All of lore.kernel.org
 help / color / mirror / Atom feed
* [GIT PULL v3 00/10] lightnvm updates for 4.18
@ 2018-06-01 14:41 Matias Bjørling
  2018-06-01 14:41 ` [GIT PULL 01/10] lightnvm: pblk: rework write error recovery path Matias Bjørling
                   ` (10 more replies)
  0 siblings, 11 replies; 12+ messages in thread
From: Matias Bjørling @ 2018-06-01 14:41 UTC (permalink / raw)
  To: axboe
  Cc: linux-block, linux-kernel, javier, igor.j.konopko,
	marcin.dziegielewski, Matias Bjørling

Hi Jens,

Please pick up the following patches.

Changes since v2:

 Rebased write error recovery patch on top of latest
 for-4.18/block

 Added (missed the first round)
  lightnvm: pblk: only try to recover lines with written smeta
  lightnvm: pblk: kick writer on new flush points
  lightnvm: pblk: take bitmap alloc. out of critical section

Changes since v1:

 Removed
  lightnvm: pblk: sync RB and RL states during GC
  lightnvm: error handling when whole line is bad
  lightnvm: pblk: handle case when mw_cunits equals to 0

 Added
  lightnvm: pblk: remove unnecessary bio_get/put

Thanks!

Hans Holmberg (5):
  lightnvm: pblk: rework write error recovery path
  lightnvm: pblk: garbage collect lines with failed writes
  lightnvm: pblk: fix smeta write error path
  lightnvm: pblk: only try to recover lines with written smeta
  lightnvm: pblk: kick writer on new flush points

Igor Konopko (2):
  lightnvm: proper error handling for pblk_bio_add_pages
  lightnvm: fix partial read error path

Javier González (2):
  lightnvm: pblk: remove unnecessary bio_get/put
  lightnvm: pblk: take bitmap alloc. out of critical section

Marcin Dziegielewski (1):
  lightnvm: pblk: add possibility to set write buffer size manually

 drivers/lightnvm/pblk-cache.c    |  10 +-
 drivers/lightnvm/pblk-core.c     | 157 +++++++++++++++-------
 drivers/lightnvm/pblk-gc.c       | 102 ++++++++------
 drivers/lightnvm/pblk-init.c     |  62 ++++++---
 drivers/lightnvm/pblk-rb.c       |  43 +-----
 drivers/lightnvm/pblk-read.c     |  69 +++++-----
 drivers/lightnvm/pblk-recovery.c | 121 +++--------------
 drivers/lightnvm/pblk-rl.c       |  29 +++-
 drivers/lightnvm/pblk-sysfs.c    |  15 ++-
 drivers/lightnvm/pblk-write.c    | 281 +++++++++++++++++++++++++--------------
 drivers/lightnvm/pblk.h          |  37 ++++--
 11 files changed, 524 insertions(+), 402 deletions(-)

-- 
2.11.0

^ permalink raw reply	[flat|nested] 12+ messages in thread
* [GIT PULL 01/10] lightnvm: pblk: rework write error recovery path
  2018-06-01 14:41 [GIT PULL v3 00/10] lightnvm updates for 4.18 Matias Bjørling
@ 2018-06-01 14:41 ` Matias Bjørling
  2018-06-01 14:41 ` [GIT PULL 02/10] lightnvm: pblk: garbage collect lines with failed writes Matias Bjørling
                   ` (9 subsequent siblings)
  10 siblings, 0 replies; 12+ messages in thread
From: Matias Bjørling @ 2018-06-01 14:41 UTC (permalink / raw)
  To: axboe
  Cc: linux-block, linux-kernel, javier, igor.j.konopko,
	marcin.dziegielewski, Hans Holmberg, Matias Bjørling

From: Hans Holmberg <hans.holmberg@cnexlabs.com>

The write error recovery path is incomplete, so rework
the write error recovery handling to do resubmits directly
from the write buffer.

When a write error occurs, the remaining sectors in the chunk are
mapped out and invalidated and the request inserted in a resubmit list.

The writer thread checks if there are any requests to resubmit,
scans and invalidates any lbas that have been overwritten by later
writes and resubmits the failed entries.

Signed-off-by: Hans Holmberg <hans.holmberg@cnexlabs.com>
Reviewed-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk-init.c     |   2 +
 drivers/lightnvm/pblk-rb.c       |  39 ------
 drivers/lightnvm/pblk-recovery.c |  91 -------------
 drivers/lightnvm/pblk-write.c    | 279 +++++++++++++++++++++++++--------------
 drivers/lightnvm/pblk.h          |  11 +-
 5 files changed, 187 insertions(+), 235 deletions(-)

diff --git a/drivers/lightnvm/pblk-init.c b/drivers/lightnvm/pblk-init.c
index f47e95c0e5da..2a3302f7e853 100644
--- a/drivers/lightnvm/pblk-init.c
+++ b/drivers/lightnvm/pblk-init.c
@@ -426,6 +426,7 @@ static int pblk_core_init(struct pblk *pblk)
 		goto free_r_end_wq;
 
 	INIT_LIST_HEAD(&pblk->compl_list);
+	INIT_LIST_HEAD(&pblk->resubmit_list);
 
 	return 0;
 
@@ -1185,6 +1186,7 @@ static void *pblk_init(struct nvm_tgt_dev *dev, struct gendisk *tdisk,
 	pblk->state = PBLK_STATE_RUNNING;
 	pblk->gc.gc_enabled = 0;
 
+	spin_lock_init(&pblk->resubmit_lock);
 	spin_lock_init(&pblk->trans_lock);
 	spin_lock_init(&pblk->lock);
 
diff --git a/drivers/lightnvm/pblk-rb.c b/drivers/lightnvm/pblk-rb.c
index 58946ffebe81..1b74ec51a4ad 100644
--- a/drivers/lightnvm/pblk-rb.c
+++ b/drivers/lightnvm/pblk-rb.c
@@ -503,45 +503,6 @@ int pblk_rb_may_write_gc(struct pblk_rb *rb, unsigned int nr_entries,
 }
 
 /*
- * The caller of this function must ensure that the backpointer will not
- * overwrite the entries passed on the list.
- */
-unsigned int pblk_rb_read_to_bio_list(struct pblk_rb *rb, struct bio *bio,
-				      struct list_head *list,
-				      unsigned int max)
-{
-	struct pblk_rb_entry *entry, *tentry;
-	struct page *page;
-	unsigned int read = 0;
-	int ret;
-
-	list_for_each_entry_safe(entry, tentry, list, index) {
-		if (read > max) {
-			pr_err("pblk: too many entries on list\n");
-			goto out;
-		}
-
-		page = virt_to_page(entry->data);
-		if (!page) {
-			pr_err("pblk: could not allocate write bio page\n");
-			goto out;
-		}
-
-		ret = bio_add_page(bio, page, rb->seg_size, 0);
-		if (ret != rb->seg_size) {
-			pr_err("pblk: could not add page to write bio\n");
-			goto out;
-		}
-
-		list_del(&entry->index);
-		read++;
-	}
-
-out:
-	return read;
-}
-
-/*
  * Read available entries on rb and add them to the given bio. To avoid a memory
  * copy, a page reference to the write buffer is used to be added to the bio.
  *
diff --git a/drivers/lightnvm/pblk-recovery.c b/drivers/lightnvm/pblk-recovery.c
index 364ad52a5bfb..788dce87043e 100644
--- a/drivers/lightnvm/pblk-recovery.c
+++ b/drivers/lightnvm/pblk-recovery.c
@@ -16,97 +16,6 @@
 
 #include "pblk.h"
 
-void pblk_submit_rec(struct work_struct *work)
-{
-	struct pblk_rec_ctx *recovery =
-			container_of(work, struct pblk_rec_ctx, ws_rec);
-	struct pblk *pblk = recovery->pblk;
-	struct nvm_rq *rqd = recovery->rqd;
-	struct pblk_c_ctx *c_ctx = nvm_rq_to_pdu(rqd);
-	struct bio *bio;
-	unsigned int nr_rec_secs;
-	unsigned int pgs_read;
-	int ret;
-
-	nr_rec_secs = bitmap_weight((unsigned long int *)&rqd->ppa_status,
-								NVM_MAX_VLBA);
-
-	bio = bio_alloc(GFP_KERNEL, nr_rec_secs);
-
-	bio->bi_iter.bi_sector = 0;
-	bio_set_op_attrs(bio, REQ_OP_WRITE, 0);
-	rqd->bio = bio;
-	rqd->nr_ppas = nr_rec_secs;
-
-	pgs_read = pblk_rb_read_to_bio_list(&pblk->rwb, bio, &recovery->failed,
-								nr_rec_secs);
-	if (pgs_read != nr_rec_secs) {
-		pr_err("pblk: could not read recovery entries\n");
-		goto err;
-	}
-
-	if (pblk_setup_w_rec_rq(pblk, rqd, c_ctx)) {
-		pr_err("pblk: could not setup recovery request\n");
-		goto err;
-	}
-
-#ifdef CONFIG_NVM_DEBUG
-	atomic_long_add(nr_rec_secs, &pblk->recov_writes);
-#endif
-
-	ret = pblk_submit_io(pblk, rqd);
-	if (ret) {
-		pr_err("pblk: I/O submission failed: %d\n", ret);
-		goto err;
-	}
-
-	mempool_free(recovery, &pblk->rec_pool);
-	return;
-
-err:
-	bio_put(bio);
-	pblk_free_rqd(pblk, rqd, PBLK_WRITE);
-}
-
-int pblk_recov_setup_rq(struct pblk *pblk, struct pblk_c_ctx *c_ctx,
-			struct pblk_rec_ctx *recovery, u64 *comp_bits,
-			unsigned int comp)
-{
-	struct nvm_rq *rec_rqd;
-	struct pblk_c_ctx *rec_ctx;
-	int nr_entries = c_ctx->nr_valid + c_ctx->nr_padded;
-
-	rec_rqd = pblk_alloc_rqd(pblk, PBLK_WRITE);
-	rec_ctx = nvm_rq_to_pdu(rec_rqd);
-
-	/* Copy completion bitmap, but exclude the first X completed entries */
-	bitmap_shift_right((unsigned long int *)&rec_rqd->ppa_status,
-				(unsigned long int *)comp_bits,
-				comp, NVM_MAX_VLBA);
-
-	/* Save the context for the entries that need to be re-written and
-	 * update current context with the completed entries.
-	 */
-	rec_ctx->sentry = pblk_rb_wrap_pos(&pblk->rwb, c_ctx->sentry + comp);
-	if (comp >= c_ctx->nr_valid) {
-		rec_ctx->nr_valid = 0;
-		rec_ctx->nr_padded = nr_entries - comp;
-
-		c_ctx->nr_padded = comp - c_ctx->nr_valid;
-	} else {
-		rec_ctx->nr_valid = c_ctx->nr_valid - comp;
-		rec_ctx->nr_padded = c_ctx->nr_padded;
-
-		c_ctx->nr_valid = comp;
-		c_ctx->nr_padded = 0;
-	}
-
-	recovery->rqd = rec_rqd;
-	recovery->pblk = pblk;
-
-	return 0;
-}
-
 int pblk_recov_check_emeta(struct pblk *pblk, struct line_emeta *emeta_buf)
 {
 	u32 crc;
diff --git a/drivers/lightnvm/pblk-write.c b/drivers/lightnvm/pblk-write.c
index aef7fa2d401d..7f9491bc64b5 100644
--- a/drivers/lightnvm/pblk-write.c
+++ b/drivers/lightnvm/pblk-write.c
@@ -103,68 +103,149 @@ static void pblk_complete_write(struct pblk *pblk, struct nvm_rq *rqd,
 	pblk_rb_sync_end(&pblk->rwb, &flags);
 }
 
