Skip to content

release/21.x: [SCEVExp] Fix early exit in ComputeEndCheck. #157410

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.

Sign up for GitHub

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

Closed
nikic wants to merge 3 commits into from
Closed

release/21.x: [SCEVExp] Fix early exit in ComputeEndCheck. #157410

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
4318653
[LV] Add more tests for interleave groups requiring predicates.
fhahn Sep 4, 2025
7fec7e6
[LV] Don't run instcombine for interleaved-accesses test.
fhahn Sep 4, 2025
120698f
[SCEVExp] Fix early exit in ComputeEndCheck. (#156910)
fhahn Sep 5, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
11 changes: 9 additions & 2 deletions llvm/lib/Transforms/Utils/ScalarEvolutionExpander.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2133,8 +2133,15 @@ Value *SCEVExpander::generateOverflowCheck(const SCEVAddRecExpr *AR,
// negative. If Step is known to be positive or negative, only create
// either 1. or 2.
auto ComputeEndCheck = [&]() -> Value * {
// Checking <u 0 is always false.
if (!Signed && Start->isZero() && SE.isKnownPositive(Step))
// Checking <u 0 is always false, if (Step * trunc ExitCount) does not wrap.
// TODO: Predicates that can be proven true/false should be discarded when
// the predicates are created, not late during expansion.
if (!Signed && Start->isZero() && SE.isKnownPositive(Step) &&
DstBits < SrcBits &&
ExitCount == SE.getZeroExtendExpr(SE.getTruncateExpr(ExitCount, ARTy),
ExitCount->getType()) &&
SE.willNotOverflow(Instruction::Mul, Signed, Step,
SE.getTruncateExpr(ExitCount, ARTy)))
return ConstantInt::getFalse(Loc->getContext());

