[linux-2.6-block.git] / drivers / md / dm-delay.c

/*
 * Copyright (C) 2005-2007 Red Hat GmbH
 *
 * A target that delays reads and/or writes and can send
 * them to different devices.
 *
 * This file is released under the GPL.
 */

#include <linux/module.h>
#include <linux/init.h>
#include <linux/blkdev.h>
#include <linux/bio.h>
#include <linux/slab.h>

#include "dm.h"
#include "dm-bio-list.h"

#define DM_MSG_PREFIX "delay"

struct delay_c {
	struct timer_list delay_timer;
	struct semaphore timer_lock;
	struct work_struct flush_expired_bios;
	struct list_head delayed_bios;
	atomic_t may_delay;
	mempool_t *delayed_pool;

	struct dm_dev *dev_read;
	sector_t start_read;
	unsigned read_delay;
	unsigned reads;

	struct dm_dev *dev_write;
	sector_t start_write;
	unsigned write_delay;
	unsigned writes;
};

struct delay_info {
	struct delay_c *context;
	struct list_head list;
	struct bio *bio;
	unsigned long expires;
};

static DEFINE_MUTEX(delayed_bios_lock);

static struct workqueue_struct *kdelayd_wq;
static struct kmem_cache *delayed_cache;

static void handle_delayed_timer(unsigned long data)
{
	struct delay_c *dc = (struct delay_c *)data;

	queue_work(kdelayd_wq, &dc->flush_expired_bios);
}

static void queue_timeout(struct delay_c *dc, unsigned long expires)
{
	down(&dc->timer_lock);

	if (!timer_pending(&dc->delay_timer) || expires < dc->delay_timer.expires)
		mod_timer(&dc->delay_timer, expires);

	up(&dc->timer_lock);
}

static void flush_bios(struct bio *bio)
{
	struct bio *n;

	while (bio) {
		n = bio->bi_next;
		bio->bi_next = NULL;
		generic_make_request(bio);
		bio = n;
	}
}

static struct bio *flush_delayed_bios(struct delay_c *dc, int flush_all)
{
	struct delay_info *delayed, *next;
	unsigned long next_expires = 0;
	int start_timer = 0;
	BIO_LIST(flush_bios);

	mutex_lock(&delayed_bios_lock);
	list_for_each_entry_safe(delayed, next, &dc->delayed_bios, list) {
		if (flush_all || time_after_eq(jiffies, delayed->expires)) {
			list_del(&delayed->list);
			bio_list_add(&flush_bios, delayed->bio);
			if ((bio_data_dir(delayed->bio) == WRITE))
				delayed->context->writes--;
			else
				delayed->context->reads--;
			mempool_free(delayed, dc->delayed_pool);
			continue;
		}

		if (!start_timer) {
			start_timer = 1;
			next_expires = delayed->expires;
		} else
			next_expires = min(next_expires, delayed->expires);
	}

	mutex_unlock(&delayed_bios_lock);

	if (start_timer)
		queue_timeout(dc, next_expires);

	return bio_list_get(&flush_bios);
}

static void flush_expired_bios(struct work_struct *work)
{
	struct delay_c *dc;

	dc = container_of(work, struct delay_c, flush_expired_bios);
	flush_bios(flush_delayed_bios(dc, 0));
}

/*
 * Mapping parameters:
 *    <device> <offset> <delay> [<write_device> <write_offset> <write_delay>]
 *
 * With separate write parameters, the first set is only used for reads.
 * Delays are specified in milliseconds.
 */
static int delay_ctr(struct dm_target *ti, unsigned int argc, char **argv)
{
	struct delay_c *dc;
	unsigned long long tmpll;

	if (argc != 3 && argc != 6) {
		ti->error = "requires exactly 3 or 6 arguments";
		return -EINVAL;
	}

	dc = kmalloc(sizeof(*dc), GFP_KERNEL);
	if (!dc) {
		ti->error = "Cannot allocate context";
		return -ENOMEM;
	}

	dc->reads = dc->writes = 0;

	if (sscanf(argv[1], "%llu", &tmpll) != 1) {
		ti->error = "Invalid device sector";
		goto bad;
	}
	dc->start_read = tmpll;

	if (sscanf(argv[2], "%u", &dc->read_delay) != 1) {
		ti->error = "Invalid delay";
		goto bad;
	}

	if (dm_get_device(ti, argv[0], dc->start_read, ti->len,
			  dm_table_get_mode(ti->table), &dc->dev_read)) {
		ti->error = "Device lookup failed";
		goto bad;
	}

