| View previous topic :: View next topic |
| Author |
Message |
DarkAudit
n00b
Joined: 27 Mar 2021
Posts: 27
|
 Posted: Thu Jul 01, 2021 3:55 am Post subject: SOLVED-Kernel 5.13.0/Radeon R6 laptop - screen won't wake up |
 |
|
System is an HP Pavilion laptop running an A10-8700P with Radeon R6 graphics. Running LightDM into either Plasma 5.22 or XFCE4.
Upgraded kernel to 5.13.0 from 5.12.13 by copying the .config per the wiki and running make olddefconfig, also per the wiki.
When the screen blanks from idle and I try to come back, the backlight lights up, but the screen stays blank. Trying to switch to a console with ctrl-alt-F1 or F2 does nothing. Ctrl-alt-delete will reboot the system, so it's not a lockup.
Reverting to the 5.12.13 kernel works as intended.
A different machine running the same kernel, but with an i7-3770 and intel graphics displays no such issues.
Is there a new setting I may have overlooked?
EDIT: it's been brought to my attention that linux-headers has yet to update to 5.13. Might that be a cause of my issue?
Last edited by DarkAudit on Mon Aug 09, 2021 12:36 am; edited 1 time in total |
|
| Back to top |
|
 |
DarkAudit
n00b
Joined: 27 Mar 2021
Posts: 27
|
| Posted: Mon Jul 05, 2021 1:32 pm Post subject: |
|
|
| I guess something was up with that kernel, because portage just handed me 5.12.14 and yoinked 5.13.0. |
|
| Back to top |
|
|
Hu
Administrator
Joined: 06 Mar 2007
Posts: 22443
|
|
| Back to top |
|
|
DarkAudit
n00b
Joined: 27 Mar 2021
Posts: 27
|
| Posted: Tue Jul 06, 2021 7:33 am Post subject: |
|
|
And has since been unmasked. New version has been compiled. Then I had to leave for work. We'll see what the result is later this morning. |
|
| Back to top |
|
|
DarkAudit
n00b
Joined: 27 Mar 2021
Posts: 27
|
| Posted: Tue Jul 06, 2021 10:52 am Post subject: |
|
|
| New version of 5.13.0, same problem. Screen blanks from idle and stays blank when moving mouse or touching keys. Backlight lights up, but that's it. |
|
| Back to top |
|
|
DarkAudit
n00b
Joined: 27 Mar 2021
Posts: 27
|
| Posted: Sat Jul 10, 2021 4:32 am Post subject: |
|
|
Nothing? Anyone? Not even a hint of what's going on?
The system isn't locking up. I can ctrl-alt-f* to another console, login as root, and reboot from there, even though I can't see what I'm typing. Essentially the system is still running, just with no display.
Running the 5.12.* kernels, everything is fine. With 5.13.*, the display does not come back if it blanks from idle.
When I configured and installed the kernel, I followed the wiki and first copied the old .config, then ran make olddefconfig. Then make && make modules install and make install.
There is an old Optiplex with Intel graphics also running 5.13.*. No such blank screen issues. I follow the same kernel upgrade process, but with the addition of genkernel due to LUKS.
I've been waiting for linux-headers-5.13 to become available regularly before installing. Would that have any effect on this issue? |
|
| Back to top |
|
|
Hu
Administrator
Joined: 06 Mar 2007
Posts: 22443
|
| Posted: Sat Jul 10, 2021 3:55 pm Post subject: |
|
|
| DarkAudit wrote: | | I've been waiting for linux-headers-5.13 to become available regularly before installing. Would that have any effect on this issue? |
Probably not. Newer linux-headers are useful when building glibc so that it can access all the features of your new kernel. Since your problem seems tied to the kernel version you run, I would say that your glibc version, and what it does or does not support, likely does not matter. This seems like a kernel bug. Since the machine does not die, you should be able to access any logs it writes after the screen goes blank. Does the kernel log anything of note? What kind of blank events break things? Some desktops will have a first level blank where they effectively create a topmost window of a single color, usually black. This window obscures all real windows, thereby blanking the monitor, but the monitor is still "powered on" as far as the graphics card and monitor are concerned. There are also deeper blank states, with varying levels of powersaving. The deepest blank can direct the graphics card to cease emitting a single. Many monitors will deactivate their backlight in this case.
Please try the following. Disable your screensaver, if any. Save all your work, in case this test results in a stuck screen. Run xset dpms force standby. Your screen should blank, but the backlight will remain on. When the system is working well, you can unblank the screen almost instantly with a keypress. Wait a few seconds after the screen blanks, then press a key to wake it. Report whether you can recover from standby. Next, set dpms force suspend. This takes a bit longer to unblank, but still less than 2 seconds for me. Report whether you can recover from suspend. Next, xset dpms force off. This is the deepest sleep, and the backlight will likely be disabled. Report whether you can recover from off. |
|
| Back to top |
|
|
DarkAudit
n00b
Joined: 27 Mar 2021
Posts: 27
|
| Posted: Sun Jul 11, 2021 1:13 am Post subject: |
|
|
| Hu wrote: |
Please try the following. Disable your screensaver, if any. Save all your work, in case this test results in a stuck screen. Run xset dpms force standby. Your screen should blank, but the backlight will remain on. When the system is working well, you can unblank the screen almost instantly with a keypress. Wait a few seconds after the screen blanks, then press a key to wake it. Report whether you can recover from standby. Next, set dpms force suspend. This takes a bit longer to unblank, but still less than 2 seconds for me. Report whether you can recover from suspend. Next, xset dpms force off. This is the deepest sleep, and the backlight will likely be disabled. Report whether you can recover from off. |
Failed on all. |
|
| Back to top |
|
|
DarkAudit
n00b
Joined: 27 Mar 2021
Posts: 27
|
| Posted: Tue Aug 03, 2021 7:49 am Post subject: |
|
|
I posted a bug report as suggested, and... nothing. No suggestions, no requests for more information, nothing. It's still the same situation. The 5.12.* kernels work fine, but the 5.13.* kernels do not recover properly from the backlight shutting off. There was one instance where I caught the screen blanking but before the backlight shut off, and the display returned properly, but if the backlight shuts off, I still get a blank screen when it comes back on.
