[PowerPC] Eliminate compares in instruction selection - Vol. 1

[nemanjai]

Summary

This patch is the first in a series of 6 patches that split https://reviews.llvm.org/D31240 into more manageable chunks as well as include some additional work that depends on the patch posted by Lei.

The first 4 are really just a split of the functionality in that patch. None of the functionality is turned on in those patches, so they include no test cases. Patches 5 and 6 add further functionality and turn on what the original patch provided. So those two patches are the ones that include the test cases.

This particular one, just provides some of the infrastructure that the others build on - none of the functions here transform any compare instructions.

[nemanjai]

Temporarily commandeering this just to update the patches as per @echristo's request.

[nemanjai]

Updated to reflect some changes in how the big patch is split up as per Eric's request. This specific patch is NFC and can probably be quickly reviewed and committed. It essentially just adds all the methods that contain functionality as stubs. The subsequent patches will then add functionality for the methods and add the test cases. As such, they should be mostly independent of each other then.

[nemanjai]

Taking over this review so that we can update/review/commit it soon.

[nemanjai]

• Removed the static utility functions that weren't being used yet
• Added codegen for one of the patterns
• Added testing for that pattern

[echristo]

Let's delete any function that just returns false or SDValue() and see what that does to some of the code.

[nemanjai]

Removed functions that do nothing.

[echristo]

Hi Nemanja!

The patch looks really good, my inline comments notwithstanding. I think they're mostly cosmetic rather than anything else with a little bit of refactoring.

Thanks!

-eric

lib/Target/PowerPC/PPCISelDAGToDAG.cpp

2497 ↗

(On Diff #97501)

The PPC64 bit is because of testing or some other reason?

2522 ↗

(On Diff #97501)

I'm typically not a fan of boolean arguments to functions (for a number of reasons). Can we refactor this to deal with not doing that? Either pass N down, or an enum with one of the 4 states?

2535–2541 ↗

(On Diff #97501)

This could probably use some more detailed comments.

2539 ↗

(On Diff #97501)

Feels like these could be separate and separately documented early returns based on the above booleans/variables?

2559–2560 ↗

(On Diff #97501)

Feels like these could be separate and separately documented early returns based on the above booleans/variables?

2570–2571 ↗

(On Diff #97501)

See my earlier comment on boolean variables. I'm pretty sure it'll help readability here as well.

2574 ↗

(On Diff #97501)

I find it handy to describe the code you expect to return for these blocks.

2597 ↗

(On Diff #97501)

Document each case with the return instructions that you're going to be emitting? (and other places)

2639–2640 ↗

(On Diff #97501)

Ditto with the boolean arguments?

[nemanjai]

Address review comments:

• remove Boolean parameters
• split and document early exits
• document the code-gen for various setcc condition codes in each case
• document the reason this isn't supported on 32-bit targets

[echristo]

Few more inline comments :)

lib/Target/PowerPC/PPCISelDAGToDAG.cpp

264–265 ↗

(On Diff #98111)

These need documenting as to what each of them means.

2496–2498 ↗

(On Diff #98111)

This sounds more like "This code is currently specialized to 64-bit registers" rather than "this transformation doesn't make sense on 32-bit" yes?

2511 ↗

(On Diff #98111)

No need for an else here. Can just leave it as an early return.

2512 ↗

(On Diff #98111)

This can probably just be an if statement or a ternary operator in the call and get rid of ConvOpts as a local.

2516–2520 ↗

(On Diff #98111)

Do we expect to do this with non-32bit quantities? It also seems that we don't do anything for NoOp and so can early return? (Reading down through addExtOrTrunc). Are you actually getting things like zext/sext 32->32?

2521 ↗

(On Diff #98111)

This can be a few lines up.

2524 ↗

(On Diff #98111)

Can probably hoist the N->getOptcode() == ISD::SIGN_EXTEND up to the beginning of the function.

2561 ↗

(On Diff #98111)

May wish to assert that the input is MVT:i32?

2568–2569 ↗

(On Diff #98111)

Let's go ahead and sink these two down to where they're used.

2586 ↗

(On Diff #98111)

Did we want an i32 assertion in this function as well?

2604–2610 ↗

(On Diff #98111)

Probably want to do something like:

if (cond) {

<stuff>
return SDValue(...)

}

assert(otherCond);
<stuff>
return SDValue(...);

2683 ↗

(On Diff #98111)

4 and 2?

2684 ↗

(On Diff #98111)

MVT InputVT

2685 ↗

(On Diff #98111)

Do you mean '||' maybe?

2690 ↗

(On Diff #98111)

Early returns if not 32bit inputs? It'll probably clean up the logic below as well. Also a bit confusing with the InputVT stuff above.

4810 ↗

(On Diff #98111)

Cut and paste? :)

[nemanjai]

lib/Target/PowerPC/PPCISelDAGToDAG.cpp

2497 ↗

(On Diff #97501)

We really only want to do this on 64-bit targets. The reason is that instructions sequences in many cases need to be different in 32-bit mode. This is a result of:

1. Registers not actually being 32 bits wide in 32-bit mode
2. The i64 values not fitting in single registers in 32-bit mode

I will add a comment to these early exits to this end.

264–265 ↗

(On Diff #98111)

OK. I'll document them.

2496–2498 ↗

(On Diff #98111)

Yes, the transformation as implemented is specialized for 64-bit targets. I'll update the comment.

2511 ↗

(On Diff #98111)

Makes sense. Not sure why I left it in.

2512 ↗

(On Diff #98111)

OK, I'll get rid of the temp.

2516–2520 ↗

(On Diff #98111)

The complete patch will cover all cases. Right now we really only do 32-bit comparisons and we may or may not want the value in a 64-bit register (depending on the result type). But eventually we'll have all the combinations:
compare 32-bit values -> produce 32-bit value
compare 32-bit values -> produce 64-bit value
compare 64-bit values -> produce 32-bit value
compare 64-bit values -> produce 64-bit value

Two of those don't require conversions, one requires an extension and the other a truncation.

2521 ↗

(On Diff #98111)

OK, makes sense.

2524 ↗

(On Diff #98111)

We don't want to increment the stats in the early exit cases.

2561 ↗

(On Diff #98111)

Will do. But at the beginning of the function. I'm not sure why the comment is at the end of the function.

2586 ↗

(On Diff #98111)

Yup.

2604–2610 ↗

(On Diff #98111)

Yeah. When I refactored, I missed this weird setup in this function.

2683 ↗

(On Diff #98111)

Yup. 4 and 2. Just the operand numbers for the condition code in the two nodes. I'll rewrite this to name the operand number based on which node it is. Something like:
int CCOpNum = Compare.getOpcode() == ISD::SELECT_CC ? 4 : 2;

2684 ↗

(On Diff #98111)

I thought we typically used the EVT class for this type of thing but sure.

2685 ↗

(On Diff #98111)

Nope. Well I suppose I could DeMorgan-ize it to something like:

if (!(InputVT == MVT::i32 || InputVT == MVT::i64))
but I don't find that any more readable.

Ultimately, I want to exit early if it isn't one of those two types.

2690 ↗

(On Diff #98111)

I suppose I can clean that up on this patch since it currently doesn't handle 64-bit inputs, but it'll just be right back as soon as I add it in one of the subsequent patches.

4810 ↗

(On Diff #98111)

That's bizarre. No idea how I did that. I'll of course remove it :).

[nemanjai]

lib/Target/PowerPC/PPCISelDAGToDAG.cpp

2684 ↗

(On Diff #98111)

On second thought, looks like it'll have to stay an EVT or else I'd have to call getSimpleType() on the result of getValueType().

[nemanjai]

Addressed more inline comments.
Fixed up some early exits, added documentation, etc.

[echristo]

Couple of small comment requests, but that's it.

Thanks!

-eric

lib/Target/PowerPC/PPCISelDAGToDAG.cpp

2496–2498 ↗

(On Diff #98111)

"We only do this for 64-bit targets for now." maybe?

2516–2520 ↗

(On Diff #98111)

Interesting. I'll just take this on faith :)

2683 ↗

(On Diff #98111)

Oh. Sure. That makes a lot more sense. Can you add a comment as well?

2684 ↗

(On Diff #98111)

*shrug* Sure :)

2685 ↗

(On Diff #98111)

k.

llvm/trunk/lib/Target/PowerPC/InstPrinter/PPCInstPrinter.cpp

if ((MI->getOpcode() == PPC::OR || MI->getOpcode() == PPC::OR8) &&
     O << "\tmr ";
     printOperand(MI, 0, O);
     O << ", ";
     printOperand(MI, 1, O);
     printAnnotation(O, Annot);
     return;
   }
   
-  if (MI->getOpcode() == PPC::RLDICR) {
+  if (MI->getOpcode() == PPC::RLDICR ||
+      MI->getOpcode() == PPC::RLDICR_32) {
     unsigned char SH = MI->getOperand(2).getImm();
     unsigned char ME = MI->getOperand(3).getImm();
     // rldicr RA, RS, SH, 63-SH == sldi RA, RS, SH
     if (63-SH == ME) {
       O << "\tsldi ";
       printOperand(MI, 0, O);
       O << ", ";
       printOperand(MI, 1, O);

llvm/trunk/lib/Target/PowerPC/PPCFastISel.cpp

switch(MI->getOpcode()) {
     case PPC::EXTSH8:
     case PPC::EXTSH8_32_64: {
       if (VT != MVT::i16 && VT != MVT::i8)
         return false;
       break;
     }
 
     case PPC::EXTSW:
+    case PPC::EXTSW_32:
     case PPC::EXTSW_32_64: {
       if (VT != MVT::i32 && VT != MVT::i16 && VT != MVT::i8)
         return false;
       break;
     }
   }
 
   // See if we can handle this address.

