[linux-2.6-block.git] / drivers / lightnvm / pblk-read.c

/*
 * Copyright (C) 2016 CNEX Labs
 * Initial release: Javier Gonzalez <javier@cnexlabs.com>
 *                  Matias Bjorling <matias@cnexlabs.com>
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public License version
 * 2 as published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it will be useful, but
 * WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * General Public License for more details.
 *
 * pblk-read.c - pblk's read path
 */

#include "pblk.h"

/*
 * There is no guarantee that the value read from cache has not been updated and
 * resides at another location in the cache. We guarantee though that if the
 * value is read from the cache, it belongs to the mapped lba. In order to
 * guarantee and order between writes and reads are ordered, a flush must be
 * issued.
 */
static int pblk_read_from_cache(struct pblk *pblk, struct bio *bio,
				sector_t lba, struct ppa_addr ppa,
				int bio_iter)
{
#ifdef CONFIG_NVM_DEBUG
	/* Callers must ensure that the ppa points to a cache address */
	BUG_ON(pblk_ppa_empty(ppa));
	BUG_ON(!pblk_addr_in_cache(ppa));
#endif

	return pblk_rb_copy_to_bio(&pblk->rwb, bio, lba,
					pblk_addr_to_cacheline(ppa), bio_iter);
}

static void pblk_read_ppalist_rq(struct pblk *pblk, struct nvm_rq *rqd,
				 unsigned long *read_bitmap)
{
	struct bio *bio = rqd->bio;
	struct ppa_addr ppas[PBLK_MAX_REQ_ADDRS];
	sector_t blba = pblk_get_lba(bio);
	int nr_secs = rqd->nr_ppas;
	int advanced_bio = 0;
	int i, j = 0;

	/* logic error: lba out-of-bounds. Ignore read request */
	if (blba + nr_secs >= pblk->rl.nr_secs) {
		WARN(1, "pblk: read lbas out of bounds\n");
		return;
	}

	pblk_lookup_l2p_seq(pblk, ppas, blba, nr_secs);

	for (i = 0; i < nr_secs; i++) {
		struct ppa_addr p = ppas[i];
		sector_t lba = blba + i;

retry:
		if (pblk_ppa_empty(p)) {
			WARN_ON(test_and_set_bit(i, read_bitmap));
			continue;
		}

		/* Try to read from write buffer. The address is later checked
		 * on the write buffer to prevent retrieving overwritten data.
		 */
		if (pblk_addr_in_cache(p)) {
			if (!pblk_read_from_cache(pblk, bio, lba, p, i)) {
				pblk_lookup_l2p_seq(pblk, &p, lba, 1);
				goto retry;
			}
			WARN_ON(test_and_set_bit(i, read_bitmap));
			advanced_bio = 1;
		} else {
			/* Read from media non-cached sectors */
			rqd->ppa_list[j++] = p;
		}

		if (advanced_bio)
			bio_advance(bio, PBLK_EXPOSED_PAGE_SIZE);
	}

#ifdef CONFIG_NVM_DEBUG
	atomic_long_add(nr_secs, &pblk->inflight_reads);
#endif
}

static int pblk_submit_read_io(struct pblk *pblk, struct nvm_rq *rqd)
{
	int err;

	rqd->flags = pblk_set_read_mode(pblk);

	err = pblk_submit_io(pblk, rqd);
	if (err)
		return NVM_IO_ERR;

	return NVM_IO_OK;
}

static void pblk_end_io_read(struct nvm_rq *rqd)
{
	struct pblk *pblk = rqd->private;
	struct nvm_tgt_dev *dev = pblk->dev;
	struct pblk_r_ctx *r_ctx = nvm_rq_to_pdu(rqd);
	struct bio *bio = rqd->bio;

	if (rqd->error)
		pblk_log_read_err(pblk, rqd);
#ifdef CONFIG_NVM_DEBUG
	else
		WARN_ONCE(bio->bi_error, "pblk: corrupted read error\n");
#endif

	if (rqd->nr_ppas > 1)
		nvm_dev_dma_free(dev->parent, rqd->ppa_list, rqd->dma_ppa_list);

	bio_put(bio);
	if (r_ctx->orig_bio) {
#ifdef CONFIG_NVM_DEBUG
		WARN_ONCE(r_ctx->orig_bio->bi_error,
						"pblk: corrupted read bio\n");
#endif
		bio_endio(r_ctx->orig_bio);
		bio_put(r_ctx->orig_bio);
	}

