Merge commit 'dlm/next'

6 files changed, 224 insertions, 66 deletions

diff --git a/fs/dlm/dlm_internal.h b/fs/dlm/dlm_internal.h
index 5a7ac33b629c..868e4c9ef127 100644
--- a/fs/dlm/dlm_internal.h
+++ b/fs/dlm/dlm_internal.h
@@ -2,7 +2,7 @@
 *******************************************************************************
 **
 **  Copyright (C) Sistina Software, Inc.  1997-2003  All rights reserved.
-**  Copyright (C) 2004-2007 Red Hat, Inc.  All rights reserved.
+**  Copyright (C) 2004-2008 Red Hat, Inc.  All rights reserved.
 **
 **  This copyrighted material is made available to anyone wishing to use,
 **  modify, copy, or redistribute it subject to the terms and conditions
@@ -441,8 +441,11 @@ struct dlm_ls {
 	uint32_t		ls_global_id;	/* global unique lockspace ID */
 	uint32_t		ls_exflags;
 	int			ls_lvblen;
-	int			ls_count;	/* reference count */
+	int			ls_count;	/* refcount of processes in
+						   the dlm using this ls */
+	int			ls_create_count; /* create/release refcount */
 	unsigned long		ls_flags;	/* LSFL_ */
+	unsigned long		ls_scan_time;
 	struct kobject		ls_kobj;
 
 	struct dlm_rsbtable	*ls_rsbtbl;
diff --git a/fs/dlm/lockspace.c b/fs/dlm/lockspace.c
index 499e16759e96..bb24f80feda0 100644
--- a/fs/dlm/lockspace.c
+++ b/fs/dlm/lockspace.c
@@ -2,7 +2,7 @@
 *******************************************************************************
 **
 **  Copyright (C) Sistina Software, Inc.  1997-2003  All rights reserved.
-**  Copyright (C) 2004-2007 Red Hat, Inc.  All rights reserved.
+**  Copyright (C) 2004-2008 Red Hat, Inc.  All rights reserved.
 **
 **  This copyrighted material is made available to anyone wishing to use,
 **  modify, copy, or redistribute it subject to the terms and conditions
@@ -23,6 +23,7 @@
 #include "lock.h"
 #include "recover.h"
 #include "requestqueue.h"
+#include "user.h"
 
 static int			ls_count;
 static struct mutex		ls_lock;
@@ -211,19 +212,41 @@ void dlm_lockspace_exit(void)
 	kset_unregister(dlm_kset);
 }
 
+static struct dlm_ls *find_ls_to_scan(void)
+{
+	struct dlm_ls *ls;
+
+	spin_lock(&lslist_lock);
+	list_for_each_entry(ls, &lslist, ls_list) {
+		if (time_after_eq(jiffies, ls->ls_scan_time +
+					    dlm_config.ci_scan_secs * HZ)) {
+			spin_unlock(&lslist_lock);
+			return ls;
+		}
+	}
+	spin_unlock(&lslist_lock);
+	return NULL;
+}
+
 static int dlm_scand(void *data)
 {
 	struct dlm_ls *ls;
+	int timeout_jiffies = dlm_config.ci_scan_secs * HZ;
 
 	while (!kthread_should_stop()) {
-		list_for_each_entry(ls, &lslist, ls_list) {
+		ls = find_ls_to_scan();
+		if (ls) {
 			if (dlm_lock_recovery_try(ls)) {
+				ls->ls_scan_time = jiffies;
 				dlm_scan_rsbs(ls);
 				dlm_scan_timeout(ls);
 				dlm_unlock_recovery(ls);
+			} else {
+				ls->ls_scan_time += HZ;
 			}
+		} else {
+			schedule_timeout_interruptible(timeout_jiffies);
 		}
-		schedule_timeout_interruptible(dlm_config.ci_scan_secs * HZ);
 	}
 	return 0;
 }
@@ -246,23 +269,6 @@ static void dlm_scand_stop(void)
 	kthread_stop(scand_task);
 }
 
