Btrfs: leave btree locks spinning more often

btrfs_mark_buffer dirty would set dirty bits in the extent_io tree
for the buffers it was dirtying.  This may require a kmalloc and it
was not atomic.  So, anyone who called btrfs_mark_buffer_dirty had to
set any btree locks they were holding to blocking first.

This commit changes dirty tracking for extent buffers to just use a flag
in the extent buffer.  Now that we have one and only one extent buffer
per page, this can be safely done without losing dirty bits along the way.

This also introduces a path->leave_spinning flag that callers of
btrfs_search_slot can use to indicate they will properly deal with a
path returned where all the locks are spinning instead of blocking.

Many of the btree search callers now expect spinning paths,
resulting in better btree concurrency overall.

Signed-off-by: Chris Mason <chris.mason@oracle.com>

1 files changed, 54 insertions, 13 deletions

diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index 1f1d89b18818..9244cd7313d4 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -668,14 +668,31 @@ static int btree_submit_bio_hook(struct inode *inode, int rw, struct bio *bio,
 static int btree_writepage(struct page *page, struct writeback_control *wbc)
 {
 	struct extent_io_tree *tree;
+	struct btrfs_root *root = BTRFS_I(page->mapping->host)->root;
+	struct extent_buffer *eb;
+	int was_dirty;
+
 	tree = &BTRFS_I(page->mapping->host)->io_tree;
+	if (!(current->flags & PF_MEMALLOC)) {
+		return extent_write_full_page(tree, page,
+					      btree_get_extent, wbc);
+	}
 
-	if (current->flags & PF_MEMALLOC) {
-		redirty_page_for_writepage(wbc, page);
-		unlock_page(page);
-		return 0;
+	redirty_page_for_writepage(wbc, page);
+	eb = btrfs_find_tree_block(root, page_offset(page),
+				      PAGE_CACHE_SIZE);
+	WARN_ON(!eb);
+
+	was_dirty = test_and_set_bit(EXTENT_BUFFER_DIRTY, &eb->bflags);
+	if (!was_dirty) {
+		spin_lock(&root->fs_info->delalloc_lock);
+		root->fs_info->dirty_metadata_bytes += PAGE_CACHE_SIZE;
+		spin_unlock(&root->fs_info->delalloc_lock);
 	}
-	return extent_write_full_page(tree, page, btree_get_extent, wbc);
+	free_extent_buffer(eb);
+
+	unlock_page(page);
+	return 0;
 }
 
 static int btree_writepages(struct address_space *mapping,
@@ -684,15 +701,15 @@ static int btree_writepages(struct address_space *mapping,
 	struct extent_io_tree *tree;
 	tree = &BTRFS_I(mapping->host)->io_tree;
 	if (wbc->sync_mode == WB_SYNC_NONE) {
+		struct btrfs_root *root = BTRFS_I(mapping->host)->root;
 		u64 num_dirty;
-		u64 start = 0;
 		unsigned long thresh = 32 * 1024 * 1024;
 
 		if (wbc->for_kupdate)
 			return 0;
 
-		num_dirty = count_range_bits(tree, &start, (u64)-1,
-					     thresh, EXTENT_DIRTY);
+		/* this is a bit racy, but that's ok */
+		num_dirty = root->fs_info->dirty_metadata_bytes;
 		if (num_dirty < thresh)
 			return 0;
 	}
@@ -859,9 +876,17 @@ int clean_tree_block(struct btrfs_trans_handle *trans, struct btrfs_root *root,
 	    root->fs_info->running_transaction->transid) {
 		btrfs_assert_tree_locked(buf);
 
-		/* ugh, clear_extent_buffer_dirty can be expensive */
-		btrfs_set_lock_blocking(buf);
+		if (test_and_clear_bit(EXTENT_BUFFER_DIRTY, &buf->bflags)) {
+			spin_lock(&root->fs_info->delalloc_lock);
+			if (root->fs_info->dirty_metadata_bytes >= buf->len)
+				root->fs_info->dirty_metadata_bytes -= buf->len;
+			else
+				WARN_ON(1);
+			spin_unlock(&root->fs_info->delalloc_lock);
+		}
 
+		/* ugh, clear_extent_buffer_dirty needs to lock the page */
+		btrfs_set_lock_blocking(buf);
 		clear_extent_buffer_dirty(&BTRFS_I(btree_inode)->io_tree,
 					  buf);
 	}
@@ -2348,8 +2373,7 @@ void btrfs_mark_buffer_dirty(struct extent_buffer *buf)
 	struct btrfs_root *root = BTRFS_I(buf->first_page->mapping->host)->root;
 	u64 transid = btrfs_header_generation(buf);
 	struct inode *btree_inode = root->fs_info->btree_inode;
-
-	btrfs_set_lock_blocking(buf);
+	int was_dirty;
 
 	btrfs_assert_tree_locked(buf);
 	if (transid != root->fs_info->generation) {
@@ -2360,7 +2384,13 @@ void btrfs_mark_buffer_dirty(struct extent_buffer *buf)
 			(unsigned long long)root->fs_info->generation);
 		WARN_ON(1);
 	}
-	set_extent_buffer_dirty(&BTRFS_I(btree_inode)->io_tree, buf);
+	was_dirty = set_extent_buffer_dirty(&BTRFS_I(btree_inode)->io_tree,
+					    buf);
+	if (!was_dirty) {
+		spin_lock(&root->fs_info->delalloc_lock);
+		root->fs_info->dirty_metadata_bytes += buf->len;
+		spin_unlock(&root->fs_info->delalloc_lock);
+	}
 }
 
 void btrfs_btree_balance_dirty(struct btrfs_root *root, unsigned long nr)
@@ -2400,6 +2430,7 @@ int btrfs_read_buffer(struct extent_buffer *buf, u64 parent_transid)
 int btree_lock_page_hook(struct page *page)
 {
 	struct inode *inode = page->mapping->host;
+	struct btrfs_root *root = BTRFS_I(inode)->root;
 	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
 	struct extent_buffer *eb;
 	unsigned long len;
@@ -2415,6 +2446,16 @@ int btree_lock_page_hook(struct page *page)
 
 	btrfs_tree_lock(eb);
 	btrfs_set_header_flag(eb, BTRFS_HEADER_FLAG_WRITTEN);
+
+	if (test_and_clear_bit(EXTENT_BUFFER_DIRTY, &eb->bflags)) {
+		spin_lock(&root->fs_info->delalloc_lock);
+		if (root->fs_info->dirty_metadata_bytes >= eb->len)
+			root->fs_info->dirty_metadata_bytes -= eb->len;
+		else
+			WARN_ON(1);
+		spin_unlock(&root->fs_info->delalloc_lock);
+	}
+
 	btrfs_tree_unlock(eb);
 	free_extent_buffer(eb);
 out: