[mirror_ubuntu-focal-kernel.git] / drivers / md / dm-zoned-reclaim.c

// SPDX-License-Identifier: GPL-2.0-only
/*
 * Copyright (C) 2017 Western Digital Corporation or its affiliates.
 *
 * This file is released under the GPL.
 */

#include "dm-zoned.h"

#include <linux/module.h>

#define	DM_MSG_PREFIX		"zoned reclaim"

struct dmz_reclaim {
	struct dmz_metadata     *metadata;
	struct dmz_dev		*dev;

	struct delayed_work	work;
	struct workqueue_struct *wq;

	struct dm_kcopyd_client	*kc;
	struct dm_kcopyd_throttle kc_throttle;
	int			kc_err;

	unsigned long		flags;

	/* Last target access time */
	unsigned long		atime;
};

/*
 * Reclaim state flags.
 */
enum {
	DMZ_RECLAIM_KCOPY,
};

/*
 * Number of seconds of target BIO inactivity to consider the target idle.
 */
#define DMZ_IDLE_PERIOD			(10UL * HZ)

/*
 * Percentage of unmapped (free) random zones below which reclaim starts
 * even if the target is busy.
 */
#define DMZ_RECLAIM_LOW_UNMAP_RND	30

/*
 * Percentage of unmapped (free) random zones above which reclaim will
 * stop if the target is busy.
 */
#define DMZ_RECLAIM_HIGH_UNMAP_RND	50

/*
 * Align a sequential zone write pointer to chunk_block.
 */
static int dmz_reclaim_align_wp(struct dmz_reclaim *zrc, struct dm_zone *zone,
				sector_t block)
{
	struct dmz_metadata *zmd = zrc->metadata;
	sector_t wp_block = zone->wp_block;
	unsigned int nr_blocks;
	int ret;

	if (wp_block == block)
		return 0;

	if (wp_block > block)
		return -EIO;

	/*
	 * Zeroout the space between the write
	 * pointer and the requested position.
	 */
	nr_blocks = block - wp_block;
	ret = blkdev_issue_zeroout(zrc->dev->bdev,
				   dmz_start_sect(zmd, zone) + dmz_blk2sect(wp_block),
				   dmz_blk2sect(nr_blocks), GFP_NOIO, 0);
	if (ret) {
		dmz_dev_err(zrc->dev,
			    "Align zone %u wp %llu to %llu (wp+%u) blocks failed %d",
			    dmz_id(zmd, zone), (unsigned long long)wp_block,
			    (unsigned long long)block, nr_blocks, ret);
		return ret;
	}

	zone->wp_block = block;

	return 0;
}

/*
 * dm_kcopyd_copy end notification.
 */
static void dmz_reclaim_kcopy_end(int read_err, unsigned long write_err,
				  void *context)
{
	struct dmz_reclaim *zrc = context;

	if (read_err || write_err)
		zrc->kc_err = -EIO;
	else
		zrc->kc_err = 0;

	clear_bit_unlock(DMZ_RECLAIM_KCOPY, &zrc->flags);
	smp_mb__after_atomic();
	wake_up_bit(&zrc->flags, DMZ_RECLAIM_KCOPY);
}

/*
 * Copy valid blocks of src_zone into dst_zone.
 */
static int dmz_reclaim_copy(struct dmz_reclaim *zrc,
			    struct dm_zone *src_zone, struct dm_zone *dst_zone)
{
	struct dmz_metadata *zmd = zrc->metadata;
	struct dmz_dev *dev = zrc->dev;
	struct dm_io_region src, dst;
	sector_t block = 0, end_block;
	sector_t nr_blocks;
	sector_t src_zone_block;
	sector_t dst_zone_block;
	unsigned long flags = 0;
	int ret;

	if (dmz_is_seq(src_zone))
		end_block = src_zone->wp_block;
	else
		end_block = dev->zone_nr_blocks;
	src_zone_block = dmz_start_block(zmd, src_zone);
	dst_zone_block = dmz_start_block(zmd, dst_zone);

	if (dmz_is_seq(dst_zone))
		set_bit(DM_KCOPYD_WRITE_SEQ, &flags);

	while (block < end_block) {
		if (dev->flags & DMZ_BDEV_DYING)
			return -EIO;

		/* Get a valid region from the source zone */
		ret = dmz_first_valid_block(zmd, src_zone, &block);
		if (ret <= 0)
			return ret;
		nr_blocks = ret;

		/*
		 * If we are writing in a sequential zone, we must make sure
		 * that writes are sequential. So Zeroout any eventual hole
		 * between writes.
		 */
		if (dmz_is_seq(dst_zone)) {
			ret = dmz_reclaim_align_wp(zrc, dst_zone, block);
			if (ret)
				return ret;
		}

		src.bdev = dev->bdev;
		src.sector = dmz_blk2sect(src_zone_block + block);
		src.count = dmz_blk2sect(nr_blocks);

		dst.bdev = dev->bdev;
		dst.sector = dmz_blk2sect(dst_zone_block + block);
		dst.count = src.count;

		/* Copy the valid region */
		set_bit(DMZ_RECLAIM_KCOPY, &zrc->flags);
		dm_kcopyd_copy(zrc->kc, &src, 1, &dst, flags,
			       dmz_reclaim_kcopy_end, zrc);

