Page MenuHomePhabricator

indutny (Fedor Indutny)
User

Projects

User does not belong to any projects.

User Details

User Since
Feb 23 2018, 2:01 PM (85 w, 5 d)

Recent Activity

Mar 3 2018

indutny committed rL326668: [CallSiteSplitting] fix use after-free.
[CallSiteSplitting] fix use after-free
Mar 3 2018, 2:37 PM
indutny added a comment to D43729: [CallSiteSplitting] properly split musttail calls.

Thank you, everyone!

Mar 3 2018, 1:43 PM
indutny committed rL326666: [CallSiteSplitting] properly split musttail calls.
[CallSiteSplitting] properly split musttail calls
Mar 3 2018, 1:42 PM
indutny closed D43729: [CallSiteSplitting] properly split musttail calls.
Mar 3 2018, 1:42 PM

Mar 2 2018

indutny updated the diff for D43729: [CallSiteSplitting] properly split musttail calls.

Address feedback.

Mar 2 2018, 11:05 AM
indutny added inline comments to D43729: [CallSiteSplitting] properly split musttail calls.
Mar 2 2018, 10:21 AM
indutny updated the diff for D43729: [CallSiteSplitting] properly split musttail calls.

Address feedback.

Mar 2 2018, 10:21 AM

Mar 1 2018

indutny updated the diff for D43729: [CallSiteSplitting] properly split musttail calls.

Proper fixup for pending clone issue.

Mar 1 2018, 10:29 PM
indutny added a comment to D43926: [ArgumentPromotion] don't break musttail invariant PR36543.

Thank you!

Mar 1 2018, 5:03 PM
indutny committed rL326521: [ArgumentPromotion] don't break musttail invariant PR36543.
[ArgumentPromotion] don't break musttail invariant PR36543
Mar 1 2018, 5:01 PM
indutny closed D43926: [ArgumentPromotion] don't break musttail invariant PR36543.
Mar 1 2018, 5:01 PM
indutny added inline comments to D43729: [CallSiteSplitting] properly split musttail calls.
Mar 1 2018, 12:38 PM

Feb 28 2018

indutny added a comment to D43926: [ArgumentPromotion] don't break musttail invariant PR36543.

Decided you might be interested in this patch since you subscribed to the bugzilla bug. Hopefully this isn't bothering you.

Feb 28 2018, 8:10 PM
indutny added a reviewer for D43926: [ArgumentPromotion] don't break musttail invariant PR36543: fhahn.
Feb 28 2018, 8:08 PM
indutny created D43926: [ArgumentPromotion] don't break musttail invariant PR36543.
Feb 28 2018, 8:04 PM
indutny updated the diff for D43923: Add myself to CREDITS.txt.

Binary insertion of my surname into the list.

Feb 28 2018, 6:38 PM
indutny updated the summary of D43923: Add myself to CREDITS.txt.
Feb 28 2018, 6:38 PM
indutny created D43923: Add myself to CREDITS.txt.
Feb 28 2018, 6:30 PM
indutny added a comment to D43708: [DAE] don't remove args of musttail target/caller.

Thank you, everyone!

Feb 28 2018, 6:22 PM
indutny added a comment to D43695: [IPSCCP] do not break musttail invariant (PR36485).

Thank you, everyone!

Feb 28 2018, 6:21 PM
indutny updated the summary of D43729: [CallSiteSplitting] properly split musttail calls.
Feb 28 2018, 3:44 PM
indutny added a comment to D43729: [CallSiteSplitting] properly split musttail calls.

There's a huge FIXME in the code that depends on @fhahn's patch: https://reviews.llvm.org/D43822 . Will remove it as soon as that patch lands. Hopefully it won't deter you from reviewing it ;)

Feb 28 2018, 3:29 PM
indutny updated the diff for D43729: [CallSiteSplitting] properly split musttail calls.

Address feedback.

Feb 28 2018, 3:29 PM
indutny updated the diff for D43859: [GlobalOpt] don't change CC of musttail calle(e|r) PR36546.

Fix test.

Feb 28 2018, 2:25 PM
indutny updated the diff for D43859: [GlobalOpt] don't change CC of musttail calle(e|r) PR36546.

Added FIXME, as requested.

Feb 28 2018, 2:03 PM
indutny added a comment to D43729: [CallSiteSplitting] properly split musttail calls.

Sounds like a plan! Will work on it today. Thank you, guys!

Feb 28 2018, 1:47 PM
indutny added a comment to D43708: [DAE] don't remove args of musttail target/caller.

@rnk not sure if it matters, but I don't have commit access yet. If you could land the change - I'd appreciate it a lot. Thank you!

Feb 28 2018, 1:13 PM
indutny added a comment to D43695: [IPSCCP] do not break musttail invariant (PR36485).

I don't have commit access yet. Would greatly appreciate if you could land it for me. Thank you!

Feb 28 2018, 1:13 PM
indutny added a comment to D43695: [IPSCCP] do not break musttail invariant (PR36485).

Thank you!

Feb 28 2018, 12:50 PM
indutny added a comment to D43729: [CallSiteSplitting] properly split musttail calls.

