[GIT PULL 00/45] lightnvm updates for 4.20.

[GIT PULL 00/45] lightnvm updates for 4.20.

[Matias Bjørling]

Hi Jens,

Please pick up the following patches for 4.20.

The changes include:

 - pblk trace support from Hans.
 - SPDX license header update from Javier.
 - Tons of refactoring patches from Javier, and me to cleanly abstract
   the 1.2 and 2.0 specs behind a common core interface.
 - Enable pblk to use a common interface to retrieve chunk metadata.
   From me.
 - Bug fixes from Hans, Javier, Jia-ju, Wei, Zhoujie, and me.

Thanks!,
Matias

Hans Holmberg (11):
  lightnvm: introduce nvm_rq_to_ppa_list
  lightnvm: pblk: allocate line map bitmaps using a mempool
  lightnvm: pblk: remove unused parameters in pblk_up_rq
  lightnvm: pblk: fix up prints in pblk_read_check_rand
  lightnvm: pblk: fix write amplificiation calculation
  lightnvm: pblk: add trace events for chunk states
  lightnvm: pblk: add trace events for line state changes
  lightnvm: pblk: add trace events for pblk state changes
  lightnvm: pblk: add tracing for chunk resets
  lightnvm: pblk: stop recreating global caches
  lightnvm: pblk: fix mapping issue on failed writes

Javier González (21):
  lightnvm: pblk: fix race condition on metadata I/O
  lightnvm: pblk: add helpers for chunk addresses
  lightnvm: pblk: improve line helpers
  lightnvm: pblk: fix comment typo
  lightnvm: pblk: remove unused variable.
  lightnvm: pblk: guarantee emeta on line close
  lightnvm: move ppa transformations to core
  lightnvm: pblk: calculate line pad distance in helper
  lightnvm: use internal allocation for chunk log page
  lightnvm: pblk: encapsulate rqd dma allocations
  lightnvm: pblk: refactor metadata paths
  lightnvm: pblk: take write semaphore on metadata
  lightnvm: pblk: recover open lines on 2.0 devices
  lightnvm: pblk: add SPDX license tag
  lightnvm: pblk: fix race on sysfs line state
  lightnvm: pblk: remove unused function
  lightnvm: pblk: encapsulate rb pointer operations
  lightnvm: pblk: move ring buffer alloc/free rb init
  lightnvm: pblk: guarantee mw_cunits on read buffer
  lightnvm: do no update csecs and sos on 1.2
  lightnvm: pblk: guarantee that backpointer is respected on writer
    stall

Jia-Ju Bai (1):
  lightnvm: pblk: fix two sleep-in-atomic-context bugs

Matias Bjørling (10):
  lightnvm: remove dependencies on BLK_DEV_NVME and PCI
  lightnvm: combine 1.2 and 2.0 command flags
  lightnvm: pblk: fix rqd.error return value in pblk_blk_erase_sync
  lightnvm: move device L2P detection to core
  lightnvm: move bad block and chunk state logic to core
  lightnvm: pblk: unify vector max req constants
  lightnvm: pblk: fix incorrect min_write_pgs
  lightnvm: pblk: remove size and out of bounds read check
  lightnvm: pblk: refactor put line fn on read completion
  lightnvm: pblk: remove debug from pblk_[down/up]_page

Wei Yongjun (1):
  lightnvm: pblk: fix error handling of pblk_lines_init()

Zhoujie Wu (1):
  lightnvm: pblk: consider max hw sectors supported for max_write_pgs

 drivers/lightnvm/Kconfig         |   3 +-
 drivers/lightnvm/core.c          | 334 ++++++++++++++----
 drivers/lightnvm/pblk-cache.c    |   1 +
 drivers/lightnvm/pblk-core.c     | 589 +++++++++++++++++--------------
 drivers/lightnvm/pblk-gc.c       |  11 +-
 drivers/lightnvm/pblk-init.c     | 321 +++++++----------
 drivers/lightnvm/pblk-map.c      |  13 +-
 drivers/lightnvm/pblk-rb.c       | 110 +++---
 drivers/lightnvm/pblk-read.c     |  86 ++---
 drivers/lightnvm/pblk-recovery.c | 477 +++++++++----------------
 drivers/lightnvm/pblk-rl.c       |   5 +-
 drivers/lightnvm/pblk-sysfs.c    |  12 +-
 drivers/lightnvm/pblk-trace.h    | 145 ++++++++
 drivers/lightnvm/pblk-write.c    |  90 ++---
 drivers/lightnvm/pblk.h          | 225 ++++--------
 drivers/nvme/host/lightnvm.c     |  32 +-
 include/linux/lightnvm.h         | 166 ++++++++-
 17 files changed, 1436 insertions(+), 1184 deletions(-)
 create mode 100644 drivers/lightnvm/pblk-trace.h

-- 
2.17.1

[GIT PULL 01/45] lightnvm: remove dependencies on BLK_DEV_NVME and PCI

[Matias Bjørling]

No need to force NVMe device driver to be compiled in if the
lightnvm subsystem is selected. Also no need for PCI to be selected
as well, as it would be selected by the device driver that hooks into
the subsystem.

Signed-off-by: Matias Bjørling <mb@lightnvm.io>
Reviewed-by: Javier González <javier@cnexlabs.com>
---
 drivers/lightnvm/Kconfig | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/drivers/lightnvm/Kconfig b/drivers/lightnvm/Kconfig
index 439bf90d084d..a872cd720967 100644
--- a/drivers/lightnvm/Kconfig
+++ b/drivers/lightnvm/Kconfig
@@ -4,8 +4,7 @@
 
 menuconfig NVM
 	bool "Open-Channel SSD target support"
-	depends on BLOCK && PCI
-	select BLK_DEV_NVME
+	depends on BLOCK
 	help
 	  Say Y here to get to enable Open-channel SSDs.
 
-- 
2.17.1

[GIT PULL 02/45] lightnvm: combine 1.2 and 2.0 command flags

[Matias Bjørling]

Add nvm_set_flags helper to enable core to appropriately
set the command flags for read/write/erase depending on which version
a drive supports.

The flags arguments can be distilled into the access hint,
scrambling, and program/erase suspend. Replace the access hint with
a "is_seq" parameter. The rest of the flags are dependent on the
command opcode, which is trivial to detect and set.

Signed-off-by: Matias Bjørling <mb@lightnvm.io>
Reviewed-by: Javier González <javier@cnexlabs.com>
---
 drivers/lightnvm/core.c          | 20 +++++++++++++++++
 drivers/lightnvm/pblk-core.c     | 13 ++++-------
 drivers/lightnvm/pblk-read.c     |  8 +------
 drivers/lightnvm/pblk-recovery.c | 14 ++++--------
 drivers/lightnvm/pblk-write.c    |  2 +-
 drivers/lightnvm/pblk.h          | 38 --------------------------------
 include/linux/lightnvm.h         |  2 ++
 7 files changed, 32 insertions(+), 65 deletions(-)

diff --git a/drivers/lightnvm/core.c b/drivers/lightnvm/core.c
index 60aa7bc5a630..68553c7ae937 100644
--- a/drivers/lightnvm/core.c
+++ b/drivers/lightnvm/core.c
@@ -752,6 +752,24 @@ int nvm_set_tgt_bb_tbl(struct nvm_tgt_dev *tgt_dev, struct ppa_addr *ppas,
 }
 EXPORT_SYMBOL(nvm_set_tgt_bb_tbl);
 
+static int nvm_set_flags(struct nvm_geo *geo, struct nvm_rq *rqd)
+{
+	int flags = 0;
+
+	if (geo->version == NVM_OCSSD_SPEC_20)
+		return 0;
+
+	if (rqd->is_seq)
+		flags |= geo->pln_mode >> 1;
+
+	if (rqd->opcode == NVM_OP_PREAD)
+		flags |= (NVM_IO_SCRAMBLE_ENABLE | NVM_IO_SUSPEND);
+	else if (rqd->opcode == NVM_OP_PWRITE)
+		flags |= NVM_IO_SCRAMBLE_ENABLE;
+
+	return flags;
+}
+
 int nvm_submit_io(struct nvm_tgt_dev *tgt_dev, struct nvm_rq *rqd)
 {
 	struct nvm_dev *dev = tgt_dev->parent;
@@ -763,6 +781,7 @@ int nvm_submit_io(struct nvm_tgt_dev *tgt_dev, struct nvm_rq *rqd)
 	nvm_rq_tgt_to_dev(tgt_dev, rqd);
 
 	rqd->dev = tgt_dev;
+	rqd->flags = nvm_set_flags(&tgt_dev->geo, rqd);
 
 	/* In case of error, fail with right address format */
 	ret = dev->ops->submit_io(dev, rqd);
@@ -783,6 +802,7 @@ int nvm_submit_io_sync(struct nvm_tgt_dev *tgt_dev, struct nvm_rq *rqd)
 	nvm_rq_tgt_to_dev(tgt_dev, rqd);
 
 	rqd->dev = tgt_dev;
+	rqd->flags = nvm_set_flags(&tgt_dev->geo, rqd);
 
 	/* In case of error, fail with right address format */
 	ret = dev->ops->submit_io_sync(dev, rqd);
diff --git a/drivers/lightnvm/pblk-core.c b/drivers/lightnvm/pblk-core.c
index 00984b486fea..72acf2f6dbd6 100644
--- a/drivers/lightnvm/pblk-core.c
+++ b/drivers/lightnvm/pblk-core.c
@@ -688,7 +688,7 @@ static int pblk_line_submit_emeta_io(struct pblk *pblk, struct pblk_line *line,
 	if (dir == PBLK_WRITE) {
 		struct pblk_sec_meta *meta_list = rqd.meta_list;
 
-		rqd.flags = pblk_set_progr_mode(pblk, PBLK_WRITE);
+		rqd.is_seq = 1;
 		for (i = 0; i < rqd.nr_ppas; ) {
 			spin_lock(&line->lock);
 			paddr = __pblk_alloc_page(pblk, line, min);
@@ -703,11 +703,9 @@ static int pblk_line_submit_emeta_io(struct pblk *pblk, struct pblk_line *line,
 		for (i = 0; i < rqd.nr_ppas; ) {
 			struct ppa_addr ppa = addr_to_gen_ppa(pblk, paddr, id);
 			int pos = pblk_ppa_to_pos(geo, ppa);
-			int read_type = PBLK_READ_RANDOM;
 
 			if (pblk_io_aligned(pblk, rq_ppas))
-				read_type = PBLK_READ_SEQUENTIAL;
-			rqd.flags = pblk_set_read_mode(pblk, read_type);
+				rqd.is_seq = 1;
 
 			while (test_bit(pos, line->blk_bitmap)) {
 				paddr += min;
@@ -787,17 +785,14 @@ static int pblk_line_submit_smeta_io(struct pblk *pblk, struct pblk_line *line,
 	__le64 *lba_list = NULL;
 	int i, ret;
 	int cmd_op, bio_op;
-	int flags;
 
 	if (dir == PBLK_WRITE) {
 		bio_op = REQ_OP_WRITE;
 		cmd_op = NVM_OP_PWRITE;
-		flags = pblk_set_progr_mode(pblk, PBLK_WRITE);
 		lba_list = emeta_to_lbas(pblk, line->emeta->buf);
 	} else if (dir == PBLK_READ_RECOV || dir == PBLK_READ) {
 		bio_op = REQ_OP_READ;
 		cmd_op = NVM_OP_PREAD;
-		flags = pblk_set_read_mode(pblk, PBLK_READ_SEQUENTIAL);
 	} else
 		return -EINVAL;
 
@@ -822,7 +817,7 @@ static int pblk_line_submit_smeta_io(struct pblk *pblk, struct pblk_line *line,
 
 	rqd.bio = bio;
 	rqd.opcode = cmd_op;
-	rqd.flags = flags;
+	rqd.is_seq = 1;
 	rqd.nr_ppas = lm->smeta_sec;
 
 	for (i = 0; i < lm->smeta_sec; i++, paddr++) {
@@ -885,7 +880,7 @@ static void pblk_setup_e_rq(struct pblk *pblk, struct nvm_rq *rqd,
 	rqd->opcode = NVM_OP_ERASE;
 	rqd->ppa_addr = ppa;
 	rqd->nr_ppas = 1;
-	rqd->flags = pblk_set_progr_mode(pblk, PBLK_ERASE);
+	rqd->is_seq = 1;
 	rqd->bio = NULL;
 }
 
diff --git a/drivers/lightnvm/pblk-read.c b/drivers/lightnvm/pblk-read.c
index 5a46d7f9302f..07b7f82c7595 100644
--- a/drivers/lightnvm/pblk-read.c
+++ b/drivers/lightnvm/pblk-read.c
@@ -93,9 +93,7 @@ static void pblk_read_ppalist_rq(struct pblk *pblk, struct nvm_rq *rqd,
 	}
 
 	if (pblk_io_aligned(pblk, nr_secs))
-		rqd->flags = pblk_set_read_mode(pblk, PBLK_READ_SEQUENTIAL);
-	else
-		rqd->flags = pblk_set_read_mode(pblk, PBLK_READ_RANDOM);
+		rqd->is_seq = 1;
 
 #ifdef CONFIG_NVM_PBLK_DEBUG
 	atomic_long_add(nr_secs, &pblk->inflight_reads);
@@ -344,7 +342,6 @@ static int pblk_setup_partial_read(struct pblk *pblk, struct nvm_rq *rqd,
 
 	rqd->bio = new_bio;
 	rqd->nr_ppas = nr_holes;
-	rqd->flags = pblk_set_read_mode(pblk, PBLK_READ_RANDOM);
 
 	pr_ctx->ppa_ptr = NULL;
 	pr_ctx->orig_bio = bio;
@@ -438,8 +435,6 @@ static void pblk_read_rq(struct pblk *pblk, struct nvm_rq *rqd, struct bio *bio,
 	} else {
 		rqd->ppa_addr = ppa;
 	}
-
-	rqd->flags = pblk_set_read_mode(pblk, PBLK_READ_RANDOM);
 }
 
 int pblk_submit_read(struct pblk *pblk, struct bio *bio)
@@ -663,7 +658,6 @@ int pblk_submit_read_gc(struct pblk *pblk, struct pblk_gc_rq *gc_rq)
 
 	rqd.opcode = NVM_OP_PREAD;
 	rqd.nr_ppas = gc_rq->secs_to_gc;
-	rqd.flags = pblk_set_read_mode(pblk, PBLK_READ_RANDOM);
 	rqd.bio = bio;
 
 	if (pblk_submit_io_sync(pblk, &rqd)) {
diff --git a/drivers/lightnvm/pblk-recovery.c b/drivers/lightnvm/pblk-recovery.c
index e232e47e1353..cf629ab016ba 100644
--- a/drivers/lightnvm/pblk-recovery.c
+++ b/drivers/lightnvm/pblk-recovery.c
@@ -159,9 +159,7 @@ static int pblk_recov_read_oob(struct pblk *pblk, struct pblk_line *line,
 	rqd->dma_meta_list = dma_meta_list;
 
 	if (pblk_io_aligned(pblk, rq_ppas))
-		rqd->flags = pblk_set_read_mode(pblk, PBLK_READ_SEQUENTIAL);
-	else
-		rqd->flags = pblk_set_read_mode(pblk, PBLK_READ_RANDOM);
+		rqd->is_seq = 1;
 
 	for (i = 0; i < rqd->nr_ppas; ) {
 		struct ppa_addr ppa;
@@ -302,7 +300,7 @@ static int pblk_recov_pad_oob(struct pblk *pblk, struct pblk_line *line,
 
 	rqd->bio = bio;
 	rqd->opcode = NVM_OP_PWRITE;
-	rqd->flags = pblk_set_progr_mode(pblk, PBLK_WRITE);
+	rqd->is_seq = 1;
 	rqd->meta_list = meta_list;
 	rqd->nr_ppas = rq_ppas;
 	rqd->ppa_list = ppa_list;
@@ -436,9 +434,7 @@ static int pblk_recov_scan_all_oob(struct pblk *pblk, struct pblk_line *line,
 	rqd->dma_meta_list = dma_meta_list;
 
 	if (pblk_io_aligned(pblk, rq_ppas))
-		rqd->flags = pblk_set_read_mode(pblk, PBLK_READ_SEQUENTIAL);
-	else
-		rqd->flags = pblk_set_read_mode(pblk, PBLK_READ_RANDOM);
+		rqd->is_seq = 1;
 
 	for (i = 0; i < rqd->nr_ppas; ) {
 		struct ppa_addr ppa;
@@ -567,9 +563,7 @@ static int pblk_recov_scan_oob(struct pblk *pblk, struct pblk_line *line,
 	rqd->dma_meta_list = dma_meta_list;
 
 	if (pblk_io_aligned(pblk, rq_ppas))
-		rqd->flags = pblk_set_read_mode(pblk, PBLK_READ_SEQUENTIAL);
-	else
-		rqd->flags = pblk_set_read_mode(pblk, PBLK_READ_RANDOM);
+		rqd->is_seq = 1;
 
 	for (i = 0; i < rqd->nr_ppas; ) {
 		struct ppa_addr ppa;
diff --git a/drivers/lightnvm/pblk-write.c b/drivers/lightnvm/pblk-write.c
index ee774a86cf1e..508c63701eda 100644
--- a/drivers/lightnvm/pblk-write.c
+++ b/drivers/lightnvm/pblk-write.c
@@ -302,7 +302,7 @@ static int pblk_alloc_w_rq(struct pblk *pblk, struct nvm_rq *rqd,
 	/* Setup write request */
 	rqd->opcode = NVM_OP_PWRITE;
 	rqd->nr_ppas = nr_secs;
-	rqd->flags = pblk_set_progr_mode(pblk, PBLK_WRITE);
+	rqd->is_seq = 1;
 	rqd->private = pblk;
 	rqd->end_io = end_io;
 
diff --git a/drivers/lightnvm/pblk.h b/drivers/lightnvm/pblk.h
index 4760af7b6499..48b3035df3c4 100644
--- a/drivers/lightnvm/pblk.h
+++ b/drivers/lightnvm/pblk.h
@@ -1255,44 +1255,6 @@ static inline u32 pblk_calc_emeta_crc(struct pblk *pblk,
 	return crc;
 }
 
-static inline int pblk_set_progr_mode(struct pblk *pblk, int type)
-{
-	struct nvm_tgt_dev *dev = pblk->dev;
-	struct nvm_geo *geo = &dev->geo;
-	int flags;
-
-	if (geo->version == NVM_OCSSD_SPEC_20)
-		return 0;
-
-	flags = geo->pln_mode >> 1;
-
-	if (type == PBLK_WRITE)
-		flags |= NVM_IO_SCRAMBLE_ENABLE;
-
-	return flags;
-}
-
-enum {
-	PBLK_READ_RANDOM	= 0,
-	PBLK_READ_SEQUENTIAL	= 1,
-};
-
-static inline int pblk_set_read_mode(struct pblk *pblk, int type)
-{
-	struct nvm_tgt_dev *dev = pblk->dev;
-	struct nvm_geo *geo = &dev->geo;
-	int flags;
-
-	if (geo->version == NVM_OCSSD_SPEC_20)
-		return 0;
-
-	flags = NVM_IO_SUSPEND | NVM_IO_SCRAMBLE_ENABLE;
-	if (type == PBLK_READ_SEQUENTIAL)
-		flags |= geo->pln_mode >> 1;
-
-	return flags;
-}
-
 static inline int pblk_io_aligned(struct pblk *pblk, int nr_secs)
 {
 	return !(nr_secs % pblk->min_write_pgs);
diff --git a/include/linux/lightnvm.h b/include/linux/lightnvm.h
index e9e0d1c7eaf5..8acc2fe277d6 100644
--- a/include/linux/lightnvm.h
+++ b/include/linux/lightnvm.h
@@ -305,6 +305,8 @@ struct nvm_rq {
 	u64 ppa_status; /* ppa media status */
 	int error;
 
+	int is_seq; /* Sequential hint flag. 1.2 only */
+
 	void *private;
 };
 
-- 
2.17.1

[GIT PULL 03/45] lightnvm: pblk: fix rqd.error return value in pblk_blk_erase_sync

[Matias Bjørling]

rqd.error is masked by the return value of pblk_submit_io_sync.
The rqd structure is then passed on to the end_io function, which
assumes that any error should lead to a chunk being marked
offline/bad. Since the pblk_submit_io_sync can fail before the
command is issued to the device, the error value maybe not correspond
to a media failure, leading to chunks being immaturely retired.

Also, the pblk_blk_erase_sync function prints an error message in case
the erase fails. Since the caller prints an error message by itself,
remove the error message in this function.

Signed-off-by: Matias Bjørling <mb@lightnvm.io>
Reviewed-by: Javier González <javier@cnexlabs.com>
Reviewed-by: Hans Holmberg <hans.holmberg@cnexlabs.com>
---
 drivers/lightnvm/pblk-core.c | 19 ++-----------------
 1 file changed, 2 insertions(+), 17 deletions(-)

diff --git a/drivers/lightnvm/pblk-core.c b/drivers/lightnvm/pblk-core.c
index 72acf2f6dbd6..72de7456845b 100644
--- a/drivers/lightnvm/pblk-core.c
+++ b/drivers/lightnvm/pblk-core.c
@@ -886,10 +886,8 @@ 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 = 0;
-
-	memset(&rqd, 0, sizeof(struct nvm_rq));
+	struct nvm_rq rqd = {NULL};
+	int ret;
 
 	pblk_setup_e_rq(pblk, &rqd, ppa);
 
@@ -897,19 +895,6 @@ static int pblk_blk_erase_sync(struct pblk *pblk, struct ppa_addr ppa)
 	 * with writes. Thus, there is no need to take the LUN semaphore.
 	 */
 	ret = pblk_submit_io_sync(pblk, &rqd);
-	if (ret) {
-		struct nvm_tgt_dev *dev = pblk->dev;
-		struct nvm_geo *geo = &dev->geo;
-
-		pblk_err(pblk, "could not sync erase line:%d,blk:%d\n",
-					pblk_ppa_to_line(ppa),
-					pblk_ppa_to_pos(geo, ppa));
-
-		rqd.error = ret;
-		goto out;
-	}
-
-out:
 	rqd.private = pblk;
 	__pblk_end_io_erase(pblk, &rqd);
 
-- 
2.17.1

[GIT PULL 04/45] lightnvm: move device L2P detection to core

[Matias Bjørling]

A 1.2 device is able to manage the logical to physical mapping
table internally or leave it to the host.

A target only supports one of those approaches, and therefore must
check on initialization. Move this check to core to avoid each target
implement the check.

Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/core.c      | 5 +++++
 drivers/lightnvm/pblk-init.c | 7 -------
 include/linux/lightnvm.h     | 6 ++++++
 3 files changed, 11 insertions(+), 7 deletions(-)

diff --git a/drivers/lightnvm/core.c b/drivers/lightnvm/core.c
index 68553c7ae937..964352720a03 100644
--- a/drivers/lightnvm/core.c
+++ b/drivers/lightnvm/core.c
@@ -355,6 +355,11 @@ static int nvm_create_tgt(struct nvm_dev *dev, struct nvm_ioctl_create *create)
 		return -EINVAL;
 	}
 
+	if ((tt->flags & NVM_TGT_F_HOST_L2P) != (dev->geo.dom & NVM_RSP_L2P)) {
+		pr_err("nvm: device is incompatible with target L2P type.\n");
+		return -EINVAL;
+	}
+
 	if (nvm_target_exists(create->tgtname)) {
 		pr_err("nvm: target name already exists (%s)\n",
 							create->tgtname);
diff --git a/drivers/lightnvm/pblk-init.c b/drivers/lightnvm/pblk-init.c
index 537e98f2b24a..039f62d05e84 100644
--- a/drivers/lightnvm/pblk-init.c
+++ b/drivers/lightnvm/pblk-init.c
@@ -1210,13 +1210,6 @@ static void *pblk_init(struct nvm_tgt_dev *dev, struct gendisk *tdisk,
 		return ERR_PTR(-EINVAL);
 	}
 
-	if (geo->version == NVM_OCSSD_SPEC_12 && geo->dom & NVM_RSP_L2P) {
-		pblk_err(pblk, "host-side L2P table not supported. (%x)\n",
-							geo->dom);
-		kfree(pblk);
-		return ERR_PTR(-EINVAL);
-	}
-
 	spin_lock_init(&pblk->resubmit_lock);
 	spin_lock_init(&pblk->trans_lock);
 	spin_lock_init(&pblk->lock);
diff --git a/include/linux/lightnvm.h b/include/linux/lightnvm.h
index 8acc2fe277d6..f4a84694e5e2 100644
--- a/include/linux/lightnvm.h
+++ b/include/linux/lightnvm.h
@@ -495,9 +495,15 @@ typedef void (nvm_tgt_exit_fn)(void *, bool);
 typedef int (nvm_tgt_sysfs_init_fn)(struct gendisk *);
 typedef void (nvm_tgt_sysfs_exit_fn)(struct gendisk *);
 
+enum {
+	NVM_TGT_F_DEV_L2P = 0,
+	NVM_TGT_F_HOST_L2P = 1 << 0,
+};
+
 struct nvm_tgt_type {
 	const char *name;
 	unsigned int version[3];
+	int flags;
 
 	/* target entry points */
 	nvm_tgt_make_rq_fn *make_rq;
-- 
2.17.1

[GIT PULL 05/45] lightnvm: pblk: fix race condition on metadata I/O

[Matias Bjørling]

From: Javier González <javier@javigon.com>

In pblk, when a new line is allocated, metadata for the previously
written line is scheduled. This is done through a fixed memory region
that is shared through time and contexts across different lines and
therefore protected by a lock. Unfortunately, this lock is not properly
covering all the metadata used for sharing this memory regions,
resulting in a race condition.

This patch fixes this race condition by protecting this metadata
properly.

Fixes: dd2a43437337 ("lightnvm: pblk: sched. metadata on write thread")
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk-write.c | 14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)

diff --git a/drivers/lightnvm/pblk-write.c b/drivers/lightnvm/pblk-write.c
index 508c63701eda..df99c45778d4 100644
--- a/drivers/lightnvm/pblk-write.c
+++ b/drivers/lightnvm/pblk-write.c
@@ -417,12 +417,11 @@ int pblk_submit_meta_io(struct pblk *pblk, struct pblk_line *meta_line)
 			rqd->ppa_list[i] = addr_to_gen_ppa(pblk, paddr, id);
 	}
 
+	spin_lock(&l_mg->close_lock);
 	emeta->mem += rq_len;
-	if (emeta->mem >= lm->emeta_len[0]) {
-		spin_lock(&l_mg->close_lock);
+	if (emeta->mem >= lm->emeta_len[0])
 		list_del(&meta_line->list);
-		spin_unlock(&l_mg->close_lock);
-	}
+	spin_unlock(&l_mg->close_lock);
 
 	pblk_down_page(pblk, rqd->ppa_list, rqd->nr_ppas);
 
@@ -491,14 +490,15 @@ static struct pblk_line *pblk_should_submit_meta_io(struct pblk *pblk,
 	struct pblk_line *meta_line;
 
 	spin_lock(&l_mg->close_lock);
-retry:
 	if (list_empty(&l_mg->emeta_list)) {
 		spin_unlock(&l_mg->close_lock);
 		return NULL;
 	}
 	meta_line = list_first_entry(&l_mg->emeta_list, struct pblk_line, list);
-	if (meta_line->emeta->mem >= lm->emeta_len[0])
-		goto retry;
+	if (meta_line->emeta->mem >= lm->emeta_len[0]) {
+		spin_unlock(&l_mg->close_lock);
+		return NULL;
+	}
 	spin_unlock(&l_mg->close_lock);
 
 	if (!pblk_valid_meta_ppa(pblk, meta_line, data_rqd))
-- 
2.17.1

[GIT PULL 06/45] lightnvm: move bad block and chunk state logic to core

[Matias Bjørling]

pblk implements two data paths for recovery line state. One for 1.2
and another for 2.0, instead of having pblk implement these, combine
them in the core to reduce complexity and make available to other
targets.

The new interface will adhere to the 2.0 chunk definition,
including managing open chunks with an active write pointer. To provide
this interface, a 1.2 device recovers the state of the chunks by
manually detecting if a chunk is either free/open/close/offline, and if
open, scanning the flash pages sequentially to find the next writeable
page. This process takes on average ~10 seconds on a device with 64 dies,
1024 blocks and 60us read access time. The process can be parallelized
but is left out for maintenance simplicity, as the 1.2 specification is
deprecated. For 2.0 devices, the logic is maintained internally in the
drive and retrieved through the 2.0 interface.

Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/core.c      | 309 ++++++++++++++++++++++++++++-------
 drivers/lightnvm/pblk-core.c |   6 +-
 drivers/lightnvm/pblk-init.c | 116 +------------
 drivers/lightnvm/pblk.h      |   2 +-
 drivers/nvme/host/lightnvm.c |   4 +-
 include/linux/lightnvm.h     |  15 +-
 6 files changed, 265 insertions(+), 187 deletions(-)

diff --git a/drivers/lightnvm/core.c b/drivers/lightnvm/core.c
index 964352720a03..8df188e0767e 100644
--- a/drivers/lightnvm/core.c
+++ b/drivers/lightnvm/core.c
@@ -717,46 +717,6 @@ static void nvm_free_rqd_ppalist(struct nvm_tgt_dev *tgt_dev,
 	nvm_dev_dma_free(tgt_dev->parent, rqd->ppa_list, rqd->dma_ppa_list);
 }
 
-int nvm_get_chunk_meta(struct nvm_tgt_dev *tgt_dev, struct nvm_chk_meta *meta,
-		struct ppa_addr ppa, int nchks)
-{
-	struct nvm_dev *dev = tgt_dev->parent;
-
-	nvm_ppa_tgt_to_dev(tgt_dev, &ppa, 1);
-
-	return dev->ops->get_chk_meta(tgt_dev->parent, meta,
-						(sector_t)ppa.ppa, nchks);
-}
-EXPORT_SYMBOL(nvm_get_chunk_meta);
-
-int nvm_set_tgt_bb_tbl(struct nvm_tgt_dev *tgt_dev, struct ppa_addr *ppas,
-		       int nr_ppas, int type)
-{
-	struct nvm_dev *dev = tgt_dev->parent;
-	struct nvm_rq rqd;
-	int ret;
-
-	if (nr_ppas > NVM_MAX_VLBA) {
-		pr_err("nvm: unable to update all blocks atomically\n");
-		return -EINVAL;
-	}
-
-	memset(&rqd, 0, sizeof(struct nvm_rq));
-
-	nvm_set_rqd_ppalist(tgt_dev, &rqd, ppas, nr_ppas);
-	nvm_rq_tgt_to_dev(tgt_dev, &rqd);
-
-	ret = dev->ops->set_bb_tbl(dev, &rqd.ppa_addr, rqd.nr_ppas, type);
-	nvm_free_rqd_ppalist(tgt_dev, &rqd);
-	if (ret) {
-		pr_err("nvm: failed bb mark\n");
-		return -EINVAL;
-	}
-
-	return 0;
-}
-EXPORT_SYMBOL(nvm_set_tgt_bb_tbl);
-
 static int nvm_set_flags(struct nvm_geo *geo, struct nvm_rq *rqd)
 {
 	int flags = 0;
@@ -830,27 +790,159 @@ void nvm_end_io(struct nvm_rq *rqd)
 }
 EXPORT_SYMBOL(nvm_end_io);
 
+static int nvm_submit_io_sync_raw(struct nvm_dev *dev, struct nvm_rq *rqd)
+{
+	if (!dev->ops->submit_io_sync)
+		return -ENODEV;
+
+	rqd->flags = nvm_set_flags(&dev->geo, rqd);
+
+	return dev->ops->submit_io_sync(dev, rqd);
+}
+
+static int nvm_bb_chunk_sense(struct nvm_dev *dev, struct ppa_addr ppa)
+{
+	struct nvm_rq rqd = { NULL };
+	struct bio bio;
+	struct bio_vec bio_vec;
+	struct page *page;
+	int ret;
+
+	page = alloc_page(GFP_KERNEL);
+	if (!page)
+		return -ENOMEM;
+
+	bio_init(&bio, &bio_vec, 1);
+	bio_add_page(&bio, page, PAGE_SIZE, 0);
+	bio_set_op_attrs(&bio, REQ_OP_READ, 0);
+
+	rqd.bio = &bio;
+	rqd.opcode = NVM_OP_PREAD;
+	rqd.is_seq = 1;
+	rqd.nr_ppas = 1;
+	rqd.ppa_addr = generic_to_dev_addr(dev, ppa);
+
+	ret = nvm_submit_io_sync_raw(dev, &rqd);
+	if (ret)
+		return ret;
+
+	__free_page(page);
+
+	return rqd.error;
+}
+
 /*
- * folds a bad block list from its plane representation to its virtual
- * block representation. The fold is done in place and reduced size is
- * returned.
+ * Scans a 1.2 chunk first and last page to determine if its state.
+ * If the chunk is found to be open, also scan it to update the write
+ * pointer.
+ */
+static int nvm_bb_chunk_scan(struct nvm_dev *dev, struct ppa_addr ppa,
+			     struct nvm_chk_meta *meta)
+{
+	struct nvm_geo *geo = &dev->geo;
+	int ret, pg, pl;
+
+	/* sense first page */
+	ret = nvm_bb_chunk_sense(dev, ppa);
+	if (ret < 0) /* io error */
+		return ret;
+	else if (ret == 0) /* valid data */
+		meta->state = NVM_CHK_ST_OPEN;
+	else if (ret > 0) {
+		/*
+		 * If empty page, the chunk is free, else it is an
+		 * actual io error. In that case, mark it offline.
+		 */
+		switch (ret) {
+		case NVM_RSP_ERR_EMPTYPAGE:
+			meta->state = NVM_CHK_ST_FREE;
+			return 0;
+		case NVM_RSP_ERR_FAILCRC:
+		case NVM_RSP_ERR_FAILECC:
+		case NVM_RSP_WARN_HIGHECC:
+			meta->state = NVM_CHK_ST_OPEN;
+			goto scan;
+		default:
+			return -ret; /* other io error */
+		}
+	}
+
+	/* sense last page */
+	ppa.g.pg = geo->num_pg - 1;
+	ppa.g.pl = geo->num_pln - 1;
+
+	ret = nvm_bb_chunk_sense(dev, ppa);
+	if (ret < 0) /* io error */
+		return ret;
+	else if (ret == 0) { /* Chunk fully written */
+		meta->state = NVM_CHK_ST_CLOSED;
+		meta->wp = geo->clba;
+		return 0;
+	} else if (ret > 0) {
+		switch (ret) {
+		case NVM_RSP_ERR_EMPTYPAGE:
+		case NVM_RSP_ERR_FAILCRC:
+		case NVM_RSP_ERR_FAILECC:
+		case NVM_RSP_WARN_HIGHECC:
+			meta->state = NVM_CHK_ST_OPEN;
+			break;
+		default:
+			return -ret; /* other io error */
+		}
+	}
+
+scan:
+	/*
+	 * chunk is open, we scan sequentially to update the write pointer.
+	 * We make the assumption that targets write data across all planes
+	 * before moving to the next page.
+	 */
+	for (pg = 0; pg < geo->num_pg; pg++) {
+		for (pl = 0; pl < geo->num_pln; pl++) {
+			ppa.g.pg = pg;
+			ppa.g.pl = pl;
+
+			ret = nvm_bb_chunk_sense(dev, ppa);
+			if (ret < 0) /* io error */
+				return ret;
+			else if (ret == 0) {
+				meta->wp += geo->ws_min;
+			} else if (ret > 0) {
+				switch (ret) {
+				case NVM_RSP_ERR_EMPTYPAGE:
+					return 0;
+				case NVM_RSP_ERR_FAILCRC:
+				case NVM_RSP_ERR_FAILECC:
+				case NVM_RSP_WARN_HIGHECC:
+					meta->wp += geo->ws_min;
+					break;
+				default:
+					return -ret; /* other io error */
+				}
+			}
+		}
+	}
+
+	return 0;
+}
+
+/*
+ * folds a bad block list from its plane representation to its
+ * chunk representation.
  *
- * If any of the planes status are bad or grown bad block, the virtual block
- * is marked bad. If not bad, the first plane state acts as the block state.
+ * If any of the planes status are bad or grown bad, the chunk is marked
+ * offline. If not bad, the first plane state acts as the chunk state.
  */
-int nvm_bb_tbl_fold(struct nvm_dev *dev, u8 *blks, int nr_blks)
+static int nvm_bb_to_chunk(struct nvm_dev *dev, struct ppa_addr ppa,
+			   u8 *blks, int nr_blks, struct nvm_chk_meta *meta)
 {
 	struct nvm_geo *geo = &dev->geo;
-	int blk, offset, pl, blktype;
-
-	if (nr_blks != geo->num_chk * geo->pln_mode)
-		return -EINVAL;
+	int ret, blk, pl, offset, blktype;
 
 	for (blk = 0; blk < geo->num_chk; blk++) {
 		offset = blk * geo->pln_mode;
 		blktype = blks[offset];
 