-/* When a write fails, we are not sure whether the block has grown bad or a page
- * range is more susceptible to write errors. If a high number of pages fail, we
- * assume that the block is bad and we mark it accordingly. In all cases, we
- * remap and resubmit the failed entries as fast as possible; if a flush is
- * waiting on a completion, the whole stack would stall otherwise.
- */
+/* Map remaining sectors in chunk, starting from ppa */
+static void pblk_map_remaining(struct pblk *pblk, struct ppa_addr *ppa)
+{
+	struct nvm_tgt_dev *dev = pblk->dev;
+	struct nvm_geo *geo = &dev->geo;
+	struct pblk_line *line;
+	struct ppa_addr map_ppa = *ppa;
+	u64 paddr;
+	int done = 0;
+
+	line = &pblk->lines[pblk_ppa_to_line(*ppa)];
+	spin_lock(&line->lock);
+
+	while (!done)  {
+		paddr = pblk_dev_ppa_to_line_addr(pblk, map_ppa);
+
+		if (!test_and_set_bit(paddr, line->map_bitmap))
+			line->left_msecs--;
+
+		if (!test_and_set_bit(paddr, line->invalid_bitmap))
+			le32_add_cpu(line->vsc, -1);
+
+		if (geo->version == NVM_OCSSD_SPEC_12) {
+			map_ppa.ppa++;
+			if (map_ppa.g.pg == geo->num_pg)
+				done = 1;
+		} else {
+			map_ppa.m.sec++;
+			if (map_ppa.m.sec == geo->clba)
+				done = 1;
+		}
+	}
+
+	spin_unlock(&line->lock);
+}
+
+static void pblk_prepare_resubmit(struct pblk *pblk, unsigned int sentry,
+				  unsigned int nr_entries)
+{
+	struct pblk_rb *rb = &pblk->rwb;
+	struct pblk_rb_entry *entry;
+	struct pblk_line *line;
+	struct pblk_w_ctx *w_ctx;
+	struct ppa_addr ppa_l2p;
+	int flags;
+	unsigned int pos, i;
+
+	spin_lock(&pblk->trans_lock);
+	pos = sentry;
+	for (i = 0; i < nr_entries; i++) {
+		entry = &rb->entries[pos];
+		w_ctx = &entry->w_ctx;
+
+		/* Check if the lba has been overwritten */
+		ppa_l2p = pblk_trans_map_get(pblk, w_ctx->lba);
+		if (!pblk_ppa_comp(ppa_l2p, entry->cacheline))
+			w_ctx->lba = ADDR_EMPTY;
+
+		/* Mark up the entry as submittable again */
+		flags = READ_ONCE(w_ctx->flags);
+		flags |= PBLK_WRITTEN_DATA;
+		/* Release flags on write context. Protect from writes */
+		smp_store_release(&w_ctx->flags, flags);
+
+		/* Decrese the reference count to the line as we will
+		 * re-map these entries
+		 */
+		line = &pblk->lines[pblk_ppa_to_line(w_ctx->ppa)];
+		kref_put(&line->ref, pblk_line_put);
+
+		pos = (pos + 1) & (rb->nr_entries - 1);
+	}
+	spin_unlock(&pblk->trans_lock);
+}
+
+static void pblk_queue_resubmit(struct pblk *pblk, struct pblk_c_ctx *c_ctx)
+{
+	struct pblk_c_ctx *r_ctx;
+
+	r_ctx = kzalloc(sizeof(struct pblk_c_ctx), GFP_KERNEL);
+	if (!r_ctx)
+		return;
+
+	r_ctx->lun_bitmap = NULL;
+	r_ctx->sentry = c_ctx->sentry;
+	r_ctx->nr_valid = c_ctx->nr_valid;
+	r_ctx->nr_padded = c_ctx->nr_padded;
+
+	spin_lock(&pblk->resubmit_lock);
+	list_add_tail(&r_ctx->list, &pblk->resubmit_list);
+	spin_unlock(&pblk->resubmit_lock);
+
+#ifdef CONFIG_NVM_DEBUG
+	atomic_long_add(c_ctx->nr_valid, &pblk->recov_writes);
+#endif
+}
+
+static void pblk_submit_rec(struct work_struct *work)
+{
+	struct pblk_rec_ctx *recovery =
+			container_of(work, struct pblk_rec_ctx, ws_rec);
+	struct pblk *pblk = recovery->pblk;
+	struct nvm_rq *rqd = recovery->rqd;
+	struct pblk_c_ctx *c_ctx = nvm_rq_to_pdu(rqd);
+	struct ppa_addr *ppa_list;
+
+	pblk_log_write_err(pblk, rqd);
+
+	if (rqd->nr_ppas == 1)
+		ppa_list = &rqd->ppa_addr;
+	else
+		ppa_list = rqd->ppa_list;
+
+	pblk_map_remaining(pblk, ppa_list);
+	pblk_queue_resubmit(pblk, c_ctx);
+
+	pblk_up_rq(pblk, rqd->ppa_list, rqd->nr_ppas, c_ctx->lun_bitmap);
+	if (c_ctx->nr_padded)
+		pblk_bio_free_pages(pblk, rqd->bio, c_ctx->nr_valid,
+							c_ctx->nr_padded);
+	bio_put(rqd->bio);
+	pblk_free_rqd(pblk, rqd, PBLK_WRITE);
+	mempool_free(recovery, &pblk->rec_pool);
+
+	atomic_dec(&pblk->inflight_io);
+}
+
+
 static void pblk_end_w_fail(struct pblk *pblk, struct nvm_rq *rqd)
 {
-	void *comp_bits = &rqd->ppa_status;
-	struct pblk_c_ctx *c_ctx = nvm_rq_to_pdu(rqd);
 	struct pblk_rec_ctx *recovery;
-	struct ppa_addr *ppa_list = rqd->ppa_list;
-	int nr_ppas = rqd->nr_ppas;
-	unsigned int c_entries;
-	int bit, ret;
-
-	if (unlikely(nr_ppas == 1))
-		ppa_list = &rqd->ppa_addr;
 
 	recovery = mempool_alloc(&pblk->rec_pool, GFP_ATOMIC);
-
-	INIT_LIST_HEAD(&recovery->failed);
-
-	bit = -1;
-	while ((bit = find_next_bit(comp_bits, nr_ppas, bit + 1)) < nr_ppas) {
-		struct pblk_rb_entry *entry;
-		struct ppa_addr ppa;
-
-		/* Logic error */
-		if (bit > c_ctx->nr_valid) {
-			WARN_ONCE(1, "pblk: corrupted write request\n");
-			mempool_free(recovery, &pblk->rec_pool);
-			goto out;
-		}
-
-		ppa = ppa_list[bit];
-		entry = pblk_rb_sync_scan_entry(&pblk->rwb, &ppa);
-		if (!entry) {
-			pr_err("pblk: could not scan entry on write failure\n");
-			mempool_free(recovery, &pblk->rec_pool);
-			goto out;
-		}
-
-		/* The list is filled first and emptied afterwards. No need for
-		 * protecting it with a lock
-		 */
-		list_add_tail(&entry->index, &recovery->failed);
+	if (!recovery) {
+		pr_err("pblk: could not allocate recovery work\n");
+		return;
 	}
 
-	c_entries = find_first_bit(comp_bits, nr_ppas);
-	ret = pblk_recov_setup_rq(pblk, c_ctx, recovery, comp_bits, c_entries);
-	if (ret) {
-		pr_err("pblk: could not recover from write failure\n");
-		mempool_free(recovery, &pblk->rec_pool);
-		goto out;
-	}
+	recovery->pblk = pblk;
+	recovery->rqd = rqd;
 
 	INIT_WORK(&recovery->ws_rec, pblk_submit_rec);
 	queue_work(pblk->close_wq, &recovery->ws_rec);
-
-out:
-	pblk_complete_write(pblk, rqd, c_ctx);
 }
 
 static void pblk_end_io_write(struct nvm_rq *rqd)
@@ -173,8 +254,8 @@ static void pblk_end_io_write(struct nvm_rq *rqd)
 	struct pblk_c_ctx *c_ctx = nvm_rq_to_pdu(rqd);
 
 	if (rqd->error) {
-		pblk_log_write_err(pblk, rqd);
-		return pblk_end_w_fail(pblk, rqd);
+		pblk_end_w_fail(pblk, rqd);
+		return;
 	}
 #ifdef CONFIG_NVM_DEBUG
 	else
@@ -266,31 +347,6 @@ static int pblk_setup_w_rq(struct pblk *pblk, struct nvm_rq *rqd,
 	return 0;
 }
 
-int pblk_setup_w_rec_rq(struct pblk *pblk, struct nvm_rq *rqd,
-			struct pblk_c_ctx *c_ctx)
-{
-	struct pblk_line_meta *lm = &pblk->lm;
-	unsigned long *lun_bitmap;
-	int ret;
-
-	lun_bitmap = kzalloc(lm->lun_bitmap_len, GFP_KERNEL);
-	if (!lun_bitmap)
-		return -ENOMEM;
-
-	c_ctx->lun_bitmap = lun_bitmap;
-
-	ret = pblk_alloc_w_rq(pblk, rqd, rqd->nr_ppas, pblk_end_io_write);
-	if (ret)
-		return ret;
-
-	pblk_map_rq(pblk, rqd, c_ctx->sentry, lun_bitmap, c_ctx->nr_valid, 0);
-
-	rqd->ppa_status = (u64)0;
-	rqd->flags = pblk_set_progr_mode(pblk, PBLK_WRITE);
-
-	return ret;
-}
-
 static int pblk_calc_secs_to_sync(struct pblk *pblk, unsigned int secs_avail,
 				  unsigned int secs_to_flush)
 {
@@ -339,6 +395,7 @@ int pblk_submit_meta_io(struct pblk *pblk, struct pblk_line *meta_line)
 	bio = pblk_bio_map_addr(pblk, data, rq_ppas, rq_len,
 					l_mg->emeta_alloc_type, GFP_KERNEL);
 	if (IS_ERR(bio)) {
+		pr_err("pblk: failed to map emeta io");
 		ret = PTR_ERR(bio);
 		goto fail_free_rqd;
 	}
@@ -515,27 +572,55 @@ static int pblk_submit_write(struct pblk *pblk)
 	unsigned int secs_avail, secs_to_sync, secs_to_com;
 	unsigned int secs_to_flush;
 	unsigned long pos;
+	unsigned int resubmit;
 
-	/* If there are no sectors in the cache, flushes (bios without data)
-	 * will be cleared on the cache threads
-	 */
-	secs_avail = pblk_rb_read_count(&pblk->rwb);
-	if (!secs_avail)
-		return 1;
-
-	secs_to_flush = pblk_rb_flush_point_count(&pblk->rwb);
-	if (!secs_to_flush && secs_avail < pblk->min_write_pgs)
-		return 1;
-
-	secs_to_sync = pblk_calc_secs_to_sync(pblk, secs_avail, secs_to_flush);
-	if (secs_to_sync > pblk->max_write_pgs) {
-		pr_err("pblk: bad buffer sync calculation\n");
-		return 1;
+	spin_lock(&pblk->resubmit_lock);
+	resubmit = !list_empty(&pblk->resubmit_list);
+	spin_unlock(&pblk->resubmit_lock);
+
+	/* Resubmit failed writes first */
+	if (resubmit) {
+		struct pblk_c_ctx *r_ctx;
+
+		spin_lock(&pblk->resubmit_lock);
+		r_ctx = list_first_entry(&pblk->resubmit_list,
+					struct pblk_c_ctx, list);
+		list_del(&r_ctx->list);
+		spin_unlock(&pblk->resubmit_lock);
+
+		secs_avail = r_ctx->nr_valid;
+		pos = r_ctx->sentry;
+
+		pblk_prepare_resubmit(pblk, pos, secs_avail);
+		secs_to_sync = pblk_calc_secs_to_sync(pblk, secs_avail,
+				secs_avail);
+
+		kfree(r_ctx);
+	} else {
+		/* If there are no sectors in the cache,
+		 * flushes (bios without data) will be cleared on
+		 * the cache threads
+		 */
+		secs_avail = pblk_rb_read_count(&pblk->rwb);
+		if (!secs_avail)
+			return 1;
+
+		secs_to_flush = pblk_rb_flush_point_count(&pblk->rwb);
+		if (!secs_to_flush && secs_avail < pblk->min_write_pgs)
+			return 1;
+
+		secs_to_sync = pblk_calc_secs_to_sync(pblk, secs_avail,
+					secs_to_flush);
+		if (secs_to_sync > pblk->max_write_pgs) {
+			pr_err("pblk: bad buffer sync calculation\n");
+			return 1;
+		}
+
+		secs_to_com = (secs_to_sync > secs_avail) ?
+			secs_avail : secs_to_sync;
+		pos = pblk_rb_read_commit(&pblk->rwb, secs_to_com);
 	}
 