-static struct dlm_ls *dlm_find_lockspace_name(char *name, int namelen)
-{
-	struct dlm_ls *ls;
-
-	spin_lock(&lslist_lock);
-
-	list_for_each_entry(ls, &lslist, ls_list) {
-		if (ls->ls_namelen == namelen &&
-		    memcmp(ls->ls_name, name, namelen) == 0)
-			goto out;
-	}
-	ls = NULL;
- out:
-	spin_unlock(&lslist_lock);
-	return ls;
-}
-
 struct dlm_ls *dlm_find_lockspace_global(uint32_t id)
 {
 	struct dlm_ls *ls;
@@ -327,6 +333,7 @@ static void remove_lockspace(struct dlm_ls *ls)
 	for (;;) {
 		spin_lock(&lslist_lock);
 		if (ls->ls_count == 0) {
+			WARN_ON(ls->ls_create_count != 0);
 			list_del(&ls->ls_list);
 			spin_unlock(&lslist_lock);
 			return;
@@ -381,7 +388,7 @@ static int new_lockspace(char *name, int namelen, void **lockspace,
 			 uint32_t flags, int lvblen)
 {
 	struct dlm_ls *ls;
-	int i, size, error = -ENOMEM;
+	int i, size, error;
 	int do_unreg = 0;
 
 	if (namelen > DLM_LOCKSPACE_LEN)
@@ -393,13 +400,38 @@ static int new_lockspace(char *name, int namelen, void **lockspace,
 	if (!try_module_get(THIS_MODULE))
 		return -EINVAL;
 
-	ls = dlm_find_lockspace_name(name, namelen);
-	if (ls) {
-		*lockspace = ls;
+	if (!dlm_user_daemon_available()) {
 		module_put(THIS_MODULE);
-		return -EEXIST;
+		return -EUNATCH;
 	}
 
+	error = 0;
+
+	spin_lock(&lslist_lock);
+	list_for_each_entry(ls, &lslist, ls_list) {
+		WARN_ON(ls->ls_create_count <= 0);
+		if (ls->ls_namelen != namelen)
+			continue;
+		if (memcmp(ls->ls_name, name, namelen))
+			continue;
+		if (flags & DLM_LSFL_NEWEXCL) {
+			error = -EEXIST;
+			break;
+		}
+		ls->ls_create_count++;
+		module_put(THIS_MODULE);
+		error = 1; /* not an error, return 0 */
+		break;
+	}
+	spin_unlock(&lslist_lock);
+
+	if (error < 0)
+		goto out;
+	if (error)
+		goto ret_zero;
+
+	error = -ENOMEM;
+
 	ls = kzalloc(sizeof(struct dlm_ls) + namelen, GFP_KERNEL);
 	if (!ls)
 		goto out;
@@ -408,6 +440,7 @@ static int new_lockspace(char *name, int namelen, void **lockspace,
 	ls->ls_lvblen = lvblen;
 	ls->ls_count = 0;
 	ls->ls_flags = 0;
+	ls->ls_scan_time = jiffies;
 
 	if (flags & DLM_LSFL_TIMEWARN)
 		set_bit(LSFL_TIMEWARN, &ls->ls_flags);
@@ -418,8 +451,9 @@ static int new_lockspace(char *name, int namelen, void **lockspace,
 		ls->ls_allocation = GFP_KERNEL;
 
 	/* ls_exflags are forced to match among nodes, and we don't
-	   need to require all nodes to have TIMEWARN or FS set */
-	ls->ls_exflags = (flags & ~(DLM_LSFL_TIMEWARN | DLM_LSFL_FS));
+	   need to require all nodes to have some flags set */
+	ls->ls_exflags = (flags & ~(DLM_LSFL_TIMEWARN | DLM_LSFL_FS |
+				    DLM_LSFL_NEWEXCL));
 
 	size = dlm_config.ci_rsbtbl_size;
 	ls->ls_rsbtbl_size = size;
@@ -510,6 +544,7 @@ static int new_lockspace(char *name, int namelen, void **lockspace,
 	down_write(&ls->ls_in_recovery);
 
 	spin_lock(&lslist_lock);
+	ls->ls_create_count = 1;
 	list_add(&ls->ls_list, &lslist);
 	spin_unlock(&lslist_lock);
 
