This is an archive of the discontinued LLVM Phabricator instance.

[LV] Unconditionally branch from middle to scalar preheader if the scalar loop must execute (try 4)
AbandonedPublic

Authored by reames on Jul 6 2021, 11:33 AM.

Details

Reviewers
None
Summary

(Not a real review, posted to save state until bots are quiet.

Resubmit after the following:

  • Fix a latent bug related to unrolling with required epilogue (see e49d65f). I believe this is the cause of the prior PPC buildbot failure.
  • Disable non-latch exits for epilogue vectorization to be safe (9ffa90d)
  • Split out assert movement (600624a) to reduce churn if this gets reverted again.

Previous commit message (try 3)

Resubmit after fixing test/Transforms/LoopVectorize/ARM/mve-gather-scatter-tailpred.ll

Previous commit message...

This is a resubmit of 3e5ce4 (which was reverted by 7fe41ac).  The original commit caused a PPC build bot failure we never really got to the bottom of.  I can't reproduce the issue, and the bot owner was non-responsive.  In the meantime, we stumbled across an issue which seems possibly related, and worked around a latent bug in 80e8025.  My best guess is that the original patch exposed that latent issue at higher frequency, but it really is just a guess.

Original commit message follows...

If we know that the scalar epilogue is required to run, modify the CFG to end the middle block with an unconditional branch to scalar preheader. This is instead of a conditional branch to either the preheader or the exit block.

The motivation to do this is to support multiple exit blocks. Specifically, the current structure forces us to identify immediate dominators and *which* exit block to branch from in the middle terminator. For the multiple exit case - where we know require scalar will hold - these questions are ill formed.

This is the last change needed to support multiple exit loops, but since the diffs are already large enough, I'm going to land this, and then enable separately. You can think of this as being NFCIish prep work, but the changes are a bit too involved for me to feel comfortable tagging the review that way.

Differential Revision: https://reviews.llvm.org/D94892

Diff Detail

Event Timeline

reames created this revision.Jul 6 2021, 11:33 AM
reames requested review of this revision.Jul 6 2021, 11:33 AM
Herald added a project: Restricted Project. · View Herald TranscriptJul 6 2021, 11:33 AM
reames abandoned this revision.Jul 7 2021, 7:53 AM

Note to future self...

If this doesn't stick, we could consider generating a switch with a case for each (dead) exit instead. We could both assert that the switch is only used with requireScalarEpilogue and insert trap insts along the dead edges. This would likely restrict miscompiles to the non-latch exit case and might be easier to debug if needed.