[PATCH 2/2] [RFC] Remove BKL from fs/locks.c

[PATCH 2/2] [RFC] Remove BKL from fs/locks.c

[Arnd Bergmann]

From: Matthew Wilcox <willy@linux.intel.com>

I've taken a patch originally written by Matthew Wilcox and
ported it to the current version. It seems that there were
originally concerns that this breaks NFS, but since Trond
has recently removed the BKL from NFS, my naive assumption
would be that it's all good now, despite not having tried to
understand what it does.

Original introduction from Willy:

   I've been promising to do this for about seven years now.

   It seems to work well enough, but I haven't run any serious stress
   tests on it.  This implementation uses one spinlock to protect both lock
   lists and all the i_flock chains.  It doesn't seem worth splitting up
   the locking any further.

   I had to move one memory allocation out from under the file_lock_lock.
   I hope I got that logic right.  I'm rather tempted to split out the
   find_conflict algorithm from that function into something that can be
   called separately for the FL_ACCESS case.

   I also have to drop and reacquire the file_lock_lock around the call
   to cond_resched().  This was done automatically for us before by the
   special BKL semantics.

   I had to change vfs_setlease() as it relied on the special BKL ability
   to recursively acquire the same lock.  The internal caller now calls
   __vfs_setlease and the exported interface acquires and releases the
   file_lock_lock around calling __vfs_setlease.

   I should probably split out the removal of interruptible_sleep_on_locked()
   as it's basically unrelated to all this.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Cc: Matthew Wilcox <willy@linux.intel.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Trond Myklebust <trond.myklebust@fys.uio.no>
Cc: "J. Bruce Fields" <bfields@fieldses.org>
Cc: Miklos Szeredi <mszeredi@suse.cz>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: John Kacur <jkacur@redhat.com>
Cc: linux-kernel@vger.kernel.org
Cc: linux-fsdevel@vger.kernel.org
---
 fs/locks.c |  110 ++++++++++++++++++++++++++++++++++++------------------------
 1 files changed, 66 insertions(+), 44 deletions(-)

diff --git a/fs/locks.c b/fs/locks.c
index ab24d49..87f1c60 100644
--- a/fs/locks.c
+++ b/fs/locks.c
@@ -140,9 +140,23 @@ int lease_break_time = 45;
 #define for_each_lock(inode, lockp) \
 	for (lockp = &inode->i_flock; *lockp != NULL; lockp = &(*lockp)->fl_next)
 
+/*
+ * Protects the two list heads below, plus the inode->i_flock list
+ */
+static DEFINE_SPINLOCK(file_lock_lock);
 static LIST_HEAD(file_lock_list);
 static LIST_HEAD(blocked_list);
 
+static inline void lock_flocks(void)
+{
+	spin_lock(&file_lock_lock);
+}
+
+static inline void unlock_flocks(void)
+{
+	spin_unlock(&file_lock_lock);
+}
+
 static struct kmem_cache *filelock_cache __read_mostly;
 
 /* Allocate an empty lock structure. */
@@ -511,9 +525,9 @@ static void __locks_delete_block(struct file_lock *waiter)
  */
 static void locks_delete_block(struct file_lock *waiter)
 {
-	lock_kernel();
+	lock_flocks();
 	__locks_delete_block(waiter);
-	unlock_kernel();
+	unlock_flocks();
 }
 
 /* Insert waiter into blocker's block list.
@@ -644,7 +658,7 @@ posix_test_lock(struct file *filp, struct file_lock *fl)
 {
 	struct file_lock *cfl;
 
-	lock_kernel();
+	lock_flocks();
 	for (cfl = filp->f_path.dentry->d_inode->i_flock; cfl; cfl = cfl->fl_next) {
 		if (!IS_POSIX(cfl))
 			continue;
@@ -657,7 +671,7 @@ posix_test_lock(struct file *filp, struct file_lock *fl)
 			fl->fl_pid = pid_vnr(cfl->fl_nspid);
 	} else
 		fl->fl_type = F_UNLCK;
-	unlock_kernel();
+	unlock_flocks();
 	return;
 }
 EXPORT_SYMBOL(posix_test_lock);
@@ -730,18 +744,16 @@ static int flock_lock_file(struct file *filp, struct file_lock *request)
 	int error = 0;
 	int found = 0;
 
