Just as a quick note for people who haven't been following the drivers
forum, we've beefed up some of the USB infrastructure (USB interfaces will
now detach themselves when they're removed) and provided a porting doc for
those who want to try and port a driver from NetBSD,

 

Porting doc is here:

 

(http://community.qnx.com/integration/viewcvs/viewcvs.cgi/trunk/sys/dev/doc/
port_bsd_usb.txt?root=core_networking
<http://community.qnx.com/integration/viewcvs/viewcvs.cgi/trunk/sys/dev/doc/
port_bsd_usb.txt?root=core_networking&rev=197&system=exsy1001&view=log>
&rev=197&system=exsy1001&view=log)

 

NetBSD driver source is here (look for the if_* source (driver files)).
Looks like there are quite a few drivers that could be ported over (both
wired and wireless)

 

http://cvsweb.netbsd.org/bsdweb.cgi/src/sys/dev/usb/?only_with_tag=netbsd-4-
0-RELEASE
<http://cvsweb.netbsd.org/bsdweb.cgi/src/sys/dev/usb/?only_with_tag=netbsd-4
-0-RELEASE> 

 

            The USB porting doc is based upon what was done to make the ASIX
driver (devnp-axe.so) operate. Hopefully, these drivers will now be as
simple to port as the PCI ones.

 

The io-net drivers for USB (devn-asix.so and devn-pegasus.so) unfortunately
trap insertion events and then attempt to register the device from a thread
that isn't a stack thread.  This means that they won't work unless the
device is already plugged in and detected when io-pkt comes up.  The en?
interface also isn't automatically detached on removal (need to do an
"ifconfig en? destroy" to get rid of it after the device is removed).

 

            Note that io-pkt still requires a separate USB enumerator (which
could be an enumerator of the human kind :>) to mount the appropriate driver
when a device is inserted.

 

Let us know if you have any feedback (positive OR negative).

 

                        Robert.