[InstCombine] Remove some uses with replaceUndefsWith() #89190
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Now that we don't accept undef splat in PatternMatch, we can remove some uses of replaceUndefsWith(). I believe in all these cases only poison splats are possible now, in which case no replacement is necessary.
|
@llvm/pr-subscribers-llvm-transforms Author: Nikita Popov (nikic) ChangesNow that we don't accept undef splat in PatternMatch, we can remove some uses of replaceUndefsWith(). I believe in all these cases only poison splats are possible now, in which case no replacement is necessary. Full diff: https://github.com/llvm/llvm-project/pull/89190.diff 11 Files Affected:
diff --git a/llvm/lib/Transforms/InstCombine/InstCombineAndOrXor.cpp b/llvm/lib/Transforms/InstCombine/InstCombineAndOrXor.cpp
index bf7c0074a38f05..e1923a3441790a 100644
--- a/llvm/lib/Transforms/InstCombine/InstCombineAndOrXor.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstCombineAndOrXor.cpp
@@ -4433,23 +4433,13 @@ Instruction *InstCombinerImpl::foldNot(BinaryOperator &I) {
// ~(C >>s Y) --> ~C >>u Y (when inverting the replicated sign bits)
Constant *C;
if (match(NotVal, m_AShr(m_Constant(C), m_Value(Y))) &&
- match(C, m_Negative())) {
- // We matched a negative constant, so propagating undef is unsafe.
- // Clamp undef elements to -1.
- Type *EltTy = Ty->getScalarType();
- C = Constant::replaceUndefsWith(C, ConstantInt::getAllOnesValue(EltTy));
+ match(C, m_Negative()))
return BinaryOperator::CreateLShr(ConstantExpr::getNot(C), Y);
- }
// ~(C >>u Y) --> ~C >>s Y (when inverting the replicated sign bits)
if (match(NotVal, m_LShr(m_Constant(C), m_Value(Y))) &&
- match(C, m_NonNegative())) {
- // We matched a non-negative constant, so propagating undef is unsafe.
- // Clamp undef elements to 0.
- Type *EltTy = Ty->getScalarType();
- C = Constant::replaceUndefsWith(C, ConstantInt::getNullValue(EltTy));
+ match(C, m_NonNegative()))
return BinaryOperator::CreateAShr(ConstantExpr::getNot(C), Y);
- }
// ~(X + C) --> ~C - X
if (match(NotVal, m_Add(m_Value(X), m_ImmConstant(C))))
diff --git a/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp b/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
index 00fb1f7dd75b78..fdadf729fbd541 100644
--- a/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
@@ -4322,25 +4322,6 @@ static Value *foldICmpWithLowBitMaskedVal(ICmpInst::Predicate Pred, Value *Op0,
if (!IsLowBitMask())
return nullptr;
- // The mask value may be a vector constant that has undefined elements. But
- // it may not be safe to propagate those undefs into the new compare, so
- // replace those elements by copying an existing, defined, and safe scalar
- // constant.
