Bug 701009 - Suspend to RAM hangs on resume
Summary: Suspend to RAM hangs on resume
Keywords:
Status: CLOSED CANTFIX
Alias: None
Product: Fedora
Classification: Fedora
Component: pm-utils
Version: 15
Hardware: Unspecified
OS: Unspecified
unspecified
unspecified
Target Milestone: ---
Assignee: Jaroslav Škarvada
QA Contact: Fedora Extras Quality Assurance
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2011-04-30 11:14 UTC by Philipp Dörfler
Modified: 2011-05-17 16:32 UTC (History)
4 users (show)

Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Clone Of:
Environment:
Last Closed: 2011-05-17 16:32:51 UTC
Type: ---


Attachments (Terms of Use)
The last 280 lines of /var/log/messages (26.75 KB, text/plain)
2011-04-30 11:14 UTC, Philipp Dörfler
no flags Details
contents of /var/log/pm-suspend.log (7.16 KB, text/plain)
2011-04-30 11:17 UTC, Philipp Dörfler
no flags Details
there's an interesting ACPI-entry related to nvidia in Xorg.0.log (19.87 KB, text/plain)
2011-04-30 11:20 UTC, Philipp Dörfler
no flags Details
Screenshot of weird desktop after resume (250.87 KB, image/png)
2011-05-02 14:13 UTC, Philipp Dörfler
no flags Details
/var/log/messages after resume (19.82 KB, text/plain)
2011-05-02 14:21 UTC, Philipp Dörfler
no flags Details
Fancy screencast of even facier nautilus (951.54 KB, video/webm)
2011-05-02 14:38 UTC, Philipp Dörfler
no flags Details

Description Philipp Dörfler 2011-04-30 11:14:59 UTC
Created attachment 495939 [details]
The last 280 lines of /var/log/messages

Description of problem:
When resuming from suspend-to-ram, gnome shell get's resumed but after entering the user's password, only the mouse cursor and the bar at the top (only displaying the date and the user name along with a locked symbol) is shown. One can move the cursor around, otherwise nothing happens. There is no background wallpaper, the background is completely black.

Pressing Ctrl + Alt + Backspace brings X to a halt completely. Switching to the ttys does not work. Logging in remotely via ssh still works, albeit the initial log-in took a long time directly after entering the password.

Version-Release number of selected component (if applicable):
Linux shodan 2.6.38.3-18.fc15.x86_64 #1 SMP Fri Apr 22 13:24:23 UTC 2011 x86_64 x86_64 x86_64 GNU/Linux

How reproducible:
Haven't checked yet, will verify later.

Steps to Reproduce:
1. In gnome shell click on the user's name, then on "suspend"
2. watch the PC suspending to RAM, the PC's LEDs are blinking
3. press the PC's power button to initiate the resume
4. wait for the resume to complete, then enter your password
5. welcome to a broken Gnome Shell
6. File a bug

Expected results:
Resume should work, IMHO entering the password should not be required, but that's not the point of this bug.
All services and applications should be restored correctly.

Additional info:
After trying to kill X and gdm I've looked into /var/log/messages and a kernel oops happened.

Comment 1 Philipp Dörfler 2011-04-30 11:17:16 UTC
Created attachment 495940 [details]
contents of /var/log/pm-suspend.log

The version of pm-utils is:
pm-utils-1.4.1-6.fc15.x86_64

Comment 2 Philipp Dörfler 2011-04-30 11:20:07 UTC
Created attachment 495941 [details]
there's an interesting ACPI-entry related to nvidia in Xorg.0.log

Comment 3 Jaroslav Škarvada 2011-05-01 20:55:05 UTC
According to backtrace it seems to hang in nvidia module. Could you try without nvidia module (e.g. with nouveau)?

Comment 4 Philipp Dörfler 2011-05-01 21:30:04 UTC
thanks for the fast reply. Tomorrow I'll try with nouveau instead of nvidia. 

Could you give me a pointer what's the quickest way to temporarily disable the nvidia module?

Thanks in advance

Comment 5 Jaroslav Škarvada 2011-05-01 22:02:11 UTC
IIRC remove rdblacklist=nouveau from your /etc/grub.conf (if used), remove 'blacklist nouveau' from the /etc/modprobe.d/blacklist.conf and add there 'blacklist nvidia', reboot and check that nvidia is not loaded:
# lsmod | grep nvidia

Comment 6 Philipp Dörfler 2011-05-02 14:11:24 UTC
Thanks for the hint. With nouveau instead of nvidia, suspend seems to work a bit better. After resume though, the desktop wallpaper is replaced by the application icons of gnome.

Comment 7 Philipp Dörfler 2011-05-02 14:13:37 UTC
Created attachment 496271 [details]
Screenshot of weird desktop after resume

Comment 8 Philipp Dörfler 2011-05-02 14:21:42 UTC
Created attachment 496273 [details]
/var/log/messages after resume

Attached /var/log/messages for the sake of completeness.

Probable unrelated but weird nonetheless:
rtkit-daemon reports it's stuff in a UTC timesone instead of the UTC+2 timezone of the others...

Comment 9 Philipp Dörfler 2011-05-02 14:35:40 UTC
Shortly afterwards I started nautilus and then the weird wallpaper was replaced by a lifesize nautilus! It was even updated. When I selected something in nautilus and moved the window, the wallpaper was updated as well and reflected the new selection. I was curious and resized nautilus wondering what would happen. And what happened? The system freezed somewhat. Strange graphical glitches appeared. I tried to record a screencast but I dunno whether the file is readable. SSHin into the machine and remotely rebooting it didn't work either.

So... is this considered a seperate bug or does it belong to this one?

Comment 10 Philipp Dörfler 2011-05-02 14:38:41 UTC
Created attachment 496276 [details]
Fancy screencast of even facier nautilus

file seems to be readable. Unfortunately the recording stops when the real fun part begins.

Comment 11 Jaroslav Škarvada 2011-05-02 14:56:31 UTC
Thanks for info, please provide /var/log/Xorg.0.log and output of the following command after the resume:
# pm-utils-bugreport-info.sh

Comment 12 Philipp Dörfler 2011-05-02 16:22:27 UTC
Hm... things happen. I thought I double-checked lsmod | grep nvidia but it seems I didn't. I'm sorry, I just discovered that despite blacklisting nvidia it still gets loaded and causes this sporadic (albeit reproducible in 90% of the cases) bug (according to Xorg.0.log).

In addition to the steps you suggested, I prefixed the line "blacklist nouveau" in /etc/modprobe.d/blacklist-nouveau with '#', too. Still, both nvidia and nouveau are loaded. I verified that nvidia is explicitly blacklisted and grub.conf looks fine to me, too. I'm a bit short of ideas of what to try...

Tomorrow I'll investigate how to disable nvidia further.

Thanks for your efforts!

Comment 13 Jaroslav Škarvada 2011-05-02 16:41:09 UTC
Strange, maybe it is deps for some other non-blacklisted third-party module. As a last resort you can try to uninstall the nvidia SW or rename/delete the module.

Comment 14 Philipp Dörfler 2011-05-17 16:10:16 UTC
I erased nvidia and now with nouveau, suspend and resume seems to work as expected. Thanks for the suggestions so far :)

Comment 15 Jaroslav Škarvada 2011-05-17 16:32:51 UTC
Thanks for info. I cannot help with the binary blob - it should be reported to nvidia. Feel free to reopen if the problem re-appear with nouveau.


Note You need to log in before you can comment on or make changes to this bug.