From: Namjae Jeon <namjae.jeon@samsung.com>
Date: Fri, 3 Jan 2014 03:10:25 +0000 (+1100)
Subject: fat: fallback to buffered write in case of fallocatded region on direct IO
X-Git-Tag: next-20140106~2^2~51
X-Git-Url: https://git.karo-electronics.de/?a=commitdiff_plain;h=d8feeea359a9b5e06e099f8955bc0390d696e6de;p=karo-tx-linux.git

fat: fallback to buffered write in case of fallocatded region on direct IO

For normal cases of direct IO write, trying to seek to location greater
than file size, makes it fall back to buffered write to fill that region.
Similarly, in case for write in Fallocated region, make it fall to
buffered write.

Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Amit Sahrawat <a.sahrawat@samsung.com>
Cc: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
---

diff --git a/fs/fat/inode.c b/fs/fat/inode.c
index 7a547ee0d567..ffa2eccc62a2 100644
--- a/fs/fat/inode.c
+++ b/fs/fat/inode.c
@@ -218,6 +218,14 @@ static ssize_t fat_direct_IO(int rw, struct kiocb *iocb,
 		loff_t size = offset + iov_length(iov, nr_segs);
 		if (MSDOS_I(inode)->mmu_private < size)
 			return 0;
+
+		/*
+		 * In case of writing in fallocated region, return 0 and
+		 * fallback to buffered write.
+		 */
+		if (MSDOS_I(inode)->i_disksize >
+		    round_up(i_size_read(inode), inode->i_sb->s_blocksize))
+			return 0;
 	}
 
 	/*