llvm/trunk/lib/Target/PowerPC/PPCISelDAGToDAG.cpp

 #include "llvm/Support/Compiler.h"
 #include "llvm/Support/Debug.h"
 #include "llvm/Support/ErrorHandling.h"
 #include "llvm/Support/KnownBits.h"
 #include "llvm/Support/MathExtras.h"
 #include "llvm/Support/raw_ostream.h"
 #include "llvm/Target/TargetInstrInfo.h"
 #include "llvm/Target/TargetRegisterInfo.h"
+#include "llvm/ADT/Statistic.h"
 #include <algorithm>
 #include <cassert>
 #include <cstdint>
 #include <iterator>
 #include <limits>
 #include <memory>
 #include <new>
 #include <tuple>
 #include <utility>
 
 using namespace llvm;
 
 #define DEBUG_TYPE "ppc-codegen"
 
+STATISTIC(NumSextSetcc,
+          "Number of (sext(setcc)) nodes expanded into GPR sequence.");
+STATISTIC(NumZextSetcc,
+          "Number of (zext(setcc)) nodes expanded into GPR sequence.");
+STATISTIC(SignExtensionsAdded,
+          "Number of sign extensions for compare inputs added.");
+STATISTIC(ZeroExtensionsAdded,
+          "Number of zero extensions for compare inputs added.");
 // FIXME: Remove this once the bug has been fixed!
 cl::opt<bool> ANDIGlueBug("expose-ppc-andi-glue-bug",
 cl::desc("expose the ANDI glue bug on PPC"), cl::Hidden);
 
 static cl::opt<bool>
     UseBitPermRewriter("ppc-use-bit-perm-rewriter", cl::init(true),
                        cl::desc("use aggressive ppc isel for bit permutations"),
                        cl::Hidden);
public:
     StringRef getPassName() const override {
       return "PowerPC DAG->DAG Pattern Instruction Selection";
     }
 
 // Include the pieces autogenerated from the target description.
 #include "PPCGenDAGISel.inc"
 
 private:
+    // Conversion type for interpreting results of a 32-bit instruction as
+    // a 64-bit value or vice versa.
+    enum ExtOrTruncConversion { Ext, Trunc };
+
+    // Modifiers to guide how an ISD::SETCC node's result is to be computed
+    // in a GPR.
+    // ZExtOrig - use the original condition code, zero-extend value
+    // ZExtInvert - invert the condition code, zero-extend value
+    // SExtOrig - use the original condition code, sign-extend value
+    // SExtInvert - invert the condition code, sign-extend value
+    enum SetccInGPROpts { ZExtOrig, ZExtInvert, SExtOrig, SExtInvert };
+
     bool trySETCC(SDNode *N);
+    bool tryEXTEND(SDNode *N);
+    SDValue signExtendInputIfNeeded(SDValue Input);
+    SDValue zeroExtendInputIfNeeded(SDValue Input);
+    SDValue addExtOrTrunc(SDValue NatWidthRes, ExtOrTruncConversion Conv);
+    SDValue get32BitZExtCompare(SDValue LHS, SDValue RHS, ISD::CondCode CC,
+                                int64_t RHSValue, SDLoc dl);
+    SDValue get32BitSExtCompare(SDValue LHS, SDValue RHS, ISD::CondCode CC,
+                                int64_t RHSValue, SDLoc dl);
+    SDValue getSETCCInGPR(SDValue Compare, SetccInGPROpts ConvOpts);
 
     void PeepholePPC64();
     void PeepholePPC64ZExt();
     void PeepholeCROps();
 
     SDValue combineToCMPB(SDNode *N);
     void foldBoolExts(SDValue &Res, SDNode *&N);
 
bool PPCDAGToDAGISel::trySETCC(SDNode *N) {
 
   // Get the specified bit.
   SDValue Tmp =
     SDValue(CurDAG->getMachineNode(PPC::RLWINM, dl, MVT::i32, Ops), 0);
   CurDAG->SelectNodeTo(N, PPC::XORI, MVT::i32, Tmp, getI32Imm(1, dl));
   return true;
 }
 
+/// If this node is a sign/zero extension of an integer comparison,
+/// it can usually be computed in GPR's rather than using comparison
+/// instructions and ISEL. We only do this on 64-bit targets for now
+/// as the code is specialized for 64-bit (it uses 64-bit instructions
+/// and assumes 64-bit registers).
+bool PPCDAGToDAGISel::tryEXTEND(SDNode *N) {
+  if (TM.getOptLevel() == CodeGenOpt::None || !TM.isPPC64())
+    return false;
+  assert((N->getOpcode() == ISD::ZERO_EXTEND ||
+          N->getOpcode() == ISD::SIGN_EXTEND) &&
+          "Expecting a zero/sign extend node!");
+
+  if (N->getOperand(0).getOpcode() != ISD::SETCC)
+    return false;
+
+  SDValue WideRes =
+    getSETCCInGPR(N->getOperand(0),
+                  N->getOpcode() == ISD::SIGN_EXTEND ?
+                  SetccInGPROpts::SExtOrig : SetccInGPROpts::ZExtOrig);
+
+  if (!WideRes)
+    return false;
+
+  SDLoc dl(N);
+  bool Inputs32Bit = N->getOperand(0).getOperand(0).getValueType() == MVT::i32;
+  bool Output32Bit = N->getValueType(0) == MVT::i32;
+
+  NumSextSetcc += N->getOpcode() == ISD::SIGN_EXTEND ? 1 : 0;
+  NumZextSetcc += N->getOpcode() == ISD::SIGN_EXTEND ? 0 : 1;
+
+  SDValue ConvOp = WideRes;
+  if (Inputs32Bit != Output32Bit)
+    ConvOp = addExtOrTrunc(WideRes, Inputs32Bit ? ExtOrTruncConversion::Ext :
+                           ExtOrTruncConversion::Trunc);
+  ReplaceNode(N, ConvOp.getNode());
+
+  return true;
+}
+
+/// If the value isn't guaranteed to be sign-extended to 64-bits, extend it.
+/// Useful when emitting comparison code for 32-bit values without using
+/// the compare instruction (which only considers the lower 32-bits).
+SDValue PPCDAGToDAGISel::signExtendInputIfNeeded(SDValue Input) {
+  assert(Input.getValueType() == MVT::i32 &&
+         "Can only sign-extend 32-bit values here.");
+  unsigned Opc = Input.getOpcode();
+
+  // The value was sign extended and then truncated to 32-bits. No need to
+  // sign extend it again.
+  if (Opc == ISD::TRUNCATE &&
+      (Input.getOperand(0).getOpcode() == ISD::AssertSext ||
+       Input.getOperand(0).getOpcode() == ISD::SIGN_EXTEND))
+    return Input;
+
+  LoadSDNode *InputLoad = dyn_cast<LoadSDNode>(Input);
+  // The input is a sign-extending load. No reason to sign-extend.
+  if (InputLoad && InputLoad->getExtensionType() == ISD::SEXTLOAD)
+    return Input;
+
+  ConstantSDNode *InputConst = dyn_cast<ConstantSDNode>(Input);
+  // We don't sign-extend constants and already sign-extended values.
+  if (InputConst || Opc == ISD::AssertSext || Opc == ISD::SIGN_EXTEND_INREG ||
+      Opc == ISD::SIGN_EXTEND)
+    return Input;
+
+  SDLoc dl(Input);
+  SignExtensionsAdded++;
+  return SDValue(CurDAG->getMachineNode(PPC::EXTSW_32, dl, MVT::i32, Input), 0);
+}
+
+/// If the value isn't guaranteed to be zero-extended to 64-bits, extend it.
+/// Useful when emitting comparison code for 32-bit values without using
+/// the compare instruction (which only considers the lower 32-bits).
+SDValue PPCDAGToDAGISel::zeroExtendInputIfNeeded(SDValue Input) {
+  assert(Input.getValueType() == MVT::i32 &&
+         "Can only zero-extend 32-bit values here.");
+  LoadSDNode *InputLoad = dyn_cast<LoadSDNode>(Input);
+  unsigned Opc = Input.getOpcode();
+
+  // No need to zero-extend loaded values (unless they're loaded with
+  // a sign-extending load).
+  if (InputLoad && InputLoad->getExtensionType() != ISD::SEXTLOAD)
+    return Input;
+
+  ConstantSDNode *InputConst = dyn_cast<ConstantSDNode>(Input);
+  bool InputZExtConst = InputConst && InputConst->getSExtValue() >= 0;
+  // An ISD::TRUNCATE will be lowered to an EXTRACT_SUBREG so we have
+  // to conservatively actually clear the high bits. We also don't need to
+  // zero-extend constants or values that are already zero-extended.
+  if (InputZExtConst || Opc == ISD::AssertZext || Opc == ISD::ZERO_EXTEND)
+    return Input;
+
+  SDLoc dl(Input);
+  ZeroExtensionsAdded++;
+  return SDValue(CurDAG->getMachineNode(PPC::RLDICL_32, dl, MVT::i32, Input,
+                                        getI64Imm(0, dl), getI64Imm(32, dl)),
+                 0);
+}
+
+// Handle a 32-bit value in a 64-bit register and vice-versa. These are of
+// course not actual zero/sign extensions that will generate machine code,
+// they're just a way to reinterpret a 32 bit value in a register as a
+// 64 bit value and vice-versa.
+SDValue PPCDAGToDAGISel::addExtOrTrunc(SDValue NatWidthRes,
+                                       ExtOrTruncConversion Conv) {
+  SDLoc dl(NatWidthRes);
+
+  // For reinterpreting 32-bit values as 64 bit values, we generate
+  // INSERT_SUBREG IMPLICIT_DEF:i64, <input>, TargetConstant:i32<1>
+  if (Conv == ExtOrTruncConversion::Ext) {
+    SDValue ImDef(CurDAG->getMachineNode(PPC::IMPLICIT_DEF, dl, MVT::i64), 0);
+    SDValue SubRegIdx =
+      CurDAG->getTargetConstant(PPC::sub_32, dl, MVT::i32);
+    return SDValue(CurDAG->getMachineNode(PPC::INSERT_SUBREG, dl, MVT::i64,
+                                          ImDef, NatWidthRes, SubRegIdx), 0);
+  }
+
+  assert(Conv == ExtOrTruncConversion::Trunc &&
+         "Unknown convertion between 32 and 64 bit values.");
+  // For reinterpreting 64-bit values as 32-bit values, we just need to
+  // EXTRACT_SUBREG (i.e. extract the low word).
+  SDValue SubRegIdx =
+    CurDAG->getTargetConstant(PPC::sub_32, dl, MVT::i32);
+  return SDValue(CurDAG->getMachineNode(PPC::EXTRACT_SUBREG, dl, MVT::i32,
+                                        NatWidthRes, SubRegIdx), 0);
+}
+
+/// Produces a zero-extended result of comparing two 32-bit values according to
+/// the passed condition code.
+SDValue PPCDAGToDAGISel::get32BitZExtCompare(SDValue LHS, SDValue RHS,
+                                             ISD::CondCode CC,
+                                             int64_t RHSValue, SDLoc dl) {
+  bool IsRHSZero = RHSValue == 0;
+  switch (CC) {
+  default: return SDValue();
+  case ISD::SETEQ: {
+    // (zext (setcc %a, %b, seteq)) -> (lshr (cntlzw (xor %a, %b)), 5)
+    // (zext (setcc %a, 0, seteq))  -> (lshr (cntlzw %a), 5)
+    SDValue Xor = IsRHSZero ? LHS :
+      SDValue(CurDAG->getMachineNode(PPC::XOR, dl, MVT::i32, LHS, RHS), 0);
+    SDValue Clz =
+      SDValue(CurDAG->getMachineNode(PPC::CNTLZW, dl, MVT::i32, Xor), 0);
+    SDValue ShiftOps[] = { Clz, getI32Imm(27, dl), getI32Imm(5, dl),
+      getI32Imm(31, dl) };
+    return SDValue(CurDAG->getMachineNode(PPC::RLWINM, dl, MVT::i32,
+                                          ShiftOps), 0);
+  }
+  }
+}
+
+/// Produces a sign-extended result of comparing two 32-bit values according to
+/// the passed condition code.
+SDValue PPCDAGToDAGISel::get32BitSExtCompare(SDValue LHS, SDValue RHS,
+                                             ISD::CondCode CC,
+                                             int64_t RHSValue, SDLoc dl) {
+  bool IsRHSZero = RHSValue == 0;
+  switch (CC) {
+  default: return SDValue();
+  case ISD::SETEQ: {
+    // (sext (setcc %a, %b, seteq)) ->
+    //   (ashr (shl (ctlz (xor %a, %b)), 58), 63)
+    // (sext (setcc %a, 0, seteq)) ->
+    //   (ashr (shl (ctlz %a), 58), 63)
+    SDValue CountInput = IsRHSZero ? LHS :
+      SDValue(CurDAG->getMachineNode(PPC::XOR, dl, MVT::i32, LHS, RHS), 0);
+    SDValue Cntlzw =
+      SDValue(CurDAG->getMachineNode(PPC::CNTLZW, dl, MVT::i32, CountInput), 0);
+    SDValue SHLOps[] = { Cntlzw, getI32Imm(58, dl), getI32Imm(0, dl) };
+    SDValue Sldi =
+      SDValue(CurDAG->getMachineNode(PPC::RLDICR_32, dl, MVT::i32, SHLOps), 0);
+    return SDValue(CurDAG->getMachineNode(PPC::SRADI_32, dl, MVT::i32, Sldi,
+                                          getI32Imm(63, dl)), 0);
+  }
+  }
+}
+
+/// Returns an equivalent of a SETCC node but with the result the same width as
+/// the inputs. This can nalso be used for SELECT_CC if either the true or false
+/// values is a power of two while the other is zero.
+SDValue PPCDAGToDAGISel::getSETCCInGPR(SDValue Compare,
+                                       SetccInGPROpts ConvOpts) {
+  assert((Compare.getOpcode() == ISD::SETCC ||
+          Compare.getOpcode() == ISD::SELECT_CC) &&
+         "An ISD::SETCC node required here.");
+
+  SDValue LHS = Compare.getOperand(0);
+  SDValue RHS = Compare.getOperand(1);
+
+  // The condition code is operand 2 for SETCC and operand 4 for SELECT_CC.
+  int CCOpNum = Compare.getOpcode() == ISD::SELECT_CC ? 4 : 2;
+  ISD::CondCode CC =
+    cast<CondCodeSDNode>(Compare.getOperand(CCOpNum))->get();
+  EVT InputVT = LHS.getValueType();
+  if (InputVT != MVT::i32)
+    return SDValue();
+
+  SDLoc dl(Compare);
+  ConstantSDNode *RHSConst = dyn_cast<ConstantSDNode>(RHS);
+  int64_t RHSValue = RHSConst ? RHSConst->getSExtValue() : INT64_MAX;
+
+  if (ConvOpts == SetccInGPROpts::ZExtInvert ||
+      ConvOpts == SetccInGPROpts::SExtInvert)
+    CC = ISD::getSetCCInverse(CC, true);
+
+  if (ISD::isSignedIntSetCC(CC)) {
+    LHS = signExtendInputIfNeeded(LHS);
+    RHS = signExtendInputIfNeeded(RHS);
+  } else if (ISD::isUnsignedIntSetCC(CC)) {
+    LHS = zeroExtendInputIfNeeded(LHS);
+    RHS = zeroExtendInputIfNeeded(RHS);
+  }
+
+  bool IsSext = ConvOpts == SetccInGPROpts::SExtOrig ||
+    ConvOpts == SetccInGPROpts::SExtInvert;
+  if (IsSext)
+    return get32BitSExtCompare(LHS, RHS, CC, RHSValue, dl);
+  return get32BitZExtCompare(LHS, RHS, CC, RHSValue, dl);
+}
+
 void PPCDAGToDAGISel::transferMemOperands(SDNode *N, SDNode *Result) {
   // Transfer memoperands.
   MachineSDNode::mmo_iterator MemOp = MF->allocateMemRefsArray(1);
   MemOp[0] = cast<MemSDNode>(N)->getMemOperand();
   cast<MachineSDNode>(Result)->setMemRefs(MemOp, MemOp + 1);
 }
 
