[f2fs-dev] [PATCH 2/3] f2fs: avoid unneeded page release for correct _count of page

[Chao Yu]

In find_fsync_dnodes() and recover_data(), our flow is like this:

->f2fs_submit_page_bio()
        -> f2fs_put_page()
                -> page_cache_release() ---- page->_count declined to zero.
->__free_pages()
        -> put_page_testzero() ---- page->_count will be declined again.

We will get a segment fault in put_page_testzero when CONFIG_DEBUG_VM
is on, or return MM with a bad page with wrong _count num.

So let's just release this page.

Signed-off-by: Chao Yu <chao2...@samsung.com>
---
 fs/f2fs/recovery.c |    9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/fs/f2fs/recovery.c b/fs/f2fs/recovery.c
index 7dda1f28..d075465 100644
--- a/fs/f2fs/recovery.c
+++ b/fs/f2fs/recovery.c
@@ -145,7 +145,7 @@ static int find_fsync_dnodes(struct f2fs_sb_info *sbi, 
struct list_head *head)
 
                err = f2fs_submit_page_bio(sbi, page, blkaddr, READ_SYNC);
                if (err)
-                       goto out;
+                       return err;
 
                lock_page(page);
 
@@ -191,9 +191,10 @@ next:
                /* check next segment */
                blkaddr = next_blkaddr_of_node(page);
        }
+
        unlock_page(page);
-out:
        __free_pages(page, 0);
+
        return err;
 }
 
@@ -388,7 +389,7 @@ static int recover_data(struct f2fs_sb_info *sbi,
 
                err = f2fs_submit_page_bio(sbi, page, blkaddr, READ_SYNC);
                if (err)
-                       goto out;
+                       return err;
 
                lock_page(page);
 
@@ -412,8 +413,8 @@ next:
                /* check next segment */
                blkaddr = next_blkaddr_of_node(page);
        }
+
        unlock_page(page);
-out:
        __free_pages(page, 0);
 
        if (!err)
-- 
1.7.9.5

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majord...@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/