[SelectionDAG] Fix shift libcall ABI mismatch in shift-amount argument

[nextsilicon-itay-bookstein]

Summary

The shift libcalls have a shift amount parameter of MVT::i32, but
sometimes ExpandIntRes_Shift may be called with a node whose
second operand is a type that is larger than that. This leads to
an ABI mismatch, and for example causes a spurious zeroing of
a register in RV32 for 64-bit shifts. Note that at present regular
shift intstructions already have their shift amount operand adapted
at SelectionDAGBuilder::visitShift time, and funnelled shifts bypass that.

[nextsilicon-itay-bookstein]

Fixed clang-format.
Updated X86 and AArch64 shift_minsize test output, and refreshed labels for RISCV shifts test output.

[nextsilicon-itay-bookstein]

Rebased on top of main.

[craig.topper]

llvm/test/CodeGen/RISCV/shifts.ll

613

Can you test fshr.i128 as well?

[nextsilicon-itay-bookstein]

Added fshr.i128 test, refreshed another small diff in AArch64

[nextsilicon-itay-bookstein]

Modified to use DAG.getLibInfo().getIntSize() as @bjope suggested in llvm-dev.

[bjope]

llvm/lib/CodeGen/SelectionDAG/LegalizeIntegerTypes.cpp

4001

AnyExt seem a bit weird as I assume the lib function will use all bits. I think getZExtOrTrunc would be a more correct choice here.

[nextsilicon-itay-bookstein]

llvm/lib/CodeGen/SelectionDAG/LegalizeIntegerTypes.cpp

4001

I agree that AnyExt is incorrect, however I'm wondering whether it should be ZExt or SExt. The source-level semantics seem to point towards SExt, because the libcalls are defined with a signed shift amount (si_int as opposed to su_int). Earlier code in this function uses ZExt (under the LegalOrCustom && TLI.shouldExpandShift(DAG, N) case), but it doesn't need to concern itself with the signature for __ashlti3 and friends. Note that this actually affects the output for llvm/test/CodeGen/X86/shift_minsize.ll functions shl128, ashr128, lshr128, in whether the movzbl is emitted or not.

[nextsilicon-itay-bookstein]

Changed to use ZExtOrTrunc. Note change in X86/shift_minsize.ll

[bjope]

llvm/lib/CodeGen/SelectionDAG/LegalizeIntegerTypes.cpp

4001

My reasoning was that the shift amount can't be negative (that would be undefined both for the SelectionDAG nodes and afaik the lib calls (e.g. __ashlti3 expects the shift amount b to be in the range 0 <= b < "bits in a tword". And at least for rotates and funnel shifts the shift amount (in SelectionDAG nodes) is treated as an unsigned. So I kind of assumed that doing a zero extend would be safe, as it would avoid passing a negative shift amount to the lib function (e.g. if the shift amount is 15 and the type we should convert from is i4, then doing a zext to a 32 bit si_int (treated as signed) would still give the value 15, while doing a sext would result in passing -1 to the lib function).

[nextsilicon-itay-bookstein]

Alright, yes, that makes a lot of sense to me. The patch already uses ZExt, so I think it could be considered ready.

[nextsilicon-itay-bookstein]

ping :)

[craig.topper]

LGTM

[nextsilicon-itay-bookstein]

😄

I have no commit access, might I ask that you commit this on my behalf?

Author: Itay Bookstein <[e-mail]>

llvm/lib/CodeGen/SelectionDAG/LegalizeIntegerTypes.cpp

else if (VT == MVT::i32)
       LC = RTLIB::SRA_I32;
     else if (VT == MVT::i64)
       LC = RTLIB::SRA_I64;
     else if (VT == MVT::i128)
       LC = RTLIB::SRA_I128;
   }
 
   if (LC != RTLIB::UNKNOWN_LIBCALL && TLI.getLibcallName(LC)) {
-    SDValue Ops[2] = { N->getOperand(0), N->getOperand(1) };
+    SDValue ShAmt = DAG.getAnyExtOrTrunc(N->getOperand(1), dl, MVT::i32);
+    SDValue Ops[2] = {N->getOperand(0), ShAmt};
     TargetLowering::MakeLibCallOptions CallOptions;
     CallOptions.setSExt(isSigned);
     SplitInteger(TLI.makeLibCall(DAG, LC, VT, Ops, CallOptions, dl).first, Lo, Hi);
     return;
   }
 
