[thirdparty/linux.git] / drivers / block / null_blk_zoned.c

// SPDX-License-Identifier: GPL-2.0
#include <linux/vmalloc.h>
#include "null_blk.h"

#define CREATE_TRACE_POINTS
#include "null_blk_trace.h"

/* zone_size in MBs to sectors. */
#define ZONE_SIZE_SHIFT		11

static inline unsigned int null_zone_no(struct nullb_device *dev, sector_t sect)
{
	return sect >> ilog2(dev->zone_size_sects);
}

int null_init_zoned_dev(struct nullb_device *dev, struct request_queue *q)
{
	sector_t dev_size = (sector_t)dev->size * 1024 * 1024;
	sector_t sector = 0;
	unsigned int i;

	if (!is_power_of_2(dev->zone_size)) {
		pr_err("zone_size must be power-of-two\n");
		return -EINVAL;
	}

	dev->zone_size_sects = dev->zone_size << ZONE_SIZE_SHIFT;
	dev->nr_zones = dev_size >>
				(SECTOR_SHIFT + ilog2(dev->zone_size_sects));
	dev->zones = kvmalloc_array(dev->nr_zones, sizeof(struct blk_zone),
			GFP_KERNEL | __GFP_ZERO);
	if (!dev->zones)
		return -ENOMEM;

	if (dev->zone_nr_conv >= dev->nr_zones) {
		dev->zone_nr_conv = dev->nr_zones - 1;
		pr_info("changed the number of conventional zones to %u",
			dev->zone_nr_conv);
	}

	for (i = 0; i <  dev->zone_nr_conv; i++) {
		struct blk_zone *zone = &dev->zones[i];

		zone->start = sector;
		zone->len = dev->zone_size_sects;
		zone->wp = zone->start + zone->len;
		zone->type = BLK_ZONE_TYPE_CONVENTIONAL;
		zone->cond = BLK_ZONE_COND_NOT_WP;

		sector += dev->zone_size_sects;
	}

	for (i = dev->zone_nr_conv; i < dev->nr_zones; i++) {
		struct blk_zone *zone = &dev->zones[i];

		zone->start = zone->wp = sector;
		zone->len = dev->zone_size_sects;
		zone->type = BLK_ZONE_TYPE_SEQWRITE_REQ;
		zone->cond = BLK_ZONE_COND_EMPTY;

		sector += dev->zone_size_sects;
	}

	q->limits.zoned = BLK_ZONED_HM;
	blk_queue_flag_set(QUEUE_FLAG_ZONE_RESETALL, q);
	blk_queue_required_elevator_features(q, ELEVATOR_F_ZBD_SEQ_WRITE);

	return 0;
}

int null_register_zoned_dev(struct nullb *nullb)
{
	struct request_queue *q = nullb->q;

	if (queue_is_mq(q))
		return blk_revalidate_disk_zones(nullb->disk);

	blk_queue_chunk_sectors(q, nullb->dev->zone_size_sects);
	q->nr_zones = blkdev_nr_zones(nullb->disk);

	return 0;
}

void null_free_zoned_dev(struct nullb_device *dev)
{
	kvfree(dev->zones);
}

int null_report_zones(struct gendisk *disk, sector_t sector,
		unsigned int nr_zones, report_zones_cb cb, void *data)
{
	struct nullb *nullb = disk->private_data;
	struct nullb_device *dev = nullb->dev;
	unsigned int first_zone, i;
	struct blk_zone zone;
	int error;

	first_zone = null_zone_no(dev, sector);
	if (first_zone >= dev->nr_zones)
		return 0;

	nr_zones = min(nr_zones, dev->nr_zones - first_zone);
	trace_nullb_report_zones(nullb, nr_zones);

	for (i = 0; i < nr_zones; i++) {
		/*
		 * Stacked DM target drivers will remap the zone information by
		 * modifying the zone information passed to the report callback.
		 * So use a local copy to avoid corruption of the device zone
		 * array.
		 */
		memcpy(&zone, &dev->zones[first_zone + i],
		       sizeof(struct blk_zone));
		error = cb(&zone, i, data);
		if (error)
			return error;
	}

	return nr_zones;
}

size_t null_zone_valid_read_len(struct nullb *nullb,
				sector_t sector, unsigned int len)
{
	struct nullb_device *dev = nullb->dev;
	struct blk_zone *zone = &dev->zones[null_zone_no(dev, sector)];
	unsigned int nr_sectors = len >> SECTOR_SHIFT;

