From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from imap.thunk.org ([74.207.234.97]:42790 "EHLO imap.thunk.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752844AbcI2BRp (ORCPT ); Wed, 28 Sep 2016 21:17:45 -0400 Date: Wed, 28 Sep 2016 20:01:30 -0400 From: "Theodore Ts'o" Subject: Re: [PATCH] common: add support for the "local" file system type Message-ID: <20160929000130.ixuytsr5euxef6vk@thunk.org> References: <20160923200526.29674-1-tytso@mit.edu> <20160926132503.GJ27776@eguan.usersys.redhat.com> <20160926151431.uhtny5tx3b436j4i@thunk.org> <20160927095529.GO27776@eguan.usersys.redhat.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20160927095529.GO27776@eguan.usersys.redhat.com> Sender: fstests-owner@vger.kernel.org To: Eryu Guan Cc: fstests@vger.kernel.org List-ID: On Tue, Sep 27, 2016 at 05:55:29PM +0800, Eryu Guan wrote: > > It has nothing to do with mount & umount, it's about adding a new > "-local" option to "check", like "-nfs" and "-overlay". And do > TEST_DEV/TEST_DIR and SCRATCH_DEV/SCRATCH_MNT validations in > _require_test and _require_scratch_nocheck, as how we do the check for > NFS and overlayfs. So we don't have to work around the validation by > specifying TEST_DEV/SCRATCH_DEV in a non-block device format (e.g. > local:/test to mimic an NFS export). I'd have to dig and check, but as I recall, there were assumptions that foo:/bar means "no block device" scattered around, and it's why we have to use a similar naming scheme for tmpfs. Is that really a problem? I had thought it was a standard convention for xfstests. If we want to avoid forcing testers to use a foo:/bar naming scheme for TEST_DEV and SCRATCH_DEV, instead of a new command-line variable, I'd much rather have a function in rc/common that could be keyed off the file system type. You have to specify whether you are using tmpfs or nfs or local in FSTYP anyway, so why not just use that? Cheers, - Ted