On Wed, Mar 19, 2014 at 02:09:07PM -0700, Andrew Morton
wrote:>
> I'm sitting on six patches which I'm reluctant to merge under my
own
> steam. Some of these I have probably send out before.
>
>
ocfs2-alloc_dinode-counts-and-group-bitmap-should-be-update-simultaneously.patch
> ocfs2-flock-drop-cross-node-lock-when-failed-locally.patch
>
ocfs2-o2net-o2net_listen_data_ready-should-do-nothing-if-socket-state-is-not-tcp_listen.patch
> ocfs2-call-ocfs2_update_inode_fsync_trans-when-updating-any-inode.patch
>
ocfs2-do-not-return-dlm_migrate_response_mastery_ref-to-avoid-endlessloop-during-umount.patch
>
ocfs2-manually-do-the-iput-once-ocfs2_add_entry-failed-in-ocfs2_symlink-and-ocfs2_mknod.patch
I can the review above patches tommorrow for us.
> And these two have been hanging around for a long time:
>
>
ocfs2-should-call-ocfs2_journal_access_di-before-ocfs2_delete_entry-in-ocfs2_orphan_del.patch
> ("Sunil nack, Mark confused")
Go ahead and drop this one please. I never figured out exactly what the
actual problem was - I'd be happy to come back to it later once we know
more.
> ocfs2-llseek-requires-ocfs2-inode-lock-for-the-file-in-seek_end.patch
> ("Sunil worried about performance, Joel had Q, Mark wanted update,
Joel
> acked, worried about perf")
The performance testing didn't really happen in a way that was useful. I
wanted to find time to do it on my own but never did. Since this fixes a bug
for a known user I would say to go for it.
--Mark
--
Mark Fasheh