Release 4.14 drivers/md/multipath.c

/*
 * multipath.c : Multiple Devices driver for Linux
 *
 * Copyright (C) 1999, 2000, 2001 Ingo Molnar, Red Hat
 *
 * Copyright (C) 1996, 1997, 1998 Ingo Molnar, Miguel de Icaza, Gadi Oxman
 *
 * MULTIPATH management functions.
 *
 * derived from raid1.c.
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 2, or (at your option)
 * any later version.
 *
 * You should have received a copy of the GNU General Public License
 * (for example /usr/src/linux/COPYING); if not, write to the Free
 * Software Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
 */

#include <linux/blkdev.h>
#include <linux/module.h>
#include <linux/raid/md_u.h>
#include <linux/seq_file.h>
#include <linux/slab.h>
#include "md.h"
#include "multipath.h"


#define MAX_WORK_PER_DISK 128


#define	NR_RESERVED_BUFS	32



static int multipath_map (struct mpconf *conf)
{
	int i, disks = conf->raid_disks;

	/*
         * Later we do read balancing on the read side
         * now we use the first available disk.
         */

	rcu_read_lock();
	for (i = 0; i < disks; i++) {
		struct md_rdev *rdev = rcu_dereference(conf->multipaths[i].rdev);
		if (rdev && test_bit(In_sync, &rdev->flags) &&
		    !test_bit(Faulty, &rdev->flags)) {
			atomic_inc(&rdev->nr_pending);
			rcu_read_unlock();
			return i;
		}
	}
	rcu_read_unlock();

	pr_crit_ratelimited("multipath_map(): no more operational IO paths?\n");
	return (-1);
}

Contributors
Person
Tokens
Prop
Commits
CommitProp
Neil Brown
57
51.35%
13
86.67%
Linus Torvalds
51
45.95%
1
6.67%
Suzanne Wood
3
2.70%
1
6.67%
Total
111
100.00%
15
100.00%



static void multipath_reschedule_retry (struct multipath_bh *mp_bh)
{
	unsigned long flags;
	struct mddev *mddev = mp_bh->mddev;
	struct mpconf *conf = mddev->private;

	spin_lock_irqsave(&conf->device_lock, flags);
	list_add(&mp_bh->retry_list, &conf->retry_list);
	spin_unlock_irqrestore(&conf->device_lock, flags);
	md_wakeup_thread(mddev->thread);
}

Contributors
Person
Tokens
Prop
Commits
CommitProp
Linus Torvalds
45
61.64%
2
25.00%
Neil Brown
26
35.62%
5
62.50%
Al Viro
2
2.74%
1
12.50%
Total
73
100.00%
8
100.00%

/*
 * multipath_end_bh_io() is called when we have finished servicing a multipathed
 * operation and are ready to return a success/failure code to the buffer
 * cache layer.
 */


static void multipath_end_bh_io(struct multipath_bh *mp_bh, blk_status_t status)
{
	struct bio *bio = mp_bh->master_bio;
	struct mpconf *conf = mp_bh->mddev->private;

	bio->bi_status = status;
	bio_endio(bio);
	mempool_free(mp_bh, conf->pool);
}

Contributors
Person
Tokens
Prop
Commits
CommitProp
Linus Torvalds
25
46.30%
2
25.00%
Neil Brown
16
29.63%
3
37.50%
Christoph Hellwig
8
14.81%
2
25.00%
Al Viro
5
9.26%
1
12.50%
Total
54
100.00%
8
100.00%



static void multipath_end_request(struct bio *bio)
{
	struct multipath_bh *mp_bh = bio->bi_private;
	struct mpconf *conf = mp_bh->mddev->private;
	struct md_rdev *rdev = conf->multipaths[mp_bh->path].rdev;

	if (!bio->bi_status)
		multipath_end_bh_io(mp_bh, 0);
	else if (!(bio->bi_opf & REQ_RAHEAD)) {
		/*
                 * oops, IO error:
                 */
		char b[BDEVNAME_SIZE];
		md_error (mp_bh->mddev, rdev);
		pr_info("multipath: %s: rescheduling sector %llu\n",
			bdevname(rdev->bdev,b),
			(unsigned long long)bio->bi_iter.bi_sector);
		multipath_reschedule_retry(mp_bh);
	} else
		multipath_end_bh_io(mp_bh, bio->bi_status);
	rdev_dec_pending(rdev, conf->mddev);
}