- Type *OpTy = M->getType();
- auto *VecC = dyn_cast<Constant>(M);
- auto *OpVTy = dyn_cast<FixedVectorType>(OpTy);
- if (OpVTy && VecC && VecC->containsUndefOrPoisonElement()) {
- Constant *SafeReplacementConstant = nullptr;
- for (unsigned i = 0, e = OpVTy->getNumElements(); i != e; ++i) {
- if (!isa<UndefValue>(VecC->getAggregateElement(i))) {
- SafeReplacementConstant = VecC->getAggregateElement(i);
- break;
- }
- }
- assert(SafeReplacementConstant && "Failed to find undef replacement");
- M = Constant::replaceUndefsWith(VecC, SafeReplacementConstant);
- }
-
return IC.Builder.CreateICmp(DstPred, X, M);
}
diff --git a/llvm/lib/Transforms/InstCombine/InstCombineMulDivRem.cpp b/llvm/lib/Transforms/InstCombine/InstCombineMulDivRem.cpp
index 53478e07394b2c..a0333b7db8f7a9 100644
--- a/llvm/lib/Transforms/InstCombine/InstCombineMulDivRem.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstCombineMulDivRem.cpp
@@ -842,9 +842,6 @@ Instruction *InstCombinerImpl::visitFMul(BinaryOperator &I) {
isKnownNeverNaN(&I, /*Depth=*/0, SQ.getWithInstruction(&I)))) {
if (FPC->isNegative())
Op0 = Builder.CreateFNegFMF(Op0, &I);
- Op1 = Constant::replaceUndefsWith(
- cast<Constant>(Op1),
- ConstantFP::get(Op1->getType()->getScalarType(), *FPC));
CallInst *CopySign = Builder.CreateIntrinsic(Intrinsic::copysign,
{I.getType()}, {Op1, Op0}, &I);
return replaceInstUsesWith(I, CopySign);
diff --git a/llvm/test/Transforms/InstCombine/binop-itofp.ll b/llvm/test/Transforms/InstCombine/binop-itofp.ll
index 19f7ca9282f291..702bbbbf7d1760 100644
--- a/llvm/test/Transforms/InstCombine/binop-itofp.ll
+++ b/llvm/test/Transforms/InstCombine/binop-itofp.ll
@@ -1070,7 +1070,7 @@ define <2 x float> @nonzero_check_on_constant_for_si_fmul_vec_w_poison(i1 %c, i1
; CHECK-NEXT: [[CONV_I_V:%.*]] = insertelement <2 x i16> poison, i16 [[CONV_I_S]], i64 0
; CHECK-NEXT: [[CONV_I:%.*]] = shufflevector <2 x i16> [[CONV_I_V]], <2 x i16> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[CONV1_I:%.*]] = sitofp <2 x i16> [[CONV_I]] to <2 x float>
-; CHECK-NEXT: [[MUL3_I_I:%.*]] = call <2 x float> @llvm.copysign.v2f32(<2 x float> zeroinitializer, <2 x float> [[CONV1_I]])
+; CHECK-NEXT: [[MUL3_I_I:%.*]] = call <2 x float> @llvm.copysign.v2f32(<2 x float> <float poison, float 0.000000e+00>, <2 x float> [[CONV1_I]])
; CHECK-NEXT: store i32 [[SEL]], ptr [[G_2345:%.*]], align 4
; CHECK-NEXT: ret <2 x float> [[MUL3_I_I]]
;
@@ -1090,9 +1090,9 @@ define <2 x float> @nonzero_check_on_constant_for_si_fmul_nz_vec_w_poison(i1 %c,
; CHECK-NEXT: [[CONV_I_S:%.*]] = trunc nuw i32 [[SEL]] to i16
; CHECK-NEXT: [[CONV_I_V:%.*]] = insertelement <2 x i16> poison, i16 [[CONV_I_S]], i64 0
; CHECK-NEXT: [[CONV_I:%.*]] = shufflevector <2 x i16> [[CONV_I_V]], <2 x i16> poison, <2 x i32> zeroinitializer
