From: Mingming Cao Subject: Re: 2.6.21-ext4-1 Date: Tue, 08 May 2007 15:05:56 -0700 Message-ID: <1178661957.4135.66.camel@dyn9047017103.beaverton.ibm.com> References: <1178571383.3933.8.camel@dyn9047017103.beaverton.ibm.com> <20070508025048.GC32602149@melbourne.sgi.com> Reply-To: cmm@us.ibm.com Mime-Version: 1.0 Content-Type: text/plain Content-Transfer-Encoding: 7bit Cc: "Theodore Ts'o" , linux-ext4@vger.kernel.org, Johann Lombardi , "Amit K. Arora" , linux-kernel@vger.kernel.org To: David Chinner Return-path: Received: from e32.co.us.ibm.com ([32.97.110.150]:42450 "EHLO e32.co.us.ibm.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751220AbXEHWGF (ORCPT ); Tue, 8 May 2007 18:06:05 -0400 In-Reply-To: <20070508025048.GC32602149@melbourne.sgi.com> Sender: linux-ext4-owner@vger.kernel.org List-Id: linux-ext4.vger.kernel.org On Tue, 2007-05-08 at 12:50 +1000, David Chinner wrote: > On Mon, May 07, 2007 at 01:56:23PM -0700, Mingming Cao wrote: > > In any case, it would be useful to add a new set of testsuites for the > > new fallocate() syscall and fsstress in LTP testsuites to automatically > > the preallocation code in ext4/XFS. > > I hacked an existing XFS test prog to do manual testing of the fallocate() > syscall. In the XFSQA suite we have various pre-alloc enhanced utils (e.g. > fsstress, fsx, etc) that we should probably update to be able to use both > fallocate and xfsctl so we can test both. > > Here's all the programs we use that have preallocation awareness: > > chook 982% grep RESVSP ltp/*.c | awk '/^ltp/ { split($1,a,":"); print a[1] ;}' | uniq > ltp/doio.c > ltp/fsstress.c > ltp/fsx.c > ltp/growfiles.c > ltp/iogen.c > chook 983% grep RESVSP src/* | awk '/^src/ { split($1,a,":"); print a[1] ;}' | uniq > src/alloc.c > src/fstest.c > src/iopat.c > src/randholes.c > src/resvtest.c > src/unwritten_mmap.c > src/unwritten_sync.c > Thanks for sharing the info. I think Amit used a fsx version with preallocation awareness test, but I don't know if we have other ltp tests that aware preallocation. I would very appreciate if you could share your preallocation test with us. > BTW, have you guys tested mmap writes into unwritten extents? ;) > I am not sure, Amit, have you done some mmap write test into uninitialized extents? Sorry, I still not quite clear what's the mapped problem you are worry about. Could you explain to me a bit more? thanks! Mingming > Cheers, > > Dave.