ext4: don't keep using page if inline conversion fails
authorDarrick J. Wong <darrick.wong@oracle.com>
Thu, 11 Sep 2014 15:45:12 +0000 (11:45 -0400)
committerTheodore Ts'o <tytso@mit.edu>
Thu, 11 Sep 2014 15:45:12 +0000 (11:45 -0400)
If inline->extent conversion fails (most probably due to ENOSPC) and
we release the temporary page that we allocated to transfer the file
contents, don't keep using the page pointer after releasing the page.
This occasionally leads to complaints about evicting locked pages or
hangs when blocksize > pagesize, because it's possible for the page to
get reallocated elsewhere in the meantime.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: Tao Ma <tm@tao.ma>
fs/ext4/inline.c

index bea662bd0ca6b921b15c1bda0aab950a2fec2c78..378aadf5e6db6bc3c9a92fd2e8d9561af382328a 100644 (file)
@@ -594,6 +594,7 @@ retry:
        if (ret) {
                unlock_page(page);
                page_cache_release(page);
+               page = NULL;
                ext4_orphan_add(handle, inode);
                up_write(&EXT4_I(inode)->xattr_sem);
                sem_held = 0;
@@ -613,7 +614,8 @@ retry:
        if (ret == -ENOSPC && ext4_should_retry_alloc(inode->i_sb, &retries))
                goto retry;
 
-       block_commit_write(page, from, to);
+       if (page)
+               block_commit_write(page, from, to);
 out:
        if (page) {
                unlock_page(page);
This page took 0.029721 seconds and 5 git commands to generate.