Skip to contentSkip to conversation

Introduce ocfs2_recover_node

PATCH New Goldwyn Rodrigues 3 participants 5 messages Nov 17, 2010 – Dec 2, 2010 OCFS2

lore
Share & tools
lore
#0 lore raw

PATCH 1/2 Introduce ocfs2_recover_node

Review comments by Wengang incorporated.

ocfs2_recover_node is a new data structure to include all information required to recover one node. The structure is maintainer as a list in the super block.

recovery_map is no longer required with ocfs2_recover_node.

---
 fs/ocfs2/journal.c |  171 ++++++++++++++++++++++++----------------------------
 fs/ocfs2/journal.h |    9 ++-
 fs/ocfs2/ocfs2.h   |    3 +-
 fs/ocfs2/super.c   |   16 -----
 4 files changed, 88 insertions(+), 111 deletions(-)
fs/ocfs2/journal.c +2 -16
index faa2303..277b810 100644
@@ -58,8 +58,7 @@ DEFINE_SPINLOCK(trans_inc_lock);
 #define ORPHAN_SCAN_SCHEDULE_TIMEOUT 300000
 
 static int ocfs2_force_read_journal(struct inode *inode);
-static int ocfs2_recover_node(struct ocfs2_super *osb,
-			      int node_num, int slot_num);
+static int ocfs2_recover_one_node(struct ocfs2_recover_node *rn);
 static int __ocfs2_recovery_thread(void *arg);
 static int ocfs2_commit_cache(struct ocfs2_super *osb);
 static int __ocfs2_wait_on_mount(struct ocfs2_super *osb, int quota);
@@ -179,24 +178,11 @@ void ocfs2_free_replay_slots(struct ocfs2_super *osb)
 
 int ocfs2_recovery_init(struct ocfs2_super *osb)
 {
-	struct ocfs2_recovery_map *rm;
-
 	mutex_init(&osb->recovery_lock);
 	osb->disable_recovery = 0;
 	osb->recovery_thread_task = NULL;
 	init_waitqueue_head(&osb->recovery_event);
-
-	rm = kzalloc(sizeof(struct ocfs2_recovery_map) +
-		     osb->max_slots * sizeof(unsigned int),
-		     GFP_KERNEL);
-	if (!rm) {
-		mlog_errno(-ENOMEM);
-		return -ENOMEM;
-	}
-
-	rm->rm_entries = (unsigned int *)((char *)rm +
-					  sizeof(struct ocfs2_recovery_map));
-	osb->recovery_map = rm;
+	INIT_LIST_HEAD(&osb->s_recovery_nodes);
 
 	return 0;
 }
@@ -212,8 +198,8 @@ static int ocfs2_recovery_thread_running(struct

ocfs2_super *osb)

void ocfs2_recovery_exit(struct ocfs2_super *osb)
{
- struct ocfs2_recovery_map *rm;
-
+ struct list_head *iter, *next;
+ struct ocfs2_recover_node *rn;
/* disable any new recovery threads and wait for any currently
* running ones to exit. Do this before setting the vol_state. */
mutex_lock(&osb->recovery_lock);
@@ -226,75 +212,67 @@ void ocfs2_recovery_exit(struct ocfs2_super *osb)
* complete. */
flush_workqueue(ocfs2_wq);

- /*
- * Now that recovery is shut down, and the osb is about to be
- * freed, the osb_lock is not taken here.
- */
- rm = osb->recovery_map;
- /* XXX: Should we bug if there are dirty entries? */
-
- kfree(rm);
+ spin_lock(&osb->osb_lock);
+ list_for_each_safe(iter, next, &osb->s_recovery_nodes) {
+ rn = list_entry(iter, struct ocfs2_recover_node,
+ rn_list);
+ list_del(&rn->rn_list);
+ kfree(rn);
+ }
+ spin_unlock(&osb->osb_lock);
}

-static int __ocfs2_recovery_map_test(struct ocfs2_super *osb, +static int __ocfs2_recovery_node_test(struct ocfs2_super *osb,
unsigned int node_num)
{
- int i;
- struct ocfs2_recovery_map *rm = osb->recovery_map;
+ struct ocfs2_recover_node *rn;
+ struct list_head *iter;

assert_spin_locked(&osb->osb_lock);
-
- for (i = 0; i < rm->rm_used; i++) {
- if (rm->rm_entries[i] == node_num)
+ list_for_each(iter, &osb->s_recovery_nodes) {
+ rn = list_entry(iter, struct ocfs2_recover_node, rn_list);
+ if (rn->rn_node_num == node_num)
return 1;
}
-
return 0;
}

/* Behaves like test-and-set. Returns the previous value */
-static int ocfs2_recovery_map_set(struct ocfs2_super *osb, +static int ocfs2_recovery_node_set(struct ocfs2_super *osb,
unsigned int node_num)
{
- struct ocfs2_recovery_map *rm = osb->recovery_map;
-
- spin_lock(&osb->osb_lock);
- if (__ocfs2_recovery_map_test(osb, node_num)) {
- spin_unlock(&osb->osb_lock);
- return 1;
- }
+ struct list_head *iter;
+ struct ocfs2_recover_node *rn = NULL, *trn;

