Re: Linux regressions report for mainline [2022-05-09] (was: Linux 5.18-rc6)

From: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
To: Hans de Goede <hdegoede@redhat.com>,
	Linus Torvalds <torvalds@linux-foundation.org>,
	"Regzbot (on behalf of Thorsten Leemhuis)" 
	<regressions@leemhuis.info>,
	Jani Nikula <jani.nikula@linux.intel.com>,
	Matthew Auld <matthew.auld@intel.com>,
	Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: LKML <linux-kernel@vger.kernel.org>,
	Linux regressions mailing list <regressions@lists.linux.dev>
Subject: Re: Linux regressions report for mainline [2022-05-09] (was: Linux 5.18-rc6)
Date: Tue, 10 May 2022 16:41:00 +0200	[thread overview]
Message-ID: <12bebc80-b792-5739-0d9d-ef1466a9b2aa@linux.intel.com> (raw)
In-Reply-To: <54664f6a-b046-1330-e794-cb533e942a94@redhat.com>

Op 09-05-2022 om 21:04 schreef Hans de Goede:
> Hi,
>
> On 5/9/22 19:20, Linus Torvalds wrote:
>> On Mon, May 9, 2022 at 3:47 AM Regzbot (on behalf of Thorsten
>> Leemhuis) <regressions@leemhuis.info> wrote:
>>> Hi Linus! Here's a quick compilation of open reports about regressions in
>>> 5.18-rc that I'm currently aware of; most of the reports are quite
>>> recent and there afaics is nothing that looks particularly worrisome.
>> Well, the Intel GPU issue seems likely to cause problems for lots of people:
>>
>>> [ *NEW* ] drm/i915: BYT rendering broken due to "Remove short-term pins from execbuf, v6"
>>> -----------------------------------------------------------------------------------------
>>> https://linux-regtracking.leemhuis.info/regzbot/regression/1366349e-f96a-3f2c-3094-f5cd1a6fa31f@redhat.com/
>>> https://lore.kernel.org/dri-devel/1366349e-f96a-3f2c-3094-f5cd1a6fa31f@redhat.com/
>>>
>>> By Hans de Goede; 0 days ago; 2 activities, latest 0 days ago.
>>> Introduced in b5cfe6f7a6e1 (v5.18-rc1)
>>>
>>> Recent activities from: Tvrtko Ursulin (1), Hans de Goede (1)
>> Although it looks possible that it mainly affects old chipsets (ie the
>> two reports are for a Bay Trail chip and a Core 2 Duo chip - I have no
>> idea how they compare).
>>
>> That probably means there are a lot of machines out there, but likely
>> not the kind that most kernel developers will be testing, so not a ton
>> of reports until it hits distro kernels etc.
>>
>> It looks like Maarten is already involved.
> This is being tracked here:
> https://gitlab.freedesktop.org/drm/intel/-/issues/5806
>
> I've just tested a patch from Maarten which fixes things for me,
> so hopefully we can get this resolved soon.
>
> Regards,
>
> Hans
>
Hey,

I'm still waiting on feedback on the last patch there, can you test it?

That should be the real fix most likely, waiting for binding to complete before using the vma.

~Maarten