* [V2 PATCH 0/2] pblk: retrieve chunk metadata on erase
@ 2018-09-11 11:35 Javier González
2018-09-11 11:35 ` [PATCH 1/2] lightnvm: pblk: add helper for printing chunk state Javier González
2018-09-11 11:35 ` [PATCH 2/2] lightnvm: pblk: retrieve chunk metadata on erase Javier González
0 siblings, 2 replies; 6+ messages in thread
From: Javier González @ 2018-09-11 11:35 UTC (permalink / raw)
To: mb; +Cc: linux-block, linux-kernel, Javier González
Hi Matias,
Here you have a new series. Please refer to for-4.20/pblk for ordering. I left
the OOB extension patch out since Igor will send a version for it. It is on my
branch though for completeness, until we ifnd the patch that should go in.
If we need to do some rebasing then, I would suggest that we do it directly on
core.
Thanks,
Javier
# Changes sinve V1:
- Remove patches folded into Matias' 1.2 chunck patch
- Rebase on top of Matias' core
Javier González (2):
lightnvm: pblk: add helper for printing chunk state
lightnvm: pblk: retrieve chunk metadata on erase
drivers/lightnvm/core.c | 44 ++++++++++++++++++++++++++++++++++++--
drivers/lightnvm/pblk-core.c | 51 +++++++++++++++++++++++++++++++++-----------
drivers/lightnvm/pblk.h | 9 ++++++++
3 files changed, 89 insertions(+), 15 deletions(-)
--
2.7.4
^ permalink raw reply [flat|nested] 6+ messages in thread
* [PATCH 1/2] lightnvm: pblk: add helper for printing chunk state
2018-09-11 11:35 [V2 PATCH 0/2] pblk: retrieve chunk metadata on erase Javier González
@ 2018-09-11 11:35 ` Javier González
2018-09-11 11:35 ` [PATCH 2/2] lightnvm: pblk: retrieve chunk metadata on erase Javier González
1 sibling, 0 replies; 6+ messages in thread
From: Javier González @ 2018-09-11 11:35 UTC (permalink / raw)
To: mb; +Cc: linux-block, linux-kernel, Javier González
Implement pblk helper for printing chunk state.
Signed-off-by: Javier González <javier@cnexlabs.com>
---
drivers/lightnvm/pblk.h | 9 +++++++++
1 file changed, 9 insertions(+)
diff --git a/drivers/lightnvm/pblk.h b/drivers/lightnvm/pblk.h
index 3d7248829013..5e4cec453a43 100644
--- a/drivers/lightnvm/pblk.h
+++ b/drivers/lightnvm/pblk.h
@@ -1230,6 +1230,15 @@ static inline int pblk_io_aligned(struct pblk *pblk, int nr_secs)
return !(nr_secs % pblk->min_write_pgs);
}
+static inline void print_chunk(struct pblk *pblk, struct nvm_chk_meta *chk,
+ char *msg, int error)
+{
+ pblk_err(pblk, "chunk: (%s: %x) s:%d,t:%d,wi:%d,slba:%llu,cnlb:%llu,wp:%llu\n",
+ msg, error,
+ chk->state, chk->type, chk->wi,
+ chk->slba, chk->cnlb, chk->wp);
+}
+
#ifdef CONFIG_NVM_PBLK_DEBUG
static inline void print_ppa(struct pblk *pblk, struct ppa_addr *p,
char *msg, int error)
--
2.7.4
^ permalink raw reply related [flat|nested] 6+ messages in thread
* [PATCH 2/2] lightnvm: pblk: retrieve chunk metadata on erase
2018-09-11 11:35 [V2 PATCH 0/2] pblk: retrieve chunk metadata on erase Javier González
2018-09-11 11:35 ` [PATCH 1/2] lightnvm: pblk: add helper for printing chunk state Javier González
@ 2018-09-11 11:35 ` Javier González
2018-09-17 8:26 ` Matias Bjørling
1 sibling, 1 reply; 6+ messages in thread
From: Javier González @ 2018-09-11 11:35 UTC (permalink / raw)
To: mb; +Cc: linux-block, linux-kernel, Javier González
On the OCSSD 2.0 spec, the device populates the metadata pointer (if
provided) when a chunk is reset. Implement this path in pblk and use it
for sanity chunk checks.
For 1.2, reset the write pointer and the state on core so that the erase
path is transparent to pblk wrt OCSSD version.