-		/* Bad blocks on any planes take precedence over other types */
 		for (pl = 0; pl < geo->pln_mode; pl++) {
 			if (blks[offset + pl] &
 					(NVM_BLK_T_BAD|NVM_BLK_T_GRWN_BAD)) {
@@ -859,23 +951,124 @@ int nvm_bb_tbl_fold(struct nvm_dev *dev, u8 *blks, int nr_blks)
 			}
 		}
 
-		blks[blk] = blktype;
+		ppa.g.blk = blk;
+
+		meta->wp = 0;
+		meta->type = NVM_CHK_TP_W_SEQ;
+		meta->wi = 0;
+		meta->slba = generic_to_dev_addr(dev, ppa).ppa;
+		meta->cnlb = dev->geo.clba;
+
+		if (blktype == NVM_BLK_T_FREE) {
+			ret = nvm_bb_chunk_scan(dev, ppa, meta);
+			if (ret)
+				return ret;
+		} else {
+			meta->state = NVM_CHK_ST_OFFLINE;
+		}
+
+		meta++;
 	}
 
-	return geo->num_chk;
+	return 0;
+}
+
+static int nvm_get_bb_meta(struct nvm_dev *dev, sector_t slba,
+			   int nchks, struct nvm_chk_meta *meta)
+{
+	struct nvm_geo *geo = &dev->geo;
+	struct ppa_addr ppa;
+	u8 *blks;
+	int ch, lun, nr_blks;
+	int ret;
+
+	ppa.ppa = slba;
+	ppa = dev_to_generic_addr(dev, ppa);
+
+	if (ppa.g.blk != 0)
+		return -EINVAL;
+
+	if ((nchks % geo->num_chk) != 0)
+		return -EINVAL;
+
+	nr_blks = geo->num_chk * geo->pln_mode;
+
+	blks = kmalloc(nr_blks, GFP_KERNEL);
+	if (!blks)
+		return -ENOMEM;
+
+	for (ch = ppa.g.ch; ch < geo->num_ch; ch++) {
+		for (lun = ppa.g.lun; lun < geo->num_lun; lun++) {
+			struct ppa_addr ppa_gen, ppa_dev;
+
+			if (!nchks)
+				goto done;
+
+			ppa_gen.ppa = 0;
+			ppa_gen.g.ch = ch;
+			ppa_gen.g.lun = lun;
+			ppa_dev = generic_to_dev_addr(dev, ppa_gen);
+
+			ret = dev->ops->get_bb_tbl(dev, ppa_dev, blks);
+			if (ret)
+				goto done;
+
+			ret = nvm_bb_to_chunk(dev, ppa_gen, blks, nr_blks,
+									meta);
+			if (ret)
+				goto done;
+
+			meta += geo->num_chk;
+			nchks -= geo->num_chk;
+		}
+	}
+done:
+	kfree(blks);
+	return ret;
 }
-EXPORT_SYMBOL(nvm_bb_tbl_fold);
 
-int nvm_get_tgt_bb_tbl(struct nvm_tgt_dev *tgt_dev, struct ppa_addr ppa,
-		       u8 *blks)
+int nvm_get_chunk_meta(struct nvm_tgt_dev *tgt_dev, struct ppa_addr ppa,
+		       int nchks, struct nvm_chk_meta *meta)
 {
 	struct nvm_dev *dev = tgt_dev->parent;
 
 	nvm_ppa_tgt_to_dev(tgt_dev, &ppa, 1);
 
-	return dev->ops->get_bb_tbl(dev, ppa, blks);
+	if (dev->geo.version == NVM_OCSSD_SPEC_12)
+		return nvm_get_bb_meta(dev, (sector_t)ppa.ppa, nchks, meta);
+
+	return dev->ops->get_chk_meta(dev, (sector_t)ppa.ppa, nchks, meta);
 }
-EXPORT_SYMBOL(nvm_get_tgt_bb_tbl);
+EXPORT_SYMBOL_GPL(nvm_get_chunk_meta);
+
+int nvm_set_chunk_meta(struct nvm_tgt_dev *tgt_dev, struct ppa_addr *ppas,
+		       int nr_ppas, int type)
+{
+	struct nvm_dev *dev = tgt_dev->parent;
+	struct nvm_rq rqd;
+	int ret;
+
+	if (dev->geo.version == NVM_OCSSD_SPEC_20)
+		return 0;
+
+	if (nr_ppas > NVM_MAX_VLBA) {
+		pr_err("nvm: unable to update all blocks atomically\n");
+		return -EINVAL;
+	}
+
+	memset(&rqd, 0, sizeof(struct nvm_rq));
+
+	nvm_set_rqd_ppalist(tgt_dev, &rqd, ppas, nr_ppas);
+	nvm_rq_tgt_to_dev(tgt_dev, &rqd);
+
+	ret = dev->ops->set_bb_tbl(dev, &rqd.ppa_addr, rqd.nr_ppas, type);
+	nvm_free_rqd_ppalist(tgt_dev, &rqd);
+	if (ret)
+		return -EINVAL;
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(nvm_set_chunk_meta);
 
 static int nvm_core_init(struct nvm_dev *dev)
 {
diff --git a/drivers/lightnvm/pblk-core.c b/drivers/lightnvm/pblk-core.c
index 72de7456845b..e0b513d07e14 100644
--- a/drivers/lightnvm/pblk-core.c
+++ b/drivers/lightnvm/pblk-core.c
@@ -27,7 +27,7 @@ static void pblk_line_mark_bb(struct work_struct *work)
 	struct ppa_addr *ppa = line_ws->priv;
 	int ret;
 
-	ret = nvm_set_tgt_bb_tbl(dev, ppa, 1, NVM_BLK_T_GRWN_BAD);
+	ret = nvm_set_chunk_meta(dev, ppa, 1, NVM_BLK_T_GRWN_BAD);
 	if (ret) {
 		struct pblk_line *line;
 		int pos;
@@ -110,7 +110,7 @@ static void pblk_end_io_erase(struct nvm_rq *rqd)
  *
  * The caller is responsible for freeing the returned structure
  */
-struct nvm_chk_meta *pblk_chunk_get_info(struct pblk *pblk)
+struct nvm_chk_meta *pblk_get_chunk_meta(struct pblk *pblk)
 {
 	struct nvm_tgt_dev *dev = pblk->dev;
 	struct nvm_geo *geo = &dev->geo;
@@ -126,7 +126,7 @@ struct nvm_chk_meta *pblk_chunk_get_info(struct pblk *pblk)
 	if (!meta)
 		return ERR_PTR(-ENOMEM);
 
-	ret = nvm_get_chunk_meta(dev, meta, ppa, geo->all_chunks);
+	ret = nvm_get_chunk_meta(dev, ppa, geo->all_chunks, meta);
 	if (ret) {
 		kfree(meta);
 		return ERR_PTR(-EIO);
diff --git a/drivers/lightnvm/pblk-init.c b/drivers/lightnvm/pblk-init.c
index 039f62d05e84..53bd52114aee 100644
--- a/drivers/lightnvm/pblk-init.c
+++ b/drivers/lightnvm/pblk-init.c
@@ -540,67 +540,6 @@ static void pblk_lines_free(struct pblk *pblk)
 	kfree(pblk->lines);
 }
 
-static int pblk_bb_get_tbl(struct nvm_tgt_dev *dev, struct pblk_lun *rlun,
-			   u8 *blks, int nr_blks)
-{
-	struct ppa_addr ppa;
-	int ret;
-
-	ppa.ppa = 0;
-	ppa.g.ch = rlun->bppa.g.ch;
-	ppa.g.lun = rlun->bppa.g.lun;
-
-	ret = nvm_get_tgt_bb_tbl(dev, ppa, blks);
-	if (ret)
-		return ret;
-
-	nr_blks = nvm_bb_tbl_fold(dev->parent, blks, nr_blks);
-	if (nr_blks < 0)
-		return -EIO;
-
-	return 0;
-}
-
-static void *pblk_bb_get_meta(struct pblk *pblk)
-{
-	struct nvm_tgt_dev *dev = pblk->dev;
-	struct nvm_geo *geo = &dev->geo;
-	u8 *meta;
-	int i, nr_blks, blk_per_lun;
-	int ret;
-
-	blk_per_lun = geo->num_chk * geo->pln_mode;
-	nr_blks = blk_per_lun * geo->all_luns;
-
-	meta = kmalloc(nr_blks, GFP_KERNEL);
-	if (!meta)
-		return ERR_PTR(-ENOMEM);
-
-	for (i = 0; i < geo->all_luns; i++) {
-		struct pblk_lun *rlun = &pblk->luns[i];
-		u8 *meta_pos = meta + i * blk_per_lun;
-
-		ret = pblk_bb_get_tbl(dev, rlun, meta_pos, blk_per_lun);
-		if (ret) {
-			kfree(meta);
-			return ERR_PTR(-EIO);
-		}
-	}
-
-	return meta;
-}
-
-static void *pblk_chunk_get_meta(struct pblk *pblk)
-{
-	struct nvm_tgt_dev *dev = pblk->dev;
-	struct nvm_geo *geo = &dev->geo;
-
-	if (geo->version == NVM_OCSSD_SPEC_12)
-		return pblk_bb_get_meta(pblk);
-	else
-		return pblk_chunk_get_info(pblk);
-}
-
 static int pblk_luns_init(struct pblk *pblk)
 {
 	struct nvm_tgt_dev *dev = pblk->dev;
@@ -699,51 +638,7 @@ static void pblk_set_provision(struct pblk *pblk, long nr_free_blks)
 	atomic_set(&pblk->rl.free_user_blocks, nr_free_blks);
 }
 
-static int pblk_setup_line_meta_12(struct pblk *pblk, struct pblk_line *line,
-				   void *chunk_meta)
-{
-	struct nvm_tgt_dev *dev = pblk->dev;
-	struct nvm_geo *geo = &dev->geo;
-	struct pblk_line_meta *lm = &pblk->lm;
-	int i, chk_per_lun, nr_bad_chks = 0;
-
-	chk_per_lun = geo->num_chk * geo->pln_mode;
-
-	for (i = 0; i < lm->blk_per_line; i++) {
-		struct pblk_lun *rlun = &pblk->luns[i];
-		struct nvm_chk_meta *chunk;
-		int pos = pblk_ppa_to_pos(geo, rlun->bppa);
-		u8 *lun_bb_meta = chunk_meta + pos * chk_per_lun;
-
-		chunk = &line->chks[pos];
-
-		/*
-		 * In 1.2 spec. chunk state is not persisted by the device. Thus
-		 * some of the values are reset each time pblk is instantiated,
-		 * so we have to assume that the block is closed.
-		 */
-		if (lun_bb_meta[line->id] == NVM_BLK_T_FREE)
-			chunk->state =  NVM_CHK_ST_CLOSED;
-		else
-			chunk->state = NVM_CHK_ST_OFFLINE;
-
-		chunk->type = NVM_CHK_TP_W_SEQ;
-		chunk->wi = 0;
-		chunk->slba = -1;
-		chunk->cnlb = geo->clba;
-		chunk->wp = 0;
-
-		if (!(chunk->state & NVM_CHK_ST_OFFLINE))
-			continue;
-
-		set_bit(pos, line->blk_bitmap);
-		nr_bad_chks++;
-	}
-
-	return nr_bad_chks;
-}
-
-static int pblk_setup_line_meta_20(struct pblk *pblk, struct pblk_line *line,
+static int pblk_setup_line_meta_chk(struct pblk *pblk, struct pblk_line *line,
 				   struct nvm_chk_meta *meta)
 {
 	struct nvm_tgt_dev *dev = pblk->dev;
@@ -790,8 +685,6 @@ static int pblk_setup_line_meta_20(struct pblk *pblk, struct pblk_line *line,
 static long pblk_setup_line_meta(struct pblk *pblk, struct pblk_line *line,
 				 void *chunk_meta, int line_id)
 {
-	struct nvm_tgt_dev *dev = pblk->dev;
-	struct nvm_geo *geo = &dev->geo;
 	struct pblk_line_mgmt *l_mg = &pblk->l_mg;
 	struct pblk_line_meta *lm = &pblk->lm;
 	long nr_bad_chks, chk_in_line;
@@ -804,10 +697,7 @@ static long pblk_setup_line_meta(struct pblk *pblk, struct pblk_line *line,
 	line->vsc = &l_mg->vsc_list[line_id];
 	spin_lock_init(&line->lock);
 
-	if (geo->version == NVM_OCSSD_SPEC_12)
-		nr_bad_chks = pblk_setup_line_meta_12(pblk, line, chunk_meta);
-	else
-		nr_bad_chks = pblk_setup_line_meta_20(pblk, line, chunk_meta);
+	nr_bad_chks = pblk_setup_line_meta_chk(pblk, line, chunk_meta);
 
 	chk_in_line = lm->blk_per_line - nr_bad_chks;
 	if (nr_bad_chks < 0 || nr_bad_chks > lm->blk_per_line ||
@@ -1058,7 +948,7 @@ static int pblk_lines_init(struct pblk *pblk)
 	if (ret)
 		goto fail_free_meta;
 
-	chunk_meta = pblk_chunk_get_meta(pblk);
+	chunk_meta = pblk_get_chunk_meta(pblk);
 	if (IS_ERR(chunk_meta)) {
 		ret = PTR_ERR(chunk_meta);
 		goto fail_free_luns;
diff --git a/drivers/lightnvm/pblk.h b/drivers/lightnvm/pblk.h
index 48b3035df3c4..579b4ea9716c 100644
--- a/drivers/lightnvm/pblk.h
+++ b/drivers/lightnvm/pblk.h
@@ -774,7 +774,7 @@ void pblk_set_sec_per_write(struct pblk *pblk, int sec_per_write);
 int pblk_setup_w_rec_rq(struct pblk *pblk, struct nvm_rq *rqd,
 			struct pblk_c_ctx *c_ctx);
 void pblk_discard(struct pblk *pblk, struct bio *bio);
-struct nvm_chk_meta *pblk_chunk_get_info(struct pblk *pblk);
+struct nvm_chk_meta *pblk_get_chunk_meta(struct pblk *pblk);
 struct nvm_chk_meta *pblk_chunk_get_off(struct pblk *pblk,
 					      struct nvm_chk_meta *lp,
 					      struct ppa_addr ppa);
diff --git a/drivers/nvme/host/lightnvm.c b/drivers/nvme/host/lightnvm.c
index 1e4f97538838..e42af7771fe5 100644
--- a/drivers/nvme/host/lightnvm.c
+++ b/drivers/nvme/host/lightnvm.c
@@ -567,8 +567,8 @@ static int nvme_nvm_set_bb_tbl(struct nvm_dev *nvmdev, struct ppa_addr *ppas,
  * Expect the lba in device format
  */
 static int nvme_nvm_get_chk_meta(struct nvm_dev *ndev,
-				 struct nvm_chk_meta *meta,
-				 sector_t slba, int nchks)
+				 sector_t slba, int nchks,
+				 struct nvm_chk_meta *meta)
 {
 	struct nvm_geo *geo = &ndev->geo;
 	struct nvme_ns *ns = ndev->q->queuedata;
diff --git a/include/linux/lightnvm.h b/include/linux/lightnvm.h
index f4a84694e5e2..0106984400bc 100644
--- a/include/linux/lightnvm.h
+++ b/include/linux/lightnvm.h
@@ -86,8 +86,8 @@ struct nvm_chk_meta;
 typedef int (nvm_id_fn)(struct nvm_dev *);
 typedef int (nvm_op_bb_tbl_fn)(struct nvm_dev *, struct ppa_addr, u8 *);
 typedef int (nvm_op_set_bb_fn)(struct nvm_dev *, struct ppa_addr *, int, int);
-typedef int (nvm_get_chk_meta_fn)(struct nvm_dev *, struct nvm_chk_meta *,
-								sector_t, int);
+typedef int (nvm_get_chk_meta_fn)(struct nvm_dev *, sector_t, int,
+							struct nvm_chk_meta *);
 typedef int (nvm_submit_io_fn)(struct nvm_dev *, struct nvm_rq *);
 typedef int (nvm_submit_io_sync_fn)(struct nvm_dev *, struct nvm_rq *);
 typedef void *(nvm_create_dma_pool_fn)(struct nvm_dev *, char *);
@@ -532,18 +532,13 @@ extern struct nvm_dev *nvm_alloc_dev(int);
 extern int nvm_register(struct nvm_dev *);
 extern void nvm_unregister(struct nvm_dev *);
 
-
-extern int nvm_get_chunk_meta(struct nvm_tgt_dev *tgt_dev,
-			      struct nvm_chk_meta *meta, struct ppa_addr ppa,
-			      int nchks);
-
-extern int nvm_set_tgt_bb_tbl(struct nvm_tgt_dev *, struct ppa_addr *,
+extern int nvm_get_chunk_meta(struct nvm_tgt_dev *, struct ppa_addr,
+			      int, struct nvm_chk_meta *);
+extern int nvm_set_chunk_meta(struct nvm_tgt_dev *, struct ppa_addr *,
 			      int, int);
 extern int nvm_submit_io(struct nvm_tgt_dev *, struct nvm_rq *);
 extern int nvm_submit_io_sync(struct nvm_tgt_dev *, struct nvm_rq *);
 extern void nvm_end_io(struct nvm_rq *);
-extern int nvm_bb_tbl_fold(struct nvm_dev *, u8 *, int);
-extern int nvm_get_tgt_bb_tbl(struct nvm_tgt_dev *, struct ppa_addr, u8 *);
 
 #else /* CONFIG_NVM */
 struct nvm_dev_ops;
-- 
2.17.1

[GIT PULL 07/45] lightnvm: pblk: unify vector max req constants

[Matias Bjørling]

Both NVM_MAX_VLBA and PBLK_MAX_REQ_ADDRS define how many LBAs that
are available in a vector command. pblk uses them interchangeably
in its implementation. Use NVM_MAX_VLBA as the main one and remove
usages of PBLK_MAX_REQ_ADDRS.

Also remove the power representation that only has one user, and
instead calculate it at runtime.

Signed-off-by: Matias Bjørling <mb@lightnvm.io>
Reviewed-by: Javier González <javier@cnexlabs.com>
---
 drivers/lightnvm/pblk-init.c |  6 ------
 drivers/lightnvm/pblk-read.c |  6 +++---
 drivers/lightnvm/pblk-rl.c   |  4 ++--
 drivers/lightnvm/pblk.h      | 10 ++++------
 4 files changed, 9 insertions(+), 17 deletions(-)

diff --git a/drivers/lightnvm/pblk-init.c b/drivers/lightnvm/pblk-init.c
index 53bd52114aee..9119c64d6f62 100644
--- a/drivers/lightnvm/pblk-init.c
+++ b/drivers/lightnvm/pblk-init.c
@@ -376,12 +376,6 @@ static int pblk_core_init(struct pblk *pblk)
 	pblk->max_write_pgs = min_t(int, max_write_ppas, NVM_MAX_VLBA);
 	pblk_set_sec_per_write(pblk, pblk->min_write_pgs);
 
-	if (pblk->max_write_pgs > PBLK_MAX_REQ_ADDRS) {
-		pblk_err(pblk, "vector list too big(%u > %u)\n",
-				pblk->max_write_pgs, PBLK_MAX_REQ_ADDRS);
-		return -EINVAL;
-	}
-
 	pblk->pad_dist = kcalloc(pblk->min_write_pgs - 1, sizeof(atomic64_t),
 								GFP_KERNEL);
 	if (!pblk->pad_dist)
diff --git a/drivers/lightnvm/pblk-read.c b/drivers/lightnvm/pblk-read.c
index 07b7f82c7595..9c61c19be5dc 100644
--- a/drivers/lightnvm/pblk-read.c
+++ b/drivers/lightnvm/pblk-read.c
@@ -43,7 +43,7 @@ static void pblk_read_ppalist_rq(struct pblk *pblk, struct nvm_rq *rqd,
 				 unsigned long *read_bitmap)
 {
 	struct pblk_sec_meta *meta_list = rqd->meta_list;
-	struct ppa_addr ppas[PBLK_MAX_REQ_ADDRS];
+	struct ppa_addr ppas[NVM_MAX_VLBA];
 	int nr_secs = rqd->nr_ppas;
 	bool advanced_bio = false;
 	int i, j = 0;
@@ -450,7 +450,7 @@ int pblk_submit_read(struct pblk *pblk, struct bio *bio)
 	int ret = NVM_IO_ERR;
 
 	/* logic error: lba out-of-bounds. Ignore read request */
-	if (blba >= pblk->rl.nr_secs || nr_secs > PBLK_MAX_REQ_ADDRS) {
+	if (blba >= pblk->rl.nr_secs || nr_secs > NVM_MAX_VLBA) {
 		WARN(1, "pblk: read lba out of bounds (lba:%llu, nr:%d)\n",
 					(unsigned long long)blba, nr_secs);
 		return NVM_IO_ERR;
@@ -547,7 +547,7 @@ static int read_ppalist_rq_gc(struct pblk *pblk, struct nvm_rq *rqd,
 			      struct pblk_line *line, u64 *lba_list,
 			      u64 *paddr_list_gc, unsigned int nr_secs)
 {
-	struct ppa_addr ppa_list_l2p[PBLK_MAX_REQ_ADDRS];
+	struct ppa_addr ppa_list_l2p[NVM_MAX_VLBA];
 	struct ppa_addr ppa_gc;
 	int valid_secs = 0;
 	int i;
diff --git a/drivers/lightnvm/pblk-rl.c b/drivers/lightnvm/pblk-rl.c
index 6a0616a6fcaf..a32790f7b7fc 100644
--- a/drivers/lightnvm/pblk-rl.c
+++ b/drivers/lightnvm/pblk-rl.c
@@ -127,7 +127,7 @@ static void __pblk_rl_update_rates(struct pblk_rl *rl,
 	} else if (free_blocks < rl->high) {
 		int shift = rl->high_pw - rl->rb_windows_pw;
 		int user_windows = free_blocks >> shift;
-		int user_max = user_windows << PBLK_MAX_REQ_ADDRS_PW;
+		int user_max = user_windows << ilog2(NVM_MAX_VLBA);
 
 		rl->rb_user_max = user_max;
 		rl->rb_gc_max = max - user_max;
@@ -228,7 +228,7 @@ void pblk_rl_init(struct pblk_rl *rl, int budget)
 	rl->rsv_blocks = min_blocks;
 
 	/* This will always be a power-of-2 */
-	rb_windows = budget / PBLK_MAX_REQ_ADDRS;
+	rb_windows = budget / NVM_MAX_VLBA;
 	rl->rb_windows_pw = get_count_order(rb_windows);
 
 	/* To start with, all buffer is available to user I/O writers */
diff --git a/drivers/lightnvm/pblk.h b/drivers/lightnvm/pblk.h
index 579b4ea9716c..64ae0c7ed3bb 100644
--- a/drivers/lightnvm/pblk.h
+++ b/drivers/lightnvm/pblk.h
@@ -37,8 +37,6 @@
 
 #define PBLK_SECTOR (512)
 #define PBLK_EXPOSED_PAGE_SIZE (4096)
-#define PBLK_MAX_REQ_ADDRS (64)
-#define PBLK_MAX_REQ_ADDRS_PW (6)
 
 #define PBLK_NR_CLOSE_JOBS (4)
 
@@ -99,8 +97,8 @@ enum {
 	PBLK_RL_LOW = 4
 };
 
-#define pblk_dma_meta_size (sizeof(struct pblk_sec_meta) * PBLK_MAX_REQ_ADDRS)
-#define pblk_dma_ppa_size (sizeof(u64) * PBLK_MAX_REQ_ADDRS)
+#define pblk_dma_meta_size (sizeof(struct pblk_sec_meta) * NVM_MAX_VLBA)
+#define pblk_dma_ppa_size (sizeof(u64) * NVM_MAX_VLBA)
 
 /* write buffer completion context */
 struct pblk_c_ctx {
@@ -218,8 +216,8 @@ struct pblk_lun {
 struct pblk_gc_rq {
 	struct pblk_line *line;
 	void *data;
-	u64 paddr_list[PBLK_MAX_REQ_ADDRS];
-	u64 lba_list[PBLK_MAX_REQ_ADDRS];
+	u64 paddr_list[NVM_MAX_VLBA];
+	u64 lba_list[NVM_MAX_VLBA];
 	int nr_secs;
 	int secs_to_gc;
 	struct list_head list;
-- 
2.17.1

[GIT PULL 08/45] lightnvm: pblk: fix incorrect min_write_pgs

[Matias Bjørling]

The calculation of pblk->min_write_pgs should only use the optimal
write size attribute provided by the drive, it does not correlate to
the memory page size of the system, which can be smaller or larger
than the LBA size reported.

Signed-off-by: Matias Bjørling <mb@lightnvm.io>
Reviewed-by: Javier González <javier@cnexlabs.com>
---
 drivers/lightnvm/pblk-init.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/lightnvm/pblk-init.c b/drivers/lightnvm/pblk-init.c
index 9119c64d6f62..8adc8ac8b03c 100644
--- a/drivers/lightnvm/pblk-init.c
+++ b/drivers/lightnvm/pblk-init.c
@@ -371,7 +371,7 @@ static int pblk_core_init(struct pblk *pblk)
 	atomic64_set(&pblk->nr_flush, 0);
 	pblk->nr_flush_rst = 0;
 
-	pblk->min_write_pgs = geo->ws_opt * (geo->csecs / PAGE_SIZE);
+	pblk->min_write_pgs = geo->ws_opt;
 	max_write_ppas = pblk->min_write_pgs * geo->all_luns;
 	pblk->max_write_pgs = min_t(int, max_write_ppas, NVM_MAX_VLBA);
 	pblk_set_sec_per_write(pblk, pblk->min_write_pgs);
-- 
2.17.1

[GIT PULL 09/45] lightnvm: pblk: remove size and out of bounds read check

[Matias Bjørling]

The I/O size and capacity checks are already done by the block layer.

Signed-off-by: Matias Bjørling <mb@lightnvm.io>
Reviewed-by: Javier González <javier@cnexlabs.com>
---
 drivers/lightnvm/pblk-read.c | 7 -------
 1 file changed, 7 deletions(-)

diff --git a/drivers/lightnvm/pblk-read.c b/drivers/lightnvm/pblk-read.c
index 9c61c19be5dc..ba3dcb6be4c3 100644
--- a/drivers/lightnvm/pblk-read.c
+++ b/drivers/lightnvm/pblk-read.c
@@ -449,13 +449,6 @@ int pblk_submit_read(struct pblk *pblk, struct bio *bio)
 	DECLARE_BITMAP(read_bitmap, NVM_MAX_VLBA);
 	int ret = NVM_IO_ERR;
 
-	/* logic error: lba out-of-bounds. Ignore read request */
-	if (blba >= pblk->rl.nr_secs || nr_secs > NVM_MAX_VLBA) {
-		WARN(1, "pblk: read lba out of bounds (lba:%llu, nr:%d)\n",
-					(unsigned long long)blba, nr_secs);
-		return NVM_IO_ERR;
-	}
-
 	generic_start_io_acct(q, REQ_OP_READ, bio_sectors(bio),
 			      &pblk->disk->part0);
 
-- 
2.17.1

[GIT PULL 10/45] lightnvm: pblk: refactor put line fn on read completion

[Matias Bjørling]

The read completion path uses the put_line variable to decide whether
the reference on a line should be released. The function name used for
that is pblk_read_put_rqd_kref, which could lead one to believe that it
is the rqd that is releasing the reference, while it is the line
reference that is put.

Rename and also split the function in two to account for either rqd or
single ppa callers and move it to core, such that it later can be used
in the write path as well.

Signed-off-by: Matias Bjørling <mb@lightnvm.io>
Reviewed-by: Javier González <javier@cnexlabs.com>
Reviewed-by: Heiner Litz <hlitz@ucsc.edu>
---
 drivers/lightnvm/pblk-core.c | 19 +++++++++++++++++++
 drivers/lightnvm/pblk-read.c | 18 +-----------------
 drivers/lightnvm/pblk.h      |  2 ++
 3 files changed, 22 insertions(+), 17 deletions(-)

diff --git a/drivers/lightnvm/pblk-core.c b/drivers/lightnvm/pblk-core.c
index e0b513d07e14..5f99cf396072 100644
--- a/drivers/lightnvm/pblk-core.c
+++ b/drivers/lightnvm/pblk-core.c
@@ -1426,6 +1426,25 @@ struct pblk_line *pblk_line_get_first_data(struct pblk *pblk)
 	return line;
 }
 
+void pblk_ppa_to_line_put(struct pblk *pblk, struct ppa_addr ppa)
+{
+	struct pblk_line *line;
+
+	line = &pblk->lines[pblk_ppa_to_line(ppa)];
+	kref_put(&line->ref, pblk_line_put_wq);
+}
+
+void pblk_rq_to_line_put(struct pblk *pblk, struct nvm_rq *rqd)
+{
+	struct ppa_addr *ppa_list;
+	int i;
+
+	ppa_list = (rqd->nr_ppas > 1) ? rqd->ppa_list : &rqd->ppa_addr;
+
+	for (i = 0; i < rqd->nr_ppas; i++)
+		pblk_ppa_to_line_put(pblk, ppa_list[i]);
+}
+
 static void pblk_stop_writes(struct pblk *pblk, struct pblk_line *line)
 {
 	lockdep_assert_held(&pblk->l_mg.free_lock);
diff --git a/drivers/lightnvm/pblk-read.c b/drivers/lightnvm/pblk-read.c
index ba3dcb6be4c3..ad1d7713bbda 100644
--- a/drivers/lightnvm/pblk-read.c
+++ b/drivers/lightnvm/pblk-read.c
@@ -165,22 +165,6 @@ static void pblk_read_check_rand(struct pblk *pblk, struct nvm_rq *rqd,
 	WARN_ONCE(j != rqd->nr_ppas, "pblk: corrupted random request\n");
 }
 
-static void pblk_read_put_rqd_kref(struct pblk *pblk, struct nvm_rq *rqd)
-{
-	struct ppa_addr *ppa_list;
-	int i;
-
-	ppa_list = (rqd->nr_ppas > 1) ? rqd->ppa_list : &rqd->ppa_addr;
-
-	for (i = 0; i < rqd->nr_ppas; i++) {
-		struct ppa_addr ppa = ppa_list[i];
-		struct pblk_line *line;
-
-		line = &pblk->lines[pblk_ppa_to_line(ppa)];
-		kref_put(&line->ref, pblk_line_put_wq);
-	}
-}
-
 static void pblk_end_user_read(struct bio *bio)
 {
 #ifdef CONFIG_NVM_PBLK_DEBUG
@@ -208,7 +192,7 @@ static void __pblk_end_io_read(struct pblk *pblk, struct nvm_rq *rqd,
 		bio_put(int_bio);
 
 	if (put_line)
-		pblk_read_put_rqd_kref(pblk, rqd);
+		pblk_rq_to_line_put(pblk, rqd);
 
 #ifdef CONFIG_NVM_PBLK_DEBUG
 	atomic_long_add(rqd->nr_ppas, &pblk->sync_reads);
diff --git a/drivers/lightnvm/pblk.h b/drivers/lightnvm/pblk.h
index 64ae0c7ed3bb..be67bbfa3d0a 100644
--- a/drivers/lightnvm/pblk.h
+++ b/drivers/lightnvm/pblk.h
@@ -787,6 +787,8 @@ struct bio *pblk_bio_map_addr(struct pblk *pblk, void *data,
 struct pblk_line *pblk_line_get(struct pblk *pblk);
 struct pblk_line *pblk_line_get_first_data(struct pblk *pblk);
 struct pblk_line *pblk_line_replace_data(struct pblk *pblk);
+void pblk_ppa_to_line_put(struct pblk *pblk, struct ppa_addr ppa);
+void pblk_rq_to_line_put(struct pblk *pblk, struct nvm_rq *rqd);
 int pblk_line_recov_alloc(struct pblk *pblk, struct pblk_line *line);
 void pblk_line_recov_close(struct pblk *pblk, struct pblk_line *line);
 struct pblk_line *pblk_line_get_data(struct pblk *pblk);
-- 
2.17.1

[GIT PULL 11/45] lightnvm: pblk: add helpers for chunk addresses

[Matias Bjørling]

From: Javier González <javier@javigon.com>

Implement helpers to go from ppas to a chunk within a line and an
address within a chunk.

These helpers will be used on the patches adding trace support in pblk,
which will be sent in this window.

Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk.h  | 19 +++++++++++++++++++
 include/linux/lightnvm.h | 19 +++++++++++++++++++
 2 files changed, 38 insertions(+)

diff --git a/drivers/lightnvm/pblk.h b/drivers/lightnvm/pblk.h
index be67bbfa3d0a..f95fe75fef6e 100644
--- a/drivers/lightnvm/pblk.h
+++ b/drivers/lightnvm/pblk.h
@@ -1034,6 +1034,25 @@ static inline struct ppa_addr addr_to_gen_ppa(struct pblk *pblk, u64 paddr,
 	return ppa;
 }
 
+static inline struct nvm_chk_meta *pblk_dev_ppa_to_chunk(struct pblk *pblk,
+							struct ppa_addr p)
+{
+	struct nvm_tgt_dev *dev = pblk->dev;
+	struct nvm_geo *geo = &dev->geo;
+	struct pblk_line *line = &pblk->lines[pblk_ppa_to_line(p)];
+	int pos = pblk_ppa_to_pos(geo, p);
+
+	return &line->chks[pos];
+}
+
+static inline u64 pblk_dev_ppa_to_chunk_addr(struct pblk *pblk,
+							struct ppa_addr p)
+{
+	struct nvm_tgt_dev *dev = pblk->dev;
+
+	return dev_to_chunk_addr(dev->parent, &pblk->addrf, p);
+}
+
 static inline u64 pblk_dev_ppa_to_line_addr(struct pblk *pblk,
 							struct ppa_addr p)
 {
diff --git a/include/linux/lightnvm.h b/include/linux/lightnvm.h
index 0106984400bc..77743a02ec0d 100644
--- a/include/linux/lightnvm.h
+++ b/include/linux/lightnvm.h
@@ -487,6 +487,25 @@ static inline struct ppa_addr dev_to_generic_addr(struct nvm_dev *dev,
 	return l;
 }
 
+static inline u64 dev_to_chunk_addr(struct nvm_dev *dev, void *addrf,
+				    struct ppa_addr p)
+{
+	struct nvm_geo *geo = &dev->geo;
+	u64 caddr;
+
+	if (geo->version == NVM_OCSSD_SPEC_12) {
+		struct nvm_addrf_12 *ppaf = (struct nvm_addrf_12 *)addrf;
+
+		caddr = (u64)p.g.pg << ppaf->pg_offset;
+		caddr |= (u64)p.g.pl << ppaf->pln_offset;
+		caddr |= (u64)p.g.sec << ppaf->sec_offset;
+	} else {
+		caddr = p.m.sec;
+	}
+
+	return caddr;
+}
+
 typedef blk_qc_t (nvm_tgt_make_rq_fn)(struct request_queue *, struct bio *);
 typedef sector_t (nvm_tgt_capacity_fn)(void *);
 typedef void *(nvm_tgt_init_fn)(struct nvm_tgt_dev *, struct gendisk *,
-- 
2.17.1

[GIT PULL 12/45] lightnvm: pblk: improve line helpers

[Matias Bjørling]

From: Javier González <javier@javigon.com>

The current helper to obtain a line from a ppa returns the line id,
which requires its users to explicitly retrieve the pointer to the line
with the id.

