[Ocfs2-devel] Backports to OCFS2 1.4

[Ocfs2-devel] Backports to OCFS2 1.4

[Sunil Mushran]

The following patches are in this series. All except the first one
have already been merged into mainline. The headers indicate the mainline
commit# and other details of the original patch.

[PATCH 01/18] ocfs2: Add helper task_pid_nr
[PATCH 02/18] ocfs2: Spelling fixes
[PATCH 03/18] ocfs2: Negotiate locking protocol versions.
[PATCH 04/18] ocfs2: Fix writeout in ocfs2_data_convert_worker()
[PATCH 05/18] ocfs2: possible cleanups
[PATCH 06/18] ocfs2: make ocfs2_downconvert_thread() static
[PATCH 07/18] ocfs2: make dlm_do_assert_master() static
[PATCH 08/18] ocfs2: Correct use of ! and & in aops.c
[PATCH 09/18] ocfs2/dlm: fix printk warning
[PATCH 10/18] ocfs2: Use dlm_print_one_lock_resource for lock resource print
[PATCH 11/18] ocfs2: Fix endian bug in o2dlm protocol negotiation.
[PATCH 12/18] ocfs2: Fix an endian bug in online resize
[PATCH 13/18] ocfs2/dlm: Add missing dlm_lock_put()s
[PATCH 14/18] ocfs2/dlm: Add missing dlm_lockres_put()s in migration path
[PATCH 15/18] ocfs2/dlm: Add missing dlm_lockres_put()s
[PATCH 16/18] ocfs2/dlm: Print message showing the recovery master
[PATCH 17/18] ocfs2/dlm: dlm_thread should not sleep while holding the dlm_spinlock
[PATCH 18/18] ocfs2: Fix NULL pointer dereferences in o2net

[Ocfs2-devel] [PATCH 07/18] ocfs2: make dlm_do_assert_master() static

[Sunil Mushran]

Mainline commit 05488bbebe3deedbc5d58a1832f563ff96bc2ef6
Author: Adrian Bunk <bunk@kernel.org>
Date: Sun, 17 Feb 2008 10:20:41 +0200

This patch makes the needlessly global dlm_do_assert_master() static.

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
---
 fs/ocfs2/dlm/dlmmaster.c |    6 +++---
 1 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/fs/ocfs2/dlm/dlmmaster.c b/fs/ocfs2/dlm/dlmmaster.c
index 8faa2e4..da627c8 100644
--- a/fs/ocfs2/dlm/dlmmaster.c
+++ b/fs/ocfs2/dlm/dlmmaster.c
@@ -1695,9 +1695,9 @@ send_response:
  * can periodically run all locks owned by this node
  * and re-assert across the cluster...
  */
-int dlm_do_assert_master(struct dlm_ctxt *dlm,
-			 struct dlm_lock_resource *res,
-			 void *nodemap, u32 flags)
+static int dlm_do_assert_master(struct dlm_ctxt *dlm,
+				struct dlm_lock_resource *res,
+				void *nodemap, u32 flags)
 {
 	struct dlm_assert_master assert;
 	int to, tmpret;
-- 
1.5.3.4

[Ocfs2-devel] [PATCH 08/18] ocfs2: Correct use of ! and & in aops.c

[Sunil Mushran]

Mainline commit 86c838b03daf35e2af6555842d04fe09a89f8d93
Author: Julia Lawall <julia@diku.dk>
Date: Tue, 26 Feb 2008 21:45:56 +0100

In mainline commit e6bafba5b4765a5a252f1b8d31cbf6d2459da337, a bug was fixed that
involved converting !x & y to !(x & y).  The code below shows the same
pattern, and thus should perhaps be fixed in the same way.

This is not tested and clearly changes the semantics, so it is only
something to consider.

Signed-off-by: Julia Lawall <julia@diku.dk>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
---
 fs/ocfs2/aops.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/fs/ocfs2/aops.c b/fs/ocfs2/aops.c
index 7db3a5d..c06b7b1 100644
--- a/fs/ocfs2/aops.c
+++ b/fs/ocfs2/aops.c
@@ -257,7 +257,7 @@ static int ocfs2_readpage_inline(struct inode *inode, struct page *page)
 	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
 
 	BUG_ON(!PageLocked(page));
-	BUG_ON(!OCFS2_I(inode)->ip_dyn_features & OCFS2_INLINE_DATA_FL);
+	BUG_ON(!(OCFS2_I(inode)->ip_dyn_features & OCFS2_INLINE_DATA_FL));
 
 	ret = ocfs2_read_block(osb, OCFS2_I(inode)->ip_blkno, &di_bh,
 			       OCFS2_BH_CACHED, inode);
-- 
1.5.3.4

[Ocfs2-devel] [PATCH 01/18] ocfs2: Add helper task_pid_nr

[Sunil Mushran]

Commit 7af5729474b5b8ad385adadab78d6e723e7655a3 in mainline introduced
helper task_pid_nr(). This patch allows one to build ocfs2 with kernels
having/not having that change.

Signed-off-by: Sunil Mushran <sunil.mushran@oracle.com>
---
 Makefile                          |    3 ++-
 configure.in                      |    5 +++++
 kapi-compat/include/task_pid_nr.h |   12 ++++++++++++
 3 files changed, 19 insertions(+), 1 deletions(-)
 create mode 100644 kapi-compat/include/task_pid_nr.h

diff --git a/Makefile b/Makefile
index 8e17890..0768cd0 100644
--- a/Makefile
+++ b/Makefile
@@ -32,7 +32,8 @@ KAPI_COMPAT_FILES = \
 	kapi-compat/include/blkcnt_t.h \
 	kapi-compat/include/read_mapping_page.h \
 	kapi-compat/include/aiovec.h \
-	kapi-compat/include/mandatory_lock.h
+	kapi-compat/include/mandatory_lock.h \
+	kapi-compat/include/task_pid_nr.h
 
 PATCH_FILES =
 
diff --git a/configure.in b/configure.in
index f3544b0..df2dcc6 100644
--- a/configure.in
+++ b/configure.in
@@ -420,6 +420,11 @@ OCFS2_CHECK_KERNEL([fops->sendfile() in fs.h], fs.h,
   HAS_FOPS_SENDFILE=yes, , [^.*ssize_t (\*sendfile)])
 AC_SUBST(HAS_FOPS_SENDFILE)
 
+task_pid_nr_compat_header=""
+OCFS2_CHECK_KERNEL([task_pid_nr in sched.h], sched.h,
+  , task_pid_nr_compat_header="task_pid_nr.h", [^static inline pid_t task_pid_nr(struct])
+KAPI_COMPAT_HEADERS="$KAPI_COMPAT_HEADERS $task_pid_nr_compat_header"
+
 # using -include has two advantages:
 #  the source doesn't need to know to include compat headers
 #  the compat header file names don't go through the search path
diff --git a/kapi-compat/include/task_pid_nr.h b/kapi-compat/include/task_pid_nr.h
new file mode 100644
index 0000000..bd69a75
--- /dev/null
+++ b/kapi-compat/include/task_pid_nr.h
@@ -0,0 +1,12 @@
+#ifndef KAPI_TASK_PID_NR_H
+#define KAPI_TASK_PID_NR_H
+
+#include <linux/types.h>
+#include <linux/sched.h>
+
+static inline pid_t task_pid_nr(struct task_struct *tsk)
+{
+        return tsk->pid;
+}
+
+#endif
-- 
1.5.3.4

[Ocfs2-devel] [PATCH 05/18] ocfs2: possible cleanups

[Sunil Mushran]

Mainline commit 006000566d4e95b8d1924addfb41094acf0d5ec2
Author: Adrian Bunk <bunk@kernel.org>
Date: Tue, 29 Jan 2008 00:11:41 +0200

This patch contains the following cleanups that are now possible:
- make the following needlessly global functions static:
  - dlmglue.c:ocfs2_process_blocked_lock()
  - heartbeat.c:ocfs2_node_map_init()
- #if 0 the following unused global function plus support functions:
  - heartbeat.c:ocfs2_node_map_is_only()

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
---
 fs/ocfs2/dlmglue.c   |    4 ++--
 fs/ocfs2/dlmglue.h   |    2 --
 fs/ocfs2/heartbeat.c |   26 +++++++++++++-------------
 fs/ocfs2/heartbeat.h |    5 -----
 4 files changed, 15 insertions(+), 22 deletions(-)

diff --git a/fs/ocfs2/dlmglue.c b/fs/ocfs2/dlmglue.c
index c8bcba4..f240079 100644
--- a/fs/ocfs2/dlmglue.c
+++ b/fs/ocfs2/dlmglue.c
@@ -3227,8 +3227,8 @@ static int ocfs2_dentry_convert_worker(struct ocfs2_lock_res *lockres,
 	return UNBLOCK_CONTINUE_POST;
 }
 
-void ocfs2_process_blocked_lock(struct ocfs2_super *osb,
-				struct ocfs2_lock_res *lockres)
+static void ocfs2_process_blocked_lock(struct ocfs2_super *osb,
+				       struct ocfs2_lock_res *lockres)
 {
 	int status;
 	struct ocfs2_unblock_ctl ctl = {0, 0,};
diff --git a/fs/ocfs2/dlmglue.h b/fs/ocfs2/dlmglue.h
index 1d5b069..e3cf902 100644
--- a/fs/ocfs2/dlmglue.h
+++ b/fs/ocfs2/dlmglue.h
@@ -109,8 +109,6 @@ void ocfs2_simple_drop_lockres(struct ocfs2_super *osb,
 			       struct ocfs2_lock_res *lockres);
 
 /* for the downconvert thread */
-void ocfs2_process_blocked_lock(struct ocfs2_super *osb,
-				struct ocfs2_lock_res *lockres);
 void ocfs2_wake_downconvert_thread(struct ocfs2_super *osb);
 
 struct ocfs2_dlm_debug *ocfs2_new_dlm_debug(void);
diff --git a/fs/ocfs2/heartbeat.c b/fs/ocfs2/heartbeat.c
index c0efd94..0758daf 100644
--- a/fs/ocfs2/heartbeat.c
+++ b/fs/ocfs2/heartbeat.c
@@ -49,10 +49,15 @@ static inline void __ocfs2_node_map_set_bit(struct ocfs2_node_map *map,
 static inline void __ocfs2_node_map_clear_bit(struct ocfs2_node_map *map,
 					      int bit);
 static inline int __ocfs2_node_map_is_empty(struct ocfs2_node_map *map);
-static void __ocfs2_node_map_dup(struct ocfs2_node_map *target,
-				 struct ocfs2_node_map *from);
-static void __ocfs2_node_map_set(struct ocfs2_node_map *target,
-				 struct ocfs2_node_map *from);
+
+/* special case -1 for now
+ * TODO: should *really* make sure the calling func never passes -1!!  */
+static void ocfs2_node_map_init(struct ocfs2_node_map *map)
+{
+	map->num_nodes = OCFS2_NODE_MAP_MAX_NODES;
+	memset(map->map, 0, BITS_TO_LONGS(OCFS2_NODE_MAP_MAX_NODES) *
+	       sizeof(unsigned long));
+}
 
 void ocfs2_init_node_maps(struct ocfs2_super *osb)
 {
@@ -136,15 +141,6 @@ void ocfs2_stop_heartbeat(struct ocfs2_super *osb)
 		mlog_errno(ret);
 }
 
