[PATCH AUTOSEL 5.4 026/330] dm table: do not allow request-based DM to stack on partitions

[PATCH AUTOSEL 5.4 026/330] dm table: do not allow request-based DM to stack on partitions

[Sasha Levin]

From: Mike Snitzer <snitzer@redhat.com>

[ Upstream commit 6ba01df72b4b63a26b4977790f58d8f775d2992c ]

Partitioned request-based devices cannot be used as underlying devices
for request-based DM because no partition offsets are added to each
incoming request.  As such, until now, stacking on partitioned devices
would _always_ result in data corruption (e.g. wiping the partition
table, writing to other partitions, etc).  Fix this by disallowing
request-based stacking on partitions.

While at it, since all .request_fn support has been removed from block
core, remove legacy dm-table code that differentiated between blk-mq and
.request_fn request-based.

Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
 drivers/md/dm-table.c | 27 ++++++++-------------------
 1 file changed, 8 insertions(+), 19 deletions(-)

diff --git a/drivers/md/dm-table.c b/drivers/md/dm-table.c
index 52e049554f5cd..2ae0c19137667 100644
--- a/drivers/md/dm-table.c
+++ b/drivers/md/dm-table.c
@@ -918,21 +918,15 @@ bool dm_table_supports_dax(struct dm_table *t,
 
 static bool dm_table_does_not_support_partial_completion(struct dm_table *t);
 
-struct verify_rq_based_data {
-	unsigned sq_count;
-	unsigned mq_count;
-};
-
-static int device_is_rq_based(struct dm_target *ti, struct dm_dev *dev,
-			      sector_t start, sector_t len, void *data)
+static int device_is_rq_stackable(struct dm_target *ti, struct dm_dev *dev,
+				  sector_t start, sector_t len, void *data)
 {
-	struct request_queue *q = bdev_get_queue(dev->bdev);
-	struct verify_rq_based_data *v = data;
+	struct block_device *bdev = dev->bdev;
+	struct request_queue *q = bdev_get_queue(bdev);
 
-	if (queue_is_mq(q))
-		v->mq_count++;
-	else
-		v->sq_count++;
+	/* request-based cannot stack on partitions! */
+	if (bdev != bdev->bd_contains)
+		return false;
 
 	return queue_is_mq(q);
 }
@@ -941,7 +935,6 @@ static int dm_table_determine_type(struct dm_table *t)
 {
 	unsigned i;
 	unsigned bio_based = 0, request_based = 0, hybrid = 0;
-	struct verify_rq_based_data v = {.sq_count = 0, .mq_count = 0};
 	struct dm_target *tgt;
 	struct list_head *devices = dm_table_get_devices(t);
 	enum dm_queue_mode live_md_type = dm_get_md_type(t->md);
@@ -1045,14 +1038,10 @@ verify_rq_based:
 
 	/* Non-request-stackable devices can't be used for request-based dm */
 	if (!tgt->type->iterate_devices ||
-	    !tgt->type->iterate_devices(tgt, device_is_rq_based, &v)) {
+	    !tgt->type->iterate_devices(tgt, device_is_rq_stackable, NULL)) {
 		DMERR("table load rejected: including non-request-stackable devices");
 		return -EINVAL;
 	}
-	if (v.sq_count > 0) {
-		DMERR("table load rejected: not all devices are blk-mq request-stackable");
-		return -EINVAL;
-	}
 
 	return 0;
 }
-- 
2.25.1

[PATCH AUTOSEL 5.4 045/330] bcache: fix a lost wake-up problem caused by mca_cannibalize_lock

[Sasha Levin]

From: Guoju Fang <fangguoju@gmail.com>

[ Upstream commit 34cf78bf34d48dddddfeeadb44f9841d7864997a ]

This patch fix a lost wake-up problem caused by the race between
mca_cannibalize_lock and bch_cannibalize_unlock.