#ifdef CONFIG_NVM_DEBUG
	atomic_long_add(rqd->nr_ppas, &pblk->sync_reads);
	atomic_long_sub(rqd->nr_ppas, &pblk->inflight_reads);
#endif

	pblk_free_rqd(pblk, rqd, READ);
}

static int pblk_fill_partial_read_bio(struct pblk *pblk, struct nvm_rq *rqd,
				      unsigned int bio_init_idx,
				      unsigned long *read_bitmap)
{
	struct bio *new_bio, *bio = rqd->bio;
	struct bio_vec src_bv, dst_bv;
	void *ppa_ptr = NULL;
	void *src_p, *dst_p;
	dma_addr_t dma_ppa_list = 0;
	int nr_secs = rqd->nr_ppas;
	int nr_holes = nr_secs - bitmap_weight(read_bitmap, nr_secs);
	int i, ret, hole;
	DECLARE_COMPLETION_ONSTACK(wait);

	new_bio = bio_alloc(GFP_KERNEL, nr_holes);
	if (!new_bio) {
		pr_err("pblk: could not alloc read bio\n");
		return NVM_IO_ERR;
	}

	if (pblk_bio_add_pages(pblk, new_bio, GFP_KERNEL, nr_holes))
		goto err;

	if (nr_holes != new_bio->bi_vcnt) {
		pr_err("pblk: malformed bio\n");
		goto err;
	}

	new_bio->bi_iter.bi_sector = 0; /* internal bio */
	bio_set_op_attrs(new_bio, REQ_OP_READ, 0);
	new_bio->bi_private = &wait;
	new_bio->bi_end_io = pblk_end_bio_sync;

	rqd->bio = new_bio;
	rqd->nr_ppas = nr_holes;
	rqd->end_io = NULL;

	if (unlikely(nr_secs > 1 && nr_holes == 1)) {
		ppa_ptr = rqd->ppa_list;
		dma_ppa_list = rqd->dma_ppa_list;
		rqd->ppa_addr = rqd->ppa_list[0];
	}

	ret = pblk_submit_read_io(pblk, rqd);
	if (ret) {
		bio_put(rqd->bio);
		pr_err("pblk: read IO submission failed\n");
		goto err;
	}

	if (!wait_for_completion_io_timeout(&wait,
				msecs_to_jiffies(PBLK_COMMAND_TIMEOUT_MS))) {
		pr_err("pblk: partial read I/O timed out\n");
	}

	if (rqd->error) {
		atomic_long_inc(&pblk->read_failed);
#ifdef CONFIG_NVM_DEBUG
		pblk_print_failed_rqd(pblk, rqd, rqd->error);
#endif
	}

	if (unlikely(nr_secs > 1 && nr_holes == 1)) {
		rqd->ppa_list = ppa_ptr;
		rqd->dma_ppa_list = dma_ppa_list;
	}

	/* Fill the holes in the original bio */
	i = 0;
	hole = find_first_zero_bit(read_bitmap, nr_secs);
	do {
		src_bv = new_bio->bi_io_vec[i++];
		dst_bv = bio->bi_io_vec[bio_init_idx + hole];

		src_p = kmap_atomic(src_bv.bv_page);
		dst_p = kmap_atomic(dst_bv.bv_page);

		memcpy(dst_p + dst_bv.bv_offset,
			src_p + src_bv.bv_offset,
			PBLK_EXPOSED_PAGE_SIZE);

		kunmap_atomic(src_p);
		kunmap_atomic(dst_p);

		mempool_free(src_bv.bv_page, pblk->page_pool);

		hole = find_next_zero_bit(read_bitmap, nr_secs, hole + 1);
	} while (hole < nr_secs);

	bio_put(new_bio);

	/* Complete the original bio and associated request */
	rqd->bio = bio;
	rqd->nr_ppas = nr_secs;
	rqd->private = pblk;

	bio_endio(bio);
	pblk_end_io_read(rqd);
	return NVM_IO_OK;

err:
	/* Free allocated pages in new bio */
	pblk_bio_free_pages(pblk, bio, 0, new_bio->bi_vcnt);
	rqd->private = pblk;
	pblk_end_io_read(rqd);
	return NVM_IO_ERR;
}

static void pblk_read_rq(struct pblk *pblk, struct nvm_rq *rqd,
			 unsigned long *read_bitmap)
{
	struct bio *bio = rqd->bio;
	struct ppa_addr ppa;
	sector_t lba = pblk_get_lba(bio);

	/* logic error: lba out-of-bounds. Ignore read request */
	if (lba >= pblk->rl.nr_secs) {
		WARN(1, "pblk: read lba out of bounds\n");
		return;
	}

	pblk_lookup_l2p_seq(pblk, &ppa, lba, 1);

#ifdef CONFIG_NVM_DEBUG
	atomic_long_inc(&pblk->inflight_reads);
#endif

retry:
	if (pblk_ppa_empty(ppa)) {
		WARN_ON(test_and_set_bit(0, read_bitmap));
		return;
	}