-/* special case -1 for now
- * TODO: should *really* make sure the calling func never passes -1!!  */
-void ocfs2_node_map_init(struct ocfs2_node_map *map)
-{
-	map->num_nodes = OCFS2_NODE_MAP_MAX_NODES;
-	memset(map->map, 0, BITS_TO_LONGS(OCFS2_NODE_MAP_MAX_NODES) *
-	       sizeof(unsigned long));
-}
-
 static inline void __ocfs2_node_map_set_bit(struct ocfs2_node_map *map,
 					    int bit)
 {
@@ -216,6 +212,8 @@ int ocfs2_node_map_is_empty(struct ocfs2_super *osb,
 	return ret;
 }
 
+#if 0
+
 static void __ocfs2_node_map_dup(struct ocfs2_node_map *target,
 				 struct ocfs2_node_map *from)
 {
@@ -254,6 +252,8 @@ static void __ocfs2_node_map_set(struct ocfs2_node_map *target,
 		target->map[i] = from->map[i];
 }
 
+#endif  /*  0  */
+
 /* Returns whether the recovery bit was actually set - it may not be
  * if a node is still marked as needing recovery */
 int ocfs2_recovery_map_set(struct ocfs2_super *osb,
diff --git a/fs/ocfs2/heartbeat.h b/fs/ocfs2/heartbeat.h
index 5685921..eac63ae 100644
--- a/fs/ocfs2/heartbeat.h
+++ b/fs/ocfs2/heartbeat.h
@@ -33,7 +33,6 @@ void ocfs2_stop_heartbeat(struct ocfs2_super *osb);
 
 /* node map functions - used to keep track of mounted and in-recovery
  * nodes. */
-void ocfs2_node_map_init(struct ocfs2_node_map *map);
 int ocfs2_node_map_is_empty(struct ocfs2_super *osb,
 			    struct ocfs2_node_map *map);
 void ocfs2_node_map_set_bit(struct ocfs2_super *osb,
@@ -57,9 +56,5 @@ int ocfs2_recovery_map_set(struct ocfs2_super *osb,
 			   int num);
 void ocfs2_recovery_map_clear(struct ocfs2_super *osb,
 			      int num);
-/* returns 1 if bit is the only bit set in target, 0 otherwise */
-int ocfs2_node_map_is_only(struct ocfs2_super *osb,
-			   struct ocfs2_node_map *target,
-			   int bit);
 
 #endif /* OCFS2_HEARTBEAT_H */
-- 
1.5.3.4

[Ocfs2-devel] [PATCH 04/18] ocfs2: Fix writeout in ocfs2_data_convert_worker()

[Sunil Mushran]

Mainline commit 1044e401af9a309637828aa3cc8f3b6409fcbf4e
Author: Mark Fasheh <mark.fasheh@oracle.com>
Date: Thu, 28 Feb 2008 17:16:03 -0800

Mainline commit f1f540688eae66c274ff1c1133b5d9c687b28f58 "optimized"
ocfs2_data_convert_worker() to "only do work for regular files".
Unfortunately, I left out a '!', which casued it to *skip* regular files.
This was hidden from testing until recently because the default data
journaling mode (data=ordered) doesn't exercise this code.

Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
---
 fs/ocfs2/dlmglue.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/fs/ocfs2/dlmglue.c b/fs/ocfs2/dlmglue.c
index 0ff457c..c8bcba4 100644
--- a/fs/ocfs2/dlmglue.c
+++ b/fs/ocfs2/dlmglue.c
@@ -3050,7 +3050,7 @@ static int ocfs2_data_convert_worker(struct ocfs2_lock_res *lockres,
        	inode = ocfs2_lock_res_inode(lockres);
 	mapping = inode->i_mapping;
 
-	if (S_ISREG(inode->i_mode))
+	if (!S_ISREG(inode->i_mode))
 		goto out;
 
 	/*
-- 
1.5.3.4

[Ocfs2-devel] [PATCH 06/18] ocfs2: make ocfs2_downconvert_thread() static

[Sunil Mushran]

Mainline commit 200bfae37a15e50e0f9aa5683958bdfc3fd55e05
Author: Adrian Bunk <bunk@kernel.org>
Date: Sun, 17 Feb 2008 10:20:38 +0200

This patch makes the needlessly global ocfs2_downconvert_thread()
static.

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
---
 fs/ocfs2/dlmglue.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/fs/ocfs2/dlmglue.c b/fs/ocfs2/dlmglue.c
index f240079..7227af5 100644
--- a/fs/ocfs2/dlmglue.c
+++ b/fs/ocfs2/dlmglue.c
@@ -3364,7 +3364,7 @@ static int ocfs2_downconvert_thread_should_wake(struct ocfs2_super *osb)
 	return should_wake;
 }
 
-int ocfs2_downconvert_thread(void *arg)
+static int ocfs2_downconvert_thread(void *arg)
 {
 	int status = 0;
 	struct ocfs2_super *osb = arg;
-- 
1.5.3.4

[Ocfs2-devel] [PATCH 02/18] ocfs2: Spelling fixes

[Sunil Mushran]

Mainline commit c78bad11fbf1272ea021f56458025dc98486d6f4
Author: Joe Perches <joe@perches.com>
Date: Sun, 3 Feb 2008 17:33:42 +0200

This patch includes the ocfs2 relevant changes from the above
mainline commit.

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Adrian Bunk <bunk@kernel.org>
---
 fs/ocfs2/alloc.c           |    2 +-
 fs/ocfs2/dir.c             |    2 +-
 fs/ocfs2/ocfs1_fs_compat.h |    2 +-
 fs/ocfs2/suballoc.c        |    2 +-
 4 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/fs/ocfs2/alloc.c b/fs/ocfs2/alloc.c
index 834a001..56f17f9 100644
--- a/fs/ocfs2/alloc.c
+++ b/fs/ocfs2/alloc.c
@@ -3338,7 +3338,7 @@ static int ocfs2_insert_path(struct inode *inode,
 	if (insert->ins_split != SPLIT_NONE) {
 		/*
 		 * We could call ocfs2_insert_at_leaf() for some types
-		 * of splits, but it's easier to just let one seperate
+		 * of splits, but it's easier to just let one separate
 		 * function sort it all out.
 		 */
 		ocfs2_split_record(inode, left_path, right_path,
diff --git a/fs/ocfs2/dir.c b/fs/ocfs2/dir.c
index 0e45fb8..a448d1e 100644
--- a/fs/ocfs2/dir.c
+++ b/fs/ocfs2/dir.c
@@ -1221,7 +1221,7 @@ static int ocfs2_expand_inline_dir(struct inode *dir, struct buffer_head *di_bh,
 	down_write(&oi->ip_alloc_sem);
 
 	/*
-	 * Prepare for worst case allocation scenario of two seperate
+	 * Prepare for worst case allocation scenario of two separate
 	 * extents.
 	 */
 	if (alloc == 2)
diff --git a/fs/ocfs2/ocfs1_fs_compat.h b/fs/ocfs2/ocfs1_fs_compat.h
index 0b499bc..dfb313b 100644
--- a/fs/ocfs2/ocfs1_fs_compat.h
+++ b/fs/ocfs2/ocfs1_fs_compat.h
@@ -77,7 +77,7 @@ struct ocfs1_disk_lock
 {
 /*00*/	__u32 curr_master;
 	__u8 file_lock;
-	__u8 compat_pad[3];  /* Not in orignal definition.  Used to
+	__u8 compat_pad[3];  /* Not in original definition.  Used to
 				make the already existing alignment
 				explicit */
 	__u64 last_write_time;
diff --git a/fs/ocfs2/suballoc.c b/fs/ocfs2/suballoc.c
index 7e397e2..72c198a 100644
--- a/fs/ocfs2/suballoc.c
+++ b/fs/ocfs2/suballoc.c
@@ -646,7 +646,7 @@ bail:
  * sync-data inodes."
  *
  * Note: OCFS2 already does this differently for metadata vs data
- * allocations, as those bitmaps are seperate and undo access is never
+ * allocations, as those bitmaps are separate and undo access is never
  * called on a metadata group descriptor.
  */
 static int ocfs2_test_bg_bit_allocatable(struct buffer_head *bg_bh,
-- 
1.5.3.4

[Ocfs2-devel] [PATCH 14/18] ocfs2/dlm: Add missing dlm_lockres_put()s in migration path

[Sunil Mushran]

Mainline commit 52987e2ab456c1a828046494aac53819b1454341
Author: Sunil Mushran <sunil.mushran@oracle.com>
Date: Sat, 1 Mar 2008 14:04:21 -0800

During migration, the recovery master node may be asked to master a lockres
it may not know about. In that case, it would not only have to create a
lockres and add it to the hash, but also remember to to do the _put_
corresponding to the kref_init in dlm_init_lockres(), as soon as the migration
is completed. Yes, we don't wait for the dlm_purge_lockres() to do that
matching put. Note the ref added for it being in the hash protects the lockres
from being freed prematurely.

This patch adds that missing put, as described above, to plug a memleak.

Signed-off-by: Sunil Mushran <sunil.mushran@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
---
 fs/ocfs2/dlm/dlmcommon.h   |    1 +
 fs/ocfs2/dlm/dlmrecovery.c |   40 ++++++++++++++++++++++++++++++++++------
 2 files changed, 35 insertions(+), 6 deletions(-)

diff --git a/fs/ocfs2/dlm/dlmcommon.h b/fs/ocfs2/dlm/dlmcommon.h
index b90ee17..57c0a08 100644
--- a/fs/ocfs2/dlm/dlmcommon.h
+++ b/fs/ocfs2/dlm/dlmcommon.h
@@ -176,6 +176,7 @@ struct dlm_mig_lockres_priv
 {
 	struct dlm_lock_resource *lockres;
 	u8 real_master;
+	u8 extra_ref;
 };
 
 struct dlm_assert_master_priv
diff --git a/fs/ocfs2/dlm/dlmrecovery.c b/fs/ocfs2/dlm/dlmrecovery.c
index 4a0e7aa..7843ec1 100644
--- a/fs/ocfs2/dlm/dlmrecovery.c
+++ b/fs/ocfs2/dlm/dlmrecovery.c
@@ -1327,6 +1327,7 @@ int dlm_mig_lockres_handler(struct o2net_msg *msg, u32 len, void *data,
 		(struct dlm_migratable_lockres *)msg->buf;
 	int ret = 0;
 	u8 real_master;
+	u8 extra_refs = 0;
 	char *buf = NULL;
 	struct dlm_work_item *item = NULL;
 	struct dlm_lock_resource *res = NULL;
@@ -1404,16 +1405,28 @@ int dlm_mig_lockres_handler(struct o2net_msg *msg, u32 len, void *data,
 		__dlm_insert_lockres(dlm, res);
 		spin_unlock(&dlm->spinlock);
 