   if (!ExpandShiftWithUnknownAmountBit(N, Lo, Hi))
     llvm_unreachable("Unsupported shift!");
 }

llvm/test/CodeGen/AArch64/shift_minsize.ll

 }
 
 define dso_local { i64, i64 } @shl128(i64 %x.coerce0, i64 %x.coerce1, i8 signext %y) minsize optsize {
 ; CHECK-LABEL: shl128:
 ; CHECK:       // %bb.0: // %entry
 ; CHECK-NEXT:    str x30, [sp, #-16]! // 8-byte Folded Spill
 ; CHECK-NEXT:    .cfi_def_cfa_offset 16
 ; CHECK-NEXT:    .cfi_offset w30, -16
-; CHECK-NEXT:    mov w2, w2
 ; CHECK-NEXT:    bl __ashlti3
 ; CHECK-NEXT:    ldr x30, [sp], #16 // 8-byte Folded Reload
 ; CHECK-NEXT:    ret
 
 entry:
   %x.sroa.2.0.insert.ext = zext i64 %x.coerce1 to i128
   %x.sroa.2.0.insert.shift = shl nuw i128 %x.sroa.2.0.insert.ext, 64
   %x.sroa.0.0.insert.ext = zext i64 %x.coerce0 to i128

 }
 
 define dso_local { i64, i64 } @ashr128(i64 %x.coerce0, i64 %x.coerce1, i8 signext %y) minsize optsize {
 ; CHECK-LABEL: ashr128:
 ; CHECK:       // %bb.0: // %entry
 ; CHECK-NEXT:    str x30, [sp, #-16]! // 8-byte Folded Spill
 ; CHECK-NEXT:    .cfi_def_cfa_offset 16
 ; CHECK-NEXT:    .cfi_offset w30, -16
-; CHECK-NEXT:    mov w2, w2
 ; CHECK-NEXT:    bl __ashrti3
 ; CHECK-NEXT:    ldr x30, [sp], #16 // 8-byte Folded Reload
 ; CHECK-NEXT:    ret
 entry:
   %x.sroa.2.0.insert.ext = zext i64 %x.coerce1 to i128
   %x.sroa.2.0.insert.shift = shl nuw i128 %x.sroa.2.0.insert.ext, 64
   %x.sroa.0.0.insert.ext = zext i64 %x.coerce0 to i128
   %x.sroa.0.0.insert.insert = or i128 %x.sroa.2.0.insert.shift, %x.sroa.0.0.insert.ext

 }
 
 define dso_local { i64, i64 } @lshr128(i64 %x.coerce0, i64 %x.coerce1, i8 signext %y) minsize optsize {
 ; CHECK-LABEL: lshr128:
 ; CHECK:       // %bb.0: // %entry
 ; CHECK-NEXT:    str x30, [sp, #-16]! // 8-byte Folded Spill
 ; CHECK-NEXT:    .cfi_def_cfa_offset 16
 ; CHECK-NEXT:    .cfi_offset w30, -16
-; CHECK-NEXT:    mov w2, w2
 ; CHECK-NEXT:    bl __lshrti3
 ; CHECK-NEXT:    ldr x30, [sp], #16 // 8-byte Folded Reload
 ; CHECK-NEXT:    ret
 entry:
   %x.sroa.2.0.insert.ext = zext i64 %x.coerce1 to i128
   %x.sroa.2.0.insert.shift = shl nuw i128 %x.sroa.2.0.insert.ext, 64
   %x.sroa.0.0.insert.ext = zext i64 %x.coerce0 to i128
   %x.sroa.0.0.insert.insert = or i128 %x.sroa.2.0.insert.shift, %x.sroa.0.0.insert.ext

llvm/test/CodeGen/RISCV/shifts.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc -mtriple=riscv32 -verify-machineinstrs < %s \
 ; RUN:   | FileCheck %s -check-prefix=RV32I
 ; RUN: llc -mtriple=riscv64 -verify-machineinstrs < %s \
 ; RUN:   | FileCheck %s -check-prefix=RV64I
 
 ; Basic shift support is tested as part of ALU.ll. This file ensures that
 ; shifts which may not be supported natively are lowered properly.
 