	/* Read must be below the write pointer position */
	if (zone->type == BLK_ZONE_TYPE_CONVENTIONAL ||
	    sector + nr_sectors <= zone->wp)
		return len;

	if (sector > zone->wp)
		return 0;

	return (zone->wp - sector) << SECTOR_SHIFT;
}

static blk_status_t null_zone_write(struct nullb_cmd *cmd, sector_t sector,
		     unsigned int nr_sectors)
{
	struct nullb_device *dev = cmd->nq->dev;
	unsigned int zno = null_zone_no(dev, sector);
	struct blk_zone *zone = &dev->zones[zno];
	blk_status_t ret;

	trace_nullb_zone_op(cmd, zno, zone->cond);

	if (zone->type == BLK_ZONE_TYPE_CONVENTIONAL)
		return null_process_cmd(cmd, REQ_OP_WRITE, sector, nr_sectors);

	switch (zone->cond) {
	case BLK_ZONE_COND_FULL:
		/* Cannot write to a full zone */
		return BLK_STS_IOERR;
	case BLK_ZONE_COND_EMPTY:
	case BLK_ZONE_COND_IMP_OPEN:
	case BLK_ZONE_COND_EXP_OPEN:
	case BLK_ZONE_COND_CLOSED:
		/* Writes must be at the write pointer position */
		if (sector != zone->wp)
			return BLK_STS_IOERR;

		if (zone->cond != BLK_ZONE_COND_EXP_OPEN)
			zone->cond = BLK_ZONE_COND_IMP_OPEN;

		ret = null_process_cmd(cmd, REQ_OP_WRITE, sector, nr_sectors);
		if (ret != BLK_STS_OK)
			return ret;

		zone->wp += nr_sectors;
		if (zone->wp == zone->start + zone->len)
			zone->cond = BLK_ZONE_COND_FULL;
		return BLK_STS_OK;
	default:
		/* Invalid zone condition */
		return BLK_STS_IOERR;
	}
}

static blk_status_t null_zone_mgmt(struct nullb_cmd *cmd, enum req_opf op,
				   sector_t sector)
{
	struct nullb_device *dev = cmd->nq->dev;
	unsigned int zone_no = null_zone_no(dev, sector);
	struct blk_zone *zone = &dev->zones[zone_no];
	size_t i;

	switch (op) {
	case REQ_OP_ZONE_RESET_ALL:
		for (i = 0; i < dev->nr_zones; i++) {
			if (zone[i].type == BLK_ZONE_TYPE_CONVENTIONAL)
				continue;
			zone[i].cond = BLK_ZONE_COND_EMPTY;
			zone[i].wp = zone[i].start;
		}
		break;
	case REQ_OP_ZONE_RESET:
		if (zone->type == BLK_ZONE_TYPE_CONVENTIONAL)
			return BLK_STS_IOERR;

		zone->cond = BLK_ZONE_COND_EMPTY;
		zone->wp = zone->start;
		break;
	case REQ_OP_ZONE_OPEN:
		if (zone->type == BLK_ZONE_TYPE_CONVENTIONAL)
			return BLK_STS_IOERR;
		if (zone->cond == BLK_ZONE_COND_FULL)
			return BLK_STS_IOERR;

		zone->cond = BLK_ZONE_COND_EXP_OPEN;
		break;
	case REQ_OP_ZONE_CLOSE:
		if (zone->type == BLK_ZONE_TYPE_CONVENTIONAL)
			return BLK_STS_IOERR;
		if (zone->cond == BLK_ZONE_COND_FULL)
			return BLK_STS_IOERR;

		if (zone->wp == zone->start)
			zone->cond = BLK_ZONE_COND_EMPTY;
		else
			zone->cond = BLK_ZONE_COND_CLOSED;
		break;
	case REQ_OP_ZONE_FINISH:
		if (zone->type == BLK_ZONE_TYPE_CONVENTIONAL)
			return BLK_STS_IOERR;

		zone->cond = BLK_ZONE_COND_FULL;
		zone->wp = zone->start + zone->len;
		break;
	default:
		return BLK_STS_NOTSUPP;
	}