+		/* Add an extra ref for this lock-less lockres lest the
+		 * dlm_thread purges it before we get the chance to add
+		 * locks to it */
+		dlm_lockres_get(res);
+
+		/* There are three refs that need to be put.
+		 * 1. Taken above.
+		 * 2. kref_init in dlm_new_lockres()->dlm_init_lockres().
+		 * 3. dlm_lookup_lockres()
+		 * The first one is handled at the end of this function. The
+		 * other two are handled in the worker thread after locks have
+		 * been attached. Yes, we don't wait for purge time to match
+		 * kref_init. The lockres will still have atleast one ref
+		 * added because it is in the hash __dlm_insert_lockres() */
+		extra_refs++;
+
 		/* now that the new lockres is inserted,
 		 * make it usable by other processes */
 		spin_lock(&res->spinlock);
 		res->state &= ~DLM_LOCK_RES_IN_PROGRESS;
 		spin_unlock(&res->spinlock);
 		wake_up(&res->wq);
-
-		/* add an extra ref for just-allocated lockres 
-		 * otherwise the lockres will be purged immediately */
-		dlm_lockres_get(res);
 	}
 
 	/* at this point we have allocated everything we need,
@@ -1443,12 +1456,17 @@ int dlm_mig_lockres_handler(struct o2net_msg *msg, u32 len, void *data,
 	dlm_init_work_item(dlm, item, dlm_mig_lockres_worker, buf);
 	item->u.ml.lockres = res; /* already have a ref */
 	item->u.ml.real_master = real_master;
+	item->u.ml.extra_ref = extra_refs;
 	spin_lock(&dlm->work_lock);
 	list_add_tail(&item->list, &dlm->work_list);
 	spin_unlock(&dlm->work_lock);
 	queue_work(dlm->dlm_worker, &dlm->dispatched_work);
 
 leave:
+	/* One extra ref taken needs to be put here */
+	if (extra_refs)
+		dlm_lockres_put(res);
+
 	dlm_put(dlm);
 	if (ret < 0) {
 		if (buf)
@@ -1464,17 +1482,19 @@ leave:
 
 static void dlm_mig_lockres_worker(struct dlm_work_item *item, void *data)
 {
-	struct dlm_ctxt *dlm = data;
+	struct dlm_ctxt *dlm;
 	struct dlm_migratable_lockres *mres;
 	int ret = 0;
 	struct dlm_lock_resource *res;
 	u8 real_master;
+	u8 extra_ref;
 
 	dlm = item->dlm;
 	mres = (struct dlm_migratable_lockres *)data;
 
 	res = item->u.ml.lockres;
 	real_master = item->u.ml.real_master;
+	extra_ref = item->u.ml.extra_ref;
 
 	if (real_master == DLM_LOCK_RES_OWNER_UNKNOWN) {
 		/* this case is super-rare. only occurs if
@@ -1517,6 +1537,12 @@ again:
 	}
 
 leave:
+	/* See comment in dlm_mig_lockres_handler() */
+	if (res) {
+		if (extra_ref)
+			dlm_lockres_put(res);
+		dlm_lockres_put(res);
+	}
 	kfree(data);
 	mlog_exit(ret);
 }
@@ -1644,7 +1670,8 @@ int dlm_master_requery_handler(struct o2net_msg *msg, u32 len, void *data,
 				/* retry!? */
 				BUG();
 			}
-		}
+		} else /* put.. incase we are not the master */
+			dlm_lockres_put(res);
 		spin_unlock(&res->spinlock);
 	}
 	spin_unlock(&dlm->spinlock);
@@ -1921,6 +1948,7 @@ void dlm_move_lockres_to_recovery_list(struct dlm_ctxt *dlm,
 		     "Recovering res %s:%.*s, is already on recovery list!\n",
 		     dlm->name, res->lockname.len, res->lockname.name);
 		list_del_init(&res->recovering);
+		dlm_lockres_put(res);
 	}
 	/* We need to hold a reference while on the recovery list */
 	dlm_lockres_get(res);
-- 
1.5.3.4

[Ocfs2-devel] [PATCH 11/18] ocfs2: Fix endian bug in o2dlm protocol negotiation.

[Sunil Mushran]

Mainline commit 0f71b7b40f55de909e40fa5ab217a5da3439c7d8
Author: Joel Becker <joel.becker@oracle.com>
Date: Tue, 12 Feb 2008 14:56:25 -0800

struct dlm_query_join_packet is made up of four one-byte fields.  They
are effectively in big-endian order already.  However, little-endian
machines swap them before putting the packet on the wire (because
query_join's response is a status, and that status is treated as a u32
on the wire).  Thus, a big-endian and little-endian machines will
treat this structure differently.

The solution is to have little-endian machines swap the structure when
converting from the structure to the u32 representation.

Signed-off-by: Joel Becker <joel.becker@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
---
 fs/ocfs2/dlm/dlmcommon.h |   20 +++++----
 fs/ocfs2/dlm/dlmdomain.c |  101 +++++++++++++++++++++++++++++----------------
 2 files changed, 76 insertions(+), 45 deletions(-)

diff --git a/fs/ocfs2/dlm/dlmcommon.h b/fs/ocfs2/dlm/dlmcommon.h
index 0f6896d..b90ee17 100644
--- a/fs/ocfs2/dlm/dlmcommon.h
+++ b/fs/ocfs2/dlm/dlmcommon.h
@@ -602,17 +602,19 @@ enum dlm_query_join_response_code {
 	JOIN_PROTOCOL_MISMATCH,
 };
 
+struct dlm_query_join_packet {
+	u8 code;	/* Response code.  dlm_minor and fs_minor
+			   are only valid if this is JOIN_OK */
+	u8 dlm_minor;	/* The minor version of the protocol the
+			   dlm is speaking. */
+	u8 fs_minor;	/* The minor version of the protocol the
+			   filesystem is speaking. */
+	u8 reserved;
+};
+
 union dlm_query_join_response {
 	u32 intval;
-	struct {
-		u8 code;	/* Response code.  dlm_minor and fs_minor
-				   are only valid if this is JOIN_OK */
-		u8 dlm_minor;	/* The minor version of the protocol the
-				   dlm is speaking. */
-		u8 fs_minor;	/* The minor version of the protocol the
-				   filesystem is speaking. */
-		u8 reserved;
-	} packet;
+	struct dlm_query_join_packet packet;
 };
 
 struct dlm_lock_request
diff --git a/fs/ocfs2/dlm/dlmdomain.c b/fs/ocfs2/dlm/dlmdomain.c
index 8861c2d..0d6050d 100644
--- a/fs/ocfs2/dlm/dlmdomain.c
+++ b/fs/ocfs2/dlm/dlmdomain.c
@@ -713,14 +713,46 @@ static int dlm_query_join_proto_check(char *proto_type, int node,
 	return rc;
 }
 
+/*
+ * struct dlm_query_join_packet is made up of four one-byte fields.  They
+ * are effectively in big-endian order already.  However, little-endian
+ * machines swap them before putting the packet on the wire (because
+ * query_join's response is a status, and that status is treated as a u32
+ * on the wire).  Thus, a big-endian and little-endian machines will treat
+ * this structure differently.
+ *
+ * The solution is to have little-endian machines swap the structure when
+ * converting from the structure to the u32 representation.  This will
+ * result in the structure having the correct format on the wire no matter
+ * the host endian format.
+ */
+static void dlm_query_join_packet_to_wire(struct dlm_query_join_packet *packet,
+					  u32 *wire)
+{
+	union dlm_query_join_response response;
+
+	response.packet = *packet;
+	*wire = cpu_to_be32(response.intval);
+}
+
+static void dlm_query_join_wire_to_packet(u32 wire,
+					  struct dlm_query_join_packet *packet)
+{
+	union dlm_query_join_response response;
+
+	response.intval = cpu_to_be32(wire);
+	*packet = response.packet;
+}
+
 static int dlm_query_join_handler(struct o2net_msg *msg, u32 len, void *data,
 				  void **ret_data)
 {
 	struct dlm_query_join_request *query;
-	union dlm_query_join_response response = {
-		.packet.code = JOIN_DISALLOW,
+	struct dlm_query_join_packet packet = {
+		.code = JOIN_DISALLOW,
 	};
 	struct dlm_ctxt *dlm = NULL;
+	u32 response;
 	u8 nodenum;
 
 	query = (struct dlm_query_join_request *) msg->buf;
@@ -737,11 +769,11 @@ static int dlm_query_join_handler(struct o2net_msg *msg, u32 len, void *data,
 		mlog(0, "node %u is not in our live map yet\n",
 		     query->node_idx);
 
-		response.packet.code = JOIN_DISALLOW;
+		packet.code = JOIN_DISALLOW;
 		goto respond;
 	}
 
-	response.packet.code = JOIN_OK_NO_MAP;
+	packet.code = JOIN_OK_NO_MAP;
 
 	spin_lock(&dlm_domain_lock);
 	dlm = __dlm_lookup_domain_full(query->domain, query->name_len);
@@ -760,7 +792,7 @@ static int dlm_query_join_handler(struct o2net_msg *msg, u32 len, void *data,
 				mlog(0, "disallow join as node %u does not "
 				     "have node %u in its nodemap\n",
 				     query->node_idx, nodenum);
-				response.packet.code = JOIN_DISALLOW;
+				packet.code = JOIN_DISALLOW;
 				goto unlock_respond;
 			}
 		}
@@ -780,23 +812,23 @@ static int dlm_query_join_handler(struct o2net_msg *msg, u32 len, void *data,
 			/*If this is a brand new context and we
 			 * haven't started our join process yet, then
 			 * the other node won the race. */
-			response.packet.code = JOIN_OK_NO_MAP;
+			packet.code = JOIN_OK_NO_MAP;
 		} else if (dlm->joining_node != DLM_LOCK_RES_OWNER_UNKNOWN) {
 			/* Disallow parallel joins. */
-			response.packet.code = JOIN_DISALLOW;
+			packet.code = JOIN_DISALLOW;
 		} else if (dlm->reco.state & DLM_RECO_STATE_ACTIVE) {
 			mlog(0, "node %u trying to join, but recovery "
 			     "is ongoing.\n", bit);
-			response.packet.code = JOIN_DISALLOW;
+			packet.code = JOIN_DISALLOW;
 		} else if (test_bit(bit, dlm->recovery_map)) {
 			mlog(0, "node %u trying to join, but it "
 			     "still needs recovery.\n", bit);
-			response.packet.code = JOIN_DISALLOW;
+			packet.code = JOIN_DISALLOW;
 		} else if (test_bit(bit, dlm->domain_map)) {
 			mlog(0, "node %u trying to join, but it "
 			     "is still in the domain! needs recovery?\n",
 			     bit);
-			response.packet.code = JOIN_DISALLOW;
+			packet.code = JOIN_DISALLOW;
 		} else {
 			/* Alright we're fully a part of this domain
 			 * so we keep some state as to who's joining
@@ -807,19 +839,15 @@ static int dlm_query_join_handler(struct o2net_msg *msg, u32 len, void *data,
 			if (dlm_query_join_proto_check("DLM", bit,
 						       &dlm->dlm_locking_proto,
 						       &query->dlm_proto)) {
-				response.packet.code =
-					JOIN_PROTOCOL_MISMATCH;
+				packet.code = JOIN_PROTOCOL_MISMATCH;
 			} else if (dlm_query_join_proto_check("fs", bit,
 							      &dlm->fs_locking_proto,
 							      &query->fs_proto)) {
-				response.packet.code =
-					JOIN_PROTOCOL_MISMATCH;
+				packet.code = JOIN_PROTOCOL_MISMATCH;
 			} else {
-				response.packet.dlm_minor =
-					query->dlm_proto.pv_minor;
-				response.packet.fs_minor =
-					query->fs_proto.pv_minor;
-				response.packet.code = JOIN_OK;
+				packet.dlm_minor = query->dlm_proto.pv_minor;
+				packet.fs_minor = query->fs_proto.pv_minor;
+				packet.code = JOIN_OK;
 				__dlm_set_joining_node(dlm, query->node_idx);
 			}
 		}
@@ -830,9 +858,10 @@ unlock_respond:
 	spin_unlock(&dlm_domain_lock);
 
 respond:
-	mlog(0, "We respond with %u\n", response.packet.code);
+	mlog(0, "We respond with %u\n", packet.code);
 