Make 2 different helpers: one returning the line id and one returning
the line directly.

Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk-core.c  | 15 ++++++---------
 drivers/lightnvm/pblk-rb.c    |  2 +-
 drivers/lightnvm/pblk-read.c  |  4 ++--
 drivers/lightnvm/pblk-write.c |  4 ++--
 drivers/lightnvm/pblk.h       | 13 +++++++++----
 5 files changed, 20 insertions(+), 18 deletions(-)

diff --git a/drivers/lightnvm/pblk-core.c b/drivers/lightnvm/pblk-core.c
index 5f99cf396072..36ac9eff8ebd 100644
--- a/drivers/lightnvm/pblk-core.c
+++ b/drivers/lightnvm/pblk-core.c
@@ -32,7 +32,7 @@ static void pblk_line_mark_bb(struct work_struct *work)
 		struct pblk_line *line;
 		int pos;
 
-		line = &pblk->lines[pblk_ppa_to_line(*ppa)];
+		line = pblk_ppa_to_line(pblk, *ppa);
 		pos = pblk_ppa_to_pos(&dev->geo, *ppa);
 
 		pblk_err(pblk, "failed to mark bb, line:%d, pos:%d\n",
@@ -80,7 +80,7 @@ static void __pblk_end_io_erase(struct pblk *pblk, struct nvm_rq *rqd)
 	struct pblk_line *line;
 	int pos;
 
-	line = &pblk->lines[pblk_ppa_to_line(rqd->ppa_addr)];
+	line = pblk_ppa_to_line(pblk, rqd->ppa_addr);
 	pos = pblk_ppa_to_pos(geo, rqd->ppa_addr);
 	chunk = &line->chks[pos];
 
@@ -192,7 +192,6 @@ void pblk_map_invalidate(struct pblk *pblk, struct ppa_addr ppa)
 {
 	struct pblk_line *line;
 	u64 paddr;
-	int line_id;
 
 #ifdef CONFIG_NVM_PBLK_DEBUG
 	/* Callers must ensure that the ppa points to a device address */
@@ -200,8 +199,7 @@ void pblk_map_invalidate(struct pblk *pblk, struct ppa_addr ppa)
 	BUG_ON(pblk_ppa_empty(ppa));
 #endif
 
-	line_id = pblk_ppa_to_line(ppa);
-	line = &pblk->lines[line_id];
+	line = pblk_ppa_to_line(pblk, ppa);
 	paddr = pblk_dev_ppa_to_line_addr(pblk, ppa);
 
 	__pblk_map_invalidate(pblk, line, paddr);
@@ -1430,7 +1428,7 @@ void pblk_ppa_to_line_put(struct pblk *pblk, struct ppa_addr ppa)
 {
 	struct pblk_line *line;
 
-	line = &pblk->lines[pblk_ppa_to_line(ppa)];
+	line = pblk_ppa_to_line(pblk, ppa);
 	kref_put(&line->ref, pblk_line_put_wq);
 }
 
@@ -1688,7 +1686,7 @@ int pblk_blk_erase_async(struct pblk *pblk, struct ppa_addr ppa)
 		struct nvm_geo *geo = &dev->geo;
 
 		pblk_err(pblk, "could not async erase line:%d,blk:%d\n",
-					pblk_ppa_to_line(ppa),
+					pblk_ppa_to_line_id(ppa),
 					pblk_ppa_to_pos(geo, ppa));
 	}
 
@@ -2059,8 +2057,7 @@ void pblk_lookup_l2p_seq(struct pblk *pblk, struct ppa_addr *ppas,
 
 		/* If the L2P entry maps to a line, the reference is valid */
 		if (!pblk_ppa_empty(ppa) && !pblk_addr_in_cache(ppa)) {
-			int line_id = pblk_ppa_to_line(ppa);
-			struct pblk_line *line = &pblk->lines[line_id];
+			struct pblk_line *line = pblk_ppa_to_line(pblk, ppa);
 
 			kref_get(&line->ref);
 		}
diff --git a/drivers/lightnvm/pblk-rb.c b/drivers/lightnvm/pblk-rb.c
index f6eec0212dfc..a7648e12f54f 100644
--- a/drivers/lightnvm/pblk-rb.c
+++ b/drivers/lightnvm/pblk-rb.c
@@ -225,7 +225,7 @@ static int __pblk_rb_update_l2p(struct pblk_rb *rb, unsigned int to_update)
 		pblk_update_map_dev(pblk, w_ctx->lba, w_ctx->ppa,
 							entry->cacheline);
 
-		line = &pblk->lines[pblk_ppa_to_line(w_ctx->ppa)];
+		line = pblk_ppa_to_line(pblk, w_ctx->ppa);
 		kref_put(&line->ref, pblk_line_put);
 		clean_wctx(w_ctx);
 		rb->l2p_update = (rb->l2p_update + 1) & (rb->nr_entries - 1);
diff --git a/drivers/lightnvm/pblk-read.c b/drivers/lightnvm/pblk-read.c
index ad1d7713bbda..49744caaa300 100644
--- a/drivers/lightnvm/pblk-read.c
+++ b/drivers/lightnvm/pblk-read.c
@@ -252,9 +252,9 @@ static void pblk_end_partial_read(struct nvm_rq *rqd)
 	i = 0;
 	hole = find_first_zero_bit(read_bitmap, nr_secs);
 	do {
-		int line_id = pblk_ppa_to_line(rqd->ppa_list[i]);
-		struct pblk_line *line = &pblk->lines[line_id];
+		struct pblk_line *line;
 
+		line = pblk_ppa_to_line(pblk, rqd->ppa_list[i]);
 		kref_put(&line->ref, pblk_line_put);
 
 		meta_list[hole].lba = lba_list_media[i];
diff --git a/drivers/lightnvm/pblk-write.c b/drivers/lightnvm/pblk-write.c
index df99c45778d4..c23b65aaa27b 100644
--- a/drivers/lightnvm/pblk-write.c
+++ b/drivers/lightnvm/pblk-write.c
@@ -113,7 +113,7 @@ static void pblk_map_remaining(struct pblk *pblk, struct ppa_addr *ppa)
 	u64 paddr;
 	int done = 0;
 
-	line = &pblk->lines[pblk_ppa_to_line(*ppa)];
+	line = pblk_ppa_to_line(pblk, *ppa);
 	spin_lock(&line->lock);
 
 	while (!done)  {
@@ -171,7 +171,7 @@ static void pblk_prepare_resubmit(struct pblk *pblk, unsigned int sentry,
 		/* Decrese the reference count to the line as we will
 		 * re-map these entries
 		 */
-		line = &pblk->lines[pblk_ppa_to_line(w_ctx->ppa)];
+		line = pblk_ppa_to_line(pblk, w_ctx->ppa);
 		kref_put(&line->ref, pblk_line_put);
 
 		pos = (pos + 1) & (rb->nr_entries - 1);
diff --git a/drivers/lightnvm/pblk.h b/drivers/lightnvm/pblk.h
index f95fe75fef6e..af9477b7e803 100644
--- a/drivers/lightnvm/pblk.h
+++ b/drivers/lightnvm/pblk.h
@@ -984,11 +984,17 @@ static inline int pblk_pad_distance(struct pblk *pblk)
 	return geo->mw_cunits * geo->all_luns * geo->ws_opt;
 }
 
-static inline int pblk_ppa_to_line(struct ppa_addr p)
+static inline int pblk_ppa_to_line_id(struct ppa_addr p)
 {
 	return p.a.blk;
 }
 
+static inline struct pblk_line *pblk_ppa_to_line(struct pblk *pblk,
+						 struct ppa_addr p)
+{
+	return &pblk->lines[pblk_ppa_to_line_id(p)];
+}
+
 static inline int pblk_ppa_to_pos(struct nvm_geo *geo, struct ppa_addr p)
 {
 	return p.a.lun * geo->num_ch + p.a.ch;
@@ -1039,7 +1045,7 @@ static inline struct nvm_chk_meta *pblk_dev_ppa_to_chunk(struct pblk *pblk,
 {
 	struct nvm_tgt_dev *dev = pblk->dev;
 	struct nvm_geo *geo = &dev->geo;
-	struct pblk_line *line = &pblk->lines[pblk_ppa_to_line(p)];
+	struct pblk_line *line = pblk_ppa_to_line(pblk, p);
 	int pos = pblk_ppa_to_pos(geo, p);
 
 	return &line->chks[pos];
@@ -1371,8 +1377,7 @@ static inline int pblk_check_io(struct pblk *pblk, struct nvm_rq *rqd)
 		int i;
 
 		for (i = 0; i < rqd->nr_ppas; i++) {
-			ppa = ppa_list[i];
-			line = &pblk->lines[pblk_ppa_to_line(ppa)];
+			line = pblk_ppa_to_line(pblk, ppa_list[i]);
 
 			spin_lock(&line->lock);
 			if (line->state != PBLK_LINESTATE_OPEN) {
-- 
2.17.1

[GIT PULL 13/45] lightnvm: pblk: fix comment typo

[Matias Bjørling]

From: Javier González <javier@javigon.com>

Fix comment typo Decrese -> Decrease

Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk-write.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/lightnvm/pblk-write.c b/drivers/lightnvm/pblk-write.c
index c23b65aaa27b..625ed5a3a020 100644
--- a/drivers/lightnvm/pblk-write.c
+++ b/drivers/lightnvm/pblk-write.c
@@ -168,7 +168,7 @@ static void pblk_prepare_resubmit(struct pblk *pblk, unsigned int sentry,
 		/* 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
+		/* Decrease the reference count to the line as we will
 		 * re-map these entries
 		 */
 		line = pblk_ppa_to_line(pblk, w_ctx->ppa);
-- 
2.17.1

[GIT PULL 14/45] lightnvm: pblk: remove unused variable.

[Matias Bjørling]

From: Javier González <javier@javigon.com>

Removed unused struct ppa_addr variable.

Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk.h | 1 -
 1 file changed, 1 deletion(-)

diff --git a/drivers/lightnvm/pblk.h b/drivers/lightnvm/pblk.h
index af9477b7e803..1865ac1560fa 100644
--- a/drivers/lightnvm/pblk.h
+++ b/drivers/lightnvm/pblk.h
@@ -1373,7 +1373,6 @@ static inline int pblk_check_io(struct pblk *pblk, struct nvm_rq *rqd)
 
 	if (rqd->opcode == NVM_OP_PWRITE) {
 		struct pblk_line *line;
-		struct ppa_addr ppa;
 		int i;
 
 		for (i = 0; i < rqd->nr_ppas; i++) {
-- 
2.17.1

[GIT PULL 15/45] lightnvm: pblk: guarantee emeta on line close

[Matias Bjørling]

From: Javier González <javier@javigon.com>

If a line is recovered from open chunks, the memory structures for
emeta have not necessarily been properly set on line initialization.
When closing a line, make sure that emeta is consistent so that the line
can be recovered on the fast path on next reboot.

Also, remove a couple of empty lines at the end of the function.

Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk-core.c | 13 +++++++++++--
 1 file changed, 11 insertions(+), 2 deletions(-)

diff --git a/drivers/lightnvm/pblk-core.c b/drivers/lightnvm/pblk-core.c
index 36ac9eff8ebd..a31417682c90 100644
--- a/drivers/lightnvm/pblk-core.c
+++ b/drivers/lightnvm/pblk-core.c
@@ -1775,6 +1775,17 @@ void pblk_line_close_meta(struct pblk *pblk, struct pblk_line *line)
 	wa->pad = cpu_to_le64(atomic64_read(&pblk->pad_wa));
 	wa->gc = cpu_to_le64(atomic64_read(&pblk->gc_wa));
 
+	if (le32_to_cpu(emeta_buf->header.identifier) != PBLK_MAGIC) {
+		emeta_buf->header.identifier = cpu_to_le32(PBLK_MAGIC);
+		memcpy(emeta_buf->header.uuid, pblk->instance_uuid, 16);
+		emeta_buf->header.id = cpu_to_le32(line->id);
+		emeta_buf->header.type = cpu_to_le16(line->type);
+		emeta_buf->header.version_major = EMETA_VERSION_MAJOR;
+		emeta_buf->header.version_minor = EMETA_VERSION_MINOR;
+		emeta_buf->header.crc = cpu_to_le32(
+			pblk_calc_meta_header_crc(pblk, &emeta_buf->header));
+	}
+
 	emeta_buf->nr_valid_lbas = cpu_to_le64(line->nr_valid_lbas);
 	emeta_buf->crc = cpu_to_le32(pblk_calc_emeta_crc(pblk, emeta_buf));
 
@@ -1792,8 +1803,6 @@ void pblk_line_close_meta(struct pblk *pblk, struct pblk_line *line)
 	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)
-- 
2.17.1

[GIT PULL 16/45] lightnvm: introduce nvm_rq_to_ppa_list

[Matias Bjørling]

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

There is a number of places in the lightnvm subsystem where the user
iterates over the ppa list. Before iterating, the user must know if it
is a single or multiple LBAs due to vector commands using either the
nvm_rq ->ppa_addr or ->ppa_list fields on command submission, which
leads to open-coding the if/else statement.

Instead of having multiple if/else's, move it into a function that can
be called by its users.

A nice side effect of this cleanup is that this patch fixes up a
bunch of cases where we don't consider the single-ppa case in pblk.

Signed-off-by: Hans Holmberg <hans.holmberg@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/core.c          | 14 ++++----------
 drivers/lightnvm/pblk-map.c      | 10 ++++++----
 drivers/lightnvm/pblk-read.c     | 11 ++++-------
 drivers/lightnvm/pblk-recovery.c |  9 ++++++---
 drivers/lightnvm/pblk-write.c    | 18 ++++++++----------
 drivers/lightnvm/pblk.h          |  4 +---
 include/linux/lightnvm.h         |  5 +++++
 7 files changed, 34 insertions(+), 37 deletions(-)

diff --git a/drivers/lightnvm/core.c b/drivers/lightnvm/core.c
index 8df188e0767e..efb976a863d2 100644
--- a/drivers/lightnvm/core.c
+++ b/drivers/lightnvm/core.c
@@ -603,22 +603,16 @@ static void nvm_ppa_dev_to_tgt(struct nvm_tgt_dev *tgt_dev,
 
 static void nvm_rq_tgt_to_dev(struct nvm_tgt_dev *tgt_dev, struct nvm_rq *rqd)
 {
-	if (rqd->nr_ppas == 1) {
-		nvm_ppa_tgt_to_dev(tgt_dev, &rqd->ppa_addr, 1);
-		return;
-	}
+	struct ppa_addr *ppa_list = nvm_rq_to_ppa_list(rqd);
 
-	nvm_ppa_tgt_to_dev(tgt_dev, rqd->ppa_list, rqd->nr_ppas);
+	nvm_ppa_tgt_to_dev(tgt_dev, ppa_list, rqd->nr_ppas);
 }
 
 static void nvm_rq_dev_to_tgt(struct nvm_tgt_dev *tgt_dev, struct nvm_rq *rqd)
 {
-	if (rqd->nr_ppas == 1) {
-		nvm_ppa_dev_to_tgt(tgt_dev, &rqd->ppa_addr, 1);
-		return;
-	}
+	struct ppa_addr *ppa_list = nvm_rq_to_ppa_list(rqd);
 
-	nvm_ppa_dev_to_tgt(tgt_dev, rqd->ppa_list, rqd->nr_ppas);
+	nvm_ppa_dev_to_tgt(tgt_dev, ppa_list, rqd->nr_ppas);
 }
 
 int nvm_register_tgt_type(struct nvm_tgt_type *tt)
diff --git a/drivers/lightnvm/pblk-map.c b/drivers/lightnvm/pblk-map.c
index 953ca31dda68..dc0efb852475 100644
--- a/drivers/lightnvm/pblk-map.c
+++ b/drivers/lightnvm/pblk-map.c
@@ -88,13 +88,14 @@ void pblk_map_rq(struct pblk *pblk, struct nvm_rq *rqd, unsigned int sentry,
 		 unsigned int off)
 {
 	struct pblk_sec_meta *meta_list = rqd->meta_list;
+	struct ppa_addr *ppa_list = nvm_rq_to_ppa_list(rqd);
 	unsigned int map_secs;
 	int min = pblk->min_write_pgs;
 	int i;
 
 	for (i = off; i < rqd->nr_ppas; i += min) {
 		map_secs = (i + min > valid_secs) ? (valid_secs % min) : min;
-		if (pblk_map_page_data(pblk, sentry + i, &rqd->ppa_list[i],
+		if (pblk_map_page_data(pblk, sentry + i, &ppa_list[i],
 					lun_bitmap, &meta_list[i], map_secs)) {
 			bio_put(rqd->bio);
 			pblk_free_rqd(pblk, rqd, PBLK_WRITE);
@@ -112,6 +113,7 @@ void pblk_map_erase_rq(struct pblk *pblk, struct nvm_rq *rqd,
 	struct nvm_geo *geo = &dev->geo;
 	struct pblk_line_meta *lm = &pblk->lm;
 	struct pblk_sec_meta *meta_list = rqd->meta_list;
+	struct ppa_addr *ppa_list = nvm_rq_to_ppa_list(rqd);
 	struct pblk_line *e_line, *d_line;
 	unsigned int map_secs;
 	int min = pblk->min_write_pgs;
@@ -119,14 +121,14 @@ void pblk_map_erase_rq(struct pblk *pblk, struct nvm_rq *rqd,
 
 	for (i = 0; i < rqd->nr_ppas; i += min) {
 		map_secs = (i + min > valid_secs) ? (valid_secs % min) : min;
-		if (pblk_map_page_data(pblk, sentry + i, &rqd->ppa_list[i],
+		if (pblk_map_page_data(pblk, sentry + i, &ppa_list[i],
 					lun_bitmap, &meta_list[i], map_secs)) {
 			bio_put(rqd->bio);
 			pblk_free_rqd(pblk, rqd, PBLK_WRITE);
 			pblk_pipeline_stop(pblk);
 		}
 
-		erase_lun = pblk_ppa_to_pos(geo, rqd->ppa_list[i]);
+		erase_lun = pblk_ppa_to_pos(geo, ppa_list[i]);
 
 		/* line can change after page map. We might also be writing the
 		 * last line.
@@ -141,7 +143,7 @@ void pblk_map_erase_rq(struct pblk *pblk, struct nvm_rq *rqd,
 			set_bit(erase_lun, e_line->erase_bitmap);
 			atomic_dec(&e_line->left_eblks);
 
-			*erase_ppa = rqd->ppa_list[i];
+			*erase_ppa = ppa_list[i];
 			erase_ppa->a.blk = e_line->id;
 
 			spin_unlock(&e_line->lock);
diff --git a/drivers/lightnvm/pblk-read.c b/drivers/lightnvm/pblk-read.c
index 49744caaa300..d3ff8c3e9010 100644
--- a/drivers/lightnvm/pblk-read.c
+++ b/drivers/lightnvm/pblk-read.c
@@ -116,10 +116,9 @@ static void pblk_read_check_seq(struct pblk *pblk, struct nvm_rq *rqd,
 
 		if (lba != blba + i) {
 #ifdef CONFIG_NVM_PBLK_DEBUG
-			struct ppa_addr *p;
+			struct ppa_addr *ppa_list = nvm_rq_to_ppa_list(rqd);
 
-			p = (nr_lbas == 1) ? &rqd->ppa_list[i] : &rqd->ppa_addr;
-			print_ppa(pblk, p, "seq", i);
+			print_ppa(pblk, &ppa_list[i], "seq", i);
 #endif
 			pblk_err(pblk, "corrupted read LBA (%llu/%llu)\n",
 							lba, (u64)blba + i);
@@ -148,11 +147,9 @@ static void pblk_read_check_rand(struct pblk *pblk, struct nvm_rq *rqd,
 
 		if (lba != meta_lba) {
 #ifdef CONFIG_NVM_PBLK_DEBUG
-			struct ppa_addr *p;
-			int nr_ppas = rqd->nr_ppas;
+			struct ppa_addr *ppa_list = nvm_rq_to_ppa_list(rqd);
 
-			p = (nr_ppas == 1) ? &rqd->ppa_list[j] : &rqd->ppa_addr;
-			print_ppa(pblk, p, "seq", j);
+			print_ppa(pblk, &ppa_list[j], "seq", j);
 #endif
 			pblk_err(pblk, "corrupted read LBA (%llu/%llu)\n",
 								lba, meta_lba);
diff --git a/drivers/lightnvm/pblk-recovery.c b/drivers/lightnvm/pblk-recovery.c
index cf629ab016ba..3bd2b6b0a359 100644
--- a/drivers/lightnvm/pblk-recovery.c
+++ b/drivers/lightnvm/pblk-recovery.c
@@ -161,6 +161,8 @@ static int pblk_recov_read_oob(struct pblk *pblk, struct pblk_line *line,
 	if (pblk_io_aligned(pblk, rq_ppas))
 		rqd->is_seq = 1;
 
+	ppa_list = nvm_rq_to_ppa_list(rqd);
+
 	for (i = 0; i < rqd->nr_ppas; ) {
 		struct ppa_addr ppa;
 		int pos;
@@ -175,7 +177,7 @@ static int pblk_recov_read_oob(struct pblk *pblk, struct pblk_line *line,
 		}
 
 		for (j = 0; j < pblk->min_write_pgs; j++, i++, r_ptr_int++)
-			rqd->ppa_list[i] =
+			ppa_list[i] =
 				addr_to_gen_ppa(pblk, r_ptr_int, line->id);
 	}
 
@@ -202,7 +204,7 @@ static int pblk_recov_read_oob(struct pblk *pblk, struct pblk_line *line,
 		if (lba == ADDR_EMPTY || lba > pblk->rl.nr_secs)
 			continue;
 
-		pblk_update_map(pblk, lba, rqd->ppa_list[i]);
+		pblk_update_map(pblk, lba, ppa_list[i]);
 	}
 
 	left_ppas -= rq_ppas;
@@ -221,10 +223,11 @@ static void pblk_recov_complete(struct kref *ref)
 
 static void pblk_end_io_recov(struct nvm_rq *rqd)
 {
+	struct ppa_addr *ppa_list = nvm_rq_to_ppa_list(rqd);
 	struct pblk_pad_rq *pad_rq = rqd->private;
 	struct pblk *pblk = pad_rq->pblk;
 
-	pblk_up_page(pblk, rqd->ppa_list, rqd->nr_ppas);
+	pblk_up_page(pblk, ppa_list, rqd->nr_ppas);
 
 	pblk_free_rqd(pblk, rqd, PBLK_WRITE_INT);
 
diff --git a/drivers/lightnvm/pblk-write.c b/drivers/lightnvm/pblk-write.c
index 625ed5a3a020..c20bb7f6d703 100644
--- a/drivers/lightnvm/pblk-write.c
+++ b/drivers/lightnvm/pblk-write.c
@@ -208,15 +208,10 @@ static void pblk_submit_rec(struct work_struct *work)
 	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;
+	struct ppa_addr *ppa_list = nvm_rq_to_ppa_list(rqd);
 
 	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);
 
@@ -273,9 +268,10 @@ static void pblk_end_io_write_meta(struct nvm_rq *rqd)
 	struct pblk_g_ctx *m_ctx = nvm_rq_to_pdu(rqd);
 	struct pblk_line *line = m_ctx->private;
 	struct pblk_emeta *emeta = line->emeta;
+	struct ppa_addr *ppa_list = nvm_rq_to_ppa_list(rqd);
 	int sync;
 
-	pblk_up_page(pblk, rqd->ppa_list, rqd->nr_ppas);
+	pblk_up_page(pblk, ppa_list, rqd->nr_ppas);
 
 	if (rqd->error) {
 		pblk_log_write_err(pblk, rqd);
@@ -375,6 +371,7 @@ int pblk_submit_meta_io(struct pblk *pblk, struct pblk_line *meta_line)
 	struct pblk_line_mgmt *l_mg = &pblk->l_mg;
 	struct pblk_line_meta *lm = &pblk->lm;
 	struct pblk_emeta *emeta = meta_line->emeta;
+	struct ppa_addr *ppa_list;
 	struct pblk_g_ctx *m_ctx;
 	struct bio *bio;
 	struct nvm_rq *rqd;
@@ -409,12 +406,13 @@ int pblk_submit_meta_io(struct pblk *pblk, struct pblk_line *meta_line)
 	if (ret)
 		goto fail_free_bio;
 
+	ppa_list = nvm_rq_to_ppa_list(rqd);
 	for (i = 0; i < rqd->nr_ppas; ) {
 		spin_lock(&meta_line->lock);
 		paddr = __pblk_alloc_page(pblk, meta_line, rq_ppas);
 		spin_unlock(&meta_line->lock);
 		for (j = 0; j < rq_ppas; j++, i++, paddr++)
-			rqd->ppa_list[i] = addr_to_gen_ppa(pblk, paddr, id);
+			ppa_list[i] = addr_to_gen_ppa(pblk, paddr, id);
 	}
 
 	spin_lock(&l_mg->close_lock);
@@ -423,7 +421,7 @@ int pblk_submit_meta_io(struct pblk *pblk, struct pblk_line *meta_line)
 		list_del(&meta_line->list);
 	spin_unlock(&l_mg->close_lock);
 
-	pblk_down_page(pblk, rqd->ppa_list, rqd->nr_ppas);
+	pblk_down_page(pblk, ppa_list, rqd->nr_ppas);
 
 	ret = pblk_submit_io(pblk, rqd);
 	if (ret) {
@@ -434,7 +432,7 @@ int pblk_submit_meta_io(struct pblk *pblk, struct pblk_line *meta_line)
 	return NVM_IO_OK;
 
 fail_rollback:
-	pblk_up_page(pblk, rqd->ppa_list, rqd->nr_ppas);
+	pblk_up_page(pblk, ppa_list, rqd->nr_ppas);
 	spin_lock(&l_mg->close_lock);
 	pblk_dealloc_page(pblk, meta_line, rq_ppas);
 	list_add(&meta_line->list, &meta_line->list);
diff --git a/drivers/lightnvm/pblk.h b/drivers/lightnvm/pblk.h
index 1865ac1560fa..60c509a00574 100644
--- a/drivers/lightnvm/pblk.h
+++ b/drivers/lightnvm/pblk.h
@@ -1362,9 +1362,7 @@ static inline int pblk_boundary_ppa_checks(struct nvm_tgt_dev *tgt_dev,
 static inline int pblk_check_io(struct pblk *pblk, struct nvm_rq *rqd)
 {
 	struct nvm_tgt_dev *dev = pblk->dev;
-	struct ppa_addr *ppa_list;
-
-	ppa_list = (rqd->nr_ppas > 1) ? rqd->ppa_list : &rqd->ppa_addr;
+	struct ppa_addr *ppa_list = nvm_rq_to_ppa_list(rqd);
 
 	if (pblk_boundary_ppa_checks(dev, ppa_list, rqd->nr_ppas)) {
 		WARN_ON(1);
diff --git a/include/linux/lightnvm.h b/include/linux/lightnvm.h
index 77743a02ec0d..50ac5b21297c 100644
--- a/include/linux/lightnvm.h
+++ b/include/linux/lightnvm.h
@@ -320,6 +320,11 @@ static inline void *nvm_rq_to_pdu(struct nvm_rq *rqdata)
 	return rqdata + 1;
 }
 
+static inline struct ppa_addr *nvm_rq_to_ppa_list(struct nvm_rq *rqd)
+{
+	return (rqd->nr_ppas > 1) ? rqd->ppa_list : &rqd->ppa_addr;
+}
+
 enum {
 	NVM_BLK_ST_FREE =	0x1,	/* Free block */
 	NVM_BLK_ST_TGT =	0x2,	/* Block in use by target */
-- 
2.17.1

[GIT PULL 17/45] lightnvm: pblk: allocate line map bitmaps using a mempool

[Matias Bjørling]

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

Line map bitmap allocations are fairly large and can fail. Allocation
failures are fatal to pblk, stopping the write pipeline. To avoid this,
allocate the bitmaps using a mempool instead.

Mempool allocations never fail if called from a process context,
and pblk *should* only allocate map bitmaps in process context,
but keep the failure handling for robustness sake.

Signed-off-by: Hans Holmberg <hans.holmberg@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk-core.c     | 22 +++++++++++++++-------
 drivers/lightnvm/pblk-init.c     | 18 ++++++++++++++++++
 drivers/lightnvm/pblk-recovery.c |  2 +-
 drivers/lightnvm/pblk.h          |  4 ++++
 4 files changed, 38 insertions(+), 8 deletions(-)

diff --git a/drivers/lightnvm/pblk-core.c b/drivers/lightnvm/pblk-core.c
index a31417682c90..e1207a4f9d54 100644
--- a/drivers/lightnvm/pblk-core.c
+++ b/drivers/lightnvm/pblk-core.c
@@ -1049,15 +1049,18 @@ static int pblk_line_init_metadata(struct pblk *pblk, struct pblk_line *line,
 static int pblk_line_alloc_bitmaps(struct pblk *pblk, struct pblk_line *line)
 {
 	struct pblk_line_meta *lm = &pblk->lm;
+	struct pblk_line_mgmt *l_mg = &pblk->l_mg;
 
-	line->map_bitmap = kzalloc(lm->sec_bitmap_len, GFP_KERNEL);
+	line->map_bitmap = mempool_alloc(l_mg->bitmap_pool, GFP_KERNEL);
 	if (!line->map_bitmap)
 		return -ENOMEM;
 
+	memset(line->map_bitmap, 0, lm->sec_bitmap_len);
+
 	/* will be initialized using bb info from map_bitmap */
-	line->invalid_bitmap = kmalloc(lm->sec_bitmap_len, GFP_KERNEL);
+	line->invalid_bitmap = mempool_alloc(l_mg->bitmap_pool, GFP_KERNEL);
 	if (!line->invalid_bitmap) {
-		kfree(line->map_bitmap);
+		mempool_free(line->map_bitmap, l_mg->bitmap_pool);
 		line->map_bitmap = NULL;
 		return -ENOMEM;
 	}
@@ -1243,7 +1246,9 @@ int pblk_line_recov_alloc(struct pblk *pblk, struct pblk_line *line)
 
 void pblk_line_recov_close(struct pblk *pblk, struct pblk_line *line)
 {
-	kfree(line->map_bitmap);
+	struct pblk_line_mgmt *l_mg = &pblk->l_mg;
+
+	mempool_free(line->map_bitmap, l_mg->bitmap_pool);
 	line->map_bitmap = NULL;
 	line->smeta = NULL;
 	line->emeta = NULL;
@@ -1261,8 +1266,11 @@ static void pblk_line_reinit(struct pblk_line *line)
 
 void pblk_line_free(struct pblk_line *line)
 {
-	kfree(line->map_bitmap);
-	kfree(line->invalid_bitmap);
+	struct pblk *pblk = line->pblk;
+	struct pblk_line_mgmt *l_mg = &pblk->l_mg;
+
+	mempool_free(line->map_bitmap, l_mg->bitmap_pool);
+	mempool_free(line->invalid_bitmap, l_mg->bitmap_pool);
 
 	pblk_line_reinit(line);
 }
@@ -1741,7 +1749,7 @@ void pblk_line_close(struct pblk *pblk, struct pblk_line *line)
 
 	list_add_tail(&line->list, move_list);
 
-	kfree(line->map_bitmap);
+	mempool_free(line->map_bitmap, l_mg->bitmap_pool);
 	line->map_bitmap = NULL;
 	line->smeta = NULL;
 	line->emeta = NULL;
diff --git a/drivers/lightnvm/pblk-init.c b/drivers/lightnvm/pblk-init.c
index 8adc8ac8b03c..76a4a271b9cf 100644
--- a/drivers/lightnvm/pblk-init.c
+++ b/drivers/lightnvm/pblk-init.c
@@ -498,6 +498,9 @@ static void pblk_line_mg_free(struct pblk *pblk)
 		pblk_mfree(l_mg->eline_meta[i]->buf, l_mg->emeta_alloc_type);
 		kfree(l_mg->eline_meta[i]);
 	}
+
+	mempool_destroy(l_mg->bitmap_pool);
+	kmem_cache_destroy(l_mg->bitmap_cache);
 }
 
 static void pblk_line_meta_free(struct pblk_line_mgmt *l_mg,
@@ -797,6 +800,17 @@ static int pblk_line_mg_init(struct pblk *pblk)
 			goto fail_free_smeta;
 	}
 
+	l_mg->bitmap_cache = kmem_cache_create("pblk_lm_bitmap",
+			lm->sec_bitmap_len, 0, 0, NULL);
+	if (!l_mg->bitmap_cache)
+		goto fail_free_smeta;
+
+	/* the bitmap pool is used for both valid and map bitmaps */
+	l_mg->bitmap_pool = mempool_create_slab_pool(PBLK_DATA_LINES * 2,
+				l_mg->bitmap_cache);
+	if (!l_mg->bitmap_pool)
+		goto fail_destroy_bitmap_cache;
+
 	/* emeta allocates three different buffers for managing metadata with
 	 * in-memory and in-media layouts
 	 */
@@ -849,6 +863,10 @@ static int pblk_line_mg_init(struct pblk *pblk)
 			kfree(l_mg->eline_meta[i]->buf);
 		kfree(l_mg->eline_meta[i]);
 	}
+
+	mempool_destroy(l_mg->bitmap_pool);
+fail_destroy_bitmap_cache:
+	kmem_cache_destroy(l_mg->bitmap_cache);
 fail_free_smeta:
 	for (i = 0; i < PBLK_DATA_LINES; i++)
 		kfree(l_mg->sline_meta[i]);
diff --git a/drivers/lightnvm/pblk-recovery.c b/drivers/lightnvm/pblk-recovery.c
index 3bd2b6b0a359..eea901d7cebc 100644
--- a/drivers/lightnvm/pblk-recovery.c
+++ b/drivers/lightnvm/pblk-recovery.c
@@ -939,7 +939,7 @@ struct pblk_line *pblk_recov_l2p(struct pblk *pblk)
 			list_move_tail(&line->list, move_list);
 			spin_unlock(&l_mg->gc_lock);
 
-			kfree(line->map_bitmap);
+			mempool_free(line->map_bitmap, l_mg->bitmap_pool);
 			line->map_bitmap = NULL;
 			line->smeta = NULL;
 			line->emeta = NULL;
diff --git a/drivers/lightnvm/pblk.h b/drivers/lightnvm/pblk.h
index 60c509a00574..9068b158de22 100644
--- a/drivers/lightnvm/pblk.h
+++ b/drivers/lightnvm/pblk.h
@@ -530,6 +530,10 @@ struct pblk_line_mgmt {
 	struct pblk_emeta *eline_meta[PBLK_DATA_LINES];
 	unsigned long meta_bitmap;
 
+	/* Cache and mempool for map/invalid bitmaps */
+	struct kmem_cache *bitmap_cache;
+	mempool_t *bitmap_pool;
+
 	/* Helpers for fast bitmap calculations */
 	unsigned long *bb_template;
 	unsigned long *bb_aux;
-- 
2.17.1

[GIT PULL 18/45] lightnvm: pblk: remove unused parameters in pblk_up_rq

[Matias Bjørling]

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

The parameters nr_ppas and ppa_list are not used, so remove them.

