1 files changed, 101 insertions, 107 deletions

diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c
index beb6c69cd1e5..20c6ac1a5de7 100644
--- a/fs/btrfs/transaction.c
+++ b/fs/btrfs/transaction.c
@@ -21,6 +21,7 @@
 #include "dev-replace.h"
 #include "qgroup.h"
 #include "block-group.h"
+#include "space-info.h"
 
 #define BTRFS_ROOT_TRANS_TAG 0
 
@@ -141,7 +142,7 @@ void btrfs_put_transaction(struct btrfs_transaction *transaction)
 						 struct btrfs_block_group,
 						 bg_list);
 			list_del_init(&cache->bg_list);
-			btrfs_put_block_group_trimming(cache);
+			btrfs_unfreeze_block_group(cache);
 			btrfs_put_block_group(cache);
 		}
 		WARN_ON(!list_empty(&transaction->dev_update_list));
@@ -221,7 +222,7 @@ void btrfs_trans_release_chunk_metadata(struct btrfs_trans_handle *trans)
 	WARN_ON_ONCE(!list_empty(&trans->new_bgs));
 
 	btrfs_block_rsv_release(fs_info, &fs_info->chunk_block_rsv,
-				trans->chunk_bytes_reserved);
+				trans->chunk_bytes_reserved, NULL);
 	trans->chunk_bytes_reserved = 0;
 }
 
@@ -243,7 +244,7 @@ loop:
 
 	cur_trans = fs_info->running_transaction;
 	if (cur_trans) {
-		if (cur_trans->aborted) {
+		if (TRANS_ABORTED(cur_trans)) {
 			spin_unlock(&fs_info->trans_lock);
 			return cur_trans->aborted;
 		}
@@ -336,6 +337,8 @@ loop:
 	list_add_tail(&cur_trans->list, &fs_info->trans_list);
 	extent_io_tree_init(fs_info, &cur_trans->dirty_pages,
 			IO_TREE_TRANS_DIRTY_PAGES, fs_info->btree_inode);
+	extent_io_tree_init(fs_info, &cur_trans->pinned_extents,
+			IO_TREE_FS_PINNED_EXTENTS, NULL);
 	fs_info->generation++;
 	cur_trans->transid = fs_info->generation;
 	fs_info->running_transaction = cur_trans;
@@ -346,10 +349,10 @@ loop:
 }
 
 /*
- * this does all the record keeping required to make sure that a reference
- * counted root is properly recorded in a given transaction.  This is required
- * to make sure the old root from before we joined the transaction is deleted
- * when the transaction commits
+ * This does all the record keeping required to make sure that a shareable root
+ * is properly recorded in a given transaction.  This is required to make sure
+ * the old root from before we joined the transaction is deleted when the
+ * transaction commits.
  */
 static int record_root_in_trans(struct btrfs_trans_handle *trans,
 			       struct btrfs_root *root,
@@ -357,7 +360,7 @@ static int record_root_in_trans(struct btrfs_trans_handle *trans,
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
 
-	if ((test_bit(BTRFS_ROOT_REF_COWS, &root->state) &&
+	if ((test_bit(BTRFS_ROOT_SHAREABLE, &root->state) &&
 	    root->last_trans < trans->transid) || force) {
 		WARN_ON(root == fs_info->extent_root);
 		WARN_ON(!force && root->commit_root != root->node);
@@ -436,7 +439,7 @@ int btrfs_record_root_in_trans(struct btrfs_trans_handle *trans,
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
 
-	if (!test_bit(BTRFS_ROOT_REF_COWS, &root->state))
+	if (!test_bit(BTRFS_ROOT_SHAREABLE, &root->state))
 		return 0;
 
 	/*
@@ -459,7 +462,7 @@ static inline int is_transaction_blocked(struct btrfs_transaction *trans)
 {
 	return (trans->state >= TRANS_STATE_COMMIT_START &&
 		trans->state < TRANS_STATE_UNBLOCKED &&
-		!trans->aborted);
+		!TRANS_ABORTED(trans));
 }
 