	if (argc == 3) {
		dc->dev_write = NULL;
		goto out;
	}

	if (sscanf(argv[4], "%llu", &tmpll) != 1) {
		ti->error = "Invalid write device sector";
		goto bad;
	}
	dc->start_write = tmpll;

	if (sscanf(argv[5], "%u", &dc->write_delay) != 1) {
		ti->error = "Invalid write delay";
		goto bad;
	}

	if (dm_get_device(ti, argv[3], dc->start_write, ti->len,
			  dm_table_get_mode(ti->table), &dc->dev_write)) {
		ti->error = "Write device lookup failed";
		dm_put_device(ti, dc->dev_read);
		goto bad;
	}

out:
	dc->delayed_pool = mempool_create_slab_pool(128, delayed_cache);
	if (!dc->delayed_pool) {
		DMERR("Couldn't create delayed bio pool.");
		goto bad;
	}

	init_timer(&dc->delay_timer);
	dc->delay_timer.function = handle_delayed_timer;
	dc->delay_timer.data = (unsigned long)dc;

	INIT_WORK(&dc->flush_expired_bios, flush_expired_bios);
	INIT_LIST_HEAD(&dc->delayed_bios);
	init_MUTEX(&dc->timer_lock);
	atomic_set(&dc->may_delay, 1);

	ti->private = dc;
	return 0;

bad:
	kfree(dc);
	return -EINVAL;
}

static void delay_dtr(struct dm_target *ti)
{
	struct delay_c *dc = ti->private;

	flush_workqueue(kdelayd_wq);

	dm_put_device(ti, dc->dev_read);

	if (dc->dev_write)
		dm_put_device(ti, dc->dev_write);

	mempool_destroy(dc->delayed_pool);
	kfree(dc);
}

static int delay_bio(struct delay_c *dc, int delay, struct bio *bio)
{
	struct delay_info *delayed;
	unsigned long expires = 0;

	if (!delay || !atomic_read(&dc->may_delay))
		return 1;

	delayed = mempool_alloc(dc->delayed_pool, GFP_NOIO);

	delayed->context = dc;
	delayed->bio = bio;
	delayed->expires = expires = jiffies + (delay * HZ / 1000);

	mutex_lock(&delayed_bios_lock);

	if (bio_data_dir(bio) == WRITE)
		dc->writes++;
	else
		dc->reads++;

	list_add_tail(&delayed->list, &dc->delayed_bios);

	mutex_unlock(&delayed_bios_lock);

	queue_timeout(dc, expires);

	return 0;
}

static void delay_presuspend(struct dm_target *ti)
{
	struct delay_c *dc = ti->private;

	atomic_set(&dc->may_delay, 0);
	del_timer_sync(&dc->delay_timer);
	flush_bios(flush_delayed_bios(dc, 1));
}

static void delay_resume(struct dm_target *ti)
{
	struct delay_c *dc = ti->private;

	atomic_set(&dc->may_delay, 1);
}

static int delay_map(struct dm_target *ti, struct bio *bio,
		     union map_info *map_context)
{
	struct delay_c *dc = ti->private;

	if ((bio_data_dir(bio) == WRITE) && (dc->dev_write)) {
		bio->bi_bdev = dc->dev_write->bdev;
		bio->bi_sector = dc->start_write +
				 (bio->bi_sector - ti->begin);

		return delay_bio(dc, dc->write_delay, bio);
	}

	bio->bi_bdev = dc->dev_read->bdev;
	bio->bi_sector = dc->start_read +
			 (bio->bi_sector - ti->begin);

	return delay_bio(dc, dc->read_delay, bio);
}

static int delay_status(struct dm_target *ti, status_type_t type,
			char *result, unsigned maxlen)
{
	struct delay_c *dc = ti->private;
	int sz = 0;

	switch (type) {
	case STATUSTYPE_INFO:
		DMEMIT("%u %u", dc->reads, dc->writes);
		break;

	case STATUSTYPE_TABLE:
		DMEMIT("%s %llu %u", dc->dev_read->name,
		       (unsigned long long) dc->start_read,
		       dc->read_delay);
		if (dc->dev_write)
			DMEMIT("%s %llu %u", dc->dev_write->name,
			       (unsigned long long) dc->start_write,
			       dc->write_delay);
		break;
	}

	return 0;
}

static struct target_type delay_target = {
	.name	     = "delay",
	.version     = {1, 0, 2},
	.module      = THIS_MODULE,
	.ctr	     = delay_ctr,
	.dtr	     = delay_dtr,
	.map	     = delay_map,
	.presuspend  = delay_presuspend,
	.resume	     = delay_resume,
	.status	     = delay_status,
};

