[Cluster-devel] [GFS2 PATCH] [TRY #2] GFS2: Prevent recovery before the local journal is set
Steven Whitehouse
swhiteho at redhat.com
Wed May 7 10:56:59 UTC 2014
Hi,
On 02/05/14 15:05, Bob Peterson wrote:
> Hi,
>
> This patch uses a completion to prevent dlm's recovery process from
> referencing and trying to recover a journal before a journal has been
> opened.
>
> Regards,
>
> Bob Peterson
> Red Hat File Systems
There is one possible issue that I can see, which is the error case. If
we have some error that occurs prior to getting to the complete_all()
will we land up with a process that has requested recovery waiting
forever? If we can resolve that, then I think this should be good to go,
Steve.
> Signed-off-by: Bob Peterson <rpeterso at redhat.com>
> ---
> diff --git a/fs/gfs2/incore.h b/fs/gfs2/incore.h
> index bdf70c1..04c062c 100644
> --- a/fs/gfs2/incore.h
> +++ b/fs/gfs2/incore.h
> @@ -730,6 +730,8 @@ struct gfs2_sbd {
> struct gfs2_holder sd_sc_gh;
> struct gfs2_holder sd_qc_gh;
>
> + struct completion sd_journal_ready;
> +
> /* Daemon stuff */
>
> struct task_struct *sd_logd_process;
> diff --git a/fs/gfs2/ops_fstype.c b/fs/gfs2/ops_fstype.c
> index 22f9540..4535156 100644
> --- a/fs/gfs2/ops_fstype.c
> +++ b/fs/gfs2/ops_fstype.c
> @@ -94,6 +94,7 @@ static struct gfs2_sbd *init_sbd(struct super_block *sb)
> INIT_LIST_HEAD(&sdp->sd_jindex_list);
> spin_lock_init(&sdp->sd_jindex_spin);
> mutex_init(&sdp->sd_jindex_mutex);
> + init_completion(&sdp->sd_journal_ready);
>
> INIT_LIST_HEAD(&sdp->sd_quota_list);
> mutex_init(&sdp->sd_quota_mutex);
> @@ -676,6 +677,7 @@ static int init_journal(struct gfs2_sbd *sdp, int undo)
>
> if (sdp->sd_args.ar_spectator) {
> sdp->sd_jdesc = gfs2_jdesc_find(sdp, 0);
> + complete_all(&sdp->sd_journal_ready);
> atomic_set(&sdp->sd_log_blks_free, sdp->sd_jdesc->jd_blocks);
> atomic_set(&sdp->sd_log_thresh1, 2*sdp->sd_jdesc->jd_blocks/5);
> atomic_set(&sdp->sd_log_thresh2, 4*sdp->sd_jdesc->jd_blocks/5);
> @@ -686,10 +688,13 @@ static int init_journal(struct gfs2_sbd *sdp, int undo)
> fs_err(sdp, "there are only %u journals (0 - %u)\n",
> gfs2_jindex_size(sdp),
> gfs2_jindex_size(sdp) - 1);
> + complete_all(&sdp->sd_journal_ready);
> goto fail_jindex;
> }
> sdp->sd_jdesc = gfs2_jdesc_find(sdp, sdp->sd_lockstruct.ls_jid);
>
> + complete_all(&sdp->sd_journal_ready);
> +
> error = gfs2_glock_nq_num(sdp, sdp->sd_lockstruct.ls_jid,
> &gfs2_journal_glops,
> LM_ST_EXCLUSIVE, LM_FLAG_NOEXP,
> diff --git a/fs/gfs2/sys.c b/fs/gfs2/sys.c
> index de25d55..02afa82 100644
> --- a/fs/gfs2/sys.c
> +++ b/fs/gfs2/sys.c
> @@ -407,6 +407,9 @@ int gfs2_recover_set(struct gfs2_sbd *sdp, unsigned jid)
> struct gfs2_jdesc *jd;
> int rv;
>
> + /* Wait for our primary journal to be initialized */
> + wait_for_completion(&sdp->sd_journal_ready);
> +
> spin_lock(&sdp->sd_jindex_spin);
> rv = -EBUSY;
> if (sdp->sd_jdesc->jd_jid == jid)
>
More information about the Cluster-devel
mailing list