Signed-off-by: Hans Holmberg <hans.holmberg@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk-core.c  | 3 +--
 drivers/lightnvm/pblk-write.c | 5 ++---
 drivers/lightnvm/pblk.h       | 3 +--
 3 files changed, 4 insertions(+), 7 deletions(-)

diff --git a/drivers/lightnvm/pblk-core.c b/drivers/lightnvm/pblk-core.c
index e1207a4f9d54..bb1a7cc24cbb 100644
--- a/drivers/lightnvm/pblk-core.c
+++ b/drivers/lightnvm/pblk-core.c
@@ -1929,8 +1929,7 @@ void pblk_up_page(struct pblk *pblk, struct ppa_addr *ppa_list, int nr_ppas)
 	up(&rlun->wr_sem);
 }
 
-void pblk_up_rq(struct pblk *pblk, struct ppa_addr *ppa_list, int nr_ppas,
-		unsigned long *lun_bitmap)
+void pblk_up_rq(struct pblk *pblk, unsigned long *lun_bitmap)
 {
 	struct nvm_tgt_dev *dev = pblk->dev;
 	struct nvm_geo *geo = &dev->geo;
diff --git a/drivers/lightnvm/pblk-write.c b/drivers/lightnvm/pblk-write.c
index c20bb7f6d703..cd579b440b56 100644
--- a/drivers/lightnvm/pblk-write.c
+++ b/drivers/lightnvm/pblk-write.c
@@ -81,8 +81,7 @@ static void pblk_complete_write(struct pblk *pblk, struct nvm_rq *rqd,
 #ifdef CONFIG_NVM_PBLK_DEBUG
 	atomic_long_sub(c_ctx->nr_valid, &pblk->inflight_writes);
 #endif
-
-	pblk_up_rq(pblk, rqd->ppa_list, rqd->nr_ppas, c_ctx->lun_bitmap);
+	pblk_up_rq(pblk, c_ctx->lun_bitmap);
 
 	pos = pblk_rb_sync_init(&pblk->rwb, &flags);
 	if (pos == c_ctx->sentry) {
@@ -215,7 +214,7 @@ static void pblk_submit_rec(struct work_struct *work)
 	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);
+	pblk_up_rq(pblk, c_ctx->lun_bitmap);
 	if (c_ctx->nr_padded)
 		pblk_bio_free_pages(pblk, rqd->bio, c_ctx->nr_valid,
 							c_ctx->nr_padded);
diff --git a/drivers/lightnvm/pblk.h b/drivers/lightnvm/pblk.h
index 9068b158de22..0ca67e8f99d5 100644
--- a/drivers/lightnvm/pblk.h
+++ b/drivers/lightnvm/pblk.h
@@ -827,8 +827,7 @@ void pblk_up_page(struct pblk *pblk, struct ppa_addr *ppa_list, int nr_ppas);
 void pblk_down_rq(struct pblk *pblk, struct ppa_addr *ppa_list, int nr_ppas,
 		  unsigned long *lun_bitmap);
 void pblk_down_page(struct pblk *pblk, struct ppa_addr *ppa_list, int nr_ppas);
-void pblk_up_rq(struct pblk *pblk, struct ppa_addr *ppa_list, int nr_ppas,
-		unsigned long *lun_bitmap);
+void pblk_up_rq(struct pblk *pblk, unsigned long *lun_bitmap);
 int pblk_bio_add_pages(struct pblk *pblk, struct bio *bio, gfp_t flags,
 		       int nr_pages);
 void pblk_bio_free_pages(struct pblk *pblk, struct bio *bio, int off,
-- 
2.17.1

[GIT PULL 19/45] lightnvm: pblk: fix up prints in pblk_read_check_rand

[Matias Bjørling]

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

The prefix when printing ppas in pblk_read_check_rand should be "rnd"
not "seq", so fix this so we can differentiate between lba missmatches
in random and sequential reads. Also change the print order so
we align with pblk_read_check_seq, printing read lba first.

Signed-off-by: Hans Holmberg <hans.holmberg@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk-read.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/lightnvm/pblk-read.c b/drivers/lightnvm/pblk-read.c
index d3ff8c3e9010..829e92857289 100644
--- a/drivers/lightnvm/pblk-read.c
+++ b/drivers/lightnvm/pblk-read.c
@@ -149,10 +149,10 @@ static void pblk_read_check_rand(struct pblk *pblk, struct nvm_rq *rqd,
 #ifdef CONFIG_NVM_PBLK_DEBUG
 			struct ppa_addr *ppa_list = nvm_rq_to_ppa_list(rqd);
 
-			print_ppa(pblk, &ppa_list[j], "seq", j);
+			print_ppa(pblk, &ppa_list[j], "rnd", j);
 #endif
 			pblk_err(pblk, "corrupted read LBA (%llu/%llu)\n",
-								lba, meta_lba);
+							meta_lba, lba);
 			WARN_ON(1);
 		}
 
-- 
2.17.1

[GIT PULL 20/45] lightnvm: pblk: fix write amplificiation calculation

[Matias Bjørling]

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

When the user data counter exceeds 32 bits, the write amplification
calculation does not provide the right value. Fix this by using
div64_u64 in stead of div64.

Fixes: 76758390f83e ("lightnvm: pblk: export write amplification counters to sysfs")
Signed-off-by: Hans Holmberg <hans.holmberg@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk-sysfs.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/drivers/lightnvm/pblk-sysfs.c b/drivers/lightnvm/pblk-sysfs.c
index 9fc3dfa168b4..e56eeeeb914e 100644
--- a/drivers/lightnvm/pblk-sysfs.c
+++ b/drivers/lightnvm/pblk-sysfs.c
@@ -337,7 +337,6 @@ static ssize_t pblk_get_write_amp(u64 user, u64 gc, u64 pad,
 {
 	int sz;
 
-
 	sz = snprintf(page, PAGE_SIZE,
 			"user:%lld gc:%lld pad:%lld WA:",
 			user, gc, pad);
@@ -349,7 +348,7 @@ static ssize_t pblk_get_write_amp(u64 user, u64 gc, u64 pad,
 		u32 wa_frac;
 
 		wa_int = (user + gc + pad) * 100000;
-		wa_int = div_u64(wa_int, user);
+		wa_int = div64_u64(wa_int, user);
 		wa_int = div_u64_rem(wa_int, 100000, &wa_frac);
 
 		sz += snprintf(page + sz, PAGE_SIZE - sz, "%llu.%05u\n",
-- 
2.17.1

[GIT PULL 21/45] lightnvm: pblk: remove debug from pblk_[down/up]_page

[Matias Bjørling]

Remove the debug only iteration within __pblk_down_page, which
then allows us to reduce the number of arguments down to pblk and
the parallel unit from the functions that calls it. Simplifying the
callers logic considerably.

Also, rename the functions pblk_[down/up]_page to
pblk_[down/up]_chunk, to communicate that it manages the write
pointer of the chunk. Note that it also protects the parallel unit
such that at most one chunk is active per parallel unit.

Signed-off-by: Matias Bjørling <mb@lightnvm.io>
Reviewed-by: Javier González <javier@cnexlabs.com>
---
 drivers/lightnvm/pblk-core.c     | 34 +++++++++-----------------------
 drivers/lightnvm/pblk-map.c      |  2 +-
 drivers/lightnvm/pblk-recovery.c |  6 +++---
 drivers/lightnvm/pblk-write.c    |  6 +++---
 drivers/lightnvm/pblk.h          |  6 +++---
 5 files changed, 19 insertions(+), 35 deletions(-)

diff --git a/drivers/lightnvm/pblk-core.c b/drivers/lightnvm/pblk-core.c
index bb1a7cc24cbb..968597d10cc2 100644
--- a/drivers/lightnvm/pblk-core.c
+++ b/drivers/lightnvm/pblk-core.c
@@ -1861,8 +1861,7 @@ void pblk_gen_run_ws(struct pblk *pblk, struct pblk_line *line, void *priv,
 	queue_work(wq, &line_ws->ws);
 }
 
-static void __pblk_down_page(struct pblk *pblk, struct ppa_addr *ppa_list,
-			     int nr_ppas, int pos)
+static void __pblk_down_chunk(struct pblk *pblk, int pos)
 {
 	struct pblk_lun *rlun = &pblk->luns[pos];
 	int ret;
@@ -1871,13 +1870,6 @@ static void __pblk_down_page(struct pblk *pblk, struct ppa_addr *ppa_list,
 	 * Only send one inflight I/O per LUN. Since we map at a page
 	 * granurality, all ppas in the I/O will map to the same LUN
 	 */
-#ifdef CONFIG_NVM_PBLK_DEBUG
-	int i;
-
-	for (i = 1; i < nr_ppas; i++)
-		WARN_ON(ppa_list[0].a.lun != ppa_list[i].a.lun ||
-				ppa_list[0].a.ch != ppa_list[i].a.ch);
-#endif
 
 	ret = down_timeout(&rlun->wr_sem, msecs_to_jiffies(30000));
 	if (ret == -ETIME || ret == -EINTR)
@@ -1885,21 +1877,21 @@ static void __pblk_down_page(struct pblk *pblk, struct ppa_addr *ppa_list,
 				-ret);
 }
 
-void pblk_down_page(struct pblk *pblk, struct ppa_addr *ppa_list, int nr_ppas)
+void pblk_down_chunk(struct pblk *pblk, struct ppa_addr ppa)
 {
 	struct nvm_tgt_dev *dev = pblk->dev;
 	struct nvm_geo *geo = &dev->geo;
-	int pos = pblk_ppa_to_pos(geo, ppa_list[0]);
+	int pos = pblk_ppa_to_pos(geo, ppa);
 
-	__pblk_down_page(pblk, ppa_list, nr_ppas, pos);
+	__pblk_down_chunk(pblk, pos);
 }
 
-void pblk_down_rq(struct pblk *pblk, struct ppa_addr *ppa_list, int nr_ppas,
+void pblk_down_rq(struct pblk *pblk, struct ppa_addr ppa,
 		  unsigned long *lun_bitmap)
 {
 	struct nvm_tgt_dev *dev = pblk->dev;
 	struct nvm_geo *geo = &dev->geo;
-	int pos = pblk_ppa_to_pos(geo, ppa_list[0]);
+	int pos = pblk_ppa_to_pos(geo, ppa);
 
 	/* If the LUN has been locked for this same request, do no attempt to
 	 * lock it again
@@ -1907,23 +1899,15 @@ void pblk_down_rq(struct pblk *pblk, struct ppa_addr *ppa_list, int nr_ppas,
 	if (test_and_set_bit(pos, lun_bitmap))
 		return;
 
-	__pblk_down_page(pblk, ppa_list, nr_ppas, pos);
+	__pblk_down_chunk(pblk, pos);
 }
 
-void pblk_up_page(struct pblk *pblk, struct ppa_addr *ppa_list, int nr_ppas)
+void pblk_up_chunk(struct pblk *pblk, struct ppa_addr ppa)
 {
 	struct nvm_tgt_dev *dev = pblk->dev;
 	struct nvm_geo *geo = &dev->geo;
 	struct pblk_lun *rlun;
-	int pos = pblk_ppa_to_pos(geo, ppa_list[0]);
-
-#ifdef CONFIG_NVM_PBLK_DEBUG
-	int i;
-
-	for (i = 1; i < nr_ppas; i++)
-		WARN_ON(ppa_list[0].a.lun != ppa_list[i].a.lun ||
-				ppa_list[0].a.ch != ppa_list[i].a.ch);
-#endif
+	int pos = pblk_ppa_to_pos(geo, ppa);
 
 	rlun = &pblk->luns[pos];
 	up(&rlun->wr_sem);
diff --git a/drivers/lightnvm/pblk-map.c b/drivers/lightnvm/pblk-map.c
index dc0efb852475..ff677ca6e4e1 100644
--- a/drivers/lightnvm/pblk-map.c
+++ b/drivers/lightnvm/pblk-map.c
@@ -79,7 +79,7 @@ static int pblk_map_page_data(struct pblk *pblk, unsigned int sentry,
 		}
 	}
 
-	pblk_down_rq(pblk, ppa_list, nr_secs, lun_bitmap);
+	pblk_down_rq(pblk, ppa_list[0], lun_bitmap);
 	return 0;
 }
 
diff --git a/drivers/lightnvm/pblk-recovery.c b/drivers/lightnvm/pblk-recovery.c
index eea901d7cebc..cbcc0828517e 100644
--- a/drivers/lightnvm/pblk-recovery.c
+++ b/drivers/lightnvm/pblk-recovery.c
@@ -227,7 +227,7 @@ static void pblk_end_io_recov(struct nvm_rq *rqd)
 	struct pblk_pad_rq *pad_rq = rqd->private;
 	struct pblk *pblk = pad_rq->pblk;
 
-	pblk_up_page(pblk, ppa_list, rqd->nr_ppas);
+	pblk_up_chunk(pblk, ppa_list[0]);
 
 	pblk_free_rqd(pblk, rqd, PBLK_WRITE_INT);
 
@@ -339,12 +339,12 @@ static int pblk_recov_pad_oob(struct pblk *pblk, struct pblk_line *line,
 	}
 
 	kref_get(&pad_rq->ref);
-	pblk_down_page(pblk, rqd->ppa_list, rqd->nr_ppas);
+	pblk_down_chunk(pblk, rqd->ppa_list[0]);
 
 	ret = pblk_submit_io(pblk, rqd);
 	if (ret) {
 		pblk_err(pblk, "I/O submission failed: %d\n", ret);
-		pblk_up_page(pblk, rqd->ppa_list, rqd->nr_ppas);
+		pblk_up_chunk(pblk, rqd->ppa_list[0]);
 		goto fail_free_bio;
 	}
 
diff --git a/drivers/lightnvm/pblk-write.c b/drivers/lightnvm/pblk-write.c
index cd579b440b56..674ba4d1a9f4 100644
--- a/drivers/lightnvm/pblk-write.c
+++ b/drivers/lightnvm/pblk-write.c
@@ -270,7 +270,7 @@ static void pblk_end_io_write_meta(struct nvm_rq *rqd)
 	struct ppa_addr *ppa_list = nvm_rq_to_ppa_list(rqd);
 	int sync;
 
-	pblk_up_page(pblk, ppa_list, rqd->nr_ppas);
+	pblk_up_chunk(pblk, ppa_list[0]);
 
 	if (rqd->error) {
 		pblk_log_write_err(pblk, rqd);
@@ -420,7 +420,7 @@ int pblk_submit_meta_io(struct pblk *pblk, struct pblk_line *meta_line)
 		list_del(&meta_line->list);
 	spin_unlock(&l_mg->close_lock);
 
-	pblk_down_page(pblk, ppa_list, rqd->nr_ppas);
+	pblk_down_chunk(pblk, ppa_list[0]);
 
 	ret = pblk_submit_io(pblk, rqd);
 	if (ret) {
@@ -431,7 +431,7 @@ int pblk_submit_meta_io(struct pblk *pblk, struct pblk_line *meta_line)
 	return NVM_IO_OK;
 
 fail_rollback:
-	pblk_up_page(pblk, ppa_list, rqd->nr_ppas);
+	pblk_up_chunk(pblk, ppa_list[0]);
 	spin_lock(&l_mg->close_lock);
 	pblk_dealloc_page(pblk, meta_line, rq_ppas);
 	list_add(&meta_line->list, &meta_line->list);
diff --git a/drivers/lightnvm/pblk.h b/drivers/lightnvm/pblk.h
index 0ca67e8f99d5..429347bcd1fa 100644
--- a/drivers/lightnvm/pblk.h
+++ b/drivers/lightnvm/pblk.h
@@ -823,10 +823,10 @@ u64 pblk_alloc_page(struct pblk *pblk, struct pblk_line *line, int nr_secs);
 u64 __pblk_alloc_page(struct pblk *pblk, struct pblk_line *line, int nr_secs);
 int pblk_calc_secs(struct pblk *pblk, unsigned long secs_avail,
 		   unsigned long secs_to_flush);
-void pblk_up_page(struct pblk *pblk, struct ppa_addr *ppa_list, int nr_ppas);
-void pblk_down_rq(struct pblk *pblk, struct ppa_addr *ppa_list, int nr_ppas,
+void pblk_down_rq(struct pblk *pblk, struct ppa_addr ppa,
 		  unsigned long *lun_bitmap);
-void pblk_down_page(struct pblk *pblk, struct ppa_addr *ppa_list, int nr_ppas);
+void pblk_down_chunk(struct pblk *pblk, struct ppa_addr ppa);
+void pblk_up_chunk(struct pblk *pblk, struct ppa_addr ppa);
 void pblk_up_rq(struct pblk *pblk, unsigned long *lun_bitmap);
 int pblk_bio_add_pages(struct pblk *pblk, struct bio *bio, gfp_t flags,
 		       int nr_pages);
-- 
2.17.1

[GIT PULL 22/45] lightnvm: pblk: add trace events for chunk states

[Matias Bjørling]

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

Introduce trace points for tracking chunk states in pblk - this is
useful for inspection of the entire state of the drive, and real handy
for both fw and pblk debugging.

Signed-off-by: Hans Holmberg <hans.holmberg@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk-core.c  | 35 ++++++++++++++++++++++-
 drivers/lightnvm/pblk-init.c  |  4 +++
 drivers/lightnvm/pblk-trace.h | 53 +++++++++++++++++++++++++++++++++++
 drivers/lightnvm/pblk-write.c | 10 +++++--
 drivers/lightnvm/pblk.h       |  8 ++++++
 5 files changed, 107 insertions(+), 3 deletions(-)
 create mode 100644 drivers/lightnvm/pblk-trace.h

diff --git a/drivers/lightnvm/pblk-core.c b/drivers/lightnvm/pblk-core.c
index 968597d10cc2..8a5158607467 100644
--- a/drivers/lightnvm/pblk-core.c
+++ b/drivers/lightnvm/pblk-core.c
@@ -16,7 +16,10 @@
  *
  */
 
+#define CREATE_TRACE_POINTS
+
 #include "pblk.h"
+#include "pblk-trace.h"
 
 static void pblk_line_mark_bb(struct work_struct *work)
 {
@@ -93,6 +96,9 @@ static void __pblk_end_io_erase(struct pblk *pblk, struct nvm_rq *rqd)
 		chunk->state = NVM_CHK_ST_FREE;
 	}
 
+	trace_pblk_chunk_state(pblk_disk_name(pblk), &rqd->ppa_addr,
+				chunk->state);
+
 	atomic_dec(&pblk->inflight_io);
 }
 
@@ -477,9 +483,30 @@ int pblk_submit_io(struct pblk *pblk, struct nvm_rq *rqd)
 	return nvm_submit_io(dev, rqd);
 }
 
+void pblk_check_chunk_state_update(struct pblk *pblk, struct nvm_rq *rqd)
+{
+	struct ppa_addr *ppa_list = nvm_rq_to_ppa_list(rqd);
+
+	int i;
+
+	for (i = 0; i < rqd->nr_ppas; i++) {
+		struct ppa_addr *ppa = &ppa_list[i];
+		struct nvm_chk_meta *chunk = pblk_dev_ppa_to_chunk(pblk, *ppa);
+		u64 caddr = pblk_dev_ppa_to_chunk_addr(pblk, *ppa);
+
+		if (caddr == 0)
+			trace_pblk_chunk_state(pblk_disk_name(pblk),
+							ppa, NVM_CHK_ST_OPEN);
+		else if (caddr == chunk->cnlb)
+			trace_pblk_chunk_state(pblk_disk_name(pblk),
+							ppa, NVM_CHK_ST_CLOSED);
+	}
+}
+
 int pblk_submit_io_sync(struct pblk *pblk, struct nvm_rq *rqd)
 {
 	struct nvm_tgt_dev *dev = pblk->dev;
+	int ret;
 
 	atomic_inc(&pblk->inflight_io);
 
@@ -488,7 +515,13 @@ int pblk_submit_io_sync(struct pblk *pblk, struct nvm_rq *rqd)
 		return NVM_IO_ERR;
 #endif
 
-	return nvm_submit_io_sync(dev, rqd);
+	ret = nvm_submit_io_sync(dev, rqd);
+
+	if (trace_pblk_chunk_state_enabled() && !ret &&
+	    rqd->opcode == NVM_OP_PWRITE)
+		pblk_check_chunk_state_update(pblk, rqd);
+
+	return ret;
 }
 
 static void pblk_bio_map_addr_endio(struct bio *bio)
diff --git a/drivers/lightnvm/pblk-init.c b/drivers/lightnvm/pblk-init.c
index 76a4a271b9cf..4f2d9b502028 100644
--- a/drivers/lightnvm/pblk-init.c
+++ b/drivers/lightnvm/pblk-init.c
@@ -19,6 +19,7 @@
  */
 
 #include "pblk.h"
+#include "pblk-trace.h"
 
 static unsigned int write_buffer_size;
 
@@ -664,6 +665,9 @@ static int pblk_setup_line_meta_chk(struct pblk *pblk, struct pblk_line *line,
 		chunk->cnlb = chunk_meta->cnlb;
 		chunk->wp = chunk_meta->wp;
 
+		trace_pblk_chunk_state(pblk_disk_name(pblk), &ppa,
+					chunk->state);
+
 		if (chunk->type & NVM_CHK_TP_SZ_SPEC) {
 			WARN_ONCE(1, "pblk: custom-sized chunks unsupported\n");
 			continue;
diff --git a/drivers/lightnvm/pblk-trace.h b/drivers/lightnvm/pblk-trace.h
new file mode 100644
index 000000000000..d985b729428f
--- /dev/null
+++ b/drivers/lightnvm/pblk-trace.h
@@ -0,0 +1,53 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#undef TRACE_SYSTEM
+#define TRACE_SYSTEM pblk
+
+#if !defined(_TRACE_PBLK_H) || defined(TRACE_HEADER_MULTI_READ)
+#define _TRACE_PBLK_H
+
+#include <linux/tracepoint.h>
+
+struct ppa_addr;
+
+#define show_chunk_flags(state) __print_flags(state, "",	\
+	{ NVM_CHK_ST_FREE,		"FREE",		},	\
+	{ NVM_CHK_ST_CLOSED,		"CLOSED",	},	\
+	{ NVM_CHK_ST_OPEN,		"OPEN",		},	\
+	{ NVM_CHK_ST_OFFLINE,		"OFFLINE",	})
+
+TRACE_EVENT(pblk_chunk_state,
+
+	TP_PROTO(const char *name, struct ppa_addr *ppa, int state),
+
+	TP_ARGS(name, ppa, state),
+
+	TP_STRUCT__entry(
+		__string(name, name)
+		__field(u64, ppa)
+		__field(int, state);
+	),
+
+	TP_fast_assign(
+		__assign_str(name, name);
+		__entry->ppa = ppa->ppa;
+		__entry->state = state;
+	),
+
+	TP_printk("dev=%s grp=%llu pu=%llu chk=%llu state=%s", __get_str(name),
+			(u64)(((struct ppa_addr *)(&__entry->ppa))->m.grp),
+			(u64)(((struct ppa_addr *)(&__entry->ppa))->m.pu),
+			(u64)(((struct ppa_addr *)(&__entry->ppa))->m.chk),
+			show_chunk_flags((int)__entry->state))
+
+);
+
+
+#endif /* !defined(_TRACE_PBLK_H) || defined(TRACE_HEADER_MULTI_READ) */
+
+/* This part must be outside protection */
+
+#undef TRACE_INCLUDE_PATH
+#define TRACE_INCLUDE_PATH ../../../drivers/lightnvm
+#undef TRACE_INCLUDE_FILE
+#define TRACE_INCLUDE_FILE pblk-trace
+#include <trace/define_trace.h>
diff --git a/drivers/lightnvm/pblk-write.c b/drivers/lightnvm/pblk-write.c
index 674ba4d1a9f4..61fe22ccc7a1 100644
--- a/drivers/lightnvm/pblk-write.c
+++ b/drivers/lightnvm/pblk-write.c
@@ -16,6 +16,7 @@
  */
 
 #include "pblk.h"
+#include "pblk-trace.h"
 
 static unsigned long pblk_end_w_bio(struct pblk *pblk, struct nvm_rq *rqd,
 				    struct pblk_c_ctx *c_ctx)
@@ -251,11 +252,13 @@ static void pblk_end_io_write(struct nvm_rq *rqd)
 	if (rqd->error) {
 		pblk_end_w_fail(pblk, rqd);
 		return;
-	}
+	} else {
+		if (trace_pblk_chunk_state_enabled())
+			pblk_check_chunk_state_update(pblk, rqd);
 #ifdef CONFIG_NVM_PBLK_DEBUG
-	else
 		WARN_ONCE(rqd->bio->bi_status, "pblk: corrupted write error\n");
 #endif
+	}
 
 	pblk_complete_write(pblk, rqd, c_ctx);
 	atomic_dec(&pblk->inflight_io);
@@ -276,6 +279,9 @@ static void pblk_end_io_write_meta(struct nvm_rq *rqd)
 		pblk_log_write_err(pblk, rqd);
 		pblk_err(pblk, "metadata I/O failed. Line %d\n", line->id);
 		line->w_err_gc->has_write_err = 1;
+	} else {
+		if (trace_pblk_chunk_state_enabled())
+			pblk_check_chunk_state_update(pblk, rqd);
 	}
 
 	sync = atomic_add_return(rqd->nr_ppas, &emeta->sync);
diff --git a/drivers/lightnvm/pblk.h b/drivers/lightnvm/pblk.h
index 429347bcd1fa..b2746099ca1d 100644
--- a/drivers/lightnvm/pblk.h
+++ b/drivers/lightnvm/pblk.h
@@ -785,6 +785,7 @@ void pblk_log_read_err(struct pblk *pblk, struct nvm_rq *rqd);
 int pblk_submit_io(struct pblk *pblk, struct nvm_rq *rqd);
 int pblk_submit_io_sync(struct pblk *pblk, struct nvm_rq *rqd);
 int pblk_submit_meta_io(struct pblk *pblk, struct pblk_line *meta_line);
+void pblk_check_chunk_state_update(struct pblk *pblk, struct nvm_rq *rqd);
 struct bio *pblk_bio_map_addr(struct pblk *pblk, void *data,
 			      unsigned int nr_secs, unsigned int len,
 			      int alloc_type, gfp_t gfp_mask);
@@ -1427,4 +1428,11 @@ static inline void pblk_setup_uuid(struct pblk *pblk)
 	uuid_le_gen(&uuid);
 	memcpy(pblk->instance_uuid, uuid.b, 16);
 }
+
+static inline char *pblk_disk_name(struct pblk *pblk)
+{
+	struct gendisk *disk = pblk->disk;
+
+	return disk->disk_name;
+}
 #endif /* PBLK_H_ */
-- 
2.17.1

[GIT PULL 23/45] lightnvm: pblk: add trace events for line state changes

[Matias Bjørling]

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

Add trace events for logging for line state changes.

Signed-off-by: Hans Holmberg <hans.holmberg@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk-core.c     | 19 +++++++++++++++++-
 drivers/lightnvm/pblk-gc.c       |  8 ++++++++
 drivers/lightnvm/pblk-init.c     |  3 +++
 drivers/lightnvm/pblk-recovery.c |  6 ++++++
 drivers/lightnvm/pblk-trace.h    | 33 ++++++++++++++++++++++++++++++++
 5 files changed, 68 insertions(+), 1 deletion(-)

diff --git a/drivers/lightnvm/pblk-core.c b/drivers/lightnvm/pblk-core.c
index 8a5158607467..ef52c1647f88 100644
--- a/drivers/lightnvm/pblk-core.c
+++ b/drivers/lightnvm/pblk-core.c
@@ -413,6 +413,9 @@ struct list_head *pblk_line_gc_list(struct pblk *pblk, struct pblk_line *line)
 		}
 	} else {
 		line->state = PBLK_LINESTATE_CORRUPT;
+		trace_pblk_line_state(pblk_disk_name(pblk), line->id,
+					line->state);
+
 		line->gc_group = PBLK_LINEGC_NONE;
 		move_list =  &l_mg->corrupt_list;
 		pblk_err(pblk, "corrupted vsc for line %d, vsc:%d (%d/%d/%d)\n",
@@ -1019,6 +1022,8 @@ static int pblk_line_init_metadata(struct pblk *pblk, struct pblk_line *line,
 		spin_lock(&l_mg->free_lock);
 		spin_lock(&line->lock);
 		line->state = PBLK_LINESTATE_BAD;
+		trace_pblk_line_state(pblk_disk_name(pblk), line->id,
+					line->state);
 		spin_unlock(&line->lock);
 
 		list_add_tail(&line->list, &l_mg->bad_list);
@@ -1166,6 +1171,8 @@ static int pblk_line_init_bb(struct pblk *pblk, struct pblk_line *line,
 		bitmap_weight(line->invalid_bitmap, lm->sec_per_line)) {
 		spin_lock(&line->lock);
 		line->state = PBLK_LINESTATE_BAD;
+		trace_pblk_line_state(pblk_disk_name(pblk), line->id,
+					line->state);
 		spin_unlock(&line->lock);
 
 		list_add_tail(&line->list, &l_mg->bad_list);
@@ -1218,6 +1225,8 @@ static int pblk_line_prepare(struct pblk *pblk, struct pblk_line *line)
 	if (line->state == PBLK_LINESTATE_NEW) {
 		blk_to_erase = pblk_prepare_new_line(pblk, line);
 		line->state = PBLK_LINESTATE_FREE;
+		trace_pblk_line_state(pblk_disk_name(pblk), line->id,
+					line->state);
 	} else {
 		blk_to_erase = blk_in_line;
 	}
@@ -1235,6 +1244,8 @@ static int pblk_line_prepare(struct pblk *pblk, struct pblk_line *line)
 	}
 
 	line->state = PBLK_LINESTATE_OPEN;
+	trace_pblk_line_state(pblk_disk_name(pblk), line->id,
+				line->state);
 
 	atomic_set(&line->left_eblks, blk_to_erase);
 	atomic_set(&line->left_seblks, blk_to_erase);
@@ -1331,6 +1342,8 @@ struct pblk_line *pblk_line_get(struct pblk *pblk)
 	if (unlikely(bit >= lm->blk_per_line)) {
 		spin_lock(&line->lock);
 		line->state = PBLK_LINESTATE_BAD;
+		trace_pblk_line_state(pblk_disk_name(pblk), line->id,
+					line->state);
 		spin_unlock(&line->lock);
 
 		list_add_tail(&line->list, &l_mg->bad_list);
@@ -1650,6 +1663,8 @@ static void __pblk_line_put(struct pblk *pblk, struct pblk_line *line)
 	spin_lock(&line->lock);
 	WARN_ON(line->state != PBLK_LINESTATE_GC);
 	line->state = PBLK_LINESTATE_FREE;
+	trace_pblk_line_state(pblk_disk_name(pblk), line->id,
+					line->state);
 	line->gc_group = PBLK_LINEGC_NONE;
 	pblk_line_free(line);
 
@@ -1779,7 +1794,6 @@ void pblk_line_close(struct pblk *pblk, struct pblk_line *line)
 	WARN_ON(line->state != PBLK_LINESTATE_OPEN);
 	line->state = PBLK_LINESTATE_CLOSED;
 	move_list = pblk_line_gc_list(pblk, line);
-
 	list_add_tail(&line->list, move_list);
 
 	mempool_free(line->map_bitmap, l_mg->bitmap_pool);
@@ -1798,6 +1812,9 @@ void pblk_line_close(struct pblk *pblk, struct pblk_line *line)
 
 	spin_unlock(&line->lock);
 	spin_unlock(&l_mg->gc_lock);
+
+	trace_pblk_line_state(pblk_disk_name(pblk), line->id,
+					line->state);
 }
 
 void pblk_line_close_meta(struct pblk *pblk, struct pblk_line *line)
diff --git a/drivers/lightnvm/pblk-gc.c b/drivers/lightnvm/pblk-gc.c
index 157c2567c9e8..b841d84c4342 100644
--- a/drivers/lightnvm/pblk-gc.c
+++ b/drivers/lightnvm/pblk-gc.c
@@ -16,8 +16,10 @@
  */
 
 #include "pblk.h"
+#include "pblk-trace.h"
 #include <linux/delay.h>
 
+
 static void pblk_gc_free_gc_rq(struct pblk_gc_rq *gc_rq)
 {
 	if (gc_rq->data)
@@ -64,6 +66,8 @@ static void pblk_put_line_back(struct pblk *pblk, struct pblk_line *line)
 	spin_lock(&line->lock);
 	WARN_ON(line->state != PBLK_LINESTATE_GC);
 	line->state = PBLK_LINESTATE_CLOSED;
+	trace_pblk_line_state(pblk_disk_name(pblk), line->id,
+					line->state);
 	move_list = pblk_line_gc_list(pblk, line);
 	spin_unlock(&line->lock);
 
@@ -405,6 +409,8 @@ void pblk_gc_free_full_lines(struct pblk *pblk)
 		spin_lock(&line->lock);
 		WARN_ON(line->state != PBLK_LINESTATE_CLOSED);
 		line->state = PBLK_LINESTATE_GC;
+		trace_pblk_line_state(pblk_disk_name(pblk), line->id,
+					line->state);
 		spin_unlock(&line->lock);
 
 		list_del(&line->list);
@@ -451,6 +457,8 @@ static void pblk_gc_run(struct pblk *pblk)
 		spin_lock(&line->lock);
 		WARN_ON(line->state != PBLK_LINESTATE_CLOSED);
 		line->state = PBLK_LINESTATE_GC;
+		trace_pblk_line_state(pblk_disk_name(pblk), line->id,
+					line->state);
 		spin_unlock(&line->lock);
 
 		list_del(&line->list);
diff --git a/drivers/lightnvm/pblk-init.c b/drivers/lightnvm/pblk-init.c
index 4f2d9b502028..fca37da16266 100644
--- a/drivers/lightnvm/pblk-init.c
+++ b/drivers/lightnvm/pblk-init.c
@@ -985,6 +985,9 @@ static int pblk_lines_init(struct pblk *pblk)
 			goto fail_free_lines;
 
 		nr_free_chks += pblk_setup_line_meta(pblk, line, chunk_meta, i);
+
+		trace_pblk_line_state(pblk_disk_name(pblk), line->id,
+								line->state);
 	}
 
 	if (!nr_free_chks) {
diff --git a/drivers/lightnvm/pblk-recovery.c b/drivers/lightnvm/pblk-recovery.c
index cbcc0828517e..0322ab915ddc 100644
--- a/drivers/lightnvm/pblk-recovery.c
+++ b/drivers/lightnvm/pblk-recovery.c
@@ -15,6 +15,7 @@
  */
 
 #include "pblk.h"
+#include "pblk-trace.h"
 
 int pblk_recov_check_emeta(struct pblk *pblk, struct line_emeta *emeta_buf)
 {
@@ -932,6 +933,8 @@ struct pblk_line *pblk_recov_l2p(struct pblk *pblk)
 
 			spin_lock(&line->lock);
 			line->state = PBLK_LINESTATE_CLOSED;
+			trace_pblk_line_state(pblk_disk_name(pblk), line->id,
+					line->state);
 			move_list = pblk_line_gc_list(pblk, line);
 			spin_unlock(&line->lock);
 
@@ -947,6 +950,9 @@ struct pblk_line *pblk_recov_l2p(struct pblk *pblk)
 			if (open_lines > 1)
 				pblk_err(pblk, "failed to recover L2P\n");
 
+			trace_pblk_line_state(pblk_disk_name(pblk), line->id,
+					line->state);
+
 			open_lines++;
 			line->meta_line = meta_line;
 			data_line = line;
diff --git a/drivers/lightnvm/pblk-trace.h b/drivers/lightnvm/pblk-trace.h
index d985b729428f..f9aa9e25e428 100644
--- a/drivers/lightnvm/pblk-trace.h
+++ b/drivers/lightnvm/pblk-trace.h
@@ -15,6 +15,16 @@ struct ppa_addr;
 	{ NVM_CHK_ST_OPEN,		"OPEN",		},	\
 	{ NVM_CHK_ST_OFFLINE,		"OFFLINE",	})
 
+#define show_line_state(state) __print_symbolic(state,		\
+	{ PBLK_LINESTATE_NEW,		"NEW",		},	\
+	{ PBLK_LINESTATE_FREE,		"FREE",		},	\
+	{ PBLK_LINESTATE_OPEN,		"OPEN",		},	\
+	{ PBLK_LINESTATE_CLOSED,	"CLOSED",	},	\
+	{ PBLK_LINESTATE_GC,		"GC",		},	\
+	{ PBLK_LINESTATE_BAD,		"BAD",		},	\
+	{ PBLK_LINESTATE_CORRUPT,	"CORRUPT"	})
+
+
 TRACE_EVENT(pblk_chunk_state,
 
 	TP_PROTO(const char *name, struct ppa_addr *ppa, int state),
@@ -41,6 +51,29 @@ TRACE_EVENT(pblk_chunk_state,
 
 );
 
+TRACE_EVENT(pblk_line_state,
+
+	TP_PROTO(const char *name, int line, int state),
+
+	TP_ARGS(name, line, state),
+
+	TP_STRUCT__entry(
+		__string(name, name)
+		__field(int, line)
+		__field(int, state);
+	),
+
+	TP_fast_assign(
+		__assign_str(name, name);
+		__entry->line = line;
+		__entry->state = state;
+	),
+
+	TP_printk("dev=%s line=%d state=%s", __get_str(name),
+			(int)__entry->line,
+			show_line_state((int)__entry->state))
+
+);
 
 #endif /* !defined(_TRACE_PBLK_H) || defined(TRACE_HEADER_MULTI_READ) */
 
-- 
2.17.1

[GIT PULL 24/45] lightnvm: pblk: add trace events for pblk state changes

[Matias Bjørling]

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

Add trace events for tracking pblk state changes.