Thank you for taking a look at it. I'll be glad to make the changes. Just one question before I start: it seems that the suggestions of @junbuml and you are to some extent contradictory. At first I was suggested to leave the function as it is, and now the suggestion is to move things back to it. Given this, what do we actually desire it to look like? Let's decide on this before making any additional changes.

Feb 28 2018, 12:39 PM
indutny updated the diff for D43859: [GlobalOpt] don't change CC of musttail calle(e|r) PR36546.

Address inline comments.

Feb 28 2018, 12:30 PM

Feb 27 2018

indutny created D43859: [GlobalOpt] don't change CC of musttail calle(e|r) PR36546.
Feb 27 2018, 10:50 PM
indutny updated the diff for D43729: [CallSiteSplitting] properly split musttail calls.

Rename function, and remove extra debug calls.

Feb 27 2018, 1:43 PM
indutny retitled D43695: [IPSCCP] do not break musttail invariant (PR36485) from [IPSCCP] mark musttail calls overdefined (PR36485) to [IPSCCP] do not break musttail invariant (PR36485).
Feb 27 2018, 12:31 PM
indutny updated the diff for D43695: [IPSCCP] do not break musttail invariant (PR36485).

I've addressed your feedback, and covered more erroneous cases in my patch now.

Feb 27 2018, 12:27 PM
indutny added a comment to D43708: [DAE] don't remove args of musttail target/caller.

Thank you for review!

Feb 27 2018, 10:42 AM
indutny updated the diff for D43708: [DAE] don't remove args of musttail target/caller.

Add FIXME.

Feb 27 2018, 10:42 AM
indutny updated the diff for D43729: [CallSiteSplitting] properly split musttail calls.

Address feedback.

Feb 27 2018, 10:39 AM

Feb 26 2018

indutny added a comment to D43695: [IPSCCP] do not break musttail invariant (PR36485).

I think I've "under-implemented" the fix here. First of all, it doesn't mark callees overdefined, which leads to zapped return value. Second, it basically kills whole optimization because of musttail call site.

Feb 26 2018, 9:52 PM
indutny updated the diff for D43708: [DAE] don't remove args of musttail target/caller.

Add extra check to the test.

Feb 26 2018, 5:56 PM
indutny added inline comments to D43708: [DAE] don't remove args of musttail target/caller.
Feb 26 2018, 5:53 PM
indutny added inline comments to D43708: [DAE] don't remove args of musttail target/caller.
Feb 26 2018, 5:29 PM
indutny added a comment to D43729: [CallSiteSplitting] properly split musttail calls.

Just wanted to say thank you for giving a thorough review! I appreciate it!

Feb 26 2018, 2:35 PM
indutny updated the diff for D43729: [CallSiteSplitting] properly split musttail calls.

Address review comments.

Feb 26 2018, 2:29 PM
indutny updated the diff for D43708: [DAE] don't remove args of musttail target/caller.

Added explanation to the test.

Feb 26 2018, 12:10 PM
indutny added inline comments to D43708: [DAE] don't remove args of musttail target/caller.
Feb 26 2018, 12:04 PM

Feb 24 2018

indutny updated the diff for D43729: [CallSiteSplitting] properly split musttail calls.

Fix non-musttail calls.

Feb 24 2018, 12:08 PM
indutny updated the diff for D43729: [CallSiteSplitting] properly split musttail calls.

Proper implementation!

Feb 24 2018, 12:02 PM
indutny updated the diff for D43729: [CallSiteSplitting] properly split musttail calls.

Move condition to proper function.

Feb 24 2018, 11:28 AM

Feb 23 2018

indutny added a comment to D43729: [CallSiteSplitting] properly split musttail calls.

A question: what is backport policy of LLVM project? I'd love to see this fix in 5.0.0 .

Feb 23 2018, 11:12 PM
indutny created D43729: [CallSiteSplitting] properly split musttail calls.
Feb 23 2018, 11:09 PM
indutny updated the diff for D43708: [DAE] don't remove args of musttail target/caller.

Use getTerminatingMustTailCall instead of iterating through all instructions.

Feb 23 2018, 7:39 PM
indutny updated the diff for D43695: [IPSCCP] do not break musttail invariant (PR36485).

Removed FileCheck from the test. It's enough if it doesn't crash, there's nothing to validate.

Feb 23 2018, 6:20 PM
indutny added a comment to D43708: [DAE] don't remove args of musttail target/caller.

A question: what is backport policy of LLVM project? I'd love to see this fix in 5.0.0 .

Feb 23 2018, 6:07 PM
indutny added a comment to D43695: [IPSCCP] do not break musttail invariant (PR36485).

A question: what is backport policy of LLVM project? I'd love to see this fix in 5.0.0 .

Feb 23 2018, 6:07 PM
indutny created D43708: [DAE] don't remove args of musttail target/caller.
Feb 23 2018, 6:03 PM
indutny updated the diff for D43695: [IPSCCP] do not break musttail invariant (PR36485).

Move test file into a proper directory.

Feb 23 2018, 3:07 PM
indutny created D43695: [IPSCCP] do not break musttail invariant (PR36485).
Feb 23 2018, 2:11 PM