On Monday 06 August 2007, Aurelien Jarno wrote:
> On Mon, Aug 06, 2007 at 08:05:29PM +0200, Michael Buesch wrote:
> > On Monday 06 August 2007, Aurelien Jarno wrote:
> > > The patch below against 2.6.23-rc1-mm2 adds support for BCM947xx CPUs.
> > > It originally comes from the OpenWrt patches.
> > >
> > > Cc: Michael Buesch <mb@xxxxxxxxx>
> > > Cc: Waldemar Brodkorb <wbx@xxxxxxxxxxx>
> > > Cc: Felix Fietkau <nbd@xxxxxxxxxxx>
> > > Cc: Florian Schirmer <jolt@xxxxxxxxxx>
> > > Signed-off-by: Aurelien Jarno <aurelien@xxxxxxxxxxx>
>
> [snip]
>
> > > --- a/arch/mips/bcm947xx/setup.c
> > > +++ b/arch/mips/bcm947xx/setup.c
> > > @@ -0,0 +1,106 @@
>
> [snip]
>
> > > +static void bcm947xx_machine_halt(void)
> > > +{
> > > + /* Disable interrupts and watchdog and spin forever */
> > > + local_irq_disable();
> > > + ssb_chipco_watchdog_timer_set(&ssb.chipco, 0);
> > > + while (1)
> > > + cpu_relax();
> > > +}
> > > +
> > > +static int bcm947xx_get_invariants(struct ssb_bus *bus, struct ssb_init_invariants *iv)
> > > +{
> >
> > No reading of NVRAM, yet?
>
> This requires CFE support, and this part is still a bit problematic now.
> There are already CFE files for SiByte support, and I think it's not a
> good idea to have too different implementations.
Yep, I forgot about that. My CFE stuff should probably partly get
merged into the existing CFE stuff, if neeeded.
So let's leave this out for now.
> I will add a comment here;
>
> > > + return 0;
> > > +}
> > > +
> > > +void __init plat_mem_setup(void)
> > > +{
> > > + int i, err;
> > > + struct ssb_mipscore *mcore;
> > > +
> > > + err = ssb_bus_ssbbus_register(&ssb, SSB_ENUM_BASE, bcm947xx_get_invariants);
> > > + if (err) {
> > > + const char *msg = "Failed to initialize SSB bus (err %d)\n";
> > > + panic(msg, err);
> >
> > We don't need the msg variable anymore.
>
> I will fix that.
>
>