-	lock_kernel();
-	if (request->fl_flags & FL_ACCESS)
-		goto find_conflict;
-
-	if (request->fl_type != F_UNLCK) {
-		error = -ENOMEM;
+	if (!(request->fl_flags & FL_ACCESS) && (request->fl_type != F_UNLCK)) {
 		new_fl = locks_alloc_lock();
-		if (new_fl == NULL)
-			goto out;
-		error = 0;
+		if (!new_fl)
+			return -ENOMEM;
 	}
 
+	lock_flocks();
+	if (request->fl_flags & FL_ACCESS)
+		goto find_conflict;
+
 	for_each_lock(inode, before) {
 		struct file_lock *fl = *before;
 		if (IS_POSIX(fl))
@@ -767,8 +779,11 @@ static int flock_lock_file(struct file *filp, struct file_lock *request)
 	 * If a higher-priority process was blocked on the old file lock,
 	 * give it the opportunity to lock the file.
 	 */
-	if (found)
+	if (found) {
+		unlock_flocks();
 		cond_resched();
+		lock_flocks();
+	}
 
 find_conflict:
 	for_each_lock(inode, before) {
@@ -794,7 +809,7 @@ find_conflict:
 	error = 0;
 
 out:
-	unlock_kernel();
+	unlock_flocks();
 	if (new_fl)
 		locks_free_lock(new_fl);
 	return error;
@@ -823,7 +838,7 @@ static int __posix_lock_file(struct inode *inode, struct file_lock *request, str
 		new_fl2 = locks_alloc_lock();
 	}
 
-	lock_kernel();
+	lock_flocks();
 	if (request->fl_type != F_UNLCK) {
 		for_each_lock(inode, before) {
 			fl = *before;
@@ -991,7 +1006,7 @@ static int __posix_lock_file(struct inode *inode, struct file_lock *request, str
 		locks_wake_up_blocks(left);
 	}
  out:
-	unlock_kernel();
+	unlock_flocks();
 	/*
 	 * Free any unused locks.
 	 */
@@ -1066,14 +1081,14 @@ int locks_mandatory_locked(struct inode *inode)
 	/*
 	 * Search the lock list for this inode for any POSIX locks.
 	 */
-	lock_kernel();
+	lock_flocks();
 	for (fl = inode->i_flock; fl != NULL; fl = fl->fl_next) {
 		if (!IS_POSIX(fl))
 			continue;
 		if (fl->fl_owner != owner)
 			break;
 	}
-	unlock_kernel();
+	unlock_flocks();
 	return fl ? -EAGAIN : 0;
 }
 
@@ -1186,7 +1201,7 @@ int __break_lease(struct inode *inode, unsigned int mode)
 
 	new_fl = lease_alloc(NULL, want_write ? F_WRLCK : F_RDLCK);
 
-	lock_kernel();
+	lock_flocks();
 
 	time_out_leases(inode);
 
@@ -1247,8 +1262,10 @@ restart:
 			break_time++;
 	}
 	locks_insert_block(flock, new_fl);
+	unlock_flocks();
 	error = wait_event_interruptible_timeout(new_fl->fl_wait,
 						!new_fl->fl_next, break_time);
+	lock_flocks();
 	__locks_delete_block(new_fl);
 	if (error >= 0) {
 		if (error == 0)
@@ -1263,7 +1280,7 @@ restart:
 	}
 
 out:
-	unlock_kernel();
+	unlock_flocks();
 	if (!IS_ERR(new_fl))
 		locks_free_lock(new_fl);
 	return error;
@@ -1319,7 +1336,7 @@ int fcntl_getlease(struct file *filp)
 	struct file_lock *fl;
 	int type = F_UNLCK;
 
-	lock_kernel();
+	lock_flocks();
 	time_out_leases(filp->f_path.dentry->d_inode);
 	for (fl = filp->f_path.dentry->d_inode->i_flock; fl && IS_LEASE(fl);
 			fl = fl->fl_next) {
@@ -1328,7 +1345,7 @@ int fcntl_getlease(struct file *filp)
 			break;
 		}
 	}
-	unlock_kernel();
+	unlock_flocks();
 	return type;
 }
 