		/* Wait for copy to complete */
		wait_on_bit_io(&zrc->flags, DMZ_RECLAIM_KCOPY,
			       TASK_UNINTERRUPTIBLE);
		if (zrc->kc_err)
			return zrc->kc_err;

		block += nr_blocks;
		if (dmz_is_seq(dst_zone))
			dst_zone->wp_block = block;
	}

	return 0;
}

/*
 * Move valid blocks of dzone buffer zone into dzone (after its write pointer)
 * and free the buffer zone.
 */
static int dmz_reclaim_buf(struct dmz_reclaim *zrc, struct dm_zone *dzone)
{
	struct dm_zone *bzone = dzone->bzone;
	sector_t chunk_block = dzone->wp_block;
	struct dmz_metadata *zmd = zrc->metadata;
	int ret;

	dmz_dev_debug(zrc->dev,
		      "Chunk %u, move buf zone %u (weight %u) to data zone %u (weight %u)",
		      dzone->chunk, dmz_id(zmd, bzone), dmz_weight(bzone),
		      dmz_id(zmd, dzone), dmz_weight(dzone));

	/* Flush data zone into the buffer zone */
	ret = dmz_reclaim_copy(zrc, bzone, dzone);
	if (ret < 0)
		return ret;

	dmz_lock_flush(zmd);

	/* Validate copied blocks */
	ret = dmz_merge_valid_blocks(zmd, bzone, dzone, chunk_block);
	if (ret == 0) {
		/* Free the buffer zone */
		dmz_invalidate_blocks(zmd, bzone, 0, zrc->dev->zone_nr_blocks);
		dmz_lock_map(zmd);
		dmz_unmap_zone(zmd, bzone);
		dmz_unlock_zone_reclaim(dzone);
		dmz_free_zone(zmd, bzone);
		dmz_unlock_map(zmd);
	}

	dmz_unlock_flush(zmd);

	return ret;
}

/*
 * Merge valid blocks of dzone into its buffer zone and free dzone.
 */
static int dmz_reclaim_seq_data(struct dmz_reclaim *zrc, struct dm_zone *dzone)
{
	unsigned int chunk = dzone->chunk;
	struct dm_zone *bzone = dzone->bzone;
	struct dmz_metadata *zmd = zrc->metadata;
	int ret = 0;

	dmz_dev_debug(zrc->dev,
		      "Chunk %u, move data zone %u (weight %u) to buf zone %u (weight %u)",
		      chunk, dmz_id(zmd, dzone), dmz_weight(dzone),
		      dmz_id(zmd, bzone), dmz_weight(bzone));

	/* Flush data zone into the buffer zone */
	ret = dmz_reclaim_copy(zrc, dzone, bzone);
	if (ret < 0)
		return ret;

	dmz_lock_flush(zmd);

	/* Validate copied blocks */
	ret = dmz_merge_valid_blocks(zmd, dzone, bzone, 0);
	if (ret == 0) {
		/*
		 * Free the data zone and remap the chunk to
		 * the buffer zone.
		 */
		dmz_invalidate_blocks(zmd, dzone, 0, zrc->dev->zone_nr_blocks);
		dmz_lock_map(zmd);
		dmz_unmap_zone(zmd, bzone);
		dmz_unmap_zone(zmd, dzone);
		dmz_unlock_zone_reclaim(dzone);
		dmz_free_zone(zmd, dzone);
		dmz_map_zone(zmd, bzone, chunk);
		dmz_unlock_map(zmd);
	}

	dmz_unlock_flush(zmd);

	return ret;
}

/*
 * Move valid blocks of the random data zone dzone into a free sequential zone.
 * Once blocks are moved, remap the zone chunk to the sequential zone.
 */
static int dmz_reclaim_rnd_data(struct dmz_reclaim *zrc, struct dm_zone *dzone)
{
	unsigned int chunk = dzone->chunk;
	struct dm_zone *szone = NULL;
	struct dmz_metadata *zmd = zrc->metadata;
	int ret;

	/* Get a free sequential zone */
	dmz_lock_map(zmd);
	szone = dmz_alloc_zone(zmd, DMZ_ALLOC_RECLAIM);
	dmz_unlock_map(zmd);
	if (!szone)
		return -ENOSPC;

	dmz_dev_debug(zrc->dev,
		      "Chunk %u, move rnd zone %u (weight %u) to seq zone %u",
		      chunk, dmz_id(zmd, dzone), dmz_weight(dzone),
		      dmz_id(zmd, szone));

	/* Flush the random data zone into the sequential zone */
	ret = dmz_reclaim_copy(zrc, dzone, szone);

	dmz_lock_flush(zmd);

	if (ret == 0) {
		/* Validate copied blocks */
		ret = dmz_copy_valid_blocks(zmd, dzone, szone);
	}
	if (ret) {
		/* Free the sequential zone */
		dmz_lock_map(zmd);
		dmz_free_zone(zmd, szone);
		dmz_unlock_map(zmd);
	} else {
		/* Free the data zone and remap the chunk */
		dmz_invalidate_blocks(zmd, dzone, 0, zrc->dev->zone_nr_blocks);
		dmz_lock_map(zmd);
		dmz_unmap_zone(zmd, dzone);
		dmz_unlock_zone_reclaim(dzone);
		dmz_free_zone(zmd, dzone);
		dmz_map_zone(zmd, szone, chunk);
		dmz_unlock_map(zmd);
	}

	dmz_unlock_flush(zmd);

	return ret;
}

/*
 * Reclaim an empty zone.
 */
static void dmz_reclaim_empty(struct dmz_reclaim *zrc, struct dm_zone *dzone)
{
	struct dmz_metadata *zmd = zrc->metadata;

	dmz_lock_flush(zmd);
	dmz_lock_map(zmd);
	dmz_unmap_zone(zmd, dzone);
	dmz_unlock_zone_reclaim(dzone);
	dmz_free_zone(zmd, dzone);
	dmz_unlock_map(zmd);
	dmz_unlock_flush(zmd);
}