	/* Try to read from write buffer. The address is later checked on the
	 * write buffer to prevent retrieving overwritten data.
	 */
	if (pblk_addr_in_cache(ppa)) {
		if (!pblk_read_from_cache(pblk, bio, lba, ppa, 0)) {
			pblk_lookup_l2p_seq(pblk, &ppa, lba, 1);
			goto retry;
		}
		WARN_ON(test_and_set_bit(0, read_bitmap));
	} else {
		rqd->ppa_addr = ppa;
	}
}

int pblk_submit_read(struct pblk *pblk, struct bio *bio)
{
	struct nvm_tgt_dev *dev = pblk->dev;
	unsigned int nr_secs = pblk_get_secs(bio);
	struct nvm_rq *rqd;
	unsigned long read_bitmap; /* Max 64 ppas per request */
	unsigned int bio_init_idx;
	int ret = NVM_IO_ERR;

	if (nr_secs > PBLK_MAX_REQ_ADDRS)
		return NVM_IO_ERR;

	bitmap_zero(&read_bitmap, nr_secs);

	rqd = pblk_alloc_rqd(pblk, READ);
	if (IS_ERR(rqd)) {
		pr_err_ratelimited("pblk: not able to alloc rqd");
		return NVM_IO_ERR;
	}

	rqd->opcode = NVM_OP_PREAD;
	rqd->bio = bio;
	rqd->nr_ppas = nr_secs;
	rqd->private = pblk;
	rqd->end_io = pblk_end_io_read;

	/* Save the index for this bio's start. This is needed in case
	 * we need to fill a partial read.
	 */
	bio_init_idx = pblk_get_bi_idx(bio);

	if (nr_secs > 1) {
		rqd->ppa_list = nvm_dev_dma_alloc(dev->parent, GFP_KERNEL,
						&rqd->dma_ppa_list);
		if (!rqd->ppa_list) {
			pr_err("pblk: not able to allocate ppa list\n");
			goto fail_rqd_free;
		}

		pblk_read_ppalist_rq(pblk, rqd, &read_bitmap);
	} else {
		pblk_read_rq(pblk, rqd, &read_bitmap);
	}

	bio_get(bio);
	if (bitmap_full(&read_bitmap, nr_secs)) {
		bio_endio(bio);
		pblk_end_io_read(rqd);
		return NVM_IO_OK;
	}

	/* All sectors are to be read from the device */
	if (bitmap_empty(&read_bitmap, rqd->nr_ppas)) {
		struct bio *int_bio = NULL;
		struct pblk_r_ctx *r_ctx = nvm_rq_to_pdu(rqd);

		/* Clone read bio to deal with read errors internally */
		int_bio = bio_clone_bioset(bio, GFP_KERNEL, fs_bio_set);
		if (!int_bio) {
			pr_err("pblk: could not clone read bio\n");
			return NVM_IO_ERR;
		}

		rqd->bio = int_bio;
		r_ctx->orig_bio = bio;

		ret = pblk_submit_read_io(pblk, rqd);
		if (ret) {
			pr_err("pblk: read IO submission failed\n");
			if (int_bio)
				bio_put(int_bio);
			return ret;
		}

		return NVM_IO_OK;
	}

	/* The read bio request could be partially filled by the write buffer,
	 * but there are some holes that need to be read from the drive.
	 */
	ret = pblk_fill_partial_read_bio(pblk, rqd, bio_init_idx, &read_bitmap);
	if (ret) {
		pr_err("pblk: failed to perform partial read\n");
		return ret;
	}

	return NVM_IO_OK;

fail_rqd_free:
	pblk_free_rqd(pblk, rqd, READ);
	return ret;
}

static int read_ppalist_rq_gc(struct pblk *pblk, struct nvm_rq *rqd,
			      struct pblk_line *line, u64 *lba_list,
			      unsigned int nr_secs)
{
	struct ppa_addr ppas[PBLK_MAX_REQ_ADDRS];
	int valid_secs = 0;
	int i;

	pblk_lookup_l2p_rand(pblk, ppas, lba_list, nr_secs);

	for (i = 0; i < nr_secs; i++) {
		if (pblk_addr_in_cache(ppas[i]) || ppas[i].g.blk != line->id ||
						pblk_ppa_empty(ppas[i])) {
			lba_list[i] = ADDR_EMPTY;
			continue;
		}

		rqd->ppa_list[valid_secs++] = ppas[i];
	}

#ifdef CONFIG_NVM_DEBUG
	atomic_long_add(valid_secs, &pblk->inflight_reads);
#endif
	return valid_secs;
}

static int read_rq_gc(struct pblk *pblk, struct nvm_rq *rqd,
		      struct pblk_line *line, sector_t lba)
{
	struct ppa_addr ppa;
	int valid_secs = 0;

	/* logic error: lba out-of-bounds */
	if (lba >= pblk->rl.nr_secs) {
		WARN(1, "pblk: read lba out of bounds\n");
		goto out;
	}

	if (lba == ADDR_EMPTY)
		goto out;

	spin_lock(&pblk->trans_lock);
	ppa = pblk_trans_map_get(pblk, lba);
	spin_unlock(&pblk->trans_lock);

	/* Ignore updated values until the moment */
	if (pblk_addr_in_cache(ppa) || ppa.g.blk != line->id ||
							pblk_ppa_empty(ppa))
		goto out;

	rqd->ppa_addr = ppa;
	valid_secs = 1;