 /* wait for commit against the current transaction to become unblocked
@@ -478,7 +481,7 @@ static void wait_current_trans(struct btrfs_fs_info *fs_info)
 
 		wait_event(fs_info->transaction_wait,
 			   cur_trans->state >= TRANS_STATE_UNBLOCKED ||
-			   cur_trans->aborted);
+			   TRANS_ABORTED(cur_trans));
 		btrfs_put_transaction(cur_trans);
 	} else {
 		spin_unlock(&fs_info->trans_lock);
@@ -501,7 +504,7 @@ static inline bool need_reserve_reloc_root(struct btrfs_root *root)
 	struct btrfs_fs_info *fs_info = root->fs_info;
 
 	if (!fs_info->reloc_ctl ||
-	    !test_bit(BTRFS_ROOT_REF_COWS, &root->state) ||
+	    !test_bit(BTRFS_ROOT_SHAREABLE, &root->state) ||
 	    root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID ||
 	    root->reloc_root)
 		return false;
@@ -521,6 +524,7 @@ start_transaction(struct btrfs_root *root, unsigned int num_items,
 	u64 num_bytes = 0;
 	u64 qgroup_reserved = 0;
 	bool reloc_reserved = false;
+	bool do_chunk_alloc = false;
 	int ret;
 
 	/* Send isn't supposed to start transactions. */
@@ -561,7 +565,8 @@ start_transaction(struct btrfs_root *root, unsigned int num_items,
 		 * refill that amount for whatever is missing in the reserve.
 		 */
 		num_bytes = btrfs_calc_insert_metadata_size(fs_info, num_items);
-		if (delayed_refs_rsv->full == 0) {
+		if (flush == BTRFS_RESERVE_FLUSH_ALL &&
+		    delayed_refs_rsv->full == 0) {
 			delayed_refs_bytes = num_bytes;
 			num_bytes <<= 1;
 		}
@@ -582,6 +587,9 @@ start_transaction(struct btrfs_root *root, unsigned int num_items,
 							  delayed_refs_bytes);
 			num_bytes -= delayed_refs_bytes;
 		}
+
+		if (rsv->space_info->force_alloc)
+			do_chunk_alloc = true;
 	} else if (num_items == 0 && flush == BTRFS_RESERVE_FLUSH_ALL &&
 		   !delayed_refs_rsv->full) {
 		/*
@@ -660,10 +668,32 @@ again:
 	}
 
 got_it:
-	btrfs_record_root_in_trans(h, root);
-
 	if (!current->journal_info)
 		current->journal_info = h;
+
+	/*
+	 * If the space_info is marked ALLOC_FORCE then we'll get upgraded to
+	 * ALLOC_FORCE the first run through, and then we won't allocate for
+	 * anybody else who races in later.  We don't care about the return
+	 * value here.
+	 */
+	if (do_chunk_alloc && num_bytes) {
+		u64 flags = h->block_rsv->space_info->flags;
+
+		btrfs_chunk_alloc(h, btrfs_get_alloc_profile(fs_info, flags),
+				  CHUNK_ALLOC_NO_FORCE);
+	}
+
+	/*
+	 * btrfs_record_root_in_trans() needs to alloc new extents, and may
+	 * call btrfs_join_transaction() while we're also starting a
+	 * transaction.
+	 *
+	 * Thus it need to be called after current->journal_info initialized,
+	 * or we can deadlock.
+	 */
+	btrfs_record_root_in_trans(h, root);
+
 	return h;
 
 join_fail:
@@ -673,7 +703,7 @@ join_fail:
 alloc_fail:
 	if (num_bytes)
 		btrfs_block_rsv_release(fs_info, &fs_info->trans_block_rsv,
-					num_bytes);
+					num_bytes, NULL);
 reserve_fail:
 	btrfs_qgroup_free_meta_pertrans(root, qgroup_reserved);
 	return ERR_PTR(ret);
@@ -688,43 +718,10 @@ struct btrfs_trans_handle *btrfs_start_transaction(struct btrfs_root *root,
 