Contributors
Person
Tokens
Prop
Commits
CommitProp
Neil Brown
72
51.43%
13
52.00%
Linus Torvalds
27
19.29%
2
8.00%
Jens Axboe
14
10.00%
3
12.00%
Christoph Hellwig
12
8.57%
3
12.00%
Andrew Morton
6
4.29%
1
4.00%
Al Viro
6
4.29%
1
4.00%
Kent Overstreet
2
1.43%
1
4.00%
Adrian Bunk
1
0.71%
1
4.00%
Total
140
100.00%
25
100.00%



static bool multipath_make_request(struct mddev *mddev, struct bio * bio)
{
	struct mpconf *conf = mddev->private;
	struct multipath_bh * mp_bh;
	struct multipath_info *multipath;

	if (unlikely(bio->bi_opf & REQ_PREFLUSH)) {
		md_flush_request(mddev, bio);
		return true;
	}

	mp_bh = mempool_alloc(conf->pool, GFP_NOIO);

	mp_bh->master_bio = bio;
	mp_bh->mddev = mddev;

	mp_bh->path = multipath_map(conf);
	if (mp_bh->path < 0) {
		bio_io_error(bio);
		mempool_free(mp_bh, conf->pool);
		return true;
	}
	multipath = conf->multipaths + mp_bh->path;

	bio_init(&mp_bh->bio, NULL, 0);
	__bio_clone_fast(&mp_bh->bio, bio);

	mp_bh->bio.bi_iter.bi_sector += multipath->rdev->data_offset;
	bio_set_dev(&mp_bh->bio, multipath->rdev->bdev);
	mp_bh->bio.bi_opf |= REQ_FAILFAST_TRANSPORT;
	mp_bh->bio.bi_end_io = multipath_end_request;
	mp_bh->bio.bi_private = mp_bh;
	mddev_check_writesame(mddev, &mp_bh->bio);
	mddev_check_write_zeroes(mddev, &mp_bh->bio);
	generic_make_request(&mp_bh->bio);
	return true;
}

Contributors
Person
Tokens
Prop
Commits
CommitProp
Neil Brown
100
43.10%
15
48.39%
Linus Torvalds
63
27.16%
2
6.45%
Christoph Hellwig
19
8.19%
4
12.90%
Ming Lei
13
5.60%
1
3.23%
Al Viro
11
4.74%
2
6.45%
Shaohua Li
10
4.31%
1
3.23%
Jens Axboe
8
3.45%
2
6.45%
Lei Ming
4
1.72%
1
3.23%
Kent Overstreet
2
0.86%
1
3.23%
Michael Christie
1
0.43%
1
3.23%
Tejun Heo
1
0.43%
1
3.23%
Total
232
100.00%
31
100.00%



static void multipath_status(struct seq_file *seq, struct mddev *mddev)
{
	struct mpconf *conf = mddev->private;
	int i;

	seq_printf (seq, " [%d/%d] [", conf->raid_disks,
		    conf->raid_disks - mddev->degraded);
	rcu_read_lock();
	for (i = 0; i < conf->raid_disks; i++) {
		struct md_rdev *rdev = rcu_dereference(conf->multipaths[i].rdev);
		seq_printf (seq, "%s", rdev && test_bit(In_sync, &rdev->flags) ? "U" : "_");
	}
	rcu_read_unlock();
	seq_printf (seq, "]");
}

Contributors
Person
Tokens
Prop
Commits
CommitProp
Linus Torvalds
64
54.70%
1
11.11%
Neil Brown
53
45.30%
8
88.89%
Total
117
100.00%
9
100.00%



static int multipath_congested(struct mddev *mddev, int bits)
{
	struct mpconf *conf = mddev->private;
	int i, ret = 0;

	rcu_read_lock();
	for (i = 0; i < mddev->raid_disks ; i++) {
		struct md_rdev *rdev = rcu_dereference(conf->multipaths[i].rdev);
		if (rdev && !test_bit(Faulty, &rdev->flags)) {
			struct request_queue *q = bdev_get_queue(rdev->bdev);

			ret |= bdi_congested(q->backing_dev_info, bits);
			/* Just like multipath_map, we just check the
                         * first available device
                         */
			break;
		}
	}
	rcu_read_unlock();
	return ret;
}