#ifdef CONFIG_NVM_DEBUG
	atomic_long_inc(&pblk->inflight_reads);
#endif

out:
	return valid_secs;
}

int pblk_submit_read_gc(struct pblk *pblk, u64 *lba_list, void *data,
			unsigned int nr_secs, unsigned int *secs_to_gc,
			struct pblk_line *line)
{
	struct nvm_tgt_dev *dev = pblk->dev;
	struct nvm_geo *geo = &dev->geo;
	struct request_queue *q = dev->q;
	struct bio *bio;
	struct nvm_rq rqd;
	int ret, data_len;
	DECLARE_COMPLETION_ONSTACK(wait);

	memset(&rqd, 0, sizeof(struct nvm_rq));

	if (nr_secs > 1) {
		rqd.ppa_list = nvm_dev_dma_alloc(dev->parent, GFP_KERNEL,
							&rqd.dma_ppa_list);
		if (!rqd.ppa_list)
			return NVM_IO_ERR;

		*secs_to_gc = read_ppalist_rq_gc(pblk, &rqd, line, lba_list,
								nr_secs);
		if (*secs_to_gc == 1) {
			struct ppa_addr ppa;

			ppa = rqd.ppa_list[0];
			nvm_dev_dma_free(dev->parent, rqd.ppa_list,
							rqd.dma_ppa_list);
			rqd.ppa_addr = ppa;
		}
	} else {
		*secs_to_gc = read_rq_gc(pblk, &rqd, line, lba_list[0]);
	}

	if (!(*secs_to_gc))
		goto out;

	data_len = (*secs_to_gc) * geo->sec_size;
	bio = bio_map_kern(q, data, data_len, GFP_KERNEL);
	if (IS_ERR(bio)) {
		pr_err("pblk: could not allocate GC bio (%lu)\n", PTR_ERR(bio));
		goto err_free_dma;
	}

	bio->bi_iter.bi_sector = 0; /* internal bio */
	bio_set_op_attrs(bio, REQ_OP_READ, 0);

	rqd.opcode = NVM_OP_PREAD;
	rqd.end_io = pblk_end_io_sync;
	rqd.private = &wait;
	rqd.nr_ppas = *secs_to_gc;
	rqd.bio = bio;

	ret = pblk_submit_read_io(pblk, &rqd);
	if (ret) {
		bio_endio(bio);
		pr_err("pblk: GC read request failed\n");
		goto err_free_dma;
	}

	if (!wait_for_completion_io_timeout(&wait,
				msecs_to_jiffies(PBLK_COMMAND_TIMEOUT_MS))) {
		pr_err("pblk: GC read I/O timed out\n");
	}

	if (rqd.error) {
		atomic_long_inc(&pblk->read_failed_gc);
#ifdef CONFIG_NVM_DEBUG
		pblk_print_failed_rqd(pblk, &rqd, rqd.error);
#endif
	}

#ifdef CONFIG_NVM_DEBUG
	atomic_long_add(*secs_to_gc, &pblk->sync_reads);
	atomic_long_add(*secs_to_gc, &pblk->recov_gc_reads);
	atomic_long_sub(*secs_to_gc, &pblk->inflight_reads);
#endif

out:
	if (rqd.nr_ppas > 1)
		nvm_dev_dma_free(dev->parent, rqd.ppa_list, rqd.dma_ppa_list);
	return NVM_IO_OK;