static int __init dm_delay_init(void)
{
	int r = -ENOMEM;

	kdelayd_wq = create_workqueue("kdelayd");
	if (!kdelayd_wq) {
		DMERR("Couldn't start kdelayd");
		goto bad_queue;
	}

	delayed_cache = kmem_cache_create("dm-delay",
					  sizeof(struct delay_info),
					  __alignof__(struct delay_info),
					  0, NULL, NULL);
	if (!delayed_cache) {
		DMERR("Couldn't create delayed bio cache.");
		goto bad_memcache;
	}

	r = dm_register_target(&delay_target);
	if (r < 0) {
		DMERR("register failed %d", r);
		goto bad_register;
	}

	return 0;

bad_register:
	kmem_cache_destroy(delayed_cache);
bad_memcache:
	destroy_workqueue(kdelayd_wq);
bad_queue:
	return r;
}

static void __exit dm_delay_exit(void)
{
	int r = dm_unregister_target(&delay_target);

	if (r < 0)
		DMERR("unregister failed %d", r);

	kmem_cache_destroy(delayed_cache);
	destroy_workqueue(kdelayd_wq);
}

/* Module hooks */
module_init(dm_delay_init);
module_exit(dm_delay_exit);

MODULE_DESCRIPTION(DM_NAME " delay target");
MODULE_AUTHOR("Heinz Mauelshagen <mauelshagen@redhat.com>");
MODULE_LICENSE("GPL");
Commit	Line	Data
26b9f228 HM	1	/*
	2	* Copyright (C) 2005-2007 Red Hat GmbH
	3	*
	4	* A target that delays reads and/or writes and can send
	5	* them to different devices.
	6	*
	7	* This file is released under the GPL.
	8	*/
	9
	10	#include <linux/module.h>
	11	#include <linux/init.h>
	12	#include <linux/blkdev.h>
	13	#include <linux/bio.h>
	14	#include <linux/slab.h>
	15
	16	#include "dm.h"
	17	#include "dm-bio-list.h"
	18
	19	#define DM_MSG_PREFIX "delay"
	20
	21	struct delay_c {
	22	struct timer_list delay_timer;
	23	struct semaphore timer_lock;
	24	struct work_struct flush_expired_bios;
	25	struct list_head delayed_bios;
	26	atomic_t may_delay;
	27	mempool_t *delayed_pool;
	28
	29	struct dm_dev *dev_read;
	30	sector_t start_read;
	31	unsigned read_delay;
	32	unsigned reads;
	33
	34	struct dm_dev *dev_write;
	35	sector_t start_write;
	36	unsigned write_delay;
	37	unsigned writes;
	38	};
	39
	40	struct delay_info {
	41	struct delay_c *context;
	42	struct list_head list;
	43	struct bio *bio;
	44	unsigned long expires;
	45	};
	46
	47	static DEFINE_MUTEX(delayed_bios_lock);
	48
	49	static struct workqueue_struct *kdelayd_wq;
	50	static struct kmem_cache *delayed_cache;
	51
	52	static void handle_delayed_timer(unsigned long data)
	53	{
	54	struct delay_c dc = (struct delay_c )data;
	55
	56	queue_work(kdelayd_wq, &dc->flush_expired_bios);
	57	}
	58
	59	static void queue_timeout(struct delay_c *dc, unsigned long expires)
	60	{
	61	down(&dc->timer_lock);
	62
	63	if (!timer_pending(&dc->delay_timer) \|\| expires < dc->delay_timer.expires)
	64	mod_timer(&dc->delay_timer, expires);
65
66	up(&dc->timer_lock);
67	}
68
69	static void flush_bios(struct bio *bio)
70	{
71	struct bio *n;
72
73	while (bio) {
74	n = bio->bi_next;
75	bio->bi_next = NULL;
76	generic_make_request(bio);
77	bio = n;
78	}
79	}
80
81	static struct bio flush_delayed_bios(struct delay_c dc, int flush_all)
82	{
83	struct delay_info delayed, next;
84	unsigned long next_expires = 0;
85	int start_timer = 0;
86	BIO_LIST(flush_bios);
87
88	mutex_lock(&delayed_bios_lock);
89	list_for_each_entry_safe(delayed, next, &dc->delayed_bios, list) {
90	if (flush_all \|\| time_after_eq(jiffies, delayed->expires)) {
91	list_del(&delayed->list);
92	bio_list_add(&flush_bios, delayed->bio);
93	if ((bio_data_dir(delayed->bio) == WRITE))
94	delayed->context->writes--;
95	else
