thr3ads.net - Ocfs2 devel - [Ocfs2-devel] [git patches] Ocfs2 and Configfs updates for 2.6.27 [Jul 2008]

If this information is useful, please help other people find it:
Share via:

Mark Fasheh

2008-Jul-14 22:01 UTC

[Ocfs2-devel] [git patches] Ocfs2 and Configfs updates for 2.6.27

I'm running a bit late with the e-mail this time around, but I think
that's
ok since there really isn't any major new features here - the bulk of the
Ocfs2 update is bug fixes, or cleanups. The same goes for configfs.

The only two things that could be described as features would be:

- Sunil has updated Ocfs2 to provide even more live cluster locking
  information via debugfs.

- Joel updated configfs's ->make_item() and ->make_group() callbacks
to
  return more descriptive error types. This update is responsible for the
  changes to drivers/net/netconsole.c and fs/dlm/config.c as shown in the
diffstat.
	--Mark

Please pull from 'upstream-linus' branch of
git://git.kernel.org/pub/scm/linux/kernel/git/mfasheh/ocfs2.git upstream-linus

to receive the following updates:

 Documentation/filesystems/configfs/configfs.txt    |   10 +-
 .../filesystems/configfs/configfs_example.c        |   14 +-
 drivers/net/netconsole.c                           |   10 +-
 fs/Kconfig                                         |    8 +
 fs/configfs/configfs_internal.h                    |    4 +
 fs/configfs/dir.c                                  |  147 ++++++++++++++------
 fs/configfs/inode.c                                |    2 +
 fs/configfs/symlink.c                              |   16 ++-
 fs/dlm/config.c                                    |   45 ++++---
 fs/ocfs2/aops.c                                    |   17 +++
 fs/ocfs2/cluster/heartbeat.c                       |   17 ++-
 fs/ocfs2/cluster/netdebug.c                        |    8 +-
 fs/ocfs2/cluster/nodemanager.c                     |   45 ++++---
 fs/ocfs2/dlmglue.c                                 |  122 ++++++++++++++++-
 fs/ocfs2/file.c                                    |    2 +-
 fs/ocfs2/journal.c                                 |    2 +-
 fs/ocfs2/localalloc.c                              |    2 +-
 fs/ocfs2/ocfs2.h                                   |   12 ++
 fs/ocfs2/ocfs2_fs.h                                |    2 +-
 fs/ocfs2/stack_user.c                              |   19 +--
 fs/ocfs2/super.c                                   |    6 +-
 include/linux/configfs.h                           |    4 +-
 22 files changed, 390 insertions(+), 124 deletions(-)

Akinobu Mita (1):
      ocfs2: use simple_read_from_buffer()

Coly Li (1):
      ocfs2: fix oops in mmap_truncate testing

Joel Becker (3):
      ocfs2: Fix CONFIG_OCFS2_DEBUG_FS #ifdefs
      ocfs2: Don't snprintf() without a format.
      configfs: Allow ->make_item() and ->make_group() to return detailed
errors.

Louis Rilling (6):
      configfs: Introduce configfs_dirent_lock
      configfs: Protect configfs_dirent s_links list mutations
      configfs: Make configfs_new_dirent() return error code instead of NULL
      configfs: Fix deadlock with racing rmdir() and rename()
      configfs: Fix failing mkdir() making racing rmdir() fail
      configfs: call drop_link() to cleanup after create_link() failure

Randy Dunlap (1):
      ocfs2: fix printk format warnings with OCFS2_FS_STATS=n

Sunil Mushran (4):
      ocfs2: Add CONFIG_OCFS2_FS_STATS config option
      ocfs2: Instrument fs cluster locks
      ocfs2: Silence an error message in ocfs2_file_aio_read()
      ocfs2/net: Silence build warnings on sparc64

Wengang Wang (1):
      ocfs2: Handle error during journal load

Mark Fasheh

2008-Jul-14 22:01 UTC

head link

[Ocfs2-devel] [PATCH 01/17] ocfs2: Add CONFIG_OCFS2_FS_STATS config option

From: Sunil Mushran <sunil.mushran at oracle.com>

This patch adds config option CONFIG_OCFS2_FS_STATS to allow building
the fs with instrumentation enabled. An upcoming patch will provide
support to instrument cluster locking, which is a crucial overhead in
a cluster file system. This config option allows users to avoid the cpu
and memory overhead that is involved in gathering such statistics.

Signed-off-by: Sunil Mushran <sunil.mushran at oracle.com>
Signed-off-by: Mark Fasheh <mfasheh at suse.com>
---
 fs/Kconfig |    8 ++++++++
 1 files changed, 8 insertions(+), 0 deletions(-)

diff --git a/fs/Kconfig b/fs/Kconfig
index 2694648..9c07a4f 100644
--- a/fs/Kconfig
+++ b/fs/Kconfig
@@ -470,6 +470,14 @@ config OCFS2_FS_USERSPACE_CLUSTER
 	  It is safe to say Y, as the clustering method is run-time
 	  selectable.
 
+config OCFS2_FS_STATS
+	bool "OCFS2 statistics"
+	depends on OCFS2_FS
+	default y
+	help
+	  This option allows some fs statistics to be captured. Enabling
+	  this option may increase the memory consumption.
+
 config OCFS2_DEBUG_MASKLOG
 	bool "OCFS2 logging support"
 	depends on OCFS2_FS
-- 
1.5.4.1

Mark Fasheh

2008-Jul-14 22:01 UTC

head link

[Ocfs2-devel] [PATCH 02/17] ocfs2: Instrument fs cluster locks

From: Sunil Mushran <sunil.mushran at oracle.com>

This patch adds code to track the number of times the fs takes
various cluster locks as well as the times associated with it.
The information is made available to users via debugfs.

This patch was originally written by Jan Kara <jack at suse.cz>.

Signed-off-by: Sunil Mushran <sunil.mushran at oracle.com>
Signed-off-by: Mark Fasheh <mfasheh at suse.com>
---
 fs/ocfs2/dlmglue.c |  122 +++++++++++++++++++++++++++++++++++++++++++++++++++-
 fs/ocfs2/ocfs2.h   |   12 +++++
 2 files changed, 133 insertions(+), 1 deletions(-)

diff --git a/fs/ocfs2/dlmglue.c b/fs/ocfs2/dlmglue.c
index 80e20d9..80537b7 100644
--- a/fs/ocfs2/dlmglue.c
+++ b/fs/ocfs2/dlmglue.c
@@ -31,6 +31,7 @@
 #include <linux/pagemap.h>
 #include <linux/debugfs.h>
 #include <linux/seq_file.h>
+#include <linux/time.h>
 
 #define MLOG_MASK_PREFIX ML_DLM_GLUE
 #include <cluster/masklog.h>
@@ -59,6 +60,9 @@ struct ocfs2_mask_waiter {
 	struct completion	mw_complete;
 	unsigned long		mw_mask;
 	unsigned long		mw_goal;
+#ifdef CONFIG_OCFS2_FS_STATS
+	unsigned long long 	mw_lock_start;
+#endif
 };
 
 static struct ocfs2_super *ocfs2_get_dentry_osb(struct ocfs2_lock_res
*lockres);
@@ -366,6 +370,75 @@ static void ocfs2_remove_lockres_tracking(struct
ocfs2_lock_res *res)
 	spin_unlock(&ocfs2_dlm_tracking_lock);
 }
 
+#ifdef CONFIG_OCFS2_FS_STATS
+static void ocfs2_init_lock_stats(struct ocfs2_lock_res *res)
+{
+	res->l_lock_num_prmode = 0;
+	res->l_lock_num_prmode_failed = 0;
+	res->l_lock_total_prmode = 0;
+	res->l_lock_max_prmode = 0;
+	res->l_lock_num_exmode = 0;
+	res->l_lock_num_exmode_failed = 0;
+	res->l_lock_total_exmode = 0;
+	res->l_lock_max_exmode = 0;
+	res->l_lock_refresh = 0;
+}
+
+static void ocfs2_update_lock_stats(struct ocfs2_lock_res *res, int level,
+				    struct ocfs2_mask_waiter *mw, int ret)
+{
+	unsigned long long *num, *sum;
+	unsigned int *max, *failed;
+	struct timespec ts = current_kernel_time();
+	unsigned long long time = timespec_to_ns(&ts) - mw->mw_lock_start;
+
+	if (level == LKM_PRMODE) {
+		num = &res->l_lock_num_prmode;
+		sum = &res->l_lock_total_prmode;
+		max = &res->l_lock_max_prmode;
+		failed = &res->l_lock_num_prmode_failed;
+	} else if (level == LKM_EXMODE) {
+		num = &res->l_lock_num_exmode;
+		sum = &res->l_lock_total_exmode;
+		max = &res->l_lock_max_exmode;
+		failed = &res->l_lock_num_exmode_failed;
+	} else
+		return;
+
+	(*num)++;
+	(*sum) += time;
+	if (time > *max)
+		*max = time;
+	if (ret)
+		(*failed)++;
+}
+
+static inline void ocfs2_track_lock_refresh(struct ocfs2_lock_res *lockres)
+{
+	lockres->l_lock_refresh++;
+}
+
+static inline void ocfs2_init_start_time(struct ocfs2_mask_waiter *mw)
+{
+	struct timespec ts = current_kernel_time();
+	mw->mw_lock_start = timespec_to_ns(&ts);
+}
+#else
+static inline void ocfs2_init_lock_stats(struct ocfs2_lock_res *res)
+{
+}
+static inline void ocfs2_update_lock_stats(struct ocfs2_lock_res *res,
+			   int level, struct ocfs2_mask_waiter *mw, int ret)
+{
+}
+static inline void ocfs2_track_lock_refresh(struct ocfs2_lock_res *lockres)
+{
+}
+static inline void ocfs2_init_start_time(struct ocfs2_mask_waiter *mw)
+{
+}
+#endif
+
 static void ocfs2_lock_res_init_common(struct ocfs2_super *osb,
 				       struct ocfs2_lock_res *res,
 				       enum ocfs2_lock_type type,
@@ -385,6 +458,8 @@ static void ocfs2_lock_res_init_common(struct ocfs2_super
*osb,
 	res->l_flags         = OCFS2_LOCK_INITIALIZED;
 
 	ocfs2_add_lockres_tracking(res, osb->osb_dlm_debug);
+
+	ocfs2_init_lock_stats(res);
 }
 
 void ocfs2_lock_res_init_once(struct ocfs2_lock_res *res)
@@ -1048,6 +1123,7 @@ static void ocfs2_init_mask_waiter(struct
ocfs2_mask_waiter *mw)
 {
 	INIT_LIST_HEAD(&mw->mw_item);
 	init_completion(&mw->mw_complete);
+	ocfs2_init_start_time(mw);
 }
 
 static int ocfs2_wait_for_mask(struct ocfs2_mask_waiter *mw)
@@ -1254,6 +1330,7 @@ out:
 			goto again;
 		mlog_errno(ret);
 	}
+	ocfs2_update_lock_stats(lockres, level, &mw, ret);
 
 	mlog_exit(ret);
 	return ret;
@@ -1983,6 +2060,7 @@ static int ocfs2_inode_lock_update(struct inode *inode,
 				le32_to_cpu(fe->i_flags));
 
 		ocfs2_refresh_inode(inode, fe);
+		ocfs2_track_lock_refresh(lockres);
 	}
 
 	status = 0;
@@ -2267,6 +2345,7 @@ int ocfs2_super_lock(struct ocfs2_super *osb,
 
 		if (status < 0)
 			mlog_errno(status);
+		ocfs2_track_lock_refresh(lockres);
 	}
 bail:
 	mlog_exit(status);
@@ -2461,7 +2540,7 @@ static void *ocfs2_dlm_seq_next(struct seq_file *m, void
*v, loff_t *pos)
 }
 
