]* %arr2, i32 0, i32 16
-  call void @llvm.memset.p0i8.i32(i8* %2, i8 0, i32 %n, i32 0, i1 false)
+  call void @llvm.memset.p0i8.i32(i8* %2, i8 1, i32 %n, i32 0, i1 false)
 
   unreachable
 }
 
 ; Check that global variables are aligned if they are large enough, but only if
 ; they are defined in this object and don't have an explicit section.
 @arr1 = global [7 x i8] c"\01\02\03\04\05\06\07", align 1
 @arr2 = global [8 x i8] c"\01\02\03\04\05\06\07\08", align 1