* [PATCH] lightnvm: pblk: recover chunk state on 1.2 devices @ 2018-06-28 9:12 Javier González 2018-06-28 9:12 ` Javier González 0 siblings, 1 reply; 12+ messages in thread From: Javier González @ 2018-06-28 9:12 UTC (permalink / raw) To: mb; +Cc: hans.holmberg, linux-block, linux-kernel, Javier González Hi Matias, As discussed, here you have the patch reconstructing the state for 1.2 devices. As mentioned in the patch itself, we make a tradeoff by doing this, as writes might fail (or be more unstable) in case that a block has been erased long before it is in used, and 1.2 does not have a way to communicate this. The only thing the device can do is force an erase on write when this happens, but if the device is responsible for this, one could argue that it could prevent double-erases - which follows better the predictability model we seek in Open-Channel. Thoughts? If we go this way, I will push a patch to the qemu 1.2 branch, as in the initialization, we need to match empty page reporting with the block metadata representing that the block is indeed erased. Ping me if you want this patch in advance for testing. Thanks, Javier Javier González (1): lightnvm: pblk: recover chunk state on 1.2 devices drivers/lightnvm/pblk-init.c | 138 ++++++++++++++++++++++++++++++++++++++----- 1 file changed, 124 insertions(+), 14 deletions(-) -- 2.7.4 ^ permalink raw reply [flat|nested] 12+ messages in thread
* [PATCH] lightnvm: pblk: recover chunk state on 1.2 devices 2018-06-28 9:12 [PATCH] lightnvm: pblk: recover chunk state on 1.2 devices Javier González @ 2018-06-28 9:12 ` Javier González 2018-06-29 11:14 ` Matias Bjørling 0 siblings, 1 reply; 12+ messages in thread From: Javier González @ 2018-06-28 9:12 UTC (permalink / raw) To: mb; +Cc: hans.holmberg, linux-block, linux-kernel, Javier González The Open-Channel 1.2 spec does not define a mechanism for the host to recover the block (chunk) state. As a consequence, a newly format device will need to reconstruct the state. Currently, pblk assumes that blocks are not erased, which might cause double-erases in case that the device does not protect itself against them (which is not specified in the spec either). This patch, reconstructs the state based on read errors. If the first sector of a block returns and empty page (NVM_RSP_ERR_EMPTYPAGE), then the block s marked free, i.e., erased and ready to be used (NVM_CHK_ST_FREE). Otherwise, the block is marked as closed (NVM_CHK_ST_CLOSED). Note that even if a block is open and not fully written, it has to be erased in order to be used again. One caveat of this approach is that blocks that have been erased at a moment in time, will always be considered as erased. However, some media might become unstable if blocks are not erased before usage. Since pblk would follow this principle after the state of all blocks fall under pblk's domain, we can consider this as an initialization problem. The trade-off would be to fall back to the old behavior and risk premature media wearing. Signed-off-by: Javier González <javier@cnexlabs.com> --- drivers/lightnvm/pblk-init.c | 138 ++++++++++++++++++++++++++++++++++++++----- 1 file changed, 124 insertions(+), 14 deletions(-) diff --git a/drivers/lightnvm/pblk-init.c b/drivers/lightnvm/pblk-init.c index 3b8aa4a64cac..ce25f1473d8e 100644 --- a/drivers/lightnvm/pblk-init.c +++ b/drivers/lightnvm/pblk-init.c @@ -697,47 +697,138 @@ static void pblk_set_provision(struct pblk *pblk, long nr_free_blks) atomic_set(&pblk->rl.free_user_blocks, nr_free_blks); } +static void pblk_state_complete(struct kref *ref) +{ + struct pblk_pad_rq *pad_rq = container_of(ref, struct pblk_pad_rq, ref); + + complete(&pad_rq->wait); +} + +static void pblk_end_io_state(struct nvm_rq *rqd) +{ + struct pblk_pad_rq *pad_rq = rqd->private; + struct pblk *pblk = pad_rq->pblk; + struct nvm_tgt_dev *dev = pblk->dev; + struct nvm_geo *geo = &dev->geo; + struct pblk_line *line; + struct nvm_chk_meta *chunk; + int pos; + + line = &pblk->lines[pblk_ppa_to_line(rqd->ppa_addr)]; + pos = pblk_ppa_to_pos(geo, rqd->ppa_addr); + + chunk = &line->chks[pos]; + + if (rqd->error == NVM_RSP_ERR_EMPTYPAGE) + chunk->state = NVM_CHK_ST_FREE; + else + chunk->state = NVM_CHK_ST_CLOSED; + + bio_put(rqd->bio); + pblk_free_rqd(pblk, rqd, PBLK_READ); + kref_put(&pad_rq->ref, pblk_state_complete); +} + +static int pblk_check_chunk_state(struct pblk *pblk, struct nvm_chk_meta *chunk, + struct ppa_addr ppa, struct pblk_pad_rq *pad_rq) +{ + struct nvm_rq *rqd; + struct bio *bio; + int ret; + + bio = bio_alloc(GFP_KERNEL, 1); + + if (pblk_bio_add_pages(pblk, bio, GFP_KERNEL, 1)) + goto fail_free_bio; + + rqd = pblk_alloc_rqd(pblk, PBLK_READ); + + rqd->bio = bio; + rqd->opcode = NVM_OP_PREAD; + rqd->flags = pblk_set_read_mode(pblk, PBLK_READ_SEQUENTIAL); + rqd->nr_ppas = 1; + rqd->ppa_addr = ppa; + rqd->end_io = pblk_end_io_state; + rqd->private = pad_rq; + + kref_get(&pad_rq->ref); + + ret = pblk_submit_io(pblk, rqd); + if (ret) { + pr_err("pblk: I/O submissin failed: %d\n", ret); + goto fail_free_rqd; + } + + return NVM_IO_OK; + +fail_free_rqd: + pblk_free_rqd(pblk, rqd, PBLK_READ); + pblk_bio_free_pages(pblk, bio, 0, bio->bi_vcnt); +fail_free_bio: + bio_put(bio); + + return NVM_IO_ERR; +} + 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; + struct pblk_pad_rq *pad_rq; int i, chk_per_lun, nr_bad_chks = 0; + pad_rq = kmalloc(sizeof(struct pblk_pad_rq), GFP_KERNEL); + if (!pad_rq) + return -1; + + pad_rq->pblk = pblk; + init_completion(&pad_rq->wait); + kref_init(&pad_rq->ref); + 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); + struct ppa_addr ppa = rlun->bppa; + int pos = pblk_ppa_to_pos(geo, ppa); 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)) + if (lun_bb_meta[line->id] != NVM_BLK_T_FREE) { + chunk->state = NVM_CHK_ST_OFFLINE; + set_bit(pos, line->blk_bitmap); + nr_bad_chks++; + continue; + } - set_bit(pos, line->blk_bitmap); - nr_bad_chks++; + /* + * In 1.2 spec. chunk state is not persisted by the device. + * Recover the state based on media response. + */ + ppa.g.blk = line->id; + pblk_check_chunk_state(pblk, chunk, ppa, pad_rq); } + kref_put(&pad_rq->ref, pblk_state_complete); + + if (!wait_for_completion_io_timeout(&pad_rq->wait, + msecs_to_jiffies(PBLK_COMMAND_TIMEOUT_MS))) { + pr_err("pblk: state recovery timed out\n"); + return -1; + } + + kfree(pad_rq); return nr_bad_chks; } @@ -1036,6 +1127,23 @@ static int pblk_line_meta_init(struct pblk *pblk) return 0; } +static void check_meta(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 i; + + for (i = 0; i < lm->blk_per_line; i++) { + struct pblk_lun *rlun = &pblk->luns[i]; + struct nvm_chk_meta *chunk; + struct ppa_addr ppa = rlun->bppa; + int pos = pblk_ppa_to_pos(geo, ppa); + + chunk = &line->chks[pos]; + } +} + static int pblk_lines_init(struct pblk *pblk) { struct pblk_line_mgmt *l_mg = &pblk->l_mg; @@ -1077,6 +1185,8 @@ static int pblk_lines_init(struct pblk *pblk) goto fail_free_lines; nr_free_chks += pblk_setup_line_meta(pblk, line, chunk_meta, i); + + check_meta(pblk, line); } if (!nr_free_chks) { -- 2.7.4 ^ permalink raw reply related [flat|nested] 12+ messages in thread
* Re: [PATCH] lightnvm: pblk: recover chunk state on 1.2 devices 2018-06-28 9:12 ` Javier González @ 2018-06-29 11:14 ` Matias Bjørling 2018-06-29 11:22 ` Javier Gonzalez 0 siblings, 1 reply; 12+ messages in thread From: Matias Bjørling @ 2018-06-29 11:14 UTC (permalink / raw) To: javier; +Cc: hans.holmberg, linux-block, linux-kernel, javier On 06/28/2018 11:12 AM, Javier González wrote: > The Open-Channel 1.2 spec does not define a mechanism for the host to > recover the block (chunk) state. As a consequence, a newly format device > will need to reconstruct the state. Currently, pblk assumes that blocks > are not erased, which might cause double-erases in case that the device > does not protect itself against them (which is not specified in the spec > either). It should not be specified in the spec. It is up to the device to handle double erases and not do it. > > This patch, reconstructs the state based on read errors. If the first > sector of a block returns and empty page (NVM_RSP_ERR_EMPTYPAGE), then > the block s marked free, i.e., erased and ready to be used > (NVM_CHK_ST_FREE). Otherwise, the block is marked as closed > (NVM_CHK_ST_CLOSED). Note that even if a block is open and not fully > written, it has to be erased in order to be used again. Should we extend it to do the scan, and update the write pointer as well? I think this kind of feature already is baked into pblk? > > One caveat of this approach is that blocks that have been erased at a > moment in time, will always be considered as erased. However, some media > might become unstable if blocks are not erased before usage. Since pblk > would follow this principle after the state of all blocks fall under > pblk's domain, we can consider this as an initialization problem. The > trade-off would be to fall back to the old behavior and risk premature > media wearing. The above is up to the device implementation to handle. We cannot expect users to understand the intrinsics of media. > > Signed-off-by: Javier González <javier@cnexlabs.com> > --- > drivers/lightnvm/pblk-init.c | 138 ++++++++++++++++++++++++++++++++++++++----- > 1 file changed, 124 insertions(+), 14 deletions(-) > > diff --git a/drivers/lightnvm/pblk-init.c b/drivers/lightnvm/pblk-init.c > index 3b8aa4a64cac..ce25f1473d8e 100644 > --- a/drivers/lightnvm/pblk-init.c > +++ b/drivers/lightnvm/pblk-init.c > @@ -697,47 +697,138 @@ static void pblk_set_provision(struct pblk *pblk, long nr_free_blks) > atomic_set(&pblk->rl.free_user_blocks, nr_free_blks); > } > > +static void pblk_state_complete(struct kref *ref) > +{ > + struct pblk_pad_rq *pad_rq = container_of(ref, struct pblk_pad_rq, ref); > + > + complete(&pad_rq->wait); > +} > + > +static void pblk_end_io_state(struct nvm_rq *rqd) > +{ > + struct pblk_pad_rq *pad_rq = rqd->private; > + struct pblk *pblk = pad_rq->pblk; > + struct nvm_tgt_dev *dev = pblk->dev; > + struct nvm_geo *geo = &dev->geo; > + struct pblk_line *line; > + struct nvm_chk_meta *chunk; > + int pos; > + > + line = &pblk->lines[pblk_ppa_to_line(rqd->ppa_addr)]; > + pos = pblk_ppa_to_pos(geo, rqd->ppa_addr); > + > + chunk = &line->chks[pos]; > + > + if (rqd->error == NVM_RSP_ERR_EMPTYPAGE) > + chunk->state = NVM_CHK_ST_FREE; > + else > + chunk->state = NVM_CHK_ST_CLOSED; > + > + bio_put(rqd->bio); > + pblk_free_rqd(pblk, rqd, PBLK_READ); > + kref_put(&pad_rq->ref, pblk_state_complete); > +} > + > +static int pblk_check_chunk_state(struct pblk *pblk, struct nvm_chk_meta *chunk, > + struct ppa_addr ppa, struct pblk_pad_rq *pad_rq) > +{ > + struct nvm_rq *rqd; > + struct bio *bio; > + int ret; > + > + bio = bio_alloc(GFP_KERNEL, 1); > + > + if (pblk_bio_add_pages(pblk, bio, GFP_KERNEL, 1)) > + goto fail_free_bio; > + > + rqd = pblk_alloc_rqd(pblk, PBLK_READ); > + > + rqd->bio = bio; > + rqd->opcode = NVM_OP_PREAD; > + rqd->flags = pblk_set_read_mode(pblk, PBLK_READ_SEQUENTIAL); > + rqd->nr_ppas = 1; > + rqd->ppa_addr = ppa; > + rqd->end_io = pblk_end_io_state; > + rqd->private = pad_rq; > + > + kref_get(&pad_rq->ref); > + > + ret = pblk_submit_io(pblk, rqd); > + if (ret) { > + pr_err("pblk: I/O submissin failed: %d\n", ret); > + goto fail_free_rqd; > + } > + > + return NVM_IO_OK; > + > +fail_free_rqd: > + pblk_free_rqd(pblk, rqd, PBLK_READ); > + pblk_bio_free_pages(pblk, bio, 0, bio->bi_vcnt); > +fail_free_bio: > + bio_put(bio); > + > + return NVM_IO_ERR; > +} > + > 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; > + struct pblk_pad_rq *pad_rq; > int i, chk_per_lun, nr_bad_chks = 0; > > + pad_rq = kmalloc(sizeof(struct pblk_pad_rq), GFP_KERNEL); > + if (!pad_rq) > + return -1; > + > + pad_rq->pblk = pblk; > + init_completion(&pad_rq->wait); > + kref_init(&pad_rq->ref); > + > 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); > + struct ppa_addr ppa = rlun->bppa; > + int pos = pblk_ppa_to_pos(geo, ppa); > 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)) > + if (lun_bb_meta[line->id] != NVM_BLK_T_FREE) { > + chunk->state = NVM_CHK_ST_OFFLINE; > + set_bit(pos, line->blk_bitmap); > + nr_bad_chks++; > + > continue; > + } > > - set_bit(pos, line->blk_bitmap); > - nr_bad_chks++; > + /* > + * In 1.2 spec. chunk state is not persisted by the device. > + * Recover the state based on media response. > + */ > + ppa.g.blk = line->id; > + pblk_check_chunk_state(pblk, chunk, ppa, pad_rq); > } > > + kref_put(&pad_rq->ref, pblk_state_complete); > + > + if (!wait_for_completion_io_timeout(&pad_rq->wait, > + msecs_to_jiffies(PBLK_COMMAND_TIMEOUT_MS))) { > + pr_err("pblk: state recovery timed out\n"); > + return -1; > + } > + > + kfree(pad_rq); > return nr_bad_chks; > } > > @@ -1036,6 +1127,23 @@ static int pblk_line_meta_init(struct pblk *pblk) > return 0; > } > > +static void check_meta(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 i; > + > + for (i = 0; i < lm->blk_per_line; i++) { > + struct pblk_lun *rlun = &pblk->luns[i]; > + struct nvm_chk_meta *chunk; > + struct ppa_addr ppa = rlun->bppa; > + int pos = pblk_ppa_to_pos(geo, ppa); > + > + chunk = &line->chks[pos]; > + } > +} > + > static int pblk_lines_init(struct pblk *pblk) > { > struct pblk_line_mgmt *l_mg = &pblk->l_mg; > @@ -1077,6 +1185,8 @@ static int pblk_lines_init(struct pblk *pblk) > goto fail_free_lines; > > nr_free_chks += pblk_setup_line_meta(pblk, line, chunk_meta, i); > + > + check_meta(pblk, line); > } > > if (!nr_free_chks) { > I'm okay with us doing this in pblk for now. Over time, someone may do the work move this (and other specific only-1.2/2.0 stuff) into the lightnvm subsystem. I don't think pblk should need to care about either 1.2 or 2.0. ^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH] lightnvm: pblk: recover chunk state on 1.2 devices 2018-06-29 11:14 ` Matias Bjørling @ 2018-06-29 11:22 ` Javier Gonzalez 2018-06-29 11:28 ` Matias Bjørling 0 siblings, 1 reply; 12+ messages in thread From: Javier Gonzalez @ 2018-06-29 11:22 UTC (permalink / raw) To: Matias Bjørling; +Cc: Hans Holmberg, linux-block, linux-kernel [-- Attachment #1: Type: text/plain, Size: 8473 bytes --] > On 29 Jun 2018, at 13.14, Matias Bjørling <mb@lightnvm.io> wrote: > > On 06/28/2018 11:12 AM, Javier González wrote: >> The Open-Channel 1.2 spec does not define a mechanism for the host to >> recover the block (chunk) state. As a consequence, a newly format device >> will need to reconstruct the state. Currently, pblk assumes that blocks >> are not erased, which might cause double-erases in case that the device >> does not protect itself