[igt-dev] ✓ Fi.CI.BAT: success for runner: Make sure output is still collected when killing test due to taint

[Petri Latvala]

On Wed, Jan 29, 2020 at 01:28:04PM +0000, Chris Wilson wrote:
> Quoting Petri Latvala (2020-01-29 13:24:49)
> > On Tue, Jan 28, 2020 at 03:56:11PM +0000, Chris Wilson wrote:
> > > Quoting Patchwork (2020-01-28 15:30:42)
> > > > == Series Details ==
> > > > 
> > > > Series: runner: Make sure output is still collected when killing test due to taint
> > > > URL   : https://patchwork.freedesktop.org/series/72657/
> > > > State : success
> > > > 
> > > > == Summary ==
> > > > 
> > > > CI Bug Log - changes from IGT_5394 -> IGTPW_4016
> > > > ====================================================
> > > > 
> > > > Summary
> > > > -------
> > > > 
> > > >   **WARNING**
> > > > 
> > > >   Minor unknown changes coming with IGTPW_4016 need to be verified
> > > >   manually.
> > > >   
> > > >   If you think the reported changes have nothing to do with the changes
> > > >   introduced in IGTPW_4016, please notify your bug team to allow them
> > > >   to document this new failure mode, which will reduce false positives in CI.
> > > > 
> > > >   External URL: https://intel-gfx-ci.01.org/tree/drm-tip/IGTPW_4016/index.html
> > > > 
> > > > Possible new issues
> > > > -------------------
> > > > 
> > > >   Here are the unknown changes that may have been introduced in IGTPW_4016:
> > > > 
> > > > ### IGT changes ###
> > > > 
> > > > #### Warnings ####
> > > > 
> > > >   * igt at gem_exec_suspend@basic-s3:
> > > >     - fi-cml-s:           [FAIL][1] ([fdo#103375]) -> [TIMEOUT][2]
> > > >    [1]: https://intel-gfx-ci.01.org/tree/drm-tip/IGT_5394/fi-cml-s/igt@gem_exec_suspend@basic-s3.html
> > > >    [2]: https://intel-gfx-ci.01.org/tree/drm-tip/IGTPW_4016/fi-cml-s/igt@gem_exec_suspend@basic-s3.html
> > > 
> > > Incomplete or Timeout?
> > > 
> > > Starting subtest: basic-S3
> > > [160.045053] Killing the test because the kernel is tainted.
> > > [160.119964] Aborting: Kernel badly tainted (0x240) (check dmesg for details):
> > >       (0x200) TAINT_WARN: WARN_ON has happened.
> > > 
> > > I'd rather have stern purple than whimsical blue.
> > 
> > I can probably make that happen and it's almost going to be easy.
> > 
> > Do you mind if I do it in a followup patch though? I'd like to get
> > this in, the questions about incomplete output are increasing every
> > day...
> 
> Sure. Could we also ticking down earlier i.e. periodic taint checking?


A possibility, yes.


-- 
Petri Latvala