You are not logged in.
Hello to everyone!
Recently when I switch from my Xorg session started with lightdm in VT7 to another Xorg session started with xinit on VT2-6, sometimes causes crash on Xorg allocated on VT7. I have a laptop with hybrid graphics but I only use the Intel card.
This is the command i use to start the xorg session with xinit:
export HOME=/home/sandia-sabrosa/games
xinit /usr/bin/openbox-session -- :3 vt$XDG_VTNRAnd my xorg session in VT7 is xfce4 started with lightdm and lightdm-gtk-greeter; and I use compiz as window manager.
I use to reproduce the crash and these are the logs: Dmesg, Journalctl, Xorg log.
Last edited by i_love_r34 (2017-08-05 00:00:14)
Offline
The Xorg log does not indicate a segfault or abort, but looks like from a running server. Either this is the wrong log or you've to specify "crash" a bit. Using compiz on VT7 and another GL client on the other server could lead to a bug in the intel driver, messing up GL contexts (typically one or both outputs would stall, either you get a static image or a black frame) - killing compiz might get you back the (unmanged then) X11 server on VT7 "reacting" then.
Offline
Apparently compiz is the "problem".
Another information I would like to add is when Xorg crash on VT7 I get this on the screen, this is only visible around 2 to 3 seconds before lightdm send me back to the login screen.
Offline
Can you read anything on that photo? Please type the actual text.
before lightdm send me back to the login screen
Might as well just mean that the session keeping process crashed (X then terminates expectedly)
We'll need an actual error message here to know *what* actually crashes.
As a workaround, you could try SIGSTOP'ing compiz and SIGCONT'inuing it whenever returning to its VT.
Offline
The photo isn't show any text, is like random pixels with diferents colors. Recently I reproduce the error and get the same image, after I get the error I reboot the system and the "image" keeps until reboot. This is the photo I take.
I tried sigstop and sigcont with the command kill and the results are with all the interface in VT7 "freeze" until I send the sigcont command.
Another information I would like to add are the lightdm logs (lightdm.log, x-0.log, seat0-greeter.log).
lightdm.log
seat0-greeter.log:
** Message: Starting lightdm-gtk-greeter 2.0.2 (Oct 17 2016, 14:22:50)
** Message: [Configuration] Reading file: /etc/lightdm/lightdm-gtk-greeter.conf
** (lightdm-gtk-greeter:1120): WARNING **: [PIDs] Failed to execute command: upstart
(lightdm-gtk-greeter:1120): Gtk-WARNING **: Drawing a gadget with negative dimensions. Did you forget to allocate a size? (node menubar owner GreeterMenuBar)x-0.log:
X.Org X Server 1.19.3
Release Date: 2017-03-15
X Protocol Version 11, Revision 0
Build Operating System: Linux 4.9.11-1-ARCH x86_64
Current Operating System: Linux illuminati 4.11.12-1-ck #1 SMP PREEMPT Sun Jul 23 15:42:02 PDT 2017 x86_64
Kernel command line: BOOT_IMAGE=/vmlinuz-linux-ck root=UUID=029999b9-6275-4f93-9eb2-331861b0d56f rw nmi_watchdog=0 quiet splash
Build Date: 07 April 2017 05:42:48PM
Current version of pixman: 0.34.0
Before reporting problems, check http://wiki.x.org
to make sure that you have the latest version.
Markers: (--) probed, (**) from config file, (==) default setting,
(++) from command line, (!!) notice, (II) informational,
(WW) warning, (EE) error, (NI) not implemented, (??) unknown.
(==) Log file: "/var/log/Xorg.0.log", Time: Thu Aug 3 23:21:15 2017
(==) Using config file: "/etc/X11/xorg.conf"
(==) Using system config directory "/usr/share/X11/xorg.conf.d"
(II) AIGLX: Suspending AIGLX clients for VT switchTo be honest, I don't know what happening here.
Offline
The log seems a bit capped?
The behavior od SIGSTOP and SIGCONT is expected and intended - the idea is to prevent compiz from doing anything as long as the GL context is on the other VT. The question is whether this stabilizes the situation.
Offline
I tried again with SIGSTOP and SIGCONT and everything was nice until I send the SIGCONT command from the terminal in VT2 and switching again I get the same error behavior. I tried with another VT without Xorg and the same error.
The log seems a bit capped?
A bit capped? Did you refer to the log that isn't indicate errors?
Offline
"capped" like it ends pretty early and does not contain most of what you'd expect to see in an Xorg log.
sleep 10; pkill -SIGCONT compizand switch the VT before the process continues (but this might be a dead end approach anyway)
Just to be sure (as you didn't really confirm it) - killing (and restarting?) compiz "resolves" the problem?
Offline
Killing compiz doesn't resolve the problem on the fly. I need to start the session with another wm like xfwm4 to avoid that problem. I tried the command you put and sometimes works fine but sometimes switching to the VT7 I get the same error before the command execution of SIGCONT.
And other thing i would like to add is the solution I find:
I installed the xf86-video-intel, using the default configuration I get the same behavior; but using DRI2 and UXA, Xorg on VT7 doesn't "crash" anymore. I tried around 30 minutes to reproduce the error and nothing.
I want to add as observations are the desktop is so slow to start, around 40 seconds (with the modesetting driver was around 10 seconds); and the compiz animations are a little slow but thats have a easy fix, CCSM > General Options > Display settings > Refresh rate to maximum (200).
I think the error was in the modesetting driver, but i'm not 100% sure.
Offline