yv12_to_rgb24_mmx seems to write too much...

Before all, I would like to thanks xvid programmers, it's a very good 
implementation of MPEG4, well documented, and the code is very clean !

(perhaps the only remarks: why not using C++ ? (C++ is not so much slow))

Well, i've created a class that encapsulates the core functions, and all 
works very well.

I encode a set of bitmaps RGB24 in a file, and decode this file.
But the function yv12_to_rgb24_mmx writes 1 octet after the output array !
The pointer used by this function (dst) is allocated like this (in another 
part of the code):
m_pdata_out	= new unsigned char[m_Width * m_Height * 3];

All the octets of this array are well written, but the octet 'm_pdata_out+ 
m_Width * m_Height * 3' is also written !
(and in VC++ 6, I can see the famous four FDFDFDFD become 00FDFDFD)

Futhermore yv12_to_rgb24_c works very well, and doesn't pass over the array.

So I think it's a bug.
Else, tell me were I'm wrong.

Thanks.