* [Ocfs2-devel] [PATCH 1/1] o2dlm: force free mles during dlm exit
@ 2010-09-21 23:27 Srinivas Eeda
2010-09-21 23:50 ` Sunil Mushran
2010-09-23 21:23 ` Joel Becker
0 siblings, 2 replies; 3+ messages in thread
From: Srinivas Eeda @ 2010-09-21 23:27 UTC (permalink / raw)
To: ocfs2-devel
While umounting, a block mle doesn't get freed if dlm is shutdown after
master request is received but before assert master. This results in unclean
shutdown of dlm domain.
This patch frees all mles that lie around after other nodes were notified about
exiting the dlm and marking dlm state as leaving. Only block mles are expected
to be around, so we log ERROR for other mles but still free them.
Signed-off-by: Srinivas Eeda <srinivas.eeda@oracle.com>
---
fs/ocfs2/dlm/dlmcommon.h | 1 +
fs/ocfs2/dlm/dlmdomain.c | 1 +
fs/ocfs2/dlm/dlmmaster.c | 40 ++++++++++++++++++++++++++++++++++++++++
3 files changed, 42 insertions(+), 0 deletions(-)
diff --git a/fs/ocfs2/dlm/dlmcommon.h b/fs/ocfs2/dlm/dlmcommon.h
index 4b6ae2c..7652989 100644
--- a/fs/ocfs2/dlm/dlmcommon.h
+++ b/fs/ocfs2/dlm/dlmcommon.h
@@ -1030,6 +1030,7 @@ int dlm_drop_lockres_ref(struct dlm_ctxt *dlm,
struct dlm_lock_resource *res);
void dlm_clean_master_list(struct dlm_ctxt *dlm,
u8 dead_node);
+void dlm_force_free_mles(struct dlm_ctxt *dlm);
int dlm_lock_basts_flushed(struct dlm_ctxt *dlm, struct dlm_lock *lock);
int __dlm_lockres_has_locks(struct dlm_lock_resource *res);
int __dlm_lockres_unused(struct dlm_lock_resource *res);
diff --git a/fs/ocfs2/dlm/dlmdomain.c b/fs/ocfs2/dlm/dlmdomain.c
index 153abb5..11a5c87 100644
--- a/fs/ocfs2/dlm/dlmdomain.c
+++ b/fs/ocfs2/dlm/dlmdomain.c
@@ -693,6 +693,7 @@ void dlm_unregister_domain(struct dlm_ctxt *dlm)
dlm_mark_domain_leaving(dlm);
dlm_leave_domain(dlm);
+ dlm_force_free_mles(dlm);
dlm_complete_dlm_shutdown(dlm);
}
dlm_put(dlm);
diff --git a/fs/ocfs2/dlm/dlmmaster.c b/fs/ocfs2/dlm/dlmmaster.c
index ffb4c68..f564b0e 100644
--- a/fs/ocfs2/dlm/dlmmaster.c
+++ b/fs/ocfs2/dlm/dlmmaster.c
@@ -3433,3 +3433,43 @@ void dlm_lockres_release_ast(struct dlm_ctxt *dlm,
wake_up(&res->wq);
wake_up(&dlm->migration_wq);
}
+
+void dlm_force_free_mles(struct dlm_ctxt *dlm)
+{
+ int i;
+ struct hlist_head *bucket;
+ struct dlm_master_list_entry *mle;
+ struct hlist_node *tmp, *list;
+
+ /*
+ * We notified all other nodes that we are exiting the domain and
+ * marked the dlm state to DLM_CTXT_LEAVING. If any mles are still
+ * around we force free them and wake any processes that are waiting
+ * on the mles
+ */
+ spin_lock(&dlm->spinlock);
+ spin_lock(&dlm->master_lock);
+
+ BUG_ON(dlm->dlm_state != DLM_CTXT_LEAVING);
+ BUG_ON((find_next_bit(dlm->domain_map, O2NM_MAX_NODES, 0) < O2NM_MAX_NODES));
+
+ for (i = 0; i < DLM_HASH_BUCKETS; i++) {
+ bucket = dlm_master_hash(dlm, i);
+ hlist_for_each_safe(list, tmp, bucket) {
