Note: This is a public test instance of Red Hat Bugzilla. The data contained within is a snapshot of the live data so any changes you make will not be reflected in the production Bugzilla. Email is disabled so feel free to test any aspect of the site that you want. File any problems you find or give feedback at bugzilla.redhat.com.
Bug 1226743 - xorg-x11-drv-intel-2.99.917-10.20150526.fc22.x86_64 causes screen corruption with Intel Mobile GM965/GL960 chipset
Summary: xorg-x11-drv-intel-2.99.917-10.20150526.fc22.x86_64 causes screen corruption ...
Keywords:
Status: CLOSED ERRATA
Alias: None
Product: Fedora
Classification: Fedora
Component: xorg-x11-drv-intel
Version: 22
Hardware: x86_64
OS: Linux
unspecified
high
Target Milestone: ---
Assignee: Dave Airlie
QA Contact: Fedora Extras Quality Assurance
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2015-06-01 00:41 UTC by Stephen Haffly
Modified: 2015-06-28 16:04 UTC (History)
9 users (show)

Fixed In Version: kernel-4.0.5-300.fc22
Doc Type: Bug Fix
Doc Text:
Clone Of:
Environment:
Last Closed: 2015-06-20 23:58:26 UTC
Type: Bug
Embargoed:


Attachments (Terms of Use)
Screen capture of corrupted screen on dv6833us laptop (660.70 KB, image/png)
2015-06-01 00:41 UTC, Stephen Haffly
no flags Details
disable WC mmaps until userspace issues are sorted out. (555 bytes, patch)
2015-06-03 05:30 UTC, Dave Airlie
no flags Details | Diff


Links
System ID Private Priority Status Summary Last Updated
Red Hat Bugzilla 1226531 0 urgent CLOSED DRI3 disabled after update to 2.99.917-9 causes massive screen corruption on older Intel cards 2022-05-16 11:32:56 UTC

Internal Links: 1226531

Description Stephen Haffly 2015-06-01 00:41:53 UTC
Created attachment 1033063 [details]
Screen capture of corrupted screen on dv6833us laptop

Description of problem: Updating to xorg-x11-drv-intel-2.99.917-10.20150526.fc22.x86_64 causes screen corruption (see attachment). xorg-x11-drv-intel-2.99.917-6.20150211.fc22.x86_64 works. Downgrading fixes the problem


Version-Release number of selected component (if applicable):
2.99.917-10.20150526.fc22.x86_64

How reproducible:
Simply allowing dnf to apply the update manifests this problem at next reboot.

Steps to Reproduce:
1. Update xorg-x11-drv-intel
2. reboot
3. problem occurs

Actual results:
video corruption

Expected results:
Should be no corruption

Additional info:
This is on an HP Pavilion dv6833us laptop with Intel Mobile GM965/GL960 Integrated Graphics Controller. lshw reports that it is using the i915 driver.

Comment 1 Dave Airlie 2015-06-01 02:53:37 UTC
https://kojipkgs.fedoraproject.org/scratch/airlied/task_9901298/

can you give that a try? its just updated to newer upstream to get a data point.

Comment 2 Stephen Haffly 2015-06-01 13:33:35 UTC
Unfortunately, this version also has the same problem. Again, downgrading to the 2.99.917-6.20150211 version works.

Would it help to install the intel-gpu-tools and/or the debug version to try to find out why this is happening? If so, I would need instructions on how to use them. If you are at Red Hat in Raleigh, I could bring the laptop to you as I live in Wake Forest.

Comment 3 Dave Airlie 2015-06-02 08:01:52 UTC
okay got more from upstream

https://kojipkgs.fedoraproject.org/scratch/airlied/task_9914452/

any better,

can you also attach 

cat /sys/kernel/debug/dri/0/i915_swizzle_info

I might have to move the bug to upstream as Chris Wilson is fairly good at reacting to things, I'll see how this package helps, he'll provide better debugging instructions than I can!

(nope not in Raleigh unfortunately)

Comment 4 Stephen Haffly 2015-06-02 16:33:42 UTC
Not only did this one also fail, but it messed up something so that I could not undo it with dnf (said was not installed). I finally found the older rpm package online and used rpm to install with the --force switch. I am just glad I thought of that since I was thinking I would have to reinstall Fedora 22.

The output of cat /sys/kernel/debug/dri/0/i915_swizzle_info is:

bit6 swizzle for X-tiling = bit9/bit10/bit11
bit6 swizzle for Y-tiling = bit9/bit11
DDC = 0x000f0002
DDC2 = 0x00000000
C0DRB3 = 0x0000
C1DRB3 = 0x0000
L-shaped memory detected

Comment 5 Dave Airlie 2015-06-02 23:34:32 UTC
okay that's no good, dnf distro-sync xorg-x11-drv-intel should return you the upstream one.


