Subject: Re: Filesystem I/O conformance testing?
To: Martin Fouts <mfouts@danger.com>
From: Antti Kantee <pooka@cs.hut.fi>
List: tech-kern
Date: 10/18/2007 16:41:09
On Wed Oct 17 2007 at 15:30:07 -0700, Martin Fouts wrote:
> I'm adding a new file system to NetBSD to support NAND and am far enough
> along to start serious testing.
> 
> I thought I'd canvas the group to see if there's any existing
> conformance test suite that anyone likes that I might have missed.
> 
> We're beating on the file system with lmbench, bonnie, and fsx, but
> we're wondering if there's anything else that we could use.  I've looked
> at the Open Posix Test Suite, but it's not really testing fs
> conformance.

There's also the tmpfs regression test in NetBSD and then there's pjd's
fstest (http://people.freebsd.org/~pjd/fstest/).  But adapting the latter
might take some effort.  I very quickly tried running it on NetBSD once
and it didn't work out-of-the-box and I didn't have time to fix it then.

A cvs checkout was already mentioned.  And simply untarring a tarball
does a surprisingly good job at finding most easy-to-encounter mistakes.

And if you're doing serious testing, you probably want unit tests as well.
On -current rump/ukfs will allow you to execute these for the kernel fs
code in userspace.

-- 
Antti Kantee <pooka@iki.fi>                     Of course he runs NetBSD
http://www.iki.fi/pooka/                          http://www.NetBSD.org/
    "la qualité la plus indispensable du cuisinier est l'exactitude"