From e6aabe0cac14a495d42f629a803c5e221089bae8 Mon Sep 17 00:00:00 2001
From: Jan Kara <jack@suse.cz>
Date: Thu, 15 Oct 2009 14:54:03 +0200
Subject: ocfs2: Always include ACL support

To become consistent with filesystems such as XFS or BTRFS, make posix
ACLs always available. This also reduces possibility of
misconfiguration on admin's side.

Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
---
 fs/ocfs2/Kconfig  | 10 +---------
 fs/ocfs2/Makefile |  7 ++-----
 fs/ocfs2/acl.h    | 22 ----------------------
 fs/ocfs2/super.c  |  9 ---------
 fs/ocfs2/xattr.c  |  4 ----
 fs/ocfs2/xattr.h  |  2 --
 6 files changed, 3 insertions(+), 51 deletions(-)

(limited to 'fs/ocfs2')

diff --git a/fs/ocfs2/Kconfig b/fs/ocfs2/Kconfig
index 701b7a3a872..0d840669698 100644
--- a/fs/ocfs2/Kconfig
+++ b/fs/ocfs2/Kconfig
@@ -6,6 +6,7 @@ config OCFS2_FS
 	select CRC32
 	select QUOTA
 	select QUOTA_TREE
+	select FS_POSIX_ACL
 	help
 	  OCFS2 is a general purpose extent based shared disk cluster file
 	  system with many similarities to ext3. It supports 64 bit inode
@@ -74,12 +75,3 @@ config OCFS2_DEBUG_FS
 	  This option will enable expensive consistency checks. Enable
 	  this option for debugging only as it is likely to decrease
 	  performance of the filesystem.
-
-config OCFS2_FS_POSIX_ACL
-	bool "OCFS2 POSIX Access Control Lists"
-	depends on OCFS2_FS
-	select FS_POSIX_ACL
-	default n
-	help
-	  Posix Access Control Lists (ACLs) support permissions for users and
-	  groups beyond the owner/group/world scheme.
diff --git a/fs/ocfs2/Makefile b/fs/ocfs2/Makefile
index 31f25ce32c9..600d2d2ade1 100644
--- a/fs/ocfs2/Makefile
+++ b/fs/ocfs2/Makefile
@@ -39,11 +39,8 @@ ocfs2-objs := \
 	ver.o			\
 	quota_local.o		\
 	quota_global.o		\
-	xattr.o
-
-ifeq ($(CONFIG_OCFS2_FS_POSIX_ACL),y)
-ocfs2-objs += acl.o
-endif
+	xattr.o			\
+	acl.o
 
 ocfs2_stackglue-objs := stackglue.o
 ocfs2_stack_o2cb-objs := stack_o2cb.o
diff --git a/fs/ocfs2/acl.h b/fs/ocfs2/acl.h
index 8f6389ed4da..5c5d31f0585 100644
--- a/fs/ocfs2/acl.h
+++ b/fs/ocfs2/acl.h
@@ -26,8 +26,6 @@ struct ocfs2_acl_entry {
 	__le32 e_id;
 };
 
-#ifdef CONFIG_OCFS2_FS_POSIX_ACL
-
 extern int ocfs2_check_acl(struct inode *, int);
 extern int ocfs2_acl_chmod(struct inode *);
 extern int ocfs2_init_acl(handle_t *, struct inode *, struct inode *,
@@ -35,24 +33,4 @@ extern int ocfs2_init_acl(handle_t *, struct inode *, struct inode *,
 			  struct ocfs2_alloc_context *,
 			  struct ocfs2_alloc_context *);
 
-#else /* CONFIG_OCFS2_FS_POSIX_ACL*/
-
-#define ocfs2_check_acl NULL
-static inline int ocfs2_acl_chmod(struct inode *inode)
-{
-	return 0;
-}
-static inline int ocfs2_init_acl(handle_t *handle,
-				 struct inode *inode,
-				 struct inode *dir,
-				 struct buffer_head *di_bh,
-				 struct buffer_head *dir_bh,
-				 struct ocfs2_alloc_context *meta_ac,
-				 struct ocfs2_alloc_context *data_ac)
-{
-	return 0;
-}
-
-#endif /* CONFIG_OCFS2_FS_POSIX_ACL*/
-
 #endif /* OCFS2_ACL_H */
diff --git a/fs/ocfs2/super.c b/fs/ocfs2/super.c
index 960673004df..da7d33a57cf 100644
--- a/fs/ocfs2/super.c
+++ b/fs/ocfs2/super.c
@@ -1413,19 +1413,12 @@ static int ocfs2_parse_options(struct super_block *sb,
 			}
 			mopt->mount_opt |= OCFS2_MOUNT_GRPQUOTA;
 			break;
-#ifdef CONFIG_OCFS2_FS_POSIX_ACL
 		case Opt_acl:
 			mopt->mount_opt |= OCFS2_MOUNT_POSIX_ACL;
 			break;
 		case Opt_noacl:
 			mopt->mount_opt &= ~OCFS2_MOUNT_POSIX_ACL;
 			break;
-#else
-		case Opt_acl:
-		case Opt_noacl:
-			printk(KERN_INFO "ocfs2 (no)acl options not supported\n");
-			break;
-#endif
 		default:
 			mlog(ML_ERROR,
 			     "Unrecognized mount option \"%s\" "
@@ -1502,12 +1495,10 @@ static int ocfs2_show_options(struct seq_file *s, struct vfsmount *mnt)
 	if (opts & OCFS2_MOUNT_INODE64)
 		seq_printf(s, ",inode64");
 
-#ifdef CONFIG_OCFS2_FS_POSIX_ACL
 	if (opts & OCFS2_MOUNT_POSIX_ACL)
 		seq_printf(s, ",acl");
 	else
 		seq_printf(s, ",noacl");
-#endif
 
 	return 0;
 }
diff --git a/fs/ocfs2/xattr.c b/fs/ocfs2/xattr.c
index fe3419068df..923e950e544 100644
--- a/fs/ocfs2/xattr.c
+++ b/fs/ocfs2/xattr.c
@@ -98,10 +98,8 @@ static struct ocfs2_xattr_def_value_root def_xv = {
 
 struct xattr_handler *ocfs2_xattr_handlers[] = {
 	&ocfs2_xattr_user_handler,
-#ifdef CONFIG_OCFS2_FS_POSIX_ACL
 	&ocfs2_xattr_acl_access_handler,
 	&ocfs2_xattr_acl_default_handler,
-#endif
 	&ocfs2_xattr_trusted_handler,
 	&ocfs2_xattr_security_handler,
 	NULL
@@ -109,12 +107,10 @@ struct xattr_handler *ocfs2_xattr_handlers[] = {
 
 static struct xattr_handler *ocfs2_xattr_handler_map[OCFS2_XATTR_MAX] = {
 	[OCFS2_XATTR_INDEX_USER]	= &ocfs2_xattr_user_handler,
-#ifdef CONFIG_OCFS2_FS_POSIX_ACL
 	[OCFS2_XATTR_INDEX_POSIX_ACL_ACCESS]
 					= &ocfs2_xattr_acl_access_handler,
 	[OCFS2_XATTR_INDEX_POSIX_ACL_DEFAULT]
 					= &ocfs2_xattr_acl_default_handler,
-#endif
 	[OCFS2_XATTR_INDEX_TRUSTED]	= &ocfs2_xattr_trusted_handler,
 	[OCFS2_XATTR_INDEX_SECURITY]	= &ocfs2_xattr_security_handler,
 };
diff --git a/fs/ocfs2/xattr.h b/fs/ocfs2/xattr.h
index 08e36389f56..abd72a47f52 100644
--- a/fs/ocfs2/xattr.h
+++ b/fs/ocfs2/xattr.h
@@ -40,10 +40,8 @@ struct ocfs2_security_xattr_info {
 extern struct xattr_handler ocfs2_xattr_user_handler;
 extern struct xattr_handler ocfs2_xattr_trusted_handler;
 extern struct xattr_handler ocfs2_xattr_security_handler;
-#ifdef CONFIG_OCFS2_FS_POSIX_ACL
 extern struct xattr_handler ocfs2_xattr_acl_access_handler;
 extern struct xattr_handler ocfs2_xattr_acl_default_handler;
-#endif
 extern struct xattr_handler *ocfs2_xattr_handlers[];
 
 ssize_t ocfs2_listxattr(struct dentry *, char *, size_t);
-- 
cgit v1.2.3


From 5297aad80cd3a3e62116e0a0b29116ccd8ae397d Mon Sep 17 00:00:00 2001
From: Jan Kara <jack@suse.cz>
Date: Thu, 15 Oct 2009 14:54:04 +0200
Subject: ocfs2: Make acl use the default

Change acl mount options handling to match the one of XFS and BTRFS and
hopefully it is also easier to use now. When admin does not specify any
acl mount option, acls are enabled if and only if the filesystem has
xattr feature enabled. If admin specifies 'acl' mount option, we fail
the mount if the filesystem does not have xattr feature and thus acls
cannot be enabled.

Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
---
 fs/ocfs2/ocfs2.h |  8 +++---
 fs/ocfs2/super.c | 82 ++++++++++++++++++++++++++++++--------------------------
 2 files changed, 49 insertions(+), 41 deletions(-)

(limited to 'fs/ocfs2')

diff --git a/fs/ocfs2/ocfs2.h b/fs/ocfs2/ocfs2.h
index eae40460242..35ad46cf89b 100644
--- a/fs/ocfs2/ocfs2.h
+++ b/fs/ocfs2/ocfs2.h
@@ -250,9 +250,11 @@ enum ocfs2_mount_options
 	OCFS2_MOUNT_LOCALFLOCKS = 1 << 5, /* No cluster aware user file locks */
 	OCFS2_MOUNT_NOUSERXATTR = 1 << 6, /* No user xattr */
 	OCFS2_MOUNT_INODE64 = 1 << 7,	/* Allow inode numbers > 2^32 */
-	OCFS2_MOUNT_POSIX_ACL = 1 << 8,	/* POSIX access control lists */
-	OCFS2_MOUNT_USRQUOTA = 1 << 9, /* We support user quotas */
-	OCFS2_MOUNT_GRPQUOTA = 1 << 10, /* We support group quotas */
+	OCFS2_MOUNT_POSIX_ACL = 1 << 8,	/* Force POSIX access control lists */
+	OCFS2_MOUNT_NO_POSIX_ACL = 1 << 9,	/* Disable POSIX access
+						   control lists */
+	OCFS2_MOUNT_USRQUOTA = 1 << 10, /* We support user quotas */
+	OCFS2_MOUNT_GRPQUOTA = 1 << 11, /* We support group quotas */
 };
 
 #define OCFS2_OSB_SOFT_RO			0x0001
diff --git a/fs/ocfs2/super.c b/fs/ocfs2/super.c
index da7d33a57cf..a5116b9b7e7 100644
--- a/fs/ocfs2/super.c
+++ b/fs/ocfs2/super.c
@@ -100,6 +100,8 @@ struct mount_options
 static int ocfs2_parse_options(struct super_block *sb, char *options,
 			       struct mount_options *mopt,
 			       int is_remount);
+static int ocfs2_check_set_options(struct super_block *sb,
+				   struct mount_options *options);
 static int ocfs2_show_options(struct seq_file *s, struct vfsmount *mnt);
 static void ocfs2_put_super(struct super_block *sb);
 static int ocfs2_mount_volume(struct super_block *sb);
@@ -600,7 +602,8 @@ static int ocfs2_remount(struct super_block *sb, int *flags, char *data)
 
 	lock_kernel();
 
-	if (!ocfs2_parse_options(sb, data, &parsed_options, 1)) {
+	if (!ocfs2_parse_options(sb, data, &parsed_options, 1) ||
+	    !ocfs2_check_set_options(sb, &parsed_options)) {
 		ret = -EINVAL;
 		goto out;
 	}
@@ -691,8 +694,6 @@ unlock_osb:
 	if (!ret) {
 		/* Only save off the new mount options in case of a successful
 		 * remount. */
-		if (!(osb->s_feature_incompat & OCFS2_FEATURE_INCOMPAT_XATTR))
-			parsed_options.mount_opt &= ~OCFS2_MOUNT_POSIX_ACL;
 		osb->s_mount_opt = parsed_options.mount_opt;
 		osb->s_atime_quantum = parsed_options.atime_quantum;
 		osb->preferred_slot = parsed_options.slot;
@@ -1011,31 +1012,16 @@ static int ocfs2_fill_super(struct super_block *sb, void *data, int silent)
 	brelse(bh);
 	bh = NULL;
 
-	if (!(osb->s_feature_incompat & OCFS2_FEATURE_INCOMPAT_XATTR))
-		parsed_options.mount_opt &= ~OCFS2_MOUNT_POSIX_ACL;
-
+	if (!ocfs2_check_set_options(sb, &parsed_options)) {
+		status = -EINVAL;
+		goto read_super_error;
+	}
 	osb->s_mount_opt = parsed_options.mount_opt;
 	osb->s_atime_quantum = parsed_options.atime_quantum;
 	osb->preferred_slot = parsed_options.slot;
 	osb->osb_commit_interval = parsed_options.commit_interval;
 	osb->local_alloc_default_bits = ocfs2_megabytes_to_clusters(sb, parsed_options.localalloc_opt);
 	osb->local_alloc_bits = osb->local_alloc_default_bits;
-	if (osb->s_mount_opt & OCFS2_MOUNT_USRQUOTA &&
-	    !OCFS2_HAS_RO_COMPAT_FEATURE(sb,
-					 OCFS2_FEATURE_RO_COMPAT_USRQUOTA)) {
-		status = -EINVAL;
-		mlog(ML_ERROR, "User quotas were requested, but this "
-		     "filesystem does not have the feature enabled.\n");
-		goto read_super_error;
-	}
-	if (osb->s_mount_opt & OCFS2_MOUNT_GRPQUOTA &&
-	    !OCFS2_HAS_RO_COMPAT_FEATURE(sb,
-					 OCFS2_FEATURE_RO_COMPAT_GRPQUOTA)) {
-		status = -EINVAL;
-		mlog(ML_ERROR, "Group quotas were requested, but this "
-		     "filesystem does not have the feature enabled.\n");
-		goto read_super_error;
-	}
 
 	status = ocfs2_verify_userspace_stack(osb, &parsed_options);
 	if (status)
@@ -1245,6 +1231,40 @@ static struct file_system_type ocfs2_fs_type = {
 	.next           = NULL
 };
 
+static int ocfs2_check_set_options(struct super_block *sb,
+				   struct mount_options *options)
+{
+	if (options->mount_opt & OCFS2_MOUNT_USRQUOTA &&
+	    !OCFS2_HAS_RO_COMPAT_FEATURE(sb,
+					 OCFS2_FEATURE_RO_COMPAT_USRQUOTA)) {
+		mlog(ML_ERROR, "User quotas were requested, but this "
+		     "filesystem does not have the feature enabled.\n");
+		return 0;
+	}
+	if (options->mount_opt & OCFS2_MOUNT_GRPQUOTA &&
+	    !OCFS2_HAS_RO_COMPAT_FEATURE(sb,
+					 OCFS2_FEATURE_RO_COMPAT_GRPQUOTA)) {
+		mlog(ML_ERROR, "Group quotas were requested, but this "
+		     "filesystem does not have the feature enabled.\n");
+		return 0;
+	}
+	if (options->mount_opt & OCFS2_MOUNT_POSIX_ACL &&
+	    !OCFS2_HAS_INCOMPAT_FEATURE(sb, OCFS2_FEATURE_INCOMPAT_XATTR)) {
+		mlog(ML_ERROR, "ACL support requested but extended attributes "
+		     "feature is not enabled\n");
+		return 0;
+	}
+	/* No ACL setting specified? Use XATTR feature... */
+	if (!(options->mount_opt & (OCFS2_MOUNT_POSIX_ACL |
+				    OCFS2_MOUNT_NO_POSIX_ACL))) {
+		if (OCFS2_HAS_INCOMPAT_FEATURE(sb, OCFS2_FEATURE_INCOMPAT_XATTR))
+			options->mount_opt |= OCFS2_MOUNT_POSIX_ACL;
+		else
+			options->mount_opt |= OCFS2_MOUNT_NO_POSIX_ACL;
+	}
+	return 1;
+}
+
 static int ocfs2_parse_options(struct super_block *sb,
 			       char *options,
 			       struct mount_options *mopt,
@@ -1392,31 +1412,17 @@ static int ocfs2_parse_options(struct super_block *sb,
 			mopt->mount_opt |= OCFS2_MOUNT_INODE64;
 			break;
 		case Opt_usrquota:
-			/* We check only on remount, otherwise features
-			 * aren't yet initialized. */
-			if (is_remount && !OCFS2_HAS_RO_COMPAT_FEATURE(sb,
-			    OCFS2_FEATURE_RO_COMPAT_USRQUOTA)) {
-				mlog(ML_ERROR, "User quota requested but "
-				     "filesystem feature is not set\n");
-				status = 0;
-				goto bail;
-			}
 			mopt->mount_opt |= OCFS2_MOUNT_USRQUOTA;
 			break;
 		case Opt_grpquota:
-			if (is_remount && !OCFS2_HAS_RO_COMPAT_FEATURE(sb,
-			    OCFS2_FEATURE_RO_COMPAT_GRPQUOTA)) {
-				mlog(ML_ERROR, "Group quota requested but "
-				     "filesystem feature is not set\n");
-				status = 0;
-				goto bail;
-			}
 			mopt->mount_opt |= OCFS2_MOUNT_GRPQUOTA;
 			break;
 		case Opt_acl:
 			mopt->mount_opt |= OCFS2_MOUNT_POSIX_ACL;
+			mopt->mount_opt &= ~OCFS2_MOUNT_NO_POSIX_ACL;
 			break;
 		case Opt_noacl:
+			mopt->mount_opt |= OCFS2_MOUNT_NO_POSIX_ACL;
 			mopt->mount_opt &= ~OCFS2_MOUNT_POSIX_ACL;
 			break;
 		default:
-- 
cgit v1.2.3


