-
Notifications
You must be signed in to change notification settings - Fork 14k
[InstCombine] Fold icmp of gep chain with base #144065
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Open
nikic
wants to merge
2
commits into
llvm:main
Choose a base branch
from
nikic:gep-icmp-multiple
base: main
Could not load branches
Branch not found: {{ refName }}
Loading
Could not load tags
Nothing to show
Loading
Are you sure you want to change the base?
Some commits from the old base branch may be removed from the timeline,
and old review comments may become outdated.
Open
+74
−13
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
Fold icmp between a chain of geps and its based pointer. Previously only a single gep was supported. This will be extended to handle the case of two gep chains with a common base in a followup.
@llvm/pr-subscribers-llvm-transforms Author: Nikita Popov (nikic) ChangesFold icmp between a chain of geps and its base pointer. Previously only a single gep was supported. This will be extended to handle the case of two gep chains with a common base in a followup. This helps to avoid regressions after #137297. Full diff: https://github.com/llvm/llvm-project/pull/144065.diff 4 Files Affected:
diff --git a/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp b/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
index c112fae351817..60cd93f9e2d42 100644
--- a/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
@@ -708,10 +708,12 @@ Instruction *InstCombinerImpl::foldGEPICmp(GEPOperator *GEPLHS, Value *RHS,
return I;
};
- Value *PtrBase = GEPLHS->getOperand(0);
- if (PtrBase == RHS && CanFold(GEPLHS->getNoWrapFlags())) {
+ CommonPointerBase Base = CommonPointerBase::compute(GEPLHS, RHS);
+ if (Base.Ptr == RHS && CanFold(Base.LHSNW)) {
// ((gep Ptr, OFFSET) cmp Ptr) ---> (OFFSET cmp 0).
- Value *Offset = EmitGEPOffset(GEPLHS);
+ Type *IdxTy = DL.getIndexType(GEPLHS->getType());
+ Value *Offset =
+ EmitGEPOffsets(Base.LHSGEPs, Base.LHSNW, IdxTy, /*RewriteGEPs=*/false);
return NewICmp(GEPLHS->getNoWrapFlags(), Offset,
Constant::getNullValue(Offset->getType()));
}
@@ -750,6 +752,7 @@ Instruction *InstCombinerImpl::foldGEPICmp(GEPOperator *GEPLHS, Value *RHS,
// If the base pointers are different, but the indices are the same, just
// compare the base pointer.
+ Value *PtrBase = GEPLHS->getOperand(0);
if (PtrBase != GEPRHS->getOperand(0)) {
bool IndicesTheSame =
GEPLHS->getNumOperands() == GEPRHS->getNumOperands() &&
diff --git a/llvm/test/Transforms/InstCombine/getelementptr.ll b/llvm/test/Transforms/InstCombine/getelementptr.ll
index 61236df80bfa6..23e682f5217cd 100644
--- a/llvm/test/Transforms/InstCombine/getelementptr.ll
+++ b/llvm/test/Transforms/InstCombine/getelementptr.ll
@@ -694,7 +694,7 @@ define i32 @test28() nounwind {
