lightnvm: pblk: fail gracefully on irrec. error

Due to user writes being decoupled from media writes because of the need
of an intermediate write buffer, irrecoverable media write errors lead
to pblk stalling; user writes fill up the buffer and end up in an
infinite retry loop.

In order to let user writes fail gracefully, it is necessary for pblk to
keep track of its own internal state and prevent further writes from
being placed into the write buffer.

This patch implements a state machine to keep track of internal errors
and, in case of failure, fail further user writes in an standard way.
Depending on the type of error, pblk will do its best to persist
buffered writes (which are already acknowledged) and close down on a
graceful manner. This way, data might be recovered by re-instantiating
pblk. Such state machine paves out the way for a state-based FTL log.

Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

1 files changed, 199 insertions, 81 deletions

diff --git a/drivers/lightnvm/pblk-core.c b/drivers/lightnvm/pblk-core.c
index 823e53f95a80..7648186bd1b1 100644
--- a/drivers/lightnvm/pblk-core.c
+++ b/drivers/lightnvm/pblk-core.c
@@ -53,6 +53,8 @@ static void __pblk_end_io_erase(struct pblk *pblk, struct nvm_rq *rqd)
 		*ppa = rqd->ppa_addr;
 		pblk_mark_bb(pblk, line, ppa);
 	}
+
+	atomic_dec(&pblk->inflight_io);
 }
 
 /* Erase completion assumes that only one block is erased at the time */
@@ -257,35 +259,25 @@ void pblk_end_io_sync(struct nvm_rq *rqd)
 	complete(waiting);
 }
 
-void pblk_flush_writer(struct pblk *pblk)
+void pblk_wait_for_meta(struct pblk *pblk)
 {
-	struct bio *bio;
-	int ret;
-	DECLARE_COMPLETION_ONSTACK(wait);
-
-	bio = bio_alloc(GFP_KERNEL, 1);
-	if (!bio)
-		return;
-
-	bio->bi_iter.bi_sector = 0; /* internal bio */
-	bio_set_op_attrs(bio, REQ_OP_WRITE, REQ_OP_FLUSH);
-	bio->bi_private = &wait;
-	bio->bi_end_io = pblk_end_bio_sync;
+	do {
+		if (!atomic_read(&pblk->inflight_io))
+			break;
 
-	ret = pblk_write_to_cache(pblk, bio, 0);
-	if (ret == NVM_IO_OK) {
-		if (!wait_for_completion_io_timeout(&wait,
-				msecs_to_jiffies(PBLK_COMMAND_TIMEOUT_MS))) {
-			pr_err("pblk: flush cache timed out\n");
-		}
-	} else if (ret != NVM_IO_DONE) {
-		pr_err("pblk: tear down bio failed\n");
-	}
+		schedule();
+	} while (1);
+}
 
-	if (bio->bi_status)
-		pr_err("pblk: flush sync write failed (%u)\n", bio->bi_status);
+static void pblk_flush_writer(struct pblk *pblk)
+{
+	pblk_rb_flush(&pblk->rwb);
+	do {
+		if (!pblk_rb_read_count(&pblk->rwb))
+			break;
 
-	bio_put(bio);
+		schedule();
+	} while (1);
 }
 
 struct list_head *pblk_line_gc_list(struct pblk *pblk, struct pblk_line *line)
@@ -425,6 +417,9 @@ int pblk_submit_io(struct pblk *pblk, struct nvm_rq *rqd)
 		}
 	}
 #endif
+
+	atomic_inc(&pblk->inflight_io);
+
 	return nvm_submit_io(dev, rqd);
 }
 