// Get the backedge taken count and truncate or extended to the AR type.
Expand Down
18 changes: 14 additions & 4 deletions llvm/test/Transforms/LoopDistribute/scev-inserted-runtime-check.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -10,8 +10,13 @@ define void @f(ptr noalias %a, ptr noalias %b, ptr noalias %c, ptr noalias %d, p
; CHECK-NEXT: br label [[FOR_BODY_LVER_CHECK:%.*]]
; CHECK: for.body.lver.check:
; CHECK-NEXT: [[TMP0:%.*]] = add i64 [[N:%.*]], -1
; CHECK-NEXT: [[TMP2:%.*]] = trunc i64 [[TMP0]] to i32
; CHECK-NEXT: [[MUL1:%.*]] = call { i32, i1 } @llvm.umul.with.overflow.i32(i32 2, i32 [[TMP2]])
; CHECK-NEXT: [[MUL_RESULT:%.*]] = extractvalue { i32, i1 } [[MUL1]], 0
; CHECK-NEXT: [[MUL_OVERFLOW:%.*]] = extractvalue { i32, i1 } [[MUL1]], 1
; CHECK-NEXT: [[TMP1:%.*]] = icmp ugt i64 [[TMP0]], 4294967295
; CHECK-NEXT: br i1 [[TMP1]], label [[FOR_BODY_PH_LVER_ORIG:%.*]], label [[FOR_BODY_PH_LDIST1:%.*]]
; CHECK-NEXT: [[TMP3:%.*]] = or i1 [[MUL_OVERFLOW]], [[TMP1]]
; CHECK-NEXT: br i1 [[TMP3]], label [[FOR_BODY_PH_LVER_ORIG:%.*]], label [[FOR_BODY_PH_LDIST1:%.*]]
; CHECK: for.body.ph.lver.orig:
; CHECK-NEXT: br label [[FOR_BODY_LVER_ORIG:%.*]]
; CHECK: for.body.lver.orig:
Expand Down Expand Up @@ -75,7 +80,7 @@ define void @f(ptr noalias %a, ptr noalias %b, ptr noalias %c, ptr noalias %d, p
; CHECK-NEXT: br i1 [[EXITCOND]], label [[FOR_END_LOOPEXIT2:%.*]], label [[FOR_BODY]]
; CHECK: for.end.loopexit:
; CHECK-NEXT: br label [[FOR_END:%.*]]
; CHECK: for.end.loopexit1:
; CHECK: for.end.loopexit2:
; CHECK-NEXT: br label [[FOR_END]]
; CHECK: for.end:
; CHECK-NEXT: ret void
Expand Down Expand Up @@ -135,8 +140,13 @@ define void @f_with_offset(ptr noalias %b, ptr noalias %c, ptr noalias %d, ptr n
; CHECK-NEXT: br label [[FOR_BODY_LVER_CHECK:%.*]]
; CHECK: for.body.lver.check:
; CHECK-NEXT: [[TMP0:%.*]] = add i64 [[N:%.*]], -1
; CHECK-NEXT: [[TMP2:%.*]] = trunc i64 [[TMP0]] to i32
; CHECK-NEXT: [[MUL1:%.*]] = call { i32, i1 } @llvm.umul.with.overflow.i32(i32 2, i32 [[TMP2]])
; CHECK-NEXT: [[MUL_RESULT:%.*]] = extractvalue { i32, i1 } [[MUL1]], 0
; CHECK-NEXT: [[MUL_OVERFLOW:%.*]] = extractvalue { i32, i1 } [[MUL1]], 1
; CHECK-NEXT: [[TMP1:%.*]] = icmp ugt i64 [[TMP0]], 4294967295
; CHECK-NEXT: br i1 [[TMP1]], label [[FOR_BODY_PH_LVER_ORIG:%.*]], label [[FOR_BODY_PH_LDIST1:%.*]]
; CHECK-NEXT: [[TMP3:%.*]] = or i1 [[MUL_OVERFLOW]], [[TMP1]]
; CHECK-NEXT: br i1 [[TMP3]], label [[FOR_BODY_PH_LVER_ORIG:%.*]], label [[FOR_BODY_PH_LDIST1:%.*]]
; CHECK: for.body.ph.lver.orig:
; CHECK-NEXT: br label [[FOR_BODY_LVER_ORIG:%.*]]
; CHECK: for.body.lver.orig:
Expand Down Expand Up @@ -200,7 +210,7 @@ define void @f_with_offset(ptr noalias %b, ptr noalias %c, ptr noalias %d, ptr n
; CHECK-NEXT: br i1 [[EXITCOND]], label [[FOR_END_LOOPEXIT2:%.*]], label [[FOR_BODY]]
; CHECK: for.end.loopexit:
; CHECK-NEXT: br label [[FOR_END:%.*]]
; CHECK: for.end.loopexit1:
; CHECK: for.end.loopexit2:
; CHECK-NEXT: br label [[FOR_END]]
; CHECK: for.end:
; CHECK-NEXT: ret void
Expand Down
5 changes: 3 additions & 2 deletions llvm/test/Transforms/LoopVectorize/X86/cost-model.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1045,8 +1045,8 @@ define i64 @live_in_known_1_via_scev() {
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x i64> [ <i64 3, i64 1, i64 1, i64 1>, [[VECTOR_PH]] ], [ [[VEC_PHI]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 4
; CHECK-NEXT: [[TMP2:%.*]] = icmp eq i32 [[INDEX_NEXT]], 8
; CHECK-NEXT: br i1 [[TMP2]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP24:![0-9]+]]
; CHECK-NEXT: [[TMP0:%.*]] = icmp eq i32 [[INDEX_NEXT]], 8
; CHECK-NEXT: br i1 [[TMP0]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP24:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: [[TMP3:%.*]] = call i64 @llvm.vector.reduce.mul.v4i64(<4 x i64> [[VEC_PHI]])
; CHECK-NEXT: br label [[EXIT:%.*]]
Expand Down Expand Up @@ -1213,6 +1213,7 @@ define i32 @g(i64 %n) {
; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i32 [[TMP1]], 4
; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[VEC_EPILOG_SCALAR_PH:%.*]], label [[VECTOR_SCEVCHECK:%.*]]
; CHECK: vector.scevcheck:
; CHECK-NEXT: [[TMP28:%.*]] = trunc i64 [[N]] to i32
; CHECK-NEXT: [[TMP2:%.*]] = icmp ugt i64 [[N]], 4294967295
; CHECK-NEXT: br i1 [[TMP2]], label [[VEC_EPILOG_SCALAR_PH]], label [[VECTOR_MAIN_LOOP_ITER_CHECK:%.*]]
; CHECK: vector.main.loop.iter.check:
Expand Down
124 changes: 0 additions & 124 deletions llvm/test/Transforms/LoopVectorize/interleaved-accesses-3.ll
View file
Open in desktop

This file was deleted.

Loading
Loading