- /* XXX: Can this be exploited? Not from o2dlm... */
- BUG_ON(rm->rm_used >= osb->max_slots);
+ trn = kzalloc(sizeof(struct ocfs2_recover_node), GFP_NOFS);
+ if (!trn)
+ return -ENOMEM;

- rm->rm_entries[rm->rm_used] = node_num;
- rm->rm_used++;
+ spin_lock(&osb->osb_lock);
+ list_for_each(iter, &osb->s_recovery_nodes) {
+ rn = list_entry(iter, struct ocfs2_recover_node,
+ rn_list);
+ if (rn->rn_node_num == node_num) {
+ spin_unlock(&osb->osb_lock);
+ kfree(trn);
+ return 1;
+ }
+ }
+ trn->rn_node_num = node_num;
+ trn->rn_osb = osb;
+ list_add_tail(&trn->rn_list, &osb->s_recovery_nodes);
spin_unlock(&osb->osb_lock);

return 0;
}

-static void ocfs2_recovery_map_clear(struct ocfs2_super *osb,
- unsigned int node_num)
+static void ocfs2_recovery_node_clear(struct ocfs2_super *osb,
+ struct ocfs2_recover_node *rn)
{
- int i;
- struct ocfs2_recovery_map *rm = osb->recovery_map;
-
spin_lock(&osb->osb_lock);
-
- for (i = 0; i < rm->rm_used; i++) {
- if (rm->rm_entries[i] == node_num)
- break;
- }
-
- if (i < rm->rm_used) {
- /* XXX: be careful with the pointer math */
- memmove(&(rm->rm_entries[i]), &(rm->rm_entries[i + 1]),
- (rm->rm_used - i - 1) * sizeof(unsigned int));
- rm->rm_used--;
- }
-
+ list_del(&rn->rn_list);
spin_unlock(&osb->osb_lock);
+ kfree(rn);
}

static int ocfs2_commit_cache(struct ocfs2_super *osb)
@@ -1092,10 +1070,9 @@ bail:
static int ocfs2_recovery_completed(struct ocfs2_super *osb) {
int empty;
- struct ocfs2_recovery_map *rm = osb->recovery_map;

spin_lock(&osb->osb_lock);
- empty = (rm->rm_used == 0);
+ empty = list_empty(&osb->s_recovery_nodes);
spin_unlock(&osb->osb_lock);

return empty;
@@ -1332,12 +1309,13 @@ void ocfs2_complete_quota_recovery(struct ocfs2_super *osb)

static int __ocfs2_recovery_thread(void *arg)
{
- int status, node_num, slot_num;
+ int status;
struct ocfs2_super *osb = arg;
- struct ocfs2_recovery_map *rm = osb->recovery_map;
int *rm_quota = NULL;
int rm_quota_used = 0, i;
struct ocfs2_quota_recovery *qrec;
+ struct list_head *iter;
+ struct ocfs2_recover_node *rn = NULL;

mlog_entry_void();

@@ -1367,20 +1345,21 @@ restart:
NULL, NULL);

spin_lock(&osb->osb_lock);
- while (rm->rm_used) {
+ list_for_each(iter, &osb->s_recovery_nodes) {
+ rn = list_entry(iter, struct ocfs2_recover_node, rn_list);
/* It's always safe to remove entry zero, as we won't
* clear it until ocfs2_recover_node() has succeeded. */
- node_num = rm->rm_entries[0];
spin_unlock(&osb->osb_lock);
- mlog(0, "checking node %d\n", node_num);
- slot_num = ocfs2_node_num_to_slot(osb, node_num);
- if (slot_num == -ENOENT) {
+ mlog(0, "checking node %d\n", rn->rn_node_num);
+ rn->rn_slot_num = ocfs2_node_num_to_slot(osb, rn->rn_node_num);
+ if (rn->rn_slot_num == -ENOENT) {
status = 0;
mlog(0, "no slot for this node, so no recovery"
"required.\n");
goto skip_recovery;
}
- mlog(0, "node %d was using slot %d\n", node_num, slot_num);
+ mlog(0, "node %d was using slot %d\n",
+ rn->rn_node_num, rn->rn_slot_num);

/* It is a bit subtle with quota recovery. We cannot do it
* immediately because we have to obtain cluster locks from
@@ -1388,18 +1367,19 @@ restart:
* then quota usage would be out of sync until some node takes
* the slot. So we remember which nodes need quota recovery
* and when everything else is done, we recover quotas. */
- for (i = 0; i < rm_quota_used && rm_quota[i] != slot_num; i++);
+ for (i = 0; i < rm_quota_used && rm_quota[i] != rn->rn_slot_num;
+ i++);
if (i == rm_quota_used)
- rm_quota[rm_quota_used++] = slot_num;
+ rm_quota[rm_quota_used++] = rn->rn_slot_num;

- status = ocfs2_recover_node(osb, node_num, slot_num);
+ status = ocfs2_recover_one_node(rn);
skip_recovery:
if (!status) {
- ocfs2_recovery_map_clear(osb, node_num);
+ ocfs2_recovery_node_clear(osb, rn);
} else {
mlog(ML_ERROR,
"Error %d recovering node %d on device (%u,%u)!\n",
- status, node_num,
+ status, rn->rn_node_num,
MAJOR(osb->sb->s_dev), MINOR(osb->sb->s_dev));
mlog(ML_ERROR, "Volume requires unmount.\n");
}
@@ -1461,6 +1441,7 @@ bail:

void ocfs2_recovery_thread(struct ocfs2_super *osb, int node_num) {
+ int ret = 0;
mlog_entry("(node_num=%d, osb->node_num = %d)\n",
node_num, osb->node_num);