/*
 * Find a candidate zone for reclaim and process it.
 */
static int dmz_do_reclaim(struct dmz_reclaim *zrc)
{
	struct dmz_metadata *zmd = zrc->metadata;
	struct dm_zone *dzone;
	struct dm_zone *rzone;
	unsigned long start;
	int ret;

	/* Get a data zone */
	dzone = dmz_get_zone_for_reclaim(zmd);
	if (IS_ERR(dzone))
		return PTR_ERR(dzone);

	start = jiffies;

	if (dmz_is_rnd(dzone)) {
		if (!dmz_weight(dzone)) {
			/* Empty zone */
			dmz_reclaim_empty(zrc, dzone);
			ret = 0;
		} else {
			/*
			 * Reclaim the random data zone by moving its
			 * valid data blocks to a free sequential zone.
			 */
			ret = dmz_reclaim_rnd_data(zrc, dzone);
		}
		rzone = dzone;

	} else {
		struct dm_zone *bzone = dzone->bzone;
		sector_t chunk_block = 0;

		ret = dmz_first_valid_block(zmd, bzone, &chunk_block);
		if (ret < 0)
			goto out;

		if (ret == 0 || chunk_block >= dzone->wp_block) {
			/*
			 * The buffer zone is empty or its valid blocks are
			 * after the data zone write pointer.
			 */
			ret = dmz_reclaim_buf(zrc, dzone);
			rzone = bzone;
		} else {
			/*
			 * Reclaim the data zone by merging it into the
			 * buffer zone so that the buffer zone itself can
			 * be later reclaimed.
			 */
			ret = dmz_reclaim_seq_data(zrc, dzone);
			rzone = dzone;
		}
	}
out:
	if (ret) {
		dmz_unlock_zone_reclaim(dzone);
		return ret;
	}

	ret = dmz_flush_metadata(zrc->metadata);
	if (ret) {
		dmz_dev_debug(zrc->dev,
			      "Metadata flush for zone %u failed, err %d\n",
			      dmz_id(zmd, rzone), ret);
		return ret;
	}

	dmz_dev_debug(zrc->dev, "Reclaimed zone %u in %u ms",
		      dmz_id(zmd, rzone), jiffies_to_msecs(jiffies - start));
	return 0;
}

/*
 * Test if the target device is idle.
 */
static inline int dmz_target_idle(struct dmz_reclaim *zrc)
{
	return time_is_before_jiffies(zrc->atime + DMZ_IDLE_PERIOD);
}

/*
 * Test if reclaim is necessary.
 */
static bool dmz_should_reclaim(struct dmz_reclaim *zrc)
{
	struct dmz_metadata *zmd = zrc->metadata;
	unsigned int nr_rnd = dmz_nr_rnd_zones(zmd);
	unsigned int nr_unmap_rnd = dmz_nr_unmap_rnd_zones(zmd);
	unsigned int p_unmap_rnd = nr_unmap_rnd * 100 / nr_rnd;

	/* Reclaim when idle */
	if (dmz_target_idle(zrc) && nr_unmap_rnd < nr_rnd)
		return true;

	/* If there are still plenty of random zones, do not reclaim */
	if (p_unmap_rnd >= DMZ_RECLAIM_HIGH_UNMAP_RND)
		return false;

	/*
	 * If the percentage of unmapped random zones is low,
	 * reclaim even if the target is busy.
	 */
	return p_unmap_rnd <= DMZ_RECLAIM_LOW_UNMAP_RND;
}

/*
 * Reclaim work function.
 */
static void dmz_reclaim_work(struct work_struct *work)
{
	struct dmz_reclaim *zrc = container_of(work, struct dmz_reclaim, work.work);
	struct dmz_metadata *zmd = zrc->metadata;
	unsigned int nr_rnd, nr_unmap_rnd;
	unsigned int p_unmap_rnd;
	int ret;

	if (dmz_bdev_is_dying(zrc->dev))
		return;

	if (!dmz_should_reclaim(zrc)) {
		mod_delayed_work(zrc->wq, &zrc->work, DMZ_IDLE_PERIOD);
		return;
	}