against them (which is not specified in the spec >> either). > > It should not be specified in the spec. It is up to the device to handle > double erases and not do it. > >> This patch, reconstructs the state based on read errors. If the first >> sector of a block returns and empty page (NVM_RSP_ERR_EMPTYPAGE), then >> the block s marked free, i.e., erased and ready to be used >> (NVM_CHK_ST_FREE). Otherwise, the block is marked as closed >> (NVM_CHK_ST_CLOSED). Note that even if a block is open and not fully >> written, it has to be erased in order to be used again. > > Should we extend it to do the scan, and update the write pointer as > well? I think this kind of feature already is baked into pblk? > This is already in place: we scan until empty page and take it from there. This patch is only for the case in which we start a pblk instance form scratch. On a device already owned by pblk, we would not have the problem we are trying to solve here because we know the state. >> One caveat of this approach is that blocks that have been erased at a >> moment in time, will always be considered as erased. However, some media >> might become unstable if blocks are not erased before usage. Since pblk >> would follow this principle after the state of all blocks fall under >> pblk's domain, we can consider this as an initialization problem. The >> trade-off would be to fall back to the old behavior and risk premature >> media wearing. > > The above is up to the device implementation to handle. We cannot > expect users to understand the intrinsics of media. > Of course. The point is that with this approach, erases are left a bit in the air and "preventable" write errors might happen, with the previous the burden was put on the device to deal with double erases. It's a tradeoff that I want to make clear before the path is taken. >> Signed-off-by: Javier González <javier@cnexlabs.com> >> --- >> drivers/lightnvm/pblk-init.c | 138 ++++++++++++++++++++++++++++++++++++++----- >> 1 file changed, 124 insertions(+), 14 deletions(-) >> diff --git a/drivers/lightnvm/pblk-init.c b/drivers/lightnvm/pblk-init.c >> index 3b8aa4a64cac..ce25f1473d8e 100644 >> --- a/drivers/lightnvm/pblk-init.c >> +++ b/drivers/lightnvm/pblk-init.c >> @@ -697,47 +697,138 @@ static void pblk_set_provision(struct pblk *pblk, long nr_free_blks) >> atomic_set(&pblk->rl.free_user_blocks, nr_free_blks); >> } >> +static void pblk_state_complete(struct kref *ref) >> +{ >> + struct pblk_pad_rq *pad_rq = container_of(ref, struct pblk_pad_rq, ref); >> + >> + complete(&pad_rq->wait); >> +} >> + >> +static void pblk_end_io_state(struct nvm_rq *rqd) >> +{ >> + struct pblk_pad_rq *pad_rq = rqd->private; >> + struct pblk *pblk = pad_rq->pblk; >> + struct nvm_tgt_dev *dev = pblk->dev; >> + struct nvm_geo *geo = &dev->geo; >> + struct pblk_line *line; >> + struct nvm_chk_meta *chunk; >> + int pos; >> + >> + line = &pblk->lines[pblk_ppa_to_line(rqd->ppa_addr)]; >> + pos = pblk_ppa_to_pos(geo, rqd->ppa_addr); >> + >> + chunk = &line->chks[pos]; >> + >> + if (rqd->error == NVM_RSP_ERR_EMPTYPAGE) >> + chunk->state = NVM_CHK_ST_FREE; >> + else >> + chunk->state = NVM_CHK_ST_CLOSED; >> + >> + bio_put(rqd->bio); >> + pblk_free_rqd(pblk, rqd, PBLK_READ); >> + kref_put(&pad_rq->ref, pblk_state_complete); >> +} >> + >> +static int pblk_check_chunk_state(struct pblk *pblk, struct nvm_chk_meta *chunk, >> + struct ppa_addr ppa, struct pblk_pad_rq *pad_rq) >> +{ >> + struct nvm_rq *rqd; >> + struct bio *bio; >> + int ret; >> + >> + bio = bio_alloc(GFP_KERNEL, 1); >> + >> + if (pblk_bio_add_pages(pblk, bio, GFP_KERNEL, 1)) >> + goto fail_free_bio; >> + >> + rqd = pblk_alloc_rqd(pblk, PBLK_READ); >> + >> + rqd->bio = bio; >> + rqd->opcode = NVM_OP_PREAD; >> + rqd->flags = pblk_set_read_mode(pblk, PBLK_READ_SEQUENTIAL); >> + rqd->nr_ppas = 1; >> + rqd->ppa_addr = ppa; >> + rqd->end_io = pblk_end_io_state; >> + rqd->private = pad_rq; >> + >> + kref_get(&pad_rq->ref); >> + >> + ret = pblk_submit_io(pblk, rqd); >> + if (ret) { >> + pr_err("pblk: I/O submissin failed: %d\n", ret); >> + goto fail_free_rqd; >> + } >> + >> + return NVM_IO_OK; >> + >> +fail_free_rqd: >> + pblk_free_rqd(pblk, rqd, PBLK_READ); >> + pblk_bio_free_pages(pblk, bio, 0, bio->bi_vcnt); >> +fail_free_bio: >> + bio_put(bio); >> + >> + return NVM_IO_ERR; >> +} >> + >> 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; >> + struct pblk_pad_rq *pad_rq; >> int i, chk_per_lun, nr_bad_chks = 0; >> + pad_rq = kmalloc(sizeof(struct pblk_pad_rq), GFP_KERNEL); >> + if (!pad_rq) >> + return -1; >> + >> + pad_rq->pblk = pblk; >> + init_completion(&pad_rq->wait); >> + kref_init(&pad_rq->ref); >> + >> 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); >> + struct ppa_addr ppa = rlun->bppa; >> + int pos = pblk_ppa_to_pos(geo, ppa); >> 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)) >> + if (lun_bb_meta[line->id] != NVM_BLK_T_FREE) { >> + chunk->state = NVM_CHK_ST_OFFLINE; >> + set_bit(pos, line->blk_bitmap); >> + nr_bad_chks++; >> + >> continue; >> + } >> - set_bit(pos, line->blk_bitmap); >> - nr_bad_chks++; >> + /* >> + * In 1.2 spec. chunk state is not persisted by the device. >> + * Recover the state based on media response. >> + */ >> + ppa.g.blk = line->id; >> + pblk_check_chunk_state(pblk, chunk, ppa, pad_rq); >> } >> + kref_put(&pad_rq->ref, pblk_state_complete); >> + >> + if (!wait_for_completion_io_timeout(&pad_rq->wait, >> + msecs_to_jiffies(PBLK_COMMAND_TIMEOUT_MS))) { >> + pr_err("pblk: state recovery timed out\n"); >> + return -1; >> + } >> + >> + kfree(pad_rq); >> return nr_bad_chks; >> } >> @@ -1036,6 +1127,23 @@ static int pblk_line_meta_init(struct pblk *pblk) >> return 0; >> } >> +static void check_meta(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 i; >> + >> + for (i = 0; i < lm->blk_per_line; i++) { >> + struct pblk_lun *rlun = &pblk->luns[i]; >> + struct nvm_chk_meta *chunk; >> + struct ppa_addr ppa = rlun->bppa; >> + int pos = pblk_ppa_to_pos(geo, ppa); >> + >> + chunk = &line->chks[pos]; >> + } >> +} >> + >> static int pblk_lines_init(struct pblk *pblk) >> { >> struct pblk_line_mgmt *l_mg = &pblk->l_mg; >> @@ -1077,6 +1185,8 @@ static int pblk_lines_init(struct pblk *pblk) >> goto fail_free_lines; >> nr_free_chks += pblk_setup_line_meta(pblk, line, chunk_meta, i); >> + >> + check_meta(pblk, line); >> } >> if (!nr_free_chks) { > > I'm okay with us doing this in pblk for now. Over time, someone may do > the work move this (and other specific only-1.2/2.0 stuff) into the > lightnvm subsystem. I don't think pblk should need to care about > either 1.2 or 2.0. That would be ideal. Thanks! [-- Attachment #2: Message signed with OpenPGP --] [-- Type: application/pgp-signature, Size: 833 bytes --] ^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH] lightnvm: pblk: recover chunk state on 1.2 devices 2018-06-29 11:22 ` Javier Gonzalez @ 2018-06-29 11:28 ` Matias Bjørling 2018-07-24 7:54 ` Javier Gonzalez 0 siblings, 1 reply; 12+ messages in thread From: Matias Bjørling @ 2018-06-29 11:28 UTC (permalink / raw) To: javier; +Cc: hans.holmberg, linux-block, linux-kernel On 06/29/2018 01:22 PM, Javier Gonzalez wrote: >> On 29 Jun 2018, at 13.14, Matias Bjørling <mb@lightnvm.io> wrote: >> >> On 06/28/2018 11:12 AM, Javier González wrote: >>> The Open-Channel 1.2 spec does