96	delayed->context->reads--;
97	mempool_free(delayed, dc->delayed_pool);
98	continue;
99	}
100
101	if (!start_timer) {
102	start_timer = 1;
103	next_expires = delayed->expires;
104	} else
105	next_expires = min(next_expires, delayed->expires);
106	}
107
108	mutex_unlock(&delayed_bios_lock);
109
110	if (start_timer)
111	queue_timeout(dc, next_expires);
112
113	return bio_list_get(&flush_bios);
114	}
115
116	static void flush_expired_bios(struct work_struct *work)
117	{
118	struct delay_c *dc;
119
120	dc = container_of(work, struct delay_c, flush_expired_bios);
121	flush_bios(flush_delayed_bios(dc, 0));
122	}
123
124	/*
125	* Mapping parameters:
126	* <device> <offset> <delay> [<write_device> <write_offset> <write_delay>]
127	*
128	* With separate write parameters, the first set is only used for reads.
129	* Delays are specified in milliseconds.
130	*/
131	static int delay_ctr(struct dm_target ti, unsigned int argc, char *argv)
132	{
133	struct delay_c *dc;
134	unsigned long long tmpll;
135
136	if (argc != 3 && argc != 6) {
137	ti->error = "requires exactly 3 or 6 arguments";
138	return -EINVAL;
139	}
140
141	dc = kmalloc(sizeof(*dc), GFP_KERNEL);
142	if (!dc) {
143	ti->error = "Cannot allocate context";
144	return -ENOMEM;
145	}
146
147	dc->reads = dc->writes = 0;
148
149	if (sscanf(argv[1], "%llu", &tmpll) != 1) {
150	ti->error = "Invalid device sector";
151	goto bad;
152	}
153	dc->start_read = tmpll;
154
155	if (sscanf(argv[2], "%u", &dc->read_delay) != 1) {
156	ti->error = "Invalid delay";
157	goto bad;
158	}
159
160	if (dm_get_device(ti, argv[0], dc->start_read, ti->len,
161	dm_table_get_mode(ti->table), &dc->dev_read)) {
162	ti->error = "Device lookup failed";
163	goto bad;
164	}
165
166	if (argc == 3) {
167	dc->dev_write = NULL;
168	goto out;
169	}
170
171	if (sscanf(argv[4], "%llu", &tmpll) != 1) {
172	ti->error = "Invalid write device sector";
173	goto bad;
174	}
175	dc->start_write = tmpll;
176
177	if (sscanf(argv[5], "%u", &dc->write_delay) != 1) {
178	ti->error = "Invalid write delay";
179	goto bad;
180	}
181
182	if (dm_get_device(ti, argv[3], dc->start_write, ti->len,
183	dm_table_get_mode(ti->table), &dc->dev_write)) {
184	ti->error = "Write device lookup failed";
185	dm_put_device(ti, dc->dev_read);
186	goto bad;
187	}
188
189	out:
190	dc->delayed_pool = mempool_create_slab_pool(128, delayed_cache);
191	if (!dc->delayed_pool) {
192	DMERR("Couldn't create delayed bio pool.");
193	goto bad;
194	}
195
196	init_timer(&dc->delay_timer);
197	dc->delay_timer.function = handle_delayed_timer;
198	dc->delay_timer.data = (unsigned long)dc;
199
200	INIT_WORK(&dc->flush_expired_bios, flush_expired_bios);
201	INIT_LIST_HEAD(&dc->delayed_bios);
202	init_MUTEX(&dc->timer_lock);
203	atomic_set(&dc->may_delay, 1);
204
205	ti->private = dc;
206	return 0;
207
208	bad:
209	kfree(dc);
210	return -EINVAL;
211	}
212
213	static void delay_dtr(struct dm_target *ti)
214	{
215	struct delay_c *dc = ti->private;
216
217	flush_workqueue(kdelayd_wq);
218
219	dm_put_device(ti, dc->dev_read);
220
221	if (dc->dev_write)
222	dm_put_device(ti, dc->dev_write);
223
224	mempool_destroy(dc->delayed_pool);
225	kfree(dc);
226	}
227
228	static int delay_bio(struct delay_c dc, int delay, struct bio bio)
229	{
230	struct delay_info *delayed;
231	unsigned long expires = 0;
232
233	if (!delay \|\| !atomic_read(&dc->may_delay))
234	return 1;
235
236	delayed = mempool_alloc(dc->delayed_pool, GFP_NOIO);
