[PATCH] f2fs: fix O_DIRECT cleanup range for append writes

[Seongjae Jeong <jsjlee1020@gmail.com>]

When an O_DIRECT write falls back to buffered I/O,
f2fs_flush_buffered_write() uses orig_pos to flush and invalidate
the page cache.

For O_APPEND writes, f2fs_write_checks() updates iocb->ki_pos to
the end of the file after orig_pos has been saved. As a result,
the cleanup range can differ from the actual buffered write range.

Save the write position after f2fs_write_checks() and use it for
the forced buffered I/O cleanup.

Fixes: 92318f20d703 ("f2fs: preserve direct write semantics when buffering is forced")
Signed-off-by: Seongjae Jeong <jsjlee1020@gmail.com>
---
 fs/f2fs/file.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c
index b99d9cdf9ba7..bf61545603e5 100644
--- a/fs/f2fs/file.c
+++ b/fs/f2fs/file.c
@@ -5313,6 +5313,7 @@ static ssize_t f2fs_file_write_iter(struct kiocb *iocb, struct iov_iter *from)
 	const loff_t pos = iocb->ki_pos;
 	const ssize_t count = iov_iter_count(from);
 	ssize_t ret;
+	loff_t bufio_start_pos;
 
 	if (unlikely(f2fs_cp_error(F2FS_I_SB(inode)))) {
 		ret = -EIO;
@@ -5343,6 +5344,8 @@ static ssize_t f2fs_file_write_iter(struct kiocb *iocb, struct iov_iter *from)
 	if (ret <= 0)
 		goto out_unlock;
 
+	bufio_start_pos = iocb->ki_pos;
+
 	/* Determine whether we will do a direct write or a buffered write. */
 	dio = f2fs_should_use_dio(inode, iocb, from);
 
@@ -5396,8 +5399,8 @@ static ssize_t f2fs_file_write_iter(struct kiocb *iocb, struct iov_iter *from)
 	 */
 	if (ret > 0 && !dio && (iocb->ki_flags & IOCB_DIRECT))
 		f2fs_flush_buffered_write(iocb->ki_filp->f_mapping,
-					  orig_pos,
-					  orig_pos + ret - 1);
+					  bufio_start_pos,
+					  bufio_start_pos + ret - 1);
 
 	return ret;
 }
-- 
2.53.0