	/*
	 * We need to start reclaiming random zones: set up zone copy
	 * throttling to either go fast if we are very low on random zones
	 * and slower if there are still some free random zones to avoid
	 * as much as possible to negatively impact the user workload.
	 */
	nr_rnd = dmz_nr_rnd_zones(zmd);
	nr_unmap_rnd = dmz_nr_unmap_rnd_zones(zmd);
	p_unmap_rnd = nr_unmap_rnd * 100 / nr_rnd;
	if (dmz_target_idle(zrc) || p_unmap_rnd < DMZ_RECLAIM_LOW_UNMAP_RND / 2) {
		/* Idle or very low percentage: go fast */
		zrc->kc_throttle.throttle = 100;
	} else {
		/* Busy but we still have some random zone: throttle */
		zrc->kc_throttle.throttle = min(75U, 100U - p_unmap_rnd / 2);
	}

	dmz_dev_debug(zrc->dev,
		      "Reclaim (%u): %s, %u%% free rnd zones (%u/%u)",
		      zrc->kc_throttle.throttle,
		      (dmz_target_idle(zrc) ? "Idle" : "Busy"),
		      p_unmap_rnd, nr_unmap_rnd, nr_rnd);

	ret = dmz_do_reclaim(zrc);
	if (ret) {
		dmz_dev_debug(zrc->dev, "Reclaim error %d\n", ret);
		if (ret == -EIO)
			/*
			 * LLD might be performing some error handling sequence
			 * at the underlying device. To not interfere, do not
			 * attempt to schedule the next reclaim run immediately.
			 */
			return;
	}

	dmz_schedule_reclaim(zrc);
}

/*
 * Initialize reclaim.
 */
int dmz_ctr_reclaim(struct dmz_dev *dev, struct dmz_metadata *zmd,
		    struct dmz_reclaim **reclaim)
{
	struct dmz_reclaim *zrc;
	int ret;

	zrc = kzalloc(sizeof(struct dmz_reclaim), GFP_KERNEL);
	if (!zrc)
		return -ENOMEM;

	zrc->dev = dev;
	zrc->metadata = zmd;
	zrc->atime = jiffies;

	/* Reclaim kcopyd client */
	zrc->kc = dm_kcopyd_client_create(&zrc->kc_throttle);
	if (IS_ERR(zrc->kc)) {
		ret = PTR_ERR(zrc->kc);
		zrc->kc = NULL;
		goto err;
	}

	/* Reclaim work */
	INIT_DELAYED_WORK(&zrc->work, dmz_reclaim_work);
	zrc->wq = alloc_ordered_workqueue("dmz_rwq_%s", WQ_MEM_RECLAIM,
					  dev->name);
	if (!zrc->wq) {
		ret = -ENOMEM;
		goto err;
	}

	*reclaim = zrc;
	queue_delayed_work(zrc->wq, &zrc->work, 0);

	return 0;
err:
	if (zrc->kc)
		dm_kcopyd_client_destroy(zrc->kc);
	kfree(zrc);

	return ret;
}

/*
 * Terminate reclaim.
 */
void dmz_dtr_reclaim(struct dmz_reclaim *zrc)
{
	cancel_delayed_work_sync(&zrc->work);
	destroy_workqueue(zrc->wq);
	dm_kcopyd_client_destroy(zrc->kc);
	kfree(zrc);
}

/*
 * Suspend reclaim.
 */
void dmz_suspend_reclaim(struct dmz_reclaim *zrc)
{
	cancel_delayed_work_sync(&zrc->work);
}

/*
 * Resume reclaim.
 */
void dmz_resume_reclaim(struct dmz_reclaim *zrc)
{
	queue_delayed_work(zrc->wq, &zrc->work, DMZ_IDLE_PERIOD);
}

/*
 * BIO accounting.
 */
void dmz_reclaim_bio_acc(struct dmz_reclaim *zrc)
{
	zrc->atime = jiffies;
}

