You are not logged in.
Hi everyone,
I installed the Linux 3.8.0-2 64 bits kernel from [testing] to see if another issue I was having with [core] Linux 3.7.9 64 bits was solved ("aplay -l" not listing the HDMI port, however the gnome sound manager does display an HDMI port, not usable of course because unknown to alsa).
However my laptop behaves quite badly now, when suspend/resume (suspend to RAM).
It's really annoying because I'm used to reboot my computer every few monthes and keep my workspaces live... and I usually use quite a lot.
Here is the description of the problem:
- Boot laptop, do some work, anything.
- Suspend
- Hit keyboard to resume... the machine wakes up BUT the screen remains black, no backlit and the screen dones not seem to display anything (usually even without backlit at least the shapes can be seen).
So the machine is started and I can see nothing. I am even able to login (I recognize the gnome classic login manager beeps), launch a terminal with keyboard shortcuts and do "sudo shutdown -H now". The HDD activity indicates the system is shutting down, the fans stop, however the power remains (power LEDs alive). I have to press power button 5s to actually stop the machine.
After reboot, reading logs I can confirm I actually logged, opened the terminal and shutdowned the machine.
Conclusion:
- Power management related issue
- Screen backlit issue
and this did not happen with previous kernel.
Hardware & drivers:
- Laptop Acer Aspire 5920G
- Video card nVidia 8600M GT, driver nouveau and all gl-related libs from [testing].
If it's a nouveau-related issue, maybe also related to ALSA still not listing the HDMI port ?
Do you have any advice ?
Regards,
marzoul
Last edited by marzoul (2013-03-05 09:17:35)
Offline
Hi again.
Now, downgrading from testing/linux-3.8.0-2 to core/linux-3.7.9-2 completely solved the suspend issue.
I downgraded ONLY the linux package.
Conclusion: testing/linux-3.8.0-2 has a real stability issue, however I am not really able to locate where and why.
Should I open a bug report somewhere ?
Offline
hi
having the same problem with intel HD3000...
Offline
Yesterday I did install testing/linux-3.8.1-1 (x86-64) because the official Linux 3.8.1 changelog indicates resolved bugs about screen backlit upon resume.
Conclusion: the problem is still there.
The screen stays still completely black upon resume with testing/linux-3.8.1-1 (x86-64).
Other observation:
alsa -l still does not display the HDMI port. However this is probably more a driver bug (nouveau) than a kernel bug.
Offline
I would like to report this as a bug in the Archlinux bugs section.
Before doing that, can anyone advise about the relevant information to attach ?
I don't have much available, except saying the screen remains completely black upon resume...
I could dig in, say, whatever system log would be useful... Does anybody known which one / how to "decrypt" them ?
(I say "decrypt" because I was used to Ubuntu's syslog, but there in /var/log/journal the files are not clear text, so I really don't know what to do with them, and the wiki does not help there...)
Offline
Just read the Linux 3.8.2 changelog: nothing related to my problem is mentioned.
So I'll just skip testing this version.
I've also updated the discussion title to more accurately represent the problem.
Last edited by marzoul (2013-03-05 13:27:21)
Offline
I can confirm the same issue.
However, in my case, plugging in the power cable of the laptop (or re-inserting) will reactivate the screen.
Samsung QX-412 on Intel HD3000.
Last edited by koz (2013-03-05 11:14:24)
Offline
I could dig in, say, whatever system log would be useful... Does anybody known which one / how to "decrypt" them ?
(I say "decrypt" because I was used to Ubuntu's syslog, but there in /var/log/journal the files are not clear text, so I really don't know what to do with them, and the wiki does not help there...)
try man journalctl
Disliking systemd intensely, but not satisfied with alternatives so focusing on taming systemd.
(A works at time B) && (time C > time B ) ≠ (A works at time C)
Offline
However, in my case, plugging in the power cable of the laptop (or re-inserting) will reactivate the screen.
Samsung QX-412 on Intel HD3000.
Ok I'll try that trick soon.
However if the bug involves a driver (I use Nouveau on Nvidia), that might not make it.
And the bug would still be open!
try man journalctl
Oww now I understand, systemd does handle the journal. Now I know where I should have searched in the wiki: systemd. That wiki search engine gives no real hint as where to go when searching 'journal', 'syslog', etc.
Thanks!
Offline
For me it has been a hit or miss situation lately. When it happens, usually closing and reopening the lid once more fixes the problem.
ᶘ ᵒᴥᵒᶅ
Offline
Did test with testing/linux-3.8.2-1 x86_64, no change since 3.8.1:
After resume, the screen stays black.
When it happens, usually closing and reopening the lid once more fixes the problem.
And this does NOT work for me...
I read the system journal rapidly, let's say it seems ok, on 3.8.2 the kernel pretends to do things about backlit... I'll dig more deeply when I have time...
Also, as I said before, I think it is NOT a simple backlit problem, because when I manually turn backlit off (keyboard shortcut) I can still see what is on the screen. But when resume on 3.8.x, NOTHING can be seen on the screen. As if it were not even part of the laptop.
I could also find what I think is a bug and always have been there: a device "BMA150 accelerometer" is found, and there is absolutely no reason such a device is present in my machine.
