[linux.git] / drivers / md / raid10.h

/* SPDX-License-Identifier: GPL-2.0 */
#ifndef _RAID10_H
#define _RAID10_H

struct raid10_info {
	struct md_rdev	*rdev, *replacement;
	sector_t	head_position;
	int		recovery_disabled;	/* matches
						 * mddev->recovery_disabled
						 * when we shouldn't try
						 * recovering this device.
						 */
};

struct r10conf {
	struct mddev		*mddev;
	struct raid10_info	*mirrors;
	struct raid10_info	*mirrors_new, *mirrors_old;
	spinlock_t		device_lock;

	/* geometry */
	struct geom {
		int		raid_disks;
		int		near_copies;  /* number of copies laid out
					       * raid0 style */
		int		far_copies;   /* number of copies laid out
					       * at large strides across drives
					       */
		int		far_offset;   /* far_copies are offset by 1
					       * stripe instead of many
					       */
		sector_t	stride;	      /* distance between far copies.
					       * This is size / far_copies unless
					       * far_offset, in which case it is
					       * 1 stripe.
					       */
		int             far_set_size; /* The number of devices in a set,
					       * where a 'set' are devices that
					       * contain far/offset copies of
					       * each other.
					       */
		int		chunk_shift; /* shift from chunks to sectors */
		sector_t	chunk_mask;
	} prev, geo;
	int			copies;	      /* near_copies * far_copies.
					       * must be <= raid_disks
					       */

	sector_t		dev_sectors;  /* temp copy of
					       * mddev->dev_sectors */
	sector_t		reshape_progress;
	sector_t		reshape_safe;
	unsigned long		reshape_checkpoint;
	sector_t		offset_diff;

	struct list_head	retry_list;
	/* A separate list of r1bio which just need raid_end_bio_io called.
	 * This mustn't happen for writes which had any errors if the superblock
	 * needs to be written.
	 */
	struct list_head	bio_end_io_list;

	/* queue pending writes and submit them on unplug */
	struct bio_list		pending_bio_list;
	int			pending_count;

	spinlock_t		resync_lock;
	atomic_t		nr_pending;
	int			nr_waiting;
	int			nr_queued;
	int			barrier;
	int			array_freeze_pending;
	sector_t		next_resync;
	int			fullsync;  /* set to 1 if a full sync is needed,
					    * (fresh device added).
					    * Cleared when a sync completes.
					    */
	int			have_replacement; /* There is at least one
						   * replacement device.
						   */
	wait_queue_head_t	wait_barrier;

	mempool_t		*r10bio_pool;
	mempool_t		*r10buf_pool;
	struct page		*tmppage;
	struct bio_set		*bio_split;

	/* When taking over an array from a different personality, we store
	 * the new thread here until we fully activate the array.
	 */
	struct md_thread	*thread;

	/*
	 * Keep track of cluster resync window to send to other nodes.
	 */
	sector_t		cluster_sync_low;
	sector_t		cluster_sync_high;
};

/*
 * this is our 'private' RAID10 bio.
 *
 * it contains information about what kind of IO operations were started
 * for this RAID10 operation, and about their status:
 */

struct r10bio {
	atomic_t		remaining; /* 'have we finished' count,
					    * used from IRQ handlers
					    */
	sector_t		sector;	/* virtual sector number */
	int			sectors;
	unsigned long		state;
	struct mddev		*mddev;
	/*
	 * original bio going to /dev/mdx
	 */
	struct bio		*master_bio;
	/*
	 * if the IO is in READ direction, then this is where we read
	 */
	int			read_slot;

	struct list_head	retry_list;
	/*
	 * if the IO is in WRITE direction, then multiple bios are used,
	 * one for each copy.
	 * When resyncing we also use one for each copy.
	 * When reconstructing, we use 2 bios, one for read, one for write.
	 * We choose the number when they are allocated.
	 * We sometimes need an extra bio to write to the replacement.
	 */
	struct r10dev {
		struct bio	*bio;
		union {
			struct bio	*repl_bio; /* used for resync and
						    * writes */
			struct md_rdev	*rdev;	   /* used for reads
						    * (read_slot >= 0) */
		};
		sector_t	addr;
		int		devnum;
	} devs[0];
};