@@ -1470,7 +1451,11 @@ void ocfs2_recovery_thread(struct ocfs2_super *osb, int node_num)

/* People waiting on recovery will wait on
* the recovery map to empty. */
- if (ocfs2_recovery_map_set(osb, node_num))
+ ret = ocfs2_recovery_node_set(osb, node_num);
+ if (ret == -ENOMEM) {
+ mlog_errno(ret);
+ goto out;
+ } else if (ret)
mlog(0, "node %d already in recovery map.\n", node_num);

mlog(0, "starting recovery thread...\n");
@@ -1681,26 +1666,27 @@ done:
* second part of a nodes recovery process (local alloc recovery) is
* far less concerning.
*/
-static int ocfs2_recover_node(struct ocfs2_super *osb,
- int node_num, int slot_num)
+static int ocfs2_recover_one_node(struct ocfs2_recover_node *rn)
{
int status = 0;
struct ocfs2_dinode *la_copy = NULL;
struct ocfs2_dinode *tl_copy = NULL;
+ struct ocfs2_super *osb = rn->rn_osb;

mlog_entry("(node_num=%d, slot_num=%d, osb->node_num = %d)\n",
- node_num, slot_num, osb->node_num);
+ rn->rn_node_num, rn->rn_slot_num, osb->node_num);

/* Should not ever be called to recover ourselves -- in that
* case we should've called ocfs2_journal_load instead. */
- BUG_ON(osb->node_num == node_num);
+ BUG_ON(osb->node_num == rn->rn_node_num);

- status = ocfs2_replay_journal(osb, node_num, slot_num);
+ status = ocfs2_replay_journal(osb, rn->rn_node_num,
+ rn->rn_slot_num);
if (status < 0) {
if (status == -EBUSY) {
mlog(0, "Skipping recovery for slot %u (node %u) "

-			     "as another node has recovered it\n", slot_num,
-			     node_num);
+			     "as another node has recovered it\n",
+			     rn->rn_slot_num, rn->rn_node_num);

status = 0;
goto done;
}
@@ -1709,7 +1695,8 @@ static int ocfs2_recover_node(struct ocfs2_super *osb,
}

/* Stamp a clean local alloc file AFTER recovering the journal... */
- status = ocfs2_begin_local_alloc_recovery(osb, slot_num, &la_copy);
+ status = ocfs2_begin_local_alloc_recovery(osb, rn->rn_slot_num,
+ &la_copy);
if (status < 0) {
mlog_errno(status);
goto done;
@@ -1718,22 +1705,24 @@ static int ocfs2_recover_node(struct ocfs2_super *osb,
/* An error from begin_truncate_log_recovery is not
* serious enough to warrant halting the rest of
* recovery. */
- status = ocfs2_begin_truncate_log_recovery(osb, slot_num, &tl_copy);
+ status = ocfs2_begin_truncate_log_recovery(osb,
+ rn->rn_slot_num, &tl_copy);
if (status < 0)
mlog_errno(status);

/* Likewise, this would be a strange but ultimately not so
* harmful place to get an error... */
- status = ocfs2_clear_slot(osb, slot_num);
+ status = ocfs2_clear_slot(osb, rn->rn_slot_num);
if (status < 0)
mlog_errno(status);

/* This will kfree the memory pointed to by la_copy and tl_copy */
- ocfs2_queue_recovery_completion(osb->journal, slot_num, la_copy,
- tl_copy, NULL);
+ ocfs2_queue_recovery_completion(osb->journal, rn->rn_slot_num,
+ la_copy, tl_copy, NULL);

status = 0;
done:
+ rn->rn_status = status;

mlog_exit(status);
return status;
@@ -1822,7 +1811,7 @@ int ocfs2_mark_dead_nodes(struct ocfs2_super *osb)
continue;
}

- if (__ocfs2_recovery_map_test(osb, node_num)) {
+ if (__ocfs2_recovery_node_test(osb, node_num)) {
spin_unlock(&osb->osb_lock); continue;
}

fs/ocfs2/journal.h +6 -3
index 43e56b9..0325d81 100644
@@ -43,9 +43,12 @@ struct ocfs2_dinode;
  * It is protected by the recovery_lock.
  */
 
-struct ocfs2_recovery_map {
-	unsigned int rm_used;
-	unsigned int *rm_entries;
+struct ocfs2_recover_node {
+	struct ocfs2_super *rn_osb;
+	int rn_node_num;
+	int rn_slot_num;
+	int rn_status;
+	struct list_head rn_list;
 };
 
 
fs/ocfs2/ocfs2.h +2 -1
index 1efea36..f70c25a 100644
@@ -337,7 +337,8 @@ struct ocfs2_super
 
 	atomic_t vol_state;
 	struct mutex recovery_lock;
-	struct ocfs2_recovery_map *recovery_map;
+	struct list_head s_recovery_nodes;
+
 	struct ocfs2_replay_map *replay_map;
 	struct task_struct *recovery_thread_task;
 	int disable_recovery;
fs/ocfs2/super.c +0 -1
index f02c0ef..478715b 100644
@@ -220,7 +220,6 @@ static const match_table_t tokens = {
 static int ocfs2_osb_dump(struct ocfs2_super *osb, char *buf, int len)
 {
 	struct ocfs2_cluster_connection *cconn = osb->cconn;
-	struct ocfs2_recovery_map *rm = osb->recovery_map;
 	struct ocfs2_orphan_scan *os = &osb->osb_orphan_scan;
 	int i, out = 0;
 