-	secs_to_com = (secs_to_sync > secs_avail) ? secs_avail : secs_to_sync;
-	pos = pblk_rb_read_commit(&pblk->rwb, secs_to_com);
-
 	bio = bio_alloc(GFP_KERNEL, secs_to_sync);
 
 	bio->bi_iter.bi_sector = 0; /* internal bio */
diff --git a/drivers/lightnvm/pblk.h b/drivers/lightnvm/pblk.h
index 9f22bbb54a6d..043a851a8e6c 100644
--- a/drivers/lightnvm/pblk.h
+++ b/drivers/lightnvm/pblk.h
@@ -128,7 +128,6 @@ struct pblk_pad_rq {
 struct pblk_rec_ctx {
 	struct pblk *pblk;
 	struct nvm_rq *rqd;
-	struct list_head failed;
 	struct work_struct ws_rec;
 };
 
@@ -664,6 +663,9 @@ struct pblk {
 
 	struct list_head compl_list;
 
+	spinlock_t resubmit_lock;	 /* Resubmit list lock */
+	struct list_head resubmit_list; /* Resubmit list for failed writes*/
+
 	mempool_t page_bio_pool;
 	mempool_t gen_ws_pool;
 	mempool_t rec_pool;
@@ -713,9 +715,6 @@ void pblk_rb_sync_l2p(struct pblk_rb *rb);
 unsigned int pblk_rb_read_to_bio(struct pblk_rb *rb, struct nvm_rq *rqd,
 				 unsigned int pos, unsigned int nr_entries,
 				 unsigned int count);
-unsigned int pblk_rb_read_to_bio_list(struct pblk_rb *rb, struct bio *bio,
-				      struct list_head *list,
-				      unsigned int max);
 int pblk_rb_copy_to_bio(struct pblk_rb *rb, struct bio *bio, sector_t lba,
 			struct ppa_addr ppa, int bio_iter, bool advanced_bio);
 unsigned int pblk_rb_read_commit(struct pblk_rb *rb, unsigned int entries);
@@ -848,13 +847,9 @@ int pblk_submit_read_gc(struct pblk *pblk, struct pblk_gc_rq *gc_rq);
 /*
  * pblk recovery
  */
-void pblk_submit_rec(struct work_struct *work);
 struct pblk_line *pblk_recov_l2p(struct pblk *pblk);
 int pblk_recov_pad(struct pblk *pblk);
 int pblk_recov_check_emeta(struct pblk *pblk, struct line_emeta *emeta);
-int pblk_recov_setup_rq(struct pblk *pblk, struct pblk_c_ctx *c_ctx,
-			struct pblk_rec_ctx *recovery, u64 *comp_bits,
-			unsigned int comp);
 
 /*
  * pblk gc
-- 
2.11.0

^ permalink raw reply related	[flat|nested] 12+ messages in thread
* [GIT PULL 02/10] lightnvm: pblk: garbage collect lines with failed writes
  2018-06-01 14:41 [GIT PULL v3 00/10] lightnvm updates for 4.18 Matias Bjørling
  2018-06-01 14:41 ` [GIT PULL 01/10] lightnvm: pblk: rework write error recovery path Matias Bjørling
@ 2018-06-01 14:41 ` Matias Bjørling
  2018-06-01 14:41 ` [GIT PULL 03/10] lightnvm: pblk: fix smeta write error path Matias Bjørling
                   ` (8 subsequent siblings)
  10 siblings, 0 replies; 12+ messages in thread
From: Matias Bjørling @ 2018-06-01 14:41 UTC (permalink / raw)
  To: axboe
  Cc: linux-block, linux-kernel, javier, igor.j.konopko,
	marcin.dziegielewski, Hans Holmberg, Matias Bjørling

From: Hans Holmberg <hans.holmberg@cnexlabs.com>

Write failures should not happen under normal circumstances,
so in order to bring the chunk back into a known state as soon
as possible, evacuate all the valid data out of the line and let the
fw judge if the block can be written to in the next reset cycle.

Do this by introducing a new gc list for lines with failed writes,
and ensure that the rate limiter allocates a small portion of
the write bandwidth to get the job done.

The lba list is saved in memory for use during gc as we
cannot gurantee that the emeta data is readable if a write
error occurred.

Signed-off-by: Hans Holmberg <hans.holmberg@cnexlabs.com>
Reviewed-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk-core.c  |  45 ++++++++++++++++++-
 drivers/lightnvm/pblk-gc.c    | 102 +++++++++++++++++++++++++++---------------
 drivers/lightnvm/pblk-init.c  |  46 ++++++++++++-------
 drivers/lightnvm/pblk-rl.c    |  29 ++++++++++--
 drivers/lightnvm/pblk-sysfs.c |  15 ++++++-
 drivers/lightnvm/pblk-write.c |   2 +
 drivers/lightnvm/pblk.h       |  25 +++++++++--
 7 files changed, 200 insertions(+), 64 deletions(-)

diff --git a/drivers/lightnvm/pblk-core.c b/drivers/lightnvm/pblk-core.c
index 56e3934712f9..e81a70cf9e9b 100644
--- a/drivers/lightnvm/pblk-core.c
+++ b/drivers/lightnvm/pblk-core.c
@@ -373,7 +373,13 @@ struct list_head *pblk_line_gc_list(struct pblk *pblk, struct pblk_line *line)
 
 	lockdep_assert_held(&line->lock);
 
-	if (!vsc) {
+	if (line->w_err_gc->has_write_err) {
+		if (line->gc_group != PBLK_LINEGC_WERR) {
+			line->gc_group = PBLK_LINEGC_WERR;
+			move_list = &l_mg->gc_werr_list;
+			pblk_rl_werr_line_in(&pblk->rl);
+		}
+	} else if (!vsc) {
 		if (line->gc_group != PBLK_LINEGC_FULL) {
 			line->gc_group = PBLK_LINEGC_FULL;
 			move_list = &l_mg->gc_full_list;
@@ -1589,8 +1595,13 @@ static void __pblk_line_put(struct pblk *pblk, struct pblk_line *line)
 	line->state = PBLK_LINESTATE_FREE;
 	line->gc_group = PBLK_LINEGC_NONE;
 	pblk_line_free(line);
+
+	if (line->w_err_gc->has_write_err) {
+		pblk_rl_werr_line_out(&pblk->rl);
+		line->w_err_gc->has_write_err = 0;
+	}
+
 	spin_unlock(&line->lock);
-
 	atomic_dec(&gc->pipeline_gc);
 
 	spin_lock(&l_mg->free_lock);
@@ -1753,11 +1764,34 @@ void pblk_line_close_meta(struct pblk *pblk, struct pblk_line *line)
 
 	spin_lock(&l_mg->close_lock);
 	spin_lock(&line->lock);
+
+	/* Update the in-memory start address for emeta, in case it has
+	 * shifted due to write errors
+	 */
+	if (line->emeta_ssec != line->cur_sec)
+		line->emeta_ssec = line->cur_sec;
+
 	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);
+
+
+}
+
+static void pblk_save_lba_list(struct pblk *pblk, struct pblk_line *line)
+{
+	struct pblk_line_meta *lm = &pblk->lm;
+	struct pblk_line_mgmt *l_mg = &pblk->l_mg;
+	unsigned int lba_list_size = lm->emeta_len[2];
+	struct pblk_w_err_gc *w_err_gc = line->w_err_gc;
+	struct pblk_emeta *emeta = line->emeta;
+
+	w_err_gc->lba_list = pblk_malloc(lba_list_size,
+					 l_mg->emeta_alloc_type, GFP_KERNEL);
+	memcpy(w_err_gc->lba_list, emeta_to_lbas(pblk, emeta->buf),
+				lba_list_size);
 }
 
 void pblk_line_close_ws(struct work_struct *work)
@@ -1766,6 +1800,13 @@ void pblk_line_close_ws(struct work_struct *work)
 									ws);
 	struct pblk *pblk = line_ws->pblk;
 	struct pblk_line *line = line_ws->line;
+	struct pblk_w_err_gc *w_err_gc = line->w_err_gc;
+
+	/* Write errors makes the emeta start address stored in smeta invalid,
+	 * so keep a copy of the lba list until we've gc'd the line
+	 */
+	if (w_err_gc->has_write_err)
+		pblk_save_lba_list(pblk, line);
 
 	pblk_line_close(pblk, line);
 	mempool_free(line_ws, &pblk->gen_ws_pool);
diff --git a/drivers/lightnvm/pblk-gc.c b/drivers/lightnvm/pblk-gc.c
index b0cc277bf972..df88f1bdd921 100644
--- a/drivers/lightnvm/pblk-gc.c
+++ b/drivers/lightnvm/pblk-gc.c
@@ -129,6 +129,53 @@ static void pblk_gc_line_ws(struct work_struct *work)
 	kfree(gc_rq_ws);
 }
 
+static __le64 *get_lba_list_from_emeta(struct pblk *pblk,
+				       struct pblk_line *line)
+{
+	struct line_emeta *emeta_buf;
+	struct pblk_line_mgmt *l_mg = &pblk->l_mg;
+	struct pblk_line_meta *lm = &pblk->lm;
+	unsigned int lba_list_size = lm->emeta_len[2];
+	__le64 *lba_list;
+	int ret;
+
+	emeta_buf = pblk_malloc(lm->emeta_len[0],
+				l_mg->emeta_alloc_type, GFP_KERNEL);
+	if (!emeta_buf)
+		return NULL;
+
+	ret = pblk_line_read_emeta(pblk, line, emeta_buf);
+	if (ret) {
+		pr_err("pblk: line %d read emeta failed (%d)\n",
+				line->id, ret);
+		pblk_mfree(emeta_buf, l_mg->emeta_alloc_type);
+		return NULL;
+	}
+
+	/* If this read fails, it means that emeta is corrupted.
+	 * For now, leave the line untouched.
+	 * TODO: Implement a recovery routine that scans and moves
+	 * all sectors on the line.
+	 */
+
+	ret = pblk_recov_check_emeta(pblk, emeta_buf);
+	if (ret) {
+		pr_err("pblk: inconsistent emeta (line %d)\n",
+				line->id);
+		pblk_mfree(emeta_buf, l_mg->emeta_alloc_type);
+		return NULL;
+	}
+
+	lba_list = pblk_malloc(lba_list_size,
+			       l_mg->emeta_alloc_type, GFP_KERNEL);
+	if (lba_list)
+		memcpy(lba_list, emeta_to_lbas(pblk, emeta_buf), lba_list_size);
+
+	pblk_mfree(emeta_buf, l_mg->emeta_alloc_type);
+
+	return lba_list;
+}
+
 static void pblk_gc_line_prepare_ws(struct work_struct *work)
 {
 	struct pblk_line_ws *line_ws = container_of(work, struct pblk_line_ws,
@@ -138,46 +185,26 @@ static void pblk_gc_line_prepare_ws(struct work_struct *work)
 	struct pblk_line_mgmt *l_mg = &pblk->l_mg;
 	struct pblk_line_meta *lm = &pblk->lm;
 	struct pblk_gc *gc = &pblk->gc;
-	struct line_emeta *emeta_buf;
 	struct pblk_line_ws *gc_rq_ws;
 	struct pblk_gc_rq *gc_rq;
 	__le64 *lba_list;
 	unsigned long *invalid_bitmap;
 	int sec_left, nr_secs, bit;
-	int ret;
 
 	invalid_bitmap = kmalloc(lm->sec_bitmap_len, GFP_KERNEL);
 	if (!invalid_bitmap)
 		goto fail_free_ws;
 
