| Message ID | 20210122080014.174391-2-damien.lemoal@wdc.com (mailing list archive) |
|---|---|
| State | Superseded |
| Headers | show |
| Series | block: add zone write granularity limit | expand |
> @@ -864,18 +891,20 @@ void blk_queue_set_zoned(struct gendisk *disk, enum blk_zoned_model model) > * partitions and zoned block device support is enabled, else > * we do nothing special as far as the block layer is concerned. > */ > - if (!IS_ENABLED(CONFIG_BLK_DEV_ZONED) || > - disk_has_partitions(disk)) > - model = BLK_ZONED_NONE; > - break; > + if (IS_ENABLED(CONFIG_BLK_DEV_ZONED) && > + !disk_has_partitions(disk)) > + break; > + model = BLK_ZONED_NONE; > + fallthrough; > case BLK_ZONED_NONE: > default: > if (WARN_ON_ONCE(model != BLK_ZONED_NONE)) > model = BLK_ZONED_NONE; > + q->limits.zone_write_granularity = 0; > break; > } > > - disk->queue->limits.zoned = model; > + q->limits.zoned = model; > } This looks a little strange. If we special case zoned vs not zoned here anyway, why not set the zone_write_granularity to the logical block size here by default.
On 2021/01/22 17:42, Christoph Hellwig wrote: >> @@ -864,18 +891,20 @@ void blk_queue_set_zoned(struct gendisk *disk, enum blk_zoned_model model) >> * partitions and zoned block device support is enabled, else >> * we do nothing special as far as the block layer is concerned. >> */ >> - if (!IS_ENABLED(CONFIG_BLK_DEV_ZONED) || >> - disk_has_partitions(disk)) >> - model = BLK_ZONED_NONE; >> - break; >> + if (IS_ENABLED(CONFIG_BLK_DEV_ZONED) && >> + !disk_has_partitions(disk)) >> + break; >> + model = BLK_ZONED_NONE; >> + fallthrough; >> case BLK_ZONED_NONE: >> default: >> if (WARN_ON_ONCE(model != BLK_ZONED_NONE)) >> model = BLK_ZONED_NONE; >> + q->limits.zone_write_granularity = 0; >> break; >> } >> >> - disk->queue->limits.zoned = model; >> + q->limits.zoned = model; >> } > > This looks a little strange. If we special case zoned vs not zoned > here anyway, why not set the zone_write_granularity to the logical > block size here by default. The convention is zone_write_granularity == 0 for the BLK_ZONED_NONE case. Hence the reset here if we force the zoned model to none for HA drives. This way, this does not create a special case for HA drives used as regular disks. Of note is that there is something a little weird in the sd_zbc.c code that needs fixing: blk_queue_set_zoned() is called before sd_zbc_read_zones() is executed and that function will check the zones of an HA drive and set the queue nr_zones and max zone append sectors, even if blk_queue_set_zoned() set the zoned model to none due to partitions. That makes the BLK_ZONED_NONE case of HA drives a little weird since zone information is visible and correct but the model says "none". As long as users separate zoned vs not-zoned cases by looking at the zoned model, this does not create any problem, but that is not pretty. Will send a separate patch to clean that up and have something consistent with regular disks for this special HA case. The above blk_queue_set_zoned() function can be used to cleanup the zones information for an HA drive that is used as a regular disk (nr_zones, zone append sectors and zone bitmaps).
Damien, > To solve this, introduce the zone_write_granularity queue limit to > indicate the alignment constraint, in bytes, of write operations into > zones of a zoned block device. This new limit is exported as a > read-only sysfs queue attribute and the helper > blk_queue_zone_write_granularity() introduced for drivers to set this > limit. The scsi disk driver is modified to use this helper to set > host-managed SMR disk zone write granularity to the disk physical > block size. The ZNS support code of the NVMe driver is also modified > to use this helper to set the new limit to the logical block size of > the namespace. The nullblk driver is similarly modified too. Looks fine. Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
On Fri, Jan 22, 2021 at 08:56:58AM +0000, Damien Le Moal wrote: > > This looks a little strange. If we special case zoned vs not zoned > > here anyway, why not set the zone_write_granularity to the logical > > block size here by default. > > The convention is zone_write_granularity == 0 for the BLK_ZONED_NONE case. Hence > the reset here if we force the zoned model to none for HA drives. This way, this > does not create a special case for HA drives used as regular disks. Just inititialize it for all cases if you initialize it for some here. That way everyone but sd already gets a right default and life becomes simpler.
