[InstCombine] Refactor the code for folding logicop and sext/zext. NFC. #137132
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
This refactoring is for more easily adding the code to preserve disjoint or in the PR llvm#136815. Both casts must have one use for folding logicop and sext/zext when the src type differ to avoid creating an extra instruction. If the src type of casts are the same, only one of the casts needs to have one use. This PR also adds more tests for the same src type.
llvmbot
added
llvm:instcombine
|
@llvm/pr-subscribers-llvm-transforms Author: Jim Lin (tclin914) ChangesThis refactoring is for more easily adding the code to preserve disjoint or in the PR #136815. Both casts must have one use for folding logicop and sext/zext when the src type differ to avoid creating an extra instruction. If the src type of casts are the same, only one of the casts needs to have one use. This PR also adds more tests for the same src type. Full diff: https://github.com/llvm/llvm-project/pull/137132.diff 2 Files Affected:
diff --git a/llvm/lib/Transforms/InstCombine/InstCombineAndOrXor.cpp b/llvm/lib/Transforms/InstCombine/InstCombineAndOrXor.cpp
index f1b225c0f238a..3c0836da3c343 100644
--- a/llvm/lib/Transforms/InstCombine/InstCombineAndOrXor.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstCombineAndOrXor.cpp
@@ -1846,34 +1846,45 @@ Instruction *InstCombinerImpl::foldCastedBitwiseLogic(BinaryOperator &I) {
if (CastOpcode != Cast1->getOpcode())
return nullptr;
- // If the source types do not match, but the casts are matching extends, we
- // can still narrow the logic op.
- if (SrcTy != Cast1->getSrcTy()) {
- Value *X, *Y;
- if (match(Cast0, m_OneUse(m_ZExtOrSExt(m_Value(X)))) &&
- match(Cast1, m_OneUse(m_ZExtOrSExt(m_Value(Y))))) {
- // Cast the narrower source to the wider source type.
- unsigned XNumBits = X->getType()->getScalarSizeInBits();
- unsigned YNumBits = Y->getType()->getScalarSizeInBits();
- if (XNumBits < YNumBits)
+ // Can't fold it profitably if no one of casts has one use.
+ if (!Cast0->hasOneUse() && !Cast1->hasOneUse())
+ return nullptr;
+
+ Value *X, *Y;
+ if (match(Cast0, m_ZExtOrSExt(m_Value(X))) &&
+ match(Cast1, m_ZExtOrSExt(m_Value(Y)))) {
+ // Cast the narrower source to the wider source type.
+ unsigned XNumBits = X->getType()->getScalarSizeInBits();
+ unsigned YNumBits = Y->getType()->getScalarSizeInBits();
+ if (XNumBits != YNumBits) {
+ // Cast the narrower source to the wider source type only if both of casts
+ // have one use to avoid creating an extra instruction.
+ if (!Cast0->hasOneUse() || !Cast1->hasOneUse())
+ return nullptr;
+
+ // If the source types do not match, but the casts are matching extends,
+ // we can still narrow the logic op.
+ if (XNumBits < YNumBits) {
X = Builder.CreateCast(CastOpcode, X, Y->getType());
- else
+ } else if (YNumBits < XNumBits) {
Y = Builder.CreateCast(CastOpcode, Y, X->getType());
- // Do the logic op in the intermediate width, then widen more.
- Value *NarrowLogic = Builder.CreateBinOp(LogicOpc, X, Y);
- return CastInst::Create(CastOpcode, NarrowLogic, DestTy);
+ }
}
- // Give up for other cast opcodes.
- return nullptr;
+ // Do the logic op in the intermediate width, then widen more.
+ Value *NarrowLogic = Builder.CreateBinOp(LogicOpc, X, Y, I.getName());
+ return CastInst::Create(CastOpcode, NarrowLogic, DestTy);
}
+ // If the src type of casts are different, give up for other cast opcodes.
