User Details
- User Since
- Nov 19 2013, 9:02 PM (374 w, 4 d)
Fri, Jan 22
Thank you for the great work that makes this default. LGTM.
Tue, Jan 12
rebase
rebase
Address comments
ping?
Thu, Jan 7
LGTM. I still think we should get back to the legacy names so that translation is avoided.
Tue, Jan 5
LGTM
Wed, Dec 30
Tue, Dec 29
Mon, Dec 28
- Remove obsolete TODO
LGTM
ping?
LGTM
Dec 22 2020
LGTM
Dec 21 2020
- Address the laying issue in 31ecf8d29d81d196374a562c6d2bd2c25a62861e
- Move llvm::registerCodeGenCallback declaration to TargetPassConfig.h
- update to the landed version.
I'll update this to the correct/landed version. Sorry for the noise.
LGTM. The next step is to use this in AMDGPU and add tests?
ping?
Dec 16 2020
Not for a concrete bug, but for consistency only. Basically, it makes the module data layout match target data layout before opt&codegen starts under all circumstances (LTO/clang).
Dec 14 2020
@dblaikie The laying issue in 31ecf8d29d81d196374a562c6d2bd2c25a62861e is addressed here. Do you mind taking another look? Thanks!
Dec 9 2020
fix the laying issue.
- change buildCodeGenPipeline signature to take pass manager references as inputs.
Dec 8 2020
Dec 7 2020
- clang/test/Misc/time-passes.c: add -fno-experimental-new-pass-manager
- llvm/test/Other/time-passes.ll: makes sure each pass has one entry.
Dec 4 2020
- fix rebase error
Dec 3 2020
It is very unfortunate that we have to manage and translate between two sets of names (one pass name and one type name). This makes me wonder if we just keep the pass name as the return value of PassInfoMixin::name and get rid of class name everywhere. Right now I couldn't think of anything is blocking us from doing that. WDYT? @asbirlea ?
Dec 2 2020
- rebase
Dec 1 2020
LGTM.
Nov 30 2020
Nov 24 2020
LGTM. Thanks!
Nov 23 2020
Nov 19 2020
FWIW, could we keep naming it ModuleOptimizerEarlyEPCallback? That sounds pretty intuitive to me and reminds users of the old name. Is the renaming for using the prefix Pipeline?
Nov 14 2020
LGTM
Nov 13 2020
Nov 12 2020
- formatting
- address comments: update comment, check output making sure bitcast is there.
This is NFC and looks like a nice cleaup. So LGTM.
Nov 10 2020
Abandoning. It doesn't feel like a trivial fix and I could workaround this offline.
Nov 5 2020
Nov 4 2020
LGTM
Nov 2 2020
- revert fixup in IRMover
- allow CGProfile metadata node representing edge to use bitcast.