On 2021/01/24 19:07, Christoph Hellwig wrote: > On Fri, Jan 22, 2021 at 08:56:58AM +0000, Damien Le Moal wrote: >>> This looks a little strange. If we special case zoned vs not zoned >>> here anyway, why not set the zone_write_granularity to the logical >>> block size here by default. >> >> The convention is zone_write_granularity == 0 for the BLK_ZONED_NONE case. Hence >> the reset here if we force the zoned model to none for HA drives. This way, this >> does not create a special case for HA drives used as regular disks. > > Just inititialize it for all cases if you initialize it for some here. > That way everyone but sd already gets a right default and life becomes > simpler. True for nullblk, and that also simplifies sd a little. But not for nvme, blk_queue_set_zoned() is not used AND nvme_update_zone_info() is called before nvme_update_disk_info() where the NS logical block size is set. So some surgery/cleanups would be needed to benefit. I could add a cleanup for this, but not entirely sure if calling nvme_update_zone_info() after nvme_update_disk_info() is OK. Thoughts ?
On 2021/01/25 14:32, Damien Le Moal wrote: > On 2021/01/24 19:07, Christoph Hellwig wrote: >> On Fri, Jan 22, 2021 at 08:56:58AM +0000, Damien Le Moal wrote: >>>> This looks a little strange. If we special case zoned vs not zoned >>>> here anyway, why not set the zone_write_granularity to the logical >>>> block size here by default. >>> >>> The convention is zone_write_granularity == 0 for the BLK_ZONED_NONE case. Hence >>> the reset here if we force the zoned model to none for HA drives. This way, this >>> does not create a special case for HA drives used as regular disks. >> >> Just inititialize it for all cases if you initialize it for some here. >> That way everyone but sd already gets a right default and life becomes >> simpler. > > True for nullblk, and that also simplifies sd a little. But not for nvme, > blk_queue_set_zoned() is not used AND nvme_update_zone_info() is called before > nvme_update_disk_info() where the NS logical block size is set. So some > surgery/cleanups would be needed to benefit. I could add a cleanup for this, but > not entirely sure if calling nvme_update_zone_info() after > nvme_update_disk_info() is OK. Thoughts ? Also, because nvme_update_zone_info() is called without the logical block size being set yet, this patch was not good for ZNS.
diff --git a/Documentation/block/queue-sysfs.rst b/Documentation/block/queue-sysfs.rst index 2638d3446b79..c8bf8bc3c03a 100644 --- a/Documentation/block/queue-sysfs.rst +++ b/Documentation/block/queue-sysfs.rst @@ -273,4 +273,11 @@ devices are described in the ZBC (Zoned Block Commands) and ZAC do not support zone commands, they will be treated as regular block devices and zoned will report "none". +zone_write_granularity (RO) +--------------------------- +This indicates the alignment constraint, in bytes, for write operations in +sequential zones of zoned block devices (devices with a zoned attributed +that reports "host-managed" or "host-aware"). This value is always 0 for +regular block devices. + Jens Axboe <jens.axboe@oracle.com>, February 2009 diff --git a/block/blk-settings.c b/block/blk-settings.c index 43990b1d148b..48872b4085d4 100644 --- a/block/blk-settings.c +++ b/block/blk-settings.c @@ -60,6 +60,7 @@ void blk_set_default_limits(struct queue_limits *lim) lim->io_opt = 0; lim->misaligned = 0; lim->zoned = BLK_ZONED_NONE; + lim->zone_write_granularity = 0; } EXPORT_SYMBOL(blk_set_default_limits); @@ -366,6 +367,28 @@ void blk_queue_physical_block_size(struct request_queue *q, unsigned int size) } EXPORT_SYMBOL(blk_queue_physical_block_size); +/** + * blk_queue_zone_write_granularity - set zone write granularity for the queue + * @q: the request queue for the zoned device + * @size: the zone write granularity size, in bytes + * + * Description: + * This should be set to the lowest possible size allowing to write in + * sequential