Bug 201505 - Resume from suspend does not power up the display
Summary: Resume from suspend does not power up the display
Status: RESOLVED OBSOLETE
Alias: None
Product: Drivers
Classification: Unclassified
Component: Video(DRI - non Intel) (show other bugs)
Hardware: All Linux
: P1 normal
Assignee: drivers_video-dri
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2018-10-24 13:53 UTC by Jan Ziak
Modified: 2020-05-11 19:57 UTC (History)
5 users (show)

See Also:
Kernel Version: 4.19
Subsystem:
Regression: No
Bisected commit-id:

Attachments
dmesg (72.91 KB, text/plain)
2018-10-24 14:36 UTC, Jan Ziak 		Details
dmesg with resume-from-suspend (80.00 KB, text/plain)
2018-10-24 15:50 UTC, Jan Ziak 		Details
bisect.log (3.41 KB, text/plain)
2018-10-25 15:04 UTC, Jan Ziak 		Details
Add an attachment (proposed patch, testcase, etc.)

Description Jan Ziak 2018-10-24 13:53:00 UTC 
Resuming the machine from suspend-to-ram no longer restores power to the display attached to an AMD GPU in Linux 4.19 (assuming the display has switched to sleep mode by itself because the computer hasn't been in use for a while).

If the display is still powered on when resuming, it ends up blank.

Linux 4.18 did not contain this bug.

Kernel: 4.19
Module: amdgpu
GPU: R9 390
Comment 1 Michel Dänzer 2018-10-24 14:16:29 UTC 
Please attach the output of dmesg.
Comment 2 Jan Ziak 2018-10-24 14:36:40 UTC 
Created attachment 279135 [details]
dmesg

dmesg without any resume-from-suspend.
Comment 3 Jan Ziak 2018-10-24 15:50:03 UTC 
Created attachment 279137 [details]
dmesg with resume-from-suspend

dmesg with resume-from-suspend. It contains 2 amdgpu failures to send a message.
Comment 4 Alex Deucher 2018-10-24 19:42:04 UTC 
Can you bisect?
Comment 5 Jan Ziak 2018-10-25 15:04:54 UTC 
Created attachment 279151 [details]
bisect.log

I bisected the issue but encountered some inconsistencies along the way:

- The bisected commit ends up with a white display after resume-from-suspend, while I was expecting a blank display.

- I started bisecting with v4.18 good and v4.19 bad. The bisected commit has date 25 may 2018, while 4.18 has been released on 12 aug 2018.
Comment 6 Jan Ziak 2020-05-11 19:57:05 UTC 
I am resolving this issue as obsolete because I stopped using R9 390 as a primary GPU some time ago. If somebody can still reproduce this bug please reopen this issue.
Note You need to log in before you can comment on or make changes to this bug.