From 57b09bb5e492c37c1e4273fe4e435ffd1d2ddbe0 Mon Sep 17 00:00:00 2001
From: Jan Kara <jack@suse.cz>
Date: Thu, 15 Oct 2009 14:54:05 +0200
Subject: ocfs2: Set MS_POSIXACL on remount

We have to set MS_POSIXACL on remount as well. Otherwise VFS
would not know we started supporting ACLs after remount and
thus ACLs would not work.

Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
---
 fs/ocfs2/super.c | 4 ++++
 1 file changed, 4 insertions(+)

(limited to 'fs/ocfs2')

diff --git a/fs/ocfs2/super.c b/fs/ocfs2/super.c
index a5116b9b7e7..45d654cb77e 100644
--- a/fs/ocfs2/super.c
+++ b/fs/ocfs2/super.c
@@ -702,6 +702,10 @@ unlock_osb:
 
 		if (!ocfs2_is_hard_readonly(osb))
 			ocfs2_set_journal_params(osb);
+
+		sb->s_flags = (sb->s_flags & ~MS_POSIXACL) |
+			((osb->s_mount_opt & OCFS2_MOUNT_POSIX_ACL) ?
+							MS_POSIXACL : 0);
 	}
 out:
 	unlock_kernel();
-- 
cgit v1.2.3


From 38a04e432768ec0b016f3c687b4de31ac111ae59 Mon Sep 17 00:00:00 2001
From: Tao Ma <tao.ma@oracle.com>
Date: Mon, 30 Nov 2009 14:32:19 +0800
Subject: ocfs2: Find proper end cpos for a leaf refcount block.

ocfs2 refcount tree is stored as an extent tree while
the leaf ocfs2_refcount_rec points to a refcount block.

The following step can trip a kernel panic.
mkfs.ocfs2 -b 512 -C 1M --fs-features=refcount $DEVICE
mount -t ocfs2 $DEVICE $MNT_DIR
FILE_NAME=$RANDOM
FILE_NAME_1=$RANDOM
FILE_REF="${FILE_NAME}_ref"
FILE_REF_1="${FILE_NAME}_ref_1"
for((i=0;i<305;i++))
do
# /mnt/1048576 is a file with 1048576 sizes.
cat /mnt/1048576 >> $MNT_DIR/$FILE_NAME
cat /mnt/1048576 >> $MNT_DIR/$FILE_NAME_1
done
for((i=0;i<3;i++))
do
cat /mnt/1048576 >> $MNT_DIR/$FILE_NAME
done

for((i=0;i<2;i++))
do
cat /mnt/1048576 >> $MNT_DIR/$FILE_NAME
cat /mnt/1048576 >> $MNT_DIR/$FILE_NAME_1
done

cat /mnt/1048576 >> $MNT_DIR/$FILE_NAME

for((i=0;i<11;i++))
do
cat /mnt/1048576 >> $MNT_DIR/$FILE_NAME
cat /mnt/1048576 >> $MNT_DIR/$FILE_NAME_1
done
reflink $MNT_DIR/$FILE_NAME $MNT_DIR/$FILE_REF
# write_f is a program which will write some bytes to a file at offset.
# write_f -f file_name -l offset -w write_bytes.
./write_f -f $MNT_DIR/$FILE_REF -l $[310*1048576] -w 4096
./write_f -f $MNT_DIR/$FILE_REF -l $[306*1048576] -w 4096
./write_f -f $MNT_DIR/$FILE_REF -l $[311*1048576] -w 4096
./write_f -f $MNT_DIR/$FILE_NAME -l $[310*1048576] -w 4096
./write_f -f $MNT_DIR/$FILE_NAME -l $[311*1048576] -w 4096
reflink $MNT_DIR/$FILE_NAME $MNT_DIR/$FILE_REF_1
./write_f -f $MNT_DIR/$FILE_NAME -l $[311*1048576] -w 4096
#kernel panic here.

The reason is that if the ocfs2_extent_rec is the last record
in a leaf extent block, the old solution fails to find the
suitable end cpos. So this patch try to walk through the b-tree,
find the next sub root and get the c_pos the next sub-tree starts
from.

btw, I have runned tristan's test case against the patched kernel
for several days and this type of kernel panic never happens again.

Signed-off-by: Tao Ma <tao.ma@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
---
 fs/ocfs2/alloc.c        |  10 ++---
 fs/ocfs2/alloc.h        |   5 +++
 fs/ocfs2/refcounttree.c | 117 ++++++++++++++++++++++++++++++++++++++++++++----
 3 files changed, 119 insertions(+), 13 deletions(-)

(limited to 'fs/ocfs2')

diff --git a/fs/ocfs2/alloc.c b/fs/ocfs2/alloc.c
index 38a42f5d59f..5661db139ca 100644
--- a/fs/ocfs2/alloc.c
+++ b/fs/ocfs2/alloc.c
@@ -1765,9 +1765,9 @@ set_and_inc:
  *
  * The array index of the subtree root is passed back.
  */
-static int ocfs2_find_subtree_root(struct ocfs2_extent_tree *et,
-				   struct ocfs2_path *left,
-				   struct ocfs2_path *right)
+int ocfs2_find_subtree_root(struct ocfs2_extent_tree *et,
+			    struct ocfs2_path *left,
+			    struct ocfs2_path *right)
 {
 	int i = 0;
 
@@ -2872,8 +2872,8 @@ out:
  * This looks similar, but is subtly different to
  * ocfs2_find_cpos_for_left_leaf().
  */
-static int ocfs2_find_cpos_for_right_leaf(struct super_block *sb,
-					  struct ocfs2_path *path, u32 *cpos)
+int ocfs2_find_cpos_for_right_leaf(struct super_block *sb,
+				   struct ocfs2_path *path, u32 *cpos)
 {
 	int i, j, ret = 0;
 	u64 blkno;
diff --git a/fs/ocfs2/alloc.h b/fs/ocfs2/alloc.h
index 9c122d57446..1db4359ccb9 100644
--- a/fs/ocfs2/alloc.h
+++ b/fs/ocfs2/alloc.h
@@ -317,4 +317,9 @@ int ocfs2_path_bh_journal_access(handle_t *handle,
 int ocfs2_journal_access_path(struct ocfs2_caching_info *ci,
 			      handle_t *handle,
 			      struct ocfs2_path *path);
+int ocfs2_find_cpos_for_right_leaf(struct super_block *sb,
+				   struct ocfs2_path *path, u32 *cpos);
+int ocfs2_find_subtree_root(struct ocfs2_extent_tree *et,
+			    struct ocfs2_path *left,
+			    struct ocfs2_path *right);
 #endif /* OCFS2_ALLOC_H */
diff --git a/fs/ocfs2/refcounttree.c b/fs/ocfs2/refcounttree.c
index 3a0df7a1b81..9c3956f24e5 100644
--- a/fs/ocfs2/refcounttree.c
+++ b/fs/ocfs2/refcounttree.c
@@ -968,6 +968,103 @@ out:
 	return 0;
 }
 