+declare i64 @llvm.fshr.i64(i64, i64, i64)
+declare i128 @llvm.fshr.i128(i128, i128, i128)
+
 define i64 @lshr64(i64 %a, i64 %b) nounwind {
 ; RV32I-LABEL: lshr64:
 ; RV32I:       # %bb.0:
 ; RV32I-NEXT:    addi a3, a2, -32
 ; RV32I-NEXT:    bltz a3, .LBB0_2
 ; RV32I-NEXT:  # %bb.1:
 ; RV32I-NEXT:    srl a0, a1, a3
 ; RV32I-NEXT:    mv a1, zero

 ; RV64I-NEXT:    srli a4, a0, 1
 ; RV64I-NEXT:    srl a3, a4, a3
 ; RV64I-NEXT:    or a1, a1, a3
 ; RV64I-NEXT:    sll a0, a0, a2
 ; RV64I-NEXT:    ret
   %1 = shl i128 %a, %b
   ret i128 %1
 }
+
+define i64 @fshr64_minsize(i64 %a, i64 %b) minsize nounwind {
+; RV32I-LABEL: fshr64_minsize:
+; RV32I:       # %bb.0:
+; RV32I-NEXT:    addi sp, sp, -32
+; RV32I-NEXT:    sw ra, 28(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw s0, 24(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw s4, 8(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    mv s0, a2
+; RV32I-NEXT:    mv s2, a1
+; RV32I-NEXT:    mv s1, a0
+; RV32I-NEXT:    andi a2, a2, 63
+; RV32I-NEXT:    call __lshrdi3@plt
+; RV32I-NEXT:    mv s3, a0
+; RV32I-NEXT:    mv s4, a1
+; RV32I-NEXT:    neg a0, s0
+; RV32I-NEXT:    andi a2, a0, 63
+; RV32I-NEXT:    mv a0, s1
+; RV32I-NEXT:    mv a1, s2
+; RV32I-NEXT:    call __ashldi3@plt
+; RV32I-NEXT:    or a0, s3, a0
+; RV32I-NEXT:    or a1, s4, a1
+; RV32I-NEXT:    lw s4, 8(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    addi sp, sp, 32
+; RV32I-NEXT:    ret
+;
+; RV64I-LABEL: fshr64_minsize:
+; RV64I:       # %bb.0:
+; RV64I-NEXT:    srl a2, a0, a1
+; RV64I-NEXT:    neg a1, a1
+; RV64I-NEXT:    sll a0, a0, a1
+; RV64I-NEXT:    or a0, a2, a0
+; RV64I-NEXT:    ret
+  %res = tail call i64 @llvm.fshr.i64(i64 %a, i64 %a, i64 %b)
+  ret i64 %res
+}
+
+define i128 @fshr128_minsize(i128 %a, i128 %b) minsize nounwind {
+; RV32I-LABEL: fshr128_minsize:
+; RV32I:       # %bb.0:
+; RV32I-NEXT:    addi sp, sp, -64
+; RV32I-NEXT:    sw ra, 60(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw s0, 56(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw s1, 52(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw s2, 48(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw s3, 44(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw s4, 40(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw s5, 36(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw s6, 32(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw s7, 28(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw s8, 24(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw s9, 20(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw s10, 16(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    sw s11, 12(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    lw s5, 0(a1)
+; RV32I-NEXT:    lw s6, 4(a1)
+; RV32I-NEXT:    lw s4, 8(a1)
+; RV32I-NEXT:    lw s3, 12(a1)
+; RV32I-NEXT:    lw s11, 0(a2)
+; RV32I-NEXT:    mv s2, a0
+; RV32I-NEXT:    andi s0, s11, 127
+; RV32I-NEXT:    addi a2, s0, -64
+; RV32I-NEXT:    mv a0, s4
+; RV32I-NEXT:    mv a1, s3
+; RV32I-NEXT:    call __lshrdi3@plt
+; RV32I-NEXT:    mv s8, a0
+; RV32I-NEXT:    sw a1, 8(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    mv a0, s5
+; RV32I-NEXT:    mv a1, s6
+; RV32I-NEXT:    mv a2, s0
+; RV32I-NEXT:    call __lshrdi3@plt
+; RV32I-NEXT:    mv s1, a0
+; RV32I-NEXT:    sw a1, 0(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    addi s9, zero, 64
+; RV32I-NEXT:    sub a2, s9, s0
+; RV32I-NEXT:    mv a0, s4
+; RV32I-NEXT:    mv a1, s3
+; RV32I-NEXT:    call __ashldi3@plt
+; RV32I-NEXT:    mv s10, a1
+; RV32I-NEXT:    bgeu s0, s9, .LBB10_2
+; RV32I-NEXT:  # %bb.1:
+; RV32I-NEXT:    or s8, s1, a0
+; RV32I-NEXT:  .LBB10_2:
+; RV32I-NEXT:    mv s7, s5
+; RV32I-NEXT:    beqz s0, .LBB10_4
+; RV32I-NEXT:  # %bb.3:
+; RV32I-NEXT:    mv s7, s8
+; RV32I-NEXT:  .LBB10_4:
+; RV32I-NEXT:    neg a0, s11
+; RV32I-NEXT:    andi s1, a0, 127
+; RV32I-NEXT:    mv a0, s5
+; RV32I-NEXT:    mv a1, s6
+; RV32I-NEXT:    mv a2, s1
+; RV32I-NEXT:    call __ashldi3@plt
+; RV32I-NEXT:    sw a1, 4(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    bgeu s1, s9, .LBB10_6
+; RV32I-NEXT:  # %bb.5:
+; RV32I-NEXT:    or s7, s7, a0
+; RV32I-NEXT:  .LBB10_6:
+; RV32I-NEXT:    bltu s0, s9, .LBB10_8
+; RV32I-NEXT:  # %bb.7:
+; RV32I-NEXT:    lw a0, 8(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    j .LBB10_9
+; RV32I-NEXT:  .LBB10_8:
