[DAG] SimplifyDemandedBits - use KnownBits comparisons to remove ISD::UMIN/UMAX ops

[RKSimon]

Summary

Use the KnownBits icmp comparisons to determine when a ISD::UMIN/UMAX op is unnecessary should either op be known to be ULE/UGE than the other.

[foad]

LGTM. Any reason not to do this for SMAX and SMIN too?

[RKSimon]

In D94532#2495699, @foad wrote:

LGTM. Any reason not to do this for SMAX and SMIN too?

One step at a time :-) The UMIN/UMAX support will help some USUBSAT matching I'm working on, also we don't have test coverage yet and I haven't gotten around to creating SMIN/SMAX tests yet.

[RKSimon]

Anyone have any final comments - otherwise I'll commit this tomorrow.

[foad]

llvm/lib/CodeGen/SelectionDAG/TargetLowering.cpp

1745

I think this check can miss some opportunities because it is slightly asymmetrical. For example, with 4-bit numbers, if LHS = 110x and RHS = x100 then ule(LHS, RHS) will return None. However ult(LHS, RLHS) would return False, so you could optimise to RHS.

It does seem a shame to have to call both ule and ult just to catch this case. I wonder if there is a better way to code it.

[RKSimon]

llvm/lib/CodeGen/SelectionDAG/TargetLowering.cpp

1745

Yeah, it seems a shame but we don't have good alternatives - adding the ult/ugt test as well doesn't improve any additional existing tests - I can add coverage if you want though?

[foad]

llvm/lib/CodeGen/SelectionDAG/TargetLowering.cpp

1745

I'd appreciate it but I certainly won't insist. I think the current patch is fine as far as it goes.

[RKSimon]

add ult/ugt handling

[foad]

LGTM, thanks!

llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp

if (!TLI.isOperationLegal(Opcode, VT) &&
     case ISD::UMIN: AltOpcode = ISD::SMIN; break;
     case ISD::UMAX: AltOpcode = ISD::SMAX; break;
     default: llvm_unreachable("Unknown MINMAX opcode");
     }
     if (TLI.isOperationLegal(AltOpcode, VT))
       return DAG.getNode(AltOpcode, SDLoc(N), VT, N0, N1);
   }
 
+  // Simplify the operands using demanded-bits information.
+  if (SimplifyDemandedBits(SDValue(N, 0)))
+    return SDValue(N, 0);
+
   return SDValue();
 }
 
 /// If this is a bitwise logic instruction and both operands have the same
 /// opcode, try to sink the other opcode after the logic instruction.
 SDValue DAGCombiner::hoistLogicOpWithSameOpcodeHands(SDNode *N) {
   SDValue N0 = N->getOperand(0), N1 = N->getOperand(1);
   EVT VT = N0.getValueType();

llvm/lib/CodeGen/SelectionDAG/TargetLowering.cpp

case ISD::ROTR: {
     if (isPowerOf2_32(BitWidth)) {
       APInt DemandedAmtBits(Op1.getScalarValueSizeInBits(), BitWidth - 1);
       if (SimplifyDemandedBits(Op1, DemandedAmtBits, DemandedElts, Known2, TLO,
                                Depth + 1))
         return true;
     }
     break;
   }
+  case ISD::UMIN: {
+    // Check if one arg is always less than or equal to the other arg.
+    SDValue Op0 = Op.getOperand(0);
+    SDValue Op1 = Op.getOperand(1);
+    KnownBits Known0 = TLO.DAG.computeKnownBits(Op0, DemandedElts, Depth + 1);
+    KnownBits Known1 = TLO.DAG.computeKnownBits(Op1, DemandedElts, Depth + 1);
+    Known = KnownBits::umin(Known0, Known1);
+    if (Optional<bool> IsULE = KnownBits::ule(Known0, Known1))
+      return TLO.CombineTo(Op, IsULE.getValue() ? Op0 : Op1);
+    break;
+  }
+  case ISD::UMAX: {
+    // Check if one arg is always greater than or equal to the other arg.
+    SDValue Op0 = Op.getOperand(0);
+    SDValue Op1 = Op.getOperand(1);
+    KnownBits Known0 = TLO.DAG.computeKnownBits(Op0, DemandedElts, Depth + 1);
+    KnownBits Known1 = TLO.DAG.computeKnownBits(Op1, DemandedElts, Depth + 1);
+    Known = KnownBits::umax(Known0, Known1);
+    if (Optional<bool> IsUGE = KnownBits::uge(Known0, Known1))
+      return TLO.CombineTo(Op, IsUGE.getValue() ? Op0 : Op1);
+    break;
+  }
   case ISD::BITREVERSE: {
     SDValue Src = Op.getOperand(0);
     APInt DemandedSrcBits = DemandedBits.reverseBits();
     if (SimplifyDemandedBits(Src, DemandedSrcBits, DemandedElts, Known2, TLO,
                              Depth + 1))
       return true;
     Known.One = Known2.One.reverseBits();
     Known.Zero = Known2.Zero.reverseBits();

llvm/test/CodeGen/AMDGPU/r600-legalize-umax-bug.ll

 ; CHECK-NEXT:    PAD
 ; CHECK-NEXT:    ALU clause starting at 6:
 ; CHECK-NEXT:     MOV T0.X, literal.x,
 ; CHECK-NEXT:     MOV T0.Y, 0.0,
 ; CHECK-NEXT:     LSHR * T1.X, KC0[2].Y, literal.x,
 ; CHECK-NEXT:    2(2.802597e-45), 0(0.000000e+00)
 ; CHECK-NEXT:     MOV * T0.W, KC0[2].Y,
 ; CHECK-NEXT:    ALU clause starting at 11:
-; CHECK-NEXT:     MAX_UINT T0.X, T0.X, literal.x,
+; CHECK-NEXT:     MOV T0.X, literal.x,
 ; CHECK-NEXT:     MOV T0.Y, 0.0,
 ; CHECK-NEXT:     LSHR * T1.X, T0.W, literal.y,
 ; CHECK-NEXT:    4(5.605194e-45), 2(2.802597e-45)
 bb:
   store i64 2, i64 addrspace(1)* %out
   %tmp = load i64, i64 addrspace(1)* %out
   br label %jump
 
 jump:                                             ; preds = %bb
   %tmp1 = icmp ugt i64 %tmp, 4
   %umax = select i1 %tmp1, i64 %tmp, i64 4
   store i64 %umax, i64 addrspace(1)* %out
   ret void
 }