; CHECK-NEXT: [[TMP0:%.*]] = sext i32 [[T12_REC]] to i64
; CHECK-NEXT: [[T12:%.*]] = getelementptr inbounds [[STRUCT_X:%.*]], ptr [[T45]], i64 [[TMP0]]
; CHECK-NEXT: [[T16:%.*]] = call i32 (ptr, ...) @printf(ptr noundef nonnull dereferenceable(1) @.str1, ptr nonnull [[T12]]) #[[ATTR0]]
-; CHECK-NEXT: [[T84:%.*]] = icmp eq ptr [[T12]], [[ORIENTATIONS]]
+; CHECK-NEXT: [[T84:%.*]] = icmp eq i32 [[INDVAR]], 0
; CHECK-NEXT: [[INDVAR_NEXT]] = add i32 [[INDVAR]], 1
; CHECK-NEXT: br i1 [[T84]], label [[BB17:%.*]], label [[BB10]]
; CHECK: bb17:
diff --git a/llvm/test/Transforms/InstCombine/icmp-gep.ll b/llvm/test/Transforms/InstCombine/icmp-gep.ll
index 7f8f1ae73948d..c2ec3bc83e066 100644
--- a/llvm/test/Transforms/InstCombine/icmp-gep.ll
+++ b/llvm/test/Transforms/InstCombine/icmp-gep.ll
@@ -757,3 +757,67 @@ define i1 @gep_diff_base_same_indices_nuw_nusw(ptr %x, ptr %y, i64 %z) {
%cmp = icmp ult ptr %gep1, %gep2
ret i1 %cmp
}
+
+define i1 @gep_multiple_eq(ptr %base, i64 %idx, i64 %idx2) {
+; CHECK-LABEL: @gep_multiple_eq(
+; CHECK-NEXT: [[GEP1_IDX1:%.*]] = add i64 [[IDX:%.*]], [[IDX2:%.*]]
+; CHECK-NEXT: [[DOTMASK:%.*]] = and i64 [[GEP1_IDX1]], 4611686018427387903
+; CHECK-NEXT: [[CMP:%.*]] = icmp eq i64 [[DOTMASK]], 0
+; CHECK-NEXT: ret i1 [[CMP]]
+;
+ %gep1 = getelementptr i32, ptr %base, i64 %idx
+ %gep2 = getelementptr i32, ptr %gep1, i64 %idx2
+ %cmp = icmp eq ptr %gep2, %base
+ ret i1 %cmp
+}
+
+define i1 @gep_multiple_eq_commuted(ptr %base, i64 %idx, i64 %idx2) {
+; CHECK-LABEL: @gep_multiple_eq_commuted(
+; CHECK-NEXT: [[GEP1_IDX1:%.*]] = add i64 [[IDX:%.*]], [[IDX2:%.*]]
+; CHECK-NEXT: [[DOTMASK:%.*]] = and i64 [[GEP1_IDX1]], 4611686018427387903
+; CHECK-NEXT: [[CMP:%.*]] = icmp eq i64 [[DOTMASK]], 0
+; CHECK-NEXT: ret i1 [[CMP]]
+;
+ %gep1 = getelementptr i32, ptr %base, i64 %idx
+ %gep2 = getelementptr i32, ptr %gep1, i64 %idx2
+ %cmp = icmp eq ptr %base, %gep2
+ ret i1 %cmp
+}
+
+define i1 @gep_mugtiple_ugt_nuw(ptr %base, i64 %idx, i64 %idx2) {
+; CHECK-LABEL: @gep_mugtiple_ugt_nuw(
+; CHECK-NEXT: [[GEP1_IDX1:%.*]] = sub i64 0, [[IDX2:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = icmp ne i64 [[IDX:%.*]], [[GEP1_IDX1]]
+; CHECK-NEXT: ret i1 [[CMP]]
+;
+ %gep1 = getelementptr nuw i32, ptr %base, i64 %idx
+ %gep2 = getelementptr nuw i32, ptr %gep1, i64 %idx2
+ %cmp = icmp ugt ptr %gep2, %base
+ ret i1 %cmp
+}
+
+define i1 @gep_mugtiple_ugt_not_all_nuw(ptr %base, i64 %idx, i64 %idx2) {
+; CHECK-LABEL: @gep_mugtiple_ugt_not_all_nuw(
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr nuw i32, ptr [[BASE:%.*]], i64 [[IDX:%.*]]
+; CHECK-NEXT: [[GEP2:%.*]] = getelementptr i32, ptr [[GEP1]], i64 [[IDX2:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = icmp ugt ptr [[GEP2]], [[BASE]]
+; CHECK-NEXT: ret i1 [[CMP]]
+;
+ %gep1 = getelementptr nuw i32, ptr %base, i64 %idx
