Re: umodem(4) with "no pointer to data interface"

[Ryo ONODERA <ryo_on%yk.rim.or.jp@localhost>]

Hi,

From: Felix Deichmann <m4j0rd0m0%gmail.com@localhost>, Date: Wed, 22 Jan 2014 
19:32:08 +0100

> Am 22.01.2014 14:26, schrieb Ryo ONODERA:
>> Each quirky USB modem has specific data interface number. In my case
>> it is 1.
> 
> This is also what Linux does (but for *all* quirky USB modems with
> NO_UNION_NORMAL, this is always 1).
> 
>> This number should be given from dev table, I feel.
> 
> Are you sure this number can and will change from modem to modem? Or
> is it always 1 (which would make things easier)?

I am not sure about all modems.
Each modem may have each number.
I cannot believe that always 1 is always true.

>> How about the following change?
> 
> NetBSD already has usb_quirks.c, and using this already present
> facility, as I suggested previously and as far as it is suitable,
> appears cleaner to me. Do you agree?
> There are also umass_quirks.c and umidi_quirks.c, but I doubt it would
> be preferred to introduce a new umodem_quirks.c if it can be done with
> already present flags at the moment.
> Maybe if more modems get supported and many new modem-specific quirks
> are there, a umodem_quirks.c will be needed.

I feel quirk and dev table are different.
It seems that the quark table is used for support attached devices.
And dev table is used for overwrite matching mechanism.

> Let's see if Brett comes up with a more idiomatic patch? :)

Yes, of course.

--
Ryo ONODERA // ryo_on%yk.rim.or.jp@localhost
PGP fingerprint = 82A2 DC91 76E0 A10A 8ABB  FD1B F404 27FA C7D1 15F3