llvm/test/CodeGen/X86/sdiv_fix_sat.ll

 ; X64-NEXT:    pushq %r13
 ; X64-NEXT:    pushq %r12
 ; X64-NEXT:    pushq %rbx
 ; X64-NEXT:    subq $24, %rsp
 ; X64-NEXT:    movq %rsi, %rdx
 ; X64-NEXT:    movq %rsi, (%rsp) # 8-byte Spill
 ; X64-NEXT:    movq %rdi, %r15
 ; X64-NEXT:    leaq (%rdi,%rdi), %rax
-; X64-NEXT:    movq %rdi, %r12
-; X64-NEXT:    sarq $63, %r12
-; X64-NEXT:    shldq $31, %rax, %r12
+; X64-NEXT:    movq %rdi, %rbx
+; X64-NEXT:    sarq $63, %rbx
+; X64-NEXT:    shldq $31, %rax, %rbx
 ; X64-NEXT:    shlq $32, %r15
-; X64-NEXT:    movq %rsi, %r13
-; X64-NEXT:    sarq $63, %r13
+; X64-NEXT:    movq %rsi, %r12
+; X64-NEXT:    sarq $63, %r12
 ; X64-NEXT:    movq %r15, %rdi
-; X64-NEXT:    movq %r12, %rsi
-; X64-NEXT:    movq %r13, %rcx
+; X64-NEXT:    movq %rbx, %rsi
+; X64-NEXT:    movq %r12, %rcx
 ; X64-NEXT:    callq __divti3@PLT
-; X64-NEXT:    movq %rax, %rbx
+; X64-NEXT:    movq %rax, %r13
 ; X64-NEXT:    movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
 ; X64-NEXT:    movq %rdx, %rbp
 ; X64-NEXT:    movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
-; X64-NEXT:    subq $1, %rbx
+; X64-NEXT:    subq $1, %r13
 ; X64-NEXT:    sbbq $0, %rbp
-; X64-NEXT:    testq %r12, %r12
+; X64-NEXT:    testq %rbx, %rbx
 ; X64-NEXT:    sets %al
-; X64-NEXT:    testq %r13, %r13
+; X64-NEXT:    testq %r12, %r12
 ; X64-NEXT:    sets %r14b
 ; X64-NEXT:    xorb %al, %r14b
 ; X64-NEXT:    movq %r15, %rdi
-; X64-NEXT:    movq %r12, %rsi
+; X64-NEXT:    movq %rbx, %rsi
 ; X64-NEXT:    movq (%rsp), %rdx # 8-byte Reload
-; X64-NEXT:    movq %r13, %rcx
+; X64-NEXT:    movq %r12, %rcx
 ; X64-NEXT:    callq __modti3@PLT
 ; X64-NEXT:    orq %rax, %rdx
 ; X64-NEXT:    setne %al
 ; X64-NEXT:    testb %r14b, %al
 ; X64-NEXT:    cmoveq {{[-0-9]+}}(%r{{[sb]}}p), %rbp # 8-byte Folded Reload
-; X64-NEXT:    cmoveq {{[-0-9]+}}(%r{{[sb]}}p), %rbx # 8-byte Folded Reload
-; X64-NEXT:    cmpq $-1, %rbx
-; X64-NEXT:    movq $-1, %rax
-; X64-NEXT:    movq $-1, %rcx
-; X64-NEXT:    cmovbq %rbx, %rcx
-; X64-NEXT:    xorl %edx, %edx
+; X64-NEXT:    cmoveq {{[-0-9]+}}(%r{{[sb]}}p), %r13 # 8-byte Folded Reload
+; X64-NEXT:    xorl %eax, %eax
 ; X64-NEXT:    testq %rbp, %rbp
-; X64-NEXT:    cmovnsq %rax, %rbx
-; X64-NEXT:    cmoveq %rcx, %rbx
-; X64-NEXT:    cmovnsq %rdx, %rbp
-; X64-NEXT:    testq %rbx, %rbx
-; X64-NEXT:    movl $0, %ecx
-; X64-NEXT:    cmovaq %rbx, %rcx
+; X64-NEXT:    movq $-1, %rcx
+; X64-NEXT:    movq $-1, %rdx
+; X64-NEXT:    cmovsq %r13, %rdx
+; X64-NEXT:    cmoveq %r13, %rdx
+; X64-NEXT:    cmovnsq %rax, %rbp
 ; X64-NEXT:    testq %rbp, %rbp
-; X64-NEXT:    cmovnsq %rbp, %rax
-; X64-NEXT:    cmovsq %rdx, %rbx
+; X64-NEXT:    cmovnsq %rbp, %rcx
+; X64-NEXT:    cmovnsq %rdx, %rax
 ; X64-NEXT:    cmpq $-1, %rbp
-; X64-NEXT:    cmoveq %rcx, %rbx
-; X64-NEXT:    shrdq $1, %rax, %rbx
-; X64-NEXT:    movq %rbx, %rax
+; X64-NEXT:    cmoveq %rdx, %rax
+; X64-NEXT:    shrdq $1, %rcx, %rax
 ; X64-NEXT:    addq $24, %rsp
 ; X64-NEXT:    popq %rbx
 ; X64-NEXT:    popq %r12
 ; X64-NEXT:    popq %r13
 ; X64-NEXT:    popq %r14
 ; X64-NEXT:    popq %r15
 ; X64-NEXT:    popq %rbp
 ; X64-NEXT:    retq
 ;
 ; X86-LABEL: func5:
 ; X86:       # %bb.0:
 ; X86-NEXT:    pushl %ebp
 ; X86-NEXT:    movl %esp, %ebp
 ; X86-NEXT:    pushl %ebx
 ; X86-NEXT:    pushl %edi
 ; X86-NEXT:    pushl %esi
 ; X86-NEXT:    andl $-8, %esp
 ; X86-NEXT:    subl $88, %esp
 ; X86-NEXT:    movl 8(%ebp), %ecx
 ; X86-NEXT:    movl 12(%ebp), %eax
-; X86-NEXT:    movl 20(%ebp), %ebx
-; X86-NEXT:    sarl $31, %ebx
-; X86-NEXT:    movl %eax, %edi
+; X86-NEXT:    movl 20(%ebp), %edi
 ; X86-NEXT:    sarl $31, %edi
-; X86-NEXT:    movl %edi, %edx
-; X86-NEXT:    movl %edi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
+; X86-NEXT:    movl %eax, %ebx
+; X86-NEXT:    sarl $31, %ebx
+; X86-NEXT:    movl %ebx, %edx
+; X86-NEXT:    movl %ebx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
 ; X86-NEXT:    shldl $31, %eax, %edx
 ; X86-NEXT:    movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
 ; X86-NEXT:    shldl $31, %ecx, %eax
 ; X86-NEXT:    movl %eax, %esi
 ; X86-NEXT:    movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
 ; X86-NEXT:    shll $31, %ecx
 ; X86-NEXT:    movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
 ; X86-NEXT:    leal {{[0-9]+}}(%esp), %eax
-; X86-NEXT:    movl %ebx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
-; X86-NEXT:    pushl %ebx
-; X86-NEXT:    pushl %ebx
+; X86-NEXT:    pushl %edi
+; X86-NEXT:    pushl %edi
 ; X86-NEXT:    pushl 20(%ebp)
 ; X86-NEXT:    pushl 16(%ebp)
-; X86-NEXT:    pushl %edi
+; X86-NEXT:    pushl %ebx
 ; X86-NEXT:    pushl %edx
 ; X86-NEXT:    pushl %esi
 ; X86-NEXT:    pushl %ecx
 ; X86-NEXT:    pushl %eax
 ; X86-NEXT:    calll __divti3
 ; X86-NEXT:    addl $32, %esp
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %ecx
+; X86-NEXT:    movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %esi
 ; X86-NEXT:    movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
-; X86-NEXT:    movl {{[0-9]+}}(%esp), %edi
-; X86-NEXT:    movl %edi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
-; X86-NEXT:    subl $1, %esi
-; X86-NEXT:    sbbl $0, %edi
+; X86-NEXT:    subl $1, %eax
+; X86-NEXT:    movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
+; X86-NEXT:    sbbl $0, %esi
+; X86-NEXT:    movl %ecx, %eax
 ; X86-NEXT:    sbbl $0, %eax
 ; X86-NEXT:    movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %ebx
 ; X86-NEXT:    movl %ebx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
 ; X86-NEXT:    sbbl $0, %ebx
-; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload
-; X86-NEXT:    testl %edx, %edx
+; X86-NEXT:    testl %edi, %edi
 ; X86-NEXT:    sets %al
 ; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload
 ; X86-NEXT:    testl %ecx, %ecx
-; X86-NEXT:    sets %ah
-; X86-NEXT:    xorb %al, %ah
-; X86-NEXT:    movb %ah, {{[-0-9]+}}(%e{{[sb]}}p) # 1-byte Spill
+; X86-NEXT:    sets %dl
+; X86-NEXT:    xorb %al, %dl
+; X86-NEXT:    movb %dl, {{[-0-9]+}}(%e{{[sb]}}p) # 1-byte Spill
 ; X86-NEXT:    leal {{[0-9]+}}(%esp), %eax
-; X86-NEXT:    pushl %edx
-; X86-NEXT:    pushl %edx
+; X86-NEXT:    pushl %edi
+; X86-NEXT:    pushl %edi
 ; X86-NEXT:    pushl 20(%ebp)
 ; X86-NEXT:    pushl 16(%ebp)
 ; X86-NEXT:    pushl %ecx
 ; X86-NEXT:    pushl {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Reload
 ; X86-NEXT:    pushl {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Reload
 ; X86-NEXT:    pushl {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Reload
 ; X86-NEXT:    pushl %eax
 ; X86-NEXT:    calll __modti3
 ; X86-NEXT:    addl $32, %esp
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    orl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %ecx
 ; X86-NEXT:    orl {{[0-9]+}}(%esp), %ecx
 ; X86-NEXT:    orl %eax, %ecx
 ; X86-NEXT:    setne %al
 ; X86-NEXT:    testb %al, {{[-0-9]+}}(%e{{[sb]}}p) # 1-byte Folded Reload
+; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload
+; X86-NEXT:    cmovel {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Folded Reload
+; X86-NEXT:    movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
 ; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload
 ; X86-NEXT:    cmovel {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Folded Reload
 ; X86-NEXT:    movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
-; X86-NEXT:    cmovel {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Folded Reload
 ; X86-NEXT:    cmovel {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Folded Reload
 ; X86-NEXT:    cmovel {{[-0-9]+}}(%e{{[sb]}}p), %ebx # 4-byte Folded Reload
 ; X86-NEXT:    testl %ebx, %ebx
-; X86-NEXT:    movl $0, %ecx
-; X86-NEXT:    cmovsl %ebx, %ecx
-; X86-NEXT:    movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
-; X86-NEXT:    movl $-1, %ecx
-; X86-NEXT:    cmovsl %esi, %ecx
-; X86-NEXT:    movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
-; X86-NEXT:    movl $2147483647, %ecx # imm = 0x7FFFFFFF
-; X86-NEXT:    cmovsl %edi, %ecx
-; X86-NEXT:    movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
-; X86-NEXT:    movl %ebx, %edx
-; X86-NEXT:    sarl $31, %edx
-; X86-NEXT:    andl %eax, %edx
+; X86-NEXT:    movl $0, %edx
+; X86-NEXT:    cmovsl %ebx, %edx
+; X86-NEXT:    movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
+; X86-NEXT:    movl $2147483647, %edx # imm = 0x7FFFFFFF
+; X86-NEXT:    cmovsl %esi, %edx
+; X86-NEXT:    movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
+; X86-NEXT:    movl $-1, %edx
+; X86-NEXT:    cmovsl %eax, %edx
+; X86-NEXT:    movl %ebx, %edi
+; X86-NEXT:    sarl $31, %edi
+; X86-NEXT:    andl %ecx, %edi
 ; X86-NEXT:    testl %ebx, %ebx
-; X86-NEXT:    cmovel %ebx, %edx
-; X86-NEXT:    cmpl $-1, %esi
-; X86-NEXT:    movl $-1, %eax
-; X86-NEXT:    cmovbl %esi, %eax
-; X86-NEXT:    cmpl $2147483647, %edi # imm = 0x7FFFFFFF
-; X86-NEXT:    movl $-1, %ecx
-; X86-NEXT:    cmovael %ecx, %esi
-; X86-NEXT:    cmovel %eax, %esi
+; X86-NEXT:    cmovel %ebx, %edi
+; X86-NEXT:    cmpl $2147483647, %esi # imm = 0x7FFFFFFF
 ; X86-NEXT:    movl $2147483647, %eax # imm = 0x7FFFFFFF
-; X86-NEXT:    cmovael %eax, %edi
+; X86-NEXT:    cmovael %eax, %esi
+; X86-NEXT:    movl $-1, %eax
+; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload
+; X86-NEXT:    cmovbl %ecx, %eax
+; X86-NEXT:    cmovel %ecx, %eax
 ; X86-NEXT:    orl {{[-0-9]+}}(%e{{[sb]}}p), %ebx # 4-byte Folded Reload
-; X86-NEXT:    cmovnel {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Folded Reload
+; X86-NEXT:    cmovnel %edx, %eax
 ; X86-NEXT:    cmovnel {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Folded Reload
-; X86-NEXT:    testl %esi, %esi
-; X86-NEXT:    movl $0, %eax
-; X86-NEXT:    cmoval %esi, %eax
-; X86-NEXT:    cmpl $-2147483648, %edi # imm = 0x80000000
+; X86-NEXT:    cmpl $-2147483648, %esi # imm = 0x80000000
 ; X86-NEXT:    movl $0, %ecx
-; X86-NEXT:    cmoval %esi, %ecx
+; X86-NEXT:    cmoval %eax, %ecx
 ; X86-NEXT:    cmovel %eax, %ecx
-; X86-NEXT:    movl $-2147483648, %eax # imm = 0x80000000
-; X86-NEXT:    cmoval %edi, %eax
+; X86-NEXT:    movl $-2147483648, %edx # imm = 0x80000000
+; X86-NEXT:    cmoval %esi, %edx
 ; X86-NEXT:    cmpl $0, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Reload
-; X86-NEXT:    movl $-2147483648, %ebx # imm = 0x80000000
-; X86-NEXT:    cmovsl %ebx, %edi
 ; X86-NEXT:    movl $0, %ebx
+; X86-NEXT:    cmovsl %ebx, %eax
+; X86-NEXT:    movl $-2147483648, %ebx # imm = 0x80000000
 ; X86-NEXT:    cmovsl %ebx, %esi
-; X86-NEXT:    andl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload
-; X86-NEXT:    cmpl $-1, %edx
-; X86-NEXT:    cmovel %ecx, %esi
-; X86-NEXT:    cmovel %eax, %edi
-; X86-NEXT:    movl %esi, %eax
-; X86-NEXT:    movl %edi, %edx
+; X86-NEXT:    andl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Folded Reload
+; X86-NEXT:    cmpl $-1, %edi
+; X86-NEXT:    cmovel %edx, %esi
+; X86-NEXT:    cmovel %ecx, %eax
+; X86-NEXT:    movl %esi, %edx
 ; X86-NEXT:    leal -12(%ebp), %esp
 ; X86-NEXT:    popl %esi
 ; X86-NEXT:    popl %edi
 ; X86-NEXT:    popl %ebx
 ; X86-NEXT:    popl %ebp
 ; X86-NEXT:    retl
   %tmp = call i64 @llvm.sdiv.fix.sat.i64(i64 %x, i64 %y, i32 31)
   ret i64 %tmp

