Very Computer

Quote:> Hi

> Accidentally  I filled up my jfs root partition.
> After that I had some crashes with jfs
> Below is the dmesg and the crash.

I think I found the problem.  The patch below should fix it.  I will try
to reproduce the problem and verify that this patch works.

Quote:> I had to reboot in kernel 2.4 to be able to delete the files.

The funny thing is that the bug exists in 2.4 too.

Quote:> Also I found in logs a line with
> _mark_inode_dirty: this cannot happen

I don't know about this.

Thanks,
Shaggy

===== fs/jfs/jfs_dtree.c 1.21 vs edited =====
--- 1.21/fs/jfs/jfs_dtree.c     Thu Mar 13 09:14:08 2003

                        d->index = cpu_to_le32(add_index(tid, inode, bn, i));
                        if (dtlck->index >= dtlck->maxcnt)
                                dtlck = (struct dt_lock *) txLinelock(dtlck);
-                       lv = dtlck->lv;
+                       lv = &dtlck->lv[dtlck->index];
                        lv->offset = stbl[i];
                        lv->length = 1;
                        dtlck->index++;

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in

More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Quote:

> I think I found the problem.  The patch below should fix it.  I will
> try to reproduce the problem and verify that this patch works.

I wasn't able to reproduce the trap (which depends on the contents of
uninitialized memory) but I was able to verify that it fixes a problem
where changes to a directory got lost.  I'm convinced the patch is
good.

--
David Kleikamp
IBM Linux Technology Center

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in

More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Dave Kleikamp said:

I applied it. It seems to work. Didn't have much time to test it.
I did a quick test with a dd=if/dev/zero of=file ; rm file
and it seems to work

Thanks

Quote:> --
> David Kleikamp
> IBM Linux Technology Center

> -

Calin

--
# fortune
fortune: write error on /dev/null --- please empty the bit bucket

-----------------------------------------
This email was sent using SquirrelMail.
   "Webmail for nuts!"
http://squirrelmail.org/

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in

More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/