block: move the nowait flag to queue_limits

Move the nowait flag into the queue_limits feature field so that it can
be set atomically with the queue frozen.

Stacking drivers are simplified in that they now can simply set the
flag, and blk_stack_limits will clear it when the features is not
supported by any of the underlying devices.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Link: https://lore.kernel.org/r/20240617060532.127975-20-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block/blk-mq-debugfs.c

@@ -96,7 +96,6 @@ static const char *const blk_queue_flag_name[] = {
 	QUEUE_FLAG_NAME(ZONE_RESETALL),
 	QUEUE_FLAG_NAME(RQ_ALLOC_TIME),
 	QUEUE_FLAG_NAME(HCTX_ACTIVE),
-	QUEUE_FLAG_NAME(NOWAIT),
 	QUEUE_FLAG_NAME(SQ_SCHED),
 	QUEUE_FLAG_NAME(SKIP_TAGSET_QUIESCE),
 };

block/blk-mq.c

@@ -4118,7 +4118,7 @@ struct request_queue *blk_mq_alloc_queue(struct blk_mq_tag_set *set,
 
 	if (!lim)
 		lim = &default_lim;
-	lim->features |= BLK_FEAT_IO_STAT;
+	lim->features |= BLK_FEAT_IO_STAT | BLK_FEAT_NOWAIT;
 
 	q = blk_alloc_queue(lim, set->numa_node);
 	if (IS_ERR(q))

block/blk-settings.c

@@ -459,6 +459,15 @@ int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
 
 	t->features |= (b->features & BLK_FEAT_INHERIT_MASK);
 
+	/*
+	 * BLK_FEAT_NOWAIT needs to be supported both by the stacking driver
+	 * and all underlying devices.  The stacking driver sets the flag
+	 * before stacking the limits, and this will clear the flag if any
+	 * of the underlying devices does not support it.
+	 */
+	if (!(b->features & BLK_FEAT_NOWAIT))
+		t->features &= ~BLK_FEAT_NOWAIT;
+
 	t->max_sectors = min_not_zero(t->max_sectors, b->max_sectors);
 	t->max_user_sectors = min_not_zero(t->max_user_sectors,
 			b->max_user_sectors);

drivers/block/brd.c

@@ -335,7 +335,8 @@ static int brd_alloc(int i)
 		.max_hw_discard_sectors	= UINT_MAX,
 		.max_discard_segments	= 1,
 		.discard_granularity	= PAGE_SIZE,
-		.features		= BLK_FEAT_SYNCHRONOUS,
+		.features		= BLK_FEAT_SYNCHRONOUS |
+					  BLK_FEAT_NOWAIT,
 	};
 
 	list_for_each_entry(brd, &brd_devices, brd_list)
@@ -367,7 +368,6 @@ static int brd_alloc(int i)
 	strscpy(disk->disk_name, buf, DISK_NAME_LEN);
 	set_capacity(disk, rd_size * 2);
 	
-	blk_queue_flag_set(QUEUE_FLAG_NOWAIT, disk->queue);
 	err = add_disk(disk);
 	if (err)
 		goto out_cleanup_disk;

drivers/md/dm-table.c

@@ -582,7 +582,7 @@ int dm_split_args(int *argc, char ***argvp, char *input)
 static void dm_set_stacking_limits(struct queue_limits *limits)
 {
 	blk_set_stacking_limits(limits);
-	limits->features |= BLK_FEAT_IO_STAT;
+	limits->features |= BLK_FEAT_IO_STAT | BLK_FEAT_NOWAIT;
 }
 
 /*
@@ -1746,12 +1746,6 @@ static bool dm_table_supports_write_zeroes(struct dm_table *t)
 	return true;
 }
 
-static int device_not_nowait_capable(struct dm_target *ti, struct dm_dev *dev,
-				     sector_t start, sector_t len, void *data)
-{
-	return !bdev_nowait(dev->bdev);
-}
-
 static bool dm_table_supports_nowait(struct dm_table *t)
 {
 	for (unsigned int i = 0; i < t->num_targets; i++) {
@@ -1759,10 +1753,6 @@ static bool dm_table_supports_nowait(struct dm_table *t)
 
 		if (!dm_target_supports_nowait(ti->type))
 			return false;
-
-		if (!ti->type->iterate_devices ||
-		    ti->type->iterate_devices(ti, device_not_nowait_capable, NULL))
-			return false;
 	}
 
 	return true;
@@ -1824,10 +1814,8 @@ int dm_table_set_restrictions(struct dm_table *t, struct request_queue *q,
 {
 	int r;
 