 // Select - Convert the specified operand from a target-independent to a
void PPCDAGToDAGISel::Select(SDNode *N) {
 
   case ISD::Constant:
     if (N->getValueType(0) == MVT::i64) {
       ReplaceNode(N, getInt64(CurDAG, N));
       return;
     }
     break;
 
+  case ISD::ZERO_EXTEND:
+  case ISD::SIGN_EXTEND:
+    if (tryEXTEND(N))
+      return;
+    break;
+
   case ISD::SETCC:
     if (trySETCC(N))
       return;
     break;
 
   case PPCISD::GlobalBaseReg:
     ReplaceNode(N, getGlobalBaseReg());
     return;

llvm/trunk/lib/Target/PowerPC/PPCInstr64Bit.td

 
 defm EXTSW  : XForm_11r<31, 986, (outs g8rc:$rA), (ins g8rc:$rS),
                         "extsw", "$rA, $rS", IIC_IntSimple,
                         [(set i64:$rA, (sext_inreg i64:$rS, i32))]>, isPPC64;
 let Interpretation64Bit = 1, isCodeGenOnly = 1 in
 defm EXTSW_32_64 : XForm_11r<31, 986, (outs g8rc:$rA), (ins gprc:$rS),
                              "extsw", "$rA, $rS", IIC_IntSimple,
                              [(set i64:$rA, (sext i32:$rS))]>, isPPC64;
+let isCodeGenOnly = 1 in
+def EXTSW_32 : XForm_11<31, 986, (outs gprc:$rA), (ins gprc:$rS),
+                        "extsw $rA, $rS", IIC_IntSimple,
+                        []>, isPPC64;
 
 defm SRADI  : XSForm_1rc<31, 413, (outs g8rc:$rA), (ins g8rc:$rS, u6imm:$SH),
                          "sradi", "$rA, $rS, $SH", IIC_IntRotateDI,
                          [(set i64:$rA, (sra i64:$rS, (i32 imm:$SH)))]>, isPPC64;
+// For fast-isel:
+let isCodeGenOnly = 1 in
+def SRADI_32  : XSForm_1<31, 413, (outs gprc:$rA), (ins gprc:$rS, u6imm:$SH),
+                         "sradi $rA, $rS, $SH", IIC_IntRotateDI, []>, isPPC64;
+
 defm CNTLZD : XForm_11r<31,  58, (outs g8rc:$rA), (ins g8rc:$rS),
                         "cntlzd", "$rA, $rS", IIC_IntGeneral,
                         [(set i64:$rA, (ctlz i64:$rS))]>;
 defm CNTTZD : XForm_11r<31, 570, (outs g8rc:$rA), (ins g8rc:$rS),
                         "cnttzd", "$rA, $rS", IIC_IntGeneral,
                         [(set i64:$rA, (cttz i64:$rS))]>, Requires<[IsISA3_0]>;
 def POPCNTD : XForm_11<31, 506, (outs g8rc:$rA), (ins g8rc:$rS),
                        "popcntd $rA, $rS", IIC_IntGeneral,
defm RLDCR  : MDSForm_1r<30, 9,
                         []>, isPPC64;
 defm RLDICL : MDForm_1r<30, 0,
                         (outs g8rc:$rA), (ins g8rc:$rS, u6imm:$SH, u6imm:$MBE),
                         "rldicl", "$rA, $rS, $SH, $MBE", IIC_IntRotateDI,
                         []>, isPPC64;
 // For fast-isel:
 let isCodeGenOnly = 1 in
 def RLDICL_32_64 : MDForm_1<30, 0,
                             (outs g8rc:$rA),
                             (ins gprc:$rS, u6imm:$SH, u6imm:$MBE),
                             "rldicl $rA, $rS, $SH, $MBE", IIC_IntRotateDI,
                             []>, isPPC64;
 // End fast-isel.
+let isCodeGenOnly = 1 in
+def RLDICL_32 : MDForm_1<30, 0,
+                         (outs gprc:$rA),
+                         (ins gprc:$rS, u6imm:$SH, u6imm:$MBE),
+                         "rldicl $rA, $rS, $SH, $MBE", IIC_IntRotateDI,
+                         []>, isPPC64;
 defm RLDICR : MDForm_1r<30, 1,
                         (outs g8rc:$rA), (ins g8rc:$rS, u6imm:$SH, u6imm:$MBE),
                         "rldicr", "$rA, $rS, $SH, $MBE", IIC_IntRotateDI,
                         []>, isPPC64;
+let isCodeGenOnly = 1 in
+def RLDICR_32 : MDForm_1<30, 1,
+                         (outs gprc:$rA), (ins gprc:$rS, u6imm:$SH, u6imm:$MBE),
+                         "rldicr $rA, $rS, $SH, $MBE", IIC_IntRotateDI,
+                         []>, isPPC64;
 defm RLDIC  : MDForm_1r<30, 2,
                         (outs g8rc:$rA), (ins g8rc:$rS, u6imm:$SH, u6imm:$MBE),
                         "rldic", "$rA, $rS, $SH, $MBE", IIC_IntRotateDI,
                         []>, isPPC64;
 