not define a mechanism for the host to >>> recover the block (chunk) state. As a consequence, a newly format device >>> will need to reconstruct the state. Currently, pblk assumes that blocks >>> are not erased, which might cause double-erases in case that the device >>> does not protect itself against them (which is not specified in the spec >>> either). >> >> It should not be specified in the spec. It is up to the device to handle >> double erases and not do it. >> >>> This patch, reconstructs the state based on read errors. If the first >>> sector of a block returns and empty page (NVM_RSP_ERR_EMPTYPAGE), then >>> the block s marked free, i.e., erased and ready to be used >>> (NVM_CHK_ST_FREE). Otherwise, the block is marked as closed >>> (NVM_CHK_ST_CLOSED). Note that even if a block is open and not fully >>> written, it has to be erased in order to be used again. >> >> Should we extend it to do the scan, and update the write pointer as >> well? I think this kind of feature already is baked into pblk? >> > > This is already in place: we scan until empty page and take it from > there. This patch is only for the case in which we start a pblk instance > form scratch. On a device already owned by pblk, we would not have the > problem we are trying to solve here because we know the state. Agree. What I meant was that when we anyway are recovering the state, we could just as well update ->wp and set to NVM_CHK_ST_OPEN and so forth for the initialization phase. > >>> One caveat of this approach is that blocks that have been erased at a >>> moment in time, will always be considered as erased. However, some media >>> might become unstable if blocks are not erased before usage. Since pblk >>> would follow this principle after the state of all blocks fall under >>> pblk's domain, we can consider this as an initialization problem. The >>> trade-off would be to fall back to the old behavior and risk premature >>> media wearing. >> >> The above is up to the device implementation to handle. We cannot >> expect users to understand the intrinsics of media. >> > > Of course. The point is that with this approach, erases are left a bit > in the air and "preventable" write errors might happen, with the previous > the burden was put on the device to deal with double erases. It's a > tradeoff that I want to make clear before the path is taken. Cool. > >>> Signed-off-by: Javier González <javier@cnexlabs.com> >>> --- >>> drivers/lightnvm/pblk-init.c | 138 ++++++++++++++++++++++++++++++++++++++----- >>> 1 file changed, 124 insertions(+), 14 deletions(-) >>> diff --git a/drivers/lightnvm/pblk-init.c b/drivers/lightnvm/pblk-init.c >>> index 3b8aa4a64cac..ce25f1473d8e 100644 >>> --- a/drivers/lightnvm/pblk-init.c >>> +++ b/drivers/lightnvm/pblk-init.c >>> @@ -697,47 +697,138 @@ static void pblk_set_provision(struct pblk *pblk, long nr_free_blks) >>> atomic_set(&pblk->rl.free_user_blocks, nr_free_blks); >>> } >>> +static void pblk_state_complete(struct kref *ref) >>> +{ >>> + struct pblk_pad_rq *pad_rq = container_of(ref, struct pblk_pad_rq, ref); >>> + >>> + complete(&pad_rq->wait); >>> +} >>> + >>> +static void pblk_end_io_state(struct nvm_rq *rqd) >>> +{ >>> + struct pblk_pad_rq *pad_rq = rqd->private; >>> + struct pblk *pblk = pad_rq->pblk; >>> + struct nvm_tgt_dev *dev = pblk->dev; >>> + struct nvm_geo *geo = &dev->geo; >>> + struct pblk_line *line; >>> + struct nvm_chk_meta *chunk; >>> + int pos; >>> + >>> + line = &pblk->lines[pblk_ppa_to_line(rqd->ppa_addr)]; >>> + pos = pblk_ppa_to_pos(geo, rqd->ppa_addr); >>> + >>> + chunk = &line->chks[pos]; >>> + >>> + if (rqd->error == NVM_RSP_ERR_EMPTYPAGE) >>> + chunk->state = NVM_CHK_ST_FREE; >>> + else >>> + chunk->state = NVM_CHK_ST_CLOSED; >>> + >>> + bio_put(rqd->bio); >>> + pblk_free_rqd(pblk, rqd, PBLK_READ); >>> + kref_put(&pad_rq->ref, pblk_state_complete); >>> +} >>> + >>> +static int pblk_check_chunk_state(struct pblk *pblk, struct nvm_chk_meta *chunk, >>> + struct ppa_addr ppa, struct pblk_pad_rq *pad_rq) >>> +{ >>> + struct nvm_rq *rqd; >>> + struct bio *bio; >>> + int ret; >>> + >>> + bio = bio_alloc(GFP_KERNEL, 1); >>> + >>> + if (pblk_bio_add_pages(pblk, bio, GFP_KERNEL, 1)) >>> + goto fail_free_bio; >>> + >>> + rqd = pblk_alloc_rqd(pblk, PBLK_READ); >>> + >>> + rqd->bio = bio; >>> + rqd->opcode = NVM_OP_PREAD; >>> + rqd->flags = pblk_set_read_mode(pblk, PBLK_READ_SEQUENTIAL); >>> + rqd->nr_ppas = 1; >>> + rqd->ppa_addr = ppa; >>> + rqd->end_io = pblk_end_io_state; >>> + rqd->private = pad_rq; >>> + >>> + kref_get(&pad_rq->ref); >>> + >>> + ret = pblk_submit_io(pblk, rqd); >>> + if (ret) { >>> + pr_err("pblk: I/O submissin failed: %d\n", ret); >>> + goto fail_free_rqd; >>> + } >>> + >>> + return NVM_IO_OK; >>> + >>> +fail_free_rqd: >>> + pblk_free_rqd(pblk, rqd, PBLK_READ); >>> + pblk_bio_free_pages(pblk, bio, 0, bio->bi_vcnt); >>> +fail_free_bio: >>> + bio_put(bio); >>> + >>> + return NVM_IO_ERR; >>> +} >>> + >>> 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; >>> + struct pblk_pad_rq *pad_rq; >>> int i, chk_per_lun, nr_bad_chks = 0; >>> + pad_rq = kmalloc(sizeof(struct pblk_pad_rq), GFP_KERNEL); >>> + if (!pad_rq) >>> + return -1; >>> + >>> + pad_rq->pblk = pblk; >>> + init_completion(&pad_rq->wait); >>> + kref_init(&pad_rq->ref); >>> + >>> 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); >>> + struct ppa_addr ppa = rlun->bppa; >>> + int pos = pblk_ppa_to_pos(geo, ppa); >>> 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)) >>> + if (lun_bb_meta[line->id] != NVM_BLK_T_FREE) { >>> + chunk->state = NVM_CHK_ST_OFFLINE; >>> + set_bit(pos, line->blk_bitmap); >>> + nr_bad_chks++; >>> + >>> continue; >>> + } >>> - set_bit(pos, line->blk_bitmap); >>> - nr_bad_chks++; >>> + /* >>> + * In 1.2 spec. chunk state is not persisted by the device. >>> + * Recover the state based on media response. >>> + */ >>> + ppa.g.blk = line->id; >>> + pblk_check_chunk_state(pblk, chunk, ppa, pad_rq); >>> } >>> + kref_put(&pad_rq->ref, pblk_state_complete); >>> + >>> + if (!wait_for_completion_io_timeout(&pad_rq->wait, >>> + msecs_to_jiffies(PBLK_COMMAND_TIMEOUT_MS))) { >>> + pr_err("pblk: state recovery timed out\n"); >>> + return -1; >>> + } >>> + >>> + kfree(pad_rq); >>> return nr_bad_chks; >>> } >>> @@ -1036,6 +1127,23 @@ static int pblk_line_meta_init(struct pblk *pblk) >>> return 0; >>> } >>> +static void check_meta(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 i; >>> + >>> + for (i = 0; i < lm->blk_per_line; i++) { >>> + struct pblk_lun *rlun = &pblk->luns[i]; >>> + struct nvm_chk_meta *chunk; >>> + struct ppa_addr ppa = rlun->bppa; >>> + int pos = pblk_ppa_to_pos(geo, ppa); >>> + >>> + chunk = &line->chks[pos]; >>> + } >>> +} >>> + >>> static int pblk_lines_init(struct pblk *pblk) >>> { >>> struct pblk_line_mgmt *l_mg = &pblk->l_mg; >>> @@ -1077,6 +1185,8 @@ static int pblk_lines_init(struct pblk *pblk) >>> goto fail_free_lines; >>> nr_free_chks += pblk_setup_line_meta(pblk, line, chunk_meta, i); >>> + >>> + check_meta(pblk, line); >>> } >>> if (!nr_free_chks) { >> >> I'm okay with us doing this in pblk for now. Over time, someone may do >> the work move this (and other specific only-1.2/2.0 stuff) into the >> lightnvm subsystem. I don't think pblk should need to care about >> either 1.2 or 2.0. > > That would be