Merge tag 'for-6.14/dm-changes' of git://git.kernel.org/pub/scm/linux/kernel/git/device-mapper/linux-dm

Pull device mapper updates from Mikulas Patocka:

- fix a spelling error in dm-raid

- change kzalloc to kcalloc

- remove useless test in alloc_multiple_bios

- disable REQ_NOWAIT for flushes

- dm-transaction-manager: use red-black trees instead of linear lists

- atomic writes support for dm-linear, dm-stripe and dm-mirror

- dm-crypt: code cleanups and two bugfixes

* tag 'for-6.14/dm-changes' of git://git.kernel.org/pub/scm/linux/kernel/git/device-mapper/linux-dm:
dm-crypt: track tag_offset in convert_context
dm-crypt: don't initialize cc_sector again
dm-crypt: don't update io->sector after kcryptd_crypt_write_io_submit()
dm-crypt: use bi_sector in bio when initialize integrity seed
dm-crypt: fully initialize clone->bi_iter in crypt_alloc_buffer()
dm-crypt: set atomic as false when calling crypt_convert() in kworker
dm-mirror: Support atomic writes
dm-io: Warn on creating multiple atomic write bios for a region
dm-stripe: Enable atomic writes
dm-linear: Enable atomic writes
dm: Ensure cloned bio is same length for atomic write
dm-table: atomic writes support
dm-transaction-manager: use red-black trees instead of linear lists
dm: disable REQ_NOWAIT for flushes
dm: remove useless test in alloc_multiple_bios
dm: change kzalloc to kcalloc
dm raid: fix spelling errors in raid_ctr()

drivers/md/dm-crypt.c

···
 	struct bio *bio_out;
 	struct bvec_iter iter_out;
 	atomic_t cc_pending;
+	unsigned int tag_offset;
 	u64 cc_sector;
 	union {
 		struct skcipher_request *req;
···
 
 	tag_len = io->cc->tuple_size * (bio_sectors(bio) >> io->cc->sector_shift);
 
-	bip->bip_iter.bi_sector = io->cc->start + io->sector;
+	bip->bip_iter.bi_sector = bio->bi_iter.bi_sector;
 
 	ret = bio_integrity_add_page(bio, virt_to_page(io->integrity_metadata),
 				     tag_len, offset_in_page(io->integrity_metadata));
···
 	if (bio_out)
 		ctx->iter_out = bio_out->bi_iter;
 	ctx->cc_sector = sector + cc->iv_offset;
+	ctx->tag_offset = 0;
 	init_completion(&ctx->restart);
 }
 
···
 static blk_status_t crypt_convert(struct crypt_config *cc,
 			 struct convert_context *ctx, bool atomic, bool reset_pending)
 {
-	unsigned int tag_offset = 0;
 	unsigned int sector_step = cc->sector_size >> SECTOR_SHIFT;
 	int r;
 
···
 		atomic_inc(&ctx->cc_pending);
 
 		if (crypt_integrity_aead(cc))
-			r = crypt_convert_block_aead(cc, ctx, ctx->r.req_aead, tag_offset);
+			r = crypt_convert_block_aead(cc, ctx, ctx->r.req_aead, ctx->tag_offset);
 		else
-			r = crypt_convert_block_skcipher(cc, ctx, ctx->r.req, tag_offset);
+			r = crypt_convert_block_skcipher(cc, ctx, ctx->r.req, ctx->tag_offset);
 
 		switch (r) {
 		/*
···
 					 * exit and continue processing in a workqueue
 					 */
 					ctx->r.req = NULL;
+					ctx->tag_offset++;
 					ctx->cc_sector += sector_step;
-					tag_offset++;
 					return BLK_STS_DEV_RESOURCE;
 				}
 			} else {
···
 		 */
 		case -EINPROGRESS:
 			ctx->r.req = NULL;
+			ctx->tag_offset++;
 			ctx->cc_sector += sector_step;
-			tag_offset++;
 			continue;
 		/*
 		 * The request was already processed (synchronously).
···
 		case 0:
 			atomic_dec(&ctx->cc_pending);
 			ctx->cc_sector += sector_step;
-			tag_offset++;
+			ctx->tag_offset++;
 			if (!atomic)
 				cond_resched();
 			continue;
···
 	clone->bi_private = io;
 	clone->bi_end_io = crypt_endio;
 	clone->bi_ioprio = io->base_bio->bi_ioprio;
+	clone->bi_iter.bi_sector = cc->start + io->sector;
 
 	remaining_size = size;
 
