Paths
| Differential D104148
[LoopUtils] Fix incorrect RT check bounds for loop-invariant mem accesses ClosedPublic Authored by mdchen on Jun 11 2021, 1:17 PM.
Details Summary void m(int *p, int *p2, int q) {
int *a = p;
int i, j, r;
i = 0;
for (; i < 63; i++) {
r = j = 0;
for (; j < 63; j++)
r -= p2[j];
a[i] = r;
}
}For the code above, expandBounds that works for LoopVectorize can
Diff Detail Event TimelineComment Actions Thanks for the patch. Some comments inline
mdchen retitled this revision from WIP: [LoopUtils] Fix incorrect runtimechecks to [LoopUtils] Fix incorrect runtime checks. mdchen marked 4 inline comments as done. mdchen added inline comments.
mdchen marked an inline comment as done.
mdchen retitled this revision from [LoopUtils] Fix incorrect runtime checks to WIP: [LoopUtils] Fix incorrect runtime checks.Jun 18 2021, 1:55 AM mdchen retitled this revision from WIP: [LoopUtils] Fix incorrect runtime checks to [LoopUtils] Fix incorrect runtime checks. mdchen retitled this revision from [LoopUtils] Fix incorrect runtime checks to [LoopUtils] Fix incorrect RT check bounds for loop-invariant mem accesses. Comment ActionsUpdated previous failed x86 cases. As a explanation for the changes, I think it's unnecessary to deal with loop-invariant cases separately Comment Actions Thanks for the update. Makes sense to me, it would just be good to reduce the noise in the test diffs
Comment Actions
Great thanks! You could probably already rebase the patch on top of D105438 and upload the new version. fhahn added a child revision: D105481: [LAA] Remove RuntimeCheckingPtrGroup::RtCheck member (NFC)..Jul 6 2021, 7:12 AM mdchen added a parent revision: D105438: [LV] Re-generate check lines of some fragile tests.Jul 7 2021, 7:06 AM
Comment Actions LGTM, thanks! It would be good to clean up llvm/test/Transforms/LoopVectorize/pr50686.ll a bit as well, to make maintaining it easier in the future.
This revision is now accepted and ready to land.Jul 10 2021, 6:47 AM This revision was landed with ongoing or failed builds.Jul 19 2021, 4:39 AM Closed by commit rGe908e063d1f9: [LoopUtils] Fix incorrect RT check bounds of loop-invariant mem accesses (authored by mdchen, committed by Peilin Guo <guopeilin1@huawei.com>). · Explain Why This revision was automatically updated to reflect the committed changes.
Revision Contents
Diff 356944 llvm/lib/Analysis/LoopAccessAnalysis.cpp
llvm/lib/Transforms/Utils/LoopUtils.cpp
llvm/test/Transforms/LoopVectorize/X86/invariant-store-vectorization.ll
llvm/test/Transforms/LoopVectorize/X86/uniform_mem_op.ll
llvm/test/Transforms/LoopVectorize/invariant-store-vectorization.ll
llvm/test/Transforms/LoopVectorize/multiple-strides-vectorization.ll
llvm/test/Transforms/LoopVectorize/pr47343-expander-lcssa-after-cfg-update.ll
llvm/test/Transforms/LoopVectorize/pr50686.ll
llvm/test/Transforms/LoopVersioningLICM/loopversioningLICM1.ll
llvm/test/Transforms/LoopVersioningLICM/loopversioningLICM2.ll
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
I think we should use the same code as in the else part to compute ScEnd. Perhaps we can just keep ScStart = ScEnd = Sc and move the increment code out of the else.
// Add the size of the pointed element to ScEnd. auto &DL = Lp->getHeader()->getModule()->getDataLayout(); Type *IdxTy = DL.getIndexType(Ptr->getType()); const SCEV *EltSizeSCEV = SE->getStoreSizeOfExpr(IdxTy, Ptr->getType()->getPointerElementType()); ScEnd = SE->getAddExpr(ScEnd, EltSizeSCEV);