zones of a zoned block device. + */ +void blk_queue_zone_write_granularity(struct request_queue *q, + unsigned int size) +{ + if (WARN_ON_ONCE(!blk_queue_is_zoned(q))) + return; + + q->limits.zone_write_granularity = size; + + if (q->limits.zone_write_granularity < q->limits.logical_block_size) + q->limits.zone_write_granularity = q->limits.logical_block_size; +} +EXPORT_SYMBOL_GPL(blk_queue_zone_write_granularity); + /** * blk_queue_alignment_offset - set physical block alignment offset * @q: the request queue for the device @@ -631,6 +654,8 @@ int blk_stack_limits(struct queue_limits *t, struct queue_limits *b, t->discard_granularity; } + t->zone_write_granularity = max(t->zone_write_granularity, + b->zone_write_granularity); t->zoned = max(t->zoned, b->zoned); return ret; } @@ -847,6 +872,8 @@ EXPORT_SYMBOL_GPL(blk_queue_can_use_dma_map_merging); */ void blk_queue_set_zoned(struct gendisk *disk, enum blk_zoned_model model) { + struct request_queue *q = disk->queue; + switch (model) { case BLK_ZONED_HM: /* @@ -864,18 +891,20 @@ void blk_queue_set_zoned(struct gendisk *disk, enum blk_zoned_model model) * partitions and zoned block device support is enabled, else * we do nothing special as far as the block layer is concerned. */ - if (!IS_ENABLED(CONFIG_BLK_DEV_ZONED) || - disk_has_partitions(disk)) - model = BLK_ZONED_NONE; - break; + if (IS_ENABLED(CONFIG_BLK_DEV_ZONED) && + !disk_has_partitions(disk)) + break; + model = BLK_ZONED_NONE; + fallthrough; case BLK_ZONED_NONE: default: if (WARN_ON_ONCE(model != BLK_ZONED_NONE)) model = BLK_ZONED_NONE; + q->limits.zone_write_granularity = 0; break; } - disk->queue->limits.zoned = model; + q->limits.zoned = model; } EXPORT_SYMBOL_GPL(blk_queue_set_zoned); diff --git a/block/blk-sysfs.c b/block/blk-sysfs.c index b513f1683af0..ae39c7f3d83d 100644 --- a/block/blk-sysfs.c +++ b/block/blk-sysfs.c @@ -219,6 +219,12 @@ static ssize_t queue_write_zeroes_max_show(struct request_queue *q, char *page) (unsigned long long)q->limits.max_write_zeroes_sectors << 9); } +static ssize_t queue_zone_write_granularity_show(struct request_queue *q, + char *page) +{ + return queue_var_show(queue_zone_write_granularity(q), page); +} + static ssize_t queue_zone_append_max_show(struct request_queue *q, char *page) { unsigned long long max_sectors = q->limits.max_zone_append_sectors; @@ -585,6 +591,7 @@ QUEUE_RO_ENTRY(queue_discard_zeroes_data, "discard_zeroes_data"); QUEUE_RO_ENTRY(queue_write_same_max, "write_same_max_bytes"); QUEUE_RO_ENTRY(queue_write_zeroes_max, "write_zeroes_max_bytes"); QUEUE_RO_ENTRY(queue_zone_append_max, "zone_append_max_bytes"); +QUEUE_RO_ENTRY(queue_zone_write_granularity, "zone_write_granularity"); QUEUE_RO_ENTRY(queue_zoned, "zoned"); QUEUE_RO_ENTRY(queue_nr_zones, "nr_zones"); @@ -639,6 +646,7 @@ static struct attribute *queue_attrs[] = { &queue_write_same_max_entry.attr, &queue_write_zeroes_max_entry.attr, &queue_zone_append_max_entry.attr, + &queue_zone_write_granularity_entry.attr, &queue_nonrot_entry.attr, &queue_zoned_entry.attr, &queue_nr_zones_entry.attr, diff --git a/drivers/block/null_blk/zoned.c b/drivers/block/null_blk/zoned.c index 535351570bb2..704d09481e0d 100644 --- a/drivers/block/null_blk/zoned.c +++ b/drivers/block/null_blk/zoned.c @@ -172,6 +172,7 @@ int null_register_zoned_dev(struct nullb *nullb) blk_queue_max_zone_append_sectors(q, dev->zone_size_sects); blk_queue_max_open_zones(q, dev->zone_max_open); blk_queue_max_active_zones(q, dev->zone_max_active); + blk_queue_zone_write_granularity(q, dev->blocksize); return 0; } diff --git a/drivers/nvme/host/zns.c b/drivers/nvme/host/zns.c index 1dfe9a3500e3..f25311ccd996 100644 --- a/drivers/nvme/host/zns.c +++ b/drivers/nvme/host/zns.c @@ -113,6 +113,7 @@ int nvme_update_zone_info(struct nvme_ns *ns, unsigned lbaf) blk_queue_flag_set(QUEUE_FLAG_ZONE_RESETALL, q); blk_queue_max_open_zones(q, le32_to_cpu(id->mor) + 1); blk_queue_max_active_zones(q, le32_to_cpu(id->mar) + 