FS#46968 - [linux] Display is broken after upgrading to linux kernel 4.3.

Attached to Project: Arch Linux
Opened by Frederic Bezies (fredbezies) - Wednesday, 04 November 2015, 16:07 GMT
Last edited by Tobias Powalowski (tpowa) - Thursday, 24 December 2015, 07:30 GMT
Task Type Bug Report
Category Kernel
Status Closed
Assigned To Tobias Powalowski (tpowa)
Thomas Bächler (brain0)
Architecture All
Severity Critical
Priority Normal
Reported Version
Due in Version Undecided
Due Date Undecided
Percent Complete 100%
Votes 3
Private No

Details

Description: Simple bug to reproduce.

If you use an Intel chipset with linux 4.3, display is completely busted. I saw that on my asus eeePC with an intel video chipset and archlinux testing on it.

Additional info:
linux 4.3-1
1:2.99.917+478+gdf72bc5-2

Steps to reproduce:

See details.

This task depends upon

Closed by  Tobias Powalowski (tpowa)
Thursday, 24 December 2015, 07:30 GMT
Reason for closing:  Fixed
Additional comments about closing:  4.3.3-2
Comment by Frederic Bezies (fredbezies) - Wednesday, 04 November 2015, 16:53 GMT
More infos.

It seems to break display with this chipset :

00:02.0 VGA compatible controller: Intel Corporation Mobile 945GSE Express Integrated Graphics Controller (rev 03)
00:02.1 Display controller: Intel Corporation Mobile 945GM/GMS/GME, 943/940GML Express Integrated Graphics Controller (rev 03)

I attached error log. Only workaround on my eeePC was to "downgrade to linux 4.2.5.

Comment by Daniel Hahler (blueyed) - Thursday, 12 November 2015, 00:40 GMT
Does not affect any/all Intel chipsets. It works on a Lenovo X250 with:

00:02.0 VGA compatible controller: Intel Corporation Broadwell-U Integrated Graphics (rev 09)
Comment by Frederic Bezies (fredbezies) - Thursday, 12 November 2015, 15:17 GMT
Display is broken on my old eeePC with intel 945 chipset. But on my Toshiba, which is also using intel chipset, no problem.

My laptop is using :

00:02.0 VGA compatible controller: Intel Corporation Mobile 4 Series Chipset Integrated Graphics Controller (rev 07)
00:02.1 Display controller: Intel Corporation Mobile 4 Series Chipset Integrated Graphics Controller (rev 07)

So, only breaking old chipsets like intel 945 one ?
Comment by Eli Schwartz (eschwartz) - Wednesday, 18 November 2015, 04:16 GMT
Same problem here on an old Dell Inspiron B130 with aur/linux-ck and testing/linux.


00:02.0 VGA compatible controller: Intel Corporation Mobile 915GM/GMS/910GML Express Graphics Controller (rev 03)
00:02.1 Display controller: Intel Corporation Mobile 915GM/GMS/910GML Express Graphics Controller (rev 03)
Comment by txtsd (txtsd) - Wednesday, 18 November 2015, 09:44 GMT
Works fine on my ancient GM965

00:02.0 VGA compatible controller: Intel Corporation Mobile GM965/GL960 Integrated Graphics Controller (primary) (rev 03)
00:02.1 Display controller: Intel Corporation Mobile GM965/GL960 Integrated Graphics Controller (secondary) (rev 03)
Comment by John (graysky) - Wednesday, 18 November 2015, 11:40 GMT
Using linux from ABS (4.3-1 with a few custom settings in the .config) works fine on the following hardware as does the corresponding linux-ck (4.3-1). Note I do not have [testing] enabled:

*Broadwell i7-4790K that uses HD Graphics 4600
*Haswell i3-4130T that uses HD Graphics 4400
*Ivy i7-3770K that uses HD Graphics 4000

@Frederic - Did you open a report upstream?
Comment by Frederic Bezies (fredbezies) - Wednesday, 18 November 2015, 11:43 GMT
I didn't report it. I will try to see if it is a known bug or not.
Comment by John (graysky) - Wednesday, 18 November 2015, 11:49 GMT
My recommendation is that you open one up against the "Video(DRI - Intel)" at https://bugzilla.kernel.org/enter_bug.cgi?product=Drivers

In the past, Chris has been very swift and accurate in diagnosing and fixing.
Comment by Frederic Bezies (fredbezies) - Wednesday, 18 November 2015, 11:58 GMT
https://bugzilla.kernel.org/show_bug.cgi?id=108141

