1 files changed, 75 insertions, 12 deletions

diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index 3e64f49c394b..e798caee978e 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -4,6 +4,7 @@
  */
 
 #include <linux/sched.h>
+#include <linux/sched/mm.h>
 #include <linux/bio.h>
 #include <linux/slab.h>
 #include <linux/buffer_head.h>
@@ -219,7 +220,9 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info,
  *
  * global::fs_devs - add, remove, updates to the global list
  *
- * does not protect: manipulation of the fs_devices::devices list!
+ * does not protect: manipulation of the fs_devices::devices list in general
+ * but in mount context it could be used to exclude list modifications by eg.
+ * scan ioctl
  *
  * btrfs_device::name - renames (write side), read is RCU
  *
@@ -232,6 +235,9 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info,
  * may be used to exclude some operations from running concurrently without any
  * modifications to the list (see write_all_supers)
  *
+ * Is not required at mount and close times, because our device list is
+ * protected by the uuid_mutex at that point.
+ *
  * balance_mutex
  * -------------
  * protects balance structures (status, state) and context accessed from
@@ -778,6 +784,11 @@ static int btrfs_free_stale_devices(const char *path,
 	return ret;
 }
 
+/*
+ * This is only used on mount, and we are protected from competing things
+ * messing with our fs_devices by the uuid_mutex, thus we do not need the
+ * fs_devices->device_list_mutex here.
+ */
 static int btrfs_open_one_device(struct btrfs_fs_devices *fs_devices,
 			struct btrfs_device *device, fmode_t flags,
 			void *holder)
@@ -1223,6 +1234,8 @@ again:
 							&device->dev_state)) {
 			if (!test_bit(BTRFS_DEV_STATE_REPLACE_TGT,
 			     &device->dev_state) &&
+			    !test_bit(BTRFS_DEV_STATE_MISSING,
+				      &device->dev_state) &&
 			     (!latest_dev ||
 			      device->generation > latest_dev->generation)) {
 				latest_dev = device;
@@ -1416,8 +1429,14 @@ int btrfs_open_devices(struct btrfs_fs_devices *fs_devices,
 	int ret;
 
 	lockdep_assert_held(&uuid_mutex);
+	/*
+	 * The device_list_mutex cannot be taken here in case opening the
+	 * underlying device takes further locks like bd_mutex.
+	 *
+	 * We also don't need the lock here as this is called during mount and
+	 * exclusion is provided by uuid_mutex
+	 */
 
-	mutex_lock(&fs_devices->device_list_mutex);
 	if (fs_devices->opened) {
 		fs_devices->opened++;
 		ret = 0;
@@ -1425,7 +1444,6 @@ int btrfs_open_devices(struct btrfs_fs_devices *fs_devices,
 		list_sort(NULL, &fs_devices->devices, devid_cmp);
 		ret = open_fs_devices(fs_devices, flags, holder);
 	}
-	mutex_unlock(&fs_devices->device_list_mutex);
 
 	return ret;
 }
@@ -2769,8 +2787,18 @@ int btrfs_init_new_device(struct btrfs_fs_info *fs_info, const char *device_path
 		ret = btrfs_commit_transaction(trans);
 	}
 
-	/* Update ctime/mtime for libblkid */
+	/*
+	 * Now that we have written a new super block to this device, check all
+	 * other fs_devices list if device_path alienates any other scanned
+	 * device.
+	 * We can ignore the return value as it typically returns -EINVAL and
+	 * only succeeds if the device was an alien.
+	 */
+	btrfs_forget_devices(device_path);
+
+	/* Update ctime/mtime for blkid or udev */
 	update_dev_time(device_path);
+
 	return ret;
 
 error_sysfs:
@@ -3271,7 +3299,7 @@ static int insert_balance_item(struct btrfs_fs_info *fs_info,
 	if (!path)
 		return -ENOMEM;
 
-	trans = btrfs_start_transaction(root, 0);
+	trans = btrfs_start_transaction_fallback_global_rsv(root, 0);
 	if (IS_ERR(trans)) {
 		btrfs_free_path(path);
 		return PTR_ERR(trans);
@@ -4234,7 +4262,22 @@ int btrfs_balance(struct btrfs_fs_info *fs_info,
 	mutex_lock(&fs_info->balance_mutex);
 	if (ret == -ECANCELED && atomic_read(&fs_info->balance_pause_req))
 		btrfs_info(fs_info, "balance: paused");
-	else if (ret == -ECANCELED && atomic_read(&fs_info->balance_cancel_req))
+	/*
+	 * Balance can be canceled by:
+	 *
+	 * - Regular cancel request
+	 *   Then ret == -ECANCELED and balance_cancel_req > 0
+	 *
+	 * - Fatal signal to "btrfs" process
+	 *   Either the signal caught by wait_reserve_ticket() and callers
+	 *   got -EINTR, or caught by btrfs_should_cancel_balance() and
+	 *   got -ECANCELED.
+	 *   Either way, in this case balance_cancel_req = 0, and
+	 *   ret == -EINTR or ret == -ECANCELED.
+	 *
+	 * So here we only check the return value to catch canceled balance.
+	 */
+	else if (ret == -ECANCELED || ret == -EINTR)
 		btrfs_info(fs_info, "balance: canceled");
 	else
 		btrfs_info(fs_info, "balance: ended with status: %d", ret);
@@ -4526,6 +4569,7 @@ static int btrfs_uuid_scan_kthread(void *data)
 			goto skip;
 		}
 update_tree:
+		btrfs_release_path(path);
 		if (!btrfs_is_empty_uuid(root_item.uuid)) {
 			ret = btrfs_uuid_tree_add(trans, root_item.uuid,
 						  BTRFS_UUID_KEY_SUBVOL,
@@ -4550,6 +4594,7 @@ update_tree:
 		}
 
