Re: request_firmware() hotplug interface, third round.

[Alan Cox]

> >  If the device losses the firmware upon suspend, the driver will have to
> >  reinitialize it as if it just got plugged, which somehow makes all
> >  devices hotplugable.
> 
> So all firmware has to be permanently in RAM anyway?

Of course not you can just go back out to user space and ask for it

> > 	- In a diskless client, it is the network card

Already insoluble because of routing daemons.

> No, still no good. It means that you get a memory leak if you unload
> a driver before firmware is provided. You need the ability to explicitely
> cancel a request for firmware.

Only if you program it wrongly. Its not exactly hard.

As to an interface. The simplest is probably

	request_firmware()
and
	request_firmware_nowait(......, workqueuehandler)

The issues brought up about it failing appear bogus too, if the hotplug
run returns a non zero exit code you know about this already.

Alan