Bug 111848 - AMDGPU and display fails after resume from suspend
Summary: AMDGPU and display fails after resume from suspend
Status: NEW
Alias: None
Product: DRI
Classification: Unclassified
Component: DRM/AMDgpu (show other bugs)
Version: unspecified
Hardware: x86-64 (AMD64) Linux (All)
: not set normal
Assignee: Default DRI bug account
QA Contact:
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2019-09-28 09:30 UTC by Martin Jørgensen
Modified: 2019-10-07 10:14 UTC (History)
1 user (show)

See Also:
i915 platform:
i915 features:


Attachments
dmesg output after wakeup (221.12 KB, text/plain)
2019-09-28 09:32 UTC, Martin Jørgensen
no flags Details
basic info (331 bytes, text/plain)
2019-09-28 09:32 UTC, Martin Jørgensen
no flags Details
demidecode output (24.27 KB, text/x-log)
2019-09-28 09:32 UTC, Martin Jørgensen
no flags Details
lsinitramfs output (114.45 KB, text/x-log)
2019-09-28 09:33 UTC, Martin Jørgensen
no flags Details
lshw log (35.03 KB, text/x-log)
2019-09-28 09:33 UTC, Martin Jørgensen
no flags Details
lsmod amdgpu (301 bytes, text/x-log)
2019-09-28 09:34 UTC, Martin Jørgensen
no flags Details
lspic -nn output (8.37 KB, text/x-log)
2019-09-28 09:34 UTC, Martin Jørgensen
no flags Details
lspci -vnn output (33.57 KB, text/x-log)
2019-09-28 09:34 UTC, Martin Jørgensen
no flags Details
modinfo amdgpu output (30.40 KB, text/x-log)
2019-09-28 09:35 UTC, Martin Jørgensen
no flags Details
package log output (792.97 KB, text/x-log)
2019-09-28 09:35 UTC, Martin Jørgensen
no flags Details
uname -a output (92 bytes, text/x-log)
2019-09-28 09:35 UTC, Martin Jørgensen
no flags Details
Xorg log (62.45 KB, text/x-log)
2019-09-28 09:36 UTC, Martin Jørgensen
no flags Details

Note You need to log in before you can comment on or make changes to this bug.
Description Martin Jørgensen 2019-09-28 09:30:48 UTC
Sometimes after my machine resumes from suspend my display stays off with no signal, and USB keyboard (Num lock) and mouse doesn't react.

This time I managed to SSH into the machine and gather some logs after it failed resuming.

I have experienced this with different kernels and under different Linux operating systems, like Debian and Gentoo, but it is not happening often enough for me to sense a pattern.

I will attach log files to the best of my abilities according to: https://amdgpu-install.readthedocs.io/en/latest/install-bugrep.html
Comment 1 Martin Jørgensen 2019-09-28 09:32:02 UTC
Created attachment 145563 [details]
dmesg output after wakeup
Comment 2 Martin Jørgensen 2019-09-28 09:32:26 UTC
Created attachment 145564 [details]
basic info
Comment 3 Martin Jørgensen 2019-09-28 09:32:55 UTC
Created attachment 145565 [details]
demidecode output
Comment 4 Martin Jørgensen 2019-09-28 09:33:15 UTC
Created attachment 145566 [details]
lsinitramfs output
Comment 5 Martin Jørgensen 2019-09-28 09:33:42 UTC
Created attachment 145567 [details]
lshw log
Comment 6 Martin Jørgensen 2019-09-28 09:34:04 UTC
Created attachment 145568 [details]
lsmod amdgpu
Comment 7 Martin Jørgensen 2019-09-28 09:34:23 UTC
Created attachment 145569 [details]
lspic -nn output
Comment 8 Martin Jørgensen 2019-09-28 09:34:40 UTC
Created attachment 145570 [details]
lspci -vnn output
Comment 9 Martin Jørgensen 2019-09-28 09:35:08 UTC
Created attachment 145571 [details]
modinfo amdgpu output
Comment 10 Martin Jørgensen 2019-09-28 09:35:27 UTC
Created attachment 145572 [details]
package log output
Comment 11 Martin Jørgensen 2019-09-28 09:35:58 UTC
Created attachment 145573 [details]
uname -a output
Comment 12 Martin Jørgensen 2019-09-28 09:36:11 UTC
Created attachment 145574 [details]
Xorg log
Comment 13 me 2019-10-03 14:20:04 UTC
Can confirm this problem on

kernel-5.2.17-200.fc30 
kernel-5.2.9-200.fc30
kernel-5.2.5-200.fc30
kernel-5.3.1-150.vanilla.knurd.1.fc30

Suspend & resume work flawlessly on

kernel-4.20.16-200.fc29
Comment 14 Alex Deucher 2019-10-03 14:31:04 UTC
Can you bisect?
Comment 15 me 2019-10-03 14:38:08 UTC
Is there a Fedora workflow for bisecting?
Also, do I need to take care of any GPU firmware or is it just the kernel?
Comment 16 Alex Deucher 2019-10-03 14:40:17 UTC
(In reply to me from comment #15)
> Is there a Fedora workflow for bisecting?
> Also, do I need to take care of any GPU firmware or is it just the kernel?

Don't worry about the firmware, just the kernel is fine.  Here's a bisect howto tailored to fedora:
https://fedoraproject.org/wiki/User:Ignatenkobrain/Kernel/Bisection
Comment 17 me 2019-10-03 14:59:36 UTC
The kernel-package.py script is broken.
I'll build kernels directly.
Git bisect says approximately 13 steps are required between v4.20 and v5.0.
Will report back in ~14 days when that's done (bug manifests only after some uptime).
Comment 18 me 2019-10-04 18:27:37 UTC
Bisection turns out to be harder than anticipated because there is another bug that makes the system freeze on suspend (not reported yet).

Anyways, this seems to be a duplicate of https://bugs.freedesktop.org/show_bug.cgi?id=111729#c7
Comment 19 me 2019-10-07 10:14:10 UTC
Potential fix (and kernel Bugzilla bug): https://bugzilla.kernel.org/show_bug.cgi?id=204241


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.