-	return response.intval;
+	dlm_query_join_packet_to_wire(&packet, &response);
+	return response;
 }
 
 static int dlm_assert_joined_handler(struct o2net_msg *msg, u32 len, void *data,
@@ -968,7 +997,8 @@ static int dlm_request_join(struct dlm_ctxt *dlm,
 {
 	int status;
 	struct dlm_query_join_request join_msg;
-	union dlm_query_join_response join_resp;
+	struct dlm_query_join_packet packet;
+	u32 join_resp;
 
 	mlog(0, "querying node %d\n", node);
 
@@ -984,11 +1014,12 @@ static int dlm_request_join(struct dlm_ctxt *dlm,
 
 	status = o2net_send_message(DLM_QUERY_JOIN_MSG, DLM_MOD_KEY, &join_msg,
 				    sizeof(join_msg), node,
-				    &join_resp.intval);
+				    &join_resp);
 	if (status < 0 && status != -ENOPROTOOPT) {
 		mlog_errno(status);
 		goto bail;
 	}
+	dlm_query_join_wire_to_packet(join_resp, &packet);
 
 	/* -ENOPROTOOPT from the net code means the other side isn't
 	    listening for our message type -- that's fine, it means
@@ -997,10 +1028,10 @@ static int dlm_request_join(struct dlm_ctxt *dlm,
 	if (status == -ENOPROTOOPT) {
 		status = 0;
 		*response = JOIN_OK_NO_MAP;
-	} else if (join_resp.packet.code == JOIN_DISALLOW ||
-		   join_resp.packet.code == JOIN_OK_NO_MAP) {
-		*response = join_resp.packet.code;
-	} else if (join_resp.packet.code == JOIN_PROTOCOL_MISMATCH) {
+	} else if (packet.code == JOIN_DISALLOW ||
+		   packet.code == JOIN_OK_NO_MAP) {
+		*response = packet.code;
+	} else if (packet.code == JOIN_PROTOCOL_MISMATCH) {
 		mlog(ML_NOTICE,
 		     "This node requested DLM locking protocol %u.%u and "
 		     "filesystem locking protocol %u.%u.  At least one of "
@@ -1012,14 +1043,12 @@ static int dlm_request_join(struct dlm_ctxt *dlm,
 		     dlm->fs_locking_proto.pv_minor,
 		     node);
 		status = -EPROTO;
-		*response = join_resp.packet.code;
-	} else if (join_resp.packet.code == JOIN_OK) {
-		*response = join_resp.packet.code;
+		*response = packet.code;
+	} else if (packet.code == JOIN_OK) {
+		*response = packet.code;
 		/* Use the same locking protocol as the remote node */
-		dlm->dlm_locking_proto.pv_minor =
-			join_resp.packet.dlm_minor;
-		dlm->fs_locking_proto.pv_minor =
-			join_resp.packet.fs_minor;
+		dlm->dlm_locking_proto.pv_minor = packet.dlm_minor;
+		dlm->fs_locking_proto.pv_minor = packet.fs_minor;
 		mlog(0,
 		     "Node %d responds JOIN_OK with DLM locking protocol "
 		     "%u.%u and fs locking protocol %u.%u\n",
@@ -1031,11 +1060,11 @@ static int dlm_request_join(struct dlm_ctxt *dlm,
 	} else {
 		status = -EINVAL;
 		mlog(ML_ERROR, "invalid response %d from node %u\n",
-		     join_resp.packet.code, node);
+		     packet.code, node);
 	}
 
 	mlog(0, "status %d, node %d response is %d\n", status, node,
-		  *response);
+	     *response);
 
 bail:
 	return status;
-- 
1.5.3.4

[Ocfs2-devel] [PATCH 15/18] ocfs2/dlm: Add missing dlm_lockres_put()s

[Sunil Mushran]

Mainline commit b31cfc0237f89c3a8bc8f31b5da996e71b543214
Author: Sunil Mushran <sunil.mushran@oracle.com>
Date: Sat, 1 Mar 2008 14:04:22 -0800

dlm_master_request_handler() forgot to put a lockres when
dlm_assert_master_worker() failed or was skipped.

Signed-off-by: Sunil Mushran <sunil.mushran@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
---
 fs/ocfs2/dlm/dlmmaster.c |   11 ++++++++++-
 1 files changed, 10 insertions(+), 1 deletions(-)

diff --git a/fs/ocfs2/dlm/dlmmaster.c b/fs/ocfs2/dlm/dlmmaster.c
index e5ceda9..c31fee7 100644
--- a/fs/ocfs2/dlm/dlmmaster.c
+++ b/fs/ocfs2/dlm/dlmmaster.c
@@ -1663,7 +1663,12 @@ way_up_top:
 		dlm_put_mle(tmpmle);
 	}
 send_response:
-
+	/*
+	 * __dlm_lookup_lockres() grabbed a reference to this lockres.
+	 * The reference is released by dlm_assert_master_worker() under
+	 * the call to dlm_dispatch_assert_master().  If
+	 * dlm_assert_master_worker() isn't called, we drop it here.
+	 */
 	if (dispatch_assert) {
 		if (response != DLM_MASTER_RESP_YES)
 			mlog(ML_ERROR, "invalid response %d\n", response);
@@ -1678,7 +1683,11 @@ send_response:
 		if (ret < 0) {
 			mlog(ML_ERROR, "failed to dispatch assert master work\n");
 			response = DLM_MASTER_RESP_ERROR;
+			dlm_lockres_put(res);
 		}
+	} else {
+		if (res)
+			dlm_lockres_put(res);
 	}
 
 	dlm_put(dlm);
-- 
1.5.3.4

[Ocfs2-devel] [PATCH 16/18] ocfs2/dlm: Print message showing the recovery master

[Sunil Mushran]

Mainline commit 535f7026fddafce6d0a0524db01a432c23a0a7b4
Author: Sunil Mushran <sunil.mushran@oracle.com>
Date: Sat, 1 Mar 2008 14:04:24 -0800

Knowing the dlm recovery master helps in debugging recovery
issues. This patch prints a message on the recovery master node.

Signed-off-by: Sunil Mushran <sunil.mushran@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
---
 fs/ocfs2/dlm/dlmrecovery.c |    6 +++---
 1 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/fs/ocfs2/dlm/dlmrecovery.c b/fs/ocfs2/dlm/dlmrecovery.c
index 7843ec1..79f111a 100644
--- a/fs/ocfs2/dlm/dlmrecovery.c
+++ b/fs/ocfs2/dlm/dlmrecovery.c
@@ -519,9 +519,9 @@ static int dlm_do_recovery(struct dlm_ctxt *dlm)
 	return 0;
 
 master_here:
-	mlog(0, "(%d) mastering recovery of %s:%u here(this=%u)!\n",
-	     dlm->dlm_reco_thread_task->pid,
-	     dlm->name, dlm->reco.dead_node, dlm->node_num);
+	mlog(ML_NOTICE, "(%d) Node %u is the Recovery Master for the Dead Node "
+	     "%u for Domain %s\n", task_pid_nr(dlm->dlm_reco_thread_task),
+	     dlm->node_num, dlm->reco.dead_node, dlm->name);
 
 	status = dlm_remaster_locks(dlm, dlm->reco.dead_node);
 	if (status < 0) {
-- 
1.5.3.4

[Ocfs2-devel] [PATCH 03/18] ocfs2: Negotiate locking protocol versions.

[Sunil Mushran]

Mainline commit d24fbcda0c4988322949df3d759f1cfb32b32953
Author: Joel Becker <Joel.Becker@oracle.com>
Date: Fri, 25 Jan 2008 17:02:21 -0800

Currently, when ocfs2 nodes connect via TCP, they advertise their
compatibility level.  If the versions do not match, two nodes cannot speak
to each other and they disconnect. As a result, this provides no forward or
backwards compatibility.

This patch implements a simple protocol negotiation at the dlm level by
introducing a major/minor version number scheme for entities that
communicate.  Specifically, o2dlm has a major/minor version for interaction
with o2dlm on other nodes, and ocfs2 itself has a major/minor version for
interacting with the filesystem on other nodes.

This will allow rolling upgrades of ocfs2 clusters when changes to the
locking or network protocols can be done in a backwards compatible manner.
In those cases, only the minor number is changed and the negotatied protocol
minor is returned from dlm join. In the far less likely event that a
required protocol change makes backwards compatibility impossible, we simply
bump the major number.

Signed-off-by: Joel Becker <joel.becker@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
---
 fs/ocfs2/cluster/tcp_internal.h |   11 ++-
 fs/ocfs2/dlm/dlmapi.h           |    7 +-
 fs/ocfs2/dlm/dlmcommon.h        |   24 +++++-
 fs/ocfs2/dlm/dlmdomain.c        |  195 +++++++++++++++++++++++++++++++++-----
 fs/ocfs2/dlm/dlmfs.c            |   15 +++-
 fs/ocfs2/dlm/userdlm.c          |    5 +-
 fs/ocfs2/dlm/userdlm.h          |    3 +-
 fs/ocfs2/dlmglue.c              |   29 ++++++-
 fs/ocfs2/dlmglue.h              |    1 +
 fs/ocfs2/ocfs2.h                |    1 +
 fs/ocfs2/ocfs2_lockingver.h     |   30 ++++++
 fs/ocfs2/super.c                |    1 +
 12 files changed, 288 insertions(+), 34 deletions(-)
 create mode 100644 fs/ocfs2/ocfs2_lockingver.h

diff --git a/fs/ocfs2/cluster/tcp_internal.h b/fs/ocfs2/cluster/tcp_internal.h
index fe1a407..b4c5586 100644
--- a/fs/ocfs2/cluster/tcp_internal.h
+++ b/fs/ocfs2/cluster/tcp_internal.h
@@ -38,6 +38,15 @@
  * locking semantics of the file system using the protocol.  It should 
  * be somewhere else, I'm sure, but right now it isn't.
  *
+ * With version 11, we separate out the filesystem locking portion.  The
+ * filesystem now has a major.minor version it negotiates.  Version 11
+ * introduces this negotiation to the o2dlm protocol, and as such the
+ * version here in tcp_internal.h should not need to be bumped for
+ * filesystem locking changes.
+ *
+ * New in version 11
+ * 	- Negotiation of filesystem locking in the dlm join.
+ *
  * New in version 10:
  * 	- Meta/data locks combined
  *
@@ -66,7 +75,7 @@
  * 	- full 64 bit i_size in the metadata lock lvbs
  * 	- introduction of "rw" lock and pushing meta/data locking down
  */
-#define O2NET_PROTOCOL_VERSION 10ULL
+#define O2NET_PROTOCOL_VERSION 11ULL
 struct o2net_handshake {
 	__be64	protocol_version;
 	__be64	connector_id;
diff --git a/fs/ocfs2/dlm/dlmapi.h b/fs/ocfs2/dlm/dlmapi.h
index cfd5cb6..b5786a7 100644
--- a/fs/ocfs2/dlm/dlmapi.h
+++ b/fs/ocfs2/dlm/dlmapi.h
@@ -193,7 +193,12 @@ enum dlm_status dlmunlock(struct dlm_ctxt *dlm,
 			  dlm_astunlockfunc_t *unlockast,
 			  void *data);
 