Consider two processes, A and B. Process A is executing
mca_cannibalize_lock, while process B takes c->btree_cache_alloc_lock
and is executing bch_cannibalize_unlock. The problem happens that after
process A executes cmpxchg and will execute prepare_to_wait. In this
timeslice process B executes wake_up, but after that process A executes
prepare_to_wait and set the state to TASK_INTERRUPTIBLE. Then process A
goes to sleep but no one will wake up it. This problem may cause bcache
device to dead.

Signed-off-by: Guoju Fang <fangguoju@gmail.com>
Signed-off-by: Coly Li <colyli@suse.de>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
 drivers/md/bcache/bcache.h |  1 +
 drivers/md/bcache/btree.c  | 12 ++++++++----
 drivers/md/bcache/super.c  |  1 +
 3 files changed, 10 insertions(+), 4 deletions(-)

diff --git a/drivers/md/bcache/bcache.h b/drivers/md/bcache/bcache.h
index 217c838a1b405..859567ad3db4e 100644
--- a/drivers/md/bcache/bcache.h
+++ b/drivers/md/bcache/bcache.h
@@ -585,6 +585,7 @@ struct cache_set {
 	 */
 	wait_queue_head_t	btree_cache_wait;
 	struct task_struct	*btree_cache_alloc_lock;
+	spinlock_t		btree_cannibalize_lock;
 
 	/*
 	 * When we free a btree node, we increment the gen of the bucket the
diff --git a/drivers/md/bcache/btree.c b/drivers/md/bcache/btree.c
index 46556bde032e2..8d06105fc9ff5 100644
--- a/drivers/md/bcache/btree.c
+++ b/drivers/md/bcache/btree.c
@@ -886,15 +886,17 @@ out:
 
 static int mca_cannibalize_lock(struct cache_set *c, struct btree_op *op)
 {
-	struct task_struct *old;
-
-	old = cmpxchg(&c->btree_cache_alloc_lock, NULL, current);
-	if (old && old != current) {
+	spin_lock(&c->btree_cannibalize_lock);
+	if (likely(c->btree_cache_alloc_lock == NULL)) {
+		c->btree_cache_alloc_lock = current;
+	} else if (c->btree_cache_alloc_lock != current) {
 		if (op)
 			prepare_to_wait(&c->btree_cache_wait, &op->wait,
 					TASK_UNINTERRUPTIBLE);
+		spin_unlock(&c->btree_cannibalize_lock);
 		return -EINTR;
 	}
+	spin_unlock(&c->btree_cannibalize_lock);
 
 	return 0;
 }
@@ -929,10 +931,12 @@ static struct btree *mca_cannibalize(struct cache_set *c, struct btree_op *op,
  */
 static void bch_cannibalize_unlock(struct cache_set *c)
 {
+	spin_lock(&c->btree_cannibalize_lock);
 	if (c->btree_cache_alloc_lock == current) {
 		c->btree_cache_alloc_lock = NULL;
 		wake_up(&c->btree_cache_wait);
 	}
+	spin_unlock(&c->btree_cannibalize_lock);
 }
 
 static struct btree *mca_alloc(struct cache_set *c, struct btree_op *op,
diff --git a/drivers/md/bcache/super.c b/drivers/md/bcache/super.c
index 2cbfcd99b7ee7..63f5ce18311bb 100644
--- a/drivers/md/bcache/super.c
+++ b/drivers/md/bcache/super.c
@@ -1798,6 +1798,7 @@ struct cache_set *bch_cache_set_alloc(struct cache_sb *sb)
 	sema_init(&c->sb_write_mutex, 1);
 	mutex_init(&c->bucket_lock);
 	init_waitqueue_head(&c->btree_cache_wait);
+	spin_lock_init(&c->btree_cannibalize_lock);
 	init_waitqueue_head(&c->bucket_wait);
 	init_waitqueue_head(&c->gc_wait);
 	sema_init(&c->uuid_write_mutex, 1);
-- 
2.25.1