The following reply was made to PR usb/149039; it has been noted by GNATS.
From: Fredrik Lindberg <f...@shapeshifter.se>
Cc: pilzablei...@web.de, bug-follo...@freebsd.org,
Subject: Re: usb/149039: [uhso] Binding problem with uhso
Date: Mon, 02 Aug 2010 08:54:12 +0200
On 08/01/2010 04:37 AM, per...@pluto.rain.com wrote:
>> The following reply was made to PR usb/149039; it has been noted
>> by GNATS.
>> From: Fredrik Lindberg<f...@shapeshifter.se>
>> To: bug-follo...@freebsd.org, pilzablei...@web.de
>> Cc: Hans Petter Selasky<hsela...@c2i.net>
>> Subject: Re: usb/149039: [uhso] Binding problem with uhso
>> Date: Sat, 31 Jul 2010 15:00:07 +0200
>> I apparently missed some interface flags (that really doesn't make
>> sense for this device, it's configured with a /32 mask so broadcast
>> etc can only be to itself) that the network stack wants to work
> Is a /32 mask even legal? Unless there's a special case involved,
> it ought to mean that there are no interfaces on the subnet other
> than this one, thus this interface has no peer to communicate with
> and might as well not exist.
> Adding net@ in hopes someone there knows what should happen.
Yes, technically a /32 mask defines only one single address, but it's
the only mask that really makes sense for this device. /32 masks are
"legal" and commonly used for the loopback address of routers.
But this is is indeed a very special case. The device has a
USB interface that accepts raw IP-packets (with no other
encapsulation). Once you have told the device to connect, it will
tell you what IP-address you have and what DNS-servers to use, but
that's it. My best guess is that the devices does PPP internally in
firmware and abstracts the point-to-point link with a IP-packet
interface. But since none of these details are available the only (as
far as I know) viable thing is to set a /32 mask and set 0.0.0.0
(default route) to be directly reachable through the interface (route
email@example.com mailing list
To unsubscribe, send any mail to "freebsd-usb-unsubscr...@freebsd.org"