 /* So that debugfs.ocfs2 can determine which format is being used */
-#define OCFS2_DLM_DEBUG_STR_VERSION 1
+#define OCFS2_DLM_DEBUG_STR_VERSION 2
 static int ocfs2_dlm_seq_show(struct seq_file *m, void *v)
 {
 	int i;
@@ -2502,6 +2581,47 @@ static int ocfs2_dlm_seq_show(struct seq_file *m, void
*v)
 	for(i = 0; i < DLM_LVB_LEN; i++)
 		seq_printf(m, "0x%x\t", lvb[i]);
 
+#ifdef CONFIG_OCFS2_FS_STATS
+# define lock_num_prmode(_l)		(_l)->l_lock_num_prmode
+# define lock_num_exmode(_l)		(_l)->l_lock_num_exmode
+# define lock_num_prmode_failed(_l)	(_l)->l_lock_num_prmode_failed
+# define lock_num_exmode_failed(_l)	(_l)->l_lock_num_exmode_failed
+# define lock_total_prmode(_l)		(_l)->l_lock_total_prmode
+# define lock_total_exmode(_l)		(_l)->l_lock_total_exmode
+# define lock_max_prmode(_l)		(_l)->l_lock_max_prmode
+# define lock_max_exmode(_l)		(_l)->l_lock_max_exmode
+# define lock_refresh(_l)		(_l)->l_lock_refresh
+#else
+# define lock_num_prmode(_l)		(0)
+# define lock_num_exmode(_l)		(0)
+# define lock_num_prmode_failed(_l)	(0)
+# define lock_num_exmode_failed(_l)	(0)
+# define lock_total_prmode(_l)		(0)
+# define lock_total_exmode(_l)		(0)
+# define lock_max_prmode(_l)		(0)
+# define lock_max_exmode(_l)		(0)
+# define lock_refresh(_l)		(0)
+#endif
+	/* The following seq_print was added in version 2 of this output */
+	seq_printf(m, "%llu\t"
+		   "%llu\t"
+		   "%u\t"
+		   "%u\t"
+		   "%llu\t"
+		   "%llu\t"
+		   "%u\t"
+		   "%u\t"
+		   "%u\t",
+		   lock_num_prmode(lockres),
+		   lock_num_exmode(lockres),
+		   lock_num_prmode_failed(lockres),
+		   lock_num_exmode_failed(lockres),
+		   lock_total_prmode(lockres),
+		   lock_total_exmode(lockres),
+		   lock_max_prmode(lockres),
+		   lock_max_exmode(lockres),
+		   lock_refresh(lockres));
+
 	/* End the line */
 	seq_printf(m, "\n");
 	return 0;
diff --git a/fs/ocfs2/ocfs2.h b/fs/ocfs2/ocfs2.h
index 3169237..1cb814b 100644
--- a/fs/ocfs2/ocfs2.h
+++ b/fs/ocfs2/ocfs2.h
@@ -132,6 +132,18 @@ struct ocfs2_lock_res {
 	wait_queue_head_t        l_event;
 
 	struct list_head         l_debug_list;
+
+#ifdef CONFIG_OCFS2_FS_STATS
+	unsigned long long	 l_lock_num_prmode; 	   /* PR acquires */
+	unsigned long long 	 l_lock_num_exmode; 	   /* EX acquires */
+	unsigned int		 l_lock_num_prmode_failed; /* Failed PR gets */
+	unsigned int		 l_lock_num_exmode_failed; /* Failed EX gets */
+	unsigned long long	 l_lock_total_prmode; 	   /* Tot wait for PR */
+	unsigned long long	 l_lock_total_exmode; 	   /* Tot wait for EX */
+	unsigned int		 l_lock_max_prmode; 	   /* Max wait for PR */
+	unsigned int		 l_lock_max_exmode; 	   /* Max wait for EX */
+	unsigned int		 l_lock_refresh;	   /* Disk refreshes */
+#endif
 };
 
 struct ocfs2_dlm_debug {
-- 
1.5.4.1

Mark Fasheh

2008-Jul-14 22:01 UTC

head link

[Ocfs2-devel] [PATCH 03/17] ocfs2: fix printk format warnings with OCFS2_FS_STATS=n

From: Randy Dunlap <randy.dunlap at oracle.com>

Fix printk format warnings when OCFS2_FS_STATS=n:

linux-next-20080528/fs/ocfs2/dlmglue.c: In function
'ocfs2_dlm_seq_show':
linux-next-20080528/fs/ocfs2/dlmglue.c:2623: warning: format '%llu'
expects type 'long long unsigned int', but argument 3 has type
'int'
linux-next-20080528/fs/ocfs2/dlmglue.c:2623: warning: format '%llu'
expects type 'long long unsigned int', but argument 4 has type
'int'
linux-next-20080528/fs/ocfs2/dlmglue.c:2623: warning: format '%llu'
expects type 'long long unsigned int', but argument 7 has type
'int'
linux-next-20080528/fs/ocfs2/dlmglue.c:2623: warning: format '%llu'
expects type 'long long unsigned int', but argument 8 has type
'int'

Signed-off-by: Randy Dunlap <randy.dunlap at oracle.com>
Signed-off-by: Mark Fasheh <mfasheh at suse.com>
---
 fs/ocfs2/dlmglue.c |    8 ++++----
 1 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/fs/ocfs2/dlmglue.c b/fs/ocfs2/dlmglue.c
index 80537b7..eae3d64 100644
--- a/fs/ocfs2/dlmglue.c
+++ b/fs/ocfs2/dlmglue.c
@@ -2592,12 +2592,12 @@ static int ocfs2_dlm_seq_show(struct seq_file *m, void
*v)
 # define lock_max_exmode(_l)		(_l)->l_lock_max_exmode
 # define lock_refresh(_l)		(_l)->l_lock_refresh
 #else
-# define lock_num_prmode(_l)		(0)
-# define lock_num_exmode(_l)		(0)
+# define lock_num_prmode(_l)		(0ULL)
+# define lock_num_exmode(_l)		(0ULL)
 # define lock_num_prmode_failed(_l)	(0)
 # define lock_num_exmode_failed(_l)	(0)
-# define lock_total_prmode(_l)		(0)
-# define lock_total_exmode(_l)		(0)
+# define lock_total_prmode(_l)		(0ULL)
+# define lock_total_exmode(_l)		(0ULL)
 # define lock_max_prmode(_l)		(0)
 # define lock_max_exmode(_l)		(0)
 # define lock_refresh(_l)		(0)
-- 
1.5.4.1

Mark Fasheh

2008-Jul-14 22:01 UTC

head link

[Ocfs2-devel] [PATCH 04/17] ocfs2: use simple_read_from_buffer()

From: Akinobu Mita <akinobu.mita at gmail.com>

Signed-off-by: Akinobu Mita <akinobu.mita at gmail.com>
Acked-by: Joel Becker <joel.becker at oracle.com>
Signed-off-by: Andrew Morton <akpm at linux-foundation.org>
Signed-off-by: Mark Fasheh <mfasheh at suse.com>
---
 fs/ocfs2/stack_user.c |   19 +++++--------------
 1 files changed, 5 insertions(+), 14 deletions(-)

diff --git a/fs/ocfs2/stack_user.c b/fs/ocfs2/stack_user.c
index c021280..24e0b19 100644
--- a/fs/ocfs2/stack_user.c
+++ b/fs/ocfs2/stack_user.c
@@ -549,26 +549,17 @@ static ssize_t ocfs2_control_read(struct file *file,
 				  size_t count,
 				  loff_t *ppos)
 {
-	char *proto_string = OCFS2_CONTROL_PROTO;
-	size_t to_write = 0;
-
-	if (*ppos >= OCFS2_CONTROL_PROTO_LEN)
-		return 0;
-
-	to_write = OCFS2_CONTROL_PROTO_LEN - *ppos;
-	if (to_write > count)
-		to_write = count;
-	if (copy_to_user(buf, proto_string + *ppos, to_write))
-		return -EFAULT;
+	ssize_t ret;
 
-	*ppos += to_write;
+	ret = simple_read_from_buffer(buf, count, ppos,
+			OCFS2_CONTROL_PROTO, OCFS2_CONTROL_PROTO_LEN);
 
 	/* Have we read the whole protocol list? */
-	if (*ppos >= OCFS2_CONTROL_PROTO_LEN)
+	if (ret > 0 && *ppos >= OCFS2_CONTROL_PROTO_LEN)
 		ocfs2_control_set_handshake_state(file,
 						  OCFS2_CONTROL_HANDSHAKE_READ);
 
-	return to_write;
+	return ret;
 }
 
 static int ocfs2_control_release(struct inode *inode, struct file *file)
-- 
1.5.4.1

Mark Fasheh

2008-Jul-14 22:01 UTC

head link

[Ocfs2-devel] [PATCH 05/17] ocfs2: Silence an error message in ocfs2_file_aio_read()

From: Sunil Mushran <sunil.mushran at oracle.com>

This patch silences an EINVAL error message in ocfs2_file_aio_read()
that is always due to a user error.

Signed-off-by: Sunil Mushran <sunil.mushran at oracle.com>
Signed-off-by: Mark Fasheh <mfasheh at suse.com>
---
 fs/ocfs2/file.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/fs/ocfs2/file.c b/fs/ocfs2/file.c
index 57e0d30..e8514e8 100644
--- a/fs/ocfs2/file.c
+++ b/fs/ocfs2/file.c
@@ -2202,7 +2202,7 @@ static ssize_t ocfs2_file_aio_read(struct kiocb *iocb,
 
 	ret = generic_file_aio_read(iocb, iov, nr_segs, iocb->ki_pos);
 	if (ret == -EINVAL)
-		mlog(ML_ERROR, "generic_file_aio_read returned -EINVAL\n");
+		mlog(0, "generic_file_aio_read returned -EINVAL\n");
 
 	/* buffered aio wouldn't have proper lock coverage today */
 	BUG_ON(ret == -EIOCBQUEUED && !(filp->f_flags & O_DIRECT));
-- 
1.5.4.1

Mark Fasheh

2008-Jul-14 22:01 UTC

head link

[Ocfs2-devel] [PATCH 06/17] ocfs2: Handle error during journal load

From: Wengang Wang <wen.gang.wang at oracle.com>

This patch ensures the mount fails if the fs is unable to load the journal.