/*
 * Start reclaim if necessary.
 */
void dmz_schedule_reclaim(struct dmz_reclaim *zrc)
{
	if (dmz_should_reclaim(zrc))
		mod_delayed_work(zrc->wq, &zrc->work, 0);
}
Commit	Line	Data
bae9a0aa	1	// SPDX-License-Identifier: GPL-2.0-only
3b1a94c8 DLM	2	/*
	3	* Copyright (C) 2017 Western Digital Corporation or its affiliates.
	4	*
	5	* This file is released under the GPL.
	6	*/
	7
	8	#include "dm-zoned.h"
	9
	10	#include <linux/module.h>
	11
	12	#define DM_MSG_PREFIX "zoned reclaim"
	13
	14	struct dmz_reclaim {
	15	struct dmz_metadata *metadata;
	16	struct dmz_dev *dev;
	17
	18	struct delayed_work work;
	19	struct workqueue_struct *wq;
	20
	21	struct dm_kcopyd_client *kc;
	22	struct dm_kcopyd_throttle kc_throttle;
	23	int kc_err;
	24
	25	unsigned long flags;
	26
	27	/* Last target access time */
	28	unsigned long atime;
	29	};
	30
	31	/*
	32	* Reclaim state flags.
	33	*/
	34	enum {
	35	DMZ_RECLAIM_KCOPY,
	36	};
	37
	38	/*
	39	* Number of seconds of target BIO inactivity to consider the target idle.
	40	*/
75d66ffb	41	#define DMZ_IDLE_PERIOD (10UL * HZ)
3b1a94c8 DLM	42
	43	/*
	44	* Percentage of unmapped (free) random zones below which reclaim starts
	45	* even if the target is busy.
	46	*/
	47	#define DMZ_RECLAIM_LOW_UNMAP_RND 30
	48
	49	/*
	50	* Percentage of unmapped (free) random zones above which reclaim will
	51	* stop if the target is busy.
	52	*/
	53	#define DMZ_RECLAIM_HIGH_UNMAP_RND 50
	54
	55	/*
	56	* Align a sequential zone write pointer to chunk_block.
	57	*/
	58	static int dmz_reclaim_align_wp(struct dmz_reclaim zrc, struct dm_zone zone,
	59	sector_t block)
	60	{
	61	struct dmz_metadata *zmd = zrc->metadata;
	62	sector_t wp_block = zone->wp_block;
	63	unsigned int nr_blocks;
	64	int ret;
	65
	66	if (wp_block == block)
	67	return 0;
	68
	69	if (wp_block > block)
	70	return -EIO;
	71
	72	/*
	73	* Zeroout the space between the write
	74	* pointer and the requested position.
	75	*/
	76	nr_blocks = block - wp_block;
	77	ret = blkdev_issue_zeroout(zrc->dev->bdev,
	78	dmz_start_sect(zmd, zone) + dmz_blk2sect(wp_block),
4218a955	79	dmz_blk2sect(nr_blocks), GFP_NOIO, 0);
3b1a94c8 DLM	80	if (ret) {
	81	dmz_dev_err(zrc->dev,
	82	"Align zone %u wp %llu to %llu (wp+%u) blocks failed %d",
	83	dmz_id(zmd, zone), (unsigned long long)wp_block,
	84	(unsigned long long)block, nr_blocks, ret);
	85	return ret;
	86	}
	87
	88	zone->wp_block = block;
	89
	90	return 0;
	91	}
	92
	93	/*
	94	* dm_kcopyd_copy end notification.
	95	*/
	96	static void dmz_reclaim_kcopy_end(int read_err, unsigned long write_err,
	97	void *context)
	98	{
	99	struct dmz_reclaim *zrc = context;
	100
	101	if (read_err \|\| write_err)
	102	zrc->kc_err = -EIO;
	103	else
	104	zrc->kc_err = 0;
	105
	106	clear_bit_unlock(DMZ_RECLAIM_KCOPY, &zrc->flags);
	107	smp_mb__after_atomic();
	108	wake_up_bit(&zrc->flags, DMZ_RECLAIM_KCOPY);
	109	}
	110
	111	/*
	112	* Copy valid blocks of src_zone into dst_zone.
	113	*/
	114	static int dmz_reclaim_copy(struct dmz_reclaim *zrc,
	115	struct dm_zone src_zone, struct dm_zone dst_zone)
	116	{
	117	struct dmz_metadata *zmd = zrc->metadata;
	118	struct dmz_dev *dev = zrc->dev;
	119	struct dm_io_region src, dst;
	120	sector_t block = 0, end_block;
	121	sector_t nr_blocks;
	122	sector_t src_zone_block;
	123	sector_t dst_zone_block;
	124	unsigned long flags = 0;
	125	int ret;
	126
	127	if (dmz_is_seq(src_zone))
	128	end_block = src_zone->wp_block;
	129	else
	130	end_block = dev->zone_nr_blocks;
	131	src_zone_block = dmz_start_block(zmd, src_zone);
	132	dst_zone_block = dmz_start_block(zmd, dst_zone);
	133
	134	if (dmz_is_seq(dst_zone))
	135	set_bit(DM_KCOPYD_WRITE_SEQ, &flags);
	136
	137	while (block < end_block) {
75d66ffb DF	138	if (dev->flags & DMZ_BDEV_DYING)
	139	return -EIO;
	140
3b1a94c8 DLM	141	/* Get a valid region from the source zone */
	142	ret = dmz_first_valid_block(zmd, src_zone, &block);
	143	if (ret <= 0)
	144	return ret;
	145	nr_blocks = ret;
	146
	147	/*
	148	* If we are writing in a sequential zone, we must make sure
	149	* that writes are sequential. So Zeroout any eventual hole
	150	* between writes.