ideal. > > Thanks! > ^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH] lightnvm: pblk: recover chunk state on 1.2 devices 2018-06-29 11:28 ` Matias Bjørling @ 2018-07-24 7:54 ` Javier Gonzalez 2018-08-03 11:57 ` Matias Bjørling 0 siblings, 1 reply; 12+ messages in thread From: Javier Gonzalez @ 2018-07-24 7:54 UTC (permalink / raw) To: Matias Bjørling; +Cc: Hans Holmberg, linux-block, linux-kernel > On 29 Jun 2018, at 13.28, Matias Bjørling <mb@lightnvm.io> wrote: > > On 06/29/2018 01:22 PM, Javier Gonzalez wrote: >>> On 29 Jun 2018, at 13.14, Matias Bjørling <mb@lightnvm.io> wrote: >>> >>> On 06/28/2018 11:12 AM, Javier González wrote: >>>> The Open-Channel 1.2 spec does not define a mechanism for the host to >>>> recover the block (chunk) state. As a consequence, a newly format device >>>> will need to reconstruct the state. Currently, pblk assumes that blocks >>>> are not erased, which might cause double-erases in case that the device >>>> does not protect itself against them (which is not specified in the spec >>>> either). >>> >>> It should not be specified in the spec. It is up to the device to handle >>> double erases and not do it. >>> >>>> This patch, reconstructs the state based on read errors. If the first >>>> sector of a block returns and empty page (NVM_RSP_ERR_EMPTYPAGE), then >>>> the block s marked free, i.e., erased and ready to be used >>>> (NVM_CHK_ST_FREE). Otherwise, the block is marked as closed >>>> (NVM_CHK_ST_CLOSED). Note that even if a block is open and not fully >>>> written, it has to be erased in order to be used again. >>> >>> Should we extend it to do the scan, and update the write pointer as >>> well? I think this kind of feature already is baked into pblk? >> This is already in place: we scan until empty page and take it from >> there. This patch is only for the case in which we start a pblk instance >> form scratch. On a device already owned by pblk, we would not have the >> problem we are trying to solve here because we know the state. > > Agree. What I meant was that when we anyway are recovering the state, > we could just as well update ->wp and set to NVM_CHK_ST_OPEN and so > forth for the initialization phase. > In 1.2 the use of chunk metadata is purely fictional. We respect the chunk state machine as we transition lines, but all the write pointers are ignored. Instead, we use the line bitmap to point to the next writable entry. This is BTW the same way we it in open lines on 2.0 too. Chunk metadata is only used to setup the bitmaps on init/recovery. From here on, we use the bitmap to find the next writable sector, without worrying about the specific per-chunk write pointer. Thus, updating chunk metadata here has no effect. Does this make sense to you? Javier ^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH] lightnvm: pblk: recover chunk state on 1.2 devices 2018-07-24 7:54 ` Javier Gonzalez @ 2018-08-03 11:57 ` Matias Bjørling 2018-08-03 12:02 ` Javier Gonzalez 0 siblings, 1 reply; 12+ messages in thread From: Matias Bjørling @ 2018-08-03 11:57 UTC (permalink / raw) To: javier; +Cc: hans.holmberg, linux-block, linux-kernel On 07/24/2018 09:54 AM, Javier Gonzalez wrote: >> On 29 Jun 2018, at 13.28, Matias Bjørling <mb@lightnvm.io> wrote: >> >> On 06/29/2018 01:22 PM, Javier Gonzalez wrote: >>>> On 29 Jun 2018, at 13.14, Matias Bjørling <mb@lightnvm.io> wrote: >>>> >>>> On 06/28/2018 11:12 AM, Javier González wrote: >>>>> The Open-Channel 1.2 spec does not define a mechanism for the host to >>>>> recover the block (chunk) state. As a consequence, a newly format device >>>>> will need to reconstruct the state. Currently, pblk assumes that blocks >>>>> are not erased, which might cause double-erases in case that the device >>>>> does not protect itself against them (which is not specified in the spec >>>>> either). >>>> >>>> It should not be specified in the spec. It is up to the device to handle >>>> double erases and not do it. >>>> >>>>> This patch, reconstructs the state based on read errors. If the first >>>>> sector of a block returns and empty page (NVM_RSP_ERR_EMPTYPAGE), then >>>>> the block s marked free, i.e., erased and ready to be used >>>>> (NVM_CHK_ST_FREE). Otherwise, the block is marked as closed >>>>> (NVM_CHK_ST_CLOSED). Note that even if a block is open and not fully >>>>> written, it has to be erased in order to be used again. >>>> >>>> Should we extend it to do the scan, and update the write pointer as >>>> well? I think this kind of feature already is baked into pblk? >>> This is already in place: we scan until empty page and take it from >>> there. This patch is only for the case in which we start a pblk instance >>> form scratch. On a device already owned by pblk, we would not have the >>> problem we are trying to solve here because we know the state. >> >> Agree. What I meant was that when we anyway are recovering the state, >> we could just as well update ->wp and set to NVM_CHK_ST_OPEN and so >> forth for the initialization phase. >> > > In 1.2 the use of chunk metadata is purely fictional. We respect the > chunk state machine as we transition lines, but all the write pointers > are ignored. Instead, we use the line bitmap to point to the next > writable entry. This is BTW the same way we it in open lines on 2.0 too. > Now I understand where you are coming from. I had the understanding that we where using the write pointer now that we moved to 2.0, looking through the code, that wasn't the case. :) Which means that pblk doesn't work with a devices that implements 2.0. Yikes... I knew I had forgot a detail when support was added into pblk. There are no empty sector marker in the 2.0 spec, since it uses the write pointer to know where it is in the chunk. So there is a bit of work to do there. Since this properly is a bit more work to do, I'll look into it after FMS. I'm also moving the explicit coding of 1.2/2.0 chunk / bad block fixing into core, so pblk can be simplfied, and doesn't have to think to manage each version separately. > Chunk metadata is only used to setup the bitmaps on init/recovery. From > here on, we use the bitmap to find the next writable sector, without > worrying about the specific per-chunk write pointer. Thus, updating > chunk metadata here has no effect. > > Does this make sense to you? > > Javier > ^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH] lightnvm: pblk: recover chunk state on 1.2 devices 2018-08-03 11:57 ` Matias Bjørling @ 2018-08-03 12:02 ` Javier Gonzalez 2018-08-03 12:30 ` Matias Bjørling 0 siblings, 1 reply; 12+ messages in thread From: Javier Gonzalez @ 2018-08-03 12:02 UTC (permalink / raw) To: Matias Bjørling; +Cc: Hans Holmberg, linux-block, linux-kernel [-- Attachment #1: Type: text/plain, Size: 4105 bytes --] > On 3 Aug 2018, at 13.57, Matias Bjørling <mb@lightnvm.io> wrote: > > On 07/24/2018 09:54 AM, Javier Gonzalez wrote: >>> On 29 Jun 2018, at 13.28, Matias Bjørling <mb@lightnvm.io> wrote: >>> >>> On 06/29/2018 01:22 PM, Javier Gonzalez wrote: >>>>> On 29 Jun 2018, at 13.14, Matias Bjørling <mb@lightnvm.io> wrote: >>>>> >>>>> On 06/28/2018 11:12 AM, Javier González wrote: >>>>>> The Open-Channel 1.2 spec does not define a mechanism for the host to >>>>>> recover the block (chunk) state. As a consequence, a newly format device >>>>>> will need to reconstruct the