err_free_dma:
	if (rqd.nr_ppas > 1)
		nvm_dev_dma_free(dev->parent, rqd.ppa_list, rqd.dma_ppa_list);
	return NVM_IO_ERR;
}
Commit	Line	Data
a4bd217b JG	1	/*
	2	* Copyright (C) 2016 CNEX Labs
	3	* Initial release: Javier Gonzalez <javier@cnexlabs.com>
	4	* Matias Bjorling <matias@cnexlabs.com>
	5	*
	6	* This program is free software; you can redistribute it and/or
	7	* modify it under the terms of the GNU General Public License version
	8	* 2 as published by the Free Software Foundation.
	9	*
	10	* This program is distributed in the hope that it will be useful, but
	11	* WITHOUT ANY WARRANTY; without even the implied warranty of
	12	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
	13	* General Public License for more details.
	14	*
	15	* pblk-read.c - pblk's read path
	16	*/
	17
	18	#include "pblk.h"
	19
	20	/*
	21	* There is no guarantee that the value read from cache has not been updated and
	22	* resides at another location in the cache. We guarantee though that if the
	23	* value is read from the cache, it belongs to the mapped lba. In order to
	24	* guarantee and order between writes and reads are ordered, a flush must be
	25	* issued.
	26	*/
	27	static int pblk_read_from_cache(struct pblk pblk, struct bio bio,
	28	sector_t lba, struct ppa_addr ppa,
	29	int bio_iter)
	30	{
	31	#ifdef CONFIG_NVM_DEBUG
	32	/* Callers must ensure that the ppa points to a cache address */
	33	BUG_ON(pblk_ppa_empty(ppa));
	34	BUG_ON(!pblk_addr_in_cache(ppa));
	35	#endif
	36
	37	return pblk_rb_copy_to_bio(&pblk->rwb, bio, lba,
	38	pblk_addr_to_cacheline(ppa), bio_iter);
	39	}
	40
	41	static void pblk_read_ppalist_rq(struct pblk pblk, struct nvm_rq rqd,
	42	unsigned long *read_bitmap)
	43	{
	44	struct bio *bio = rqd->bio;
	45	struct ppa_addr ppas[PBLK_MAX_REQ_ADDRS];
	46	sector_t blba = pblk_get_lba(bio);
	47	int nr_secs = rqd->nr_ppas;
	48	int advanced_bio = 0;
	49	int i, j = 0;
	50
	51	/* logic error: lba out-of-bounds. Ignore read request */
2a79efd8 DC	52	if (blba + nr_secs >= pblk->rl.nr_secs) {
2a79efd8 DC	53	WARN(1, "pblk: read lbas out of bounds\n");
a4bd217b JG	54	return;
	55	}
	56
	57	pblk_lookup_l2p_seq(pblk, ppas, blba, nr_secs);
	58
	59	for (i = 0; i < nr_secs; i++) {
	60	struct ppa_addr p = ppas[i];
	61	sector_t lba = blba + i;
	62
	63	retry:
	64	if (pblk_ppa_empty(p)) {
	65	WARN_ON(test_and_set_bit(i, read_bitmap));
	66	continue;
	67	}
	68
	69	/* Try to read from write buffer. The address is later checked
	70	* on the write buffer to prevent retrieving overwritten data.
	71	*/
	72	if (pblk_addr_in_cache(p)) {
	73	if (!pblk_read_from_cache(pblk, bio, lba, p, i)) {
	74	pblk_lookup_l2p_seq(pblk, &p, lba, 1);
	75	goto retry;
	76	}
	77	WARN_ON(test_and_set_bit(i, read_bitmap));
	78	advanced_bio = 1;
	79	} else {
	80	/* Read from media non-cached sectors */
	81	rqd->ppa_list[j++] = p;
	82	}
	83
	84	if (advanced_bio)
	85	bio_advance(bio, PBLK_EXPOSED_PAGE_SIZE);
	86	}
	87
	88	#ifdef CONFIG_NVM_DEBUG
	89	atomic_long_add(nr_secs, &pblk->inflight_reads);
	90	#endif
	91	}
	92
	93	static int pblk_submit_read_io(struct pblk pblk, struct nvm_rq rqd)
	94	{
	95	int err;
	96
	97	rqd->flags = pblk_set_read_mode(pblk);
	98
	99	err = pblk_submit_io(pblk, rqd);
	100	if (err)
	101	return NVM_IO_ERR;
	102
	103	return NVM_IO_OK;
	104	}
	105
	106	static void pblk_end_io_read(struct nvm_rq *rqd)
	107	{
	108	struct pblk *pblk = rqd->private;
	109	struct nvm_tgt_dev *dev = pblk->dev;
	110	struct pblk_r_ctx *r_ctx = nvm_rq_to_pdu(rqd);
	111	struct bio *bio = rqd->bio;
	112
	113	if (rqd->error)
	114	pblk_log_read_err(pblk, rqd);
	115	#ifdef CONFIG_NVM_DEBUG
	116	else
	117	WARN_ONCE(bio->bi_error, "pblk: corrupted read error\n");
