Btrfs: fix btrfs_release_extent_buffer_page with the right usage of num_extent_pages
authorWang Sheng-Hui <shhuiw@gmail.com>
Fri, 6 Apr 2012 06:35:31 +0000 (14:35 +0800)
committerJosef Bacik <josef@redhat.com>
Fri, 11 May 2012 14:56:38 +0000 (10:56 -0400)
num_extent_pages returns the number of pages in the specific range, not
the index of the last page in the eb range.

btrfs_release_extent_buffer_page is called with start_idx set 0 in current
codes, so it's not a problem yet. But the logic is indeed wrong.

Fix it here.

Signed-off-by: Wang Sheng-Hui <shhuiw@gmail.com>
fs/btrfs/extent_io.c

index aeb98ceda51cc8ba116c540f377dc67eca882366..455daec2b6cefc329dd1ae63a27d1b2de8f524d9 100644 (file)
@@ -3981,11 +3981,13 @@ static void btrfs_release_extent_buffer_page(struct extent_buffer *eb,
                                                unsigned long start_idx)
 {
        unsigned long index;
+       unsigned long num_pages;
        struct page *page;
 
        BUG_ON(extent_buffer_under_io(eb));
 
-       index = num_extent_pages(eb->start, eb->len);
+       num_pages = num_extent_pages(eb->start, eb->len);
+       index = start_idx + num_pages;
        if (start_idx >= index)
                return;