Signed-off-by: Javier González <javier@cnexlabs.com>
---
drivers/lightnvm/core.c | 44 ++++++++++++++++++++++++++++++++++++--
drivers/lightnvm/pblk-core.c | 51 +++++++++++++++++++++++++++++++++-----------
2 files changed, 80 insertions(+), 15 deletions(-)
diff --git a/drivers/lightnvm/core.c b/drivers/lightnvm/core.c
index efb976a863d2..dceaae4e795f 100644
--- a/drivers/lightnvm/core.c
+++ b/drivers/lightnvm/core.c
@@ -750,9 +750,40 @@ int nvm_submit_io(struct nvm_tgt_dev *tgt_dev, struct nvm_rq *rqd)
}
EXPORT_SYMBOL(nvm_submit_io);
+/* Take only addresses in generic format */
+static void nvm_set_chunk_state_12(struct nvm_dev *dev, 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;
+ struct nvm_chk_meta *chunk;
+
+ chunk = ((struct nvm_chk_meta *)rqd->meta_list) + i;
+
+ if (rqd->error)
+ chunk->state = NVM_CHK_ST_OFFLINE;
+ else
+ chunk->state = NVM_CHK_ST_FREE;
+
+ chunk->wp = 0;
+ chunk->wi = 0;
+ chunk->type = NVM_CHK_TP_W_SEQ;
+ chunk->cnlb = dev->geo.clba;
+
+ /* recalculate slba for the chunk */
+ ppa = ppa_list[i];
+ ppa.g.pg = ppa.g.pl = ppa.g.sec = 0;
+
+ chunk->slba = generic_to_dev_addr(dev, ppa).ppa;
+ }
+}
+
int nvm_submit_io_sync(struct nvm_tgt_dev *tgt_dev, struct nvm_rq *rqd)
{
struct nvm_dev *dev = tgt_dev->parent;
+ struct nvm_geo *geo = &dev->geo;
int ret;
if (!dev->ops->submit_io_sync)
@@ -765,8 +796,12 @@ int nvm_submit_io_sync(struct nvm_tgt_dev *tgt_dev, struct nvm_rq *rqd)
/* In case of error, fail with right address format */
ret = dev->ops->submit_io_sync(dev, rqd);
+
nvm_rq_dev_to_tgt(tgt_dev, rqd);
+ if (geo->version == NVM_OCSSD_SPEC_12 && rqd->opcode == NVM_OP_ERASE)
+ nvm_set_chunk_state_12(dev, rqd);
+
return ret;
}
EXPORT_SYMBOL(nvm_submit_io_sync);
@@ -775,10 +810,15 @@ void nvm_end_io(struct nvm_rq *rqd)
{
struct nvm_tgt_dev *tgt_dev = rqd->dev;
- /* Convert address space */
- if (tgt_dev)
+ if (tgt_dev) {
+ /* Convert address space */
nvm_rq_dev_to_tgt(tgt_dev, rqd);
+ if (tgt_dev->geo.version == NVM_OCSSD_SPEC_12 &&
+ rqd->opcode == NVM_OP_ERASE)
+ nvm_set_chunk_state_12(tgt_dev->parent, rqd);
+ }
+
if (rqd->end_io)
rqd->end_io(rqd);
}
diff --git a/drivers/lightnvm/pblk-core.c b/drivers/lightnvm/pblk-core.c
index 417d12b274da..80f0ec756672 100644
--- a/drivers/lightnvm/pblk-core.c
+++ b/drivers/lightnvm/pblk-core.c
@@ -79,7 +79,7 @@ static void __pblk_end_io_erase(struct pblk *pblk, struct nvm_rq *rqd)
{
struct nvm_tgt_dev *dev = pblk->dev;
struct nvm_geo *geo = &dev->geo;
- struct nvm_chk_meta *chunk;
+ struct nvm_chk_meta *chunk, *dev_chunk;
struct pblk_line *line;
int pos;
@@ -89,22 +89,39 @@ static void __pblk_end_io_erase(struct pblk *pblk, struct nvm_rq *rqd)
atomic_dec(&line->left_seblks);
+ /* pblk submits a single erase per command */
+ dev_chunk = rqd->meta_list;
+
+ if (dev_chunk->slba != chunk->slba || dev_chunk->wp)
+ print_chunk(pblk, chunk, "corrupted erase chunk", 0);
+
+ memcpy(chunk, dev_chunk, sizeof(struct nvm_chk_meta));
+
if (rqd->error) {
trace_pblk_chunk_reset(pblk_disk_name(pblk),
&rqd->ppa_addr, PBLK_CHUNK_RESET_FAILED);
- chunk->state = NVM_CHK_ST_OFFLINE;
+#ifdef CONFIG_NVM_PBLK_DEBUG
+ if (chunk->state != NVM_CHK_ST_OFFLINE)
+ print_chunk(pblk, chunk,
+ "corrupted erase chunk state", 0);
+#endif
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;
+#ifdef CONFIG_NVM_PBLK_DEBUG
+ if (chunk->state != NVM_CHK_ST_FREE)
+ print_chunk(pblk, chunk,
+ "corrupted erase chunk state", 0);
+#endif
}
trace_pblk_chunk_state(pblk_disk_name(pblk), &rqd->ppa_addr,
chunk->state);
+ pblk_free_rqd_meta(pblk, rqd);
atomic_dec(&pblk->inflight_io);
}
@@ -952,14 +969,16 @@ int pblk_line_emeta_read(struct pblk *pblk, struct pblk_line *line,
return ret;
}
-static void pblk_setup_e_rq(struct pblk *pblk, struct nvm_rq *rqd,
- struct ppa_addr ppa)
+static int pblk_setup_e_rq(struct pblk *pblk, struct nvm_rq *rqd,
+ struct ppa_addr ppa)
{
rqd->opcode = NVM_OP_ERASE;
rqd->ppa_addr = ppa;
rqd->nr_ppas = 1;
rqd->is_seq = 1;
rqd->bio = NULL;
+
+ return pblk_alloc_rqd_meta(pblk, rqd);
}
static int pblk_blk_erase_sync(struct pblk *pblk, struct ppa_addr ppa)
@@ -967,10 +986,12 @@ static int pblk_blk_erase_sync(struct pblk *pblk, struct ppa_addr ppa)
struct nvm_rq rqd = {NULL};
int ret;
+ ret = pblk_setup_e_rq(pblk, &rqd, ppa);
+ if (ret)
+ return ret;
+
trace_pblk_chunk_reset(pblk_disk_name(pblk), &ppa,
- PBLK_CHUNK_RESET_START);
-
- pblk_setup_e_rq(pblk, &rqd, 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.