state. Currently, pblk assumes that blocks >>>>>> are not erased, which might cause double-erases in case that the device >>>>>> does not protect itself against them (which is not specified in the spec >>>>>> either). >>>>> >>>>> It should not be specified in the spec. It is up to the device to handle >>>>> double erases and not do it. >>>>> >>>>>> This patch, reconstructs the state based on read errors. If the first >>>>>> sector of a block returns and empty page (NVM_RSP_ERR_EMPTYPAGE), then >>>>>> the block s marked free, i.e., erased and ready to be used >>>>>> (NVM_CHK_ST_FREE). Otherwise, the block is marked as closed >>>>>> (NVM_CHK_ST_CLOSED). Note that even if a block is open and not fully >>>>>> written, it has to be erased in order to be used again. >>>>> >>>>> Should we extend it to do the scan, and update the write pointer as >>>>> well? I think this kind of feature already is baked into pblk? >>>> This is already in place: we scan until empty page and take it from >>>> there. This patch is only for the case in which we start a pblk instance >>>> form scratch. On a device already owned by pblk, we would not have the >>>> problem we are trying to solve here because we know the state. >>> >>> Agree. What I meant was that when we anyway are recovering the state, >>> we could just as well update ->wp and set to NVM_CHK_ST_OPEN and so >>> forth for the initialization phase. >> In 1.2 the use of chunk metadata is purely fictional. We respect the >> chunk state machine as we transition lines, but all the write pointers >> are ignored. Instead, we use the line bitmap to point to the next >> writable entry. This is BTW the same way we it in open lines on 2.0 too. > > Now I understand where you are coming from. I had the understanding > that we where using the write pointer now that we moved to 2.0, > looking through the code, that wasn't the case. :) Which means that > pblk doesn't work with a devices that implements 2.0. Yikes... I knew > I had forgot a detail when support was added into pblk. > I think you misunderstood; pblk does support 2.0 devices. What happens is that we transform the per chunk WP in 2.0 into the line bitmap to simplify the lookup. The point being that we do not need to create a fictional chunk for 1.2 devices since we do the translation to the bitmap directly. Does this make sense? > There are no empty sector marker in the 2.0 spec, since it uses the > write pointer to know where it is in the chunk. So there is a bit of > work to do there. > Yes. And for 2.0 devices we go and look at the WP, but for 1.2 devices we need to scan. > Since this properly is a bit more work to do, I'll look into it after FMS. > Look the comments above. All we need for 2.0 support is in place. We can talk about it f2f. > I'm also moving the explicit coding of 1.2/2.0 chunk / bad block > fixing into core, so pblk can be simplfied, and doesn't have to think > to manage each version separately. > Good. I have a patch I was expecting to send after FMS for moving chunk / bad block out of pblk for the same reason. If you're doing the same thing I can stop looking into it... > >> Chunk metadata is only used to setup the bitmaps on init/recovery. From >> here on, we use the bitmap to find the next writable sector, without >> worrying about the specific per-chunk write pointer. Thus, updating >> chunk metadata here has no effect. >> Does this make sense to you? >> Javier [-- Attachment #2: Message signed with OpenPGP --] [-- Type: application/pgp-signature, Size: 833 bytes --] ^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH] lightnvm: pblk: recover chunk state on 1.2 devices 2018-08-03 12:02 ` Javier Gonzalez @ 2018-08-03 12:30 ` Matias Bjørling 2018-08-03 12:37 ` Javier Gonzalez 0 siblings, 1 reply; 12+ messages in thread From: Matias Bjørling @ 2018-08-03 12:30 UTC (permalink / raw) To: javier; +Cc: hans.holmberg, linux-block, linux-kernel On 08/03/2018 02:02 PM, Javier Gonzalez wrote: >> On 3 Aug 2018, at 13.57, Matias Bjørling <mb@lightnvm.io> wrote: >> >> On 07/24/2018 09:54 AM, Javier Gonzalez wrote: >>>> On 29 Jun 2018, at 13.28, Matias Bjørling <mb@lightnvm.io> wrote: >>>> >>>> On 06/29/2018 01:22 PM, Javier Gonzalez wrote: >>>>>> On 29 Jun 2018, at 13.14, Matias Bjørling <mb@lightnvm.io> wrote: >>>>>> >>>>>> On 06/28/2018 11:12 AM, Javier González wrote: >>>>>>> The Open-Channel 1.2 spec does not define a mechanism for the host to >>>>>>> recover the block (chunk) state. As a consequence, a newly format device >>>>>>> will need to reconstruct the state. Currently, pblk assumes that blocks >>>>>>> are not erased, which might cause double-erases in case that the device >>>>>>> does not protect itself against them (which is not specified in the spec >>>>>>> either). >>>>>> >>>>>> It should not be specified in the spec. It is up to the device to handle >>>>>> double erases and not do it. >>>>>> >>>>>>> This patch, reconstructs the state based on read errors. If the first >>>>>>> sector of a block returns and empty page (NVM_RSP_ERR_EMPTYPAGE), then >>>>>>> the block s marked free, i.e., erased and ready to be used >>>>>>> (NVM_CHK_ST_FREE). Otherwise, the block is marked as closed >>>>>>> (NVM_CHK_ST_CLOSED). Note that even if a block is open and not fully >>>>>>> written, it has to be erased in order to be used again. >>>>>> >>>>>> Should we extend it to do the scan, and update the write pointer as >>>>>> well? I think this kind of feature already is baked into pblk? >>>>> This is already in place: we scan until empty page and take it from >>>>> there. This patch is only for the case in which we start a pblk instance >>>>> form scratch. On a device already owned by pblk, we would not have the >>>>> problem we are trying to solve here because we know the state. >>>> >>>> Agree. What I meant was that when we anyway are recovering the state, >>>> we could just as well update ->wp and set to NVM_CHK_ST_OPEN and so >>>> forth for the initialization phase. >>> In 1.2 the use of chunk metadata is purely fictional. We respect the >>> chunk state machine as we transition lines, but all the write pointers >>> are ignored. Instead, we use the line bitmap to point to the next >>> writable entry. This is BTW the same way we it in open lines on 2.0 too. >> >> Now I understand where you are coming from. I had the understanding >> that we where using the write pointer now that we moved to 2.0, >> looking through the code, that wasn't the case. :) Which means that >> pblk doesn't work with a devices that implements 2.0. Yikes... I knew >> I had forgot a detail when support was added into pblk. >> > > I think you misunderstood; pblk does support 2.0 devices. What happens > is that we transform the per chunk WP in 2.0 into the line bitmap to > simplify the lookup. The point being that we do not need to create a > fictional chunk for 1.2 devices since we do the translation to the > bitmap directly. Does this make sense? The chunk->wp isn't used anywhere. So it can't take wp into account. It uses the EMPTYPAGE marker from 1.2 instead. See pblk-core and pblk-recovery. > >> There are no empty sector marker in the 2.0 spec, since it uses the >> write pointer to know where it is in the chunk. So there is a bit of >> work to do there. >> > > Yes. And for 2.0 devices we go and look at the