-	emeta_buf = pblk_malloc(lm->emeta_len[0], l_mg->emeta_alloc_type,
-								GFP_KERNEL);
-	if (!emeta_buf) {
-		pr_err("pblk: cannot use GC emeta\n");
-		goto fail_free_bitmap;
-	}
-
-	ret = pblk_line_read_emeta(pblk, line, emeta_buf);
-	if (ret) {
-		pr_err("pblk: line %d read emeta failed (%d)\n", line->id, ret);
-		goto fail_free_emeta;
-	}
-
-	/* If this read fails, it means that emeta is corrupted. For now, leave
-	 * the line untouched. TODO: Implement a recovery routine that scans and
-	 * moves all sectors on the line.
-	 */
-
-	ret = pblk_recov_check_emeta(pblk, emeta_buf);
-	if (ret) {
-		pr_err("pblk: inconsistent emeta (line %d)\n", line->id);
-		goto fail_free_emeta;
-	}
-
-	lba_list = emeta_to_lbas(pblk, emeta_buf);
-	if (!lba_list) {
-		pr_err("pblk: could not interpret emeta (line %d)\n", line->id);
-		goto fail_free_emeta;
+	if (line->w_err_gc->has_write_err) {
+		lba_list = line->w_err_gc->lba_list;
+		line->w_err_gc->lba_list = NULL;
+	} else {
+		lba_list = get_lba_list_from_emeta(pblk, line);
+		if (!lba_list) {
+			pr_err("pblk: could not interpret emeta (line %d)\n",
+					line->id);
+			goto fail_free_ws;
+		}
 	}
 
 	spin_lock(&line->lock);
@@ -187,14 +214,14 @@ static void pblk_gc_line_prepare_ws(struct work_struct *work)
 
 	if (sec_left < 0) {
 		pr_err("pblk: corrupted GC line (%d)\n", line->id);
-		goto fail_free_emeta;
+		goto fail_free_lba_list;
 	}
 
 	bit = -1;
 next_rq:
 	gc_rq = kmalloc(sizeof(struct pblk_gc_rq), GFP_KERNEL);
 	if (!gc_rq)
-		goto fail_free_emeta;
+		goto fail_free_lba_list;
 
 	nr_secs = 0;
 	do {
@@ -240,7 +267,7 @@ static void pblk_gc_line_prepare_ws(struct work_struct *work)
 		goto next_rq;
 
 out:
-	pblk_mfree(emeta_buf, l_mg->emeta_alloc_type);
+	pblk_mfree(lba_list, l_mg->emeta_alloc_type);
 	kfree(line_ws);
 	kfree(invalid_bitmap);
 
@@ -251,9 +278,8 @@ static void pblk_gc_line_prepare_ws(struct work_struct *work)
 
 fail_free_gc_rq:
 	kfree(gc_rq);
-fail_free_emeta:
-	pblk_mfree(emeta_buf, l_mg->emeta_alloc_type);
-fail_free_bitmap:
+fail_free_lba_list:
+	pblk_mfree(lba_list, l_mg->emeta_alloc_type);
 	kfree(invalid_bitmap);
 fail_free_ws:
 	kfree(line_ws);
@@ -349,12 +375,14 @@ static struct pblk_line *pblk_gc_get_victim_line(struct pblk *pblk,
 static bool pblk_gc_should_run(struct pblk_gc *gc, struct pblk_rl *rl)
 {
 	unsigned int nr_blocks_free, nr_blocks_need;
+	unsigned int werr_lines = atomic_read(&rl->werr_lines);
 
 	nr_blocks_need = pblk_rl_high_thrs(rl);
 	nr_blocks_free = pblk_rl_nr_free_blks(rl);
 
 	/* This is not critical, no need to take lock here */
-	return ((gc->gc_active) && (nr_blocks_need > nr_blocks_free));
+	return ((werr_lines > 0) ||
+		((gc->gc_active) && (nr_blocks_need > nr_blocks_free)));
 }
 
 void pblk_gc_free_full_lines(struct pblk *pblk)
diff --git a/drivers/lightnvm/pblk-init.c b/drivers/lightnvm/pblk-init.c
index 2a3302f7e853..783887b891b7 100644
--- a/drivers/lightnvm/pblk-init.c
+++ b/drivers/lightnvm/pblk-init.c
@@ -493,11 +493,17 @@ static void pblk_line_mg_free(struct pblk *pblk)
 	}
 }
 
-static void pblk_line_meta_free(struct pblk_line *line)
+static void pblk_line_meta_free(struct pblk_line_mgmt *l_mg,
+				struct pblk_line *line)
 {
+	struct pblk_w_err_gc *w_err_gc = line->w_err_gc;
+
 	kfree(line->blk_bitmap);
 	kfree(line->erase_bitmap);
 	kfree(line->chks);
+
+	pblk_mfree(w_err_gc->lba_list, l_mg->emeta_alloc_type);
+	kfree(w_err_gc);
 }
 
 static void pblk_lines_free(struct pblk *pblk)
@@ -511,7 +517,7 @@ static void pblk_lines_free(struct pblk *pblk)
 		line = &pblk->lines[i];
 
 		pblk_line_free(line);
-		pblk_line_meta_free(line);
+		pblk_line_meta_free(l_mg, line);
 	}
 	spin_unlock(&l_mg->free_lock);
 
@@ -813,20 +819,28 @@ static int pblk_alloc_line_meta(struct pblk *pblk, struct pblk_line *line)
 		return -ENOMEM;
 
 	line->erase_bitmap = kzalloc(lm->blk_bitmap_len, GFP_KERNEL);
-	if (!line->erase_bitmap) {
-		kfree(line->blk_bitmap);
-		return -ENOMEM;
-	}
+	if (!line->erase_bitmap)
+		goto free_blk_bitmap;
+
 
 	line->chks = kmalloc(lm->blk_per_line * sizeof(struct nvm_chk_meta),
 								GFP_KERNEL);
-	if (!line->chks) {
-		kfree(line->erase_bitmap);
-		kfree(line->blk_bitmap);
-		return -ENOMEM;
-	}
+	if (!line->chks)
+		goto free_erase_bitmap;
+
+	line->w_err_gc = kzalloc(sizeof(struct pblk_w_err_gc), GFP_KERNEL);
+	if (!line->w_err_gc)
+		goto free_chks;
 
 	return 0;
+
+free_chks:
+	kfree(line->chks);
+free_erase_bitmap:
+	kfree(line->erase_bitmap);
+free_blk_bitmap:
+	kfree(line->blk_bitmap);
+	return -ENOMEM;
 }
 
 static int pblk_line_mg_init(struct pblk *pblk)
@@ -851,12 +865,14 @@ static int pblk_line_mg_init(struct pblk *pblk)
 	INIT_LIST_HEAD(&l_mg->gc_mid_list);
 	INIT_LIST_HEAD(&l_mg->gc_low_list);
 	INIT_LIST_HEAD(&l_mg->gc_empty_list);
+	INIT_LIST_HEAD(&l_mg->gc_werr_list);
 
 	INIT_LIST_HEAD(&l_mg->emeta_list);
 
-	l_mg->gc_lists[0] = &l_mg->gc_high_list;
-	l_mg->gc_lists[1] = &l_mg->gc_mid_list;
-	l_mg->gc_lists[2] = &l_mg->gc_low_list;
+	l_mg->gc_lists[0] = &l_mg->gc_werr_list;
+	l_mg->gc_lists[1] = &l_mg->gc_high_list;
+	l_mg->gc_lists[2] = &l_mg->gc_mid_list;
+	l_mg->gc_lists[3] = &l_mg->gc_low_list;
 
 	spin_lock_init(&l_mg->free_lock);
 	spin_lock_init(&l_mg->close_lock);
@@ -1063,7 +1079,7 @@ static int pblk_lines_init(struct pblk *pblk)
 
 fail_free_lines:
 	while (--i >= 0)
-		pblk_line_meta_free(&pblk->lines[i]);
+		pblk_line_meta_free(l_mg, &pblk->lines[i]);
 	kfree(pblk->lines);
 fail_free_chunk_meta:
 	kfree(chunk_meta);
diff --git a/drivers/lightnvm/pblk-rl.c b/drivers/lightnvm/pblk-rl.c
index 883a7113b19d..6a0616a6fcaf 100644
--- a/drivers/lightnvm/pblk-rl.c
+++ b/drivers/lightnvm/pblk-rl.c
@@ -73,6 +73,16 @@ void pblk_rl_user_in(struct pblk_rl *rl, int nr_entries)
 	pblk_rl_kick_u_timer(rl);
 }
 
+void pblk_rl_werr_line_in(struct pblk_rl *rl)
+{
+	atomic_inc(&rl->werr_lines);
+}
+
+void pblk_rl_werr_line_out(struct pblk_rl *rl)
+{
+	atomic_dec(&rl->werr_lines);
+}
+
 void pblk_rl_gc_in(struct pblk_rl *rl, int nr_entries)
 {
 	atomic_add(nr_entries, &rl->rb_gc_cnt);
@@ -99,11 +109,21 @@ static void __pblk_rl_update_rates(struct pblk_rl *rl,
 {
 	struct pblk *pblk = container_of(rl, struct pblk, rl);
 	int max = rl->rb_budget;
+	int werr_gc_needed = atomic_read(&rl->werr_lines);
 
 	if (free_blocks >= rl->high) {
-		rl->rb_user_max = max;
-		rl->rb_gc_max = 0;
-		rl->rb_state = PBLK_RL_HIGH;
+		if (werr_gc_needed) {
+			/* Allocate a small budget for recovering
+			 * lines with write errors
+			 */
+			rl->rb_gc_max = 1 << rl->rb_windows_pw;
+			rl->rb_user_max = max - rl->rb_gc_max;
+			rl->rb_state = PBLK_RL_WERR;
+		} else {
+			rl->rb_user_max = max;
+			rl->rb_gc_max = 0;
+			rl->rb_state = PBLK_RL_OFF;
+		}
 	} else if (free_blocks < rl->high) {
 		int shift = rl->high_pw - rl->rb_windows_pw;
 		int user_windows = free_blocks >> shift;
@@ -124,7 +144,7 @@ static void __pblk_rl_update_rates(struct pblk_rl *rl,
 		rl->rb_state = PBLK_RL_LOW;
 	}
 
-	if (rl->rb_state == (PBLK_RL_MID | PBLK_RL_LOW))
+	if (rl->rb_state != PBLK_RL_OFF)
 		pblk_gc_should_start(pblk);
 	else
 		pblk_gc_should_stop(pblk);
@@ -221,6 +241,7 @@ void pblk_rl_init(struct pblk_rl *rl, int budget)
 	atomic_set(&rl->rb_user_cnt, 0);
 	atomic_set(&rl->rb_gc_cnt, 0);
 	atomic_set(&rl->rb_space, -1);
+	atomic_set(&rl->werr_lines, 0);
 
 	timer_setup(&rl->u_timer, pblk_rl_u_timer, 0);
 
diff --git a/drivers/lightnvm/pblk-sysfs.c b/drivers/lightnvm/pblk-sysfs.c
index e61909af23a5..88a0a7c407aa 100644
--- a/drivers/lightnvm/pblk-sysfs.c
+++ b/drivers/lightnvm/pblk-sysfs.c
@@ -173,6 +173,8 @@ static ssize_t pblk_sysfs_lines(struct pblk *pblk, char *page)
 	int free_line_cnt = 0, closed_line_cnt = 0, emeta_line_cnt = 0;
 	int d_line_cnt = 0, l_line_cnt = 0;
 	int gc_full = 0, gc_high = 0, gc_mid = 0, gc_low = 0, gc_empty = 0;
+	int gc_werr = 0;
+
 	int bad = 0, cor = 0;
 	int msecs = 0, cur_sec = 0, vsc = 0, sec_in_line = 0;
 	int map_weight = 0, meta_weight = 0;
@@ -237,6 +239,15 @@ static ssize_t pblk_sysfs_lines(struct pblk *pblk, char *page)
 		gc_empty++;
 	}
 
