[PATCH] Platform bus support

[Richard Hughes]

On Tue, 2005-09-27 at 17:37 +0200, Pierre Ossman wrote:
> Richard Hughes wrote:
> 
> >
> >Bang on.
> >
> >[root at localhost hal]# egrep -ir
> >"CONFIG_SERIAL_8250_NR_UARTS" /usr/src/kernels/2.6.12-1.1456_FC4-i686/.config
> >CONFIG_SERIAL_8250_NR_UARTS=32
> >
> >This is a kernel oddity I guess. I thought this all went out with the
> >MAKEDEV stuff.
> The 8250 is still legacy and very enumeration-unfriendly. We'll just
> have to hope it goes the way of the dodo.

This isn't an old laptop tho...

> >In my opinion, HAL has to hide the kernel funnies. Is there any way to
> >test if the 8250-derived serial port is actually existing and add this
> >logic to HAL? It's madness to add 32 devices that don't actually exist.
> It would be nice. An ioctl might be the way to go. In any case, hal
> should still show all of these for now. We do not want to hide ports
> that might actually be there. Excluding ports would be a follow-up patch.

Hmm, over 35 new devices in my tree for (no offence) for no real
personal gain doesn't seem a winner for me. How would you use the ioctl
in the existing framework? I'm sure 8250 isn't the only driver that does
this.

> >p.s. is the fact I'm not running a bleeding edge kernel anything to do
> >with this?
> No, default is still 4. So how you got it to 32 is a mystery. :)

Guess it's a redhat thing.

Richard.