@@ -1774,11 +1795,15 @@ void pblk_line_put_wq(struct kref *ref)
int pblk_blk_erase_async(struct pblk *pblk, struct ppa_addr ppa)
{
struct nvm_rq *rqd;
- int err;
+ int ret;
rqd = pblk_alloc_rqd(pblk, PBLK_ERASE);
- pblk_setup_e_rq(pblk, rqd, ppa);
+ ret = pblk_setup_e_rq(pblk, rqd, ppa);
+ if (ret) {
+ pblk_free_rqd(pblk, rqd, PBLK_ERASE);
+ return ret;
+ }
rqd->end_io = pblk_end_io_erase;
rqd->private = pblk;
@@ -1789,8 +1814,8 @@ int pblk_blk_erase_async(struct pblk *pblk, struct ppa_addr ppa)
/* The write thread schedules erases so that it minimizes disturbances
* with writes. Thus, there is no need to take the LUN semaphore.
*/
- err = pblk_submit_io(pblk, rqd);
- if (err) {
+ ret = pblk_submit_io(pblk, rqd);
+ if (ret) {
struct nvm_tgt_dev *dev = pblk->dev;
struct nvm_geo *geo = &dev->geo;
@@ -1799,7 +1824,7 @@ int pblk_blk_erase_async(struct pblk *pblk, struct ppa_addr ppa)
pblk_ppa_to_pos(geo, ppa));
}
- return err;
+ return ret;
}
struct pblk_line *pblk_line_get_data(struct pblk *pblk)
--
2.7.4
^ permalink raw reply related [flat|nested] 6+ messages in thread
* Re: [PATCH 2/2] lightnvm: pblk: retrieve chunk metadata on erase
2018-09-11 11:35 ` [PATCH 2/2] lightnvm: pblk: retrieve chunk metadata on erase Javier González
@ 2018-09-17 8:26 ` Matias Bjørling
2018-09-17 10:17 ` Javier Gonzalez
0 siblings, 1 reply; 6+ messages in thread
From: Matias Bjørling @ 2018-09-17 8:26 UTC (permalink / raw)
To: javier; +Cc: linux-block, linux-kernel, javier
On 09/11/2018 01:35 PM, Javier González wrote:
> On the OCSSD 2.0 spec, the device populates the metadata pointer (if
> provided) when a chunk is reset. Implement this path in pblk and use it
> for sanity chunk checks.
>
> For 1.2, reset the write pointer and the state on core so that the erase
> path is transparent to pblk wrt OCSSD version.
>
> Signed-off-by: Javier González <javier@cnexlabs.com>
> ---
> drivers/lightnvm/core.c | 44 ++++++++++++++++++++++++++++++++++++--
> drivers/lightnvm/pblk-core.c | 51 +++++++++++++++++++++++++++++++++-----------
> 2 files changed, 80 insertions(+), 15 deletions(-)
>
> diff --git a/drivers/lightnvm/core.c b/drivers/lightnvm/core.c
> index efb976a863d2..dceaae4e795f 100644
> --- a/drivers/lightnvm/core.c
> +++ b/drivers/lightnvm/core.c
> @@ -750,9 +750,40 @@ int nvm_submit_io(struct nvm_tgt_dev *tgt_dev, struct nvm_rq *rqd)
> }
> EXPORT_SYMBOL(nvm_submit_io);
>
> +/* Take only addresses in generic format */
> +static void nvm_set_chunk_state_12(struct nvm_dev *dev, 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;
> + struct nvm_chk_meta *chunk;
> +
> + chunk = ((struct nvm_chk_meta *)rqd->meta_list) + i;
> +
> + if (rqd->error)
> + chunk->state = NVM_CHK_ST_OFFLINE;
> + else
> + chunk->state = NVM_CHK_ST_FREE;
> +
> + chunk->wp = 0;
> + chunk->wi = 0;
> + chunk->type = NVM_CHK_TP_W_SEQ;
> + chunk->cnlb = dev->geo.clba;
> +
> + /* recalculate slba for the chunk */
> + ppa = ppa_list[i];
> + ppa.g.pg = ppa.g.pl = ppa.g.sec = 0;
> +
> + chunk->slba = generic_to_dev_addr(dev, ppa).ppa;
> + }
> +}
> +
> int nvm_submit_io_sync(struct nvm_tgt_dev *tgt_dev, struct nvm_rq *rqd)
> {
> struct nvm_dev *dev = tgt_dev->parent;
> + struct nvm_geo *geo = &dev->geo;
> int ret;
>
> if (!dev->ops->submit_io_sync)
> @@ -765,8 +796,12 @@ int nvm_submit_io_sync(struct nvm_tgt_dev *tgt_dev, struct nvm_rq *rqd)
>
> /* In case of error, fail with right address format */
> ret = dev->ops->submit_io_sync(dev, rqd);
> +
> nvm_rq_dev_to_tgt(tgt_dev, rqd);
>
> + if (geo->version == NVM_OCSSD_SPEC_12 && rqd->opcode == NVM_OP_ERASE)
> + nvm_set_chunk_state_12(dev, rqd);
> +
> return ret;
> }
