PCI API

[Kate Feng]

I agree with what Peter Dufault said.
Providing hot adding  in BSP is a safe provision,  at least till
MOTLoad change its firmware.  Will it  or wo'nt it ?  I do not
care because I am using RTEMS.


On Fri, 25 Feb 2005, at 15:15:48 , he wrote :

> If it was enabled by MOTLoad (the monitor on the MVME5500),
> then it has a bug similar to RTEMS mvme5500 (arbitrarily limiting number
of
> busses) since the bus and device don't show up with
> pciShow (and other pciFoo) commands in MOTLoad, but after
> increasing the number of busses in the RTEMS source it is
> showing up in RTEMS using BSP_pciFindDevice.


Regards,
Kate


> On Mar 2, 2005, at 8:58 PM, Till Straumann wrote:
>
> > I don't think 'hot adding' is needed or in any way easy to do. AFAIK
> > you rely on MOTLoad [or whatever it's called]
> > to perform the PCI configuration and the firmware is not going to
> > support any 'hot adding' [hot == while system
> > is powered-on / running].
>
> Considering how the cards mount on the VME card, hot adding is
> definitely an intellectual exercise only in relation to the MVME5500.
> >
> > OTOH, if a DEC21150 is plugged into the [cold] system and powered up,
> > MOTLoad will configure PCI
> > and you'll end up with 4 segments in total, the 1GHz ending up on
> > segment 3.
> >
>
> I don't think MOTLoad is that smart.  When I add a PMCSPAN (DEC21150)
> with cards in it, MOTLoad displays the DEC21150 but doesn't show
> anything about the devices on it, though it displays the devices on
> BUS0 and BUS1. The manuals sure sound as if everything is hardwired in
> place in the topology I showed in my ASCII art.  If there is a way to
> set everything up with solder I wouldn't be surprised to find out that
> is how it is configured.
>
> If anyone adds a bridge device off the PMCSPAN they'll have to plan to
> get the code in powerpc/shared working to setup the bridge and
> downstream devices.
>
> Peter Dufault
> HD Associates, Inc.