Signed-off-by: Wengang Wang <wen.gang.wang at oracle.com>
Acked-by: Sunil Mushran <sunil.mushran at oracle.com>
Signed-off-by: Mark Fasheh <mfasheh at suse.com>
---
 fs/ocfs2/super.c |    6 +++++-
 1 files changed, 5 insertions(+), 1 deletions(-)

diff --git a/fs/ocfs2/super.c b/fs/ocfs2/super.c
index df63ba2..ccecfe5 100644
--- a/fs/ocfs2/super.c
+++ b/fs/ocfs2/super.c
@@ -1703,7 +1703,11 @@ static int ocfs2_check_volume(struct ocfs2_super *osb)
 	local = ocfs2_mount_local(osb);
 
 	/* will play back anything left in the journal. */
-	ocfs2_journal_load(osb->journal, local);
+	status = ocfs2_journal_load(osb->journal, local);
+	if (status < 0) {
+		mlog(ML_ERROR, "ocfs2 journal load failed! %d\n", status);
+		goto finally;
+	}
 
 	if (dirty) {
 		/* recover my local alloc if we didn't unmount cleanly. */
-- 
1.5.4.1

Mark Fasheh

2008-Jul-14 22:01 UTC

head link

[Ocfs2-devel] [PATCH 07/17] ocfs2/net: Silence build warnings on sparc64

From: Sunil Mushran <sunil.mushran at oracle.com>

suseconds_t is type long on most arches except sparc64 where it is type int.
This patch silences the following warnings that are generated when building
on it.

netdebug.c: In function 'nst_seq_show':
netdebug.c:152: warning: format '%lu' expects type 'long unsigned
int', but argument 13 has type 'suseconds_t'
netdebug.c:152: warning: format '%lu' expects type 'long unsigned
int', but argument 15 has type 'suseconds_t'
netdebug.c:152: warning: format '%lu' expects type 'long unsigned
int', but argument 17 has type 'suseconds_t'
netdebug.c: In function 'sc_seq_show':
netdebug.c:332: warning: format '%lu' expects type 'long unsigned
int', but argument 19 has type 'suseconds_t'
netdebug.c:332: warning: format '%lu' expects type 'long unsigned
int', but argument 21 has type 'suseconds_t'
netdebug.c:332: warning: format '%lu' expects type 'long unsigned
int', but argument 23 has type 'suseconds_t'
netdebug.c:332: warning: format '%lu' expects type 'long unsigned
int', but argument 25 has type 'suseconds_t'
netdebug.c:332: warning: format '%lu' expects type 'long unsigned
int', but argument 27 has type 'suseconds_t'
netdebug.c:332: warning: format '%lu' expects type 'long unsigned
int', but argument 29 has type 'suseconds_t'

Signed-off-by: Sunil Mushran <sunil.mushran at oracle.com>
Signed-off-by: Mark Fasheh <mfasheh at suse.com>
---
 fs/ocfs2/cluster/netdebug.c |    8 +++++---
 1 files changed, 5 insertions(+), 3 deletions(-)

diff --git a/fs/ocfs2/cluster/netdebug.c b/fs/ocfs2/cluster/netdebug.c
index 7bf3c0e..d8bfa0e 100644
--- a/fs/ocfs2/cluster/netdebug.c
+++ b/fs/ocfs2/cluster/netdebug.c
@@ -146,8 +146,10 @@ static int nst_seq_show(struct seq_file *seq, void *v)
 			   nst->st_task->comm, nst->st_node,
 			   nst->st_sc, nst->st_id, nst->st_msg_type,
 			   nst->st_msg_key,
-			   nst->st_sock_time.tv_sec, nst->st_sock_time.tv_usec,
-			   nst->st_send_time.tv_sec, nst->st_send_time.tv_usec,
+			   nst->st_sock_time.tv_sec,
+			   (unsigned long)nst->st_sock_time.tv_usec,
+			   nst->st_send_time.tv_sec,
+			   (unsigned long)nst->st_send_time.tv_usec,
 			   nst->st_status_time.tv_sec,
 			   nst->st_status_time.tv_usec);
 	}
@@ -274,7 +276,7 @@ static void *sc_seq_next(struct seq_file *seq, void *v,
loff_t *pos)
 	return sc; /* unused, just needs to be null when done */
 }
 
-#define TV_SEC_USEC(TV) TV.tv_sec, TV.tv_usec
+#define TV_SEC_USEC(TV) TV.tv_sec, (unsigned long)TV.tv_usec
 
 static int sc_seq_show(struct seq_file *seq, void *v)
 {
-- 
1.5.4.1

Mark Fasheh

2008-Jul-14 22:01 UTC

head link

[Ocfs2-devel] [PATCH 08/17] ocfs2: Fix CONFIG_OCFS2_DEBUG_FS #ifdefs

From: Joel Becker <joel.becker at oracle.com>

A couple places use OCFS2_DEBUG_FS where they really mean
CONFIG_OCFS2_DEBUG_FS.

Reported-by: Robert P. J. Day <rpjday at crashcourse.ca>
Signed-off-by: Joel Becker <joel.becker at oracle.com>
---
 fs/ocfs2/journal.c    |    2 +-
 fs/ocfs2/localalloc.c |    2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/fs/ocfs2/journal.c b/fs/ocfs2/journal.c
index 9698338..a8c19cb 100644
--- a/fs/ocfs2/journal.c
+++ b/fs/ocfs2/journal.c
@@ -329,7 +329,7 @@ int ocfs2_extend_trans(handle_t *handle, int nblocks)
 
 	mlog(0, "Trying to extend transaction by %d blocks\n", nblocks);
 
-#ifdef OCFS2_DEBUG_FS
+#ifdef CONFIG_OCFS2_DEBUG_FS
 	status = 1;
 #else
 	status = journal_extend(handle, nblocks);
diff --git a/fs/ocfs2/localalloc.c b/fs/ocfs2/localalloc.c
index be774bd..28e492e 100644
--- a/fs/ocfs2/localalloc.c
+++ b/fs/ocfs2/localalloc.c
@@ -498,7 +498,7 @@ int ocfs2_reserve_local_alloc_bits(struct ocfs2_super *osb,
 
 	alloc = (struct ocfs2_dinode *) osb->local_alloc_bh->b_data;
 
-#ifdef OCFS2_DEBUG_FS
+#ifdef CONFIG_OCFS2_DEBUG_FS
 	if (le32_to_cpu(alloc->id1.bitmap1.i_used) ! 	   
ocfs2_local_alloc_count_bits(alloc)) {
 		ocfs2_error(osb->sb, "local alloc inode %llu says it has "
-- 
1.5.4.1

Mark Fasheh

2008-Jul-14 22:01 UTC

head link

[Ocfs2-devel] [PATCH 09/17] ocfs2: Don't snprintf() without a format.

From: Joel Becker <joel.becker at oracle.com>

Some system files are per-slot.  Their names include the slot number.
ocfs2_sprintf_system_inode_name() uses the system inode definitions to
fill in the slot number with snprintf().

For global system files, there is no node number, and the name was
printed as a format with no arguments.  -Wformat-nonliteral and
-Wformat-security don't like this.  Instead, use a static "%s"
format
and the name as the argument.

Signed-off-by: Joel Becker <joel.becker at oracle.com>
---
 fs/ocfs2/ocfs2_fs.h |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/fs/ocfs2/ocfs2_fs.h b/fs/ocfs2/ocfs2_fs.h
index 52c4266..3f19451 100644
--- a/fs/ocfs2/ocfs2_fs.h
+++ b/fs/ocfs2/ocfs2_fs.h
@@ -901,7 +901,7 @@ static inline int ocfs2_sprintf_system_inode_name(char *buf,
int len,
          * list has a copy per slot.
          */
 	if (type <= OCFS2_LAST_GLOBAL_SYSTEM_INODE)
-		chars = snprintf(buf, len,
+		chars = snprintf(buf, len, "%s",
 				 ocfs2_system_inodes[type].si_name);
 	else
 		chars = snprintf(buf, len,
-- 
1.5.4.1

Mark Fasheh

2008-Jul-14 22:01 UTC

head link

[Ocfs2-devel] [PATCH 10/17] configfs: Introduce configfs_dirent_lock

From: Louis Rilling <Louis.Rilling at kerlabs.com>

This patch introduces configfs_dirent_lock spinlock to protect configfs_dirent
traversals against linkage mutations (add/del/move). This will allow
configfs_detach_prep() to avoid locking i_mutexes.

Locking rules for configfs_dirent linkage mutations are the same plus the
requirement of taking configfs_dirent_lock. For configfs_dirent walking, one can
either take appropriate i_mutex as before, or take configfs_dirent_lock.

The spinlock could actually be a mutex, but the critical sections are either
O(1) or should not be too long (default groups walking in last patch).

ChangeLog:
  - Clarify the comment on configfs_dirent_lock usage
  - Move sd->s_element init before linking the new dirent
  - In lseek(), do not release configfs_dirent_lock before the dirent is
    relinked.

