From mboxrd@z Thu Jan 1 00:00:00 1970 From: Tao Ma Date: Fri, 09 Jul 2010 11:38:33 +0800 Subject: [Ocfs2-devel] [PATCH] ocfs2: make xattr extension work with new local alloc reservation. In-Reply-To: <20100708185134.GD17445@mail.oracle.com> References: <4C1AE141.60707@oracle.com> <1276830171-5908-1-git-send-email-tao.ma@oracle.com> <20100708185134.GD17445@mail.oracle.com> Message-ID: <4C3699B9.3010204@oracle.com> List-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: ocfs2-devel@oss.oracle.com On 07/09/2010 02:51 AM, Joel Becker wrote: > On Fri, Jun 18, 2010 at 11:02:50AM +0800, Tao Ma wrote: >> @@ -735,8 +736,9 @@ static int ocfs2_xattr_extend_allocation(struct inode *inode, >> ctxt->data_ac, >> ctxt->meta_ac, >> &why); > > Btw, this code was already buggy. > ocfs2_xattr_extend_allocation() calls ocfs2_add_clusters_in_btree(), > which can return with RESTART_TRANS just because the filesystem is > fragmented. We would just fail with EAGAIN in that case, which makes no > sense to a user. > So this fix actually matters to older kernels and non-reflink > operations too. Would you agree it should go to the stable tree? If > so, add the Cc: to your commit message. yes, the old kernel should be affected by this and we should have it in stable tree. But I just have one concern: should I add stable to cc list the first time I create the patch? I am afraid not since it may be changed may times and have many revisions. So my suggestion is that we go on the normal process, and when my patch get acked, I will resend it, add your ack and cc to stable. Agree? The same goes to other's patches. Regards, Tao