@@ -267,21 +266,6 @@ static int ocfs2_osb_dump(struct ocfs2_super

*osb, char *buf, int len)
osb->dc_work_sequence);
spin_unlock(&osb->dc_task_lock);

- spin_lock(&osb->osb_lock);
- out += snprintf(buf + out, len - out, "%10s => Pid: %d Nodes:",
- "Recovery",
- (osb->recovery_thread_task ?
- task_pid_nr(osb->recovery_thread_task) : -1));
- if (rm->rm_used == 0)
- out += snprintf(buf + out, len - out, " None\n");
- else {
- for (i = 0; i < rm->rm_used; i++)
- out += snprintf(buf + out, len - out, " %d",
- rm->rm_entries[i]);
- out += snprintf(buf + out, len - out, "\n");
- }
- spin_unlock(&osb->osb_lock);
-
out += snprintf(buf + out, len - out,
"%10s => Pid: %d Interval: %lu Needs: %d\n", "Commit", (osb->commit_task ? task_pid_nr(osb->commit_task) : -1),

Signature · 5 lines
1.7.1


-- 
Goldwyn
#1 lore raw

PATCH 1/2 Introduce ocfs2_recover_node

Quoted 4 lines · Goldwyn Rodrigues

Review comments by Wengang incorporated.

ocfs2_recover_node is a new data structure to include all information required to recover one node. The structure is maintainer as a list in the super block.

ocfs2_recover_node suggests action. How about calling it ocfs2_recovery_list.

Quoted 63 lines

recovery_map is no longer required with ocfs2_recover_node.

---
  fs/ocfs2/journal.c |  171 ++++++++++++++++++++++++----------------------------
  fs/ocfs2/journal.h |    9 ++-
  fs/ocfs2/ocfs2.h   |    3 +-
  fs/ocfs2/super.c   |   16 -----
  4 files changed, 88 insertions(+), 111 deletions(-)
fs/ocfs2/journal.c +4 -18
index faa2303..277b810 100644
@@ -58,8 +58,7 @@ DEFINE_SPINLOCK(trans_inc_lock);
  #define ORPHAN_SCAN_SCHEDULE_TIMEOUT 300000
 
  static int ocfs2_force_read_journal(struct inode *inode);
-static int ocfs2_recover_node(struct ocfs2_super *osb,
-			      int node_num, int slot_num);
+static int ocfs2_recover_one_node(struct ocfs2_recover_node *rn);
  static int __ocfs2_recovery_thread(void *arg);
  static int ocfs2_commit_cache(struct ocfs2_super *osb);
  static int __ocfs2_wait_on_mount(struct ocfs2_super *osb, int quota);
@@ -179,24 +178,11 @@ void ocfs2_free_replay_slots(struct ocfs2_super *osb)
 
  int ocfs2_recovery_init(struct ocfs2_super *osb)
  {
-	struct ocfs2_recovery_map *rm;
-
  	mutex_init(&osb->recovery_lock);
  	osb->disable_recovery = 0;
  	osb->recovery_thread_task = NULL;
  	init_waitqueue_head(&osb->recovery_event);
-
-	rm = kzalloc(sizeof(struct ocfs2_recovery_map) +
-		     osb->max_slots * sizeof(unsigned int),
-		     GFP_KERNEL);
-	if (!rm) {
-		mlog_errno(-ENOMEM);
-		return -ENOMEM;
-	}
-
-	rm->rm_entries = (unsigned int *)((char *)rm +
-					  sizeof(struct ocfs2_recovery_map));
-	osb->recovery_map = rm;
+	INIT_LIST_HEAD(&osb->s_recovery_nodes);
 
  	return 0;
  }
@@ -212,8 +198,8 @@ static int ocfs2_recovery_thread_running(struct
 ocfs2_super *osb)
 
  void ocfs2_recovery_exit(struct ocfs2_super *osb)
  {
-	struct ocfs2_recovery_map *rm;
-
+	struct list_head *iter, *next;
+	struct ocfs2_recover_node *rn;
  	/* disable any new recovery threads and wait for any currently
  	 * running ones to exit. Do this before setting the vol_state. */
  	mutex_lock(&osb->recovery_lock);

fs/ocfs2/journal.c void ocfs2_recovery_exit(struct ocfs2_super *osb) · line 234

Quoted 9 lines
fs/ocfs2/journal.c +0 -6
@@ -226,75 +212,67 @@ void ocfs2_recovery_exit(struct ocfs2_super *osb)
  	 * complete. */
  	flush_workqueue(ocfs2_wq);
 
-	/*
-	 * Now that recovery is shut down, and the osb is about to be
-	 * freed,  the osb_lock is not taken here.
-	 */
-	rm = osb->recovery_map;
-	/* XXX: Should we bug if there are dirty entries? */

This comment is still relevant.