Signed-off-by: Louis Rilling <Louis.Rilling at kerlabs.com>
Signed-off-by: Joel Becker <joel.becker at oracle.com>
---
 fs/configfs/configfs_internal.h |    3 +++
 fs/configfs/dir.c               |   28 +++++++++++++++++++++++++++-
 fs/configfs/inode.c             |    2 ++
 fs/configfs/symlink.c           |    2 ++
 4 files changed, 34 insertions(+), 1 deletions(-)

diff --git a/fs/configfs/configfs_internal.h b/fs/configfs/configfs_internal.h
index cca9860..5a33b58 100644
--- a/fs/configfs/configfs_internal.h
+++ b/fs/configfs/configfs_internal.h
@@ -26,6 +26,7 @@
 
 #include <linux/slab.h>
 #include <linux/list.h>
+#include <linux/spinlock.h>
 
 struct configfs_dirent {
 	atomic_t		s_count;
@@ -49,6 +50,8 @@ struct configfs_dirent {
 #define CONFIGFS_USET_DROPPING	0x0100
 #define CONFIGFS_NOT_PINNED	(CONFIGFS_ITEM_ATTR)
 
+extern spinlock_t configfs_dirent_lock;
+
 extern struct vfsmount * configfs_mount;
 extern struct kmem_cache *configfs_dir_cachep;
 
diff --git a/fs/configfs/dir.c b/fs/configfs/dir.c
index a48dc7d..2619f48 100644
--- a/fs/configfs/dir.c
+++ b/fs/configfs/dir.c
@@ -35,6 +35,14 @@
 #include "configfs_internal.h"
 
 DECLARE_RWSEM(configfs_rename_sem);
+/*
+ * Protects mutations of configfs_dirent linkage together with proper i_mutex
+ * Mutators of configfs_dirent linkage must *both* have the proper inode locked
+ * and configfs_dirent_lock locked, in that order.
+ * This allows one to safely traverse configfs_dirent trees without having to
+ * lock inodes.
+ */
+DEFINE_SPINLOCK(configfs_dirent_lock);
 
 static void configfs_d_iput(struct dentry * dentry,
 			    struct inode * inode)
@@ -79,8 +87,10 @@ static struct configfs_dirent *configfs_new_dirent(struct
configfs_dirent * pare
 	atomic_set(&sd->s_count, 1);
 	INIT_LIST_HEAD(&sd->s_links);
 	INIT_LIST_HEAD(&sd->s_children);
-	list_add(&sd->s_sibling, &parent_sd->s_children);
 	sd->s_element = element;
+	spin_lock(&configfs_dirent_lock);
+	list_add(&sd->s_sibling, &parent_sd->s_children);
+	spin_unlock(&configfs_dirent_lock);
 
 	return sd;
 }
@@ -173,7 +183,9 @@ static int create_dir(struct config_item * k, struct dentry
* p,
 		} else {
 			struct configfs_dirent *sd = d->d_fsdata;
 			if (sd) {
+				spin_lock(&configfs_dirent_lock);
 				list_del_init(&sd->s_sibling);
+				spin_unlock(&configfs_dirent_lock);
 				configfs_put(sd);
 			}
 		}
@@ -224,7 +236,9 @@ int configfs_create_link(struct configfs_symlink *sl,
 		else {
 			struct configfs_dirent *sd = dentry->d_fsdata;
 			if (sd) {
+				spin_lock(&configfs_dirent_lock);
 				list_del_init(&sd->s_sibling);
+				spin_unlock(&configfs_dirent_lock);
 				configfs_put(sd);
 			}
 		}
@@ -238,7 +252,9 @@ static void remove_dir(struct dentry * d)
 	struct configfs_dirent * sd;
 
 	sd = d->d_fsdata;
+	spin_lock(&configfs_dirent_lock);
 	list_del_init(&sd->s_sibling);
+	spin_unlock(&configfs_dirent_lock);
 	configfs_put(sd);
 	if (d->d_inode)
 		simple_rmdir(parent->d_inode,d);
@@ -410,7 +426,9 @@ static void detach_attrs(struct config_item * item)
 	list_for_each_entry_safe(sd, tmp, &parent_sd->s_children, s_sibling) {
 		if (!sd->s_element || !(sd->s_type & CONFIGFS_NOT_PINNED))
 			continue;
+		spin_lock(&configfs_dirent_lock);
 		list_del_init(&sd->s_sibling);
+		spin_unlock(&configfs_dirent_lock);
 		configfs_drop_dentry(sd, dentry);
 		configfs_put(sd);
 	}
@@ -1268,7 +1286,9 @@ static int configfs_dir_close(struct inode *inode, struct
file *file)
 	struct configfs_dirent * cursor = file->private_data;
 
 	mutex_lock(&dentry->d_inode->i_mutex);
+	spin_lock(&configfs_dirent_lock);
 	list_del_init(&cursor->s_sibling);
+	spin_unlock(&configfs_dirent_lock);
 	mutex_unlock(&dentry->d_inode->i_mutex);
 
 	release_configfs_dirent(cursor);
@@ -1308,7 +1328,9 @@ static int configfs_readdir(struct file * filp, void *
dirent, filldir_t filldir
 			/* fallthrough */
 		default:
 			if (filp->f_pos == 2) {
+				spin_lock(&configfs_dirent_lock);
 				list_move(q, &parent_sd->s_children);
+				spin_unlock(&configfs_dirent_lock);
 			}
 			for (p=q->next; p!= &parent_sd->s_children; p=p->next) {
 				struct configfs_dirent *next;
@@ -1331,7 +1353,9 @@ static int configfs_readdir(struct file * filp, void *
dirent, filldir_t filldir
 						 dt_type(next)) < 0)
 					return 0;
 
+				spin_lock(&configfs_dirent_lock);
 				list_move(q, p);
+				spin_unlock(&configfs_dirent_lock);
 				p = q;
 				filp->f_pos++;
 			}
@@ -1362,6 +1386,7 @@ static loff_t configfs_dir_lseek(struct file * file,
loff_t offset, int origin)
 			struct list_head *p;
 			loff_t n = file->f_pos - 2;
 
+			spin_lock(&configfs_dirent_lock);
 			list_del(&cursor->s_sibling);
 			p = sd->s_children.next;
 			while (n && p != &sd->s_children) {
@@ -1373,6 +1398,7 @@ static loff_t configfs_dir_lseek(struct file * file,
loff_t offset, int origin)
 				p = p->next;
 			}
 			list_add_tail(&cursor->s_sibling, p);
+			spin_unlock(&configfs_dirent_lock);
 		}
 	}
 	mutex_unlock(&dentry->d_inode->i_mutex);
diff --git a/fs/configfs/inode.c b/fs/configfs/inode.c
index b9a1d81..4803ccc 100644
--- a/fs/configfs/inode.c
+++ b/fs/configfs/inode.c
@@ -247,7 +247,9 @@ void configfs_hash_and_remove(struct dentry * dir, const
char * name)
 		if (!sd->s_element)
 			continue;
 		if (!strcmp(configfs_get_name(sd), name)) {
+			spin_lock(&configfs_dirent_lock);
 			list_del_init(&sd->s_sibling);
+			spin_unlock(&configfs_dirent_lock);
 			configfs_drop_dentry(sd, dir);
 			configfs_put(sd);
 			break;
diff --git a/fs/configfs/symlink.c b/fs/configfs/symlink.c
index 2a731ef..676c84c 100644
--- a/fs/configfs/symlink.c
+++ b/fs/configfs/symlink.c
@@ -169,7 +169,9 @@ int configfs_unlink(struct inode *dir, struct dentry
*dentry)
 	parent_item = configfs_get_config_item(dentry->d_parent);
 	type = parent_item->ci_type;
 
+	spin_lock(&configfs_dirent_lock);
 	list_del_init(&sd->s_sibling);
+	spin_unlock(&configfs_dirent_lock);
 	configfs_drop_dentry(sd, dentry->d_parent);
 	dput(dentry);
 	configfs_put(sd);
-- 
1.5.4.1

Mark Fasheh

2008-Jul-14 22:01 UTC

head link

[Ocfs2-devel] [PATCH 11/17] configfs: Protect configfs_dirent s_links list mutations

From: Louis Rilling <Louis.Rilling at kerlabs.com>

Symlinks to a config_item are listed under its configfs_dirent s_links, but the
list mutations are not protected by any common lock.

This patch uses the configfs_dirent_lock spinlock to add the necessary
protection.

Note: we should also protect the list_empty() test in configfs_detach_prep() but
1/ the lock should not be released immediately because nothing would prevent the
list from being filled after a successful list_empty() test, making the problem
tricky,
2/ this will be solved by the rmdir() vs rename() deadlock bugfix.

Signed-off-by: Louis Rilling <Louis.Rilling at kerlabs.com>
Signed-off-by: Joel Becker <joel.becker at oracle.com>
---
 fs/configfs/dir.c     |    5 +++--
 fs/configfs/symlink.c |    8 ++++++--
 2 files changed, 9 insertions(+), 4 deletions(-)

diff --git a/fs/configfs/dir.c b/fs/configfs/dir.c
index 2619f48..a08e5c2 100644
--- a/fs/configfs/dir.c
+++ b/fs/configfs/dir.c
@@ -37,10 +37,11 @@
 DECLARE_RWSEM(configfs_rename_sem);
 /*
  * Protects mutations of configfs_dirent linkage together with proper i_mutex
+ * Also protects mutations of symlinks linkage to target configfs_dirent
  * Mutators of configfs_dirent linkage must *both* have the proper inode locked
  * and configfs_dirent_lock locked, in that order.
- * This allows one to safely traverse configfs_dirent trees without having to
- * lock inodes.
+ * This allows one to safely traverse configfs_dirent trees and symlinks
without
+ * having to lock inodes.
  */
 DEFINE_SPINLOCK(configfs_dirent_lock);
 
diff --git a/fs/configfs/symlink.c b/fs/configfs/symlink.c
index 676c84c..faeb441 100644
--- a/fs/configfs/symlink.c
+++ b/fs/configfs/symlink.c
@@ -77,12 +77,15 @@ static int create_link(struct config_item *parent_item,
 	sl = kmalloc(sizeof(struct configfs_symlink), GFP_KERNEL);
 	if (sl) {
 		sl->sl_target = config_item_get(item);
-		/* FIXME: needs a lock, I'd bet */
+		spin_lock(&configfs_dirent_lock);
 		list_add(&sl->sl_list, &target_sd->s_links);
+		spin_unlock(&configfs_dirent_lock);
 		ret = configfs_create_link(sl, parent_item->ci_dentry,
 					   dentry);
 		if (ret) {
+			spin_lock(&configfs_dirent_lock);
 			list_del_init(&sl->sl_list);
+			spin_unlock(&configfs_dirent_lock);
 			config_item_put(item);
 			kfree(sl);
 		}
@@ -186,8 +189,9 @@ int configfs_unlink(struct inode *dir, struct dentry
*dentry)
 		type->ct_item_ops->drop_link(parent_item,
 					       sl->sl_target);
 
-	/* FIXME: Needs lock */
+	spin_lock(&configfs_dirent_lock);
 	list_del_init(&sl->sl_list);
+	spin_unlock(&configfs_dirent_lock);
 
 	/* Put reference from create_link() */
 	config_item_put(sl->sl_target);
-- 
1.5.4.1

Mark Fasheh

2008-Jul-14 22:01 UTC

head link

[Ocfs2-devel] [PATCH 12/17] configfs: Make configfs_new_dirent() return error code instead of NULL

From: Louis Rilling <Louis.Rilling at kerlabs.com>

This patch makes configfs_new_dirent return negative error code instead of NULL,
which will be useful in the next patch to differentiate ENOMEM from ENOENT.