> EXPORT_SYMBOL(nvm_submit_io_sync);
> @@ -775,10 +810,15 @@ void nvm_end_io(struct nvm_rq *rqd)
> {
> struct nvm_tgt_dev *tgt_dev = rqd->dev;
>
> - /* Convert address space */
> - if (tgt_dev)
> + if (tgt_dev) {
> + /* Convert address space */
> nvm_rq_dev_to_tgt(tgt_dev, rqd);
>
> + if (tgt_dev->geo.version == NVM_OCSSD_SPEC_12 &&
> + rqd->opcode == NVM_OP_ERASE)
> + nvm_set_chunk_state_12(tgt_dev->parent, rqd);
> + }
> +
> if (rqd->end_io)
> rqd->end_io(rqd);
> }
> diff --git a/drivers/lightnvm/pblk-core.c b/drivers/lightnvm/pblk-core.c
> index 417d12b274da..80f0ec756672 100644
> --- a/drivers/lightnvm/pblk-core.c
> +++ b/drivers/lightnvm/pblk-core.c
> @@ -79,7 +79,7 @@ static void __pblk_end_io_erase(struct pblk *pblk, struct nvm_rq *rqd)
> {
> struct nvm_tgt_dev *dev = pblk->dev;
> struct nvm_geo *geo = &dev->geo;
> - struct nvm_chk_meta *chunk;
> + struct nvm_chk_meta *chunk, *dev_chunk;
> struct pblk_line *line;
> int pos;
>
> @@ -89,22 +89,39 @@ static void __pblk_end_io_erase(struct pblk *pblk, struct nvm_rq *rqd)
>
> atomic_dec(&line->left_seblks);
>
> + /* pblk submits a single erase per command */
> + dev_chunk = rqd->meta_list;
> +
> + if (dev_chunk->slba != chunk->slba || dev_chunk->wp)
> + print_chunk(pblk, chunk, "corrupted erase chunk", 0);
> +
> + memcpy(chunk, dev_chunk, sizeof(struct nvm_chk_meta));
> +
> if (rqd->error) {
> trace_pblk_chunk_reset(pblk_disk_name(pblk),
> &rqd->ppa_addr, PBLK_CHUNK_RESET_FAILED);
>
> - chunk->state = NVM_CHK_ST_OFFLINE;
> +#ifdef CONFIG_NVM_PBLK_DEBUG
> + if (chunk->state != NVM_CHK_ST_OFFLINE)
> + print_chunk(pblk, chunk,
> + "corrupted erase chunk state", 0);
> +#endif
> 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;
> +#ifdef CONFIG_NVM_PBLK_DEBUG
> + if (chunk->state != NVM_CHK_ST_FREE)
> + print_chunk(pblk, chunk,
> + "corrupted erase chunk state", 0);
> +#endif > }
>
> trace_pblk_chunk_state(pblk_disk_name(pblk), &rqd->ppa_addr,
> chunk->state);
>
> + pblk_free_rqd_meta(pblk, rqd);
> atomic_dec(&pblk->inflight_io);
> }
>
> @@ -952,14 +969,16 @@ int pblk_line_emeta_read(struct pblk *pblk, struct pblk_line *line,
> return ret;
> }
>
> -static void pblk_setup_e_rq(struct pblk *pblk, struct nvm_rq *rqd,
> - struct ppa_addr ppa)
> +static int pblk_setup_e_rq(struct pblk *pblk, struct nvm_rq *rqd,
> + struct ppa_addr ppa)
> {
> rqd->opcode = NVM_OP_ERASE;
> rqd->ppa_addr = ppa;
> rqd->nr_ppas = 1;
> rqd->is_seq = 1;
> rqd->bio = NULL;
> +
> + return pblk_alloc_rqd_meta(pblk, rqd);
> }
>
> static int pblk_blk_erase_sync(struct pblk *pblk, struct ppa_addr ppa)
> @@ -967,10 +986,12 @@ static int pblk_blk_erase_sync(struct pblk *pblk, struct ppa_addr ppa)
> struct nvm_rq rqd = {NULL};
> int ret;
>
> + ret = pblk_setup_e_rq(pblk, &rqd, ppa);
> + if (ret)
> + return ret;
> +
> trace_pblk_chunk_reset(pblk_disk_name(pblk), &ppa,
> - PBLK_CHUNK_RESET_START);
> -
> - pblk_setup_e_rq(pblk, &rqd, 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.
> @@ -1774,11 +1795,15 @@ void pblk_line_put_wq(struct kref *ref)
> int pblk_blk_erase_async(struct pblk *pblk, struct ppa_addr ppa)
> {
> struct nvm_rq *rqd;
> - int err;
> + int ret;
>
> rqd = pblk_alloc_rqd(pblk, PBLK_ERASE);
>
> - pblk_setup_e_rq(pblk, rqd, ppa);
> + ret = pblk_setup_e_rq(pblk, rqd, ppa);
> + if (ret) {
> + pblk_free_rqd(pblk, rqd, PBLK_ERASE);
> + return ret;
> + }
>
> rqd->end_io = pblk_end_io_erase;
> rqd->private = pblk;
> @@ -1789,8 +1814,8 @@ int pblk_blk_erase_async(struct pblk *pblk, struct ppa_addr ppa)
> /* The write thread schedules erases so that it minimizes disturbances
> * with writes. Thus, there is no need to take the LUN semaphore.