	trace_nullb_zone_op(cmd, zone_no, zone->cond);
	return BLK_STS_OK;
}

blk_status_t null_process_zoned_cmd(struct nullb_cmd *cmd, enum req_opf op,
				    sector_t sector, sector_t nr_sectors)
{
	switch (op) {
	case REQ_OP_WRITE:
		return null_zone_write(cmd, sector, nr_sectors);
	case REQ_OP_ZONE_RESET:
	case REQ_OP_ZONE_RESET_ALL:
	case REQ_OP_ZONE_OPEN:
	case REQ_OP_ZONE_CLOSE:
	case REQ_OP_ZONE_FINISH:
		return null_zone_mgmt(cmd, op, sector);
	default:
		return null_process_cmd(cmd, op, sector, nr_sectors);
	}
}
Commit	Line	Data
ca4b2a01 MB	1	// SPDX-License-Identifier: GPL-2.0
	2	#include <linux/vmalloc.h>
	3	#include "null_blk.h"
	4
766c3297 CK	5	#define CREATE_TRACE_POINTS
	6	#include "null_blk_trace.h"
	7
ca4b2a01 MB	8	/* zone_size in MBs to sectors. */
	9	#define ZONE_SIZE_SHIFT 11
	10
	11	static inline unsigned int null_zone_no(struct nullb_device *dev, sector_t sect)
	12	{
	13	return sect >> ilog2(dev->zone_size_sects);
	14	}
	15
d205bde7	16	int null_init_zoned_dev(struct nullb_device dev, struct request_queue q)
ca4b2a01 MB	17	{
	18	sector_t dev_size = (sector_t)dev->size * 1024 * 1024;
	19	sector_t sector = 0;
	20	unsigned int i;
	21
	22	if (!is_power_of_2(dev->zone_size)) {
9c7eddf1	23	pr_err("zone_size must be power-of-two\n");
ca4b2a01 MB	24	return -EINVAL;
	25	}
	26
	27	dev->zone_size_sects = dev->zone_size << ZONE_SIZE_SHIFT;
	28	dev->nr_zones = dev_size >>
	29	(SECTOR_SHIFT + ilog2(dev->zone_size_sects));
	30	dev->zones = kvmalloc_array(dev->nr_zones, sizeof(struct blk_zone),
	31	GFP_KERNEL \| __GFP_ZERO);
	32	if (!dev->zones)
	33	return -ENOMEM;
	34
ea2c18e1 MS	35	if (dev->zone_nr_conv >= dev->nr_zones) {
ea2c18e1 MS	36	dev->zone_nr_conv = dev->nr_zones - 1;
9c7eddf1	37	pr_info("changed the number of conventional zones to %u",
ea2c18e1 MS	38	dev->zone_nr_conv);
	39	}
	40
	41	for (i = 0; i < dev->zone_nr_conv; i++) {
	42	struct blk_zone *zone = &dev->zones[i];
	43
	44	zone->start = sector;
	45	zone->len = dev->zone_size_sects;
	46	zone->wp = zone->start + zone->len;
	47	zone->type = BLK_ZONE_TYPE_CONVENTIONAL;
	48	zone->cond = BLK_ZONE_COND_NOT_WP;
	49
	50	sector += dev->zone_size_sects;
	51	}
	52
	53	for (i = dev->zone_nr_conv; i < dev->nr_zones; i++) {
ca4b2a01 MB	54	struct blk_zone *zone = &dev->zones[i];
	55
	56	zone->start = zone->wp = sector;
	57	zone->len = dev->zone_size_sects;
	58	zone->type = BLK_ZONE_TYPE_SEQWRITE_REQ;
	59	zone->cond = BLK_ZONE_COND_EMPTY;
	60
	61	sector += dev->zone_size_sects;
	62	}
	63
d205bde7 DLM	64	q->limits.zoned = BLK_ZONED_HM;
	65	blk_queue_flag_set(QUEUE_FLAG_ZONE_RESETALL, q);
	66	blk_queue_required_elevator_features(q, ELEVATOR_F_ZBD_SEQ_WRITE);
	67
	68	return 0;
	69	}
	70
	71	int null_register_zoned_dev(struct nullb *nullb)
	72	{
	73	struct request_queue *q = nullb->q;
	74
	75	if (queue_is_mq(q))
	76	return blk_revalidate_disk_zones(nullb->disk);
	77
	78	blk_queue_chunk_sectors(q, nullb->dev->zone_size_sects);
	79	q->nr_zones = blkdev_nr_zones(nullb->disk);
