[PATCH 01/52] ARM: introduce atag_offset to replace boot_params

[Nicolas Pitre]

On Tue, 5 Jul 2011, Stephen Boyd wrote:

> On 07/05/2011 07:46 PM, Nicolas Pitre wrote:
> > diff --git a/arch/arm/include/asm/mach/arch.h b/arch/arm/include/asm/mach/arch.h
> > index 7b79a00..1bdf6e1 100644
> > --- a/arch/arm/include/asm/mach/arch.h
> > +++ b/arch/arm/include/asm/mach/arch.h
> > @@ -18,6 +18,7 @@ struct machine_desc {
> >  	unsigned int		nr;		/* architecture number	*/
> >  	const char		*name;		/* architecture name	*/
> >  	unsigned long		boot_params;	/* tagged list		*/
> > +	unsigned long		atag_offset;	/* tagged list (relative) */
> >  	const char		**dt_compat;	/* array of device tree
> >  						 * 'compatible' strings	*/
> >  
> > diff --git a/arch/arm/kernel/setup.c b/arch/arm/kernel/setup.c
> > index e0db84d..4cc3e2b 100644
> > --- a/arch/arm/kernel/setup.c
> > +++ b/arch/arm/kernel/setup.c
> > @@ -816,6 +816,8 @@ static struct machine_desc * __init setup_machine_tags(unsigned int nr)
> >  
> >  	if (__atags_pointer)
> >  		tags = phys_to_virt(__atags_pointer);
> > +	else if(mdesc->atag_offset)
> > +		tags = PAGE_OFFSET + mdesc->atag_offset;
> >  	else if (mdesc->boot_params) {
> >  #ifdef CONFIG_MMU
> >  		/*
> 
> Would it make sense to drop the atag_offset in most board files and
> default it to 0x100 in the MACHINE_START macro itself? It's almost a
> universal value and would cut down on hundreds of lines.

Well, no more than 50 lines actually.

And there are those cases where that value is not set to anything, and 
therefore I'd prefer we don't start always assuming something at 0x*100 
for them.  IOW, this is meant to be functionally transparent.


Nicolas