+/*
+ * Find the end range for a leaf refcount block indicated by
+ * el->l_recs[index].e_blkno.
+ */
+static int ocfs2_get_refcount_cpos_end(struct ocfs2_caching_info *ci,
+				       struct buffer_head *ref_root_bh,
+				       struct ocfs2_extent_block *eb,
+				       struct ocfs2_extent_list *el,
+				       int index,  u32 *cpos_end)
+{
+	int ret, i, subtree_root;
+	u32 cpos;
+	u64 blkno;
+	struct super_block *sb = ocfs2_metadata_cache_get_super(ci);
+	struct ocfs2_path *left_path = NULL, *right_path = NULL;
+	struct ocfs2_extent_tree et;
+	struct ocfs2_extent_list *tmp_el;
+
+	if (index < le16_to_cpu(el->l_next_free_rec) - 1) {
+		/*
+		 * We have a extent rec after index, so just use the e_cpos
+		 * of the next extent rec.
+		 */
+		*cpos_end = le32_to_cpu(el->l_recs[index+1].e_cpos);
+		return 0;
+	}
+
+	if (!eb || (eb && !eb->h_next_leaf_blk)) {
+		/*
+		 * We are the last extent rec, so any high cpos should
+		 * be stored in this leaf refcount block.
+		 */
+		*cpos_end = UINT_MAX;
+		return 0;
+	}
+
+	/*
+	 * If the extent block isn't the last one, we have to find
+	 * the subtree root between this extent block and the next
+	 * leaf extent block and get the corresponding e_cpos from
+	 * the subroot. Otherwise we may corrupt the b-tree.
+	 */
+	ocfs2_init_refcount_extent_tree(&et, ci, ref_root_bh);
+
+	left_path = ocfs2_new_path_from_et(&et);
+	if (!left_path) {
+		ret = -ENOMEM;
+		mlog_errno(ret);
+		goto out;
+	}
+
+	cpos = le32_to_cpu(eb->h_list.l_recs[index].e_cpos);
+	ret = ocfs2_find_path(ci, left_path, cpos);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	right_path = ocfs2_new_path_from_path(left_path);
+	if (!right_path) {
+		ret = -ENOMEM;
+		mlog_errno(ret);
+		goto out;
+	}
+
+	ret = ocfs2_find_cpos_for_right_leaf(sb, left_path, &cpos);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	ret = ocfs2_find_path(ci, right_path, cpos);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	subtree_root = ocfs2_find_subtree_root(&et, left_path,
+					       right_path);
+
+	tmp_el = left_path->p_node[subtree_root].el;
+	blkno = left_path->p_node[subtree_root+1].bh->b_blocknr;
+	for (i = 0; i < le32_to_cpu(tmp_el->l_next_free_rec); i++) {
+		if (le64_to_cpu(tmp_el->l_recs[i].e_blkno) == blkno) {
+			*cpos_end = le32_to_cpu(tmp_el->l_recs[i+1].e_cpos);
+			break;
+		}
+	}
+
+	BUG_ON(i == le32_to_cpu(tmp_el->l_next_free_rec));
+
+out:
+	ocfs2_free_path(left_path);
+	ocfs2_free_path(right_path);
+	return ret;
+}
+
 /*
  * Given a cpos and len, try to find the refcount record which contains cpos.
  * 1. If cpos can be found in one refcount record, return the record.
@@ -983,10 +1080,10 @@ static int ocfs2_get_refcount_rec(struct ocfs2_caching_info *ci,
 				  struct buffer_head **ret_bh)
 {
 	int ret = 0, i, found;
-	u32 low_cpos;
+	u32 low_cpos, uninitialized_var(cpos_end);
 	struct ocfs2_extent_list *el;
-	struct ocfs2_extent_rec *tmp, *rec = NULL;
-	struct ocfs2_extent_block *eb;
+	struct ocfs2_extent_rec *rec = NULL;
+	struct ocfs2_extent_block *eb = NULL;
 	struct buffer_head *eb_bh = NULL, *ref_leaf_bh = NULL;
 	struct super_block *sb = ocfs2_metadata_cache_get_super(ci);
 	struct ocfs2_refcount_block *rb =
@@ -1034,12 +1131,16 @@ static int ocfs2_get_refcount_rec(struct ocfs2_caching_info *ci,
 		}
 	}
 
-	/* adjust len when we have ocfs2_extent_rec after it. */
-	if (found && i < le16_to_cpu(el->l_next_free_rec) - 1) {
-		tmp = &el->l_recs[i+1];
+	if (found) {
+		ret = ocfs2_get_refcount_cpos_end(ci, ref_root_bh,
+						  eb, el, i, &cpos_end);
+		if (ret) {
+			mlog_errno(ret);
+			goto out;
+		}
 
-		if (le32_to_cpu(tmp->e_cpos) < cpos + len)
-			len = le32_to_cpu(tmp->e_cpos) - cpos;
+		if (cpos_end < low_cpos + len)
+			len = cpos_end - low_cpos;
 	}
 
 	ret = ocfs2_read_refcount_block(ci, le64_to_cpu(rec->e_blkno),
-- 
cgit v1.2.3


From 12d4cec988a8a20927a9795dbf2c6d9ef2f85baa Mon Sep 17 00:00:00 2001
From: Tao Ma <tao.ma@oracle.com>
Date: Mon, 30 Nov 2009 15:08:40 +0800
Subject: ocfs2: refcounttree.c cleanup.

sparse check finds some endian problem and some other minor issues.
There is an obsolete function which should be removed.
So this patch resolve all these.

Signed-off-by: Tao Ma <tao.ma@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
---
 fs/ocfs2/refcounttree.c | 33 ++++++++-------------------------
 1 file changed, 8 insertions(+), 25 deletions(-)

(limited to 'fs/ocfs2')

diff --git a/fs/ocfs2/refcounttree.c b/fs/ocfs2/refcounttree.c
index 9c3956f24e5..c7ffba86e9c 100644
--- a/fs/ocfs2/refcounttree.c
+++ b/fs/ocfs2/refcounttree.c
@@ -276,7 +276,7 @@ static void ocfs2_erase_refcount_tree_from_list(struct ocfs2_super *osb,
 	spin_unlock(&osb->osb_lock);
 }
 
-void ocfs2_kref_remove_refcount_tree(struct kref *kref)
+static void ocfs2_kref_remove_refcount_tree(struct kref *kref)
 {
 	struct ocfs2_refcount_tree *tree =
 		container_of(kref, struct ocfs2_refcount_tree, rf_getcnt);
@@ -524,23 +524,6 @@ out:
 	return ret;
 }
 
-int ocfs2_lock_refcount_tree_by_inode(struct inode *inode, int rw,
-				      struct ocfs2_refcount_tree **ret_tree,
-				      struct buffer_head **ref_bh)
-{
-	int ret;
-	u64 ref_blkno;
-
-	ret = ocfs2_get_refcount_block(inode, &ref_blkno);
-	if (ret) {
-		mlog_errno(ret);
-		return ret;
-	}
-
-	return ocfs2_lock_refcount_tree(OCFS2_SB(inode->i_sb), ref_blkno,
-					rw, ret_tree, ref_bh);
-}
-
 void ocfs2_unlock_refcount_tree(struct ocfs2_super *osb,
 				struct ocfs2_refcount_tree *tree, int rw)
 {
@@ -1519,7 +1502,7 @@ static int ocfs2_divide_leaf_refcount_block(struct buffer_head *ref_leaf_bh,
 
 	/* change old and new rl_used accordingly. */
 	le16_add_cpu(&rl->rl_used, -num_moved);
-	new_rl->rl_used = cpu_to_le32(num_moved);
+	new_rl->rl_used = cpu_to_le16(num_moved);
 
 	sort(&rl->rl_recs, le16_to_cpu(rl->rl_used),
 	     sizeof(struct ocfs2_refcount_rec),
@@ -1898,7 +1881,8 @@ static int ocfs2_split_refcount_rec(handle_t *handle,
 		recs_need++;
 
 	/* If the leaf block don't have enough record, expand it. */
-	if (le16_to_cpu(rf_list->rl_used) + recs_need > rf_list->rl_count) {
+	if (le16_to_cpu(rf_list->rl_used) + recs_need >
+					 le16_to_cpu(rf_list->rl_count)) {
 		struct ocfs2_refcount_rec tmp_rec;
 		u64 cpos = le64_to_cpu(orig_rec->r_cpos);
 		len = le32_to_cpu(orig_rec->r_clusters);
@@ -1960,7 +1944,7 @@ static int ocfs2_split_refcount_rec(handle_t *handle,
 		memcpy(tail_rec, orig_rec, sizeof(struct ocfs2_refcount_rec));
 		le64_add_cpu(&tail_rec->r_cpos,
 			     le32_to_cpu(tail_rec->r_clusters) - len);
-		tail_rec->r_clusters = le32_to_cpu(len);
+		tail_rec->r_clusters = cpu_to_le32(len);
 	}
 
 	/*
@@ -3941,8 +3925,7 @@ static int ocfs2_add_refcounted_extent(struct inode *inode,
 	}
 
 	ret = ocfs2_insert_extent(handle, et, cpos,
-			cpu_to_le64(ocfs2_clusters_to_blocks(inode->i_sb,
-							     p_cluster)),
+			ocfs2_clusters_to_blocks(inode->i_sb, p_cluster),
 			num_clusters, ext_flags, meta_ac);
 	if (ret) {
 		mlog_errno(ret);
@@ -4354,8 +4337,8 @@ static int ocfs2_user_path_parent(const char __user *path,
  * @new_dentry:        target dentry
  * @preserve:  if true, preserve all file attributes
  */
-int ocfs2_vfs_reflink(struct dentry *old_dentry, struct inode *dir,
-		      struct dentry *new_dentry, bool preserve)
+static int ocfs2_vfs_reflink(struct dentry *old_dentry, struct inode *dir,
+			     struct dentry *new_dentry, bool preserve)
 {
 	struct inode *inode = old_dentry->d_inode;
 	int error;
-- 
cgit v1.2.3


From f6656d26d17b2598f43cd41be088853fa2a03397 Mon Sep 17 00:00:00 2001
From: Sunil Mushran <sunil.mushran@oracle.com>
Date: Tue, 17 Nov 2009 16:29:19 -0800
Subject: ocfs2/cluster: Make fence method configurable - v2

By default, o2cb fences the box by calling emergency_restart(). While this
scheme works well in production, it comes in the way during testing as it
does not let the tester take stack/core dumps for analysis.

