f2fs: keep isize once block is reserved cross EOF
authorChao Yu <yuchao0@huawei.com>
Sun, 5 Nov 2017 13:53:30 +0000 (21:53 +0800)
committerJaegeuk Kim <jaegeuk@kernel.org>
Tue, 19 Dec 2017 03:38:30 +0000 (19:38 -0800)
Without FADVISE_KEEP_SIZE_BIT, we will try to recover file size
according to last non-hole block, so in fallocate(), we must set
FADVISE_KEEP_SIZE_BIT flag once we have preallocated block cross
EOF, instead of when all preallocation is success. Otherwise, file
size will be incorrect due to lack of this flag.

Simple testcase to reproduce this:

1. echo 2 > /sys/fs/f2fs/<device>/inject_type
2. echo 10 > /sys/fs/f2fs/<device>/inject_rate
3. run tests/generic/392
4. disable fault injection
5. do remount

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

index 6034f693f92fa12dd930019748ca07cb2f763c5b..ab481f63b8b189143742d2a4a1cbbfa8db4fb2f9 100644 (file)
@@ -1473,8 +1473,12 @@ static int expand_inode_data(struct inode *inode, loff_t offset,
                new_size = ((loff_t)pg_end << PAGE_SHIFT) + off_end;
        }
 
-       if (!(mode & FALLOC_FL_KEEP_SIZE) && i_size_read(inode) < new_size)
-               f2fs_i_size_write(inode, new_size);
+       if (new_size > i_size_read(inode)) {
+               if (mode & FALLOC_FL_KEEP_SIZE)
+                       file_set_keep_isize(inode);
+               else
+                       f2fs_i_size_write(inode, new_size);
+       }
 
        return err;
 }
@@ -1521,8 +1525,6 @@ static long f2fs_fallocate(struct file *file, int mode,
        if (!ret) {
                inode->i_mtime = inode->i_ctime = current_time(inode);
                f2fs_mark_inode_dirty_sync(inode, false);
-               if (mode & FALLOC_FL_KEEP_SIZE)
-                       file_set_keep_isize(inode);
                f2fs_update_time(F2FS_I_SB(inode), REQ_TIME);
        }