[Btrfs-devel][PATCH]Fix u32 overflow in dirty_and_release_pages.

[Yan Zheng]

Hello,

When calculating the size of inline extent,  inode->i_size should also
be take into consideration, otherwise sys_write may drop some data
silently.  You can test this bug by:

#dd if=/dev/zero bs=4k count=1 of=test_file
#dd if=/dev/zero bs=2k count=1 of=test_file conv=notrunc

Previous patch is included in this one, please ignore it.

Regards
YZ
---
diff -r 2456df534a5f file.c
--- a/file.c	Thu Nov 01 19:45:34 2007 -0400
+++ b/file.c	Mon Nov 05 18:44:36 2007 +0800
@@ -239,7 +239,7 @@ static int dirty_and_release_pages(struc
 	u64 start_pos;
 	u64 end_of_last_block;
 	u64 end_pos = pos + write_bytes;
-	u32 inline_size;
+	u64 inline_size;
 	loff_t isize = i_size_read(inode);

 	em = alloc_extent_map(GFP_NOFS);
@@ -328,9 +328,11 @@ static int dirty_and_release_pages(struc
 					 aligned_end, aligned_end, &hint_byte);
 		if (err)
 			goto failed;
+		if (isize > inline_size)
+			inline_size = min_t(u64, isize, aligned_end);
+		inline_size -= start_pos;
 		err = insert_inline_extent(trans, root, inode, start_pos,
-					   end_pos - start_pos, pages, 0,
-					   num_pages);
+					   inline_size, pages, 0, num_pages);
 		BUG_ON(err);
 	}
 	if (end_pos > isize) {