	151	*/
	152	if (dmz_is_seq(dst_zone)) {
	153	ret = dmz_reclaim_align_wp(zrc, dst_zone, block);
	154	if (ret)
	155	return ret;
	156	}
	157
	158	src.bdev = dev->bdev;
	159	src.sector = dmz_blk2sect(src_zone_block + block);
	160	src.count = dmz_blk2sect(nr_blocks);
	161
	162	dst.bdev = dev->bdev;
	163	dst.sector = dmz_blk2sect(dst_zone_block + block);
	164	dst.count = src.count;
	165
	166	/* Copy the valid region */
	167	set_bit(DMZ_RECLAIM_KCOPY, &zrc->flags);
7209049d MS	168	dm_kcopyd_copy(zrc->kc, &src, 1, &dst, flags,
7209049d MS	169	dmz_reclaim_kcopy_end, zrc);
3b1a94c8 DLM	170
	171	/* Wait for copy to complete */
	172	wait_on_bit_io(&zrc->flags, DMZ_RECLAIM_KCOPY,
	173	TASK_UNINTERRUPTIBLE);
	174	if (zrc->kc_err)
	175	return zrc->kc_err;
	176
	177	block += nr_blocks;
	178	if (dmz_is_seq(dst_zone))
	179	dst_zone->wp_block = block;
	180	}
	181
	182	return 0;
	183	}
	184
	185	/*
	186	* Move valid blocks of dzone buffer zone into dzone (after its write pointer)
	187	* and free the buffer zone.
	188	*/
	189	static int dmz_reclaim_buf(struct dmz_reclaim zrc, struct dm_zone dzone)
	190	{
	191	struct dm_zone *bzone = dzone->bzone;
	192	sector_t chunk_block = dzone->wp_block;
	193	struct dmz_metadata *zmd = zrc->metadata;
	194	int ret;
	195
	196	dmz_dev_debug(zrc->dev,
	197	"Chunk %u, move buf zone %u (weight %u) to data zone %u (weight %u)",
	198	dzone->chunk, dmz_id(zmd, bzone), dmz_weight(bzone),
	199	dmz_id(zmd, dzone), dmz_weight(dzone));
	200
	201	/* Flush data zone into the buffer zone */
	202	ret = dmz_reclaim_copy(zrc, bzone, dzone);
	203	if (ret < 0)
	204	return ret;
	205
	206	dmz_lock_flush(zmd);
	207
	208	/* Validate copied blocks */
	209	ret = dmz_merge_valid_blocks(zmd, bzone, dzone, chunk_block);
	210	if (ret == 0) {
	211	/* Free the buffer zone */
	212	dmz_invalidate_blocks(zmd, bzone, 0, zrc->dev->zone_nr_blocks);
	213	dmz_lock_map(zmd);
	214	dmz_unmap_zone(zmd, bzone);
	215	dmz_unlock_zone_reclaim(dzone);
	216	dmz_free_zone(zmd, bzone);
	217	dmz_unlock_map(zmd);
	218	}
	219
	220	dmz_unlock_flush(zmd);
	221
b234c6d7	222	return ret;
3b1a94c8 DLM	223	}
	224
	225	/*
	226	* Merge valid blocks of dzone into its buffer zone and free dzone.
	227	*/
	228	static int dmz_reclaim_seq_data(struct dmz_reclaim zrc, struct dm_zone dzone)
	229	{
	230	unsigned int chunk = dzone->chunk;
	231	struct dm_zone *bzone = dzone->bzone;
	232	struct dmz_metadata *zmd = zrc->metadata;
	233	int ret = 0;
	234
	235	dmz_dev_debug(zrc->dev,
	236	"Chunk %u, move data zone %u (weight %u) to buf zone %u (weight %u)",
	237	chunk, dmz_id(zmd, dzone), dmz_weight(dzone),
	238	dmz_id(zmd, bzone), dmz_weight(bzone));
	239
	240	/* Flush data zone into the buffer zone */
	241	ret = dmz_reclaim_copy(zrc, dzone, bzone);
	242	if (ret < 0)
	243	return ret;
	244
	245	dmz_lock_flush(zmd);
	246
	247	/* Validate copied blocks */
	248	ret = dmz_merge_valid_blocks(zmd, dzone, bzone, 0);
	249	if (ret == 0) {
	250	/*
	251	* Free the data zone and remap the chunk to
	252	* the buffer zone.
	253	*/
	254	dmz_invalidate_blocks(zmd, dzone, 0, zrc->dev->zone_nr_blocks);
	255	dmz_lock_map(zmd);
	256	dmz_unmap_zone(zmd, bzone);
	257	dmz_unmap_zone(zmd, dzone);
	258	dmz_unlock_zone_reclaim(dzone);
	259	dmz_free_zone(zmd, dzone);
	260	dmz_map_zone(zmd, bzone, chunk);
	261	dmz_unlock_map(zmd);
	262	}
	263
	264	dmz_unlock_flush(zmd);
	265
b234c6d7	266	return ret;
3b1a94c8 DLM	267	}
	268
	269	/*
	270	* Move valid blocks of the random data zone dzone into a free sequential zone.
	271	* Once blocks are moved, remap the zone chunk to the sequential zone.
	272	*/
	273	static int dmz_reclaim_rnd_data(struct dmz_reclaim zrc, struct dm_zone dzone)
	274	{
	275	unsigned int chunk = dzone->chunk;
	276	struct dm_zone *szone = NULL;
	277	struct dmz_metadata *zmd = zrc->metadata;
	278	int ret;
	279
	280	/* Get a free sequential zone */
	281	dmz_lock_map(zmd);
	282	szone = dmz_alloc_zone(zmd, DMZ_ALLOC_RECLAIM);
	283	dmz_unlock_map(zmd);
	284	if (!szone)
	285	return -ENOSPC;
	286
	287	dmz_dev_debug(zrc->dev,
	288	"Chunk %u, move rnd zone %u (weight %u) to seq zone %u",
