Page MenuHomePhabricator

jaykang10 (JinGu Kang)
User

Projects

User does not belong to any projects.

User Details

User Since
Jun 18 2014, 2:14 AM (356 w, 21 h)

Recent Activity

Fri, Apr 9

jaykang10 updated the diff for D99354: [SimpleLoopUnswitch] Port partially invariant unswitch from LoopUnswitch to SimpleLoopUnswitch.

Following comments of @fhahn, updated patch.

Fri, Apr 9, 4:08 AM · Restricted Project
jaykang10 added a comment to D99354: [SimpleLoopUnswitch] Port partially invariant unswitch from LoopUnswitch to SimpleLoopUnswitch.

Does this patch give the expected speedup on omnetpp?

Fri, Apr 9, 3:23 AM · Restricted Project

Thu, Apr 8

jaykang10 added inline comments to D99490: [NFC][LoopUnswitch] Move hasPartialIVCondition to LoopUtils.
Thu, Apr 8, 5:15 AM · Restricted Project
jaykang10 added a comment to D99354: [SimpleLoopUnswitch] Port partially invariant unswitch from LoopUnswitch to SimpleLoopUnswitch.

@fhahn Can you review this change when you have time please?

Thu, Apr 8, 2:27 AM · Restricted Project

Wed, Apr 7

jaykang10 added a comment to D99774: [LoopUtils] Populate sibling loops in reverse program order on new pass manager.

Traversing forward for the loops can cause a value in an earlier loop to be evaluated, which could help a later loop, right? I'm now not sure that it's such a clear tradeoff.
For example, scev-expander-preserve-lcssa.ll looks worse now.

It could be interesting to revisit a previous loop if a loop was deleted.

I think there are more knowledgeable people about this sort of stuff on llvm-dev, could you make a post there?

Wed, Apr 7, 12:06 PM · Restricted Project
jaykang10 added a comment to D99774: [LoopUtils] Populate sibling loops in reverse program order on new pass manager.

@aeubanks Can we push this change please?

Wed, Apr 7, 9:09 AM · Restricted Project
jaykang10 updated the diff for D99354: [SimpleLoopUnswitch] Port partially invariant unswitch from LoopUnswitch to SimpleLoopUnswitch.

Rebased

Wed, Apr 7, 8:34 AM · Restricted Project
jaykang10 committed rG798b0fd36b48: [NPM] Fix typo inisLTOPreLink for loop rotate (authored by jaykang10).
[NPM] Fix typo inisLTOPreLink for loop rotate
Wed, Apr 7, 7:10 AM
jaykang10 closed D100033: [NPM] Fix typo inisLTOPreLink for loop rotate .
Wed, Apr 7, 7:09 AM · Restricted Project
jaykang10 retitled D100033: [NPM] Fix typo inisLTOPreLink for loop rotate from [LoopRotate] Fix typo for PrepareForLTO to [NPM] Fix typo inisLTOPreLink for loop rotate .
Wed, Apr 7, 6:51 AM · Restricted Project
jaykang10 requested review of D100033: [NPM] Fix typo inisLTOPreLink for loop rotate .
Wed, Apr 7, 6:33 AM · Restricted Project

Tue, Apr 6

jaykang10 added a comment to D99774: [LoopUtils] Populate sibling loops in reverse program order on new pass manager.

Actually, can you give a concrete example? The example in the description isn't clear enough

There will probably be cases where one way is better and cases where the other way is better. Might just be better to try to fix your use case directly in the passes themselves

@aeubanks Thanks for comments. Let me try to create a example for it.

Tue, Apr 6, 11:12 AM · Restricted Project
jaykang10 updated the diff for D99774: [LoopUtils] Populate sibling loops in reverse program order on new pass manager.

Added a test

Tue, Apr 6, 10:56 AM · Restricted Project
jaykang10 added a reviewer for D99774: [LoopUtils] Populate sibling loops in reverse program order on new pass manager: sanwou01.
Tue, Apr 6, 5:34 AM · Restricted Project
jaykang10 added a reviewer for D99354: [SimpleLoopUnswitch] Port partially invariant unswitch from LoopUnswitch to SimpleLoopUnswitch: chandlerc.
Tue, Apr 6, 5:28 AM · Restricted Project
jaykang10 added a comment to D99774: [LoopUtils] Populate sibling loops in reverse program order on new pass manager.