The journal line is:
kernel: input: Acer BMA150 accelerometer as /devices/virtual/input/input11
Don't know if this could be related to alsa not finding my hdli port ? (device ID conflict, whatever...)
Offline
With kernel 3.8.2, the screen is dead permanently for me as well on resume, closing/reopening does nothing.
ᶘ ᵒᴥᵒᶅ
Offline
Hi guys.
Using the stable stock arch kernel, closing the lid in console or X works as expected. Console / X session comes back in around 1 seconds
But using the testing 3.8.2 (was using the -ck kernel):
* In console (no X)
If I close the lid or do a systemctl suspend, the laptop goes to suspend mode (entry in the journal etc .. everything is ok).
Then opening the lid, the system wakes up (I can access it via ssh or the previous session is restored) quite instantaneously. However it takes roughly more than a minute before I can see the screen of the laptop....
* In X (started with startx, launching a simple fvwm)
I close the lid, everything goes fine to suspend. I open the lid or press the power button ... and then again remotely no problem, the system is back online. However locally the screen never goes back ... I have to reboot! Killing X / telinit 3 doesn't do anything
I did litemotiv's trick and ... it works in console but you have to wait the system goes back to suspend mode. On X ... well it doesn't
Using journalctrl I see this event
Mar 07 13:34:55 30L3 kernel: xhci_hcd 0000:00:14.0: WARN Event TRB for slot 1 ep 0 with no TDs queued?
I'm using arch 64bits on a Lenovo x230 Tablet with KMS
-=<>=-
Archlinux French site : http://www.archlinux.fr
Offline
I noticed that I am not in the group power.
This has no effect right now on 3.7.10, under Gnome Shell desktop (good old fallback style).
Is this group relevant for the current issue? @litemotiv are you in this group?
I'll try and reboot under Linux 3.8.2, in this group.
Offline
I noticed that I am not in the group power.
This has no effect right now on 3.7.10, under Gnome Shell desktop (good old fallback style).
Is this group relevant for the current issue? @litemotiv are you in this group?
I'll try and reboot under Linux 3.8.2, in this group.
I found out that this bugreport applies to my situation: https://bugs.archlinux.org/task/34153
Meaning that manually echoing a non-zero value to /sys/class/backlight/intel_backlight/brightness (as root) brings back the screen.
For good measure, these are my groups: wheel video audio optical power users
ᶘ ᵒᴥᵒᶅ
Offline
I'm not in the group power only in users and plugdev (for fingerprint-gui) groups.
Wonder if switching to this group will solve something
-=<>=-
Archlinux French site : http://www.archlinux.fr
Offline
Hi again, now I'm in the power group and I logged everything.
I uploaded the logs on my server: http://marzoul.no-ip.info/publicshare/3.8.x.tar.xz
The archive contains a little readme about the procedure.
The interesting logs are logs-3.7.10-1 and logs-3.8.2-1 which correspond to sessions including boot, suspend and resume.
The "-nodate" logs are obtained by removing the beginning of the lines to use diff.
The "-downgrade" logs are from the session when I downgraded Linux, if interesting in any way.
Well, I read them and compared them a bit, found nothing terrible.
In particular, in the 3.8.2 session, during resume the kernel says backlit is restored.
I tried closing and reopening the lid, the kernel says it is exactly that I did, the screen is removed and restored in the system.
Si I still don't know...
EDIT: the archive is updated to linux 3.8.3 logs and I will update it until the bug is solved.
Last edited by marzoul (2013-03-16 16:24:44)
Offline
It looks like it broke in the 3.8 kernel; fedora is reverting part of the commit that leads to the breakage - perhaps worth a shot?
Offline
This is about Intel driver and it is now fixed in linux 3.8.3 according to the changelog.
It should have no impact on NVidia chips.
I confirm the bug still exists with this 3.8.3 version.
Offline
For me(with intel gpu), then v3.8.3 blanks my screen during normal startup, whereas v3.8.2 didn't have any issues and which i've now downgraded to again...
Offline
abrouwers wrote:This is about Intel driver and it is now fixed in linux 3.8.3 according to the changelog.
Where do you see anything about it being fixed ? It looks like it's still affected; fedora is still applying their patch, which fixes it for me.
Offline
Where do you see anything about it being fixed ? It looks like it's still affected; fedora is still applying their patch, which fixes it for me.
In the Linux 3.8.3 changelog, search for i915, there is load of bug fixes about Intel graphics, for resume and reactivation of parts.
https://www.kernel.org/pub/linux/kernel … eLog-3.8.3
If your bug is still not solved, these bugfixes weren't about this exctly... too bad :-S
Offline
With Linux 3.8.4-1 x86-64 the bug is still there: the screen does not wake up upon resume.
EDIT: ditto for Linux 3.8.5
Last edited by marzoul (2013-03-31 21:42:27)
Offline
This problem should have been solved with 3.8.6.
Offline
Tested yesterday: no it is not solved for me!
Additionally the Linux changelog says nothing remotely related to it.
Actually the Nouveau developpers advised me to try Linux 3.9rc, but I still not tried due to compilation time.
Offline