On Tue, 7 Jul 2009 12:36:31 +0100 Ethan Grammatikidis <eeke...@fastmail.fm> wrote:
> On Mon, 6 Jul 2009 23:50:14 -0400 > erik quanstrom <quans...@quanstro.net> wrote: > > > On Mon Jul 6 19:41:36 EDT 2009, jrm8...@gmail.com wrote: > > > On Mon, Jul 6, 2009 at 7:37 PM, ron minnich<rminn...@gmail.com> wrote: > > > > I just ported the linux driver > > > > > > I'm interested in how hard this is, and how it might be made easier. > > > > that depends. i've found that porting a driver or working backwards > > from an example is often harder than just writing a new driver. > > this is because the hard part is understanding how the hardware works, > > not in coding that knowledge up. and unfortunately, looking at a > > linux driver hasn't been very instructive to me. > > > > i'm sure one could create a compatability layer for certain driver types > > along the lines of ndiswrapper. but given the instability of linux > > internal interfaces, this might be finished about the same time as > > duke nukem forever. > > What about emulating a usb-ethernet device in software? (Assuming the beagle > board's usb interface can operate in gadget mode as opposed to host mode.) > The Linux kernel does a good job of emulating a usb ethernet device, I use it > extensively. I rather assumed the Bitsy port had this already, but perhaps > not. Huh, I just realised that if both ends of the fake ethernet link run Plan 9 then the only advantage of what I suggested would be that both ends of the link are known quantities with open source code. Now I wonder if it might be simpler to implement 9p directly over USB. -- Ethan Grammatikidis Those who are slower at parsing information must necessarily be faster at problem-solving.