Actually, can you give a concrete example? The example in the description isn't clear enough

There will probably be cases where one way is better and cases where the other way is better. Might just be better to try to fix your use case directly in the passes themselves

Tue, Apr 6, 4:55 AM · Restricted Project
jaykang10 updated the diff for D99774: [LoopUtils] Populate sibling loops in reverse program order on new pass manager.

Fixed failed tests

Tue, Apr 6, 4:54 AM · Restricted Project
jaykang10 added a comment to D99354: [SimpleLoopUnswitch] Port partially invariant unswitch from LoopUnswitch to SimpleLoopUnswitch.

Any objection to push this change please?

Tue, Apr 6, 4:28 AM · Restricted Project

Fri, Apr 2

jaykang10 updated the summary of D99774: [LoopUtils] Populate sibling loops in reverse program order on new pass manager.
Fri, Apr 2, 12:57 AM · Restricted Project
jaykang10 updated the summary of D99774: [LoopUtils] Populate sibling loops in reverse program order on new pass manager.
Fri, Apr 2, 12:50 AM · Restricted Project
jaykang10 added a comment to D99354: [SimpleLoopUnswitch] Port partially invariant unswitch from LoopUnswitch to SimpleLoopUnswitch.

Any comments please?

Fri, Apr 2, 12:46 AM · Restricted Project

Thu, Apr 1

jaykang10 requested review of D99774: [LoopUtils] Populate sibling loops in reverse program order on new pass manager.
Thu, Apr 1, 3:19 PM · Restricted Project

Tue, Mar 30

jaykang10 updated the diff for D99354: [SimpleLoopUnswitch] Port partially invariant unswitch from LoopUnswitch to SimpleLoopUnswitch.
Tue, Mar 30, 1:49 AM · Restricted Project

Mon, Mar 29

jaykang10 updated the diff for D99354: [SimpleLoopUnswitch] Port partially invariant unswitch from LoopUnswitch to SimpleLoopUnswitch.

Rebase

Mon, Mar 29, 6:20 AM · Restricted Project
jaykang10 committed rGab728717033a: [SimpleLoopUnswitch] Fix wrong assertions in partial-unswitch.ll (authored by jaykang10).
[SimpleLoopUnswitch] Fix wrong assertions in partial-unswitch.ll
Mon, Mar 29, 6:05 AM
jaykang10 added a comment to D99493: [SimpleLoopUnswitch] Add tests to check partially invariant unswitch.

Sorry for mistake... I will push the change with trunk.

Mon, Mar 29, 6:01 AM · Restricted Project
jaykang10 updated the diff for D99354: [SimpleLoopUnswitch] Port partially invariant unswitch from LoopUnswitch to SimpleLoopUnswitch.

Rebased

Mon, Mar 29, 5:50 AM · Restricted Project
jaykang10 added a comment to D99493: [SimpleLoopUnswitch] Add tests to check partially invariant unswitch.

Looks like this test fails: http://45.33.8.238/linux/42881/step_12.txt

Mon, Mar 29, 5:46 AM · Restricted Project
jaykang10 committed rGe4abb64100e4: [LoopUnswitch] Use reference variables instead of pointer one (authored by jaykang10).
[LoopUnswitch] Use reference variables instead of pointer one
Mon, Mar 29, 5:36 AM
jaykang10 closed D99496: [LoopUnswitch] Use reference variables instead of pointer one.
Mon, Mar 29, 5:36 AM · Restricted Project
jaykang10 added a comment to D99496: [LoopUnswitch] Use reference variables instead of pointer one.

@fhahn If you need something more for this one, please let me know.

Mon, Mar 29, 5:17 AM · Restricted Project
jaykang10 committed rG07142b304004: [SimpleLoopUnswitch] Add tests to check partially invariant unswitch (authored by jaykang10).
[SimpleLoopUnswitch] Add tests to check partially invariant unswitch
Mon, Mar 29, 5:08 AM
jaykang10 closed D99493: [SimpleLoopUnswitch] Add tests to check partially invariant unswitch.
Mon, Mar 29, 5:07 AM · Restricted Project
jaykang10 added a comment to D99493: [SimpleLoopUnswitch] Add tests to check partially invariant unswitch.

LGTM, thanks!

Thanks for review @fhahn. I have a question. Is it ok to push this test now? or I have to wait for https://reviews.llvm.org/D99354?