+	list_for_each_entry(line, &l_mg->gc_werr_list, list) {
+		if (line->type == PBLK_LINETYPE_DATA)
+			d_line_cnt++;
+		else if (line->type == PBLK_LINETYPE_LOG)
+			l_line_cnt++;
+		closed_line_cnt++;
+		gc_werr++;
+	}
+
 	list_for_each_entry(line, &l_mg->bad_list, list)
 		bad++;
 	list_for_each_entry(line, &l_mg->corrupt_list, list)
@@ -275,8 +286,8 @@ static ssize_t pblk_sysfs_lines(struct pblk *pblk, char *page)
 					l_mg->nr_lines);
 
 	sz += snprintf(page + sz, PAGE_SIZE - sz,
-		"GC: full:%d, high:%d, mid:%d, low:%d, empty:%d, queue:%d\n",
-			gc_full, gc_high, gc_mid, gc_low, gc_empty,
+		"GC: full:%d, high:%d, mid:%d, low:%d, empty:%d, werr: %d, queue:%d\n",
+			gc_full, gc_high, gc_mid, gc_low, gc_empty, gc_werr,
 			atomic_read(&pblk->gc.read_inflight_gc));
 
 	sz += snprintf(page + sz, PAGE_SIZE - sz,
diff --git a/drivers/lightnvm/pblk-write.c b/drivers/lightnvm/pblk-write.c
index 7f9491bc64b5..f353e52941f5 100644
--- a/drivers/lightnvm/pblk-write.c
+++ b/drivers/lightnvm/pblk-write.c
@@ -136,6 +136,7 @@ static void pblk_map_remaining(struct pblk *pblk, struct ppa_addr *ppa)
 		}
 	}
 
+	line->w_err_gc->has_write_err = 1;
 	spin_unlock(&line->lock);
 }
 
@@ -279,6 +280,7 @@ static void pblk_end_io_write_meta(struct nvm_rq *rqd)
 	if (rqd->error) {
 		pblk_log_write_err(pblk, rqd);
 		pr_err("pblk: metadata I/O failed. Line %d\n", line->id);
+		line->w_err_gc->has_write_err = 1;
 	}
 
 	sync = atomic_add_return(rqd->nr_ppas, &emeta->sync);
diff --git a/drivers/lightnvm/pblk.h b/drivers/lightnvm/pblk.h
index 043a851a8e6c..725bc1bfa849 100644
--- a/drivers/lightnvm/pblk.h
+++ b/drivers/lightnvm/pblk.h
@@ -89,12 +89,14 @@ struct pblk_sec_meta {
 /* The number of GC lists and the rate-limiter states go together. This way the
  * rate-limiter can dictate how much GC is needed based on resource utilization.
  */
-#define PBLK_GC_NR_LISTS 3
+#define PBLK_GC_NR_LISTS 4
 
 enum {
-	PBLK_RL_HIGH = 1,
-	PBLK_RL_MID = 2,
-	PBLK_RL_LOW = 3,
+	PBLK_RL_OFF = 0,
+	PBLK_RL_WERR = 1,
+	PBLK_RL_HIGH = 2,
+	PBLK_RL_MID = 3,
+	PBLK_RL_LOW = 4
 };
 
 #define pblk_dma_meta_size (sizeof(struct pblk_sec_meta) * PBLK_MAX_REQ_ADDRS)
@@ -278,6 +280,8 @@ struct pblk_rl {
 	int rb_user_active;
 	int rb_gc_active;
 
+	atomic_t werr_lines;	/* Number of write error lines that needs gc */
+
 	struct timer_list u_timer;
 
 	unsigned long long nr_secs;
@@ -311,6 +315,7 @@ enum {
 	PBLK_LINEGC_MID = 23,
 	PBLK_LINEGC_HIGH = 24,
 	PBLK_LINEGC_FULL = 25,
+	PBLK_LINEGC_WERR = 26
 };
 
 #define PBLK_MAGIC 0x70626c6b /*pblk*/
@@ -412,6 +417,11 @@ struct pblk_smeta {
 	struct line_smeta *buf;		/* smeta buffer in persistent format */
 };
 
+struct pblk_w_err_gc {
+	int has_write_err;
+	__le64 *lba_list;
+};
+
 struct pblk_line {
 	struct pblk *pblk;
 	unsigned int id;		/* Line number corresponds to the
@@ -457,6 +467,8 @@ struct pblk_line {
 
 	struct kref ref;		/* Write buffer L2P references */
 
+	struct pblk_w_err_gc *w_err_gc;	/* Write error gc recovery metadata */
+
 	spinlock_t lock;		/* Necessary for invalid_bitmap only */
 };
 
@@ -488,6 +500,8 @@ struct pblk_line_mgmt {
 	struct list_head gc_mid_list;	/* Full lines ready to GC, mid isc */
 	struct list_head gc_low_list;	/* Full lines ready to GC, low isc */
 
+	struct list_head gc_werr_list;  /* Write err recovery list */
+
 	struct list_head gc_full_list;	/* Full lines ready to GC, no valid */
 	struct list_head gc_empty_list;	/* Full lines close, all valid */
 
@@ -890,6 +904,9 @@ void pblk_rl_free_lines_dec(struct pblk_rl *rl, struct pblk_line *line,
 			    bool used);
 int pblk_rl_is_limit(struct pblk_rl *rl);
 
+void pblk_rl_werr_line_in(struct pblk_rl *rl);
+void pblk_rl_werr_line_out(struct pblk_rl *rl);
+
 /*
  * pblk sysfs
  */
-- 
2.11.0

^ permalink raw reply related	[flat|nested] 12+ messages in thread
* [GIT PULL 03/10] lightnvm: pblk: fix smeta write error path
  2018-06-01 14:41 [GIT PULL v3 00/10] lightnvm updates for 4.18 Matias Bjørling
  2018-06-01 14:41 ` [GIT PULL 01/10] lightnvm: pblk: rework write error recovery path Matias Bjørling
  2018-06-01 14:41 ` [GIT PULL 02/10] lightnvm: pblk: garbage collect lines with failed writes Matias Bjørling
@ 2018-06-01 14:41 ` Matias Bjørling
  2018-06-01 14:41 ` [GIT PULL 04/10] lightnvm: proper error handling for pblk_bio_add_pages Matias Bjørling
                   ` (7 subsequent siblings)
  10 siblings, 0 replies; 12+ messages in thread
From: Matias Bjørling @ 2018-06-01 14:41 UTC (permalink / raw)
  To: axboe
  Cc: linux-block, linux-kernel, javier, igor.j.konopko,
	marcin.dziegielewski, Hans Holmberg, Matias Bjørling

From: Hans Holmberg <hans.holmberg@cnexlabs.com>

Smeta write errors were previously ignored. Skip these
lines instead and throw them back on the free
list, so the chunks will go through a reset cycle
before we attempt to use the line again.

Signed-off-by: Hans Holmberg <hans.holmberg@cnexlabs.com>
Reviewed-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk-core.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/drivers/lightnvm/pblk-core.c b/drivers/lightnvm/pblk-core.c
index e81a70cf9e9b..549341b24e3c 100644
--- a/drivers/lightnvm/pblk-core.c
+++ b/drivers/lightnvm/pblk-core.c
@@ -849,9 +849,10 @@ static int pblk_line_submit_smeta_io(struct pblk *pblk, struct pblk_line *line,
 	atomic_dec(&pblk->inflight_io);
 
 	if (rqd.error) {
-		if (dir == PBLK_WRITE)
+		if (dir == PBLK_WRITE) {
 			pblk_log_write_err(pblk, &rqd);
-		else if (dir == PBLK_READ)
+			ret = 1;
+		} else if (dir == PBLK_READ)
 			pblk_log_read_err(pblk, &rqd);
 	}
 
@@ -1101,7 +1102,7 @@ static int pblk_line_init_bb(struct pblk *pblk, struct pblk_line *line,
 
 	if (init && pblk_line_submit_smeta_io(pblk, line, off, PBLK_WRITE)) {
 		pr_debug("pblk: line smeta I/O failed. Retry\n");
-		return 1;
+		return 0;
 	}
 
 	bitmap_copy(line->invalid_bitmap, line->map_bitmap, lm->sec_per_line);
-- 
2.11.0

^ permalink raw reply related	[flat|nested] 12+ messages in thread
* [GIT PULL 04/10] lightnvm: proper error handling for pblk_bio_add_pages
  2018-06-01 14:41 [GIT PULL v3 00/10] lightnvm updates for 4.18 Matias Bjørling
                   ` (2 preceding siblings ...)
  2018-06-01 14:41 ` [GIT PULL 03/10] lightnvm: pblk: fix smeta write error path Matias Bjørling
@ 2018-06-01 14:41 ` Matias Bjørling
  2018-06-01 14:41 ` [GIT PULL 05/10] lightnvm: fix partial read error path Matias Bjørling
                   ` (6 subsequent siblings)
  10 siblings, 0 replies; 12+ messages in thread
From: Matias Bjørling @ 2018-06-01 14:41 UTC (permalink / raw)
  To: axboe
  Cc: linux-block, linux-kernel, javier, igor.j.konopko,
	marcin.dziegielewski, Matias Bjørling

From: Igor Konopko <igor.j.konopko@intel.com>

Currently in case of error caused by bio_pc_add_page in
pblk_bio_add_pages two issues occur when calling from
pblk_rb_read_to_bio(). First one is in pblk_bio_free_pages, since we
are trying to free pages not allocated from our mempool. Second one
is the warn from dma_pool_free, that we are trying to free NULL
pointer dma.

This commit fix both issues.

Signed-off-by: Igor Konopko <igor.j.konopko@intel.com>
Signed-off-by: Marcin Dziegielewski <marcin.dziegielewski@intel.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk-core.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/drivers/lightnvm/pblk-core.c b/drivers/lightnvm/pblk-core.c
index 549341b24e3c..99036befb571 100644
--- a/drivers/lightnvm/pblk-core.c
+++ b/drivers/lightnvm/pblk-core.c
@@ -278,7 +278,9 @@ void pblk_free_rqd(struct pblk *pblk, struct nvm_rq *rqd, int type)
 		return;
 	}
 
-	nvm_dev_dma_free(dev->parent, rqd->meta_list, rqd->dma_meta_list);
+	if (rqd->meta_list)
+		nvm_dev_dma_free(dev->parent, rqd->meta_list,
+				rqd->dma_meta_list);
 	mempool_free(rqd, pool);
 }
 
@@ -316,7 +318,7 @@ int pblk_bio_add_pages(struct pblk *pblk, struct bio *bio, gfp_t flags,
 
 	return 0;
 err:
-	pblk_bio_free_pages(pblk, bio, 0, i - 1);
+	pblk_bio_free_pages(pblk, bio, (bio->bi_vcnt - i), i);
 	return -1;
 }
 
-- 
2.11.0

^ permalink raw reply related	[flat|nested] 12+ messages in thread
* [GIT PULL 05/10] lightnvm: fix partial read error path
  2018-06-01 14:41 [GIT PULL v3 00/10] lightnvm updates for 4.18 Matias Bjørling
                   ` (3 preceding siblings ...)
  2018-06-01 14:41 ` [GIT PULL 04/10] lightnvm: proper error handling for pblk_bio_add_pages Matias Bjørling
@ 2018-06-01 14:41 ` Matias Bjørling
  2018-06-01 14:41 ` [GIT PULL 06/10] lightnvm: pblk: add possibility to set write buffer size manually Matias Bjørling
                   ` (5 subsequent siblings)
  10 siblings, 0 replies; 12+ messages in thread
From: Matias Bjørling @ 2018-06-01 14:41 UTC (permalink / raw)
  To: axboe
  Cc: linux-block, linux-kernel, javier, igor.j.konopko,
	marcin.dziegielewski, Matias Bjørling

From: Igor Konopko <igor.j.konopko@intel.com>

When error occurs during bio_add_page on partial read path, pblk
tries to free pages twice.