···
 			crypt_dec_pending(io);
 			return 1;
 		}
-		clone->bi_iter.bi_sector = cc->start + io->sector;
 		crypt_convert_init(cc, &io->ctx, clone, clone, io->sector);
 		io->saved_bi_iter = clone->bi_iter;
 		dm_submit_bio_remap(io->base_bio, clone);
···
 	clone = bio_alloc_clone(cc->dev->bdev, io->base_bio, gfp, &cc->bs);
 	if (!clone)
 		return 1;
+
+	clone->bi_iter.bi_sector = cc->start + io->sector;
 	clone->bi_private = io;
 	clone->bi_end_io = crypt_endio;
 
 	crypt_inc_pending(io);
-
-	clone->bi_iter.bi_sector = cc->start + io->sector;
 
 	if (dm_crypt_integrity_io_alloc(io, clone)) {
 		crypt_dec_pending(io);
···
 	/* crypt_convert should have filled the clone bio */
 	BUG_ON(io->ctx.iter_out.bi_size);
 
-	clone->bi_iter.bi_sector = cc->start + io->sector;
-
 	if ((likely(!async) && test_bit(DM_CRYPT_NO_OFFLOAD, &cc->flags)) ||
 	    test_bit(DM_CRYPT_NO_WRITE_WORKQUEUE, &cc->flags)) {
 		dm_submit_bio_remap(io->base_bio, clone);
···
 	struct crypt_config *cc = io->cc;
 	struct convert_context *ctx = &io->ctx;
 	int crypt_finished;
-	sector_t sector = io->sector;
 	blk_status_t r;
 
 	wait_for_completion(&ctx->restart);
 	reinit_completion(&ctx->restart);
 
-	r = crypt_convert(cc, &io->ctx, true, false);
+	r = crypt_convert(cc, &io->ctx, false, false);
 	if (r)
 		io->error = r;
 	crypt_finished = atomic_dec_and_test(&ctx->cc_pending);
···
 	}
 
 	/* Encryption was already finished, submit io now */
-	if (crypt_finished) {
+	if (crypt_finished)
 		kcryptd_crypt_write_io_submit(io, 0);
-		io->sector = sector;
-	}
 
 	crypt_dec_pending(io);
 }
···
 	struct convert_context *ctx = &io->ctx;
 	struct bio *clone;
 	int crypt_finished;
-	sector_t sector = io->sector;
 	blk_status_t r;
 
 	/*
 	 * Prevent io from disappearing until this function completes.
 	 */
 	crypt_inc_pending(io);
-	crypt_convert_init(cc, ctx, NULL, io->base_bio, sector);
+	crypt_convert_init(cc, ctx, NULL, io->base_bio, io->sector);
 
 	clone = crypt_alloc_buffer(io, io->base_bio->bi_iter.bi_size);
 	if (unlikely(!clone)) {
···
 		io->ctx.bio_in = clone;
 		io->ctx.iter_in = clone->bi_iter;
 	}
-
-	sector += bio_sectors(clone);
 
 	crypt_inc_pending(io);
 	r = crypt_convert(cc, ctx,
···
 	}
 
 	/* Encryption was already finished, submit io now */
-	if (crypt_finished) {
+	if (crypt_finished)
 		kcryptd_crypt_write_io_submit(io, 0);
-		io->sector = sector;
-	}
 
 dec:
 	crypt_dec_pending(io);
···
 	wait_for_completion(&io->ctx.restart);
 	reinit_completion(&io->ctx.restart);
 
-	r = crypt_convert(cc, &io->ctx, true, false);
+	r = crypt_convert(cc, &io->ctx, false, false);
 	if (r)
 		io->error = r;
 