-struct dlm_ctxt * dlm_register_domain(const char *domain, u32 key);
+struct dlm_protocol_version {
+	u8 pv_major;
+	u8 pv_minor;
+};
+struct dlm_ctxt * dlm_register_domain(const char *domain, u32 key,
+				      struct dlm_protocol_version *fs_proto);
 
 void dlm_unregister_domain(struct dlm_ctxt *dlm);
 
diff --git a/fs/ocfs2/dlm/dlmcommon.h b/fs/ocfs2/dlm/dlmcommon.h
index cddf85a..0f6896d 100644
--- a/fs/ocfs2/dlm/dlmcommon.h
+++ b/fs/ocfs2/dlm/dlmcommon.h
@@ -142,6 +142,12 @@ struct dlm_ctxt
 	spinlock_t work_lock;
 	struct list_head dlm_domain_handlers;
 	struct list_head	dlm_eviction_callbacks;
+
+	/* The filesystem specifies this at domain registration.  We
+	 * cache it here to know what to tell other nodes. */
+	struct dlm_protocol_version fs_locking_proto;
+	/* This is the inter-dlm communication version */
+	struct dlm_protocol_version dlm_locking_proto;
 };
 
 static inline struct hlist_head *dlm_lockres_hash(struct dlm_ctxt *dlm, unsigned i)
@@ -589,10 +595,24 @@ struct dlm_proxy_ast
 #define DLM_PROXY_AST_MAX_LEN  (sizeof(struct dlm_proxy_ast)+DLM_LVB_LEN)
 
 #define DLM_MOD_KEY (0x666c6172)
-enum dlm_query_join_response {
+enum dlm_query_join_response_code {
 	JOIN_DISALLOW = 0,
 	JOIN_OK,
 	JOIN_OK_NO_MAP,
+	JOIN_PROTOCOL_MISMATCH,
+};
+
+union dlm_query_join_response {
+	u32 intval;
+	struct {
+		u8 code;	/* Response code.  dlm_minor and fs_minor
+				   are only valid if this is JOIN_OK */
+		u8 dlm_minor;	/* The minor version of the protocol the
+				   dlm is speaking. */
+		u8 fs_minor;	/* The minor version of the protocol the
+				   filesystem is speaking. */
+		u8 reserved;
+	} packet;
 };
 
 struct dlm_lock_request
@@ -633,6 +653,8 @@ struct dlm_query_join_request
 	u8 node_idx;
 	u8 pad1[2];
 	u8 name_len;
+	struct dlm_protocol_version dlm_proto;
+	struct dlm_protocol_version fs_proto;
 	u8 domain[O2NM_MAX_NAME_LEN];
 	u8 node_map[BITS_TO_BYTES(O2NM_MAX_NODES)];
 };
diff --git a/fs/ocfs2/dlm/dlmdomain.c b/fs/ocfs2/dlm/dlmdomain.c
index 10b3e60..3cd509b 100644
--- a/fs/ocfs2/dlm/dlmdomain.c
+++ b/fs/ocfs2/dlm/dlmdomain.c
@@ -123,6 +123,17 @@ DEFINE_SPINLOCK(dlm_domain_lock);
 LIST_HEAD(dlm_domains);
 static DECLARE_WAIT_QUEUE_HEAD(dlm_domain_events);
 
+/*
+ * The supported protocol version for DLM communication.  Running domains
+ * will have a negotiated version with the same major number and a minor
+ * number equal or smaller.  The dlm_ctxt->dlm_locking_proto field should
+ * be used to determine what a running domain is actually using.
+ */
+static const struct dlm_protocol_version dlm_protocol = {
+	.pv_major = 1,
+	.pv_minor = 0,
+};
+
 #define DLM_DOMAIN_BACKOFF_MS 200
 
 static int dlm_query_join_handler(struct o2net_msg *msg, u32 len, void *data,
@@ -133,6 +144,8 @@ static int dlm_cancel_join_handler(struct o2net_msg *msg, u32 len, void *data,
 				   void **ret_data);
 static int dlm_exit_domain_handler(struct o2net_msg *msg, u32 len, void *data,
 				   void **ret_data);
+static int dlm_protocol_compare(struct dlm_protocol_version *existing,
+				struct dlm_protocol_version *request);
 
 static void dlm_unregister_domain_handlers(struct dlm_ctxt *dlm);
 
@@ -668,11 +681,45 @@ void dlm_unregister_domain(struct dlm_ctxt *dlm)
 }
 EXPORT_SYMBOL_GPL(dlm_unregister_domain);
 
+static int dlm_query_join_proto_check(char *proto_type, int node,
+				      struct dlm_protocol_version *ours,
+				      struct dlm_protocol_version *request)
+{
+	int rc;
+	struct dlm_protocol_version proto = *request;
+
+	if (!dlm_protocol_compare(ours, &proto)) {
+		mlog(0,
+		     "node %u wanted to join with %s locking protocol "
+		     "%u.%u, we respond with %u.%u\n",
+		     node, proto_type,
+		     request->pv_major,
+		     request->pv_minor,
+		     proto.pv_major, proto.pv_minor);
+		request->pv_minor = proto.pv_minor;
+		rc = 0;
+	} else {
+		mlog(ML_NOTICE,
+		     "Node %u wanted to join with %s locking "
+		     "protocol %u.%u, but we have %u.%u, disallowing\n",
+		     node, proto_type,
+		     request->pv_major,
+		     request->pv_minor,
+		     ours->pv_major,
+		     ours->pv_minor);
+		rc = 1;
+	}
+
+	return rc;
+}
+
 static int dlm_query_join_handler(struct o2net_msg *msg, u32 len, void *data,
 				  void **ret_data)
 {
 	struct dlm_query_join_request *query;
-	enum dlm_query_join_response response;
+	union dlm_query_join_response response = {
+		.packet.code = JOIN_DISALLOW,
+	};
 	struct dlm_ctxt *dlm = NULL;
 	u8 nodenum;
 
@@ -690,11 +737,11 @@ static int dlm_query_join_handler(struct o2net_msg *msg, u32 len, void *data,
 		mlog(0, "node %u is not in our live map yet\n",
 		     query->node_idx);
 
-		response = JOIN_DISALLOW;
+		response.packet.code = JOIN_DISALLOW;
 		goto respond;
 	}
 
-	response = JOIN_OK_NO_MAP;
+	response.packet.code = JOIN_OK_NO_MAP;
 
 	spin_lock(&dlm_domain_lock);
 	dlm = __dlm_lookup_domain_full(query->domain, query->name_len);
@@ -713,7 +760,7 @@ static int dlm_query_join_handler(struct o2net_msg *msg, u32 len, void *data,
 				mlog(0, "disallow join as node %u does not "
 				     "have node %u in its nodemap\n",
 				     query->node_idx, nodenum);
-				response = JOIN_DISALLOW;
+				response.packet.code = JOIN_DISALLOW;
 				goto unlock_respond;
 			}
 		}
@@ -733,30 +780,48 @@ static int dlm_query_join_handler(struct o2net_msg *msg, u32 len, void *data,
 			/*If this is a brand new context and we
 			 * haven't started our join process yet, then
 			 * the other node won the race. */
-			response = JOIN_OK_NO_MAP;
+			response.packet.code = JOIN_OK_NO_MAP;
 		} else if (dlm->joining_node != DLM_LOCK_RES_OWNER_UNKNOWN) {
 			/* Disallow parallel joins. */
-			response = JOIN_DISALLOW;
+			response.packet.code = JOIN_DISALLOW;
 		} else if (dlm->reco.state & DLM_RECO_STATE_ACTIVE) {
 			mlog(0, "node %u trying to join, but recovery "
 			     "is ongoing.\n", bit);
-			response = JOIN_DISALLOW;
+			response.packet.code = JOIN_DISALLOW;
 		} else if (test_bit(bit, dlm->recovery_map)) {
 			mlog(0, "node %u trying to join, but it "
 			     "still needs recovery.\n", bit);
-			response = JOIN_DISALLOW;
+			response.packet.code = JOIN_DISALLOW;
 		} else if (test_bit(bit, dlm->domain_map)) {
 			mlog(0, "node %u trying to join, but it "
 			     "is still in the domain! needs recovery?\n",
 			     bit);
-			response = JOIN_DISALLOW;
+			response.packet.code = JOIN_DISALLOW;
 		} else {
 			/* Alright we're fully a part of this domain
 			 * so we keep some state as to who's joining
 			 * and indicate to him that needs to be fixed
 			 * up. */
-			response = JOIN_OK;
-			__dlm_set_joining_node(dlm, query->node_idx);
+
+			/* Make sure we speak compatible locking protocols.  */
+			if (dlm_query_join_proto_check("DLM", bit,
+						       &dlm->dlm_locking_proto,
+						       &query->dlm_proto)) {
+				response.packet.code =
+					JOIN_PROTOCOL_MISMATCH;
+			} else if (dlm_query_join_proto_check("fs", bit,
+							      &dlm->fs_locking_proto,
+							      &query->fs_proto)) {
+				response.packet.code =
+					JOIN_PROTOCOL_MISMATCH;
+			} else {
+				response.packet.dlm_minor =
+					query->dlm_proto.pv_minor;
+				response.packet.fs_minor =
+					query->fs_proto.pv_minor;
+				response.packet.code = JOIN_OK;
+				__dlm_set_joining_node(dlm, query->node_idx);
+			}
 		}
 
 		spin_unlock(&dlm->spinlock);
@@ -765,9 +830,9 @@ unlock_respond:
 	spin_unlock(&dlm_domain_lock);
 
 respond:
-	mlog(0, "We respond with %u\n", response);
+	mlog(0, "We respond with %u\n", response.packet.code);
 
-	return response;
+	return response.intval;
 }
 
 static int dlm_assert_joined_handler(struct o2net_msg *msg, u32 len, void *data,
@@ -899,10 +964,11 @@ static int dlm_send_join_cancels(struct dlm_ctxt *dlm,
 
 static int dlm_request_join(struct dlm_ctxt *dlm,
 			    int node,
-			    enum dlm_query_join_response *response)
+			    enum dlm_query_join_response_code *response)
 {
-	int status, retval;
+	int status;
 	struct dlm_query_join_request join_msg;
+	union dlm_query_join_response join_resp;
 
 	mlog(0, "querying node %d\n", node);
 