fs/ocfs2/journal.c void ocfs2_recovery_exit(struct ocfs2_super *osb) · line 222

Quoted 10 lines
fs/ocfs2/journal.c +8 -2
-
-	kfree(rm);
+	spin_lock(&osb->osb_lock);
+	list_for_each_safe(iter, next,&osb->s_recovery_nodes) {
+			rn = list_entry(iter, struct ocfs2_recover_node,
+					rn_list);
+			list_del(&rn->rn_list);
+			kfree(rn);
+	}
+	spin_unlock(&osb->osb_lock);

extra indentation

Quoted 34 lines
fs/ocfs2/journal.c +7 -13
  }
 
-static int __ocfs2_recovery_map_test(struct ocfs2_super *osb,
+static int __ocfs2_recovery_node_test(struct ocfs2_super *osb,
  				     unsigned int node_num)
  {
-	int i;
-	struct ocfs2_recovery_map *rm = osb->recovery_map;
+	struct ocfs2_recover_node *rn;
+	struct list_head *iter;
 
  	assert_spin_locked(&osb->osb_lock);
-
-	for (i = 0; i<  rm->rm_used; i++) {
-		if (rm->rm_entries[i] == node_num)
+	list_for_each(iter,&osb->s_recovery_nodes) {
+		rn = list_entry(iter, struct ocfs2_recover_node, rn_list);
+		if (rn->rn_node_num == node_num)
  			return 1;
  	}
-
  	return 0;
  }
 
  /* Behaves like test-and-set.  Returns the previous value */
-static int ocfs2_recovery_map_set(struct ocfs2_super *osb,
+static int ocfs2_recovery_node_set(struct ocfs2_super *osb,
  				  unsigned int node_num)
  {
-	struct ocfs2_recovery_map *rm = osb->recovery_map;
-
-	spin_lock(&osb->osb_lock);
-	if (__ocfs2_recovery_map_test(osb, node_num)) {
-		spin_unlock(&osb->osb_lock);

fs/ocfs2/journal.c void ocfs2_recovery_exit(struct ocfs2_super *osb) · line 249

Quoted 10 lines
fs/ocfs2/journal.c +5 -4
-		return 1;
-	}
+	struct list_head *iter;
+	struct ocfs2_recover_node *rn = NULL, *trn;
 
-	/* XXX: Can this be exploited? Not from o2dlm... */
-	BUG_ON(rm->rm_used>= osb->max_slots);
+	trn = kzalloc(sizeof(struct ocfs2_recover_node), GFP_NOFS);
+	if (!trn)
+		return -ENOMEM;

Is this enomem case handled properly up the stack? This is new. Previously we allocated memory upfront. Wondering why we cannot so the same here.

fs/ocfs2/journal.c void ocfs2_recovery_exit(struct ocfs2_super *osb) · line 259

Quoted 12 lines
fs/ocfs2/journal.c +10 -2
-	rm->rm_entries[rm->rm_used] = node_num;
-	rm->rm_used++;
+	spin_lock(&osb->osb_lock);
+	list_for_each(iter,&osb->s_recovery_nodes) {
+			rn = list_entry(iter, struct ocfs2_recover_node,
+					rn_list);
+			if (rn->rn_node_num == node_num) {
+				spin_unlock(&osb->osb_lock);
+				kfree(trn);
+				return 1;
+			}
+	}

extra indentation

Quoted 23 lines
fs/ocfs2/journal.c +5 -11
+	trn->rn_node_num = node_num;
+	trn->rn_osb = osb;
+	list_add_tail(&trn->rn_list,&osb->s_recovery_nodes);
  	spin_unlock(&osb->osb_lock);
 
  	return 0;
  }
 
-static void ocfs2_recovery_map_clear(struct ocfs2_super *osb,
-				     unsigned int node_num)
+static void ocfs2_recovery_node_clear(struct ocfs2_super *osb,
+			struct ocfs2_recover_node *rn)
  {
-	int i;
-	struct ocfs2_recovery_map *rm = osb->recovery_map;
-
  	spin_lock(&osb->osb_lock);
-
-	for (i = 0; i<  rm->rm_used; i++) {
-		if (rm->rm_entries[i] == node_num)
-			break;
-	}
-

fs/ocfs2/journal.c void ocfs2_recovery_exit(struct ocfs2_super *osb) · line 274

Quoted 10 lines
fs/ocfs2/journal.c +2 -7
-	if (i<  rm->rm_used) {
-		/* XXX: be careful with the pointer math */
-		memmove(&(rm->rm_entries[i]),&(rm->rm_entries[i + 1]),
-			(rm->rm_used - i - 1) * sizeof(unsigned int));
-		rm->rm_used--;
-	}
-
+	list_del(&rn->rn_list);
  	spin_unlock(&osb->osb_lock);
+	kfree(rn);

Aah...we did memmove. That won't work here. So list manipulation it is.