@@ -548,7 +583,7 @@ static int new_lockspace(char *name, int namelen, void **lockspace,
 	dlm_create_debug_file(ls);
 
 	log_debug(ls, "join complete");
-
+ ret_zero:
 	*lockspace = ls;
 	return 0;
 
@@ -635,13 +670,34 @@ static int release_lockspace(struct dlm_ls *ls, int force)
 	struct dlm_lkb *lkb;
 	struct dlm_rsb *rsb;
 	struct list_head *head;
-	int i;
-	int busy = lockspace_busy(ls);
+	int i, busy, rv;
+
+	busy = lockspace_busy(ls);
+
+	spin_lock(&lslist_lock);
+	if (ls->ls_create_count == 1) {
+		if (busy > force)
+			rv = -EBUSY;
+		else {
+			/* remove_lockspace takes ls off lslist */
+			ls->ls_create_count = 0;
+			rv = 0;
+		}
+	} else if (ls->ls_create_count > 1) {
+		rv = --ls->ls_create_count;
+	} else {
+		rv = -EINVAL;
+	}
+	spin_unlock(&lslist_lock);
+
+	if (rv) {
+		log_debug(ls, "release_lockspace no remove %d", rv);
+		return rv;
+	}
 
-	if (busy > force)
-		return -EBUSY;
+	dlm_device_deregister(ls);
 
-	if (force < 3)
+	if (force < 3 && dlm_user_daemon_available())
 		do_uevent(ls, 0);
 
 	dlm_recoverd_stop(ls);
@@ -720,15 +776,10 @@ static int release_lockspace(struct dlm_ls *ls, int force)
 	dlm_clear_members(ls);
 	dlm_clear_members_gone(ls);
 	kfree(ls->ls_node_array);
+	log_debug(ls, "release_lockspace final free");
 	kobject_put(&ls->ls_kobj);
 	/* The ls structure will be freed when the kobject is done with */
 
-	mutex_lock(&ls_lock);
-	ls_count--;
-	if (!ls_count)
-		threads_stop();
-	mutex_unlock(&ls_lock);
-
 	module_put(THIS_MODULE);
 	return 0;
 }
@@ -750,11 +801,38 @@ static int release_lockspace(struct dlm_ls *ls, int force)
 int dlm_release_lockspace(void *lockspace, int force)
 {
 	struct dlm_ls *ls;
+	int error;
 
 	ls = dlm_find_lockspace_local(lockspace);
 	if (!ls)
 		return -EINVAL;
 	dlm_put_lockspace(ls);
-	return release_lockspace(ls, force);
+
+	mutex_lock(&ls_lock);
+	error = release_lockspace(ls, force);
+	if (!error) {
+		ls_count--;
+		if (!ls_count)
+			threads_stop();
+	}
+	mutex_unlock(&ls_lock);
+
+	return error;
+}
+
+void dlm_stop_lockspaces(void)
+{
+	struct dlm_ls *ls;
+
+ restart:
+	spin_lock(&lslist_lock);
+	list_for_each_entry(ls, &lslist, ls_list) {
+		if (!test_bit(LSFL_RUNNING, &ls->ls_flags))
+			continue;
+		spin_unlock(&lslist_lock);
+		dlm_ls_stop(ls);
+		goto restart;
+	}
+	spin_unlock(&lslist_lock);
 }
 
diff --git a/fs/dlm/lockspace.h b/fs/dlm/lockspace.h
index 891eabbdd021..f879f87901f8 100644
--- a/fs/dlm/lockspace.h
+++ b/fs/dlm/lockspace.h
@@ -20,6 +20,7 @@ struct dlm_ls *dlm_find_lockspace_global(uint32_t id);
 struct dlm_ls *dlm_find_lockspace_local(void *id);
 struct dlm_ls *dlm_find_lockspace_device(int minor);
 void dlm_put_lockspace(struct dlm_ls *ls);
+void dlm_stop_lockspaces(void);
 