Signed-off-by: Louis Rilling <Louis.Rilling at kerlabs.com>
Signed-off-by: Joel Becker <joel.becker at oracle.com>
---
 fs/configfs/dir.c |    9 +++++----
 1 files changed, 5 insertions(+), 4 deletions(-)

diff --git a/fs/configfs/dir.c b/fs/configfs/dir.c
index a08e5c2..918a332 100644
--- a/fs/configfs/dir.c
+++ b/fs/configfs/dir.c
@@ -30,6 +30,7 @@
 #include <linux/mount.h>
 #include <linux/module.h>
 #include <linux/slab.h>
+#include <linux/err.h>
 
 #include <linux/configfs.h>
 #include "configfs_internal.h"
@@ -83,7 +84,7 @@ static struct configfs_dirent *configfs_new_dirent(struct
configfs_dirent * pare
 
 	sd = kmem_cache_zalloc(configfs_dir_cachep, GFP_KERNEL);
 	if (!sd)
-		return NULL;
+		return ERR_PTR(-ENOMEM);
 
 	atomic_set(&sd->s_count, 1);
 	INIT_LIST_HEAD(&sd->s_links);
@@ -129,8 +130,8 @@ int configfs_make_dirent(struct configfs_dirent * parent_sd,
 	struct configfs_dirent * sd;
 
 	sd = configfs_new_dirent(parent_sd, element);
-	if (!sd)
-		return -ENOMEM;
+	if (IS_ERR(sd))
+		return PTR_ERR(sd);
 
 	sd->s_mode = mode;
 	sd->s_type = type;
@@ -1277,7 +1278,7 @@ static int configfs_dir_open(struct inode *inode, struct
file *file)
 	file->private_data = configfs_new_dirent(parent_sd, NULL);
 	mutex_unlock(&dentry->d_inode->i_mutex);
 
-	return file->private_data ? 0 : -ENOMEM;
+	return IS_ERR(file->private_data) ? PTR_ERR(file->private_data) : 0;
 
 }
 
-- 
1.5.4.1

Mark Fasheh

2008-Jul-14 22:01 UTC

head link

[Ocfs2-devel] [PATCH 13/17] configfs: Fix deadlock with racing rmdir() and rename()

From: Louis Rilling <Louis.Rilling at kerlabs.com>

This patch fixes the deadlock between racing sys_rename() and configfs_rmdir().

The idea is to avoid locking i_mutexes of default groups in
configfs_detach_prep(), and rely instead on the new configfs_dirent_lock to
protect against configfs_dirent's linkage mutations. To ensure that an
mkdir()
racing with rmdir() will not create new items in a to-be-removed default group,
we make configfs_new_dirent() check for the CONFIGFS_USET_DROPPING flag right
before linking the new dirent, and return error if the flag is set. This makes
racing mkdir()/symlink()/dir_open() fail in places where errors could already
happen, resp. in (attach_item()|attach_group())/create_link()/new_dirent().

configfs_depend() remains safe since it locks all the path from configfs root,
and is thus mutually exclusive with rmdir().

An advantage of this is that now detach_groups() unconditionnaly takes the
default groups i_mutex, which makes it more consistent with populate_groups().

Signed-off-by: Louis Rilling <Louis.Rilling at kerlabs.com>
Signed-off-by: Joel Becker <joel.becker at oracle.com>
---
 fs/configfs/dir.c |   45 +++++++++++++++++++++++++--------------------
 1 files changed, 25 insertions(+), 20 deletions(-)

diff --git a/fs/configfs/dir.c b/fs/configfs/dir.c
index 918a332..d5b5985 100644
--- a/fs/configfs/dir.c
+++ b/fs/configfs/dir.c
@@ -43,6 +43,10 @@ DECLARE_RWSEM(configfs_rename_sem);
  * and configfs_dirent_lock locked, in that order.
  * This allows one to safely traverse configfs_dirent trees and symlinks
without
  * having to lock inodes.
+ *
+ * Protects setting of CONFIGFS_USET_DROPPING: checking the flag
+ * unlocked is not reliable unless in detach_groups() called from
+ * rmdir()/unregister() and from configfs_attach_group()
  */
 DEFINE_SPINLOCK(configfs_dirent_lock);
 
@@ -91,6 +95,11 @@ static struct configfs_dirent *configfs_new_dirent(struct
configfs_dirent * pare
 	INIT_LIST_HEAD(&sd->s_children);
 	sd->s_element = element;
 	spin_lock(&configfs_dirent_lock);
+	if (parent_sd->s_type & CONFIGFS_USET_DROPPING) {
+		spin_unlock(&configfs_dirent_lock);
+		kmem_cache_free(configfs_dir_cachep, sd);
+		return ERR_PTR(-ENOENT);
+	}
 	list_add(&sd->s_sibling, &parent_sd->s_children);
 	spin_unlock(&configfs_dirent_lock);
 
@@ -349,11 +358,11 @@ static struct dentry * configfs_lookup(struct inode *dir,
 
 /*
  * Only subdirectories count here.  Files (CONFIGFS_NOT_PINNED) are
- * attributes and are removed by rmdir().  We recurse, taking i_mutex
- * on all children that are candidates for default detach.  If the
- * result is clean, then configfs_detach_group() will handle dropping
- * i_mutex.  If there is an error, the caller will clean up the i_mutex
- * holders via configfs_detach_rollback().
+ * attributes and are removed by rmdir().  We recurse, setting
+ * CONFIGFS_USET_DROPPING on all children that are candidates for
+ * default detach.
+ * If there is an error, the caller will reset the flags via
+ * configfs_detach_rollback().
  */
 static int configfs_detach_prep(struct dentry *dentry)
 {
@@ -370,8 +379,7 @@ static int configfs_detach_prep(struct dentry *dentry)
 		if (sd->s_type & CONFIGFS_NOT_PINNED)
 			continue;
 		if (sd->s_type & CONFIGFS_USET_DEFAULT) {
-			mutex_lock(&sd->s_dentry->d_inode->i_mutex);
-			/* Mark that we've taken i_mutex */
+			/* Mark that we're trying to drop the group */
 			sd->s_type |= CONFIGFS_USET_DROPPING;
 
 			/*
@@ -392,7 +400,7 @@ out:
 }
 
 /*
- * Walk the tree, dropping i_mutex wherever CONFIGFS_USET_DROPPING is
+ * Walk the tree, resetting CONFIGFS_USET_DROPPING wherever it was
  * set.
  */
 static void configfs_detach_rollback(struct dentry *dentry)
@@ -403,11 +411,7 @@ static void configfs_detach_rollback(struct dentry *dentry)
 	list_for_each_entry(sd, &parent_sd->s_children, s_sibling) {
 		if (sd->s_type & CONFIGFS_USET_DEFAULT) {
 			configfs_detach_rollback(sd->s_dentry);
-
-			if (sd->s_type & CONFIGFS_USET_DROPPING) {
-				sd->s_type &= ~CONFIGFS_USET_DROPPING;
-				mutex_unlock(&sd->s_dentry->d_inode->i_mutex);
-			}
+			sd->s_type &= ~CONFIGFS_USET_DROPPING;
 		}
 	}
 }
@@ -486,16 +490,12 @@ static void detach_groups(struct config_group *group)
 
 		child = sd->s_dentry;
 
+		mutex_lock(&child->d_inode->i_mutex);
+
 		configfs_detach_group(sd->s_element);
 		child->d_inode->i_flags |= S_DEAD;
 
-		/*
-		 * From rmdir/unregister, a configfs_detach_prep() pass
-		 * has taken our i_mutex for us.  Drop it.
-		 * From mkdir/register cleanup, there is no sem held.
-		 */
-		if (sd->s_type & CONFIGFS_USET_DROPPING)
-			mutex_unlock(&child->d_inode->i_mutex);
+		mutex_unlock(&child->d_inode->i_mutex);
 
 		d_delete(child);
 		dput(child);
@@ -1181,12 +1181,15 @@ static int configfs_rmdir(struct inode *dir, struct
dentry *dentry)
 		return -EINVAL;
 	}
 
+	spin_lock(&configfs_dirent_lock);
 	ret = configfs_detach_prep(dentry);
 	if (ret) {
 		configfs_detach_rollback(dentry);
+		spin_unlock(&configfs_dirent_lock);
 		config_item_put(parent_item);
 		return ret;
 	}
+	spin_unlock(&configfs_dirent_lock);
 
 	/* Get a working ref for the duration of this function */
 	item = configfs_get_config_item(dentry);
@@ -1476,9 +1479,11 @@ void configfs_unregister_subsystem(struct
configfs_subsystem *subsys)
 	mutex_lock_nested(&configfs_sb->s_root->d_inode->i_mutex,
 			  I_MUTEX_PARENT);
 	mutex_lock_nested(&dentry->d_inode->i_mutex, I_MUTEX_CHILD);
+	spin_lock(&configfs_dirent_lock);
 	if (configfs_detach_prep(dentry)) {
 		printk(KERN_ERR "configfs: Tried to unregister non-empty
subsystem!\n");
 	}
+	spin_unlock(&configfs_dirent_lock);
 	configfs_detach_group(&group->cg_item);
 	dentry->d_inode->i_flags |= S_DEAD;
 	mutex_unlock(&dentry->d_inode->i_mutex);
-- 
1.5.4.1

Mark Fasheh

2008-Jul-14 22:01 UTC

head link

[Ocfs2-devel] [PATCH 14/17] configfs: Fix failing mkdir() making racing rmdir() fail

From: Louis Rilling <Louis.Rilling at kerlabs.com>

When fixing the rename() vs rmdir() deadlock, we stopped locking default
groups'
inodes in configfs_detach_prep(), letting racing mkdir() in default groups
proceed concurrently. This enables races like below happen, which leads to a
failing mkdir() making rmdir() fail, despite the group to remove having no
user-created directory under it in the end.

	process A: 			process B:
	/* PWD=A/B */
	mkdir("C")
	  make_item("C")
	  attach_group("C")
					rmdir("A")
					  detach_prep("A")
					    detach_prep("B")
					      error because of "C"
					  return -ENOTEMPTY
	    attach_group("C/D")
	      error (eg -ENOMEM)
	  return -ENOMEM

This patch prevents such scenarii by making rmdir() wait as long as
detach_prep() fails because a racing mkdir() is in the middle of attach_group().
To achieve this, mkdir() sets a flag CONFIGFS_USET_IN_MKDIR in parent's
configfs_dirent before calling attach_group(), and clears the flag once
attach_group() is done. detach_prep() fails with -EAGAIN whenever the flag is
hit and returns the guilty inode's mutex so that rmdir() can wait on it.

