[linux.git] / drivers / md / dm-raid.c

/*
 * Copyright (C) 2010-2011 Neil Brown
 * Copyright (C) 2010-2011 Red Hat, Inc. All rights reserved.
 *
 * This file is released under the GPL.
 */

#include <linux/slab.h>

#include "md.h"
#include "raid5.h"
#include "dm.h"
#include "bitmap.h"

#define DM_MSG_PREFIX "raid"

/*
 * If the MD doesn't support MD_SYNC_STATE_FORCED yet, then
 * make it so the flag doesn't set anything.
 */
#ifndef MD_SYNC_STATE_FORCED
#define MD_SYNC_STATE_FORCED 0
#endif

struct raid_dev {
	/*
	 * Two DM devices, one to hold metadata and one to hold the
	 * actual data/parity.  The reason for this is to not confuse
	 * ti->len and give more flexibility in altering size and
	 * characteristics.
	 *
	 * While it is possible for this device to be associated
	 * with a different physical device than the data_dev, it
	 * is intended for it to be the same.
	 *    |--------- Physical Device ---------|
	 *    |- meta_dev -|------ data_dev ------|
	 */
	struct dm_dev *meta_dev;
	struct dm_dev *data_dev;
	struct mdk_rdev_s rdev;
};

/*
 * Flags for rs->print_flags field.
 */
#define DMPF_DAEMON_SLEEP      0x1
#define DMPF_MAX_WRITE_BEHIND  0x2
#define DMPF_SYNC              0x4
#define DMPF_NOSYNC            0x8
#define DMPF_STRIPE_CACHE      0x10
#define DMPF_MIN_RECOVERY_RATE 0x20
#define DMPF_MAX_RECOVERY_RATE 0x40

struct raid_set {
	struct dm_target *ti;

	uint64_t print_flags;

	struct mddev_s md;
	struct raid_type *raid_type;
	struct dm_target_callbacks callbacks;

	struct raid_dev dev[0];
};

/* Supported raid types and properties. */
static struct raid_type {
	const char *name;		/* RAID algorithm. */
	const char *descr;		/* Descriptor text for logging. */
	const unsigned parity_devs;	/* # of parity devices. */
	const unsigned minimal_devs;	/* minimal # of devices in set. */
	const unsigned level;		/* RAID level. */
	const unsigned algorithm;	/* RAID algorithm. */
} raid_types[] = {
	{"raid4",    "RAID4 (dedicated parity disk)",	1, 2, 5, ALGORITHM_PARITY_0},
	{"raid5_la", "RAID5 (left asymmetric)",		1, 2, 5, ALGORITHM_LEFT_ASYMMETRIC},
	{"raid5_ra", "RAID5 (right asymmetric)",	1, 2, 5, ALGORITHM_RIGHT_ASYMMETRIC},
	{"raid5_ls", "RAID5 (left symmetric)",		1, 2, 5, ALGORITHM_LEFT_SYMMETRIC},
	{"raid5_rs", "RAID5 (right symmetric)",		1, 2, 5, ALGORITHM_RIGHT_SYMMETRIC},
	{"raid6_zr", "RAID6 (zero restart)",		2, 4, 6, ALGORITHM_ROTATING_ZERO_RESTART},
	{"raid6_nr", "RAID6 (N restart)",		2, 4, 6, ALGORITHM_ROTATING_N_RESTART},
	{"raid6_nc", "RAID6 (N continue)",		2, 4, 6, ALGORITHM_ROTATING_N_CONTINUE}
};

static struct raid_type *get_raid_type(char *name)
{
	int i;

	for (i = 0; i < ARRAY_SIZE(raid_types); i++)
		if (!strcmp(raid_types[i].name, name))
			return &raid_types[i];

	return NULL;
}

static struct raid_set *context_alloc(struct dm_target *ti, struct raid_type *raid_type, unsigned raid_devs)
{
	unsigned i;
	struct raid_set *rs;
	sector_t sectors_per_dev;

	if (raid_devs <= raid_type->parity_devs) {
		ti->error = "Insufficient number of devices";
		return ERR_PTR(-EINVAL);
	}

	sectors_per_dev = ti->len;
	if (sector_div(sectors_per_dev, (raid_devs - raid_type->parity_devs))) {
		ti->error = "Target length not divisible by number of data devices";
		return ERR_PTR(-EINVAL);
	}

	rs = kzalloc(sizeof(*rs) + raid_devs * sizeof(rs->dev[0]), GFP_KERNEL);
	if (!rs) {
		ti->error = "Cannot allocate raid context";
		return ERR_PTR(-ENOMEM);
	}

	mddev_init(&rs->md);

	rs->ti = ti;
	rs->raid_type = raid_type;
	rs->md.raid_disks = raid_devs;
	rs->md.level = raid_type->level;
	rs->md.new_level = rs->md.level;
	rs->md.dev_sectors = sectors_per_dev;
	rs->md.layout = raid_type->algorithm;
	rs->md.new_layout = rs->md.layout;
	rs->md.delta_disks = 0;
	rs->md.recovery_cp = 0;

	for (i = 0; i < raid_devs; i++)
		md_rdev_init(&rs->dev[i].rdev);

	/*
	 * Remaining items to be initialized by further RAID params:
	 *  rs->md.persistent
	 *  rs->md.external
	 *  rs->md.chunk_sectors
	 *  rs->md.new_chunk_sectors
	 */

	return rs;
}

static void context_free(struct raid_set *rs)
{
	int i;

	for (i = 0; i < rs->md.raid_disks; i++)
		if (rs->dev[i].data_dev)
			dm_put_device(rs->ti, rs->dev[i].data_dev);

	kfree(rs);
}

/*
 * For every device we have two words
 *  <meta_dev>: meta device name or '-' if missing
 *  <data_dev>: data device name or '-' if missing
 *
 * This code parses those words.
 */
static int dev_parms(struct raid_set *rs, char **argv)
{
	int i;
	int rebuild = 0;
	int metadata_available = 0;
	int ret = 0;

	for (i = 0; i < rs->md.raid_disks; i++, argv += 2) {
		rs->dev[i].rdev.raid_disk = i;

		rs->dev[i].meta_dev = NULL;
		rs->dev[i].data_dev = NULL;

