Bug 1916674 - X11 lockup in i915 with benign logs due to deadlock with all 5.10.1+ kernels
Summary: X11 lockup in i915 with benign logs due to deadlock with all 5.10.1+ kernels
Keywords:
Status: CLOSED ERRATA
Alias: None
Product: Fedora
Classification: Fedora
Component: kernel
Version: 33
Hardware: Unspecified
OS: Unspecified
unspecified
urgent
Target Milestone: ---
Assignee: Kernel Maintainer List
QA Contact: Fedora Extras Quality Assurance
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2021-01-15 11:55 UTC by Arcadiy Ivanov
Modified: 2021-02-15 02:16 UTC (History)
24 users (show)

Fixed In Version: kernel-5.10.15-200.fc33 kernel-5.10.15-100.fc32
Clone Of:
Environment:
Last Closed: 2021-02-14 01:23:28 UTC
Type: Bug
Embargoed:


Attachments (Terms of Use)

Description Arcadiy Ivanov 2021-01-15 11:55:21 UTC
The issue described in detail is here:

https://gitlab.freedesktop.org/drm/intel/-/issues/2905

Proposed fix: 

drm/i915: Add ww locking to pin_to_display_plane, v2
https://www.spinics.net/lists/intel-gfx/msg246625.html

The fix is neither confirmed nor currently scheduled for stable.

Comment 1 Arcadiy Ivanov 2021-01-15 14:15:21 UTC
Correction to the above note: 

"drm/i915: Add ww locking to pin_to_display_plane, v2" is the **likely cause**, not the fix. Sorry for the confusion.

There is neither a confirmed fix for the issue nor any ETA from the drm/i915 team.

Comment 2 Arcadiy Ivanov 2021-02-08 17:35:57 UTC
The proposed fix is slated to be included in 5.10.15+
https://lore.kernel.org/stable/20210208145818.395353822@linuxfoundation.org/
The proposed fix patch is "[PATCH 5.10 080/120] drm/i915/gem: Drop lru bumping on display unpinning Greg Kroah-Hartman"

Comment 3 Arcadiy Ivanov 2021-02-10 16:20:59 UTC
The fix is released in 5.10.15.

Comment 4 Arcadiy Ivanov 2021-02-10 16:33:10 UTC
@jforbes, please build .15 ASAP as without this fix the i915 in DRI3 is unusable in 5.10.1+. Thanks!

Comment 5 Fedora Update System 2021-02-10 22:14:18 UTC
FEDORA-2021-76aaa904e2 has been submitted as an update to Fedora 33. https://bodhi.fedoraproject.org/updates/FEDORA-2021-76aaa904e2

Comment 6 Fedora Update System 2021-02-10 22:14:30 UTC
FEDORA-2021-f8ede2fdfc has been submitted as an update to Fedora 32. https://bodhi.fedoraproject.org/updates/FEDORA-2021-f8ede2fdfc

Comment 7 Fedora Update System 2021-02-11 02:00:16 UTC
FEDORA-2021-76aaa904e2 has been pushed to the Fedora 33 testing repository.
Soon you'll be able to install the update with the following command:
`sudo dnf upgrade --enablerepo=updates-testing --advisory=FEDORA-2021-76aaa904e2`
You can provide feedback for this update here: https://bodhi.fedoraproject.org/updates/FEDORA-2021-76aaa904e2

See also https://fedoraproject.org/wiki/QA:Updates_Testing for more information on how to test updates.

Comment 8 Fedora Update System 2021-02-11 02:23:23 UTC
FEDORA-2021-f8ede2fdfc has been pushed to the Fedora 32 testing repository.
Soon you'll be able to install the update with the following command:
`sudo dnf upgrade --enablerepo=updates-testing --advisory=FEDORA-2021-f8ede2fdfc`
You can provide feedback for this update here: https://bodhi.fedoraproject.org/updates/FEDORA-2021-f8ede2fdfc

See also https://fedoraproject.org/wiki/QA:Updates_Testing for more information on how to test updates.

Comment 9 Arcadiy Ivanov 2021-02-11 20:30:14 UTC
I confirm the issue has been resolved (>24hrs with no hangs).

Comment 10 Fedora Update System 2021-02-14 01:23:28 UTC
FEDORA-2021-76aaa904e2 has been pushed to the Fedora 33 stable repository.
If problem still persists, please make note of it in this bug report.

Comment 11 Fedora Update System 2021-02-14 01:45:37 UTC
FEDORA-2021-f8ede2fdfc has been pushed to the Fedora 32 stable repository.
If problem still persists, please make note of it in this bug report.

Comment 12 Alexander Zaitsev 2021-02-14 22:43:58 UTC
On Fedora 33 the issue is still reproducible:

dmesg -T piece:

[Вс фев 14 23:17:58 2021] i915 0000:00:02.0: [drm] GPU HANG: ecode 7:1:bd9bffff, in chrome [9300]
[Вс фев 14 23:17:58 2021] GPU hangs can indicate a bug anywhere in the entire gfx stack, including userspace.
[Вс фев 14 23:17:58 2021] Please file a _new_ bug report at https://gitlab.freedesktop.org/drm/intel/issues/new.
[Вс фев 14 23:17:58 2021] Please see https://gitlab.freedesktop.org/drm/intel/-/wikis/How-to-file-i915-bugs for details.
[Вс фев 14 23:17:58 2021] drm/i915 developers can then reassign to the right component if it's not a kernel issue.
[Вс фев 14 23:17:58 2021] The GPU crash dump is required to analyze GPU hangs, so please always attach it.
[Вс фев 14 23:17:58 2021] GPU crash dump saved to /sys/class/drm/card0/error
[Вс фев 14 23:17:58 2021] i915 0000:00:02.0: [drm] Resetting chip for stopped heartbeat on rcs0
[Вс фев 14 23:17:58 2021] i915 0000:00:02.0: [drm] chrome[9300] context reset due to GPU hang
[Вс фев 14 23:18:04 2021] i915 0000:00:02.0: [drm] GPU HANG: ecode 7:1:bd9bffff, in chrome [59153]
[Вс фев 14 23:18:04 2021] i915 0000:00:02.0: [drm] Resetting chip for stopped heartbeat on rcs0
[Вс фев 14 23:18:04 2021] i915 0000:00:02.0: [drm] chrome[59153] context reset due to GPU hang

uname -r:

5.10.15-200.fc33.x86_64

related lspci:

00:02.0 VGA compatible controller: Intel Corporation 3rd Gen Core processor Graphics Controller (rev 09)

Comment 13 Arcadiy Ivanov 2021-02-15 02:16:28 UTC
(In reply to Alexander Zaitsev from comment #12)
> On Fedora 33 the issue is still reproducible:
> 
> dmesg -T piece:
> 
> [Вс фев 14 23:17:58 2021] i915 0000:00:02.0: [drm] GPU HANG: ecode
> 7:1:bd9bffff, in chrome [9300]

It's not reproducible because this is not the same issue. This issue does not describe a GPU crash/hang, but a full irreversible lockup. Please file a new issue at https://gitlab.freedesktop.org/drm/intel/-/issues.


Note You need to log in before you can comment on or make changes to this bug.