 struct btrfs_trans_handle *btrfs_start_transaction_fallback_global_rsv(
 					struct btrfs_root *root,
-					unsigned int num_items,
-					int min_factor)
+					unsigned int num_items)
 {
-	struct btrfs_fs_info *fs_info = root->fs_info;
-	struct btrfs_trans_handle *trans;
-	u64 num_bytes;
-	int ret;
-
-	/*
-	 * We have two callers: unlink and block group removal.  The
-	 * former should succeed even if we will temporarily exceed
-	 * quota and the latter operates on the extent root so
-	 * qgroup enforcement is ignored anyway.
-	 */
-	trans = start_transaction(root, num_items, TRANS_START,
-				  BTRFS_RESERVE_FLUSH_ALL, false);
-	if (!IS_ERR(trans) || PTR_ERR(trans) != -ENOSPC)
-		return trans;
-
-	trans = btrfs_start_transaction(root, 0);
-	if (IS_ERR(trans))
-		return trans;
-
-	num_bytes = btrfs_calc_insert_metadata_size(fs_info, num_items);
-	ret = btrfs_cond_migrate_bytes(fs_info, &fs_info->trans_block_rsv,
-				       num_bytes, min_factor);
-	if (ret) {
-		btrfs_end_transaction(trans);
-		return ERR_PTR(ret);
-	}
-
-	trans->block_rsv = &fs_info->trans_block_rsv;
-	trans->bytes_reserved = num_bytes;
-	trace_btrfs_space_reservation(fs_info, "transaction",
-				      trans->transid, num_bytes, 1);
-
-	return trans;
+	return start_transaction(root, num_items, TRANS_START,
+				 BTRFS_RESERVE_FLUSH_ALL_STEAL, false);
 }
 
 struct btrfs_trans_handle *btrfs_join_transaction(struct btrfs_root *root)
@@ -896,7 +893,7 @@ static void btrfs_trans_release_metadata(struct btrfs_trans_handle *trans)
 	trace_btrfs_space_reservation(fs_info, "transaction",
 				      trans->transid, trans->bytes_reserved, 0);
 	btrfs_block_rsv_release(fs_info, trans->block_rsv,
-				trans->bytes_reserved);
+				trans->bytes_reserved, NULL);
 	trans->bytes_reserved = 0;
 }
 
@@ -937,10 +934,13 @@ static int __btrfs_end_transaction(struct btrfs_trans_handle *trans,
 	if (throttle)
 		btrfs_run_delayed_iputs(info);
 
-	if (trans->aborted ||
+	if (TRANS_ABORTED(trans) ||
 	    test_bit(BTRFS_FS_STATE_ERROR, &info->fs_state)) {
 		wake_up_process(info->transaction_kthread);
-		err = -EIO;
+		if (TRANS_ABORTED(trans))
+			err = trans->aborted;
+		else
+			err = -EROFS;
 	}
 
 	kmem_cache_free(btrfs_trans_handle_cachep, trans);
@@ -1262,8 +1262,10 @@ void btrfs_add_dead_root(struct btrfs_root *root)
 	struct btrfs_fs_info *fs_info = root->fs_info;
 
 	spin_lock(&fs_info->trans_lock);
-	if (list_empty(&root->root_list))
+	if (list_empty(&root->root_list)) {
+		btrfs_grab_root(root);
 		list_add_tail(&root->root_list, &fs_info->dead_roots);
+	}
 	spin_unlock(&fs_info->trans_lock);
 }
 
@@ -1477,7 +1479,6 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 	u64 index = 0;
 	u64 objectid;
 	u64 root_flags;
-	uuid_le new_uuid;
 
 	ASSERT(pending->path);
 	path = pending->path;
@@ -1570,8 +1571,7 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 
 	btrfs_set_root_generation_v2(new_root_item,
 			trans->transid);
-	uuid_le_gen(&new_uuid);
-	memcpy(new_root_item->uuid, new_uuid.b, BTRFS_UUID_SIZE);
+	generate_random_guid(new_root_item->uuid);
 	memcpy(new_root_item->parent_uuid, root->root_item.uuid,
 			BTRFS_UUID_SIZE);
 	if (!(root_flags & BTRFS_ROOT_SUBVOL_RDONLY)) {
@@ -1633,7 +1633,7 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 	}
 