		/*
		 * There are no offsets, since there is a separate device
		 * for data and metadata.
		 */
		rs->dev[i].rdev.data_offset = 0;
		rs->dev[i].rdev.mddev = &rs->md;

		if (strcmp(argv[0], "-")) {
			rs->ti->error = "Metadata devices not supported";
			return -EINVAL;
		}

		if (!strcmp(argv[1], "-")) {
			if (!test_bit(In_sync, &rs->dev[i].rdev.flags) &&
			    (!rs->dev[i].rdev.recovery_offset)) {
				rs->ti->error = "Drive designated for rebuild not specified";
				return -EINVAL;
			}

			continue;
		}

		ret = dm_get_device(rs->ti, argv[1],
				    dm_table_get_mode(rs->ti->table),
				    &rs->dev[i].data_dev);
		if (ret) {
			rs->ti->error = "RAID device lookup failure";
			return ret;
		}

		rs->dev[i].rdev.bdev = rs->dev[i].data_dev->bdev;
		list_add(&rs->dev[i].rdev.same_set, &rs->md.disks);
		if (!test_bit(In_sync, &rs->dev[i].rdev.flags))
			rebuild++;
	}

	if (metadata_available) {
		rs->md.external = 0;
		rs->md.persistent = 1;
		rs->md.major_version = 2;
	} else if (rebuild && !rs->md.recovery_cp) {
		/*
		 * Without metadata, we will not be able to tell if the array
		 * is in-sync or not - we must assume it is not.  Therefore,
		 * it is impossible to rebuild a drive.
		 *
		 * Even if there is metadata, the on-disk information may
		 * indicate that the array is not in-sync and it will then
		 * fail at that time.
		 *
		 * User could specify 'nosync' option if desperate.
		 */
		DMERR("Unable to rebuild drive while array is not in-sync");
		rs->ti->error = "RAID device lookup failure";
		return -EINVAL;
	}

	return 0;
}

/*
 * Possible arguments are...
 * RAID456:
 *	<chunk_size> [optional_args]
 *
 * Optional args:
 *    [[no]sync]			Force or prevent recovery of the entire array
 *    [rebuild <idx>]			Rebuild the drive indicated by the index
 *    [daemon_sleep <ms>]		Time between bitmap daemon work to clear bits
 *    [min_recovery_rate <kB/sec/disk>]	Throttle RAID initialization
 *    [max_recovery_rate <kB/sec/disk>]	Throttle RAID initialization
 *    [max_write_behind <sectors>]	See '-write-behind=' (man mdadm)
 *    [stripe_cache <sectors>]		Stripe cache size for higher RAIDs
 */
static int parse_raid_params(struct raid_set *rs, char **argv,
			     unsigned num_raid_params)
{
	unsigned i, rebuild_cnt = 0;
	unsigned long value;
	char *key;

	/*
	 * First, parse the in-order required arguments
	 */
	if ((strict_strtoul(argv[0], 10, &value) < 0) ||
	    !is_power_of_2(value) || (value < 8)) {
		rs->ti->error = "Bad chunk size";
		return -EINVAL;
	}

	rs->md.new_chunk_sectors = rs->md.chunk_sectors = value;
	argv++;
	num_raid_params--;

	/*
	 * Second, parse the unordered optional arguments
	 */
	for (i = 0; i < rs->md.raid_disks; i++)
		set_bit(In_sync, &rs->dev[i].rdev.flags);

	for (i = 0; i < num_raid_params; i++) {
		if (!strcmp(argv[i], "nosync")) {
			rs->md.recovery_cp = MaxSector;
			rs->print_flags |= DMPF_NOSYNC;
			rs->md.flags |= MD_SYNC_STATE_FORCED;
			continue;
		}
		if (!strcmp(argv[i], "sync")) {
			rs->md.recovery_cp = 0;
			rs->print_flags |= DMPF_SYNC;
			rs->md.flags |= MD_SYNC_STATE_FORCED;
			continue;
		}

		/* The rest of the optional arguments come in key/value pairs */
		if ((i + 1) >= num_raid_params) {
			rs->ti->error = "Wrong number of raid parameters given";
			return -EINVAL;
		}

		key = argv[i++];
		if (strict_strtoul(argv[i], 10, &value) < 0) {
			rs->ti->error = "Bad numerical argument given in raid params";
			return -EINVAL;
		}

		if (!strcmp(key, "rebuild")) {
			if (++rebuild_cnt > rs->raid_type->parity_devs) {
				rs->ti->error = "Too many rebuild drives given";
				return -EINVAL;
			}
			if (value > rs->md.raid_disks) {
				rs->ti->error = "Invalid rebuild index given";
				return -EINVAL;
			}
			clear_bit(In_sync, &rs->dev[value].rdev.flags);
			rs->dev[value].rdev.recovery_offset = 0;
		} else if (!strcmp(key, "max_write_behind")) {
			rs->print_flags |= DMPF_MAX_WRITE_BEHIND;

			/*
			 * In device-mapper, we specify things in sectors, but
			 * MD records this value in kB
			 */
			value /= 2;
			if (value > COUNTER_MAX) {
				rs->ti->error = "Max write-behind limit out of range";
				return -EINVAL;
			}
			rs->md.bitmap_info.max_write_behind = value;
		} else if (!strcmp(key, "daemon_sleep")) {
			rs->print_flags |= DMPF_DAEMON_SLEEP;
			if (!value || (value > MAX_SCHEDULE_TIMEOUT)) {
				rs->ti->error = "daemon sleep period out of range";
				return -EINVAL;
			}
			rs->md.bitmap_info.daemon_sleep = value;
		} else if (!strcmp(key, "stripe_cache")) {
			rs->print_flags |= DMPF_STRIPE_CACHE;