WP, but for 1.2 devices we > need to scan. > >> Since this properly is a bit more work to do, I'll look into it after FMS. >> > > Look the comments above. All we need for 2.0 support is in place. We can > talk about it f2f. > >> I'm also moving the explicit coding of 1.2/2.0 chunk / bad block >> fixing into core, so pblk can be simplfied, and doesn't have to think >> to manage each version separately. >> > > Good. I have a patch I was expecting to send after FMS for moving chunk > / bad block out of pblk for the same reason. If you're doing the same > thing I can stop looking into it... I am, will post when done. > >> >>> Chunk metadata is only used to setup the bitmaps on init/recovery. From >>> here on, we use the bitmap to find the next writable sector, without >>> worrying about the specific per-chunk write pointer. Thus, updating >>> chunk metadata here has no effect. >>> Does this make sense to you? >>> Javier ^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH] lightnvm: pblk: recover chunk state on 1.2 devices 2018-08-03 12:30 ` Matias Bjørling @ 2018-08-03 12:37 ` Javier Gonzalez 2018-08-03 12:46 ` Matias Bjørling 0 siblings, 1 reply; 12+ messages in thread From: Javier Gonzalez @ 2018-08-03 12:37 UTC (permalink / raw) To: Matias Bjørling; +Cc: Hans Holmberg, linux-block, linux-kernel [-- Attachment #1: Type: text/plain, Size: 4327 bytes --] > On 3 Aug 2018, at 14.30, Matias Bjørling <mb@lightnvm.io> wrote: > > On 08/03/2018 02:02 PM, Javier Gonzalez wrote: >>> On 3 Aug 2018, at 13.57, Matias Bjørling <mb@lightnvm.io> wrote: >>> >>> On 07/24/2018 09:54 AM, Javier Gonzalez wrote: >>>>> On 29 Jun 2018, at 13.28, Matias Bjørling <mb@lightnvm.io> wrote: >>>>> >>>>> On 06/29/2018 01:22 PM, Javier Gonzalez wrote: >>>>>>> On 29 Jun 2018, at 13.14, Matias Bjørling <mb@lightnvm.io> wrote: >>>>>>> >>>>>>> On 06/28/2018 11:12 AM, Javier González wrote: >>>>>>>> The Open-Channel 1.2 spec does not define a mechanism for the host to >>>>>>>> recover the block (chunk) state. As a consequence, a newly format device >>>>>>>> will need to reconstruct the state. Currently, pblk assumes that blocks >>>>>>>> are not erased, which might cause double-erases in case that the device >>>>>>>> does not protect itself against them (which is not specified in the spec >>>>>>>> either). >>>>>>> >>>>>>> It should not be specified in the spec. It is up to the device to handle >>>>>>> double erases and not do it. >>>>>>> >>>>>>>> This patch, reconstructs the state based on read errors. If the first >>>>>>>> sector of a block returns and empty page (NVM_RSP_ERR_EMPTYPAGE), then >>>>>>>> the block s marked free, i.e., erased and ready to be used >>>>>>>> (NVM_CHK_ST_FREE). Otherwise, the block is marked as closed >>>>>>>> (NVM_CHK_ST_CLOSED). Note that even if a block is open and not fully >>>>>>>> written, it has to be erased in order to be used again. >>>>>>> >>>>>>> Should we extend it to do the scan, and update the write pointer as >>>>>>> well? I think this kind of feature already is baked into pblk? >>>>>> This is already in place: we scan until empty page and take it from >>>>>> there. This patch is only for the case in which we start a pblk instance >>>>>> form scratch. On a device already owned by pblk, we would not have the >>>>>> problem we are trying to solve here because we know the state. >>>>> >>>>> Agree. What I meant was that when we anyway are recovering the state, >>>>> we could just as well update ->wp and set to NVM_CHK_ST_OPEN and so >>>>> forth for the initialization phase. >>>> In 1.2 the use of chunk metadata is purely fictional. We respect the >>>> chunk state machine as we transition lines, but all the write pointers >>>> are ignored. Instead, we use the line bitmap to point to the next >>>> writable entry. This is BTW the same way we it in open lines on 2.0 too. >>> >>> Now I understand where you are coming from. I had the understanding >>> that we where using the write pointer now that we moved to 2.0, >>> looking through the code, that wasn't the case. :) Which means that >>> pblk doesn't work with a devices that implements 2.0. Yikes... I knew >>> I had forgot a detail when support was added into pblk. >> I think you misunderstood; pblk does support 2.0 devices. What happens >> is that we transform the per chunk WP in 2.0 into the line bitmap to >> simplify the lookup. The point being that we do not need to create a >> fictional chunk for 1.2 devices since we do the translation to the >> bitmap directly. Does this make sense? > > The chunk->wp isn't used anywhere. So it can't take wp into account. > It uses the EMPTYPAGE marker from 1.2 instead. See pblk-core and > pblk-recovery. > I see that the patches for this are still internal. Will post for 4.20 >>> There are no empty sector marker in the 2.0 spec, since it uses the >>> write pointer to know where it is in the chunk. So there is a bit of >>> work to do there. >> Yes. And for 2.0 devices we go and look at the WP, but for 1.2 devices we >> need to scan. >>> Since this properly is a bit more work to do, I'll look into it after FMS. >> Look the comments above. All we need for 2.0 support is in place. We can >> talk about it f2f. >>> I'm also moving the explicit coding of 1.2/2.0 chunk / bad block >>> fixing into core, so pblk can be simplfied, and doesn't have to think >>> to manage each version separately. >> Good. I have a patch I was expecting to send after FMS for moving chunk >> / bad block out of pblk for the same reason. If you're doing the same >> thing I can stop looking into it... > > I am, will post when done. Cool! Javier [-- Attachment #2: Message signed with OpenPGP --] [-- Type: application/pgp-signature, Size: 833 bytes --] ^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH] lightnvm: pblk: recover chunk state on 1.2 devices 2018-08-03 12:37 ` Javier Gonzalez @ 2018-08-03 12:46 ` Matias Bjørling 2018-08-03 12:53 ` Javier Gonzalez 0 siblings, 1 reply; 12+ messages in thread From: Matias Bjørling @ 2018-08-03 12:46 UTC (permalink / raw) To: javier; +Cc: hans.holmberg, linux-block, linux-kernel On 08/03/2018 02:37 PM, Javier Gonzalez wrote: >> On 3 Aug 2018, at 14.30, Matias Bjørling <mb@lightnvm.io> wrote: >> >> On 08/03/2018 02:02 PM, Javier Gonzalez wrote: >>>> On 3 Aug 2018, at 13.57, Matias Bjørling <mb@lightnvm.io> wrote: >>>> >>>> On 07/24/2018 09:54 AM, Javier Gonzalez wrote: >>>>>> On 29 Jun 2018, at 13.28, Matias Bjørling <mb@lightnvm.io> wrote: >>>>>> >>>>>> On 06/29/2018 01:22 PM, Javier Gonzalez wrote: >>>>>>>> On 29 Jun 2018, at 13.14, Matias Bjørling <mb@lightnvm.io> wrote: >>>>>>>> >>>>>>>> On 06/28/2018 11:12 AM, Javier González wrote: >>>>>>>>> The Open-Channel 1.2 spec does not define a mechanism for the host to >>>>>>>>> recover the block (chunk) state. As a consequence, a newly format device >>>>>>>>> will need to reconstruct the state. Currently, pblk assumes that blocks >>>>>>>>> are not erased, which might cause double-erases in case that the device >>>>>>>>> does not protect itself against them (which is not specified in the spec >>>>>>>>> either). >>>>>>>> >>>>>>>> It should not be specified in the spec. It is up to the device to handle >>>>>>>> double erases and not do it. >>>>>>>> >>>>>>>>> This patch, reconstructs the state based on read errors. If the first >>>>>>>>> sector