-; CHECK-NEXT: [[CONV1_I:%.*]] = sitofp <2 x i16> [[CONV_I]] to <2 x float>
+; CHECK-NEXT: [[MUL3_I_I:%.*]] = sitofp <2 x i16> [[CONV_I]] to <2 x float>
; CHECK-NEXT: store i32 [[SEL]], ptr [[G_2345:%.*]], align 4
-; CHECK-NEXT: ret <2 x float> [[CONV1_I]]
+; CHECK-NEXT: ret <2 x float> [[MUL3_I_I]]
;
%sel = select i1 %c, i32 65529, i32 53264
%conv.i.s = trunc i32 %sel to i16
@@ -1112,7 +1112,7 @@ define <2 x float> @nonzero_check_on_constant_for_si_fmul_negz_vec_w_poison(i1 %
; CHECK-NEXT: [[CONV_I:%.*]] = shufflevector <2 x i16> [[CONV_I_V]], <2 x i16> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[CONV1_I:%.*]] = sitofp <2 x i16> [[CONV_I]] to <2 x float>
; CHECK-NEXT: [[TMP1:%.*]] = fneg <2 x float> [[CONV1_I]]
-; CHECK-NEXT: [[MUL3_I_I:%.*]] = call <2 x float> @llvm.copysign.v2f32(<2 x float> zeroinitializer, <2 x float> [[TMP1]])
+; CHECK-NEXT: [[MUL3_I_I:%.*]] = call <2 x float> @llvm.copysign.v2f32(<2 x float> <float poison, float -0.000000e+00>, <2 x float> [[TMP1]])
; CHECK-NEXT: store i32 [[SEL]], ptr [[G_2345:%.*]], align 4
; CHECK-NEXT: ret <2 x float> [[MUL3_I_I]]
;
diff --git a/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-ne-to-icmp-ugt.ll b/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-ne-to-icmp-ugt.ll
index 95e6d5ac6a5f81..9b28129dd9e172 100644
--- a/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-ne-to-icmp-ugt.ll
+++ b/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-ne-to-icmp-ugt.ll
@@ -81,7 +81,7 @@ define <2 x i1> @p2_vec_nonsplat_edgecase1(<2 x i8> %x) {
define <3 x i1> @p3_vec_splat_poison(<3 x i8> %x) {
; CHECK-LABEL: @p3_vec_splat_poison(
-; CHECK-NEXT: [[RET:%.*]] = icmp ugt <3 x i8> [[X:%.*]], <i8 3, i8 3, i8 3>
+; CHECK-NEXT: [[RET:%.*]] = icmp ugt <3 x i8> [[X:%.*]], <i8 3, i8 poison, i8 3>
; CHECK-NEXT: ret <3 x i1> [[RET]]
;
%tmp0 = and <3 x i8> %x, <i8 3, i8 poison, i8 3>
@@ -91,7 +91,7 @@ define <3 x i1> @p3_vec_splat_poison(<3 x i8> %x) {
define <3 x i1> @p3_vec_nonsplat_poison(<3 x i8> %x) {
; CHECK-LABEL: @p3_vec_nonsplat_poison(
-; CHECK-NEXT: [[RET:%.*]] = icmp ugt <3 x i8> [[X:%.*]], <i8 -1, i8 -1, i8 3>
+; CHECK-NEXT: [[RET:%.*]] = icmp ugt <3 x i8> [[X:%.*]], <i8 -1, i8 poison, i8 3>
; CHECK-NEXT: ret <3 x i1> [[RET]]
;
%tmp0 = and <3 x i8> %x, <i8 -1, i8 poison, i8 3>
diff --git a/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-sgt-to-icmp-sgt.ll b/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-sgt-to-icmp-sgt.ll
index f1333fed2c5179..d1dd411ee86b31 100644
--- a/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-sgt-to-icmp-sgt.ll
+++ b/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-sgt-to-icmp-sgt.ll
@@ -75,7 +75,7 @@ define <2 x i1> @p2_vec_nonsplat_edgecase() {
define <3 x i1> @p3_vec_splat_poison() {
; CHECK-LABEL: @p3_vec_splat_poison(
; CHECK-NEXT: [[X:%.*]] = call <3 x i8> @gen3x8()