/* bits for r10bio.state */
enum r10bio_state {
	R10BIO_Uptodate,
	R10BIO_IsSync,
	R10BIO_IsRecover,
	R10BIO_IsReshape,
	R10BIO_Degraded,
/* Set ReadError on bios that experience a read error
 * so that raid10d knows what to do with them.
 */
	R10BIO_ReadError,
/* If a write for this request means we can clear some
 * known-bad-block records, we set this flag.
 */
	R10BIO_MadeGood,
	R10BIO_WriteError,
/* During a reshape we might be performing IO on the
 * 'previous' part of the array, in which case this
 * flag is set
 */
	R10BIO_Previous,
/* failfast devices did receive failfast requests. */
	R10BIO_FailFast,
};
#endif
Commit	Line	Data
b2441318	1	/* SPDX-License-Identifier: GPL-2.0 */
1da177e4 LT	2	#ifndef _RAID10_H
	3	#define _RAID10_H
	4
dc280d98	5	struct raid10_info {
69335ef3	6	struct md_rdev rdev, replacement;
1da177e4	7	sector_t head_position;
2bb77736 N	8	int recovery_disabled; /* matches
	9	* mddev->recovery_disabled
	10	* when we shouldn't try
	11	* recovering this device.
	12	*/
1da177e4 LT	13	};
1da177e4 LT	14
e879a879	15	struct r10conf {
fd01b88c	16	struct mddev *mddev;
dc280d98 JB	17	struct raid10_info *mirrors;
dc280d98 JB	18	struct raid10_info mirrors_new, mirrors_old;
1da177e4 LT	19	spinlock_t device_lock;
	20
	21	/* geometry */
5cf00fcd N	22	struct geom {
	23	int raid_disks;
	24	int near_copies; /* number of copies laid out
69335ef3	25	* raid0 style */
5cf00fcd	26	int far_copies; /* number of copies laid out
1da177e4 LT	27	* at large strides across drives
1da177e4 LT	28	*/
5cf00fcd	29	int far_offset; /* far_copies are offset by 1
69335ef3	30	* stripe instead of many
c93983bf	31	*/
5cf00fcd	32	sector_t stride; /* distance between far copies.
c93983bf N	33	* This is size / far_copies unless
	34	* far_offset, in which case it is
	35	* 1 stripe.
1da177e4	36	*/
475901af JB	37	int far_set_size; /* The number of devices in a set,
	38	* where a 'set' are devices that
	39	* contain far/offset copies of
	40	* each other.
	41	*/
5cf00fcd N	42	int chunk_shift; /* shift from chunks to sectors */
5cf00fcd N	43	sector_t chunk_mask;
f8c9e74f	44	} prev, geo;
5cf00fcd N	45	int copies; /* near_copies * far_copies.
	46	* must be <= raid_disks
	47	*/
1da177e4	48
69335ef3 N	49	sector_t dev_sectors; /* temp copy of
69335ef3 N	50	* mddev->dev_sectors */
f8c9e74f	51	sector_t reshape_progress;
3ea7daa5 N	52	sector_t reshape_safe;
	53	unsigned long reshape_checkpoint;
	54	sector_t offset_diff;
dab8b292	55
1da177e4	56	struct list_head retry_list;
95af587e N	57	/* A separate list of r1bio which just need raid_end_bio_io called.
	58	* This mustn't happen for writes which had any errors if the superblock
	59	* needs to be written.
	60	*/
	61	struct list_head bio_end_io_list;
	62
6cce3b23 N	63	/* queue pending writes and submit them on unplug */
6cce3b23 N	64	struct bio_list pending_bio_list;
34db0cd6	65	int pending_count;
1da177e4 LT	66
1da177e4 LT	67	spinlock_t resync_lock;
0e5313e2	68	atomic_t nr_pending;
69335ef3 N	69	int nr_waiting;
	70	int nr_queued;
	71	int barrier;
0e5313e2	72	int array_freeze_pending;