> */
> - err = pblk_submit_io(pblk, rqd);
> - if (err) {
> + ret = pblk_submit_io(pblk, rqd);
> + if (ret) {
> struct nvm_tgt_dev *dev = pblk->dev;
> struct nvm_geo *geo = &dev->geo;
>
> @@ -1799,7 +1824,7 @@ int pblk_blk_erase_async(struct pblk *pblk, struct ppa_addr ppa)
> pblk_ppa_to_pos(geo, ppa));
> }
>
> - return err;
> + return ret;
> }
>
> struct pblk_line *pblk_line_get_data(struct pblk *pblk)
>
I'll lean towards not taking this patch. The usecase adds extra logic,
overhead, complexity, and its only usecase for debugging.
I am not an advocate of defensive coding in the fast path. For example,
if an SLBA is inequal with the chunk slba, the drive is buggy, and does
not behave according to the spec. and therefore does not warrant the
checks. These type of bugs can be caught in the general drive testing suite.
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH 2/2] lightnvm: pblk: retrieve chunk metadata on erase
2018-09-17 8:26 ` Matias Bjørling
@ 2018-09-17 10:17 ` Javier Gonzalez
0 siblings, 0 replies; 6+ messages in thread
From: Javier Gonzalez @ 2018-09-17 10:17 UTC (permalink / raw)
To: Matias Bjørling; +Cc: linux-block, linux-kernel
[-- Attachment #1: Type: text/plain, Size: 8155 bytes --]
> On 17 Sep 2018, at 10.26, Matias Bjørling <mb@lightnvm.io> wrote:
>
> On 09/11/2018 01:35 PM, Javier González wrote:
>> On the OCSSD 2.0 spec, the device populates the metadata pointer (if
>> provided) when a chunk is reset. Implement this path in pblk and use it
>> for sanity chunk checks.
>> For 1.2, reset the write pointer and the state on core so that the erase
>> path is transparent to pblk wrt OCSSD version.
>> Signed-off-by: Javier González <javier@cnexlabs.com>
>> ---
>> drivers/lightnvm/core.c | 44 ++++++++++++++++++++++++++++++++++++--
>> drivers/lightnvm/pblk-core.c | 51 +++++++++++++++++++++++++++++++++-----------
>> 2 files changed, 80 insertions(+), 15 deletions(-)
>> diff --git a/drivers/lightnvm/core.c b/drivers/lightnvm/core.c
>> index efb976a863d2..dceaae4e795f 100644
>> --- a/drivers/lightnvm/core.c
>> +++ b/drivers/lightnvm/core.c
>> @@ -750,9 +750,40 @@ int nvm_submit_io(struct nvm_tgt_dev *tgt_dev, struct nvm_rq *rqd)
>> }
>> EXPORT_SYMBOL(nvm_submit_io);
>> +/* Take only addresses in generic format */
>> +static void nvm_set_chunk_state_12(struct nvm_dev *dev, 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;
>> + struct nvm_chk_meta *chunk;
>> +
>> + chunk = ((struct nvm_chk_meta *)rqd->meta_list) + i;
>> +
>> + if (rqd->error)
>> + chunk->state = NVM_CHK_ST_OFFLINE;
>> + else
>> + chunk->state = NVM_CHK_ST_FREE;
>> +
>> + chunk->wp = 0;
>> + chunk->wi = 0;
>> + chunk->type = NVM_CHK_TP_W_SEQ;
>> + chunk->cnlb = dev->geo.clba;
>> +
>> + /* recalculate slba for the chunk */
>> + ppa = ppa_list[i];
>> + ppa.g.pg = ppa.g.pl = ppa.g.sec = 0;
>> +
>> + chunk->slba = generic_to_dev_addr(dev, ppa).ppa;
>> + }
>> +}
>> +
>> int nvm_submit_io_sync(struct nvm_tgt_dev *tgt_dev, struct nvm_rq *rqd)
>> {
>> struct nvm_dev *dev = tgt_dev->parent;
>> + struct nvm_geo *geo = &dev->geo;
>> int ret;
>> if (!dev->ops->submit_io_sync)
>> @@ -765,8 +796,12 @@ int nvm_submit_io_sync(struct nvm_tgt_dev *tgt_dev, struct nvm_rq *rqd)
>> /* In case of error, fail with right address format */
>> ret = dev->ops->submit_io_sync(dev, rqd);
>> +
>> nvm_rq_dev_to_tgt(tgt_dev, rqd);
>> + if (geo->version == NVM_OCSSD_SPEC_12 && rqd->opcode == NVM_OP_ERASE)
>> + nvm_set_chunk_state_12(dev, rqd);
>> +
>> return ret;
>> }
>> EXPORT_SYMBOL(nvm_submit_io_sync);
>> @@ -775,10 +810,15 @@ void nvm_end_io(struct nvm_rq *rqd)
>> {
>> struct nvm_tgt_dev *tgt_dev = rqd->dev;
>> - /* Convert address space */
>> - if (tgt_dev)
>> + if (tgt_dev) {
>> + /* Convert address space */
>> nvm_rq_dev_to_tgt(tgt_dev, rqd);
>> + if (tgt_dev->geo.version == NVM_OCSSD_SPEC_12 &&
>> + rqd->opcode == NVM_OP_ERASE)
