aboutsummaryrefslogtreecommitdiff
path: root/fs/ocfs2/dlm
diff options
context:
space:
mode:
Diffstat (limited to 'fs/ocfs2/dlm')
-rw-r--r--fs/ocfs2/dlm/dlmcommon.h5
-rw-r--r--fs/ocfs2/dlm/dlmdebug.c2
-rw-r--r--fs/ocfs2/dlm/dlmdomain.c41
-rw-r--r--fs/ocfs2/dlm/dlmlock.c2
-rw-r--r--fs/ocfs2/dlm/dlmmaster.c71
-rw-r--r--fs/ocfs2/dlm/dlmrecovery.c42
-rw-r--r--fs/ocfs2/dlm/dlmthread.c13
-rw-r--r--fs/ocfs2/dlm/dlmunlock.c18
8 files changed, 143 insertions, 51 deletions
diff --git a/fs/ocfs2/dlm/dlmcommon.h b/fs/ocfs2/dlm/dlmcommon.h
index e0517762fcc..fae17c640df 100644
--- a/fs/ocfs2/dlm/dlmcommon.h
+++ b/fs/ocfs2/dlm/dlmcommon.h
@@ -108,7 +108,6 @@ static inline int dlm_is_recovery_lock(const char *lock_name, int name_len)
struct dlm_recovery_ctxt
{
struct list_head resources;
- struct list_head received;
struct list_head node_data;
u8 new_master;
u8 dead_node;
@@ -332,6 +331,7 @@ struct dlm_lock_resource
u16 state;
char lvb[DLM_LVB_LEN];
unsigned int inflight_locks;
+ unsigned int inflight_assert_workers;
unsigned long refmap[BITS_TO_LONGS(O2NM_MAX_NODES)];
};
@@ -911,6 +911,9 @@ void dlm_lockres_drop_inflight_ref(struct dlm_ctxt *dlm,
void dlm_lockres_grab_inflight_ref(struct dlm_ctxt *dlm,
struct dlm_lock_resource *res);
+void __dlm_lockres_grab_inflight_worker(struct dlm_ctxt *dlm,
+ struct dlm_lock_resource *res);
+
void dlm_queue_ast(struct dlm_ctxt *dlm, struct dlm_lock *lock);
void dlm_queue_bast(struct dlm_ctxt *dlm, struct dlm_lock *lock);
void __dlm_queue_ast(struct dlm_ctxt *dlm, struct dlm_lock *lock);
diff --git a/fs/ocfs2/dlm/dlmdebug.c b/fs/ocfs2/dlm/dlmdebug.c
index e33cd7a3c58..18f13c2e4a1 100644
--- a/fs/ocfs2/dlm/dlmdebug.c
+++ b/fs/ocfs2/dlm/dlmdebug.c
@@ -338,7 +338,7 @@ void dlm_print_one_mle(struct dlm_master_list_entry *mle)
#ifdef CONFIG_DEBUG_FS
-static struct dentry *dlm_debugfs_root = NULL;
+static struct dentry *dlm_debugfs_root;
#define DLM_DEBUGFS_DIR "o2dlm"
#define DLM_DEBUGFS_DLM_STATE "dlm_state"
diff --git a/fs/ocfs2/dlm/dlmdomain.c b/fs/ocfs2/dlm/dlmdomain.c
index 33660a4a52f..39efc5057a3 100644
--- a/fs/ocfs2/dlm/dlmdomain.c
+++ b/fs/ocfs2/dlm/dlmdomain.c
@@ -959,6 +959,14 @@ static int dlm_assert_joined_handler(struct o2net_msg *msg, u32 len, void *data,
* domain. Set him in the map and clean up our
* leftover join state. */
BUG_ON(dlm->joining_node != assert->node_idx);
+
+ if (dlm->reco.state & DLM_RECO_STATE_ACTIVE) {
+ mlog(0, "dlm recovery is ongoing, disallow join\n");
+ spin_unlock(&dlm->spinlock);
+ spin_unlock(&dlm_domain_lock);
+ return -EAGAIN;
+ }
+
set_bit(assert->node_idx, dlm->domain_map);
clear_bit(assert->node_idx, dlm->exit_domain_map);
__dlm_set_joining_node(dlm, DLM_LOCK_RES_OWNER_UNKNOWN);
@@ -1123,7 +1131,6 @@ static int dlm_query_region_handler(struct o2net_msg *msg, u32 len,
struct dlm_ctxt *dlm = NULL;
char *local = NULL;
int status = 0;