1da177e4	73	sector_t next_resync;
6cce3b23 N	74	int fullsync; /* set to 1 if a full sync is needed,
	75	* (fresh device added).
	76	* Cleared when a sync completes.
	77	*/
69335ef3 N	78	int have_replacement; /* There is at least one
	79	* replacement device.
	80	*/
0a27ec96	81	wait_queue_head_t wait_barrier;
1da177e4	82
69335ef3 N	83	mempool_t *r10bio_pool;
69335ef3 N	84	mempool_t *r10buf_pool;
4443ae10	85	struct page *tmppage;
fc9977dd	86	struct bio_set *bio_split;
dab8b292 TM	87
	88	/* When taking over an array from a different personality, we store
	89	* the new thread here until we fully activate the array.
	90	*/
2b8bf345	91	struct md_thread *thread;
8db87912 GJ	92
	93	/*
	94	* Keep track of cluster resync window to send to other nodes.
	95	*/
	96	sector_t cluster_sync_low;
	97	sector_t cluster_sync_high;
1da177e4 LT	98	};
1da177e4 LT	99
1da177e4 LT	100	/*
	101	* this is our 'private' RAID10 bio.
	102	*
	103	* it contains information about what kind of IO operations were started
	104	* for this RAID10 operation, and about their status:
	105	*/
	106
9f2c9d12	107	struct r10bio {
1da177e4 LT	108	atomic_t remaining; /* 'have we finished' count,
	109	* used from IRQ handlers
	110	*/
	111	sector_t sector; /* virtual sector number */
	112	int sectors;
	113	unsigned long state;
fd01b88c	114	struct mddev *mddev;
1da177e4 LT	115	/*
	116	* original bio going to /dev/mdx
	117	*/
	118	struct bio *master_bio;
	119	/*
	120	* if the IO is in READ direction, then this is where we read
	121	*/
	122	int read_slot;
	123
	124	struct list_head retry_list;
	125	/*
	126	* if the IO is in WRITE direction, then multiple bios are used,
	127	* one for each copy.
	128	* When resyncing we also use one for each copy.
	129	* When reconstructing, we use 2 bios, one for read, one for write.
	130	* We choose the number when they are allocated.
69335ef3	131	* We sometimes need an extra bio to write to the replacement.
1da177e4	132	*/
e0ee7785	133	struct r10dev {
69335ef3 N	134	struct bio *bio;
	135	union {
	136	struct bio repl_bio; / used for resync and
	137	* writes */
	138	struct md_rdev rdev; / used for reads
	139	* (read_slot >= 0) */
	140	};
	141	sector_t addr;
	142	int devnum;
1da177e4 LT	143	} devs[0];
	144	};
	145
	146	/* bits for r10bio.state */
69335ef3 N	147	enum r10bio_state {
	148	R10BIO_Uptodate,
	149	R10BIO_IsSync,
	150	R10BIO_IsRecover,
3ea7daa5	151	R10BIO_IsReshape,
69335ef3	152	R10BIO_Degraded,
856e08e2 N	153	/* Set ReadError on bios that experience a read error
	154	* so that raid10d knows what to do with them.
	155	*/
69335ef3	156	R10BIO_ReadError,
749c55e9 N	157	/* If a write for this request means we can clear some
	158	* known-bad-block records, we set this flag.
	159	*/
69335ef3 N	160	R10BIO_MadeGood,
69335ef3 N	161	R10BIO_WriteError,
f8c9e74f N	162	/* During a reshape we might be performing IO on the
	163	* 'previous' part of the array, in which case this
	164	* flag is set
	165	*/
	166	R10BIO_Previous,
8d3ca83d N	167	/* failfast devices did receive failfast requests. */
8d3ca83d N	168	R10BIO_FailFast,
69335ef3	169	};
1da177e4	170	#endif