			/*
			 * In device-mapper, we specify things in sectors, but
			 * MD records this value in kB
			 */
			value /= 2;

			if (rs->raid_type->level < 5) {
				rs->ti->error = "Inappropriate argument: stripe_cache";
				return -EINVAL;
			}
			if (raid5_set_cache_size(&rs->md, (int)value)) {
				rs->ti->error = "Bad stripe_cache size";
				return -EINVAL;
			}
		} else if (!strcmp(key, "min_recovery_rate")) {
			rs->print_flags |= DMPF_MIN_RECOVERY_RATE;
			if (value > INT_MAX) {
				rs->ti->error = "min_recovery_rate out of range";
				return -EINVAL;
			}
			rs->md.sync_speed_min = (int)value;
		} else if (!strcmp(key, "max_recovery_rate")) {
			rs->print_flags |= DMPF_MAX_RECOVERY_RATE;
			if (value > INT_MAX) {
				rs->ti->error = "max_recovery_rate out of range";
				return -EINVAL;
			}
			rs->md.sync_speed_max = (int)value;
		} else {
			DMERR("Unable to parse RAID parameter: %s", key);
			rs->ti->error = "Unable to parse RAID parameters";
			return -EINVAL;
		}
	}

	/* Assume there are no metadata devices until the drives are parsed */
	rs->md.persistent = 0;
	rs->md.external = 1;

	return 0;
}

static void do_table_event(struct work_struct *ws)
{
	struct raid_set *rs = container_of(ws, struct raid_set, md.event_work);

	dm_table_event(rs->ti->table);
}

static int raid_is_congested(struct dm_target_callbacks *cb, int bits)
{
	struct raid_set *rs = container_of(cb, struct raid_set, callbacks);

	return md_raid5_congested(&rs->md, bits);
}

static void raid_unplug(struct dm_target_callbacks *cb)
{
	struct raid_set *rs = container_of(cb, struct raid_set, callbacks);

	md_raid5_unplug_device(rs->md.private);
}

/*
 * Construct a RAID4/5/6 mapping:
 * Args:
 *	<raid_type> <#raid_params> <raid_params>		\
 *	<#raid_devs> { <meta_dev1> <dev1> .. <meta_devN> <devN> }
 *
 * ** metadata devices are not supported yet, use '-' instead **
 *
 * <raid_params> varies by <raid_type>.  See 'parse_raid_params' for
 * details on possible <raid_params>.
 */
static int raid_ctr(struct dm_target *ti, unsigned argc, char **argv)
{
	int ret;
	struct raid_type *rt;
	unsigned long num_raid_params, num_raid_devs;
	struct raid_set *rs = NULL;

	/* Must have at least <raid_type> <#raid_params> */
	if (argc < 2) {
		ti->error = "Too few arguments";
		return -EINVAL;
	}

	/* raid type */
	rt = get_raid_type(argv[0]);
	if (!rt) {
		ti->error = "Unrecognised raid_type";
		return -EINVAL;
	}
	argc--;
	argv++;

	/* number of RAID parameters */
	if (strict_strtoul(argv[0], 10, &num_raid_params) < 0) {
		ti->error = "Cannot understand number of RAID parameters";
		return -EINVAL;
	}
	argc--;
	argv++;

	/* Skip over RAID params for now and find out # of devices */
	if (num_raid_params + 1 > argc) {
		ti->error = "Arguments do not agree with counts given";
		return -EINVAL;
	}

	if ((strict_strtoul(argv[num_raid_params], 10, &num_raid_devs) < 0) ||
	    (num_raid_devs >= INT_MAX)) {
		ti->error = "Cannot understand number of raid devices";
		return -EINVAL;
	}

	rs = context_alloc(ti, rt, (unsigned)num_raid_devs);
	if (IS_ERR(rs))
		return PTR_ERR(rs);

	ret = parse_raid_params(rs, argv, (unsigned)num_raid_params);
	if (ret)
		goto bad;

	ret = -EINVAL;

	argc -= num_raid_params + 1; /* +1: we already have num_raid_devs */
	argv += num_raid_params + 1;

	if (argc != (num_raid_devs * 2)) {
		ti->error = "Supplied RAID devices does not match the count given";
		goto bad;
	}

	ret = dev_parms(rs, argv);
	if (ret)
		goto bad;

	INIT_WORK(&rs->md.event_work, do_table_event);
	ti->split_io = rs->md.chunk_sectors;
	ti->private = rs;

	mutex_lock(&rs->md.reconfig_mutex);
	ret = md_run(&rs->md);
	rs->md.in_sync = 0; /* Assume already marked dirty */
	mutex_unlock(&rs->md.reconfig_mutex);

	if (ret) {
		ti->error = "Fail to run raid array";
		goto bad;
	}

	rs->callbacks.congested_fn = raid_is_congested;
	rs->callbacks.unplug_fn = raid_unplug;
	dm_table_add_target_callbacks(ti->table, &rs->callbacks);

	return 0;

bad:
	context_free(rs);

	return ret;
}

static void raid_dtr(struct dm_target *ti)
{
	struct raid_set *rs = ti->private;

	list_del_init(&rs->callbacks.list);
	md_stop(&rs->md);
	context_free(rs);
}

static int raid_map(struct dm_target *ti, struct bio *bio, union map_info *map_context)
{
	struct raid_set *rs = ti->private;
	mddev_t *mddev = &rs->md;

	mddev->pers->make_request(mddev, bio);

	return DM_MAPIO_SUBMITTED;
}

static int raid_status(struct dm_target *ti, status_type_t type,
		       char *result, unsigned maxlen)
{
	struct raid_set *rs = ti->private;
	unsigned raid_param_cnt = 1; /* at least 1 for chunksize */
	unsigned sz = 0;
	int i;
	sector_t sync;

	switch (type) {
	case STATUSTYPE_INFO:
		DMEMIT("%s %d ", rs->raid_type->name, rs->md.raid_disks);

		for (i = 0; i < rs->md.raid_disks; i++) {
			if (test_bit(Faulty, &rs->dev[i].rdev.flags))
				DMEMIT("D");
			else if (test_bit(In_sync, &rs->dev[i].rdev.flags))
				DMEMIT("A");
			else
				DMEMIT("a");
		}

		if (test_bit(MD_RECOVERY_RUNNING, &rs->md.recovery))
			sync = rs->md.curr_resync_completed;
		else
			sync = rs->md.recovery_cp;

		if (sync > rs->md.resync_max_sectors)
			sync = rs->md.resync_max_sectors;