	80
ca4b2a01 MB	81	return 0;
	82	}
	83
d205bde7	84	void null_free_zoned_dev(struct nullb_device *dev)
ca4b2a01 MB	85	{
	86	kvfree(dev->zones);
	87	}
	88
7fc8fb51	89	int null_report_zones(struct gendisk *disk, sector_t sector,
d4100351	90	unsigned int nr_zones, report_zones_cb cb, void *data)
ca4b2a01	91	{
e76239a3 CH	92	struct nullb *nullb = disk->private_data;
e76239a3 CH	93	struct nullb_device *dev = nullb->dev;
d4100351 CH	94	unsigned int first_zone, i;
	95	struct blk_zone zone;
	96	int error;
ca4b2a01	97
d4100351 CH	98	first_zone = null_zone_no(dev, sector);
	99	if (first_zone >= dev->nr_zones)
	100	return 0;
ca4b2a01	101
d4100351	102	nr_zones = min(nr_zones, dev->nr_zones - first_zone);
766c3297 CK	103	trace_nullb_report_zones(nullb, nr_zones);
766c3297 CK	104
d4100351 CH	105	for (i = 0; i < nr_zones; i++) {
	106	/*
	107	* Stacked DM target drivers will remap the zone information by
	108	* modifying the zone information passed to the report callback.
	109	* So use a local copy to avoid corruption of the device zone
	110	* array.
	111	*/
	112	memcpy(&zone, &dev->zones[first_zone + i],
	113	sizeof(struct blk_zone));
	114	error = cb(&zone, i, data);
	115	if (error)
	116	return error;
	117	}
ca4b2a01	118
d4100351	119	return nr_zones;
ca4b2a01 MB	120	}
ca4b2a01 MB	121
dd85b492 AJ	122	size_t null_zone_valid_read_len(struct nullb *nullb,
	123	sector_t sector, unsigned int len)
	124	{
	125	struct nullb_device *dev = nullb->dev;
	126	struct blk_zone *zone = &dev->zones[null_zone_no(dev, sector)];
	127	unsigned int nr_sectors = len >> SECTOR_SHIFT;
	128
	129	/* Read must be below the write pointer position */
	130	if (zone->type == BLK_ZONE_TYPE_CONVENTIONAL \|\|
	131	sector + nr_sectors <= zone->wp)
	132	return len;
	133
	134	if (sector > zone->wp)
	135	return 0;
	136
	137	return (zone->wp - sector) << SECTOR_SHIFT;
	138	}
	139
fceb5d1b	140	static blk_status_t null_zone_write(struct nullb_cmd *cmd, sector_t sector,
b228ba1c	141	unsigned int nr_sectors)
ca4b2a01 MB	142	{
ca4b2a01 MB	143	struct nullb_device *dev = cmd->nq->dev;
ca4b2a01 MB	144	unsigned int zno = null_zone_no(dev, sector);
ca4b2a01 MB	145	struct blk_zone *zone = &dev->zones[zno];
9dd44c7e DLM	146	blk_status_t ret;
	147
	148	trace_nullb_zone_op(cmd, zno, zone->cond);
	149
	150	if (zone->type == BLK_ZONE_TYPE_CONVENTIONAL)
	151	return null_process_cmd(cmd, REQ_OP_WRITE, sector, nr_sectors);
ca4b2a01 MB	152
	153	switch (zone->cond) {
	154	case BLK_ZONE_COND_FULL:
	155	/* Cannot write to a full zone */
fceb5d1b	156	return BLK_STS_IOERR;
ca4b2a01 MB	157	case BLK_ZONE_COND_EMPTY:
ca4b2a01 MB	158	case BLK_ZONE_COND_IMP_OPEN:
16c731fe DLM	159	case BLK_ZONE_COND_EXP_OPEN:
16c731fe DLM	160	case BLK_ZONE_COND_CLOSED:
ca4b2a01	161	/* Writes must be at the write pointer position */
fceb5d1b CK	162	if (sector != zone->wp)
fceb5d1b CK	163	return BLK_STS_IOERR;
ca4b2a01	164
16c731fe	165	if (zone->cond != BLK_ZONE_COND_EXP_OPEN)