118	#endif
119
120	if (rqd->nr_ppas > 1)
121	nvm_dev_dma_free(dev->parent, rqd->ppa_list, rqd->dma_ppa_list);
122
123	bio_put(bio);
124	if (r_ctx->orig_bio) {
125	#ifdef CONFIG_NVM_DEBUG
126	WARN_ONCE(r_ctx->orig_bio->bi_error,
127	"pblk: corrupted read bio\n");
128	#endif
129	bio_endio(r_ctx->orig_bio);
130	bio_put(r_ctx->orig_bio);
131	}
132
133	#ifdef CONFIG_NVM_DEBUG
134	atomic_long_add(rqd->nr_ppas, &pblk->sync_reads);
135	atomic_long_sub(rqd->nr_ppas, &pblk->inflight_reads);
136	#endif
137
138	pblk_free_rqd(pblk, rqd, READ);
139	}
140
141	static int pblk_fill_partial_read_bio(struct pblk pblk, struct nvm_rq rqd,
142	unsigned int bio_init_idx,
143	unsigned long *read_bitmap)
144	{
145	struct bio new_bio, bio = rqd->bio;
146	struct bio_vec src_bv, dst_bv;
147	void *ppa_ptr = NULL;
148	void src_p, dst_p;
149	dma_addr_t dma_ppa_list = 0;
150	int nr_secs = rqd->nr_ppas;
151	int nr_holes = nr_secs - bitmap_weight(read_bitmap, nr_secs);
152	int i, ret, hole;
153	DECLARE_COMPLETION_ONSTACK(wait);
154
155	new_bio = bio_alloc(GFP_KERNEL, nr_holes);
156	if (!new_bio) {
157	pr_err("pblk: could not alloc read bio\n");
158	return NVM_IO_ERR;
159	}
160
161	if (pblk_bio_add_pages(pblk, new_bio, GFP_KERNEL, nr_holes))
162	goto err;
163
164	if (nr_holes != new_bio->bi_vcnt) {
165	pr_err("pblk: malformed bio\n");
166	goto err;
167	}
168
169	new_bio->bi_iter.bi_sector = 0; /* internal bio */
170	bio_set_op_attrs(new_bio, REQ_OP_READ, 0);
171	new_bio->bi_private = &wait;
172	new_bio->bi_end_io = pblk_end_bio_sync;
173
174	rqd->bio = new_bio;
175	rqd->nr_ppas = nr_holes;
176	rqd->end_io = NULL;
177
178	if (unlikely(nr_secs > 1 && nr_holes == 1)) {
179	ppa_ptr = rqd->ppa_list;
180	dma_ppa_list = rqd->dma_ppa_list;
181	rqd->ppa_addr = rqd->ppa_list[0];
182	}
183
184	ret = pblk_submit_read_io(pblk, rqd);
185	if (ret) {
186	bio_put(rqd->bio);
187	pr_err("pblk: read IO submission failed\n");
188	goto err;
189	}
190
191	if (!wait_for_completion_io_timeout(&wait,
192	msecs_to_jiffies(PBLK_COMMAND_TIMEOUT_MS))) {
193	pr_err("pblk: partial read I/O timed out\n");
194	}
195
196	if (rqd->error) {
197	atomic_long_inc(&pblk->read_failed);
198	#ifdef CONFIG_NVM_DEBUG
199	pblk_print_failed_rqd(pblk, rqd, rqd->error);
200	#endif
201	}
202
203	if (unlikely(nr_secs > 1 && nr_holes == 1)) {
204	rqd->ppa_list = ppa_ptr;
205	rqd->dma_ppa_list = dma_ppa_list;
206	}
207
208	/* Fill the holes in the original bio */
209	i = 0;
210	hole = find_first_zero_bit(read_bitmap, nr_secs);
211	do {
212	src_bv = new_bio->bi_io_vec[i++];
213	dst_bv = bio->bi_io_vec[bio_init_idx + hole];
214
215	src_p = kmap_atomic(src_bv.bv_page);
216	dst_p = kmap_atomic(dst_bv.bv_page);
217
218	memcpy(dst_p + dst_bv.bv_offset,
219	src_p + src_bv.bv_offset,
220	PBLK_EXPOSED_PAGE_SIZE);
221
222	kunmap_atomic(src_p);
223	kunmap_atomic(dst_p);
224
225	mempool_free(src_bv.bv_page, pblk->page_pool);
226
227	hole = find_next_zero_bit(read_bitmap, nr_secs, hole + 1);
228	} while (hole < nr_secs);
229
230	bio_put(new_bio);
231
232	/* Complete the original bio and associated request */
233	rqd->bio = bio;
234	rqd->nr_ppas = nr_secs;
235	rqd->private = pblk;
236
237	bio_endio(bio);
238	pblk_end_io_read(rqd);
239	return NVM_IO_OK;
240
241	err:
242	/* Free allocated pages in new bio */
243	pblk_bio_free_pages(pblk, bio, 0, new_bio->bi_vcnt);
244	rqd->private = pblk;
245	pblk_end_io_read(rqd);
246	return NVM_IO_ERR;
247	}
248
249	static void pblk_read_rq(struct pblk pblk, struct nvm_rq rqd,
250	unsigned long *read_bitmap)
251	{
252	struct bio *bio = rqd->bio;
253	struct ppa_addr ppa;
254	sector_t lba = pblk_get_lba(bio);
255
256	/* logic error: lba out-of-bounds. Ignore read request */