···
 	crypt_inc_pending(io);
 
 	if (io->ctx.aead_recheck) {
-		io->ctx.cc_sector = io->sector + cc->iv_offset;
 		r = crypt_convert(cc, &io->ctx,
 				  test_bit(DM_CRYPT_NO_READ_WORKQUEUE, &cc->flags), true);
 	} else {

drivers/md/dm-io.c

···
 
 		atomic_inc(&io->count);
 		submit_bio(bio);
+		WARN_ON_ONCE(opf & REQ_ATOMIC && remaining);
 	} while (remaining);
 }
 

drivers/md/dm-linear.c

···
 
 static struct target_type linear_target = {
 	.name   = "linear",
-	.version = {1, 4, 0},
+	.version = {1, 5, 0},
 	.features = DM_TARGET_PASSES_INTEGRITY | DM_TARGET_NOWAIT |
-		    DM_TARGET_ZONED_HM | DM_TARGET_PASSES_CRYPTO,
+		    DM_TARGET_ZONED_HM | DM_TARGET_PASSES_CRYPTO |
+		    DM_TARGET_ATOMIC_WRITES,
 	.report_zones = linear_report_zones,
 	.module = THIS_MODULE,
 	.ctr    = linear_ctr,

drivers/md/dm-ps-io-affinity.c

···
 	if (!s)
 		return -ENOMEM;
 
-	s->path_map = kzalloc(nr_cpu_ids * sizeof(struct path_info *),
+	s->path_map = kcalloc(nr_cpu_ids, sizeof(struct path_info *),
 			      GFP_KERNEL);
 	if (!s->path_map)
 		goto free_selector;

drivers/md/dm-raid.c

···
 		if (reshape_sectors || rs_is_raid1(rs)) {
 			/*
 			 * We can only prepare for a reshape here, because the
-			 * raid set needs to run to provide the repective reshape
+			 * raid set needs to run to provide the respective reshape
 			 * check functions via its MD personality instance.
 			 *
 			 * So do the reshape check after md_run() succeeded.

drivers/md/dm-raid1.c

···
 	unsigned int i;
 	struct dm_io_region io[MAX_NR_MIRRORS], *dest = io;
 	struct mirror *m;
-	blk_opf_t op_flags = bio->bi_opf & (REQ_FUA | REQ_PREFLUSH);
+	blk_opf_t op_flags = bio->bi_opf & (REQ_FUA | REQ_PREFLUSH | REQ_ATOMIC);
 	struct dm_io_request io_req = {
 		.bi_opf = REQ_OP_WRITE | op_flags,
 		.mem.type = DM_IO_BIO,
···
 
 static struct target_type mirror_target = {
 	.name	 = "mirror",
-	.version = {1, 14, 0},
+	.version = {1, 15, 0},
 	.module	 = THIS_MODULE,
+	.features = DM_TARGET_ATOMIC_WRITES,
 	.ctr	 = mirror_ctr,
 	.dtr	 = mirror_dtr,
 	.map	 = mirror_map,

drivers/md/dm-stripe.c

···
 
 static struct target_type stripe_target = {
 	.name   = "striped",
-	.version = {1, 6, 0},
-	.features = DM_TARGET_PASSES_INTEGRITY | DM_TARGET_NOWAIT,
+	.version = {1, 7, 0},
+	.features = DM_TARGET_PASSES_INTEGRITY | DM_TARGET_NOWAIT |
+		    DM_TARGET_ATOMIC_WRITES,
 	.module = THIS_MODULE,
 	.ctr    = stripe_ctr,
 	.dtr    = stripe_dtr,

drivers/md/dm-table.c

···
 	return true;
 }
 
+static int device_not_atomic_write_capable(struct dm_target *ti,
+			struct dm_dev *dev, sector_t start,
+			sector_t len, void *data)
+{
+	return !bdev_can_atomic_write(dev->bdev);
+}
+
+static bool dm_table_supports_atomic_writes(struct dm_table *t)
+{
+	for (unsigned int i = 0; i < t->num_targets; i++) {
+		struct dm_target *ti = dm_table_get_target(t, i);
+
+		if (!dm_target_supports_atomic_writes(ti->type))
+			return false;
+
+		if (!ti->type->iterate_devices)
+			return false;
+
+		if (ti->type->iterate_devices(ti,
+			device_not_atomic_write_capable, NULL)) {
+			return false;
+		}
+	}
+	return true;
+}
+
 int dm_table_set_restrictions(struct dm_table *t, struct request_queue *q,
 			      struct queue_limits *limits)
 {
···
 		if (r)
 			return r;
 	}
+
+	if (dm_table_supports_atomic_writes(t))
+		limits->features |= BLK_FEAT_ATOMIC_WRITES;
 
 	r = queue_limits_set(q, limits);
 	if (r)

drivers/md/dm.c

···
 
 static void alloc_multiple_bios(struct bio_list *blist, struct clone_info *ci,
 				struct dm_target *ti, unsigned int num_bios,
-				unsigned *len, gfp_t gfp_flag)
+				unsigned *len)
 {
 	struct bio *bio;
-	int try = (gfp_flag & GFP_NOWAIT) ? 0 : 1;
+	int try;
 
