aboutsummaryrefslogtreecommitdiff
path: root/fs/ocfs2/dlm/dlmrecovery.c
diff options
context:
space:
mode:
Diffstat (limited to 'fs/ocfs2/dlm/dlmrecovery.c')
-rw-r--r--fs/ocfs2/dlm/dlmrecovery.c82
1 files changed, 58 insertions, 24 deletions
diff --git a/fs/ocfs2/dlm/dlmrecovery.c b/fs/ocfs2/dlm/dlmrecovery.c
index 01ebfd0bdad..45067faf569 100644
--- a/fs/ocfs2/dlm/dlmrecovery.c
+++ b/fs/ocfs2/dlm/dlmrecovery.c
@@ -55,9 +55,6 @@
static void dlm_do_local_recovery_cleanup(struct dlm_ctxt *dlm, u8 dead_node);
static int dlm_recovery_thread(void *data);
-void dlm_complete_recovery_thread(struct dlm_ctxt *dlm);
-int dlm_launch_recovery_thread(struct dlm_ctxt *dlm);
-void dlm_kick_recovery_thread(struct dlm_ctxt *dlm);
static int dlm_do_recovery(struct dlm_ctxt *dlm);
static int dlm_pick_recovery_master(struct dlm_ctxt *dlm);
@@ -540,7 +537,10 @@ master_here:
/* success! see if any other nodes need recovery */
mlog(0, "DONE mastering recovery of %s:%u here(this=%u)!\n",
dlm->name, dlm->reco.dead_node, dlm->node_num);
- dlm_reset_recovery(dlm);
+ spin_lock(&dlm->spinlock);
+ __dlm_reset_recovery(dlm);
+ dlm->reco.state &= ~DLM_RECO_STATE_FINALIZE;
+ spin_unlock(&dlm->spinlock);
}
dlm_end_recovery(dlm);
@@ -698,6 +698,14 @@ static int dlm_remaster_locks(struct dlm_ctxt *dlm, u8 dead_node)
if (all_nodes_done) {
int ret;
+ /* Set this flag on recovery master to avoid
+ * a new recovery for another dead node start
+ * before the recovery is not done. That may
+ * cause recovery hung.*/
+ spin_lock(&dlm->spinlock);
+ dlm->reco.state |= DLM_RECO_STATE_FINALIZE;
+ spin_unlock(&dlm->spinlock);
+
/* all nodes are now in DLM_RECO_NODE_DATA_DONE state
* just send a finalize message to everyone and
* clean up */
@@ -789,7 +797,8 @@ static int dlm_request_all_locks(struct dlm_ctxt *dlm, u8 request_from,
u8 dead_node)
{
struct dlm_lock_request lr;
- enum dlm_status ret;
+ int ret;
+ int status;
mlog(0, "\n");
@@ -802,15 +811,16 @@ static int dlm_request_all_locks(struct dlm_ctxt *dlm, u8 request_from,
lr.dead_node = dead_node;
// send message
- ret = DLM_NOLOCKMGR;
ret = o2net_send_message(DLM_LOCK_REQUEST_MSG, dlm->key,
- &lr, sizeof(lr), request_from, NULL);
+ &lr, sizeof(lr), request_from, &status);
/* negative status is handled by caller */
if (ret < 0)
mlog(ML_ERROR, "%s: Error %d send LOCK_REQUEST to node %u "
"to recover dead node %u\n", dlm->name, ret,
request_from, dead_node);
+ else
+ ret = status;
// return from here, then
// sleep until all received or error
return ret;
@@ -1408,6 +1418,7 @@ int dlm_mig_lockres_handler(struct o2net_msg *msg, u32 len, void *data,
mres->lockname_len, mres->lockname);
ret = -EFAULT;
spin_unlock(&res->spinlock);
+ dlm_lockres_put(res);
goto leave;
}
res->state |= DLM_LOCK_RES_MIGRATING;
@@ -1498,10 +1509,8 @@ leave:
dlm_put(dlm);
if (ret < 0) {
- if (buf)
- kfree(buf);
- if (item)
- kfree(item);
+ kfree(buf);
+ kfree(item);
mlog_errno(ret);
}
@@ -1699,7 +1708,8 @@ int dlm_master_requery_handler(struct o2net_msg *msg, u32 len, void *data,
mlog_errno(-ENOMEM);
/* retry!? */
BUG();
- }
+ } else
+ __dlm_lockres_grab_inflight_worker(dlm, res);
} else /* put.. incase we are not the master */
dlm_lockres_put(res);
spin_unlock(&res->spinlock);
@@ -1752,13 +1762,13 @@ static int dlm_process_recovery_data(struct dlm_ctxt *dlm,
struct dlm_migratable_lockres *mres)
{
struct dlm_migratable_lock *ml;
- struct list_head *queue;
+ struct list_head *queue, *iter;
struct list_head *tmpq = NULL;
struct dlm_lock *newlock = NULL;
struct dlm_lockstatus *lksb = NULL;
int ret = 0;
int i, j, bad;
- struct dlm_lock *lock = NULL;
+ struct dlm_lock *lock;
u8 from = O2NM_MAX_NODES;
unsigned int added = 0;
__be64 c;
@@ -1793,14 +1803,16 @@ static int dlm_process_recovery_data(struct dlm_ctxt *dlm,
/* MIGRATION ONLY! */