I've got a test kernel that might work better,

https://kojipkgs.fedoraproject.org/scratch/airlied/task_9915290/

it disables a feature of the i915 kernel driver that is possibly causing the problems.

Comment 6 Stephen Haffly 2015-06-03 02:17:13 UTC
Success!

I installed the test kernel and then did a dnf update xorg-x11-drv-intel, which brought in xorg-x11-drv-intel-2.99.917-10.20150526.fc22.x86_64, which is the version in which I originally noticed the problem and for which I opened this bug report. I am happy to say that with the test kernel, things appear to be working normally.

This leads to the question being whether the fault is in this driver or in the kernel i915 driver.

How do I prevent this in the future? Do I exclude kernel updates, exclude xorg-x11-drv-intel updates, both, or neither?

Comment 7 Dave Airlie 2015-06-03 05:30:53 UTC
Created attachment 1034128 [details]
disable WC mmaps until userspace issues are sorted out.

Comment 8 Stephen Haffly 2015-06-03 15:52:40 UTC
I did notice one video glitch. I was using Firefox and one of the graphics on the web page suddenly had the broken appearance. However, when I refreshed the page, the graphic reverted to normal. It only happened the one time so far. Otherwise, the test kernel has continued to work well.

As for disabling WC mmaps, I don't think that is something I can do unless I build my own kernels. Is that correct? Will future kernel updates already have this fix applied?

Comment 9 Josh Boyer 2015-06-04 11:28:00 UTC
(In reply to Dave Airlie from comment #7)
> Created attachment 1034128 [details]
> disable WC mmaps until userspace issues are sorted out.

Applied on all branches.

Comment 10 Fedora Update System 2015-06-09 12:57:52 UTC
kernel-4.0.5-300.fc22 has been submitted as an update for Fedora 22.
https://admin.fedoraproject.org/updates/kernel-4.0.5-300.fc22

Comment 11 Fedora Update System 2015-06-09 12:59:25 UTC
kernel-4.0.5-200.fc21 has been submitted as an update for Fedora 21.
https://admin.fedoraproject.org/updates/kernel-4.0.5-200.fc21

Comment 12 Fedora Update System 2015-06-10 19:08:59 UTC
Package kernel-4.0.5-200.fc21:
* should fix your issue,
* was pushed to the Fedora 21 testing repository,
* should be available at your local mirror within two days.
Update it with:
# su -c 'yum update --enablerepo=updates-testing kernel-4.0.5-200.fc21'
as soon as you are able to, then reboot.
Please go to the following url:
https://admin.fedoraproject.org/updates/FEDORA-2015-9704/kernel-4.0.5-200.fc21
then log in and leave karma (feedback).

Comment 13 Stephen Haffly 2015-06-14 04:28:58 UTC
This thread in Fedora Forums sugggests forcing UXA grephics instead of SNA graphics as a way to fix the video corruption problem. http://forums.fedoraforum.org/showthread.php?t=304718

However, this was still associated with kernel changes. The UXA architecture, while slower than the SNA architecture, seems to be more reliable on older, pre-SandyBridge hardware.

Comment 14 Sergio Belkin 2015-06-16 14:48:00 UTC
I've tested the kernel 4.0.5 and the same error persists:

[mar jun 16 11:44:37 2015] [drm] stuck on render ring
[mar jun 16 11:44:37 2015] [drm] GPU HANG: ecode 3:0:0x717fffc1, in Xorg [992], reason: Ring hung, action: reset
[mar jun 16 11:44:37 2015] [drm] GPU hangs can indicate a bug anywhere in the entire gfx stack, including userspace.
[mar jun 16 11:44:37 2015] [drm] Please file a _new_ bug report on bugs.freedesktop.org against DRI -> DRM/Intel
[mar jun 16 11:44:37 2015] [drm] drm/i915 developers can then reassign to the right component if it's not a kernel issue.
[mar jun 16 11:44:37 2015] [drm] The gpu crash dump is required to analyze gpu hangs, so please always attach it.
[mar jun 16 11:44:37 2015] [drm] GPU crash dump saved to /sys/class/drm/card0/error
[mar jun 16 11:44:37 2015] drm/i915: Resetting chip after gpu hang

Comment 15 Fedora Update System 2015-06-20 23:58:26 UTC
kernel-4.0.5-200.fc21 has been pushed to the Fedora 21 stable repository.  If problems still persist, please make note of it in this bug report.

Comment 16 Fedora Update System 2015-06-21 00:31:45 UTC
kernel-4.0.5-300.fc22 has been pushed to the Fedora 22 stable repository.  If problems still persist, please make note of it in this bug report.

Comment 17 Neil Romig 2015-06-28 16:04:40 UTC
I still have this problem with kernel kernel-4.0.5-300.fc22.x86_64 and xorg-x11-drv-intel-2.99.917-12.20150615.fc22.x86_64. At least it looks exactly this bug with the rectangular blocks of lines corrupting the display.


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