 skip:
+		btrfs_release_path(path);
 		if (trans) {
 			ret = btrfs_end_transaction(trans);
 			trans = NULL;
@@ -4557,7 +4602,6 @@ skip:
 				break;
 		}
 
-		btrfs_release_path(path);
 		if (key.offset < (u64)-1) {
 			key.offset++;
 		} else if (key.type < BTRFS_ROOT_ITEM_KEY) {
@@ -5632,12 +5676,13 @@ void btrfs_put_bbio(struct btrfs_bio *bbio)
  * replace.
  */
 static int __btrfs_map_block_for_discard(struct btrfs_fs_info *fs_info,
-					 u64 logical, u64 length,
+					 u64 logical, u64 *length_ret,
 					 struct btrfs_bio **bbio_ret)
 {
 	struct extent_map *em;
 	struct map_lookup *map;
 	struct btrfs_bio *bbio;
+	u64 length = *length_ret;
 	u64 offset;
 	u64 stripe_nr;
 	u64 stripe_nr_end;
@@ -5670,7 +5715,8 @@ static int __btrfs_map_block_for_discard(struct btrfs_fs_info *fs_info,
 	}
 
 	offset = logical - em->start;
-	length = min_t(u64, em->len - offset, length);
+	length = min_t(u64, em->start + em->len - logical, length);
+	*length_ret = length;
 
 	stripe_len = map->stripe_len;
 	/*
@@ -6085,7 +6131,7 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info,
 
 	if (op == BTRFS_MAP_DISCARD)
 		return __btrfs_map_block_for_discard(fs_info, logical,
-						     *length, bbio_ret);
+						     length, bbio_ret);
 
 	ret = btrfs_get_io_geometry(fs_info, op, logical, *length, &geom);
 	if (ret < 0)
@@ -6665,8 +6711,17 @@ static struct btrfs_device *add_missing_dev(struct btrfs_fs_devices *fs_devices,
 					    u64 devid, u8 *dev_uuid)
 {
 	struct btrfs_device *device;
+	unsigned int nofs_flag;
 
+	/*
+	 * We call this under the chunk_mutex, so we want to use NOFS for this
+	 * allocation, however we don't want to change btrfs_alloc_device() to
+	 * always do NOFS because we use it in a lot of other GFP_KERNEL safe
+	 * places.
+	 */
+	nofs_flag = memalloc_nofs_save();
 	device = btrfs_alloc_device(NULL, &devid, dev_uuid);
+	memalloc_nofs_restore(nofs_flag);
 	if (IS_ERR(device))
 		return device;
 
@@ -7255,7 +7310,14 @@ int btrfs_read_chunk_tree(struct btrfs_fs_info *fs_info)
 	 * otherwise we don't need it.
 	 */
 	mutex_lock(&uuid_mutex);
-	mutex_lock(&fs_info->chunk_mutex);
+
+	/*
+	 * It is possible for mount and umount to race in such a way that
+	 * we execute this code path, but open_fs_devices failed to clear
+	 * total_rw_bytes. We certainly want it cleared before reading the
+	 * device items, so clear it here.
+	 */
+	fs_info->fs_devices->total_rw_bytes = 0;
 
 	/*
 	 * Read all device items, and then all the chunk items. All
@@ -7292,7 +7354,9 @@ int btrfs_read_chunk_tree(struct btrfs_fs_info *fs_info)
 		} else if (found_key.type == BTRFS_CHUNK_ITEM_KEY) {
 			struct btrfs_chunk *chunk;
 			chunk = btrfs_item_ptr(leaf, slot, struct btrfs_chunk);
+			mutex_lock(&fs_info->chunk_mutex);
 			ret = read_one_chunk(&found_key, leaf, chunk);
+			mutex_unlock(&fs_info->chunk_mutex);
 			if (ret)
 				goto error;
 		}
@@ -7322,7 +7386,6 @@ int btrfs_read_chunk_tree(struct btrfs_fs_info *fs_info)
 	}
 	ret = 0;
 error:
-	mutex_unlock(&fs_info->chunk_mutex);
 	mutex_unlock(&uuid_mutex);
 
 	btrfs_free_path(path);