ca4b2a01 MB	166	zone->cond = BLK_ZONE_COND_IMP_OPEN;
ca4b2a01 MB	167
9dd44c7e DLM	168	ret = null_process_cmd(cmd, REQ_OP_WRITE, sector, nr_sectors);
	169	if (ret != BLK_STS_OK)
	170	return ret;
	171
b228ba1c	172	zone->wp += nr_sectors;
ca4b2a01 MB	173	if (zone->wp == zone->start + zone->len)
ca4b2a01 MB	174	zone->cond = BLK_ZONE_COND_FULL;
9dd44c7e	175	return BLK_STS_OK;
ca4b2a01 MB	176	default:
ca4b2a01 MB	177	/* Invalid zone condition */
fceb5d1b	178	return BLK_STS_IOERR;
ca4b2a01 MB	179	}
	180	}
	181
da644b2c AJ	182	static blk_status_t null_zone_mgmt(struct nullb_cmd *cmd, enum req_opf op,
da644b2c AJ	183	sector_t sector)
ca4b2a01 MB	184	{
ca4b2a01 MB	185	struct nullb_device *dev = cmd->nq->dev;
766c3297 CK	186	unsigned int zone_no = null_zone_no(dev, sector);
766c3297 CK	187	struct blk_zone *zone = &dev->zones[zone_no];
a61dbfb1 CK	188	size_t i;
a61dbfb1 CK	189
da644b2c	190	switch (op) {
a61dbfb1 CK	191	case REQ_OP_ZONE_RESET_ALL:
	192	for (i = 0; i < dev->nr_zones; i++) {
	193	if (zone[i].type == BLK_ZONE_TYPE_CONVENTIONAL)
	194	continue;
	195	zone[i].cond = BLK_ZONE_COND_EMPTY;
	196	zone[i].wp = zone[i].start;
	197	}
	198	break;
	199	case REQ_OP_ZONE_RESET:
fceb5d1b CK	200	if (zone->type == BLK_ZONE_TYPE_CONVENTIONAL)
fceb5d1b CK	201	return BLK_STS_IOERR;
ca4b2a01	202
a61dbfb1 CK	203	zone->cond = BLK_ZONE_COND_EMPTY;
	204	zone->wp = zone->start;
	205	break;
da644b2c AJ	206	case REQ_OP_ZONE_OPEN:
	207	if (zone->type == BLK_ZONE_TYPE_CONVENTIONAL)
	208	return BLK_STS_IOERR;
	209	if (zone->cond == BLK_ZONE_COND_FULL)
	210	return BLK_STS_IOERR;
	211
	212	zone->cond = BLK_ZONE_COND_EXP_OPEN;
	213	break;
	214	case REQ_OP_ZONE_CLOSE:
	215	if (zone->type == BLK_ZONE_TYPE_CONVENTIONAL)
	216	return BLK_STS_IOERR;
	217	if (zone->cond == BLK_ZONE_COND_FULL)
	218	return BLK_STS_IOERR;
	219
c7d776f8 DLM	220	if (zone->wp == zone->start)
	221	zone->cond = BLK_ZONE_COND_EMPTY;
	222	else
	223	zone->cond = BLK_ZONE_COND_CLOSED;
da644b2c AJ	224	break;
	225	case REQ_OP_ZONE_FINISH:
	226	if (zone->type == BLK_ZONE_TYPE_CONVENTIONAL)
	227	return BLK_STS_IOERR;
	228
	229	zone->cond = BLK_ZONE_COND_FULL;
	230	zone->wp = zone->start + zone->len;
	231	break;
a61dbfb1	232	default:
79a85e21	233	return BLK_STS_NOTSUPP;
ea2c18e1	234	}
766c3297 CK	235
766c3297 CK	236	trace_nullb_zone_op(cmd, zone_no, zone->cond);
fceb5d1b CK	237	return BLK_STS_OK;
	238	}
	239
9dd44c7e DLM	240	blk_status_t null_process_zoned_cmd(struct nullb_cmd *cmd, enum req_opf op,
9dd44c7e DLM	241	sector_t sector, sector_t nr_sectors)
fceb5d1b CK	242	{
	243	switch (op) {
	244	case REQ_OP_WRITE:
	245	return null_zone_write(cmd, sector, nr_sectors);
	246	case REQ_OP_ZONE_RESET:
	247	case REQ_OP_ZONE_RESET_ALL:
da644b2c AJ	248	case REQ_OP_ZONE_OPEN:
	249	case REQ_OP_ZONE_CLOSE:
	250	case REQ_OP_ZONE_FINISH:
	251	return null_zone_mgmt(cmd, op, sector);
fceb5d1b	252	default:
9dd44c7e	253	return null_process_cmd(cmd, op, sector, nr_sectors);
fceb5d1b	254	}
ca4b2a01	255	}