+ if (SrcTy != Cast1->getSrcTy())
+ return nullptr;
+
Value *Cast0Src = Cast0->getOperand(0);
Value *Cast1Src = Cast1->getOperand(0);
// fold logic(cast(A), cast(B)) -> cast(logic(A, B))
- if ((Cast0->hasOneUse() || Cast1->hasOneUse()) &&
- shouldOptimizeCast(Cast0) && shouldOptimizeCast(Cast1)) {
+ if (shouldOptimizeCast(Cast0) && shouldOptimizeCast(Cast1)) {
Value *NewOp = Builder.CreateBinOp(LogicOpc, Cast0Src, Cast1Src,
I.getName());
return CastInst::Create(CastOpcode, NewOp, DestTy);
diff --git a/llvm/test/Transforms/InstCombine/and-xor-or.ll b/llvm/test/Transforms/InstCombine/and-xor-or.ll
index 5a58995f6c315..74d1f2119a2ad 100644
--- a/llvm/test/Transforms/InstCombine/and-xor-or.ll
+++ b/llvm/test/Transforms/InstCombine/and-xor-or.ll
@@ -2,6 +2,8 @@
; RUN: opt < %s -passes=instcombine -S | FileCheck %s
declare void @use(i32)
+declare void @use_i16(i8)
+declare void @use_2xi16(<2 x i16>)
declare void @use_i8(i8)
declare void @use_i1(i1)
@@ -4250,8 +4252,8 @@ define i16 @and_zext_zext(i8 %x, i4 %y) {
; CHECK-LABEL: define {{[^@]+}}@and_zext_zext
; CHECK-SAME: (i8 [[X:%.*]], i4 [[Y:%.*]]) {
; CHECK-NEXT: [[TMP1:%.*]] = zext i4 [[Y]] to i8
-; CHECK-NEXT: [[TMP2:%.*]] = and i8 [[X]], [[TMP1]]
-; CHECK-NEXT: [[R:%.*]] = zext nneg i8 [[TMP2]] to i16
+; CHECK-NEXT: [[R1:%.*]] = and i8 [[X]], [[TMP1]]
+; CHECK-NEXT: [[R:%.*]] = zext nneg i8 [[R1]] to i16
; CHECK-NEXT: ret i16 [[R]]
;
%zx = zext i8 %x to i16
@@ -4260,12 +4262,41 @@ define i16 @and_zext_zext(i8 %x, i4 %y) {
ret i16 %r
}
+define i16 @and_zext_zext_2(i8 %x, i8 %y) {
+; CHECK-LABEL: define {{[^@]+}}@and_zext_zext_2
+; CHECK-SAME: (i8 [[X:%.*]], i8 [[Y:%.*]]) {
+; CHECK-NEXT: [[R1:%.*]] = and i8 [[X]], [[Y]]
+; CHECK-NEXT: [[R:%.*]] = zext i8 [[R1]] to i16
+; CHECK-NEXT: ret i16 [[R]]
+;
+ %zx = zext i8 %x to i16
+ %zy = zext i8 %y to i16
+ %r = and i16 %zx, %zy
+ ret i16 %r
+}
+
+define i16 @and_zext_zext_2_use1(i8 %x, i8 %y) {
+; CHECK-LABEL: define {{[^@]+}}@and_zext_zext_2_use1
+; CHECK-SAME: (i8 [[X:%.*]], i8 [[Y:%.*]]) {
+; CHECK-NEXT: [[ZX:%.*]] = zext i8 [[X]] to i16
+; CHECK-NEXT: call void @use_i16(i16 [[ZX]])
+; CHECK-NEXT: [[R1:%.*]] = and i8 [[X]], [[Y]]
+; CHECK-NEXT: [[R:%.*]] = zext i8 [[R1]] to i16
+; CHECK-NEXT: ret i16 [[R]]
+;
+ %zx = zext i8 %x to i16
+ call void @use_i16(i16 %zx)
+ %zy = zext i8 %y to i16
+ %r = and i16 %zx, %zy
+ ret i16 %r
+}
+
define i16 @or_zext_zext(i8 %x, i4 %y) {
; CHECK-LABEL: define {{[^@]+}}@or_zext_zext
; CHECK-SAME: (i8 [[X:%.*]], i4 [[Y:%.*]]) {
; CHECK-NEXT: [[TMP1:%.*]] = zext i4 [[Y]] to i8
-; CHECK-NEXT: [[TMP2:%.*]] = or i8 [[X]], [[TMP1]]
-; CHECK-NEXT: [[R:%.*]] = zext i8 [[TMP2]] to i16
+; CHECK-NEXT: [[R1:%.*]] = or i8 [[X]], [[TMP1]]
