On Fri, 27 Jul 2012 11:48:39 -0500 Steve French <smfrench@xxxxxxxxx> wrote: > Presumably to address the cifs equivalent of NFS bug (drop_nlink warning): > > https://bugzilla.redhat.com/show_bug.cgi?id=822703 > > your patch which wraps drop_nlink in i_lock spinlock (as Jeff noted > addresses only part of the problem) > http://git.altlinux.org/people/piastry/public/cifs-2.6.git?p=cifs-2.6.git;a=patch;h=df2d6b1fbf2401c5ee04f2ac143ea0954e3a87a6 > > shouldn't it also wrap the three places in fs/cifs/inode.c which do set_nlink? > It might not hurt to serialize all inode attribute updates under the i_lock. If two updates are racing and the CPU ends up reording things, you could end up with an inode that has a mix of attrs returned by calls. The NFS client did this quite some time ago... -- Jeff Layton <jlayton@xxxxxxxxxx> -- To unsubscribe from this list: send the line "unsubscribe linux-cifs" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
