Btrfs: write out free space cache

This is a simple bit, just dump the free space cache out to our preallocated
inode when we're writing out dirty block groups.  There are a bunch of changes
in inode.c in order to account for special cases.  Mostly when we're doing the
writeout we're holding trans_mutex, so we need to use the nolock transacation
functions.  Also we can't do asynchronous completions since the async thread
could be blocked on already completed IO waiting for the transaction lock.  This
has been tested with xfstests and btrfs filesystem balance, as well as my ENOSPC
tests.  Thanks,

Signed-off-by: Josef Bacik <josef@redhat.com>

1 files changed, 48 insertions, 12 deletions

diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 1af1ea88e8a8..f2fb974ed8f0 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -764,6 +764,7 @@ static noinline int cow_file_range(struct inode *inode,
 	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
 	int ret = 0;
 
+	BUG_ON(root == root->fs_info->tree_root);
 	trans = btrfs_join_transaction(root, 1);
 	BUG_ON(!trans);
 	btrfs_set_trans_block_group(trans, inode);
@@ -1035,10 +1036,16 @@ static noinline int run_delalloc_nocow(struct inode *inode,
 	int type;
 	int nocow;
 	int check_prev = 1;
+	bool nolock = false;
 
 	path = btrfs_alloc_path();
 	BUG_ON(!path);
-	trans = btrfs_join_transaction(root, 1);
+	if (root == root->fs_info->tree_root) {
+		nolock = true;
+		trans = btrfs_join_transaction_nolock(root, 1);
+	} else {
+		trans = btrfs_join_transaction(root, 1);
+	}
 	BUG_ON(!trans);
 
 	cow_start = (u64)-1;
@@ -1211,8 +1218,13 @@ out_check:
 		BUG_ON(ret);
 	}
 
-	ret = btrfs_end_transaction(trans, root);
-	BUG_ON(ret);
+	if (nolock) {
+		ret = btrfs_end_transaction_nolock(trans, root);
+		BUG_ON(ret);
+	} else {
+		ret = btrfs_end_transaction(trans, root);
+		BUG_ON(ret);
+	}
 	btrfs_free_path(path);
 	return 0;
 }
@@ -1289,6 +1301,8 @@ static int btrfs_set_bit_hook(struct inode *inode,
 	if (!(state->state & EXTENT_DELALLOC) && (*bits & EXTENT_DELALLOC)) {
 		struct btrfs_root *root = BTRFS_I(inode)->root;
 		u64 len = state->end + 1 - state->start;
+		int do_list = (root->root_key.objectid !=
+			       BTRFS_ROOT_TREE_OBJECTID);
 
 		if (*bits & EXTENT_FIRST_DELALLOC)
 			*bits &= ~EXTENT_FIRST_DELALLOC;
@@ -1298,7 +1312,7 @@ static int btrfs_set_bit_hook(struct inode *inode,
 		spin_lock(&root->fs_info->delalloc_lock);
 		BTRFS_I(inode)->delalloc_bytes += len;
 		root->fs_info->delalloc_bytes += len;
-		if (list_empty(&BTRFS_I(inode)->delalloc_inodes)) {
+		if (do_list && list_empty(&BTRFS_I(inode)->delalloc_inodes)) {
 			list_add_tail(&BTRFS_I(inode)->delalloc_inodes,
 				      &root->fs_info->delalloc_inodes);
 		}
@@ -1321,6 +1335,8 @@ static int btrfs_clear_bit_hook(struct inode *inode,
 	if ((state->state & EXTENT_DELALLOC) && (*bits & EXTENT_DELALLOC)) {
 		struct btrfs_root *root = BTRFS_I(inode)->root;
 		u64 len = state->end + 1 - state->start;
+		int do_list = (root->root_key.objectid !=
+			       BTRFS_ROOT_TREE_OBJECTID);
 
 		if (*bits & EXTENT_FIRST_DELALLOC)
 			*bits &= ~EXTENT_FIRST_DELALLOC;
@@ -1330,14 +1346,15 @@ static int btrfs_clear_bit_hook(struct inode *inode,
 		if (*bits & EXTENT_DO_ACCOUNTING)
 			btrfs_delalloc_release_metadata(inode, len);
 