		DMEMIT(" %llu/%llu",
		       (unsigned long long) sync,
		       (unsigned long long) rs->md.resync_max_sectors);

		break;
	case STATUSTYPE_TABLE:
		/* The string you would use to construct this array */
		for (i = 0; i < rs->md.raid_disks; i++)
			if (rs->dev[i].data_dev &&
			    !test_bit(In_sync, &rs->dev[i].rdev.flags))
				raid_param_cnt++; /* for rebuilds */

		raid_param_cnt += (hweight64(rs->print_flags) * 2);
		if (rs->print_flags & (DMPF_SYNC | DMPF_NOSYNC))
			raid_param_cnt--;

		DMEMIT("%s %u %u", rs->raid_type->name,
		       raid_param_cnt, rs->md.chunk_sectors);

		if ((rs->print_flags & DMPF_SYNC) &&
		    (rs->md.recovery_cp == MaxSector))
			DMEMIT(" sync");
		if (rs->print_flags & DMPF_NOSYNC)
			DMEMIT(" nosync");

		for (i = 0; i < rs->md.raid_disks; i++)
			if (rs->dev[i].data_dev &&
			    !test_bit(In_sync, &rs->dev[i].rdev.flags))
				DMEMIT(" rebuild %u", i);

		if (rs->print_flags & DMPF_DAEMON_SLEEP)
			DMEMIT(" daemon_sleep %lu",
			       rs->md.bitmap_info.daemon_sleep);

		if (rs->print_flags & DMPF_MIN_RECOVERY_RATE)
			DMEMIT(" min_recovery_rate %d", rs->md.sync_speed_min);

		if (rs->print_flags & DMPF_MAX_RECOVERY_RATE)
			DMEMIT(" max_recovery_rate %d", rs->md.sync_speed_max);

		if (rs->print_flags & DMPF_MAX_WRITE_BEHIND)
			DMEMIT(" max_write_behind %lu",
			       rs->md.bitmap_info.max_write_behind);

		if (rs->print_flags & DMPF_STRIPE_CACHE) {
			raid5_conf_t *conf = rs->md.private;

			/* convert from kiB to sectors */
			DMEMIT(" stripe_cache %d",
			       conf ? conf->max_nr_stripes * 2 : 0);
		}

		DMEMIT(" %d", rs->md.raid_disks);
		for (i = 0; i < rs->md.raid_disks; i++) {
			DMEMIT(" -"); /* metadata device */

			if (rs->dev[i].data_dev)
				DMEMIT(" %s", rs->dev[i].data_dev->name);
			else
				DMEMIT(" -");
		}
	}

	return 0;
}

static int raid_iterate_devices(struct dm_target *ti, iterate_devices_callout_fn fn, void *data)
{
	struct raid_set *rs = ti->private;
	unsigned i;
	int ret = 0;

	for (i = 0; !ret && i < rs->md.raid_disks; i++)
		if (rs->dev[i].data_dev)
			ret = fn(ti,
				 rs->dev[i].data_dev,
				 0, /* No offset on data devs */
				 rs->md.dev_sectors,
				 data);

	return ret;
}

static void raid_io_hints(struct dm_target *ti, struct queue_limits *limits)
{
	struct raid_set *rs = ti->private;
	unsigned chunk_size = rs->md.chunk_sectors << 9;
	raid5_conf_t *conf = rs->md.private;

	blk_limits_io_min(limits, chunk_size);
	blk_limits_io_opt(limits, chunk_size * (conf->raid_disks - conf->max_degraded));
}

static void raid_presuspend(struct dm_target *ti)
{
	struct raid_set *rs = ti->private;

	md_stop_writes(&rs->md);
}

static void raid_postsuspend(struct dm_target *ti)
{
	struct raid_set *rs = ti->private;

	mddev_suspend(&rs->md);
}

static void raid_resume(struct dm_target *ti)
{
	struct raid_set *rs = ti->private;

	mddev_resume(&rs->md);
}

static struct target_type raid_target = {
	.name = "raid",
	.version = {1, 0, 0},
	.module = THIS_MODULE,
	.ctr = raid_ctr,
	.dtr = raid_dtr,
	.map = raid_map,
	.status = raid_status,
	.iterate_devices = raid_iterate_devices,
	.io_hints = raid_io_hints,
	.presuspend = raid_presuspend,
	.postsuspend = raid_postsuspend,
	.resume = raid_resume,
};

static int __init dm_raid_init(void)
{
	return dm_register_target(&raid_target);
}

static void __exit dm_raid_exit(void)
{
	dm_unregister_target(&raid_target);
}

module_init(dm_raid_init);
module_exit(dm_raid_exit);

