i915 intel_frontbuffer_put panic, patch for review/testing

[Pete Wright]

On 1/28/21 5:49 PM, Ed Maste wrote:
> For the last few months I have been experiencing occasional panics
> from intel_frontbuffer_put(), while running OBS-studio. An excerpt
> from the backtrace:
>
> intel_frontbuffer_put()
> intel_user_framebuffer_destroy()
> _drm atomic_helper_plane_destroy_state()
> drm_atomic_helper_plane_destroy_state()
> drm_atomic_state_default_clear()
> intel_atomic_state_clear()
> __drm_atomic_state_free()
> intel_atomic_cleanup_work()
> linux_work_fn()
>
> I found a freedesktop.org bug report for an issue that looked
> related[1]. Unfortunately the patch attached there does not apply to
> Linux 5.4 corresponding to what we have in drm-kmod[2].
>
> I've backported the patch and applied it against the 5.4-lts branch in
> the drm-kmod repo, and the issue has not recurred. My patch is in a
> pull request at [3], and available as a patch file at [4]. Review
> and/or testing is very much appreciated.
I would be happy to test on my laptop running 13, is there an easy way 
to download a branch and have it build using the ports infrastructure?

  I believe ports using github for distfiles expect a tag not branch.  i 
can test out building it locally outside of the ports tree, but wanted 
to see if i could be lazy first :)

-pete

-- 
Pete Wright
pete at nomadlogic.org
@nomadlogicLA