I think you should push it and rebase D99354

Mon, Mar 29, 5:05 AM · Restricted Project
jaykang10 added a comment to D99493: [SimpleLoopUnswitch] Add tests to check partially invariant unswitch.

LGTM, thanks!

Mon, Mar 29, 4:41 AM · Restricted Project
jaykang10 updated the diff for D99496: [LoopUnswitch] Use reference variables instead of pointer one.

Following comment of @fhahn, updated code.

Mon, Mar 29, 4:38 AM · Restricted Project
jaykang10 added inline comments to D99496: [LoopUnswitch] Use reference variables instead of pointer one.
Mon, Mar 29, 4:35 AM · Restricted Project
jaykang10 requested review of D99496: [LoopUnswitch] Use reference variables instead of pointer one.
Mon, Mar 29, 4:23 AM · Restricted Project
jaykang10 added inline comments to D99490: [NFC][LoopUnswitch] Move hasPartialIVCondition to LoopUtils.
Mon, Mar 29, 3:29 AM · Restricted Project
jaykang10 updated the diff for D99354: [SimpleLoopUnswitch] Port partially invariant unswitch from LoopUnswitch to SimpleLoopUnswitch.

Following comment of @lebedev.ri, moved the test to separate patch.

Mon, Mar 29, 3:21 AM · Restricted Project
jaykang10 requested review of D99493: [SimpleLoopUnswitch] Add tests to check partially invariant unswitch.
Mon, Mar 29, 3:16 AM · Restricted Project
jaykang10 added inline comments to D99354: [SimpleLoopUnswitch] Port partially invariant unswitch from LoopUnswitch to SimpleLoopUnswitch.
Mon, Mar 29, 2:51 AM · Restricted Project
jaykang10 committed rGcfe87d4eddfc: [NFC][LoopUnswitch] Move hasPartialIVCondition to LoopUtils (authored by jaykang10).
[NFC][LoopUnswitch] Move hasPartialIVCondition to LoopUtils
Mon, Mar 29, 2:44 AM
jaykang10 closed D99490: [NFC][LoopUnswitch] Move hasPartialIVCondition to LoopUtils.
Mon, Mar 29, 2:44 AM · Restricted Project
jaykang10 added inline comments to D99354: [SimpleLoopUnswitch] Port partially invariant unswitch from LoopUnswitch to SimpleLoopUnswitch.
Mon, Mar 29, 2:42 AM · Restricted Project
jaykang10 updated the diff for D99354: [SimpleLoopUnswitch] Port partially invariant unswitch from LoopUnswitch to SimpleLoopUnswitch.

Following comment of @lebedev.ri, split previous patch into two patches. This one works on top of https://reviews.llvm.org/D99490

Mon, Mar 29, 2:27 AM · Restricted Project
jaykang10 requested review of D99490: [NFC][LoopUnswitch] Move hasPartialIVCondition to LoopUtils.
Mon, Mar 29, 2:17 AM · Restricted Project
jaykang10 added a comment to D99354: [SimpleLoopUnswitch] Port partially invariant unswitch from LoopUnswitch to SimpleLoopUnswitch.

Thank you for looking into this!
Unhelpful comment: have you considered splitting this into two patches, moving code from LoopUnswitch to a common place, and enhancing SimpleLoopUnswitch ?

Mon, Mar 29, 2:00 AM · Restricted Project
jaykang10 updated the diff for D99354: [SimpleLoopUnswitch] Port partially invariant unswitch from LoopUnswitch to SimpleLoopUnswitch.
Mon, Mar 29, 1:12 AM · Restricted Project

Sun, Mar 28

jaykang10 updated the diff for D99354: [SimpleLoopUnswitch] Port partially invariant unswitch from LoopUnswitch to SimpleLoopUnswitch.

Moved hasPartialIVCondition to LoopUtils.h following comment of @fhahn

Sun, Mar 28, 4:42 AM · Restricted Project

Fri, Mar 26

jaykang10 added a comment to D99354: [SimpleLoopUnswitch] Port partially invariant unswitch from LoopUnswitch to SimpleLoopUnswitch.

Thanks for putting up this patch! Would it be possible to move the detection code (hasPartialIVCondition) somewhere, so both version of unswitching can use the same code?

