[RFC 6/6] ARM: vexpress: DT-based support for Coretile Express A5x2 (V2P-CA5s)

Pawel Moll pawel.moll at arm.com
Tue Nov 8 11:11:17 EST 2011


On Tue, 2011-11-08 at 14:34 +0000, Rob Herring wrote:
> I really think v2p-ca5s.c and v2p-ca9.c should be combined into 1 file.
> There's a lot of duplication and the parts that are different can go in
> the DT.

As I mentioned in the cover letter - that's the goal. My (minor) worry
is how to treat such file... The A5 and A9 are very similar indeed, but
A7/A15 are slightly different. So there would be "a5_and_a9.c" and
"a7_and_a15.c" and "64bit_chip.c"... I'll see how it works.

> > +                     uart0: uart at 090000 {
> 
> Use generic names:
> 
> s/uart/serial/

Ok, will do.

> > +     memory {
> 
> memory at 80000000

Ok.

> > +#ifdef CONFIG_SMP
> > +static void v2p_ca5s_init_cpu_map(void)
> > +{
> > +     int i, ncores = scu_get_core_count(V2TILE_PERIPH_P2V(A5_MPCORE_SCU));
> 
> This works on A9, does it work on A5? 
> 
>         /* Get SCU base */
>         asm("mrc p15, 4, %0, c15, c0, 0" : "=r" (base));

Cool, didn't know about this :-) And as this is a PERIPHBASE value
(http://infocenter.arm.com/help/topic/com.arm.doc.ddi0407g/CACCJFCJ.html) it could help with TWD as well. 

> If not, an SCU binding should be created.

I'll do one or the other.

Thanks!

Pawel






More information about the linux-arm-kernel mailing list