-	if (dm_table_supports_nowait(t))
-		blk_queue_flag_set(QUEUE_FLAG_NOWAIT, q);
-	else
-		blk_queue_flag_clear(QUEUE_FLAG_NOWAIT, q);
+	if (!dm_table_supports_nowait(t))
+		limits->features &= ~BLK_FEAT_NOWAIT;
 
 	if (!dm_table_supports_discards(t)) {
 		limits->max_hw_discard_sectors = 0;

drivers/md/md.c

@@ -5788,7 +5788,7 @@ struct mddev *md_alloc(dev_t dev, char *name)
 	int error;
 	struct queue_limits lim = {
 		.features		= BLK_FEAT_WRITE_CACHE | BLK_FEAT_FUA |
-					  BLK_FEAT_IO_STAT,
+					  BLK_FEAT_IO_STAT | BLK_FEAT_NOWAIT,
 	};
 
 	/*
@@ -6150,13 +6150,6 @@ int md_run(struct mddev *mddev)
 		}
 	}
 
-	if (!mddev_is_dm(mddev)) {
-		struct request_queue *q = mddev->gendisk->queue;
-
-		/* Set the NOWAIT flags if all underlying devices support it */
-		if (nowait)
-			blk_queue_flag_set(QUEUE_FLAG_NOWAIT, q);
-	}
 	if (pers->sync_request) {
 		if (mddev->kobj.sd &&
 		    sysfs_create_group(&mddev->kobj, &md_redundancy_group))
@@ -7115,15 +7108,6 @@ static int hot_add_disk(struct mddev *mddev, dev_t dev)
 	set_bit(MD_SB_CHANGE_DEVS, &mddev->sb_flags);
 	if (!mddev->thread)
 		md_update_sb(mddev, 1);
-	/*
-	 * If the new disk does not support REQ_NOWAIT,
-	 * disable on the whole MD.
-	 */
-	if (!bdev_nowait(rdev->bdev)) {
-		pr_info("%s: Disabling nowait because %pg does not support nowait\n",
-			mdname(mddev), rdev->bdev);
-		blk_queue_flag_clear(QUEUE_FLAG_NOWAIT, mddev->gendisk->queue);
-	}
 	/*
 	 * Kick recovery, maybe this spare has to be added to the
 	 * array immediately.

drivers/nvme/host/multipath.c

@@ -538,7 +538,7 @@ int nvme_mpath_alloc_disk(struct nvme_ctrl *ctrl, struct nvme_ns_head *head)
 
 	blk_set_stacking_limits(&lim);
 	lim.dma_alignment = 3;
-	lim.features |= BLK_FEAT_IO_STAT;
+	lim.features |= BLK_FEAT_IO_STAT | BLK_FEAT_NOWAIT;
 	if (head->ids.csi != NVME_CSI_ZNS)
 		lim.max_zone_append_sectors = 0;
 
@@ -550,7 +550,6 @@ int nvme_mpath_alloc_disk(struct nvme_ctrl *ctrl, struct nvme_ns_head *head)
 	sprintf(head->disk->disk_name, "nvme%dn%d",
 			ctrl->subsys->instance, head->instance);
 
-	blk_queue_flag_set(QUEUE_FLAG_NOWAIT, head->disk->queue);
 	/*
 	 * This assumes all controllers that refer to a namespace either
 	 * support poll queues or not.  That is not a strict guarantee,

include/linux/blkdev.h

@@ -304,6 +304,9 @@ enum {
 
 	/* always completes in submit context */
 	BLK_FEAT_SYNCHRONOUS			= (1u << 6),
+
+	/* supports REQ_NOWAIT */
+	BLK_FEAT_NOWAIT				= (1u << 7),
 };
 
 /*
@@ -580,12 +583,10 @@ struct request_queue {
 #define QUEUE_FLAG_ZONE_RESETALL 26	/* supports Zone Reset All */
 #define QUEUE_FLAG_RQ_ALLOC_TIME 27	/* record rq->alloc_time_ns */
 #define QUEUE_FLAG_HCTX_ACTIVE	28	/* at least one blk-mq hctx is active */
-#define QUEUE_FLAG_NOWAIT       29	/* device supports NOWAIT */
 #define QUEUE_FLAG_SQ_SCHED     30	/* single queue style io dispatch */
 #define QUEUE_FLAG_SKIP_TAGSET_QUIESCE	31 /* quiesce_tagset skip the queue*/
 
-#define QUEUE_FLAG_MQ_DEFAULT	((1UL << QUEUE_FLAG_SAME_COMP) |	\
-				 (1UL << QUEUE_FLAG_NOWAIT))
+#define QUEUE_FLAG_MQ_DEFAULT	(1UL << QUEUE_FLAG_SAME_COMP)
 
 void blk_queue_flag_set(unsigned int flag, struct request_queue *q);
 void blk_queue_flag_clear(unsigned int flag, struct request_queue *q);
@@ -1348,7 +1349,7 @@ static inline bool bdev_fua(struct block_device *bdev)
 
 static inline bool bdev_nowait(struct block_device *bdev)
 {
-	return test_bit(QUEUE_FLAG_NOWAIT, &bdev_get_queue(bdev)->queue_flags);
+	return bdev->bd_disk->queue->limits.features & BLK_FEAT_NOWAIT;
 }
 
 static inline bool bdev_is_zoned(struct block_device *bdev)