+; CHECK-NEXT: [[R:%.*]] = zext i8 [[R1]] to i16
; CHECK-NEXT: ret i16 [[R]]
;
%zx = zext i8 %x to i16
@@ -4274,12 +4305,41 @@ define i16 @or_zext_zext(i8 %x, i4 %y) {
ret i16 %r
}
+define i16 @or_zext_zext_2(i8 %x, i8 %y) {
+; CHECK-LABEL: define {{[^@]+}}@or_zext_zext_2
+; CHECK-SAME: (i8 [[X:%.*]], i8 [[Y:%.*]]) {
+; CHECK-NEXT: [[R1:%.*]] = or i8 [[Y]], [[X]]
+; CHECK-NEXT: [[R:%.*]] = zext i8 [[R1]] to i16
+; CHECK-NEXT: ret i16 [[R]]
+;
+ %zx = zext i8 %x to i16
+ %zy = zext i8 %y to i16
+ %r = or i16 %zy, %zx
+ ret i16 %r
+}
+
+define i16 @or_zext_zext_2_use1(i8 %x, i8 %y) {
+; CHECK-LABEL: define {{[^@]+}}@or_zext_zext_2_use1
+; CHECK-SAME: (i8 [[X:%.*]], i8 [[Y:%.*]]) {
+; CHECK-NEXT: [[ZX:%.*]] = zext i8 [[X]] to i16
+; CHECK-NEXT: call void @use_i16(i16 [[ZX]])
+; CHECK-NEXT: [[R1:%.*]] = or i8 [[Y]], [[X]]
+; CHECK-NEXT: [[R:%.*]] = zext i8 [[R1]] to i16
+; CHECK-NEXT: ret i16 [[R]]
+;
+ %zx = zext i8 %x to i16
+ call void @use_i16(i16 %zx)
+ %zy = zext i8 %y to i16
+ %r = or i16 %zy, %zx
+ ret i16 %r
+}
+
define <2 x i16> @xor_zext_zext(<2 x i8> %x, <2 x i4> %y) {
; CHECK-LABEL: define {{[^@]+}}@xor_zext_zext
; CHECK-SAME: (<2 x i8> [[X:%.*]], <2 x i4> [[Y:%.*]]) {
; CHECK-NEXT: [[TMP1:%.*]] = zext <2 x i4> [[Y]] to <2 x i8>
-; CHECK-NEXT: [[TMP2:%.*]] = xor <2 x i8> [[X]], [[TMP1]]
-; CHECK-NEXT: [[R:%.*]] = zext <2 x i8> [[TMP2]] to <2 x i16>
+; CHECK-NEXT: [[R1:%.*]] = xor <2 x i8> [[X]], [[TMP1]]
+; CHECK-NEXT: [[R:%.*]] = zext <2 x i8> [[R1]] to <2 x i16>
; CHECK-NEXT: ret <2 x i16> [[R]]
;
%zx = zext <2 x i8> %x to <2 x i16>
@@ -4288,12 +4348,41 @@ define <2 x i16> @xor_zext_zext(<2 x i8> %x, <2 x i4> %y) {
ret <2 x i16> %r
}
+define <2 x i16> @xor_zext_zext_2(<2 x i8> %x, <2 x i8> %y) {
+; CHECK-LABEL: define {{[^@]+}}@xor_zext_zext_2
+; CHECK-SAME: (<2 x i8> [[X:%.*]], <2 x i8> [[Y:%.*]]) {
+; CHECK-NEXT: [[R1:%.*]] = xor <2 x i8> [[X]], [[Y]]
+; CHECK-NEXT: [[R:%.*]] = zext <2 x i8> [[R1]] to <2 x i16>
+; CHECK-NEXT: ret <2 x i16> [[R]]
+;
+ %zx = zext <2 x i8> %x to <2 x i16>
+ %zy = zext <2 x i8> %y to <2 x i16>
+ %r = xor <2 x i16> %zx, %zy
+ ret <2 x i16> %r
+}
+
+define <2 x i16> @xor_zext_zext_2_use1(<2 x i8> %x, <2 x i8> %y) {
+; CHECK-LABEL: define {{[^@]+}}@xor_zext_zext_2_use1
+; CHECK-SAME: (<2 x i8> [[X:%.*]], <2 x i8> [[Y:%.*]]) {
+; CHECK-NEXT: [[ZX:%.*]] = zext <2 x i8> [[X]] to <2 x i16>
+; CHECK-NEXT: call void @use_2xi16(<2 x i16> [[ZX]])
+; CHECK-NEXT: [[R1:%.*]] = xor <2 x i8> [[X]], [[Y]]
+; CHECK-NEXT: [[R:%.*]] = zext <2 x i8> [[R1]] to <2 x i16>
+; CHECK-NEXT: ret <2 x i16> [[R]]
+;
+ %zx = zext <2 x i8> %x to <2 x i16>
+ call void @use_2xi16(<2 x i16> %zx)
+ %zy = zext <2 x i8> %y to <2 x i16>
+ %r = xor <2 x i16> %zx, %zy
+ ret <2 x i16> %r
+}
+
