[ValueTracking] Add support for insertelement in isKnownNonZero
#87703
Conversation
This file contains 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
Inserts don't modify the data, so if all elements that end up in the destination are non-zero the result is non-zero.
|
@llvm/pr-subscribers-llvm-analysis Author: None (goldsteinn) Changes
Full diff: https://github.com/llvm/llvm-project/pull/87703.diff 2 Files Affected:
diff --git a/llvm/lib/Analysis/ValueTracking.cpp b/llvm/lib/Analysis/ValueTracking.cpp
index 5ad4da43bca7db..fce22e3af12027 100644
--- a/llvm/lib/Analysis/ValueTracking.cpp
+++ b/llvm/lib/Analysis/ValueTracking.cpp
@@ -2751,6 +2751,29 @@ static bool isKnownNonZeroFromOperator(const Operator *I,
return isKnownNonZero(U.get(), DemandedElts, NewDepth, RecQ);
});
}
+ case Instruction::InsertElement: {
+ if (isa<ScalableVectorType>(I->getType()))
+ break;
+
+ const Value *Vec = I->getOperand(0);
+ const Value *Elt = I->getOperand(1);
+ auto *CIdx = dyn_cast<ConstantInt>(I->getOperand(2));
+
+ unsigned NumElts = DemandedElts.getBitWidth();
+ APInt DemandedVecElts = DemandedElts;
+ bool SkipElt = false;
+ // If we know the index we are inserting too, clear it from Vec check.
+ if (CIdx && CIdx->getValue().ult(NumElts)) {
+ DemandedVecElts.clearBit(CIdx->getZExtValue());
+ SkipElt = !DemandedElts[CIdx->getZExtValue()];
+ }
+
+ // Result is zero if Elt is non-zero and rest of the demanded elts in Vec
+ // are non-zero.
+ return (SkipElt || isKnownNonZero(Elt, Depth, Q)) &&
+ (DemandedVecElts.isZero() ||
+ isKnownNonZero(Vec, DemandedVecElts, Depth, Q));
+ }
case Instruction::ExtractElement:
if (const auto *EEI = dyn_cast<ExtractElementInst>(I)) {
const Value *Vec = EEI->getVectorOperand();
diff --git a/llvm/test/Transforms/InstSimplify/known-non-zero.ll b/llvm/test/Transforms/InstSimplify/known-non-zero.ll
index b647f11af4461d..a54309be3e4163 100644
--- a/llvm/test/Transforms/InstSimplify/known-non-zero.ll
+++ b/llvm/test/Transforms/InstSimplify/known-non-zero.ll
@@ -166,3 +166,87 @@ A:
B:
ret i1 0
}
+
+define <2 x i1> @insert_nonzero0(<2 x i8> %xx, i8 %yy) {
+; CHECK-LABEL: @insert_nonzero0(
+; CHECK-NEXT: ret <2 x i1> zeroinitializer
+;
+ %x = add nuw <2 x i8> %xx, <i8 1, i8 0>
+ %y = add nuw i8 %yy, 1
+
+ %ins = insertelement <2 x i8> %x, i8 %y, i32 1
+ %r = icmp eq <2 x i8> %ins, zeroinitializer
+ ret <2 x i1> %r
+}
+
+define <2 x i1> @insert_nonzero1(<2 x i8> %xx, i8 %yy) {
+; CHECK-LABEL: @insert_nonzero1(
+; CHECK-NEXT: ret <2 x i1> zeroinitializer
+;
+ %x = add nuw <2 x i8> %xx, <i8 0, i8 1>
+ %y = add nuw i8 %yy, 1
+
+ %ins = insertelement <2 x i8> %x, i8 %y, i32 0
+ %r = icmp eq <2 x i8> %ins, zeroinitializer
+ ret <2 x i1> %r
+}
+
+define <2 x i1> @insert_nonzero_fail(<2 x i8> %xx, i8 %yy) {
+; CHECK-LABEL: @insert_nonzero_fail(
+; CHECK-NEXT: [[X:%.*]] = add nuw <2 x i8> [[XX:%.*]], <i8 1, i8 0>
+; CHECK-NEXT: [[Y:%.*]] = add nuw i8 [[YY:%.*]], 1
+; CHECK-NEXT: [[INS:%.*]] = insertelement <2 x i8> [[X]], i8 [[Y]], i32 0
