Re: scsi-related kernel panic messing up shutdown

[Bruce Perens <bruce@Pixar.com>]

Once the EXT2 FS panics, I think it doesn't trust itself to write any of its
filesystems. Since all of your filesystems are EXT2 ones, there isn't much
point in doing anything but pressing the reset switch. The other errors are
probably because of the inability to write temp files, etc.

There are two things that should be repaired in the kernel here. One is that
the driver should be able to spin up the cartrige drive. Another is that
EXT2 should be more robust about hardware errors, protecting one filesystem
from writes until unmount rather than refusing to write _any_ filesystem.

I think we have some of the EXT2 FS developers here, don't we? If one of them
doesn't join in, we'll have to nudge them.

     Thanks

     Bruce


bhogan@rahul.net said:
> --------------------------- quote -------------------------------- # 
> mv ispell*.deb /E/DEBIAN2.binary/text Kernel panic: EXT2-fs panic 
> (device 8/37): set_inode_dtime: Cannot load inode table block - 
> inode=20352, inode_block=81928

> # shutdown -h now Broadcast message from root (tty1) Wed Sep 27 
> 16:04:01 1995...

> # really? bash: really?: command not found ------------------------ 
> unquote ---------------------------------

>   This happens every time I have a kernel panic on my Fujitsu M2511a 
> m.o. drive -- which is to say almost any time I access that drive 
> after it has been idle for (say) 20 minutes.

>   At this point, because `shutdown' has not in fact shut down, all I 
> can do is hit reset and hope for the best.

>   I would much prefer to have a guaranteed safe exit if possible.

>   Bill

--
-- Attention Radio Amateurs: For information on "Linux for Hams",
-- read the WWW page http://www.hams.com/LinuxForHams, 
-- or e-mail the word "help" to info@hams.com