@@ -910,12 +976,15 @@ static int dlm_request_join(struct dlm_ctxt *dlm,
 	join_msg.node_idx = dlm->node_num;
 	join_msg.name_len = strlen(dlm->name);
 	memcpy(join_msg.domain, dlm->name, join_msg.name_len);
+	join_msg.dlm_proto = dlm->dlm_locking_proto;
+	join_msg.fs_proto = dlm->fs_locking_proto;
 
 	/* copy live node map to join message */
 	byte_copymap(join_msg.node_map, dlm->live_nodes_map, O2NM_MAX_NODES);
 
 	status = o2net_send_message(DLM_QUERY_JOIN_MSG, DLM_MOD_KEY, &join_msg,
-				    sizeof(join_msg), node, &retval);
+				    sizeof(join_msg), node,
+				    &join_resp.intval);
 	if (status < 0 && status != -ENOPROTOOPT) {
 		mlog_errno(status);
 		goto bail;
@@ -928,14 +997,41 @@ static int dlm_request_join(struct dlm_ctxt *dlm,
 	if (status == -ENOPROTOOPT) {
 		status = 0;
 		*response = JOIN_OK_NO_MAP;
-	} else if (retval == JOIN_DISALLOW ||
-		   retval == JOIN_OK ||
-		   retval == JOIN_OK_NO_MAP) {
-		*response = retval;
+	} else if (join_resp.packet.code == JOIN_DISALLOW ||
+		   join_resp.packet.code == JOIN_OK_NO_MAP) {
+		*response = join_resp.packet.code;
+	} else if (join_resp.packet.code == JOIN_PROTOCOL_MISMATCH) {
+		mlog(ML_NOTICE,
+		     "This node requested DLM locking protocol %u.%u and "
+		     "filesystem locking protocol %u.%u.  At least one of "
+		     "the protocol versions on node %d is not compatible, "
+		     "disconnecting\n",
+		     dlm->dlm_locking_proto.pv_major,
+		     dlm->dlm_locking_proto.pv_minor,
+		     dlm->fs_locking_proto.pv_major,
+		     dlm->fs_locking_proto.pv_minor,
+		     node);
+		status = -EPROTO;
+		*response = join_resp.packet.code;
+	} else if (join_resp.packet.code == JOIN_OK) {
+		*response = join_resp.packet.code;
+		/* Use the same locking protocol as the remote node */
+		dlm->dlm_locking_proto.pv_minor =
+			join_resp.packet.dlm_minor;
+		dlm->fs_locking_proto.pv_minor =
+			join_resp.packet.fs_minor;
+		mlog(0,
+		     "Node %d responds JOIN_OK with DLM locking protocol "
+		     "%u.%u and fs locking protocol %u.%u\n",
+		     node,
+		     dlm->dlm_locking_proto.pv_major,
+		     dlm->dlm_locking_proto.pv_minor,
+		     dlm->fs_locking_proto.pv_major,
+		     dlm->fs_locking_proto.pv_minor);
 	} else {
 		status = -EINVAL;
-		mlog(ML_ERROR, "invalid response %d from node %u\n", retval,
-		     node);
+		mlog(ML_ERROR, "invalid response %d from node %u\n",
+		     join_resp.packet.code, node);
 	}
 
 	mlog(0, "status %d, node %d response is %d\n", status, node,
@@ -1008,7 +1104,7 @@ struct domain_join_ctxt {
 
 static int dlm_should_restart_join(struct dlm_ctxt *dlm,
 				   struct domain_join_ctxt *ctxt,
-				   enum dlm_query_join_response response)
+				   enum dlm_query_join_response_code response)
 {
 	int ret;
 
@@ -1034,7 +1130,7 @@ static int dlm_try_to_join_domain(struct dlm_ctxt *dlm)
 {
 	int status = 0, tmpstat, node;
 	struct domain_join_ctxt *ctxt;
-	enum dlm_query_join_response response = JOIN_DISALLOW;
+	enum dlm_query_join_response_code response = JOIN_DISALLOW;
 
 	mlog_entry("%p", dlm);
 
@@ -1450,10 +1546,38 @@ leave:
 }
 
 /*
- * dlm_register_domain: one-time setup per "domain"
+ * Compare a requested locking protocol version against the current one.
+ *
+ * If the major numbers are different, they are incompatible.
+ * If the current minor is greater than the request, they are incompatible.
+ * If the current minor is less than or equal to the request, they are
+ * compatible, and the requester should run at the current minor version.
+ */
+static int dlm_protocol_compare(struct dlm_protocol_version *existing,
+				struct dlm_protocol_version *request)
+{
+	if (existing->pv_major != request->pv_major)
+		return 1;
+
+	if (existing->pv_minor > request->pv_minor)
+		return 1;
+
+	if (existing->pv_minor < request->pv_minor)
+		request->pv_minor = existing->pv_minor;
+
+	return 0;
+}
+
+/*
+ * dlm_register_domain: one-time setup per "domain".
+ *
+ * The filesystem passes in the requested locking version via proto.
+ * If registration was successful, proto will contain the negotiated
+ * locking protocol.
  */
 struct dlm_ctxt * dlm_register_domain(const char *domain,
-			       u32 key)
+			       u32 key,
+			       struct dlm_protocol_version *fs_proto)
 {
 	int ret;
 	struct dlm_ctxt *dlm = NULL;
@@ -1496,6 +1620,15 @@ retry:
 			goto retry;
 		}
 
+		if (dlm_protocol_compare(&dlm->fs_locking_proto, fs_proto)) {
+			mlog(ML_ERROR,
+			     "Requested locking protocol version is not "
+			     "compatible with already registered domain "
+			     "\"%s\"\n", domain);
+			ret = -EPROTO;
+			goto leave;
+		}
+
 		__dlm_get(dlm);
 		dlm->num_joins++;
 
@@ -1526,6 +1659,13 @@ retry:
 	list_add_tail(&dlm->list, &dlm_domains);
 	spin_unlock(&dlm_domain_lock);
 
+	/*
+	 * Pass the locking protocol version into the join.  If the join
+	 * succeeds, it will have the negotiated protocol set.
+	 */
+	dlm->dlm_locking_proto = dlm_protocol;
+	dlm->fs_locking_proto = *fs_proto;
+
 	ret = dlm_join_domain(dlm);
 	if (ret) {
 		mlog_errno(ret);
@@ -1533,6 +1673,9 @@ retry:
 		goto leave;
 	}
 
+	/* Tell the caller what locking protocol we negotiated */
+	*fs_proto = dlm->fs_locking_proto;
+
 	ret = 0;
 leave:
 	if (new_ctxt)
diff --git a/fs/ocfs2/dlm/dlmfs.c b/fs/ocfs2/dlm/dlmfs.c
index f7e0263..c30fff5 100644
--- a/fs/ocfs2/dlm/dlmfs.c
+++ b/fs/ocfs2/dlm/dlmfs.c
@@ -60,6 +60,8 @@
 #define MLOG_MASK_PREFIX ML_DLMFS
 #include "cluster/masklog.h"
 
+#include "ocfs2_lockingver.h"
+
 #ifdef SOP_IS_NOT_CONST
 static struct super_operations dlmfs_ops;
 #else
@@ -84,6 +86,16 @@ static struct kmem_cache *dlmfs_inode_cache;
 struct workqueue_struct *user_dlm_worker;
 
 /*
+ * This is the userdlmfs locking protocol version.
+ *
+ * See fs/ocfs2/dlmglue.c for more details on locking versions.
+ */
+static const struct dlm_protocol_version user_locking_protocol = {
+	.pv_major = OCFS2_LOCKING_PROTOCOL_MAJOR,
+	.pv_minor = OCFS2_LOCKING_PROTOCOL_MINOR,
+};
+
+/*
  * decodes a set of open flags into a valid lock level and a set of flags.
  * returns < 0 if we have invalid flags
  * flags which mean something to us:
@@ -431,6 +443,7 @@ static int dlmfs_mkdir(struct inode * dir,
 	struct qstr *domain = &dentry->d_name;
 	struct dlmfs_inode_private *ip;
 	struct dlm_ctxt *dlm;
+	struct dlm_protocol_version proto = user_locking_protocol;
 
 	mlog(0, "mkdir %.*s\n", domain->len, domain->name);
 
@@ -450,7 +463,7 @@ static int dlmfs_mkdir(struct inode * dir,
 
 	ip = DLMFS_I(inode);
 
-	dlm = user_dlm_register_context(domain);
+	dlm = user_dlm_register_context(domain, &proto);
 	if (IS_ERR(dlm)) {
 		status = PTR_ERR(dlm);
 		mlog(ML_ERROR, "Error %d could not register domain \"%.*s\"\n",
diff --git a/fs/ocfs2/dlm/userdlm.c b/fs/ocfs2/dlm/userdlm.c
index 6340da3..2181e75 100644
--- a/fs/ocfs2/dlm/userdlm.c
+++ b/fs/ocfs2/dlm/userdlm.c
@@ -645,7 +645,8 @@ bail:
 	return status;
 }
 
-struct dlm_ctxt *user_dlm_register_context(struct qstr *name)
+struct dlm_ctxt *user_dlm_register_context(struct qstr *name,
+					   struct dlm_protocol_version *proto)
 {
 	struct dlm_ctxt *dlm;
 	u32 dlm_key;
@@ -661,7 +662,7 @@ struct dlm_ctxt *user_dlm_register_context(struct qstr *name)
 
 	snprintf(domain, name->len + 1, "%.*s", name->len, name->name);
 
-	dlm = dlm_register_domain(domain, dlm_key);
+	dlm = dlm_register_domain(domain, dlm_key, proto);
 	if (IS_ERR(dlm))
 		mlog_errno(PTR_ERR(dlm));
 
diff --git a/fs/ocfs2/dlm/userdlm.h b/fs/ocfs2/dlm/userdlm.h
index c400e93..39ec277 100644
--- a/fs/ocfs2/dlm/userdlm.h
+++ b/fs/ocfs2/dlm/userdlm.h
@@ -83,7 +83,8 @@ void user_dlm_write_lvb(struct inode *inode,
 void user_dlm_read_lvb(struct inode *inode,
 		       char *val,
 		       unsigned int len);
-struct dlm_ctxt *user_dlm_register_context(struct qstr *name);
+struct dlm_ctxt *user_dlm_register_context(struct qstr *name,
+					   struct dlm_protocol_version *proto);
 void user_dlm_unregister_context(struct dlm_ctxt *dlm);
 
 struct dlmfs_inode_private {
diff --git a/fs/ocfs2/dlmglue.c b/fs/ocfs2/dlmglue.c
index df6a8fd..0ff457c 100644
--- a/fs/ocfs2/dlmglue.c
+++ b/fs/ocfs2/dlmglue.c
@@ -43,6 +43,7 @@
 #include <cluster/masklog.h>
 
 #include "ocfs2.h"
+#include "ocfs2_lockingver.h"
 