>> + nvm_set_chunk_state_12(tgt_dev->parent, rqd);
>> + }
>> +
>> if (rqd->end_io)
>> rqd->end_io(rqd);
>> }
>> diff --git a/drivers/lightnvm/pblk-core.c b/drivers/lightnvm/pblk-core.c
>> index 417d12b274da..80f0ec756672 100644
>> --- a/drivers/lightnvm/pblk-core.c
>> +++ b/drivers/lightnvm/pblk-core.c
>> @@ -79,7 +79,7 @@ static void __pblk_end_io_erase(struct pblk *pblk, struct nvm_rq *rqd)
>> {
>> struct nvm_tgt_dev *dev = pblk->dev;
>> struct nvm_geo *geo = &dev->geo;
>> - struct nvm_chk_meta *chunk;
>> + struct nvm_chk_meta *chunk, *dev_chunk;
>> struct pblk_line *line;
>> int pos;
>> @@ -89,22 +89,39 @@ static void __pblk_end_io_erase(struct pblk *pblk, struct nvm_rq *rqd)
>> atomic_dec(&line->left_seblks);
>> + /* pblk submits a single erase per command */
>> + dev_chunk = rqd->meta_list;
>> +
>> + if (dev_chunk->slba != chunk->slba || dev_chunk->wp)
>> + print_chunk(pblk, chunk, "corrupted erase chunk", 0);
>> +
>> + memcpy(chunk, dev_chunk, sizeof(struct nvm_chk_meta));
>> +
>> if (rqd->error) {
>> trace_pblk_chunk_reset(pblk_disk_name(pblk),
>> &rqd->ppa_addr, PBLK_CHUNK_RESET_FAILED);
>> - chunk->state = NVM_CHK_ST_OFFLINE;
>> +#ifdef CONFIG_NVM_PBLK_DEBUG
>> + if (chunk->state != NVM_CHK_ST_OFFLINE)
>> + print_chunk(pblk, chunk,
>> + "corrupted erase chunk state", 0);
>> +#endif
>> 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;
>> +#ifdef CONFIG_NVM_PBLK_DEBUG
>> + if (chunk->state != NVM_CHK_ST_FREE)
>> + print_chunk(pblk, chunk,
>> + "corrupted erase chunk state", 0);
>> +#endif > }
>> trace_pblk_chunk_state(pblk_disk_name(pblk), &rqd->ppa_addr,
>> chunk->state);
>> + pblk_free_rqd_meta(pblk, rqd);
>> atomic_dec(&pblk->inflight_io);
>> }
>> @@ -952,14 +969,16 @@ int pblk_line_emeta_read(struct pblk *pblk, struct pblk_line *line,
>> return ret;
>> }
>> -static void pblk_setup_e_rq(struct pblk *pblk, struct nvm_rq *rqd,
>> - struct ppa_addr ppa)
>> +static int pblk_setup_e_rq(struct pblk *pblk, struct nvm_rq *rqd,
>> + struct ppa_addr ppa)
>> {
>> rqd->opcode = NVM_OP_ERASE;
>> rqd->ppa_addr = ppa;
>> rqd->nr_ppas = 1;
>> rqd->is_seq = 1;
>> rqd->bio = NULL;
>> +
>> + return pblk_alloc_rqd_meta(pblk, rqd);
>> }
>> static int pblk_blk_erase_sync(struct pblk *pblk, struct ppa_addr ppa)
>> @@ -967,10 +986,12 @@ static int pblk_blk_erase_sync(struct pblk *pblk, struct ppa_addr ppa)
>> struct nvm_rq rqd = {NULL};
>> int ret;
>> + ret = pblk_setup_e_rq(pblk, &rqd, ppa);
>> + if (ret)
>> + return ret;
>> +
>> trace_pblk_chunk_reset(pblk_disk_name(pblk), &ppa,
>> - PBLK_CHUNK_RESET_START);
>> -
>> - pblk_setup_e_rq(pblk, &rqd, 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.
>> @@ -1774,11 +1795,15 @@ void pblk_line_put_wq(struct kref *ref)
>> int pblk_blk_erase_async(struct pblk *pblk, struct ppa_addr ppa)
>> {
>> struct nvm_rq *rqd;
>> - int err;
>> + int ret;
>> rqd = pblk_alloc_rqd(pblk, PBLK_ERASE);
>> - pblk_setup_e_rq(pblk, rqd, ppa);
>> + ret = pblk_setup_e_rq(pblk, rqd, ppa);
>> + if (ret) {
>> + pblk_free_rqd(pblk, rqd, PBLK_ERASE);
>> + return ret;
>> + }
>> rqd->end_io = pblk_end_io_erase;
>> rqd->private = pblk;
>> @@ -1789,8 +1814,8 @@ int pblk_blk_erase_async(struct pblk *pblk, struct ppa_addr ppa)
>> /* The write thread schedules erases so that it minimizes disturbances
>> * with writes. Thus, there is no need to take the LUN semaphore.
>> */
>> - err = pblk_submit_io(pblk, rqd);
>> - if (err) {
>> + ret = pblk_submit_io(pblk, rqd);
>> + if (ret) {
>> struct nvm_tgt_dev *dev = pblk->dev;
>> struct nvm_geo *geo = &dev->geo;
>> @@ -1799,7 +1824,7 @@ int pblk_blk_erase_async(struct pblk *pblk, struct ppa_addr ppa)
>> pblk_ppa_to_pos(geo, ppa));
>> }
>> - return err;
>> + return ret;
>> }
>> struct pblk_line *pblk_line_get_data(struct pblk *pblk)
>
> I'll lean towards not taking this patch. The usecase adds extra logic,
> overhead, complexity, and its only usecase for debugging.
