[Intel-gfx] Potential BUG in drm/i915/execlists: Reset RING registers upon resume

[Jani Nikula]

On Wed, 12 Apr 2017, Greg Kroah-Hartman <gregkh at linuxfoundation.org> wrote:
> On Mon, Mar 20, 2017 at 11:24:38AM -0400, Eric Blau wrote:
>> On Mon, Mar 20, 2017 at 11:13 AM, Greg Kroah-Hartman
>> <gregkh at linuxfoundation.org> wrote:
>> > On Mon, Mar 20, 2017 at 05:01:34PM +0200, Jani Nikula wrote:
>> >> On Tue, 14 Mar 2017, Eric Blau <eblau at eblau.com> wrote:
>> >> > That's funny. I have a MacBook Pro 12,1 from late 2015. Hibernate
>> >> > failed for me in 4.9.6 through 4.9.8 (possibly earlier as well, I do
>> >> > no recall) without the patch. The patch you reference fixed my problem
>> >> > and apparently many others based on the bug reports:
>> >> >
>> >> > https://bugs.freedesktop.org/show_bug.cgi?id=96526
>> >> >
>> >> > I applied the patch myself in 4.9.6 through 4.9.8 and hibernate works
>> >> > for me. I have run vanilla 4.9.9 and 4.10.1 and in both hibernate
>> >> > works.
>> >>
>> >> So I'm quite surprised
>> >>
>> >> commit f2a0409a08502d64fbe3990354dff5902b08d2fb
>> >> Author: Chris Wilson <chris at chris-wilson.co.uk>
>> >> Date:   Wed Sep 21 14:51:08 2016 +0100
>> >>
>> >>     drm/i915/execlists: Reset RING registers upon resume
>> >>
>> >>     commit bafb2f7d4755bf1571bd5e9a03b97f3fc4fe69ae upstream.
>> >>
>> >> ended up in v4.9.9 considering that the upstream commit is not annotated
>> >> for stable, it has no Fixes: tags, and I can't find any backport
>> >> requests or even notifications for it in my mails. (Admittedly I'm not
>> >> subscribed on stable@, but I'd expect our lists, maintainers or
>> >> developers be Cc'd.)
>> >
>> > Eric sent this to the stable list a few times, my mistake for not seeing
>> > he hadn't also cc:ed the developer list.
>> >
>> > Want me to revert it?
>> >
>> > thanks,
>> >
>> > greg k-h
>> 
>> My apologies. I'm new to requesting stable patches, but I thought
>> sending to the stable mailing lists would keep the proper folks in the
>> loop. Sorry about that.
>> 
>> The bug that the commit fixes has quite a long documentation trail:
>> 
>> [BAT execlists] Sporadic - gem_exec_suspend basic-s4 GPU hang after resume
>> https://bugs.freedesktop.org/show_bug.cgi?id=96526
>> 
>> I've had problems with that one and this one that is not fixed in 4.10.x stable:
>> 
>> [Regression BDW] kernel panic in Intel i915 module, complete system
>> freeze in 4.10-rc2
>> https://bugs.freedesktop.org/show_bug.cgi?id=99295
>
> So, what needs to be done here?
>
> Do I need to revert something in 4.9? Add something?  Do something else?
>
> totally confused,

I did reply to you [1]:

>> Want me to revert it?
>
> Chris says that would be the correct thing to do. I'm no expert in the
> area, but IIUC having just that commit backported is not enough,
> something more would be needed, but we're not sure yet what exactly and
> if those dependencies can easily be backported.

"it" being commit f2a0409a0850 ("drm/i915/execlists: Reset RING
registers upon resume") in 4.9.9+.

BR,
Jani.


[1] http://mid.mail-archive.com/87a88gar9c.fsf@intel.com

-- 
Jani Nikula, Intel Open Source Technology Center