[deliverable/linux.git] / fs / btrfs / locking.c

/*
 * Copyright (C) 2008 Oracle.  All rights reserved.
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public
 * License v2 as published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * General Public License for more details.
 *
 * You should have received a copy of the GNU General Public
 * License along with this program; if not, write to the
 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
 * Boston, MA 021110-1307, USA.
 */
#include <linux/sched.h>
#include <linux/pagemap.h>
#include <linux/spinlock.h>
#include <linux/page-flags.h>
#include <asm/bug.h>
#include "ctree.h"
#include "extent_io.h"
#include "locking.h"

void btrfs_assert_tree_read_locked(struct extent_buffer *eb);

/*
 * if we currently have a spinning reader or writer lock
 * (indicated by the rw flag) this will bump the count
 * of blocking holders and drop the spinlock.
 */
void btrfs_set_lock_blocking_rw(struct extent_buffer *eb, int rw)
{
	if (eb->lock_nested) {
		read_lock(&eb->lock);
		if (eb->lock_nested && current->pid == eb->lock_owner) {
			read_unlock(&eb->lock);
			return;
		}
		read_unlock(&eb->lock);
	}
	if (rw == BTRFS_WRITE_LOCK) {
		if (atomic_read(&eb->blocking_writers) == 0) {
			WARN_ON(atomic_read(&eb->spinning_writers) != 1);
			atomic_dec(&eb->spinning_writers);
			btrfs_assert_tree_locked(eb);
			atomic_inc(&eb->blocking_writers);
			write_unlock(&eb->lock);
		}
	} else if (rw == BTRFS_READ_LOCK) {
		btrfs_assert_tree_read_locked(eb);
		atomic_inc(&eb->blocking_readers);
		WARN_ON(atomic_read(&eb->spinning_readers) == 0);
		atomic_dec(&eb->spinning_readers);
		read_unlock(&eb->lock);
	}
	return;
}

/*
 * if we currently have a blocking lock, take the spinlock
 * and drop our blocking count
 */
void btrfs_clear_lock_blocking_rw(struct extent_buffer *eb, int rw)
{
	if (eb->lock_nested) {
		read_lock(&eb->lock);
		if (&eb->lock_nested && current->pid == eb->lock_owner) {
			read_unlock(&eb->lock);
			return;
		}
		read_unlock(&eb->lock);
	}
	if (rw == BTRFS_WRITE_LOCK_BLOCKING) {
		BUG_ON(atomic_read(&eb->blocking_writers) != 1);
		write_lock(&eb->lock);
		WARN_ON(atomic_read(&eb->spinning_writers));
		atomic_inc(&eb->spinning_writers);
		if (atomic_dec_and_test(&eb->blocking_writers))
			wake_up(&eb->write_lock_wq);
	} else if (rw == BTRFS_READ_LOCK_BLOCKING) {
		BUG_ON(atomic_read(&eb->blocking_readers) == 0);
		read_lock(&eb->lock);
		atomic_inc(&eb->spinning_readers);
		if (atomic_dec_and_test(&eb->blocking_readers))
			wake_up(&eb->read_lock_wq);
	}
	return;
}

/*
 * take a spinning read lock.  This will wait for any blocking
 * writers
 */
void btrfs_tree_read_lock(struct extent_buffer *eb)
{
again:
	read_lock(&eb->lock);
	if (atomic_read(&eb->blocking_writers) &&
	    current->pid == eb->lock_owner) {
		/*
		 * This extent is already write-locked by our thread. We allow
		 * an additional read lock to be added because it's for the same
		 * thread. btrfs_find_all_roots() depends on this as it may be
		 * called on a partly (write-)locked tree.
		 */
		BUG_ON(eb->lock_nested);
		eb->lock_nested = 1;
		read_unlock(&eb->lock);
		return;
	}
	read_unlock(&eb->lock);
	wait_event(eb->write_lock_wq, atomic_read(&eb->blocking_writers) == 0);
	read_lock(&eb->lock);
	if (atomic_read(&eb->blocking_writers)) {
		read_unlock(&eb->lock);
		goto again;
	}
	atomic_inc(&eb->read_locks);
	atomic_inc(&eb->spinning_readers);
}

