Merge branch 'flock' of git://git.kernel.org/pub/scm/linux/kernel/git/arnd/bkl

* 'flock' of git://git.kernel.org/pub/scm/linux/kernel/git/arnd/bkl:
  locks: turn lock_flocks into a spinlock
  fasync: re-organize fasync entry insertion to allow it under a spinlock
  locks/nfsd: allocate file lock outside of spinlock
  lockd: fix nlmsvc_notify_blocked locking
  lockd: push lock_flocks down

9 files changed, 108 insertions, 70 deletions

diff --git a/fs/Kconfig b/fs/Kconfig
index b5e582bd769..97673c95548 100644
--- a/fs/Kconfig
+++ b/fs/Kconfig
@@ -53,7 +53,6 @@ config EXPORTFS
 config FILE_LOCKING
 	bool "Enable POSIX file locking API" if EMBEDDED
 	default y
-	select BKL # while lockd still uses it.
 	help
 	  This option enables standard file locking support, required
           for filesystems like NFS and for the flock() system
diff --git a/fs/fcntl.c b/fs/fcntl.c
index f8cc34f542c..dcdbc6f5c33 100644
--- a/fs/fcntl.c
+++ b/fs/fcntl.c
@@ -640,7 +640,7 @@ static void fasync_free_rcu(struct rcu_head *head)
  * match the state "is the filp on a fasync list".
  *
  */
-static int fasync_remove_entry(struct file *filp, struct fasync_struct **fapp)
+int fasync_remove_entry(struct file *filp, struct fasync_struct **fapp)
 {
 	struct fasync_struct *fa, **fp;
 	int result = 0;
@@ -666,21 +666,28 @@ static int fasync_remove_entry(struct file *filp, struct fasync_struct **fapp)
 	return result;
 }
 
+struct fasync_struct *fasync_alloc(void)
+{
+	return kmem_cache_alloc(fasync_cache, GFP_KERNEL);
+}
+
 /*
- * Add a fasync entry. Return negative on error, positive if
- * added, and zero if did nothing but change an existing one.
- *
- * NOTE! It is very important that the FASYNC flag always
- * match the state "is the filp on a fasync list".
+ * NOTE! This can be used only for unused fasync entries:
+ * entries that actually got inserted on the fasync list
+ * need to be released by rcu - see fasync_remove_entry.
  */
-static int fasync_add_entry(int fd, struct file *filp, struct fasync_struct **fapp)
+void fasync_free(struct fasync_struct *new)
 {
-	struct fasync_struct *new, *fa, **fp;
-	int result = 0;
+	kmem_cache_free(fasync_cache, new);
+}
 
-	new = kmem_cache_alloc(fasync_cache, GFP_KERNEL);
-	if (!new)
-		return -ENOMEM;
+/*
+ * Insert a new entry into the fasync list.  Return the pointer to the
+ * old one if we didn't use the new one.
+ */
+struct fasync_struct *fasync_insert_entry(int fd, struct file *filp, struct fasync_struct **fapp, struct fasync_struct *new)
+{
+        struct fasync_struct *fa, **fp;
 
 	spin_lock(&filp->f_lock);
 	spin_lock(&fasync_lock);
@@ -691,8 +698,6 @@ static int fasync_add_entry(int fd, struct file *filp, struct fasync_struct **fa
 		spin_lock_irq(&fa->fa_lock);
 		fa->fa_fd = fd;
 		spin_unlock_irq(&fa->fa_lock);
-
-		kmem_cache_free(fasync_cache, new);
 		goto out;
 	}
 
@@ -702,13 +707,42 @@ static int fasync_add_entry(int fd, struct file *filp, struct fasync_struct **fa
 	new->fa_fd = fd;
 	new->fa_next = *fapp;
 	rcu_assign_pointer(*fapp, new);
-	result = 1;
 	filp->f_flags |= FASYNC;
 
 out:
 	spin_unlock(&fasync_lock);
 	spin_unlock(&filp->f_lock);
-	return result;
+	return fa;
+}
+
+/*
+ * Add a fasync entry. Return negative on error, positive if
+ * added, and zero if did nothing but change an existing one.
+ *
+ * NOTE! It is very important that the FASYNC flag always
+ * match the state "is the filp on a fasync list".
+ */
+static int fasync_add_entry(int fd, struct file *filp, struct fasync_struct **fapp)
+{
+	struct fasync_struct *new;
+
+	new = fasync_alloc();
+	if (!new)
+		return -ENOMEM;
+
+	/*
+	 * fasync_insert_entry() returns the old (update) entry if
+	 * it existed.
+	 *
+	 * So free the (unused) new entry and return 0 to let the
+	 * caller know that we didn't add any new fasync entries.
+	 */
+	if (fasync_insert_entry(fd, filp, fapp, new)) {
+		fasync_free(new);
+		return 0;
+	}
+
+	return 1;
 }
 