This patch allows user to dynamically change the fence method to panic() by:
# echo "panic" > /sys/kernel/config/cluster/<clustername>/fence_method

Signed-off-by: Sunil Mushran <sunil.mushran@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
---
 fs/ocfs2/cluster/nodemanager.c | 51 ++++++++++++++++++++++++++++++++++++++++++
 fs/ocfs2/cluster/nodemanager.h |  7 ++++++
 fs/ocfs2/cluster/quorum.c      | 16 +++++++++++--
 3 files changed, 72 insertions(+), 2 deletions(-)

(limited to 'fs/ocfs2')

diff --git a/fs/ocfs2/cluster/nodemanager.c b/fs/ocfs2/cluster/nodemanager.c
index 7ee6188bc79..c81142e3ef8 100644
--- a/fs/ocfs2/cluster/nodemanager.c
+++ b/fs/ocfs2/cluster/nodemanager.c
@@ -35,6 +35,10 @@
  * cluster references throughout where nodes are looked up */
 struct o2nm_cluster *o2nm_single_cluster = NULL;
 
+char *o2nm_fence_method_desc[O2NM_FENCE_METHODS] = {
+		"reset",	/* O2NM_FENCE_RESET */
+		"panic",	/* O2NM_FENCE_PANIC */
+};
 
 struct o2nm_node *o2nm_get_node_by_num(u8 node_num)
 {
@@ -579,6 +583,43 @@ static ssize_t o2nm_cluster_attr_reconnect_delay_ms_write(
 	return o2nm_cluster_attr_write(page, count,
 	                               &cluster->cl_reconnect_delay_ms);
 }
+
+static ssize_t o2nm_cluster_attr_fence_method_read(
+	struct o2nm_cluster *cluster, char *page)
+{
+	ssize_t ret = 0;
+
+	if (cluster)
+		ret = sprintf(page, "%s\n",
+			      o2nm_fence_method_desc[cluster->cl_fence_method]);
+	return ret;
+}
+
+static ssize_t o2nm_cluster_attr_fence_method_write(
+	struct o2nm_cluster *cluster, const char *page, size_t count)
+{
+	unsigned int i;
+
+	if (page[count - 1] != '\n')
+		goto bail;
+
+	for (i = 0; i < O2NM_FENCE_METHODS; ++i) {
+		if (count != strlen(o2nm_fence_method_desc[i]) + 1)
+			continue;
+		if (strncasecmp(page, o2nm_fence_method_desc[i], count - 1))
+			continue;
+		if (cluster->cl_fence_method != i) {
+			printk(KERN_INFO "ocfs2: Changing fence method to %s\n",
+			       o2nm_fence_method_desc[i]);
+			cluster->cl_fence_method = i;
+		}
+		return count;
+	}
+
+bail:
+	return -EINVAL;
+}
+
 static struct o2nm_cluster_attribute o2nm_cluster_attr_idle_timeout_ms = {
 	.attr	= { .ca_owner = THIS_MODULE,
 		    .ca_name = "idle_timeout_ms",
@@ -603,10 +644,19 @@ static struct o2nm_cluster_attribute o2nm_cluster_attr_reconnect_delay_ms = {
 	.store	= o2nm_cluster_attr_reconnect_delay_ms_write,
 };
 
+static struct o2nm_cluster_attribute o2nm_cluster_attr_fence_method = {
+	.attr	= { .ca_owner = THIS_MODULE,
+		    .ca_name = "fence_method",
+		    .ca_mode = S_IRUGO | S_IWUSR },
+	.show	= o2nm_cluster_attr_fence_method_read,
+	.store	= o2nm_cluster_attr_fence_method_write,
+};
+
 static struct configfs_attribute *o2nm_cluster_attrs[] = {
 	&o2nm_cluster_attr_idle_timeout_ms.attr,
 	&o2nm_cluster_attr_keepalive_delay_ms.attr,
 	&o2nm_cluster_attr_reconnect_delay_ms.attr,
+	&o2nm_cluster_attr_fence_method.attr,
 	NULL,
 };
 static ssize_t o2nm_cluster_show(struct config_item *item,
@@ -778,6 +828,7 @@ static struct config_group *o2nm_cluster_group_make_group(struct config_group *g
 	cluster->cl_reconnect_delay_ms = O2NET_RECONNECT_DELAY_MS_DEFAULT;
 	cluster->cl_idle_timeout_ms    = O2NET_IDLE_TIMEOUT_MS_DEFAULT;
 	cluster->cl_keepalive_delay_ms = O2NET_KEEPALIVE_DELAY_MS_DEFAULT;
+	cluster->cl_fence_method       = O2NM_FENCE_RESET;
 
 	ret = &cluster->cl_group;
 	o2nm_single_cluster = cluster;
diff --git a/fs/ocfs2/cluster/nodemanager.h b/fs/ocfs2/cluster/nodemanager.h
index c992ea0da4a..09ea2d388bb 100644
--- a/fs/ocfs2/cluster/nodemanager.h
+++ b/fs/ocfs2/cluster/nodemanager.h
@@ -33,6 +33,12 @@
 #include <linux/configfs.h>
 #include <linux/rbtree.h>
 
+enum o2nm_fence_method {
+	O2NM_FENCE_RESET	= 0,
+	O2NM_FENCE_PANIC,
+	O2NM_FENCE_METHODS,	/* Number of fence methods */
+};
+
 struct o2nm_node {
 	spinlock_t		nd_lock;
 	struct config_item	nd_item;
@@ -58,6 +64,7 @@ struct o2nm_cluster {
 	unsigned int		cl_idle_timeout_ms;
 	unsigned int		cl_keepalive_delay_ms;
 	unsigned int		cl_reconnect_delay_ms;
+	enum o2nm_fence_method	cl_fence_method;
 
 	/* this bitmap is part of a hack for disk bitmap.. will go eventually. - zab */
 	unsigned long	cl_nodes_bitmap[BITS_TO_LONGS(O2NM_MAX_NODES)];
diff --git a/fs/ocfs2/cluster/quorum.c b/fs/ocfs2/cluster/quorum.c
index bbacf7da48a..639024033fc 100644
--- a/fs/ocfs2/cluster/quorum.c
+++ b/fs/ocfs2/cluster/quorum.c
@@ -74,8 +74,20 @@ static void o2quo_fence_self(void)
 	 * threads can still schedule, etc, etc */
 	o2hb_stop_all_regions();
 
-	printk("ocfs2 is very sorry to be fencing this system by restarting\n");
-	emergency_restart();
+	switch (o2nm_single_cluster->cl_fence_method) {
+	case O2NM_FENCE_PANIC:
+		panic("*** ocfs2 is very sorry to be fencing this system by "
+		      "panicing ***\n");
+		break;
+	default:
+		WARN_ON(o2nm_single_cluster->cl_fence_method >=
+			O2NM_FENCE_METHODS);
+	case O2NM_FENCE_RESET:
+		printk(KERN_ERR "*** ocfs2 is very sorry to be fencing this "
+		       "system by restarting ***\n");
+		emergency_restart();
+		break;
+	};
 }
 
 /* Indicate that a timeout occured on a hearbeat region write. The
-- 
cgit v1.2.3


From aad1b15310b9bcd59fa81ab8f2b1513b59553ea8 Mon Sep 17 00:00:00 2001
From: Tiger Yang <tiger.yang@oracle.com>
Date: Thu, 19 Nov 2009 10:17:46 +0800
Subject: ocfs2: return -EAGAIN instead of EAGAIN in dlm

We used to return positive EAGAIN to indicate a retry action
is needed in dlm_begin_reco_handler(). Now we return negative
-EAGAIN to erase the confusion caused by this error code.