 let Interpretation64Bit = 1, isCodeGenOnly = 1 in {
 defm RLWINM8 : MForm_2r<21, (outs g8rc:$rA),
                         (ins g8rc:$rS, u5imm:$SH, u5imm:$MB, u5imm:$ME),

llvm/trunk/lib/Target/PowerPC/PPCInstrInfo.td

 def CLRLSLDIo : PPCAsmPseudo<"clrlsldi. $rA, $rS, $b, $n",
                              (ins g8rc:$rA, g8rc:$rS, u6imm:$b, u6imm:$n)>;
 
 def : InstAlias<"rotldi $rA, $rS, $n", (RLDICL g8rc:$rA, g8rc:$rS, u6imm:$n, 0)>;
 def : InstAlias<"rotldi. $rA, $rS, $n", (RLDICLo g8rc:$rA, g8rc:$rS, u6imm:$n, 0)>;
 def : InstAlias<"rotld $rA, $rS, $rB", (RLDCL g8rc:$rA, g8rc:$rS, gprc:$rB, 0)>;
 def : InstAlias<"rotld. $rA, $rS, $rB", (RLDCLo g8rc:$rA, g8rc:$rS, gprc:$rB, 0)>;
 def : InstAlias<"clrldi $rA, $rS, $n", (RLDICL g8rc:$rA, g8rc:$rS, 0, u6imm:$n)>;
+def : InstAlias<"clrldi $rA, $rS, $n",
+                (RLDICL_32 gprc:$rA, gprc:$rS, 0, u6imm:$n)>;
 def : InstAlias<"clrldi. $rA, $rS, $n", (RLDICLo g8rc:$rA, g8rc:$rS, 0, u6imm:$n)>;
 
 def RLWINMbm : PPCAsmPseudo<"rlwinm $rA, $rS, $n, $b",
                             (ins g8rc:$rA, g8rc:$rS, u5imm:$n, i32imm:$b)>;
 def RLWINMobm : PPCAsmPseudo<"rlwinm. $rA, $rS, $n, $b",
                             (ins g8rc:$rA, g8rc:$rS, u5imm:$n, i32imm:$b)>;
 def RLWIMIbm : PPCAsmPseudo<"rlwimi $rA, $rS, $n, $b",
                            (ins g8rc:$rA, g8rc:$rS, u5imm:$n, i32imm:$b)>;

llvm/trunk/test/CodeGen/PowerPC/setcc-logic.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc < %s -verify-machineinstrs -mtriple=powerpc64le-unknown-unknown | FileCheck %s
 
 define zeroext i1 @all_bits_clear(i32 %P, i32 %Q)  {
 ; CHECK-LABEL: all_bits_clear:
 ; CHECK:       # BB#0:
 ; CHECK-NEXT:    or 3, 3, 4
 ; CHECK-NEXT:    cntlzw 3, 3
-; CHECK-NEXT:    rlwinm 3, 3, 27, 31, 31
+; CHECK-NEXT:    srwi 3, 3, 5
 ; CHECK-NEXT:    blr
   %a = icmp eq i32 %P, 0
   %b = icmp eq i32 %Q, 0
   %c = and i1 %a, %b
   ret i1 %c
 }
 
 define zeroext i1 @all_sign_bits_clear(i32 %P, i32 %Q)  {
 ; CHECK-LABEL: all_sign_bits_clear:
 ; CHECK:       # BB#0:
 ; CHECK-NEXT:    or 3, 3, 4
 ; CHECK-NEXT:    nor 3, 3, 3
 ; CHECK-NEXT:    srwi 3, 3, 31
 ; CHECK-NEXT:    blr
   %a = icmp sgt i32 %P, -1
   %b = icmp sgt i32 %Q, -1
   %c = and i1 %a, %b
   ret i1 %c
 }
 
 define zeroext i1 @all_bits_set(i32 %P, i32 %Q)  {
 ; CHECK-LABEL: all_bits_set:
 ; CHECK:       # BB#0:
+; CHECK-NEXT:    li 5, -1
 ; CHECK-NEXT:    and 3, 3, 4
-; CHECK-NEXT:    li 5, 0
-; CHECK-NEXT:    li 12, 1
-; CHECK-NEXT:    cmpwi 0, 3, -1
-; CHECK-NEXT:    isel 3, 12, 5, 2
+; CHECK-NEXT:    xor 3, 3, 5
+; CHECK-NEXT:    cntlzw 3, 3
+; CHECK-NEXT:    srwi 3, 3, 5
 ; CHECK-NEXT:    blr
   %a = icmp eq i32 %P, -1
   %b = icmp eq i32 %Q, -1
   %c = and i1 %a, %b
   ret i1 %c
 }
 
 define zeroext i1 @all_sign_bits_set(i32 %P, i32 %Q)  {

 
 define zeroext i1 @and_eq(i16 zeroext  %a, i16 zeroext %b, i16 zeroext %c, i16 zeroext %d) {
 ; CHECK-LABEL: and_eq:
 ; CHECK:       # BB#0:
 ; CHECK-NEXT:    xor 5, 5, 6
 ; CHECK-NEXT:    xor 3, 3, 4
 ; CHECK-NEXT:    or 3, 3, 5
 ; CHECK-NEXT:    cntlzw 3, 3
-; CHECK-NEXT:    rlwinm 3, 3, 27, 31, 31
+; CHECK-NEXT:    srwi 3, 3, 5
 ; CHECK-NEXT:    blr
   %cmp1 = icmp eq i16 %a, %b
   %cmp2 = icmp eq i16 %c, %d
   %and = and i1 %cmp1, %cmp2
   ret i1 %and
 }
 
 define zeroext i1 @or_ne(i32 %a, i32 %b, i32 %c, i32 %d) {

llvm/trunk/test/CodeGen/PowerPC/testComparesieqsc.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -verify-machineinstrs -mtriple=powerpc64-unknown-linux-gnu -O2 \
+; RUN:   -ppc-asm-full-reg-names -mcpu=pwr8 < %s | FileCheck %s \
+; RUN:  --implicit-check-not cmpw --implicit-check-not cmpd --implicit-check-not cmpl
+; RUN: llc -verify-machineinstrs -mtriple=powerpc64le-unknown-linux-gnu -O2 \
+; RUN:   -ppc-asm-full-reg-names -mcpu=pwr8 < %s | FileCheck %s \
+; RUN:  --implicit-check-not cmpw --implicit-check-not cmpd --implicit-check-not cmpl
+; ModuleID = 'ComparisonTestCases/testComparesieqsc.c'
+
+@glob = common local_unnamed_addr global i8 0, align 1
+
+; Function Attrs: norecurse nounwind readnone
+define signext i32 @test_ieqsc(i8 signext %a, i8 signext %b) {
+; CHECK-LABEL: test_ieqsc:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i8 %a, %b
+  %conv2 = zext i1 %cmp to i32
+  ret i32 %conv2
+}
+
+; Function Attrs: norecurse nounwind readnone
+define signext i32 @test_ieqsc_sext(i8 signext %a, i8 signext %b) {
+; CHECK-LABEL: test_ieqsc_sext:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i8 %a, %b
+  %sub = sext i1 %cmp to i32
+  ret i32 %sub
+}
+
+; Function Attrs: norecurse nounwind readnone
+define signext i32 @test_ieqsc_z(i8 signext %a) {
+; CHECK-LABEL: test_ieqsc_z:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i8 %a, 0
+  %conv1 = zext i1 %cmp to i32
+  ret i32 %conv1
+}
+
+; Function Attrs: norecurse nounwind readnone
+define signext i32 @test_ieqsc_sext_z(i8 signext %a) {
+; CHECK-LABEL: test_ieqsc_sext_z:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i8 %a, 0
+  %sub = sext i1 %cmp to i32
+  ret i32 %sub
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_ieqsc_store(i8 signext %a, i8 signext %b) {
+; CHECK-LABEL: test_ieqsc_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    addis r5, r2, .LC0@toc@ha
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    ld r12, .LC0@toc@l(r5)
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    stb r3, 0(r12)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i8 %a, %b
+  %conv3 = zext i1 %cmp to i8
+  store i8 %conv3, i8* @glob, align 1
+  ret void
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_ieqsc_sext_store(i8 signext %a, i8 signext %b) {
+; CHECK-LABEL: test_ieqsc_sext_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    addis r5, r2, .LC0@toc@ha
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    ld r4, .LC0@toc@l(r5)
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    stb r3, 0(r4)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i8 %a, %b
+  %conv3 = sext i1 %cmp to i8
+  store i8 %conv3, i8* @glob, align 1
+  ret void
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_ieqsc_z_store(i8 signext %a) {
+; CHECK-LABEL: test_ieqsc_z_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    addis r4, r2, .LC0@toc@ha
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    ld r4, .LC0@toc@l(r4)
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    stb r3, 0(r4)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i8 %a, 0
+  %conv2 = zext i1 %cmp to i8
+  store i8 %conv2, i8* @glob, align 1
+  ret void
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_ieqsc_sext_z_store(i8 signext %a) {
+; CHECK-LABEL: test_ieqsc_sext_z_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    addis r4, r2, .LC0@toc@ha
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    ld r4, .LC0@toc@l(r4)
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    stb r3, 0(r4)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i8 %a, 0
+  %conv2 = sext i1 %cmp to i8
+  store i8 %conv2, i8* @glob, align 1
+  ret void
+}