Signed-off-by: Louis Rilling <Louis.Rilling at kerlabs.com>
Signed-off-by: Joel Becker <joel.becker at oracle.com>
---
 fs/configfs/configfs_internal.h |    1 +
 fs/configfs/dir.c               |   53 +++++++++++++++++++++++++++++++-------
 2 files changed, 44 insertions(+), 10 deletions(-)

diff --git a/fs/configfs/configfs_internal.h b/fs/configfs/configfs_internal.h
index 5a33b58..da015c1 100644
--- a/fs/configfs/configfs_internal.h
+++ b/fs/configfs/configfs_internal.h
@@ -48,6 +48,7 @@ struct configfs_dirent {
 #define CONFIGFS_USET_DIR	0x0040
 #define CONFIGFS_USET_DEFAULT	0x0080
 #define CONFIGFS_USET_DROPPING	0x0100
+#define CONFIGFS_USET_IN_MKDIR	0x0200
 #define CONFIGFS_NOT_PINNED	(CONFIGFS_ITEM_ATTR)
 
 extern spinlock_t configfs_dirent_lock;
diff --git a/fs/configfs/dir.c b/fs/configfs/dir.c
index d5b5985..614e382 100644
--- a/fs/configfs/dir.c
+++ b/fs/configfs/dir.c
@@ -364,7 +364,7 @@ static struct dentry * configfs_lookup(struct inode *dir,
  * If there is an error, the caller will reset the flags via
  * configfs_detach_rollback().
  */
-static int configfs_detach_prep(struct dentry *dentry)
+static int configfs_detach_prep(struct dentry *dentry, struct mutex
**wait_mutex)
 {
 	struct configfs_dirent *parent_sd = dentry->d_fsdata;
 	struct configfs_dirent *sd;
@@ -379,6 +379,12 @@ static int configfs_detach_prep(struct dentry *dentry)
 		if (sd->s_type & CONFIGFS_NOT_PINNED)
 			continue;
 		if (sd->s_type & CONFIGFS_USET_DEFAULT) {
+			/* Abort if racing with mkdir() */
+			if (sd->s_type & CONFIGFS_USET_IN_MKDIR) {
+				if (wait_mutex)
+					*wait_mutex = &sd->s_dentry->d_inode->i_mutex;
+				return -EAGAIN;
+			}
 			/* Mark that we're trying to drop the group */
 			sd->s_type |= CONFIGFS_USET_DROPPING;
 
@@ -386,7 +392,7 @@ static int configfs_detach_prep(struct dentry *dentry)
 			 * Yup, recursive.  If there's a problem, blame
 			 * deep nesting of default_groups
 			 */
-			ret = configfs_detach_prep(sd->s_dentry);
+			ret = configfs_detach_prep(sd->s_dentry, wait_mutex);
 			if (!ret)
 				continue;
 		} else
@@ -1113,11 +1119,26 @@ static int configfs_mkdir(struct inode *dir, struct
dentry *dentry, int mode)
 	 */
 	module_got = 1;
 
+	/*
+	 * Make racing rmdir() fail if it did not tag parent with
+	 * CONFIGFS_USET_DROPPING
+	 * Note: if CONFIGFS_USET_DROPPING is already set, attach_group() will
+	 * fail and let rmdir() terminate correctly
+	 */
+	spin_lock(&configfs_dirent_lock);
+	/* This will make configfs_detach_prep() fail */
+	sd->s_type |= CONFIGFS_USET_IN_MKDIR;
+	spin_unlock(&configfs_dirent_lock);
+
 	if (group)
 		ret = configfs_attach_group(parent_item, item, dentry);
 	else
 		ret = configfs_attach_item(parent_item, item, dentry);
 
+	spin_lock(&configfs_dirent_lock);
+	sd->s_type &= ~CONFIGFS_USET_IN_MKDIR;
+	spin_unlock(&configfs_dirent_lock);
+
 out_unlink:
 	if (ret) {
 		/* Tear down everything we built up */
@@ -1182,13 +1203,25 @@ static int configfs_rmdir(struct inode *dir, struct
dentry *dentry)
 	}
 
 	spin_lock(&configfs_dirent_lock);
-	ret = configfs_detach_prep(dentry);
-	if (ret) {
-		configfs_detach_rollback(dentry);
-		spin_unlock(&configfs_dirent_lock);
-		config_item_put(parent_item);
-		return ret;
-	}
+	do {
+		struct mutex *wait_mutex;
+
+		ret = configfs_detach_prep(dentry, &wait_mutex);
+		if (ret) {
+			configfs_detach_rollback(dentry);
+			spin_unlock(&configfs_dirent_lock);
+			if (ret != -EAGAIN) {
+				config_item_put(parent_item);
+				return ret;
+			}
+
+			/* Wait until the racing operation terminates */
+			mutex_lock(wait_mutex);
+			mutex_unlock(wait_mutex);
+
+			spin_lock(&configfs_dirent_lock);
+		}
+	} while (ret == -EAGAIN);
 	spin_unlock(&configfs_dirent_lock);
 
 	/* Get a working ref for the duration of this function */
@@ -1480,7 +1513,7 @@ void configfs_unregister_subsystem(struct
configfs_subsystem *subsys)
 			  I_MUTEX_PARENT);
 	mutex_lock_nested(&dentry->d_inode->i_mutex, I_MUTEX_CHILD);
 	spin_lock(&configfs_dirent_lock);
-	if (configfs_detach_prep(dentry)) {
+	if (configfs_detach_prep(dentry, NULL)) {
 		printk(KERN_ERR "configfs: Tried to unregister non-empty
subsystem!\n");
 	}
 	spin_unlock(&configfs_dirent_lock);
-- 
1.5.4.1

Mark Fasheh

2008-Jul-14 22:01 UTC

head link

[Ocfs2-devel] [PATCH 15/17] configfs: Allow ->make_item() and ->make_group() to return detailed errors.

From: Joel Becker <joel.becker at oracle.com>

The configfs operations ->make_item() and ->make_group() currently
return a new item/group.  A return of NULL signifies an error.  Because
of this, -ENOMEM is the only return code bubbled up the stack.

Multiple folks have requested the ability to return specific error codes
when these operations fail.  This patch adds that ability by changing the
->make_item/group() ops to return an int.

Also updated are the in-kernel users of configfs.

Signed-off-by: Joel Becker <joel.becker at oracle.com>
---
 Documentation/filesystems/configfs/configfs.txt    |   10 +++--
 .../filesystems/configfs/configfs_example.c        |   14 ++++---
 drivers/net/netconsole.c                           |   10 +++--
 fs/configfs/dir.c                                  |   13 +++---
 fs/dlm/config.c                                    |   45 ++++++++++++-------
 fs/ocfs2/cluster/heartbeat.c                       |   17 ++++---
 fs/ocfs2/cluster/nodemanager.c                     |   45 ++++++++++++-------
 include/linux/configfs.h                           |    4 +-
 8 files changed, 94 insertions(+), 64 deletions(-)

diff --git a/Documentation/filesystems/configfs/configfs.txt
b/Documentation/filesystems/configfs/configfs.txt
index 44c97e6..15838d7 100644
--- a/Documentation/filesystems/configfs/configfs.txt
+++ b/Documentation/filesystems/configfs/configfs.txt
@@ -233,10 +233,12 @@ accomplished via the group operations specified on the
group's
 config_item_type.
 
 	struct configfs_group_operations {
-		struct config_item *(*make_item)(struct config_group *group,
-						 const char *name);
-		struct config_group *(*make_group)(struct config_group *group,
-						   const char *name);
+		int (*make_item)(struct config_group *group,
+				 const char *name,
+				 struct config_item **new_item);
+		int (*make_group)(struct config_group *group,
+				  const char *name,
+				  struct config_group **new_group);
 		int (*commit_item)(struct config_item *item);
 		void (*disconnect_notify)(struct config_group *group,
 					  struct config_item *item);
diff --git a/Documentation/filesystems/configfs/configfs_example.c
b/Documentation/filesystems/configfs/configfs_example.c
index 25151fd..0b422ac 100644
--- a/Documentation/filesystems/configfs/configfs_example.c
+++ b/Documentation/filesystems/configfs/configfs_example.c
@@ -273,13 +273,13 @@ static inline struct simple_children
*to_simple_children(struct config_item *ite
 	return item ? container_of(to_config_group(item), struct simple_children,
group) : NULL;
 }
 
-static struct config_item *simple_children_make_item(struct config_group
*group, const char *name)
+static int simple_children_make_item(struct config_group *group, const char
*name, struct config_item **new_item)
 {
 	struct simple_child *simple_child;
 
 	simple_child = kzalloc(sizeof(struct simple_child), GFP_KERNEL);
 	if (!simple_child)
-		return NULL;
+		return -ENOMEM;
 
 
 	config_item_init_type_name(&simple_child->item, name,
@@ -287,7 +287,8 @@ static struct config_item *simple_children_make_item(struct
config_group *group,
 
 	simple_child->storeme = 0;
 
-	return &simple_child->item;
+	*new_item = &simple_child->item;
+	return 0;
 }
 
 static struct configfs_attribute simple_children_attr_description = {
@@ -359,20 +360,21 @@ static struct configfs_subsystem simple_children_subsys =
{
  * children of its own.
  */
 
-static struct config_group *group_children_make_group(struct config_group
*group, const char *name)
+static int group_children_make_group(struct config_group *group, const char
*name, struct config_group **new_group)
 {
 	struct simple_children *simple_children;
 
 	simple_children = kzalloc(sizeof(struct simple_children),
 				  GFP_KERNEL);
 	if (!simple_children)
-		return NULL;
+		return -ENOMEM;
 
 
 	config_group_init_type_name(&simple_children->group, name,
 				    &simple_children_type);
 
-	return &simple_children->group;
+	*new_group = &simple_children->group;
+	return 0;
 }
 
 static struct configfs_attribute group_children_attr_description = {
diff --git a/drivers/net/netconsole.c b/drivers/net/netconsole.c
index 665341e..387a133 100644
--- a/drivers/net/netconsole.c
+++ b/drivers/net/netconsole.c
@@ -585,8 +585,9 @@ static struct config_item_type netconsole_target_type = {
  * Group operations and type for netconsole_subsys.
  */
 
-static struct config_item *make_netconsole_target(struct config_group *group,
-						  const char *name)
+static int make_netconsole_target(struct config_group *group,
+				  const char *name,
+				  struct config_item **new_item)
 {
 	unsigned long flags;
 	struct netconsole_target *nt;
@@ -598,7 +599,7 @@ static struct config_item *make_netconsole_target(struct
config_group *group,
 	nt = kzalloc(sizeof(*nt), GFP_KERNEL);
 	if (!nt) {
 		printk(KERN_ERR "netconsole: failed to allocate memory\n");
-		return NULL;
+		return -ENOMEM;
 	}
 
 	nt->np.name = "netconsole";
@@ -615,7 +616,8 @@ static struct config_item *make_netconsole_target(struct
config_group *group,
 	list_add(&nt->list, &target_list);
 	spin_unlock_irqrestore(&target_list_lock, flags);
 