/*
 * returns 1 if we get the read lock and 0 if we don't
 * this won't wait for blocking writers
 */
int btrfs_try_tree_read_lock(struct extent_buffer *eb)
{
	if (atomic_read(&eb->blocking_writers))
		return 0;

	read_lock(&eb->lock);
	if (atomic_read(&eb->blocking_writers)) {
		read_unlock(&eb->lock);
		return 0;
	}
	atomic_inc(&eb->read_locks);
	atomic_inc(&eb->spinning_readers);
	return 1;
}

/*
 * returns 1 if we get the read lock and 0 if we don't
 * this won't wait for blocking writers or readers
 */
int btrfs_try_tree_write_lock(struct extent_buffer *eb)
{
	if (atomic_read(&eb->blocking_writers) ||
	    atomic_read(&eb->blocking_readers))
		return 0;
	write_lock(&eb->lock);
	if (atomic_read(&eb->blocking_writers) ||
	    atomic_read(&eb->blocking_readers)) {
		write_unlock(&eb->lock);
		return 0;
	}
	atomic_inc(&eb->write_locks);
	atomic_inc(&eb->spinning_writers);
	eb->lock_owner = current->pid;
	return 1;
}

/*
 * drop a spinning read lock
 */
void btrfs_tree_read_unlock(struct extent_buffer *eb)
{
	if (eb->lock_nested) {
		read_lock(&eb->lock);
		if (eb->lock_nested && current->pid == eb->lock_owner) {
			eb->lock_nested = 0;
			read_unlock(&eb->lock);
			return;
		}
		read_unlock(&eb->lock);
	}
	btrfs_assert_tree_read_locked(eb);
	WARN_ON(atomic_read(&eb->spinning_readers) == 0);
	atomic_dec(&eb->spinning_readers);
	atomic_dec(&eb->read_locks);
	read_unlock(&eb->lock);
}

/*
 * drop a blocking read lock
 */
void btrfs_tree_read_unlock_blocking(struct extent_buffer *eb)
{
	if (eb->lock_nested) {
		read_lock(&eb->lock);
		if (eb->lock_nested && current->pid == eb->lock_owner) {
			eb->lock_nested = 0;
			read_unlock(&eb->lock);
			return;
		}
		read_unlock(&eb->lock);
	}
	btrfs_assert_tree_read_locked(eb);
	WARN_ON(atomic_read(&eb->blocking_readers) == 0);
	if (atomic_dec_and_test(&eb->blocking_readers))
		wake_up(&eb->read_lock_wq);
	atomic_dec(&eb->read_locks);
}

/*
 * take a spinning write lock.  This will wait for both
 * blocking readers or writers
 */
void btrfs_tree_lock(struct extent_buffer *eb)
{
again:
	wait_event(eb->read_lock_wq, atomic_read(&eb->blocking_readers) == 0);
	wait_event(eb->write_lock_wq, atomic_read(&eb->blocking_writers) == 0);
	write_lock(&eb->lock);
	if (atomic_read(&eb->blocking_readers)) {
		write_unlock(&eb->lock);
		wait_event(eb->read_lock_wq,
			   atomic_read(&eb->blocking_readers) == 0);
		goto again;
	}
	if (atomic_read(&eb->blocking_writers)) {
		write_unlock(&eb->lock);
		wait_event(eb->write_lock_wq,
			   atomic_read(&eb->blocking_writers) == 0);
		goto again;
	}
	WARN_ON(atomic_read(&eb->spinning_writers));
	atomic_inc(&eb->spinning_writers);
	atomic_inc(&eb->write_locks);
	eb->lock_owner = current->pid;
}

/*
 * drop a spinning or a blocking write lock.
 */
void btrfs_tree_unlock(struct extent_buffer *eb)
{
	int blockers = atomic_read(&eb->blocking_writers);

