- User Since
- Nov 23 2015, 1:58 PM (121 w, 2 d)
Tue, Mar 20
Mon, Mar 19
This looks good. Can you convert your example into a test case?
Sat, Mar 17
Fri, Mar 16
This looks like a nice improvement modulo a few issues:
Wed, Mar 14
Is there any chance of the right shift needing this case as well?
Fri, Mar 9
Tweak negation as suggested by James.
Looks overall reasonable. Is there any performance impact of EnforceNodeIdInvariant?
Cleanup by folding EnforceNodeIdInvariant into ISel's replace functions
Fri, Mar 2
Change to a global fix for all NodeId ordering violation.
Tue, Feb 27
This is the last patch preventing the relanding of https://reviews.llvm.org/D41293.
Mon, Feb 26
As verifying topological order is preserved is not practical, add EXPENSIVE_CHECK checking that pruning doesn't cause use to miss finding a predecessor.
Thu, Feb 22
Wed, Feb 21
Rephrase comments to be more clear and use less conservative node id. The comments now reference pictoral comments from D43154.
Tue, Feb 20
Address James' comment and test case requests.
Feb 13 2018
Remove unrelated format changes.
Feb 12 2018
Fix PR number in test and commit message.
Add a missing check between ST and Zn nodes.
Feb 9 2018
Already commited in rL324359
Feb 7 2018
Feb 6 2018
Feb 5 2018
Address cleanup comments.
Feb 3 2018
Related question: Should the check be TLI.isLoadExtLegalOrCustom?
Feb 1 2018
This was landed in rL323539.
Jan 31 2018
LGTM modulo comment nit.
This looks good modulo test cases.
Failed to commit typo fix (N vs. M) in last rebase.
Jan 30 2018
Update because of PR35316. Also tweak findNonImmUse to use hasPredecessorHelper like other functions.
Jan 29 2018
You're missing the optsize test case but otherwise LGTM.
address reviewer comments
Okay. This looks good. Just one more quick sanity check: Is this patch rebased off of D42615 or without it? I just want to makre sure that we've not missed any test changes.
Jan 26 2018
This seems reasonable to me. The test outputs all look better or more optimizable (Have you looked at why the ands aren't being colleasced into a vector?)
Jan 25 2018
Turn back on nodeid-based pruning on X86 Target.
Jan 24 2018
Jan 23 2018
Found bug when ImmedUse has non-direct uses of Def. Added test and fixup.
Jan 22 2018
Address RKSimon's comments and add trivial early exit check in findNonImmUse.
Jan 19 2018
Jan 18 2018
Unless I'm crazy the entirety of addcarry2.ll already passes on trunk. I think we hit regular isel patterns unless the carry out of the ADC/SBB is used.
Jan 17 2018
Add ADC immediate tests. I haven't worked out a way to the appropriate SUB immediate nodes via LLVM IR as they're realized via add.
Jan 16 2018
Rebasing past D42128.
Jan 12 2018
In testing I discovered the follow on patch to X86's load-op-store pattern check needed to be folded in. Added discovered test as well.
Jan 11 2018
Jan 10 2018
Address comments and rebase.
Jan 9 2018
Jan 8 2018
Jan 5 2018
Correct failing on non-constant indices.
Jan 3 2018
Rebase and fix NodeId pruning to also ignore 0 ids (from legalization) and turn pruning on for checks only
Factored out the indexed memory operation correctness check into D417101.