-	return &nt->item;
+	*new_item = &nt->item;
+	return 0;
 }
 
 static void drop_netconsole_target(struct config_group *group,
diff --git a/fs/configfs/dir.c b/fs/configfs/dir.c
index 614e382..0e64312 100644
--- a/fs/configfs/dir.c
+++ b/fs/configfs/dir.c
@@ -1073,25 +1073,24 @@ static int configfs_mkdir(struct inode *dir, struct
dentry *dentry, int mode)
 	group = NULL;
 	item = NULL;
 	if (type->ct_group_ops->make_group) {
-		group = type->ct_group_ops->make_group(to_config_group(parent_item),
name);
-		if (group) {
+		ret = type->ct_group_ops->make_group(to_config_group(parent_item),
name, &group);
+		if (!ret) {
 			link_group(to_config_group(parent_item), group);
 			item = &group->cg_item;
 		}
 	} else {
-		item = type->ct_group_ops->make_item(to_config_group(parent_item),
name);
-		if (item)
+		ret = type->ct_group_ops->make_item(to_config_group(parent_item), name,
&item);
+		if (!ret)
 			link_obj(parent_item, item);
 	}
 	mutex_unlock(&subsys->su_mutex);
 
 	kfree(name);
-	if (!item) {
+	if (ret) {
 		/*
-		 * If item == NULL, then link_obj() was never called.
+		 * If ret != 0, then link_obj() was never called.
 		 * There are no extra references to clean up.
 		 */
-		ret = -ENOMEM;
 		goto out_put;
 	}
 
diff --git a/fs/dlm/config.c b/fs/dlm/config.c
index eac23bd..492d8ca 100644
--- a/fs/dlm/config.c
+++ b/fs/dlm/config.c
@@ -41,16 +41,20 @@ struct comm;
 struct nodes;
 struct node;
 
-static struct config_group *make_cluster(struct config_group *, const char *);
+static int make_cluster(struct config_group *, const char *,
+			struct config_group **);
 static void drop_cluster(struct config_group *, struct config_item *);
 static void release_cluster(struct config_item *);
-static struct config_group *make_space(struct config_group *, const char *);
+static int make_space(struct config_group *, const char *,
+		      struct config_group **);
 static void drop_space(struct config_group *, struct config_item *);
 static void release_space(struct config_item *);
-static struct config_item *make_comm(struct config_group *, const char *);
+static int make_comm(struct config_group *, const char *,
+		     struct config_item **);
 static void drop_comm(struct config_group *, struct config_item *);
 static void release_comm(struct config_item *);
-static struct config_item *make_node(struct config_group *, const char *);
+static int make_node(struct config_group *, const char *,
+		     struct config_item **);
 static void drop_node(struct config_group *, struct config_item *);
 static void release_node(struct config_item *);
 
@@ -392,8 +396,8 @@ static struct node *to_node(struct config_item *i)
 	return i ? container_of(i, struct node, item) : NULL;
 }
 
-static struct config_group *make_cluster(struct config_group *g,
-					 const char *name)
+static int make_cluster(struct config_group *g, const char *name,
+			struct config_group **new_g)
 {
 	struct cluster *cl = NULL;
 	struct spaces *sps = NULL;
@@ -431,14 +435,15 @@ static struct config_group *make_cluster(struct
config_group *g,
 
 	space_list = &sps->ss_group;
 	comm_list = &cms->cs_group;
-	return &cl->group;
+	*new_g = &cl->group;
+	return 0;
 
  fail:
 	kfree(cl);
 	kfree(gps);
 	kfree(sps);
 	kfree(cms);
-	return NULL;
+	return -ENOMEM;
 }
 
 static void drop_cluster(struct config_group *g, struct config_item *i)
@@ -466,7 +471,8 @@ static void release_cluster(struct config_item *i)
 	kfree(cl);
 }
 
-static struct config_group *make_space(struct config_group *g, const char
*name)
+static int make_space(struct config_group *g, const char *name,
+		      struct config_group **new_g)
 {
 	struct space *sp = NULL;
 	struct nodes *nds = NULL;
@@ -489,13 +495,14 @@ static struct config_group *make_space(struct config_group
*g, const char *name)
 	INIT_LIST_HEAD(&sp->members);
 	mutex_init(&sp->members_lock);
 	sp->members_count = 0;
-	return &sp->group;
+	*new_g = &sp->group;
+	return 0;
 
  fail:
 	kfree(sp);
 	kfree(gps);
 	kfree(nds);
-	return NULL;
+	return -ENOMEM;
 }
 
 static void drop_space(struct config_group *g, struct config_item *i)
@@ -522,19 +529,21 @@ static void release_space(struct config_item *i)
 	kfree(sp);
 }
 
-static struct config_item *make_comm(struct config_group *g, const char *name)
+static int make_comm(struct config_group *g, const char *name,
+		     struct config_item **new_i)
 {
 	struct comm *cm;
 
 	cm = kzalloc(sizeof(struct comm), GFP_KERNEL);
 	if (!cm)
-		return NULL;
+		return -ENOMEM;
 
 	config_item_init_type_name(&cm->item, name, &comm_type);
 	cm->nodeid = -1;
 	cm->local = 0;
 	cm->addr_count = 0;
-	return &cm->item;
+	*new_i = &cm->item;
+	return 0;
 }
 
 static void drop_comm(struct config_group *g, struct config_item *i)
@@ -554,14 +563,15 @@ static void release_comm(struct config_item *i)
 	kfree(cm);
 }
 
-static struct config_item *make_node(struct config_group *g, const char *name)
+static int make_node(struct config_group *g, const char *name,
+		     struct config_item **new_i)
 {
 	struct space *sp = to_space(g->cg_item.ci_parent);
 	struct node *nd;
 
 	nd = kzalloc(sizeof(struct node), GFP_KERNEL);
 	if (!nd)
-		return NULL;
+		return -ENOMEM;
 
 	config_item_init_type_name(&nd->item, name, &node_type);
 	nd->nodeid = -1;
@@ -573,7 +583,8 @@ static struct config_item *make_node(struct config_group *g,
const char *name)
 	sp->members_count++;
 	mutex_unlock(&sp->members_lock);
 
-	return &nd->item;
+	*new_i = &nd->item;
+	return 0;
 }
 
 static void drop_node(struct config_group *g, struct config_item *i)
diff --git a/fs/ocfs2/cluster/heartbeat.c b/fs/ocfs2/cluster/heartbeat.c
index f02ccb3..443d108 100644
--- a/fs/ocfs2/cluster/heartbeat.c
+++ b/fs/ocfs2/cluster/heartbeat.c
@@ -1489,25 +1489,28 @@ static struct o2hb_heartbeat_group
*to_o2hb_heartbeat_group(struct config_group
 		: NULL;
 }
 
-static struct config_item *o2hb_heartbeat_group_make_item(struct config_group
*group,
-							  const char *name)
+static int o2hb_heartbeat_group_make_item(struct config_group *group,
+					  const char *name,
+					  struct config_item **new_item)
 {
 	struct o2hb_region *reg = NULL;
-	struct config_item *ret = NULL;
+	int ret = 0;
 
 	reg = kzalloc(sizeof(struct o2hb_region), GFP_KERNEL);
-	if (reg == NULL)
-		goto out; /* ENOMEM */
+	if (reg == NULL) {
+		ret = -ENOMEM;
+		goto out;
+	}
 
 	config_item_init_type_name(&reg->hr_item, name, &o2hb_region_type);
 
-	ret = &reg->hr_item;
+	*new_item = &reg->hr_item;
 
 	spin_lock(&o2hb_live_lock);
 	list_add_tail(&reg->hr_all_item, &o2hb_all_regions);
 	spin_unlock(&o2hb_live_lock);
 out:
-	if (ret == NULL)
+	if (ret)
 		kfree(reg);
 
 	return ret;
diff --git a/fs/ocfs2/cluster/nodemanager.c b/fs/ocfs2/cluster/nodemanager.c
index cfdb08b..b364b70 100644
--- a/fs/ocfs2/cluster/nodemanager.c
+++ b/fs/ocfs2/cluster/nodemanager.c
@@ -644,27 +644,32 @@ out:
 	return ret;
 }
 
-static struct config_item *o2nm_node_group_make_item(struct config_group
*group,
-						     const char *name)
+static int o2nm_node_group_make_item(struct config_group *group,
+				     const char *name,
+				     struct config_item **new_item)
 {
 	struct o2nm_node *node = NULL;
-	struct config_item *ret = NULL;
+	int ret = 0;
 
-	if (strlen(name) > O2NM_MAX_NAME_LEN)
-		goto out; /* ENAMETOOLONG */
+	if (strlen(name) > O2NM_MAX_NAME_LEN) {
+		ret = -ENAMETOOLONG;
+		goto out;
+	}
 
 	node = kzalloc(sizeof(struct o2nm_node), GFP_KERNEL);
-	if (node == NULL)
-		goto out; /* ENOMEM */
+	if (node == NULL) {
+		ret = -ENOMEM;
+		goto out;
+	}
 
 	strcpy(node->nd_name, name); /* use item.ci_namebuf instead? */
 	config_item_init_type_name(&node->nd_item, name, &o2nm_node_type);
 	spin_lock_init(&node->nd_lock);
 
-	ret = &node->nd_item;
+	*new_item = &node->nd_item;
 
 out:
-	if (ret == NULL)
+	if (ret)
 		kfree(node);
 
 	return ret;
@@ -751,25 +756,31 @@ static struct o2nm_cluster_group
*to_o2nm_cluster_group(struct config_group *gro
 }
 #endif
 
-static struct config_group *o2nm_cluster_group_make_group(struct config_group
*group,
-							  const char *name)
+static int o2nm_cluster_group_make_group(struct config_group *group,
+					 const char *name,
+					 struct config_group **new_group)
 {
 	struct o2nm_cluster *cluster = NULL;
 	struct o2nm_node_group *ns = NULL;
-	struct config_group *o2hb_group = NULL, *ret = NULL;
+	struct config_group *o2hb_group = NULL;
 	void *defs = NULL;
+	int ret = 0;
 
 	/* this runs under the parent dir's i_mutex; there can be only
 	 * one caller in here at a time */
-	if (o2nm_single_cluster)
-		goto out; /* ENOSPC */
+	if (o2nm_single_cluster) {
+		ret = -ENOSPC;
+		goto out;
+	}
 
 	cluster = kzalloc(sizeof(struct o2nm_cluster), GFP_KERNEL);
 	ns = kzalloc(sizeof(struct o2nm_node_group), GFP_KERNEL);
 	defs = kcalloc(3, sizeof(struct config_group *), GFP_KERNEL);
 	o2hb_group = o2hb_alloc_hb_set();
-	if (cluster == NULL || ns == NULL || o2hb_group == NULL || defs == NULL)
+	if (cluster == NULL || ns == NULL || o2hb_group == NULL || defs == NULL) {
+		ret = -ENOMEM;
 		goto out;
+	}
 
 	config_group_init_type_name(&cluster->cl_group, name,
 				    &o2nm_cluster_type);
@@ -786,11 +797,11 @@ static struct config_group
*o2nm_cluster_group_make_group(struct config_group *g
 	cluster->cl_idle_timeout_ms    = O2NET_IDLE_TIMEOUT_MS_DEFAULT;
 	cluster->cl_keepalive_delay_ms = O2NET_KEEPALIVE_DELAY_MS_DEFAULT;
 
-	ret = &cluster->cl_group;
+	*new_group = &cluster->cl_group;
 	o2nm_single_cluster = cluster;
 
 out:
-	if (ret == NULL) {
+	if (ret) {
 		kfree(cluster);
 		kfree(ns);
 		o2hb_free_hb_set(o2hb_group);
diff --git a/include/linux/configfs.h b/include/linux/configfs.h
index 3ae65b1..0488f93 100644
--- a/include/linux/configfs.h
+++ b/include/linux/configfs.h
@@ -165,8 +165,8 @@ struct configfs_item_operations {
 };
 
 struct configfs_group_operations {
-	struct config_item *(*make_item)(struct config_group *group, const char
*name);
-	struct config_group *(*make_group)(struct config_group *group, const char
*name);
+	int (*make_item)(struct config_group *group, const char *name, struct
config_item **new_item);
+	int (*make_group)(struct config_group *group, const char *name, struct
config_group **new_group);
 	int (*commit_item)(struct config_item *item);
 	void (*disconnect_notify)(struct config_group *group, struct config_item
*item);
 	void (*drop_item)(struct config_group *group, struct config_item *item);
-- 
1.5.4.1

Mark Fasheh

2008-Jul-14 22:01 UTC

head link

[Ocfs2-devel] [PATCH 16/17] configfs: call drop_link() to cleanup after create_link() failure

From: Louis Rilling <Louis.Rilling at kerlabs.com>

When allow_link() succeeds but create_link() fails, the subsystem is not
informed of the failure.