Signed-off-by: Igor Konopko <igor.j.konopko@intel.com>
Signed-off-by: Marcin Dziegielewski <marcin.dziegielewski@intel.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk-read.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/drivers/lightnvm/pblk-read.c b/drivers/lightnvm/pblk-read.c
index 38f89a84ce5e..f770fd113ffe 100644
--- a/drivers/lightnvm/pblk-read.c
+++ b/drivers/lightnvm/pblk-read.c
@@ -256,7 +256,7 @@ static int pblk_partial_read_bio(struct pblk *pblk, struct nvm_rq *rqd,
 	new_bio = bio_alloc(GFP_KERNEL, nr_holes);
 
 	if (pblk_bio_add_pages(pblk, new_bio, GFP_KERNEL, nr_holes))
-		goto err;
+		goto err_add_pages;
 
 	if (nr_holes != new_bio->bi_vcnt) {
 		pr_err("pblk: malformed bio\n");
@@ -347,10 +347,10 @@ static int pblk_partial_read_bio(struct pblk *pblk, struct nvm_rq *rqd,
 	return NVM_IO_OK;
 
 err:
-	pr_err("pblk: failed to perform partial read\n");
-
 	/* Free allocated pages in new bio */
-	pblk_bio_free_pages(pblk, bio, 0, new_bio->bi_vcnt);
+	pblk_bio_free_pages(pblk, new_bio, 0, new_bio->bi_vcnt);
+err_add_pages:
+	pr_err("pblk: failed to perform partial read\n");
 	__pblk_end_io_read(pblk, rqd, false);
 	return NVM_IO_ERR;
 }
-- 
2.11.0

^ permalink raw reply related	[flat|nested] 12+ messages in thread
* [GIT PULL 06/10] lightnvm: pblk: add possibility to set write buffer size manually
  2018-06-01 14:41 [GIT PULL v3 00/10] lightnvm updates for 4.18 Matias Bjørling
                   ` (4 preceding siblings ...)
  2018-06-01 14:41 ` [GIT PULL 05/10] lightnvm: fix partial read error path Matias Bjørling
@ 2018-06-01 14:41 ` Matias Bjørling
  2018-06-01 14:41 ` [GIT PULL 07/10] lightnvm: pblk: remove unnecessary bio_get/put Matias Bjørling
                   ` (4 subsequent siblings)
  10 siblings, 0 replies; 12+ messages in thread
From: Matias Bjørling @ 2018-06-01 14:41 UTC (permalink / raw)
  To: axboe
  Cc: linux-block, linux-kernel, javier, igor.j.konopko,
	marcin.dziegielewski, Matias Bjørling

From: Marcin Dziegielewski <marcin.dziegielewski@intel.com>

In some cases, users can want set write buffer size manually, e.g. to
adjust it to specific workload. This patch provides the possibility
to set write buffer size via module parameter feature.

Signed-off-by: Marcin Dziegielewski <marcin.dziegielewski@intel.com>
Signed-off-by: Igor Konopko <igor.j.konopko@intel.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk-init.c | 14 ++++++++++++--
 1 file changed, 12 insertions(+), 2 deletions(-)

diff --git a/drivers/lightnvm/pblk-init.c b/drivers/lightnvm/pblk-init.c
index 783887b891b7..ce561f5d48ce 100644
--- a/drivers/lightnvm/pblk-init.c
+++ b/drivers/lightnvm/pblk-init.c
@@ -20,6 +20,11 @@
 
 #include "pblk.h"
 
+unsigned int write_buffer_size;
+
+module_param(write_buffer_size, uint, 0644);
+MODULE_PARM_DESC(write_buffer_size, "number of entries in a write buffer");
+
 static struct kmem_cache *pblk_ws_cache, *pblk_rec_cache, *pblk_g_rq_cache,
 				*pblk_w_rq_cache;
 static DECLARE_RWSEM(pblk_lock);
@@ -172,10 +177,15 @@ static int pblk_rwb_init(struct pblk *pblk)
 	struct nvm_tgt_dev *dev = pblk->dev;
 	struct nvm_geo *geo = &dev->geo;
 	struct pblk_rb_entry *entries;
-	unsigned long nr_entries;
+	unsigned long nr_entries, buffer_size;
 	unsigned int power_size, power_seg_sz;
 
-	nr_entries = pblk_rb_calculate_size(pblk->pgs_in_buffer);
+	if (write_buffer_size && (write_buffer_size > pblk->pgs_in_buffer))
+		buffer_size = write_buffer_size;
+	else
+		buffer_size = pblk->pgs_in_buffer;
+
+	nr_entries = pblk_rb_calculate_size(buffer_size);
 
 	entries = vzalloc(nr_entries * sizeof(struct pblk_rb_entry));
 	if (!entries)
-- 
2.11.0

^ permalink raw reply related	[flat|nested] 12+ messages in thread
* [GIT PULL 07/10] lightnvm: pblk: remove unnecessary bio_get/put
  2018-06-01 14:41 [GIT PULL v3 00/10] lightnvm updates for 4.18 Matias Bjørling
                   ` (5 preceding siblings ...)
  2018-06-01 14:41 ` [GIT PULL 06/10] lightnvm: pblk: add possibility to set write buffer size manually Matias Bjørling
@ 2018-06-01 14:41 ` Matias Bjørling
  2018-06-01 14:41 ` [GIT PULL 08/10] lightnvm: pblk: only try to recover lines with written smeta Matias Bjørling
                   ` (3 subsequent siblings)
  10 siblings, 0 replies; 12+ messages in thread
From: Matias Bjørling @ 2018-06-01 14:41 UTC (permalink / raw)
  To: axboe
  Cc: linux-block, linux-kernel, javier, igor.j.konopko,
	marcin.dziegielewski, Javier González, Matias Bjørling

From: Javier González <javier@javigon.com>

In the read path, pblk gets a reference to the incoming bio and puts it
after ending the bio. Though this behavior is correct, it is unnecessary
since pblk is the one putting the bio, therefore, it cannot disappear
underneath it.

Removing this reference, allows to clean up rqd->bio and avoids pointer
bouncing for the different read paths. Now, the incoming bio always
resides in the read context and pblk's internal bios (if any) reside in
rqd->bio.

Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk-read.c | 65 +++++++++++++++++++-------------------------
 1 file changed, 28 insertions(+), 37 deletions(-)

diff --git a/drivers/lightnvm/pblk-read.c b/drivers/lightnvm/pblk-read.c
index f770fd113ffe..18694694e5f0 100644
--- a/drivers/lightnvm/pblk-read.c
+++ b/drivers/lightnvm/pblk-read.c
@@ -39,10 +39,10 @@ static int pblk_read_from_cache(struct pblk *pblk, struct bio *bio,
 }
 
 static void pblk_read_ppalist_rq(struct pblk *pblk, struct nvm_rq *rqd,
-				 sector_t blba, unsigned long *read_bitmap)
+				 struct bio *bio, sector_t blba,
+				 unsigned long *read_bitmap)
 {
 	struct pblk_sec_meta *meta_list = rqd->meta_list;
-	struct bio *bio = rqd->bio;
 	struct ppa_addr ppas[PBLK_MAX_REQ_ADDRS];
 	int nr_secs = rqd->nr_ppas;
 	bool advanced_bio = false;
@@ -189,7 +189,6 @@ static void pblk_end_user_read(struct bio *bio)
 	WARN_ONCE(bio->bi_status, "pblk: corrupted read bio\n");
 #endif
 	bio_endio(bio);
-	bio_put(bio);
 }
 
 static void __pblk_end_io_read(struct pblk *pblk, struct nvm_rq *rqd,
@@ -197,23 +196,18 @@ static void __pblk_end_io_read(struct pblk *pblk, struct nvm_rq *rqd,
 {
 	struct nvm_tgt_dev *dev = pblk->dev;
 	struct pblk_g_ctx *r_ctx = nvm_rq_to_pdu(rqd);
-	struct bio *bio = rqd->bio;
+	struct bio *int_bio = rqd->bio;
 	unsigned long start_time = r_ctx->start_time;
 
 	generic_end_io_acct(dev->q, READ, &pblk->disk->part0, start_time);
 
 	if (rqd->error)
 		pblk_log_read_err(pblk, rqd);
-#ifdef CONFIG_NVM_DEBUG
-	else
-		WARN_ONCE(bio->bi_status, "pblk: corrupted read error\n");
-#endif
 
 	pblk_read_check_seq(pblk, rqd, r_ctx->lba);
 
-	bio_put(bio);
-	if (r_ctx->private)
-		pblk_end_user_read((struct bio *)r_ctx->private);
+	if (int_bio)
+		bio_put(int_bio);
 
 	if (put_line)
 		pblk_read_put_rqd_kref(pblk, rqd);
@@ -230,16 +224,19 @@ static void __pblk_end_io_read(struct pblk *pblk, struct nvm_rq *rqd,
 static void pblk_end_io_read(struct nvm_rq *rqd)
 {
 	struct pblk *pblk = rqd->private;
+	struct pblk_g_ctx *r_ctx = nvm_rq_to_pdu(rqd);
+	struct bio *bio = (struct bio *)r_ctx->private;
 
+	pblk_end_user_read(bio);
 	__pblk_end_io_read(pblk, rqd, true);
 }
 
-static int pblk_partial_read_bio(struct pblk *pblk, struct nvm_rq *rqd,
-				 unsigned int bio_init_idx,
-				 unsigned long *read_bitmap)
+static int pblk_partial_read(struct pblk *pblk, struct nvm_rq *rqd,
+			     struct bio *orig_bio, unsigned int bio_init_idx,
+			     unsigned long *read_bitmap)
 {
-	struct bio *new_bio, *bio = rqd->bio;
 	struct pblk_sec_meta *meta_list = rqd->meta_list;
+	struct bio *new_bio;
 	struct bio_vec src_bv, dst_bv;
 	void *ppa_ptr = NULL;
 	void *src_p, *dst_p;
@@ -256,11 +253,11 @@ static int pblk_partial_read_bio(struct pblk *pblk, struct nvm_rq *rqd,
 	new_bio = bio_alloc(GFP_KERNEL, nr_holes);
 
 	if (pblk_bio_add_pages(pblk, new_bio, GFP_KERNEL, nr_holes))
-		goto err_add_pages;
+		goto fail_add_pages;
 
 	if (nr_holes != new_bio->bi_vcnt) {
 		pr_err("pblk: malformed bio\n");
-		goto err;
+		goto fail;
 	}
 
 	for (i = 0; i < nr_secs; i++)
@@ -283,7 +280,7 @@ static int pblk_partial_read_bio(struct pblk *pblk, struct nvm_rq *rqd,
 	if (ret) {
 		bio_put(rqd->bio);
 		pr_err("pblk: sync read IO submission failed\n");
-		goto err;
+		goto fail;
 	}
 
 	if (rqd->error) {
@@ -319,7 +316,7 @@ static int pblk_partial_read_bio(struct pblk *pblk, struct nvm_rq *rqd,
 		meta_list[hole].lba = lba_list_media[i];
 
 		src_bv = new_bio->bi_io_vec[i++];
-		dst_bv = bio->bi_io_vec[bio_init_idx + hole];
+		dst_bv = orig_bio->bi_io_vec[bio_init_idx + hole];
 
 		src_p = kmap_atomic(src_bv.bv_page);
 		dst_p = kmap_atomic(dst_bv.bv_page);
@@ -338,28 +335,26 @@ static int pblk_partial_read_bio(struct pblk *pblk, struct nvm_rq *rqd,
 
 	bio_put(new_bio);
 
-	/* Complete the original bio and associated request */
-	bio_endio(bio);
-	rqd->bio = bio;
+	/* restore original request */
+	rqd->bio = NULL;
 	rqd->nr_ppas = nr_secs;
 
 	__pblk_end_io_read(pblk, rqd, false);
-	return NVM_IO_OK;
+	return NVM_IO_DONE;
 
-err:
+fail:
 	/* Free allocated pages in new bio */
 	pblk_bio_free_pages(pblk, new_bio, 0, new_bio->bi_vcnt);
-err_add_pages:
+fail_add_pages:
 	pr_err("pblk: failed to perform partial read\n");
 	__pblk_end_io_read(pblk, rqd, false);
 	return NVM_IO_ERR;
 }
 
-static void pblk_read_rq(struct pblk *pblk, struct nvm_rq *rqd,
+static void pblk_read_rq(struct pblk *pblk, struct nvm_rq *rqd, struct bio *bio,
 			 sector_t lba, unsigned long *read_bitmap)
 {
 	struct pblk_sec_meta *meta_list = rqd->meta_list;
-	struct bio *bio = rqd->bio;
 	struct ppa_addr ppa;
 
 	pblk_lookup_l2p_seq(pblk, &ppa, lba, 1);
@@ -423,14 +418,15 @@ int pblk_submit_read(struct pblk *pblk, struct bio *bio)
 	rqd = pblk_alloc_rqd(pblk, PBLK_READ);
 
 	rqd->opcode = NVM_OP_PREAD;
-	rqd->bio = bio;
 	rqd->nr_ppas = nr_secs;
+	rqd->bio = NULL; /* cloned bio if needed */
 	rqd->private = pblk;
 	rqd->end_io = pblk_end_io_read;
 
 	r_ctx = nvm_rq_to_pdu(rqd);
 	r_ctx->start_time = jiffies;
 	r_ctx->lba = blba;
+	r_ctx->private = bio; /* original bio */
 