 /*
diff --git a/fs/lockd/svc.c b/fs/lockd/svc.c
index b13aabc1229..abfff9d7979 100644
--- a/fs/lockd/svc.c
+++ b/fs/lockd/svc.c
@@ -22,7 +22,6 @@
 #include <linux/in.h>
 #include <linux/uio.h>
 #include <linux/smp.h>
-#include <linux/smp_lock.h>
 #include <linux/mutex.h>
 #include <linux/kthread.h>
 #include <linux/freezer.h>
@@ -130,15 +129,6 @@ lockd(void *vrqstp)
 
 	dprintk("NFS locking service started (ver " LOCKD_VERSION ").\n");
 
-	/*
-	 * FIXME: it would be nice if lockd didn't spend its entire life
-	 * running under the BKL. At the very least, it would be good to
-	 * have someone clarify what it's intended to protect here. I've
-	 * seen some handwavy posts about posix locking needing to be
-	 * done under the BKL, but it's far from clear.
-	 */
-	lock_kernel();
-
 	if (!nlm_timeout)
 		nlm_timeout = LOCKD_DFLT_TIMEO;
 	nlmsvc_timeout = nlm_timeout * HZ;
@@ -195,7 +185,6 @@ lockd(void *vrqstp)
 	if (nlmsvc_ops)
 		nlmsvc_invalidate_all();
 	nlm_shutdown_hosts();
-	unlock_kernel();
 	return 0;
 }
 
diff --git a/fs/lockd/svclock.c b/fs/lockd/svclock.c
index 6f1ef000975..c462d346acb 100644
--- a/fs/lockd/svclock.c
+++ b/fs/lockd/svclock.c
@@ -700,14 +700,16 @@ nlmsvc_notify_blocked(struct file_lock *fl)
 	struct nlm_block	*block;
 
 	dprintk("lockd: VFS unblock notification for block %p\n", fl);
+	spin_lock(&nlm_blocked_lock);
 	list_for_each_entry(block, &nlm_blocked, b_list) {
 		if (nlm_compare_locks(&block->b_call->a_args.lock.fl, fl)) {
-			nlmsvc_insert_block(block, 0);
+			nlmsvc_insert_block_locked(block, 0);
+			spin_unlock(&nlm_blocked_lock);
 			svc_wake_up(block->b_daemon);
 			return;
 		}
 	}
-
+	spin_unlock(&nlm_blocked_lock);
 	printk(KERN_WARNING "lockd: notification for unknown block!\n");
 }
 
diff --git a/fs/lockd/svcsubs.c b/fs/lockd/svcsubs.c
index d0ef94cfb3d..1ca0679c80b 100644
--- a/fs/lockd/svcsubs.c
+++ b/fs/lockd/svcsubs.c
@@ -170,6 +170,7 @@ nlm_traverse_locks(struct nlm_host *host, struct nlm_file *file,
 
 again:
 	file->f_locks = 0;
+	lock_flocks(); /* protects i_flock list */
 	for (fl = inode->i_flock; fl; fl = fl->fl_next) {
 		if (fl->fl_lmops != &nlmsvc_lock_operations)
 			continue;
@@ -181,6 +182,7 @@ again:
 		if (match(lockhost, host)) {
 			struct file_lock lock = *fl;
 
+			unlock_flocks();
 			lock.fl_type  = F_UNLCK;
 			lock.fl_start = 0;
 			lock.fl_end   = OFFSET_MAX;
@@ -192,6 +194,7 @@ again:
 			goto again;
 		}
 	}
+	unlock_flocks();
 
 	return 0;
 }
@@ -226,10 +229,14 @@ nlm_file_inuse(struct nlm_file *file)
 	if (file->f_count || !list_empty(&file->f_blocks) || file->f_shares)
 		return 1;
 