- int locked = 0;
qr = (struct dlm_query_region *) msg->buf;
@@ -1132,10 +1139,8 @@ static int dlm_query_region_handler(struct o2net_msg *msg, u32 len,
/* buffer used in dlm_mast_regions() */
local = kmalloc(sizeof(qr->qr_regions), GFP_KERNEL);
- if (!local) {
- status = -ENOMEM;
- goto bail;
- }
+ if (!local)
+ return -ENOMEM;
status = -EINVAL;
@@ -1144,16 +1149,15 @@ static int dlm_query_region_handler(struct o2net_msg *msg, u32 len,
if (!dlm) {
mlog(ML_ERROR, "Node %d queried hb regions on domain %s "
"before join domain\n", qr->qr_node, qr->qr_domain);
- goto bail;
+ goto out_domain_lock;
}
spin_lock(&dlm->spinlock);
- locked = 1;
if (dlm->joining_node != qr->qr_node) {
mlog(ML_ERROR, "Node %d queried hb regions on domain %s "
"but joining node is %d\n", qr->qr_node, qr->qr_domain,
dlm->joining_node);
- goto bail;
+ goto out_dlm_lock;
}
/* Support for global heartbeat was added in 1.1 */
@@ -1163,14 +1167,15 @@ static int dlm_query_region_handler(struct o2net_msg *msg, u32 len,
"but active dlm protocol is %d.%d\n", qr->qr_node,
qr->qr_domain, dlm->dlm_locking_proto.pv_major,
dlm->dlm_locking_proto.pv_minor);
- goto bail;
+ goto out_dlm_lock;
}
status = dlm_match_regions(dlm, qr, local, sizeof(qr->qr_regions));
-bail:
- if (locked)
- spin_unlock(&dlm->spinlock);
+out_dlm_lock:
+ spin_unlock(&dlm->spinlock);
+
+out_domain_lock:
spin_unlock(&dlm_domain_lock);
kfree(local);
@@ -1520,6 +1525,7 @@ static int dlm_send_one_join_assert(struct dlm_ctxt *dlm,
unsigned int node)
{
int status;
+ int ret;
struct dlm_assert_joined assert_msg;
mlog(0, "Sending join assert to node %u\n", node);
@@ -1531,11 +1537,13 @@ static int dlm_send_one_join_assert(struct dlm_ctxt *dlm,
status = o2net_send_message(DLM_ASSERT_JOINED_MSG, DLM_MOD_KEY,
&assert_msg, sizeof(assert_msg), node,
- NULL);
+ &ret);
if (status < 0)
mlog(ML_ERROR, "Error %d when sending message %u (key 0x%x) to "
"node %u\n", status, DLM_ASSERT_JOINED_MSG, DLM_MOD_KEY,
node);
+ else
+ status = ret;
return status;
}
@@ -1877,19 +1885,19 @@ static int dlm_join_domain(struct dlm_ctxt *dlm)
goto bail;
}
- status = dlm_debug_init(dlm);
+ status = dlm_launch_thread(dlm);
if (status < 0) {
mlog_errno(status);
goto bail;
}
- status = dlm_launch_thread(dlm);
+ status = dlm_launch_recovery_thread(dlm);
if (status < 0) {
mlog_errno(status);
goto bail;
}
- status = dlm_launch_recovery_thread(dlm);
+ status = dlm_debug_init(dlm);
if (status < 0) {
mlog_errno(status);
goto bail;
@@ -2026,7 +2034,6 @@ static struct dlm_ctxt *dlm_alloc_ctxt(const char *domain,
INIT_LIST_HEAD(&dlm->list);
INIT_LIST_HEAD(&dlm->dirty_list);
INIT_LIST_HEAD(&dlm->reco.resources);
- INIT_LIST_HEAD(&dlm->reco.received);
INIT_LIST_HEAD(&dlm->reco.node_data);
INIT_LIST_HEAD(&dlm->purge_list);
INIT_LIST_HEAD(&dlm->dlm_domain_handlers);
diff --git a/fs/ocfs2/dlm/dlmlock.c b/fs/ocfs2/dlm/dlmlock.c