define i16 @and_sext_sext(i8 %x, i4 %y) {
; CHECK-LABEL: define {{[^@]+}}@and_sext_sext
; CHECK-SAME: (i8 [[X:%.*]], i4 [[Y:%.*]]) {
; CHECK-NEXT: [[TMP1:%.*]] = sext i4 [[Y]] to i8
-; CHECK-NEXT: [[TMP2:%.*]] = and i8 [[X]], [[TMP1]]
-; CHECK-NEXT: [[R:%.*]] = sext i8 [[TMP2]] to i16
+; CHECK-NEXT: [[R1:%.*]] = and i8 [[X]], [[TMP1]]
+; CHECK-NEXT: [[R:%.*]] = sext i8 [[R1]] to i16
; CHECK-NEXT: ret i16 [[R]]
;
%sx = sext i8 %x to i16
@@ -4302,12 +4391,41 @@ define i16 @and_sext_sext(i8 %x, i4 %y) {
ret i16 %r
}
+define i16 @and_sext_sext_2(i8 %x, i8 %y) {
+; CHECK-LABEL: define {{[^@]+}}@and_sext_sext_2
+; CHECK-SAME: (i8 [[X:%.*]], i8 [[Y:%.*]]) {
+; CHECK-NEXT: [[R1:%.*]] = and i8 [[Y]], [[X]]
+; CHECK-NEXT: [[R:%.*]] = sext i8 [[R1]] to i16
+; CHECK-NEXT: ret i16 [[R]]
+;
+ %sx = sext i8 %x to i16
+ %sy = sext i8 %y to i16
+ %r = and i16 %sy, %sx
+ ret i16 %r
+}
+
+define i16 @and_sext_sext_2_use1(i8 %x, i8 %y) {
+; CHECK-LABEL: define {{[^@]+}}@and_sext_sext_2_use1
+; CHECK-SAME: (i8 [[X:%.*]], i8 [[Y:%.*]]) {
+; CHECK-NEXT: [[SX:%.*]] = sext i8 [[X]] to i16
+; CHECK-NEXT: call void @use_i16(i16 [[SX]])
+; CHECK-NEXT: [[R1:%.*]] = and i8 [[Y]], [[X]]
+; CHECK-NEXT: [[R:%.*]] = sext i8 [[R1]] to i16
+; CHECK-NEXT: ret i16 [[R]]
+;
+ %sx = sext i8 %x to i16
+ call void @use_i16(i16 %sx)
+ %sy = sext i8 %y to i16
+ %r = and i16 %sy, %sx
+ ret i16 %r
+}
+
define i16 @or_sext_sext(i8 %x, i4 %y) {
; CHECK-LABEL: define {{[^@]+}}@or_sext_sext
; CHECK-SAME: (i8 [[X:%.*]], i4 [[Y:%.*]]) {
; CHECK-NEXT: [[TMP1:%.*]] = sext i4 [[Y]] to i8
-; CHECK-NEXT: [[TMP2:%.*]] = or i8 [[X]], [[TMP1]]
-; CHECK-NEXT: [[R:%.*]] = sext i8 [[TMP2]] to i16
+; CHECK-NEXT: [[R1:%.*]] = or i8 [[X]], [[TMP1]]
+; CHECK-NEXT: [[R:%.*]] = sext i8 [[R1]] to i16
; CHECK-NEXT: ret i16 [[R]]
;
%sx = sext i8 %x to i16
@@ -4316,12 +4434,41 @@ define i16 @or_sext_sext(i8 %x, i4 %y) {
ret i16 %r
}
+define i16 @or_sext_sext_2(i8 %x, i8 %y) {
+; CHECK-LABEL: define {{[^@]+}}@or_sext_sext_2
+; CHECK-SAME: (i8 [[X:%.*]], i8 [[Y:%.*]]) {
+; CHECK-NEXT: [[R1:%.*]] = or i8 [[X]], [[Y]]
+; CHECK-NEXT: [[R:%.*]] = sext i8 [[R1]] to i16
+; CHECK-NEXT: ret i16 [[R]]
+;
+ %sx = sext i8 %x to i16
+ %sy = sext i8 %y to i16
+ %r = or i16 %sx, %sy
+ ret i16 %r
+}
+
+define i16 @or_sext_sext_2_use1(i8 %x, i8 %y) {
+; CHECK-LABEL: define {{[^@]+}}@or_sext_sext_2_use1
+; CHECK-SAME: (i8 [[X:%.*]], i8 [[Y:%.*]]) {
+; CHECK-NEXT: [[SX:%.*]] = sext i8 [[X]] to i16
+; CHECK-NEXT: call void @use_i16(i16 [[SX]])
+; CHECK-NEXT: [[R1:%.*]] = or i8 [[X]], [[Y]]
+; CHECK-NEXT: [[R:%.*]] = sext i8 [[R1]] to i16
+; CHECK-NEXT: ret i16 [[R]]
+;
+ %sx = sext i8 %x to i16
+ call void @use_i16(i16 %sx)
+ %sy = sext i8 %y to i16
+ %r = or i16 %sx, %sy
