[Intel-gfx] ✗ Fi.CI.IGT: failure for tests/core_hotunplug: Fixes and enhancements (rev4)

[Janusz Krzysztofik]

Hi Petri,

On Mon, 2020-08-24 at 12:26 +0300, Petri Latvala wrote:
> On Mon, Aug 24, 2020 at 10:42:10AM +0200, Janusz Krzysztofik wrote:
> > On Fri, 2020-08-21 at 17:40 +0000, Patchwork wrote:
> > > Patch Details
> > > Series:	tests/core_hotunplug: Fixes and enhancements (rev4)
> > > URL:	https://patchwork.freedesktop.org/series/79671/
> > > State:	failure
> > > Details:	https://intel-gfx-ci.01.org/tree/drm-tip/IGTPW_4901/index.html
> > > CI Bug Log - changes from CI_DRM_8913_full -> IGTPW_4901_full
> > > Summary
> > > FAILURE
> > > 
> > > Serious unknown changes coming with IGTPW_4901_full absolutely need to be
> > > verified manually.
> > > 
> > > If you think the reported changes have nothing to do with the changes
> > > introduced in IGTPW_4901_full, 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_4901/index.html
> > > 
> > > Possible new issues
> > > Here are the unknown changes that may have been introduced in IGTPW_4901_full:
> > > 
> > > IGT changes
> > > Possible regressions
> > > {igt at core_hotunplug@hotrebind-lateclose} (NEW):
> > > 
> > > shard-snb: NOTRUN -> FAIL
> > > 
> > > shard-glk: NOTRUN -> FAIL
> > > 
> > > shard-kbl: NOTRUN -> FAIL
> > 
> > This is an existing but formerly not reported GPU hang driver issue
> > exhibited by the test, not a regression.  The issue needs to be fixed
> > in the driver for the test to succeed.  As one can see from CI reports,
> > the test succesfuly recovers from that condition and subsequent tests
> > are not affected.
> >   
> > > {igt at core_hotunplug@hotunbind-lateclose} (NEW):
> > > 
> > > shard-hsw: NOTRUN -> INCOMPLETE +3 similar issues
> > 
> > This is a known driver issue, already reported by igt at device
> > _reset at unbind-reset-rebind.  This test shows clearly that the issue has
> > nothing to do with device reset, only with driver unbind-rebind cycle. 
> > The driver needs to be fixed for the kernel warning not be triggered
> > and the tests succeed.
> 
> Is there an ETA for the driver fix?

Not that I know of.  I'll have a look at it.

Thanks,
Janusz

> 
> > I think the *bind* subtests in their current shape are perfectly ready
> > for inclusion in CI runs.
> 
> Agreed otherwise, except for the known incomplete. Introducing new
> incompletes without a fix in sight is real bad.
> 
>