Re: [PATCH 2/3] afs: Fix afs_write_end() to handle short writes

From: David Howells <dhowells@redhat.com>
To: Matthew Wilcox <willy@infradead.org>
Cc: dhowells@redhat.com, jlayton@kernel.org,
	linux-afs@lists.infradead.org, ceph-devel@vger.kernel.org,
	linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH 2/3] afs: Fix afs_write_end() to handle short writes
Date: Mon, 14 Jun 2021 16:38:21 +0100	[thread overview]
Message-ID: <475131.1623685101@warthog.procyon.org.uk> (raw)
In-Reply-To: <YMdpxbYafHnE0F8N@casper.infradead.org>

Matthew Wilcox <willy@infradead.org> wrote:

> > -	ASSERT(PageUptodate(page));
> > -
> >  	if (PagePrivate(page)) {
> >  		priv = page_private(page);
> >  		f = afs_page_dirty_from(page, priv);
> 
> Why are you removing this assertion?  Does AFS now support dirty,
> partially-uptodate pages?  If so, a subsequent read() to that
> page is going to need to be careful to only read the parts of the page
> from the server that haven't been written ...

Because the previous hunk in the patch:

	+	if (!PageUptodate(page)) {
	+		if (copied < len) {
	+			copied = 0;
	+			goto out;
	+		}
	+
	+		SetPageUptodate(page);
	+	}

means you can't get there unless PageUptodate() is true by that point.

David