Re: [Fastboot] [PATCH 2/5] Use the APIC to determine the hardware processor id - i386

[Vivek Goyal <vgoyal@in.ibm.com>]

On Thu, Mar 01, 2007 at 04:16:59PM +0900, Fernando Luis Vázquez Cao wrote:
> Use the APIC to determine the hardware processor id when APIC support
> has been selected, independently of whether CONFIG_SMP is set or not.
> 
> Signed-off-by: Fernando Luis Vazquez Cao <fernando@oss.ntt.co.jp>
> ---
> 
> diff -urNp linux-2.6.21-rc2/include/asm-i386/smp.h linux-2.6.21-rc2-hwcpuid/include/asm-i386/smp.h
> --- linux-2.6.21-rc2/include/asm-i386/smp.h	2007-03-01 14:02:21.000000000 +0900
> +++ linux-2.6.21-rc2-hwcpuid/include/asm-i386/smp.h	2007-03-01 14:08:50.000000000 +0900
> @@ -74,20 +74,6 @@ static inline int num_booting_cpus(void)
>  	return cpus_weight(cpu_callout_map);
>  }
>  
> -#ifdef CONFIG_X86_LOCAL_APIC
> -
> -#ifdef APIC_DEFINITION
> -extern int hard_smp_processor_id(void);
> -#else
> -#include <mach_apicdef.h>
> -static inline int hard_smp_processor_id(void)
> -{
> -	/* we don't want to mark this access volatile - bad code generation */
> -	return GET_APIC_ID(*(unsigned long *)(APIC_BASE+APIC_ID));
> -}
> -#endif
> -#endif
> -
>  extern int safe_smp_processor_id(void);
>  extern int __cpu_disable(void);
>  extern void __cpu_die(unsigned int cpu);
> @@ -102,10 +88,23 @@ extern unsigned int num_processors;
>  
>  #define NO_PROC_ID		0xFF		/* No processor magic marker */
>  
> -#endif
> +#endif /* CONFIG_SMP */
>  
>  #ifndef __ASSEMBLY__
>  
> +#ifdef CONFIG_X86_LOCAL_APIC
> +#ifdef APIC_DEFINITION
> +extern int hard_smp_processor_id(void);
> +#else
> +#include <mach_apicdef.h>
> +static inline int hard_smp_processor_id(void)
> +{
> +	/* we don't want to mark this access volatile - bad code generation */
> +	return GET_APIC_ID(*(unsigned long *)(APIC_BASE+APIC_ID));
> +}
> +#endif /* APIC_DEFINITION */
> +#endif /* CONFIG_X86_LOCAL_APIC */
> +

I think compilation will fail if CONFIG_X86_LOCAL_APIC=n and I build a UP
kernel? There is no definition of hard_smp_processor_id in that case.

Otherwise, as a concept it seems to make sense that hard_smp_processor_id()
is not necessarily zero on UP systems.

Thanks
Vivek