+ %gep2 = getelementptr i32, ptr %gep1, i64 %idx2
+ %cmp = icmp ugt ptr %gep2, %base
+ ret i1 %cmp
+}
+
+define i1 @gep_mugtiple_ugt_inbounds_nusw(ptr %base, i64 %idx, i64 %idx2) {
+; CHECK-LABEL: @gep_mugtiple_ugt_inbounds_nusw(
+; CHECK-NEXT: [[GEP1_IDX1:%.*]] = add i64 [[IDX:%.*]], [[IDX2:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[GEP1_IDX1]], 2
+; CHECK-NEXT: [[CMP:%.*]] = icmp sgt i64 [[TMP1]], 0
+; CHECK-NEXT: ret i1 [[CMP]]
+;
+ %gep1 = getelementptr inbounds i32, ptr %base, i64 %idx
+ %gep2 = getelementptr nusw i32, ptr %gep1, i64 %idx2
+ %cmp = icmp ugt ptr %gep2, %base
+ ret i1 %cmp
+}
diff --git a/llvm/test/Transforms/InstCombine/pr39908.ll b/llvm/test/Transforms/InstCombine/pr39908.ll
index ca143f417fb27..5d13a331c6d2e 100644
--- a/llvm/test/Transforms/InstCombine/pr39908.ll
+++ b/llvm/test/Transforms/InstCombine/pr39908.ll
@@ -7,9 +7,7 @@ target datalayout = "p:32:32"
define i1 @test(ptr %p, i32 %n) {
; CHECK-LABEL: @test(
-; CHECK-NEXT: [[END:%.*]] = getelementptr inbounds [0 x %S], ptr [[P:%.*]], i32 0, i32 [[N:%.*]], i32 0, i32 0
-; CHECK-NEXT: [[LAST:%.*]] = getelementptr inbounds i8, ptr [[END]], i32 -8
-; CHECK-NEXT: [[CMP:%.*]] = icmp eq ptr [[LAST]], [[P]]
+; CHECK-NEXT: [[CMP:%.*]] = icmp eq i32 [[N:%.*]], 1
; CHECK-NEXT: ret i1 [[CMP]]
;
%end = getelementptr inbounds [0 x %S], ptr %p, i32 0, i32 %n, i32 0, i32 0
@@ -22,9 +20,7 @@ define i1 @test(ptr %p, i32 %n) {
define i1 @test64(ptr %p, i64 %n) {
; CHECK-LABEL: @test64(
; CHECK-NEXT: [[TMP1:%.*]] = trunc i64 [[N:%.*]] to i32
-; CHECK-NEXT: [[END:%.*]] = getelementptr inbounds [0 x %S], ptr [[P:%.*]], i32 0, i32 [[TMP1]], i32 0, i32 0
-; CHECK-NEXT: [[LAST:%.*]] = getelementptr inbounds i8, ptr [[END]], i32 -8
-; CHECK-NEXT: [[CMP:%.*]] = icmp eq ptr [[LAST]], [[P]]
+; CHECK-NEXT: [[CMP:%.*]] = icmp eq i32 [[TMP1]], 1
; CHECK-NEXT: ret i1 [[CMP]]
;
%end = getelementptr inbounds [0 x %S], ptr %p, i64 0, i64 %n, i32 0, i64 0
@@ -37,9 +33,7 @@ define i1 @test64(ptr %p, i64 %n) {
define i1 @test64_overflow(ptr %p, i64 %n) {
; CHECK-LABEL: @test64_overflow(
; CHECK-NEXT: [[TMP1:%.*]] = trunc i64 [[N:%.*]] to i32
-; CHECK-NEXT: [[END:%.*]] = getelementptr inbounds [0 x %S], ptr [[P:%.*]], i32 0, i32 [[TMP1]], i32 0, i32 0
-; CHECK-NEXT: [[LAST:%.*]] = getelementptr inbounds i8, ptr [[END]], i32 -8
-; CHECK-NEXT: [[CMP:%.*]] = icmp eq ptr [[LAST]], [[P]]
+; CHECK-NEXT: [[CMP:%.*]] = icmp eq i32 [[TMP1]], 1
; CHECK-NEXT: ret i1 [[CMP]]
;
%end = getelementptr inbounds [0 x %S], ptr %p, i64 0, i64 %n, i32 0, i64 8589934592
|
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.
Fold icmp between a chain of geps and its base pointer. Previously only a single gep was supported.
This will be extended to handle the case of two gep chains with a common base in a followup.
This helps to avoid regressions after #137297.