I ran across this in testing and the above solved it for me... Odd.
| LinHES Forums http://forums.linhes.org/ |
|
| R5.5 Xorg Soft Lockup upon Reboot http://forums.linhes.org/viewtopic.php?f=1&t=18485 |
Page 1 of 1 |
| Author: | akulcsar [ Tue Jul 08, 2008 7:14 pm ] |
| Post subject: | R5.5 Xorg Soft Lockup upon Reboot |
I'm getting a weird hang whenever I try to reboot my R5.5 server. When attempting a reboot, the runlevel switch begins normally. When it reaches "Attempting to kill all other processes," Xorg hangs with a soft lockup. While initiating the reboot process from console 1, I get a message similar to the following every ten seconds or so: Code: BUG: soft lockup - CPU#0 stuck for 11s! [Xorg:3488] PID: 3488, comm: Xorg EIP: 0060:[<c0568f1a>] CPU: 0 EIP is at _spin_unlock_irq+0xa/0x30 ... The message goes on to provide a mini-dump of the process. The message varies a bit as xorg tries to unstick itself. It never does...I've let it sit for about an hour before giving up and pushing the reset button. If, instead of just conducting a reboot as normal, I do part of the process manually, then the box reboots smoothly: Code: # telinit 1
# pkill xinit # reboot VESA drivers do not load properly (odd, I thought). I have tried an upgrade to the current xserver-xorg; that did not help, either. This was a manual install into /dev/hda1 (all files into one partition). Auto-detected driver is via (old S3 Unichrome chip). CD was a proper burn and passed the checksum verification. Nothing like this in previous versions of KnoppMyth. Everything else seems to work just fine. Ideas? |
|
| Author: | tjc [ Tue Jul 08, 2008 7:18 pm ] |
| Post subject: | |
Have you checked the APIC and ACPI settings in your BIOS? |
|
| Author: | akulcsar [ Tue Jul 08, 2008 8:02 pm ] |
| Post subject: | |
tjc: Feeling really stupid right now...poked around the forum and the wiki and haven't found anything useful. How should they be set? |
|
| Author: | cecil [ Tue Jul 08, 2008 10:35 pm ] |
| Post subject: | |
Code: vi /etc/lilo.conf Find the append line and addCode: hpet=disable Write and quitCode: lilo -v After it comes up, reboot again. Does it work?
reboot |
|
| Author: | akulcsar [ Tue Jul 08, 2008 11:22 pm ] |
| Post subject: | |
No...still hanging on reboot. I'm working on getting my diskless frontends up and running. Will post whether or not they are affected, too. |
|
| Author: | akulcsar [ Wed Jul 09, 2008 12:00 am ] |
| Post subject: | |
Well, this is interesting.... Diskless frontend with an Epia MII 10000 produces the same result. VESA drivers do not function properly; openchrome drivers hang with a similar response. Odd. Not the end of the world; just really odd. |
|
| Author: | cecil [ Wed Jul 09, 2008 2:41 pm ] |
| Post subject: | |
I ran across this in testing and the above solved it for me... Odd.
|
|
| Author: | sparky34 [ Thu Jul 10, 2008 1:23 pm ] |
| Post subject: | |
cecil wrote: Code: vi /etc/lilo.conf Find the append line and addCode: hpet=disable Write and quitCode: lilo -v After it comes up, reboot again. Does it work?reboot I'm having the same issue and trying the 'hpet=disable' workaround hasn't worked for me. I get the Xorg soft lockup on 2 systems. One is an EPIA M10000 and the other an ECS system. I first noticed when Mythwelcome goes to shutdown it causes both the master frontend/backend and the slave backend to enter this state. I've also seen it happen with 'halt -p', 'reboot', 'shutdown -h now', etc. Since it's happening in Xorg is there any sort of debug we can gather that might help pinpoint this? I'd imagine that a lot of people might end up seeing this issue I'm just clueless as to where to begin. |
|
| Author: | akulcsar [ Thu Jul 10, 2008 9:54 pm ] |
| Post subject: | |
sparky34: You can watch the shutdown or restart progress by switching to a text console (e.g. Ctrl-Alt-F1), logging in, and initiating it from there. I gave a snippit in my original post, but you'd be able to see the full memory dump in all of its glory.  Scouring Google and xorg's web site has not brought me much luck, though.
|
|
| Author: | akulcsar [ Thu Jul 10, 2008 10:02 pm ] |
| Post subject: | |
HA. OK...one of my frontends has an nVidia video chipset in it (6100, I believe; could be a 6150). This problem DOES NOT manifest itself with the nVidia drivers, and the hpet=disable kernel flag is not needed. Running it diskless, though that should not matter. So, I'm guessing the current beta incarnation of Xorg hates older video chipsets?
Anyone using either a generic VESA driver (yuck), the opencrhome driver, or the via driver getting this version of xorg to work? |
|
| Author: | chunter2 [ Thu Jul 10, 2008 10:08 pm ] |
| Post subject: | |
I had this problem in testing too. I noticed it with onboard intel graphics chips. Changing the driver in xorg.conf to vesa instead of intel worked for me. I leave my boxes on all time time so it's not a major issue for me. Not the optimal solution since the intel driver is much faster than the standard vesa driver. |
|
| Author: | iemai [ Fri Jul 25, 2008 5:24 am ] |
| Post subject: | Workaround |
I have the same problem on an EPIA SP-13000. But I have a workaround. It seems that the kernel has some problem with Xorg, so I wrote this file: /etc/rc6.d/K10xinit with permissions 755 which contains: Code: #!/bin/bash
pkill xinit So when you reboot or poweroff the first thing the system do is to kill the X server and on my system doesn't hang anymore on reboot or powerdown. |
|
| Author: | akulcsar [ Fri Aug 01, 2008 9:19 pm ] |
| Post subject: | |
hrm...intersting fix, iemai. I would not have expected something that simple to work, but it did for all of my boxes that were experiencing the problem. To be complete, you'd want to add it to rc0.d, also, so that shutdown works properly. Thanks for the post! Nice work! |
|
| Page 1 of 1 | All times are UTC - 6 hours |
| Powered by phpBB® Forum Software © phpBB Group http://www.phpbb.com/ |
|