 	/* Save the index for this bio's start. This is needed in case
 	 * we need to fill a partial read.
@@ -448,17 +444,15 @@ int pblk_submit_read(struct pblk *pblk, struct bio *bio)
 		rqd->ppa_list = rqd->meta_list + pblk_dma_meta_size;
 		rqd->dma_ppa_list = rqd->dma_meta_list + pblk_dma_meta_size;
 
-		pblk_read_ppalist_rq(pblk, rqd, blba, &read_bitmap);
+		pblk_read_ppalist_rq(pblk, rqd, bio, blba, &read_bitmap);
 	} else {
-		pblk_read_rq(pblk, rqd, blba, &read_bitmap);
+		pblk_read_rq(pblk, rqd, bio, blba, &read_bitmap);
 	}
 
-	bio_get(bio);
 	if (bitmap_full(&read_bitmap, nr_secs)) {
-		bio_endio(bio);
 		atomic_inc(&pblk->inflight_io);
 		__pblk_end_io_read(pblk, rqd, false);
-		return NVM_IO_OK;
+		return NVM_IO_DONE;
 	}
 
 	/* All sectors are to be read from the device */
@@ -473,13 +467,10 @@ int pblk_submit_read(struct pblk *pblk, struct bio *bio)
 		}
 
 		rqd->bio = int_bio;
-		r_ctx->private = bio;
 
 		if (pblk_submit_io(pblk, rqd)) {
 			pr_err("pblk: read IO submission failed\n");
 			ret = NVM_IO_ERR;
-			if (int_bio)
-				bio_put(int_bio);
 			goto fail_end_io;
 		}
 
@@ -489,7 +480,7 @@ int pblk_submit_read(struct pblk *pblk, struct bio *bio)
 	/* 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.
 	 */
-	return pblk_partial_read_bio(pblk, rqd, bio_init_idx, &read_bitmap);
+	return pblk_partial_read(pblk, rqd, bio, bio_init_idx, &read_bitmap);
 
 fail_rqd_free:
 	pblk_free_rqd(pblk, rqd, PBLK_READ);
-- 
2.11.0

^ permalink raw reply related	[flat|nested] 12+ messages in thread
* [GIT PULL 08/10] lightnvm: pblk: only try to recover lines with written smeta
  2018-06-01 14:41 [GIT PULL v3 00/10] lightnvm updates for 4.18 Matias Bjørling
                   ` (6 preceding siblings ...)
  2018-06-01 14:41 ` [GIT PULL 07/10] lightnvm: pblk: remove unnecessary bio_get/put Matias Bjørling
@ 2018-06-01 14:41 ` Matias Bjørling
  2018-06-01 14:41 ` [GIT PULL 09/10] lightnvm: pblk: kick writer on new flush points Matias Bjørling
                   ` (2 subsequent siblings)
  10 siblings, 0 replies; 12+ messages in thread
From: Matias Bjørling @ 2018-06-01 14:41 UTC (permalink / raw)
  To: axboe
  Cc: linux-block, linux-kernel, javier, igor.j.konopko,
	marcin.dziegielewski, Hans Holmberg, Matias Bjørling

From: Hans Holmberg <hans.holmberg@cnexlabs.com>

When switching between different lun configurations, there is no
guarantee that all lines that contain closed/open chunks have some
valid data to recover.

Check that the smeta chunk has been written to instead. Also
skip bad lines (that does not have enough good chunks).

Signed-off-by: Hans Holmberg <hans.holmberg@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk-recovery.c | 30 +++++++++++++++++++++---------
 1 file changed, 21 insertions(+), 9 deletions(-)

diff --git a/drivers/lightnvm/pblk-recovery.c b/drivers/lightnvm/pblk-recovery.c
index 788dce87043e..598342833d0d 100644
--- a/drivers/lightnvm/pblk-recovery.c
+++ b/drivers/lightnvm/pblk-recovery.c
@@ -774,18 +774,30 @@ static void pblk_recov_wa_counters(struct pblk *pblk,
 }
 
 static int pblk_line_was_written(struct pblk_line *line,
-			    struct pblk_line_meta *lm)
+			    struct pblk *pblk)
 {
 
-	int i;
-	int state_mask = NVM_CHK_ST_OFFLINE | NVM_CHK_ST_FREE;
+	struct pblk_line_meta *lm = &pblk->lm;
+	struct nvm_tgt_dev *dev = pblk->dev;
+	struct nvm_geo *geo = &dev->geo;
+	struct nvm_chk_meta *chunk;
+	struct ppa_addr bppa;
+	int smeta_blk;
 
-	for (i = 0; i < lm->blk_per_line; i++) {
-		if (!(line->chks[i].state & state_mask))
-			return 1;
-	}
+	if (line->state == PBLK_LINESTATE_BAD)
+		return 0;
 
-	return 0;
+	smeta_blk = find_first_zero_bit(line->blk_bitmap, lm->blk_per_line);
+	if (smeta_blk >= lm->blk_per_line)
+		return 0;
+
+	bppa = pblk->luns[smeta_blk].bppa;
+	chunk = &line->chks[pblk_ppa_to_pos(geo, bppa)];
+
+	if (chunk->state & NVM_CHK_ST_FREE)
+		return 0;
+
+	return 1;
 }
 
 struct pblk_line *pblk_recov_l2p(struct pblk *pblk)
@@ -824,7 +836,7 @@ struct pblk_line *pblk_recov_l2p(struct pblk *pblk)
 		line->lun_bitmap = ((void *)(smeta_buf)) +
 						sizeof(struct line_smeta);
 
-		if (!pblk_line_was_written(line, lm))
+		if (!pblk_line_was_written(line, pblk))
 			continue;
 
 		/* Lines that cannot be read are assumed as not written here */
-- 
2.11.0

^ permalink raw reply related	[flat|nested] 12+ messages in thread
* [GIT PULL 09/10] lightnvm: pblk: kick writer on new flush points
  2018-06-01 14:41 [GIT PULL v3 00/10] lightnvm updates for 4.18 Matias Bjørling
                   ` (7 preceding siblings ...)
  2018-06-01 14:41 ` [GIT PULL 08/10] lightnvm: pblk: only try to recover lines with written smeta Matias Bjørling
@ 2018-06-01 14:41 ` Matias Bjørling
  2018-06-01 14:41 ` [GIT PULL 10/10] lightnvm: pblk: take bitmap alloc. out of critical section Matias Bjørling
  2018-06-01 15:03 ` [GIT PULL v3 00/10] lightnvm updates for 4.18 Jens Axboe
  10 siblings, 0 replies; 12+ messages in thread
From: Matias Bjørling @ 2018-06-01 14:41 UTC (permalink / raw)
  To: axboe
  Cc: linux-block, linux-kernel, javier, igor.j.konopko,
	marcin.dziegielewski, Hans Holmberg, Matias Bjørling

From: Hans Holmberg <hans.holmberg@cnexlabs.com>

Unless we kick the writer directly when setting a new flush point, the
user risks having to wait for up to one second (the default timeout for
the write thread to be kicked) for the IO to complete.

Signed-off-by: Hans Holmberg <hans.holmberg@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk-cache.c | 10 ++++++----
 drivers/lightnvm/pblk-core.c  |  2 +-
 drivers/lightnvm/pblk-rb.c    |  4 ++--
 drivers/lightnvm/pblk.h       |  1 +
 4 files changed, 10 insertions(+), 7 deletions(-)

diff --git a/drivers/lightnvm/pblk-cache.c b/drivers/lightnvm/pblk-cache.c
index 29a23111b31c..b1c6d7eb6115 100644
--- a/drivers/lightnvm/pblk-cache.c
+++ b/drivers/lightnvm/pblk-cache.c
@@ -44,13 +44,15 @@ int pblk_write_to_cache(struct pblk *pblk, struct bio *bio, unsigned long flags)
 		goto out;
 	}
 
-	if (unlikely(!bio_has_data(bio)))
-		goto out;
-
 	pblk_ppa_set_empty(&w_ctx.ppa);
 	w_ctx.flags = flags;
-	if (bio->bi_opf & REQ_PREFLUSH)
+	if (bio->bi_opf & REQ_PREFLUSH) {
 		w_ctx.flags |= PBLK_FLUSH_ENTRY;
+		pblk_write_kick(pblk);
+	}
+
+	if (unlikely(!bio_has_data(bio)))
+		goto out;
 
 	for (i = 0; i < nr_entries; i++) {
 		void *data = bio_data(bio);
diff --git a/drivers/lightnvm/pblk-core.c b/drivers/lightnvm/pblk-core.c
index 99036befb571..a5750534efed 100644
--- a/drivers/lightnvm/pblk-core.c
+++ b/drivers/lightnvm/pblk-core.c
@@ -322,7 +322,7 @@ int pblk_bio_add_pages(struct pblk *pblk, struct bio *bio, gfp_t flags,
 	return -1;
 }
 
-static void pblk_write_kick(struct pblk *pblk)
+void pblk_write_kick(struct pblk *pblk)
 {
 	wake_up_process(pblk->writer_ts);
 	mod_timer(&pblk->wtimer, jiffies + msecs_to_jiffies(1000));
diff --git a/drivers/lightnvm/pblk-rb.c b/drivers/lightnvm/pblk-rb.c
index 1b74ec51a4ad..00cd1f20a196 100644
--- a/drivers/lightnvm/pblk-rb.c
+++ b/drivers/lightnvm/pblk-rb.c
@@ -349,7 +349,7 @@ void pblk_rb_write_entry_gc(struct pblk_rb *rb, void *data,
 }
 
 static int pblk_rb_flush_point_set(struct pblk_rb *rb, struct bio *bio,
-				  unsigned int pos)
+				   unsigned int pos)
 {
 	struct pblk_rb_entry *entry;
 	unsigned int sync, flush_point;
@@ -419,7 +419,7 @@ void pblk_rb_flush(struct pblk_rb *rb)
 	if (pblk_rb_flush_point_set(rb, NULL, mem))
 		return;
 
-	pblk_write_should_kick(pblk);
+	pblk_write_kick(pblk);
 }
 
 static int pblk_rb_may_write_flush(struct pblk_rb *rb, unsigned int nr_entries,
diff --git a/drivers/lightnvm/pblk.h b/drivers/lightnvm/pblk.h
index 725bc1bfa849..34cc1d64a9d4 100644
--- a/drivers/lightnvm/pblk.h
+++ b/drivers/lightnvm/pblk.h
@@ -851,6 +851,7 @@ void pblk_map_rq(struct pblk *pblk, struct nvm_rq *rqd, unsigned int sentry,
 int pblk_write_ts(void *data);
 void pblk_write_timer_fn(struct timer_list *t);
 void pblk_write_should_kick(struct pblk *pblk);
+void pblk_write_kick(struct pblk *pblk);
 
 /*
  * pblk read path
-- 
2.11.0

^ permalink raw reply related	[flat|nested] 12+ messages in thread
* [GIT PULL 10/10] lightnvm: pblk: take bitmap alloc. out of critical section
  2018-06-01 14:41 [GIT PULL v3 00/10] lightnvm updates for 4.18 Matias Bjørling
                   ` (8 preceding siblings ...)
  2018-06-01 14:41 ` [GIT PULL 09/10] lightnvm: pblk: kick writer on new flush points Matias Bjørling
@ 2018-06-01 14:41 ` Matias Bjørling
  2018-06-01 15:03 ` [GIT PULL v3 00/10] lightnvm updates for 4.18 Jens Axboe
  10 siblings, 0 replies; 12+ messages in thread
From: Matias Bjørling @ 2018-06-01 14:41 UTC (permalink / raw)
  To: axboe
  Cc: linux-block, linux-kernel, javier, igor.j.konopko,
	marcin.dziegielewski, Javier González, Matias Bjørling

From: Javier González <javier@javigon.com>

pblk allocates line bitmaps within the line lock unnecessarily. In order
to take pressure out of the fast patch, allocate line bitmaps outside
of this lock and refactor accordingly.

Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk-core.c | 97 +++++++++++++++++++++++++-------------------
 1 file changed, 56 insertions(+), 41 deletions(-)

diff --git a/drivers/lightnvm/pblk-core.c b/drivers/lightnvm/pblk-core.c
index a5750534efed..ed9cc977c8b3 100644
--- a/drivers/lightnvm/pblk-core.c
+++ b/drivers/lightnvm/pblk-core.c
@@ -1067,6 +1067,25 @@ static int pblk_line_init_metadata(struct pblk *pblk, struct pblk_line *line,
 	return 1;
 }
 
+static int pblk_line_alloc_bitmaps(struct pblk *pblk, struct pblk_line *line)
+{
+	struct pblk_line_meta *lm = &pblk->lm;
+
+	line->map_bitmap = kzalloc(lm->sec_bitmap_len, GFP_KERNEL);
+	if (!line->map_bitmap)
+		return -ENOMEM;
+
+	/* will be initialized using bb info from map_bitmap */
+	line->invalid_bitmap = kmalloc(lm->sec_bitmap_len, GFP_KERNEL);
+	if (!line->invalid_bitmap) {
+		kfree(line->map_bitmap);
+		line->map_bitmap = NULL;
+		return -ENOMEM;
+	}
+
+	return 0;
+}
+
 /* For now lines are always assumed full lines. Thus, smeta former and current
  * lun bitmaps are omitted.
  */
@@ -1171,18 +1190,7 @@ static int pblk_line_prepare(struct pblk *pblk, struct pblk_line *line)
 {
 	struct pblk_line_meta *lm = &pblk->lm;
 	int blk_in_line = atomic_read(&line->blk_in_line);
-	int blk_to_erase, ret;
-
-	line->map_bitmap = kzalloc(lm->sec_bitmap_len, GFP_ATOMIC);
-	if (!line->map_bitmap)
-		return -ENOMEM;
-
-	/* will be initialized using bb info from map_bitmap */
-	line->invalid_bitmap = kmalloc(lm->sec_bitmap_len, GFP_ATOMIC);
-	if (!line->invalid_bitmap) {
-		ret = -ENOMEM;
-		goto fail_free_map_bitmap;
-	}
+	int blk_to_erase;
 
 	/* Bad blocks do not need to be erased */
 	bitmap_copy(line->erase_bitmap, line->blk_bitmap, lm->blk_per_line);
@@ -1200,15 +1208,15 @@ static int pblk_line_prepare(struct pblk *pblk, struct pblk_line *line)
 	}
 
 	if (blk_in_line < lm->min_blk_line) {
-		ret = -EAGAIN;
-		goto fail_free_invalid_bitmap;
+		spin_unlock(&line->lock);
+		return -EAGAIN;
 	}
 
 	if (line->state != PBLK_LINESTATE_FREE) {
 		WARN(1, "pblk: corrupted line %d, state %d\n",
 							line->id, line->state);
-		ret = -EINTR;
-		goto fail_free_invalid_bitmap;
+		spin_unlock(&line->lock);
+		return -EINTR;
 	}
 
 	line->state = PBLK_LINESTATE_OPEN;
@@ -1222,16 +1230,6 @@ static int pblk_line_prepare(struct pblk *pblk, struct pblk_line *line)
 	kref_init(&line->ref);
 
 	return 0;
-
-fail_free_invalid_bitmap:
-	spin_unlock(&line->lock);
-	kfree(line->invalid_bitmap);
-	line->invalid_bitmap = NULL;
-fail_free_map_bitmap:
-	kfree(line->map_bitmap);
-	line->map_bitmap = NULL;
-
-	return ret;
 }
 
 int pblk_line_recov_alloc(struct pblk *pblk, struct pblk_line *line)
@@ -1251,13 +1249,16 @@ int pblk_line_recov_alloc(struct pblk *pblk, struct pblk_line *line)
 	}
 	spin_unlock(&l_mg->free_lock);
 
-	pblk_rl_free_lines_dec(&pblk->rl, line, true);
+	ret = pblk_line_alloc_bitmaps(pblk, line);
+	if (ret)
+		return ret;
 
 	if (!pblk_line_init_bb(pblk, line, 0)) {
 		list_add(&line->list, &l_mg->free_list);
 		return -EINTR;
 	}
 
+	pblk_rl_free_lines_dec(&pblk->rl, line, true);
 	return 0;
 }
 
@@ -1269,6 +1270,24 @@ void pblk_line_recov_close(struct pblk *pblk, struct pblk_line *line)
 	line->emeta = NULL;
 }
 
+static void pblk_line_reinit(struct pblk_line *line)
+{
+	*line->vsc = cpu_to_le32(EMPTY_ENTRY);
+
+	line->map_bitmap = NULL;
+	line->invalid_bitmap = NULL;
+	line->smeta = NULL;
+	line->emeta = NULL;
+}
+
+void pblk_line_free(struct pblk_line *line)
+{
+	kfree(line->map_bitmap);
+	kfree(line->invalid_bitmap);
+
+	pblk_line_reinit(line);
+}
+
 struct pblk_line *pblk_line_get(struct pblk *pblk)
 {
 	struct pblk_line_mgmt *l_mg = &pblk->l_mg;
@@ -1335,11 +1354,14 @@ static struct pblk_line *pblk_line_retry(struct pblk *pblk,
 		return NULL;
 	}
 
+	retry_line->map_bitmap = line->map_bitmap;
+	retry_line->invalid_bitmap = line->invalid_bitmap;
 	retry_line->smeta = line->smeta;
 	retry_line->emeta = line->emeta;
 	retry_line->meta_line = line->meta_line;
 
-	pblk_line_free(line);
+	pblk_line_reinit(line);
+
 	l_mg->data_line = retry_line;
 	spin_unlock(&l_mg->free_lock);
 
@@ -1392,6 +1414,9 @@ struct pblk_line *pblk_line_get_first_data(struct pblk *pblk)
 	}
 	spin_unlock(&l_mg->free_lock);
 
+	if (pblk_line_alloc_bitmaps(pblk, line))
+		return NULL;
+
 	if (pblk_line_erase(pblk, line)) {
 		line = pblk_line_retry(pblk, line);
 		if (!line)
@@ -1536,6 +1561,9 @@ struct pblk_line *pblk_line_replace_data(struct pblk *pblk)
 		goto retry_erase;
 	}
 
+	if (pblk_line_alloc_bitmaps(pblk, new))
+		return NULL;
+
 retry_setup:
 	if (!pblk_line_init_metadata(pblk, new, cur)) {
 		new = pblk_line_retry(pblk, new);
@@ -1575,19 +1603,6 @@ struct pblk_line *pblk_line_replace_data(struct pblk *pblk)
 	return new;
 }
 
-void pblk_line_free(struct pblk_line *line)
-{
-	kfree(line->map_bitmap);
-	kfree(line->invalid_bitmap);
-
-	*line->vsc = cpu_to_le32(EMPTY_ENTRY);
-
-	line->map_bitmap = NULL;
-	line->invalid_bitmap = NULL;
-	line->smeta = NULL;
-	line->emeta = NULL;
-}
-
 static void __pblk_line_put(struct pblk *pblk, struct pblk_line *line)
 {
 	struct pblk_line_mgmt *l_mg = &pblk->l_mg;
-- 
2.11.0

^ permalink raw reply related	[flat|nested] 12+ messages in thread
* Re: [GIT PULL v3 00/10] lightnvm updates for 4.18
  2018-06-01 14:41 [GIT PULL v3 00/10] lightnvm updates for 4.18 Matias Bjørling
                   ` (9 preceding siblings ...)
  2018-06-01 14:41 ` [GIT PULL 10/10] lightnvm: pblk: take bitmap alloc. out of critical section Matias Bjørling
@ 2018-06-01 15:03 ` Jens Axboe
  10 siblings, 0 replies; 12+ messages in thread
From: Jens Axboe @ 2018-06-01 15:03 UTC (permalink / raw)
  To: Matias Bjørling
  Cc: linux-block, linux-kernel, javier, igor.j.konopko, marcin.dziegielewski

On 6/1/18 8:41 AM, Matias Bjørling wrote:
> Hi Jens,
> 
> Please pick up the following patches.
> 
> Changes since v2:
> 
>  Rebased write error recovery patch on top of latest
>  for-4.18/block
> 
>  Added (missed the first round)
>   lightnvm: pblk: only try to recover lines with written smeta
>   lightnvm: pblk: kick writer on new flush points
>   lightnvm: pblk: take bitmap alloc. out of critical section
> 
> Changes since v1:
> 
>  Removed
>   lightnvm: pblk: sync RB and RL states during GC
>   lightnvm: error handling when whole line is bad
>   lightnvm: pblk: handle case when mw_cunits equals to 0
> 
>  Added
>   lightnvm: pblk: remove unnecessary bio_get/put
> 
> Thanks!
> 
> Hans Holmberg (5):
>   lightnvm: pblk: rework write error recovery path
>   lightnvm: pblk: garbage collect lines with failed writes
>   lightnvm: pblk: fix smeta write error path
>   lightnvm: pblk: only try to recover lines with written smeta
>   lightnvm: pblk: kick writer on new flush points
> 
> Igor Konopko (2):
>   lightnvm: proper error handling for pblk_bio_add_pages
>   lightnvm: fix partial read error path
> 
> Javier González (2):
>   lightnvm: pblk: remove unnecessary bio_get/put
>   lightnvm: pblk: take bitmap alloc. out of critical section
> 
> Marcin Dziegielewski (1):
>   lightnvm: pblk: add possibility to set write buffer size manually
> 
>  drivers/lightnvm/pblk-cache.c    |  10 +-
>  drivers/lightnvm/pblk-core.c     | 157 +++++++++++++++-------
>  drivers/lightnvm/pblk-gc.c       | 102 ++++++++------
>  drivers/lightnvm/pblk-init.c     |  62 ++++++---
>  drivers/lightnvm/pblk-rb.c       |  43 +-----
>  drivers/lightnvm/pblk-read.c     |  69 +++++-----
>  drivers/lightnvm/pblk-recovery.c | 121 +++--------------
>  drivers/lightnvm/pblk-rl.c       |  29 +++-
>  drivers/lightnvm/pblk-sysfs.c    |  15 ++-
>  drivers/lightnvm/pblk-write.c    | 281 +++++++++++++++++++++++++--------------
>  drivers/lightnvm/pblk.h          |  37 ++++--
>  11 files changed, 524 insertions(+), 402 deletions(-)

Applied, thanks.

-- 
Jens Axboe

^ permalink raw reply	[flat|nested] 12+ messages in thread
end of thread, other threads:[~2018-06-01 15:03 UTC | newest]

Thread overview: 12+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2018-06-01 14:41 [GIT PULL v3 00/10] lightnvm updates for 4.18 Matias Bjørling
2018-06-01 14:41 ` [GIT PULL 01/10] lightnvm: pblk: rework write error recovery path Matias Bjørling
2018-06-01 14:41 ` [GIT PULL 02/10] lightnvm: pblk: garbage collect lines with failed writes Matias Bjørling
2018-06-01 14:41 ` [GIT PULL 03/10] lightnvm: pblk: fix smeta write error path Matias Bjørling
2018-06-01 14:41 ` [GIT PULL 04/10] lightnvm: proper error handling for pblk_bio_add_pages Matias Bjørling
2018-06-01 14:41 ` [GIT PULL 05/10] lightnvm: fix partial read error path Matias Bjørling
2018-06-01 14:41 ` [GIT PULL 06/10] lightnvm: pblk: add possibility to set write buffer size manually Matias Bjørling
2018-06-01 14:41 ` [GIT PULL 07/10] lightnvm: pblk: remove unnecessary bio_get/put Matias Bjørling
2018-06-01 14:41 ` [GIT PULL 08/10] lightnvm: pblk: only try to recover lines with written smeta Matias Bjørling
2018-06-01 14:41 ` [GIT PULL 09/10] lightnvm: pblk: kick writer on new flush points Matias Bjørling
2018-06-01 14:41 ` [GIT PULL 10/10] lightnvm: pblk: take bitmap alloc. out of critical section Matias Bjørling
2018-06-01 15:03 ` [GIT PULL v3 00/10] lightnvm updates for 4.18 Jens Axboe

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.