Date:   Tue, 21 Dec 2021 00:48:49 -0800
From:   Christoph Hellwig <hch@infradead.org>
To:     NeilBrown <neilb@suse.de>
Cc:     Anna Schumaker <anna.schumaker@netapp.com>,
        Trond Myklebust <trond.myklebust@hammerspace.com>,
        Chuck Lever <chuck.lever@oracle.com>,
        Andrew Morton <akpm@linux-foundation.org>,
        Mel Gorman <mgorman@suse.de>,
        Christoph Hellwig <hch@infradead.org>,
        David Howells <dhowells@redhat.com>,
        Linux NFS Mailing List <linux-nfs@vger.kernel.org>,
        linux-mm@kvack.org,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
        "Darrick J. Wong" <djwong@kernel.org>, linux-xfs@vger.kernel.org,
        fstests@vger.kernel.org
Subject: Re: [PATCH 00/18 V2] Repair SWAP-over-NFS
Message-ID: <YcGU8W6+hEfRAVY9@infradead.org>
References: <163969801519.20885.3977673503103544412.stgit@noble.brown>
 <CAFX2Jfn8jER-aV_ttiAe1tkh8f+m=5-whEBTWbHO1uVwf=B4bw@mail.gmail.com>
 <163994803576.25899.6298619065481174544@noble.neil.brown.name>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <163994803576.25899.6298619065481174544@noble.neil.brown.name>
Precedence: bulk

On Mon, Dec 20, 2021 at 08:07:15AM +1100, NeilBrown wrote:
> > Thanks for fixing swap-over-NFS! Looks like it passes all the
> > swap-related xfstests except for generic/357 on NFS v4.2. This test
> > checks that we get -EINVAL on a reflinked swapfile, but I'm not sure
> > if there is a way to check for that on the client side but if you have
> > any ideas it would be nice to get that test passing while you're at
> > it!
> 
> Thanks for testing!.
> I think that testing that swap fails on a reflinked file is bogus.  This
> isn't an important part of the API, it is just an internal
> implementation detail.
> I certainly understand that it could be problematic implementing swap on
> a reflinked file within XFS and it is perfectly acceptable to fail such
> a request.  But if one day someone decided to implement it - should that
> be seen as a regression?

Yes, there is really no fundamental reason not to support swap to
reflinked files, especially for NFS.  We'll need some kind of opt-in/out
for this test.