-		if (root->root_key.objectid != BTRFS_DATA_RELOC_TREE_OBJECTID)
+		if (root->root_key.objectid != BTRFS_DATA_RELOC_TREE_OBJECTID
+		    && do_list)
 			btrfs_free_reserved_data_space(inode, len);
 
 		spin_lock(&root->fs_info->delalloc_lock);
 		root->fs_info->delalloc_bytes -= len;
 		BTRFS_I(inode)->delalloc_bytes -= len;
 
-		if (BTRFS_I(inode)->delalloc_bytes == 0 &&
+		if (do_list && BTRFS_I(inode)->delalloc_bytes == 0 &&
 		    !list_empty(&BTRFS_I(inode)->delalloc_inodes)) {
 			list_del_init(&BTRFS_I(inode)->delalloc_inodes);
 		}
@@ -1426,7 +1443,10 @@ static int btrfs_submit_bio_hook(struct inode *inode, int rw, struct bio *bio,
 
 	skip_sum = BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM;
 
-	ret = btrfs_bio_wq_end_io(root->fs_info, bio, 0);
+	if (root == root->fs_info->tree_root)
+		ret = btrfs_bio_wq_end_io(root->fs_info, bio, 2);
+	else
+		ret = btrfs_bio_wq_end_io(root->fs_info, bio, 0);
 	BUG_ON(ret);
 
 	if (!(rw & REQ_WRITE)) {
@@ -1662,6 +1682,7 @@ static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end)
 	struct extent_state *cached_state = NULL;
 	int compressed = 0;
 	int ret;
+	bool nolock = false;
 
 	ret = btrfs_dec_test_ordered_pending(inode, &ordered_extent, start,
 					     end - start + 1);
@@ -1669,11 +1690,17 @@ static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end)
 		return 0;
 	BUG_ON(!ordered_extent);
 
+	nolock = (root == root->fs_info->tree_root);
+
 	if (test_bit(BTRFS_ORDERED_NOCOW, &ordered_extent->flags)) {
 		BUG_ON(!list_empty(&ordered_extent->list));
 		ret = btrfs_ordered_update_i_size(inode, 0, ordered_extent);
 		if (!ret) {
-			trans = btrfs_join_transaction(root, 1);
+			if (nolock)
+				trans = btrfs_join_transaction_nolock(root, 1);
+			else
+				trans = btrfs_join_transaction(root, 1);
+			BUG_ON(!trans);
 			btrfs_set_trans_block_group(trans, inode);
 			trans->block_rsv = &root->fs_info->delalloc_block_rsv;
 			ret = btrfs_update_inode(trans, root, inode);
@@ -1686,7 +1713,10 @@ static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end)
 			 ordered_extent->file_offset + ordered_extent->len - 1,
 			 0, &cached_state, GFP_NOFS);
 
-	trans = btrfs_join_transaction(root, 1);
+	if (nolock)
+		trans = btrfs_join_transaction_nolock(root, 1);
+	else
+		trans = btrfs_join_transaction(root, 1);
 	btrfs_set_trans_block_group(trans, inode);
 	trans->block_rsv = &root->fs_info->delalloc_block_rsv;
 
@@ -1725,9 +1755,15 @@ static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end)
 	ret = btrfs_update_inode(trans, root, inode);
 	BUG_ON(ret);
 out:
-	btrfs_delalloc_release_metadata(inode, ordered_extent->len);
-	if (trans)
-		btrfs_end_transaction(trans, root);
+	if (nolock) {
+		if (trans)
+			btrfs_end_transaction_nolock(trans, root);
+	} else {
+		btrfs_delalloc_release_metadata(inode, ordered_extent->len);
+		if (trans)
+			btrfs_end_transaction(trans, root);
+	}
+
 	/* once for us */
 	btrfs_put_ordered_extent(ordered_extent);
 	/* once for the tree */