+; RV32I-NEXT:    lw a0, 0(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    or a0, a0, s10
+; RV32I-NEXT:  .LBB10_9:
+; RV32I-NEXT:    mv s8, s6
+; RV32I-NEXT:    beqz s0, .LBB10_11
+; RV32I-NEXT:  # %bb.10:
+; RV32I-NEXT:    mv s8, a0
+; RV32I-NEXT:  .LBB10_11:
+; RV32I-NEXT:    sub a2, s9, s1
+; RV32I-NEXT:    mv a0, s5
+; RV32I-NEXT:    mv a1, s6
+; RV32I-NEXT:    call __lshrdi3@plt
+; RV32I-NEXT:    mv s10, a0
+; RV32I-NEXT:    sw a1, 8(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    mv a0, s4
+; RV32I-NEXT:    mv a1, s3
+; RV32I-NEXT:    mv a2, s1
+; RV32I-NEXT:    call __ashldi3@plt
+; RV32I-NEXT:    mv s11, a0
+; RV32I-NEXT:    sw a1, 0(sp) # 4-byte Folded Spill
+; RV32I-NEXT:    addi a2, s1, -64
+; RV32I-NEXT:    mv a0, s5
+; RV32I-NEXT:    mv a1, s6
+; RV32I-NEXT:    call __ashldi3@plt
+; RV32I-NEXT:    mv s5, a1
+; RV32I-NEXT:    bgeu s1, s9, .LBB10_13
+; RV32I-NEXT:  # %bb.12:
+; RV32I-NEXT:    lw a0, 4(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    or s8, s8, a0
+; RV32I-NEXT:    or a0, s11, s10
+; RV32I-NEXT:  .LBB10_13:
+; RV32I-NEXT:    mv s6, s4
+; RV32I-NEXT:    beqz s1, .LBB10_15
+; RV32I-NEXT:  # %bb.14:
+; RV32I-NEXT:    mv s6, a0
+; RV32I-NEXT:  .LBB10_15:
+; RV32I-NEXT:    mv a0, s4
+; RV32I-NEXT:    mv a1, s3
+; RV32I-NEXT:    mv a2, s0
+; RV32I-NEXT:    call __lshrdi3@plt
+; RV32I-NEXT:    bltu s0, s9, .LBB10_21
+; RV32I-NEXT:  # %bb.16:
+; RV32I-NEXT:    bltu s1, s9, .LBB10_22
+; RV32I-NEXT:  .LBB10_17:
+; RV32I-NEXT:    bnez s1, .LBB10_23
+; RV32I-NEXT:  .LBB10_18:
+; RV32I-NEXT:    bgeu s0, s9, .LBB10_20
+; RV32I-NEXT:  .LBB10_19:
+; RV32I-NEXT:    or s3, s3, a1
+; RV32I-NEXT:  .LBB10_20:
+; RV32I-NEXT:    sw s8, 4(s2)
+; RV32I-NEXT:    sw s7, 0(s2)
+; RV32I-NEXT:    sw s3, 12(s2)
+; RV32I-NEXT:    sw s6, 8(s2)
+; RV32I-NEXT:    lw s11, 12(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s10, 16(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s9, 20(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s8, 24(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s7, 28(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s6, 32(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s5, 36(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s4, 40(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s3, 44(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s2, 48(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s1, 52(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw s0, 56(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw ra, 60(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    addi sp, sp, 64
+; RV32I-NEXT:    ret
+; RV32I-NEXT:  .LBB10_21:
+; RV32I-NEXT:    or s6, s6, a0
+; RV32I-NEXT:    bgeu s1, s9, .LBB10_17
+; RV32I-NEXT:  .LBB10_22:
+; RV32I-NEXT:    lw a0, 8(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    lw a2, 0(sp) # 4-byte Folded Reload
+; RV32I-NEXT:    or s5, a2, a0
+; RV32I-NEXT:    beqz s1, .LBB10_18
+; RV32I-NEXT:  .LBB10_23:
+; RV32I-NEXT:    mv s3, s5
+; RV32I-NEXT:    bltu s0, s9, .LBB10_19
+; RV32I-NEXT:    j .LBB10_20
+;
+; RV64I-LABEL: fshr128_minsize:
+; RV64I:       # %bb.0:
+; RV64I-NEXT:    addi sp, sp, -48
+; RV64I-NEXT:    sd ra, 40(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s0, 32(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s1, 24(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s2, 16(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s3, 8(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    sd s4, 0(sp) # 8-byte Folded Spill
+; RV64I-NEXT:    mv s0, a2
+; RV64I-NEXT:    mv s2, a1
+; RV64I-NEXT:    mv s1, a0
+; RV64I-NEXT:    andi a2, a2, 127
+; RV64I-NEXT:    call __lshrti3@plt
+; RV64I-NEXT:    mv s3, a0
+; RV64I-NEXT:    mv s4, a1
+; RV64I-NEXT:    neg a0, s0
+; RV64I-NEXT:    andi a2, a0, 127
+; RV64I-NEXT:    mv a0, s1
+; RV64I-NEXT:    mv a1, s2
+; RV64I-NEXT:    call __ashlti3@plt
+; RV64I-NEXT:    or a0, s3, a0
+; RV64I-NEXT:    or a1, s4, a1
+; RV64I-NEXT:    ld s4, 0(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s3, 8(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s2, 16(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s1, 24(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld s0, 32(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    ld ra, 40(sp) # 8-byte Folded Reload
+; RV64I-NEXT:    addi sp, sp, 48
+; RV64I-NEXT:    ret
+  %res = tail call i128 @llvm.fshr.i128(i128 %a, i128 %a, i128 %b)
+  ret i128 %res
+}