BUG_ON(!(mres->flags & DLM_MRES_MIGRATION));
+ lock = NULL;
spin_lock(&res->spinlock);
for (j = DLM_GRANTED_LIST; j <= DLM_BLOCKED_LIST; j++) {
tmpq = dlm_list_idx_to_ptr(res, j);
- list_for_each_entry(lock, tmpq, list) {
- if (lock->ml.cookie != ml->cookie)
- lock = NULL;
- else
+ list_for_each(iter, tmpq) {
+ lock = list_entry(iter,
+ struct dlm_lock, list);
+ if (lock->ml.cookie == ml->cookie)
break;
+ lock = NULL;
}
if (lock)
break;
@@ -1888,6 +1900,13 @@ static int dlm_process_recovery_data(struct dlm_ctxt *dlm,
if (ml->type == LKM_NLMODE)
goto skip_lvb;
+ /*
+ * If the lock is in the blocked list it can't have a valid lvb,
+ * so skip it
+ */
+ if (ml->list == DLM_BLOCKED_LIST)
+ goto skip_lvb;
+
if (!dlm_lvb_is_empty(mres->lvb)) {
if (lksb->flags & DLM_LKSB_PUT_LVB) {
/* other node was trying to update
@@ -1968,7 +1987,15 @@ skip_lvb:
}
if (!bad) {
dlm_lock_get(newlock);
- list_add_tail(&newlock->list, queue);
+ if (mres->flags & DLM_MRES_RECOVERY &&
+ ml->list == DLM_CONVERTING_LIST &&
+ newlock->ml.type >
+ newlock->ml.convert_type) {
+ /* newlock is doing downconvert, add it to the
+ * head of converting list */
+ list_add(&newlock->list, queue);
+ } else
+ list_add_tail(&newlock->list, queue);
mlog(0, "%s:%.*s: added lock for node %u, "
"setting refmap bit\n", dlm->name,
res->lockname.len, res->lockname.name, ml->node);
@@ -2083,7 +2110,6 @@ static void dlm_finish_local_lockres_recovery(struct dlm_ctxt *dlm,
u8 dead_node, u8 new_master)
{
int i;
- struct hlist_node *hash_iter;
struct hlist_head *bucket;
struct dlm_lock_resource *res, *next;
@@ -2114,7 +2140,7 @@ static void dlm_finish_local_lockres_recovery(struct dlm_ctxt *dlm,
* if necessary */
for (i = 0; i < DLM_HASH_BUCKETS; i++) {
bucket = dlm_lockres_hash(dlm, i);
- hlist_for_each_entry(res, hash_iter, bucket, hash_node) {
+ hlist_for_each_entry(res, bucket, hash_node) {
if (!(res->state & DLM_LOCK_RES_RECOVERING))
continue;
@@ -2273,7 +2299,6 @@ static void dlm_free_dead_locks(struct dlm_ctxt *dlm,
static void dlm_do_local_recovery_cleanup(struct dlm_ctxt *dlm, u8 dead_node)
{
- struct hlist_node *iter;
struct dlm_lock_resource *res;
int i;
struct hlist_head *bucket;
@@ -2299,7 +2324,7 @@ static void dlm_do_local_recovery_cleanup(struct dlm_ctxt *dlm, u8 dead_node)
*/
for (i = 0; i < DLM_HASH_BUCKETS; i++) {
bucket = dlm_lockres_hash(dlm, i);
- hlist_for_each_entry(res, iter, bucket, hash_node) {
+ hlist_for_each_entry(res, bucket, hash_node) {
/* always prune any $RECOVERY entries for dead nodes,
* otherwise hangs can occur during later recovery */
if (dlm_is_recovery_lock(res->lockname.name,
@@ -2335,6 +2360,14 @@ static void dlm_do_local_recovery_cleanup(struct dlm_ctxt *dlm, u8 dead_node)
} else if (res->owner == dlm->node_num) {
dlm_free_dead_locks(dlm, res, dead_node);
__dlm_lockres_calc_usage(dlm, res);
+ } else if (res->owner == DLM_LOCK_RES_OWNER_UNKNOWN) {
+ if (test_bit(dead_node, res->refmap)) {
+ mlog(0, "%s:%.*s: dead node %u had a ref, but had "
+ "no locks and had not purged before dying\n",
+ dlm->name, res->lockname.len,
+ res->lockname.name, dead_node);
+ dlm_lockres_clear_refmap_bit(dlm, res, dead_node);
+ }
}
spin_unlock(&res->spinlock);
}
@@ -2699,6 +2732,7 @@ int dlm_begin_reco_handler(struct o2net_msg *msg, u32 len, void *data,
dlm->name, br->node_idx, br->dead_node,
dlm->reco.dead_node, dlm->reco.new_master);
spin_unlock(&dlm->spinlock);
+ dlm_put(dlm);
return -EAGAIN;
}
spin_unlock(&dlm->spinlock);
@@ -2870,8 +2904,8 @@ int dlm_finalize_reco_handler(struct o2net_msg *msg, u32 len, void *data,
BUG();
}
dlm->reco.state &= ~DLM_RECO_STATE_FINALIZE;
+ __dlm_reset_recovery(dlm);
spin_unlock(&dlm->spinlock);
- dlm_reset_recovery(dlm);
dlm_kick_recovery_thread(dlm);
break;
default:
generated by cgit v1.2.3-70-g09d2 (git 2.46.0) at 2025-11-09 08:41:16 +0000