From: Christoph Hellwig Subject: Re: Port xfstests 145, 161, 175, 176, 185? Date: Thu, 9 Jun 2011 07:08:26 -0400 Message-ID: <20110609110826.GA14899@infradead.org> References: <4DEFC41A.9070701@linux.vnet.ibm.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Ext4 Developers List , linux-fsdevel , xfs-oss To: Allison Henderson Return-path: Content-Disposition: inline In-Reply-To: <4DEFC41A.9070701@linux.vnet.ibm.com> Sender: linux-fsdevel-owner@vger.kernel.org List-Id: linux-ext4.vger.kernel.org On Wed, Jun 08, 2011 at 11:48:58AM -0700, Allison Henderson wrote: > During one of my reviews for the punch hole tests patch set it was > mentioned that it would be helpful to take the xfstests 145, 161, > 175, 176, 185 and modify them such that they can run with out > requiring the dmapi. These tests contain some more interesting > punch hole tests, but they dont normally run unless there is support > for dmapi. > > I did take a peek at them and I was thinking that if we decide to do > this, we would probably need to do something like introduce a new > set of source code that is similar to what is seen under the dmapi > folder, but modified to use a generic interface instead of the dmapi > libraries. We could try to merge them into a single code path, but > I think that may introduce more complexities than would be > desirable. Most of it should be doable using xfs_io. If it's nessecary to write new source files because of e.g. concurrency tests that we can't easily do from xfs_io please add new source files to the src/ directory. Also, please don't rewrite the actual existing dmapi test cases, but add new ones testing these patterns using the fallocate interface, as the coverage for the dmapi interface still is useful.