>
> I am not an advocate of defensive coding in the fast path. For
> example, if an SLBA is inequal with the chunk slba, the drive is
> buggy, and does not behave according to the spec. and therefore does
> not warrant the checks. These type of bugs can be caught in the
> general drive testing suite.
The use case for this patch is to retrieve the wear-leveling information
and the CNLB, which only the drive knows about. Also, this will help
removing the bitmap metadata per line and maintain a per-chunk metadata
instead, reusing the chunk structure retrieved on erase. I want to have
this in before pushing the rest of the patches as they will be depending
on it.
As for the checks, they are sanity that I believe do not hurt, but I'm
ok with removing them (or putting them under debug flags).
Javier
[-- Attachment #2: Message signed with OpenPGP --]
[-- Type: application/pgp-signature, Size: 833 bytes --]
^ permalink raw reply [flat|nested] 6+ messages in thread
* [PATCH 2/2] lightnvm: pblk: retrieve chunk metadata on erase
2018-10-04 7:13 [V2 PATCH 0/2] " Javier González
@ 2018-10-04 7:13 ` Javier González
0 siblings, 0 replies; 6+ messages in thread
From: Javier González @ 2018-10-04 7:13 UTC (permalink / raw)
To: mb; +Cc: linux-block, linux-kernel, Javier González
On the OCSSD 2.0 spec, the device populates the metadata pointer (if
provided) when a chunk is reset. Implement this path in pblk. This is
the base for implementing wear-leveling and supporting variable size
chunks (e.g., due to the device mapping out certain sectors).
For 1.2, reset the write pointer and the state on core so that the erase
path is transparent to pblk wrt OCSSD version.
Signed-off-by: Javier González <javier@cnexlabs.com>
---
drivers/lightnvm/core.c | 44 ++++++++++++++++++++++++++++++++++--
drivers/lightnvm/pblk-core.c | 54 +++++++++++++++++++++++++++++++++-----------
2 files changed, 83 insertions(+), 15 deletions(-)
diff --git a/drivers/lightnvm/core.c b/drivers/lightnvm/core.c
index efb976a863d2..dceaae4e795f 100644
--- a/drivers/lightnvm/core.c
+++ b/drivers/lightnvm/core.c
@@ -750,9 +750,40 @@ int nvm_submit_io(struct nvm_tgt_dev *tgt_dev, struct nvm_rq *rqd)
}
EXPORT_SYMBOL(nvm_submit_io);
+/* Take only addresses in generic format */
+static void nvm_set_chunk_state_12(struct nvm_dev *dev, 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;
+ struct nvm_chk_meta *chunk;
+
+ chunk = ((struct nvm_chk_meta *)rqd->meta_list) + i;
+
+ if (rqd->error)
+ chunk->state = NVM_CHK_ST_OFFLINE;
+ else
+ chunk->state = NVM_CHK_ST_FREE;
+
+ chunk->wp = 0;
+ chunk->wi = 0;
+ chunk->type = NVM_CHK_TP_W_SEQ;
+ chunk->cnlb = dev->geo.clba;
+
+ /* recalculate slba for the chunk */
+ ppa = ppa_list[i];
+ ppa.g.pg = ppa.g.pl = ppa.g.sec = 0;
+
+ chunk->slba = generic_to_dev_addr(dev, ppa).ppa;
+ }
+}
+
int nvm_submit_io_sync(struct nvm_tgt_dev *tgt_dev, struct nvm_rq *rqd)
{
struct nvm_dev *dev = tgt_dev->parent;
+ struct nvm_geo *geo = &dev->geo;
int ret;
if (!dev->ops->submit_io_sync)
@@ -765,8 +796,12 @@ int nvm_submit_io_sync(struct nvm_tgt_dev *tgt_dev, struct nvm_rq *rqd)
/* In case of error, fail with right address format */
ret = dev->ops->submit_io_sync(dev, rqd);
+
nvm_rq_dev_to_tgt(tgt_dev, rqd);
+ if (geo->version == NVM_OCSSD_SPEC_12 && rqd->opcode == NVM_OP_ERASE)
+ nvm_set_chunk_state_12(dev, rqd);
+
return ret;
}
EXPORT_SYMBOL(nvm_submit_io_sync);
@@ -775,10 +810,15 @@ void nvm_end_io(struct nvm_rq *rqd)
{
struct nvm_tgt_dev *tgt_dev = rqd->dev;
- /* Convert address space */
- if (tgt_dev)
+ if (tgt_dev) {
+ /* Convert address space */
nvm_rq_dev_to_tgt(tgt_dev, rqd);
+ if (tgt_dev->geo.version == NVM_OCSSD_SPEC_12 &&
+ rqd->opcode == NVM_OP_ERASE)
+ nvm_set_chunk_state_12(tgt_dev->parent, rqd);
+ }
+
if (rqd->end_io)
rqd->end_io(rqd);
}
diff --git a/drivers/lightnvm/pblk-core.c b/drivers/lightnvm/pblk-core.c
index 6944aac43b01..2507c4838283 100644
--- a/drivers/lightnvm/pblk-core.c
