Re: 2.4.0 kernel paging error

Mark Hindley (mh15@st-andrews.ac.uk)
Thu, 11 Jan 2001 08:23:33 +0000 (GMT)


Philipp Rumpf writes:
> On Wed, Jan 10, 2001 at 05:55:05PM +0100, Daniel Phillips wrote:
> > Mark Hindley wrote:
> > > I am running 2.4.0 final. I got the following failed paging request which
> > > produced a complete freeze.
> > >
> > > As you can see it was precipitated by cron starting to run some
> > > housekeeping stuff overnight.
> > >
> > > Has anyone else had prblems?
> >
> > It looks real. It was executing this line of clear_inode in fs/inode.c:
> >
> > 380 if (inode->i_sb && inode->i_sb->s_op && inode->i_sb->s_op->clear_inode)
> > ^^^^^^^^^^^^^
> > and it blew up here ----->
>
> > Unable to handle kernel paging request at virtual address c4870840
>
> I'm pretty sure this is a vmalloc/module address, which would mean ->s_op
> points to a module that has been unloaded. This sounds consistent with the
> "cron starting to run some housekeeping stuff" above.
>
> Mark, which file systems are you using ?

I have ext2 (builtin) and autoload [v]fat, msdos, hfs, iso9660 and
occasionally (but not recently) udf as modules for removable media.

vfat is loaded all the time as I have a couple of partitions from
another OS I used to use permantently mounted RO.

As far as I can remember hfs and iso9660 were the only other ones I had used
just before this happened.

As I use the kernel module autoloader I also have a cron entry for rmmod -a
which runs every so often to clear out the unused modules. Although
the logs record rmmod running, they don't say what if any modules were
removed.

Mark
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/