Quoted 74 lines
fs/ocfs2/journal.c +15 -13
  }
 
  static int ocfs2_commit_cache(struct ocfs2_super *osb)
@@ -1092,10 +1070,9 @@ bail:
  static int ocfs2_recovery_completed(struct ocfs2_super *osb)
  {
  	int empty;
-	struct ocfs2_recovery_map *rm = osb->recovery_map;
 
  	spin_lock(&osb->osb_lock);
-	empty = (rm->rm_used == 0);
+	empty = list_empty(&osb->s_recovery_nodes);
  	spin_unlock(&osb->osb_lock);
 
  	return empty;
@@ -1332,12 +1309,13 @@ void ocfs2_complete_quota_recovery(struct
 ocfs2_super *osb)
 
  static int __ocfs2_recovery_thread(void *arg)
  {
-	int status, node_num, slot_num;
+	int status;
  	struct ocfs2_super *osb = arg;
-	struct ocfs2_recovery_map *rm = osb->recovery_map;
  	int *rm_quota = NULL;
  	int rm_quota_used = 0, i;
  	struct ocfs2_quota_recovery *qrec;
+	struct list_head *iter;
+	struct ocfs2_recover_node *rn = NULL;
 
  	mlog_entry_void();
@@ -1367,20 +1345,21 @@ restart:
  					NULL, NULL);
 
  	spin_lock(&osb->osb_lock);
-	while (rm->rm_used) {
+	list_for_each(iter,&osb->s_recovery_nodes) {
+		rn = list_entry(iter, struct ocfs2_recover_node, rn_list);
  		/* It's always safe to remove entry zero, as we won't
  		 * clear it until ocfs2_recover_node() has succeeded. */
-		node_num = rm->rm_entries[0];
  		spin_unlock(&osb->osb_lock);
-		mlog(0, "checking node %d\n", node_num);
-		slot_num = ocfs2_node_num_to_slot(osb, node_num);
-		if (slot_num == -ENOENT) {
+		mlog(0, "checking node %d\n", rn->rn_node_num);
+		rn->rn_slot_num = ocfs2_node_num_to_slot(osb, rn->rn_node_num);
+		if (rn->rn_slot_num == -ENOENT) {
  			status = 0;
  			mlog(0, "no slot for this node, so no recovery"
  			     "required.\n");
  			goto skip_recovery;
  		}
-		mlog(0, "node %d was using slot %d\n", node_num, slot_num);
+		mlog(0, "node %d was using slot %d\n",
+				rn->rn_node_num, rn->rn_slot_num);
 
  		/* It is a bit subtle with quota recovery. We cannot do it
  		 * immediately because we have to obtain cluster locks from
@@ -1388,18 +1367,19 @@ restart:
  		 * then quota usage would be out of sync until some node takes
  		 * the slot. So we remember which nodes need quota recovery
  		 * and when everything else is done, we recover quotas. */
-		for (i = 0; i<  rm_quota_used&&  rm_quota[i] != slot_num; i++);
+		for (i = 0; i<  rm_quota_used&&  rm_quota[i] != rn->rn_slot_num;
+					i++);
  		if (i == rm_quota_used)
-			rm_quota[rm_quota_used++] = slot_num;
+			rm_quota[rm_quota_used++] = rn->rn_slot_num;
 
-		status = ocfs2_recover_node(osb, node_num, slot_num);
+		status = ocfs2_recover_one_node(rn);
  skip_recovery:

fs/ocfs2/journal.c restart: · line 1384

Quoted 10 lines
fs/ocfs2/journal.c +2 -2
  		if (!status) {
-			ocfs2_recovery_map_clear(osb, node_num);
+			ocfs2_recovery_node_clear(osb, rn);
  		} else {
  			mlog(ML_ERROR,
  			     "Error %d recovering node %d on device (%u,%u)!\n",
-			     status, node_num,
+			     status, rn->rn_node_num,
  			     MAJOR(osb->sb->s_dev), MINOR(osb->sb->s_dev));
  			mlog(ML_ERROR, "Volume requires unmount.\n");

Hmm... this mlog could do with some clean up. But that can be a separate change.

Quoted 114 lines
fs/ocfs2/journal.c +6 -1
  		}
@@ -1461,6 +1441,7 @@ bail:
 
  void ocfs2_recovery_thread(struct ocfs2_super *osb, int node_num)
  {
+	int ret = 0;
  	mlog_entry("(node_num=%d, osb->node_num = %d)\n",
  		   node_num, osb->node_num);
 
@@ -1470,7 +1451,11 @@ void ocfs2_recovery_thread(struct ocfs2_super
 *osb, int node_num)
 
  	/* People waiting on recovery will wait on
  	 * the recovery map to empty. */
-	if (ocfs2_recovery_map_set(osb, node_num))
+	ret = ocfs2_recovery_node_set(osb, node_num);
+	if (ret == -ENOMEM) {
+		mlog_errno(ret);
+		goto out;
+	} else if (ret)
  		mlog(0, "node %d already in recovery map.\n", node_num);
 
  	mlog(0, "starting recovery thread...\n");
fs/ocfs2/journal.c +17 -13
@@ -1681,26 +1666,27 @@ done:
   * second part of a nodes recovery process (local alloc recovery) is
   * far less concerning.
   */
-static int ocfs2_recover_node(struct ocfs2_super *osb,
-			      int node_num, int slot_num)
+static int ocfs2_recover_one_node(struct ocfs2_recover_node *rn)
  {
  	int status = 0;
  	struct ocfs2_dinode *la_copy = NULL;
  	struct ocfs2_dinode *tl_copy = NULL;
+	struct ocfs2_super *osb = rn->rn_osb;
 
  	mlog_entry("(node_num=%d, slot_num=%d, osb->node_num = %d)\n",
-		   node_num, slot_num, osb->node_num);
+		   rn->rn_node_num, rn->rn_slot_num, osb->node_num);
 
