[LinuxBIOS] More USB Host to Host Adapters

Peter Stuge stuge-linuxbios at cdy.org
Fri Dec 1 11:36:14 CET 2006


On Fri, Dec 01, 2006 at 11:14:44AM +0100, Stefan Reinauer wrote:
> * Bari Ari <bari at onelabs.com> [061201 02:56]:
> > Lu, Yinghai wrote:
> > > There should not work as USB debug device. I guess.
> > 
> > Lack of Linux support seems to be the problem.
> > 
> > I've pulled a couple apart and they use the PLX or Maxim USB
> > controllers with a small micro to pass the data back and forth.

True, but they also enumerate as completely different types of
devices.


> the usb host to host adapter also needs to explixitly support the
> USB debug interface.

Yes, specifically the device must comply to the Debug Class Device
Specification.

This is what I learned by taking apart the PLX debug device at the
symposium:

Two NET2270 chips, controlled by a UBICOM micro. I would assume the
micro does only basic setup and then the 2270s just talk to each
other.

The device has two quite different USB interfaces. One is for the
host system, the other is for the target (AKA remote in debug class
terminology).

The target interface does not work as a normal USB interface, it will
not enumerate if plugged in to a USB controller in normal mode.

The host interface works as a normal USB interface, with one IN and
one OUT endpoint. (Bulk IIRC, but see debug device class spec.)


The target end of the device will only work when connected to a USB
controller that is in debug mode.


No other devices than debug class devices work when the USB
controller is in debug mode.


I sent a probably-working program to the list a while back that uses
libusb to access the host end of the device. It only handles reads
from the target. I'll add it to trac.

Stefan, you're in a unique position to test my code and Eric's code
for driving the target end of the device, along with your device. :)
Please report any progress.


//Peter




More information about the coreboot mailing list