Contributors
Person
Tokens
Prop
Commits
CommitProp
Neil Brown
113
98.26%
6
85.71%
Jens Axboe
2
1.74%
1
14.29%
Total
115
100.00%
7
100.00%

/*
 * Careful, this can execute in IRQ contexts as well!
 */


static void multipath_error (struct mddev *mddev, struct md_rdev *rdev)
{
	struct mpconf *conf = mddev->private;
	char b[BDEVNAME_SIZE];

	if (conf->raid_disks - mddev->degraded <= 1) {
		/*
                 * Uh oh, we can do nothing if this is our last path, but
                 * first check if this is a queued request for a device
                 * which has just failed.
                 */
		pr_warn("multipath: only one IO path left and IO error.\n");
		/* leave it active... it's all we have */
		return;
	}
	/*
         * Mark disk as unusable
         */
	if (test_and_clear_bit(In_sync, &rdev->flags)) {
		unsigned long flags;
		spin_lock_irqsave(&conf->device_lock, flags);
		mddev->degraded++;
		spin_unlock_irqrestore(&conf->device_lock, flags);
	}
	set_bit(Faulty, &rdev->flags);
	set_bit(MD_SB_CHANGE_DEVS, &mddev->sb_flags);
	pr_err("multipath: IO failure on %s, disabling IO path.\n"
	       "multipath: Operation continuing on %d IO paths.\n",
	       bdevname(rdev->bdev, b),
	       conf->raid_disks - mddev->degraded);
}

Contributors
Person
Tokens
Prop
Commits
CommitProp
Neil Brown
103
73.57%
14
82.35%
Linus Torvalds
34
24.29%
1
5.88%
Shaohua Li
2
1.43%
1
5.88%
Al Viro
1
0.71%
1
5.88%
Total
140
100.00%
17
100.00%



static void print_multipath_conf (struct mpconf *conf)
{
	int i;
	struct multipath_info *tmp;

	pr_debug("MULTIPATH conf printout:\n");
	if (!conf) {
		pr_debug("(conf==NULL)\n");
		return;
	}
	pr_debug(" --- wd:%d rd:%d\n", conf->raid_disks - conf->mddev->degraded,
		 conf->raid_disks);

	for (i = 0; i < conf->raid_disks; i++) {
		char b[BDEVNAME_SIZE];
		tmp = conf->multipaths + i;
		if (tmp->rdev)
			pr_debug(" disk%d, o:%d, dev:%s\n",
				 i,!test_bit(Faulty, &tmp->rdev->flags),
				 bdevname(tmp->rdev->bdev,b));
	}
}

Contributors
Person
Tokens
Prop
Commits
CommitProp
Linus Torvalds
83
66.94%
1
7.69%
Neil Brown
40
32.26%
11
84.62%
Al Viro
1
0.81%
1
7.69%
Total
124
100.00%
13
100.00%



static int multipath_add_disk(struct mddev *mddev, struct md_rdev *rdev)
{
	struct mpconf *conf = mddev->private;
	struct request_queue *q;
	int err = -EEXIST;
	int path;
	struct multipath_info *p;
	int first = 0;
	int last = mddev->raid_disks - 1;

	if (rdev->raid_disk >= 0)
		first = last = rdev->raid_disk;

	print_multipath_conf(conf);

	for (path = first; path <= last; path++)
		if ((p=conf->multipaths+path)->rdev == NULL) {
			q = rdev->bdev->bd_disk->queue;
			disk_stack_limits(mddev->gendisk, rdev->bdev,
					  rdev->data_offset << 9);

