Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs

Pull btrfs update from Chris Mason:
 "From a feature point of view, most of the code here comes from Miao
  Xie and others at Fujitsu to implement scrubbing and replacing devices
  on raid56.  This has been in development for a while, and it's a big
  improvement.

  Filipe and Josef have a great assortment of fixes, many of which solve
  problems corruptions either after a crash or in error conditions.  I
  still have a round two from Filipe for next week that solves
  corruptions with discard and block group removal"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs: (62 commits)
  Btrfs: make get_caching_control unconditionally return the ctl
  Btrfs: fix unprotected deletion from pending_chunks list
  Btrfs: fix fs mapping extent map leak
  Btrfs: fix memory leak after block remove + trimming
  Btrfs: make btrfs_abort_transaction consider existence of new block groups
  Btrfs: fix race between writing free space cache and trimming
  Btrfs: fix race between fs trimming and block group remove/allocation
  Btrfs, replace: enable dev-replace for raid56
  Btrfs: fix freeing used extents after removing empty block group
  Btrfs: fix crash caused by block group removal
  Btrfs: fix invalid block group rbtree access after bg is removed
  Btrfs, raid56: fix use-after-free problem in the final device replace procedure on raid56
  Btrfs, replace: write raid56 parity into the replace target device
  Btrfs, replace: write dirty pages into the replace target device
  Btrfs, raid56: support parity scrub on raid56
  Btrfs, raid56: use a variant to record the operation type
  Btrfs, scrub: repair the common data on RAID5/6 if it is corrupted
  Btrfs, raid56: don't change bbio and raid_map
  Btrfs: remove unnecessary code of stripe_index assignment in __btrfs_map_block
  Btrfs: remove noused bbio_ret in __btrfs_map_block in condition
  ...

1 files changed, 58 insertions, 32 deletions

diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index d47289c715c8..0144790e296e 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -53,16 +53,6 @@ static void btrfs_dev_stat_print_on_load(struct btrfs_device *device);
 DEFINE_MUTEX(uuid_mutex);
 static LIST_HEAD(fs_uuids);
 
-static void lock_chunks(struct btrfs_root *root)
-{
-	mutex_lock(&root->fs_info->chunk_mutex);
-}
-
-static void unlock_chunks(struct btrfs_root *root)
-{
-	mutex_unlock(&root->fs_info->chunk_mutex);
-}
-
 static struct btrfs_fs_devices *__alloc_fs_devices(void)
 {
 	struct btrfs_fs_devices *fs_devs;
@@ -1068,9 +1058,11 @@ static int contains_pending_extent(struct btrfs_trans_handle *trans,
 				   u64 *start, u64 len)
 {
 	struct extent_map *em;
+	struct list_head *search_list = &trans->transaction->pending_chunks;
 	int ret = 0;
 
-	list_for_each_entry(em, &trans->transaction->pending_chunks, list) {
+again:
+	list_for_each_entry(em, search_list, list) {
 		struct map_lookup *map;
 		int i;
 
@@ -1087,6 +1079,10 @@ static int contains_pending_extent(struct btrfs_trans_handle *trans,
 			ret = 1;
 		}
 	}
+	if (search_list == &trans->transaction->pending_chunks) {
+		search_list = &trans->root->fs_info->pinned_chunks;
+		goto again;
+	}
 
 	return ret;
 }
@@ -1800,8 +1796,8 @@ error_undo:
 	goto error_brelse;
 }
 
-void btrfs_rm_dev_replace_srcdev(struct btrfs_fs_info *fs_info,
-				 struct btrfs_device *srcdev)
+void btrfs_rm_dev_replace_remove_srcdev(struct btrfs_fs_info *fs_info,
+					struct btrfs_device *srcdev)
 {
 	struct btrfs_fs_devices *fs_devices;
 