1); + blk_queue_zone_write_granularity(q, q->limits.logical_block_size); free_data: kfree(id); return status; diff --git a/drivers/scsi/sd_zbc.c b/drivers/scsi/sd_zbc.c index cf07b7f93579..10e9f33cc069 100644 --- a/drivers/scsi/sd_zbc.c +++ b/drivers/scsi/sd_zbc.c @@ -789,6 +789,16 @@ int sd_zbc_read_zones(struct scsi_disk *sdkp, unsigned char *buf) blk_queue_max_active_zones(q, 0); nr_zones = round_up(sdkp->capacity, zone_blocks) >> ilog2(zone_blocks); + /* + * Per ZBC and ZAC specifications, writes in sequential write required + * zones of host-managed devices must be aligned to the device physical + * block size. + */ + if (blk_queue_zoned_model(q) == BLK_ZONED_HM) + blk_queue_zone_write_granularity(q, sdkp->physical_block_size); + else if (blk_queue_zoned_model(q) == BLK_ZONED_HA) + blk_queue_zone_write_granularity(q, sdkp->device->sector_size); + /* READ16/WRITE16 is mandatory for ZBC disks */ sdkp->device->use_16_for_rw = 1; sdkp->device->use_10_for_rw = 0; diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h index f94ee3089e01..142e3b34be75 100644 --- a/include/linux/blkdev.h +++ b/include/linux/blkdev.h @@ -337,6 +337,7 @@ struct queue_limits { unsigned int max_zone_append_sectors; unsigned int discard_granularity; unsigned int discard_alignment; + unsigned int zone_write_granularity; unsigned short max_segments; unsigned short max_integrity_segments; @@ -1161,6 +1162,8 @@ extern void blk_queue_logical_block_size(struct request_queue *, unsigned int); extern void blk_queue_max_zone_append_sectors(struct request_queue *q, unsigned int max_zone_append_sectors); extern void blk_queue_physical_block_size(struct request_queue *, unsigned int); +void blk_queue_zone_write_granularity(struct request_queue *q, + unsigned int size); extern void blk_queue_alignment_offset(struct request_queue *q, unsigned int alignment); void blk_queue_update_readahead(struct request_queue *q); @@ -1474,6 +1477,18 @@ static inline int bdev_io_opt(struct block_device *bdev) return queue_io_opt(bdev_get_queue(bdev)); } +static inline unsigned int +queue_zone_write_granularity(const struct request_queue *q) +{ + return q->limits.zone_write_granularity; +} + +static inline unsigned int +bdev_zone_write_granularity(struct block_device *bdev) +{ + return queue_zone_write_granularity(bdev_get_queue(bdev)); +} + static inline int queue_alignment_offset(const struct request_queue *q) { if (q->limits.misaligned)
Per ZBC and ZAC specifications, host-managed SMR hard-disks mandate that all writes into sequential write required zones be aligned to the device physical block size. However, NVMe ZNS does not have this constraint and allows write operations into sequential zones to be logical block size aligned. This inconsistency does not help with portability of software across device types. To solve this, introduce the zone_write_granularity queue limit to indicate the alignment constraint, in bytes, of write operations into zones of a zoned block device. This new limit is exported as a read-only sysfs queue attribute and the helper blk_queue_zone_write_granularity() introduced for drivers to set this limit. The scsi disk driver is modified to use this helper to set host-managed SMR disk zone write granularity to the disk physical block size. The ZNS support code of the NVMe driver is also modified to use this helper to set the new limit to the logical block size of the namespace. The nullblk driver is similarly modified too. The accessor functions queue_zone_write_granularity() and bdev_zone_write_granularity() are also introduced. Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> --- Documentation/block/queue-sysfs.rst | 7 ++++++ block/blk-settings.c | 39 +++++++++++++++++++++++++---- block/blk-sysfs.c | 8 ++++++ drivers/block/null_blk/zoned.c | 1 + drivers/nvme/host/zns.c | 1 + drivers/scsi/sd_zbc.c | 10 ++++++++ include/linux/blkdev.h | 15 +++++++++++ 7 files changed, 76 insertions(+), 5 deletions(-)