Bug#452739: Crash
Øystein Gisnås wrote: Note that there are two lockups/hangs/crashes, both with output to the log. The first one happens sporadically and causes gdm to restart. Isn't that most probably a crash? Yes, if gdm restarts, that's probably a real crash. /var/log/Xorg.0.log.old should contain a possibly interesting backtrace at the end (Xorg0.log is saved in .old when X restarts). Remember to save Xorg.0.log.old next time such a gdm restart occurs. The second one, when the new X server starts, may well be a lockup as you say. Wouldn't it be better to look at the first one to find the cause here? The first clue in the log is: Error in I830WaitLpRing(), timeout for 2 seconds. Yes, that's a lockup and we need to look at the first crash first. The lockup could just be caused by the first crash leaving the hardware page table in a bad state (even if we might say that the driver should be able to cleanup this mess on startup). Brice
Bug#452739: Crash
请问如何退订BUG这个右键呀 2007-12-08 linuxwork 发件人: Brice Goglin 发送时间: 2007-12-08 21:03:06 收件人: Øystein_Gisnås; [EMAIL PROTECTED] 抄送: Moshe Gorohovsky 主题: Bug#452739: Crash 貀stein Gisn錽 wrote: Note that there are two lockups/hangs/crashes, both with output to the log. The first one happens sporadically and causes gdm to restart. Isn't that most probably a crash? Yes, if gdm restarts, that's probably a real crash. /var/log/Xorg.0.log.old should contain a possibly interesting backtrace at the end (Xorg0.log is saved in .old when X restarts). Remember to save Xorg.0.log.old next time such a gdm restart occurs. The second one, when the new X server starts, may well be a lockup as you say. Wouldn't it be better to look at the first one to find the cause here? The first clue in the log is: Error in I830WaitLpRing(), timeout for 2 seconds. Yes, that's a lockup and we need to look at the first crash first. The lockup could just be caused by the first crash leaving the hardware page table in a bad state (even if we might say that the driver should be able to cleanup this mess on startup). Brice
Bug#452739: Crash
2007/12/3, Brice Goglin [EMAIL PROTECTED]: Øystein Gisnås wrote: Just a minute after I sent the previous report, X.org http://X.org crashed again. This time I managed to stop gdm from starting again. The attached Xorg.0.log is the same as in the previous report, but this time with details about the crash towards the end. Would it give more info if I had the -dbg package installed? This is not a crash (segfault of the X server or so), this looks like a lockup (the server is stuck somewhere and stop processing anything else). So -dbg would probably not help, unless you can catch a segfault with gdb. Note that there are two lockups/hangs/crashes, both with output to the log. The first one happens sporadically and causes gdm to restart. Isn't that most probably a crash? The second one, when the new X server starts, may well be a lockup as you say. Wouldn't it be better to look at the first one to find the cause here? The first clue in the log is: Error in I830WaitLpRing(), timeout for 2 seconds. I haven't had the problem again (after ~10 hours of use) since I downgraded to 2.1.1-4. Cheers, Øystein I tried to downgrade xserver-xorg-video-intel to 2.1.0-2, but I get a conflict with xserver-xorg-core. Can someone suggest an xserver-xorg-core-version I can use with 2.1.0-2, and where to download that? You probably want 2.1.1-4, it is available from http://snapshot.debian.net/archive/2007/09/17/debian/pool/main/x/xserver-xorg-video-intel/ You won't have to downgrade xserver-xorg-core if you take this one. It this problem easy to reproduce? Did you switch to/from a virtual console right before it happened? You seem to have some VT switch related messages in your log. It is not even sure that the page table error is related to the problem. Can you check whether the page table error appears when the lockup occurs and not earlier? I might have switched to terminal, but I think I have had sessions that crashed without any switches to terminal. I can't force the crash to happen exactly when I want, but it happens quite regularly about every 2 hours. Cheers, Øystein
Bug#452739: Crash
Moshe Gorohovsky wrote: Hi Brice, 1. The crash is not easy to reproduce. It occurs once or twice a day on Debian unstable, all packages upgraded on 20071203. 2. I have not switched between text VTs and X VT before a crash. 3. page table error appears only after a crash. Now I realize that the log file I had supplied is from gdm restart attempt after a crash. I am searching how to disable gdm automatic restarts. I will supply more information if I catch a crash and its log file. How to allow gdm startup at boot and disable gdm or X.org restarts after a crash? If what you want is actually the log of the previous X session (before gdm restarted), then it should be in /var/log/Xorg.0.log.old. According to the upstream dev, page table error can be caused by the previous X session leaving bad page table entries and the new session finding them. So the log of the previous session would probably be interesting. Brice -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#452739: Crash