+++ b/drivers/lightnvm/pblk-core.c
@@ -80,7 +80,7 @@ static void __pblk_end_io_erase(struct pblk *pblk, struct nvm_rq *rqd)
{
struct nvm_tgt_dev *dev = pblk->dev;
struct nvm_geo *geo = &dev->geo;
- struct nvm_chk_meta *chunk;
+ struct nvm_chk_meta *chunk, *dev_chunk;
struct pblk_line *line;
int pos;
@@ -90,22 +90,42 @@ static void __pblk_end_io_erase(struct pblk *pblk, struct nvm_rq *rqd)
atomic_dec(&line->left_seblks);
+ /* pblk submits a single erase per command */
+ dev_chunk = rqd->meta_list;
+
+ chunk->state = dev_chunk->state;
+ chunk->type = dev_chunk->type;
+ chunk->wi = dev_chunk->wi;
+ chunk->cnlb = dev_chunk->cnlb;
+ chunk->wp = dev_chunk->wp;
+
if (rqd->error) {
trace_pblk_chunk_reset(pblk_disk_name(pblk),
&rqd->ppa_addr, PBLK_CHUNK_RESET_FAILED);
- chunk->state = NVM_CHK_ST_OFFLINE;
+#ifdef CONFIG_NVM_PBLK_DEBUG
+ if (chunk->state != NVM_CHK_ST_OFFLINE)
+ print_chunk(pblk, chunk,
+ "corrupted erase chunk state", rqd->error);
+#endif
+
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;
+#ifdef CONFIG_NVM_PBLK_DEBUG
+ if (chunk->state != NVM_CHK_ST_FREE || chunk->wp ||
+ dev_chunk->slba != chunk->slba)
+ print_chunk(pblk, chunk,
+ "corrupted erase chunk state", rqd->error);
+#endif
}
trace_pblk_chunk_state(pblk_disk_name(pblk), &rqd->ppa_addr,
chunk->state);
+ pblk_free_rqd_meta(pblk, rqd);
atomic_dec(&pblk->inflight_io);
}
@@ -923,14 +943,16 @@ int pblk_line_emeta_read(struct pblk *pblk, struct pblk_line *line,
return ret;
}
-static void pblk_setup_e_rq(struct pblk *pblk, struct nvm_rq *rqd,
- struct ppa_addr ppa)
+static int pblk_setup_e_rq(struct pblk *pblk, struct nvm_rq *rqd,
+ struct ppa_addr ppa)
{
rqd->opcode = NVM_OP_ERASE;
rqd->ppa_addr = ppa;
rqd->nr_ppas = 1;
rqd->is_seq = 1;
rqd->bio = NULL;
+
+ return pblk_alloc_rqd_meta(pblk, rqd);
}
static int pblk_blk_erase_sync(struct pblk *pblk, struct ppa_addr ppa)
@@ -938,10 +960,12 @@ static int pblk_blk_erase_sync(struct pblk *pblk, struct ppa_addr ppa)
struct nvm_rq rqd = {NULL};
int ret;
+ ret = pblk_setup_e_rq(pblk, &rqd, ppa);
+ if (ret)
+ return ret;
+
trace_pblk_chunk_reset(pblk_disk_name(pblk), &ppa,
- PBLK_CHUNK_RESET_START);
-
- pblk_setup_e_rq(pblk, &rqd, 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.
@@ -1746,11 +1770,15 @@ void pblk_line_put_wq(struct kref *ref)
int pblk_blk_erase_async(struct pblk *pblk, struct ppa_addr ppa)
{
struct nvm_rq *rqd;
- int err;
+ int ret;
rqd = pblk_alloc_rqd(pblk, PBLK_ERASE);
- pblk_setup_e_rq(pblk, rqd, ppa);
+ ret = pblk_setup_e_rq(pblk, rqd, ppa);
+ if (ret) {
+ pblk_free_rqd(pblk, rqd, PBLK_ERASE);
+ return ret;
+ }
rqd->end_io = pblk_end_io_erase;
rqd->private = pblk;
@@ -1761,8 +1789,8 @@ int pblk_blk_erase_async(struct pblk *pblk, struct ppa_addr ppa)
/* The write thread schedules erases so that it minimizes disturbances
* with writes. Thus, there is no need to take the LUN semaphore.
*/
- err = pblk_submit_io(pblk, rqd);
- if (err) {
+ ret = pblk_submit_io(pblk, rqd);
+ if (ret) {
struct nvm_tgt_dev *dev = pblk->dev;
struct nvm_geo *geo = &dev->geo;
@@ -1771,7 +1799,7 @@ int pblk_blk_erase_async(struct pblk *pblk, struct ppa_addr ppa)
pblk_ppa_to_pos(geo, ppa));
}
- return err;
+ return ret;
}
struct pblk_line *pblk_line_get_data(struct pblk *pblk)
--
2.7.4
^ permalink raw reply related [flat|nested] 6+ messages in thread
end of thread, other threads:[~2018-10-04 7:14 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2018-09-11 11:35 [V2 PATCH 0/2] pblk: retrieve chunk metadata on erase Javier González
2018-09-11 11:35 ` [PATCH 1/2] lightnvm: pblk: add helper for printing chunk state Javier González
2018-09-11 11:35 ` [PATCH 2/2] lightnvm: pblk: retrieve chunk metadata on erase Javier González
2018-09-17 8:26 ` Matias Bjørling
2018-09-17 10:17 ` Javier Gonzalez
2018-10-04 7:13 [V2 PATCH 0/2] " Javier González
2018-10-04 7:13 ` [PATCH 2/2] " Javier González
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).