[InstCombine] Extend foldSelectInstWithICmpConst to handle minmax
#125346
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
|
@llvm/pr-subscribers-llvm-transforms Author: Yingwei Zheng (dtcxzyw) ChangesThis patch extends f6bb156 to handle minmax intrinsics. Full diff: https://github.com/llvm/llvm-project/pull/125346.diff 2 Files Affected:
diff --git a/llvm/lib/Transforms/InstCombine/InstCombineSelect.cpp b/llvm/lib/Transforms/InstCombine/InstCombineSelect.cpp
index 29c5cef84ccdb7..382078e85a17b6 100644
--- a/llvm/lib/Transforms/InstCombine/InstCombineSelect.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstCombineSelect.cpp
@@ -1771,26 +1771,38 @@ static Value *foldSelectInstWithICmpConst(SelectInst &SI, ICmpInst *ICI,
return Builder.CreateBinaryIntrinsic(Intrinsic::smin, V, TVal);
}
- BinaryOperator *BO;
+ // Fold icmp(X) ? f(X) : C to f(X) when f(X) is guaranteed to be equal to C
+ // for all X in the exact range of the inverse predicate.
+ Instruction *Op;
const APInt *C;
CmpInst::Predicate CPred;
- if (match(&SI, m_Select(m_Specific(ICI), m_APInt(C), m_BinOp(BO))))
+ if (match(&SI, m_Select(m_Specific(ICI), m_APInt(C), m_Instruction(Op))))
CPred = ICI->getPredicate();
- else if (match(&SI, m_Select(m_Specific(ICI), m_BinOp(BO), m_APInt(C))))
+ else if (match(&SI, m_Select(m_Specific(ICI), m_Instruction(Op), m_APInt(C))))
CPred = ICI->getInversePredicate();
else
return nullptr;
- const APInt *BinOpC;
- if (!match(BO, m_BinOp(m_Specific(V), m_APInt(BinOpC))))
- return nullptr;
-
- ConstantRange R = ConstantRange::makeExactICmpRegion(CPred, *CmpC)
- .binaryOp(BO->getOpcode(), *BinOpC);
- if (R == *C) {
- BO->dropPoisonGeneratingFlags();
- return BO;
+ ConstantRange InvDomCR = ConstantRange::makeExactICmpRegion(CPred, *CmpC);
+ const APInt *OpC;
+ if (match(Op, m_BinOp(m_Specific(V), m_APInt(OpC)))) {
+ ConstantRange R = InvDomCR.binaryOp(
+ static_cast<Instruction::BinaryOps>(Op->getOpcode()), *OpC);
+ if (R == *C) {
+ Op->dropPoisonGeneratingFlags();
+ return Op;
+ }
+ }
+ if (auto *MMI = dyn_cast<MinMaxIntrinsic>(Op);
+ MMI && MMI->getLHS() == V && match(MMI->getRHS(), m_APInt(OpC))) {
+ ConstantRange R = ConstantRange::intrinsic(MMI->getIntrinsicID(),
+ {InvDomCR, ConstantRange(*OpC)});
+ if (R == *C) {
+ MMI->dropPoisonGeneratingAnnotations();
+ return MMI;
+ }
}
+
return nullptr;
}
diff --git a/llvm/test/Transforms/InstCombine/select-min-max.ll b/llvm/test/Transforms/InstCombine/select-min-max.ll
index f13223284e6a68..0430fcd5ad3700 100644
--- a/llvm/test/Transforms/InstCombine/select-min-max.ll
+++ b/llvm/test/Transforms/InstCombine/select-min-max.ll