			err = md_integrity_add_rdev(rdev, mddev);
			if (err)
				break;
			spin_lock_irq(&conf->device_lock);
			mddev->degraded--;
			rdev->raid_disk = path;
			set_bit(In_sync, &rdev->flags);
			spin_unlock_irq(&conf->device_lock);
			rcu_assign_pointer(p->rdev, rdev);
			err = 0;
			break;
		}

	print_multipath_conf(conf);

	return err;
}

Contributors
Person
Tokens
Prop
Commits
CommitProp
Neil Brown
129
61.43%
14
66.67%
Linus Torvalds
18
8.57%
1
4.76%
Dan J Williams
14
6.67%
1
4.76%
Al Viro
13
6.19%
1
4.76%
Jesper Juhl
12
5.71%
1
4.76%
Martin K. Petersen
11
5.24%
1
4.76%
Andrew Morton
9
4.29%
1
4.76%
Suzanne Wood
4
1.90%
1
4.76%
Total
210
100.00%
21
100.00%



static int multipath_remove_disk(struct mddev *mddev, struct md_rdev *rdev)
{
	struct mpconf *conf = mddev->private;
	int err = 0;
	int number = rdev->raid_disk;
	struct multipath_info *p = conf->multipaths + number;

	print_multipath_conf(conf);

	if (rdev == p->rdev) {
		if (test_bit(In_sync, &rdev->flags) ||
		    atomic_read(&rdev->nr_pending)) {
			pr_warn("hot-remove-disk, slot %d is identified but is still operational!\n", number);
			err = -EBUSY;
			goto abort;
		}
		p->rdev = NULL;
		if (!test_bit(RemoveSynchronized, &rdev->flags)) {
			synchronize_rcu();
			if (atomic_read(&rdev->nr_pending)) {
				/* lost the race, try later */
				err = -EBUSY;
				p->rdev = rdev;
				goto abort;
			}
		}
		err = md_integrity_register(mddev);
	}
abort:

	print_multipath_conf(conf);
	return err;
}

Contributors
Person
Tokens
Prop
Commits
CommitProp
Neil Brown
88
52.38%
11
64.71%
Al Viro
39
23.21%
2
11.76%
Linus Torvalds
30
17.86%
1
5.88%
Andre Noll
8
4.76%
1
5.88%
Martin K. Petersen
2
1.19%
1
5.88%
Paul E. McKenney
1
0.60%
1
5.88%
Total
168
100.00%
17
100.00%

/*
 * This is a kernel thread which:
 *
 *      1.      Retries failed read operations on working multipaths.
 *      2.      Updates the raid superblock when problems encounter.
 *      3.      Performs writes following reads for array syncronising.
 */



static void multipathd(struct md_thread *thread)
{
	struct mddev *mddev = thread->mddev;
	struct multipath_bh *mp_bh;
	struct bio *bio;
	unsigned long flags;
	struct mpconf *conf = mddev->private;
	struct list_head *head = &conf->retry_list;

	md_check_recovery(mddev);
	for (;;) {
		char b[BDEVNAME_SIZE];
		spin_lock_irqsave(&conf->device_lock, flags);
		if (list_empty(head))
			break;
		mp_bh = list_entry(head->prev, struct multipath_bh, retry_list);
		list_del(head->prev);
		spin_unlock_irqrestore(&conf->device_lock, flags);

		bio = &mp_bh->bio;
		bio->bi_iter.bi_sector = mp_bh->master_bio->bi_iter.bi_sector;

		if ((mp_bh->path = multipath_map (conf))<0) {
			pr_err("multipath: %s: unrecoverable IO read error for block %llu\n",
			       bio_devname(bio, b),
			       (unsigned long long)bio->bi_iter.bi_sector);
			multipath_end_bh_io(mp_bh, BLK_STS_IOERR);
		} else {
			pr_err("multipath: %s: redirecting sector %llu to another IO path\n",
			       bio_devname(bio, b),
			       (unsigned long long)bio->bi_iter.bi_sector);
			*bio = *(mp_bh->master_bio);
			bio->bi_iter.bi_sector +=
				conf->multipaths[mp_bh->path].rdev->data_offset;
			bio_set_dev(bio, conf->multipaths[mp_bh->path].rdev->bdev);
			bio->bi_opf |= REQ_FAILFAST_TRANSPORT;
			bio->bi_end_io = multipath_end_request;
			bio->bi_private = mp_bh;
			generic_make_request(bio);
		}
	}
	spin_unlock_irqrestore(&conf->device_lock, flags);
}