 #endif				/* __LOCKSPACE_DOT_H__ */
 
diff --git a/fs/dlm/user.c b/fs/dlm/user.c
index 34f14a14fb4e..c76f474c8c04 100644
--- a/fs/dlm/user.c
+++ b/fs/dlm/user.c
@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2006-2007 Red Hat, Inc.  All rights reserved.
+ * Copyright (C) 2006-2008 Red Hat, Inc.  All rights reserved.
  *
  * This copyrighted material is made available to anyone wishing to use,
  * modify, copy, or redistribute it subject to the terms and conditions
@@ -27,6 +27,8 @@
 
 static const char name_prefix[] = "dlm";
 static const struct file_operations device_fops;
+static atomic_t dlm_monitor_opened;
+static int dlm_monitor_unused = 1;
 
 #ifdef CONFIG_COMPAT
 
@@ -340,10 +342,15 @@ static int device_user_deadlock(struct dlm_user_proc *proc,
 	return error;
 }
 
-static int create_misc_device(struct dlm_ls *ls, char *name)
+static int dlm_device_register(struct dlm_ls *ls, char *name)
 {
 	int error, len;
 
+	/* The device is already registered.  This happens when the
+	   lockspace is created multiple times from userspace. */
+	if (ls->ls_device.name)
+		return 0;
+
 	error = -ENOMEM;
 	len = strlen(name) + strlen(name_prefix) + 2;
 	ls->ls_device.name = kzalloc(len, GFP_KERNEL);
@@ -363,6 +370,22 @@ fail:
 	return error;
 }
 
+int dlm_device_deregister(struct dlm_ls *ls)
+{
+	int error;
+
+	/* The device is not registered.  This happens when the lockspace
+	   was never used from userspace, or when device_create_lockspace()
+	   calls dlm_release_lockspace() after the register fails. */
+	if (!ls->ls_device.name)
+		return 0;
+
+	error = misc_deregister(&ls->ls_device);
+	if (!error)
+		kfree(ls->ls_device.name);
+	return error;
+}
+
 static int device_user_purge(struct dlm_user_proc *proc,
 			     struct dlm_purge_params *params)
 {
@@ -397,7 +420,7 @@ static int device_create_lockspace(struct dlm_lspace_params *params)
 	if (!ls)
 		return -ENOENT;
 
-	error = create_misc_device(ls, params->name);
+	error = dlm_device_register(ls, params->name);
 	dlm_put_lockspace(ls);
 
 	if (error)
@@ -421,31 +444,22 @@ static int device_remove_lockspace(struct dlm_lspace_params *params)
 	if (!ls)
 		return -ENOENT;
 
-	/* Deregister the misc device first, so we don't have
-	 * a device that's not attached to a lockspace. If
-	 * dlm_release_lockspace fails then we can recreate it
-	 */
-	error = misc_deregister(&ls->ls_device);
-	if (error) {
-		dlm_put_lockspace(ls);
-		goto out;
-	}
-	kfree(ls->ls_device.name);
-
 	if (params->flags & DLM_USER_LSFLG_FORCEFREE)
 		force = 2;
 
 	lockspace = ls->ls_local_handle;
+	dlm_put_lockspace(ls);
 
-	/* dlm_release_lockspace waits for references to go to zero,
-	   so all processes will need to close their device for the ls
-	   before the release will procede */
+	/* The final dlm_release_lockspace waits for references to go to
+	   zero, so all processes will need to close their device for the
+	   ls before the release will proceed.  release also calls the
+	   device_deregister above.  Converting a positive return value
+	   from release to zero means that userspace won't know when its
+	   release was the final one, but it shouldn't need to know. */
 
-	dlm_put_lockspace(ls);
 	error = dlm_release_lockspace(lockspace, force);
-	if (error)
-		create_misc_device(ls, ls->ls_name);
- out:
+	if (error > 0)
+		error = 0;
 	return error;
 }
 
@@ -878,6 +892,26 @@ static unsigned int device_poll(struct file *file, poll_table *wait)
 	return 0;
 }
 