MODULE_DESCRIPTION(DM_NAME " raid4/5/6 target");
MODULE_ALIAS("dm-raid4");
MODULE_ALIAS("dm-raid5");
MODULE_ALIAS("dm-raid6");
MODULE_AUTHOR("Neil Brown <[email protected]>");
MODULE_LICENSE("GPL");
Commit	Line	Data
9d09e663 N	1	/*
	2	* Copyright (C) 2010-2011 Neil Brown
	3	* Copyright (C) 2010-2011 Red Hat, Inc. All rights reserved.
	4	*
	5	* This file is released under the GPL.
	6	*/
	7
	8	#include <linux/slab.h>
	9
	10	#include "md.h"
	11	#include "raid5.h"
	12	#include "dm.h"
	13	#include "bitmap.h"
	14
	15	#define DM_MSG_PREFIX "raid"
	16
	17	/*
	18	* If the MD doesn't support MD_SYNC_STATE_FORCED yet, then
	19	* make it so the flag doesn't set anything.
	20	*/
	21	#ifndef MD_SYNC_STATE_FORCED
	22	#define MD_SYNC_STATE_FORCED 0
	23	#endif
	24
	25	struct raid_dev {
	26	/*
	27	* Two DM devices, one to hold metadata and one to hold the
	28	* actual data/parity. The reason for this is to not confuse
	29	* ti->len and give more flexibility in altering size and
	30	* characteristics.
	31	*
	32	* While it is possible for this device to be associated
	33	* with a different physical device than the data_dev, it
	34	* is intended for it to be the same.
	35	* \|--------- Physical Device ---------\|
	36	* \|- meta_dev -\|------ data_dev ------\|
	37	*/
	38	struct dm_dev *meta_dev;
	39	struct dm_dev *data_dev;
	40	struct mdk_rdev_s rdev;
	41	};
	42
	43	/*
	44	* Flags for rs->print_flags field.
	45	*/
	46	#define DMPF_DAEMON_SLEEP 0x1
	47	#define DMPF_MAX_WRITE_BEHIND 0x2
	48	#define DMPF_SYNC 0x4
	49	#define DMPF_NOSYNC 0x8
	50	#define DMPF_STRIPE_CACHE 0x10
	51	#define DMPF_MIN_RECOVERY_RATE 0x20
	52	#define DMPF_MAX_RECOVERY_RATE 0x40
	53
	54	struct raid_set {
	55	struct dm_target *ti;
	56
	57	uint64_t print_flags;
	58
	59	struct mddev_s md;
	60	struct raid_type *raid_type;
	61	struct dm_target_callbacks callbacks;
	62
	63	struct raid_dev dev[0];
	64	};
65
66	/* Supported raid types and properties. */
67	static struct raid_type {
68	const char name; / RAID algorithm. */
69	const char descr; / Descriptor text for logging. */
70	const unsigned parity_devs; /* # of parity devices. */
71	const unsigned minimal_devs; /* minimal # of devices in set. */
72	const unsigned level; /* RAID level. */
73	const unsigned algorithm; /* RAID algorithm. */
74	} raid_types[] = {
75	{"raid4", "RAID4 (dedicated parity disk)", 1, 2, 5, ALGORITHM_PARITY_0},
76	{"raid5_la", "RAID5 (left asymmetric)", 1, 2, 5, ALGORITHM_LEFT_ASYMMETRIC},
77	{"raid5_ra", "RAID5 (right asymmetric)", 1, 2, 5, ALGORITHM_RIGHT_ASYMMETRIC},
78	{"raid5_ls", "RAID5 (left symmetric)", 1, 2, 5, ALGORITHM_LEFT_SYMMETRIC},
79	{"raid5_rs", "RAID5 (right symmetric)", 1, 2, 5, ALGORITHM_RIGHT_SYMMETRIC},
80	{"raid6_zr", "RAID6 (zero restart)", 2, 4, 6, ALGORITHM_ROTATING_ZERO_RESTART},
81	{"raid6_nr", "RAID6 (N restart)", 2, 4, 6, ALGORITHM_ROTATING_N_RESTART},
82	{"raid6_nc", "RAID6 (N continue)", 2, 4, 6, ALGORITHM_ROTATING_N_CONTINUE}
83	};
84
85	static struct raid_type get_raid_type(char name)
86	{
87	int i;
88
89	for (i = 0; i < ARRAY_SIZE(raid_types); i++)
90	if (!strcmp(raid_types[i].name, name))
91	return &raid_types[i];
92
93	return NULL;
94	}
95
96	static struct raid_set context_alloc(struct dm_target ti, struct raid_type *raid_type, unsigned raid_devs)
97	{
98	unsigned i;
99	struct raid_set *rs;
100	sector_t sectors_per_dev;
101
102	if (raid_devs <= raid_type->parity_devs) {
103	ti->error = "Insufficient number of devices";
104	return ERR_PTR(-EINVAL);
105	}
106
107	sectors_per_dev = ti->len;
108	if (sector_div(sectors_per_dev, (raid_devs - raid_type->parity_devs))) {
109	ti->error = "Target length not divisible by number of data devices";
110	return ERR_PTR(-EINVAL);
111	}
112
113	rs = kzalloc(sizeof(rs) + raid_devs sizeof(rs->dev[0]), GFP_KERNEL);
114	if (!rs) {
115	ti->error = "Cannot allocate raid context";
116	return ERR_PTR(-ENOMEM);
117	}
118
119	mddev_init(&rs->md);
120
121	rs->ti = ti;
122	rs->raid_type = raid_type;
123	rs->md.raid_disks = raid_devs;
124	rs->md.level = raid_type->level;
125	rs->md.new_level = rs->md.level;
126	rs->md.dev_sectors = sectors_per_dev;
127	rs->md.layout = raid_type->algorithm;
128	rs->md.new_layout = rs->md.layout;
129	rs->md.delta_disks = 0;
130	rs->md.recovery_cp = 0;
131
132	for (i = 0; i < raid_devs; i++)
133	md_rdev_init(&rs->dev[i].rdev);
134
135	/*
136	* Remaining items to be initialized by further RAID params:
137	* rs->md.persistent
138	* rs->md.external
139	* rs->md.chunk_sectors
140	* rs->md.new_chunk_sectors
141	*/
142
143	return rs;
144	}
145
146	static void context_free(struct raid_set *rs)
147	{
148	int i;
149
150	for (i = 0; i < rs->md.raid_disks; i++)
151	if (rs->dev[i].data_dev)
152	dm_put_device(rs->ti, rs->dev[i].data_dev);
153
154	kfree(rs);
155	}
156
157	/*
158	* For every device we have two words
159	* <meta_dev>: meta device name or '-' if missing
160	* <data_dev>: data device name or '-' if missing
161	*
162	* This code parses those words.