@@ -676,6 +671,7 @@ next_rq:
 				msecs_to_jiffies(PBLK_COMMAND_TIMEOUT_MS))) {
 		pr_err("pblk: emeta I/O timed out\n");
 	}
+	atomic_dec(&pblk->inflight_io);
 	reinit_completion(&wait);
 
 	if (likely(pblk->l_mg.emeta_alloc_type == PBLK_VMALLOC_META))
@@ -791,6 +787,7 @@ static int pblk_line_submit_smeta_io(struct pblk *pblk, struct pblk_line *line,
 				msecs_to_jiffies(PBLK_COMMAND_TIMEOUT_MS))) {
 		pr_err("pblk: smeta I/O timed out\n");
 	}
+	atomic_dec(&pblk->inflight_io);
 
 	if (rqd.error) {
 		if (dir == WRITE)
@@ -832,7 +829,7 @@ static void pblk_setup_e_rq(struct pblk *pblk, struct nvm_rq *rqd,
 static int pblk_blk_erase_sync(struct pblk *pblk, struct ppa_addr ppa)
 {
 	struct nvm_rq rqd;
-	int ret;
+	int ret = 0;
 	DECLARE_COMPLETION_ONSTACK(wait);
 
 	memset(&rqd, 0, sizeof(struct nvm_rq));
@@ -867,14 +864,14 @@ out:
 	rqd.private = pblk;
 	__pblk_end_io_erase(pblk, &rqd);
 
-	return 0;
+	return ret;
 }
 
 int pblk_line_erase(struct pblk *pblk, struct pblk_line *line)
 {
 	struct pblk_line_meta *lm = &pblk->lm;
 	struct ppa_addr ppa;
-	int bit = -1;
+	int ret, bit = -1;
 
 	/* Erase only good blocks, one at a time */
 	do {
@@ -893,9 +890,10 @@ int pblk_line_erase(struct pblk *pblk, struct pblk_line *line)
 		WARN_ON(test_and_set_bit(bit, line->erase_bitmap));
 		spin_unlock(&line->lock);
 
-		if (pblk_blk_erase_sync(pblk, ppa)) {
+		ret = pblk_blk_erase_sync(pblk, ppa);
+		if (ret) {
 			pr_err("pblk: failed to erase line %d\n", line->id);
-			return -ENOMEM;
+			return ret;
 		}
 	} while (1);
 
@@ -908,6 +906,8 @@ static void pblk_line_setup_metadata(struct pblk_line *line,
 {
 	int meta_line;
 
+	lockdep_assert_held(&l_mg->free_lock);
+
 retry_meta:
 	meta_line = find_first_zero_bit(&l_mg->meta_bitmap, PBLK_DATA_LINES);
 	if (meta_line == PBLK_DATA_LINES) {
@@ -1039,7 +1039,6 @@ static int pblk_line_init_bb(struct pblk *pblk, struct pblk_line *line,
 	/* Mark smeta metadata sectors as bad sectors */
 	bit = find_first_zero_bit(line->blk_bitmap, lm->blk_per_line);
 	off = bit * geo->sec_per_pl;
-retry_smeta:
 	bitmap_set(line->map_bitmap, off, lm->smeta_sec);
 	line->sec_in_line -= lm->smeta_sec;
 	line->smeta_ssec = off;
@@ -1047,8 +1046,7 @@ retry_smeta:
 
 	if (init && pblk_line_submit_smeta_io(pblk, line, off, WRITE)) {
 		pr_debug("pblk: line smeta I/O failed. Retry\n");
-		off += geo->sec_per_pl;
-		goto retry_smeta;
+		return 1;
 	}
 
 	bitmap_copy(line->invalid_bitmap, line->map_bitmap, lm->sec_per_line);
@@ -1110,10 +1108,14 @@ static int pblk_line_prepare(struct pblk *pblk, struct pblk_line *line)
 
 	spin_lock(&line->lock);
 	if (line->state != PBLK_LINESTATE_FREE) {
+		mempool_free(line->invalid_bitmap, pblk->line_meta_pool);
+		mempool_free(line->map_bitmap, pblk->line_meta_pool);
 		spin_unlock(&line->lock);
-		WARN(1, "pblk: corrupted line state\n");
-		return -EINTR;
+		WARN(1, "pblk: corrupted line %d, state %d\n",
+							line->id, line->state);
+		return -EAGAIN;
 	}
+
 	line->state = PBLK_LINESTATE_OPEN;
 
 	atomic_set(&line->left_eblks, blk_in_line);
@@ -1169,15 +1171,15 @@ struct pblk_line *pblk_line_get(struct pblk *pblk)
 {
 	struct pblk_line_mgmt *l_mg = &pblk->l_mg;
 	struct pblk_line_meta *lm = &pblk->lm;
-	struct pblk_line *line = NULL;
-	int bit;
+	struct pblk_line *line;
+	int ret, bit;
 
 	lockdep_assert_held(&l_mg->free_lock);
 