-	for (; try < 2; try++) {
+	for (try = 0; try < 2; try++) {
 		int bio_nr;
 
 		if (try && num_bios > 1)
···
 }
 
 static unsigned int __send_duplicate_bios(struct clone_info *ci, struct dm_target *ti,
-					  unsigned int num_bios, unsigned int *len,
-					  gfp_t gfp_flag)
+					  unsigned int num_bios, unsigned int *len)
 {
 	struct bio_list blist = BIO_EMPTY_LIST;
 	struct bio *clone;
···
 	 * Using alloc_multiple_bios(), even if num_bios is 1, to consistently
 	 * support allocating using GFP_NOWAIT with GFP_NOIO fallback.
 	 */
-	alloc_multiple_bios(&blist, ci, ti, num_bios, len, gfp_flag);
+	alloc_multiple_bios(&blist, ci, ti, num_bios, len);
 	while ((clone = bio_list_pop(&blist))) {
 		if (num_bios > 1)
 			dm_tio_set_flag(clone_to_tio(clone), DM_TIO_IS_DUPLICATE_BIO);
···
 
 			atomic_add(ti->num_flush_bios, &ci->io->io_count);
 			bios = __send_duplicate_bios(ci, ti, ti->num_flush_bios,
-						     NULL, GFP_NOWAIT);
+						     NULL);
 			atomic_sub(ti->num_flush_bios - bios, &ci->io->io_count);
 		}
 	} else {
···
 		    __max_io_len(ti, ci->sector, max_granularity, max_sectors));
 
 	atomic_add(num_bios, &ci->io->io_count);
-	bios = __send_duplicate_bios(ci, ti, num_bios, &len, GFP_NOIO);
+	bios = __send_duplicate_bios(ci, ti, num_bios, &len);
 	/*
 	 * alloc_io() takes one extra reference for submission, so the
 	 * reference won't reach 0 without the following (+1) subtraction
···
 	ci->submit_as_polled = !!(ci->bio->bi_opf & REQ_POLLED);
 
 	len = min_t(sector_t, max_io_len(ti, ci->sector), ci->sector_count);
+	if (ci->bio->bi_opf & REQ_ATOMIC && len != ci->sector_count)
+		return BLK_STS_IOERR;
+
 	setup_split_accounting(ci, len);
 
 	if (unlikely(ci->bio->bi_opf & REQ_NOWAIT)) {
···
 			 * not go crazy with the clone allocation.
 			 */
 			alloc_multiple_bios(&blist, ci, ti, min(nr_reset, 32),
-					    NULL, GFP_NOIO);
+					    NULL);
 		}
 
 		/* Get a clone and change it to a regular reset operation. */
···
 	unsigned int bios;
 
 	atomic_add(1, &ci->io->io_count);
-	bios = __send_duplicate_bios(ci, ti, 1, NULL, GFP_NOIO);
+	bios = __send_duplicate_bios(ci, ti, 1, NULL);
 	atomic_sub(1 - bios, &ci->io->io_count);
 
 	ci->sector_count = 0;
···
 
 	/* Only support nowait for normal IO */
 	if (unlikely(bio->bi_opf & REQ_NOWAIT) && !is_abnormal) {
+		/*
+		 * Don't support NOWAIT for FLUSH because it may allocate
+		 * multiple bios and there's no easy way how to undo the
+		 * allocations.
+		 */
+		if (bio->bi_opf & REQ_PREFLUSH) {
+			bio_wouldblock_error(bio);
+			return;
+		}
 		io = alloc_io(md, bio, GFP_NOWAIT);
 		if (unlikely(!io)) {
 			/* Unable to do anything without dm_io. */

drivers/md/persistent-data/dm-transaction-manager.c

···
 #include <linux/export.h>
 #include <linux/mutex.h>
 #include <linux/hash.h>
+#include <linux/rbtree.h>
 #include <linux/slab.h>
 #include <linux/device-mapper.h>
 
···
 /*----------------------------------------------------------------*/
 
 struct shadow_info {
-	struct hlist_node hlist;
+	struct rb_node node;
 	dm_block_t where;
 };
 