Fri, Mar 26, 2:20 AM · Restricted Project
jaykang10 added reviewers for D99354: [SimpleLoopUnswitch] Port partially invariant unswitch from LoopUnswitch to SimpleLoopUnswitch: jdoerfert, jonpa.
Fri, Mar 26, 1:37 AM · Restricted Project

Thu, Mar 25

jaykang10 committed rG3fd64cc7a361: [ValueTracking] Handle two PHIs in isKnownNonEqual() (authored by jaykang10).
[ValueTracking] Handle two PHIs in isKnownNonEqual()
Thu, Mar 25, 3:57 PM
jaykang10 closed D98422: [ValueTracking] Handle two PHIs in isKnownNonEqual().
Thu, Mar 25, 3:57 PM · Restricted Project
jaykang10 added a comment to D98422: [ValueTracking] Handle two PHIs in isKnownNonEqual().

LGTM, thanks for pulling this through!

Thu, Mar 25, 3:39 PM · Restricted Project
jaykang10 updated the diff for D98422: [ValueTracking] Handle two PHIs in isKnownNonEqual().

Allowed only one pair of phi operands for full recursion.

Thu, Mar 25, 3:32 PM · Restricted Project
jaykang10 added inline comments to D98422: [ValueTracking] Handle two PHIs in isKnownNonEqual().
Thu, Mar 25, 3:31 PM · Restricted Project
jaykang10 added inline comments to D98422: [ValueTracking] Handle two PHIs in isKnownNonEqual().
Thu, Mar 25, 1:27 PM · Restricted Project
jaykang10 updated the diff for D98422: [ValueTracking] Handle two PHIs in isKnownNonEqual().

Used MaxAnalysisRecursionDepth - 1 for limit of recursion with phi.

Thu, Mar 25, 12:24 PM · Restricted Project
jaykang10 added a comment to D98422: [ValueTracking] Handle two PHIs in isKnownNonEqual().

This looks mostly good. Only remaining issue is that we need to aggressively limit recursion when it comes to phi nodes. What most other ValueTracking code does is to use MaxAnalysisRecursionDepth - 1 as the depth for recursion over phis, which means that only one more level of recursion is allowed. If that still covers your motivating case, then I'd suggest to follow that pattern. If that doesn't cover your motivating case, then the alternative here would be to only allow full recursion over one pair of phi operands, and only check that the others are trivially non-equal (i.e. distinct ConstantInts). That should definitely cover your case, and would also be in line with existing practice for isKnownNonEqual(). Either way is fine by me.

Thu, Mar 25, 12:22 PM · Restricted Project
jaykang10 updated the diff for D98422: [ValueTracking] Handle two PHIs in isKnownNonEqual().

Following comment of @nikic, updated code.

Thu, Mar 25, 11:16 AM · Restricted Project
jaykang10 added inline comments to D98422: [ValueTracking] Handle two PHIs in isKnownNonEqual().
Thu, Mar 25, 11:15 AM · Restricted Project
jaykang10 requested review of D99354: [SimpleLoopUnswitch] Port partially invariant unswitch from LoopUnswitch to SimpleLoopUnswitch.
Thu, Mar 25, 10:30 AM · Restricted Project
jaykang10 updated the diff for D98422: [ValueTracking] Handle two PHIs in isKnownNonEqual().

Following comment of @lebedev.ri, updated code.

Thu, Mar 25, 10:16 AM · Restricted Project
jaykang10 added inline comments to D98422: [ValueTracking] Handle two PHIs in isKnownNonEqual().
Thu, Mar 25, 10:10 AM · Restricted Project
jaykang10 updated the diff for D98422: [ValueTracking] Handle two PHIs in isKnownNonEqual().

Following comments of @nikic, updated code.

Thu, Mar 25, 1:55 AM · Restricted Project
jaykang10 added inline comments to D98422: [ValueTracking] Handle two PHIs in isKnownNonEqual().
Thu, Mar 25, 1:50 AM · Restricted Project

Tue, Mar 23

jaykang10 updated the diff for D98422: [ValueTracking] Handle two PHIs in isKnownNonEqual().
Tue, Mar 23, 4:32 PM · Restricted Project
jaykang10 added a comment to D99069: [ValueTracking] Handle increasing mul recurrence in isKnownNonZero().