+ mle = hlist_entry(list, struct dlm_master_list_entry,
+ master_hash_node);
+ if (mle->type != DLM_MLE_BLOCK) {
+ mlog(ML_ERROR, "bad mle: %p\n", mle);
+ dlm_print_one_mle(mle);
+ }
+ atomic_set(&mle->woken, 1);
+ wake_up(&mle->wq);
+
+ __dlm_unlink_mle(dlm, mle);
+ __dlm_mle_detach_hb_events(dlm, mle);
+ __dlm_put_mle(mle);
+ }
+ }
+ spin_unlock(&dlm->master_lock);
+ spin_unlock(&dlm->spinlock);
+}
--
1.5.6.5
^ permalink raw reply related [flat|nested] 3+ messages in thread
* [Ocfs2-devel] [PATCH 1/1] o2dlm: force free mles during dlm exit
2010-09-21 23:27 [Ocfs2-devel] [PATCH 1/1] o2dlm: force free mles during dlm exit Srinivas Eeda
@ 2010-09-21 23:50 ` Sunil Mushran
2010-09-23 21:23 ` Joel Becker
1 sibling, 0 replies; 3+ messages in thread
From: Sunil Mushran @ 2010-09-21 23:50 UTC (permalink / raw)
To: ocfs2-devel
I hope this has been tested.
Acked-by: Sunil Mushran <sunil.mushran@oracle.com>
On 09/21/2010 04:27 PM, Srinivas Eeda wrote:
> While umounting, a block mle doesn't get freed if dlm is shutdown after
> master request is received but before assert master. This results in unclean
> shutdown of dlm domain.
>
> This patch frees all mles that lie around after other nodes were notified about
> exiting the dlm and marking dlm state as leaving. Only block mles are expected
> to be around, so we log ERROR for other mles but still free them.
>
> Signed-off-by: Srinivas Eeda<srinivas.eeda@oracle.com>
> ---
> fs/ocfs2/dlm/dlmcommon.h | 1 +
> fs/ocfs2/dlm/dlmdomain.c | 1 +
> fs/ocfs2/dlm/dlmmaster.c | 40 ++++++++++++++++++++++++++++++++++++++++
> 3 files changed, 42 insertions(+), 0 deletions(-)
>
> diff --git a/fs/ocfs2/dlm/dlmcommon.h b/fs/ocfs2/dlm/dlmcommon.h
> index 4b6ae2c..7652989 100644
> --- a/fs/ocfs2/dlm/dlmcommon.h
> +++ b/fs/ocfs2/dlm/dlmcommon.h
> @@ -1030,6 +1030,7 @@ int dlm_drop_lockres_ref(struct dlm_ctxt *dlm,
> struct dlm_lock_resource *res);
> void dlm_clean_master_list(struct dlm_ctxt *dlm,
> u8 dead_node);
> +void dlm_force_free_mles(struct dlm_ctxt *dlm);
> int dlm_lock_basts_flushed(struct dlm_ctxt *dlm, struct dlm_lock *lock);
> int __dlm_lockres_has_locks(struct dlm_lock_resource *res);
> int __dlm_lockres_unused(struct dlm_lock_resource *res);
> diff --git a/fs/ocfs2/dlm/dlmdomain.c b/fs/ocfs2/dlm/dlmdomain.c
> index 153abb5..11a5c87 100644
> --- a/fs/ocfs2/dlm/dlmdomain.c
> +++ b/fs/ocfs2/dlm/dlmdomain.c
> @@ -693,6 +693,7 @@ void dlm_unregister_domain(struct dlm_ctxt *dlm)
>
> dlm_mark_domain_leaving(dlm);
> dlm_leave_domain(dlm);
> + dlm_force_free_mles(dlm);
> dlm_complete_dlm_shutdown(dlm);
> }
> dlm_put(dlm);
> diff --git a/fs/ocfs2/dlm/dlmmaster.c b/fs/ocfs2/dlm/dlmmaster.c
> index ffb4c68..f564b0e 100644
> --- a/fs/ocfs2/dlm/dlmmaster.c
> +++ b/fs/ocfs2/dlm/dlmmaster.c
> @@ -3433,3 +3433,43 @@ void dlm_lockres_release_ast(struct dlm_ctxt *dlm,
