ext2fs related panic

[Poul-Henning Kamp]

In message <1099829438.890.11.camel at taxman>, Stefan Ehmann writes:
>On Sat, 2004-11-06 at 19:25, Poul-Henning Kamp wrote:
>> Ok, can you try the stuff in current, I've made two commits
>> and you want both.
>
>Thanks, these fix the previous panics reported by me.
>
>I was able to produce yet another panic:
>
>mount /mnt/stuff
>mount -u -o rw /mnt/stuff
>touch /mnt/stuff/foo
>mount -u -o ro /mnt/stuff
>umount /mnt/stuff
>
>At this point I get this:
>g_vfs_done():ad1s1[WRITE(offset=4096, length=4096)]error = 1
>g_vfs_done():ad1s1[WRITE(offset=8192, length=4096)]error = 1
>g_vfs_done():ad1s1[WRITE(offset=20480, length=4096)]error = 1
>g_vfs_done():ad1s1[WRITE(offset=27380416512, length=4096)]error = 1
>
>A few seconds later I get this:
>Fatal trap 12: page fault while in kernel mode
>current process = 55 (syncer)

this is a long-standing bug in ext2fs which were previously masked
by other issues:  When downgrading a filesystem from r/w to r/o
the dirty buffers are not flushed.

-- 
Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
phk at FreeBSD.ORG         | TCP/IP since RFC 956
FreeBSD committer       | BSD since 4.3-tahoe    
Never attribute to malice what can adequately be explained by incompetence.