index 5d32f7511f7..66c2a491f68 100644
--- a/fs/ocfs2/dlm/dlmlock.c
+++ b/fs/ocfs2/dlm/dlmlock.c
@@ -52,7 +52,7 @@
#define MLOG_MASK_PREFIX ML_DLM
#include "cluster/masklog.h"
-static struct kmem_cache *dlm_lock_cache = NULL;
+static struct kmem_cache *dlm_lock_cache;
static DEFINE_SPINLOCK(dlm_cookie_lock);
static u64 dlm_next_cookie = 1;
diff --git a/fs/ocfs2/dlm/dlmmaster.c b/fs/ocfs2/dlm/dlmmaster.c
index af3f7aa73e1..82abf0cc9a1 100644
--- a/fs/ocfs2/dlm/dlmmaster.c
+++ b/fs/ocfs2/dlm/dlmmaster.c
@@ -82,9 +82,9 @@ static inline int dlm_mle_equal(struct dlm_ctxt *dlm,
return 1;
}
-static struct kmem_cache *dlm_lockres_cache = NULL;
-static struct kmem_cache *dlm_lockname_cache = NULL;
-static struct kmem_cache *dlm_mle_cache = NULL;
+static struct kmem_cache *dlm_lockres_cache;
+static struct kmem_cache *dlm_lockname_cache;
+static struct kmem_cache *dlm_mle_cache;
static void dlm_mle_release(struct kref *kref);
static void dlm_init_mle(struct dlm_master_list_entry *mle,
@@ -472,11 +472,15 @@ bail:
void dlm_destroy_master_caches(void)
{
- if (dlm_lockname_cache)
+ if (dlm_lockname_cache) {
kmem_cache_destroy(dlm_lockname_cache);
+ dlm_lockname_cache = NULL;
+ }
- if (dlm_lockres_cache)
+ if (dlm_lockres_cache) {
kmem_cache_destroy(dlm_lockres_cache);
+ dlm_lockres_cache = NULL;
+ }
}
static void dlm_lockres_release(struct kref *kref)
@@ -577,6 +581,7 @@ static void dlm_init_lockres(struct dlm_ctxt *dlm,
atomic_set(&res->asts_reserved, 0);
res->migration_pending = 0;
res->inflight_locks = 0;
+ res->inflight_assert_workers = 0;
res->dlm = dlm;
@@ -679,6 +684,43 @@ void dlm_lockres_drop_inflight_ref(struct dlm_ctxt *dlm,
wake_up(&res->wq);
}
+void __dlm_lockres_grab_inflight_worker(struct dlm_ctxt *dlm,
+ struct dlm_lock_resource *res)
+{
+ assert_spin_locked(&res->spinlock);
+ res->inflight_assert_workers++;
+ mlog(0, "%s:%.*s: inflight assert worker++: now %u\n",
+ dlm->name, res->lockname.len, res->lockname.name,
+ res->inflight_assert_workers);
+}
+
+static void dlm_lockres_grab_inflight_worker(struct dlm_ctxt *dlm,
+ struct dlm_lock_resource *res)
+{
+ spin_lock(&res->spinlock);
+ __dlm_lockres_grab_inflight_worker(dlm, res);
+ spin_unlock(&res->spinlock);
+}
+
+static void __dlm_lockres_drop_inflight_worker(struct dlm_ctxt *dlm,
+ struct dlm_lock_resource *res)
+{
+ assert_spin_locked(&res->spinlock);
+ BUG_ON(res->inflight_assert_workers == 0);
+ res->inflight_assert_workers--;
+ mlog(0, "%s:%.*s: inflight assert worker--: now %u\n",
+ dlm->name, res->lockname.len, res->lockname.name,
+ res->inflight_assert_workers);
+}
+
+static void dlm_lockres_drop_inflight_worker(struct dlm_ctxt *dlm,
+ struct dlm_lock_resource *res)
+{
+ spin_lock(&res->spinlock);
+ __dlm_lockres_drop_inflight_worker(dlm, res);
+ spin_unlock(&res->spinlock);
+}
+
/*
* lookup a lock resource by name.
* may already exist in the hashtable.