@@ -1341,7 +1358,7 @@ int fcntl_getlease(struct file *filp)
  *	The (input) flp->fl_lmops->fl_break function is required
  *	by break_lease().
  *
- *	Called with kernel lock held.
+ *	Called with file_lock_lock held.
  */
 int generic_setlease(struct file *filp, long arg, struct file_lock **flp)
 {
@@ -1436,7 +1453,15 @@ out:
 }
 EXPORT_SYMBOL(generic_setlease);
 
- /**
+static int __vfs_setlease(struct file *filp, long arg, struct file_lock **lease)
+{
+	if (filp->f_op && filp->f_op->setlease)
+		return filp->f_op->setlease(filp, arg, lease);
+	else
+		return generic_setlease(filp, arg, lease);
+}
+
+/**
  *	vfs_setlease        -       sets a lease on an open file
  *	@filp: file pointer
  *	@arg: type of lease to obtain
@@ -1467,12 +1492,9 @@ int vfs_setlease(struct file *filp, long arg, struct file_lock **lease)
 {
 	int error;
 
-	lock_kernel();
-	if (filp->f_op && filp->f_op->setlease)
-		error = filp->f_op->setlease(filp, arg, lease);
-	else
-		error = generic_setlease(filp, arg, lease);
-	unlock_kernel();
+	lock_flocks();
+	error = __vfs_setlease(filp, arg, lease);
+	unlock_flocks();
 
 	return error;
 }
@@ -1499,9 +1521,9 @@ int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
 	if (error)
 		return error;
 
-	lock_kernel();
+	lock_flocks();
 
-	error = vfs_setlease(filp, arg, &flp);
+	error = __vfs_setlease(filp, arg, &flp);
 	if (error || arg == F_UNLCK)
 		goto out_unlock;
 
@@ -1516,7 +1538,7 @@ int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
 
 	error = __f_setown(filp, task_pid(current), PIDTYPE_PID, 0);
 out_unlock:
-	unlock_kernel();
+	unlock_flocks();
 	return error;
 }
 
@@ -2020,7 +2042,7 @@ void locks_remove_flock(struct file *filp)
 			fl.fl_ops->fl_release_private(&fl);
 	}
 
-	lock_kernel();
+	lock_flocks();
 	before = &inode->i_flock;
 
 	while ((fl = *before) != NULL) {
@@ -2038,7 +2060,7 @@ void locks_remove_flock(struct file *filp)
  		}
 		before = &fl->fl_next;
 	}
-	unlock_kernel();
+	unlock_flocks();
 }
 
 /**
@@ -2053,12 +2075,12 @@ posix_unblock_lock(struct file *filp, struct file_lock *waiter)
 {
 	int status = 0;
 
-	lock_kernel();
+	lock_flocks();
 	if (waiter->fl_next)
 		__locks_delete_block(waiter);
 	else
 		status = -ENOENT;
-	unlock_kernel();
+	unlock_flocks();
 	return status;
 }
 
@@ -2172,7 +2194,7 @@ static int locks_show(struct seq_file *f, void *v)
 
 static void *locks_start(struct seq_file *f, loff_t *pos)
 {
-	lock_kernel();
+	lock_flocks();
 	f->private = (void *)1;
 	return seq_list_start(&file_lock_list, *pos);
 }
@@ -2184,7 +2206,7 @@ static void *locks_next(struct seq_file *f, void *v, loff_t *pos)
 
 static void locks_stop(struct seq_file *f, void *v)
 {
-	unlock_kernel();
+	unlock_flocks();
 }
 
 static const struct seq_operations locks_seq_operations = {
@@ -2231,7 +2253,7 @@ int lock_may_read(struct inode *inode, loff_t start, unsigned long len)
 {
 	struct file_lock *fl;
 	int result = 1;
-	lock_kernel();
+	lock_flocks();
 	for (fl = inode->i_flock; fl != NULL; fl = fl->fl_next) {
 		if (IS_POSIX(fl)) {
 			if (fl->fl_type == F_RDLCK)
@@ -2248,7 +2270,7 @@ int lock_may_read(struct inode *inode, loff_t start, unsigned long len)
 		result = 0;
 		break;
 	}
-	unlock_kernel();
+	unlock_flocks();
 	return result;
 }
 