2a79efd8 DC	257	if (lba >= pblk->rl.nr_secs) {
2a79efd8 DC	258	WARN(1, "pblk: read lba out of bounds\n");
a4bd217b JG	259	return;
	260	}
	261
	262	pblk_lookup_l2p_seq(pblk, &ppa, lba, 1);
	263
	264	#ifdef CONFIG_NVM_DEBUG
	265	atomic_long_inc(&pblk->inflight_reads);
	266	#endif
	267
	268	retry:
	269	if (pblk_ppa_empty(ppa)) {
	270	WARN_ON(test_and_set_bit(0, read_bitmap));
	271	return;
	272	}
	273
	274	/* Try to read from write buffer. The address is later checked on the
	275	* write buffer to prevent retrieving overwritten data.
	276	*/
	277	if (pblk_addr_in_cache(ppa)) {
	278	if (!pblk_read_from_cache(pblk, bio, lba, ppa, 0)) {
	279	pblk_lookup_l2p_seq(pblk, &ppa, lba, 1);
	280	goto retry;
	281	}
	282	WARN_ON(test_and_set_bit(0, read_bitmap));
	283	} else {
	284	rqd->ppa_addr = ppa;
	285	}
	286	}
	287
	288	int pblk_submit_read(struct pblk pblk, struct bio bio)
	289	{
	290	struct nvm_tgt_dev *dev = pblk->dev;
5bf1e1ee	291	unsigned int nr_secs = pblk_get_secs(bio);
a4bd217b JG	292	struct nvm_rq *rqd;
	293	unsigned long read_bitmap; /* Max 64 ppas per request */
	294	unsigned int bio_init_idx;
	295	int ret = NVM_IO_ERR;
	296
	297	if (nr_secs > PBLK_MAX_REQ_ADDRS)
	298	return NVM_IO_ERR;
	299
	300	bitmap_zero(&read_bitmap, nr_secs);
	301
	302	rqd = pblk_alloc_rqd(pblk, READ);
	303	if (IS_ERR(rqd)) {
	304	pr_err_ratelimited("pblk: not able to alloc rqd");
	305	return NVM_IO_ERR;
	306	}
	307
	308	rqd->opcode = NVM_OP_PREAD;
	309	rqd->bio = bio;
	310	rqd->nr_ppas = nr_secs;
	311	rqd->private = pblk;
	312	rqd->end_io = pblk_end_io_read;
	313
	314	/* Save the index for this bio's start. This is needed in case
	315	* we need to fill a partial read.
	316	*/
	317	bio_init_idx = pblk_get_bi_idx(bio);
	318
	319	if (nr_secs > 1) {
	320	rqd->ppa_list = nvm_dev_dma_alloc(dev->parent, GFP_KERNEL,
	321	&rqd->dma_ppa_list);
	322	if (!rqd->ppa_list) {
	323	pr_err("pblk: not able to allocate ppa list\n");
	324	goto fail_rqd_free;
	325	}
	326
	327	pblk_read_ppalist_rq(pblk, rqd, &read_bitmap);
	328	} else {
	329	pblk_read_rq(pblk, rqd, &read_bitmap);
	330	}
	331
	332	bio_get(bio);
	333	if (bitmap_full(&read_bitmap, nr_secs)) {
	334	bio_endio(bio);
	335	pblk_end_io_read(rqd);
	336	return NVM_IO_OK;
	337	}
	338
	339	/* All sectors are to be read from the device */
	340	if (bitmap_empty(&read_bitmap, rqd->nr_ppas)) {
	341	struct bio *int_bio = NULL;
	342	struct pblk_r_ctx *r_ctx = nvm_rq_to_pdu(rqd);
	343
	344	/* Clone read bio to deal with read errors internally */
	345	int_bio = bio_clone_bioset(bio, GFP_KERNEL, fs_bio_set);
	346	if (!int_bio) {
	347	pr_err("pblk: could not clone read bio\n");
	348	return NVM_IO_ERR;
	349	}
	350
	351	rqd->bio = int_bio;
	352	r_ctx->orig_bio = bio;
	353
	354	ret = pblk_submit_read_io(pblk, rqd);
	355	if (ret) {
356	pr_err("pblk: read IO submission failed\n");
357	if (int_bio)
358	bio_put(int_bio);
359	return ret;
360	}
361
362	return NVM_IO_OK;
363	}
364
365	/* The read bio request could be partially filled by the write buffer,
366	* but there are some holes that need to be read from the drive.
367	*/
368	ret = pblk_fill_partial_read_bio(pblk, rqd, bio_init_idx, &read_bitmap);
369	if (ret) {
370	pr_err("pblk: failed to perform partial read\n");
371	return ret;
372	}
373
374	return NVM_IO_OK;
375
376	fail_rqd_free:
377	pblk_free_rqd(pblk, rqd, READ);
378	return ret;
379	}
380
381	static int read_ppalist_rq_gc(struct pblk pblk, struct nvm_rq rqd,
382	struct pblk_line line, u64 lba_list,
383	unsigned int nr_secs)
384	{
385	struct ppa_addr ppas[PBLK_MAX_REQ_ADDRS];
386	int valid_secs = 0;
387	int i;
388
389	pblk_lookup_l2p_rand(pblk, ppas, lba_list, nr_secs);
390
391	for (i = 0; i < nr_secs; i++) {
392	if (pblk_addr_in_cache(ppas[i]) \|\| ppas[i].g.blk != line->id \|\|
393	pblk_ppa_empty(ppas[i])) {
394	lba_list[i] = ADDR_EMPTY;
395	continue;
396	}
397
398	rqd->ppa_list[valid_secs++] = ppas[i];
399	}
400
401	#ifdef CONFIG_NVM_DEBUG
402	atomic_long_add(valid_secs, &pblk->inflight_reads);
403	#endif
404	return valid_secs;
405	}
406
407	static int read_rq_gc(struct pblk pblk, struct nvm_rq rqd,
408	struct pblk_line *line, sector_t lba)
409	{
410	struct ppa_addr ppa;
411	int valid_secs = 0;
412
413	/* logic error: lba out-of-bounds */
2a79efd8 DC	414	if (lba >= pblk->rl.nr_secs) {
2a79efd8 DC	415	WARN(1, "pblk: read lba out of bounds\n");
a4bd217b JG	416	goto out;
	417	}
	418
	419	if (lba == ADDR_EMPTY)
	420	goto out;
	421
	422	spin_lock(&pblk->trans_lock);
	423	ppa = pblk_trans_map_get(pblk, lba);
	424	spin_unlock(&pblk->trans_lock);
	425
	426	/* Ignore updated values until the moment */
	427	if (pblk_addr_in_cache(ppa) \|\| ppa.g.blk != line->id \|\|
	428	pblk_ppa_empty(ppa))
	429	goto out;
	430
	431	rqd->ppa_addr = ppa;
	432	valid_secs = 1;
	433
	434	#ifdef CONFIG_NVM_DEBUG
	435	atomic_long_inc(&pblk->inflight_reads);
	436	#endif
	437
	438	out:
	439	return valid_secs;
	440	}
	441
	442	int pblk_submit_read_gc(struct pblk pblk, u64 lba_list, void *data,
	443	unsigned int nr_secs, unsigned int *secs_to_gc,
	444	struct pblk_line *line)
	445	{
	446	struct nvm_tgt_dev *dev = pblk->dev;
	447	struct nvm_geo *geo = &dev->geo;
	448	struct request_queue *q = dev->q;
	449	struct bio *bio;
	450	struct nvm_rq rqd;
	451	int ret, data_len;
	452	DECLARE_COMPLETION_ONSTACK(wait);
	453
	454	memset(&rqd, 0, sizeof(struct nvm_rq));
	455
	456	if (nr_secs > 1) {
	457	rqd.ppa_list = nvm_dev_dma_alloc(dev->parent, GFP_KERNEL,
	458	&rqd.dma_ppa_list);
	459	if (!rqd.ppa_list)
	460	return NVM_IO_ERR;
	461
	462	*secs_to_gc = read_ppalist_rq_gc(pblk, &rqd, line, lba_list,
	463	nr_secs);
	464	if (*secs_to_gc == 1) {
	465	struct ppa_addr ppa;
	466
	467	ppa = rqd.ppa_list[0];
	468	nvm_dev_dma_free(dev->parent, rqd.ppa_list,
	469	rqd.dma_ppa_list);
	470	rqd.ppa_addr = ppa;
	471	}
	472	} else {
	473	*secs_to_gc = read_rq_gc(pblk, &rqd, line, lba_list[0]);
	474	}
	475
	476	if (!(*secs_to_gc))
	477	goto out;
	478
	479	data_len = (secs_to_gc) geo->sec_size;
480	bio = bio_map_kern(q, data, data_len, GFP_KERNEL);
481	if (IS_ERR(bio)) {
482	pr_err("pblk: could not allocate GC bio (%lu)\n", PTR_ERR(bio));
483	goto err_free_dma;
484	}
485
486	bio->bi_iter.bi_sector = 0; /* internal bio */
487	bio_set_op_attrs(bio, REQ_OP_READ, 0);
488
489	rqd.opcode = NVM_OP_PREAD;
490	rqd.end_io = pblk_end_io_sync;
491	rqd.private = &wait;
492	rqd.nr_ppas = *secs_to_gc;
493	rqd.bio = bio;
494
495	ret = pblk_submit_read_io(pblk, &rqd);
496	if (ret) {
497	bio_endio(bio);
498	pr_err("pblk: GC read request failed\n");
499	goto err_free_dma;
500	}
501
502	if (!wait_for_completion_io_timeout(&wait,
503	msecs_to_jiffies(PBLK_COMMAND_TIMEOUT_MS))) {
504	pr_err("pblk: GC read I/O timed out\n");
505	}
506
507	if (rqd.error) {
508	atomic_long_inc(&pblk->read_failed_gc);
509	#ifdef CONFIG_NVM_DEBUG
510	pblk_print_failed_rqd(pblk, &rqd, rqd.error);
511	#endif
512	}
513
514	#ifdef CONFIG_NVM_DEBUG
515	atomic_long_add(*secs_to_gc, &pblk->sync_reads);
516	atomic_long_add(*secs_to_gc, &pblk->recov_gc_reads);
517	atomic_long_sub(*secs_to_gc, &pblk->inflight_reads);
518	#endif
519
520	out:
521	if (rqd.nr_ppas > 1)
522	nvm_dev_dma_free(dev->parent, rqd.ppa_list, rqd.dma_ppa_list);
523	return NVM_IO_OK;
524
525	err_free_dma:
526	if (rqd.nr_ppas > 1)
527	nvm_dev_dma_free(dev->parent, rqd.ppa_list, rqd.dma_ppa_list);
528	return NVM_IO_ERR;
529	}