f2fs: don't trigger read IO for beyond EOF page
authorChao Yu <yuchao0@huawei.com>
Thu, 7 Mar 2019 09:31:30 +0000 (17:31 +0800)
committerJaegeuk Kim <jaegeuk@kernel.org>
Wed, 13 Mar 2019 01:59:19 +0000 (18:59 -0700)
In f2fs_mpage_readpages(), if page is beyond EOF, we should just
zero out it, but previously, before checking previous mapping
info, we missed to check filesize boundary, fix it.

Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
fs/f2fs/data.c

index fa6318549af582e034bfe62aa68b7335148a7d9b..912fdcc2dbb8a5a90bfb8dfb937388f90a5c2a39 100644 (file)
@@ -1551,6 +1551,9 @@ static int f2fs_mpage_readpages(struct address_space *mapping,
                if (last_block > last_block_in_file)
                        last_block = last_block_in_file;
 
+               /* just zeroing out page which is beyond EOF */
+               if (block_in_file >= last_block)
+                       goto zero_out;
                /*
                 * Map blocks using the previous result first.
                 */
@@ -1563,16 +1566,11 @@ static int f2fs_mpage_readpages(struct address_space *mapping,
                 * Then do more f2fs_map_blocks() calls until we are
                 * done with this page.
                 */
-               map.m_flags = 0;
+               map.m_lblk = block_in_file;
+               map.m_len = last_block - block_in_file;
 
-               if (block_in_file < last_block) {
-                       map.m_lblk = block_in_file;
-                       map.m_len = last_block - block_in_file;
-
-                       if (f2fs_map_blocks(inode, &map, 0,
-                                               F2FS_GET_BLOCK_DEFAULT))
-                               goto set_error_page;
-               }
+               if (f2fs_map_blocks(inode, &map, 0, F2FS_GET_BLOCK_DEFAULT))
+                       goto set_error_page;
 got_it:
                if ((map.m_flags & F2FS_MAP_MAPPED)) {
                        block_nr = map.m_pblk + block_in_file - map.m_lblk;
@@ -1587,6 +1585,7 @@ got_it:
                                                                DATA_GENERIC))
                                goto set_error_page;
                } else {
+zero_out:
                        zero_user_segment(page, 0, PAGE_SIZE);
                        if (!PageUptodate(page))
                                SetPageUptodate(page);