 ; X86-NEXT:    movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %edx
 ; X86-NEXT:    movl %edx, %ecx
 ; X86-NEXT:    sbbl $0, %ecx
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
 ; X86-NEXT:    sbbl $0, %eax
 ; X86-NEXT:    movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
-; X86-NEXT:    movl {{[0-9]+}}(%esp), %edi
-; X86-NEXT:    movl %edi, %eax
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %esi
+; X86-NEXT:    movl %esi, %eax
 ; X86-NEXT:    sbbl $0, %eax
 ; X86-NEXT:    movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
 ; X86-NEXT:    testl %ebx, %ebx
 ; X86-NEXT:    sets %bl
 ; X86-NEXT:    cmpl $0, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Reload
 ; X86-NEXT:    sets %bh
 ; X86-NEXT:    xorb %bl, %bh
-; X86-NEXT:    movl {{[0-9]+}}(%esp), %esi
-; X86-NEXT:    orl {{[0-9]+}}(%esp), %esi
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %edi
+; X86-NEXT:    orl {{[0-9]+}}(%esp), %edi
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    orl {{[0-9]+}}(%esp), %eax
-; X86-NEXT:    orl %esi, %eax
+; X86-NEXT:    orl %edi, %eax
 ; X86-NEXT:    setne %al
 ; X86-NEXT:    testb %bh, %al
 ; X86-NEXT:    cmovel %edx, %ecx
 ; X86-NEXT:    movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
 ; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload
 ; X86-NEXT:    cmovel {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Folded Reload
 ; X86-NEXT:    movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
 ; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload
 ; X86-NEXT:    cmovel {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Folded Reload
 ; X86-NEXT:    movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
 ; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload
-; X86-NEXT:    cmovel %edi, %eax
+; X86-NEXT:    cmovel %esi, %eax
 ; X86-NEXT:    movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
 ; X86-NEXT:    subl $1, %eax
 ; X86-NEXT:    movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %edi
-; X86-NEXT:    movl %edi, %eax
-; X86-NEXT:    sbbl $0, %eax
-; X86-NEXT:    movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
+; X86-NEXT:    movl %edi, %esi
+; X86-NEXT:    sbbl $0, %esi
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
 ; X86-NEXT:    sbbl $0, %eax
-; X86-NEXT:    movl %eax, %esi
+; X86-NEXT:    movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
 ; X86-NEXT:    sbbl $0, %eax
 ; X86-NEXT:    movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
 ; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload
 ; X86-NEXT:    testl %ecx, %ecx
 ; X86-NEXT:    sets %al
 ; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload

