Bug 208743 - suspend and hibernate periodically fail. suspend significantly more often than hibernate.
Summary: suspend and hibernate periodically fail. suspend significantly more often tha...
Status: NEW
Alias: None
Product: Drivers
Classification: Unclassified
Component: Video(DRI - non Intel) (show other bugs)
Hardware: x86-64 Linux
: P1 normal
Assignee: drivers_video-dri
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2020-07-30 04:46 UTC by arjunkc
Modified: 2021-03-19 20:39 UTC (History)
1 user (show)

See Also:
Kernel Version: 4.19.0-9,5.4.31
Subsystem:
Regression: No
Bisected commit-id:

Attachments
kernel crash log (739.49 KB, text/plain)
2020-07-30 04:46 UTC, arjunkc 		Details
dmesg (85.50 KB, text/plain)
2020-07-30 04:47 UTC, arjunkc 		Details
lscpi (39.07 KB, text/plain)
2020-07-30 04:47 UTC, arjunkc 		Details
modules (8.42 KB, text/plain)
2020-07-30 04:47 UTC, arjunkc 		Details
cmdline (150 bytes, text/plain)
2020-07-30 04:50 UTC, arjunkc 		Details
Add an attachment (proposed patch, testcase, etc.)

Description arjunkc 2020-07-30 04:46:30 UTC 
Created attachment 290685 [details]
kernel crash log

I have had this problem for a long long time on amd discrete gpus (currently Vega). Suspend and hibernate fail every few cycles. I have looked at [204241/(https://bugzilla.kernel.org/show_bug.cgi?id=204241) but none of the patches have helped; indeed the patches appear to have already applied post 5.4.3.

Using the internal video (intel i915) stops all suspend and resume failures, so it's very likely to be the amdgpu drivers.

I can test kernel versions up to 5.7 since my root filesystem is on zfs, and openzfs does not compile on 5.8.

Attached logs follow.
Comment 1 arjunkc 2020-07-30 04:47:00 UTC 
Created attachment 290687 [details]
dmesg
Comment 2 arjunkc 2020-07-30 04:47:14 UTC 
Created attachment 290689 [details]
lscpi
Comment 3 arjunkc 2020-07-30 04:47:29 UTC 
Created attachment 290691 [details]
modules
Comment 4 arjunkc 2020-07-30 04:50:17 UTC 
Created attachment 290693 [details]
cmdline
Note You need to log in before you can comment on or make changes to this bug.