The complete lack of similar reports suggests operator error, but what? What besides following the wiki procedures of copying the previous kernel's .config and running make olddefconfig prior to compiling should I have done? Another machine also running ~amd64 with intel graphics has no such issues, and the kernel was updated the same way. |
|
| Back to top |
|
|
Hu
Administrator
Joined: 06 Mar 2007
Posts: 22443
|
| Posted: Tue Aug 03, 2021 3:40 pm Post subject: |
|
|
| An unusual local configuration might be triggering this, but it sounds like a bug to me. The system shouldn't become unrecoverable just because you set a configuration option incorrectly. You may need to bisect the kernel to find the specific commit that introduced this. That is unfortunate when the search range is v5.12..v5.13, but if no one has better ideas, that may be necessary. |
|
| Back to top |
|
|
swimmer
Veteran
Joined: 15 Jul 2002
Posts: 1330
Location: Netherlands
|
| Posted: Tue Aug 03, 2021 6:45 pm Post subject: |
|
|
I can confirm that this happens to me as well with *all* 5.13 kernels I've tried. I'm building all my kernels with a script I've created and this works pretty well on all other machines and even on the same machine with all other kernel versions including 5.14-rc*
The laptop in question is a Lenovo ideapad 530S-14ARR with the following specs:
| Code: |
AMD Ryzen 5 2500U with Radeon Vega Mobile Gfx
OpenGL: renderer: AMD Radeon Vega 8 Graphics (RAVEN DRM 3.42.0 5.14.0-rc4-git LLVM 12.0.1) v: 4.6 Mesa 21.2.0-rc2)
|
|
|
| Back to top |
|
|
Hu
Administrator
Joined: 06 Mar 2007
Posts: 22443
|
| Posted: Tue Aug 03, 2021 7:15 pm Post subject: |
|
|
| If it works on v5.14-rc*, then this was probably a kernel regression that was fixed upstream, but either not flagged for a backport (possibly because the fix was incidental to some other change, and not known to fix this issue) or has not been backported yet. If someone affected can find the commit that fixed this, that commit could be brought to the attention of the stable-kernel project for inclusion in future v5.13.x releases. |
|
| Back to top |
|
|
DarkAudit
n00b
Joined: 27 Mar 2021
Posts: 27
|
| Posted: Wed Aug 04, 2021 12:44 am Post subject: |
|
|
| Hu wrote: | | The system shouldn't become unrecoverable just because you set a configuration option incorrectly. |
That's the weird thing. other than no display, everything seems to run fine. It's like working with one's eyes closed. For example, I can ctrl-alt-f1 to a console, log in as root, and reboot. I just can't see what I'm doing. |
|
| Back to top |
|
|
Hu
Administrator
Joined: 06 Mar 2007
Posts: 22443
|
| Posted: Wed Aug 04, 2021 1:35 am Post subject: |
|
|
| swimmer says it works properly again in v5.14-rc*. Does that series fix your problem as well, DarkAudit? |
|
| Back to top |
|
|
DarkAudit
n00b
Joined: 27 Mar 2021
Posts: 27
|
| Posted: Wed Aug 04, 2021 2:27 am Post subject: |
|
|
| Hu wrote: | | swimmer says it works properly again in v5.14-rc*. Does that series fix your problem as well, DarkAudit? |
I'm on my way to work, so I'll have to emerge and compile in the morning to find out. |
|
| Back to top |
|
|
swimmer
Veteran
Joined: 15 Jul 2002
Posts: 1330
Location: Netherlands
|
|
| Back to top |
|
|
DarkAudit
n00b
Joined: 27 Mar 2021
Posts: 27
|
| Posted: Wed Aug 04, 2021 12:24 pm Post subject: |
|
|
Annnnd... confirmed.
Emerged latest git-sources.
Eselected that kernel version.
Copied 5.12.19 .config over.
Make olddefconfig.
Compiled. Installed. Told grub where to find it.
Rebooted to new 5.14 kernel. Logged into my XFCE session. Waited.
Screen blanked and backlight turned off. Moved mouse. Screen came back as it should.
Yay! |
|
| Back to top |
|
|
swimmer
Veteran
Joined: 15 Jul 2002
Posts: 1330
Location: Netherlands
|
| Posted: Wed Aug 04, 2021 12:30 pm Post subject: |
|
|
Yay indeed  |
|
| Back to top |
|
|
DarkAudit
n00b
Joined: 27 Mar 2021
Posts: 27
|
| Posted: Mon Aug 09, 2021 12:35 am Post subject: |
|
|
| Finally fixed as of gentoo-sources-5.13.9 with the ASSR patch. |
|
| Back to top |
|
|
Hu
Administrator
Joined: 06 Mar 2007
Posts: 22443
|
|
| Back to top |
|
|
|
Desktop Environments