+	lock_flocks();
 	for (fl = inode->i_flock; fl; fl = fl->fl_next) {
-		if (fl->fl_lmops == &nlmsvc_lock_operations)
+		if (fl->fl_lmops == &nlmsvc_lock_operations) {
+			unlock_flocks();
 			return 1;
+		}
 	}
+	unlock_flocks();
 	file->f_locks = 0;
 	return 0;
 }
diff --git a/fs/locks.c b/fs/locks.c
index 4de3a266681..50ec15927aa 100644
--- a/fs/locks.c
+++ b/fs/locks.c
@@ -142,6 +142,7 @@ int lease_break_time = 45;
 
 static LIST_HEAD(file_lock_list);
 static LIST_HEAD(blocked_list);
+static DEFINE_SPINLOCK(file_lock_lock);
 
 /*
  * Protects the two list heads above, plus the inode->i_flock list
@@ -149,23 +150,24 @@ static LIST_HEAD(blocked_list);
  */
 void lock_flocks(void)
 {
-	lock_kernel();
+	spin_lock(&file_lock_lock);
 }
 EXPORT_SYMBOL_GPL(lock_flocks);
 
 void unlock_flocks(void)
 {
-	unlock_kernel();
+	spin_unlock(&file_lock_lock);
 }
 EXPORT_SYMBOL_GPL(unlock_flocks);
 
 static struct kmem_cache *filelock_cache __read_mostly;
 
 /* Allocate an empty lock structure. */
-static struct file_lock *locks_alloc_lock(void)
+struct file_lock *locks_alloc_lock(void)
 {
 	return kmem_cache_alloc(filelock_cache, GFP_KERNEL);
 }
+EXPORT_SYMBOL_GPL(locks_alloc_lock);
 
 void locks_release_private(struct file_lock *fl)
 {
@@ -1365,7 +1367,6 @@ int fcntl_getlease(struct file *filp)
 int generic_setlease(struct file *filp, long arg, struct file_lock **flp)
 {
 	struct file_lock *fl, **before, **my_before = NULL, *lease;
-	struct file_lock *new_fl = NULL;
 	struct dentry *dentry = filp->f_path.dentry;
 	struct inode *inode = dentry->d_inode;
 	int error, rdlease_count = 0, wrlease_count = 0;
@@ -1385,11 +1386,6 @@ int generic_setlease(struct file *filp, long arg, struct file_lock **flp)
 	lease = *flp;
 
 	if (arg != F_UNLCK) {
-		error = -ENOMEM;
-		new_fl = locks_alloc_lock();
-		if (new_fl == NULL)
-			goto out;
-
 		error = -EAGAIN;
 		if ((arg == F_RDLCK) && (atomic_read(&inode->i_writecount) > 0))
 			goto out;
@@ -1434,7 +1430,6 @@ int generic_setlease(struct file *filp, long arg, struct file_lock **flp)
 		goto out;
 	}
 
-	error = 0;
 	if (arg == F_UNLCK)
 		goto out;
 
@@ -1442,15 +1437,11 @@ int generic_setlease(struct file *filp, long arg, struct file_lock **flp)
 	if (!leases_enable)
 		goto out;
 
-	locks_copy_lock(new_fl, lease);
-	locks_insert_lock(before, new_fl);
-
-	*flp = new_fl;
+	locks_insert_lock(before, lease);
 	return 0;
 
 out:
-	if (new_fl != NULL)
-		locks_free_lock(new_fl);
+	locks_free_lock(lease);
 	return error;
 }
 EXPORT_SYMBOL(generic_setlease);
@@ -1514,26 +1505,38 @@ EXPORT_SYMBOL_GPL(vfs_setlease);
  */
 int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
 {
-	struct file_lock fl, *flp = &fl;
+	struct file_lock *fl;
+	struct fasync_struct *new;
 	struct inode *inode = filp->f_path.dentry->d_inode;
 	int error;
 
-	locks_init_lock(&fl);
-	error = lease_init(filp, arg, &fl);
-	if (error)
-		return error;
+	fl = lease_alloc(filp, arg);
+	if (IS_ERR(fl))
+		return PTR_ERR(fl);
 
+	new = fasync_alloc();
+	if (!new) {
+		locks_free_lock(fl);
+		return -ENOMEM;
+	}
 	lock_flocks();
-
-	error = __vfs_setlease(filp, arg, &flp);
+	error = __vfs_setlease(filp, arg, &fl);
 	if (error || arg == F_UNLCK)
 		goto out_unlock;
 