of a block returns and empty page (NVM_RSP_ERR_EMPTYPAGE), then >>>>>>>>> the block s marked free, i.e., erased and ready to be used >>>>>>>>> (NVM_CHK_ST_FREE). Otherwise, the block is marked as closed >>>>>>>>> (NVM_CHK_ST_CLOSED). Note that even if a block is open and not fully >>>>>>>>> written, it has to be erased in order to be used again. >>>>>>>> >>>>>>>> Should we extend it to do the scan, and update the write pointer as >>>>>>>> well? I think this kind of feature already is baked into pblk? >>>>>>> This is already in place: we scan until empty page and take it from >>>>>>> there. This patch is only for the case in which we start a pblk instance >>>>>>> form scratch. On a device already owned by pblk, we would not have the >>>>>>> problem we are trying to solve here because we know the state. >>>>>> >>>>>> Agree. What I meant was that when we anyway are recovering the state, >>>>>> we could just as well update ->wp and set to NVM_CHK_ST_OPEN and so >>>>>> forth for the initialization phase. >>>>> In 1.2 the use of chunk metadata is purely fictional. We respect the >>>>> chunk state machine as we transition lines, but all the write pointers >>>>> are ignored. Instead, we use the line bitmap to point to the next >>>>> writable entry. This is BTW the same way we it in open lines on 2.0 too. >>>> >>>> Now I understand where you are coming from. I had the understanding >>>> that we where using the write pointer now that we moved to 2.0, >>>> looking through the code, that wasn't the case. :) Which means that >>>> pblk doesn't work with a devices that implements 2.0. Yikes... I knew >>>> I had forgot a detail when support was added into pblk. >>> I think you misunderstood; pblk does support 2.0 devices. What happens >>> is that we transform the per chunk WP in 2.0 into the line bitmap to >>> simplify the lookup. The point being that we do not need to create a >>> fictional chunk for 1.2 devices since we do the translation to the >>> bitmap directly. Does this make sense? >> >> The chunk->wp isn't used anywhere. So it can't take wp into account. >> It uses the EMPTYPAGE marker from 1.2 instead. See pblk-core and >> pblk-recovery. >> > > I see that the patches for this are still internal. Will post for 4.20 > Thanks. Please also put a Fixes: on, so it gets backported appropriately. ^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH] lightnvm: pblk: recover chunk state on 1.2 devices 2018-08-03 12:46 ` Matias Bjørling @ 2018-08-03 12:53 ` Javier Gonzalez 0 siblings, 0 replies; 12+ messages in thread From: Javier Gonzalez @ 2018-08-03 12:53 UTC (permalink / raw) To: Matias Bjørling; +Cc: Hans Holmberg, linux-block, linux-kernel [-- Attachment #1: Type: text/plain, Size: 3783 bytes --] > On 3 Aug 2018, at 14.46, Matias Bjørling <mb@lightnvm.io> wrote: > > On 08/03/2018 02:37 PM, Javier Gonzalez wrote: >>> On 3 Aug 2018, at 14.30, Matias Bjørling <mb@lightnvm.io> wrote: >>> >>> On 08/03/2018 02:02 PM, Javier Gonzalez wrote: >>>>> On 3 Aug 2018, at 13.57, Matias Bjørling <mb@lightnvm.io> wrote: >>>>> >>>>> On 07/24/2018 09:54 AM, Javier Gonzalez wrote: >>>>>>> On 29 Jun 2018, at 13.28, Matias Bjørling <mb@lightnvm.io> wrote: >>>>>>> >>>>>>> On 06/29/2018 01:22 PM, Javier Gonzalez wrote: >>>>>>>>> On 29 Jun 2018, at 13.14, Matias Bjørling <mb@lightnvm.io> wrote: >>>>>>>>> >>>>>>>>> On 06/28/2018 11:12 AM, Javier González wrote: >>>>>>>>>> The Open-Channel 1.2 spec does not define a mechanism for the host to >>>>>>>>>> recover the block (chunk) state. As a consequence, a newly format device >>>>>>>>>> will need to reconstruct the state. Currently, pblk assumes that blocks >>>>>>>>>> are not erased, which might cause double-erases in case that the device >>>>>>>>>> does not protect itself against them (which is not specified in the spec >>>>>>>>>> either). >>>>>>>>> >>>>>>>>> It should not be specified in the spec. It is up to the device to handle >>>>>>>>> double erases and not do it. >>>>>>>>> >>>>>>>>>> This patch, reconstructs the state based on read errors. If the first >>>>>>>>>> sector of a block returns and empty page (NVM_RSP_ERR_EMPTYPAGE), then >>>>>>>>>> the block s marked free, i.e., erased and ready to be used >>>>>>>>>> (NVM_CHK_ST_FREE). Otherwise, the block is marked as closed >>>>>>>>>> (NVM_CHK_ST_CLOSED). Note that even if a block is open and not fully >>>>>>>>>> written, it has to be erased in order to be used again. >>>>>>>>> >>>>>>>>> Should we extend it to do the scan, and update the write pointer as >>>>>>>>> well? I think this kind of feature already is baked into pblk? >>>>>>>> This is already in place: we scan until empty page and take it from >>>>>>>> there. This patch is only for the case in which we start a pblk instance >>>>>>>> form scratch. On a device already owned by pblk, we would not have the >>>>>>>> problem we are trying to solve here because we know the state. >>>>>>> >>>>>>> Agree. What I meant was that when we anyway are recovering the state, >>>>>>> we could just as well update ->wp and set to NVM_CHK_ST_OPEN and so >>>>>>> forth for the initialization phase. >>>>>> In 1.2 the use of chunk metadata is purely fictional. We respect the >>>>>> chunk state machine as we transition lines, but all the write pointers >>>>>> are ignored. Instead, we use the line bitmap to point to the next >>>>>> writable entry. This is BTW the same way we it in open lines on 2.0 too. >>>>> >>>>> Now I understand where you are coming from. I had the understanding >>>>> that we where using the write pointer now that we moved to 2.0, >>>>> looking through the code, that wasn't the case. :) Which means that >>>>> pblk doesn't work with a devices that implements 2.0. Yikes... I knew >>>>> I had forgot a detail when support was added into pblk. >>>> I think you misunderstood; pblk does support 2.0 devices. What happens >>>> is that we transform the per chunk WP in 2.0 into the line bitmap to >>>> simplify the lookup. The point being that we do not need to create a >>>> fictional chunk for 1.2 devices since we do the translation to the >>>> bitmap directly. Does this make sense? >>> >>> The chunk->wp isn't used anywhere. So it can't take wp into account. >>> It uses the EMPTYPAGE marker from 1.2 instead. See pblk-core and >>> pblk-recovery. >> I see that the patches for this are still internal. Will post for 4.20 > > Thanks. Please also put a Fixes: on, so it gets backported appropriately. Sure. I will. [-- Attachment #2: Message signed with OpenPGP --] [-- Type: application/pgp-signature, Size: 833 bytes --] ^ permalink raw reply [flat|nested] 12+ messages in thread
end of thread, other threads:[~2018-08-03 12:55 UTC | newest] Thread overview: 12+ messages (download: mbox.gz / follow: Atom feed) -- links below jump to the message on this page -- 2018-06-28 9:12 [PATCH] lightnvm: pblk: recover chunk state on 1.2 devices Javier González 2018-06-28 9:12 ` Javier González 2018-06-29 11:14 ` Matias Bjørling 2018-06-29 11:22 ` Javier Gonzalez 2018-06-29 11:28 ` Matias Bjørling 2018-07-24 7:54 ` Javier Gonzalez 2018-08-03 11:57 ` Matias Bjørling 2018-08-03 12:02 ` Javier Gonzalez 2018-08-03 12:30 ` Matias Bjørling 2018-08-03 12:37 ` Javier Gonzalez 2018-08-03 12:46 ` Matias Bjørling 2018-08-03 12:53 ` Javier Gonzalez
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).