-; CHECK-NEXT: [[RET:%.*]] = icmp sgt <3 x i8> [[X]], <i8 3, i8 3, i8 3>
+; CHECK-NEXT: [[RET:%.*]] = icmp sgt <3 x i8> [[X]], <i8 3, i8 poison, i8 3>
; CHECK-NEXT: ret <3 x i1> [[RET]]
;
%x = call <3 x i8> @gen3x8()
@@ -87,7 +87,7 @@ define <3 x i1> @p3_vec_splat_poison() {
define <3 x i1> @p3_vec_nonsplat_poison() {
; CHECK-LABEL: @p3_vec_nonsplat_poison(
; CHECK-NEXT: [[X:%.*]] = call <3 x i8> @gen3x8()
-; CHECK-NEXT: [[RET:%.*]] = icmp sgt <3 x i8> [[X]], <i8 15, i8 3, i8 15>
+; CHECK-NEXT: [[RET:%.*]] = icmp sgt <3 x i8> [[X]], <i8 15, i8 3, i8 poison>
; CHECK-NEXT: ret <3 x i1> [[RET]]
;
%x = call <3 x i8> @gen3x8()
diff --git a/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-slt-to-icmp-sgt.ll b/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-slt-to-icmp-sgt.ll
index be6e3d0306bcdf..8415204a4915a2 100644
--- a/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-slt-to-icmp-sgt.ll
+++ b/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-slt-to-icmp-sgt.ll
@@ -60,7 +60,7 @@ define <2 x i1> @p2_vec_nonsplat_edgecase(<2 x i8> %x) {
define <3 x i1> @p3_vec_splat_poison(<3 x i8> %x) {
; CHECK-LABEL: @p3_vec_splat_poison(
-; CHECK-NEXT: [[RET:%.*]] = icmp sgt <3 x i8> [[X:%.*]], <i8 3, i8 3, i8 3>
+; CHECK-NEXT: [[RET:%.*]] = icmp sgt <3 x i8> [[X:%.*]], <i8 3, i8 poison, i8 3>
; CHECK-NEXT: ret <3 x i1> [[RET]]
;
%tmp0 = and <3 x i8> %x, <i8 3, i8 poison, i8 3>
@@ -70,7 +70,7 @@ define <3 x i1> @p3_vec_splat_poison(<3 x i8> %x) {
define <3 x i1> @p3_vec_nonsplat_poison(<3 x i8> %x) {
; CHECK-LABEL: @p3_vec_nonsplat_poison(
-; CHECK-NEXT: [[RET:%.*]] = icmp sgt <3 x i8> [[X:%.*]], <i8 15, i8 15, i8 3>
+; CHECK-NEXT: [[RET:%.*]] = icmp sgt <3 x i8> [[X:%.*]], <i8 poison, i8 15, i8 3>
; CHECK-NEXT: ret <3 x i1> [[RET]]
;
%tmp0 = and <3 x i8> %x, <i8 poison, i8 15, i8 3>
diff --git a/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-ugt-to-icmp-ugt.ll b/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-ugt-to-icmp-ugt.ll
index 6f6ba95a81c767..e8781407cb647f 100644
--- a/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-ugt-to-icmp-ugt.ll
+++ b/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-ugt-to-icmp-ugt.ll
@@ -98,7 +98,7 @@ define <2 x i1> @p2_vec_nonsplat_edgecase1() {
define <3 x i1> @p3_vec_splat_poison() {
; CHECK-LABEL: @p3_vec_splat_poison(
; CHECK-NEXT: [[X:%.*]] = call <3 x i8> @gen3x8()
-; CHECK-NEXT: [[RET:%.*]] = icmp ugt <3 x i8> [[X]], <i8 3, i8 3, i8 3>
+; CHECK-NEXT: [[RET:%.*]] = icmp ugt <3 x i8> [[X]], <i8 3, i8 poison, i8 3>
; CHECK-NEXT: ret <3 x i1> [[RET]]
;
%x = call <3 x i8> @gen3x8()
@@ -110,7 +110,7 @@ define <3 x i1> @p3_vec_splat_poison() {
define <3 x i1> @p3_vec_nonsplat_poison() {
; CHECK-LABEL: @p3_vec_nonsplat_poison(
; CHECK-NEXT: [[X:%.*]] = call <3 x i8> @gen3x8()
