From: Herbert Pötzl (herbert_at_13thfloor.at)
Date: Fri 25 Jul 2003 - 13:16:22 BST
On Fri, Jul 25, 2003 at 01:42:47PM +0200, Georges Toth wrote:
> > so, if its okay with you, I'll send you the patches
> > and some basic 'regression' tests I do ... sometimes ;)
>
> that's ok for me :-)
> just tell me what to do and i'll do it :-D
> (what kernel, which patches, what tools, ...)
okay here you go ...
attached two patches (one labeled -mq the other -bq)
what I would like you to do is the following:
- test -mq extensively as it introduces a new
way the quota hash is handled ...
I'm especially interrested in
+ race conditions on UP and SMP
+ any BUG() traces of course
+ unusual behaviour ;)
- think about some brilliant tests, which actually
verify that quota works as expected ...
basically what to check:
+ quota change on create/unlink
+ quota transfer on chown/chgrp
+ special link/symlink/mkdev handling
+ quota enforcement (soft + hard)
+ ...
- usually a good thing (TM) would be to start with
some 'defined' environment, do some heavy automated
testing and build a md5hash on the quota report.
- this should be compared with the unpatched kernel
(same release though) and _any_ differences should
be reported ...
- later the proven test scripts should be used on
the -mq -bq combo, to spot any differences there
I would suggest 2.4.22-pre8, both patches should
apply without fuzz or reject ... -bq goes ontop -mq
I consider using quotatools 3.09 a good idea, although
3.08 should suffice, v0 and v2 quota formats should
be tested ...
if you have any questions *grin*, feel free to ask!
TIA,
Herbert
> --
> regards,
> Georges Toth