 ; X86-NEXT:    addl $32, %esp
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    orl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %ecx
 ; X86-NEXT:    orl {{[0-9]+}}(%esp), %ecx
 ; X86-NEXT:    orl %eax, %ecx
 ; X86-NEXT:    setne %al
 ; X86-NEXT:    testb %bl, %al
+; X86-NEXT:    cmovel %edi, %esi
+; X86-NEXT:    movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
 ; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload
-; X86-NEXT:    cmovel %edi, %eax
+; X86-NEXT:    cmovel {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Folded Reload
 ; X86-NEXT:    movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
 ; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload
 ; X86-NEXT:    cmovel {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Folded Reload
 ; X86-NEXT:    movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
-; X86-NEXT:    cmovel {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Folded Reload
-; X86-NEXT:    movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
 ; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload
 ; X86-NEXT:    cmovel {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Folded Reload
 ; X86-NEXT:    movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
-; X86-NEXT:    movl {{[0-9]+}}(%esp), %edx
-; X86-NEXT:    movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
-; X86-NEXT:    subl $1, %edx
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %ecx
+; X86-NEXT:    movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
+; X86-NEXT:    subl $1, %ecx
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %esi
 ; X86-NEXT:    movl %esi, %eax
 ; X86-NEXT:    sbbl $0, %eax
 ; X86-NEXT:    movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
 ; X86-NEXT:    sbbl $0, %eax
 ; X86-NEXT:    movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
-; X86-NEXT:    movl {{[0-9]+}}(%esp), %ecx
-; X86-NEXT:    movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
-; X86-NEXT:    sbbl $0, %ecx
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %edx
+; X86-NEXT:    movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
+; X86-NEXT:    sbbl $0, %edx
 ; X86-NEXT:    cmpl $0, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Reload
 ; X86-NEXT:    sets %bl
 ; X86-NEXT:    cmpl $0, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Reload
 ; X86-NEXT:    sets %bh
 ; X86-NEXT:    xorb %bl, %bh
-; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
-; X86-NEXT:    orl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %edi
 ; X86-NEXT:    orl {{[0-9]+}}(%esp), %edi
-; X86-NEXT:    orl %eax, %edi
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; X86-NEXT:    orl {{[0-9]+}}(%esp), %eax
+; X86-NEXT:    orl %edi, %eax
 ; X86-NEXT:    setne %al
 ; X86-NEXT:    testb %bh, %al
 ; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload
 ; X86-NEXT:    cmovel %esi, %eax
 ; X86-NEXT:    movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
 ; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload
 ; X86-NEXT:    cmovel {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Folded Reload
 ; X86-NEXT:    movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
-; X86-NEXT:    cmovel {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload
-; X86-NEXT:    movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
 ; X86-NEXT:    cmovel {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Folded Reload
-; X86-NEXT:    testl %ecx, %ecx
+; X86-NEXT:    movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
+; X86-NEXT:    cmovel {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload
+; X86-NEXT:    testl %edx, %edx
 ; X86-NEXT:    movl $0, %eax
-; X86-NEXT:    cmovsl %ecx, %eax
+; X86-NEXT:    cmovsl %edx, %eax
 ; X86-NEXT:    movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
 ; X86-NEXT:    movl $-1, %eax
-; X86-NEXT:    cmovsl %edx, %eax
+; X86-NEXT:    cmovsl %ecx, %eax
 ; X86-NEXT:    movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
 ; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload
-; X86-NEXT:    movl %eax, %edx
-; X86-NEXT:    sarl $31, %edx
-; X86-NEXT:    movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
-; X86-NEXT:    movl %edx, %esi
+; X86-NEXT:    movl %eax, %ecx
+; X86-NEXT:    sarl $31, %ecx
+; X86-NEXT:    movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
+; X86-NEXT:    movl %ecx, %esi
 ; X86-NEXT:    andl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Folded Reload
 ; X86-NEXT:    testl %eax, %eax
 ; X86-NEXT:    cmovel %eax, %esi
 ; X86-NEXT:    movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
-; X86-NEXT:    movl $0, %edx
-; X86-NEXT:    cmovsl %eax, %edx
-; X86-NEXT:    movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
+; X86-NEXT:    movl $0, %ecx
+; X86-NEXT:    cmovsl %eax, %ecx
+; X86-NEXT:    movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
 ; X86-NEXT:    movl $-1, %eax
 ; X86-NEXT:    cmovsl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Folded Reload
 ; X86-NEXT:    movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
 ; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload
-; X86-NEXT:    movl %eax, %ebx
-; X86-NEXT:    sarl $31, %ebx
-; X86-NEXT:    movl %ebx, %edx
-; X86-NEXT:    andl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload
+; X86-NEXT:    movl %eax, %edi
+; X86-NEXT:    sarl $31, %edi
+; X86-NEXT:    movl %edi, %ecx
+; X86-NEXT:    andl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Folded Reload
 ; X86-NEXT:    testl %eax, %eax
-; X86-NEXT:    cmovel %eax, %edx
-; X86-NEXT:    movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
-; X86-NEXT:    movl $0, %edx
-; X86-NEXT:    cmovsl %eax, %edx
-; X86-NEXT:    movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
+; X86-NEXT:    cmovel %eax, %ecx
+; X86-NEXT:    movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
+; X86-NEXT:    movl $0, %ecx
+; X86-NEXT:    cmovsl %eax, %ecx
+; X86-NEXT:    movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
 ; X86-NEXT:    movl $-1, %eax
 ; X86-NEXT:    cmovsl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Folded Reload
 ; X86-NEXT:    movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
-; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload
-; X86-NEXT:    movl %eax, %edi
-; X86-NEXT:    sarl $31, %edi
-; X86-NEXT:    movl %edi, %edx
-; X86-NEXT:    andl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload
-; X86-NEXT:    testl %eax, %eax
-; X86-NEXT:    cmovel %eax, %edx
-; X86-NEXT:    movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
-; X86-NEXT:    movl $0, %esi
-; X86-NEXT:    cmovsl %eax, %esi
+; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Reload
+; X86-NEXT:    movl %esi, %ebx
+; X86-NEXT:    sarl $31, %ebx
+; X86-NEXT:    movl %ebx, %eax
+; X86-NEXT:    andl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Folded Reload
+; X86-NEXT:    testl %esi, %esi
+; X86-NEXT:    cmovel %esi, %eax
+; X86-NEXT:    movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
+; X86-NEXT:    movl $0, %ecx
+; X86-NEXT:    cmovsl %esi, %ecx
 ; X86-NEXT:    movl $-1, %eax
-; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload
-; X86-NEXT:    cmovsl %edx, %eax
+; X86-NEXT:    cmovsl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Folded Reload
 ; X86-NEXT:    movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
-; X86-NEXT:    movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
-; X86-NEXT:    movl %ecx, %eax
+; X86-NEXT:    movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
+; X86-NEXT:    movl %edx, %eax
 ; X86-NEXT:    sarl $31, %eax
 ; X86-NEXT:    movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
 ; X86-NEXT:    andl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Folded Reload
-; X86-NEXT:    testl %ecx, %ecx
-; X86-NEXT:    cmovel %ecx, %eax
-; X86-NEXT:    movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
-; X86-NEXT:    movl %edx, %ecx
-; X86-NEXT:    cmpl $-1, %edx
-; X86-NEXT:    movl $-1, %eax
-; X86-NEXT:    cmovael %eax, %ecx
-; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload
-; X86-NEXT:    cmpl $1, %edx
-; X86-NEXT:    movl $0, %eax
-; X86-NEXT:    sbbl %eax, %eax
-; X86-NEXT:    notl %eax
-; X86-NEXT:    orl %ecx, %eax
 ; X86-NEXT:    testl %edx, %edx
-; X86-NEXT:    movl $0, %ecx
-; X86-NEXT:    cmovbl %edx, %ecx
-; X86-NEXT:    andl %edx, %edi
-; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload
+; X86-NEXT:    cmovel %edx, %eax
+; X86-NEXT:    movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
+; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload
+; X86-NEXT:    andl %eax, %ebx
+; X86-NEXT:    cmpl $1, %eax
+; X86-NEXT:    movl $0, %edx
+; X86-NEXT:    sbbl %edx, %edx
+; X86-NEXT:    notl %edx
 ; X86-NEXT:    orl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload
-; X86-NEXT:    cmovel %ecx, %edi
-; X86-NEXT:    cmovnel {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Folded Reload
-; X86-NEXT:    testl %eax, %eax
+; X86-NEXT:    orl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Folded Reload
+; X86-NEXT:    cmovnel {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload
+; X86-NEXT:    movl $0, %esi
+; X86-NEXT:    cmovel %esi, %ebx
+; X86-NEXT:    cmpl $-1, %ebx
+; X86-NEXT:    movl $0, %eax
+; X86-NEXT:    cmovel %edx, %eax
+; X86-NEXT:    testl %ecx, %ecx
+; X86-NEXT:    cmovsl %esi, %edx
+; X86-NEXT:    movl $-1, %esi
+; X86-NEXT:    cmovsl %esi, %ebx
