kern/85503: panic: wrong dirclust using msdosfs in RELENG_6

[Dmitry Pryanishnikov]

Hello!

On Sat, 3 Sep 2005, Poul-Henning Kamp wrote:
> Uhm, did none of you guys see my email about how this must be
> done correctly the say way NFS does it correctly ?

  Me, me! ;) I understood the right way to do it yesterday, but haven't
verified it yet. Of course, I've missed vfs_hash_cmp_t *fn, void *arg
arguments of vfs_hash_get(). To say more, your rev 1.88 of msdosfs_denode.c
_must_ work correctly after addition of two casts to (uint64_t) in your
64-bit inode calculations! Once I get some free time, I'll check my
assumption and will post an appropriate follow-up to my PR. I really
want that people won't be "surprised" by 6.0-RELEASE crashing by just
accessing their large FAT32 partitions.

> Making the hashes be 64bit is pointless since no filesystems will
> have that many inodes and it still doesn't solve the problem properly.

  While I agree with you about hashes, I still think that enlarging media
sizes will lead to increasing number of filesystems that will support
> 4 Gfiles / fs.

Sincerely, Dmitry
-- 
Atlantis ISP, System Administrator
e-mail:  dmitry at atlantis.dp.ua
nic-hdl: LYNX-RIPE