-	error = fasync_helper(fd, filp, 1, &flp->fl_fasync);
+	/*
+	 * fasync_insert_entry() returns the old entry if any.
+	 * If there was no old entry, then it used 'new' and
+	 * inserted it into the fasync list. Clear new so that
+	 * we don't release it here.
+	 */
+	if (!fasync_insert_entry(fd, filp, &fl->fl_fasync, new))
+		new = NULL;
+
 	if (error < 0) {
 		/* remove lease just inserted by setlease */
-		flp->fl_type = F_UNLCK | F_INPROGRESS;
-		flp->fl_break_time = jiffies - 10;
+		fl->fl_type = F_UNLCK | F_INPROGRESS;
+		fl->fl_break_time = jiffies - 10;
 		time_out_leases(inode);
 		goto out_unlock;
 	}
@@ -1541,6 +1544,8 @@ int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
 	error = __f_setown(filp, task_pid(current), PIDTYPE_PID, 0);
 out_unlock:
 	unlock_flocks();
+	if (new)
+		fasync_free(new);
 	return error;
 }
 
diff --git a/fs/nfs/Kconfig b/fs/nfs/Kconfig
index fd667652c50..ba306658a6d 100644
--- a/fs/nfs/Kconfig
+++ b/fs/nfs/Kconfig
@@ -1,7 +1,6 @@
 config NFS_FS
 	tristate "NFS client support"
 	depends on INET && FILE_LOCKING
-	depends on BKL # fix as soon as lockd is done
 	select LOCKD
 	select SUNRPC
 	select NFS_ACL_SUPPORT if NFS_V3_ACL
diff --git a/fs/nfsd/Kconfig b/fs/nfsd/Kconfig
index 31a78fce473..18b3e8975fe 100644
--- a/fs/nfsd/Kconfig
+++ b/fs/nfsd/Kconfig
@@ -2,7 +2,6 @@ config NFSD
 	tristate "NFS server support"
 	depends on INET
 	depends on FILE_LOCKING
-	depends on BKL # fix as soon as lockd is done
 	select LOCKD
 	select SUNRPC
 	select EXPORTFS
diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c
index 9019e8ec9dc..56347e0ac88 100644
--- a/fs/nfsd/nfs4state.c
+++ b/fs/nfsd/nfs4state.c
@@ -2614,7 +2614,7 @@ nfs4_open_delegation(struct svc_fh *fh, struct nfsd4_open *open, struct nfs4_sta
 	struct nfs4_delegation *dp;
 	struct nfs4_stateowner *sop = stp->st_stateowner;
 	int cb_up = atomic_read(&sop->so_client->cl_cb_set);
-	struct file_lock fl, *flp = &fl;
+	struct file_lock *fl;
 	int status, flag = 0;
 
 	flag = NFS4_OPEN_DELEGATE_NONE;
@@ -2648,20 +2648,24 @@ nfs4_open_delegation(struct svc_fh *fh, struct nfsd4_open *open, struct nfs4_sta
 		flag = NFS4_OPEN_DELEGATE_NONE;
 		goto out;
 	}
-	locks_init_lock(&fl);
-	fl.fl_lmops = &nfsd_lease_mng_ops;
-	fl.fl_flags = FL_LEASE;
-	fl.fl_type = flag == NFS4_OPEN_DELEGATE_READ? F_RDLCK: F_WRLCK;
-	fl.fl_end = OFFSET_MAX;
-	fl.fl_owner =  (fl_owner_t)dp;
-	fl.fl_file = find_readable_file(stp->st_file);
-	BUG_ON(!fl.fl_file);
-	fl.fl_pid = current->tgid;
+	status = -ENOMEM;
+	fl = locks_alloc_lock();
+	if (!fl)
+		goto out;
+	locks_init_lock(fl);
+	fl->fl_lmops = &nfsd_lease_mng_ops;
+	fl->fl_flags = FL_LEASE;
+	fl->fl_type = flag == NFS4_OPEN_DELEGATE_READ? F_RDLCK: F_WRLCK;
+	fl->fl_end = OFFSET_MAX;
+	fl->fl_owner =  (fl_owner_t)dp;
+	fl->fl_file = find_readable_file(stp->st_file);
+	BUG_ON(!fl->fl_file);
+	fl->fl_pid = current->tgid;
 
 	/* vfs_setlease checks to see if delegation should be handed out.
 	 * the lock_manager callbacks fl_mylease and fl_change are used
 	 */
-	if ((status = vfs_setlease(fl.fl_file, fl.fl_type, &flp))) {
+	if ((status = vfs_setlease(fl->fl_file, fl->fl_type, &fl))) {
 		dprintk("NFSD: setlease failed [%d], no delegation\n", status);
 		unhash_delegation(dp);
 		flag = NFS4_OPEN_DELEGATE_NONE;