	289	chunk, dmz_id(zmd, dzone), dmz_weight(dzone),
	290	dmz_id(zmd, szone));
	291
	292	/* Flush the random data zone into the sequential zone */
	293	ret = dmz_reclaim_copy(zrc, dzone, szone);
	294
	295	dmz_lock_flush(zmd);
	296
	297	if (ret == 0) {
	298	/* Validate copied blocks */
	299	ret = dmz_copy_valid_blocks(zmd, dzone, szone);
	300	}
	301	if (ret) {
	302	/* Free the sequential zone */
	303	dmz_lock_map(zmd);
	304	dmz_free_zone(zmd, szone);
	305	dmz_unlock_map(zmd);
	306	} else {
	307	/* Free the data zone and remap the chunk */
	308	dmz_invalidate_blocks(zmd, dzone, 0, zrc->dev->zone_nr_blocks);
	309	dmz_lock_map(zmd);
	310	dmz_unmap_zone(zmd, dzone);
	311	dmz_unlock_zone_reclaim(dzone);
	312	dmz_free_zone(zmd, dzone);
	313	dmz_map_zone(zmd, szone, chunk);
	314	dmz_unlock_map(zmd);
	315	}
	316
	317	dmz_unlock_flush(zmd);
	318
b234c6d7	319	return ret;
3b1a94c8 DLM	320	}
	321
	322	/*
	323	* Reclaim an empty zone.
	324	*/
	325	static void dmz_reclaim_empty(struct dmz_reclaim zrc, struct dm_zone dzone)
	326	{
	327	struct dmz_metadata *zmd = zrc->metadata;
	328
	329	dmz_lock_flush(zmd);
	330	dmz_lock_map(zmd);
	331	dmz_unmap_zone(zmd, dzone);
	332	dmz_unlock_zone_reclaim(dzone);
	333	dmz_free_zone(zmd, dzone);
	334	dmz_unlock_map(zmd);
	335	dmz_unlock_flush(zmd);
	336	}
	337
	338	/*
	339	* Find a candidate zone for reclaim and process it.
	340	*/
b234c6d7	341	static int dmz_do_reclaim(struct dmz_reclaim *zrc)
3b1a94c8 DLM	342	{
	343	struct dmz_metadata *zmd = zrc->metadata;
	344	struct dm_zone *dzone;
	345	struct dm_zone *rzone;
	346	unsigned long start;
	347	int ret;
	348
	349	/* Get a data zone */
	350	dzone = dmz_get_zone_for_reclaim(zmd);
b234c6d7 DF	351	if (IS_ERR(dzone))
b234c6d7 DF	352	return PTR_ERR(dzone);
3b1a94c8 DLM	353
	354	start = jiffies;
	355
	356	if (dmz_is_rnd(dzone)) {
	357	if (!dmz_weight(dzone)) {
	358	/* Empty zone */
	359	dmz_reclaim_empty(zrc, dzone);
	360	ret = 0;
	361	} else {
	362	/*
	363	* Reclaim the random data zone by moving its
	364	* valid data blocks to a free sequential zone.
	365	*/
	366	ret = dmz_reclaim_rnd_data(zrc, dzone);
	367	}
	368	rzone = dzone;
	369
	370	} else {
	371	struct dm_zone *bzone = dzone->bzone;
	372	sector_t chunk_block = 0;
	373
	374	ret = dmz_first_valid_block(zmd, bzone, &chunk_block);
	375	if (ret < 0)
	376	goto out;
	377
	378	if (ret == 0 \|\| chunk_block >= dzone->wp_block) {
	379	/*
	380	* The buffer zone is empty or its valid blocks are
	381	* after the data zone write pointer.
	382	*/
	383	ret = dmz_reclaim_buf(zrc, dzone);
	384	rzone = bzone;
	385	} else {
	386	/*
	387	* Reclaim the data zone by merging it into the
	388	* buffer zone so that the buffer zone itself can
	389	* be later reclaimed.
	390	*/
	391	ret = dmz_reclaim_seq_data(zrc, dzone);
	392	rzone = dzone;
	393	}
	394	}
	395	out:
	396	if (ret) {
	397	dmz_unlock_zone_reclaim(dzone);
b234c6d7	398	return ret;
3b1a94c8 DLM	399	}
3b1a94c8 DLM	400
b234c6d7 DF	401	ret = dmz_flush_metadata(zrc->metadata);
	402	if (ret) {
	403	dmz_dev_debug(zrc->dev,
	404	"Metadata flush for zone %u failed, err %d\n",
	405	dmz_id(zmd, rzone), ret);
	406	return ret;
	407	}
3b1a94c8 DLM	408
	409	dmz_dev_debug(zrc->dev, "Reclaimed zone %u in %u ms",
	410	dmz_id(zmd, rzone), jiffies_to_msecs(jiffies - start));
b234c6d7	411	return 0;
3b1a94c8 DLM	412	}
	413
	414	/*
	415	* Test if the target device is idle.
	416	*/
	417	static inline int dmz_target_idle(struct dmz_reclaim *zrc)
	418	{
	419	return time_is_before_jiffies(zrc->atime + DMZ_IDLE_PERIOD);
	420	}
	421
	422	/*
	423	* Test if reclaim is necessary.
	424	*/
	425	static bool dmz_should_reclaim(struct dmz_reclaim *zrc)
	426	{
	427	struct dmz_metadata *zmd = zrc->metadata;
	428	unsigned int nr_rnd = dmz_nr_rnd_zones(zmd);
	429	unsigned int nr_unmap_rnd = dmz_nr_unmap_rnd_zones(zmd);
	430	unsigned int p_unmap_rnd = nr_unmap_rnd * 100 / nr_rnd;
	431
	432	/* Reclaim when idle */
	433	if (dmz_target_idle(zrc) && nr_unmap_rnd < nr_rnd)
	434	return true;
	435
	436	/* If there are still plenty of random zones, do not reclaim */
	437	if (p_unmap_rnd >= DMZ_RECLAIM_HIGH_UNMAP_RND)