Signed-off-by: Tiger Yang <tiger.yang@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
---
 fs/ocfs2/dlm/dlmrecovery.c | 18 +++++++++---------
 1 file changed, 9 insertions(+), 9 deletions(-)

(limited to 'fs/ocfs2')

diff --git a/fs/ocfs2/dlm/dlmrecovery.c b/fs/ocfs2/dlm/dlmrecovery.c
index d9fa3d22e17..2f9e4e19a4f 100644
--- a/fs/ocfs2/dlm/dlmrecovery.c
+++ b/fs/ocfs2/dlm/dlmrecovery.c
@@ -2589,6 +2589,14 @@ retry:
 			     "begin reco msg (%d)\n", dlm->name, nodenum, ret);
 			ret = 0;
 		}
+		if (ret == -EAGAIN) {
+			mlog(0, "%s: trying to start recovery of node "
+			     "%u, but node %u is waiting for last recovery "
+			     "to complete, backoff for a bit\n", dlm->name,
+			     dead_node, nodenum);
+			msleep(100);
+			goto retry;
+		}
 		if (ret < 0) {
 			struct dlm_lock_resource *res;
 			/* this is now a serious problem, possibly ENOMEM 
@@ -2608,14 +2616,6 @@ retry:
 			 * another ENOMEM */
 			msleep(100);
 			goto retry;
-		} else if (ret == EAGAIN) {
-			mlog(0, "%s: trying to start recovery of node "
-			     "%u, but node %u is waiting for last recovery "
-			     "to complete, backoff for a bit\n", dlm->name,
-			     dead_node, nodenum);
-			/* TODO Look into replacing msleep with cond_resched() */
-			msleep(100);
-			goto retry;
 		}
 	}
 
@@ -2639,7 +2639,7 @@ int dlm_begin_reco_handler(struct o2net_msg *msg, u32 len, void *data,
 		     dlm->name, br->node_idx, br->dead_node,
 		     dlm->reco.dead_node, dlm->reco.new_master);
 		spin_unlock(&dlm->spinlock);
-		return EAGAIN;
+		return -EAGAIN;
 	}
 	spin_unlock(&dlm->spinlock);
 
-- 
cgit v1.2.3


From f5befbbbf9b7af033c94a1df6acaf79e77a3ceb6 Mon Sep 17 00:00:00 2001
From: Jeff Liu <jeff.liu@oracle.com>
Date: Tue, 15 Dec 2009 14:08:28 +0800
Subject: ocfs2-devel: remove redundant OCFS2_MOUNT_POSIX_ACL check in
 ocfs2_get_acl_nolock()

osb->s_mount_opt has already been checked against OCFS2_MOUNT_POSIX_ACL_CHECK before
calling ocfs2_get_acl_nolock() in ocfs2_init_acl() && ocfs2_get_acl(), so remove it.

Signed-off-by: Jeff Liu <jeff.liu@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
---
 fs/ocfs2/acl.c | 4 ----
 1 file changed, 4 deletions(-)

(limited to 'fs/ocfs2')

diff --git a/fs/ocfs2/acl.c b/fs/ocfs2/acl.c
index fbeaec76210..200b88109b2 100644
--- a/fs/ocfs2/acl.c
+++ b/fs/ocfs2/acl.c
@@ -98,15 +98,11 @@ static struct posix_acl *ocfs2_get_acl_nolock(struct inode *inode,
 					      int type,
 					      struct buffer_head *di_bh)
 {
-	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
 	int name_index;
 	char *value = NULL;
 	struct posix_acl *acl;
 	int retval;
 
-	if (!(osb->s_mount_opt & OCFS2_MOUNT_POSIX_ACL))
-		return NULL;
-
 	switch (type) {
 	case ACL_TYPE_ACCESS:
 		name_index = OCFS2_XATTR_INDEX_POSIX_ACL_ACCESS;
-- 
cgit v1.2.3


From 3a05d7961e6b5fb77660849b56a22feca5d1e0c5 Mon Sep 17 00:00:00 2001
From: Coly Li <coly.li@suse.de>
Date: Fri, 4 Dec 2009 02:02:35 +0800
Subject: ocfs2: explicit declare uninitialized var in user_cluster_connect()

This patch explicitly declares an uninitialized local variable in user_cluster_connect(), to remove a compiling warning.

Signed-off-by: Coly Li <coly.li@suse.de>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
---
 fs/ocfs2/stack_user.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'fs/ocfs2')

diff --git a/fs/ocfs2/stack_user.c b/fs/ocfs2/stack_user.c
index ff4c798a563..da78a2a334f 100644
--- a/fs/ocfs2/stack_user.c
+++ b/fs/ocfs2/stack_user.c
@@ -814,7 +814,7 @@ static int fs_protocol_compare(struct ocfs2_protocol_version *existing,
 static int user_cluster_connect(struct ocfs2_cluster_connection *conn)
 {
 	dlm_lockspace_t *fsdlm;
-	struct ocfs2_live_connection *control;
+	struct ocfs2_live_connection *uninitialized_var(control);
 	int rc = 0;
 
 	BUG_ON(conn == NULL);
-- 
cgit v1.2.3


From 936545401624217955df4dd44bb71615900e5397 Mon Sep 17 00:00:00 2001
From: Coly Li <coly.li@suse.de>
Date: Sun, 6 Dec 2009 22:38:53 +0800
Subject: ocfs2: replace u8 by __u8 in ocfs2_fs.h

This patch replaces date type 'u8' with '__u8', which follows the coding style of ocfs2_fs.h, and portable to user space
for ocfs2-tools.

Signed-off-by: Coly Li <coly.li@suse.de>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
---
 fs/ocfs2/ocfs2_fs.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'fs/ocfs2')

diff --git a/fs/ocfs2/ocfs2_fs.h b/fs/ocfs2/ocfs2_fs.h
index e9431e4a5e7..1a1a679e51b 100644
--- a/fs/ocfs2/ocfs2_fs.h
+++ b/fs/ocfs2/ocfs2_fs.h
@@ -1202,7 +1202,7 @@ struct ocfs2_local_disk_dqinfo {
 /* Header of one chunk of a quota file */
 struct ocfs2_local_disk_chunk {
 	__le32 dqc_free;	/* Number of free entries in the bitmap */
-	u8 dqc_bitmap[0];	/* Bitmap of entries in the corresponding
+	__u8 dqc_bitmap[0];	/* Bitmap of entries in the corresponding
 				 * chunk of quota file */
 };
 
-- 
cgit v1.2.3


From faf8b70f79edf56fedd531dfcd41f3e73c3e9696 Mon Sep 17 00:00:00 2001
From: Sunil Mushran <sunil.mushran@oracle.com>
Date: Thu, 3 Dec 2009 12:46:52 -0800
Subject: ocfs2: Use FIEMAP_EXTENT_SHARED

Adds FIEMAP_EXTENT_SHARED flag to refcounted extents.

Signed-off-by: Sunil Mushran <sunil.mushran@oracle.com>
Acked-by: Mark Fasheh <mfasheh@suse.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
---
 fs/ocfs2/extent_map.c | 2 ++
 1 file changed, 2 insertions(+)

(limited to 'fs/ocfs2')

diff --git a/fs/ocfs2/extent_map.c b/fs/ocfs2/extent_map.c
index 843db64e9d4..cdce5f8c1cf 100644
--- a/fs/ocfs2/extent_map.c
+++ b/fs/ocfs2/extent_map.c
@@ -786,6 +786,8 @@ int ocfs2_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
 		fe_flags = 0;
 		if (rec.e_flags & OCFS2_EXT_UNWRITTEN)
 			fe_flags |= FIEMAP_EXTENT_UNWRITTEN;
+		if (rec.e_flags & OCFS2_EXT_REFCOUNTED)
+			fe_flags |= FIEMAP_EXTENT_SHARED;
 		if (is_last)
 			fe_flags |= FIEMAP_EXTENT_LAST;
 		len_bytes = (u64)le16_to_cpu(rec.e_leaf_clusters) << osb->s_clustersize_bits;
-- 
cgit v1.2.3


From 55f4946ed23cbf78efe9f818c4b55e890fe2beff Mon Sep 17 00:00:00 2001
From: Tristan Ye <tristan.ye@oracle.com>
Date: Thu, 17 Dec 2009 18:42:16 +0800
Subject: Ocfs2: Should ocfs2 support fiemap for S_IFDIR inode?

Let userspace have a chance to get the extent info of a
directory just like extN did.