+; CHECK-NEXT: [[R:%.*]] = icmp eq <2 x i8> [[INS]], zeroinitializer
+; CHECK-NEXT: ret <2 x i1> [[R]]
+;
+ %x = add nuw <2 x i8> %xx, <i8 1, i8 0>
+ %y = add nuw i8 %yy, 1
+
+ %ins = insertelement <2 x i8> %x, i8 %y, i32 0
+ %r = icmp eq <2 x i8> %ins, zeroinitializer
+ ret <2 x i1> %r
+}
+
+define <2 x i1> @insert_nonzero_fail2(<2 x i8> %xx, i8 %yy) {
+; CHECK-LABEL: @insert_nonzero_fail2(
+; CHECK-NEXT: [[X:%.*]] = add nuw <2 x i8> [[XX:%.*]], <i8 0, i8 1>
+; CHECK-NEXT: [[Y:%.*]] = add i8 [[YY:%.*]], 1
+; CHECK-NEXT: [[INS:%.*]] = insertelement <2 x i8> [[X]], i8 [[Y]], i32 0
+; CHECK-NEXT: [[R:%.*]] = icmp eq <2 x i8> [[INS]], zeroinitializer
+; CHECK-NEXT: ret <2 x i1> [[R]]
+;
+ %x = add nuw <2 x i8> %xx, <i8 0, i8 1>
+ %y = add i8 %yy, 1
+
+ %ins = insertelement <2 x i8> %x, i8 %y, i32 0
+ %r = icmp eq <2 x i8> %ins, zeroinitializer
+ ret <2 x i1> %r
+}
+
+define <2 x i1> @insert_nonzero_any_idx(<2 x i8> %xx, i8 %yy, i32 %idx) {
+; CHECK-LABEL: @insert_nonzero_any_idx(
+; CHECK-NEXT: ret <2 x i1> zeroinitializer
+;
+ %x = add nuw <2 x i8> %xx, <i8 1, i8 1>
+ %y = add nuw i8 %yy, 1
+
+ %ins = insertelement <2 x i8> %x, i8 %y, i32 %idx
+ %r = icmp eq <2 x i8> %ins, zeroinitializer
+ ret <2 x i1> %r
+}
+
+define <2 x i1> @insert_nonzero_any_idx_fail(<2 x i8> %xx, i8 %yy, i32 %idx) {
+; CHECK-LABEL: @insert_nonzero_any_idx_fail(
+; CHECK-NEXT: [[X:%.*]] = add nuw <2 x i8> [[XX:%.*]], <i8 1, i8 0>
+; CHECK-NEXT: [[Y:%.*]] = add nuw i8 [[YY:%.*]], 1
+; CHECK-NEXT: [[INS:%.*]] = insertelement <2 x i8> [[X]], i8 [[Y]], i32 [[IDX:%.*]]
+; CHECK-NEXT: [[R:%.*]] = icmp eq <2 x i8> [[INS]], zeroinitializer
+; CHECK-NEXT: ret <2 x i1> [[R]]
+;
+ %x = add nuw <2 x i8> %xx, <i8 1, i8 0>
+ %y = add nuw i8 %yy, 1
+
+ %ins = insertelement <2 x i8> %x, i8 %y, i32 %idx
+ %r = icmp eq <2 x i8> %ins, zeroinitializer
+ ret <2 x i1> %r
+}
|
|
@llvm/pr-subscribers-llvm-transforms Author: None (goldsteinn) Changes
Full diff: https://github.com/llvm/llvm-project/pull/87703.diff 2 Files Affected:
diff --git a/llvm/lib/Analysis/ValueTracking.cpp b/llvm/lib/Analysis/ValueTracking.cpp
index 5ad4da43bca7db..fce22e3af12027 100644
--- a/llvm/lib/Analysis/ValueTracking.cpp
+++ b/llvm/lib/Analysis/ValueTracking.cpp
@@ -2751,6 +2751,29 @@ static bool isKnownNonZeroFromOperator(const Operator *I,
return isKnownNonZero(U.get(), DemandedElts, NewDepth, RecQ);
});
}
+ case Instruction::InsertElement: {
+ if (isa<ScalableVectorType>(I->getType()))
+ break;
+
+ const Value *Vec = I->getOperand(0);
+ const Value *Elt = I->getOperand(1);
+ auto *CIdx = dyn_cast<ConstantInt>(I->getOperand(2));
+
+ unsigned NumElts = DemandedElts.getBitWidth();
+ APInt DemandedVecElts = DemandedElts;
+ bool SkipElt = false;
+ // If we know the index we are inserting too, clear it from Vec check.
+ if (CIdx && CIdx->getValue().ult(NumElts)) {
+ DemandedVecElts.clearBit(CIdx->getZExtValue());
+ SkipElt = !DemandedElts[CIdx->getZExtValue()];
+ }
+
+ // Result is zero if Elt is non-zero and rest of the demanded elts in Vec
+ // are non-zero.
