no poweroff after hibernation is done / garbled graphics on resume

[AndreaBorgia-Abo]

Using 5.04 I noticed that occasionally the power stays on after hibernation is requested. I'd say the scripts work correctly: if I force a poweroff by holding the power button, at the next wakeup the system comes out of hibernation just fine.
This is new to me: ever since hibernation started working reliably on this HW (let's say after 4.20), I never had this glitch.
Going back to 5.02 to see if it's kernel related...

[AndreaBorgia-Abo]

Yup, apparently 5.02 has no problem shutting off at the end.

[M-Bab]

Is it fixed in 5.0.9?

[AndreaBorgia-Abo]

Is it fixed in 5.0.9?

Installing now, I'll update the ticket again in a day or two.

[AndreaBorgia-Abo]

While 5.0.9 did at least once power off correctly (however the problem was intermittent, so that doesn't count for much), it also brought back another issue which I thought was not kernel-related: frozen X. Going back to 5.0.2.

[AndreaBorgia-Abo]

Hmm, even 5.0.2 does something funny: it does power off but it also powers on again if I hit the space bar and then X is frozen as described in the link above. Thing is, I normally don't use the keyboard to wake up the system, just the power button, but I might have occasionally hit the keyboard while covering it. I'll try again with 5.0.9 asap.

[AndreaBorgia-Abo]

I'm having weird reboot problems, similar to a previous report except that in this case I get a reboot instead. Because the system was otherwise idle and temps are normal, I'm going to rollback all the way 5.0.1 and see it this changes anything.

[M-Bab]

Try again with 5.0.11 - there were some commits that could be related to that. If its still not fixed try to extract maximum log information and I can report it upstream to AMD.

[AndreaBorgia-Abo]

I've used 5.0.1 (with matching firmware) for a couple of days without reboots but also without trying to suspend (default UEFI values!). I've also unplugged and reinserted the RAM modules, the SSD and the power connectors, just to rule out a flaky connection. If it still behaves nicely, I'll try suspending. Once I have a confirmed working setup that I can reproduce, I'll try 5.0.11 as you suggested.

[AndreaBorgia-Abo]

Ok, 5.0.1 is (again) confirmed working for my setup, including suspend, with firmware 2.50.
Installing 5.0.11 and fw 2.60... this apparently powers off at the end of the hibernation sequence, however it also has the "frozen X" issue: mouse moves but can't click, keyboard can only be used to switch to console. Terminating X and restarting fixes the issue (but of course negates the whole point of suspending).
@M-Bab I cannot say the issue is completely fixed because I only tried once but there seem to be two distinct issues here: should I open a new bug or would you rather keep them together here?

[AndreaBorgia-Abo]

I'll try the new 5.0.17 and report back.

[AndreaBorgia-Abo]

Update for 5.0.19, sorry I never had the time to test 5.0.17!
Problem no.1) at the end of the hibernation sequence, instead of powering off (as 5.0.1 does) or just hanging there (as 5.0.2 does), it reboots.
Problem no.2) at the end of the resume sequence, there is no login prompt, either in console or in X. The display is corrupted when in X and seems ok in console, though only "resuming from hiberation is visible". At least the reboot keycombo works.
Back to 5.0.1 for daily use.

[AndreaBorgia-Abo]

Update for 5.2.0-r6:

1. poweroff at end of hibernation: seems to work but tested only once because...
2. upon resume, only the mouse pointer is visible (and partially corrupted, too), nothing else from X is shown.
   Went back to console, rebooted, back to 5.0.1

[M-Bab]

Sounds sad but at least they somehow seem to be working on this. Last chance for the 5.2.0 kernel then I would like to report this upstream to AMD.

• What GPU are you using?
• What Cable type to the monitor?
• What OS?
• And do you have any helpful logs on the case?

[AndreaBorgia-Abo]

• Ryzen 3 2200G (embedded GPU) on a ASRock B450M-HDV mobo
• HDMI cable (always used this same cable)
• Debian / Testing

Regarding the logs, first I will test 5.2.0 (not that I expect radical changes from rc6).
Please let me know what you would like to have (dmesg output, what else?) and how to enable extra logging in the kernel if required. Also, would it be enough to collect all logs for 5.2.0 or would you rather have logs for all problematic versions?