block: pre-calculate max_zone_append_sectors

max_zone_append_sectors differs from all other queue limits in that the
final value used is not stored in the queue_limits but needs to be
obtained using queue_limits_max_zone_append_sectors helper.  This not
only adds (tiny) extra overhead to the I/O path, but also can be easily
forgotten in file system code.

Add a new max_hw_zone_append_sectors value to queue_limits which is
set by the driver, and calculate max_zone_append_sectors from that and
the other inputs in blk_validate_zoned_limits, similar to how
max_sectors is calculated to fix this.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20241104073955.112324-3-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
This commit is contained in:
Christoph Hellwig 2024-11-04 08:39:32 +01:00 committed by Jens Axboe
parent e494c3dce6
commit 2a8f6153e1
12 changed files with 27 additions and 57 deletions

View File

@ -607,7 +607,7 @@ static inline blk_status_t blk_check_zone_append(struct request_queue *q,
return BLK_STS_IOERR;
/* Make sure the BIO is small enough and will not get split */
if (nr_sectors > queue_max_zone_append_sectors(q))
if (nr_sectors > q->limits.max_zone_append_sectors)
return BLK_STS_IOERR;
bio->bi_opf |= REQ_NOMERGE;

View File

@ -388,11 +388,10 @@ struct bio *bio_split_rw(struct bio *bio, const struct queue_limits *lim,
struct bio *bio_split_zone_append(struct bio *bio,
const struct queue_limits *lim, unsigned *nr_segs)
{
unsigned int max_sectors = queue_limits_max_zone_append_sectors(lim);
int split_sectors;
split_sectors = bio_split_rw_at(bio, lim, nr_segs,
max_sectors << SECTOR_SHIFT);
lim->max_zone_append_sectors << SECTOR_SHIFT);
if (WARN_ON_ONCE(split_sectors > 0))
split_sectors = -EINVAL;
return bio_submit_split(bio, split_sectors);

View File

@ -91,17 +91,16 @@ static int blk_validate_zoned_limits(struct queue_limits *lim)
if (lim->zone_write_granularity < lim->logical_block_size)
lim->zone_write_granularity = lim->logical_block_size;
if (lim->max_zone_append_sectors) {
/*
* The Zone Append size is limited by the maximum I/O size
* and the zone size given that it can't span zones.
*/
lim->max_zone_append_sectors =
min3(lim->max_hw_sectors,
lim->max_zone_append_sectors,
lim->chunk_sectors);
}
/*
* The Zone Append size is limited by the maximum I/O size and the zone
* size given that it can't span zones.
*
* If no max_hw_zone_append_sectors limit is provided, the block layer
* will emulated it, else we're also bound by the hardware limit.
*/
lim->max_zone_append_sectors =
min_not_zero(lim->max_hw_zone_append_sectors,
min(lim->chunk_sectors, lim->max_hw_sectors));
return 0;
}
@ -527,8 +526,8 @@ int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
t->max_dev_sectors = min_not_zero(t->max_dev_sectors, b->max_dev_sectors);
t->max_write_zeroes_sectors = min(t->max_write_zeroes_sectors,
b->max_write_zeroes_sectors);
t->max_zone_append_sectors = min(queue_limits_max_zone_append_sectors(t),
queue_limits_max_zone_append_sectors(b));
t->max_hw_zone_append_sectors = min(t->max_hw_zone_append_sectors,
b->max_hw_zone_append_sectors);
t->seg_boundary_mask = min_not_zero(t->seg_boundary_mask,
b->seg_boundary_mask);