 #include "alloc.h"
 #include "dcache.h"
@@ -258,6 +259,31 @@ static struct ocfs2_lock_res_ops ocfs2_flock_lops = {
 	.flags		= 0,
 };
 
+/*
+ * This is the filesystem locking protocol version.
+ *
+ * Whenever the filesystem does new things with locks (adds or removes a
+ * lock, orders them differently, does different things underneath a lock),
+ * the version must be changed.  The protocol is negotiated when joining
+ * the dlm domain.  A node may join the domain if its major version is
+ * identical to all other nodes and its minor version is greater than
+ * or equal to all other nodes.  When its minor version is greater than
+ * the other nodes, it will run at the minor version specified by the
+ * other nodes.
+ *
+ * If a locking change is made that will not be compatible with older
+ * versions, the major number must be increased and the minor version set
+ * to zero.  If a change merely adds a behavior that can be disabled when
+ * speaking to older versions, the minor version must be increased.  If a
+ * change adds a fully backwards compatible change (eg, LVB changes that
+ * are just ignored by older versions), the version does not need to be
+ * updated.
+ */
+const struct dlm_protocol_version ocfs2_locking_protocol = {
+	.pv_major = OCFS2_LOCKING_PROTOCOL_MAJOR,
+	.pv_minor = OCFS2_LOCKING_PROTOCOL_MINOR,
+};
+
 static inline int ocfs2_is_inode_lock(struct ocfs2_lock_res *lockres)
 {
 	return lockres->l_type == OCFS2_LOCK_TYPE_META ||
@@ -2514,7 +2540,8 @@ int ocfs2_dlm_init(struct ocfs2_super *osb)
 	dlm_key = crc32_le(0, osb->uuid_str, strlen(osb->uuid_str));
 
 	/* for now, uuid == domain */
-	dlm = dlm_register_domain(osb->uuid_str, dlm_key);
+	dlm = dlm_register_domain(osb->uuid_str, dlm_key,
+				  &osb->osb_locking_proto);
 	if (IS_ERR(dlm)) {
 		status = PTR_ERR(dlm);
 		mlog_errno(status);
diff --git a/fs/ocfs2/dlmglue.h b/fs/ocfs2/dlmglue.h
index 5f17243..1d5b069 100644
--- a/fs/ocfs2/dlmglue.h
+++ b/fs/ocfs2/dlmglue.h
@@ -116,4 +116,5 @@ void ocfs2_wake_downconvert_thread(struct ocfs2_super *osb);
 struct ocfs2_dlm_debug *ocfs2_new_dlm_debug(void);
 void ocfs2_put_dlm_debug(struct ocfs2_dlm_debug *dlm_debug);
 
+extern const struct dlm_protocol_version ocfs2_locking_protocol;
 #endif	/* DLMGLUE_H */
diff --git a/fs/ocfs2/ocfs2.h b/fs/ocfs2/ocfs2.h
index b9eafc0..e927d55 100644
--- a/fs/ocfs2/ocfs2.h
+++ b/fs/ocfs2/ocfs2.h
@@ -254,6 +254,7 @@ struct ocfs2_super
 	struct ocfs2_lock_res osb_rename_lockres;
 	struct dlm_eviction_cb osb_eviction_cb;
 	struct ocfs2_dlm_debug *osb_dlm_debug;
+	struct dlm_protocol_version osb_locking_proto;
 
 	struct dentry *osb_debug_root;
 
diff --git a/fs/ocfs2/ocfs2_lockingver.h b/fs/ocfs2/ocfs2_lockingver.h
new file mode 100644
index 0000000..82d5eea
--- /dev/null
+++ b/fs/ocfs2/ocfs2_lockingver.h
@@ -0,0 +1,30 @@
+/* -*- mode: c; c-basic-offset: 8; -*-
+ * vim: noexpandtab sw=8 ts=8 sts=0:
+ *
+ * ocfs2_lockingver.h
+ *
+ * Defines OCFS2 Locking version values.
+ *
+ * Copyright (C) 2008 Oracle.  All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public
+ * License, version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * General Public License for more details.
+ */
+
+#ifndef OCFS2_LOCKINGVER_H
+#define OCFS2_LOCKINGVER_H
+
+/*
+ * The protocol version for ocfs2 cluster locking.  See dlmglue.c for
+ * more details.
+ */
+#define OCFS2_LOCKING_PROTOCOL_MAJOR 1
+#define OCFS2_LOCKING_PROTOCOL_MINOR 0
+
+#endif  /* OCFS2_LOCKINGVER_H */
diff --git a/fs/ocfs2/super.c b/fs/ocfs2/super.c
index 29096b0..f41618d 100644
--- a/fs/ocfs2/super.c
+++ b/fs/ocfs2/super.c
@@ -1411,6 +1411,7 @@ static int ocfs2_initialize_super(struct super_block *sb,
 	sb->s_fs_info = osb;
 	sb->s_op = &ocfs2_sops;
 	sb->s_export_op = &ocfs2_export_ops;
+	osb->osb_locking_proto = ocfs2_locking_protocol;
 	sb->s_time_gran = 1;
 	sb->s_flags |= MS_NOATIME;
 	/* this is needed to support O_LARGEFILE */
-- 
1.5.3.4

[Ocfs2-devel] [PATCH 12/18] ocfs2: Fix an endian bug in online resize

[Sunil Mushran]

Mainline commit 4338ab6a750303cbae4cc76cc7de5edba6598ebe
Author: Tao Ma <tao.ma@oracle.com>
Date: Mon, 3 Mar 2008 10:53:02 +0800

In ocfs2_group_add, 'cr' is a disk field of type 'ocfs2_chain_rec', and we
were putting cpu byteorder values into it. Swap things to the right endian
before storing.

Signed-off-by: Tao Ma <tao.ma@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
---
 fs/ocfs2/resize.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/fs/ocfs2/resize.c b/fs/ocfs2/resize.c
index 37835ff..8166968 100644
--- a/fs/ocfs2/resize.c
+++ b/fs/ocfs2/resize.c
@@ -597,7 +597,7 @@ int ocfs2_group_add(struct inode *inode, struct ocfs2_new_group_input *input)
 		memset(cr, 0, sizeof(struct ocfs2_chain_rec));
 	}
 
-	cr->c_blkno = le64_to_cpu(input->group);
+	cr->c_blkno = cpu_to_le64(input->group);
 	le32_add_cpu(&cr->c_total, input->clusters * cl_bpc);
 	le32_add_cpu(&cr->c_free, input->frees * cl_bpc);
 
-- 
1.5.3.4

[Ocfs2-devel] [PATCH 09/18] ocfs2/dlm: fix printk warning

[Sunil Mushran]

Mainline commit 3a4780a85d4a160a471ed887bfce58b414f556b1
Author: Andrew Morton <akpm@linux-foundation.org>
Date: Fri, 29 Feb 2008 01:56:06 -0800

fs/ocfs2/dlm/dlmdomain.c: In function 'dlm_send_join_cancels':
fs/ocfs2/dlm/dlmdomain.c:983: warning: format '%u' expects type 'unsigned int', but argument 7 has type 'long unsigned int'

Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
---
 fs/ocfs2/dlm/dlmdomain.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/fs/ocfs2/dlm/dlmdomain.c b/fs/ocfs2/dlm/dlmdomain.c
index 3cd509b..8861c2d 100644
--- a/fs/ocfs2/dlm/dlmdomain.c
+++ b/fs/ocfs2/dlm/dlmdomain.c
@@ -937,7 +937,7 @@ static int dlm_send_join_cancels(struct dlm_ctxt *dlm,
 			 sizeof(unsigned long))) {
 		mlog(ML_ERROR,
 		     "map_size %u != BITS_TO_LONGS(O2NM_MAX_NODES) %u\n",
-		     map_size, BITS_TO_LONGS(O2NM_MAX_NODES));
+		     map_size, (unsigned)BITS_TO_LONGS(O2NM_MAX_NODES));
 		return -EINVAL;
 	}
 
-- 
1.5.3.4

[Ocfs2-devel] [PATCH 10/18] ocfs2: Use dlm_print_one_lock_resource for lock resource print

[Sunil Mushran]

Mainline commit 2af37ce82d199d1d8cd6286f42f37d321627a807
Author: Tao Ma <tao.ma@oracle.com>
Date: Thu, 28 Feb 2008 10:41:55 +0800

__dlm_print_one_lock_resource must be called with spin_lock
the res->spinlock. While in some cases, we use it without this
precondition and lead to the failure of assert_spin_locked.
So call dlm_print_one_lock_resource instead.

Signed-off-by: Tao Ma <tao.ma@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
---
 fs/ocfs2/dlm/dlmconvert.c  |    2 +-
 fs/ocfs2/dlm/dlmmaster.c   |    4 ++--
 fs/ocfs2/dlm/dlmrecovery.c |    2 +-
 3 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/fs/ocfs2/dlm/dlmconvert.c b/fs/ocfs2/dlm/dlmconvert.c
index ecb4d99..75997b4 100644
--- a/fs/ocfs2/dlm/dlmconvert.c
+++ b/fs/ocfs2/dlm/dlmconvert.c
@@ -487,7 +487,7 @@ int dlm_convert_lock_handler(struct o2net_msg *msg, u32 len, void *data,
 			       "cookie=%u:%llu\n",
 		     dlm_get_lock_cookie_node(be64_to_cpu(cnv->cookie)),
 		     dlm_get_lock_cookie_seq(be64_to_cpu(cnv->cookie)));
-		__dlm_print_one_lock_resource(res);
+		dlm_print_one_lock_resource(res);
 		goto leave;
 	}
 
diff --git a/fs/ocfs2/dlm/dlmmaster.c b/fs/ocfs2/dlm/dlmmaster.c
index da627c8..5b27581 100644
--- a/fs/ocfs2/dlm/dlmmaster.c
+++ b/fs/ocfs2/dlm/dlmmaster.c
@@ -2348,7 +2348,7 @@ int dlm_deref_lockres_handler(struct o2net_msg *msg, u32 len, void *data,
 			mlog(ML_ERROR, "%s:%.*s: node %u trying to drop ref "
 		     	"but it is already dropped!\n", dlm->name,
 		     	res->lockname.len, res->lockname.name, node);
-			__dlm_print_one_lock_resource(res);
+			dlm_print_one_lock_resource(res);
 		}
 		ret = 0;
 		goto done;
@@ -2408,7 +2408,7 @@ static void dlm_deref_lockres_worker(struct dlm_work_item *item, void *data)
 		mlog(ML_ERROR, "%s:%.*s: node %u trying to drop ref "
 		     "but it is already dropped!\n", dlm->name,
 		     res->lockname.len, res->lockname.name, node);
-		__dlm_print_one_lock_resource(res);
+		dlm_print_one_lock_resource(res);
 	}
 
 	dlm_lockres_put(res);
diff --git a/fs/ocfs2/dlm/dlmrecovery.c b/fs/ocfs2/dlm/dlmrecovery.c
index 796fb73..3004f95 100644
--- a/fs/ocfs2/dlm/dlmrecovery.c
+++ b/fs/ocfs2/dlm/dlmrecovery.c
@@ -1191,7 +1191,7 @@ static int dlm_add_lock_to_array(struct dlm_lock *lock,
 			    (ml->type == LKM_EXMODE ||
 			     memcmp(mres->lvb, lock->lksb->lvb, DLM_LVB_LEN))) {
 				mlog(ML_ERROR, "mismatched lvbs!\n");
-				__dlm_print_one_lock_resource(lock->lockres);
+				dlm_print_one_lock_resource(lock->lockres);
 				BUG();
 			}
 			memcpy(mres->lvb, lock->lksb->lvb, DLM_LVB_LEN);
-- 
1.5.3.4

[Ocfs2-devel] [PATCH 17/18] ocfs2/dlm: dlm_thread should not sleep while holding the dlm_spinlock

[Sunil Mushran]

Mainline commit c824c3c723f2e37a00b3b739a55b28de595fd72e
Author: Sunil Mushran <sunil.mushran@oracle.com>
Date: Sat, 1 Mar 2008 14:04:25 -0800

This patch addresses the bug in which the dlm_thread could go to sleep
while holding the dlm_spinlock.