	438	return false;
	439
	440	/*
ad1bd578	441	* If the percentage of unmapped random zones is low,
3b1a94c8 DLM	442	* reclaim even if the target is busy.
	443	*/
	444	return p_unmap_rnd <= DMZ_RECLAIM_LOW_UNMAP_RND;
	445	}
	446
	447	/*
	448	* Reclaim work function.
	449	*/
	450	static void dmz_reclaim_work(struct work_struct *work)
	451	{
	452	struct dmz_reclaim *zrc = container_of(work, struct dmz_reclaim, work.work);
	453	struct dmz_metadata *zmd = zrc->metadata;
	454	unsigned int nr_rnd, nr_unmap_rnd;
	455	unsigned int p_unmap_rnd;
b234c6d7	456	int ret;
3b1a94c8	457
75d66ffb DF	458	if (dmz_bdev_is_dying(zrc->dev))
	459	return;
	460
3b1a94c8 DLM	461	if (!dmz_should_reclaim(zrc)) {
	462	mod_delayed_work(zrc->wq, &zrc->work, DMZ_IDLE_PERIOD);
	463	return;
	464	}
	465
	466	/*
	467	* We need to start reclaiming random zones: set up zone copy
	468	* throttling to either go fast if we are very low on random zones
	469	* and slower if there are still some free random zones to avoid
	470	* as much as possible to negatively impact the user workload.
	471	*/
	472	nr_rnd = dmz_nr_rnd_zones(zmd);
	473	nr_unmap_rnd = dmz_nr_unmap_rnd_zones(zmd);
	474	p_unmap_rnd = nr_unmap_rnd * 100 / nr_rnd;
	475	if (dmz_target_idle(zrc) \|\| p_unmap_rnd < DMZ_RECLAIM_LOW_UNMAP_RND / 2) {
	476	/* Idle or very low percentage: go fast */
	477	zrc->kc_throttle.throttle = 100;
	478	} else {
	479	/* Busy but we still have some random zone: throttle */
	480	zrc->kc_throttle.throttle = min(75U, 100U - p_unmap_rnd / 2);
	481	}
	482
	483	dmz_dev_debug(zrc->dev,
	484	"Reclaim (%u): %s, %u%% free rnd zones (%u/%u)",
	485	zrc->kc_throttle.throttle,
	486	(dmz_target_idle(zrc) ? "Idle" : "Busy"),
	487	p_unmap_rnd, nr_unmap_rnd, nr_rnd);
	488
b234c6d7	489	ret = dmz_do_reclaim(zrc);
75d66ffb	490	if (ret) {
b234c6d7	491	dmz_dev_debug(zrc->dev, "Reclaim error %d\n", ret);
75d66ffb DF	492	if (ret == -EIO)
	493	/*
	494	* LLD might be performing some error handling sequence
	495	* at the underlying device. To not interfere, do not
	496	* attempt to schedule the next reclaim run immediately.
	497	*/
	498	return;
	499	}
3b1a94c8 DLM	500
	501	dmz_schedule_reclaim(zrc);
	502	}
	503
	504	/*
	505	* Initialize reclaim.
	506	*/
	507	int dmz_ctr_reclaim(struct dmz_dev dev, struct dmz_metadata zmd,
	508	struct dmz_reclaim **reclaim)
	509	{
	510	struct dmz_reclaim *zrc;
	511	int ret;
	512
	513	zrc = kzalloc(sizeof(struct dmz_reclaim), GFP_KERNEL);
	514	if (!zrc)
	515	return -ENOMEM;
	516
	517	zrc->dev = dev;
	518	zrc->metadata = zmd;
	519	zrc->atime = jiffies;
	520
	521	/* Reclaim kcopyd client */
	522	zrc->kc = dm_kcopyd_client_create(&zrc->kc_throttle);
	523	if (IS_ERR(zrc->kc)) {
	524	ret = PTR_ERR(zrc->kc);
	525	zrc->kc = NULL;
	526	goto err;
	527	}
	528
	529	/* Reclaim work */
	530	INIT_DELAYED_WORK(&zrc->work, dmz_reclaim_work);
	531	zrc->wq = alloc_ordered_workqueue("dmz_rwq_%s", WQ_MEM_RECLAIM,
	532	dev->name);
	533	if (!zrc->wq) {
	534	ret = -ENOMEM;
	535	goto err;
	536	}
	537
	538	*reclaim = zrc;
	539	queue_delayed_work(zrc->wq, &zrc->work, 0);
	540
	541	return 0;
	542	err:
	543	if (zrc->kc)
	544	dm_kcopyd_client_destroy(zrc->kc);
	545	kfree(zrc);
	546
	547	return ret;
	548	}
	549
	550	/*
	551	* Terminate reclaim.
	552	*/
	553	void dmz_dtr_reclaim(struct dmz_reclaim *zrc)
	554	{
	555	cancel_delayed_work_sync(&zrc->work);
	556	destroy_workqueue(zrc->wq);
	557	dm_kcopyd_client_destroy(zrc->kc);
	558	kfree(zrc);
	559	}
	560
	561	/*
	562	* Suspend reclaim.
	563	*/
564	void dmz_suspend_reclaim(struct dmz_reclaim *zrc)
565	{
566	cancel_delayed_work_sync(&zrc->work);
567	}
568
569	/*
570	* Resume reclaim.
571	*/
572	void dmz_resume_reclaim(struct dmz_reclaim *zrc)
573	{
574	queue_delayed_work(zrc->wq, &zrc->work, DMZ_IDLE_PERIOD);
575	}
576
577	/*
578	* BIO accounting.
579	*/
580	void dmz_reclaim_bio_acc(struct dmz_reclaim *zrc)
581	{
582	zrc->atime = jiffies;
583	}
584
585	/*
586	* Start reclaim if necessary.
587	*/
588	void dmz_schedule_reclaim(struct dmz_reclaim *zrc)
589	{
590	if (dmz_should_reclaim(zrc))
591	mod_delayed_work(zrc->wq, &zrc->work, 0);
592	}
593