Signed-off-by: Hans Holmberg <hans.holmberg@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk-core.c  |  3 +++
 drivers/lightnvm/pblk-init.c  |  1 +
 drivers/lightnvm/pblk-trace.h | 28 ++++++++++++++++++++++++++++
 3 files changed, 32 insertions(+)

diff --git a/drivers/lightnvm/pblk-core.c b/drivers/lightnvm/pblk-core.c
index ef52c1647f88..c482ef7dd4b5 100644
--- a/drivers/lightnvm/pblk-core.c
+++ b/drivers/lightnvm/pblk-core.c
@@ -1503,6 +1503,7 @@ static void pblk_stop_writes(struct pblk *pblk, struct pblk_line *line)
 
 	pblk_set_space_limit(pblk);
 	pblk->state = PBLK_STATE_STOPPING;
+	trace_pblk_state(pblk_disk_name(pblk), pblk->state);
 }
 
 static void pblk_line_close_meta_sync(struct pblk *pblk)
@@ -1552,6 +1553,7 @@ void __pblk_pipeline_flush(struct pblk *pblk)
 		return;
 	}
 	pblk->state = PBLK_STATE_RECOVERING;
+	trace_pblk_state(pblk_disk_name(pblk), pblk->state);
 	spin_unlock(&l_mg->free_lock);
 
 	pblk_flush_writer(pblk);
@@ -1573,6 +1575,7 @@ void __pblk_pipeline_stop(struct pblk *pblk)
 
 	spin_lock(&l_mg->free_lock);
 	pblk->state = PBLK_STATE_STOPPED;
+	trace_pblk_state(pblk_disk_name(pblk), pblk->state);
 	l_mg->data_line = NULL;
 	l_mg->data_next = NULL;
 	spin_unlock(&l_mg->free_lock);
diff --git a/drivers/lightnvm/pblk-init.c b/drivers/lightnvm/pblk-init.c
index fca37da16266..9aebdee8e4c9 100644
--- a/drivers/lightnvm/pblk-init.c
+++ b/drivers/lightnvm/pblk-init.c
@@ -1109,6 +1109,7 @@ static void *pblk_init(struct nvm_tgt_dev *dev, struct gendisk *tdisk,
 	pblk->dev = dev;
 	pblk->disk = tdisk;
 	pblk->state = PBLK_STATE_RUNNING;
+	trace_pblk_state(pblk_disk_name(pblk), pblk->state);
 	pblk->gc.gc_enabled = 0;
 
 	if (!(geo->version == NVM_OCSSD_SPEC_12 ||
diff --git a/drivers/lightnvm/pblk-trace.h b/drivers/lightnvm/pblk-trace.h
index f9aa9e25e428..c171d0450c07 100644
--- a/drivers/lightnvm/pblk-trace.h
+++ b/drivers/lightnvm/pblk-trace.h
@@ -25,6 +25,13 @@ struct ppa_addr;
 	{ PBLK_LINESTATE_CORRUPT,	"CORRUPT"	})
 
 
+#define show_pblk_state(state) __print_symbolic(state,		\
+	{ PBLK_STATE_RUNNING,		"RUNNING",	},	\
+	{ PBLK_STATE_STOPPING,		"STOPPING",	},	\
+	{ PBLK_STATE_RECOVERING,	"RECOVERING",	},	\
+	{ PBLK_STATE_STOPPED,		"STOPPED"	})
+
+
 TRACE_EVENT(pblk_chunk_state,
 
 	TP_PROTO(const char *name, struct ppa_addr *ppa, int state),
@@ -75,6 +82,27 @@ TRACE_EVENT(pblk_line_state,
 
 );
 
+TRACE_EVENT(pblk_state,
+
+	TP_PROTO(const char *name, int state),
+
+	TP_ARGS(name, state),
+
+	TP_STRUCT__entry(
+		__string(name, name)
+		__field(int, state);
+	),
+
+	TP_fast_assign(
+		__assign_str(name, name);
+		__entry->state = state;
+	),
+
+	TP_printk("dev=%s state=%s", __get_str(name),
+			show_pblk_state((int)__entry->state))
+
+);
+
 #endif /* !defined(_TRACE_PBLK_H) || defined(TRACE_HEADER_MULTI_READ) */
 
 /* This part must be outside protection */
-- 
2.17.1

[GIT PULL 25/45] lightnvm: pblk: add tracing for chunk resets

[Matias Bjørling]

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

Trace state of chunk resets.

Signed-off-by: Hans Holmberg <hans.holmberg@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk-core.c  | 12 ++++++++++++
 drivers/lightnvm/pblk-trace.h | 31 +++++++++++++++++++++++++++++++
 drivers/lightnvm/pblk.h       |  6 ++++++
 3 files changed, 49 insertions(+)

diff --git a/drivers/lightnvm/pblk-core.c b/drivers/lightnvm/pblk-core.c
index c482ef7dd4b5..84f3b4912b92 100644
--- a/drivers/lightnvm/pblk-core.c
+++ b/drivers/lightnvm/pblk-core.c
@@ -90,9 +90,15 @@ static void __pblk_end_io_erase(struct pblk *pblk, struct nvm_rq *rqd)
 	atomic_dec(&line->left_seblks);
 
 	if (rqd->error) {
+		trace_pblk_chunk_reset(pblk_disk_name(pblk),
+				&rqd->ppa_addr, PBLK_CHUNK_RESET_FAILED);
+
 		chunk->state = NVM_CHK_ST_OFFLINE;
 		pblk_mark_bb(pblk, line, rqd->ppa_addr);
 	} else {
+		trace_pblk_chunk_reset(pblk_disk_name(pblk),
+				&rqd->ppa_addr, PBLK_CHUNK_RESET_DONE);
+
 		chunk->state = NVM_CHK_ST_FREE;
 	}
 
@@ -923,6 +929,9 @@ static int pblk_blk_erase_sync(struct pblk *pblk, struct ppa_addr ppa)
 	struct nvm_rq rqd = {NULL};
 	int ret;
 
+	trace_pblk_chunk_reset(pblk_disk_name(pblk), &ppa,
+				PBLK_CHUNK_RESET_START);
+
 	pblk_setup_e_rq(pblk, &rqd, ppa);
 
 	/* The write thread schedules erases so that it minimizes disturbances
@@ -1736,6 +1745,9 @@ int pblk_blk_erase_async(struct pblk *pblk, struct ppa_addr ppa)
 	rqd->end_io = pblk_end_io_erase;
 	rqd->private = pblk;
 
+	trace_pblk_chunk_reset(pblk_disk_name(pblk),
+				&ppa, PBLK_CHUNK_RESET_START);
+
 	/* The write thread schedules erases so that it minimizes disturbances
 	 * with writes. Thus, there is no need to take the LUN semaphore.
 	 */
diff --git a/drivers/lightnvm/pblk-trace.h b/drivers/lightnvm/pblk-trace.h
index c171d0450c07..679e5c458ca6 100644
--- a/drivers/lightnvm/pblk-trace.h
+++ b/drivers/lightnvm/pblk-trace.h
@@ -31,6 +31,37 @@ struct ppa_addr;
 	{ PBLK_STATE_RECOVERING,	"RECOVERING",	},	\
 	{ PBLK_STATE_STOPPED,		"STOPPED"	})
 
+#define show_chunk_erase_state(state) __print_symbolic(state,	\
+	{ PBLK_CHUNK_RESET_START,	"START",	},	\
+	{ PBLK_CHUNK_RESET_DONE,	"OK",		},	\
+	{ PBLK_CHUNK_RESET_FAILED,	"FAILED"	})
+
+
+TRACE_EVENT(pblk_chunk_reset,
+
+	TP_PROTO(const char *name, struct ppa_addr *ppa, int state),
+
+	TP_ARGS(name, ppa, state),
+
+	TP_STRUCT__entry(
+		__string(name, name)
+		__field(u64, ppa)
+		__field(int, state);
+	),
+
+	TP_fast_assign(
+		__assign_str(name, name);
+		__entry->ppa = ppa->ppa;
+		__entry->state = state;
+	),
+
+	TP_printk("dev=%s grp=%llu pu=%llu chk=%llu state=%s", __get_str(name),
+			(u64)(((struct ppa_addr *)(&__entry->ppa))->m.grp),
+			(u64)(((struct ppa_addr *)(&__entry->ppa))->m.pu),
+			(u64)(((struct ppa_addr *)(&__entry->ppa))->m.chk),
+			show_chunk_erase_state((int)__entry->state))
+
+);
 
 TRACE_EVENT(pblk_chunk_state,
 
diff --git a/drivers/lightnvm/pblk.h b/drivers/lightnvm/pblk.h
index b2746099ca1d..7f4e46dfb0d7 100644
--- a/drivers/lightnvm/pblk.h
+++ b/drivers/lightnvm/pblk.h
@@ -79,6 +79,12 @@ enum {
 	PBLK_BLK_ST_CLOSED =	0x2,
 };
 
+enum {
+	PBLK_CHUNK_RESET_START,
+	PBLK_CHUNK_RESET_DONE,
+	PBLK_CHUNK_RESET_FAILED,
+};
+
 struct pblk_sec_meta {
 	u64 reserved;
 	__le64 lba;
-- 
2.17.1

[GIT PULL 26/45] lightnvm: move ppa transformations to core

[Matias Bjørling]

From: Javier González <javier@javigon.com>

Continuing the effort of moving 1.2 and 2.0 specific code to core, move
64_to_32 and 32_to_64 ppa helpers from pblk to core.

Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk.h  | 78 ++-----------------------------------
 include/linux/lightnvm.h | 83 ++++++++++++++++++++++++++++++++++++++++
 2 files changed, 87 insertions(+), 74 deletions(-)

diff --git a/drivers/lightnvm/pblk.h b/drivers/lightnvm/pblk.h
index 7f4e46dfb0d7..a2cc581951ef 100644
--- a/drivers/lightnvm/pblk.h
+++ b/drivers/lightnvm/pblk.h
@@ -1102,86 +1102,16 @@ static inline u64 pblk_dev_ppa_to_line_addr(struct pblk *pblk,
 
 static inline struct ppa_addr pblk_ppa32_to_ppa64(struct pblk *pblk, u32 ppa32)
 {
-	struct ppa_addr ppa64;
+	struct nvm_tgt_dev *dev = pblk->dev;
 
-	ppa64.ppa = 0;
-
-	if (ppa32 == -1) {
-		ppa64.ppa = ADDR_EMPTY;
-	} else if (ppa32 & (1U << 31)) {
-		ppa64.c.line = ppa32 & ((~0U) >> 1);
-		ppa64.c.is_cached = 1;
-	} else {
-		struct nvm_tgt_dev *dev = pblk->dev;
-		struct nvm_geo *geo = &dev->geo;
-
-		if (geo->version == NVM_OCSSD_SPEC_12) {
-			struct nvm_addrf_12 *ppaf =
-					(struct nvm_addrf_12 *)&pblk->addrf;
-
-			ppa64.g.ch = (ppa32 & ppaf->ch_mask) >>
-							ppaf->ch_offset;
-			ppa64.g.lun = (ppa32 & ppaf->lun_mask) >>
-							ppaf->lun_offset;
-			ppa64.g.blk = (ppa32 & ppaf->blk_mask) >>
-							ppaf->blk_offset;
-			ppa64.g.pg = (ppa32 & ppaf->pg_mask) >>
-							ppaf->pg_offset;
-			ppa64.g.pl = (ppa32 & ppaf->pln_mask) >>
-							ppaf->pln_offset;
-			ppa64.g.sec = (ppa32 & ppaf->sec_mask) >>
-							ppaf->sec_offset;
-		} else {
-			struct nvm_addrf *lbaf = &pblk->addrf;
-
-			ppa64.m.grp = (ppa32 & lbaf->ch_mask) >>
-							lbaf->ch_offset;
-			ppa64.m.pu = (ppa32 & lbaf->lun_mask) >>
-							lbaf->lun_offset;
-			ppa64.m.chk = (ppa32 & lbaf->chk_mask) >>
-							lbaf->chk_offset;
-			ppa64.m.sec = (ppa32 & lbaf->sec_mask) >>
-							lbaf->sec_offset;
-		}
-	}
-
-	return ppa64;
+	return nvm_ppa32_to_ppa64(dev->parent, &pblk->addrf, ppa32);
 }
 
 static inline u32 pblk_ppa64_to_ppa32(struct pblk *pblk, struct ppa_addr ppa64)
 {
-	u32 ppa32 = 0;
+	struct nvm_tgt_dev *dev = pblk->dev;
 
-	if (ppa64.ppa == ADDR_EMPTY) {
-		ppa32 = ~0U;
-	} else if (ppa64.c.is_cached) {
-		ppa32 |= ppa64.c.line;
-		ppa32 |= 1U << 31;
-	} else {
-		struct nvm_tgt_dev *dev = pblk->dev;
-		struct nvm_geo *geo = &dev->geo;
-
-		if (geo->version == NVM_OCSSD_SPEC_12) {
-			struct nvm_addrf_12 *ppaf =
-					(struct nvm_addrf_12 *)&pblk->addrf;
-
-			ppa32 |= ppa64.g.ch << ppaf->ch_offset;
-			ppa32 |= ppa64.g.lun << ppaf->lun_offset;
-			ppa32 |= ppa64.g.blk << ppaf->blk_offset;
-			ppa32 |= ppa64.g.pg << ppaf->pg_offset;
-			ppa32 |= ppa64.g.pl << ppaf->pln_offset;
-			ppa32 |= ppa64.g.sec << ppaf->sec_offset;
-		} else {
-			struct nvm_addrf *lbaf = &pblk->addrf;
-
-			ppa32 |= ppa64.m.grp << lbaf->ch_offset;
-			ppa32 |= ppa64.m.pu << lbaf->lun_offset;
-			ppa32 |= ppa64.m.chk << lbaf->chk_offset;
-			ppa32 |= ppa64.m.sec << lbaf->sec_offset;
-		}
-	}
-
-	return ppa32;
+	return nvm_ppa64_to_ppa32(dev->parent, &pblk->addrf, ppa64);
 }
 
 static inline struct ppa_addr pblk_trans_map_get(struct pblk *pblk,
diff --git a/include/linux/lightnvm.h b/include/linux/lightnvm.h
index 50ac5b21297c..eb7300c20f24 100644
--- a/include/linux/lightnvm.h
+++ b/include/linux/lightnvm.h
@@ -511,6 +511,89 @@ static inline u64 dev_to_chunk_addr(struct nvm_dev *dev, void *addrf,
 	return caddr;
 }
 
+static inline struct ppa_addr nvm_ppa32_to_ppa64(struct nvm_dev *dev,
+						 void *addrf, u32 ppa32)
+{
+	struct ppa_addr ppa64;
+
+	ppa64.ppa = 0;
+
+	if (ppa32 == -1) {
+		ppa64.ppa = ADDR_EMPTY;
+	} else if (ppa32 & (1U << 31)) {
+		ppa64.c.line = ppa32 & ((~0U) >> 1);
+		ppa64.c.is_cached = 1;
+	} else {
+		struct nvm_geo *geo = &dev->geo;
+
+		if (geo->version == NVM_OCSSD_SPEC_12) {
+			struct nvm_addrf_12 *ppaf = addrf;
+
+			ppa64.g.ch = (ppa32 & ppaf->ch_mask) >>
+							ppaf->ch_offset;
+			ppa64.g.lun = (ppa32 & ppaf->lun_mask) >>
+							ppaf->lun_offset;
+			ppa64.g.blk = (ppa32 & ppaf->blk_mask) >>
+							ppaf->blk_offset;
+			ppa64.g.pg = (ppa32 & ppaf->pg_mask) >>
+							ppaf->pg_offset;
+			ppa64.g.pl = (ppa32 & ppaf->pln_mask) >>
+							ppaf->pln_offset;
+			ppa64.g.sec = (ppa32 & ppaf->sec_mask) >>
+							ppaf->sec_offset;
+		} else {
+			struct nvm_addrf *lbaf = addrf;
+
+			ppa64.m.grp = (ppa32 & lbaf->ch_mask) >>
+							lbaf->ch_offset;
+			ppa64.m.pu = (ppa32 & lbaf->lun_mask) >>
+							lbaf->lun_offset;
+			ppa64.m.chk = (ppa32 & lbaf->chk_mask) >>
+							lbaf->chk_offset;
+			ppa64.m.sec = (ppa32 & lbaf->sec_mask) >>
+							lbaf->sec_offset;
+		}
+	}
+
+	return ppa64;
+}
+
+static inline u32 nvm_ppa64_to_ppa32(struct nvm_dev *dev,
+				     void *addrf, struct ppa_addr ppa64)
+{
+	u32 ppa32 = 0;
+
+	if (ppa64.ppa == ADDR_EMPTY) {
+		ppa32 = ~0U;
+	} else if (ppa64.c.is_cached) {
+		ppa32 |= ppa64.c.line;
+		ppa32 |= 1U << 31;
+	} else {
+		struct nvm_geo *geo = &dev->geo;
+
+		if (geo->version == NVM_OCSSD_SPEC_12) {
+			struct nvm_addrf_12 *ppaf = addrf;
+
+			ppa32 |= ppa64.g.ch << ppaf->ch_offset;
+			ppa32 |= ppa64.g.lun << ppaf->lun_offset;
+			ppa32 |= ppa64.g.blk << ppaf->blk_offset;
+			ppa32 |= ppa64.g.pg << ppaf->pg_offset;
+			ppa32 |= ppa64.g.pl << ppaf->pln_offset;
+			ppa32 |= ppa64.g.sec << ppaf->sec_offset;
+		} else {
+			struct nvm_addrf *lbaf = addrf;
+
+			ppa32 |= ppa64.m.grp << lbaf->ch_offset;
+			ppa32 |= ppa64.m.pu << lbaf->lun_offset;
+			ppa32 |= ppa64.m.chk << lbaf->chk_offset;
+			ppa32 |= ppa64.m.sec << lbaf->sec_offset;
+		}
+	}
+
+	return ppa32;
+}
+
+
 typedef blk_qc_t (nvm_tgt_make_rq_fn)(struct request_queue *, struct bio *);
 typedef sector_t (nvm_tgt_capacity_fn)(void *);
 typedef void *(nvm_tgt_init_fn)(struct nvm_tgt_dev *, struct gendisk *,
-- 
2.17.1

[GIT PULL 27/45] lightnvm: pblk: calculate line pad distance in helper

[Matias Bjørling]

From: Javier González <javier@javigon.com>

If a line is padded, calculate the pad distance directly on the helper
being used for this purpose.

Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk-recovery.c | 13 ++++++++++---
 drivers/lightnvm/pblk.h          |  8 --------
 2 files changed, 10 insertions(+), 11 deletions(-)

diff --git a/drivers/lightnvm/pblk-recovery.c b/drivers/lightnvm/pblk-recovery.c
index 0322ab915ddc..8036c3eb6372 100644
--- a/drivers/lightnvm/pblk-recovery.c
+++ b/drivers/lightnvm/pblk-recovery.c
@@ -380,6 +380,15 @@ static int pblk_recov_pad_oob(struct pblk *pblk, struct pblk_line *line,
 	return ret;
 }
 
+static int pblk_pad_distance(struct pblk *pblk, struct pblk_line *line)
+{
+	struct nvm_tgt_dev *dev = pblk->dev;
+	struct nvm_geo *geo = &dev->geo;
+	int distance = geo->mw_cunits * geo->all_luns * geo->ws_opt;
+
+	return (distance > line->left_msecs) ? line->left_msecs : distance;
+}
+
 /* When this function is called, it means that not all upper pages have been
  * written in a page that contains valid data. In order to recover this data, we
  * first find the write pointer on the device, then we pad all necessary
@@ -495,9 +504,7 @@ static int pblk_recov_scan_all_oob(struct pblk *pblk, struct pblk_line *line,
 		line->left_msecs += nr_error_bits;
 		bitmap_clear(line->map_bitmap, line->cur_sec, nr_error_bits);
 
-		pad_secs = pblk_pad_distance(pblk);
-		if (pad_secs > line->left_msecs)
-			pad_secs = line->left_msecs;
+		pad_secs = pblk_pad_distance(pblk, line);
 
 		ret = pblk_recov_pad_oob(pblk, line, pad_secs);
 		if (ret)
diff --git a/drivers/lightnvm/pblk.h b/drivers/lightnvm/pblk.h
index a2cc581951ef..d123cff82589 100644
--- a/drivers/lightnvm/pblk.h
+++ b/drivers/lightnvm/pblk.h
@@ -986,14 +986,6 @@ static inline int pblk_line_vsc(struct pblk_line *line)
 	return le32_to_cpu(*line->vsc);
 }
 
-static inline int pblk_pad_distance(struct pblk *pblk)
-{
-	struct nvm_tgt_dev *dev = pblk->dev;
-	struct nvm_geo *geo = &dev->geo;
-
-	return geo->mw_cunits * geo->all_luns * geo->ws_opt;
-}
-
 static inline int pblk_ppa_to_line_id(struct ppa_addr p)
 {
 	return p.a.blk;
-- 
2.17.1

[GIT PULL 28/45] lightnvm: pblk: stop recreating global caches

[Matias Bjørling]

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

Pblk should not create a set of global caches every time
a pblk instance is created. The global caches should be
made available only when there is one or more pblk instances.

This patch bundles the global caches together with a kref
keeping track of whether the caches should be available or not.

Also, turn the global pblk lock into a mutex that explicitly
protects the caches (as this was the only purpose of the lock).

Signed-off-by: Hans Holmberg <hans.holmberg@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk-init.c | 132 +++++++++++++++++++++++------------
 1 file changed, 86 insertions(+), 46 deletions(-)

diff --git a/drivers/lightnvm/pblk-init.c b/drivers/lightnvm/pblk-init.c
index 9aebdee8e4c9..fb66bc84d5ca 100644
--- a/drivers/lightnvm/pblk-init.c
+++ b/drivers/lightnvm/pblk-init.c
@@ -26,9 +26,24 @@ static 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);
+struct pblk_global_caches {
+	struct kmem_cache	*ws;
+	struct kmem_cache	*rec;
+	struct kmem_cache	*g_rq;
+	struct kmem_cache	*w_rq;
+
+	struct kref		kref;
+
+	struct mutex		mutex; /* Ensures consistency between
+					* caches and kref
+					*/
+};
+
+static struct pblk_global_caches pblk_caches = {
+	.mutex = __MUTEX_INITIALIZER(pblk_caches.mutex),
+	.kref = KREF_INIT(0),
+};
+
 struct bio_set pblk_bio_set;
 
 static int pblk_rw_io(struct request_queue *q, struct pblk *pblk,
@@ -307,53 +322,80 @@ static int pblk_set_addrf(struct pblk *pblk)
 	return 0;
 }
 
-static int pblk_init_global_caches(struct pblk *pblk)
+static int pblk_create_global_caches(void)
 {
-	down_write(&pblk_lock);
-	pblk_ws_cache = kmem_cache_create("pblk_blk_ws",
+
+	pblk_caches.ws = kmem_cache_create("pblk_blk_ws",
 				sizeof(struct pblk_line_ws), 0, 0, NULL);
-	if (!pblk_ws_cache) {
-		up_write(&pblk_lock);
+	if (!pblk_caches.ws)
 		return -ENOMEM;
-	}
 
-	pblk_rec_cache = kmem_cache_create("pblk_rec",
+	pblk_caches.rec = kmem_cache_create("pblk_rec",
 				sizeof(struct pblk_rec_ctx), 0, 0, NULL);
-	if (!pblk_rec_cache) {
-		kmem_cache_destroy(pblk_ws_cache);
-		up_write(&pblk_lock);
-		return -ENOMEM;
-	}
+	if (!pblk_caches.rec)
+		goto fail_destroy_ws;
 
-	pblk_g_rq_cache = kmem_cache_create("pblk_g_rq", pblk_g_rq_size,
+	pblk_caches.g_rq = kmem_cache_create("pblk_g_rq", pblk_g_rq_size,
 				0, 0, NULL);
-	if (!pblk_g_rq_cache) {
-		kmem_cache_destroy(pblk_ws_cache);
-		kmem_cache_destroy(pblk_rec_cache);
-		up_write(&pblk_lock);
-		return -ENOMEM;
-	}
+	if (!pblk_caches.g_rq)
+		goto fail_destroy_rec;
 
-	pblk_w_rq_cache = kmem_cache_create("pblk_w_rq", pblk_w_rq_size,
+	pblk_caches.w_rq = kmem_cache_create("pblk_w_rq", pblk_w_rq_size,
 				0, 0, NULL);
-	if (!pblk_w_rq_cache) {
-		kmem_cache_destroy(pblk_ws_cache);
-		kmem_cache_destroy(pblk_rec_cache);
-		kmem_cache_destroy(pblk_g_rq_cache);
-		up_write(&pblk_lock);
-		return -ENOMEM;
-	}
-	up_write(&pblk_lock);
+	if (!pblk_caches.w_rq)
+		goto fail_destroy_g_rq;
 
 	return 0;
+
+fail_destroy_g_rq:
+	kmem_cache_destroy(pblk_caches.g_rq);
+fail_destroy_rec:
+	kmem_cache_destroy(pblk_caches.rec);
+fail_destroy_ws:
+	kmem_cache_destroy(pblk_caches.ws);
+
+	return -ENOMEM;
+}
+
+static int pblk_get_global_caches(void)
+{
+	int ret;
+
+	mutex_lock(&pblk_caches.mutex);
+
+	if (kref_read(&pblk_caches.kref) > 0) {
+		kref_get(&pblk_caches.kref);
+		mutex_unlock(&pblk_caches.mutex);
+		return 0;
+	}
+
+	ret = pblk_create_global_caches();
+
+	if (!ret)
+		kref_get(&pblk_caches.kref);
+
+	mutex_unlock(&pblk_caches.mutex);
+
+	return ret;
+}
+
+static void pblk_destroy_global_caches(struct kref *ref)
+{
+	struct pblk_global_caches *c;
+
+	c = container_of(ref, struct pblk_global_caches, kref);
+
+	kmem_cache_destroy(c->ws);
+	kmem_cache_destroy(c->rec);
+	kmem_cache_destroy(c->g_rq);
+	kmem_cache_destroy(c->w_rq);
 }
 
-static void pblk_free_global_caches(struct pblk *pblk)
+static void pblk_put_global_caches(void)
 {
-	kmem_cache_destroy(pblk_ws_cache);
-	kmem_cache_destroy(pblk_rec_cache);
-	kmem_cache_destroy(pblk_g_rq_cache);
-	kmem_cache_destroy(pblk_w_rq_cache);
+	mutex_lock(&pblk_caches.mutex);
+	kref_put(&pblk_caches.kref, pblk_destroy_global_caches);
+	mutex_unlock(&pblk_caches.mutex);
 }
 
 static int pblk_core_init(struct pblk *pblk)
@@ -382,7 +424,7 @@ static int pblk_core_init(struct pblk *pblk)
 	if (!pblk->pad_dist)
 		return -ENOMEM;
 
-	if (pblk_init_global_caches(pblk))
+	if (pblk_get_global_caches())
 		goto fail_free_pad_dist;
 
 	/* Internal bios can be at most the sectors signaled by the device. */
@@ -391,27 +433,27 @@ static int pblk_core_init(struct pblk *pblk)
 		goto free_global_caches;
 
 	ret = mempool_init_slab_pool(&pblk->gen_ws_pool, PBLK_GEN_WS_POOL_SIZE,
-				     pblk_ws_cache);
+				     pblk_caches.ws);
 	if (ret)
 		goto free_page_bio_pool;
 
 	ret = mempool_init_slab_pool(&pblk->rec_pool, geo->all_luns,
-				     pblk_rec_cache);
+				     pblk_caches.rec);
 	if (ret)
 		goto free_gen_ws_pool;
 
 	ret = mempool_init_slab_pool(&pblk->r_rq_pool, geo->all_luns,
-				     pblk_g_rq_cache);
+				     pblk_caches.g_rq);
 	if (ret)
 		goto free_rec_pool;
 
 	ret = mempool_init_slab_pool(&pblk->e_rq_pool, geo->all_luns,
-				     pblk_g_rq_cache);
+				     pblk_caches.g_rq);
 	if (ret)
 		goto free_r_rq_pool;
 
 	ret = mempool_init_slab_pool(&pblk->w_rq_pool, geo->all_luns,
-				     pblk_w_rq_cache);
+				     pblk_caches.w_rq);
 	if (ret)
 		goto free_e_rq_pool;
 
@@ -457,7 +499,7 @@ static int pblk_core_init(struct pblk *pblk)
 free_page_bio_pool:
 	mempool_exit(&pblk->page_bio_pool);
 free_global_caches:
-	pblk_free_global_caches(pblk);
+	pblk_put_global_caches();
 fail_free_pad_dist:
 	kfree(pblk->pad_dist);
 	return -ENOMEM;
@@ -481,7 +523,7 @@ static void pblk_core_free(struct pblk *pblk)
 	mempool_exit(&pblk->e_rq_pool);
 	mempool_exit(&pblk->w_rq_pool);
 
-	pblk_free_global_caches(pblk);
+	pblk_put_global_caches();
 	kfree(pblk->pad_dist);
 }
 
@@ -1074,7 +1116,6 @@ static void pblk_exit(void *private, bool graceful)
 {
 	struct pblk *pblk = private;
 
-	down_write(&pblk_lock);
 	pblk_gc_exit(pblk, graceful);
 	pblk_tear_down(pblk, graceful);
 
@@ -1083,7 +1124,6 @@ static void pblk_exit(void *private, bool graceful)
 #endif
 
 	pblk_free(pblk);
-	up_write(&pblk_lock);
 }
 
 static sector_t pblk_capacity(void *private)
-- 
2.17.1

[GIT PULL 29/45] lightnvm: pblk: fix mapping issue on failed writes

[Matias Bjørling]

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

On 1.2-devices, the mapping-out of remaning sectors in the
failed-write's block can result in an infinite loop,
stalling the write pipeline, fix this.

Fixes: 6a3abf5beef6 ("lightnvm: pblk: rework write error recovery path")
Signed-off-by: Hans Holmberg <hans.holmberg@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk-write.c | 12 +-----------
 include/linux/lightnvm.h      | 36 +++++++++++++++++++++++++++++++++++
 2 files changed, 37 insertions(+), 11 deletions(-)

diff --git a/drivers/lightnvm/pblk-write.c b/drivers/lightnvm/pblk-write.c
index 61fe22ccc7a1..9554febee480 100644
--- a/drivers/lightnvm/pblk-write.c
+++ b/drivers/lightnvm/pblk-write.c
@@ -106,8 +106,6 @@ static void pblk_complete_write(struct pblk *pblk, struct nvm_rq *rqd,
 /* 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;
@@ -125,15 +123,7 @@ static void pblk_map_remaining(struct pblk *pblk, struct ppa_addr *ppa)
 		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;
-		}
+		done = nvm_next_ppa_in_chk(pblk->dev, &map_ppa);
 	}
 
 	line->w_err_gc->has_write_err = 1;
diff --git a/include/linux/lightnvm.h b/include/linux/lightnvm.h
index eb7300c20f24..2fdeac1a420d 100644
--- a/include/linux/lightnvm.h
+++ b/include/linux/lightnvm.h
@@ -593,6 +593,42 @@ static inline u32 nvm_ppa64_to_ppa32(struct nvm_dev *dev,
 	return ppa32;
 }
 
+static inline int nvm_next_ppa_in_chk(struct nvm_tgt_dev *dev,
+				      struct ppa_addr *ppa)
+{
+	struct nvm_geo *geo = &dev->geo;
+	int last = 0;
+
+	if (geo->version == NVM_OCSSD_SPEC_12) {
+		int sec = ppa->g.sec;
+
+		sec++;
+		if (sec == geo->ws_min) {
+			int pg = ppa->g.pg;
+
+			sec = 0;
+			pg++;
+			if (pg == geo->num_pg) {
+				int pl = ppa->g.pl;
+
+				pg = 0;
+				pl++;
+				if (pl == geo->num_pln)
+					last = 1;
+
+				ppa->g.pl = pl;
+			}
+			ppa->g.pg = pg;
+		}
+		ppa->g.sec = sec;
+	} else {
+		ppa->m.sec++;
+		if (ppa->m.sec == geo->clba)
+			last = 1;
+	}
+
+	return last;
+}
 
 typedef blk_qc_t (nvm_tgt_make_rq_fn)(struct request_queue *, struct bio *);
 typedef sector_t (nvm_tgt_capacity_fn)(void *);
-- 
2.17.1

[GIT PULL 30/45] lightnvm: pblk: fix two sleep-in-atomic-context bugs

[Matias Bjørling]

From: Jia-Ju Bai <baijiaju1990@gmail.com>

The driver may sleep with holding a spinlock.

The function call paths (from bottom to top) in Linux-4.16 are:

[FUNC] nvm_dev_dma_alloc(GFP_KERNEL)
drivers/lightnvm/pblk-core.c, 754:
	nvm_dev_dma_alloc in pblk_line_submit_smeta_io
drivers/lightnvm/pblk-core.c, 1048:
	pblk_line_submit_smeta_io in pblk_line_init_bb
drivers/lightnvm/pblk-core.c, 1434:
	pblk_line_init_bb in pblk_line_replace_data
drivers/lightnvm/pblk-recovery.c, 980:
	pblk_line_replace_data in pblk_recov_l2p
drivers/lightnvm/pblk-recovery.c, 976:
	spin_lock in pblk_recov_l2p

[FUNC] bio_map_kern(GFP_KERNEL)
drivers/lightnvm/pblk-core.c, 762:
	bio_map_kern in pblk_line_submit_smeta_io
drivers/lightnvm/pblk-core.c, 1048:
	pblk_line_submit_smeta_io in pblk_line_init_bb
drivers/lightnvm/pblk-core.c, 1434:
	pblk_line_init_bb in pblk_line_replace_data
drivers/lightnvm/pblk-recovery.c, 980:
	pblk_line_replace_data in pblk_recov_l2p
drivers/lightnvm/pblk-recovery.c, 976:
	spin_lock in pblk_recov_l2p

To fix these bugs, the call to pblk_line_replace_data()
is moved out of the spinlock protection.

These bugs are found by my static analysis tool DSAC.