-retry_get:
+retry:
 	if (list_empty(&l_mg->free_list)) {
 		pr_err("pblk: no free lines\n");
-		goto out;
+		return NULL;
 	}
 
 	line = list_first_entry(&l_mg->free_list, struct pblk_line, list);
@@ -1193,17 +1195,22 @@ retry_get:
 		list_add_tail(&line->list, &l_mg->bad_list);
 
 		pr_debug("pblk: line %d is bad\n", line->id);
-		goto retry_get;
+		goto retry;
 	}
 
-	if (pblk_line_prepare(pblk, line)) {
-		pr_err("pblk: failed to prepare line %d\n", line->id);
-		list_add(&line->list, &l_mg->free_list);
-		l_mg->nr_free_lines++;
-		return NULL;
+	ret = pblk_line_prepare(pblk, line);
+	if (ret) {
+		if (ret == -EAGAIN) {
+			list_add(&line->list, &l_mg->corrupt_list);
+			goto retry;
+		} else {
+			pr_err("pblk: failed to prepare line %d\n", line->id);
+			list_add(&line->list, &l_mg->free_list);
+			l_mg->nr_free_lines++;
+			return NULL;
+		}
 	}
 
-out:
 	return line;
 }
 
@@ -1213,6 +1220,7 @@ static struct pblk_line *pblk_line_retry(struct pblk *pblk,
 	struct pblk_line_mgmt *l_mg = &pblk->l_mg;
 	struct pblk_line *retry_line;
 
+retry:
 	spin_lock(&l_mg->free_lock);
 	retry_line = pblk_line_get(pblk);
 	if (!retry_line) {
@@ -1229,18 +1237,21 @@ static struct pblk_line *pblk_line_retry(struct pblk *pblk,
 	l_mg->data_line = retry_line;
 	spin_unlock(&l_mg->free_lock);
 
-	if (pblk_line_erase(pblk, retry_line)) {
-		spin_lock(&l_mg->free_lock);
-		l_mg->data_line = NULL;
-		spin_unlock(&l_mg->free_lock);
-		return NULL;
-	}
-
 	pblk_rl_free_lines_dec(&pblk->rl, retry_line);
 
+	if (pblk_line_erase(pblk, retry_line))
+		goto retry;
+
 	return retry_line;
 }
 
+static void pblk_set_space_limit(struct pblk *pblk)
+{
+	struct pblk_rl *rl = &pblk->rl;
+
+	atomic_set(&rl->rb_space, 0);
+}
+
 struct pblk_line *pblk_line_get_first_data(struct pblk *pblk)
 {
 	struct pblk_line_mgmt *l_mg = &pblk->l_mg;
@@ -1262,20 +1273,31 @@ struct pblk_line *pblk_line_get_first_data(struct pblk *pblk)
 
 	/* Allocate next line for preparation */
 	l_mg->data_next = pblk_line_get(pblk);
-	if (l_mg->data_next) {
+	if (!l_mg->data_next) {
+		/* If we cannot get a new line, we need to stop the pipeline.
+		 * Only allow as many writes in as we can store safely and then
+		 * fail gracefully
+		 */
+		pblk_set_space_limit(pblk);
+
+		l_mg->data_next = NULL;
+	} else {
 		l_mg->data_next->seq_nr = l_mg->d_seq_nr++;
 		l_mg->data_next->type = PBLK_LINETYPE_DATA;
 		is_next = 1;
 	}
 	spin_unlock(&l_mg->free_lock);
 