llvm/trunk/test/CodeGen/PowerPC/testComparesieqsi.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -verify-machineinstrs -mtriple=powerpc64-unknown-linux-gnu -O2 \
+; RUN:   -ppc-asm-full-reg-names -mcpu=pwr8 < %s | FileCheck %s \
+; RUN:  --implicit-check-not cmpw --implicit-check-not cmpd --implicit-check-not cmpl
+; RUN: llc -verify-machineinstrs -mtriple=powerpc64le-unknown-linux-gnu -O2 \
+; RUN:   -ppc-asm-full-reg-names -mcpu=pwr8 < %s | FileCheck %s \
+; RUN:  --implicit-check-not cmpw --implicit-check-not cmpd --implicit-check-not cmpl
+; ModuleID = 'ComparisonTestCases/testComparesieqsi.c'
+
+@glob = common local_unnamed_addr global i32 0, align 4
+
+; Function Attrs: norecurse nounwind readnone
+define signext i32 @test_ieqsi(i32 signext %a, i32 signext %b) {
+; CHECK-LABEL: test_ieqsi:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i32 %a, %b
+  %conv = zext i1 %cmp to i32
+  ret i32 %conv
+}
+
+; Function Attrs: norecurse nounwind readnone
+define signext i32 @test_ieqsi_sext(i32 signext %a, i32 signext %b) {
+; CHECK-LABEL: test_ieqsi_sext:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i32 %a, %b
+  %sub = sext i1 %cmp to i32
+  ret i32 %sub
+}
+
+; Function Attrs: norecurse nounwind readnone
+define signext i32 @test_ieqsi_z(i32 signext %a) {
+; CHECK-LABEL: test_ieqsi_z:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i32 %a, 0
+  %conv = zext i1 %cmp to i32
+  ret i32 %conv
+}
+
+; Function Attrs: norecurse nounwind readnone
+define signext i32 @test_ieqsi_sext_z(i32 signext %a) {
+; CHECK-LABEL: test_ieqsi_sext_z:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i32 %a, 0
+  %sub = sext i1 %cmp to i32
+  ret i32 %sub
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_ieqsi_store(i32 signext %a, i32 signext %b) {
+; CHECK-LABEL: test_ieqsi_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    addis r5, r2, .LC0@toc@ha
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    ld r12, .LC0@toc@l(r5)
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    stw r3, 0(r12)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i32 %a, %b
+  %conv = zext i1 %cmp to i32
+  store i32 %conv, i32* @glob, align 4
+  ret void
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_ieqsi_sext_store(i32 signext %a, i32 signext %b) {
+; CHECK-LABEL: test_ieqsi_sext_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    addis r5, r2, .LC0@toc@ha
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    ld r4, .LC0@toc@l(r5)
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    stw r3, 0(r4)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i32 %a, %b
+  %sub = sext i1 %cmp to i32
+  store i32 %sub, i32* @glob, align 4
+  ret void
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_ieqsi_z_store(i32 signext %a) {
+; CHECK-LABEL: test_ieqsi_z_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    addis r4, r2, .LC0@toc@ha
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    ld r4, .LC0@toc@l(r4)
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    stw r3, 0(r4)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i32 %a, 0
+  %conv = zext i1 %cmp to i32
+  store i32 %conv, i32* @glob, align 4
+  ret void
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_ieqsi_sext_z_store(i32 signext %a) {
+; CHECK-LABEL: test_ieqsi_sext_z_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    addis r4, r2, .LC0@toc@ha
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    ld r4, .LC0@toc@l(r4)
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    stw r3, 0(r4)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i32 %a, 0
+  %sub = sext i1 %cmp to i32
+  store i32 %sub, i32* @glob, align 4
+  ret void
+}

llvm/trunk/test/CodeGen/PowerPC/testComparesieqss.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -verify-machineinstrs -mtriple=powerpc64-unknown-linux-gnu -O2 \
+; RUN:   -ppc-asm-full-reg-names -mcpu=pwr8 < %s | FileCheck %s \
+; RUN:  --implicit-check-not cmpw --implicit-check-not cmpd --implicit-check-not cmpl
+; RUN: llc -verify-machineinstrs -mtriple=powerpc64le-unknown-linux-gnu -O2 \
+; RUN:   -ppc-asm-full-reg-names -mcpu=pwr8 < %s | FileCheck %s \
+; RUN:  --implicit-check-not cmpw --implicit-check-not cmpd --implicit-check-not cmpl
+; ModuleID = 'ComparisonTestCases/testComparesieqss.c'
+
+@glob = common local_unnamed_addr global i16 0, align 2
+
+; Function Attrs: norecurse nounwind readnone
+define signext i32 @test_ieqss(i16 signext %a, i16 signext %b) {
+; CHECK-LABEL: test_ieqss:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i16 %a, %b
+  %conv2 = zext i1 %cmp to i32
+  ret i32 %conv2
+}
+
+; Function Attrs: norecurse nounwind readnone
+define signext i32 @test_ieqss_sext(i16 signext %a, i16 signext %b) {
+; CHECK-LABEL: test_ieqss_sext:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i16 %a, %b
+  %sub = sext i1 %cmp to i32
+  ret i32 %sub
+}
+
+; Function Attrs: norecurse nounwind readnone
+define signext i32 @test_ieqss_z(i16 signext %a) {
+; CHECK-LABEL: test_ieqss_z:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i16 %a, 0
+  %conv1 = zext i1 %cmp to i32
+  ret i32 %conv1
+}
+
+; Function Attrs: norecurse nounwind readnone
+define signext i32 @test_ieqss_sext_z(i16 signext %a) {
+; CHECK-LABEL: test_ieqss_sext_z:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i16 %a, 0
+  %sub = sext i1 %cmp to i32
+  ret i32 %sub
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_ieqss_store(i16 signext %a, i16 signext %b) {
+; CHECK-LABEL: test_ieqss_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    addis r5, r2, .LC0@toc@ha
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    ld r12, .LC0@toc@l(r5)
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    sth r3, 0(r12)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i16 %a, %b
+  %conv3 = zext i1 %cmp to i16
+  store i16 %conv3, i16* @glob, align 2
+  ret void
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_ieqss_sext_store(i16 signext %a, i16 signext %b) {
+; CHECK-LABEL: test_ieqss_sext_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    addis r5, r2, .LC0@toc@ha
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    ld r4, .LC0@toc@l(r5)
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    sth r3, 0(r4)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i16 %a, %b
+  %conv3 = sext i1 %cmp to i16
+  store i16 %conv3, i16* @glob, align 2
+  ret void
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_ieqss_z_store(i16 signext %a) {
+; CHECK-LABEL: test_ieqss_z_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    addis r4, r2, .LC0@toc@ha
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    ld r4, .LC0@toc@l(r4)
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    sth r3, 0(r4)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i16 %a, 0
+  %conv2 = zext i1 %cmp to i16
+  store i16 %conv2, i16* @glob, align 2
+  ret void
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_ieqss_sext_z_store(i16 signext %a) {
+; CHECK-LABEL: test_ieqss_sext_z_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    addis r4, r2, .LC0@toc@ha
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    ld r4, .LC0@toc@l(r4)
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    sth r3, 0(r4)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i16 %a, 0
+  %conv2 = sext i1 %cmp to i16
+  store i16 %conv2, i16* @glob, align 2
+  ret void
+}

llvm/trunk/test/CodeGen/PowerPC/testComparesiequc.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -verify-machineinstrs -mtriple=powerpc64-unknown-linux-gnu -O2 \
+; RUN:   -ppc-asm-full-reg-names -mcpu=pwr8 < %s | FileCheck %s \
+; RUN:  --implicit-check-not cmpw --implicit-check-not cmpd --implicit-check-not cmpl
+; RUN: llc -verify-machineinstrs -mtriple=powerpc64le-unknown-linux-gnu -O2 \
+; RUN:   -ppc-asm-full-reg-names -mcpu=pwr8 < %s | FileCheck %s \
+; RUN:  --implicit-check-not cmpw --implicit-check-not cmpd --implicit-check-not cmpl
+; ModuleID = 'ComparisonTestCases/testComparesiequc.c'
+
+@glob = common local_unnamed_addr global i8 0, align 1
+
+; Function Attrs: norecurse nounwind readnone
+define signext i32 @test_iequc(i8 zeroext %a, i8 zeroext %b) {
+; CHECK-LABEL: test_iequc:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i8 %a, %b
+  %conv2 = zext i1 %cmp to i32
+  ret i32 %conv2
+}
+
+; Function Attrs: norecurse nounwind readnone
+define signext i32 @test_iequc_sext(i8 zeroext %a, i8 zeroext %b) {
+; CHECK-LABEL: test_iequc_sext:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i8 %a, %b
+  %sub = sext i1 %cmp to i32
+  ret i32 %sub
+}
+
+; Function Attrs: norecurse nounwind readnone
+define signext i32 @test_iequc_z(i8 zeroext %a) {
+; CHECK-LABEL: test_iequc_z:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i8 %a, 0
+  %conv1 = zext i1 %cmp to i32
+  ret i32 %conv1
+}
+
+; Function Attrs: norecurse nounwind readnone
+define signext i32 @test_iequc_sext_z(i8 zeroext %a) {
+; CHECK-LABEL: test_iequc_sext_z:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i8 %a, 0
+  %sub = sext i1 %cmp to i32
+  ret i32 %sub
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_iequc_store(i8 zeroext %a, i8 zeroext %b) {
+; CHECK-LABEL: test_iequc_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    addis r5, r2, .LC0@toc@ha
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    ld r12, .LC0@toc@l(r5)
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    stb r3, 0(r12)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i8 %a, %b
+  %conv3 = zext i1 %cmp to i8
+  store i8 %conv3, i8* @glob, align 1
+  ret void
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_iequc_sext_store(i8 zeroext %a, i8 zeroext %b) {
+; CHECK-LABEL: test_iequc_sext_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    addis r5, r2, .LC0@toc@ha
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    ld r4, .LC0@toc@l(r5)
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    stb r3, 0(r4)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i8 %a, %b
+  %conv3 = sext i1 %cmp to i8
+  store i8 %conv3, i8* @glob, align 1
+  ret void
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_iequc_z_store(i8 zeroext %a) {
+; CHECK-LABEL: test_iequc_z_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    addis r4, r2, .LC0@toc@ha
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    ld r4, .LC0@toc@l(r4)
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    stb r3, 0(r4)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i8 %a, 0
+  %conv2 = zext i1 %cmp to i8
+  store i8 %conv2, i8* @glob, align 1
+  ret void
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_iequc_sext_z_store(i8 zeroext %a) {
+; CHECK-LABEL: test_iequc_sext_z_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    addis r4, r2, .LC0@toc@ha
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    ld r4, .LC0@toc@l(r4)
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    stb r3, 0(r4)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i8 %a, 0
+  %conv2 = sext i1 %cmp to i8
+  store i8 %conv2, i8* @glob, align 1
+  ret void
+}

