Bug 14218

Summary: Console graphics is not restored on server kill
Product: xorg Reporter: Joel <k00_fol>
Component: Driver/intelAssignee: Jesse Barnes <jbarnes>
Status: RESOLVED FIXED QA Contact: Xorg Project Team <xorg-team>
Severity: normal    
Priority: medium CC: ben.pineau, kdekorte
Version: git   
Hardware: Other   
OS: All   
Whiteboard:
i915 platform: i915 features:
Attachments:
Description Flags
Xorg.0.log
none
output of lspci -vvvxxxx prior to suspend
none
output of lspci -vvvxxxx after wakeup
none
diff output between fail and success case. none

Description Joel 2008-01-23 15:44:29 UTC
Created attachment 13888 [details]
Xorg.0.log

Test running the server with 
"X -config /root/xorg.conf.new"
and killing the server with Ctrl-Alt-Backspace results in a black screen (loss of video signal). 

The console still live in the background an it's possible to re-start the server (with graphics as expected) via some blind typing.

------------

Experienced with git 2008-01-23, on G35 graphics hardware.
x86_64 linux system
No strange entries is given in syslog, kernel.log or in dmesg
Comment 1 Gordon Jin 2008-01-24 01:33:52 UTC
I can also reproduce this on G965.
Comment 2 Kevin DeKorte 2008-01-25 06:31:41 UTC
Using todays drm and xf86-video-intel (Jan 25, 2008). This problem seems to have gone away.

I switched to init 3
ran startx
Hit Ctrl-Alt-Backspace

and I got my console back.

With Jan 24th version I don't think this worked. (didn't test that exact case, but I had issues switching to VT
Comment 3 Kevin DeKorte 2008-01-25 07:29:43 UTC
benpi and I have been chatting and we've found all kinds of weirdness with switch to console and when X is killed off

1. (G35 chipset) if I have VGA and HDMI->DVI cables plugged in switch to console does not work

2. remove hdmi->dvi cable (cable has no signal anyway see bug #13625) and switch to console works after reboot

3. When switch to console works, suspend the machine...and then wake up. After that switch to console does not work

4. we used lspci -vvvxxxx and got some dumps of some registers and values are clearly different after wakeup
Comment 4 Kevin DeKorte 2008-01-25 07:32:03 UTC
Created attachment 13937 [details]
output of lspci -vvvxxxx prior to suspend
Comment 5 Kevin DeKorte 2008-01-25 07:32:32 UTC
Created attachment 13938 [details]
output of lspci -vvvxxxx after wakeup
Comment 6 Joel 2008-01-25 07:33:37 UTC
I'm not 100% sure how yet, but I can NO LONGER re-produce this bug.

The only thing I can think of that should matter is that I fixed a trivial
module loading problem. (i.e. ran depmod -a)

Full(?) changelog on my system since yesterday 
* running "depmod -a"
* Installing dbus-1.1.4
* Installing current Cmake
* re-installing module-init-tools

By making a diff output of my Xorg logs, there's only one line that stands out
in the recent logs that wasn't there in yesterdays failures: 
(II) [drm] loaded kernel module for "i915" driver.

Also the drmOpenDevices now fail two times before succeding, yesterday all
three resulted in " 9 (OK)"

Attaching diff output to make all differences easy to spot
(changed lines are marked "!", added and removed "+" or "-")
Comment 7 Joel 2008-01-25 07:34:47 UTC
Created attachment 13939 [details] [review]
diff output between fail and success case.

Created with: "diff -b failing working -p > fail_to_working_log "
Comment 8 Kevin DeKorte 2008-01-25 07:37:40 UTC
Actually the HDMI->DVI no display bug on G35 is bug # 13968
Comment 9 Benjamin Pineau 2008-01-25 08:06:04 UTC
I can reliably, at will, reproduce the bug (with G35 hardware) *if and only if* my  machine was suspended-to-ram before.

If the machine was not suspended (ie. after a fresh reboot), stopping or restarting Xorg works well and the console is not broken. 

I have others post-suspend-to-ram similar symptoms, by the way, like switch to console won't work. This is reported in bz #1429. ( http://bugs.freedesktop.org/show_bug.cgi?id=14249 ).
Comment 10 Gordon Jin 2008-01-27 21:48:12 UTC
I'm closing as it seems everyone can't reproduce this bug with a fresh boot.
Comment 11 Kevin DeKorte 2008-01-28 05:43:44 UTC
Gordon, that is incorrect. I can reproduce this bug if I have have a both my VGA monitor and another monitor connected via the HDMI->DVI adapter plugged in on my G35. If I don't have the HDMI monitor connected it works fine.

Comment 12 Gordon Jin 2008-01-28 18:09:47 UTC
Kevin, thanks for your effort on this. But "VGA + HDMI->DVI" is a different issue than the original report. I don't want the comments in this bug distracts readers to track the new issue. Could you file a new bug for the specific new issue? That will makes thing clear. Thanks.

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.