	BUG_ON(blockers > 1);

	btrfs_assert_tree_locked(eb);
	atomic_dec(&eb->write_locks);

	if (blockers) {
		WARN_ON(atomic_read(&eb->spinning_writers));
		atomic_dec(&eb->blocking_writers);
		smp_wmb();
		wake_up(&eb->write_lock_wq);
	} else {
		WARN_ON(atomic_read(&eb->spinning_writers) != 1);
		atomic_dec(&eb->spinning_writers);
		write_unlock(&eb->lock);
	}
}

void btrfs_assert_tree_locked(struct extent_buffer *eb)
{
	BUG_ON(!atomic_read(&eb->write_locks));
}

void btrfs_assert_tree_read_locked(struct extent_buffer *eb)
{
	BUG_ON(!atomic_read(&eb->read_locks));
}
Commit	Line	Data
925baedd CM	1	/*
	2	* Copyright (C) 2008 Oracle. All rights reserved.
	3	*
	4	* This program is free software; you can redistribute it and/or
	5	* modify it under the terms of the GNU General Public
	6	* License v2 as published by the Free Software Foundation.
	7	*
	8	* This program is distributed in the hope that it will be useful,
	9	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	10	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
	11	* General Public License for more details.
	12	*
	13	* You should have received a copy of the GNU General Public
	14	* License along with this program; if not, write to the
	15	* Free Software Foundation, Inc., 59 Temple Place - Suite 330,
	16	* Boston, MA 021110-1307, USA.
	17	*/
	18	#include <linux/sched.h>
925baedd CM	19	#include <linux/pagemap.h>
	20	#include <linux/spinlock.h>
	21	#include <linux/page-flags.h>
4881ee5a	22	#include <asm/bug.h>
925baedd CM	23	#include "ctree.h"
	24	#include "extent_io.h"
	25	#include "locking.h"
	26
bd681513	27	void btrfs_assert_tree_read_locked(struct extent_buffer *eb);
d397712b	28
b4ce94de	29	/*
bd681513 CM	30	* if we currently have a spinning reader or writer lock
	31	* (indicated by the rw flag) this will bump the count
	32	* of blocking holders and drop the spinlock.
b4ce94de	33	*/
bd681513	34	void btrfs_set_lock_blocking_rw(struct extent_buffer *eb, int rw)
925baedd	35	{
5b25f70f AJ	36	if (eb->lock_nested) {
	37	read_lock(&eb->lock);
	38	if (eb->lock_nested && current->pid == eb->lock_owner) {
	39	read_unlock(&eb->lock);
	40	return;
	41	}
	42	read_unlock(&eb->lock);
	43	}
bd681513 CM	44	if (rw == BTRFS_WRITE_LOCK) {
	45	if (atomic_read(&eb->blocking_writers) == 0) {
	46	WARN_ON(atomic_read(&eb->spinning_writers) != 1);
	47	atomic_dec(&eb->spinning_writers);
	48	btrfs_assert_tree_locked(eb);
	49	atomic_inc(&eb->blocking_writers);
	50	write_unlock(&eb->lock);
	51	}
	52	} else if (rw == BTRFS_READ_LOCK) {
	53	btrfs_assert_tree_read_locked(eb);
	54	atomic_inc(&eb->blocking_readers);
	55	WARN_ON(atomic_read(&eb->spinning_readers) == 0);
	56	atomic_dec(&eb->spinning_readers);
	57	read_unlock(&eb->lock);
b4ce94de	58	}
bd681513	59	return;
b4ce94de	60	}
f9efa9c7	61
b4ce94de	62	/*
bd681513 CM	63	* if we currently have a blocking lock, take the spinlock
bd681513 CM	64	* and drop our blocking count
b4ce94de	65	*/
bd681513	66	void btrfs_clear_lock_blocking_rw(struct extent_buffer *eb, int rw)
b4ce94de	67	{
5b25f70f AJ	68	if (eb->lock_nested) {
	69	read_lock(&eb->lock);
	70	if (&eb->lock_nested && current->pid == eb->lock_owner) {