View File

@ -131,6 +131,7 @@ QUEUE_SYSFS_LIMIT_SHOW_SECTORS_TO_BYTES(max_hw_discard_sectors)
QUEUE_SYSFS_LIMIT_SHOW_SECTORS_TO_BYTES(max_write_zeroes_sectors)
QUEUE_SYSFS_LIMIT_SHOW_SECTORS_TO_BYTES(atomic_write_max_sectors)
QUEUE_SYSFS_LIMIT_SHOW_SECTORS_TO_BYTES(atomic_write_boundary_sectors)
QUEUE_SYSFS_LIMIT_SHOW_SECTORS_TO_BYTES(max_zone_append_sectors)
#define QUEUE_SYSFS_LIMIT_SHOW_SECTORS_TO_KB(_field) \
static ssize_t queue_##_field##_show(struct gendisk *disk, char *page) \
@ -178,18 +179,6 @@ static ssize_t queue_max_discard_sectors_store(struct gendisk *disk,
return ret;
}
/*
* For zone append queue_max_zone_append_sectors does not just return the
* underlying queue limits, but actually contains a calculation. Because of
* that we can't simply use QUEUE_SYSFS_LIMIT_SHOW_SECTORS_TO_BYTES here.
*/
static ssize_t queue_zone_append_max_show(struct gendisk *disk, char *page)
{
return sprintf(page, "%llu\n",
(u64)queue_max_zone_append_sectors(disk->queue) <<
SECTOR_SHIFT);
}
static ssize_t
queue_max_sectors_store(struct gendisk *disk, const char *page, size_t count)
{
@ -479,7 +468,7 @@ QUEUE_RO_ENTRY(queue_atomic_write_unit_min, "atomic_write_unit_min_bytes");
QUEUE_RO_ENTRY(queue_write_same_max, "write_same_max_bytes");
QUEUE_RO_ENTRY(queue_max_write_zeroes_sectors, "write_zeroes_max_bytes");
QUEUE_RO_ENTRY(queue_zone_append_max, "zone_append_max_bytes");
QUEUE_RO_ENTRY(queue_max_zone_append_sectors, "zone_append_max_bytes");
QUEUE_RO_ENTRY(queue_zone_write_granularity, "zone_write_granularity");
QUEUE_RO_ENTRY(queue_zoned, "zoned");
@ -607,7 +596,7 @@ static struct attribute *queue_attrs[] = {
&queue_atomic_write_unit_max_entry.attr,
&queue_write_same_max_entry.attr,
&queue_max_write_zeroes_sectors_entry.attr,
&queue_zone_append_max_entry.attr,
&queue_max_zone_append_sectors_entry.attr,
&queue_zone_write_granularity_entry.attr,
&queue_rotational_entry.attr,
&queue_zoned_entry.attr,

View File

@ -166,7 +166,7 @@ int null_init_zoned_dev(struct nullb_device *dev,
lim->features |= BLK_FEAT_ZONED;
lim->chunk_sectors = dev->zone_size_sects;
lim->max_zone_append_sectors = dev->zone_append_max_sectors;
lim->max_hw_zone_append_sectors = dev->zone_append_max_sectors;
lim->max_open_zones = dev->zone_max_open;
lim->max_active_zones = dev->zone_max_active;
return 0;

View File

@ -2270,7 +2270,7 @@ static int ublk_ctrl_start_dev(struct ublk_device *ub, struct io_uring_cmd *cmd)
lim.features |= BLK_FEAT_ZONED;
lim.max_active_zones = p->max_active_zones;
lim.max_open_zones = p->max_open_zones;
lim.max_zone_append_sectors = p->max_zone_append_sectors;
lim.max_hw_zone_append_sectors = p->max_zone_append_sectors;
}
if (ub->params.basic.attrs & UBLK_ATTR_VOLATILE_CACHE) {

View File

@ -784,7 +784,7 @@ static int virtblk_read_zoned_limits(struct virtio_blk *vblk,
wg, v);
return -ENODEV;
}
lim->max_zone_append_sectors = v;
lim->max_hw_zone_append_sectors = v;
dev_dbg(&vdev->dev, "max append sectors = %u\n", v);
return 0;

