On Tue, Oct 05, 2021 at 12:49:53PM -0000, Patchwork wrote: > == Series Details == > > Series: drm/i915: Improve DP link training further (rev4) > URL : https://patchwork.freedesktop.org/series/95405/ > State : failure > > == Summary == > > CI Bug Log - changes from CI_DRM_10683 -> Patchwork_21247 > ==================================================== > > Summary > ------- > > **FAILURE** > > Serious unknown changes coming with Patchwork_21247 absolutely need to be > verified manually. > > If you think the reported changes have nothing to do with the changes > introduced in Patchwork_21247, 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/Patchwork_21247/index.html > > Possible new issues > ------------------- > > Here are the unknown changes that may have been introduced in Patchwork_21247: > > ### IGT changes ### > > #### Possible regressions #### > > * igt@core_hotunplug@unbind-rebind: > - fi-tgl-u2: NOTRUN -> [INCOMPLETE][1] > [1]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_21247/fi-tgl-u2/igt@core_hotunplug@xxxxxxxxxxxxxxxxxx That looks like a straight up race during unbind. We turn off interrupts first, only then we cancel the hpd stuff (including the modeset_retry work) which presuambly was already running and told fb_helper that stuff happened, then the fb_helper proceeds to do a modeset while interrupts are already off and we're in the middle of tearing down the driver, and all hell breaks loose as a result. Unrelated to these patches, I *think*. I'll hit retest anyway to make sure. -- Ville Syrjälä Intel