163	*/
164	static int dev_parms(struct raid_set rs, char *argv)
165	{
166	int i;
167	int rebuild = 0;
168	int metadata_available = 0;
169	int ret = 0;
170
171	for (i = 0; i < rs->md.raid_disks; i++, argv += 2) {
172	rs->dev[i].rdev.raid_disk = i;
173
174	rs->dev[i].meta_dev = NULL;
175	rs->dev[i].data_dev = NULL;
176
177	/*
178	* There are no offsets, since there is a separate device
179	* for data and metadata.
180	*/
181	rs->dev[i].rdev.data_offset = 0;
182	rs->dev[i].rdev.mddev = &rs->md;
183
184	if (strcmp(argv[0], "-")) {
185	rs->ti->error = "Metadata devices not supported";
186	return -EINVAL;
187	}
188
189	if (!strcmp(argv[1], "-")) {
190	if (!test_bit(In_sync, &rs->dev[i].rdev.flags) &&
191	(!rs->dev[i].rdev.recovery_offset)) {
192	rs->ti->error = "Drive designated for rebuild not specified";
193	return -EINVAL;
194	}
195
196	continue;
197	}
198
199	ret = dm_get_device(rs->ti, argv[1],
200	dm_table_get_mode(rs->ti->table),
201	&rs->dev[i].data_dev);
202	if (ret) {
203	rs->ti->error = "RAID device lookup failure";
204	return ret;
205	}
206
207	rs->dev[i].rdev.bdev = rs->dev[i].data_dev->bdev;
208	list_add(&rs->dev[i].rdev.same_set, &rs->md.disks);
209	if (!test_bit(In_sync, &rs->dev[i].rdev.flags))
210	rebuild++;
211	}
212
213	if (metadata_available) {
214	rs->md.external = 0;
215	rs->md.persistent = 1;
216	rs->md.major_version = 2;
217	} else if (rebuild && !rs->md.recovery_cp) {
218	/*
219	* Without metadata, we will not be able to tell if the array
220	* is in-sync or not - we must assume it is not. Therefore,
221	* it is impossible to rebuild a drive.
222	*
223	* Even if there is metadata, the on-disk information may
224	* indicate that the array is not in-sync and it will then
225	* fail at that time.
226	*
227	* User could specify 'nosync' option if desperate.
228	*/
229	DMERR("Unable to rebuild drive while array is not in-sync");
230	rs->ti->error = "RAID device lookup failure";
231	return -EINVAL;
232	}
233
234	return 0;
235	}
236
237	/*
238	* Possible arguments are...
239	* RAID456:
240	* <chunk_size> [optional_args]
241	*
242	* Optional args:
243	* [[no]sync] Force or prevent recovery of the entire array
244	* [rebuild <idx>] Rebuild the drive indicated by the index
245	* [daemon_sleep <ms>] Time between bitmap daemon work to clear bits
246	* [min_recovery_rate <kB/sec/disk>] Throttle RAID initialization
247	* [max_recovery_rate <kB/sec/disk>] Throttle RAID initialization
248	* [max_write_behind <sectors>] See '-write-behind=' (man mdadm)
249	* [stripe_cache <sectors>] Stripe cache size for higher RAIDs
250	*/
251	static int parse_raid_params(struct raid_set rs, char *argv,
252	unsigned num_raid_params)
253	{
254	unsigned i, rebuild_cnt = 0;
255	unsigned long value;
256	char *key;
257
258	/*
259	* First, parse the in-order required arguments
260	*/
261	if ((strict_strtoul(argv[0], 10, &value) < 0) \|\|
262	!is_power_of_2(value) \|\| (value < 8)) {
263	rs->ti->error = "Bad chunk size";
264	return -EINVAL;
265	}
266
267	rs->md.new_chunk_sectors = rs->md.chunk_sectors = value;
268	argv++;
269	num_raid_params--;
270
271	/*
272	* Second, parse the unordered optional arguments
273	*/
274	for (i = 0; i < rs->md.raid_disks; i++)
275	set_bit(In_sync, &rs->dev[i].rdev.flags);
276
277	for (i = 0; i < num_raid_params; i++) {
278	if (!strcmp(argv[i], "nosync")) {
279	rs->md.recovery_cp = MaxSector;
280	rs->print_flags \|= DMPF_NOSYNC;
281	rs->md.flags \|= MD_SYNC_STATE_FORCED;
282	continue;
283	}
284	if (!strcmp(argv[i], "sync")) {
285	rs->md.recovery_cp = 0;
286	rs->print_flags \|= DMPF_SYNC;
287	rs->md.flags \|= MD_SYNC_STATE_FORCED;
288	continue;
289	}
290
291	/* The rest of the optional arguments come in key/value pairs */
292	if ((i + 1) >= num_raid_params) {
293	rs->ti->error = "Wrong number of raid parameters given";
294	return -EINVAL;
295	}
296
297	key = argv[i++];
298	if (strict_strtoul(argv[i], 10, &value) < 0) {
299	rs->ti->error = "Bad numerical argument given in raid params";
300	return -EINVAL;
301	}
302
303	if (!strcmp(key, "rebuild")) {
304	if (++rebuild_cnt > rs->raid_type->parity_devs) {
305	rs->ti->error = "Too many rebuild drives given";
306	return -EINVAL;
307	}
308	if (value > rs->md.raid_disks) {
309	rs->ti->error = "Invalid rebuild index given";
310	return -EINVAL;
311	}
312	clear_bit(In_sync, &rs->dev[value].rdev.flags);
313	rs->dev[value].rdev.recovery_offset = 0;
314	} else if (!strcmp(key, "max_write_behind")) {
315	rs->print_flags \|= DMPF_MAX_WRITE_BEHIND;
316
317	/*
318	* In device-mapper, we specify things in sectors, but
319	* MD records this value in kB
320	*/
321	value /= 2;
322	if (value > COUNTER_MAX) {
323	rs->ti->error = "Max write-behind limit out of range";
324	return -EINVAL;
325	}
326	rs->md.bitmap_info.max_write_behind = value;
327	} else if (!strcmp(key, "daemon_sleep")) {
328	rs->print_flags \|= DMPF_DAEMON_SLEEP;
329	if (!value \|\| (value > MAX_SCHEDULE_TIMEOUT)) {
