Subject: Re: ext2fs support broken in -current
To: Kevin Lahey <email@example.com>
From: Kevin Lahey <firstname.lastname@example.org>
Date: 11/13/2004 21:23:47
In message <email@example.com>"Kevin Lahey" writes
>In message <firstname.lastname@example.org>
>Markus W Kilbinger writes
>>writing onto ext2 partitions on i386 and mipsel (qube2) produce
>> panic: kernel diagnostic assertion "vp->v_size == ip->i_size" failed:
>> file "../../../../ufs/ext2fs/ext2fs_readwrite.c", line 252
>> syncing disks... panic: lockmgr: locking against myself
>Ummm, I managed to fix the problem with 'df', and rather
>to my surprise, it seems to have fixed up my problem writing
>to the filesystem as well. The 'df' problem was that statvfs
>was returning a logorithmic fragment size, rather than an explicit
>fragment size in bytes:
Alas, this only postponed the agony of the assert; I could write
a few files after the patch, but I still eventually got a crash.
I think that the problem may actually *be* the assert.
I checked, and ext2fs seemed to have been working just fine before
the commit that put in the assert; when I added the assert to
that working code, the assert triggered. Removing the assert from
the code doesn't seem to exposed any more bugs.
I don't really feel comfortable enough with the filesystem code
to say all this with any confidence. Could someone else take a look
and tell me if I'm crazy?