Re: kern/39211: ubsa device does no longer work since January 2008

[Frank Wille <frank%phoenix.owl.de@localhost>]

Matthias Drochner wrote:

> You are right, the code was messed up when "uhmodem"
> was added. Generally, if a USB driver attaches to an
> interface, it is not allowed to set the configuration
> later because this potentially invalidates all interfaces.
> So ubsa should attach as whole device as it did before.
> I'm not so sure about the config index. Your error message
> suggests that your device supports only one configuration
> and thus can only be set to index 0. But the code before
> the uhmodem addition seems to use index 1...
> (I've actually never seen a USB device with more than
> a single configuration.)
> 
> Could you try the appended patch, as a starting point?

Yes, changing "ubsa at usbifif" into "ubsa at usbdevif" works fine. This was
one of my first attempts to fix this bug as well. But I'm anxious that
uhmodem needs ubsa to be at usbifif, and will be broken now as a consequence
(I cannot test uhmodem).

Otherwise I would be happy with this fix.

The only harmless problems which are left, but I think they were also
present last year, are:

1. 12x "ubsa0: ubsa_request: STALLED" when connecting to ucom0.
   1x "ubsa0: ubsa_request: STALLED" when disconnecting.

2. Sometimes ucom immediately detaches and attaches again, after being
   attached:
  ubsa0 at uhub5 port 1
  ubsa0: Vodafone Vodafone Mobile Connect Card - 3G, rev 2.00/0.00, addr 2
  ucom0 at ubsa0 portno 0
  ubsa0: at uhub5 port 1 (addr 2) disconnected
  ucom0: detached
  ubsa0: detached
  ubsa0 at uhub5 port 1
  ubsa0: Vodafone Vodafone Mobile Connect Card - 3G, rev 2.00/0.00, addr 2
  ucom0 at ubsa0 portno 0

Those have no influence on the usability of the card. So very low priority
to fix that.


-- 
Frank Wille