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
Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Link: https://lore.kernel.org/r/20241108154657.845768-2-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

Author: Christoph Hellwig

block/blk-core.c

@@ -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;

block/blk-merge.c

@@ -392,11 +392,10 @@ 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);

block/blk-settings.c

@@ -50,7 +50,7 @@ void blk_set_stacking_limits(struct queue_limits *lim)
 	lim->max_sectors = UINT_MAX;
 	lim->max_dev_sectors = UINT_MAX;
 	lim->max_write_zeroes_sectors = UINT_MAX;
-	lim->max_zone_append_sectors = UINT_MAX;
+	lim->max_hw_zone_append_sectors = UINT_MAX;
 	lim->max_user_discard_sectors = UINT_MAX;
 }
 EXPORT_SYMBOL(blk_set_stacking_limits);
@@ -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);

block/blk-sysfs.c

@@ -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,

drivers/block/null_blk/zoned.c

@@ -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;

drivers/block/ublk_drv.c

@@ -2279,7 +2279,7 @@ 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) {

drivers/block/virtio_blk.c

@@ -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;

drivers/md/dm-zone.c

@@ -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;

drivers/nvme/host/multipath.c

@@ -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))

drivers/nvme/host/zns.c

@@ -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);
 }