-; CHECK-NEXT: [[RET:%.*]] = icmp ugt <3 x i8> [[X]], <i8 3, i8 3, i8 15>
+; CHECK-NEXT: [[RET:%.*]] = icmp ugt <3 x i8> [[X]], <i8 3, i8 poison, i8 15>
; CHECK-NEXT: ret <3 x i1> [[RET]]
;
%x = call <3 x i8> @gen3x8()
diff --git a/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-ult-to-icmp-ugt.ll b/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-ult-to-icmp-ugt.ll
index 008fc6d2d6eda9..70fb34f4992899 100644
--- a/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-ult-to-icmp-ugt.ll
+++ b/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-ult-to-icmp-ugt.ll
@@ -82,7 +82,7 @@ define <2 x i1> @p2_vec_nonsplat_edgecase1(<2 x i8> %x) {
define <3 x i1> @p3_vec_splat_poison(<3 x i8> %x) {
; CHECK-LABEL: @p3_vec_splat_poison(
-; CHECK-NEXT: [[RET:%.*]] = icmp ugt <3 x i8> [[X:%.*]], <i8 3, i8 3, i8 3>
+; CHECK-NEXT: [[RET:%.*]] = icmp ugt <3 x i8> [[X:%.*]], <i8 3, i8 poison, i8 3>
; CHECK-NEXT: ret <3 x i1> [[RET]]
;
%tmp0 = and <3 x i8> %x, <i8 3, i8 poison, i8 3>
@@ -92,7 +92,7 @@ define <3 x i1> @p3_vec_splat_poison(<3 x i8> %x) {
define <3 x i1> @p3_vec_nonsplat_poison(<3 x i8> %x) {
; CHECK-LABEL: @p3_vec_nonsplat_poison(
-; CHECK-NEXT: [[RET:%.*]] = icmp ugt <3 x i8> [[X:%.*]], <i8 7, i8 31, i8 7>
+; CHECK-NEXT: [[RET:%.*]] = icmp ugt <3 x i8> [[X:%.*]], <i8 7, i8 31, i8 poison>
; CHECK-NEXT: ret <3 x i1> [[RET]]
;
%tmp0 = and <3 x i8> %x, <i8 7, i8 31, i8 poison>
diff --git a/llvm/test/Transforms/InstCombine/fmul.ll b/llvm/test/Transforms/InstCombine/fmul.ll
index e9c86a1270493e..1526956c5b241e 100644
--- a/llvm/test/Transforms/InstCombine/fmul.ll
+++ b/llvm/test/Transforms/InstCombine/fmul.ll
@@ -633,11 +633,11 @@ define float @log2half(float %x, float %y) {
define float @log2half_commute(float %x1, float %y) {
; CHECK-LABEL: @log2half_commute(
-; CHECK-NEXT: [[X1:%.*]] = fmul fast float [[X2:%.*]], 0x3FC24924A0000000
+; CHECK-NEXT: [[X:%.*]] = fmul fast float [[X1:%.*]], 0x3FC24924A0000000
; CHECK-NEXT: [[TMP1:%.*]] = call fast float @llvm.log2.f32(float [[Y:%.*]])
-; CHECK-NEXT: [[TMP2:%.*]] = fmul fast float [[TMP1]], [[X1]]
-; CHECK-NEXT: [[TMP3:%.*]] = fsub fast float [[TMP2]], [[X1]]
-; CHECK-NEXT: ret float [[TMP3]]
+; CHECK-NEXT: [[TMP2:%.*]] = fmul fast float [[TMP1]], [[X]]
+; CHECK-NEXT: [[MUL:%.*]] = fsub fast float [[TMP2]], [[X]]
+; CHECK-NEXT: ret float [[MUL]]
;
%x = fdiv fast float %x1, 7.0 ; thwart complexity-based canonicalization
%halfy = fmul fast float %y, 0.5
@@ -687,8 +687,8 @@ define float @fdiv_constant_numerator_fmul_fast(float %x) {
define float @fdiv_constant_numerator_fmul_precdiv(float %x) {
; CHECK-LABEL: @fdiv_constant_numerator_fmul_precdiv(
; CHECK-NEXT: [[T1:%.*]] = fdiv float 2.000000e+03, [[X:%.*]]
-; CHECK-NEXT: [[T4:%.*]] = fmul reassoc float [[T1]], 6.000000e+03
-; CHECK-NEXT: ret float [[T4]]
