> > > Here's the thought process I originally used:
> > > - Multiple configurations are rare
> > > - Devices won't change substantially between configurations (ie
> > > won't go from a keyboard to a cdrom)
> >
> > That is very likely an invalid assumption.
> > A device might go from webcam to storage, just to give an example.
>
> That's kind of the point, there is no best choice in that case that we
> can make either through the core, or through a driver.
>
> If that's the case, a driver has no business changing the configuration
> and the core would only change if it really feels it knows better and
> only at enumeration time.
Yes, but the user may want to change the configuration later on.
That we also have to support. In fact we currently do support
changing configuration, but if you do so, it's likely to oops the kernel.
Should that ioctl for 2.4.20 be made to require the direct hardware
capability? It seems to be a security hole.
> > No problem.
> > IMHO selecting a configuration is a policy decision and should be left
> > to user space.
>
> That's prefectly reasonable. Should we just use the first configuration
> when a device is plugged in?
Yes, what we do now is good enough for the common case.
We can leave it to user space and if we are unlucky enough we might
need an in kernel blacklist for absolutely broken devices.
> Because a user may not have told the system to change the configuration?
> If the change was made directly because of a user decision, I have no
> problem with it.
There may be some broken devices which need it during normal
operation. For that I've seperated logical and physical configuration
change in my latest patch.
Regards
Oliver
-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