  	/* Should not ever be called to recover ourselves -- in that
  	 * case we should've called ocfs2_journal_load instead. */
-	BUG_ON(osb->node_num == node_num);
+	BUG_ON(osb->node_num == rn->rn_node_num);
 
-	status = ocfs2_replay_journal(osb, node_num, slot_num);
+	status = ocfs2_replay_journal(osb, rn->rn_node_num,
+			rn->rn_slot_num);
  	if (status<  0) {
  		if (status == -EBUSY) {
  			mlog(0, "Skipping recovery for slot %u (node %u) "
-			     "as another node has recovered it\n", slot_num,
-			     node_num);
+			     "as another node has recovered it\n",
+			     rn->rn_slot_num, rn->rn_node_num);
  			status = 0;
  			goto done;
  		}
@@ -1709,7 +1695,8 @@ static int ocfs2_recover_node(struct ocfs2_super *osb,
  	}
 
  	/* Stamp a clean local alloc file AFTER recovering the journal... */
-	status = ocfs2_begin_local_alloc_recovery(osb, slot_num,&la_copy);
+	status = ocfs2_begin_local_alloc_recovery(osb, rn->rn_slot_num,
+			&la_copy);
  	if (status<  0) {
  		mlog_errno(status);
  		goto done;
@@ -1718,22 +1705,24 @@ static int ocfs2_recover_node(struct ocfs2_super *osb,
  	/* An error from begin_truncate_log_recovery is not
  	 * serious enough to warrant halting the rest of
  	 * recovery. */
-	status = ocfs2_begin_truncate_log_recovery(osb, slot_num,&tl_copy);
+	status = ocfs2_begin_truncate_log_recovery(osb,
+			rn->rn_slot_num,&tl_copy);
  	if (status<  0)
  		mlog_errno(status);
 
  	/* Likewise, this would be a strange but ultimately not so
  	 * harmful place to get an error... */
-	status = ocfs2_clear_slot(osb, slot_num);
+	status = ocfs2_clear_slot(osb, rn->rn_slot_num);
  	if (status<  0)
  		mlog_errno(status);
 
  	/* This will kfree the memory pointed to by la_copy and tl_copy */
-	ocfs2_queue_recovery_completion(osb->journal, slot_num, la_copy,
-					tl_copy, NULL);
+	ocfs2_queue_recovery_completion(osb->journal, rn->rn_slot_num,
+				la_copy, tl_copy, NULL);
 
  	status = 0;
  done:
+	rn->rn_status = status;
 
  	mlog_exit(status);
  	return status;
@@ -1822,7 +1811,7 @@ int ocfs2_mark_dead_nodes(struct ocfs2_super *osb)
  			continue;
  		}
 
-		if (__ocfs2_recovery_map_test(osb, node_num)) {
+		if (__ocfs2_recovery_node_test(osb, node_num)) {
  			spin_unlock(&osb->osb_lock);
  			continue;
  		}
fs/ocfs2/journal.h +0 -0
index 43e56b9..0325d81 100644
@@ -43,9 +43,12 @@ struct ocfs2_dinode;
   * It is protected by the recovery_lock.
   */
 

fs/ocfs2/journal.h struct ocfs2_dinode; · line 52

Quoted 10 lines
fs/ocfs2/journal.h +6 -3
-struct ocfs2_recovery_map {
-	unsigned int rm_used;
-	unsigned int *rm_entries;
+struct ocfs2_recover_node {
+	struct ocfs2_super *rn_osb;
+	int rn_node_num;
+	int rn_slot_num;
+	int rn_status;
+	struct list_head rn_list;
  };

See comment above.

fs/ocfs2/ocfs2.h struct ocfs2_super · line 340

Quoted 11 lines
fs/ocfs2/ocfs2.h +1 -1
index 1efea36..f70c25a 100644
@@ -337,7 +337,8 @@ struct ocfs2_super
 
  	atomic_t vol_state;
  	struct mutex recovery_lock;
-	struct ocfs2_recovery_map *recovery_map;
+	struct list_head s_recovery_nodes;

s_recovery_list will fit better.

Quoted 39 lines
fs/ocfs2/ocfs2.h +1 -0
+
  	struct ocfs2_replay_map *replay_map;
  	struct task_struct *recovery_thread_task;
  	int disable_recovery;
fs/ocfs2/super.c +0 -16
index f02c0ef..478715b 100644
@@ -220,7 +220,6 @@ static const match_table_t tokens = {
  static int ocfs2_osb_dump(struct ocfs2_super *osb, char *buf, int len)
  {
  	struct ocfs2_cluster_connection *cconn = osb->cconn;
-	struct ocfs2_recovery_map *rm = osb->recovery_map;
  	struct ocfs2_orphan_scan *os =&osb->osb_orphan_scan;
  	int i, out = 0;
 
@@ -267,21 +266,6 @@ static int ocfs2_osb_dump(struct ocfs2_super
 *osb, char *buf, int len)
  			osb->dc_work_sequence);
  	spin_unlock(&osb->dc_task_lock);
 
