On Wednesday 28 January 2004 15:35, Andrew de Quincey wrote:
> On Wednesday 28 January 2004 15:30, Andreas Oberritter wrote:
> > On Wed, 2004-01-28 at 14:53, Andrew de Quincey wrote:
> > > My favoured solution is the struct device* one... that would mean
> > > frontends could also determine the PCI/USB device IDs of the device
> > > they're attached to... say two different manufacturers use the same
> > > frontend chip, but it needs to be configured differently for their
> > > devices, this could be handled easily.
> >
> > There is no such thing like PCI bus or USB between Frontend and CPU
> > inside most STBs, so you should not make the drivers depend on
> > CONFIG_PCI or CONFIG_USB etc.
>
> Yeah, thats another good point. Its looking more and more like a callback
> function is the route to go....

This seems to be my day for second thoughts.

struct device* doesn't depend on any specific bus type, so including it 
dvb_adapter doesn't have any dependencies.

Even if I did add a callback, it would need to have a piece of private data 
_as well as_ the callback added to dvb_adapter, so it knew which instance of 
a device was reqesting the firmware.

I'm back to thinking adding struct device* is the best and clearest route. 
I'll post a patch so people can see what I'm suggesting clearly.


-- 
Info:
To unsubscribe send a mail to [EMAIL PROTECTED] with "unsubscribe linux-dvb" as 
subject.

Reply via email to