Signed-off-by: Sunil Mushran <sunil.mushran@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
---
 fs/ocfs2/dlm/dlmthread.c |    6 ++++--
 1 files changed, 4 insertions(+), 2 deletions(-)

diff --git a/fs/ocfs2/dlm/dlmthread.c b/fs/ocfs2/dlm/dlmthread.c
index cebd089..4060bb3 100644
--- a/fs/ocfs2/dlm/dlmthread.c
+++ b/fs/ocfs2/dlm/dlmthread.c
@@ -176,12 +176,14 @@ static int dlm_purge_lockres(struct dlm_ctxt *dlm,
 	     res->lockname.name, master);
 
 	if (!master) {
+		/* drop spinlock...  retake below */
+		spin_unlock(&dlm->spinlock);
+
 		spin_lock(&res->spinlock);
 		/* This ensures that clear refmap is sent after the set */
 		__dlm_wait_on_lockres_flags(res, DLM_LOCK_RES_SETREF_INPROG);
 		spin_unlock(&res->spinlock);
-		/* drop spinlock to do messaging, retake below */
-		spin_unlock(&dlm->spinlock);
+
 		/* clear our bit from the master's refmap, ignore errors */
 		ret = dlm_drop_lockres_ref(dlm, res);
 		if (ret < 0) {
-- 
1.5.3.4

[Ocfs2-devel] [PATCH 18/18] ocfs2: Fix NULL pointer dereferences in o2net

[Sunil Mushran]

Mainline commit cdef59a94c2fc962ada379d4240d556db7b56d55
Author: Tao Ma <tao.ma@oracle.com>
Date: Wed, 5 Mar 2008 15:49:55 +0800

In some situations, ocfs2_set_nn_state might get called with sc = NULL and
valid = 0. If sc = NULL, we can't dereference it to get the o2nm_node
member. Instead, do what o2net_initialize_handshake does and use NULL when
calling o2net_reconnect_delay and o2net_idle_timeout.

Signed-off-by: Tao Ma <tao.ma@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
---
 fs/ocfs2/cluster/tcp.c |    9 ++++-----
 1 files changed, 4 insertions(+), 5 deletions(-)

diff --git a/fs/ocfs2/cluster/tcp.c b/fs/ocfs2/cluster/tcp.c
index 6dbdee0..090018a 100644
--- a/fs/ocfs2/cluster/tcp.c
+++ b/fs/ocfs2/cluster/tcp.c
@@ -451,9 +451,9 @@ static void o2net_set_nn_state(struct o2net_node *nn,
 		/* delay if we're withing a RECONNECT_DELAY of the
 		 * last attempt */
 		delay = (nn->nn_last_connect_attempt +
-			 msecs_to_jiffies(o2net_reconnect_delay(sc->sc_node)))
+			 msecs_to_jiffies(o2net_reconnect_delay(NULL)))
 			- jiffies;
-		if (delay > msecs_to_jiffies(o2net_reconnect_delay(sc->sc_node)))
+		if (delay > msecs_to_jiffies(o2net_reconnect_delay(NULL)))
 			delay = 0;
 		mlog(ML_CONN, "queueing conn attempt in %lu jiffies\n", delay);
 		queue_delayed_work(o2net_wq, &nn->nn_connect_work, delay);
@@ -1583,12 +1583,11 @@ static void o2net_connect_expired(kapi_work_struct_t *work)
 
 	spin_lock(&nn->nn_lock);
 	if (!nn->nn_sc_valid) {
-		struct o2nm_node *node = nn->nn_sc->sc_node;
 		mlog(ML_ERROR, "no connection established with node %u after "
 		     "%u.%u seconds, giving up and returning errors.\n",
 		     o2net_num_from_nn(nn),
-		     o2net_idle_timeout(node) / 1000,
-		     o2net_idle_timeout(node) % 1000);
+		     o2net_idle_timeout(NULL) / 1000,
+		     o2net_idle_timeout(NULL) % 1000);
 
 		o2net_set_nn_state(nn, NULL, 0, -ENOTCONN);
 	}
-- 
1.5.3.4

[Ocfs2-devel] how to use run_lvb_torture.py ?

[Coly Li]

When I tried to run run_lvb_torture.py, there were some parameters I don't know how to use.

From the usage information,
run_lvb_torture.py [-d | --dlmfs <dlmfs path>] [-i | --iteractions <iteractions>] [-H | --hbdev
<heartbeat device>] [-l | --logfile <logfile>] [-n | --nodelist <nodelist>] <domain> <lockname>
[-h|--help]

1) what is heartbeat device? Should I create a separated ocfs2 partition as a heartbeat device ?
2) what is domain and lockname ? how to set it ?

I tried to google it and check the code, since I do not have clear conception for domain/lockname,
the information was very limited for me.

Thanks in advance.

-- 
Coly Li
SuSE PRC Labs

[Ocfs2-devel] how to use run_lvb_torture.py ?

[Marcos E. Matsunaga]

Coly,

You don't have to specify the hbdev. You will have the partition mounted 
to run this test.

Make sure you set /dlm directory to 777. The user needs to have full 
access to /dlm.

The domain is the UUID of the volume you are using.

debugfs.ocfs2 -R stats <device> |grep UUID

As for the lockname, I usually choose the biggest file I have on the 
partition.

To get the lockname:


debugfs.ocfs2 -R "encode <filepath>" <device>



Regards,

Marcos Eduardo Matsunaga

Oracle USA
Linux Engineering

?The statements and opinions expressed here are my own and do not
necessarily represent those of Oracle Corporation.?



Coly Li wrote:
> When I tried to run run_lvb_torture.py, there were some parameters I don't know how to use.
>
> >From the usage information,
> run_lvb_torture.py [-d | --dlmfs <dlmfs path>] [-i | --iteractions <iteractions>] [-H | --hbdev
> <heartbeat device>] [-l | --logfile <logfile>] [-n | --nodelist <nodelist>] <domain> <lockname>
> [-h|--help]
>
> 1) what is heartbeat device? Should I create a separated ocfs2 partition as a heartbeat device ?
> 2) what is domain and lockname ? how to set it ?
>
> I tried to google it and check the code, since I do not have clear conception for domain/lockname,
> the information was very limited for me.
>
> Thanks in advance.
>
>   
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://oss.oracle.com/pipermail/ocfs2-devel/attachments/20080317/2472680d/attachment.html

[Ocfs2-devel] [PATCH 13/18] ocfs2/dlm: Add missing dlm_lock_put()s

[Sunil Mushran]

Mainline commit 2c5c54aca9d0263f81bd4886232835ba31f7635a
Author: Sunil Mushran <sunil.mushran@oracle.com>
Date: Sat, 1 Mar 2008 14:04:20 -0800

Normally locks for remote nodes are freed when that node sends an UNLOCK
message to the master. The master node tags an DLM_UNLOCK_FREE_LOCK action
to do an extra put on the lock at the end.

However, there are times when the master node has to free the locks for the
remote nodes forcibly.

Two cases when this happens are:
1. When the master has migrated the lockres plus all locks to another node.
2. When the master is clearing all the locks of a dead node.

It was in the above two conditions that the dlm was missing the extra put.

Signed-off-by: Sunil Mushran <sunil.mushran@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
---
 fs/ocfs2/dlm/dlmmaster.c   |    3 +++
 fs/ocfs2/dlm/dlmrecovery.c |    9 +++++++++
 2 files changed, 12 insertions(+), 0 deletions(-)

diff --git a/fs/ocfs2/dlm/dlmmaster.c b/fs/ocfs2/dlm/dlmmaster.c
index 5b27581..e5ceda9 100644
--- a/fs/ocfs2/dlm/dlmmaster.c
+++ b/fs/ocfs2/dlm/dlmmaster.c
@@ -2933,6 +2933,9 @@ static void dlm_remove_nonlocal_locks(struct dlm_ctxt *dlm,
 				dlm_lockres_clear_refmap_bit(lock->ml.node, res);
 				list_del_init(&lock->list);
 				dlm_lock_put(lock);
+				/* In a normal unlock, we would have added a
+				 * DLM_UNLOCK_FREE_LOCK action. Force it. */
+				dlm_lock_put(lock);
 			}
 		}
 		queue++;
diff --git a/fs/ocfs2/dlm/dlmrecovery.c b/fs/ocfs2/dlm/dlmrecovery.c
index 3004f95..4a0e7aa 100644
--- a/fs/ocfs2/dlm/dlmrecovery.c
+++ b/fs/ocfs2/dlm/dlmrecovery.c
@@ -2130,11 +2130,16 @@ static void dlm_free_dead_locks(struct dlm_ctxt *dlm,
 	assert_spin_locked(&dlm->spinlock);
 	assert_spin_locked(&res->spinlock);
 
+	/* We do two dlm_lock_put(). One for removing from list and the other is
+	 * to force the DLM_UNLOCK_FREE_LOCK action so as to free the locks */
+
 	/* TODO: check pending_asts, pending_basts here */
 	list_for_each_entry_safe(lock, next, &res->granted, list) {
 		if (lock->ml.node == dead_node) {
 			list_del_init(&lock->list);
 			dlm_lock_put(lock);
+			/* Can't schedule DLM_UNLOCK_FREE_LOCK - do manually */
+			dlm_lock_put(lock);
 			freed++;
 		}
 	}
@@ -2142,6 +2147,8 @@ static void dlm_free_dead_locks(struct dlm_ctxt *dlm,
 		if (lock->ml.node == dead_node) {
 			list_del_init(&lock->list);
 			dlm_lock_put(lock);
+			/* Can't schedule DLM_UNLOCK_FREE_LOCK - do manually */
+			dlm_lock_put(lock);
 			freed++;
 		}
 	}
@@ -2149,6 +2156,8 @@ static void dlm_free_dead_locks(struct dlm_ctxt *dlm,
 		if (lock->ml.node == dead_node) {
 			list_del_init(&lock->list);
 			dlm_lock_put(lock);
+			/* Can't schedule DLM_UNLOCK_FREE_LOCK - do manually */
+			dlm_lock_put(lock);
 			freed++;
 		}
 	}
-- 
1.5.3.4