+ ret i16 %r
+}
+
define i16 @xor_sext_sext(i8 %x, i4 %y) {
; CHECK-LABEL: define {{[^@]+}}@xor_sext_sext
; CHECK-SAME: (i8 [[X:%.*]], i4 [[Y:%.*]]) {
; CHECK-NEXT: [[TMP1:%.*]] = sext i4 [[Y]] to i8
-; CHECK-NEXT: [[TMP2:%.*]] = xor i8 [[X]], [[TMP1]]
-; CHECK-NEXT: [[R:%.*]] = sext i8 [[TMP2]] to i16
+; CHECK-NEXT: [[R1:%.*]] = xor i8 [[X]], [[TMP1]]
+; CHECK-NEXT: [[R:%.*]] = sext i8 [[R1]] to i16
; CHECK-NEXT: ret i16 [[R]]
;
%sx = sext i8 %x to i16
@@ -4330,6 +4477,35 @@ define i16 @xor_sext_sext(i8 %x, i4 %y) {
ret i16 %r
}
+define i16 @xor_sext_sext_2(i8 %x, i8 %y) {
+; CHECK-LABEL: define {{[^@]+}}@xor_sext_sext_2
+; CHECK-SAME: (i8 [[X:%.*]], i8 [[Y:%.*]]) {
+; CHECK-NEXT: [[R1:%.*]] = xor i8 [[X]], [[Y]]
+; CHECK-NEXT: [[R:%.*]] = sext i8 [[R1]] to i16
+; CHECK-NEXT: ret i16 [[R]]
+;
+ %sx = sext i8 %x to i16
+ %sy = sext i8 %y to i16
+ %r = xor i16 %sx, %sy
+ ret i16 %r
+}
+
+define i16 @xor_sext_sext_2_use1(i8 %x, i8 %y) {
+; CHECK-LABEL: define {{[^@]+}}@xor_sext_sext_2_use1
+; CHECK-SAME: (i8 [[X:%.*]], i8 [[Y:%.*]]) {
+; CHECK-NEXT: [[SX:%.*]] = sext i8 [[X]] to i16
+; CHECK-NEXT: call void @use_i16(i16 [[SX]])
+; CHECK-NEXT: [[R1:%.*]] = xor i8 [[X]], [[Y]]
+; CHECK-NEXT: [[R:%.*]] = sext i8 [[R1]] to i16
+; CHECK-NEXT: ret i16 [[R]]
+;
+ %sx = sext i8 %x to i16
+ call void @use_i16(i16 %sx)
+ %sy = sext i8 %y to i16
+ %r = xor i16 %sx, %sy
+ ret i16 %r
+}
+
; negative test - mismatched casts
define i16 @and_zext_sext(i8 %x, i4 %y) {
@@ -4364,6 +4540,24 @@ define i32 @and_zext_zext_use1(i8 %x, i4 %y) {
ret i32 %r
}
+define i32 @and_zext_zext_use2(i8 %x, i8 %y) {
+; CHECK-LABEL: define {{[^@]+}}@and_zext_zext_use2
+; CHECK-SAME: (i8 [[X:%.*]], i8 [[Y:%.*]]) {
+; CHECK-NEXT: [[ZX:%.*]] = zext i8 [[X]] to i32
+; CHECK-NEXT: call void @use(i32 [[ZX]])
+; CHECK-NEXT: [[ZY:%.*]] = zext i8 [[Y]] to i32
+; CHECK-NEXT: call void @use(i32 [[ZY]])
+; CHECK-NEXT: [[R:%.*]] = and i32 [[ZX]], [[ZY]]
+; CHECK-NEXT: ret i32 [[R]]
+;
+ %zx = zext i8 %x to i32
+ call void @use(i32 %zx)
+ %zy = zext i8 %y to i32
+ call void @use(i32 %zy)
+ %r = and i32 %zx, %zy
+ ret i32 %r
+}
+
; negative test - don't create an extra instruction
define i32 @or_sext_sext_use1(i8 %x, i4 %y) {
@@ -4382,6 +4576,24 @@ define i32 @or_sext_sext_use1(i8 %x, i4 %y) {
ret i32 %r
}
+define i32 @or_sext_sext_use2(i8 %x, i8 %y) {
+; CHECK-LABEL: define {{[^@]+}}@or_sext_sext_use2
+; CHECK-SAME: (i8 [[X:%.*]], i8 [[Y:%.*]]) {
+; CHECK-NEXT: [[SX:%.*]] = sext i8 [[X]] to i32
+; CHECK-NEXT: call void @use(i32 [[SX]])
+; CHECK-NEXT: [[SY:%.*]] = sext i8 [[Y]] to i32
+; CHECK-NEXT: call void @use(i32 [[SY]])
+; CHECK-NEXT: [[R:%.*]] = or i32 [[SX]], [[SY]]
+; CHECK-NEXT: ret i32 [[R]]
+;
+ %sx = sext i8 %x to i32