Signed-off-by: Tristan Ye <tristan.ye@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
---
 fs/ocfs2/namei.c | 1 +
 1 file changed, 1 insertion(+)

(limited to 'fs/ocfs2')

diff --git a/fs/ocfs2/namei.c b/fs/ocfs2/namei.c
index f010b22b1c4..0b9f35e09bf 100644
--- a/fs/ocfs2/namei.c
+++ b/fs/ocfs2/namei.c
@@ -2326,4 +2326,5 @@ const struct inode_operations ocfs2_dir_iops = {
 	.getxattr	= generic_getxattr,
 	.listxattr	= ocfs2_listxattr,
 	.removexattr	= generic_removexattr,
+	.fiemap         = ocfs2_fiemap,
 };
-- 
cgit v1.2.3


From c7d260afcbc0f9f4074e74796d4e3646875c08e1 Mon Sep 17 00:00:00 2001
From: Tao Ma <tao.ma@oracle.com>
Date: Fri, 18 Dec 2009 10:24:54 +0800
Subject: ocfs2: Add reflinked file's inode to inode hash eariler.

We used to add reflinked file's inode to inode hash when
we add it to the dest dir. But actually there is a race.
Consider the following sequence.
1. reflink happens and create the inode in orphan dir.
2. reflink thread is scheduled out because of some io.
3. recovery begins to work and calls ocfs2_recover_orphans.
   It calls ocfs2_iget and get a new inode and i_count = 1.
   It calls iput then and delete inode. the buffer's
   uptodate state is cleared.

This patch move insert_inode_hash to the create function so
that it can be found by step 3 and prevented from deleting
because i_count > 1.

This resolves the bug
http://oss.oracle.com/bugzilla/show_bug.cgi?id=1183.

Signed-off-by: Tao Ma <tao.ma@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
---
 fs/ocfs2/namei.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'fs/ocfs2')

diff --git a/fs/ocfs2/namei.c b/fs/ocfs2/namei.c
index f010b22b1c4..5ac4d52e090 100644
--- a/fs/ocfs2/namei.c
+++ b/fs/ocfs2/namei.c
@@ -2136,6 +2136,7 @@ int ocfs2_create_inode_in_orphan(struct inode *dir,
 	if (status < 0)
 		mlog_errno(status);
 
+	insert_inode_hash(inode);
 leave:
 	if (status < 0 && did_quota_inode)
 		vfs_dq_free_inode(inode);
@@ -2284,7 +2285,6 @@ int ocfs2_mv_orphaned_inode_to_new(struct inode *dir,
 		goto out_commit;
 	}
 
-	insert_inode_hash(inode);
 	dentry->d_op = &ocfs2_dentry_ops;
 	d_instantiate(dentry, inode);
 	status = 0;
-- 
cgit v1.2.3


From 10cf1a02f444fdcd50be47cce3fa8bf08251dd9c Mon Sep 17 00:00:00 2001
From: Tao Ma <tao.ma@oracle.com>
Date: Fri, 18 Dec 2009 10:24:55 +0800
Subject: ocfs2: Set i_nlink properly during reflink.

We create a file in orphan dir for reflink so that if there
is any error, we don't create any wrong dentry in the dir.
But actually the file in orphan dir should be i_nlink = 0
so that it can be replayed and freed successfully.

This patch first set i_nlink to 0 when creating the file in
orphan dir and then set it to 1(reflink now only works for
regular file) when we move it to the dest dir.

Signed-off-by: Tao Ma <tao.ma@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
---
 fs/ocfs2/namei.c | 3 +++
 1 file changed, 3 insertions(+)

(limited to 'fs/ocfs2')

diff --git a/fs/ocfs2/namei.c b/fs/ocfs2/namei.c
index 5ac4d52e090..3e9b46002f2 100644
--- a/fs/ocfs2/namei.c
+++ b/fs/ocfs2/namei.c
@@ -2108,6 +2108,7 @@ int ocfs2_create_inode_in_orphan(struct inode *dir,
 	}
 	did_quota_inode = 1;
 
+	inode->i_nlink = 0;
 	/* do the real work now. */
 	status = ocfs2_mknod_locked(osb, dir, inode,
 				    0, &new_di_bh, parent_di_bh, handle,
@@ -2268,6 +2269,8 @@ int ocfs2_mv_orphaned_inode_to_new(struct inode *dir,
 	di = (struct ocfs2_dinode *)di_bh->b_data;
 	le32_add_cpu(&di->i_flags, -OCFS2_ORPHANED_FL);
 	di->i_orphaned_slot = 0;
+	inode->i_nlink = 1;
+	ocfs2_set_links_count(di, inode->i_nlink);
 	ocfs2_journal_dirty(handle, di_bh);
 
 	status = ocfs2_add_entry(handle, dentry, inode,
-- 
cgit v1.2.3


From 86239d59e268a35a49a00fce9c8512a506267f5c Mon Sep 17 00:00:00 2001
From: Tristan Ye <tristan.ye@oracle.com>
Date: Tue, 22 Dec 2009 09:11:58 +0800
Subject: Ocfs2: Let ocfs2 support fiemap for symlink and fast symlink.

For fast symlink, it can be treated the same as inlined files since
the data extent we want to return of both case all were stored in
metadata block. For symlink, it can be simply treated the same as we
did for regular files.

Signed-off-by: Tristan Ye <tristan.ye@oracle.com>
Acked-by: Sunil Mushran <sunil.mushran@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
---
 fs/ocfs2/extent_map.c | 23 +++++++++++++++++++----
 fs/ocfs2/symlink.c    |  2 ++
 2 files changed, 21 insertions(+), 4 deletions(-)

(limited to 'fs/ocfs2')

diff --git a/fs/ocfs2/extent_map.c b/fs/ocfs2/extent_map.c
index cdce5f8c1cf..d35a27f4523 100644
--- a/fs/ocfs2/extent_map.c
+++ b/fs/ocfs2/extent_map.c
@@ -37,6 +37,7 @@
 #include "extent_map.h"
 #include "inode.h"
 #include "super.h"
+#include "symlink.h"
 
 #include "buffer_head_io.h"
 
@@ -703,6 +704,12 @@ out:
 	return ret;
 }
 
+/*
+ * The ocfs2_fiemap_inline() may be a little bit misleading, since
+ * it not only handles the fiemap for inlined files, but also deals
+ * with the fast symlink, cause they have no difference for extent
+ * mapping per se.
+ */
 static int ocfs2_fiemap_inline(struct inode *inode, struct buffer_head *di_bh,
 			       struct fiemap_extent_info *fieinfo,
 			       u64 map_start)
@@ -715,11 +722,18 @@ static int ocfs2_fiemap_inline(struct inode *inode, struct buffer_head *di_bh,
 	struct ocfs2_inode_info *oi = OCFS2_I(inode);
 
 	di = (struct ocfs2_dinode *)di_bh->b_data;
-	id_count = le16_to_cpu(di->id2.i_data.id_count);
+	if (ocfs2_inode_is_fast_symlink(inode))
+		id_count = ocfs2_fast_symlink_chars(inode->i_sb);
+	else
+		id_count = le16_to_cpu(di->id2.i_data.id_count);
 
 	if (map_start < id_count) {
 		phys = oi->ip_blkno << inode->i_sb->s_blocksize_bits;
-		phys += offsetof(struct ocfs2_dinode, id2.i_data.id_data);
+		if (ocfs2_inode_is_fast_symlink(inode))
+			phys += offsetof(struct ocfs2_dinode, id2.i_symlink);
+		else
+			phys += offsetof(struct ocfs2_dinode,
+					 id2.i_data.id_data);
 
 		ret = fiemap_fill_next_extent(fieinfo, 0, phys, id_count,
 					      flags);
@@ -756,9 +770,10 @@ int ocfs2_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
 	down_read(&OCFS2_I(inode)->ip_alloc_sem);
 
 	/*
-	 * Handle inline-data separately.
+	 * Handle inline-data and fast symlink separately.
 	 */
-	if (OCFS2_I(inode)->ip_dyn_features & OCFS2_INLINE_DATA_FL) {
+	if ((OCFS2_I(inode)->ip_dyn_features & OCFS2_INLINE_DATA_FL) ||
+	    ocfs2_inode_is_fast_symlink(inode)) {
 		ret = ocfs2_fiemap_inline(inode, di_bh, fieinfo, map_start);
 		goto out_unlock;
 	}
diff --git a/fs/ocfs2/symlink.c b/fs/ocfs2/symlink.c
index e3421030a69..49b133ccbf1 100644
--- a/fs/ocfs2/symlink.c
+++ b/fs/ocfs2/symlink.c
@@ -163,6 +163,7 @@ const struct inode_operations ocfs2_symlink_inode_operations = {
 	.getxattr	= generic_getxattr,
 	.listxattr	= ocfs2_listxattr,
 	.removexattr	= generic_removexattr,
+	.fiemap		= ocfs2_fiemap,
 };
 const struct inode_operations ocfs2_fast_symlink_inode_operations = {
 	.readlink	= ocfs2_readlink,
@@ -174,4 +175,5 @@ const struct inode_operations ocfs2_fast_symlink_inode_operations = {
 	.getxattr	= generic_getxattr,
 	.listxattr	= ocfs2_listxattr,
 	.removexattr	= generic_removexattr,
+	.fiemap		= ocfs2_fiemap,
 };
-- 
cgit v1.2.3


From b31d308ddcfb73349582a0196b5f65dad3cf5d17 Mon Sep 17 00:00:00 2001
From: Tao Ma <tao.ma@oracle.com>
Date: Tue, 22 Dec 2009 10:32:15 +0800
Subject: ocfs2/trivial: Use proper mask for 2 places in hearbeat.c

I just noticed today that there are 2 places of "mlog(0,...)"
in  fs/ocfs2/cluster/heartbeat.c, but actually have no default
mask prefix in that file.
So change them to mlog(ML_HEARTBEAT,...).

