Re: [PATCH v2 3/6] TI816X: Update common OMAP machine specific sources

[Tony Lindgren <tony@atomide.com>]

* Pedanekar, Hemant <hemantp@ti.com> [101022 10:58]:
> Tony Lindgren wrote on Friday, September 17, 2010 3:55 AM:
> 
> > * Hemant Pedanekar <hemantp@ti.com> [100811 10:03]:
> >> This patch updates the common machine spcific source files with support
> >> for TI816X. 
> >> 
> >> Note that the nr_irqs is overridden in INTC driver since the number of
> >> IRQs on TI816X are different (128) comapared to other OMAPs (96).
> > 
> > <snip>
> > 
> >> --- a/arch/arm/mach-omap2/id.c
> >> +++ b/arch/arm/mach-omap2/id.c
> >> @@ -419,6 +441,9 @@ void __init omap2_check_revision(void)
> >>  	} else if (cpu_is_omap44xx()) {
> >>  		omap4_check_revision();
> >>  		return;
> >> +	} else if (cpu_is_ti816x()) {
> >> +		ti816x_check_revision();
> >> +		return;
> >>  	} else {
> >>  		pr_err("OMAP revision unknown, please fix!\n");
> >>  	}
> > 
> > This does not look right, at this point you should just know
> > the processor class set by the set_globals call.
> > 
> > Please take a look at plat-omap/common.c and add an entry for
> > omap2_set_globals_ti816x. This gets called very early from
> > the board-*.c
> > file, so you can initialize things.
> 
> Tony,
> 
> Do you mean following changes in the patch I sent earlier?

No, I'm mostly wondering how come you can't use cpu_is_omap34xx
for ti816x and then detect based on the values set in
omap2_set_globals. Then cpu_is_ti816x just becomes a set of
features.

In general, we should initialize more things in set_globals
if necessary rather than sprinkle cpu_is_omap tests all over
the place to add support for new omaps.

Regards,

Tony