Federico G. Schwindt wrote:
quoted text
>> Please test this on other modems which supported in umsm(4).
> 
>   comments below.

thanks.


quoted text
>> @@ -182,6 +187,7 @@ umsm_attach(struct device *parent, struc
>>  	sc->sc_iface = uaa->iface;
>>  
>>  	id = usbd_get_interface_descriptor(sc->sc_iface);
>> +	sc->sc_iface_no = id->bInterfaceNumber;
> 
>   is there any reason to set this before the check (that could return),
> and not after?

oops, this line should be located after "if (id == NULL ...)" block.

quoted text
>>  	if (id == NULL || id->bInterfaceClass == UICLASS_MASS) {
>>  		/*
>> @@ -233,6 +239,8 @@ umsm_attach(struct device *parent, struc
>>  		return;
>>  	}
>>  
>> +	sc->sc_dtr = -1;
>> +
> 
>   shouldn't be the same for rts?

initialize sc_rts to -1 will be more safe. Should change umodem.c too?


quoted text
>> +	} else if (buf->bNotification == UCDC_N_CONNECTION_SPEED_CHANGE) {
>> +		/* discard it */
>> +	} else {
>> +		DPRINTF(("%s: umsm_intr: unknown notify message (0x%02x)\n",
>> +		    sc->sc_dev.dv_xname, buf->bNotification));
>> +		return;
>> +	}
> 
>   why not else if (buf->bNotification != UCDC_N_CONNECTION_SPEED_CHANGE)
> instead of an empty if?

the modems I checked are using only UCDC_N_SERIAL_STATE and 
UCDC_N_CONNECTION_SPEED_CHANGE, but I'm not sure other modems' beheivior.
I want to know another notify message types if othere modems would used.
To avoid bogus DPRINTF message, I put empty if clause for 
UCDC_N_CONNECTION_SPEED_CHANGE.


-- 
------------------------------------------------------------
Yojiro UO
yuo@iijlab.net