Revert "block: pre-calculate max_zone_append_sectors"

This causes issue on, at least, nvme-mpath where my boot fails with:

WARNING: CPU: 354 PID: 2729 at block/blk-settings.c:75 blk_validate_limits+0x356/0x380
Modules linked in: tg3(+) nvme usbcore scsi_mod ptp i2c_piix4 libphy nvme_core crc32c_intel scsi_common usb_common pps_core i2c_smbus
CPU: 354 UID: 0 PID: 2729 Comm: kworker/u2061:1 Not tainted 6.12.0-rc6+ torvalds#181
Hardware name: Dell Inc. PowerEdge R7625/06444F, BIOS 1.8.3 04/02/2024
Workqueue: async async_run_entry_fn
RIP: 0010:blk_validate_limits+0x356/0x380
Code: f6 47 01 04 75 28 83 bf 94 00 00 00 00 75 39 83 bf 98 00 00 00 00 75 34 83 7f 68 00 75 32 31 c0 83 7f 5c 00 0f 84 9b fd ff ff <0f> 0b eb 13 0f 0b eb 0f 48 c7 c0 74 12 58 92 48 89 c7 e8 13 76 46
RSP: 0018:ffffa8a1dfb93b30 EFLAGS: 00010286
RAX: 0000000000000000 RBX: ffff9232829c8388 RCX: 0000000000000088
RDX: 0000000000000080 RSI: 0000000000000200 RDI: ffffa8a1dfb93c38
RBP: 000000000000000c R08: 00000000ffffffff R09: 000000000000ffff
R10: 0000000000000000 R11: 0000000000000000 R12: ffff9232829b9000
R13: ffff9232829b9010 R14: ffffa8a1dfb93c38 R15: ffffa8a1dfb93c38
FS:  0000000000000000(0000) GS:ffff923867c80000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 000055c1b92480a8 CR3: 0000002484ff0002 CR4: 0000000000370ef0
Call Trace:
 <TASK>
 ? __warn+0xca/0x1a0
 ? blk_validate_limits+0x356/0x380
 ? report_bug+0x11a/0x1a0
 ? handle_bug+0x5e/0x90
 ? exc_invalid_op+0x16/0x40
 ? asm_exc_invalid_op+0x16/0x20
 ? blk_validate_limits+0x356/0x380
 blk_alloc_queue+0x7a/0x250
 __blk_alloc_disk+0x39/0x80
 nvme_mpath_alloc_disk+0x13d/0x1b0 [nvme_core]
 nvme_scan_ns+0xcc7/0x1010 [nvme_core]
 async_run_entry_fn+0x27/0x120
 process_scheduled_works+0x1a0/0x360
 worker_thread+0x2bc/0x350
 ? pr_cont_work+0x1b0/0x1b0
 kthread+0x111/0x120
 ? kthread_unuse_mm+0x90/0x90
 ret_from_fork+0x30/0x40
 ? kthread_unuse_mm+0x90/0x90
 ret_from_fork_asm+0x11/0x20
 </TASK>
---[ end trace 0000000000000000 ]---

presumably due to max_zone_append_sectors not being cleared to zero,
resulting in blk_validate_zoned_limits() complaining and failing.

This reverts commit 2a8f615.

Signed-off-by: Jens Axboe <[email protected]>

Author: axboe

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 > q->limits.max_zone_append_sectors)
+	if (nr_sectors > queue_max_zone_append_sectors(q))
 		return BLK_STS_IOERR;
 
 	bio->bi_opf |= REQ_NOMERGE;

block/blk-merge.c

@@ -388,10 +388,11 @@ 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,
-			lim->max_zone_append_sectors << SECTOR_SHIFT);
+			max_sectors << SECTOR_SHIFT);
 	if (WARN_ON_ONCE(split_sectors > 0))
 		split_sectors = -EINVAL;
 	return bio_submit_split(bio, split_sectors);

block/blk-settings.c

@@ -91,16 +91,17 @@ 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;
 
-	/*
-	 * 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));
+	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);
+	}
+
 	return 0;
 }
 
@@ -526,8 +527,8 @@ 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_hw_zone_append_sectors = min(t->max_hw_zone_append_sectors,
-					b->max_hw_zone_append_sectors);
+	t->max_zone_append_sectors = min(queue_limits_max_zone_append_sectors(t),
+					 queue_limits_max_zone_append_sectors(b));
 
 	t->seg_boundary_mask = min_not_zero(t->seg_boundary_mask,
 					    b->seg_boundary_mask);

block/blk-sysfs.c

@@ -131,7 +131,6 @@ 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)	\
@@ -179,6 +178,18 @@ 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)
 {
@@ -468,7 +479,7 @@ 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_max_zone_append_sectors, "zone_append_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");
@@ -596,7 +607,7 @@ 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_max_zone_append_sectors_entry.attr,
+	&queue_zone_append_max_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_hw_zone_append_sectors = dev->zone_append_max_sectors;
+	lim->max_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

@@ -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_hw_zone_append_sectors = p->max_zone_append_sectors;
+		lim.max_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_hw_zone_append_sectors = v;
+	lim->max_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_hw_zone_append_sectors = 0;
+		lim->max_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_hw_zone_append_sectors = 0;
+		lim->max_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_hw_zone_append_sectors = 0;
+		lim.max_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_hw_zone_append_sectors = ns->ctrl->max_zone_append;
+	lim->max_zone_append_sectors = ns->ctrl->max_zone_append;
 	lim->chunk_sectors = ns->head->zsze =
 		nvme_lba_to_sect(ns->head, zi->zone_size);
 }

drivers/scsi/sd_zbc.c

@@ -634,6 +634,8 @@ 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;
 

include/linux/blkdev.h

@@ -375,7 +375,6 @@ 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;
@@ -1205,9 +1204,25 @@ 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_hw_zone_append_sectors;
+	return blk_queue_is_zoned(q) && !q->limits.max_zone_append_sectors;
 }
 
 static inline bool bdev_emulates_zone_append(struct block_device *bdev)
@@ -1218,7 +1233,7 @@ 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 bdev_limits(bdev)->max_zone_append_sectors;
+	return queue_max_zone_append_sectors(bdev_get_queue(bdev));
 }
 
 static inline unsigned int bdev_max_segments(struct block_device *bdev)