[llvm] r329287 - ARM: Do not spill CSR to stack on entry to noreturn functions

[Vitaly Buka via llvm-commits]

9355 was  not green
http://lab.llvm.org:8011/builders/sanitizer-x86_64-linux-android/builds/9355

The bot was never green after that patch.
Following are just my manual reruns on different revisions.
Apr 05 17:47 329365 failure #9374
<http://lab.llvm.org:8011/builders/sanitizer-x86_64-linux-android/builds/9374>
Failed
annotate failed run lit tests [arm/aosp_marlin-userdebug/opr4.170623.016]
Apr 05 17:19 329365 success #9373
<http://lab.llvm.org:8011/builders/sanitizer-x86_64-linux-android/builds/9373>
Build
successful
Apr 05 16:56 329363 failure #9372
<http://lab.llvm.org:8011/builders/sanitizer-x86_64-linux-android/builds/9372>
Failed
annotate failed run lit tests [arm/aosp_marlin-userdebug/opr4.170623.016]
Apr 05 16:28 329363 success #9371
<http://lab.llvm.org:8011/builders/sanitizer-x86_64-linux-android/builds/9371>
Build
successful
Apr 05 15:58 329359 success #9370
<http://lab.llvm.org:8011/builders/sanitizer-x86_64-linux-android/builds/9370>
Build
successful
Any see:

http://lab.llvm.org:8011/builders/sanitizer-x86_64-linux-android/builds/9372
(on rev 329287) is RED
http://lab.llvm.org:8011/builders/sanitizer-x86_64-linux-android/builds/9371
(on rev 329286) is GREEN

So I am 100% sure that this is r329287


On Fri, Apr 6, 2018 at 8:26 AM Tim Northover <t.p.northover at gmail.com>
wrote:

> Well, I eventually managed to reproduce it but then noticed 2 things:
>
> 1. The bot magically fixed itself after what looks like a list of
> unrelated commits (first green build after the failure is 9355).
> 2. Reverting the patch locally didn't fix the issue for me.
>
> So I don't know what's going on but until I see firm evidence
> otherwise I'm going to assume it wasn't this patch.
>
> Cheers.
>
> Tim.
>
> On 6 April 2018 at 07:55, Tim Northover <t.p.northover at gmail.com> wrote:
> > Hi Vitaly,
> >
> > On 6 April 2018 at 01:38, Vitaly Buka via llvm-commits
> > <llvm-commits at lists.llvm.org> wrote:
> >> This patch breaks UBSAN tests on Android
> >
> > If I squint a bit I can sort of see how this patch might cause a
> > failure like that, but I reverted it locally and saw no CodeGen change
> > in the failing file. I'll try to reproduce it locally today, but my
> > environments will be pretty approximate (iOS and Debian on ARM). Do
> > you have access to get artefacts off the builder if that fails?
> >
> > Cheers.
> >
> > Tim.
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20180406/10a85817/attachment.html>