LGTM

Tue, Mar 23, 4:09 PM · Restricted Project
jaykang10 committed rG2e2740b859cf: [ValueTracking] Handle increasing mul recurrence in isKnownNonZero() (authored by jaykang10).
[ValueTracking] Handle increasing mul recurrence in isKnownNonZero()
Tue, Mar 23, 4:07 PM
jaykang10 closed D99069: [ValueTracking] Handle increasing mul recurrence in isKnownNonZero().
Tue, Mar 23, 4:06 PM · Restricted Project
jaykang10 added a comment to D98422: [ValueTracking] Handle two PHIs in isKnownNonEqual().

It works on top of https://reviews.llvm.org/D99069.

Tue, Mar 23, 6:39 AM · Restricted Project
jaykang10 updated the diff for D99069: [ValueTracking] Handle increasing mul recurrence in isKnownNonZero().

Added more tests

Tue, Mar 23, 3:15 AM · Restricted Project
jaykang10 added inline comments to D99069: [ValueTracking] Handle increasing mul recurrence in isKnownNonZero().
Tue, Mar 23, 3:13 AM · Restricted Project
jaykang10 added a comment to D99069: [ValueTracking] Handle increasing mul recurrence in isKnownNonZero().

This looks good to me, but I think needs a couple more tests:

  • Positive test for nuw instead of nsw.
  • Negative test for zero start value.
  • Negative test for negative multiply. (We can't really test zero multiply, as it will get folded away.)
  • Negative test for negative start value -- alternatively, you can slightly extend this patch and split the start value check between add and mul: Add requires strictly positive start, but for mul non-zero is sufficient, I believe.
Tue, Mar 23, 3:11 AM · Restricted Project
jaykang10 closed D97949: [AArch64] Add missing intrinsics for vrnd.
Tue, Mar 23, 1:20 AM · Restricted Project

Mon, Mar 22

jaykang10 updated the diff for D99069: [ValueTracking] Handle increasing mul recurrence in isKnownNonZero().

Following comments of @nikic, updated code.

Mon, Mar 22, 11:56 AM · Restricted Project
jaykang10 added inline comments to D99069: [ValueTracking] Handle increasing mul recurrence in isKnownNonZero().
Mon, Mar 22, 11:53 AM · Restricted Project
jaykang10 added a comment to D99069: [ValueTracking] Handle increasing mul recurrence in isKnownNonZero().

It might make sense to rephrase this code in terms of matchSimpleRecurrence().

Ah! Thanks for letting me know. I did not know that. Let me try.

Mon, Mar 22, 9:14 AM · Restricted Project
jaykang10 updated the diff for D99069: [ValueTracking] Handle increasing mul recurrence in isKnownNonZero().

Use matchSimpleRecurrence

Mon, Mar 22, 5:31 AM · Restricted Project
jaykang10 added a comment to D99069: [ValueTracking] Handle increasing mul recurrence in isKnownNonZero().

It might make sense to rephrase this code in terms of matchSimpleRecurrence().

Mon, Mar 22, 4:21 AM · Restricted Project
jaykang10 updated the diff for D98422: [ValueTracking] Handle two PHIs in isKnownNonEqual().

Following patch and comment of @nikic, use isKnownNonEqual instead of checking add/mul.

Mon, Mar 22, 4:11 AM · Restricted Project
jaykang10 added a comment to D98422: [ValueTracking] Handle two PHIs in isKnownNonEqual().

This is still trying to handle the phi and the add/mul together. Both of those should be handled separately. I've applied https://github.com/llvm/llvm-project/commit/d11d5d1c5f5a8bafc28be98f43c15a3452abb98b to add the necessary handling for multiplies, and I think with that this should work if you just do the recursive isKnownNonEqual checks without any additional handling.

Mon, Mar 22, 4:08 AM · Restricted Project
jaykang10 requested review of D99069: [ValueTracking] Handle increasing mul recurrence in isKnownNonZero().
Mon, Mar 22, 4:07 AM · Restricted Project

Fri, Mar 19

jaykang10 updated the diff for D98422: [ValueTracking] Handle two PHIs in isKnownNonEqual().

Fixed a bug.

  • During optimization, there are unreachable blocks with PHI and they have no predecessors. Do not handle it.
Fri, Mar 19, 11:35 AM · Restricted Project
jaykang10 updated the diff for D98422: [ValueTracking] Handle two PHIs in isKnownNonEqual().

Fixed a bug

Fri, Mar 19, 8:30 AM · Restricted Project
jaykang10 added a comment to D98422: [ValueTracking] Handle two PHIs in isKnownNonEqual().

