Received-SPF: pass (google.com: domain of linux-nfs-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Thu, 22 Sep 2022 15:36:52 +0100
From:   Al Viro <viro@zeniv.linux.org.uk>
To:     Christoph Hellwig <hch@infradead.org>
Cc:     Jan Kara <jack@suse.cz>, John Hubbard <jhubbard@nvidia.com>,
        Andrew Morton <akpm@linux-foundation.org>,
        Jens Axboe <axboe@kernel.dk>,
        Miklos Szeredi <miklos@szeredi.hu>,
        "Darrick J . Wong" <djwong@kernel.org>,
        Trond Myklebust <trond.myklebust@hammerspace.com>,
        Anna Schumaker <anna@kernel.org>,
        David Hildenbrand <david@redhat.com>,
        Logan Gunthorpe <logang@deltatee.com>,
        linux-block@vger.kernel.org, linux-fsdevel@vger.kernel.org,
        linux-xfs@vger.kernel.org, linux-nfs@vger.kernel.org,
        linux-mm@kvack.org, LKML <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH v2 4/7] iov_iter: new iov_iter_pin_pages*() routines
Message-ID: <YyxzBPZRp/uulRmf@ZenIV>
References: <20220831041843.973026-1-jhubbard@nvidia.com>
 <20220831041843.973026-5-jhubbard@nvidia.com>
 <YxbtF1O8+kXhTNaj@infradead.org>
 <103fe662-3dc8-35cb-1a68-dda8af95c518@nvidia.com>
 <Yxb7YQWgjHkZet4u@infradead.org>
 <20220906102106.q23ovgyjyrsnbhkp@quack3>
 <YxhaJktqtHw3QTSG@infradead.org>
 <YyFPtTtxYozCuXvu@ZenIV>
 <YyxxyMk0IR2hMjgv@infradead.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <YyxxyMk0IR2hMjgv@infradead.org>
Sender: Al Viro <viro@ftp.linux.org.uk>
Precedence: bulk

On Thu, Sep 22, 2022 at 07:31:36AM -0700, Christoph Hellwig wrote:
> On Wed, Sep 14, 2022 at 04:51:17AM +0100, Al Viro wrote:
> > Unless I'm misreading Jan, the question is whether they should get or
> > pin.
> 
> And I think the answer is:  inside ->read_iter or ->write_iter they
> should neither get or pin.  The callers of it need to pin the pages
> if they are pagecache pages that can potentially be written to through
> shared mappings, else a get would be enough.  But the method instance
> should not have to care and just be able to rely on the caller making
> sure they do not go away.

The interesting part, AFAICS, is where do we _unpin_ them and how do
we keep track which pages (obtained from iov_iter_get_pages et.al.)
need to be unpinned.

> > I'm really tempted to slap
> > 	if (WARN_ON(i->data_source))
> > 		return 0;
> > into copy_to_iter() et.al., along with its opposite for copy_from_iter().
> 
> Ys, I think that would be useful.  And we could use something more
> descriptive than READ/WRITE to start with.

See #work.iov_iter; done, but it took a bit of fixing the places that
create iov_iter instances.