f2fs: fix to recover i_size from roll-forward

If user requests many data writes and fsync together, the last updated i_size
should be stored to the inode block consistently.

But, previous write_end just marks the inode as dirty and doesn't update its
metadata into its inode block.
After that, fsync just writes the inode block with newly updated data index
excluding inode metadata updates.

So, this patch introduces write_end in which updates inode block too when the
i_size is changed.

Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>

fs/f2fs/data.c

@@ -701,6 +701,27 @@ static int f2fs_write_begin(struct file *file, struct address_space *mapping,
 	return err;
 }
 
+static int f2fs_write_end(struct file *file,
+			struct address_space *mapping,
+			loff_t pos, unsigned len, unsigned copied,
+			struct page *page, void *fsdata)
+{
+	struct inode *inode = page->mapping->host;
+
+	SetPageUptodate(page);
+	set_page_dirty(page);
+
+	if (pos + copied > i_size_read(inode)) {
+		i_size_write(inode, pos + copied);
+		mark_inode_dirty(inode);
+		update_inode_page(inode);
+	}
+
+	unlock_page(page);
+	page_cache_release(page);
+	return copied;
+}
+
 static ssize_t f2fs_direct_IO(int rw, struct kiocb *iocb,
 		const struct iovec *iov, loff_t offset, unsigned long nr_segs)
 {
@@ -757,7 +778,7 @@ const struct address_space_operations f2fs_dblock_aops = {
 	.writepage	= f2fs_write_data_page,
 	.writepages	= f2fs_write_data_pages,
 	.write_begin	= f2fs_write_begin,
-	.write_end	= nobh_write_end,
+	.write_end	= f2fs_write_end,
 	.set_page_dirty	= f2fs_set_data_page_dirty,
 	.invalidatepage	= f2fs_invalidate_data_page,
 	.releasepage	= f2fs_release_data_page,