+ call void @use(i32 %sx)
+ %sy = sext i8 %y to i32
+ call void @use(i32 %sy)
+ %r = or i32 %sx, %sy
+ ret i32 %r
+}
+
define i1 @PR56294(i8 %x) {
; CHECK-LABEL: define {{[^@]+}}@PR56294
; CHECK-SAME: (i8 [[X:%.*]]) {
|
nikic
changed the title
tclin914
changed the title
|
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/24/builds/7734 Here is the relevant piece of the build log for the reference |
IanWood1
pushed a commit
to IanWood1/llvm-project
that referenced
this pull request
May 6, 2025
…C. (llvm#137132) This refactoring is for more easily adding the code to preserve disjoint or in the PR llvm#136815. Both casts must have one use for folding logicop and sext/zext when the src type differ to avoid creating an extra instruction. If the src type of casts are the same, only one of the casts needs to have one use. This PR also adds more tests for the same src type.
IanWood1
pushed a commit
to IanWood1/llvm-project
that referenced
this pull request
May 6, 2025
…C. (llvm#137132) This refactoring is for more easily adding the code to preserve disjoint or in the PR llvm#136815. Both casts must have one use for folding logicop and sext/zext when the src type differ to avoid creating an extra instruction. If the src type of casts are the same, only one of the casts needs to have one use. This PR also adds more tests for the same src type.
IanWood1
pushed a commit
to IanWood1/llvm-project
that referenced
this pull request
May 6, 2025
…C. (llvm#137132) This refactoring is for more easily adding the code to preserve disjoint or in the PR llvm#136815. Both casts must have one use for folding logicop and sext/zext when the src type differ to avoid creating an extra instruction. If the src type of casts are the same, only one of the casts needs to have one use. This PR also adds more tests for the same src type.
Ankur-0429
pushed a commit
to Ankur-0429/llvm-project
that referenced
this pull request
May 9, 2025
…C. (llvm#137132) This refactoring is for more easily adding the code to preserve disjoint or in the PR llvm#136815. Both casts must have one use for folding logicop and sext/zext when the src type differ to avoid creating an extra instruction. If the src type of casts are the same, only one of the casts needs to have one use. This PR also adds more tests for the same src type.
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 refactoring is for more easily adding the code to preserve disjoint or in the PR #136815.
Both casts must have one use for folding logicop and sext/zext when the src type differ to avoid creating an extra instruction. If the src type of casts are the same, only one of the casts needs to have one use. This PR also adds more tests for the same src type.