330	rs->ti->error = "daemon sleep period out of range";
331	return -EINVAL;
332	}
333	rs->md.bitmap_info.daemon_sleep = value;
334	} else if (!strcmp(key, "stripe_cache")) {
335	rs->print_flags \|= DMPF_STRIPE_CACHE;
336
337	/*
338	* In device-mapper, we specify things in sectors, but
339	* MD records this value in kB
340	*/
341	value /= 2;
342
343	if (rs->raid_type->level < 5) {
344	rs->ti->error = "Inappropriate argument: stripe_cache";
345	return -EINVAL;
346	}
347	if (raid5_set_cache_size(&rs->md, (int)value)) {
348	rs->ti->error = "Bad stripe_cache size";
349	return -EINVAL;
350	}
351	} else if (!strcmp(key, "min_recovery_rate")) {
352	rs->print_flags \|= DMPF_MIN_RECOVERY_RATE;
353	if (value > INT_MAX) {
354	rs->ti->error = "min_recovery_rate out of range";
355	return -EINVAL;
356	}
357	rs->md.sync_speed_min = (int)value;
358	} else if (!strcmp(key, "max_recovery_rate")) {
359	rs->print_flags \|= DMPF_MAX_RECOVERY_RATE;
360	if (value > INT_MAX) {
361	rs->ti->error = "max_recovery_rate out of range";
362	return -EINVAL;
363	}
364	rs->md.sync_speed_max = (int)value;
365	} else {
366	DMERR("Unable to parse RAID parameter: %s", key);
367	rs->ti->error = "Unable to parse RAID parameters";
368	return -EINVAL;
369	}
370	}
371
372	/* Assume there are no metadata devices until the drives are parsed */
373	rs->md.persistent = 0;
374	rs->md.external = 1;
375
376	return 0;
377	}
378
379	static void do_table_event(struct work_struct *ws)
380	{
381	struct raid_set *rs = container_of(ws, struct raid_set, md.event_work);
382
383	dm_table_event(rs->ti->table);
384	}
385
386	static int raid_is_congested(struct dm_target_callbacks *cb, int bits)
387	{
388	struct raid_set *rs = container_of(cb, struct raid_set, callbacks);
389
390	return md_raid5_congested(&rs->md, bits);
391	}
392
393	static void raid_unplug(struct dm_target_callbacks *cb)
394	{
395	struct raid_set *rs = container_of(cb, struct raid_set, callbacks);
396
397	md_raid5_unplug_device(rs->md.private);
398	}
399
400	/*
401	* Construct a RAID4/5/6 mapping:
402	* Args:
403	* <raid_type> <#raid_params> <raid_params> \
404	* <#raid_devs> { <meta_dev1> <dev1> .. <meta_devN> <devN> }
405	*
406	* metadata devices are not supported yet, use '-' instead
407	*
408	* <raid_params> varies by <raid_type>. See 'parse_raid_params' for
409	* details on possible <raid_params>.
410	*/
411	static int raid_ctr(struct dm_target ti, unsigned argc, char *argv)
412	{
413	int ret;
414	struct raid_type *rt;
415	unsigned long num_raid_params, num_raid_devs;
416	struct raid_set *rs = NULL;
417
418	/* Must have at least <raid_type> <#raid_params> */
419	if (argc < 2) {
420	ti->error = "Too few arguments";
421	return -EINVAL;
422	}
423
424	/* raid type */
425	rt = get_raid_type(argv[0]);
426	if (!rt) {
427	ti->error = "Unrecognised raid_type";
428	return -EINVAL;
429	}
430	argc--;
431	argv++;
432
433	/* number of RAID parameters */
434	if (strict_strtoul(argv[0], 10, &num_raid_params) < 0) {
435	ti->error = "Cannot understand number of RAID parameters";
436	return -EINVAL;
437	}
438	argc--;
439	argv++;
440
441	/* Skip over RAID params for now and find out # of devices */
442	if (num_raid_params + 1 > argc) {
443	ti->error = "Arguments do not agree with counts given";
444	return -EINVAL;
445	}
446
447	if ((strict_strtoul(argv[num_raid_params], 10, &num_raid_devs) < 0) \|\|
448	(num_raid_devs >= INT_MAX)) {
449	ti->error = "Cannot understand number of raid devices";
450	return -EINVAL;
451	}
452
453	rs = context_alloc(ti, rt, (unsigned)num_raid_devs);
454	if (IS_ERR(rs))
455	return PTR_ERR(rs);
456
457	ret = parse_raid_params(rs, argv, (unsigned)num_raid_params);
458	if (ret)
459	goto bad;
460
461	ret = -EINVAL;
462
463	argc -= num_raid_params + 1; /* +1: we already have num_raid_devs */
464	argv += num_raid_params + 1;
465
466	if (argc != (num_raid_devs * 2)) {
467	ti->error = "Supplied RAID devices does not match the count given";
468	goto bad;
469	}
470
471	ret = dev_parms(rs, argv);
472	if (ret)
473	goto bad;
474
475	INIT_WORK(&rs->md.event_work, do_table_event);
476	ti->split_io = rs->md.chunk_sectors;
477	ti->private = rs;
478
479	mutex_lock(&rs->md.reconfig_mutex);
480	ret = md_run(&rs->md);
481	rs->md.in_sync = 0; /* Assume already marked dirty */
482	mutex_unlock(&rs->md.reconfig_mutex);
483
484	if (ret) {
485	ti->error = "Fail to run raid array";
486	goto bad;
487	}
488
489	rs->callbacks.congested_fn = raid_is_congested;
490	rs->callbacks.unplug_fn = raid_unplug;
491	dm_table_add_target_callbacks(ti->table, &rs->callbacks);
492
493	return 0;
494
495	bad:
496	context_free(rs);
497
498	return ret;
499	}
500
501	static void raid_dtr(struct dm_target *ti)
502	{
503	struct raid_set *rs = ti->private;
504
505	list_del_init(&rs->callbacks.list);
506	md_stop(&rs->md);
507	context_free(rs);
508	}
509
510	static int raid_map(struct dm_target ti, struct bio bio, union map_info *map_context)
511	{
512	struct raid_set *rs = ti->private;