+; X86-NEXT:    andl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Folded Reload
+; X86-NEXT:    cmpl $-1, %ecx
+; X86-NEXT:    cmovnel %ebx, %ecx
+; X86-NEXT:    cmovel %eax, %edx
+; X86-NEXT:    shrdl $1, %ecx, %edx
+; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload
+; X86-NEXT:    andl %eax, %edi
+; X86-NEXT:    cmpl $1, %eax
 ; X86-NEXT:    movl $0, %ecx
-; X86-NEXT:    cmoval %eax, %ecx
+; X86-NEXT:    sbbl %ecx, %ecx
+; X86-NEXT:    notl %ecx
+; X86-NEXT:    orl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Folded Reload
+; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload
+; X86-NEXT:    orl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Folded Reload
+; X86-NEXT:    cmovnel {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Folded Reload
+; X86-NEXT:    movl $0, %esi
+; X86-NEXT:    cmovel %esi, %edi
 ; X86-NEXT:    cmpl $-1, %edi
-; X86-NEXT:    movl $0, %edx
-; X86-NEXT:    cmovnel %edx, %ecx
-; X86-NEXT:    testl %esi, %esi
-; X86-NEXT:    movl $-1, %edx
-; X86-NEXT:    cmovsl %edx, %edi
-; X86-NEXT:    movl $0, %edx
-; X86-NEXT:    cmovsl %edx, %eax
-; X86-NEXT:    andl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Folded Reload
-; X86-NEXT:    cmpl $-1, %esi
+; X86-NEXT:    movl $0, %eax
 ; X86-NEXT:    cmovel %ecx, %eax
-; X86-NEXT:    cmovnel %edi, %esi
-; X86-NEXT:    shldl $31, %eax, %esi
-; X86-NEXT:    movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
+; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %ebx # 4-byte Reload
+; X86-NEXT:    testl %ebx, %ebx
+; X86-NEXT:    cmovsl %esi, %ecx
+; X86-NEXT:    movl $-1, %esi
+; X86-NEXT:    cmovsl %esi, %edi
+; X86-NEXT:    andl {{[-0-9]+}}(%e{{[sb]}}p), %ebx # 4-byte Folded Reload
+; X86-NEXT:    cmpl $-1, %ebx
+; X86-NEXT:    cmovnel %edi, %ebx
+; X86-NEXT:    cmovel %eax, %ecx
+; X86-NEXT:    shrdl $1, %ebx, %ecx
 ; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload
+; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload
+; X86-NEXT:    andl %eax, %edi
+; X86-NEXT:    cmpl $1, %eax
+; X86-NEXT:    movl $0, %esi
+; X86-NEXT:    sbbl %esi, %esi
+; X86-NEXT:    notl %esi
+; X86-NEXT:    orl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Folded Reload
+; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload
+; X86-NEXT:    orl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Folded Reload
+; X86-NEXT:    cmovnel {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Folded Reload
+; X86-NEXT:    movl %edi, %eax
+; X86-NEXT:    movl $0, %edi
+; X86-NEXT:    cmovel %edi, %eax
 ; X86-NEXT:    cmpl $-1, %eax
-; X86-NEXT:    movl $-1, %ecx
-; X86-NEXT:    cmovael %ecx, %eax
-; X86-NEXT:    movl %eax, %ecx
-; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Reload
-; X86-NEXT:    cmpl $1, %esi
+; X86-NEXT:    movl %eax, %edi
 ; X86-NEXT:    movl $0, %eax
-; X86-NEXT:    sbbl %eax, %eax
-; X86-NEXT:    notl %eax
-; X86-NEXT:    orl %ecx, %eax
-; X86-NEXT:    testl %esi, %esi
-; X86-NEXT:    movl $0, %ecx
-; X86-NEXT:    cmovbl %esi, %ecx
-; X86-NEXT:    andl %esi, %ebx
-; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload
-; X86-NEXT:    orl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload
-; X86-NEXT:    cmovel %ecx, %ebx
-; X86-NEXT:    cmovnel {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Folded Reload
-; X86-NEXT:    testl %eax, %eax
-; X86-NEXT:    movl $0, %ecx
-; X86-NEXT:    cmoval %eax, %ecx
+; X86-NEXT:    cmovel %esi, %eax
+; X86-NEXT:    movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
+; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %ebx # 4-byte Reload
+; X86-NEXT:    testl %ebx, %ebx
+; X86-NEXT:    movl $0, %eax
+; X86-NEXT:    cmovsl %eax, %esi
+; X86-NEXT:    movl $-1, %eax
+; X86-NEXT:    cmovsl %eax, %edi
+; X86-NEXT:    andl {{[-0-9]+}}(%e{{[sb]}}p), %ebx # 4-byte Folded Reload
 ; X86-NEXT:    cmpl $-1, %ebx
+; X86-NEXT:    cmovnel %edi, %ebx
+; X86-NEXT:    cmovel {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Folded Reload
+; X86-NEXT:    shrdl $1, %ebx, %esi
+; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload
+; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %ebx # 4-byte Reload
+; X86-NEXT:    andl %eax, %ebx
+; X86-NEXT:    cmpl $1, %eax
 ; X86-NEXT:    movl $0, %edi
-; X86-NEXT:    cmovnel %edi, %ecx
-; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Reload
-; X86-NEXT:    testl %esi, %esi
-; X86-NEXT:    movl $-1, %edx
-; X86-NEXT:    cmovsl %edx, %ebx
-; X86-NEXT:    cmovsl %edi, %eax
-; X86-NEXT:    andl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Folded Reload
-; X86-NEXT:    cmpl $-1, %esi
-; X86-NEXT:    cmovel %ecx, %eax
-; X86-NEXT:    cmovnel %ebx, %esi
-; X86-NEXT:    shldl $31, %eax, %esi
-; X86-NEXT:    movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
+; X86-NEXT:    sbbl %edi, %edi
+; X86-NEXT:    notl %edi
+; X86-NEXT:    orl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Folded Reload
 ; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload
+; X86-NEXT:    orl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Folded Reload
+; X86-NEXT:    cmovnel {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Folded Reload
+; X86-NEXT:    movl %ebx, %eax
+; X86-NEXT:    movl $0, %ebx
+; X86-NEXT:    cmovel %ebx, %eax
 ; X86-NEXT:    cmpl $-1, %eax
-; X86-NEXT:    cmovael %edx, %eax
-; X86-NEXT:    movl $-1, %ebx
-; X86-NEXT:    movl %eax, %ecx
-; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload
-; X86-NEXT:    cmpl $1, %edx
-; X86-NEXT:    movl $0, %eax
-; X86-NEXT:    sbbl %eax, %eax
-; X86-NEXT:    notl %eax
-; X86-NEXT:    orl %ecx, %eax
-; X86-NEXT:    testl %edx, %edx
-; X86-NEXT:    movl $0, %ecx
-; X86-NEXT:    cmovbl %edx, %ecx
-; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload
-; X86-NEXT:    andl %edx, %edi
-; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload
-; X86-NEXT:    orl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload
-; X86-NEXT:    cmovel %ecx, %edi
-; X86-NEXT:    cmovnel {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Folded Reload
-; X86-NEXT:    testl %eax, %eax
-; X86-NEXT:    movl $0, %ecx
-; X86-NEXT:    cmoval %eax, %ecx
-; X86-NEXT:    cmpl $-1, %edi
-; X86-NEXT:    movl $0, %edx
-; X86-NEXT:    cmovnel %edx, %ecx
-; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Reload
-; X86-NEXT:    testl %esi, %esi
+; X86-NEXT:    movl $0, %ebx
+; X86-NEXT:    cmovel %edi, %ebx
+; X86-NEXT:    movl %ebx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
+; X86-NEXT:    cmpl $0, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Reload
+; X86-NEXT:    movl $0, %ebx
 ; X86-NEXT:    cmovsl %ebx, %edi
-; X86-NEXT:    cmovsl %edx, %eax
-; X86-NEXT:    andl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Folded Reload
-; X86-NEXT:    cmpl $-1, %esi
-; X86-NEXT:    cmovel %ecx, %eax
-; X86-NEXT:    cmovnel %edi, %esi
-; X86-NEXT:    shldl $31, %eax, %esi
-; X86-NEXT:    movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
+; X86-NEXT:    movl $-1, %ebx
+; X86-NEXT:    cmovsl %ebx, %eax
+; X86-NEXT:    movl %eax, %ebx
 ; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload
+; X86-NEXT:    andl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Folded Reload
 ; X86-NEXT:    cmpl $-1, %eax
-; X86-NEXT:    cmovael %ebx, %eax
-; X86-NEXT:    movl $-1, %esi
-; X86-NEXT:    movl %eax, %ecx
-; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload
-; X86-NEXT:    cmpl $1, %edx
-; X86-NEXT:    movl $0, %eax
-; X86-NEXT:    sbbl %eax, %eax
-; X86-NEXT:    notl %eax
-; X86-NEXT:    orl %ecx, %eax
-; X86-NEXT:    testl %edx, %edx
-; X86-NEXT:    movl $0, %ecx
-; X86-NEXT:    cmovbl %edx, %ecx
-; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %ebx # 4-byte Reload
-; X86-NEXT:    andl %edx, %ebx
-; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload
-; X86-NEXT:    orl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload
-; X86-NEXT:    cmovel %ecx, %ebx
-; X86-NEXT:    cmovnel {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Folded Reload
-; X86-NEXT:    testl %eax, %eax
-; X86-NEXT:    movl $0, %ecx
-; X86-NEXT:    cmoval %eax, %ecx
-; X86-NEXT:    cmpl $-1, %ebx
-; X86-NEXT:    movl $0, %edi
-; X86-NEXT:    cmovnel %edi, %ecx
-; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload
-; X86-NEXT:    testl %edx, %edx
-; X86-NEXT:    cmovsl %esi, %ebx
-; X86-NEXT:    movl %ebx, %esi
-; X86-NEXT:    cmovsl %edi, %eax
-; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %ebx # 4-byte Reload
-; X86-NEXT:    andl %edx, %ebx
-; X86-NEXT:    cmpl $-1, %ebx
-; X86-NEXT:    cmovel %ecx, %eax
-; X86-NEXT:    cmovnel %esi, %ebx
-; X86-NEXT:    shldl $31, %eax, %ebx
+; X86-NEXT:    cmovnel %ebx, %eax
+; X86-NEXT:    cmovel {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Folded Reload
+; X86-NEXT:    shrdl $1, %eax, %edi
 ; X86-NEXT:    movl 8(%ebp), %eax
-; X86-NEXT:    movl %ebx, 12(%eax)
-; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload
-; X86-NEXT:    movl %ecx, 8(%eax)
-; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload
+; X86-NEXT:    movl %edi, 12(%eax)
+; X86-NEXT:    movl %esi, 8(%eax)
 ; X86-NEXT:    movl %ecx, 4(%eax)
-; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload
-; X86-NEXT:    movl %ecx, (%eax)
+; X86-NEXT:    movl %edx, (%eax)
 ; X86-NEXT:    leal -12(%ebp), %esp
 ; X86-NEXT:    popl %esi
 ; X86-NEXT:    popl %edi
 ; X86-NEXT:    popl %ebx
 ; X86-NEXT:    popl %ebp
 ; X86-NEXT:    retl $4
   %tmp = call <4 x i32> @llvm.sdiv.fix.sat.v4i32(<4 x i32> %x, <4 x i32> %y, i32 31)
   ret <4 x i32> %tmp
 }

