Re: umodeswitch

[christos%astron.com@localhost (Christos Zoulas)]

In article <1p7infd.zofefgbhxppgM%manu%netbsd.org@localhost>,
Emmanuel Dreyfus <manu%netbsd.org@localhost> wrote:
>Emmanuel Dreyfus <manu%netbsd.org@localhost> wrote:
>
>> I had some success with ZTE MF112 by just telling umass to not attach it
>> using umass_quirks (patch below). I got a frienly OK when telling it ATZ
>> on /dev/ttyU2 at 230400 bps. I need to add a SIM to perform further
>> testing.
>> 
>> Is the approach sane?
>
>It is not. I got mislead by the fact that the device maintains state
>across reboots, and is only reset by a power cycle.
>
>Supporting ZTE MF112 only requires this change:
>
>--- u3g.c.orig
>+++ ./u3g.c
>@@ -253,4 +253,5 @@
>        { USB_VENDOR_ZTE, USB_PRODUCT_ZTE_MF628 },
>        { USB_VENDOR_ZTE, USB_PRODUCT_ZTE_MF820D },
>+       { USB_VENDOR_ZTE, USB_PRODUCT_ZTE_MF112 },
> 
>        /* 4G Systems */
>
>But it does not work out of the box. It needs two device scan for u3g to
>attach, either booting twice after a power cycle, or
>detaching/rescanning after a single boot  from cold state (drvctl -d
>umodeswitch0 ; drvctl -a usbdevif -r uhub1).
>
>Once u3g attaches the device, it keep doing so across reboots, until the
>next power cycle. 
>
>Any hint on how that could be fixed?

See FreeBSD's u3g.c driver. Perhaps we can accommodate sending the 
initialization command in the umodeswitch.c driver?

christos