Hi Brice, 1. The crash is not easy to reproduce. It occurs once or twice a day on Debian unstable, all packages upgraded on 20071203. 2. I have not switched between text VTs and X VT before a crash. 3. page table error appears only after a crash. Now I realize that the log file I had supplied is from gdm restart attempt after a crash. I am searching how to disable gdm automatic restarts. I will supply more information if I catch a crash and its log file. How to allow gdm startup at boot and disable gdm or X.org restarts after a crash? Moshe. Brice Goglin wrote: Øystein Gisnås wrote: Just a minute after I sent the previous report, X.org http://X.org crashed again. This time I managed to stop gdm from starting again. The attached Xorg.0.log is the same as in the previous report, but this time with details about the crash towards the end. Would it give more info if I had the -dbg package installed? This is not a crash (segfault of the X server or so), this looks like a lockup (the server is stuck somewhere and stop processing anything else). So -dbg would probably not help, unless you can catch a segfault with gdb. I tried to downgrade xserver-xorg-video-intel to 2.1.0-2, but I get a conflict with xserver-xorg-core. Can someone suggest an xserver-xorg-core-version I can use with 2.1.0-2, and where to download that? You probably want 2.1.1-4, it is available from http://snapshot.debian.net/archive/2007/09/17/debian/pool/main/x/xserver-xorg-video-intel/ You won't have to downgrade xserver-xorg-core if you take this one. It this problem easy to reproduce? Did you switch to/from a virtual console right before it happened? You seem to have some VT switch related messages in your log. It is not even sure that the page table error is related to the problem. Can you check whether the page table error appears when the lockup occurs and not earlier? Brice -- Moshe Gorohovsky A6 CC A7 E1 C2 BD 8C 1B 30 8E A4 C3 4C 09 88 47 Tk Open Systems Ltd. --- - [EMAIL PROTECTED] - tel: +972.52.882.0101, http://www.tkos.co.il -
Bug#452739: Crash
Hi Brice, Thanks. Please see below. Brice Goglin wrote: If what you want is actually the log of the previous X session (before gdm restarted), then it should be in /var/log/Xorg.0.log.old. According to the upstream dev, page table error can be caused by the previous X session leaving bad page table entries and the new session finding them. So the log of the previous session would probably be interesting. Brice page table error appeared in Xorg.0.log.old file. This is a log file for X.org failed restart attempt after a crash. This restart attempt left a screen blank, and page table error line in the log file. text VTs were also blank. I restarted the laptop. Then, reportbug tool grabbed Xorg.0.log file after successful X.org startup. I will try to catch Xorg.0.log after a crash. Now I am searching how to prevent gdm or X.org from restarting, so I can save log files after a crash. Probably, I will disable gdm in /etc/rc*.d/ directories and start gdm manually from a command line. Moshe -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#452739: Crash
Øystein Gisnås wrote: Just a minute after I sent the previous report, X.org http://X.org crashed again. This time I managed to stop gdm from starting again. The attached Xorg.0.log is the same as in the previous report, but this time with details about the crash towards the end. Would it give more info if I had the -dbg package installed? This is not a crash (segfault of the X server or so), this looks like a lockup (the server is stuck somewhere and stop processing anything else). So -dbg would probably not help, unless you can catch a segfault with gdb. I tried to downgrade xserver-xorg-video-intel to 2.1.0-2, but I get a conflict with xserver-xorg-core. Can someone suggest an xserver-xorg-core-version I can use with 2.1.0-2, and where to download that? You probably want 2.1.1-4, it is available from http://snapshot.debian.net/archive/2007/09/17/debian/pool/main/x/xserver-xorg-video-intel/ You won't have to downgrade xserver-xorg-core if you take this one. It this problem easy to reproduce? Did you switch to/from a virtual console right before it happened? You seem to have some VT switch related messages in your log. It is not even sure that the page table error is related to the problem. Can you check whether the page table error appears when the lockup occurs and not earlier? Brice