On Tue, 2002-11-12 at 04:55, V Ganesh wrote:
> I submitted this fix to greg quite a while ago (august 16th to be precise)
> and it was merged into greg's 2.4 and 2.5 trees. I guess the problem is that
> there hasn't been a 2.4.20 yet, so most people haven't seen it.

Ok, good!

> apologies for slack email response - this was the holiday season in india.

Ah, no problems! I just have gained some momentum with my involvment in
the SynCE project, that's why I'm releasing and mailing and things...
:-)
 
> while we're on the topic, I'd appreciate some advice on whether to add module
> options (a la usbserial) to take the usb vendor and device id. this seems useful
> so people can test new and unheard of wince devices without having to patch
> the kernel.
> 
> OTOH, wes ([EMAIL PROTECTED]) figured out that the wceusb.inf
> file, which is part of the activesync package, contained the usb ids of all
> activesync-supported devices. the ipaq driver most likely supports
> all of them, since the activesync protocol is probably identical for all of
> them.
> there are a _ton_ of them in activesync 3.6, 287 to be precise. these fall broadly 
>into two categories:
> assorted vendors:     30
> microsoft:            257
> there's a contiguous range from vid=0x45e,prod=0x400 to 0x4ff.

Credits to this guy with the long e-mail address! :-) Looking in the
ActiveSync distribution, the MS Windows driver is of course called
wceusbsh.sys. Sounds like "wceusbsh" would be a better name than "ipaq
for the Linux driver? The suffix "sh" seem to mean "serial host".

> should I add all of these into the in-kernel device table? sounds excessive.

I'm not very deep into USB, but at least I got some ideas.

I'm thinking of using linux-hotplug with SynCE, I wrote a little about
it yesterday:

http://cvs.sourceforge.net/cgi-bin/viewcvs.cgi/synce/twogood-files/usb-connection.txt?rev=HEAD&content-type=text/vnd.viewcvs-markup

First we let /etc/hotplug/usb.usermap contain all those IDs (or all but
Microsoft's IDs because they're probably only used by them) and direct
them to a hotplug script provided by SynCE (/etc/hotplug/usb/synce).

Second, the hotplug script for SynCE load the kernel module, specifying
vendor and product ID:s to the module as you suggested above? Would that
work?

By the way, why isn't there a common device class/subclass or interface
class/subclass for these devices?

-- 
Regards,
            -\- David Eriksson -/-

     SynCE - http://synce.sourceforge.net
    CalcEm - http://calcem.sourceforge.net
  Desquirr - http://desquirr.sourceforge.net



-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel

Reply via email to