[PATCH] D52125: X86, AArch64, ARM: Do not attach debug location to spill/reload instructions

[Matt Morehouse via llvm-commits]

Bot went green after revert:
http://lab.llvm.org:8011/builders/sanitizer-x86_64-linux-android/builds/15666

+Evgenii Stepanov <eugenis at google.com> for debugging advice.

On Tue, Oct 2, 2018 at 11:55 AM Matthias Braun via llvm-commits <
llvm-commits at lists.llvm.org> wrote:

> Yes, it seems only some of the failures where GlobalISel related and some
> are still around.
>
> Trying to understand what's going on here, since I suspect I am not able
> to run the hwasan tests myself (or am I when I am in an apple universe?):
>
> It seems that the tool prints every memory read and write. I suspect some
> of the reads and writes that get picked up are actually spills and reloads
> which after my change have slightly different debug locations attached
> (they really shouldn't have any debug information attached, but now happen
> to pick up the location of whatever is before the spill/reload instead of
> whatever is after). If those tests are indeed accidentally picking up
> spills/reloads instead of the load/stores they want to be testing for, then
> how can we fix this?
>
> > On Oct 2, 2018, at 11:36 AM, Daniel Sanders <daniel_l_sanders at apple.com>
> wrote:
> >
> > I reverted the GlobalISel change (r343521) yesterday so it shouldn't be
> causing any further failures. I only see hwasan failures on the latest
> build.
> >
> >> On 2 Oct 2018, at 11:19, Matthias Braun <matze at braunis.de> wrote:
> >>
> >> +CC dsanders
> >>
> >>> On Oct 2, 2018, at 11:17 AM, Matthias Braun <matze at braunis.de> wrote:
> >>>
> >>> Feel free to revert for now, but are you sure it is actually this
> commit?
> >>>
> >>> It's hard to say for sure what happenes with the earlier hwasan tests,
> but the later tests that failed are crashing in GlobalISel related code,
> which should all run before the code that I changed...
> >>>
> >>>
> >>>> On Oct 2, 2018, at 11:06 AM, Matt Morehouse via Phabricator via
> llvm-commits <llvm-commits at lists.llvm.org> wrote:
> >>>>
> >>>> morehouse added a comment.
> >>>>
> >>>> This change is breaking some symbolization tests on the Android bot:
> http://lab.llvm.org:8011/builders/sanitizer-x86_64-linux-android/builds/15630
> >>>>
> >>>> Please fix or revert.
> >>>>
> >>>>
> >>>> Repository:
> >>>> rL LLVM
> >>>>
> >>>> https://reviews.llvm.org/D52125
> >>>>
> >>>>
> >>>>
> >>>> _______________________________________________
> >>>> llvm-commits mailing list
> >>>> llvm-commits at lists.llvm.org
> >>>> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-commits
> >>>
> >>
> >
>
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at lists.llvm.org
> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-commits
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20181002/10934904/attachment.html>