Signed-off-by: Jia-Ju Bai <baijiaju1990@gmail.com>
Reviewed-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk-recovery.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/drivers/lightnvm/pblk-recovery.c b/drivers/lightnvm/pblk-recovery.c
index 8036c3eb6372..2526722304bb 100644
--- a/drivers/lightnvm/pblk-recovery.c
+++ b/drivers/lightnvm/pblk-recovery.c
@@ -966,12 +966,14 @@ struct pblk_line *pblk_recov_l2p(struct pblk *pblk)
 		}
 	}
 
-	spin_lock(&l_mg->free_lock);
 	if (!open_lines) {
+		spin_lock(&l_mg->free_lock);
 		WARN_ON_ONCE(!test_and_clear_bit(meta_line,
 							&l_mg->meta_bitmap));
+		spin_unlock(&l_mg->free_lock);
 		pblk_line_replace_data(pblk);
 	} else {
+		spin_lock(&l_mg->free_lock);
 		/* Allocate next line for preparation */
 		l_mg->data_next = pblk_line_get(pblk);
 		if (l_mg->data_next) {
@@ -979,8 +981,8 @@ struct pblk_line *pblk_recov_l2p(struct pblk *pblk)
 			l_mg->data_next->type = PBLK_LINETYPE_DATA;
 			is_next = 1;
 		}
+		spin_unlock(&l_mg->free_lock);
 	}
-	spin_unlock(&l_mg->free_lock);
 
 	if (is_next)
 		pblk_line_erase(pblk, l_mg->data_next);
-- 
2.17.1

[GIT PULL 31/45] lightnvm: use internal allocation for chunk log page

[Matias Bjørling]

From: Javier González <javier@javigon.com>

The lightnvm subsystem provides helpers to retrieve chunk metadata,
where the target needs to provide a buffer to store the metadata. An
implicit assumption is that this buffer is contiguous and can be used to
retrieve the data from the device. If the device exposes too many
chunks, then kmalloc might fail, thus failing instance creation.

This patch removes this assumption by implementing an internal buffer in
the lightnvm subsystem to retrieve chunk metadata. Targets can then
use virtual memory allocations. Since this is a target API change, adapt
pblk accordingly.

Signed-off-by: Javier González <javier@cnexlabs.com>
Reviewed-by: Hans Holmberg <hans.holmberg@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk-core.c |  4 ++--
 drivers/lightnvm/pblk-init.c |  2 +-
 drivers/nvme/host/lightnvm.c | 25 +++++++++++++++++--------
 3 files changed, 20 insertions(+), 11 deletions(-)

diff --git a/drivers/lightnvm/pblk-core.c b/drivers/lightnvm/pblk-core.c
index 84f3b4912b92..875f3cf615ac 100644
--- a/drivers/lightnvm/pblk-core.c
+++ b/drivers/lightnvm/pblk-core.c
@@ -120,7 +120,7 @@ static void pblk_end_io_erase(struct nvm_rq *rqd)
 /*
  * Get information for all chunks from the device.
  *
- * The caller is responsible for freeing the returned structure
+ * The caller is responsible for freeing (vmalloc) the returned structure
  */
 struct nvm_chk_meta *pblk_get_chunk_meta(struct pblk *pblk)
 {
@@ -134,7 +134,7 @@ struct nvm_chk_meta *pblk_get_chunk_meta(struct pblk *pblk)
 	ppa.ppa = 0;
 
 	len = geo->all_chunks * sizeof(*meta);
-	meta = kzalloc(len, GFP_KERNEL);
+	meta = vzalloc(len);
 	if (!meta)
 		return ERR_PTR(-ENOMEM);
 
diff --git a/drivers/lightnvm/pblk-init.c b/drivers/lightnvm/pblk-init.c
index fb66bc84d5ca..7ef8249108f0 100644
--- a/drivers/lightnvm/pblk-init.c
+++ b/drivers/lightnvm/pblk-init.c
@@ -1039,7 +1039,7 @@ static int pblk_lines_init(struct pblk *pblk)
 
 	pblk_set_provision(pblk, nr_free_chks);
 
-	kfree(chunk_meta);
+	vfree(chunk_meta);
 	return 0;
 
 fail_free_lines:
diff --git a/drivers/nvme/host/lightnvm.c b/drivers/nvme/host/lightnvm.c
index e42af7771fe5..7d0a4d3b0a48 100644
--- a/drivers/nvme/host/lightnvm.c
+++ b/drivers/nvme/host/lightnvm.c
@@ -573,7 +573,7 @@ static int nvme_nvm_get_chk_meta(struct nvm_dev *ndev,
 	struct nvm_geo *geo = &ndev->geo;
 	struct nvme_ns *ns = ndev->q->queuedata;
 	struct nvme_ctrl *ctrl = ns->ctrl;
-	struct nvme_nvm_chk_meta *dev_meta = (struct nvme_nvm_chk_meta *)meta;
+	struct nvme_nvm_chk_meta *dev_meta, *dev_meta_off;
 	struct ppa_addr ppa;
 	size_t left = nchks * sizeof(struct nvme_nvm_chk_meta);
 	size_t log_pos, offset, len;
@@ -585,6 +585,10 @@ static int nvme_nvm_get_chk_meta(struct nvm_dev *ndev,
 	 */
 	max_len = min_t(unsigned int, ctrl->max_hw_sectors << 9, 256 * 1024);
 
+	dev_meta = kmalloc(max_len, GFP_KERNEL);
+	if (!dev_meta)
+		return -ENOMEM;
+
 	/* Normalize lba address space to obtain log offset */
 	ppa.ppa = slba;
 	ppa = dev_to_generic_addr(ndev, ppa);
@@ -598,6 +602,9 @@ static int nvme_nvm_get_chk_meta(struct nvm_dev *ndev,
 	while (left) {
 		len = min_t(unsigned int, left, max_len);
 
+		memset(dev_meta, 0, max_len);
+		dev_meta_off = dev_meta;
+
 		ret = nvme_get_log(ctrl, ns->head->ns_id,
 				NVME_NVM_LOG_REPORT_CHUNK, 0, dev_meta, len,
 				offset);
@@ -607,21 +614,23 @@ static int nvme_nvm_get_chk_meta(struct nvm_dev *ndev,
 		}
 
 		for (i = 0; i < len; i += sizeof(struct nvme_nvm_chk_meta)) {
-			meta->state = dev_meta->state;
-			meta->type = dev_meta->type;
-			meta->wi = dev_meta->wi;
-			meta->slba = le64_to_cpu(dev_meta->slba);
-			meta->cnlb = le64_to_cpu(dev_meta->cnlb);
-			meta->wp = le64_to_cpu(dev_meta->wp);
+			meta->state = dev_meta_off->state;
+			meta->type = dev_meta_off->type;
+			meta->wi = dev_meta_off->wi;
+			meta->slba = le64_to_cpu(dev_meta_off->slba);
+			meta->cnlb = le64_to_cpu(dev_meta_off->cnlb);
+			meta->wp = le64_to_cpu(dev_meta_off->wp);
 
 			meta++;
-			dev_meta++;
+			dev_meta_off++;
 		}
 
 		offset += len;
 		left -= len;
 	}
 
+	kfree(dev_meta);
+
 	return ret;
 }
 
-- 
2.17.1

[GIT PULL 32/45] lightnvm: pblk: encapsulate rqd dma allocations

[Matias Bjørling]

From: Javier González <javier@javigon.com>

dma allocations for ppa_list and meta_list in rqd are replicated in
several places across the pblk codebase. Make helpers to encapsulate
creation and deletion to simplify the code.

Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk-core.c     | 51 ++++++++++++++++++++++----------
 drivers/lightnvm/pblk-read.c     | 31 +++++++------------
 drivers/lightnvm/pblk-recovery.c | 30 +++++++------------
 drivers/lightnvm/pblk-write.c    | 15 ++--------
 drivers/lightnvm/pblk.h          |  2 ++
 5 files changed, 59 insertions(+), 70 deletions(-)

diff --git a/drivers/lightnvm/pblk-core.c b/drivers/lightnvm/pblk-core.c
index 875f3cf615ac..8ae40855d4c9 100644
--- a/drivers/lightnvm/pblk-core.c
+++ b/drivers/lightnvm/pblk-core.c
@@ -237,6 +237,33 @@ static void pblk_invalidate_range(struct pblk *pblk, sector_t slba,
 	spin_unlock(&pblk->trans_lock);
 }
 
+int pblk_alloc_rqd_meta(struct pblk *pblk, struct nvm_rq *rqd)
+{
+	struct nvm_tgt_dev *dev = pblk->dev;
+
+	rqd->meta_list = nvm_dev_dma_alloc(dev->parent, GFP_KERNEL,
+							&rqd->dma_meta_list);
+	if (!rqd->meta_list)
+		return -ENOMEM;
+
+	if (rqd->nr_ppas == 1)
+		return 0;
+
+	rqd->ppa_list = rqd->meta_list + pblk_dma_meta_size;
+	rqd->dma_ppa_list = rqd->dma_meta_list + pblk_dma_meta_size;
+
+	return 0;
+}
+
+void pblk_free_rqd_meta(struct pblk *pblk, struct nvm_rq *rqd)
+{
+	struct nvm_tgt_dev *dev = pblk->dev;
+
+	if (rqd->meta_list)
+		nvm_dev_dma_free(dev->parent, rqd->meta_list,
+				rqd->dma_meta_list);
+}
+
 /* Caller must guarantee that the request is a valid type */
 struct nvm_rq *pblk_alloc_rqd(struct pblk *pblk, int type)
 {
@@ -268,7 +295,6 @@ struct nvm_rq *pblk_alloc_rqd(struct pblk *pblk, int type)
 /* Typically used on completion path. Cannot guarantee request consistency */
 void pblk_free_rqd(struct pblk *pblk, struct nvm_rq *rqd, int type)
 {
-	struct nvm_tgt_dev *dev = pblk->dev;
 	mempool_t *pool;
 
 	switch (type) {
@@ -289,9 +315,7 @@ void pblk_free_rqd(struct pblk *pblk, struct nvm_rq *rqd, int type)
 		return;
 	}
 
-	if (rqd->meta_list)
-		nvm_dev_dma_free(dev->parent, rqd->meta_list,
-				rqd->dma_meta_list);
+	pblk_free_rqd_meta(pblk, rqd);
 	mempool_free(rqd, pool);
 }
 
@@ -838,18 +862,14 @@ static int pblk_line_submit_smeta_io(struct pblk *pblk, struct pblk_line *line,
 
 	memset(&rqd, 0, sizeof(struct nvm_rq));
 
-	rqd.meta_list = nvm_dev_dma_alloc(dev->parent, GFP_KERNEL,
-							&rqd.dma_meta_list);
-	if (!rqd.meta_list)
-		return -ENOMEM;
-
-	rqd.ppa_list = rqd.meta_list + pblk_dma_meta_size;
-	rqd.dma_ppa_list = rqd.dma_meta_list + pblk_dma_meta_size;
+	ret = pblk_alloc_rqd_meta(pblk, &rqd);
+	if (ret)
+		return ret;
 
 	bio = bio_map_kern(dev->q, line->smeta, lm->smeta_len, GFP_KERNEL);
 	if (IS_ERR(bio)) {
 		ret = PTR_ERR(bio);
-		goto free_ppa_list;
+		goto clear_rqd;
 	}
 
 	bio->bi_iter.bi_sector = 0; /* internal bio */
@@ -881,7 +901,7 @@ static int pblk_line_submit_smeta_io(struct pblk *pblk, struct pblk_line *line,
 	if (ret) {
 		pblk_err(pblk, "smeta I/O submission failed: %d\n", ret);
 		bio_put(bio);
-		goto free_ppa_list;
+		goto clear_rqd;
 	}
 
 	atomic_dec(&pblk->inflight_io);
@@ -894,9 +914,8 @@ static int pblk_line_submit_smeta_io(struct pblk *pblk, struct pblk_line *line,
 			pblk_log_read_err(pblk, &rqd);
 	}
 
-free_ppa_list:
-	nvm_dev_dma_free(dev->parent, rqd.meta_list, rqd.dma_meta_list);
-
+clear_rqd:
+	pblk_free_rqd_meta(pblk, &rqd);
 	return ret;
 }
 
diff --git a/drivers/lightnvm/pblk-read.c b/drivers/lightnvm/pblk-read.c
index 829e92857289..f5fe01d3a07f 100644
--- a/drivers/lightnvm/pblk-read.c
+++ b/drivers/lightnvm/pblk-read.c
@@ -453,21 +453,13 @@ int pblk_submit_read(struct pblk *pblk, struct bio *bio)
 	 */
 	bio_init_idx = pblk_get_bi_idx(bio);
 
-	rqd->meta_list = nvm_dev_dma_alloc(dev->parent, GFP_KERNEL,
-							&rqd->dma_meta_list);
-	if (!rqd->meta_list) {
-		pblk_err(pblk, "not able to allocate ppa list\n");
+	if (pblk_alloc_rqd_meta(pblk, rqd))
 		goto fail_rqd_free;
-	}
-
-	if (nr_secs > 1) {
-		rqd->ppa_list = rqd->meta_list + pblk_dma_meta_size;
-		rqd->dma_ppa_list = rqd->dma_meta_list + pblk_dma_meta_size;
 
+	if (nr_secs > 1)
 		pblk_read_ppalist_rq(pblk, rqd, bio, blba, read_bitmap);
-	} else {
+	else
 		pblk_read_rq(pblk, rqd, bio, blba, read_bitmap);
-	}
 
 	if (bitmap_full(read_bitmap, nr_secs)) {
 		atomic_inc(&pblk->inflight_io);
@@ -594,15 +586,11 @@ int pblk_submit_read_gc(struct pblk *pblk, struct pblk_gc_rq *gc_rq)
 
 	memset(&rqd, 0, sizeof(struct nvm_rq));
 
-	rqd.meta_list = nvm_dev_dma_alloc(dev->parent, GFP_KERNEL,
-							&rqd.dma_meta_list);
-	if (!rqd.meta_list)
-		return -ENOMEM;
+	ret = pblk_alloc_rqd_meta(pblk, &rqd);
+	if (ret)
+		return ret;
 
 	if (gc_rq->nr_secs > 1) {
-		rqd.ppa_list = rqd.meta_list + pblk_dma_meta_size;
-		rqd.dma_ppa_list = rqd.dma_meta_list + pblk_dma_meta_size;
-
 		gc_rq->secs_to_gc = read_ppalist_rq_gc(pblk, &rqd, gc_rq->line,
 							gc_rq->lba_list,
 							gc_rq->paddr_list,
@@ -623,7 +611,8 @@ int pblk_submit_read_gc(struct pblk *pblk, struct pblk_gc_rq *gc_rq)
 						PBLK_VMALLOC_META, GFP_KERNEL);
 	if (IS_ERR(bio)) {
 		pblk_err(pblk, "could not allocate GC bio (%lu)\n",
-				PTR_ERR(bio));
+								PTR_ERR(bio));
+		ret = PTR_ERR(bio);
 		goto err_free_dma;
 	}
 
@@ -658,12 +647,12 @@ int pblk_submit_read_gc(struct pblk *pblk, struct pblk_gc_rq *gc_rq)
 #endif
 
 out:
-	nvm_dev_dma_free(dev->parent, rqd.meta_list, rqd.dma_meta_list);
+	pblk_free_rqd_meta(pblk, &rqd);
 	return ret;
 
 err_free_bio:
 	bio_put(bio);
 err_free_dma:
-	nvm_dev_dma_free(dev->parent, rqd.meta_list, rqd.dma_meta_list);
+	pblk_free_rqd_meta(pblk, &rqd);
 	return ret;
 }
diff --git a/drivers/lightnvm/pblk-recovery.c b/drivers/lightnvm/pblk-recovery.c
index 2526722304bb..218292979953 100644
--- a/drivers/lightnvm/pblk-recovery.c
+++ b/drivers/lightnvm/pblk-recovery.c
@@ -241,13 +241,11 @@ static int pblk_recov_pad_oob(struct pblk *pblk, struct pblk_line *line,
 {
 	struct nvm_tgt_dev *dev = pblk->dev;
 	struct nvm_geo *geo = &dev->geo;
-	struct ppa_addr *ppa_list;
 	struct pblk_sec_meta *meta_list;
 	struct pblk_pad_rq *pad_rq;
 	struct nvm_rq *rqd;
 	struct bio *bio;
 	void *data;
-	dma_addr_t dma_ppa_list, dma_meta_list;
 	__le64 *lba_list = emeta_to_lbas(pblk, line->emeta->buf);
 	u64 w_ptr = line->cur_sec;
 	int left_line_ppas, rq_ppas, rq_len;
@@ -281,20 +279,11 @@ static int pblk_recov_pad_oob(struct pblk *pblk, struct pblk_line *line,
 
 	rq_len = rq_ppas * geo->csecs;
 
-	meta_list = nvm_dev_dma_alloc(dev->parent, GFP_KERNEL, &dma_meta_list);
-	if (!meta_list) {
-		ret = -ENOMEM;
-		goto fail_free_pad;
-	}
-
-	ppa_list = (void *)(meta_list) + pblk_dma_meta_size;
-	dma_ppa_list = dma_meta_list + pblk_dma_meta_size;
-
 	bio = pblk_bio_map_addr(pblk, data, rq_ppas, rq_len,
 						PBLK_VMALLOC_META, GFP_KERNEL);
 	if (IS_ERR(bio)) {
 		ret = PTR_ERR(bio);
-		goto fail_free_meta;
+		goto fail_free_pad;
 	}
 
 	bio->bi_iter.bi_sector = 0; /* internal bio */
@@ -302,17 +291,19 @@ static int pblk_recov_pad_oob(struct pblk *pblk, struct pblk_line *line,
 
 	rqd = pblk_alloc_rqd(pblk, PBLK_WRITE_INT);
 
+	ret = pblk_alloc_rqd_meta(pblk, rqd);
+	if (ret)
+		goto fail_free_rqd;
+
 	rqd->bio = bio;
 	rqd->opcode = NVM_OP_PWRITE;
 	rqd->is_seq = 1;
-	rqd->meta_list = meta_list;
 	rqd->nr_ppas = rq_ppas;
-	rqd->ppa_list = ppa_list;
-	rqd->dma_ppa_list = dma_ppa_list;
-	rqd->dma_meta_list = dma_meta_list;
 	rqd->end_io = pblk_end_io_recov;
 	rqd->private = pad_rq;
 
+	meta_list = rqd->meta_list;
+
 	for (i = 0; i < rqd->nr_ppas; ) {
 		struct ppa_addr ppa;
 		int pos;
@@ -346,7 +337,7 @@ static int pblk_recov_pad_oob(struct pblk *pblk, struct pblk_line *line,
 	if (ret) {
 		pblk_err(pblk, "I/O submission failed: %d\n", ret);
 		pblk_up_chunk(pblk, rqd->ppa_list[0]);
-		goto fail_free_bio;
+		goto fail_free_rqd;
 	}
 
 	left_line_ppas -= rq_ppas;
@@ -370,10 +361,9 @@ static int pblk_recov_pad_oob(struct pblk *pblk, struct pblk_line *line,
 	kfree(pad_rq);
 	return ret;
 
-fail_free_bio:
+fail_free_rqd:
+	pblk_free_rqd(pblk, rqd, PBLK_WRITE_INT);
 	bio_put(bio);
-fail_free_meta:
-	nvm_dev_dma_free(dev->parent, meta_list, dma_meta_list);
 fail_free_pad:
 	kfree(pad_rq);
 	vfree(data);
diff --git a/drivers/lightnvm/pblk-write.c b/drivers/lightnvm/pblk-write.c
index 9554febee480..a276f25d3931 100644
--- a/drivers/lightnvm/pblk-write.c
+++ b/drivers/lightnvm/pblk-write.c
@@ -285,11 +285,8 @@ static void pblk_end_io_write_meta(struct nvm_rq *rqd)
 }
 
 static int pblk_alloc_w_rq(struct pblk *pblk, struct nvm_rq *rqd,
-			   unsigned int nr_secs,
-			   nvm_end_io_fn(*end_io))
+			   unsigned int nr_secs, nvm_end_io_fn(*end_io))
 {
-	struct nvm_tgt_dev *dev = pblk->dev;
-
 	/* Setup write request */
 	rqd->opcode = NVM_OP_PWRITE;
 	rqd->nr_ppas = nr_secs;
@@ -297,15 +294,7 @@ static int pblk_alloc_w_rq(struct pblk *pblk, struct nvm_rq *rqd,
 	rqd->private = pblk;
 	rqd->end_io = end_io;
 
-	rqd->meta_list = nvm_dev_dma_alloc(dev->parent, GFP_KERNEL,
-							&rqd->dma_meta_list);
-	if (!rqd->meta_list)
-		return -ENOMEM;
-
-	rqd->ppa_list = rqd->meta_list + pblk_dma_meta_size;
-	rqd->dma_ppa_list = rqd->dma_meta_list + pblk_dma_meta_size;
-
-	return 0;
+	return pblk_alloc_rqd_meta(pblk, rqd);
 }
 
 static int pblk_setup_w_rq(struct pblk *pblk, struct nvm_rq *rqd,
diff --git a/drivers/lightnvm/pblk.h b/drivers/lightnvm/pblk.h
index d123cff82589..b06ab0edab69 100644
--- a/drivers/lightnvm/pblk.h
+++ b/drivers/lightnvm/pblk.h
@@ -778,6 +778,8 @@ ssize_t pblk_rb_sysfs(struct pblk_rb *rb, char *buf);
  */
 struct nvm_rq *pblk_alloc_rqd(struct pblk *pblk, int type);
 void pblk_free_rqd(struct pblk *pblk, struct nvm_rq *rqd, int type);
+int pblk_alloc_rqd_meta(struct pblk *pblk, struct nvm_rq *rqd);
+void pblk_free_rqd_meta(struct pblk *pblk, struct nvm_rq *rqd);
 void pblk_set_sec_per_write(struct pblk *pblk, int sec_per_write);
 int pblk_setup_w_rec_rq(struct pblk *pblk, struct nvm_rq *rqd,
 			struct pblk_c_ctx *c_ctx);
-- 
2.17.1

[GIT PULL 33/45] lightnvm: pblk: refactor metadata paths

[Matias Bjørling]

From: Javier González <javier@javigon.com>

pblk maintains two different metadata paths for smeta and emeta, which
store metadata at the start of the line and at the end of the line,
respectively. Until now, these path has been common for writing and
retrieving metadata, however, as these paths diverge, the common code
becomes less clear and unnecessary complicated.

In preparation for further changes to the metadata write path, this
patch separates the write and read paths for smeta and emeta and
removes the synchronous emeta path as it not used anymore (emeta is
scheduled asynchronously to prevent jittering due to internal I/Os).

Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk-core.c     | 325 ++++++++++++++-----------------
 drivers/lightnvm/pblk-gc.c       |   2 +-
 drivers/lightnvm/pblk-recovery.c |   4 +-
 drivers/lightnvm/pblk.h          |   4 +-
 4 files changed, 155 insertions(+), 180 deletions(-)

diff --git a/drivers/lightnvm/pblk-core.c b/drivers/lightnvm/pblk-core.c
index 8ae40855d4c9..49cef93e328e 100644
--- a/drivers/lightnvm/pblk-core.c
+++ b/drivers/lightnvm/pblk-core.c
@@ -685,12 +685,129 @@ u64 pblk_lookup_page(struct pblk *pblk, struct pblk_line *line)
 	return paddr;
 }
 
-/*
- * Submit emeta to one LUN in the raid line at the time to avoid a deadlock when
- * taking the per LUN semaphore.
- */
-static int pblk_line_submit_emeta_io(struct pblk *pblk, struct pblk_line *line,
-				     void *emeta_buf, u64 paddr, int dir)
+u64 pblk_line_smeta_start(struct pblk *pblk, struct pblk_line *line)
+{
+	struct nvm_tgt_dev *dev = pblk->dev;
+	struct nvm_geo *geo = &dev->geo;
+	struct pblk_line_meta *lm = &pblk->lm;
+	int bit;
+
+	/* This usually only happens on bad lines */
+	bit = find_first_zero_bit(line->blk_bitmap, lm->blk_per_line);
+	if (bit >= lm->blk_per_line)
+		return -1;
+
+	return bit * geo->ws_opt;
+}
+
+int pblk_line_smeta_read(struct pblk *pblk, struct pblk_line *line)
+{
+	struct nvm_tgt_dev *dev = pblk->dev;
+	struct pblk_line_meta *lm = &pblk->lm;
+	struct bio *bio;
+	struct nvm_rq rqd;
+	u64 paddr = pblk_line_smeta_start(pblk, line);
+	int i, ret;
+
+	memset(&rqd, 0, sizeof(struct nvm_rq));
+
+	ret = pblk_alloc_rqd_meta(pblk, &rqd);
+	if (ret)
+		return ret;
+
+	bio = bio_map_kern(dev->q, line->smeta, lm->smeta_len, GFP_KERNEL);
+	if (IS_ERR(bio)) {
+		ret = PTR_ERR(bio);
+		goto clear_rqd;
+	}
+
+	bio->bi_iter.bi_sector = 0; /* internal bio */
+	bio_set_op_attrs(bio, REQ_OP_READ, 0);
+
+	rqd.bio = bio;
+	rqd.opcode = NVM_OP_PREAD;
+	rqd.nr_ppas = lm->smeta_sec;
+	rqd.is_seq = 1;
+
+	for (i = 0; i < lm->smeta_sec; i++, paddr++)
+		rqd.ppa_list[i] = addr_to_gen_ppa(pblk, paddr, line->id);
+
+	ret = pblk_submit_io_sync(pblk, &rqd);
+	if (ret) {
+		pblk_err(pblk, "smeta I/O submission failed: %d\n", ret);
+		bio_put(bio);
+		goto clear_rqd;
+	}
+
+	atomic_dec(&pblk->inflight_io);
+
+	if (rqd.error)
+		pblk_log_read_err(pblk, &rqd);
+
+clear_rqd:
+	pblk_free_rqd_meta(pblk, &rqd);
+	return ret;
+}
+
+static int pblk_line_smeta_write(struct pblk *pblk, struct pblk_line *line,
+				 u64 paddr)
+{
+	struct nvm_tgt_dev *dev = pblk->dev;
+	struct pblk_line_meta *lm = &pblk->lm;
+	struct bio *bio;
+	struct nvm_rq rqd;
+	__le64 *lba_list = emeta_to_lbas(pblk, line->emeta->buf);
+	__le64 addr_empty = cpu_to_le64(ADDR_EMPTY);
+	int i, ret;
+
+	memset(&rqd, 0, sizeof(struct nvm_rq));
+
+	ret = pblk_alloc_rqd_meta(pblk, &rqd);
+	if (ret)
+		return ret;
+
+	bio = bio_map_kern(dev->q, line->smeta, lm->smeta_len, GFP_KERNEL);
+	if (IS_ERR(bio)) {
+		ret = PTR_ERR(bio);
+		goto clear_rqd;
+	}
+
+	bio->bi_iter.bi_sector = 0; /* internal bio */
+	bio_set_op_attrs(bio, REQ_OP_WRITE, 0);
+
+	rqd.bio = bio;
+	rqd.opcode = NVM_OP_PWRITE;
+	rqd.nr_ppas = lm->smeta_sec;
+	rqd.is_seq = 1;
+
+	for (i = 0; i < lm->smeta_sec; i++, paddr++) {
+		struct pblk_sec_meta *meta_list = rqd.meta_list;
+
+		rqd.ppa_list[i] = addr_to_gen_ppa(pblk, paddr, line->id);
+		meta_list[i].lba = lba_list[paddr] = addr_empty;
+	}
+
+	ret = pblk_submit_io_sync(pblk, &rqd);
+	if (ret) {
+		pblk_err(pblk, "smeta I/O submission failed: %d\n", ret);
+		bio_put(bio);
+		goto clear_rqd;
+	}
+
+	atomic_dec(&pblk->inflight_io);
+
+	if (rqd.error) {
+		pblk_log_write_err(pblk, &rqd);
+		ret = -EIO;
+	}
+
+clear_rqd:
+	pblk_free_rqd_meta(pblk, &rqd);
+	return ret;
+}
+
+int pblk_line_emeta_read(struct pblk *pblk, struct pblk_line *line,
+			 void *emeta_buf)
 {
 	struct nvm_tgt_dev *dev = pblk->dev;
 	struct nvm_geo *geo = &dev->geo;
@@ -699,24 +816,15 @@ static int pblk_line_submit_emeta_io(struct pblk *pblk, struct pblk_line *line,
 	void *ppa_list, *meta_list;
 	struct bio *bio;
 	struct nvm_rq rqd;
+	u64 paddr = line->emeta_ssec;
 	dma_addr_t dma_ppa_list, dma_meta_list;
 	int min = pblk->min_write_pgs;
 	int left_ppas = lm->emeta_sec[0];
-	int id = line->id;
+	int line_id = line->id;
 	int rq_ppas, rq_len;
-	int cmd_op, bio_op;
 	int i, j;
 	int ret;
 
-	if (dir == PBLK_WRITE) {
-		bio_op = REQ_OP_WRITE;
-		cmd_op = NVM_OP_PWRITE;
-	} else if (dir == PBLK_READ) {
-		bio_op = REQ_OP_READ;
-		cmd_op = NVM_OP_PREAD;
-	} else
-		return -EINVAL;
-
 	meta_list = nvm_dev_dma_alloc(dev->parent, GFP_KERNEL,
 							&dma_meta_list);
 	if (!meta_list)
@@ -739,64 +847,43 @@ static int pblk_line_submit_emeta_io(struct pblk *pblk, struct pblk_line *line,
 	}
 
 	bio->bi_iter.bi_sector = 0; /* internal bio */
-	bio_set_op_attrs(bio, bio_op, 0);
+	bio_set_op_attrs(bio, REQ_OP_READ, 0);
 
 	rqd.bio = bio;
 	rqd.meta_list = meta_list;
 	rqd.ppa_list = ppa_list;
 	rqd.dma_meta_list = dma_meta_list;
 	rqd.dma_ppa_list = dma_ppa_list;
-	rqd.opcode = cmd_op;
+	rqd.opcode = NVM_OP_PREAD;
 	rqd.nr_ppas = rq_ppas;
 
-	if (dir == PBLK_WRITE) {
-		struct pblk_sec_meta *meta_list = rqd.meta_list;
+	for (i = 0; i < rqd.nr_ppas; ) {
+		struct ppa_addr ppa = addr_to_gen_ppa(pblk, paddr, line_id);
+		int pos = pblk_ppa_to_pos(geo, ppa);
 
-		rqd.is_seq = 1;
-		for (i = 0; i < rqd.nr_ppas; ) {
-			spin_lock(&line->lock);
-			paddr = __pblk_alloc_page(pblk, line, min);
-			spin_unlock(&line->lock);
-			for (j = 0; j < min; j++, i++, paddr++) {
-				meta_list[i].lba = cpu_to_le64(ADDR_EMPTY);
-				rqd.ppa_list[i] =
-					addr_to_gen_ppa(pblk, paddr, id);
-			}
-		}
-	} else {
-		for (i = 0; i < rqd.nr_ppas; ) {
-			struct ppa_addr ppa = addr_to_gen_ppa(pblk, paddr, id);
-			int pos = pblk_ppa_to_pos(geo, ppa);
+		if (pblk_io_aligned(pblk, rq_ppas))
+			rqd.is_seq = 1;
 
-			if (pblk_io_aligned(pblk, rq_ppas))
-				rqd.is_seq = 1;
-
-			while (test_bit(pos, line->blk_bitmap)) {
-				paddr += min;
-				if (pblk_boundary_paddr_checks(pblk, paddr)) {
-					pblk_err(pblk, "corrupt emeta line:%d\n",
-								line->id);
-					bio_put(bio);
-					ret = -EINTR;
-					goto free_rqd_dma;
-				}
-
-				ppa = addr_to_gen_ppa(pblk, paddr, id);
-				pos = pblk_ppa_to_pos(geo, ppa);
-			}
-
-			if (pblk_boundary_paddr_checks(pblk, paddr + min)) {
-				pblk_err(pblk, "corrupt emeta line:%d\n",
-								line->id);
+		while (test_bit(pos, line->blk_bitmap)) {
+			paddr += min;
+			if (pblk_boundary_paddr_checks(pblk, paddr)) {
 				bio_put(bio);
 				ret = -EINTR;
 				goto free_rqd_dma;
 			}
 
-			for (j = 0; j < min; j++, i++, paddr++)
-				rqd.ppa_list[i] =
-					addr_to_gen_ppa(pblk, paddr, line->id);
+			ppa = addr_to_gen_ppa(pblk, paddr, line_id);
+			pos = pblk_ppa_to_pos(geo, ppa);
 		}
+
+		if (pblk_boundary_paddr_checks(pblk, paddr + min)) {
+			bio_put(bio);
+			ret = -EINTR;
+			goto free_rqd_dma;
+		}
+
+		for (j = 0; j < min; j++, i++, paddr++)
+			rqd.ppa_list[i] = addr_to_gen_ppa(pblk, paddr, line_id);
 	}
 
 	ret = pblk_submit_io_sync(pblk, &rqd);
@@ -808,131 +895,19 @@ static int pblk_line_submit_emeta_io(struct pblk *pblk, struct pblk_line *line,
 
 	atomic_dec(&pblk->inflight_io);
 
-	if (rqd.error) {
-		if (dir == PBLK_WRITE)
-			pblk_log_write_err(pblk, &rqd);
-		else
-			pblk_log_read_err(pblk, &rqd);
-	}
+	if (rqd.error)
+		pblk_log_read_err(pblk, &rqd);
 
 	emeta_buf += rq_len;
 	left_ppas -= rq_ppas;
 	if (left_ppas)
 		goto next_rq;
+
 free_rqd_dma:
 	nvm_dev_dma_free(dev->parent, rqd.meta_list, rqd.dma_meta_list);
 	return ret;
 }
 
-u64 pblk_line_smeta_start(struct pblk *pblk, struct pblk_line *line)
-{
-	struct nvm_tgt_dev *dev = pblk->dev;
-	struct nvm_geo *geo = &dev->geo;
-	struct pblk_line_meta *lm = &pblk->lm;
-	int bit;
-
-	/* This usually only happens on bad lines */
-	bit = find_first_zero_bit(line->blk_bitmap, lm->blk_per_line);
-	if (bit >= lm->blk_per_line)
-		return -1;
-
-	return bit * geo->ws_opt;
-}
-
-static int pblk_line_submit_smeta_io(struct pblk *pblk, struct pblk_line *line,
-				     u64 paddr, int dir)
-{
-	struct nvm_tgt_dev *dev = pblk->dev;
-	struct pblk_line_meta *lm = &pblk->lm;
-	struct bio *bio;
-	struct nvm_rq rqd;
-	__le64 *lba_list = NULL;
-	int i, ret;
-	int cmd_op, bio_op;
-
-	if (dir == PBLK_WRITE) {
-		bio_op = REQ_OP_WRITE;
-		cmd_op = NVM_OP_PWRITE;
-		lba_list = emeta_to_lbas(pblk, line->emeta->buf);
-	} else if (dir == PBLK_READ_RECOV || dir == PBLK_READ) {
-		bio_op = REQ_OP_READ;
-		cmd_op = NVM_OP_PREAD;
-	} else
-		return -EINVAL;
-
-	memset(&rqd, 0, sizeof(struct nvm_rq));
-
-	ret = pblk_alloc_rqd_meta(pblk, &rqd);
-	if (ret)
-		return ret;
-
-	bio = bio_map_kern(dev->q, line->smeta, lm->smeta_len, GFP_KERNEL);
-	if (IS_ERR(bio)) {
-		ret = PTR_ERR(bio);
-		goto clear_rqd;
-	}
-
-	bio->bi_iter.bi_sector = 0; /* internal bio */
-	bio_set_op_attrs(bio, bio_op, 0);
-
-	rqd.bio = bio;
-	rqd.opcode = cmd_op;
-	rqd.is_seq = 1;
-	rqd.nr_ppas = lm->smeta_sec;
-
-	for (i = 0; i < lm->smeta_sec; i++, paddr++) {
-		struct pblk_sec_meta *meta_list = rqd.meta_list;
-
-		rqd.ppa_list[i] = addr_to_gen_ppa(pblk, paddr, line->id);
-
-		if (dir == PBLK_WRITE) {
-			__le64 addr_empty = cpu_to_le64(ADDR_EMPTY);
-
-			meta_list[i].lba = lba_list[paddr] = addr_empty;
-		}
-	}
-
-	/*
-	 * This I/O is sent by the write thread when a line is replace. Since
-	 * the write thread is the only one sending write and erase commands,
-	 * there is no need to take the LUN semaphore.
-	 */
-	ret = pblk_submit_io_sync(pblk, &rqd);
-	if (ret) {
-		pblk_err(pblk, "smeta I/O submission failed: %d\n", ret);
-		bio_put(bio);
-		goto clear_rqd;
-	}
-
-	atomic_dec(&pblk->inflight_io);
-
-	if (rqd.error) {
-		if (dir == PBLK_WRITE) {
-			pblk_log_write_err(pblk, &rqd);
-			ret = 1;
-		} else if (dir == PBLK_READ)
-			pblk_log_read_err(pblk, &rqd);
-	}
-
-clear_rqd:
-	pblk_free_rqd_meta(pblk, &rqd);
-	return ret;
-}
-
-int pblk_line_read_smeta(struct pblk *pblk, struct pblk_line *line)
-{
-	u64 bpaddr = pblk_line_smeta_start(pblk, line);
-
-	return pblk_line_submit_smeta_io(pblk, line, bpaddr, PBLK_READ_RECOV);
-}
-
-int pblk_line_read_emeta(struct pblk *pblk, struct pblk_line *line,
-			 void *emeta_buf)
-{
-	return pblk_line_submit_emeta_io(pblk, line, emeta_buf,
-						line->emeta_ssec, PBLK_READ);
-}
-
 static void pblk_setup_e_rq(struct pblk *pblk, struct nvm_rq *rqd,
 			    struct ppa_addr ppa)
 {
@@ -1169,7 +1144,7 @@ static int pblk_line_init_bb(struct pblk *pblk, struct pblk_line *line,
 	line->smeta_ssec = off;
 	line->cur_sec = off + lm->smeta_sec;
 