	71	read_unlock(&eb->lock);
	72	return;
	73	}
	74	read_unlock(&eb->lock);
	75	}
bd681513 CM	76	if (rw == BTRFS_WRITE_LOCK_BLOCKING) {
	77	BUG_ON(atomic_read(&eb->blocking_writers) != 1);
	78	write_lock(&eb->lock);
	79	WARN_ON(atomic_read(&eb->spinning_writers));
	80	atomic_inc(&eb->spinning_writers);
	81	if (atomic_dec_and_test(&eb->blocking_writers))
	82	wake_up(&eb->write_lock_wq);
	83	} else if (rw == BTRFS_READ_LOCK_BLOCKING) {
	84	BUG_ON(atomic_read(&eb->blocking_readers) == 0);
	85	read_lock(&eb->lock);
	86	atomic_inc(&eb->spinning_readers);
	87	if (atomic_dec_and_test(&eb->blocking_readers))
	88	wake_up(&eb->read_lock_wq);
b4ce94de	89	}
bd681513	90	return;
b4ce94de CM	91	}
	92
	93	/*
bd681513 CM	94	* take a spinning read lock. This will wait for any blocking
bd681513 CM	95	* writers
b4ce94de	96	*/
bd681513	97	void btrfs_tree_read_lock(struct extent_buffer *eb)
b4ce94de	98	{
bd681513	99	again:
5b25f70f AJ	100	read_lock(&eb->lock);
	101	if (atomic_read(&eb->blocking_writers) &&
	102	current->pid == eb->lock_owner) {
	103	/*
	104	* This extent is already write-locked by our thread. We allow
	105	* an additional read lock to be added because it's for the same
	106	* thread. btrfs_find_all_roots() depends on this as it may be
	107	* called on a partly (write-)locked tree.
	108	*/
	109	BUG_ON(eb->lock_nested);
	110	eb->lock_nested = 1;
	111	read_unlock(&eb->lock);
	112	return;
	113	}
	114	read_unlock(&eb->lock);
bd681513 CM	115	wait_event(eb->write_lock_wq, atomic_read(&eb->blocking_writers) == 0);
	116	read_lock(&eb->lock);
	117	if (atomic_read(&eb->blocking_writers)) {
	118	read_unlock(&eb->lock);
bd681513	119	goto again;
b4ce94de	120	}
bd681513 CM	121	atomic_inc(&eb->read_locks);
bd681513 CM	122	atomic_inc(&eb->spinning_readers);
b4ce94de CM	123	}
	124
	125	/*
bd681513 CM	126	* returns 1 if we get the read lock and 0 if we don't
bd681513 CM	127	* this won't wait for blocking writers
b4ce94de	128	*/
bd681513	129	int btrfs_try_tree_read_lock(struct extent_buffer *eb)
b4ce94de	130	{
bd681513 CM	131	if (atomic_read(&eb->blocking_writers))
bd681513 CM	132	return 0;
b4ce94de	133
bd681513 CM	134	read_lock(&eb->lock);
	135	if (atomic_read(&eb->blocking_writers)) {
	136	read_unlock(&eb->lock);
	137	return 0;
b9473439	138	}
bd681513 CM	139	atomic_inc(&eb->read_locks);
	140	atomic_inc(&eb->spinning_readers);
	141	return 1;
b4ce94de CM	142	}
	143
	144	/*
bd681513 CM	145	* returns 1 if we get the read lock and 0 if we don't
bd681513 CM	146	* this won't wait for blocking writers or readers
b4ce94de	147	*/
bd681513	148	int btrfs_try_tree_write_lock(struct extent_buffer *eb)
b4ce94de	149	{
bd681513 CM	150	if (atomic_read(&eb->blocking_writers) \|\|
	151	atomic_read(&eb->blocking_readers))
	152	return 0;
	153	write_lock(&eb->lock);
	154	if (atomic_read(&eb->blocking_writers) \|\|
	155	atomic_read(&eb->blocking_readers)) {
	156	write_unlock(&eb->lock);
	157	return 0;
	158	}
	159	atomic_inc(&eb->write_locks);
	160	atomic_inc(&eb->spinning_writers);
5b25f70f	161	eb->lock_owner = current->pid;
b4ce94de CM	162	return 1;
	163	}
	164
	165	/*