Signed-off-by: Tao Ma <tao.ma@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
---
 fs/ocfs2/cluster/heartbeat.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

(limited to 'fs/ocfs2')

diff --git a/fs/ocfs2/cluster/heartbeat.c b/fs/ocfs2/cluster/heartbeat.c
index c452d116b89..eda5b8bcddd 100644
--- a/fs/ocfs2/cluster/heartbeat.c
+++ b/fs/ocfs2/cluster/heartbeat.c
@@ -176,7 +176,8 @@ static void o2hb_write_timeout(struct work_struct *work)
 
 static void o2hb_arm_write_timeout(struct o2hb_region *reg)
 {
-	mlog(0, "Queue write timeout for %u ms\n", O2HB_MAX_WRITE_TIMEOUT_MS);
+	mlog(ML_HEARTBEAT, "Queue write timeout for %u ms\n",
+	     O2HB_MAX_WRITE_TIMEOUT_MS);
 
 	cancel_delayed_work(&reg->hr_write_timeout_work);
 	reg->hr_last_timeout_start = jiffies;
@@ -874,7 +875,8 @@ static int o2hb_thread(void *data)
 		do_gettimeofday(&after_hb);
 		elapsed_msec = o2hb_elapsed_msecs(&before_hb, &after_hb);
 
-		mlog(0, "start = %lu.%lu, end = %lu.%lu, msec = %u\n",
+		mlog(ML_HEARTBEAT,
+		     "start = %lu.%lu, end = %lu.%lu, msec = %u\n",
 		     before_hb.tv_sec, (unsigned long) before_hb.tv_usec,
 		     after_hb.tv_sec, (unsigned long) after_hb.tv_usec,
 		     elapsed_msec);
-- 
cgit v1.2.3


From 8ff6af881deccca4f88e03f2fdadb1aac42e9489 Mon Sep 17 00:00:00 2001
From: Tao Ma <tao.ma@oracle.com>
Date: Wed, 23 Dec 2009 14:31:15 +0800
Subject: ocfs2/trivial: Use le16_to_cpu for a disk value in xattr.c

In ocfs2_value_metas_in_xattr_header, we should Use
le16_to_cpu for ocfs2_extent_list.l_next_free_rec.

Signed-off-by: Tao Ma <tao.ma@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
---
 fs/ocfs2/xattr.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'fs/ocfs2')

diff --git a/fs/ocfs2/xattr.c b/fs/ocfs2/xattr.c
index 923e950e544..932c0773160 100644
--- a/fs/ocfs2/xattr.c
+++ b/fs/ocfs2/xattr.c
@@ -6062,7 +6062,7 @@ static int ocfs2_value_metas_in_xattr_header(struct super_block *sb,
 		 * to the extent block, so just calculate a maximum record num.
 		 */
 		if (!xv->xr_list.l_tree_depth)
-			*num_recs += xv->xr_list.l_next_free_rec;
+			*num_recs += le16_to_cpu(xv->xr_list.l_next_free_rec);
 		else
 			*num_recs += ocfs2_clusters_for_bytes(sb,
 							      XATTR_SIZE_MAX);
-- 
cgit v1.2.3


From 86470e98cc52532006f47a6cb96d3ebe1ba7ad97 Mon Sep 17 00:00:00 2001
From: Tao Ma <tao.ma@oracle.com>
Date: Thu, 3 Dec 2009 21:55:05 +0800
Subject: ocfs2: Handle O_DIRECT when writing to a refcounted cluster.

In case of writing to a refcounted cluster with O_DIRECT,
we need to fall back to buffer write. And when it is finished,
we need to flush the page and the journal as we did for other
O_DIRECT writes.

This patch fix oss bug 1191.
http://oss.oracle.com/bugzilla/show_bug.cgi?id=1191

Signed-off-by: Tao Ma <tao.ma@oracle.com>
Tested-by: Tristan Ye <tristan.ye@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
---
 fs/ocfs2/file.c | 21 +++++++++++++++------
 1 file changed, 15 insertions(+), 6 deletions(-)

(limited to 'fs/ocfs2')

diff --git a/fs/ocfs2/file.c b/fs/ocfs2/file.c
index 3d30a1c974a..06ccf6a86d3 100644
--- a/fs/ocfs2/file.c
+++ b/fs/ocfs2/file.c
@@ -1772,7 +1772,8 @@ static int ocfs2_prepare_inode_for_write(struct dentry *dentry,
 					 loff_t *ppos,
 					 size_t count,
 					 int appending,
-					 int *direct_io)
+					 int *direct_io,
+					 int *has_refcount)
 {
 	int ret = 0, meta_level = 0;
 	struct inode *inode = dentry->d_inode;
@@ -1833,6 +1834,8 @@ static int ocfs2_prepare_inode_for_write(struct dentry *dentry,
 							       saved_pos,
 							       count,
 							       &meta_level);
+			if (has_refcount)
+				*has_refcount = 1;
 		}
 
 		if (ret < 0) {
@@ -1856,6 +1859,10 @@ static int ocfs2_prepare_inode_for_write(struct dentry *dentry,
 			break;
 		}
 
+		if (has_refcount && *has_refcount == 1) {
+			*direct_io = 0;
+			break;
+		}
 		/*
 		 * Allowing concurrent direct writes means
 		 * i_size changes wouldn't be synchronized, so
@@ -1899,7 +1906,7 @@ static ssize_t ocfs2_file_aio_write(struct kiocb *iocb,
 				    loff_t pos)
 {
 	int ret, direct_io, appending, rw_level, have_alloc_sem  = 0;
-	int can_do_direct;
+	int can_do_direct, has_refcount = 0;
 	ssize_t written = 0;
 	size_t ocount;		/* original count */
 	size_t count;		/* after file limit checks */
@@ -1942,7 +1949,7 @@ relock:
 	can_do_direct = direct_io;
 	ret = ocfs2_prepare_inode_for_write(file->f_path.dentry, ppos,
 					    iocb->ki_left, appending,
-					    &can_do_direct);
+					    &can_do_direct, &has_refcount);
 	if (ret < 0) {
 		mlog_errno(ret);
 		goto out;
@@ -2006,14 +2013,16 @@ out_dio:
 	/* buffered aio wouldn't have proper lock coverage today */
 	BUG_ON(ret == -EIOCBQUEUED && !(file->f_flags & O_DIRECT));
 
-	if ((file->f_flags & O_DSYNC && !direct_io) || IS_SYNC(inode)) {
+	if ((file->f_flags & O_DSYNC && !direct_io) || IS_SYNC(inode) ||
+	    (file->f_flags & O_DIRECT && has_refcount)) {
 		ret = filemap_fdatawrite_range(file->f_mapping, pos,
 					       pos + count - 1);
 		if (ret < 0)
 			written = ret;
 
 		if (!ret && (old_size != i_size_read(inode) ||
-		    old_clusters != OCFS2_I(inode)->ip_clusters)) {
+		    old_clusters != OCFS2_I(inode)->ip_clusters ||
+		    has_refcount)) {
 			ret = jbd2_journal_force_commit(osb->journal->j_journal);
 			if (ret < 0)
 				written = ret;
@@ -2062,7 +2071,7 @@ static int ocfs2_splice_to_file(struct pipe_inode_info *pipe,
 	int ret;
 
 	ret = ocfs2_prepare_inode_for_write(out->f_path.dentry,	&sd->pos,
-					    sd->total_len, 0, NULL);
+					    sd->total_len, 0, NULL, NULL);
 	if (ret < 0) {
 		mlog_errno(ret);
 		return ret;
-- 
cgit v1.2.3