+	if (pblk_line_erase(pblk, line)) {
+		line = pblk_line_retry(pblk, line);
+		if (!line)
+			return NULL;
+	}
+
 	pblk_rl_free_lines_dec(&pblk->rl, line);
 	if (is_next)
 		pblk_rl_free_lines_dec(&pblk->rl, l_mg->data_next);
 
-	if (pblk_line_erase(pblk, line))
-		return NULL;
-
 retry_setup:
 	if (!pblk_line_init_metadata(pblk, line, NULL)) {
 		line = pblk_line_retry(pblk, line);
@@ -1296,7 +1318,47 @@ retry_setup:
 	return line;
 }
 
-struct pblk_line *pblk_line_replace_data(struct pblk *pblk)
+static void pblk_stop_writes(struct pblk *pblk, struct pblk_line *line)
+{
+	lockdep_assert_held(&pblk->l_mg.free_lock);
+
+	pblk_set_space_limit(pblk);
+	pblk->state = PBLK_STATE_STOPPING;
+}
+
+void pblk_pipeline_stop(struct pblk *pblk)
+{
+	struct pblk_line_mgmt *l_mg = &pblk->l_mg;
+	int ret;
+
+	spin_lock(&l_mg->free_lock);
+	if (pblk->state == PBLK_STATE_RECOVERING ||
+					pblk->state == PBLK_STATE_STOPPED) {
+		spin_unlock(&l_mg->free_lock);
+		return;
+	}
+	pblk->state = PBLK_STATE_RECOVERING;
+	spin_unlock(&l_mg->free_lock);
+
+	pblk_flush_writer(pblk);
+	pblk_wait_for_meta(pblk);
+
+	ret = pblk_recov_pad(pblk);
+	if (ret) {
+		pr_err("pblk: could not close data on teardown(%d)\n", ret);
+		return;
+	}
+
+	pblk_line_close_meta_sync(pblk);
+
+	spin_lock(&l_mg->free_lock);
+	pblk->state = PBLK_STATE_STOPPED;
+	l_mg->data_line = NULL;
+	l_mg->data_next = NULL;
+	spin_unlock(&l_mg->free_lock);
+}
+
+void pblk_line_replace_data(struct pblk *pblk)
 {
 	struct pblk_line_mgmt *l_mg = &pblk->l_mg;
 	struct pblk_line *cur, *new;
@@ -1306,42 +1368,38 @@ struct pblk_line *pblk_line_replace_data(struct pblk *pblk)
 	cur = l_mg->data_line;
 	new = l_mg->data_next;
 	if (!new)
-		return NULL;
+		return;
 	l_mg->data_line = new;
 
-retry_line:
+	spin_lock(&l_mg->free_lock);
+	if (pblk->state != PBLK_STATE_RUNNING) {
+		l_mg->data_line = NULL;
+		l_mg->data_next = NULL;
+		spin_unlock(&l_mg->free_lock);
+		return;
+	}
+
+	pblk_line_setup_metadata(new, l_mg, &pblk->lm);
+	spin_unlock(&l_mg->free_lock);
+
+retry_erase:
 	left_seblks = atomic_read(&new->left_seblks);
 	if (left_seblks) {
 		/* If line is not fully erased, erase it */
 		if (atomic_read(&new->left_eblks)) {
 			if (pblk_line_erase(pblk, new))
-				return NULL;
+				return;
 		} else {
 			io_schedule();
 		}
-		goto retry_line;
+		goto retry_erase;
 	}
 
-	spin_lock(&l_mg->free_lock);
-	/* Allocate next line for preparation */
-	l_mg->data_next = pblk_line_get(pblk);
-	if (l_mg->data_next) {
-		l_mg->data_next->seq_nr = l_mg->d_seq_nr++;
-		l_mg->data_next->type = PBLK_LINETYPE_DATA;
-		is_next = 1;
-	}
-
-	pblk_line_setup_metadata(new, l_mg, &pblk->lm);
-	spin_unlock(&l_mg->free_lock);
-
-	if (is_next)
-		pblk_rl_free_lines_dec(&pblk->rl, l_mg->data_next);
-
 retry_setup:
 	if (!pblk_line_init_metadata(pblk, new, cur)) {
 		new = pblk_line_retry(pblk, new);
 		if (!new)
-			return NULL;
+			return;
 
