Btrfs devel - May 2011 - [BUG] btrfs 2.6.39(vanilla): files created with a particular pattern result in error 400

Hi, since I switched to vanilla kernel version 2.6.39(from latest
stable 2.6.38), a lot of errors started appearing on my btrfs
partitions, without any hardware failure, power outages or crashes.
The errors look all the same "root 5 inode <inode_number> errors
400"
for all the partitions. I''ve produced a little c program that is
consistently produces these seemingly faulty files on my system. I''ve
done so by tracing vuze(a torrent app), which I noticed was the app
that created those files. If I cp those files, the result doesn''t have
those errors.
the code creates files which result in this btrfsck output:

root 5 inode 705986 errors 400
found 339059101696 bytes used err is 1
total csum bytes: 330286628
total tree bytes: 845594624
total fs tree bytes: 250171392
btree space waste bytes: 191048782
file data blocks allocated: 402010976256
 referenced 331993735168

P.S. I''m not subscribed to the list.

The code is in the attachment and appended here(don''t know if gmail is
gonna mess it up, so I''m attacching it too):

#include <unistd.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <stdio.h>

#define FILE_SIZE 325443890
#define WRT_BYTES 4096

int main(void)
{
	struct stat stat_buf;
	int fd = 0, flag = 0;
	const size_t cnt = FILE_SIZE / WRT_BYTES;
	char arr[WRT_BYTES];
	
	fd = open("this_file_name.mp4",
		O_RDWR|O_CREAT, 0666);
	
	fstat(fd, &stat_buf);
	
	flag = fcntl(fd, F_GETFD);
	fcntl(fd, F_SETFD, flag | FD_CLOEXEC);
	
	lseek(fd, 0, SEEK_CUR);
	
	ftruncate(fd, FILE_SIZE);
	
	lseek(fd, 0, SEEK_SET);
	
	for (int i = 0; i < cnt; ++i) {
		write(fd, arr, WRT_BYTES);
	}
	
	write(fd, arr, FILE_SIZE - (WRT_BYTES * cnt));
	
	close(fd);
	
	return 0;
}

--
Giangiacomo