Bug 92171 - [NVE7] Reboot causes nouveau corruption/ghosting
Summary: [NVE7] Reboot causes nouveau corruption/ghosting
Alias: None
Product: xorg
Classification: Unclassified
Component: Driver/nouveau (show other bugs)
Version: unspecified
Hardware: x86-64 (AMD64) Linux (All)
: medium major
Assignee: Nouveau Project
QA Contact: Xorg Project Team
Depends on:
Reported: 2015-09-28 23:01 UTC by Bryan O'Donoghue
Modified: 2019-12-04 09:04 UTC (History)
0 users

See Also:
i915 platform:
i915 features:

dmesg-4.3.0-rc1-broken (85.79 KB, text/plain)
2015-09-28 23:01 UTC, Bryan O'Donoghue
no flags Details
dmesg-4.3.0-rc3-broken (87.99 KB, text/plain)
2015-09-28 23:02 UTC, Bryan O'Donoghue
no flags Details
dmesg-4.3.0-rc3-broken-onescreen-good (88.94 KB, text/plain)
2015-09-28 23:03 UTC, Bryan O'Donoghue
no flags Details
artifact-on-primary-screen (174.82 KB, image/jpeg)
2015-09-29 06:44 UTC, Bryan O'Donoghue
no flags Details
4.3-rc3-drm-debug-included (186.62 KB, text/plain)
2015-09-29 11:23 UTC, Bryan O'Donoghue
no flags Details

Description Bryan O'Donoghue 2015-09-28 23:01:23 UTC
Created attachment 118503 [details]

Running kernel 4.3-rcX I can usually boot to a good nouveau session, however if I reboot the system the display is squashed and ghosted across the screen.

I have a capture of the kernel log showing a FIFO that has failed to reset. Rebooting again doesn't clear the error - nor does shutting down and starting again - I typically need to go and boot an older kernel ubuntu 3.13.0-58-generic in this case. I haven't tested shutting down the system for a prolonged period after the corruption yet.

Attached is the kernel log of

dmesg-4.3.0-rc1.txt (totally broken both screens)
dmesg-4.3.0-rc3.txt (totally broken both screens)
dmesg-4.3.0-rc3-onescreen-good.txt (totally broken on one screen)

Also note this error is readily replicated with one screen present - the laptop display. I'll attach a dump of that too.
Comment 1 Bryan O'Donoghue 2015-09-28 23:02:47 UTC
Created attachment 118504 [details]
Comment 2 Bryan O'Donoghue 2015-09-28 23:03:23 UTC
Created attachment 118505 [details]
Comment 3 Ilia Mirkin 2015-09-28 23:07:15 UTC
How are the screens plugged in? Can you provide the dmesg when booted with

nouveau.debug=debug drm.debug=0xe

This looks like a non-optimus setup, is that right?
Comment 4 Bryan O'Donoghue 2015-09-29 06:43:22 UTC
Yes, that's a non-optimus setup.

I'll get you that capture. Right now I have this weird artifact on my primary laptop screen that won't go away after booting the system a number of times @ the 4.3.0-rcX kernel - I'm assuming the system isn't actually resetting the nvidia when I power down (some sort of warm boot) - so I've disconnected the power and I'm letting the battery run flat.

I'm attaching a picture of the artifact that persisting through resets (note it's still there even after leaving the system powered off overnight)
Comment 5 Bryan O'Donoghue 2015-09-29 06:44:10 UTC
Created attachment 118512 [details]
Comment 6 Ilia Mirkin 2015-09-29 07:10:20 UTC
Also useful to know if the issue happens with 4.2. The nouveau driver got basically rewritten in 4.3, a regression easily could have crept in.
Comment 7 Bryan O'Donoghue 2015-09-29 11:22:20 UTC
Just to confirm.

1. First boot is normal
2. Issue a reboot
3. Get corrupted screen
4. Need to completely allow power to drain (so presumably there is some warm boot foolery going on here with the BIOS when I just do a power down)

Requested dmesg attached.
Comment 8 Bryan O'Donoghue 2015-09-29 11:22:54 UTC
I'll compile a 4.2 tag and let you know.
Comment 9 Bryan O'Donoghue 2015-09-29 11:23:52 UTC
Created attachment 118515 [details]
Comment 10 Martin Peres 2019-12-04 09:04:18 UTC
-- GitLab Migration Automatic Message --

This bug has been migrated to freedesktop.org's GitLab instance and has been closed from further activity.

You can subscribe and participate further through the new bug through this link to our GitLab instance: https://gitlab.freedesktop.org/xorg/driver/xf86-video-nouveau/issues/218.

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.