@@ -1829,6 +1825,12 @@ void btrfs_rm_dev_replace_srcdev(struct btrfs_fs_info *fs_info,
 
 	if (srcdev->bdev)
 		fs_devices->open_devices--;
+}
+
+void btrfs_rm_dev_replace_free_srcdev(struct btrfs_fs_info *fs_info,
+				      struct btrfs_device *srcdev)
+{
+	struct btrfs_fs_devices *fs_devices = srcdev->fs_devices;
 
 	call_rcu(&srcdev->rcu, free_device);
 
@@ -2647,18 +2649,12 @@ int btrfs_remove_chunk(struct btrfs_trans_handle *trans,
 		}
 	}
 
-	ret = btrfs_remove_block_group(trans, extent_root, chunk_offset);
+	ret = btrfs_remove_block_group(trans, extent_root, chunk_offset, em);
 	if (ret) {
 		btrfs_abort_transaction(trans, extent_root, ret);
 		goto out;
 	}
 
-	write_lock(&em_tree->lock);
-	remove_extent_mapping(em_tree, em);
-	write_unlock(&em_tree->lock);
-
-	/* once for the tree */
-	free_extent_map(em);
 out:
 	/* once for us */
 	free_extent_map(em);
@@ -4505,6 +4501,8 @@ error_del_extent:
 	free_extent_map(em);
 	/* One for the tree reference */
 	free_extent_map(em);
+	/* One for the pending_chunks list reference */
+	free_extent_map(em);
 error:
 	kfree(devices_info);
 	return ret;
@@ -4881,13 +4879,15 @@ static inline int parity_smaller(u64 a, u64 b)
 static void sort_parity_stripes(struct btrfs_bio *bbio, u64 *raid_map)
 {
 	struct btrfs_bio_stripe s;
+	int real_stripes = bbio->num_stripes - bbio->num_tgtdevs;
 	int i;
 	u64 l;
 	int again = 1;
+	int m;
 
 	while (again) {
 		again = 0;
-		for (i = 0; i < bbio->num_stripes - 1; i++) {
+		for (i = 0; i < real_stripes - 1; i++) {
 			if (parity_smaller(raid_map[i], raid_map[i+1])) {
 				s = bbio->stripes[i];
 				l = raid_map[i];
@@ -4895,6 +4895,14 @@ static void sort_parity_stripes(struct btrfs_bio *bbio, u64 *raid_map)
 				raid_map[i] = raid_map[i+1];
 				bbio->stripes[i+1] = s;
 				raid_map[i+1] = l;
+
+				if (bbio->tgtdev_map) {
+					m = bbio->tgtdev_map[i];
+					bbio->tgtdev_map[i] =
+							bbio->tgtdev_map[i + 1];
+					bbio->tgtdev_map[i + 1] = m;
+				}
+
 				again = 1;
 			}
 		}
@@ -4923,6 +4931,7 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info, int rw,
 	int ret = 0;
 	int num_stripes;
 	int max_errors = 0;
+	int tgtdev_indexes = 0;
 	struct btrfs_bio *bbio = NULL;
 	struct btrfs_dev_replace *dev_replace = &fs_info->dev_replace;
 	int dev_replace_is_ongoing = 0;
@@ -5161,15 +5170,14 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info, int rw,
 				BTRFS_BLOCK_GROUP_RAID6)) {
 		u64 tmp;
 
-		if (bbio_ret && ((rw & REQ_WRITE) || mirror_num > 1)
-		    && raid_map_ret) {
+		if (raid_map_ret &&
+		    ((rw & (REQ_WRITE | REQ_GET_READ_MIRRORS)) ||
+		     mirror_num > 1)) {
 			int i, rot;
 
 			/* push stripe_nr back to the start of the full stripe */
 			stripe_nr = raid56_full_stripe_start;
-			do_div(stripe_nr, stripe_len);
-
-			stripe_index = do_div(stripe_nr, nr_data_stripes(map));
+			do_div(stripe_nr, stripe_len * nr_data_stripes(map));
 