llvm/trunk/test/CodeGen/PowerPC/testComparesiequi.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -verify-machineinstrs -mtriple=powerpc64-unknown-linux-gnu -O2 \
+; RUN:   -ppc-asm-full-reg-names -mcpu=pwr8 < %s | FileCheck %s \
+; RUN:  --implicit-check-not cmpw --implicit-check-not cmpd --implicit-check-not cmpl
+; RUN: llc -verify-machineinstrs -mtriple=powerpc64le-unknown-linux-gnu -O2 \
+; RUN:   -ppc-asm-full-reg-names -mcpu=pwr8 < %s | FileCheck %s \
+; RUN:  --implicit-check-not cmpw --implicit-check-not cmpd --implicit-check-not cmpl
+; ModuleID = 'ComparisonTestCases/testComparesiequi.c'
+
+@glob = common local_unnamed_addr global i32 0, align 4
+
+; Function Attrs: norecurse nounwind readnone
+define signext i32 @test_iequi(i32 zeroext %a, i32 zeroext %b) {
+; CHECK-LABEL: test_iequi:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i32 %a, %b
+  %conv = zext i1 %cmp to i32
+  ret i32 %conv
+}
+
+; Function Attrs: norecurse nounwind readnone
+define signext i32 @test_iequi_sext(i32 zeroext %a, i32 zeroext %b) {
+; CHECK-LABEL: test_iequi_sext:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i32 %a, %b
+  %sub = sext i1 %cmp to i32
+  ret i32 %sub
+}
+
+; Function Attrs: norecurse nounwind readnone
+define signext i32 @test_iequi_z(i32 zeroext %a) {
+; CHECK-LABEL: test_iequi_z:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i32 %a, 0
+  %conv = zext i1 %cmp to i32
+  ret i32 %conv
+}
+
+; Function Attrs: norecurse nounwind readnone
+define signext i32 @test_iequi_sext_z(i32 zeroext %a) {
+; CHECK-LABEL: test_iequi_sext_z:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i32 %a, 0
+  %sub = sext i1 %cmp to i32
+  ret i32 %sub
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_iequi_store(i32 zeroext %a, i32 zeroext %b) {
+; CHECK-LABEL: test_iequi_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    addis r5, r2, .LC0@toc@ha
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    ld r12, .LC0@toc@l(r5)
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    stw r3, 0(r12)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i32 %a, %b
+  %conv = zext i1 %cmp to i32
+  store i32 %conv, i32* @glob, align 4
+  ret void
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_iequi_sext_store(i32 zeroext %a, i32 zeroext %b) {
+; CHECK-LABEL: test_iequi_sext_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    addis r5, r2, .LC0@toc@ha
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    ld r4, .LC0@toc@l(r5)
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    stw r3, 0(r4)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i32 %a, %b
+  %sub = sext i1 %cmp to i32
+  store i32 %sub, i32* @glob, align 4
+  ret void
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_iequi_z_store(i32 zeroext %a) {
+; CHECK-LABEL: test_iequi_z_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    addis r4, r2, .LC0@toc@ha
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    ld r4, .LC0@toc@l(r4)
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    stw r3, 0(r4)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i32 %a, 0
+  %conv = zext i1 %cmp to i32
+  store i32 %conv, i32* @glob, align 4
+  ret void
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_iequi_sext_z_store(i32 zeroext %a) {
+; CHECK-LABEL: test_iequi_sext_z_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    addis r4, r2, .LC0@toc@ha
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    ld r4, .LC0@toc@l(r4)
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    stw r3, 0(r4)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i32 %a, 0
+  %sub = sext i1 %cmp to i32
+  store i32 %sub, i32* @glob, align 4
+  ret void
+}

llvm/trunk/test/CodeGen/PowerPC/testComparesiequs.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -verify-machineinstrs -mtriple=powerpc64-unknown-linux-gnu -O2 \
+; RUN:   -ppc-asm-full-reg-names -mcpu=pwr8 < %s | FileCheck %s \
+; RUN:  --implicit-check-not cmpw --implicit-check-not cmpd --implicit-check-not cmpl
+; RUN: llc -verify-machineinstrs -mtriple=powerpc64le-unknown-linux-gnu -O2 \
+; RUN:   -ppc-asm-full-reg-names -mcpu=pwr8 < %s | FileCheck %s \
+; RUN:  --implicit-check-not cmpw --implicit-check-not cmpd --implicit-check-not cmpl
+; ModuleID = 'ComparisonTestCases/testComparesiequs.c'
+
+@glob = common local_unnamed_addr global i16 0, align 2
+
+; Function Attrs: norecurse nounwind readnone
+define signext i32 @test_iequs(i16 zeroext %a, i16 zeroext %b) {
+; CHECK-LABEL: test_iequs:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i16 %a, %b
+  %conv2 = zext i1 %cmp to i32
+  ret i32 %conv2
+}
+
+; Function Attrs: norecurse nounwind readnone
+define signext i32 @test_iequs_sext(i16 zeroext %a, i16 zeroext %b) {
+; CHECK-LABEL: test_iequs_sext:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i16 %a, %b
+  %sub = sext i1 %cmp to i32
+  ret i32 %sub
+}
+
+; Function Attrs: norecurse nounwind readnone
+define signext i32 @test_iequs_z(i16 zeroext %a) {
+; CHECK-LABEL: test_iequs_z:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i16 %a, 0
+  %conv1 = zext i1 %cmp to i32
+  ret i32 %conv1
+}
+
+; Function Attrs: norecurse nounwind readnone
+define signext i32 @test_iequs_sext_z(i16 zeroext %a) {
+; CHECK-LABEL: test_iequs_sext_z:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i16 %a, 0
+  %sub = sext i1 %cmp to i32
+  ret i32 %sub
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_iequs_store(i16 zeroext %a, i16 zeroext %b) {
+; CHECK-LABEL: test_iequs_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    addis r5, r2, .LC0@toc@ha
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    ld r12, .LC0@toc@l(r5)
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    sth r3, 0(r12)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i16 %a, %b
+  %conv3 = zext i1 %cmp to i16
+  store i16 %conv3, i16* @glob, align 2
+  ret void
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_iequs_sext_store(i16 zeroext %a, i16 zeroext %b) {
+; CHECK-LABEL: test_iequs_sext_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    addis r5, r2, .LC0@toc@ha
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    ld r4, .LC0@toc@l(r5)
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    sth r3, 0(r4)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i16 %a, %b
+  %conv3 = sext i1 %cmp to i16
+  store i16 %conv3, i16* @glob, align 2
+  ret void
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_iequs_z_store(i16 zeroext %a) {
+; CHECK-LABEL: test_iequs_z_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    addis r4, r2, .LC0@toc@ha
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    ld r4, .LC0@toc@l(r4)
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    sth r3, 0(r4)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i16 %a, 0
+  %conv2 = zext i1 %cmp to i16
+  store i16 %conv2, i16* @glob, align 2
+  ret void
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_iequs_sext_z_store(i16 zeroext %a) {
+; CHECK-LABEL: test_iequs_sext_z_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    addis r4, r2, .LC0@toc@ha
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    ld r4, .LC0@toc@l(r4)
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    sth r3, 0(r4)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i16 %a, 0
+  %conv2 = sext i1 %cmp to i16
+  store i16 %conv2, i16* @glob, align 2
+  ret void
+}

llvm/trunk/test/CodeGen/PowerPC/testCompareslleqsc.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -verify-machineinstrs -mtriple=powerpc64-unknown-linux-gnu -O2 \
+; RUN:   -ppc-asm-full-reg-names -mcpu=pwr8 < %s | FileCheck %s \
+; RUN:  --implicit-check-not cmpw --implicit-check-not cmpd --implicit-check-not cmpl
+; RUN: llc -verify-machineinstrs -mtriple=powerpc64le-unknown-linux-gnu -O2 \
+; RUN:   -ppc-asm-full-reg-names -mcpu=pwr8 < %s | FileCheck %s \
+; RUN:  --implicit-check-not cmpw --implicit-check-not cmpd --implicit-check-not cmpl
+; ModuleID = 'ComparisonTestCases/testCompareslleqsc.c'
+
+@glob = common local_unnamed_addr global i8 0, align 1
+
+; Function Attrs: norecurse nounwind readnone
+define i64 @test_lleqsc(i8 signext %a, i8 signext %b) {
+; CHECK-LABEL: test_lleqsc:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i8 %a, %b
+  %conv3 = zext i1 %cmp to i64
+  ret i64 %conv3
+}
+
+; Function Attrs: norecurse nounwind readnone
+define i64 @test_lleqsc_sext(i8 signext %a, i8 signext %b) {
+; CHECK-LABEL: test_lleqsc_sext:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i8 %a, %b
+  %conv3 = sext i1 %cmp to i64
+  ret i64 %conv3
+}
+
+; Function Attrs: norecurse nounwind readnone
+define i64 @test_lleqsc_z(i8 signext %a) {
+; CHECK-LABEL: test_lleqsc_z:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i8 %a, 0
+  %conv2 = zext i1 %cmp to i64
+  ret i64 %conv2
+}
+
+; Function Attrs: norecurse nounwind readnone
+define i64 @test_lleqsc_sext_z(i8 signext %a) {
+; CHECK-LABEL: test_lleqsc_sext_z:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i8 %a, 0
+  %conv2 = sext i1 %cmp to i64
+  ret i64 %conv2
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_lleqsc_store(i8 signext %a, i8 signext %b) {
+; CHECK-LABEL: test_lleqsc_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    addis r5, r2, .LC0@toc@ha
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    ld r12, .LC0@toc@l(r5)
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    stb r3, 0(r12)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i8 %a, %b
+  %conv3 = zext i1 %cmp to i8
+  store i8 %conv3, i8* @glob, align 1
+  ret void
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_lleqsc_sext_store(i8 signext %a, i8 signext %b) {
+; CHECK-LABEL: test_lleqsc_sext_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    addis r5, r2, .LC0@toc@ha
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    ld r4, .LC0@toc@l(r5)
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    stb r3, 0(r4)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i8 %a, %b
+  %conv3 = sext i1 %cmp to i8
+  store i8 %conv3, i8* @glob, align 1
+  ret void
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_lleqsc_z_store(i8 signext %a) {
+; CHECK-LABEL: test_lleqsc_z_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    addis r4, r2, .LC0@toc@ha
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    ld r4, .LC0@toc@l(r4)
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    stb r3, 0(r4)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i8 %a, 0
+  %conv2 = zext i1 %cmp to i8
+  store i8 %conv2, i8* @glob, align 1
+  ret void
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_lleqsc_sext_z_store(i8 signext %a) {
+; CHECK-LABEL: test_lleqsc_sext_z_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    addis r4, r2, .LC0@toc@ha
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    ld r4, .LC0@toc@l(r4)
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    stb r3, 0(r4)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i8 %a, 0
+  %conv2 = sext i1 %cmp to i8
+  store i8 %conv2, i8* @glob, align 1
+  ret void
+}

