From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from ipmail06.adl2.internode.on.net ([150.101.137.129]:5264 "EHLO ipmail06.adl2.internode.on.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752810AbcBKDHy (ORCPT ); Wed, 10 Feb 2016 22:07:54 -0500 Date: Thu, 11 Feb 2016 14:07:51 +1100 From: Dave Chinner Subject: Re: [PATCH 05/12] xfstests: do not unmount tmpfs during remount Message-ID: <20160211030751.GE19486@dastard> References: <1455069001-17846-1-git-send-email-tytso@mit.edu> <1455069001-17846-6-git-send-email-tytso@mit.edu> <20160210060716.GV19486@dastard> <20160210160732.GE26922@thunk.org> <20160210230700.GA19486@dastard> <20160210232826.GK26922@thunk.org> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20160210232826.GK26922@thunk.org> Sender: fstests-owner@vger.kernel.org To: Theodore Ts'o Cc: fstests@vger.kernel.org, hughd@google.com, Junho Ryu List-ID: On Wed, Feb 10, 2016 at 06:28:26PM -0500, Theodore Ts'o wrote: > On Thu, Feb 11, 2016 at 10:07:00AM +1100, Dave Chinner wrote: > > No, it's not really the options that are the problem here. The > > problem is -o remount vs unmount/mount and what the test is actually > > expecting. > > > > I'd say "_scratch_remount" should do "-o remount" unconditionally > > (least surprise) and the current _scratch_remount should be changed > > to something like _scratch_cycle_mount(). That way both can take > > options, but it's clear they do different things. tmpfs can simply > > implement them the same way. > > Well, I can do that, but it's going to be a huge patch --- the vast > majority of the calls to _scratch_remount in the tree (over 100) would > have to be changed to _scratch_cycle_mount, because they are just > doing a _scratch_umount / _scratch_mount without taking any arguments > to change the mount option. Yup, but we do this sort of tree-wide cleanup fairly often if it makes sense. In this case, it's just an initial patch taht does sed -i -e 's/_scratch_remount/_scratch_cycle_mount/' .... And, let's put things in context: changing 108 lines of code is a pretty damn small patch in the greater scheme of things. Indeed, it's smaller than most patches that add a new regression test. A "huge" patch is something like the series Darrick posted earlier in the week - something like 20 patches, including somewhere in the order of 30 new tests, a couple of new binary test programs, a heap of cleanups across all 80-90 existing reflink/dedupe tests, and a bunch of bug fixes to go with them. IOWs, s/_scratch_remount/_scratch_cycle_mount/ is the sort of no-brainer change that takes less time to write, test and review than it did for me to write this email.... Cheers, Dave. -- Dave Chinner david@fromorbit.com