> wake_up(&res->wq);
> wake_up(&dlm->migration_wq);
> }
> +
> +void dlm_force_free_mles(struct dlm_ctxt *dlm)
> +{
> + int i;
> + struct hlist_head *bucket;
> + struct dlm_master_list_entry *mle;
> + struct hlist_node *tmp, *list;
> +
> + /*
> + * We notified all other nodes that we are exiting the domain and
> + * marked the dlm state to DLM_CTXT_LEAVING. If any mles are still
> + * around we force free them and wake any processes that are waiting
> + * on the mles
> + */
> + spin_lock(&dlm->spinlock);
> + spin_lock(&dlm->master_lock);
> +
> + BUG_ON(dlm->dlm_state != DLM_CTXT_LEAVING);
> + BUG_ON((find_next_bit(dlm->domain_map, O2NM_MAX_NODES, 0)< O2NM_MAX_NODES));
> +
> + for (i = 0; i< DLM_HASH_BUCKETS; i++) {
> + bucket = dlm_master_hash(dlm, i);
> + hlist_for_each_safe(list, tmp, bucket) {
> + mle = hlist_entry(list, struct dlm_master_list_entry,
> + master_hash_node);
> + if (mle->type != DLM_MLE_BLOCK) {
> + mlog(ML_ERROR, "bad mle: %p\n", mle);
> + dlm_print_one_mle(mle);
> + }
> + atomic_set(&mle->woken, 1);
> + wake_up(&mle->wq);
> +
> + __dlm_unlink_mle(dlm, mle);
> + __dlm_mle_detach_hb_events(dlm, mle);
> + __dlm_put_mle(mle);
> + }
> + }
> + spin_unlock(&dlm->master_lock);
> + spin_unlock(&dlm->spinlock);
> +}
>
^ permalink raw reply [flat|nested] 3+ messages in thread
* [Ocfs2-devel] [PATCH 1/1] o2dlm: force free mles during dlm exit
2010-09-21 23:27 [Ocfs2-devel] [PATCH 1/1] o2dlm: force free mles during dlm exit Srinivas Eeda
2010-09-21 23:50 ` Sunil Mushran
@ 2010-09-23 21:23 ` Joel Becker
1 sibling, 0 replies; 3+ messages in thread
From: Joel Becker @ 2010-09-23 21:23 UTC (permalink / raw)
To: ocfs2-devel
On Tue, Sep 21, 2010 at 04:27:26PM -0700, Srinivas Eeda wrote:
> While umounting, a block mle doesn't get freed if dlm is shutdown after
> master request is received but before assert master. This results in unclean
> shutdown of dlm domain.
>
> This patch frees all mles that lie around after other nodes were notified about
> exiting the dlm and marking dlm state as leaving. Only block mles are expected
> to be around, so we log ERROR for other mles but still free them.
>
> Signed-off-by: Srinivas Eeda <srinivas.eeda@oracle.com>
This patch is now in the 'fixes' branch of ocfs2.git.
Joel
--
To spot the expert, pick the one who predicts the job will take the
longest and cost the most.
Joel Becker
Consulting Software Developer
Oracle
E-mail: joel.becker at oracle.com
Phone: (650) 506-8127
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2010-09-23 21:23 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2010-09-21 23:27 [Ocfs2-devel] [PATCH 1/1] o2dlm: force free mles during dlm exit Srinivas Eeda
2010-09-21 23:50 ` Sunil Mushran
2010-09-23 21:23 ` Joel Becker
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).