llvm/trunk/test/CodeGen/PowerPC/testCompareslleqsi.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -verify-machineinstrs -mtriple=powerpc64-unknown-linux-gnu -O2 \
+; RUN:   -ppc-asm-full-reg-names -mcpu=pwr8 < %s | FileCheck %s \
+; RUN:  --implicit-check-not cmpw --implicit-check-not cmpd --implicit-check-not cmpl
+; RUN: llc -verify-machineinstrs -mtriple=powerpc64le-unknown-linux-gnu -O2 \
+; RUN:   -ppc-asm-full-reg-names -mcpu=pwr8 < %s | FileCheck %s \
+; RUN:  --implicit-check-not cmpw --implicit-check-not cmpd --implicit-check-not cmpl
+
+@glob = common local_unnamed_addr global i32 0, align 4
+
+; Function Attrs: norecurse nounwind readnone
+define i64 @test_lleqsi(i32 signext %a, i32 signext %b) {
+; CHECK-LABEL: test_lleqsi:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i32 %a, %b
+  %conv1 = zext i1 %cmp to i64
+  ret i64 %conv1
+}
+
+; Function Attrs: norecurse nounwind readnone
+define i64 @test_lleqsi_sext(i32 signext %a, i32 signext %b) {
+; CHECK-LABEL: test_lleqsi_sext:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i32 %a, %b
+  %conv1 = sext i1 %cmp to i64
+  ret i64 %conv1
+}
+
+; Function Attrs: norecurse nounwind readnone
+define i64 @test_lleqsi_z(i32 signext %a) {
+; CHECK-LABEL: test_lleqsi_z:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i32 %a, 0
+  %conv1 = zext i1 %cmp to i64
+  ret i64 %conv1
+}
+
+; Function Attrs: norecurse nounwind readnone
+define i64 @test_lleqsi_sext_z(i32 signext %a) {
+; CHECK-LABEL: test_lleqsi_sext_z:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i32 %a, 0
+  %conv1 = sext i1 %cmp to i64
+  ret i64 %conv1
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_lleqsi_store(i32 signext %a, i32 signext %b) {
+; CHECK-LABEL: test_lleqsi_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    addis r5, r2, .LC0@toc@ha
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    ld r12, .LC0@toc@l(r5)
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    stw r3, 0(r12)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i32 %a, %b
+  %conv = zext i1 %cmp to i32
+  store i32 %conv, i32* @glob, align 4
+  ret void
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_lleqsi_sext_store(i32 signext %a, i32 signext %b) {
+; CHECK-LABEL: test_lleqsi_sext_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    addis r5, r2, .LC0@toc@ha
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    ld r4, .LC0@toc@l(r5)
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    stw r3, 0(r4)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i32 %a, %b
+  %sub = sext i1 %cmp to i32
+  store i32 %sub, i32* @glob, align 4
+  ret void
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_lleqsi_z_store(i32 signext %a) {
+; CHECK-LABEL: test_lleqsi_z_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    addis r4, r2, .LC0@toc@ha
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    ld r4, .LC0@toc@l(r4)
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    stw r3, 0(r4)
+; CHECK-NEXT:    blr
+; CHECKNEXT:    blr
+entry:
+  %cmp = icmp eq i32 %a, 0
+  %conv = zext i1 %cmp to i32
+  store i32 %conv, i32* @glob, align 4
+  ret void
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_lleqsi_sext_z_store(i32 signext %a) {
+; CHECK-LABEL: test_lleqsi_sext_z_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    addis r4, r2, .LC0@toc@ha
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    ld r4, .LC0@toc@l(r4)
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    stw r3, 0(r4)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i32 %a, 0
+  %sub = sext i1 %cmp to i32
+  store i32 %sub, i32* @glob, align 4
+  ret void
+}

llvm/trunk/test/CodeGen/PowerPC/testCompareslleqss.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -verify-machineinstrs -mtriple=powerpc64-unknown-linux-gnu -O2 \
+; RUN:   -ppc-asm-full-reg-names -mcpu=pwr8 < %s | FileCheck %s \
+; RUN:  --implicit-check-not cmpw --implicit-check-not cmpd --implicit-check-not cmpl
+; RUN: llc -verify-machineinstrs -mtriple=powerpc64le-unknown-linux-gnu -O2 \
+; RUN:   -ppc-asm-full-reg-names -mcpu=pwr8 < %s | FileCheck %s \
+; RUN:  --implicit-check-not cmpw --implicit-check-not cmpd --implicit-check-not cmpl
+
+@glob = common local_unnamed_addr global i16 0, align 2
+
+; Function Attrs: norecurse nounwind readnone
+define i64 @test_lleqss(i16 signext %a, i16 signext %b) {
+; CHECK-LABEL: test_lleqss:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i16 %a, %b
+  %conv3 = zext i1 %cmp to i64
+  ret i64 %conv3
+}
+
+; Function Attrs: norecurse nounwind readnone
+define i64 @test_lleqss_sext(i16 signext %a, i16 signext %b) {
+; CHECK-LABEL: test_lleqss_sext:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i16 %a, %b
+  %conv3 = sext i1 %cmp to i64
+  ret i64 %conv3
+}
+
+; Function Attrs: norecurse nounwind readnone
+define i64 @test_lleqss_z(i16 signext %a) {
+; CHECK-LABEL: test_lleqss_z:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i16 %a, 0
+  %conv2 = zext i1 %cmp to i64
+  ret i64 %conv2
+}
+
+; Function Attrs: norecurse nounwind readnone
+define i64 @test_lleqss_sext_z(i16 signext %a) {
+; CHECK-LABEL: test_lleqss_sext_z:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i16 %a, 0
+  %conv2 = sext i1 %cmp to i64
+  ret i64 %conv2
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_lleqss_store(i16 signext %a, i16 signext %b) {
+; CHECK-LABEL: test_lleqss_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    addis r5, r2, .LC0@toc@ha
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    ld r12, .LC0@toc@l(r5)
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    sth r3, 0(r12)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i16 %a, %b
+  %conv3 = zext i1 %cmp to i16
+  store i16 %conv3, i16* @glob, align 2
+  ret void
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_lleqss_sext_store(i16 signext %a, i16 signext %b) {
+; CHECK-LABEL: test_lleqss_sext_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    addis r5, r2, .LC0@toc@ha
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    ld r4, .LC0@toc@l(r5)
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    sth r3, 0(r4)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i16 %a, %b
+  %conv3 = sext i1 %cmp to i16
+  store i16 %conv3, i16* @glob, align 2
+  ret void
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_lleqss_z_store(i16 signext %a) {
+; CHECK-LABEL: test_lleqss_z_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    addis r4, r2, .LC0@toc@ha
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    ld r4, .LC0@toc@l(r4)
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    sth r3, 0(r4)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i16 %a, 0
+  %conv2 = zext i1 %cmp to i16
+  store i16 %conv2, i16* @glob, align 2
+  ret void
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_lleqss_sext_z_store(i16 signext %a) {
+; CHECK-LABEL: test_lleqss_sext_z_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    addis r4, r2, .LC0@toc@ha
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    ld r4, .LC0@toc@l(r4)
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    sth r3, 0(r4)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i16 %a, 0
+  %conv2 = sext i1 %cmp to i16
+  store i16 %conv2, i16* @glob, align 2
+  ret void
+}