@@ -301,3 +301,66 @@ define i8 @not_smin_swap(i8 %i41, i8 %i43) {
%spec.select = select i1 %i44, i8 %i46, i8 0
ret i8 %spec.select
}
+
+define i8 @sel_umin_constant(i8 %x) {
+; CHECK-LABEL: @sel_umin_constant(
+; CHECK-NEXT: [[UMIN:%.*]] = call i8 @llvm.umin.i8(i8 [[X:%.*]], i8 16)
+; CHECK-NEXT: ret i8 [[UMIN]]
+;
+ %cmp = icmp sgt i8 %x, -1
+ %umin = call i8 @llvm.umin.i8(i8 %x, i8 16)
+ %sel = select i1 %cmp, i8 %umin, i8 16
+ ret i8 %sel
+}
+
+define i8 @sel_constant_smax_with_range_attr(i8 %x) {
+; CHECK-LABEL: @sel_constant_smax_with_range_attr(
+; CHECK-NEXT: [[SEL:%.*]] = call i8 @llvm.smax.i8(i8 [[X:%.*]], i8 16)
+; CHECK-NEXT: ret i8 [[SEL]]
+;
+ %cmp = icmp slt i8 %x, 0
+ %smax = call range(i8 8, 16) i8 @llvm.smax.i8(i8 %x, i8 16)
+ %sel = select i1 %cmp, i8 16, i8 %smax
+ ret i8 %sel
+}
+
+; Negative tests
+
+define i8 @sel_umin_constant_mismatch(i8 %x) {
+; CHECK-LABEL: @sel_umin_constant_mismatch(
+; CHECK-NEXT: [[CMP:%.*]] = icmp sgt i8 [[X:%.*]], -1
+; CHECK-NEXT: [[UMIN:%.*]] = call i8 @llvm.umin.i8(i8 [[X]], i8 16)
+; CHECK-NEXT: [[SEL:%.*]] = select i1 [[CMP]], i8 [[UMIN]], i8 15
+; CHECK-NEXT: ret i8 [[SEL]]
+;
+ %cmp = icmp sgt i8 %x, -1
+ %umin = call i8 @llvm.umin.i8(i8 %x, i8 16)
+ %sel = select i1 %cmp, i8 %umin, i8 15
+ ret i8 %sel
+}
+
+define i8 @sel_umin_constant_op_mismatch(i8 %x, i8 %y) {
+; CHECK-LABEL: @sel_umin_constant_op_mismatch(
+; CHECK-NEXT: [[CMP:%.*]] = icmp sgt i8 [[X:%.*]], -1
+; CHECK-NEXT: [[UMIN:%.*]] = call i8 @llvm.umin.i8(i8 [[Y:%.*]], i8 16)
+; CHECK-NEXT: [[SEL:%.*]] = select i1 [[CMP]], i8 [[UMIN]], i8 16
+; CHECK-NEXT: ret i8 [[SEL]]
+;
+ %cmp = icmp sgt i8 %x, -1
+ %umin = call i8 @llvm.umin.i8(i8 %y, i8 16)
+ %sel = select i1 %cmp, i8 %umin, i8 16
+ ret i8 %sel
+}
+
+define i8 @sel_umin_non_constant(i8 %x, i8 %y) {
+; CHECK-LABEL: @sel_umin_non_constant(
+; CHECK-NEXT: [[UMIN:%.*]] = call i8 @llvm.umin.i8(i8 [[X:%.*]], i8 16)
+; CHECK-NEXT: [[CMP1:%.*]] = icmp slt i8 [[X]], 0
+; CHECK-NEXT: [[SEL:%.*]] = select i1 [[CMP1]], i8 [[Y:%.*]], i8 [[UMIN]]
+; CHECK-NEXT: ret i8 [[SEL]]
+;
+ %cmp = icmp sgt i8 %x, -1
+ %umin = call i8 @llvm.umin.i8(i8 %x, i8 16)
+ %sel = select i1 %cmp, i8 %umin, i8 %y
+ ret i8 %sel
+}
|
This was referenced Feb 1, 2025
|
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/73/builds/12880 Here is the relevant piece of the build log for the reference |
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.
This patch extends f6bb156 to handle minmax intrinsics.
Motivating case: https://alive2.llvm.org/ce/z/JFKbYn
Addresses a regression caused by #121958.
It also works for
*.sat. But no real-world benefit is observed.