-	spin_lock(&osb->osb_lock);
-	out += snprintf(buf + out, len - out, "%10s =>  Pid: %d  Nodes:",
-			"Recovery",
-			(osb->recovery_thread_task ?
-			 task_pid_nr(osb->recovery_thread_task) : -1));
-	if (rm->rm_used == 0)
-		out += snprintf(buf + out, len - out, " None\n");
-	else {
-		for (i = 0; i<  rm->rm_used; i++)
-			out += snprintf(buf + out, len - out, " %d",
-					rm->rm_entries[i]);
-		out += snprintf(buf + out, len - out, "\n");
-	}
-	spin_unlock(&osb->osb_lock);
-
  	out += snprintf(buf + out, len - out,
  			"%10s =>  Pid: %d  Interval: %lu  Needs: %d\n", "Commit",

(osb->commit_task ? task_pid_nr(osb->commit_task) : -1),

#2 lore raw

PATCH 1/2 Introduce ocfs2_recover_node

In reply to Goldwyn Rodrigues

Review void ocfs2_recovery_thread(struct ocfs2_super · line 1460

Quoted 12 lines · Goldwyn Rodrigues
(quoted diff) +5 -1
@@ -1470,7 +1451,11 @@ void ocfs2_recovery_thread(struct ocfs2_super
 *osb, int node_num)
 
 	/* People waiting on recovery will wait on
 	 * the recovery map to empty. */
-	if (ocfs2_recovery_map_set(osb, node_num))
+	ret = ocfs2_recovery_node_set(osb, node_num);
+	if (ret == -ENOMEM) {
+		mlog_errno(ret);
+		goto out;
+	} else if (ret)
 		mlog(0, "node %d already in recovery map.\n", node_num);

This is a broken change. If we get -ENOMEM, we won't block
other processes. We can't have that happen. There are two possible solutions. First, like Sunil said, we can preallocate max_slots recovery entries. Seems pretty sane. The other solution would be to set an in-recovery flag that others can check, so even when the recovery list is empty because of a failed allocation, other processes still block. I prefer the preallocation because it doesn't fail recovery.

Joel

Signature · 9 lines
"In a crisis, don't hide behind anything or anybody. They're going
 to find you anyway."
	- Paul "Bear" Bryant

Joel Becker
Senior Development Manager
Oracle
E-mail: joel.becker at oracle.com
Phone: (650) 506-8127
#3 lore raw

PATCH 1/2 Introduce ocfs2_recover_node

Quoted 4 lines · Joel Becker
Quoted 3 lines · Goldwyn Rodrigues
(quoted diff) +0 -0
@@ -1470,7 +1451,11 @@ void ocfs2_recovery_thread(struct ocfs2_super
 *osb, int node_num)
 
Quoted 16 lines

Review void ocfs2_recovery_thread(struct ocfs2_super · line 1460

Quoted 9 lines
(quoted diff) +5 -1
  	/* People waiting on recovery will wait on
  	 * the recovery map to empty. */
-	if (ocfs2_recovery_map_set(osb, node_num))
+	ret = ocfs2_recovery_node_set(osb, node_num);
+	if (ret == -ENOMEM) {
+		mlog_errno(ret);
+		goto out;
+	} else if (ret)
  		mlog(0, "node %d already in recovery map.\n", node_num);

This is a broken change. If we get -ENOMEM, we won't block
other processes. We can't have that happen. There are two possible solutions. First, like Sunil said, we can preallocate max_slots recovery entries. Seems pretty sane. The other solution would be to set an in-recovery flag that others can check, so even when the recovery list is empty because of a failed allocation, other processes still block. I prefer the preallocation because it doesn't fail recovery.

You could stick with the list but preallocate max_slots items during init. Attach all those to the s_init_reco_list. During set, move a item to s_active_reco_list. Clear moves it back.

#4 lore raw

PATCH 1/2 Introduce ocfs2_recover_node

Quoted 22 lines · Sunil Mushran
Quoted 4 lines · Joel Becker
Quoted 3 lines · Goldwyn Rodrigues
(quoted diff) +0 -0
@@ -1470,7 +1451,11 @@ void ocfs2_recovery_thread(struct ocfs2_super
 *osb, int node_num)
 
Quoted 16 lines

Review void ocfs2_recovery_thread(struct ocfs2_super · line 1460

Quoted 9 lines
(quoted diff) +5 -1
  	/* People waiting on recovery will wait on
  	 * the recovery map to empty. */
-	if (ocfs2_recovery_map_set(osb, node_num))
+	ret = ocfs2_recovery_node_set(osb, node_num);
+	if (ret == -ENOMEM) {
+		mlog_errno(ret);
+		goto out;
+	} else if (ret)
  		mlog(0, "node %d already in recovery map.\n", node_num);

This is a broken change. If we get -ENOMEM, we won't block
other processes. We can't have that happen. There are two possible solutions. First, like Sunil said, we can preallocate max_slots recovery entries. Seems pretty sane. The other solution would be to set an in-recovery flag that others can check, so even when the recovery list is empty because of a failed allocation, other processes still block. I prefer the preallocation because it doesn't fail recovery.

Quoted 3 lines

You could stick with the list but preallocate max_slots items during init. Attach all those to the s_init_reco_list. During set, move a item to s_active_reco_list. Clear moves it back.

Exactly.

Joel

Signature · 11 lines
"You must remember this:
 A kiss is just a kiss,
 A sigh is just a sigh.
 The fundamental rules apply
 As time goes by."

Joel Becker
Senior Development Manager
Oracle
E-mail: joel.becker at oracle.com
Phone: (650) 506-8127