llvm/trunk/test/CodeGen/PowerPC/testComparesllequc.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -verify-machineinstrs -mtriple=powerpc64-unknown-linux-gnu -O2 \
+; RUN:   -ppc-asm-full-reg-names -mcpu=pwr8 < %s | FileCheck %s \
+; RUN:  --implicit-check-not cmpw --implicit-check-not cmpd --implicit-check-not cmpl
+; RUN: llc -verify-machineinstrs -mtriple=powerpc64le-unknown-linux-gnu -O2 \
+; RUN:   -ppc-asm-full-reg-names -mcpu=pwr8 < %s | FileCheck %s \
+; RUN:  --implicit-check-not cmpw --implicit-check-not cmpd --implicit-check-not cmpl
+
+@glob = common local_unnamed_addr global i8 0, align 1
+
+; Function Attrs: norecurse nounwind readnone
+define i64 @test_llequc(i8 zeroext %a, i8 zeroext %b) {
+; CHECK-LABEL: test_llequc:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i8 %a, %b
+  %conv3 = zext i1 %cmp to i64
+  ret i64 %conv3
+}
+
+; Function Attrs: norecurse nounwind readnone
+define i64 @test_llequc_sext(i8 zeroext %a, i8 zeroext %b) {
+; CHECK-LABEL: test_llequc_sext:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i8 %a, %b
+  %conv3 = sext i1 %cmp to i64
+  ret i64 %conv3
+}
+
+; Function Attrs: norecurse nounwind readnone
+define i64 @test_llequc_z(i8 zeroext %a) {
+; CHECK-LABEL: test_llequc_z:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i8 %a, 0
+  %conv2 = zext i1 %cmp to i64
+  ret i64 %conv2
+}
+
+; Function Attrs: norecurse nounwind readnone
+define i64 @test_llequc_sext_z(i8 zeroext %a) {
+; CHECK-LABEL: test_llequc_sext_z:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i8 %a, 0
+  %conv2 = sext i1 %cmp to i64
+  ret i64 %conv2
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_llequc_store(i8 zeroext %a, i8 zeroext %b) {
+; CHECK-LABEL: test_llequc_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    addis r5, r2, .LC0@toc@ha
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    ld r12, .LC0@toc@l(r5)
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    stb r3, 0(r12)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i8 %a, %b
+  %conv3 = zext i1 %cmp to i8
+  store i8 %conv3, i8* @glob, align 1
+  ret void
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_llequc_sext_store(i8 zeroext %a, i8 zeroext %b) {
+; CHECK-LABEL: test_llequc_sext_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    addis r5, r2, .LC0@toc@ha
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    ld r4, .LC0@toc@l(r5)
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    stb r3, 0(r4)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i8 %a, %b
+  %conv3 = sext i1 %cmp to i8
+  store i8 %conv3, i8* @glob, align 1
+  ret void
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_llequc_z_store(i8 zeroext %a) {
+; CHECK-LABEL: test_llequc_z_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    addis r4, r2, .LC0@toc@ha
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    ld r4, .LC0@toc@l(r4)
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    stb r3, 0(r4)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i8 %a, 0
+  %conv2 = zext i1 %cmp to i8
+  store i8 %conv2, i8* @glob, align 1
+  ret void
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_llequc_sext_z_store(i8 zeroext %a) {
+; CHECK-LABEL: test_llequc_sext_z_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    addis r4, r2, .LC0@toc@ha
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    ld r4, .LC0@toc@l(r4)
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    stb r3, 0(r4)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i8 %a, 0
+  %conv2 = sext i1 %cmp to i8
+  store i8 %conv2, i8* @glob, align 1
+  ret void
+}

llvm/trunk/test/CodeGen/PowerPC/testComparesllequi.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -verify-machineinstrs -mtriple=powerpc64-unknown-linux-gnu -O2 \
+; RUN:   -ppc-asm-full-reg-names -mcpu=pwr8 < %s | FileCheck %s \
+; RUN:  --implicit-check-not cmpw --implicit-check-not cmpd --implicit-check-not cmpl
+; RUN: llc -verify-machineinstrs -mtriple=powerpc64le-unknown-linux-gnu -O2 \
+; RUN:   -ppc-asm-full-reg-names -mcpu=pwr8 < %s | FileCheck %s \
+; RUN:  --implicit-check-not cmpw --implicit-check-not cmpd --implicit-check-not cmpl
+
+@glob = common local_unnamed_addr global i32 0, align 4
+
+; Function Attrs: norecurse nounwind readnone
+define i64 @test_llequi(i32 zeroext %a, i32 zeroext %b) {
+; CHECK-LABEL: test_llequi:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i32 %a, %b
+  %conv1 = zext i1 %cmp to i64
+  ret i64 %conv1
+}
+
+; Function Attrs: norecurse nounwind readnone
+define i64 @test_llequi_sext(i32 zeroext %a, i32 zeroext %b) {
+; CHECK-LABEL: test_llequi_sext:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i32 %a, %b
+  %conv1 = sext i1 %cmp to i64
+  ret i64 %conv1
+}
+
+; Function Attrs: norecurse nounwind readnone
+define i64 @test_llequi_z(i32 zeroext %a) {
+; CHECK-LABEL: test_llequi_z:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i32 %a, 0
+  %conv1 = zext i1 %cmp to i64
+  ret i64 %conv1
+}
+
+; Function Attrs: norecurse nounwind readnone
+define i64 @test_llequi_sext_z(i32 zeroext %a) {
+; CHECK-LABEL: test_llequi_sext_z:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i32 %a, 0
+  %conv1 = sext i1 %cmp to i64
+  ret i64 %conv1
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_llequi_store(i32 zeroext %a, i32 zeroext %b) {
+; CHECK-LABEL: test_llequi_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    addis r5, r2, .LC0@toc@ha
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    ld r12, .LC0@toc@l(r5)
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    stw r3, 0(r12)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i32 %a, %b
+  %conv = zext i1 %cmp to i32
+  store i32 %conv, i32* @glob, align 4
+  ret void
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_llequi_sext_store(i32 zeroext %a, i32 zeroext %b) {
+; CHECK-LABEL: test_llequi_sext_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    addis r5, r2, .LC0@toc@ha
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    ld r4, .LC0@toc@l(r5)
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    stw r3, 0(r4)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i32 %a, %b
+  %sub = sext i1 %cmp to i32
+  store i32 %sub, i32* @glob, align 4
+  ret void
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_llequi_z_store(i32 zeroext %a) {
+; CHECK-LABEL: test_llequi_z_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    addis r4, r2, .LC0@toc@ha
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    ld r4, .LC0@toc@l(r4)
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    stw r3, 0(r4)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i32 %a, 0
+  %conv = zext i1 %cmp to i32
+  store i32 %conv, i32* @glob, align 4
+  ret void
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_llequi_sext_z_store(i32 zeroext %a) {
+; CHECK-LABEL: test_llequi_sext_z_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    addis r4, r2, .LC0@toc@ha
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    ld r4, .LC0@toc@l(r4)
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    stw r3, 0(r4)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i32 %a, 0
+  %sub = sext i1 %cmp to i32
+  store i32 %sub, i32* @glob, align 4
+  ret void
+}

llvm/trunk/test/CodeGen/PowerPC/testComparesllequs.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -verify-machineinstrs -mtriple=powerpc64-unknown-linux-gnu -O2 \
+; RUN:   -ppc-asm-full-reg-names -mcpu=pwr8 < %s | FileCheck %s \
+; RUN:  --implicit-check-not cmpw --implicit-check-not cmpd --implicit-check-not cmpl
+; RUN: llc -verify-machineinstrs -mtriple=powerpc64le-unknown-linux-gnu -O2 \
+; RUN:   -ppc-asm-full-reg-names -mcpu=pwr8 < %s | FileCheck %s \
+; RUN:  --implicit-check-not cmpw --implicit-check-not cmpd --implicit-check-not cmpl
+
+@glob = common local_unnamed_addr global i16 0, align 2
+
+; Function Attrs: norecurse nounwind readnone
+define i64 @test_llequs(i16 zeroext %a, i16 zeroext %b) {
+; CHECK-LABEL: test_llequs:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i16 %a, %b
+  %conv3 = zext i1 %cmp to i64
+  ret i64 %conv3
+}
+
+; Function Attrs: norecurse nounwind readnone
+define i64 @test_llequs_sext(i16 zeroext %a, i16 zeroext %b) {
+; CHECK-LABEL: test_llequs_sext:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i16 %a, %b
+  %conv3 = sext i1 %cmp to i64
+  ret i64 %conv3
+}
+
+; Function Attrs: norecurse nounwind readnone
+define i64 @test_llequs_z(i16 zeroext %a) {
+; CHECK-LABEL: test_llequs_z:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i16 %a, 0
+  %conv2 = zext i1 %cmp to i64
+  ret i64 %conv2
+}
+
+; Function Attrs: norecurse nounwind readnone
+define i64 @test_llequs_sext_z(i16 zeroext %a) {
+; CHECK-LABEL: test_llequs_sext_z:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i16 %a, 0
+  %conv2 = sext i1 %cmp to i64
+  ret i64 %conv2
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_llequs_store(i16 zeroext %a, i16 zeroext %b) {
+; CHECK-LABEL: test_llequs_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    addis r5, r2, .LC0@toc@ha
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    ld r12, .LC0@toc@l(r5)
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    sth r3, 0(r12)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i16 %a, %b
+  %conv3 = zext i1 %cmp to i16
+  store i16 %conv3, i16* @glob, align 2
+  ret void
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_llequs_sext_store(i16 zeroext %a, i16 zeroext %b) {
+; CHECK-LABEL: test_llequs_sext_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    xor r3, r3, r4
+; CHECK-NEXT:    addis r5, r2, .LC0@toc@ha
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    ld r4, .LC0@toc@l(r5)
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    sth r3, 0(r4)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i16 %a, %b
+  %conv3 = sext i1 %cmp to i16
+  store i16 %conv3, i16* @glob, align 2
+  ret void
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_llequs_z_store(i16 zeroext %a) {
+; CHECK-LABEL: test_llequs_z_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    addis r4, r2, .LC0@toc@ha
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    ld r4, .LC0@toc@l(r4)
+; CHECK-NEXT:    srwi r3, r3, 5
+; CHECK-NEXT:    sth r3, 0(r4)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i16 %a, 0
+  %conv2 = zext i1 %cmp to i16
+  store i16 %conv2, i16* @glob, align 2
+  ret void
+}
+
+; Function Attrs: norecurse nounwind
+define void @test_llequs_sext_z_store(i16 zeroext %a) {
+; CHECK-LABEL: test_llequs_sext_z_store:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    addis r4, r2, .LC0@toc@ha
+; CHECK-NEXT:    cntlzw r3, r3
+; CHECK-NEXT:    ld r4, .LC0@toc@l(r4)
+; CHECK-NEXT:    rldicr r3, r3, 58, 0
+; CHECK-NEXT:    sradi r3, r3, 63
+; CHECK-NEXT:    sth r3, 0(r4)
+; CHECK-NEXT:    blr
+entry:
+  %cmp = icmp eq i16 %a, 0
+  %conv2 = sext i1 %cmp to i16
+  store i16 %conv2, i16* @glob, align 2
+  ret void
+}