Bug 689044 - [NV86] Suspend with attached external monitor: failure.
Summary: [NV86] Suspend with attached external monitor: failure.
Keywords:
Status: CLOSED UPSTREAM
Alias: None
Product: Fedora
Classification: Fedora
Component: xorg-x11-server
Version: 15
Hardware: x86_64
OS: Linux
unspecified
medium
Target Milestone: ---
Assignee: Ben Skeggs
QA Contact: Fedora Extras Quality Assurance
URL:
Whiteboard: [cat:suspend]
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2011-03-18 22:09 UTC by Marco Vittorini Orgeas
Modified: 2018-04-11 06:53 UTC (History)
7 users (show)

Fixed In Version:
Clone Of:
Environment:
Last Closed: 2011-11-22 16:34:24 UTC
Type: ---
Embargoed:


Attachments (Terms of Use)
pm-suspend.log (5.60 KB, text/plain)
2011-03-18 22:10 UTC, Marco Vittorini Orgeas
no flags Details
/var/log/messages (230.48 KB, text/plain)
2011-03-18 22:11 UTC, Marco Vittorini Orgeas
no flags Details
Xorg.0.log (79.17 KB, text/plain)
2011-03-18 22:11 UTC, Marco Vittorini Orgeas
no flags Details
dmesg output (124.06 KB, text/plain)
2011-03-26 12:40 UTC, Marco Vittorini Orgeas
no flags Details
dmesg output when laptop monitor is off (17.24 KB, text/plain)
2011-05-04 20:25 UTC, Marco Vittorini Orgeas
no flags Details
dmesg output when laptop monitor is off (1.38 MB, text/plain)
2011-05-23 12:30 UTC, Marco Vittorini Orgeas
no flags Details
Xorg.0.log updated (76.43 KB, text/plain)
2011-05-23 12:35 UTC, Marco Vittorini Orgeas
no flags Details

Description Marco Vittorini Orgeas 2011-03-18 22:09:42 UTC
Description of problem:
I've a second monitor (VGA) attached to the laptop.
Xorg server, with a second monitor attached, defaults to span the visible area on both screens (i.e. the LDVS screen of the laptop and the external attached monitor).
I work on the VGA so I put off the LDVS one through Preferences --> Monitor (like I do in F14).   
If I issue a suspend from the menu on the right upper corner, the system starts to suspend,the monitor goes blank with the mouse pointer freezed, the laptop led start blinking and unfortunately keeps blinking with the fan running and the system is freezed and the only meaning to reboot is by shut down with the power button.

If I suspend WITHOUT touching the display preferences (as above), the laptop regularly suspend and I can also resume.

Comment 1 Marco Vittorini Orgeas 2011-03-18 22:10:29 UTC
Created attachment 486329 [details]
pm-suspend.log

Comment 2 Marco Vittorini Orgeas 2011-03-18 22:11:02 UTC
Created attachment 486330 [details]
/var/log/messages

Comment 3 Marco Vittorini Orgeas 2011-03-18 22:11:36 UTC
Created attachment 486331 [details]
Xorg.0.log

Comment 4 Marco Vittorini Orgeas 2011-03-18 22:12:16 UTC
Smolt profile:
http://www.smolts.org/client/show/pub_cc9c7be0-2d64-43b6-8ee4-65f9565de32e

Comment 5 Matěj Cepl 2011-03-23 23:07:28 UTC
Could we get also output of dmesg after suspend/resume cycle (probably not with multi-head configuration), please?

Thank you

Comment 6 Marco Vittorini Orgeas 2011-03-26 12:40:53 UTC
Created attachment 487732 [details]
dmesg output

Attaching dmseg output of a suspend/resume cycle with both monitor on.
As already said, as soon as I put the LDVS monitor off, suspend will fail.
If you need dmesg output when LDVS is off, don't know maybe you can suggest a command to "live" buffer its output into some file...I haven't investigated in this direction yet.

Comment 7 Marco Vittorini Orgeas 2011-04-29 14:13:43 UTC
As of today, the problem still persists.
This is very unfortunate because it prevents me to upgrade to F15 for day to day computing, as I work on the VGA monitor and I need to suspend/resume quite often during the day!
Do you need more informations about it ?

Comment 8 Marco Vittorini Orgeas 2011-05-04 20:25:34 UTC
Created attachment 496910 [details]
dmesg output when laptop monitor is off

Comment 9 Marco Vittorini Orgeas 2011-05-23 12:30:46 UTC
Created attachment 500413 [details]
dmesg output when laptop monitor is off

Added drm.debug=14 log_buf_len=16M options to kernel.

Comment 10 Marco Vittorini Orgeas 2011-05-23 12:35:59 UTC
Created attachment 500415 [details]
Xorg.0.log updated

Comment 11 Marco Vittorini Orgeas 2011-05-23 12:40:46 UTC
I've no experience with the nouveau source code, but scrolling the dmesg output and looking at the drm-nouveau output,it seems to me that there is no evidence that the nouveau driver is to blame for the problem: should the bug be moved to some other component ?

Comment 12 csirkus 2011-05-24 15:24:02 UTC
I can confirm that this is not a nouveau bug. I have the same issue with the r600g driver.

Comment 13 Marco Vittorini Orgeas 2011-05-31 13:42:51 UTC
I made a few tests:
1. The wrong behaviour is identical on XFCE,LXDE and KDE spins.
2. If x11 nouveau driver is replaced with the official binary driver from nvidia, I can suspend/resume successfully.
This is probably due to the fact that nvidia driver doesn't use at all the X11 randr module to manage multiple monitors.It does rely to its own internal module to manage multiple attached displays.
3. Because of precedent point I am quite confident that the bug is not kernel related.
4. I have even tested the latest ubuntu release, which suffers the same problem.Interestingly enough, ubuntu in his latest version has update xorg 1.9 -> 1.10 http://www.ubuntuupdates.org/pm/xorg-server
5. F14 behaves correctly.
6. F15 has update xorg to 1.10 https://admin.fedoraproject.org/updates/xorg-x11-server?_csrf_token=b946fb964f1dc97cc205219925006ffbd5a45d12

Thus I feel to say: this is definitively a xorg-x11-server-Xorg server bug. More specifically the issue is related to the randr portion of x11 server.
Enough to suggest to proceed with a bisection to find exactly where the problem lies.

In the meanwhile to use F15, I have found that if the laptop lid is put "down enough" to shut it off when booting (i.e. on the first X11 start) until you log-in inside your system, than you can suspend/resume and move the lid without problems even after the first suspend/resume cycle.

Comment 14 Vitaliy 2011-08-10 18:50:16 UTC
I am so glad to find out this report. I have exactly the same issue. At least I am not alone. Unfortunately I couldn't find where to start from to trace this issue. Thanks Marco for this bugreport!

Comment 15 Marco Vittorini Orgeas 2011-11-22 16:34:24 UTC
Upgrading to linux 3.1, nouveau problems are gone:

http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=646057


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