And after googling "drivers/gpu/drm/drm_atomic_helper.c:674" I found this :

https://lkml.org/lkml/2015/11/12/467
Comment by Frederic Bezies (fredbezies) - Wednesday, 18 November 2015, 13:22 GMT
Looks like my invalid bug (I choose the wrong component) was a duplicate of this one : https://bugs.freedesktop.org/show_bug.cgi?id=92655
Comment by Eli Schwartz (eschwartz) - Wednesday, 18 November 2015, 21:42 GMT
From the freedesktop.org bug (comment #4), this patch fixed linux-ck for me: http://lists.freedesktop.org/archives/intel-gfx/2015-November/079894.html
Comment by Frederic Bezies (fredbezies) - Wednesday, 18 November 2015, 21:43 GMT
Thanks for the feedback. I hope it could be integrated for kernel 4.3.1 :D
Comment by Larry Johnson (keepitsimpleengineer) - Tuesday, 24 November 2015, 15:53 GMT
I uas a Lenovo S12 (Intel) with:
00:02.0 VGA compatible controller: Intel Corporation Mobile 945GSE Express Integrated Graphics Controller (rev 03)
00:02.1 Display controller: Intel Corporation Mobile 945GM/GMS/GME, 943/940GML Express Integrated Graphics Controller (rev 03)

With linux-4.3-1 and linux-ck-4.3.1 running archlinux i686 I experience this bug.

However, with linux-ck-4.3.1 running manjaro i686 there is no such bug.

UPDATE: After reinstalling manjaro (12/20) for reasons unrelated to this bug, I now find the 4.3 kernel unusable.

UPDATE #2: graysky's 2015/12/24 update to linux-ck 4.3.3-2 was installed and now works. (Lenovo Ideapad S12)
Comment by Frederic Bezies (fredbezies) - Tuesday, 24 November 2015, 16:04 GMT
Smells like there is a different .config file used for Manjaro linux kernel package.
Comment by Larry Johnson (keepitsimpleengineer) - Tuesday, 24 November 2015, 18:55 GMT
Both 4.3.2. -ck packages came from graysky's repo-ck and likely on the same day... .. .
Comment by So As (Archilious) - Thursday, 03 December 2015, 23:01 GMT
On AMD E450 with Radeon HD6320 running all 'testing' Arch Linux x86_64 including kernel 4.3.0-1, no such bugs, running smooth. Everything's fine.

Comment by So As (Archilious) - Sunday, 06 December 2015, 05:20 GMT
On Sony Vaio with Intel(R) Core(TM) i3-2310M CPU :
0:02.0 VGA compatible controller: Intel Corporation 2nd Generation Core Processor Family Integrated Graphics Controller (rev 09)
4.3.0-1-ARCH #1 SMP PREEMPT Mon Nov 2 16:52:35 CET 2015 x86_64 GNU/Linux

No display problems.
Comment by Frederic Bezies (fredbezies) - Thursday, 10 December 2015, 13:13 GMT
Linux 4.3.1-1 is still breaking display on old intel chipset. Could it be possible to "backport" patch proposed in kernel bug report ? https://bugs.freedesktop.org/show_bug.cgi?id=92655 -> https://bugs.freedesktop.org/attachment.cgi?id=119541

Thanks for your answer.
Comment by Eli Schwartz (eschwartz) - Thursday, 10 December 2015, 16:03 GMT Comment by Frederic Bezies (fredbezies) - Tuesday, 15 December 2015, 14:59 GMT
Too bad kernel 4.3.3-1 is still busted. Maybe in 4.3.4 with a little luck ?
Comment by Frederic Bezies (fredbezies) - Tuesday, 15 December 2015, 15:02 GMT
Just wondering : could it be possible to move this kernel to core and ask users with old Intel Chipset (like the 945GSE one) to switch to linux lts kernel ? Or is there any old blocker to move this kernel in core ?
Comment by Eli Schwartz (eschwartz) - Tuesday, 15 December 2015, 18:23 GMT
I thought this was the blocker. (Well, I don't know, there could be more.)

Deliberately releasing a buggy kernel to core and suggesting affected people use linux-lts seems to be defeating the purpose of *having* a testing repo.

...

The fix appears to be in linux-next.
Comment by Frederic Bezies (fredbezies) - Tuesday, 15 December 2015, 18:38 GMT
I understand this is against the purpose of testing repositories. In linux-next ? Let's hope it will be in next release and that next release won't be in a month like it takes for linux 4.3.1 patch.

I feel a little guilty for having filled such a blocker, even if kernel developers are also guilty here.
Comment by cfr (cfr42) - Wednesday, 23 December 2015, 22:10 GMT
As a user of the stable repo, I am more than happy if people using testing file bugs which block the release of buggy kernels to core! I always thought that was the point....

[I'm only here because I wanted to be sure my updates hadn't gotten messed up somehow since I realised it was a while since I'd had a new kernel. I am pleased to discover that there is a perfectly good reason for this.]

Loading...