-	if (init && pblk_line_submit_smeta_io(pblk, line, off, PBLK_WRITE)) {
+	if (init && pblk_line_smeta_write(pblk, line, off)) {
 		pblk_debug(pblk, "line smeta I/O failed. Retry\n");
 		return 0;
 	}
diff --git a/drivers/lightnvm/pblk-gc.c b/drivers/lightnvm/pblk-gc.c
index b841d84c4342..e05d06bd5b83 100644
--- a/drivers/lightnvm/pblk-gc.c
+++ b/drivers/lightnvm/pblk-gc.c
@@ -148,7 +148,7 @@ static __le64 *get_lba_list_from_emeta(struct pblk *pblk,
 	if (!emeta_buf)
 		return NULL;
 
-	ret = pblk_line_read_emeta(pblk, line, emeta_buf);
+	ret = pblk_line_emeta_read(pblk, line, emeta_buf);
 	if (ret) {
 		pblk_err(pblk, "line %d read emeta failed (%d)\n",
 				line->id, ret);
diff --git a/drivers/lightnvm/pblk-recovery.c b/drivers/lightnvm/pblk-recovery.c
index 218292979953..6c57eb00a7f1 100644
--- a/drivers/lightnvm/pblk-recovery.c
+++ b/drivers/lightnvm/pblk-recovery.c
@@ -836,7 +836,7 @@ struct pblk_line *pblk_recov_l2p(struct pblk *pblk)
 			continue;
 
 		/* Lines that cannot be read are assumed as not written here */
-		if (pblk_line_read_smeta(pblk, line))
+		if (pblk_line_smeta_read(pblk, line))
 			continue;
 
 		crc = pblk_calc_smeta_crc(pblk, smeta_buf);
@@ -906,7 +906,7 @@ struct pblk_line *pblk_recov_l2p(struct pblk *pblk)
 		line->emeta = emeta;
 		memset(line->emeta->buf, 0, lm->emeta_len[0]);
 
-		if (pblk_line_read_emeta(pblk, line, line->emeta->buf)) {
+		if (pblk_line_emeta_read(pblk, line, line->emeta->buf)) {
 			pblk_recov_l2p_from_oob(pblk, line);
 			goto next;
 		}
diff --git a/drivers/lightnvm/pblk.h b/drivers/lightnvm/pblk.h
index b06ab0edab69..02e2c02b0cf4 100644
--- a/drivers/lightnvm/pblk.h
+++ b/drivers/lightnvm/pblk.h
@@ -819,8 +819,8 @@ void pblk_gen_run_ws(struct pblk *pblk, struct pblk_line *line, void *priv,
 		     void (*work)(struct work_struct *), gfp_t gfp_mask,
 		     struct workqueue_struct *wq);
 u64 pblk_line_smeta_start(struct pblk *pblk, struct pblk_line *line);
-int pblk_line_read_smeta(struct pblk *pblk, struct pblk_line *line);
-int pblk_line_read_emeta(struct pblk *pblk, struct pblk_line *line,
+int pblk_line_smeta_read(struct pblk *pblk, struct pblk_line *line);
+int pblk_line_emeta_read(struct pblk *pblk, struct pblk_line *line,
 			 void *emeta_buf);
 int pblk_blk_erase_async(struct pblk *pblk, struct ppa_addr erase_ppa);
 void pblk_line_put(struct kref *ref);
-- 
2.17.1

[GIT PULL 34/45] lightnvm: pblk: take write semaphore on metadata

[Matias Bjørling]

From: Javier González <javier@javigon.com>

pblk guarantees write ordering at a chunk level through a per open chunk
semaphore. At this point, since we only have an open I/O stream for both
user and GC data, the semaphore is per parallel unit.

For the metadata I/O that is synchronous, the semaphore is not needed as
ordering is guaranteed. However, if the metadata scheme changes or
multiple streams are open, this guarantee might not be preserved.

This patch makes sure that all writes go through the semaphore, even for
synchronous I/O. This is consistent with pblk's write I/O model. It also
simplifies maintenance since changes in the metadata scheme could cause
ordering issues.

Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk-core.c | 16 +++++++++++++++-
 drivers/lightnvm/pblk.h      |  1 +
 2 files changed, 16 insertions(+), 1 deletion(-)

diff --git a/drivers/lightnvm/pblk-core.c b/drivers/lightnvm/pblk-core.c
index 49cef93e328e..a3ce4a36dd33 100644
--- a/drivers/lightnvm/pblk-core.c
+++ b/drivers/lightnvm/pblk-core.c
@@ -557,6 +557,20 @@ int pblk_submit_io_sync(struct pblk *pblk, struct nvm_rq *rqd)
 	return ret;
 }
 
+int pblk_submit_io_sync_sem(struct pblk *pblk, struct nvm_rq *rqd)
+{
+	struct ppa_addr *ppa_list;
+	int ret;
+
+	ppa_list = (rqd->nr_ppas > 1) ? rqd->ppa_list : &rqd->ppa_addr;
+
+	pblk_down_chunk(pblk, ppa_list[0]);
+	ret = pblk_submit_io_sync(pblk, rqd);
+	pblk_up_chunk(pblk, ppa_list[0]);
+
+	return ret;
+}
+
 static void pblk_bio_map_addr_endio(struct bio *bio)
 {
 	bio_put(bio);
@@ -787,7 +801,7 @@ static int pblk_line_smeta_write(struct pblk *pblk, struct pblk_line *line,
 		meta_list[i].lba = lba_list[paddr] = addr_empty;
 	}
 
-	ret = pblk_submit_io_sync(pblk, &rqd);
+	ret = pblk_submit_io_sync_sem(pblk, &rqd);
 	if (ret) {
 		pblk_err(pblk, "smeta I/O submission failed: %d\n", ret);
 		bio_put(bio);
diff --git a/drivers/lightnvm/pblk.h b/drivers/lightnvm/pblk.h
index 02e2c02b0cf4..4c015c457197 100644
--- a/drivers/lightnvm/pblk.h
+++ b/drivers/lightnvm/pblk.h
@@ -792,6 +792,7 @@ void pblk_log_write_err(struct pblk *pblk, struct nvm_rq *rqd);
 void pblk_log_read_err(struct pblk *pblk, struct nvm_rq *rqd);
 int pblk_submit_io(struct pblk *pblk, struct nvm_rq *rqd);
 int pblk_submit_io_sync(struct pblk *pblk, struct nvm_rq *rqd);
+int pblk_submit_io_sync_sem(struct pblk *pblk, struct nvm_rq *rqd);
 int pblk_submit_meta_io(struct pblk *pblk, struct pblk_line *meta_line);
 void pblk_check_chunk_state_update(struct pblk *pblk, struct nvm_rq *rqd);
 struct bio *pblk_bio_map_addr(struct pblk *pblk, void *data,
-- 
2.17.1

[GIT PULL 35/45] lightnvm: pblk: recover open lines on 2.0 devices

[Matias Bjørling]

From: Javier González <javier@javigon.com>

In the OCSSD 2.0 spec, each chunk reports its write pointer. This means
that pblk does not need to scan open lines to find the write pointer,
but instead, it can retrieve it directly (and verify it).

This patch uses the write pointer on open lines to (i) recover the line
up until the last written lba and (ii) reconstruct the map bitmap and
rest of line metadata so that the line can be used for new data.

Since the 1.2 path in lightnvm core has been re-implemented to populate
the chunk structure and thus recover the write pointer on
initialization, this patch removes 1.2 specific recovery, as the 2.0
path can be reused.

Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk-recovery.c | 406 ++++++++++---------------------
 1 file changed, 124 insertions(+), 282 deletions(-)

diff --git a/drivers/lightnvm/pblk-recovery.c b/drivers/lightnvm/pblk-recovery.c
index 6c57eb00a7f1..fccf65bc70b3 100644
--- a/drivers/lightnvm/pblk-recovery.c
+++ b/drivers/lightnvm/pblk-recovery.c
@@ -86,15 +86,39 @@ static int pblk_recov_l2p_from_emeta(struct pblk *pblk, struct pblk_line *line)
 	return 0;
 }
 
-static int pblk_calc_sec_in_line(struct pblk *pblk, struct pblk_line *line)
+static void pblk_update_line_wp(struct pblk *pblk, struct pblk_line *line,
+				u64 written_secs)
+{
+	int i;
+
+	for (i = 0; i < written_secs; i += pblk->min_write_pgs)
+		pblk_alloc_page(pblk, line, pblk->min_write_pgs);
+}
+
+static u64 pblk_sec_in_open_line(struct pblk *pblk, struct pblk_line *line)
 {
-	struct nvm_tgt_dev *dev = pblk->dev;
-	struct nvm_geo *geo = &dev->geo;
 	struct pblk_line_meta *lm = &pblk->lm;
 	int nr_bb = bitmap_weight(line->blk_bitmap, lm->blk_per_line);
+	u64 written_secs = 0;
+	int valid_chunks = 0;
+	int i;
 
-	return lm->sec_per_line - lm->smeta_sec - lm->emeta_sec[0] -
-				nr_bb * geo->clba;
+	for (i = 0; i < lm->blk_per_line; i++) {
+		struct nvm_chk_meta *chunk = &line->chks[i];
+
+		if (chunk->state & NVM_CHK_ST_OFFLINE)
+			continue;
+
+		written_secs += chunk->wp;
+		valid_chunks++;
+	}
+
+	if (lm->blk_per_line - nr_bb != valid_chunks)
+		pblk_err(pblk, "recovery line %d is bad\n", line->id);
+
+	pblk_update_line_wp(pblk, line, written_secs - lm->smeta_sec);
+
+	return written_secs;
 }
 
 struct pblk_recov_alloc {
@@ -106,115 +130,6 @@ struct pblk_recov_alloc {
 	dma_addr_t dma_meta_list;
 };
 
-static int pblk_recov_read_oob(struct pblk *pblk, struct pblk_line *line,
-			       struct pblk_recov_alloc p, u64 r_ptr)
-{
-	struct nvm_tgt_dev *dev = pblk->dev;
-	struct nvm_geo *geo = &dev->geo;
-	struct ppa_addr *ppa_list;
-	struct pblk_sec_meta *meta_list;
-	struct nvm_rq *rqd;
-	struct bio *bio;
-	void *data;
-	dma_addr_t dma_ppa_list, dma_meta_list;
-	u64 r_ptr_int;
-	int left_ppas;
-	int rq_ppas, rq_len;
-	int i, j;
-	int ret = 0;
-
-	ppa_list = p.ppa_list;
-	meta_list = p.meta_list;
-	rqd = p.rqd;
-	data = p.data;
-	dma_ppa_list = p.dma_ppa_list;
-	dma_meta_list = p.dma_meta_list;
-
-	left_ppas = line->cur_sec - r_ptr;
-	if (!left_ppas)
-		return 0;
-
-	r_ptr_int = r_ptr;
-
-next_read_rq:
-	memset(rqd, 0, pblk_g_rq_size);
-
-	rq_ppas = pblk_calc_secs(pblk, left_ppas, 0);
-	if (!rq_ppas)
-		rq_ppas = pblk->min_write_pgs;
-	rq_len = rq_ppas * geo->csecs;
-
-	bio = bio_map_kern(dev->q, data, rq_len, GFP_KERNEL);
-	if (IS_ERR(bio))
-		return PTR_ERR(bio);
-
-	bio->bi_iter.bi_sector = 0; /* internal bio */
-	bio_set_op_attrs(bio, REQ_OP_READ, 0);
-
-	rqd->bio = bio;
-	rqd->opcode = NVM_OP_PREAD;
-	rqd->meta_list = meta_list;
-	rqd->nr_ppas = rq_ppas;
-	rqd->ppa_list = ppa_list;
-	rqd->dma_ppa_list = dma_ppa_list;
-	rqd->dma_meta_list = dma_meta_list;
-
-	if (pblk_io_aligned(pblk, rq_ppas))
-		rqd->is_seq = 1;
-
-	ppa_list = nvm_rq_to_ppa_list(rqd);
-
-	for (i = 0; i < rqd->nr_ppas; ) {
-		struct ppa_addr ppa;
-		int pos;
-
-		ppa = addr_to_gen_ppa(pblk, r_ptr_int, line->id);
-		pos = pblk_ppa_to_pos(geo, ppa);
-
-		while (test_bit(pos, line->blk_bitmap)) {
-			r_ptr_int += pblk->min_write_pgs;
-			ppa = addr_to_gen_ppa(pblk, r_ptr_int, line->id);
-			pos = pblk_ppa_to_pos(geo, ppa);
-		}
-
-		for (j = 0; j < pblk->min_write_pgs; j++, i++, r_ptr_int++)
-			ppa_list[i] =
-				addr_to_gen_ppa(pblk, r_ptr_int, line->id);
-	}
-
-	/* If read fails, more padding is needed */
-	ret = pblk_submit_io_sync(pblk, rqd);
-	if (ret) {
-		pblk_err(pblk, "I/O submission failed: %d\n", ret);
-		return ret;
-	}
-
-	atomic_dec(&pblk->inflight_io);
-
-	/* At this point, the read should not fail. If it does, it is a problem
-	 * we cannot recover from here. Need FTL log.
-	 */
-	if (rqd->error && rqd->error != NVM_RSP_WARN_HIGHECC) {
-		pblk_err(pblk, "L2P recovery failed (%d)\n", rqd->error);
-		return -EINTR;
-	}
-
-	for (i = 0; i < rqd->nr_ppas; i++) {
-		u64 lba = le64_to_cpu(meta_list[i].lba);
-
-		if (lba == ADDR_EMPTY || lba > pblk->rl.nr_secs)
-			continue;
-
-		pblk_update_map(pblk, lba, ppa_list[i]);
-	}
-
-	left_ppas -= rq_ppas;
-	if (left_ppas > 0)
-		goto next_read_rq;
-
-	return 0;
-}
-
 static void pblk_recov_complete(struct kref *ref)
 {
 	struct pblk_pad_rq *pad_rq = container_of(ref, struct pblk_pad_rq, ref);
@@ -236,8 +151,9 @@ static void pblk_end_io_recov(struct nvm_rq *rqd)
 	kref_put(&pad_rq->ref, pblk_recov_complete);
 }
 
-static int pblk_recov_pad_oob(struct pblk *pblk, struct pblk_line *line,
-			      int left_ppas)
+/* pad line using line bitmap.  */
+static int pblk_recov_pad_line(struct pblk *pblk, struct pblk_line *line,
+			       int left_ppas)
 {
 	struct nvm_tgt_dev *dev = pblk->dev;
 	struct nvm_geo *geo = &dev->geo;
@@ -379,143 +295,42 @@ static int pblk_pad_distance(struct pblk *pblk, struct pblk_line *line)
 	return (distance > line->left_msecs) ? line->left_msecs : distance;
 }
 
-/* When this function is called, it means that not all upper pages have been
- * written in a page that contains valid data. In order to recover this data, we
- * first find the write pointer on the device, then we pad all necessary
- * sectors, and finally attempt to read the valid data
- */
-static int pblk_recov_scan_all_oob(struct pblk *pblk, struct pblk_line *line,
-				   struct pblk_recov_alloc p)
+static int pblk_line_wp_is_unbalanced(struct pblk *pblk,
+				      struct pblk_line *line)
 {
 	struct nvm_tgt_dev *dev = pblk->dev;
 	struct nvm_geo *geo = &dev->geo;
-	struct ppa_addr *ppa_list;
-	struct pblk_sec_meta *meta_list;
-	struct nvm_rq *rqd;
-	struct bio *bio;
-	void *data;
-	dma_addr_t dma_ppa_list, dma_meta_list;
-	u64 w_ptr = 0, r_ptr;
-	int rq_ppas, rq_len;
-	int i, j;
-	int ret = 0;
-	int rec_round;
-	int left_ppas = pblk_calc_sec_in_line(pblk, line) - line->cur_sec;
-
-	ppa_list = p.ppa_list;
-	meta_list = p.meta_list;
-	rqd = p.rqd;
-	data = p.data;
-	dma_ppa_list = p.dma_ppa_list;
-	dma_meta_list = p.dma_meta_list;
-
-	/* we could recover up until the line write pointer */
-	r_ptr = line->cur_sec;
-	rec_round = 0;
-
-next_rq:
-	memset(rqd, 0, pblk_g_rq_size);
-
-	rq_ppas = pblk_calc_secs(pblk, left_ppas, 0);
-	if (!rq_ppas)
-		rq_ppas = pblk->min_write_pgs;
-	rq_len = rq_ppas * geo->csecs;
-
-	bio = bio_map_kern(dev->q, data, rq_len, GFP_KERNEL);
-	if (IS_ERR(bio))
-		return PTR_ERR(bio);
-
-	bio->bi_iter.bi_sector = 0; /* internal bio */
-	bio_set_op_attrs(bio, REQ_OP_READ, 0);
-
-	rqd->bio = bio;
-	rqd->opcode = NVM_OP_PREAD;
-	rqd->meta_list = meta_list;
-	rqd->nr_ppas = rq_ppas;
-	rqd->ppa_list = ppa_list;
-	rqd->dma_ppa_list = dma_ppa_list;
-	rqd->dma_meta_list = dma_meta_list;
-
-	if (pblk_io_aligned(pblk, rq_ppas))
-		rqd->is_seq = 1;
-
-	for (i = 0; i < rqd->nr_ppas; ) {
-		struct ppa_addr ppa;
-		int pos;
-
-		w_ptr = pblk_alloc_page(pblk, line, pblk->min_write_pgs);
-		ppa = addr_to_gen_ppa(pblk, w_ptr, line->id);
+	struct pblk_line_meta *lm = &pblk->lm;
+	struct pblk_lun *rlun;
+	struct nvm_chk_meta *chunk;
+	struct ppa_addr ppa;
+	u64 line_wp;
+	int pos, i;
+
+	rlun = &pblk->luns[0];
+	ppa = rlun->bppa;
+	pos = pblk_ppa_to_pos(geo, ppa);
+	chunk = &line->chks[pos];
+
+	line_wp = chunk->wp;
+
+	for (i = 1; i < lm->blk_per_line; i++) {
+		rlun = &pblk->luns[i];
+		ppa = rlun->bppa;
 		pos = pblk_ppa_to_pos(geo, ppa);
+		chunk = &line->chks[pos];
 
-		while (test_bit(pos, line->blk_bitmap)) {
-			w_ptr += pblk->min_write_pgs;
-			ppa = addr_to_gen_ppa(pblk, w_ptr, line->id);
-			pos = pblk_ppa_to_pos(geo, ppa);
-		}
-
-		for (j = 0; j < pblk->min_write_pgs; j++, i++, w_ptr++)
-			rqd->ppa_list[i] =
-				addr_to_gen_ppa(pblk, w_ptr, line->id);
-	}
-
-	ret = pblk_submit_io_sync(pblk, rqd);
-	if (ret) {
-		pblk_err(pblk, "I/O submission failed: %d\n", ret);
-		return ret;
+		if (chunk->wp > line_wp)
+			return 1;
+		else if (chunk->wp < line_wp)
+			line_wp = chunk->wp;
 	}
 
-	atomic_dec(&pblk->inflight_io);
-
-	/* This should not happen since the read failed during normal recovery,
-	 * but the media works funny sometimes...
-	 */
-	if (!rec_round++ && !rqd->error) {
-		rec_round = 0;
-		for (i = 0; i < rqd->nr_ppas; i++, r_ptr++) {
-			u64 lba = le64_to_cpu(meta_list[i].lba);
-
-			if (lba == ADDR_EMPTY || lba > pblk->rl.nr_secs)
-				continue;
-
-			pblk_update_map(pblk, lba, rqd->ppa_list[i]);
-		}
-	}
-
-	/* Reached the end of the written line */
-	if (rqd->error == NVM_RSP_ERR_EMPTYPAGE) {
-		int pad_secs, nr_error_bits, bit;
-		int ret;
-
-		bit = find_first_bit((void *)&rqd->ppa_status, rqd->nr_ppas);
-		nr_error_bits = rqd->nr_ppas - bit;
-
-		/* Roll back failed sectors */
-		line->cur_sec -= nr_error_bits;
-		line->left_msecs += nr_error_bits;
-		bitmap_clear(line->map_bitmap, line->cur_sec, nr_error_bits);
-
-		pad_secs = pblk_pad_distance(pblk, line);
-
-		ret = pblk_recov_pad_oob(pblk, line, pad_secs);
-		if (ret)
-			pblk_err(pblk, "OOB padding failed (err:%d)\n", ret);
-
-		ret = pblk_recov_read_oob(pblk, line, p, r_ptr);
-		if (ret)
-			pblk_err(pblk, "OOB read failed (err:%d)\n", ret);
-
-		left_ppas = 0;
-	}
-
-	left_ppas -= rq_ppas;
-	if (left_ppas > 0)
-		goto next_rq;
-
-	return ret;
+	return 0;
 }
 
 static int pblk_recov_scan_oob(struct pblk *pblk, struct pblk_line *line,
-			       struct pblk_recov_alloc p, int *done)
+			       struct pblk_recov_alloc p)
 {
 	struct nvm_tgt_dev *dev = pblk->dev;
 	struct nvm_geo *geo = &dev->geo;
@@ -525,11 +340,16 @@ static int pblk_recov_scan_oob(struct pblk *pblk, struct pblk_line *line,
 	struct bio *bio;
 	void *data;
 	dma_addr_t dma_ppa_list, dma_meta_list;
-	u64 paddr;
+	__le64 *lba_list;
+	u64 paddr = 0;
+	bool padded = false;
 	int rq_ppas, rq_len;
 	int i, j;
-	int ret = 0;
-	int left_ppas = pblk_calc_sec_in_line(pblk, line);
+	int ret;
+	u64 left_ppas = pblk_sec_in_open_line(pblk, line);
+
+	if (pblk_line_wp_is_unbalanced(pblk, line))
+		pblk_warn(pblk, "recovering unbalanced line (%d)\n", line->id);
 
 	ppa_list = p.ppa_list;
 	meta_list = p.meta_list;
@@ -538,7 +358,7 @@ static int pblk_recov_scan_oob(struct pblk *pblk, struct pblk_line *line,
 	dma_ppa_list = p.dma_ppa_list;
 	dma_meta_list = p.dma_meta_list;
 
-	*done = 1;
+	lba_list = emeta_to_lbas(pblk, line->emeta->buf);
 
 next_rq:
 	memset(rqd, 0, pblk_g_rq_size);
@@ -566,11 +386,11 @@ static int pblk_recov_scan_oob(struct pblk *pblk, struct pblk_line *line,
 	if (pblk_io_aligned(pblk, rq_ppas))
 		rqd->is_seq = 1;
 
+retry_rq:
 	for (i = 0; i < rqd->nr_ppas; ) {
 		struct ppa_addr ppa;
 		int pos;
 
-		paddr = pblk_alloc_page(pblk, line, pblk->min_write_pgs);
 		ppa = addr_to_gen_ppa(pblk, paddr, line->id);
 		pos = pblk_ppa_to_pos(geo, ppa);
 
@@ -580,9 +400,9 @@ static int pblk_recov_scan_oob(struct pblk *pblk, struct pblk_line *line,
 			pos = pblk_ppa_to_pos(geo, ppa);
 		}
 
-		for (j = 0; j < pblk->min_write_pgs; j++, i++, paddr++)
+		for (j = 0; j < pblk->min_write_pgs; j++, i++)
 			rqd->ppa_list[i] =
-				addr_to_gen_ppa(pblk, paddr, line->id);
+				addr_to_gen_ppa(pblk, paddr + j, line->id);
 	}
 
 	ret = pblk_submit_io_sync(pblk, rqd);
@@ -594,31 +414,33 @@ static int pblk_recov_scan_oob(struct pblk *pblk, struct pblk_line *line,
 
 	atomic_dec(&pblk->inflight_io);
 
-	/* Reached the end of the written line */
+	/* If a read fails, do a best effort by padding the line and retrying */
 	if (rqd->error) {
-		int nr_error_bits, bit;
+		int pad_distance, ret;
 
-		bit = find_first_bit((void *)&rqd->ppa_status, rqd->nr_ppas);
-		nr_error_bits = rqd->nr_ppas - bit;
+		if (padded) {
+			pblk_log_read_err(pblk, rqd);
+			return -EINTR;
+		}
 
-		/* Roll back failed sectors */
-		line->cur_sec -= nr_error_bits;
-		line->left_msecs += nr_error_bits;
-		bitmap_clear(line->map_bitmap, line->cur_sec, nr_error_bits);
+		pad_distance = pblk_pad_distance(pblk, line);
+		ret = pblk_recov_pad_line(pblk, line, pad_distance);
+		if (ret)
+			return ret;
 
-		left_ppas = 0;
-		rqd->nr_ppas = bit;
-
-		if (rqd->error != NVM_RSP_ERR_EMPTYPAGE)
-			*done = 0;
+		padded = true;
+		goto retry_rq;
 	}
 
 	for (i = 0; i < rqd->nr_ppas; i++) {
 		u64 lba = le64_to_cpu(meta_list[i].lba);
 
+		lba_list[paddr++] = cpu_to_le64(lba);
+
 		if (lba == ADDR_EMPTY || lba > pblk->rl.nr_secs)
 			continue;
 
+		line->nr_valid_lbas++;
 		pblk_update_map(pblk, lba, rqd->ppa_list[i]);
 	}
 
@@ -626,7 +448,11 @@ static int pblk_recov_scan_oob(struct pblk *pblk, struct pblk_line *line,
 	if (left_ppas > 0)
 		goto next_rq;
 
-	return ret;
+#ifdef CONFIG_NVM_PBLK_DEBUG
+	WARN_ON(padded && !pblk_line_is_full(line));
+#endif
+
+	return 0;
 }
 
 /* Scan line for lbas on out of bound area */
@@ -640,7 +466,7 @@ static int pblk_recov_l2p_from_oob(struct pblk *pblk, struct pblk_line *line)
 	struct pblk_recov_alloc p;
 	void *data;
 	dma_addr_t dma_ppa_list, dma_meta_list;
-	int done, ret = 0;
+	int ret = 0;
 
 	meta_list = nvm_dev_dma_alloc(dev->parent, GFP_KERNEL, &dma_meta_list);
 	if (!meta_list)
@@ -655,7 +481,8 @@ static int pblk_recov_l2p_from_oob(struct pblk *pblk, struct pblk_line *line)
 		goto free_meta_list;
 	}
 
-	rqd = pblk_alloc_rqd(pblk, PBLK_READ);
+	rqd = mempool_alloc(&pblk->r_rq_pool, GFP_KERNEL);
+	memset(rqd, 0, pblk_g_rq_size);
 
 	p.ppa_list = ppa_list;
 	p.meta_list = meta_list;
@@ -664,24 +491,17 @@ static int pblk_recov_l2p_from_oob(struct pblk *pblk, struct pblk_line *line)
 	p.dma_ppa_list = dma_ppa_list;
 	p.dma_meta_list = dma_meta_list;
 
-	ret = pblk_recov_scan_oob(pblk, line, p, &done);
+	ret = pblk_recov_scan_oob(pblk, line, p);
 	if (ret) {
-		pblk_err(pblk, "could not recover L2P from OOB\n");
+		pblk_err(pblk, "could not recover L2P form OOB\n");
 		goto out;
 	}
 
-	if (!done) {
-		ret = pblk_recov_scan_all_oob(pblk, line, p);
-		if (ret) {
-			pblk_err(pblk, "could not recover L2P from OOB\n");
-			goto out;
-		}
-	}
-
 	if (pblk_line_is_full(line))
 		pblk_line_recov_close(pblk, line);
 
 out:
+	mempool_free(rqd, &pblk->r_rq_pool);
 	kfree(data);
 free_meta_list:
 	nvm_dev_dma_free(dev->parent, meta_list, dma_meta_list);
@@ -770,7 +590,7 @@ static void pblk_recov_wa_counters(struct pblk *pblk,
 }
 
 static int pblk_line_was_written(struct pblk_line *line,
-			    struct pblk *pblk)
+				 struct pblk *pblk)
 {
 
 	struct pblk_line_meta *lm = &pblk->lm;
@@ -796,6 +616,18 @@ static int pblk_line_was_written(struct pblk_line *line,
 	return 1;
 }
 
+static bool pblk_line_is_open(struct pblk *pblk, struct pblk_line *line)
+{
+	struct pblk_line_meta *lm = &pblk->lm;
+	int i;
+
+	for (i = 0; i < lm->blk_per_line; i++)
+		if (line->chks[i].state & NVM_CHK_ST_OPEN)
+			return true;
+
+	return false;
+}
+
 struct pblk_line *pblk_recov_l2p(struct pblk *pblk)
 {
 	struct pblk_line_meta *lm = &pblk->lm;
@@ -906,6 +738,11 @@ struct pblk_line *pblk_recov_l2p(struct pblk *pblk)
 		line->emeta = emeta;
 		memset(line->emeta->buf, 0, lm->emeta_len[0]);
 
+		if (pblk_line_is_open(pblk, line)) {
+			pblk_recov_l2p_from_oob(pblk, line);
+			goto next;
+		}
+
 		if (pblk_line_emeta_read(pblk, line, line->emeta->buf)) {
 			pblk_recov_l2p_from_oob(pblk, line);
 			goto next;
@@ -944,15 +781,20 @@ struct pblk_line *pblk_recov_l2p(struct pblk *pblk)
 			line->smeta = NULL;
 			line->emeta = NULL;
 		} else {
-			if (open_lines > 1)
-				pblk_err(pblk, "failed to recover L2P\n");
+			spin_lock(&line->lock);
+			line->state = PBLK_LINESTATE_OPEN;
+			spin_unlock(&line->lock);
+
+			line->emeta->mem = 0;
+			atomic_set(&line->emeta->sync, 0);
 
 			trace_pblk_line_state(pblk_disk_name(pblk), line->id,
 					line->state);
 
-			open_lines++;
-			line->meta_line = meta_line;
 			data_line = line;
+			line->meta_line = meta_line;
+
+			open_lines++;
 		}
 	}
 
@@ -1000,7 +842,7 @@ int pblk_recov_pad(struct pblk *pblk)
 	left_msecs = line->left_msecs;
 	spin_unlock(&l_mg->free_lock);
 
-	ret = pblk_recov_pad_oob(pblk, line, left_msecs);
+	ret = pblk_recov_pad_line(pblk, line, left_msecs);
 	if (ret) {
 		pblk_err(pblk, "tear down padding failed (%d)\n", ret);
 		return ret;
-- 
2.17.1

[GIT PULL 36/45] lightnvm: pblk: add SPDX license tag

[Matias Bjørling]

From: Javier González <javier@javigon.com>

Add GLP-2.0 SPDX license tag to all pblk files

Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk-cache.c    | 1 +
 drivers/lightnvm/pblk-core.c     | 1 +
 drivers/lightnvm/pblk-gc.c       | 1 +
 drivers/lightnvm/pblk-init.c     | 1 +
 drivers/lightnvm/pblk-map.c      | 1 +
 drivers/lightnvm/pblk-rb.c       | 1 +
 drivers/lightnvm/pblk-read.c     | 1 +
 drivers/lightnvm/pblk-recovery.c | 1 +
 drivers/lightnvm/pblk-rl.c       | 1 +
 drivers/lightnvm/pblk-sysfs.c    | 1 +
 drivers/lightnvm/pblk-write.c    | 1 +
 drivers/lightnvm/pblk.h          | 1 +
 12 files changed, 12 insertions(+)

diff --git a/drivers/lightnvm/pblk-cache.c b/drivers/lightnvm/pblk-cache.c
index f565a56b898a..c9fa26f95659 100644
--- a/drivers/lightnvm/pblk-cache.c
+++ b/drivers/lightnvm/pblk-cache.c
@@ -1,3 +1,4 @@
+// SPDX-License-Identifier: GPL-2.0
 /*
  * Copyright (C) 2016 CNEX Labs
  * Initial release: Javier Gonzalez <javier@cnexlabs.com>
diff --git a/drivers/lightnvm/pblk-core.c b/drivers/lightnvm/pblk-core.c
index a3ce4a36dd33..4045a9b1ee74 100644
--- a/drivers/lightnvm/pblk-core.c
+++ b/drivers/lightnvm/pblk-core.c
@@ -1,3 +1,4 @@
+// SPDX-License-Identifier: GPL-2.0
 /*
  * Copyright (C) 2016 CNEX Labs
  * Initial release: Javier Gonzalez <javier@cnexlabs.com>
diff --git a/drivers/lightnvm/pblk-gc.c b/drivers/lightnvm/pblk-gc.c
index e05d06bd5b83..2fa118c8eb71 100644
--- a/drivers/lightnvm/pblk-gc.c
+++ b/drivers/lightnvm/pblk-gc.c
@@ -1,3 +1,4 @@
+// SPDX-License-Identifier: GPL-2.0
 /*
  * Copyright (C) 2016 CNEX Labs
  * Initial release: Javier Gonzalez <javier@cnexlabs.com>
diff --git a/drivers/lightnvm/pblk-init.c b/drivers/lightnvm/pblk-init.c
index 7ef8249108f0..f84c428a76f1 100644
--- a/drivers/lightnvm/pblk-init.c
+++ b/drivers/lightnvm/pblk-init.c
@@ -1,3 +1,4 @@
+// SPDX-License-Identifier: GPL-2.0
 /*
  * Copyright (C) 2015 IT University of Copenhagen (rrpc.c)
  * Copyright (C) 2016 CNEX Labs
diff --git a/drivers/lightnvm/pblk-map.c b/drivers/lightnvm/pblk-map.c
index ff677ca6e4e1..6dcbd44e3acb 100644
--- a/drivers/lightnvm/pblk-map.c
+++ b/drivers/lightnvm/pblk-map.c
@@ -1,3 +1,4 @@
+// SPDX-License-Identifier: GPL-2.0
 /*
  * Copyright (C) 2016 CNEX Labs
  * Initial release: Javier Gonzalez <javier@cnexlabs.com>
diff --git a/drivers/lightnvm/pblk-rb.c b/drivers/lightnvm/pblk-rb.c
index a7648e12f54f..c26eab2ba8bd 100644
--- a/drivers/lightnvm/pblk-rb.c
+++ b/drivers/lightnvm/pblk-rb.c
@@ -1,3 +1,4 @@
+// SPDX-License-Identifier: GPL-2.0
 /*
  * Copyright (C) 2016 CNEX Labs
  * Initial release: Javier Gonzalez <javier@cnexlabs.com>
diff --git a/drivers/lightnvm/pblk-read.c b/drivers/lightnvm/pblk-read.c
index f5fe01d3a07f..9fba614adeeb 100644
--- a/drivers/lightnvm/pblk-read.c
+++ b/drivers/lightnvm/pblk-read.c
@@ -1,3 +1,4 @@
+// SPDX-License-Identifier: GPL-2.0
 /*
  * Copyright (C) 2016 CNEX Labs
  * Initial release: Javier Gonzalez <javier@cnexlabs.com>
diff --git a/drivers/lightnvm/pblk-recovery.c b/drivers/lightnvm/pblk-recovery.c
index fccf65bc70b3..5740b7509bd8 100644
--- a/drivers/lightnvm/pblk-recovery.c
+++ b/drivers/lightnvm/pblk-recovery.c
@@ -1,3 +1,4 @@
+// SPDX-License-Identifier: GPL-2.0
 /*
  * Copyright (C) 2016 CNEX Labs
  * Initial: Javier Gonzalez <javier@cnexlabs.com>
diff --git a/drivers/lightnvm/pblk-rl.c b/drivers/lightnvm/pblk-rl.c
index a32790f7b7fc..db55a1c89997 100644
--- a/drivers/lightnvm/pblk-rl.c
+++ b/drivers/lightnvm/pblk-rl.c
@@ -1,3 +1,4 @@
+// SPDX-License-Identifier: GPL-2.0
 /*
  * Copyright (C) 2016 CNEX Labs
  * Initial release: Javier Gonzalez <javier@cnexlabs.com>
diff --git a/drivers/lightnvm/pblk-sysfs.c b/drivers/lightnvm/pblk-sysfs.c
index e56eeeeb914e..cba83ac43e62 100644
--- a/drivers/lightnvm/pblk-sysfs.c
+++ b/drivers/lightnvm/pblk-sysfs.c
@@ -1,3 +1,4 @@
+// SPDX-License-Identifier: GPL-2.0
 /*
  * Copyright (C) 2016 CNEX Labs
  * Initial release: Javier Gonzalez <javier@cnexlabs.com>
diff --git a/drivers/lightnvm/pblk-write.c b/drivers/lightnvm/pblk-write.c
index a276f25d3931..277abc8633f7 100644
--- a/drivers/lightnvm/pblk-write.c
+++ b/drivers/lightnvm/pblk-write.c
@@ -1,3 +1,4 @@
+// SPDX-License-Identifier: GPL-2.0
 /*
  * Copyright (C) 2016 CNEX Labs
  * Initial release: Javier Gonzalez <javier@cnexlabs.com>
diff --git a/drivers/lightnvm/pblk.h b/drivers/lightnvm/pblk.h
index 4c015c457197..e710f6db2a35 100644
--- a/drivers/lightnvm/pblk.h
+++ b/drivers/lightnvm/pblk.h
@@ -1,3 +1,4 @@
+/* SPDX-License-Identifier: GPL-2.0 */
 /*
  * Copyright (C) 2015 IT University of Copenhagen (rrpc.h)
  * Copyright (C) 2016 CNEX Labs
-- 
2.17.1

[GIT PULL 37/45] lightnvm: pblk: fix race on sysfs line state

[Matias Bjørling]

From: Javier González <javier@javigon.com>

pblk exposes a sysfs interface that represents its internal state. Part
of this state is the map bitmap for the current open line, which should
be protected by the line lock to avoid a race when freeing the line
metadata. Currently, it is not.