bd681513 CM	166	* drop a spinning read lock
	167	*/
	168	void btrfs_tree_read_unlock(struct extent_buffer *eb)
	169	{
5b25f70f AJ	170	if (eb->lock_nested) {
	171	read_lock(&eb->lock);
	172	if (eb->lock_nested && current->pid == eb->lock_owner) {
	173	eb->lock_nested = 0;
	174	read_unlock(&eb->lock);
	175	return;
	176	}
	177	read_unlock(&eb->lock);
	178	}
bd681513 CM	179	btrfs_assert_tree_read_locked(eb);
	180	WARN_ON(atomic_read(&eb->spinning_readers) == 0);
	181	atomic_dec(&eb->spinning_readers);
	182	atomic_dec(&eb->read_locks);
	183	read_unlock(&eb->lock);
	184	}
	185
	186	/*
	187	* drop a blocking read lock
	188	*/
	189	void btrfs_tree_read_unlock_blocking(struct extent_buffer *eb)
	190	{
5b25f70f AJ	191	if (eb->lock_nested) {
	192	read_lock(&eb->lock);
	193	if (eb->lock_nested && current->pid == eb->lock_owner) {
	194	eb->lock_nested = 0;
	195	read_unlock(&eb->lock);
	196	return;
	197	}
	198	read_unlock(&eb->lock);
	199	}
bd681513 CM	200	btrfs_assert_tree_read_locked(eb);
	201	WARN_ON(atomic_read(&eb->blocking_readers) == 0);
	202	if (atomic_dec_and_test(&eb->blocking_readers))
	203	wake_up(&eb->read_lock_wq);
	204	atomic_dec(&eb->read_locks);
	205	}
	206
	207	/*
	208	* take a spinning write lock. This will wait for both
	209	* blocking readers or writers
b4ce94de	210	*/
143bede5	211	void btrfs_tree_lock(struct extent_buffer *eb)
b4ce94de	212	{
bd681513 CM	213	again:
	214	wait_event(eb->read_lock_wq, atomic_read(&eb->blocking_readers) == 0);
	215	wait_event(eb->write_lock_wq, atomic_read(&eb->blocking_writers) == 0);
	216	write_lock(&eb->lock);
	217	if (atomic_read(&eb->blocking_readers)) {
	218	write_unlock(&eb->lock);
	219	wait_event(eb->read_lock_wq,
	220	atomic_read(&eb->blocking_readers) == 0);
	221	goto again;
f9efa9c7	222	}
bd681513 CM	223	if (atomic_read(&eb->blocking_writers)) {
	224	write_unlock(&eb->lock);
	225	wait_event(eb->write_lock_wq,
	226	atomic_read(&eb->blocking_writers) == 0);
	227	goto again;
	228	}
	229	WARN_ON(atomic_read(&eb->spinning_writers));
	230	atomic_inc(&eb->spinning_writers);
	231	atomic_inc(&eb->write_locks);
5b25f70f	232	eb->lock_owner = current->pid;
925baedd CM	233	}
925baedd CM	234
bd681513 CM	235	/*
	236	* drop a spinning or a blocking write lock.
	237	*/
143bede5	238	void btrfs_tree_unlock(struct extent_buffer *eb)
925baedd	239	{
bd681513 CM	240	int blockers = atomic_read(&eb->blocking_writers);
	241
	242	BUG_ON(blockers > 1);
	243
	244	btrfs_assert_tree_locked(eb);
	245	atomic_dec(&eb->write_locks);
	246
	247	if (blockers) {
	248	WARN_ON(atomic_read(&eb->spinning_writers));
	249	atomic_dec(&eb->blocking_writers);
	250	smp_wmb();
	251	wake_up(&eb->write_lock_wq);
	252	} else {
	253	WARN_ON(atomic_read(&eb->spinning_writers) != 1);
	254	atomic_dec(&eb->spinning_writers);
	255	write_unlock(&eb->lock);
	256	}
925baedd CM	257	}
925baedd CM	258
b9447ef8	259	void btrfs_assert_tree_locked(struct extent_buffer *eb)
925baedd	260	{
bd681513 CM	261	BUG_ON(!atomic_read(&eb->write_locks));
	262	}
	263
	264	void btrfs_assert_tree_read_locked(struct extent_buffer *eb)
	265	{
	266	BUG_ON(!atomic_read(&eb->read_locks));
925baedd	267	}