llvm/test/CodeGen/X86/shift_minsize.ll

; CHECK-WIN-NEXT:    retq
   ret i32 %b
 }
 
 define dso_local { i64, i64 } @shl128(i64 %x.coerce0, i64 %x.coerce1, i8 signext %y) minsize optsize {
 ; CHECK-LABEL: shl128:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    pushq %rax
 ; CHECK-NEXT:    .cfi_def_cfa_offset 16
-; CHECK-NEXT:    movzbl %dl, %edx
 ; CHECK-NEXT:    callq __ashlti3@PLT
 ; CHECK-NEXT:    popq %rcx
 ; CHECK-NEXT:    .cfi_def_cfa_offset 8
 ; CHECK-NEXT:    retq
 ;
 ; CHECK-WIN-LABEL: shl128:
 ; CHECK-WIN:       # %bb.0: # %entry
 ; CHECK-WIN-NEXT:    movq %rcx, %r9
entry:
   ret { i64, i64 } %.fca.1.insert
 }
 
 define dso_local { i64, i64 } @lshr128(i64 %x.coerce0, i64 %x.coerce1, i8 signext %y) minsize optsize {
 ; CHECK-LABEL: lshr128:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    pushq %rax
 ; CHECK-NEXT:    .cfi_def_cfa_offset 16
-; CHECK-NEXT:    movzbl %dl, %edx
 ; CHECK-NEXT:    callq __lshrti3@PLT
 ; CHECK-NEXT:    popq %rcx
 ; CHECK-NEXT:    .cfi_def_cfa_offset 8
 ; CHECK-NEXT:    retq
 ;
 ; CHECK-WIN-LABEL: lshr128:
 ; CHECK-WIN:       # %bb.0: # %entry
 ; CHECK-WIN-NEXT:    movq %rcx, %rax