@@ -2271,7 +2293,7 @@ int lock_may_write(struct inode *inode, loff_t start, unsigned long len)
 {
 	struct file_lock *fl;
 	int result = 1;
-	lock_kernel();
+	lock_flocks();
 	for (fl = inode->i_flock; fl != NULL; fl = fl->fl_next) {
 		if (IS_POSIX(fl)) {
 			if ((fl->fl_end < start) || (fl->fl_start > (start + len)))
@@ -2286,7 +2308,7 @@ int lock_may_write(struct inode *inode, loff_t start, unsigned long len)
 		result = 0;
 		break;
 	}
-	unlock_kernel();
+	unlock_flocks();
 	return result;
 }
 
-- 
1.7.0.4

Re: [PATCH 2/2] [RFC] Remove BKL from fs/locks.c

[Trond Myklebust]

On Wed, 2010-04-14 at 22:36 +0200, Arnd Bergmann wrote: 
> From: Matthew Wilcox <willy@linux.intel.com>
> 
> I've taken a patch originally written by Matthew Wilcox and
> ported it to the current version. It seems that there were
> originally concerns that this breaks NFS, but since Trond
> has recently removed the BKL from NFS, my naive assumption
> would be that it's all good now, despite not having tried to
> understand what it does.

Hi Arnd,

We still need to fix up the bits in NFS that dereference inode->i_flock.
On the client side, those are mainly the bits that deal with lock
recovery when the NFS server has rebooted or restarted.

AFAICS, there are two places in the NFSv4 client that need to be changed
to call lock_flocks(): nfs_delegation_claim_locks(), and
nfs4_reclaim_locks(). In both cases, the replacement is trivial.

For NFSv3, I think we are already safe, since AFAICS the host->h_rwsem
already provides exclusion between file locking and lock recovery
attempts. I think we should therefore be able to immediately remove the
BKL in fs/lockd/clntlock.c:reclaimer().

I'm not as sure about how sensitive the NFS server is to the switch from
BKL -> lock_flocks(). Perhaps Bruce can comment...

Cheers
  Trond

Re: [PATCH 2/2] [RFC] Remove BKL from fs/locks.c

[J. Bruce Fields]

On Wed, Apr 14, 2010 at 04:52:14PM -0400, Trond Myklebust wrote:
> On Wed, 2010-04-14 at 22:36 +0200, Arnd Bergmann wrote: 
> > From: Matthew Wilcox <willy@linux.intel.com>
> > 
> > I've taken a patch originally written by Matthew Wilcox and
> > ported it to the current version. It seems that there were
> > originally concerns that this breaks NFS, but since Trond
> > has recently removed the BKL from NFS, my naive assumption
> > would be that it's all good now, despite not having tried to
> > understand what it does.
> 
> Hi Arnd,
> 
> We still need to fix up the bits in NFS that dereference inode->i_flock.
> On the client side, those are mainly the bits that deal with lock
> recovery when the NFS server has rebooted or restarted.
> 
> AFAICS, there are two places in the NFSv4 client that need to be changed
> to call lock_flocks(): nfs_delegation_claim_locks(), and
> nfs4_reclaim_locks(). In both cases, the replacement is trivial.
> 
> For NFSv3, I think we are already safe, since AFAICS the host->h_rwsem
> already provides exclusion between file locking and lock recovery
> attempts. I think we should therefore be able to immediately remove the
> BKL in fs/lockd/clntlock.c:reclaimer().
> 
> I'm not as sure about how sensitive the NFS server is to the switch from
> BKL -> lock_flocks(). Perhaps Bruce can comment...

There's a callback from the lease code, the implementation of which
sleeps in the NFSv4 server case.  I've got patches for the next merge
window which remove that sleep (by just deferring work to a workqueue).

release_lockowner() traverses the lock list.  I need to fix that.

The lockd thread is entirely under the BKL--it takes it at the top of
fs/lockd/svc.c:lockd(), and drops it at the bottom.  It's possible
there may be code that lockd runs that assumes mutual exclusion with
code in locks.c, but I don't know.

--b.

