diff options
| author | Chris Mason | 2008-07-22 17:18:08 +0200 |
|---|---|---|
| committer | Chris Mason | 2008-09-25 17:04:05 +0200 |
| commit | a61e6f29dc7c9d56a776a518eed92bbc61848263 (patch) | |
| tree | 6856ad3e5822eacfc2ac94693b6023ef81567fd2 /fs/btrfs/extent_io.c | |
| parent | Btrfs: Index extent buffers in an rbtree (diff) | |
| download | kernel-qcow2-linux-a61e6f29dc7c9d56a776a518eed92bbc61848263.tar.gz kernel-qcow2-linux-a61e6f29dc7c9d56a776a518eed92bbc61848263.tar.xz kernel-qcow2-linux-a61e6f29dc7c9d56a776a518eed92bbc61848263.zip | |
Btrfs: Use a mutex in the extent buffer for tree block locking
This replaces the use of the page cache lock bit for locking, which wasn't
suitable for block size < page size and couldn't be used recursively.
The mutexes alone don't fix either problem, but they are the first step.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Diffstat (limited to 'fs/btrfs/extent_io.c')
| -rw-r--r-- | fs/btrfs/extent_io.c | 9 |
1 files changed, 9 insertions, 0 deletions
diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c index 32bb4ed3723d..7380449cb5b3 100644 --- a/fs/btrfs/extent_io.c +++ b/fs/btrfs/extent_io.c @@ -2690,6 +2690,7 @@ static struct extent_buffer *__alloc_extent_buffer(struct extent_io_tree *tree, eb = kmem_cache_zalloc(extent_buffer_cache, mask); eb->start = start; eb->len = len; + mutex_init(&eb->mutex); spin_lock_irqsave(&leak_lock, flags); list_add(&eb->leak_list, &buffers); spin_unlock_irqrestore(&leak_lock, flags); @@ -2837,6 +2838,7 @@ int clear_extent_buffer_dirty(struct extent_io_tree *tree, for (i = 0; i < num_pages; i++) { page = extent_buffer_page(eb, i); + lock_page(page); if (i == 0) set_page_extent_head(page, eb->len); else @@ -2854,6 +2856,7 @@ int clear_extent_buffer_dirty(struct extent_io_tree *tree, end = start + PAGE_CACHE_SIZE - 1; if (test_range_bit(tree, start, end, EXTENT_DIRTY, 0)) { + unlock_page(page); continue; } } @@ -2865,6 +2868,7 @@ int clear_extent_buffer_dirty(struct extent_io_tree *tree, PAGECACHE_TAG_DIRTY); } read_unlock_irq(&page->mapping->tree_lock); + unlock_page(page); } return 0; } @@ -2893,12 +2897,17 @@ int set_extent_buffer_dirty(struct extent_io_tree *tree, * on us if the page isn't already dirty. */ if (i == 0) { + lock_page(page); set_page_extent_head(page, eb->len); } else if (PagePrivate(page) && page->private != EXTENT_PAGE_PRIVATE) { + lock_page(page); set_page_extent_mapped(page); + unlock_page(page); } __set_page_dirty_nobuffers(extent_buffer_page(eb, i)); + if (i == 0) + unlock_page(page); } return set_extent_dirty(tree, eb->start, eb->start + eb->len - 1, GFP_NOFS); |