 		goto retry_setup;
 	}
@@ -1349,12 +1407,30 @@ retry_setup:
 	if (!pblk_line_init_bb(pblk, new, 1)) {
 		new = pblk_line_retry(pblk, new);
 		if (!new)
-			return NULL;
+			return;
 
 		goto retry_setup;
 	}
 
-	return new;
+	/* Allocate next line for preparation */
+	spin_lock(&l_mg->free_lock);
+	l_mg->data_next = pblk_line_get(pblk);
+	if (!l_mg->data_next) {
+		/* If we cannot get a new line, we need to stop the pipeline.
+		 * Only allow as many writes in as we can store safely and then
+		 * fail gracefully
+		 */
+		pblk_stop_writes(pblk, new);
+		l_mg->data_next = NULL;
+	} else {
+		l_mg->data_next->seq_nr = l_mg->d_seq_nr++;
+		l_mg->data_next->type = PBLK_LINETYPE_DATA;
+		is_next = 1;
+	}
+	spin_unlock(&l_mg->free_lock);
+
+	if (is_next)
+		pblk_rl_free_lines_dec(&pblk->rl, l_mg->data_next);
 }
 
 void pblk_line_free(struct pblk *pblk, struct pblk_line *line)
@@ -1438,6 +1514,46 @@ int pblk_line_is_full(struct pblk_line *line)
 	return (line->left_msecs == 0);
 }
 
+void pblk_line_close_meta_sync(struct pblk *pblk)
+{
+	struct pblk_line_mgmt *l_mg = &pblk->l_mg;
+	struct pblk_line_meta *lm = &pblk->lm;
+	struct pblk_line *line, *tline;
+	LIST_HEAD(list);
+
+	spin_lock(&l_mg->close_lock);
+	if (list_empty(&l_mg->emeta_list)) {
+		spin_unlock(&l_mg->close_lock);
+		return;
+	}
+
+	list_cut_position(&list, &l_mg->emeta_list, l_mg->emeta_list.prev);
+	spin_unlock(&l_mg->close_lock);
+
+	list_for_each_entry_safe(line, tline, &list, list) {
+		struct pblk_emeta *emeta = line->emeta;
+
+		while (emeta->mem < lm->emeta_len[0]) {
+			int ret;
+
+			ret = pblk_submit_meta_io(pblk, line);
+			if (ret) {
+				pr_err("pblk: sync meta line %d failed (%d)\n",
+							line->id, ret);
+				return;
+			}
+		}
+	}
+
+	pblk_wait_for_meta(pblk);
+}
+
+static void pblk_line_should_sync_meta(struct pblk *pblk)
+{
+	if (pblk_rl_is_limit(&pblk->rl))
+		pblk_line_close_meta_sync(pblk);
+}
+
 void pblk_line_close(struct pblk *pblk, struct pblk_line *line)
 {
 	struct pblk_line_mgmt *l_mg = &pblk->l_mg;
@@ -1477,7 +1593,7 @@ void pblk_line_close_meta(struct pblk *pblk, struct pblk_line *line)
 	struct pblk_emeta *emeta = line->emeta;
 	struct line_emeta *emeta_buf = emeta->buf;
 
-	/* No need for exact vsc value; avoid a big line lock and tak aprox. */
+	/* No need for exact vsc value; avoid a big line lock and take aprox. */
 	memcpy(emeta_to_vsc(pblk, emeta_buf), l_mg->vsc_list, lm->vsc_list_len);
 	memcpy(emeta_to_bb(emeta_buf), line->blk_bitmap, lm->blk_bitmap_len);
 
@@ -1489,6 +1605,8 @@ void pblk_line_close_meta(struct pblk *pblk, struct pblk_line *line)
 	list_add_tail(&line->list, &l_mg->emeta_list);
 	spin_unlock(&line->lock);
 	spin_unlock(&l_mg->close_lock);
+
+	pblk_line_should_sync_meta(pblk);
 }
 
 void pblk_line_close_ws(struct work_struct *work)