513	mddev_t *mddev = &rs->md;
514
515	mddev->pers->make_request(mddev, bio);
516
517	return DM_MAPIO_SUBMITTED;
518	}
519
520	static int raid_status(struct dm_target *ti, status_type_t type,
521	char *result, unsigned maxlen)
522	{
523	struct raid_set *rs = ti->private;
524	unsigned raid_param_cnt = 1; /* at least 1 for chunksize */
525	unsigned sz = 0;
526	int i;
527	sector_t sync;
528
529	switch (type) {
530	case STATUSTYPE_INFO:
531	DMEMIT("%s %d ", rs->raid_type->name, rs->md.raid_disks);
532
533	for (i = 0; i < rs->md.raid_disks; i++) {
534	if (test_bit(Faulty, &rs->dev[i].rdev.flags))
535	DMEMIT("D");
536	else if (test_bit(In_sync, &rs->dev[i].rdev.flags))
537	DMEMIT("A");
538	else
539	DMEMIT("a");
540	}
541
542	if (test_bit(MD_RECOVERY_RUNNING, &rs->md.recovery))
543	sync = rs->md.curr_resync_completed;
544	else
545	sync = rs->md.recovery_cp;
546
547	if (sync > rs->md.resync_max_sectors)
548	sync = rs->md.resync_max_sectors;
549
550	DMEMIT(" %llu/%llu",
551	(unsigned long long) sync,
552	(unsigned long long) rs->md.resync_max_sectors);
553
554	break;
555	case STATUSTYPE_TABLE:
556	/* The string you would use to construct this array */
557	for (i = 0; i < rs->md.raid_disks; i++)
558	if (rs->dev[i].data_dev &&
559	!test_bit(In_sync, &rs->dev[i].rdev.flags))
560	raid_param_cnt++; /* for rebuilds */
561
562	raid_param_cnt += (hweight64(rs->print_flags) * 2);
563	if (rs->print_flags & (DMPF_SYNC \| DMPF_NOSYNC))
564	raid_param_cnt--;
565
566	DMEMIT("%s %u %u", rs->raid_type->name,
567	raid_param_cnt, rs->md.chunk_sectors);
568
569	if ((rs->print_flags & DMPF_SYNC) &&
570	(rs->md.recovery_cp == MaxSector))
571	DMEMIT(" sync");
572	if (rs->print_flags & DMPF_NOSYNC)
573	DMEMIT(" nosync");
574
575	for (i = 0; i < rs->md.raid_disks; i++)
576	if (rs->dev[i].data_dev &&
577	!test_bit(In_sync, &rs->dev[i].rdev.flags))
578	DMEMIT(" rebuild %u", i);
579
580	if (rs->print_flags & DMPF_DAEMON_SLEEP)
581	DMEMIT(" daemon_sleep %lu",
582	rs->md.bitmap_info.daemon_sleep);
583
584	if (rs->print_flags & DMPF_MIN_RECOVERY_RATE)
585	DMEMIT(" min_recovery_rate %d", rs->md.sync_speed_min);
586
587	if (rs->print_flags & DMPF_MAX_RECOVERY_RATE)
588	DMEMIT(" max_recovery_rate %d", rs->md.sync_speed_max);
589
590	if (rs->print_flags & DMPF_MAX_WRITE_BEHIND)
591	DMEMIT(" max_write_behind %lu",
592	rs->md.bitmap_info.max_write_behind);
593
594	if (rs->print_flags & DMPF_STRIPE_CACHE) {
595	raid5_conf_t *conf = rs->md.private;
596
597	/* convert from kiB to sectors */
598	DMEMIT(" stripe_cache %d",
599	conf ? conf->max_nr_stripes * 2 : 0);
600	}
601
602	DMEMIT(" %d", rs->md.raid_disks);
603	for (i = 0; i < rs->md.raid_disks; i++) {
604	DMEMIT(" -"); /* metadata device */
605
606	if (rs->dev[i].data_dev)
607	DMEMIT(" %s", rs->dev[i].data_dev->name);
608	else
609	DMEMIT(" -");
610	}
611	}
612
613	return 0;
614	}
615
616	static int raid_iterate_devices(struct dm_target ti, iterate_devices_callout_fn fn, void data)
617	{
618	struct raid_set *rs = ti->private;
619	unsigned i;
620	int ret = 0;
621
622	for (i = 0; !ret && i < rs->md.raid_disks; i++)
623	if (rs->dev[i].data_dev)
624	ret = fn(ti,
625	rs->dev[i].data_dev,
626	0, /* No offset on data devs */
627	rs->md.dev_sectors,
628	data);
629
630	return ret;
631	}
632
633	static void raid_io_hints(struct dm_target ti, struct queue_limits limits)
634	{
635	struct raid_set *rs = ti->private;
636	unsigned chunk_size = rs->md.chunk_sectors << 9;
637	raid5_conf_t *conf = rs->md.private;
638
639	blk_limits_io_min(limits, chunk_size);
640	blk_limits_io_opt(limits, chunk_size * (conf->raid_disks - conf->max_degraded));
641	}
642
643	static void raid_presuspend(struct dm_target *ti)
644	{
645	struct raid_set *rs = ti->private;
646
647	md_stop_writes(&rs->md);
648	}
649
650	static void raid_postsuspend(struct dm_target *ti)
651	{
652	struct raid_set *rs = ti->private;
653
654	mddev_suspend(&rs->md);
655	}
656
657	static void raid_resume(struct dm_target *ti)
658	{
659	struct raid_set *rs = ti->private;
660
661	mddev_resume(&rs->md);
662	}
663
664	static struct target_type raid_target = {
665	.name = "raid",
666	.version = {1, 0, 0},
667	.module = THIS_MODULE,
668	.ctr = raid_ctr,
669	.dtr = raid_dtr,
670	.map = raid_map,
671	.status = raid_status,
672	.iterate_devices = raid_iterate_devices,
673	.io_hints = raid_io_hints,
674	.presuspend = raid_presuspend,
675	.postsuspend = raid_postsuspend,
676	.resume = raid_resume,
677	};
678
679	static int __init dm_raid_init(void)
680	{
681	return dm_register_target(&raid_target);
682	}
683
684	static void __exit dm_raid_exit(void)
685	{
686	dm_unregister_target(&raid_target);
687	}
688
689	module_init(dm_raid_init);
690	module_exit(dm_raid_exit);
691
692	MODULE_DESCRIPTION(DM_NAME " raid4/5/6 target");
693	MODULE_ALIAS("dm-raid4");
694	MODULE_ALIAS("dm-raid5");
695	MODULE_ALIAS("dm-raid6");
696	MODULE_AUTHOR("Neil Brown <[email protected]>");
697	MODULE_LICENSE("GPL");