···
 	struct dm_space_map *sm;
 
 	spinlock_t lock;
-	struct hlist_head buckets[DM_HASH_SIZE];
+	struct rb_root buckets[DM_HASH_SIZE];
 
 	struct prefetch_set prefetches;
 };
···
 {
 	int r = 0;
 	unsigned int bucket = dm_hash_block(b, DM_HASH_MASK);
-	struct shadow_info *si;
+	struct rb_node **node;
 
 	spin_lock(&tm->lock);
-	hlist_for_each_entry(si, tm->buckets + bucket, hlist)
-		if (si->where == b) {
+	node = &tm->buckets[bucket].rb_node;
+	while (*node) {
+		struct shadow_info *si =
+			rb_entry(*node, struct shadow_info, node);
+		if (b == si->where) {
 			r = 1;
 			break;
 		}
+		if (b < si->where)
+			node = &si->node.rb_left;
+		else
+			node = &si->node.rb_right;
+	}
 	spin_unlock(&tm->lock);
 
 	return r;
···
 
 	si = kmalloc(sizeof(*si), GFP_NOIO);
 	if (si) {
+		struct rb_node **node, *parent;
 		si->where = b;
 		bucket = dm_hash_block(b, DM_HASH_MASK);
+
 		spin_lock(&tm->lock);
-		hlist_add_head(&si->hlist, tm->buckets + bucket);
+		node = &tm->buckets[bucket].rb_node;
+		parent = NULL;
+		while (*node) {
+			struct shadow_info *si =
+				rb_entry(*node, struct shadow_info, node);
+			parent = *node;
+			if (b < si->where)
+				node = &si->node.rb_left;
+			else
+				node = &si->node.rb_right;
+		}
+		rb_link_node(&si->node, parent, node);
+		rb_insert_color(&si->node, &tm->buckets[bucket]);
 		spin_unlock(&tm->lock);
 	}
 }
 
 static void wipe_shadow_table(struct dm_transaction_manager *tm)
 {
-	struct shadow_info *si;
-	struct hlist_node *tmp;
-	struct hlist_head *bucket;
-	int i;
+	unsigned int i;
 
 	spin_lock(&tm->lock);
 	for (i = 0; i < DM_HASH_SIZE; i++) {
-		bucket = tm->buckets + i;
-		hlist_for_each_entry_safe(si, tmp, bucket, hlist)
+		while (!RB_EMPTY_ROOT(&tm->buckets[i])) {
+			struct shadow_info *si =
+				rb_entry(tm->buckets[i].rb_node, struct shadow_info, node);
+			rb_erase(&si->node, &tm->buckets[i]);
 			kfree(si);
-
-		INIT_HLIST_HEAD(bucket);
+		}
 	}
-
 	spin_unlock(&tm->lock);
 }
 
···
 static struct dm_transaction_manager *dm_tm_create(struct dm_block_manager *bm,
 						   struct dm_space_map *sm)
 {
-	int i;
+	unsigned int i;
 	struct dm_transaction_manager *tm;
 
 	tm = kmalloc(sizeof(*tm), GFP_KERNEL);
···
 
 	spin_lock_init(&tm->lock);
 	for (i = 0; i < DM_HASH_SIZE; i++)
-		INIT_HLIST_HEAD(tm->buckets + i);
+		tm->buckets[i] = RB_ROOT;
 
 	prefetch_init(&tm->prefetches);
 

include/linux/device-mapper.h

···
 #define dm_target_supports_mixed_zoned_model(type) (false)
 #endif
 
+#define DM_TARGET_ATOMIC_WRITES		0x00000400
+#define dm_target_supports_atomic_writes(type) ((type)->features & DM_TARGET_ATOMIC_WRITES)
+
 struct dm_target {
 	struct dm_table *table;
 	struct target_type *type;

include/uapi/linux/dm-ioctl.h

···
 #define DM_DEV_SET_GEOMETRY	_IOWR(DM_IOCTL, DM_DEV_SET_GEOMETRY_CMD, struct dm_ioctl)
 
 #define DM_VERSION_MAJOR	4
-#define DM_VERSION_MINOR	48
+#define DM_VERSION_MINOR	49
 #define DM_VERSION_PATCHLEVEL	0
-#define DM_VERSION_EXTRA	"-ioctl (2023-03-01)"
+#define DM_VERSION_EXTRA	"-ioctl (2025-01-17)"
 
 /* Status bits */
 #define DM_READONLY_FLAG	(1 << 0) /* In/Out */