View File

@ -344,7 +344,7 @@ int dm_set_zones_restrictions(struct dm_table *t, struct request_queue *q,
clear_bit(DMF_EMULATE_ZONE_APPEND, &md->flags);
} else {
set_bit(DMF_EMULATE_ZONE_APPEND, &md->flags);
lim->max_zone_append_sectors = 0;
lim->max_hw_zone_append_sectors = 0;
}
/*
@ -379,7 +379,7 @@ int dm_set_zones_restrictions(struct dm_table *t, struct request_queue *q,
if (!zlim.mapped_nr_seq_zones) {
lim->max_open_zones = 0;
lim->max_active_zones = 0;
lim->max_zone_append_sectors = 0;
lim->max_hw_zone_append_sectors = 0;
lim->zone_write_granularity = 0;
lim->chunk_sectors = 0;
lim->features &= ~BLK_FEAT_ZONED;

View File

@ -636,7 +636,7 @@ int nvme_mpath_alloc_disk(struct nvme_ctrl *ctrl, struct nvme_ns_head *head)
if (head->ids.csi == NVME_CSI_ZNS)
lim.features |= BLK_FEAT_ZONED;
else
lim.max_zone_append_sectors = 0;
lim.max_hw_zone_append_sectors = 0;
head->disk = blk_alloc_disk(&lim, ctrl->numa_node);
if (IS_ERR(head->disk))

View File

@ -111,7 +111,7 @@ void nvme_update_zone_info(struct nvme_ns *ns, struct queue_limits *lim,
lim->features |= BLK_FEAT_ZONED;
lim->max_open_zones = zi->max_open_zones;
lim->max_active_zones = zi->max_active_zones;
lim->max_zone_append_sectors = ns->ctrl->max_zone_append;
lim->max_hw_zone_append_sectors = ns->ctrl->max_zone_append;
lim->chunk_sectors = ns->head->zsze =
nvme_lba_to_sect(ns->head, zi->zone_size);
}

View File

@ -634,8 +634,6 @@ int sd_zbc_read_zones(struct scsi_disk *sdkp, struct queue_limits *lim,
lim->max_open_zones = sdkp->zones_max_open;
lim->max_active_zones = 0;
lim->chunk_sectors = logical_to_sectors(sdkp->device, zone_blocks);
/* Enable block layer zone append emulation */
lim->max_zone_append_sectors = 0;
return 0;

View File

@ -375,6 +375,7 @@ struct queue_limits {
unsigned int max_user_discard_sectors;
unsigned int max_secure_erase_sectors;
unsigned int max_write_zeroes_sectors;
unsigned int max_hw_zone_append_sectors;
unsigned int max_zone_append_sectors;
unsigned int discard_granularity;
unsigned int discard_alignment;
@ -1204,25 +1205,9 @@ static inline unsigned int queue_max_segment_size(const struct request_queue *q)
return q->limits.max_segment_size;
}
static inline unsigned int
queue_limits_max_zone_append_sectors(const struct queue_limits *l)
{
unsigned int max_sectors = min(l->chunk_sectors, l->max_hw_sectors);
return min_not_zero(l->max_zone_append_sectors, max_sectors);
}
static inline unsigned int queue_max_zone_append_sectors(struct request_queue *q)
{
if (!blk_queue_is_zoned(q))
return 0;
return queue_limits_max_zone_append_sectors(&q->limits);
}
static inline bool queue_emulates_zone_append(struct request_queue *q)
{
return blk_queue_is_zoned(q) && !q->limits.max_zone_append_sectors;
return blk_queue_is_zoned(q) && !q->limits.max_hw_zone_append_sectors;
}
static inline bool bdev_emulates_zone_append(struct block_device *bdev)
@ -1233,7 +1218,7 @@ static inline bool bdev_emulates_zone_append(struct block_device *bdev)
static inline unsigned int
bdev_max_zone_append_sectors(struct block_device *bdev)
{
return queue_max_zone_append_sectors(bdev_get_queue(bdev));
return bdev_limits(bdev)->max_zone_append_sectors;
}
static inline unsigned int bdev_max_segments(struct block_device *bdev)