This patch makes sure that the line state is consistent and NULL
bitmap pointers are not dereferenced.

Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk-core.c  | 5 +++--
 drivers/lightnvm/pblk-sysfs.c | 8 +++++++-
 2 files changed, 10 insertions(+), 3 deletions(-)

diff --git a/drivers/lightnvm/pblk-core.c b/drivers/lightnvm/pblk-core.c
index 4045a9b1ee74..6944aac43b01 100644
--- a/drivers/lightnvm/pblk-core.c
+++ b/drivers/lightnvm/pblk-core.c
@@ -1611,13 +1611,14 @@ struct pblk_line *pblk_line_replace_data(struct pblk *pblk)
 	struct pblk_line *cur, *new = NULL;
 	unsigned int left_seblks;
 
-	cur = l_mg->data_line;
 	new = l_mg->data_next;
 	if (!new)
 		goto out;
+
+	spin_lock(&l_mg->free_lock);
+	cur = l_mg->data_line;
 	l_mg->data_line = new;
 
-	spin_lock(&l_mg->free_lock);
 	pblk_line_setup_metadata(new, l_mg, &pblk->lm);
 	spin_unlock(&l_mg->free_lock);
 
diff --git a/drivers/lightnvm/pblk-sysfs.c b/drivers/lightnvm/pblk-sysfs.c
index cba83ac43e62..2d2818155aa8 100644
--- a/drivers/lightnvm/pblk-sysfs.c
+++ b/drivers/lightnvm/pblk-sysfs.c
@@ -263,8 +263,14 @@ static ssize_t pblk_sysfs_lines(struct pblk *pblk, char *page)
 		sec_in_line = l_mg->data_line->sec_in_line;
 		meta_weight = bitmap_weight(&l_mg->meta_bitmap,
 							PBLK_DATA_LINES);
-		map_weight = bitmap_weight(l_mg->data_line->map_bitmap,
+
+		spin_lock(&l_mg->data_line->lock);
+		if (l_mg->data_line->map_bitmap)
+			map_weight = bitmap_weight(l_mg->data_line->map_bitmap,
 							lm->sec_per_line);
+		else
+			map_weight = 0;
+		spin_unlock(&l_mg->data_line->lock);
 	}
 	spin_unlock(&l_mg->free_lock);
 
-- 
2.17.1

[GIT PULL 38/45] lightnvm: pblk: remove unused function

[Matias Bjørling]

From: Javier González <javier@javigon.com>

Removed unused function in pblk-rb.c

Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk-rb.c | 26 --------------------------
 drivers/lightnvm/pblk.h    |  2 --
 2 files changed, 28 deletions(-)

diff --git a/drivers/lightnvm/pblk-rb.c b/drivers/lightnvm/pblk-rb.c
index c26eab2ba8bd..82829e8151db 100644
--- a/drivers/lightnvm/pblk-rb.c
+++ b/drivers/lightnvm/pblk-rb.c
@@ -729,32 +729,6 @@ unsigned int pblk_rb_flush_point_count(struct pblk_rb *rb)
 	return (submitted < to_flush) ? (to_flush - submitted) : 0;
 }
 
-/*
- * Scan from the current position of the sync pointer to find the entry that
- * corresponds to the given ppa. This is necessary since write requests can be
- * completed out of order. The assumption is that the ppa is close to the sync
- * pointer thus the search will not take long.
- *
- * The caller of this function must guarantee that the sync pointer will no
- * reach the entry while it is using the metadata associated with it. With this
- * assumption in mind, there is no need to take the sync lock.
- */
-struct pblk_rb_entry *pblk_rb_sync_scan_entry(struct pblk_rb *rb,
-					      struct ppa_addr *ppa)
-{
-	unsigned int sync, subm, count;
-	unsigned int i;
-
-	sync = READ_ONCE(rb->sync);
-	subm = READ_ONCE(rb->subm);
-	count = pblk_rb_ring_count(subm, sync, rb->nr_entries);
-
-	for (i = 0; i < count; i++)
-		sync = (sync + 1) & (rb->nr_entries - 1);
-
-	return NULL;
-}
-
 int pblk_rb_tear_down_check(struct pblk_rb *rb)
 {
 	struct pblk_rb_entry *entry;
diff --git a/drivers/lightnvm/pblk.h b/drivers/lightnvm/pblk.h
index e710f6db2a35..22586b78198a 100644
--- a/drivers/lightnvm/pblk.h
+++ b/drivers/lightnvm/pblk.h
@@ -760,8 +760,6 @@ unsigned int pblk_rb_read_commit(struct pblk_rb *rb, unsigned int entries);
 
 unsigned int pblk_rb_sync_init(struct pblk_rb *rb, unsigned long *flags);
 unsigned int pblk_rb_sync_advance(struct pblk_rb *rb, unsigned int nr_entries);
-struct pblk_rb_entry *pblk_rb_sync_scan_entry(struct pblk_rb *rb,
-					      struct ppa_addr *ppa);
 void pblk_rb_sync_end(struct pblk_rb *rb, unsigned long *flags);
 unsigned int pblk_rb_flush_point_count(struct pblk_rb *rb);
 
-- 
2.17.1

[GIT PULL 39/45] lightnvm: pblk: encapsulate rb pointer operations

[Matias Bjørling]

From: Javier González <javier@javigon.com>

pblk's read/write buffer is always a power-of-2, thus wrapping up the
buffer can be done with a bit mask. Since this is an implementation
detail internal to the write buffer, make a helper that hides pointer
increment + wrap, and allows to transparently relax this assumption in
the future.

Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk-rb.c    | 21 +++++++++++++--------
 drivers/lightnvm/pblk-write.c |  7 ++-----
 drivers/lightnvm/pblk.h       |  2 ++
 3 files changed, 17 insertions(+), 13 deletions(-)

diff --git a/drivers/lightnvm/pblk-rb.c b/drivers/lightnvm/pblk-rb.c
index 82829e8151db..e46d8cb9d28b 100644
--- a/drivers/lightnvm/pblk-rb.c
+++ b/drivers/lightnvm/pblk-rb.c
@@ -169,6 +169,12 @@ static unsigned int pblk_rb_space(struct pblk_rb *rb)
 	return pblk_rb_ring_space(rb, mem, sync, rb->nr_entries);
 }
 
+unsigned int pblk_rb_ptr_wrap(struct pblk_rb *rb, unsigned int p,
+			      unsigned int nr_entries)
+{
+	return (p + nr_entries) & (rb->nr_entries - 1);
+}
+
 /*
  * Buffer count is calculated with respect to the submission entry signaling the
  * entries that are available to send to the media
@@ -195,8 +201,7 @@ unsigned int pblk_rb_read_commit(struct pblk_rb *rb, unsigned int nr_entries)
 
 	subm = READ_ONCE(rb->subm);
 	/* Commit read means updating submission pointer */
-	smp_store_release(&rb->subm,
-				(subm + nr_entries) & (rb->nr_entries - 1));
+	smp_store_release(&rb->subm, pblk_rb_ptr_wrap(rb, subm, nr_entries));
 
 	return subm;
 }
@@ -229,7 +234,7 @@ static int __pblk_rb_update_l2p(struct pblk_rb *rb, unsigned int to_update)
 		line = pblk_ppa_to_line(pblk, w_ctx->ppa);
 		kref_put(&line->ref, pblk_line_put);
 		clean_wctx(w_ctx);
-		rb->l2p_update = (rb->l2p_update + 1) & (rb->nr_entries - 1);
+		rb->l2p_update = pblk_rb_ptr_wrap(rb, rb->l2p_update, 1);
 	}
 
 	pblk_rl_out(&pblk->rl, user_io, gc_io);
@@ -408,7 +413,7 @@ static int pblk_rb_may_write(struct pblk_rb *rb, unsigned int nr_entries,
 		return 0;
 
 	/* Protect from read count */
-	smp_store_release(&rb->mem, (*pos + nr_entries) & (rb->nr_entries - 1));
+	smp_store_release(&rb->mem, pblk_rb_ptr_wrap(rb, *pos, nr_entries));
 	return 1;
 }
 
@@ -432,7 +437,7 @@ static int pblk_rb_may_write_flush(struct pblk_rb *rb, unsigned int nr_entries,
 	if (!__pblk_rb_may_write(rb, nr_entries, pos))
 		return 0;
 
-	mem = (*pos + nr_entries) & (rb->nr_entries - 1);
+	mem = pblk_rb_ptr_wrap(rb, *pos, nr_entries);
 	*io_ret = NVM_IO_DONE;
 
 	if (bio->bi_opf & REQ_PREFLUSH) {
@@ -572,7 +577,7 @@ unsigned int pblk_rb_read_to_bio(struct pblk_rb *rb, struct nvm_rq *rqd,
 		/* Release flags on context. Protect from writes */
 		smp_store_release(&entry->w_ctx.flags, flags);
 
-		pos = (pos + 1) & (rb->nr_entries - 1);
+		pos = pblk_rb_ptr_wrap(rb, pos, 1);
 	}
 
 	if (pad) {
@@ -652,7 +657,7 @@ int pblk_rb_copy_to_bio(struct pblk_rb *rb, struct bio *bio, sector_t lba,
 
 struct pblk_w_ctx *pblk_rb_w_ctx(struct pblk_rb *rb, unsigned int pos)
 {
-	unsigned int entry = pos & (rb->nr_entries - 1);
+	unsigned int entry = pblk_rb_ptr_wrap(rb, pos, 0);
 
 	return &rb->entries[entry].w_ctx;
 }
@@ -698,7 +703,7 @@ unsigned int pblk_rb_sync_advance(struct pblk_rb *rb, unsigned int nr_entries)
 		}
 	}
 
-	sync = (sync + nr_entries) & (rb->nr_entries - 1);
+	sync = pblk_rb_ptr_wrap(rb, sync, nr_entries);
 
 	/* Protect from counts */
 	smp_store_release(&rb->sync, sync);
diff --git a/drivers/lightnvm/pblk-write.c b/drivers/lightnvm/pblk-write.c
index 277abc8633f7..fa8726493b39 100644
--- a/drivers/lightnvm/pblk-write.c
+++ b/drivers/lightnvm/pblk-write.c
@@ -140,12 +140,11 @@ static void pblk_prepare_resubmit(struct pblk *pblk, unsigned int sentry,
 	struct pblk_w_ctx *w_ctx;
 	struct ppa_addr ppa_l2p;
 	int flags;
-	unsigned int pos, i;
+	unsigned int i;
 
 	spin_lock(&pblk->trans_lock);
-	pos = sentry;
 	for (i = 0; i < nr_entries; i++) {
-		entry = &rb->entries[pos];
+		entry = &rb->entries[pblk_rb_ptr_wrap(rb, sentry, i)];
 		w_ctx = &entry->w_ctx;
 
 		/* Check if the lba has been overwritten */
@@ -164,8 +163,6 @@ static void pblk_prepare_resubmit(struct pblk *pblk, unsigned int sentry,
 		 */
 		line = pblk_ppa_to_line(pblk, w_ctx->ppa);
 		kref_put(&line->ref, pblk_line_put);
-
-		pos = (pos + 1) & (rb->nr_entries - 1);
 	}
 	spin_unlock(&pblk->trans_lock);
 }
diff --git a/drivers/lightnvm/pblk.h b/drivers/lightnvm/pblk.h
index 22586b78198a..7660811aa8af 100644
--- a/drivers/lightnvm/pblk.h
+++ b/drivers/lightnvm/pblk.h
@@ -760,6 +760,8 @@ unsigned int pblk_rb_read_commit(struct pblk_rb *rb, unsigned int entries);
 
 unsigned int pblk_rb_sync_init(struct pblk_rb *rb, unsigned long *flags);
 unsigned int pblk_rb_sync_advance(struct pblk_rb *rb, unsigned int nr_entries);
+unsigned int pblk_rb_ptr_wrap(struct pblk_rb *rb, unsigned int p,
+			      unsigned int nr_entries);
 void pblk_rb_sync_end(struct pblk_rb *rb, unsigned long *flags);
 unsigned int pblk_rb_flush_point_count(struct pblk_rb *rb);
 
-- 
2.17.1

[GIT PULL 40/45] lightnvm: pblk: move ring buffer alloc/free rb init

[Matias Bjørling]

From: Javier González <javier@javigon.com>

pblk's read/write buffer currently takes a buffer and its size and uses
it to create the metadata around it to use it as a ring buffer. This
puts the responsibility of allocating/freeing ring buffer memory on the
ring buffer user. Instead, move it inside of the ring buffer helpers
(pblk-rb.c). This simplifies creation/destruction routines.

Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk-init.c | 18 ++----------
 drivers/lightnvm/pblk-rb.c   | 53 ++++++++++++++++++++++--------------
 drivers/lightnvm/pblk.h      |  7 ++---
 3 files changed, 38 insertions(+), 40 deletions(-)

diff --git a/drivers/lightnvm/pblk-init.c b/drivers/lightnvm/pblk-init.c
index f84c428a76f1..b2c49fc006c9 100644
--- a/drivers/lightnvm/pblk-init.c
+++ b/drivers/lightnvm/pblk-init.c
@@ -185,17 +185,14 @@ static void pblk_rwb_free(struct pblk *pblk)
 	if (pblk_rb_tear_down_check(&pblk->rwb))
 		pblk_err(pblk, "write buffer error on tear down\n");
 
-	pblk_rb_data_free(&pblk->rwb);
-	vfree(pblk_rb_entries_ref(&pblk->rwb));
+	pblk_rb_free(&pblk->rwb);
 }
 
 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, buffer_size;
-	unsigned int power_size, power_seg_sz;
+	unsigned long buffer_size;
 	int pgs_in_buffer;
 
 	pgs_in_buffer = max(geo->mw_cunits, geo->ws_opt) * geo->all_luns;
@@ -205,16 +202,7 @@ static int pblk_rwb_init(struct pblk *pblk)
 	else
 		buffer_size = pgs_in_buffer;
 
-	nr_entries = pblk_rb_calculate_size(buffer_size);
-
-	entries = vzalloc(array_size(nr_entries, sizeof(struct pblk_rb_entry)));
-	if (!entries)
-		return -ENOMEM;
-
-	power_size = get_count_order(nr_entries);
-	power_seg_sz = get_count_order(geo->csecs);
-
-	return pblk_rb_init(&pblk->rwb, entries, power_size, power_seg_sz);
+	return pblk_rb_init(&pblk->rwb, buffer_size, geo->csecs);
 }
 
 /* Minimum pages needed within a lun */
diff --git a/drivers/lightnvm/pblk-rb.c b/drivers/lightnvm/pblk-rb.c
index e46d8cb9d28b..f653faa6a9ed 100644
--- a/drivers/lightnvm/pblk-rb.c
+++ b/drivers/lightnvm/pblk-rb.c
@@ -23,7 +23,7 @@
 
 static DECLARE_RWSEM(pblk_rb_lock);
 
-void pblk_rb_data_free(struct pblk_rb *rb)
+static void pblk_rb_data_free(struct pblk_rb *rb)
 {
 	struct pblk_rb_pages *p, *t;
 
@@ -36,22 +36,46 @@ void pblk_rb_data_free(struct pblk_rb *rb)
 	up_write(&pblk_rb_lock);
 }
 
+void pblk_rb_free(struct pblk_rb *rb)
+{
+	pblk_rb_data_free(rb);
+	vfree(rb->entries);
+}
+
+/*
+ * pblk_rb_calculate_size -- calculate the size of the write buffer
+ */
+static unsigned int pblk_rb_calculate_size(unsigned int nr_entries)
+{
+	/* Alloc a write buffer that can at least fit 128 entries */
+	return (1 << max(get_count_order(nr_entries), 7));
+}
+
 /*
  * Initialize ring buffer. The data and metadata buffers must be previously
  * allocated and their size must be a power of two
  * (Documentation/core-api/circular-buffers.rst)
  */
-int pblk_rb_init(struct pblk_rb *rb, struct pblk_rb_entry *rb_entry_base,
-		 unsigned int power_size, unsigned int power_seg_sz)
+int pblk_rb_init(struct pblk_rb *rb, unsigned int size, unsigned int seg_size)
 {
 	struct pblk *pblk = container_of(rb, struct pblk, rwb);
+	struct pblk_rb_entry *entries;
 	unsigned int init_entry = 0;
-	unsigned int alloc_order = power_size;
 	unsigned int max_order = MAX_ORDER - 1;
-	unsigned int order, iter;
+	unsigned int power_size, power_seg_sz;
+	unsigned int alloc_order, order, iter;
+	unsigned int nr_entries;
+
+	nr_entries = pblk_rb_calculate_size(size);
+	entries = vzalloc(array_size(nr_entries, sizeof(struct pblk_rb_entry)));
+	if (!entries)
+		return -ENOMEM;
+
+	power_size = get_count_order(size);
+	power_seg_sz = get_count_order(seg_size);
 
 	down_write(&pblk_rb_lock);
-	rb->entries = rb_entry_base;
+	rb->entries = entries;
 	rb->seg_size = (1 << power_seg_sz);
 	rb->nr_entries = (1 << power_size);
 	rb->mem = rb->subm = rb->sync = rb->l2p_update = 0;
@@ -62,6 +86,7 @@ int pblk_rb_init(struct pblk_rb *rb, struct pblk_rb_entry *rb_entry_base,
 
 	INIT_LIST_HEAD(&rb->pages);
 
+	alloc_order = power_size;
 	if (alloc_order >= max_order) {
 		order = max_order;
 		iter = (1 << (alloc_order - max_order));
@@ -80,6 +105,7 @@ int pblk_rb_init(struct pblk_rb *rb, struct pblk_rb_entry *rb_entry_base,
 		page_set = kmalloc(sizeof(struct pblk_rb_pages), GFP_KERNEL);
 		if (!page_set) {
 			up_write(&pblk_rb_lock);
+			vfree(entries);
 			return -ENOMEM;
 		}
 
@@ -89,6 +115,7 @@ int pblk_rb_init(struct pblk_rb *rb, struct pblk_rb_entry *rb_entry_base,
 			kfree(page_set);
 			pblk_rb_data_free(rb);
 			up_write(&pblk_rb_lock);
+			vfree(entries);
 			return -ENOMEM;
 		}
 		kaddr = page_address(page_set->pages);
@@ -125,20 +152,6 @@ int pblk_rb_init(struct pblk_rb *rb, struct pblk_rb_entry *rb_entry_base,
 	return 0;
 }
 
-/*
- * pblk_rb_calculate_size -- calculate the size of the write buffer
- */
-unsigned int pblk_rb_calculate_size(unsigned int nr_entries)
-{
-	/* Alloc a write buffer that can at least fit 128 entries */
-	return (1 << max(get_count_order(nr_entries), 7));
-}
-
-void *pblk_rb_entries_ref(struct pblk_rb *rb)
-{
-	return rb->entries;
-}
-
 static void clean_wctx(struct pblk_w_ctx *w_ctx)
 {
 	int flags;
diff --git a/drivers/lightnvm/pblk.h b/drivers/lightnvm/pblk.h
index 7660811aa8af..0f98ea24ee59 100644
--- a/drivers/lightnvm/pblk.h
+++ b/drivers/lightnvm/pblk.h
@@ -734,10 +734,7 @@ struct pblk_line_ws {
 /*
  * pblk ring buffer operations
  */
-int pblk_rb_init(struct pblk_rb *rb, struct pblk_rb_entry *rb_entry_base,
-		 unsigned int power_size, unsigned int power_seg_sz);
-unsigned int pblk_rb_calculate_size(unsigned int nr_entries);
-void *pblk_rb_entries_ref(struct pblk_rb *rb);
+int pblk_rb_init(struct pblk_rb *rb, unsigned int size, unsigned int seg_sz);
 int pblk_rb_may_write_user(struct pblk_rb *rb, struct bio *bio,
 			   unsigned int nr_entries, unsigned int *pos);
 int pblk_rb_may_write_gc(struct pblk_rb *rb, unsigned int nr_entries,
@@ -771,7 +768,7 @@ unsigned int pblk_rb_wrap_pos(struct pblk_rb *rb, unsigned int pos);
 
 int pblk_rb_tear_down_check(struct pblk_rb *rb);
 int pblk_rb_pos_oob(struct pblk_rb *rb, u64 pos);
-void pblk_rb_data_free(struct pblk_rb *rb);
+void pblk_rb_free(struct pblk_rb *rb);
 ssize_t pblk_rb_sysfs(struct pblk_rb *rb, char *buf);
 
 /*
-- 
2.17.1

[GIT PULL 41/45] lightnvm: pblk: guarantee mw_cunits on read buffer

[Matias Bjørling]

From: Javier González <javier@javigon.com>

OCSSD 2.0 defines the amount of data that the host must buffer per chunk
to guarantee reads through the geometry field mw_cunits. This value is
the base that pblk uses to determine the size of its read buffer.
Currently, this size is set to be the closes power-of-2 to mw_cunits
times the number of parallel units available to the pblk instance for
each open line (currently one). When an entry (4KB) is put in the
buffer, the L2P table points to it. As the buffer wraps up, the L2P is
updated to point to addresses on the device, thus guaranteeing mw_cunits
at a chunk level.

However, given that pblk cannot write to the device under ws_min
(normally ws_opt), there might be a window in which the buffer starts
wrapping up and updating L2P entries before the mw_cunits value in a
chunk has been surpassed.

In order not to violate the mw_cunits constrain in this case, account
for ws_opt on the read buffer creation.

Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk-init.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/lightnvm/pblk-init.c b/drivers/lightnvm/pblk-init.c
index b2c49fc006c9..e0db0cb3122d 100644
--- a/drivers/lightnvm/pblk-init.c
+++ b/drivers/lightnvm/pblk-init.c
@@ -195,7 +195,8 @@ static int pblk_rwb_init(struct pblk *pblk)
 	unsigned long buffer_size;
 	int pgs_in_buffer;
 
-	pgs_in_buffer = max(geo->mw_cunits, geo->ws_opt) * geo->all_luns;
+	pgs_in_buffer = (max(geo->mw_cunits, geo->ws_opt) + geo->ws_opt)
+								* geo->all_luns;
 
 	if (write_buffer_size && (write_buffer_size > pgs_in_buffer))
 		buffer_size = write_buffer_size;
-- 
2.17.1

[GIT PULL 42/45] lightnvm: do no update csecs and sos on 1.2

[Matias Bjørling]

From: Javier González <javier@javigon.com>

1.2 devices exposes their data and metadata size through the separate
identify command. Make sure that the NVMe LBA format does not override
these values.

Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/nvme/host/lightnvm.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/drivers/nvme/host/lightnvm.c b/drivers/nvme/host/lightnvm.c
index 7d0a4d3b0a48..a4f3b263cd6c 100644
--- a/drivers/nvme/host/lightnvm.c
+++ b/drivers/nvme/host/lightnvm.c
@@ -977,6 +977,9 @@ void nvme_nvm_update_nvm_info(struct nvme_ns *ns)
 	struct nvm_dev *ndev = ns->ndev;
 	struct nvm_geo *geo = &ndev->geo;
 
+	if (geo->version == NVM_OCSSD_SPEC_12)
+		return;
+
 	geo->csecs = 1 << ns->lba_shift;
 	geo->sos = ns->ms;
 }
-- 
2.17.1

[GIT PULL 43/45] lightnvm: pblk: fix error handling of pblk_lines_init()

[Matias Bjørling]

From: Wei Yongjun <weiyongjun1@huawei.com>

In the too many bad blocks error handling case, we should release all
the allocated resources, otherwise it will cause memory leak.

Fixes: 2deeefc02dff ("lightnvm: pblk: fail gracefully on line alloc. failure")
Signed-off-by: Wei Yongjun <weiyongjun1@huawei.com>
Reviewed-by: Hans Holmberg <hans.holmberg@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk-init.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/lightnvm/pblk-init.c b/drivers/lightnvm/pblk-init.c
index e0db0cb3122d..e3573880dbda 100644
--- a/drivers/lightnvm/pblk-init.c
+++ b/drivers/lightnvm/pblk-init.c
@@ -1024,7 +1024,8 @@ static int pblk_lines_init(struct pblk *pblk)
 
 	if (!nr_free_chks) {
 		pblk_err(pblk, "too many bad blocks prevent for sane instance\n");
-		return -EINTR;
+		ret = -EINTR;
+		goto fail_free_lines;
 	}
 
 	pblk_set_provision(pblk, nr_free_chks);
-- 
2.17.1

[GIT PULL 44/45] lightnvm: pblk: consider max hw sectors supported for max_write_pgs

[Matias Bjørling]

From: Zhoujie Wu <zjwu@marvell.com>

When do GC, the number of read/write sectors are determined
by max_write_pgs(see gc_rq preparation in pblk_gc_line_prepare_ws).

Due to max_write_pgs doesn't consider max hw sectors
supported by nvme controller(128K), which leads to GC
tries to read 64 * 4K in one command, and see below error
caused by pblk_bio_map_addr in function pblk_submit_read_gc.

[ 2923.005376] pblk: could not add page to bio
[ 2923.005377] pblk: could not allocate GC bio (18446744073709551604)

Signed-off-by: Zhoujie Wu <zjwu@marvell.com>
Reviewed-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk-init.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/lightnvm/pblk-init.c b/drivers/lightnvm/pblk-init.c
index e3573880dbda..e5239aba806b 100644
--- a/drivers/lightnvm/pblk-init.c
+++ b/drivers/lightnvm/pblk-init.c
@@ -407,6 +407,8 @@ static int pblk_core_init(struct pblk *pblk)
 	pblk->min_write_pgs = geo->ws_opt;
 	max_write_ppas = pblk->min_write_pgs * geo->all_luns;
 	pblk->max_write_pgs = min_t(int, max_write_ppas, NVM_MAX_VLBA);
+	pblk->max_write_pgs = min_t(int, pblk->max_write_pgs,
+		queue_max_hw_sectors(dev->q) / (geo->csecs >> SECTOR_SHIFT));
 	pblk_set_sec_per_write(pblk, pblk->min_write_pgs);
 
 	pblk->pad_dist = kcalloc(pblk->min_write_pgs - 1, sizeof(atomic64_t),
-- 
2.17.1

[GIT PULL 45/45] lightnvm: pblk: guarantee that backpointer is respected on writer stall

[Matias Bjørling]

From: Javier González <javier@javigon.com>

pblk's write buffer must guarantee that it respects the device's
constrains for reads (i.e., mw_cunits). This is done by maintaining a
backpointer that updates the L2P table as entries wrap up, making them
point to the media instead of pointing to the write buffer.

This mechanism can race in case that the write thread stalls, as the
write pointer will protect the last written entry, thus disregarding the
read constrains.

This patch adds an extra check on wrap up, making sure that the
threshold is respected at all times, preventing new entries to overwrite
committed data, also in case of write thread stall.

Reported-by: Heiner Litz <hlitz@ucsc.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Reviewed-by: Heiner Litz <hlitz@ucsc.edu>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
---
 drivers/lightnvm/pblk-init.c | 5 +++--
 drivers/lightnvm/pblk-rb.c   | 9 +++++++--
 drivers/lightnvm/pblk.h      | 8 +++++++-
 3 files changed, 17 insertions(+), 5 deletions(-)

diff --git a/drivers/lightnvm/pblk-init.c b/drivers/lightnvm/pblk-init.c
index e5239aba806b..13822594647c 100644
--- a/drivers/lightnvm/pblk-init.c
+++ b/drivers/lightnvm/pblk-init.c
@@ -193,8 +193,9 @@ static int pblk_rwb_init(struct pblk *pblk)
 	struct nvm_tgt_dev *dev = pblk->dev;
 	struct nvm_geo *geo = &dev->geo;
 	unsigned long buffer_size;
-	int pgs_in_buffer;
+	int pgs_in_buffer, threshold;
 
+	threshold = geo->mw_cunits * geo->all_luns;
 	pgs_in_buffer = (max(geo->mw_cunits, geo->ws_opt) + geo->ws_opt)
 								* geo->all_luns;
 
@@ -203,7 +204,7 @@ static int pblk_rwb_init(struct pblk *pblk)
 	else
 		buffer_size = pgs_in_buffer;
 
-	return pblk_rb_init(&pblk->rwb, buffer_size, geo->csecs);
+	return pblk_rb_init(&pblk->rwb, buffer_size, threshold, geo->csecs);
 }
 
 /* Minimum pages needed within a lun */
diff --git a/drivers/lightnvm/pblk-rb.c b/drivers/lightnvm/pblk-rb.c
index f653faa6a9ed..b1f4b51783f4 100644
--- a/drivers/lightnvm/pblk-rb.c
+++ b/drivers/lightnvm/pblk-rb.c
@@ -56,7 +56,8 @@ static unsigned int pblk_rb_calculate_size(unsigned int nr_entries)
  * allocated and their size must be a power of two
  * (Documentation/core-api/circular-buffers.rst)
  */
-int pblk_rb_init(struct pblk_rb *rb, unsigned int size, unsigned int seg_size)
+int pblk_rb_init(struct pblk_rb *rb, unsigned int size, unsigned int threshold,
+		 unsigned int seg_size)
 {
 	struct pblk *pblk = container_of(rb, struct pblk, rwb);
 	struct pblk_rb_entry *entries;
@@ -79,6 +80,7 @@ int pblk_rb_init(struct pblk_rb *rb, unsigned int size, unsigned int seg_size)
 	rb->seg_size = (1 << power_seg_sz);
 	rb->nr_entries = (1 << power_size);
 	rb->mem = rb->subm = rb->sync = rb->l2p_update = 0;
+	rb->back_thres = threshold;
 	rb->flush_point = EMPTY_ENTRY;
 
 	spin_lock_init(&rb->w_lock);
@@ -404,11 +406,14 @@ static int __pblk_rb_may_write(struct pblk_rb *rb, unsigned int nr_entries,
 {
 	unsigned int mem;
 	unsigned int sync;
+	unsigned int threshold;
 
 	sync = READ_ONCE(rb->sync);
 	mem = READ_ONCE(rb->mem);
 
-	if (pblk_rb_ring_space(rb, mem, sync, rb->nr_entries) < nr_entries)
+	threshold = nr_entries + rb->back_thres;
+
+	if (pblk_rb_ring_space(rb, mem, sync, rb->nr_entries) < threshold)
 		return 0;
 
 	if (pblk_rb_update_l2p(rb, nr_entries, mem, sync))
diff --git a/drivers/lightnvm/pblk.h b/drivers/lightnvm/pblk.h
index 0f98ea24ee59..02bb2e98f8a9 100644
--- a/drivers/lightnvm/pblk.h
+++ b/drivers/lightnvm/pblk.h
@@ -203,6 +203,11 @@ struct pblk_rb {
 					 * will be 4KB
 					 */
 
+	unsigned int back_thres;	/* Threshold that shall be maintained by
+					 * the backpointer in order to respect
+					 * geo->mw_cunits on a per chunk basis
+					 */
+
 	struct list_head pages;		/* List of data pages */
 
 	spinlock_t w_lock;		/* Write lock */
@@ -734,7 +739,8 @@ struct pblk_line_ws {
 /*
  * pblk ring buffer operations
  */
-int pblk_rb_init(struct pblk_rb *rb, unsigned int size, unsigned int seg_sz);
+int pblk_rb_init(struct pblk_rb *rb, unsigned int size, unsigned int threshold,
+		 unsigned int seg_sz);
 int pblk_rb_may_write_user(struct pblk_rb *rb, struct bio *bio,
 			   unsigned int nr_entries, unsigned int *pos);
 int pblk_rb_may_write_gc(struct pblk_rb *rb, unsigned int nr_entries,
-- 
2.17.1

Re: [GIT PULL 00/45] lightnvm updates for 4.20.

[Jens Axboe]

On 10/9/18 5:11 AM, Matias Bjørling wrote:
> Hi Jens,
> 
> Please pick up the following patches for 4.20.
> 
> The changes include:
> 
>  - pblk trace support from Hans.
>  - SPDX license header update from Javier.
>  - Tons of refactoring patches from Javier, and me to cleanly abstract
>    the 1.2 and 2.0 specs behind a common core interface.
>  - Enable pblk to use a common interface to retrieve chunk metadata.
>    From me.
>  - Bug fixes from Hans, Javier, Jia-ju, Wei, Zhoujie, and me.

Applied, thanks Matias.

-- 
Jens Axboe