llvm/test/CodeGen/X86/udiv_fix_sat.ll

 ; X64-NEXT:    leaq (%rdi,%rdi), %rsi
 ; X64-NEXT:    movq %rdi, %rax
 ; X64-NEXT:    shrq $63, %rax
 ; X64-NEXT:    shrdq $33, %rax, %rsi
 ; X64-NEXT:    shlq $32, %rdi
 ; X64-NEXT:    xorl %ebx, %ebx
 ; X64-NEXT:    xorl %ecx, %ecx
 ; X64-NEXT:    callq __udivti3@PLT
-; X64-NEXT:    cmpq $-1, %rax
-; X64-NEXT:    movq $-1, %rcx
-; X64-NEXT:    cmovbq %rax, %rcx
 ; X64-NEXT:    cmpq $1, %rdx
-; X64-NEXT:    movl $1, %esi
-; X64-NEXT:    cmovbq %rdx, %rsi
 ; X64-NEXT:    sbbq %rbx, %rbx
 ; X64-NEXT:    notq %rbx
 ; X64-NEXT:    orq %rax, %rbx
 ; X64-NEXT:    cmpq $1, %rdx
-; X64-NEXT:    cmoveq %rcx, %rbx
-; X64-NEXT:    shrdq $1, %rsi, %rbx
+; X64-NEXT:    movl $1, %ecx
+; X64-NEXT:    cmovbq %rdx, %rcx
+; X64-NEXT:    cmoveq %rax, %rbx
+; X64-NEXT:    shrdq $1, %rcx, %rbx
 ; X64-NEXT:    movq %rbx, %rax
 ; X64-NEXT:    popq %rbx
 ; X64-NEXT:    retq
 ;
 ; X86-LABEL: func5:
 ; X86:       # %bb.0:
 ; X86-NEXT:    pushl %ebp
 ; X86-NEXT:    movl %esp, %ebp

 ; X86-NEXT:    pushl 16(%ebp)
 ; X86-NEXT:    pushl $0
 ; X86-NEXT:    pushl %edx
 ; X86-NEXT:    pushl %ecx
 ; X86-NEXT:    pushl %eax
 ; X86-NEXT:    pushl %esi
 ; X86-NEXT:    calll __udivti3
 ; X86-NEXT:    addl $32, %esp
-; X86-NEXT:    movl (%esp), %eax
+; X86-NEXT:    movl (%esp), %ecx
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %edx
-; X86-NEXT:    cmpl $-1, %eax
-; X86-NEXT:    movl $-1, %ecx
-; X86-NEXT:    movl $-1, %esi
-; X86-NEXT:    cmovbl %eax, %esi
 ; X86-NEXT:    cmpl $-1, %edx
+; X86-NEXT:    movl %ecx, %eax
 ; X86-NEXT:    cmovel %edx, %eax
-; X86-NEXT:    cmovel %esi, %eax
-; X86-NEXT:    cmovael %ecx, %edx
-; X86-NEXT:    movl {{[0-9]+}}(%esp), %esi
-; X86-NEXT:    orl {{[0-9]+}}(%esp), %esi
+; X86-NEXT:    cmovel %ecx, %eax
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %ecx
+; X86-NEXT:    orl {{[0-9]+}}(%esp), %ecx
+; X86-NEXT:    movl $-1, %ecx
 ; X86-NEXT:    cmovnel %ecx, %edx
 ; X86-NEXT:    cmovnel %ecx, %eax
 ; X86-NEXT:    leal -4(%ebp), %esp
 ; X86-NEXT:    popl %esi
 ; X86-NEXT:    popl %ebp
 ; X86-NEXT:    retl
   %tmp = call i64 @llvm.udiv.fix.sat.i64(i64 %x, i64 %y, i32 31)
   ret i64 %tmp