 	key.offset = (u64)-1;
-	pending->snap = btrfs_read_fs_root_no_name(fs_info, &key);
+	pending->snap = btrfs_get_new_fs_root(fs_info, objectid, pending->anon_dev);
 	if (IS_ERR(pending->snap)) {
 		ret = PTR_ERR(pending->snap);
 		btrfs_abort_transaction(trans, ret);
@@ -1682,7 +1682,8 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 		btrfs_abort_transaction(trans, ret);
 		goto fail;
 	}
-	ret = btrfs_uuid_tree_add(trans, new_uuid.b, BTRFS_UUID_KEY_SUBVOL,
+	ret = btrfs_uuid_tree_add(trans, new_root_item->uuid,
+				  BTRFS_UUID_KEY_SUBVOL,
 				  objectid);
 	if (ret) {
 		btrfs_abort_transaction(trans, ret);
@@ -1794,7 +1795,8 @@ static void wait_current_trans_commit_start(struct btrfs_fs_info *fs_info,
 					    struct btrfs_transaction *trans)
 {
 	wait_event(fs_info->transaction_blocked_wait,
-		   trans->state >= TRANS_STATE_COMMIT_START || trans->aborted);
+		   trans->state >= TRANS_STATE_COMMIT_START ||
+		   TRANS_ABORTED(trans));
 }
 
 /*
@@ -1806,7 +1808,8 @@ static void wait_current_trans_commit_start_and_unblock(
 					struct btrfs_transaction *trans)
 {
 	wait_event(fs_info->transaction_wait,
-		   trans->state >= TRANS_STATE_UNBLOCKED || trans->aborted);
+		   trans->state >= TRANS_STATE_UNBLOCKED ||
+		   TRANS_ABORTED(trans));
 }
 
 /*
@@ -2026,7 +2029,7 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
 	trans->dirty = true;
 
 	/* Stop the commit early if ->aborted is set */
-	if (unlikely(READ_ONCE(cur_trans->aborted))) {
+	if (TRANS_ABORTED(cur_trans)) {
 		ret = cur_trans->aborted;
 		btrfs_end_transaction(trans);
 		return ret;
@@ -2100,7 +2103,7 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
 
 		wait_for_commit(cur_trans);
 
-		if (unlikely(cur_trans->aborted))
+		if (TRANS_ABORTED(cur_trans))
 			ret = cur_trans->aborted;
 
 		btrfs_put_transaction(cur_trans);
@@ -2119,7 +2122,7 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
 			spin_unlock(&fs_info->trans_lock);
 
 			wait_for_commit(prev_trans);
-			ret = prev_trans->aborted;
+			ret = READ_ONCE(prev_trans->aborted);
 
 			btrfs_put_transaction(prev_trans);
 			if (ret)
@@ -2173,8 +2176,7 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
 	wait_event(cur_trans->writer_wait,
 		   atomic_read(&cur_trans->num_writers) == 1);
 
-	/* ->aborted might be set after the previous check, so check it */
-	if (unlikely(READ_ONCE(cur_trans->aborted))) {
+	if (TRANS_ABORTED(cur_trans)) {
 		ret = cur_trans->aborted;
 		goto scrub_continue;
 	}
@@ -2191,10 +2193,8 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
 	 * core function of the snapshot creation.
 	 */
 	ret = create_pending_snapshots(trans);
-	if (ret) {
-		mutex_unlock(&fs_info->reloc_mutex);
-		goto scrub_continue;
-	}
+	if (ret)
+		goto unlock_reloc;
 
 	/*
 	 * We insert the dir indexes of the snapshots and update the inode
@@ -2207,16 +2207,12 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
 	 * the nodes and leaves.
 	 */
 	ret = btrfs_run_delayed_items(trans);
-	if (ret) {
-		mutex_unlock(&fs_info->reloc_mutex);
-		goto scrub_continue;
-	}
+	if (ret)
+		goto unlock_reloc;
 
 	ret = btrfs_run_delayed_refs(trans, (unsigned long)-1);
-	if (ret) {
-		mutex_unlock(&fs_info->reloc_mutex);
-		goto scrub_continue;
-	}
+	if (ret)
+		goto unlock_reloc;
 
 	/*
 	 * make sure none of the code above managed to slip in a
@@ -2242,11 +2238,8 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
 	mutex_lock(&fs_info->tree_log_mutex);
 
 	ret = commit_fs_roots(trans);
-	if (ret) {
-		mutex_unlock(&fs_info->tree_log_mutex);
-		mutex_unlock(&fs_info->reloc_mutex);
-		goto scrub_continue;
-	}
+	if (ret)
+		goto unlock_tree_log;
 