Contributors
Person
Tokens
Prop
Commits
CommitProp
Neil Brown
121
41.16%
11
47.83%
Linus Torvalds
92
31.29%
2
8.70%
Heiko Carstens
26
8.84%
1
4.35%
Al Viro
15
5.10%
2
8.70%
Shaohua Li
11
3.74%
1
4.35%
Andrew Morton
10
3.40%
1
4.35%
Kent Overstreet
10
3.40%
1
4.35%
Christoph Hellwig
8
2.72%
3
13.04%
Jens Axboe
1
0.34%
1
4.35%
Total
294
100.00%
23
100.00%



static sector_t multipath_size(struct mddev *mddev, sector_t sectors, int raid_disks)
{
	WARN_ONCE(sectors || raid_disks,
		  "%s does not support generic reshape\n", __func__);

	return mddev->dev_sectors;
}

Contributors
Person
Tokens
Prop
Commits
CommitProp
Dan J Williams
31
93.94%
1
50.00%
Neil Brown
2
6.06%
1
50.00%
Total
33
100.00%
2
100.00%



static int multipath_run (struct mddev *mddev)
{
	struct mpconf *conf;
	int disk_idx;
	struct multipath_info *disk;
	struct md_rdev *rdev;
	int working_disks;

	if (md_check_no_bitmap(mddev))
		return -EINVAL;

	if (mddev->level != LEVEL_MULTIPATH) {
		pr_warn("multipath: %s: raid level not set to multipath IO (%d)\n",
			mdname(mddev), mddev->level);
		goto out;
	}
	/*
         * copy the already verified devices into our private MULTIPATH
         * bookkeeping area. [whatever we allocate in multipath_run(),
         * should be freed in multipath_free()]
         */

	conf = kzalloc(sizeof(struct mpconf), GFP_KERNEL);
	mddev->private = conf;
	if (!conf)
		goto out;

	conf->multipaths = kzalloc(sizeof(struct multipath_info)*mddev->raid_disks,
				   GFP_KERNEL);
	if (!conf->multipaths)
		goto out_free_conf;

	working_disks = 0;
	rdev_for_each(rdev, mddev) {
		disk_idx = rdev->raid_disk;
		if (disk_idx < 0 ||
		    disk_idx >= mddev->raid_disks)
			continue;

		disk = conf->multipaths + disk_idx;
		disk->rdev = rdev;
		disk_stack_limits(mddev->gendisk, rdev->bdev,
				  rdev->data_offset << 9);

		if (!test_bit(Faulty, &rdev->flags))
			working_disks++;
	}

	conf->raid_disks = mddev->raid_disks;
	conf->mddev = mddev;
	spin_lock_init(&conf->device_lock);
	INIT_LIST_HEAD(&conf->retry_list);

	if (!working_disks) {
		pr_warn("multipath: no operational IO paths for %s\n",
			mdname(mddev));
		goto out_free_conf;
	}
	mddev->degraded = conf->raid_disks - working_disks;

	conf->pool = mempool_create_kmalloc_pool(NR_RESERVED_BUFS,
						 sizeof(struct multipath_bh));
	if (conf->pool == NULL)
		goto out_free_conf;

	mddev->thread = md_register_thread(multipathd, mddev,
					   "multipath");
	if (!mddev->thread)
		goto out_free_conf;

	pr_info("multipath: array %s active with %d out of %d IO paths\n",
		mdname(mddev), conf->raid_disks - mddev->degraded,
		mddev->raid_disks);
	/*
         * Ok, everything is just fine now
         */
	md_set_array_sectors(mddev, multipath_size(mddev, 0, 0));

	if (md_integrity_register(mddev))
		goto out_free_conf;

	return 0;

out_free_conf:
	mempool_destroy(conf->pool);
	kfree(conf->multipaths);
	kfree(conf);
	mddev->private = NULL;
out:
	return -EIO;
}

