Discussion:
How could I record the information when kernel panic.
disnoir
2013-04-24 17:18:24 UTC
Permalink
There is a random probability that cause the kernel panic (debian with
kernel 3.2.41), when I execute:
"sudo umount /media/data /media/big".

Things are like this:

I use "udisks --mount /dev/sda3" to mount "data" which is a ntfs filesystem.
similarly, I use "udisks --mount /dev/wd5000bpvt/big" to mount "big"
which is a ext4 filesystem.

After finish using these filesystems, I use "sudo umount /media/data
/media/big" to umount the filesystems, and the hole system just suddenly
halt (mouse is not moving).

using alt-sysrq-b to reboot the system. nothing record in the
/var/log/messages.

So, How could I record the kernel error/info, when system was halt ?

Thanks a lot!


--
To unsubscribe from this list: send the line "unsubscribe linux-newbie" in
the body of a message to ***@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.linux-learn.org/faqs
Brandon Philips
2013-04-26 17:33:22 UTC
Permalink
Post by disnoir
So, How could I record the kernel error/info, when system was halt ?
Your best bet is to use one of the following:
0) Use sysrq to investigate a bit:
https://www.kernel.org/doc/Documentation/sysrq.txt
1) Hook a serial console up if your system has one to another system
with a null modem cable and console=ttyS0,<baud> passed in to the
kernel at boot
2) Try netconsole if you don't have a serial console:
https://www.kernel.org/doc/Documentation/networking/netconsole.txt

I would recommend trying a newer kernel and see if your bug is fixed.
If it is you could bisect down to the commit that fixes it and let the
debian kernel maintainers know.

Good luck,

Brandon
--
To unsubscribe from this list: send the line "unsubscribe linux-newbie" in
the body of a message to ***@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.linux-learn.org/faqs
disnoir
2013-04-27 04:02:49 UTC
Permalink
Post by Brandon Philips
Post by disnoir
So, How could I record the kernel error/info, when system was halt ?
https://www.kernel.org/doc/Documentation/sysrq.txt
1) Hook a serial console up if your system has one to another system
with a null modem cable and console=ttyS0,<baud> passed in to the
kernel at boot
https://www.kernel.org/doc/Documentation/networking/netconsole.txt
I would recommend trying a newer kernel and see if your bug is fixed.
If it is you could bisect down to the commit that fixes it and let the
debian kernel maintainers know.
Good luck,
Brandon
Thank you for your reply.
I use ttyUSB0 to record the dmesg, cause this is a notebook
(asus-k42je), which has no serial port, so I use a
serial-port-to-USB-convertor.
I found that if I add console=ttyUSB0,115200n8 to kernel parameter, the
kernel will fail to boot, and also without any information.
so I get into system on normal configuration (without console=...), and
modify rsyslog.conf to log kernel info to /dev/ttyUSB0.
That works, use another pc to watch that serial port, all the
information in dmesg will show in that serial port either.
But, when I test that if will also work during the kernel-panic ( I try
"echo c > /proc/sysrq-trigger" ), nothing record into serial.
I guess kernel crash will cause rsyslogd no working.
So I use kdump-tools, kexec-tools to record a kernel dump, and this
could work even on kernel-panic.
Now, I just need to wait the problem happen again.

:-)

disnoir
--
To unsubscribe from this list: send the line "unsubscribe linux-newbie" in
the body of a message to ***@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.linux-learn.org/faqs

Continue reading on narkive:
Loading...