+ return (SkipElt || isKnownNonZero(Elt, Depth, Q)) &&
+ (DemandedVecElts.isZero() ||
+ isKnownNonZero(Vec, DemandedVecElts, Depth, Q));
+ }
case Instruction::ExtractElement:
if (const auto *EEI = dyn_cast<ExtractElementInst>(I)) {
const Value *Vec = EEI->getVectorOperand();
diff --git a/llvm/test/Transforms/InstSimplify/known-non-zero.ll b/llvm/test/Transforms/InstSimplify/known-non-zero.ll
index b647f11af4461d..a54309be3e4163 100644
--- a/llvm/test/Transforms/InstSimplify/known-non-zero.ll
+++ b/llvm/test/Transforms/InstSimplify/known-non-zero.ll
@@ -166,3 +166,87 @@ A:
B:
ret i1 0
}
+
+define <2 x i1> @insert_nonzero0(<2 x i8> %xx, i8 %yy) {
+; CHECK-LABEL: @insert_nonzero0(
+; CHECK-NEXT: ret <2 x i1> zeroinitializer
+;
+ %x = add nuw <2 x i8> %xx, <i8 1, i8 0>
+ %y = add nuw i8 %yy, 1
+
+ %ins = insertelement <2 x i8> %x, i8 %y, i32 1
+ %r = icmp eq <2 x i8> %ins, zeroinitializer
+ ret <2 x i1> %r
+}
+
+define <2 x i1> @insert_nonzero1(<2 x i8> %xx, i8 %yy) {
+; CHECK-LABEL: @insert_nonzero1(
+; CHECK-NEXT: ret <2 x i1> zeroinitializer
+;
+ %x = add nuw <2 x i8> %xx, <i8 0, i8 1>
+ %y = add nuw i8 %yy, 1
+
+ %ins = insertelement <2 x i8> %x, i8 %y, i32 0
+ %r = icmp eq <2 x i8> %ins, zeroinitializer
+ ret <2 x i1> %r
+}
+
+define <2 x i1> @insert_nonzero_fail(<2 x i8> %xx, i8 %yy) {
+; CHECK-LABEL: @insert_nonzero_fail(
+; CHECK-NEXT: [[X:%.*]] = add nuw <2 x i8> [[XX:%.*]], <i8 1, i8 0>
+; CHECK-NEXT: [[Y:%.*]] = add nuw i8 [[YY:%.*]], 1
+; CHECK-NEXT: [[INS:%.*]] = insertelement <2 x i8> [[X]], i8 [[Y]], i32 0
+; CHECK-NEXT: [[R:%.*]] = icmp eq <2 x i8> [[INS]], zeroinitializer
+; CHECK-NEXT: ret <2 x i1> [[R]]
+;
+ %x = add nuw <2 x i8> %xx, <i8 1, i8 0>
+ %y = add nuw i8 %yy, 1
+
+ %ins = insertelement <2 x i8> %x, i8 %y, i32 0
+ %r = icmp eq <2 x i8> %ins, zeroinitializer
+ ret <2 x i1> %r
+}
+
+define <2 x i1> @insert_nonzero_fail2(<2 x i8> %xx, i8 %yy) {
+; CHECK-LABEL: @insert_nonzero_fail2(
+; CHECK-NEXT: [[X:%.*]] = add nuw <2 x i8> [[XX:%.*]], <i8 0, i8 1>
+; CHECK-NEXT: [[Y:%.*]] = add i8 [[YY:%.*]], 1
+; CHECK-NEXT: [[INS:%.*]] = insertelement <2 x i8> [[X]], i8 [[Y]], i32 0
+; CHECK-NEXT: [[R:%.*]] = icmp eq <2 x i8> [[INS]], zeroinitializer
+; CHECK-NEXT: ret <2 x i1> [[R]]
+;
+ %x = add nuw <2 x i8> %xx, <i8 0, i8 1>
+ %y = add i8 %yy, 1
+
+ %ins = insertelement <2 x i8> %x, i8 %y, i32 0
+ %r = icmp eq <2 x i8> %ins, zeroinitializer
+ ret <2 x i1> %r
+}
+
+define <2 x i1> @insert_nonzero_any_idx(<2 x i8> %xx, i8 %yy, i32 %idx) {
+; CHECK-LABEL: @insert_nonzero_any_idx(
+; CHECK-NEXT: ret <2 x i1> zeroinitializer
+;
+ %x = add nuw <2 x i8> %xx, <i8 1, i8 1>
+ %y = add nuw i8 %yy, 1
+
+ %ins = insertelement <2 x i8> %x, i8 %y, i32 %idx
+ %r = icmp eq <2 x i8> %ins, zeroinitializer
+ ret <2 x i1> %r
+}
+
+define <2 x i1> @insert_nonzero_any_idx_fail(<2 x i8> %xx, i8 %yy, i32 %idx) {
+; CHECK-LABEL: @insert_nonzero_any_idx_fail(
+; CHECK-NEXT: [[X:%.*]] = add nuw <2 x i8> [[XX:%.*]], <i8 1, i8 0>
+; CHECK-NEXT: [[Y:%.*]] = add nuw i8 [[YY:%.*]], 1
+; CHECK-NEXT: [[INS:%.*]] = insertelement <2 x i8> [[X]], i8 [[Y]], i32 [[IDX:%.*]]
+; CHECK-NEXT: [[R:%.*]] = icmp eq <2 x i8> [[INS]], zeroinitializer
+; CHECK-NEXT: ret <2 x i1> [[R]]
+;
+ %x = add nuw <2 x i8> %xx, <i8 1, i8 0>
+ %y = add nuw i8 %yy, 1
+
+ %ins = insertelement <2 x i8> %x, i8 %y, i32 %idx
+ %r = icmp eq <2 x i8> %ins, zeroinitializer
+ ret <2 x i1> %r
+}
|
goldsteinn
changed the title
insertelement in isKnownNonZero
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.
insertelementinisKnownNonZero; NFCinsertelementinisKnownNonZero