ext4: fix data corruption in data=journal mode
authorJan Kara <jack@suse.cz>
Fri, 27 Jan 2017 19:35:38 +0000 (14:35 -0500)
committerTheodore Ts'o <tytso@mit.edu>
Fri, 27 Jan 2017 19:35:38 +0000 (14:35 -0500)
commit3b136499e906460919f0d21a49db1aaccf0ae963
tree9da71e2dcd4f4d90b21ed5ec7a4c9cb44c0b0194
parentcd648b8a8fd5071d232242d5ee7ee3c0815776af
ext4: fix data corruption in data=journal mode

ext4_journalled_write_end() did not propely handle all the cases when
generic_perform_write() did not copy all the data into the target page
and could mark buffers with uninitialized contents as uptodate and dirty
leading to possible data corruption (which would be quickly fixed by
generic_perform_write() retrying the write but still). Fix the problem
by carefully handling the case when the page that is written to is not
uptodate.

CC: stable@vger.kernel.org
Reported-by: Al Viro <viro@ZenIV.linux.org.uk>
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
fs/ext4/inode.c