>>I think it'd be fine to have both device and interface level
>>hotplug events for USB, which might be implied by doing that
>>kind of thing at the driverfs level.  That'd get rid of the
>>major technical wart in the current USB hotplug (the stuff
>>that "usbmodules" with usbfs addresses).
> 
> There will be a /sbin/hotplug event at the following moments in time:
>       - bus type event when device is found.  This matches the
>         existing /sbin/hotplug calls for PCI and USB

I'm not sure what a "bus type" event is, but I think it's
what I called a "device level" event above (one per device).

Still seems to me that "interface level" events would still
be good to have for USB -- handling the "multiple interfaces"
case better.

What I'd have called a "bus type" event is at the level of
hotplugging a new bus:  an adapter for USB, Firewire, IR,
or what have you.  Of course adapters are just specialized
devices, so any distinction is arbitrary...


>       - class type event when a driver registers a type of device is
>         seen (i.e. input, tty, disk, etc.)  This is the new stuff that
>         will allow us to move all device naming logic out of the
>         kernel entirely.  See Pat's documentation for more
>         explanation of this (he posted a link to it on lkml a few
>         hours ago.)
>       - class type event when the driver unregisters a type of device.

These "class type" events would seem to be what I've always thought
of as "logical device" events.  Or if you will:  they're "bus type"
events on synthetic busses that hold only input devices, ttys, disks,
and so forth.



>       - bus type even when the device is removed (like today.)
> 
> So the class events will be when a driver binds to a USB interface.
> 
> Does that make sense?

Not yet, but then I've got to read Patrick's stuff (I grabbed his
docs, one webpage with HTML would be a lot more practical than so
many different/unorganized text files).

- Dave





-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
No, I will not fix your computer.
http://thinkgeek.com/sf
_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel

Reply via email to