From mboxrd@z Thu Jan 1 00:00:00 1970 From: Christoph Hellwig Date: Fri, 25 Nov 2016 00:10:58 -0800 Subject: [Ocfs2-devel] [PATCH 5/7] ocfs2: don't eat io errors during _dio_end_io_write In-Reply-To: <147987579593.2678.14305853637089494031.stgit@birch.djwong.org> References: <147987576127.2678.7138661302696363830.stgit@birch.djwong.org> <147987579593.2678.14305853637089494031.stgit@birch.djwong.org> Message-ID: <20161125081058.GA22465@infradead.org> List-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: "Darrick J. Wong" Cc: mfasheh@versity.com, jlbec@evilplan.org, linux-fsdevel@vger.kernel.org, darwin.xupt@gmail.com, zren@suse.com, ocfs2-devel@oss.oracle.com On Tue, Nov 22, 2016 at 08:36:35PM -0800, Darrick J. Wong wrote: > ocfs2_dio_end_io_write eats whatever errors may happen, > which means that write errors do not propagate to userspace. > Fix that. Btw, can you also fix up the ocfs2 calling conventions for ->end_io so that it also does the unlock for the error case in ocfs2_dio_end_io? I looked at all the inode locking in ocfs2, but couldn't convince me that either the current code nor the version with the locking changed is correct..