This patch fixes this by calling drop_link() on create_link() failures.

Signed-off-by: Louis Rilling <Louis.Rilling at kerlabs.com>
Signed-off-by: Joel Becker <joel.becker at oracle.com>
---
 fs/configfs/symlink.c |    6 +++++-
 1 files changed, 5 insertions(+), 1 deletions(-)

diff --git a/fs/configfs/symlink.c b/fs/configfs/symlink.c
index faeb441..0004d18 100644
--- a/fs/configfs/symlink.c
+++ b/fs/configfs/symlink.c
@@ -140,8 +140,12 @@ int configfs_symlink(struct inode *dir, struct dentry
*dentry, const char *symna
 		goto out_put;
 
 	ret = type->ct_item_ops->allow_link(parent_item, target_item);
-	if (!ret)
+	if (!ret) {
 		ret = create_link(parent_item, target_item, dentry);
+		if (ret && type->ct_item_ops->drop_link)
+			type->ct_item_ops->drop_link(parent_item,
+						     target_item);
+	}
 
 	config_item_put(target_item);
 	path_put(&nd.path);
-- 
1.5.4.1

Mark Fasheh

2008-Jul-14 22:01 UTC

head link

[Ocfs2-devel] [PATCH 17/17] [PATCH] ocfs2: fix oops in mmap_truncate testing

From: Coly Li <coyli at suse.de>

This patch fixes a mmap_truncate bug which was found by ocfs2 test suite.

In an ocfs2 cluster more than 1 node, run program mmap_truncate, which races
mmap writes and truncates from multiple processes. While the test is
running, a stat from another node forces writeout, causing an oops in
ocfs2_get_block() because it sees a buffer to write which isn't allocated.

This patch fixed the bug by clear dirty and uptodate bits in buffer, leave
the buffer unmapped and return.

Fix is suggested by Mark Fasheh, and I code up the patch.

Signed-off-by: Coly Li <coyli at suse.de>
Signed-off-by: Mark Fasheh <mfasheh at suse.com>
---
 fs/ocfs2/aops.c |   17 +++++++++++++++++
 1 files changed, 17 insertions(+), 0 deletions(-)

diff --git a/fs/ocfs2/aops.c b/fs/ocfs2/aops.c
index 17964c0..b74a702 100644
--- a/fs/ocfs2/aops.c
+++ b/fs/ocfs2/aops.c
@@ -179,6 +179,23 @@ static int ocfs2_get_block(struct inode *inode, sector_t
iblock,
 			"ino %lu, iblock %llu\n", inode->i_ino,
 			(unsigned long long)iblock);
 
+	/*
+	 * ocfs2 never allocates in this function - the only time we
+	 * need to use BH_New is when we're extending i_size on a file
+	 * system which doesn't support holes, in which case BH_New
+	 * allows block_prepare_write() to zero.
+	 *
+	 * If we see this on a sparse file system, then a truncate has
+	 * raced us and removed the cluster. In this case, we clear
+	 * the buffers dirty and uptodate bits and let the buffer code
+	 * ignore it as a hole.
+	 */
+	if (create && p_blkno == 0 && ocfs2_sparse_alloc(osb)) {
+		clear_buffer_dirty(bh_result);
+		clear_buffer_uptodate(bh_result);
+		goto bail;
+	}
+
 	/* Treat the unwritten extent as a hole for zeroing purposes. */
 	if (p_blkno && !(ext_flags & OCFS2_EXT_UNWRITTEN))
 		map_bh(bh_result, inode->i_sb, p_blkno);
-- 
1.5.4.1

Mark Fasheh

2008-Jul-17 17:51 UTC

head link

[Ocfs2-devel] [git patches] Ocfs2 and Configfs updates for 2.6.27

These patches were sent out earlier this week for review. The only change I
made since then was a small update to Coly's bugfix to remove some code
which it obsoleted.

Original message and revised diffstat follow:

I'm running a bit late with the e-mail this time around, but I think
that's
ok since there really isn't any major new features here - the bulk of the
Ocfs2 update is bug fixes, or cleanups. The same goes for configfs.

The only two things that could be described as features would be:

- Sunil has updated Ocfs2 to provide even more live cluster locking
  information via debugfs.

- Joel updated configfs's ->make_item() and ->make_group() callbacks
to
  return more descriptive error types. This update is responsible for the
  changes to drivers/net/netconsole.c and fs/dlm/config.c as shown in the
diffstat.

Please pull from 'upstream-linus' branch of
git://git.kernel.org/pub/scm/linux/kernel/git/mfasheh/ocfs2.git upstream-linus

to receive the following updates:

 Documentation/filesystems/configfs/configfs.txt    |   10 +-
 .../filesystems/configfs/configfs_example.c        |   14 +-
 drivers/net/netconsole.c                           |   10 +-
 fs/Kconfig                                         |    8 +
 fs/configfs/configfs_internal.h                    |    4 +
 fs/configfs/dir.c                                  |  147 ++++++++++++++------
 fs/configfs/inode.c                                |    2 +
 fs/configfs/symlink.c                              |   16 ++-
 fs/dlm/config.c                                    |   45 ++++---
 fs/ocfs2/aops.c                                    |   13 ++-
 fs/ocfs2/cluster/heartbeat.c                       |   17 ++-
 fs/ocfs2/cluster/netdebug.c                        |    8 +-
 fs/ocfs2/cluster/nodemanager.c                     |   45 ++++---
 fs/ocfs2/dlmglue.c                                 |  122 ++++++++++++++++-
 fs/ocfs2/file.c                                    |    2 +-
 fs/ocfs2/journal.c                                 |    2 +-
 fs/ocfs2/localalloc.c                              |    2 +-
 fs/ocfs2/ocfs2.h                                   |   12 ++
 fs/ocfs2/ocfs2_fs.h                                |    2 +-
 fs/ocfs2/stack_user.c                              |   19 +--
 fs/ocfs2/super.c                                   |    6 +-
 include/linux/configfs.h                           |    4 +-
 22 files changed, 383 insertions(+), 127 deletions(-)

Akinobu Mita (1):
      ocfs2: use simple_read_from_buffer()

Coly Li (1):
      ocfs2: fix oops in mmap_truncate testing

Joel Becker (3):
      ocfs2: Fix CONFIG_OCFS2_DEBUG_FS #ifdefs
      ocfs2: Don't snprintf() without a format.
      configfs: Allow ->make_item() and ->make_group() to return detailed
errors.

Louis Rilling (6):
      configfs: Introduce configfs_dirent_lock
      configfs: Protect configfs_dirent s_links list mutations
      configfs: Make configfs_new_dirent() return error code instead of NULL
      configfs: Fix deadlock with racing rmdir() and rename()
      configfs: Fix failing mkdir() making racing rmdir() fail
      configfs: call drop_link() to cleanup after create_link() failure

Randy Dunlap (1):
      ocfs2: fix printk format warnings with OCFS2_FS_STATS=n

Sunil Mushran (4):
      ocfs2: Add CONFIG_OCFS2_FS_STATS config option
      ocfs2: Instrument fs cluster locks
      ocfs2: Silence an error message in ocfs2_file_aio_read()
      ocfs2/net: Silence build warnings on sparc64

Wengang Wang (1):
      ocfs2: Handle error during journal load

--
Mark Fasheh

Seemingly Similar Threads

Search for more maybe matching threads

Ocfs2 devel - Jul 2008 - [git patches] Ocfs2 and Configfs updates for 2.6.27

[Ocfs2-devel] [git patches] Ocfs2 and Configfs updates for 2.6.27

[Ocfs2-devel] [PATCH 01/17] ocfs2: Add CONFIG_OCFS2_FS_STATS config option

[Ocfs2-devel] [PATCH 02/17] ocfs2: Instrument fs cluster locks

[Ocfs2-devel] [PATCH 03/17] ocfs2: fix printk format warnings with OCFS2_FS_STATS=n

[Ocfs2-devel] [PATCH 04/17] ocfs2: use simple_read_from_buffer()

[Ocfs2-devel] [PATCH 05/17] ocfs2: Silence an error message in ocfs2_file_aio_read()

[Ocfs2-devel] [PATCH 06/17] ocfs2: Handle error during journal load

[Ocfs2-devel] [PATCH 07/17] ocfs2/net: Silence build warnings on sparc64

[Ocfs2-devel] [PATCH 08/17] ocfs2: Fix CONFIG_OCFS2_DEBUG_FS #ifdefs

[Ocfs2-devel] [PATCH 09/17] ocfs2: Don't snprintf() without a format.

[Ocfs2-devel] [PATCH 10/17] configfs: Introduce configfs_dirent_lock

[Ocfs2-devel] [PATCH 11/17] configfs: Protect configfs_dirent s_links list mutations

[Ocfs2-devel] [PATCH 12/17] configfs: Make configfs_new_dirent() return error code instead of NULL

[Ocfs2-devel] [PATCH 13/17] configfs: Fix deadlock with racing rmdir() and rename()

[Ocfs2-devel] [PATCH 14/17] configfs: Fix failing mkdir() making racing rmdir() fail

[Ocfs2-devel] [PATCH 15/17] configfs: Allow ->make_item() and ->make_group() to return detailed errors.

[Ocfs2-devel] [PATCH 16/17] configfs: call drop_link() to cleanup after create_link() failure

[Ocfs2-devel] [PATCH 17/17] [PATCH] ocfs2: fix oops in mmap_truncate testing

[Ocfs2-devel] [git patches] Ocfs2 and Configfs updates for 2.6.27

Seemingly Similar Threads