[Bug 98690] [SKL bisected] System freeze when starting X using kernel 4.9-rc1 or later

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=98690

--- Comment #24 from Peter Ujfalusi <peter.ujfalusi at gmail.com> ---
(In reply to Paul Donohue from comment #23)
> Is there anything simple we can do to find the relevant code path?  Maybe
> using SystemTap?

The only scenario which looks broken is when drm_kms_helper_poll_init() is
called after drm_helper_probe_single_connector_modes(). I don't see any other
path.

So the interesting thing is what happens during the 1s after we scheduled the
work in drm_kms_helper_poll_enable_locked() because the
dev->mode_config.delayed_event was true. Between we schedule the work and the
work is actually run.
Then compare the events when we don't have the 1s delay when scheduling the
work to handle the delayed event.

The fact the nouveau tries to disable the poll right after it initialized it
means that it does not want it to execute just now. This is why I was surprised
that the second patch did not worked as it did prevented the work to be
scheduled when the poll_init() was called. Later on a poll enable will pick up
the delayed event and it will schedule the work right away, but this looks to
be not quite right for Optimus.

I'm a bit puzzled.

Péter

-- 
You are receiving this mail because:
You are on the CC list for the bug.
You are the assignee for the bug.
You are the QA Contact for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/intel-gfx-bugs/attachments/20170104/f7068a16/attachment.html>