More Displaylink stuff
floe at butterbrot.org
Thu Dec 25 06:03:17 PST 2008
> So both encrypted VNC or RDP variants are candidates. It doesn't sound
> like VNC because you'd expect a series of small messages first - key
The communication with the device is basically one-way. Just about the
only thing which is sent back from the device is the monitor EDID, so
there's no kind of two-way handshake.
What I forgot to mention is that I have already successfully replayed a
recorded bunch of USB packets from the Windows driver using pyusb, and
have gotten a static image of my Windows desktop background. So there's
definitely no key negotiation.
>> However, assuming for a moment that they just took a big bunch of VNC code
>> and hacked their driver out of that, that could possibly be a reason.
> or if they are using standard RDP code from Windows likewise.
Heh, good point. What would be the licensing situation in that case?
> There are probably more samples of that in the Linux kernel. Lots of
> device firmware downloads are of them form
> [offset] [some fixed number of bytes] [checksum]
No, I believe it's not really a firmware. I really think it's the
DDR initialization (see below). The ASIC probably is not very
smart and does little more than decode packets and write them into
the framebuffer at the right location. I guess it contains a small
flash memory for the firmware.
> Do you know what chip is in the device ?
There are 3 chips inside:
- Chrontel CH7301 DVI Transmitter
- Hynix HY5DU281622 128 MBit DDR SDRAM
- DisplayLink DL120 (<- this one connects to USB)
The last one, of course, is the problem - it's just one huge ASIC and
there's no practical way to look inside (short of an electron microscope,
of course :-).
"_Nothing_ brightens up my morning. Coffee simply provides a shade of
grey just above the pitch-black of the infinite depths of the _abyss_."
More information about the xorg