Re: iPAQ and USB cryptographic tokens?

Ooops! I addressed this just to Martin the first time! (Sorry Martin.
You get to see this twice!)

Martin Heimes wrote:
>
> > We'd like to use USB-connected cryptographic smart tokens with things
> > like the iPAQ or the Sharp Zaurus. From a brief study of the USB spec,
> > it looks as if one could easily hack a cable to connect the handheld
> > to a female "A" connector into which a token could be plugged. The
> > question is whether the USB driver software under Linux would know
> > what to do. Has anybody tried this? Is there any hope?
>
> USB is not a point-to-point, that is AFAIK the main problem. The ipaq is a
> USB slave from the hardware side of view, so if your USB crypto token provides
> the USB host things might work. OTOH if you find a PCMCIA (attention: no
> cardbus type will work in the ipaq sleeves) or CF USB card you get the required
> (hardware) USB host function; and you still need a driver for that card.
> That's all i know about, see
> http://pcmcia-cs.sourceforge.net/ftp/doc/PCMCIA-HOWTO.html and http://www.linux-usb.org/

Yes, but a USB slave can both read and write on the wire. Assuming
that we are controlling both the horizontal and the vertical here, what
stops two slaves from talking to each other? What stops the iPAQ from
pretending to be a master? Is there some bit of hardware that only
the master has?

I should probably just go wade through the 700 pages of the USB spec
to answer my questions. If nobody has been here before me, I guess
that's what I'll have to do. Oh, woe is me! :-)

Paul Allen

-- 
Boeing Phantom Works                   \ Paul L. Allen, (425) 865-3297
Math & Computing Technology              \ paul.l.allen@boeing.com
POB 3707 M/S 7L-40, Seattle, WA 98124-2207 \ Prototype Systems Group