@nikic @lebedev.ri I have updated the patch for basic cases. If you feel something weird or inefficient, please let me know.

Fri, Mar 19, 7:06 AM · Restricted Project
jaykang10 updated the diff for D98422: [ValueTracking] Handle two PHIs in isKnownNonEqual().

Cut down this patch to basic case.

Fri, Mar 19, 7:04 AM · Restricted Project

Thu, Mar 18

jaykang10 added a comment to D98422: [ValueTracking] Handle two PHIs in isKnownNonEqual().

I think the main problem here is that the review tries doing too much. Yes, the BasicAA test case can be handled as a combination of simple phi handling plus some additional mul handling. But from the code comments and discussion the intention is to also handle some kind of additional case involving add recurrences.

It would be great if this patch can be cut down to just basic phi-phi handling, with dedicated ValueTracking tests -- even if that does not cover the motivational case in isolation.

As @reames has mentioned before, the code should also only recurse into one pair of operands, and require trivially non-equal constants for the rest. At least we have been following this rule in the rest of the isKnownNonEqual code.

Thu, Mar 18, 1:56 PM · Restricted Project
jaykang10 added a comment to D98422: [ValueTracking] Handle two PHIs in isKnownNonEqual().

I don't understand what the problem here is?

Sure, it's not great that we need to write yet another bicycle,
but it seems to me that the "only" change isKnownNonEqual() needs to support the motivational case
is to implement a variation of isAddOfNonZero() but for multiplications.

Thu, Mar 18, 1:42 PM · Restricted Project
jaykang10 added a comment to D98422: [ValueTracking] Handle two PHIs in isKnownNonEqual().

@jaykang10 Using SCEV in BasicAA is not possible (SCEVAA uses SCEV).

Thu, Mar 18, 12:24 PM · Restricted Project
jaykang10 added a comment to D98422: [ValueTracking] Handle two PHIs in isKnownNonEqual().

I think we could check the IV of loop roughly as below.

 if (Q.DT != nullptr && Q.DT->dominates(PN1->getParent(), Pred)) {
   const SCEV *IVSCEV = nullptr;
   if (PN1 == IV2) 
     IVSCEV = SE.getSCEV(IV2);
   else if (PN2 == IV1) {
     IVSCEV = SE.getSCEV(IV1);

   if (isa<SCEVAddRecExpr>(IVSCEV))
     continue;
}

In this case, we could say the IV is not same among iterations... but it needs ScalarEvolution... @nikic @reames How do you think about it? I wonder whether it is acceptable to add ScalarEvolution to BasicAA and ValueTracking or not...

Thu, Mar 18, 9:26 AM · Restricted Project

Wed, Mar 17

jaykang10 added a comment to D98422: [ValueTracking] Handle two PHIs in isKnownNonEqual().

I think we could check the IV of loop roughly as below.

Wed, Mar 17, 4:07 PM · Restricted Project

Tue, Mar 16

jaykang10 updated the diff for D98422: [ValueTracking] Handle two PHIs in isKnownNonEqual().

Updated code following comments of @nikic

Tue, Mar 16, 2:29 AM · Restricted Project

Mar 15 2021

jaykang10 added inline comments to D98422: [ValueTracking] Handle two PHIs in isKnownNonEqual().
Mar 15 2021, 3:12 PM · Restricted Project
jaykang10 added inline comments to D98422: [ValueTracking] Handle two PHIs in isKnownNonEqual().
Mar 15 2021, 2:45 PM · Restricted Project
jaykang10 added inline comments to D98422: [ValueTracking] Handle two PHIs in isKnownNonEqual().
Mar 15 2021, 12:06 PM · Restricted Project
jaykang10 added inline comments to D98422: [ValueTracking] Handle two PHIs in isKnownNonEqual().
Mar 15 2021, 10:44 AM · Restricted Project
jaykang10 added a comment to D98422: [ValueTracking] Handle two PHIs in isKnownNonEqual().

Following @lebedev.ri comments, moved the change to a new helper function and updated redundant code.

Did you upload the right diff? I think not all inline comments were addressed.

Mar 15 2021, 10:43 AM · Restricted Project
jaykang10 updated the diff for D98422: [ValueTracking] Handle two PHIs in isKnownNonEqual().

Following @lebedev.ri comments, moved the change to a new helper function and updated redundant code.

Mar 15 2021, 7:14 AM · Restricted Project