Contributors
Person
Tokens
Prop
Commits
CommitProp
Linus Torvalds
187
47.10%
2
5.71%
Neil Brown
145
36.52%
22
62.86%
Andrew Morton
16
4.03%
2
5.71%
Andre Noll
15
3.78%
2
5.71%
Martin K. Petersen
13
3.27%
2
5.71%
Dan J Williams
11
2.77%
2
5.71%
Matthew Dobson
5
1.26%
1
2.86%
Thomas Gleixner
4
1.01%
1
2.86%
Sage Weil
1
0.25%
1
2.86%
Total
397
100.00%
35
100.00%



static void multipath_free(struct mddev *mddev, void *priv)
{
	struct mpconf *conf = priv;

	mempool_destroy(conf->pool);
	kfree(conf->multipaths);
	kfree(conf);
}

Contributors
Person
Tokens
Prop
Commits
CommitProp
Neil Brown
21
51.22%
5
71.43%
Linus Torvalds
19
46.34%
1
14.29%
Al Viro
1
2.44%
1
14.29%
Total
41
100.00%
7
100.00%


static struct md_personality multipath_personality =
{
	.name		= "multipath",
	.level		= LEVEL_MULTIPATH,
	.owner		= THIS_MODULE,
	.make_request	= multipath_make_request,
	.run		= multipath_run,
	.free		= multipath_free,
	.status		= multipath_status,
	.error_handler	= multipath_error,
	.hot_add_disk	= multipath_add_disk,
	.hot_remove_disk= multipath_remove_disk,
	.size		= multipath_size,
	.congested	= multipath_congested,
};



static int __init multipath_init (void)
{
	return register_md_personality (&multipath_personality);
}

Contributors
Person
Tokens
Prop
Commits
CommitProp
Linus Torvalds
15
93.75%
1
50.00%
Al Viro
1
6.25%
1
50.00%
Total
16
100.00%
2
100.00%



static void __exit multipath_exit (void)
{
	unregister_md_personality (&multipath_personality);
}

Contributors
Person
Tokens
Prop
Commits
CommitProp
Linus Torvalds
12
80.00%
1
33.33%
Neil Brown
2
13.33%
1
33.33%
Al Viro
1
6.67%
1
33.33%
Total
15
100.00%
3
100.00%


module_init(multipath_init);

module_exit(multipath_exit);
MODULE_LICENSE("GPL");
MODULE_DESCRIPTION("simple multi-path personality for MD");
MODULE_ALIAS("md-personality-7"); /* MULTIPATH */
MODULE_ALIAS("md-multipath");
MODULE_ALIAS("md-level--4");
Overall Contributors
Person
Tokens
Prop
Commits
CommitProp
Neil Brown
1156
47.85%
69
59.48%
Linus Torvalds
816
33.77%
3
2.59%
Al Viro
101
4.18%
5
4.31%
Dan J Williams
61
2.52%
3
2.59%
Christoph Hellwig
48
1.99%
6
5.17%
Andrew Morton
41
1.70%
3
2.59%
Martin K. Petersen
26
1.08%
2
1.72%
Heiko Carstens
26
1.08%
1
0.86%
Jens Axboe
25
1.03%
5
4.31%
Shaohua Li
23
0.95%
3
2.59%
Andre Noll
23
0.95%
2
1.72%
Kent Overstreet
14
0.58%
1
0.86%
Ming Lei
13
0.54%
1
0.86%
Jesper Juhl
12
0.50%
1
0.86%
Suzanne Wood
7
0.29%
1
0.86%
Matthew Dobson
5
0.21%
1
0.86%
Tejun Heo
4
0.17%
2
1.72%
Thomas Gleixner
4
0.17%
1
0.86%
Lei Ming
4
0.17%
1
0.86%
Paul Gortmaker
3
0.12%
1
0.86%
Adrian Bunk
1
0.04%
1
0.86%
Michael Christie
1
0.04%
1
0.86%
Sage Weil
1
0.04%
1
0.86%
Paul E. McKenney
1
0.04%
1
0.86%
Total
2416
100.00%
116
100.00%
Directory: drivers/md

Information contained on this website is for historical information purposes only and does not indicate or represent copyright ownership.