 ; X64-NEXT:    retq
 ;
 ; X86-LABEL: vec:
 ; X86:       # %bb.0:
 ; X86-NEXT:    pushl %ebp
 ; X86-NEXT:    pushl %ebx
 ; X86-NEXT:    pushl %edi
 ; X86-NEXT:    pushl %esi
-; X86-NEXT:    subl $16, %esp
-; X86-NEXT:    movl {{[0-9]+}}(%esp), %esi
+; X86-NEXT:    subl $12, %esp
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %edi
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %ecx
 ; X86-NEXT:    xorl %eax, %eax
 ; X86-NEXT:    addl %ecx, %ecx
 ; X86-NEXT:    setb %al
 ; X86-NEXT:    shldl $31, %ecx, %eax
 ; X86-NEXT:    shll $31, %ecx
 ; X86-NEXT:    pushl $0
 ; X86-NEXT:    pushl {{[0-9]+}}(%esp)
 ; X86-NEXT:    pushl %eax
 ; X86-NEXT:    pushl %ecx
 ; X86-NEXT:    calll __udivdi3
 ; X86-NEXT:    addl $16, %esp
-; X86-NEXT:    cmpl $-1, %eax
-; X86-NEXT:    movl $-1, %ecx
-; X86-NEXT:    cmovbl %eax, %ecx
 ; X86-NEXT:    cmpl $1, %edx
-; X86-NEXT:    movl $0, %edi
-; X86-NEXT:    sbbl %edi, %edi
-; X86-NEXT:    notl %edi
-; X86-NEXT:    orl %eax, %edi
-; X86-NEXT:    movl %edi, %ebx
-; X86-NEXT:    xorl %eax, %eax
-; X86-NEXT:    addl %esi, %esi
-; X86-NEXT:    setb %al
+; X86-NEXT:    movl $0, %ecx
+; X86-NEXT:    sbbl %ecx, %ecx
+; X86-NEXT:    notl %ecx
+; X86-NEXT:    orl %eax, %ecx
+; X86-NEXT:    movl %ecx, %esi
+; X86-NEXT:    xorl %ecx, %ecx
+; X86-NEXT:    addl %edi, %edi
+; X86-NEXT:    setb %cl
 ; X86-NEXT:    cmpl $1, %edx
-; X86-NEXT:    movl {{[0-9]+}}(%esp), %edi
-; X86-NEXT:    cmovel %ecx, %ebx
-; X86-NEXT:    movl %ebx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
-; X86-NEXT:    movl $1, %ecx
-; X86-NEXT:    cmovael %ecx, %edx
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %ebx
+; X86-NEXT:    movl $1, %ebp
+; X86-NEXT:    cmovael %ebp, %edx
 ; X86-NEXT:    movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
-; X86-NEXT:    shldl $31, %esi, %eax
-; X86-NEXT:    shll $31, %esi
-; X86-NEXT:    pushl $0
-; X86-NEXT:    pushl {{[0-9]+}}(%esp)
-; X86-NEXT:    pushl %eax
-; X86-NEXT:    pushl %esi
-; X86-NEXT:    calll __udivdi3
-; X86-NEXT:    addl $16, %esp
-; X86-NEXT:    cmpl $-1, %eax
-; X86-NEXT:    movl $-1, %ecx
-; X86-NEXT:    cmovbl %eax, %ecx
-; X86-NEXT:    cmpl $1, %edx
-; X86-NEXT:    movl $1, %esi
-; X86-NEXT:    cmovbl %edx, %esi
+; X86-NEXT:    movl $1, %ebp
+; X86-NEXT:    cmovel %eax, %esi
 ; X86-NEXT:    movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
-; X86-NEXT:    movl $0, %esi
-; X86-NEXT:    sbbl %esi, %esi
-; X86-NEXT:    notl %esi
-; X86-NEXT:    orl %eax, %esi
-; X86-NEXT:    xorl %eax, %eax
-; X86-NEXT:    addl %edi, %edi
-; X86-NEXT:    setb %al
-; X86-NEXT:    cmpl $1, %edx
-; X86-NEXT:    movl {{[0-9]+}}(%esp), %ebp
-; X86-NEXT:    cmovel %ecx, %esi
-; X86-NEXT:    shldl $31, %edi, %eax
+; X86-NEXT:    shldl $31, %edi, %ecx
 ; X86-NEXT:    shll $31, %edi
 ; X86-NEXT:    pushl $0
 ; X86-NEXT:    pushl {{[0-9]+}}(%esp)
-; X86-NEXT:    pushl %eax
+; X86-NEXT:    pushl %ecx
 ; X86-NEXT:    pushl %edi
 ; X86-NEXT:    calll __udivdi3
 ; X86-NEXT:    addl $16, %esp
-; X86-NEXT:    cmpl $-1, %eax
-; X86-NEXT:    movl $-1, %ebx
-; X86-NEXT:    cmovbl %eax, %ebx
 ; X86-NEXT:    cmpl $1, %edx
 ; X86-NEXT:    movl $0, %edi
 ; X86-NEXT:    sbbl %edi, %edi
 ; X86-NEXT:    notl %edi
 ; X86-NEXT:    orl %eax, %edi
 ; X86-NEXT:    xorl %ecx, %ecx
-; X86-NEXT:    addl %ebp, %ebp
+; X86-NEXT:    addl %ebx, %ebx
 ; X86-NEXT:    setb %cl
 ; X86-NEXT:    cmpl $1, %edx
-; X86-NEXT:    movl %edx, %eax
-; X86-NEXT:    movl $1, %edx
-; X86-NEXT:    cmovael %edx, %eax
-; X86-NEXT:    movl %eax, (%esp) # 4-byte Spill
-; X86-NEXT:    cmovel %ebx, %edi
-; X86-NEXT:    shldl $31, %ebp, %ecx
-; X86-NEXT:    shll $31, %ebp
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %esi
+; X86-NEXT:    cmovael %ebp, %edx
+; X86-NEXT:    movl %edx, (%esp) # 4-byte Spill
+; X86-NEXT:    cmovel %eax, %edi
+; X86-NEXT:    shldl $31, %ebx, %ecx
+; X86-NEXT:    shll $31, %ebx
 ; X86-NEXT:    pushl $0
 ; X86-NEXT:    pushl {{[0-9]+}}(%esp)
 ; X86-NEXT:    pushl %ecx
-; X86-NEXT:    pushl %ebp
+; X86-NEXT:    pushl %ebx
 ; X86-NEXT:    calll __udivdi3
 ; X86-NEXT:    addl $16, %esp
-; X86-NEXT:    cmpl $-1, %eax
-; X86-NEXT:    movl $-1, %ecx
-; X86-NEXT:    cmovbl %eax, %ecx
+; X86-NEXT:    movl %edx, %ebx
 ; X86-NEXT:    cmpl $1, %edx
-; X86-NEXT:    movl $1, %ebx
-; X86-NEXT:    cmovbl %edx, %ebx
 ; X86-NEXT:    movl $0, %ebp
 ; X86-NEXT:    sbbl %ebp, %ebp
 ; X86-NEXT:    notl %ebp
 ; X86-NEXT:    orl %eax, %ebp
+; X86-NEXT:    xorl %ecx, %ecx
+; X86-NEXT:    addl %esi, %esi
+; X86-NEXT:    setb %cl
 ; X86-NEXT:    cmpl $1, %edx
-; X86-NEXT:    cmovel %ecx, %ebp
-; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload
-; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload
-; X86-NEXT:    shrdl $1, %eax, %ecx
+; X86-NEXT:    movl $1, %edx
+; X86-NEXT:    cmovael %edx, %ebx
+; X86-NEXT:    cmovel %eax, %ebp
+; X86-NEXT:    shldl $31, %esi, %ecx
+; X86-NEXT:    shll $31, %esi
+; X86-NEXT:    pushl $0
+; X86-NEXT:    pushl {{[0-9]+}}(%esp)
+; X86-NEXT:    pushl %ecx
+; X86-NEXT:    pushl %esi
+; X86-NEXT:    calll __udivdi3
+; X86-NEXT:    addl $16, %esp
+; X86-NEXT:    cmpl $1, %edx
+; X86-NEXT:    movl $0, %ecx
+; X86-NEXT:    sbbl %ecx, %ecx
+; X86-NEXT:    notl %ecx
+; X86-NEXT:    orl %eax, %ecx
+; X86-NEXT:    cmpl $1, %edx
+; X86-NEXT:    movl $1, %esi
+; X86-NEXT:    cmovbl %edx, %esi
+; X86-NEXT:    cmovel %eax, %ecx
+; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload
 ; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload
-; X86-NEXT:    shrdl $1, %eax, %esi
+; X86-NEXT:    shrdl $1, %eax, %edx
 ; X86-NEXT:    movl (%esp), %eax # 4-byte Reload
 ; X86-NEXT:    shrdl $1, %eax, %edi
 ; X86-NEXT:    shrdl $1, %ebx, %ebp
+; X86-NEXT:    shrdl $1, %esi, %ecx
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
-; X86-NEXT:    movl %ebp, 12(%eax)
-; X86-NEXT:    movl %edi, 8(%eax)
-; X86-NEXT:    movl %esi, 4(%eax)
-; X86-NEXT:    movl %ecx, (%eax)
-; X86-NEXT:    addl $16, %esp
+; X86-NEXT:    movl %ecx, 12(%eax)
+; X86-NEXT:    movl %ebp, 8(%eax)
+; X86-NEXT:    movl %edi, 4(%eax)
+; X86-NEXT:    movl %edx, (%eax)
+; X86-NEXT:    addl $12, %esp
 ; X86-NEXT:    popl %esi
 ; X86-NEXT:    popl %edi
 ; X86-NEXT:    popl %ebx
 ; X86-NEXT:    popl %ebp
 ; X86-NEXT:    retl $4
   %tmp = call <4 x i32> @llvm.udiv.fix.sat.v4i32(<4 x i32> %x, <4 x i32> %y, i32 31)
   ret <4 x i32> %tmp
 }