 	/*
 	 * Since the transaction is done, we can apply the pending changes
@@ -2264,39 +2257,28 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
 	 * new delayed refs. Must handle them or qgroup can be wrong.
 	 */
 	ret = btrfs_run_delayed_refs(trans, (unsigned long)-1);
-	if (ret) {
-		mutex_unlock(&fs_info->tree_log_mutex);
-		mutex_unlock(&fs_info->reloc_mutex);
-		goto scrub_continue;
-	}
+	if (ret)
+		goto unlock_tree_log;
 
 	/*
 	 * Since fs roots are all committed, we can get a quite accurate
 	 * new_roots. So let's do quota accounting.
 	 */
 	ret = btrfs_qgroup_account_extents(trans);
-	if (ret < 0) {
-		mutex_unlock(&fs_info->tree_log_mutex);
-		mutex_unlock(&fs_info->reloc_mutex);
-		goto scrub_continue;
-	}
+	if (ret < 0)
+		goto unlock_tree_log;
 
 	ret = commit_cowonly_roots(trans);
-	if (ret) {
-		mutex_unlock(&fs_info->tree_log_mutex);
-		mutex_unlock(&fs_info->reloc_mutex);
-		goto scrub_continue;
-	}
+	if (ret)
+		goto unlock_tree_log;
 
 	/*
 	 * The tasks which save the space cache and inode cache may also
 	 * update ->aborted, check it.
 	 */
-	if (unlikely(READ_ONCE(cur_trans->aborted))) {
+	if (TRANS_ABORTED(cur_trans)) {
 		ret = cur_trans->aborted;
-		mutex_unlock(&fs_info->tree_log_mutex);
-		mutex_unlock(&fs_info->reloc_mutex);
-		goto scrub_continue;
+		goto unlock_tree_log;
 	}
 
 	btrfs_prepare_extent_commit(fs_info);
@@ -2343,6 +2325,10 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
 	if (ret) {
 		btrfs_handle_fs_error(fs_info, ret,
 				      "Error while writing out transaction");
+		/*
+		 * reloc_mutex has been unlocked, tree_log_mutex is still held
+		 * but we can't jump to unlock_tree_log causing double unlock
+		 */
 		mutex_unlock(&fs_info->tree_log_mutex);
 		goto scrub_continue;
 	}
@@ -2368,7 +2354,6 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
 	 */
 	cur_trans->state = TRANS_STATE_COMPLETED;
 	wake_up(&cur_trans->commit_wait);
-	clear_bit(BTRFS_FS_NEED_ASYNC_COMMIT, &fs_info->flags);
 
 	spin_lock(&fs_info->trans_lock);
 	list_del_init(&cur_trans->list);
@@ -2391,6 +2376,10 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
 
 	return ret;
 
+unlock_tree_log:
+	mutex_unlock(&fs_info->tree_log_mutex);
+unlock_reloc:
+	mutex_unlock(&fs_info->reloc_mutex);
 scrub_continue:
 	btrfs_scrub_continue(fs_info);
 cleanup_transaction:
@@ -2434,13 +2423,18 @@ int btrfs_clean_one_deleted_snapshot(struct btrfs_root *root)
 	btrfs_debug(fs_info, "cleaner removing %llu", root->root_key.objectid);
 
 	btrfs_kill_all_delayed_nodes(root);
+	if (root->ino_cache_inode) {
+		iput(root->ino_cache_inode);
+		root->ino_cache_inode = NULL;
+	}
 
 	if (btrfs_header_backref_rev(root->node) <
 			BTRFS_MIXED_BACKREF_REV)
-		ret = btrfs_drop_snapshot(root, NULL, 0, 0);
+		ret = btrfs_drop_snapshot(root, 0, 0);
 	else
-		ret = btrfs_drop_snapshot(root, NULL, 1, 0);
+		ret = btrfs_drop_snapshot(root, 1, 0);
 
+	btrfs_put_root(root);
 	return (ret < 0) ? 0 : 1;
 }