@@ -1599,7 +1641,8 @@ send_response:
mlog(ML_ERROR, "failed to dispatch assert master work\n");
response = DLM_MASTER_RESP_ERROR;
dlm_lockres_put(res);
- }
+ } else
+ dlm_lockres_grab_inflight_worker(dlm, res);
} else {
if (res)
dlm_lockres_put(res);
@@ -2114,6 +2157,8 @@ static void dlm_assert_master_worker(struct dlm_work_item *item, void *data)
dlm_lockres_release_ast(dlm, res);
put:
+ dlm_lockres_drop_inflight_worker(dlm, res);
+
dlm_lockres_put(res);
mlog(0, "finished with dlm_assert_master_worker\n");
@@ -3084,11 +3129,15 @@ static int dlm_add_migration_mle(struct dlm_ctxt *dlm,
/* remove it so that only one mle will be found */
__dlm_unlink_mle(dlm, tmp);
__dlm_mle_detach_hb_events(dlm, tmp);
- ret = DLM_MIGRATE_RESPONSE_MASTERY_REF;
- mlog(0, "%s:%.*s: master=%u, newmaster=%u, "
- "telling master to get ref for cleared out mle "
- "during migration\n", dlm->name, namelen, name,
- master, new_master);
+ if (tmp->type == DLM_MLE_MASTER) {
+ ret = DLM_MIGRATE_RESPONSE_MASTERY_REF;
+ mlog(0, "%s:%.*s: master=%u, newmaster=%u, "
+ "telling master to get ref "
+ "for cleared out mle during "
+ "migration\n", dlm->name,
+ namelen, name, master,
+ new_master);
+ }
}
spin_unlock(&tmp->spinlock);
}
diff --git a/fs/ocfs2/dlm/dlmrecovery.c b/fs/ocfs2/dlm/dlmrecovery.c
index 7035af09cc0..45067faf569 100644
--- a/fs/ocfs2/dlm/dlmrecovery.c
+++ b/fs/ocfs2/dlm/dlmrecovery.c
@@ -537,7 +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);
@@ -695,6 +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 */
@@ -1697,7 +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);
@@ -1750,13 +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;
@@ -1791,14 +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;
@@ -1973,7 +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);
@@ -2882,8 +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:
diff --git a/fs/ocfs2/dlm/dlmthread.c b/fs/ocfs2/dlm/dlmthread.c
index 9db869de829..69aac6f088a 100644
--- a/fs/ocfs2/dlm/dlmthread.c
+++ b/fs/ocfs2/dlm/dlmthread.c
@@ -259,12 +259,15 @@ static void dlm_run_purge_list(struct dlm_ctxt *dlm,
* refs on it. */
unused = __dlm_lockres_unused(lockres);
if (!unused ||
- (lockres->state & DLM_LOCK_RES_MIGRATING)) {
+ (lockres->state & DLM_LOCK_RES_MIGRATING) ||
+ (lockres->inflight_assert_workers != 0)) {
mlog(0, "%s: res %.*s is in use or being remastered, "
- "used %d, state %d\n", dlm->name,
- lockres->lockname.len, lockres->lockname.name,
- !unused, lockres->state);
- list_move_tail(&dlm->purge_list, &lockres->purge);
+ "used %d, state %d, assert master workers %u\n",
+ dlm->name, lockres->lockname.len,
+ lockres->lockname.name,
+ !unused, lockres->state,
+ lockres->inflight_assert_workers);
+ list_move_tail(&lockres->purge, &dlm->purge_list);
spin_unlock(&lockres->spinlock);
continue;
}
diff --git a/fs/ocfs2/dlm/dlmunlock.c b/fs/ocfs2/dlm/dlmunlock.c
index 5698b52cf5c..2e3c9dbab68 100644
--- a/fs/ocfs2/dlm/dlmunlock.c
+++ b/fs/ocfs2/dlm/dlmunlock.c
@@ -191,7 +191,9 @@ static enum dlm_status dlmunlock_common(struct dlm_ctxt *dlm,
DLM_UNLOCK_CLEAR_CONVERT_TYPE);
} else if (status == DLM_RECOVERING ||
status == DLM_MIGRATING ||
- status == DLM_FORWARD) {
+ status == DLM_FORWARD ||
+ status == DLM_NOLOCKMGR
+ ) {
/* must clear the actions because this unlock
* is about to be retried. cannot free or do
* any list manipulation. */
@@ -200,7 +202,8 @@ static enum dlm_status dlmunlock_common(struct dlm_ctxt *dlm,
res->lockname.name,
status==DLM_RECOVERING?"recovering":
(status==DLM_MIGRATING?"migrating":
- "forward"));
+ (status == DLM_FORWARD ? "forward" :
+ "nolockmanager")));
actions = 0;
}
if (flags & LKM_CANCEL)
@@ -364,7 +367,10 @@ static enum dlm_status dlm_send_remote_unlock_request(struct dlm_ctxt *dlm,
* updated state to the recovery master. this thread
* just needs to finish out the operation and call
* the unlockast. */
- ret = DLM_NORMAL;
+ if (dlm_is_node_dead(dlm, owner))
+ ret = DLM_NORMAL;
+ else
+ ret = DLM_NOLOCKMGR;
} else {
/* something bad. this will BUG in ocfs2 */
ret = dlm_err_to_dlm_status(tmpret);
@@ -638,7 +644,9 @@ retry:
if (status == DLM_RECOVERING ||
status == DLM_MIGRATING ||
- status == DLM_FORWARD) {
+ status == DLM_FORWARD ||
+ status == DLM_NOLOCKMGR) {
+
/* We want to go away for a tiny bit to allow recovery
* / migration to complete on this resource. I don't
* know of any wait queue we could sleep on as this
@@ -650,7 +658,7 @@ retry:
msleep(50);
mlog(0, "retrying unlock due to pending recovery/"
- "migration/in-progress\n");
+ "migration/in-progress/reconnect\n");
goto retry;
}
generated by cgit v1.2.3-70-g09d2 (git 2.46.0) at 2025-12-12 12:46:17 +0000