+int dlm_user_daemon_available(void)
+{
+	/* dlm_controld hasn't started (or, has started, but not
+	   properly populated configfs) */
+
+	if (!dlm_our_nodeid())
+		return 0;
+
+	/* This is to deal with versions of dlm_controld that don't
+	   know about the monitor device.  We assume that if the
+	   dlm_controld was started (above), but the monitor device
+	   was never opened, that it's an old version.  dlm_controld
+	   should open the monitor device before populating configfs. */
+
+	if (dlm_monitor_unused)
+		return 1;
+
+	return atomic_read(&dlm_monitor_opened) ? 1 : 0;
+}
+
 static int ctl_device_open(struct inode *inode, struct file *file)
 {
 	cycle_kernel_lock();
@@ -890,6 +924,22 @@ static int ctl_device_close(struct inode *inode, struct file *file)
 	return 0;
 }
 
+static int monitor_device_open(struct inode *inode, struct file *file)
+{
+	atomic_inc(&dlm_monitor_opened);
+	dlm_monitor_unused = 0;
+	return 0;
+}
+
+static int monitor_device_close(struct inode *inode, struct file *file)
+{
+	if (atomic_dec_and_test(&dlm_monitor_opened)) {
+		log_print("no userland control daemon, stopping lockspaces");
+		dlm_stop_lockspaces();
+	}
+	return 0;
+}
+
 static const struct file_operations device_fops = {
 	.open    = device_open,
 	.release = device_close,
@@ -913,19 +963,42 @@ static struct miscdevice ctl_device = {
 	.minor = MISC_DYNAMIC_MINOR,
 };
 
+static const struct file_operations monitor_device_fops = {
+	.open    = monitor_device_open,
+	.release = monitor_device_close,
+	.owner   = THIS_MODULE,
+};
+
+static struct miscdevice monitor_device = {
+	.name  = "dlm-monitor",
+	.fops  = &monitor_device_fops,
+	.minor = MISC_DYNAMIC_MINOR,
+};
+
 int __init dlm_user_init(void)
 {
 	int error;
 
+	atomic_set(&dlm_monitor_opened, 0);
+
 	error = misc_register(&ctl_device);
-	if (error)
+	if (error) {
 		log_print("misc_register failed for control device");
+		goto out;
+	}
 
+	error = misc_register(&monitor_device);
+	if (error) {
+		log_print("misc_register failed for monitor device");
+		misc_deregister(&ctl_device);
+	}
+ out:
 	return error;
 }
 
 void dlm_user_exit(void)
 {
 	misc_deregister(&ctl_device);
+	misc_deregister(&monitor_device);
 }
 
diff --git a/fs/dlm/user.h b/fs/dlm/user.h
index d38e9f3e4151..35eb6a13d616 100644
--- a/fs/dlm/user.h
+++ b/fs/dlm/user.h
@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2006 Red Hat, Inc.  All rights reserved.
+ * Copyright (C) 2006-2008 Red Hat, Inc.  All rights reserved.
  *
  * This copyrighted material is made available to anyone wishing to use,
  * modify, copy, or redistribute it subject to the terms and conditions
@@ -12,5 +12,7 @@
 void dlm_user_add_ast(struct dlm_lkb *lkb, int type);
 int dlm_user_init(void);
 void dlm_user_exit(void);
+int dlm_device_deregister(struct dlm_ls *ls);
+int dlm_user_daemon_available(void);
 
 #endif
diff --git a/fs/gfs2/locking/dlm/mount.c b/fs/gfs2/locking/dlm/mount.c
index 09d78c216f48..0c4cbe6c8285 100644
--- a/fs/gfs2/locking/dlm/mount.c
+++ b/fs/gfs2/locking/dlm/mount.c
@@ -144,7 +144,8 @@ static int gdlm_mount(char *table_name, char *host_data,
 
 	error = dlm_new_lockspace(ls->fsname, strlen(ls->fsname),
 				  &ls->dlm_lockspace,
-				  DLM_LSFL_FS | (nodir ? DLM_LSFL_NODIR : 0),
+				  DLM_LSFL_FS | DLM_LSFL_NEWEXCL |
+				  (nodir ? DLM_LSFL_NODIR : 0),
 				  GDLM_LVB_SIZE);
 	if (error) {
 		log_error("dlm_new_lockspace error %d", error);