System died because of ide-scsi
Hello.
I'm not sure where to report bugs in ide-scsi in kernel 2.4, so trying here ...
I've just suffered a system misbehaviour that ended with a system crash on a production server running Linux kernel 2.4.24 SMP. This server acts as terminal server here, running 10-15 KDE sessions.
The problem happened when someone started CD recording operation while CD was mounted. (It's interesting why he was able to do that, but that's another story).
After that, system because very unresponsive (up to 10 seconds to process a key press). System load meter was able to refresh it's screen for several times, and showed both CPUs were "100% SYS".
It wass possible to switch server console from X to text console (athough it took up to mitute to process Ctrl+Alt+F1). On the text cousole, about onse a second the following line was printed:
ide-scsi: Strange, packet command initiated yet DRQ isn't asserted
I could log in as root on the console and type several commands (again, up to 10 seconds to process a key press). I tried to reset the device using hdparm, and that caused kernel oops.
I don't feel I can debug this problem myself. However, if it will help, I will provide all system information on request.
Quoted 4 lines · Nikita V. Youshchenko
Hello.
I'm not sure where to report bugs in ide-scsi in kernel 2.4, so trying here ...
[...]
Quoted 5 lines
ide-scsi: Strange, packet command initiated yet DRQ isn't asserted
I could log in as root on the console and type several commands (again, up to 10 seconds to process a key press). I tried to reset the device using hdparm, and that caused kernel oops.
Hm, I reported what looks like a similar problem some time ago. At least the symptom, but not what caused it. Though I didn't get an oops.
See my previous mail:
http://www.ussg.iu.edu/hypermail/linux/kernel/0312.2/0792.html
Also got "ide-scsi: Strange, packet command initiated yet DRQ isn't asserted" and afaict this happened the first time since I used 2.4.23. Maybe it is the same problem?
If you got an oops you should probably post it. After running through ksymoops of course. :)
Bye,
Oliver
Signature · 1 line
Oliver Feiler <kiza@(kcore.de|lionking.org|gmx[pro].net)>
Hello,
Quoted 2 lines
See my previous mail:
http://www.ussg.iu.edu/hypermail/linux/kernel/0312.2/0792.html
I don't think it is the same.
Quoted 2 lines
If you got an oops you should probably post it. After running through ksymoops of course. :)
Unfortunately the oops dump was lost (it was on the console of the crashed server, and the server was rebooted immidiatly.
However, I found some information, including some traces, in system logs. It is attached.