Re: [PATCH 2/2] [RFC] Remove BKL from fs/locks.c

[Brad Boyer]

On Wed, Apr 14, 2010 at 10:36:24PM +0200, Arnd Bergmann wrote:
> From: Matthew Wilcox <willy@linux.intel.com>
> 
> I've taken a patch originally written by Matthew Wilcox and
> ported it to the current version. It seems that there were
> originally concerns that this breaks NFS, but since Trond
> has recently removed the BKL from NFS, my naive assumption
> would be that it's all good now, despite not having tried to
> understand what it does.
> 
> Original introduction from Willy:
> 
>    I've been promising to do this for about seven years now.
> 
>    It seems to work well enough, but I haven't run any serious stress
>    tests on it.  This implementation uses one spinlock to protect both lock
>    lists and all the i_flock chains.  It doesn't seem worth splitting up
>    the locking any further.
> 
>    I had to move one memory allocation out from under the file_lock_lock.
>    I hope I got that logic right.  I'm rather tempted to split out the
>    find_conflict algorithm from that function into something that can be
>    called separately for the FL_ACCESS case.
> 
>    I also have to drop and reacquire the file_lock_lock around the call
>    to cond_resched().  This was done automatically for us before by the
>    special BKL semantics.
> 
>    I had to change vfs_setlease() as it relied on the special BKL ability
>    to recursively acquire the same lock.  The internal caller now calls
>    __vfs_setlease and the exported interface acquires and releases the
>    file_lock_lock around calling __vfs_setlease.
> 
>    I should probably split out the removal of interruptible_sleep_on_locked()
>    as it's basically unrelated to all this.

Don't we need to have access to this new lock from modules? In particular,
nfsd/lockd currently call lock_kernel to be able to safely access i_flock.
This would seem to imply that they would either need new functions inside
locks.c to do the same work or export the new lock functions. It seems
easiest to just do EXPORT_SYMBOL on the new lock/unlock functions added
in this patch, but I do understand if that isn't desireable.

	Brad Boyer
	flar@allandria.com

Re: [PATCH 2/2] [RFC] Remove BKL from fs/locks.c

[Steven Whitehouse]

Hi,

Some comments...

On Wed, 2010-04-14 at 22:36 +0200, Arnd Bergmann wrote:
> From: Matthew Wilcox <willy@linux.intel.com>
> 
> I've taken a patch originally written by Matthew Wilcox and
> ported it to the current version. It seems that there were
> originally concerns that this breaks NFS, but since Trond
> has recently removed the BKL from NFS, my naive assumption
> would be that it's all good now, despite not having tried to
> understand what it does.
[snip]
>  fs/locks.c |  110 ++++++++++++++++++++++++++++++++++++------------------------
>  1 files changed, 66 insertions(+), 44 deletions(-)
> 
> diff --git a/fs/locks.c b/fs/locks.c
> index ab24d49..87f1c60 100644
> --- a/fs/locks.c
> +++ b/fs/locks.c
> @@ -140,9 +140,23 @@ int lease_break_time = 45;
>  #define for_each_lock(inode, lockp) \
>  	for (lockp = &inode->i_flock; *lockp != NULL; lockp = &(*lockp)->fl_next)
>  
> +/*
> + * Protects the two list heads below, plus the inode->i_flock list
> + */
> +static DEFINE_SPINLOCK(file_lock_lock);
>  static LIST_HEAD(file_lock_list);
>  static LIST_HEAD(blocked_list);
>  
> +static inline void lock_flocks(void)
> +{
> +	spin_lock(&file_lock_lock);
> +}
> +
> +static inline void unlock_flocks(void)
> +{
> +	spin_unlock(&file_lock_lock);
> +}
> +
>  static struct kmem_cache *filelock_cache __read_mostly;
>  
Why not just put the spin lock calls inline?

[snip]
>  	for_each_lock(inode, before) {
>  		struct file_lock *fl = *before;
>  		if (IS_POSIX(fl))
> @@ -767,8 +779,11 @@ static int flock_lock_file(struct file *filp, struct file_lock *request)
>  	 * If a higher-priority process was blocked on the old file lock,
>  	 * give it the opportunity to lock the file.
>  	 */
> -	if (found)
> +	if (found) {
> +		unlock_flocks();
>  		cond_resched();
> +		lock_flocks();
> +	}
>  
Use cond_resched_lock() here perhaps?