 			/* RAID[56] write or recovery. Return all stripes */
 			num_stripes = map->num_stripes;
@@ -5235,14 +5243,19 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info, int rw,
 			num_alloc_stripes <<= 1;
 		if (rw & REQ_GET_READ_MIRRORS)
 			num_alloc_stripes++;
+		tgtdev_indexes = num_stripes;
 	}
-	bbio = kzalloc(btrfs_bio_size(num_alloc_stripes), GFP_NOFS);
+
+	bbio = kzalloc(btrfs_bio_size(num_alloc_stripes, tgtdev_indexes),
+		       GFP_NOFS);
 	if (!bbio) {
 		kfree(raid_map);
 		ret = -ENOMEM;
 		goto out;
 	}
 	atomic_set(&bbio->error, 0);
+	if (dev_replace_is_ongoing)
+		bbio->tgtdev_map = (int *)(bbio->stripes + num_alloc_stripes);
 
 	if (rw & REQ_DISCARD) {
 		int factor = 0;
@@ -5327,6 +5340,7 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info, int rw,
 	if (rw & (REQ_WRITE | REQ_GET_READ_MIRRORS))
 		max_errors = btrfs_chunk_max_errors(map);
 
+	tgtdev_indexes = 0;
 	if (dev_replace_is_ongoing && (rw & (REQ_WRITE | REQ_DISCARD)) &&
 	    dev_replace->tgtdev != NULL) {
 		int index_where_to_add;
@@ -5355,8 +5369,10 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info, int rw,
 				new->physical = old->physical;
 				new->length = old->length;
 				new->dev = dev_replace->tgtdev;
+				bbio->tgtdev_map[i] = index_where_to_add;
 				index_where_to_add++;
 				max_errors++;
+				tgtdev_indexes++;
 			}
 		}
 		num_stripes = index_where_to_add;
@@ -5402,7 +5418,9 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info, int rw,
 				tgtdev_stripe->length =
 					bbio->stripes[index_srcdev].length;
 				tgtdev_stripe->dev = dev_replace->tgtdev;
+				bbio->tgtdev_map[index_srcdev] = num_stripes;
 
+				tgtdev_indexes++;
 				num_stripes++;
 			}
 		}
@@ -5412,6 +5430,7 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info, int rw,
 	bbio->num_stripes = num_stripes;
 	bbio->max_errors = max_errors;
 	bbio->mirror_num = mirror_num;
+	bbio->num_tgtdevs = tgtdev_indexes;
 
 	/*
 	 * this is the case that REQ_READ && dev_replace_is_ongoing &&
@@ -5443,6 +5462,16 @@ int btrfs_map_block(struct btrfs_fs_info *fs_info, int rw,
 				 mirror_num, NULL);
 }
 
+/* For Scrub/replace */
+int btrfs_map_sblock(struct btrfs_fs_info *fs_info, int rw,
+		     u64 logical, u64 *length,
+		     struct btrfs_bio **bbio_ret, int mirror_num,
+		     u64 **raid_map_ret)
+{
+	return __btrfs_map_block(fs_info, rw, logical, length, bbio_ret,
+				 mirror_num, raid_map_ret);
+}
+
 int btrfs_rmap_block(struct btrfs_mapping_tree *map_tree,
 		     u64 chunk_start, u64 physical, u64 devid,
 		     u64 **logical, int *naddrs, int *stripe_len)
@@ -5812,12 +5841,9 @@ int btrfs_map_bio(struct btrfs_root *root, int rw, struct bio *bio,
 		} else {
 			ret = raid56_parity_recover(root, bio, bbio,
 						    raid_map, map_length,
-						    mirror_num);
+						    mirror_num, 1);
 		}
-		/*
-		 * FIXME, replace dosen't support raid56 yet, please fix
-		 * it in the future.
-		 */
+
 		btrfs_bio_counter_dec(root->fs_info);
 		return ret;
 	}