+; CHECK-NEXT: [[T3:%.*]] = fmul reassoc float [[T1]], 6.000000e+03
+; CHECK-NEXT: ret float [[T3]]
;
%t1 = fdiv float 2.0e+3, %x
%t3 = fmul reassoc float %t1, 6.0e+3
@@ -1288,7 +1288,7 @@ define half @mul_zero_nnan(half %x) {
define <2 x float> @mul_zero_nnan_vec_poison(<2 x float> %x) {
; CHECK-LABEL: @mul_zero_nnan_vec_poison(
-; CHECK-NEXT: [[R:%.*]] = call nnan <2 x float> @llvm.copysign.v2f32(<2 x float> zeroinitializer, <2 x float> [[X:%.*]])
+; CHECK-NEXT: [[R:%.*]] = call nnan <2 x float> @llvm.copysign.v2f32(<2 x float> <float 0.000000e+00, float poison>, <2 x float> [[X:%.*]])
; CHECK-NEXT: ret <2 x float> [[R]]
;
%r = fmul nnan <2 x float> %x, <float 0.0, float poison>
@@ -1389,7 +1389,7 @@ define <3 x float> @mul_neg_zero_nnan_ninf_vec(<3 x float> nofpclass(inf nan) %a
; CHECK-LABEL: @mul_neg_zero_nnan_ninf_vec(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[TMP0:%.*]] = fneg <3 x float> [[A:%.*]]
-; CHECK-NEXT: [[RET:%.*]] = call <3 x float> @llvm.copysign.v3f32(<3 x float> zeroinitializer, <3 x float> [[TMP0]])
+; CHECK-NEXT: [[RET:%.*]] = call <3 x float> @llvm.copysign.v3f32(<3 x float> <float -0.000000e+00, float poison, float poison>, <3 x float> [[TMP0]])
; CHECK-NEXT: ret <3 x float> [[RET]]
;
entry:
diff --git a/llvm/test/Transforms/InstCombine/vector-xor.ll b/llvm/test/Transforms/InstCombine/vector-xor.ll
index ee593b5d15e8e3..5c96f1a691ed03 100644
--- a/llvm/test/Transforms/InstCombine/vector-xor.ll
+++ b/llvm/test/Transforms/InstCombine/vector-xor.ll
@@ -140,7 +140,7 @@ define <4 x i32> @test_v4i32_not_ashr_negative_const(<4 x i32> %a0) {
define <4 x i32> @test_v4i32_not_ashr_negative_const_poison(<4 x i32> %a0) {
; CHECK-LABEL: @test_v4i32_not_ashr_negative_const_poison(
-; CHECK-NEXT: [[TMP1:%.*]] = lshr <4 x i32> <i32 2, i32 4, i32 0, i32 8>, [[A0:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = lshr <4 x i32> <i32 2, i32 4, i32 poison, i32 8>, [[A0:%.*]]
; CHECK-NEXT: ret <4 x i32> [[TMP1]]
;
%1 = ashr <4 x i32> <i32 -3, i32 -5, i32 poison, i32 -9>, %a0
@@ -172,7 +172,7 @@ define <4 x i32> @test_v4i32_not_lshr_nonnegative_const(<4 x i32> %a0) {
define <4 x i32> @test_v4i32_not_lshr_nonnegative_const_poison(<4 x i32> %a0) {
; CHECK-LABEL: @test_v4i32_not_lshr_nonnegative_const_poison(
-; CHECK-NEXT: [[TMP1:%.*]] = ashr <4 x i32> <i32 -4, i32 -6, i32 -1, i32 -10>, [[A0:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = ashr <4 x i32> <i32 -4, i32 -6, i32 poison, i32 -10>, [[A0:%.*]]
; CHECK-NEXT: ret <4 x i32> [[TMP1]]
;
%1 = lshr <4 x i32> <i32 3, i32 5, i32 poison, i32 9>, %a0
|
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Now that we don't accept undef splat in PatternMatch, we can remove some uses of replaceUndefsWith(). I believe in all these cases only poison splats are possible now, in which case no replacement is necessary.