[snip]

> @@ -1341,7 +1358,7 @@ int fcntl_getlease(struct file *filp)
>   *	The (input) flp->fl_lmops->fl_break function is required
>   *	by break_lease().
>   *
> - *	Called with kernel lock held.
> + *	Called with file_lock_lock held.
>   */
>  int generic_setlease(struct file *filp, long arg, struct file_lock **flp)
>  {
> @@ -1436,7 +1453,15 @@ out:
>  }
>  EXPORT_SYMBOL(generic_setlease);
>  
> - /**
> +static int __vfs_setlease(struct file *filp, long arg, struct file_lock **lease)
> +{
> +	if (filp->f_op && filp->f_op->setlease)
> +		return filp->f_op->setlease(filp, arg, lease);
> +	else
> +		return generic_setlease(filp, arg, lease);
> +}
> +
> +/**
>   *	vfs_setlease        -       sets a lease on an open file
>   *	@filp: file pointer
>   *	@arg: type of lease to obtain
> @@ -1467,12 +1492,9 @@ int vfs_setlease(struct file *filp, long arg, struct file_lock **lease)
>  {
>  	int error;
>  
> -	lock_kernel();
> -	if (filp->f_op && filp->f_op->setlease)
> -		error = filp->f_op->setlease(filp, arg, lease);
> -	else
> -		error = generic_setlease(filp, arg, lease);
> -	unlock_kernel();
> +	lock_flocks();
> +	error = __vfs_setlease(filp, arg, lease);
> +	unlock_flocks();
>  
This looks to me like generic_setlease() or whatever fs specific
->setlease() there might be will be called under a spin lock. That
doesn't look right to me.

Rather than adding locking here, why not push the BKL down into
generic_setlease() and ->setlease() first, and then eliminate it on a
case by case basis later on? That is the pattern that has been followed
elsewhere in the kernel.

I might have some further comment on this, but thats as far as I've got
at the moment,

Steve.

Re: [PATCH 2/2] [RFC] Remove BKL from fs/locks.c

[Arnd Bergmann]

On Thursday 15 April 2010, Steven Whitehouse wrote:

> Some comments...

I'll wait for Willy to comment on most of these, except

> On Wed, 2010-04-14 at 22:36 +0200, Arnd Bergmann wrote:
> > @@ -1467,12 +1492,9 @@ int vfs_setlease(struct file *filp, long arg, struct file_lock **lease)
> >  {
> >  	int error;
> >  
> > -	lock_kernel();
> > -	if (filp->f_op && filp->f_op->setlease)
> > -		error = filp->f_op->setlease(filp, arg, lease);
> > -	else
> > -		error = generic_setlease(filp, arg, lease);
> > -	unlock_kernel();
> > +	lock_flocks();
> > +	error = __vfs_setlease(filp, arg, lease);
> > +	unlock_flocks();
> >  
> This looks to me like generic_setlease() or whatever fs specific
> ->setlease() there might be will be called under a spin lock. That
> doesn't look right to me.
> 
> Rather than adding locking here, why not push the BKL down into
> generic_setlease() and ->setlease() first, and then eliminate it on a
> case by case basis later on? That is the pattern that has been followed
> elsewhere in the kernel.

Sounds fair. Besides generic_setlease (which is in this file as well),
the only non-trivial one is cifs_setlease (Cc'ing Steve French now)
and that calls generic_setlease in the end.

If we can show that cifs_setlease does not need locking, the new lock
could be put into generic_setlease directly.

	Arnd

Re: [PATCH 2/2] [RFC] Remove BKL from fs/locks.c

[Arnd Bergmann]

On Wednesday 14 April 2010 23:04:03 J. Bruce Fields wrote:
> The lockd thread is entirely under the BKL--it takes it at the top of
> fs/lockd/svc.c:lockd(), and drops it at the bottom.  It's possible
> there may be code that lockd runs that assumes mutual exclusion with
> code in locks.c, but I don't know.

That one seems interesting as the lockd thread calls lots of sleeping
functions as well as file lock code that takes the BKL again, both
of which of course is not allowed when converting to a spinlock.

I just attempted to blindly convert lockd to taking lock_flocks()
in place of the BKL and releasing it where necessary, but quickly gave
up because this seems rather pointless. It basically never does
anything interesting between places where it would need to drop the
lock.

	Arnd