237
238	delayed->context = dc;
239	delayed->bio = bio;
240	delayed->expires = expires = jiffies + (delay * HZ / 1000);
241
242	mutex_lock(&delayed_bios_lock);
243
244	if (bio_data_dir(bio) == WRITE)
245	dc->writes++;
246	else
247	dc->reads++;
248
249	list_add_tail(&delayed->list, &dc->delayed_bios);
250
251	mutex_unlock(&delayed_bios_lock);
252
253	queue_timeout(dc, expires);
254
255	return 0;
256	}
257
258	static void delay_presuspend(struct dm_target *ti)
259	{
260	struct delay_c *dc = ti->private;
261
262	atomic_set(&dc->may_delay, 0);
263	del_timer_sync(&dc->delay_timer);
264	flush_bios(flush_delayed_bios(dc, 1));
265	}
266
267	static void delay_resume(struct dm_target *ti)
268	{
269	struct delay_c *dc = ti->private;
270
271	atomic_set(&dc->may_delay, 1);
272	}
273
274	static int delay_map(struct dm_target ti, struct bio bio,
275	union map_info *map_context)
276	{
277	struct delay_c *dc = ti->private;
278
279	if ((bio_data_dir(bio) == WRITE) && (dc->dev_write)) {
280	bio->bi_bdev = dc->dev_write->bdev;
281	bio->bi_sector = dc->start_write +
282	(bio->bi_sector - ti->begin);
283
284	return delay_bio(dc, dc->write_delay, bio);
285	}
286
287	bio->bi_bdev = dc->dev_read->bdev;
288	bio->bi_sector = dc->start_read +
289	(bio->bi_sector - ti->begin);
290
291	return delay_bio(dc, dc->read_delay, bio);
292	}
293
294	static int delay_status(struct dm_target *ti, status_type_t type,
295	char *result, unsigned maxlen)
296	{
297	struct delay_c *dc = ti->private;
298	int sz = 0;
299
300	switch (type) {
301	case STATUSTYPE_INFO:
302	DMEMIT("%u %u", dc->reads, dc->writes);
303	break;
304
305	case STATUSTYPE_TABLE:
306	DMEMIT("%s %llu %u", dc->dev_read->name,
307	(unsigned long long) dc->start_read,
308	dc->read_delay);
309	if (dc->dev_write)
310	DMEMIT("%s %llu %u", dc->dev_write->name,
311	(unsigned long long) dc->start_write,
312	dc->write_delay);
313	break;
314	}
315
316	return 0;
317	}
318
319	static struct target_type delay_target = {
320	.name = "delay",
321	.version = {1, 0, 2},
322	.module = THIS_MODULE,
323	.ctr = delay_ctr,
324	.dtr = delay_dtr,
325	.map = delay_map,
326	.presuspend = delay_presuspend,
327	.resume = delay_resume,
328	.status = delay_status,
329	};
330
331	static int __init dm_delay_init(void)
332	{
333	int r = -ENOMEM;
334
335	kdelayd_wq = create_workqueue("kdelayd");
336	if (!kdelayd_wq) {
337	DMERR("Couldn't start kdelayd");
338	goto bad_queue;
339	}
340
341	delayed_cache = kmem_cache_create("dm-delay",
342	sizeof(struct delay_info),
343	__alignof__(struct delay_info),
344	0, NULL, NULL);
345	if (!delayed_cache) {
346	DMERR("Couldn't create delayed bio cache.");
347	goto bad_memcache;
348	}
349
350	r = dm_register_target(&delay_target);
351	if (r < 0) {
352	DMERR("register failed %d", r);
353	goto bad_register;
354	}
355
356	return 0;
357
358	bad_register:
359	kmem_cache_destroy(delayed_cache);
360	bad_memcache:
361	destroy_workqueue(kdelayd_wq);
362	bad_queue:
363	return r;
364	}
365
366	static void __exit dm_delay_exit(void)
367	{
368	int r = dm_unregister_target(&delay_target);
369
370	if (r < 0)
371	DMERR("unregister failed %d", r);
372
373	kmem_cache_destroy(delayed_cache);
374	destroy_workqueue(kdelayd_wq);
375	}
376
377	/* Module hooks */
378	module_init(dm_delay_init);
379	module_exit(dm_delay_exit);
380
381	MODULE_DESCRIPTION(DM_NAME " delay target");
382	MODULE_AUTHOR("Heinz Mauelshagen <mauelshagen@redhat.com>");
383	MODULE_LICENSE("GPL");