Re: [PATCH] x86: don't map io_apic two times.

[Ingo Molnar <mingo@elte.hu>]

* Yinghai Lu <yinghai@kernel.org> wrote:

> mp_register_ioapic() already map it.
> 
> [ Impact: don't map io apic address again ]
> 
> Signed-off-by: Yinghai Lu <yinghai@kernel.org>
> 
> ---
>  arch/x86/kernel/apic/io_apic.c |    7 ++++++-
>  1 file changed, 6 insertions(+), 1 deletion(-)
> 
> Index: linux-2.6/arch/x86/kernel/apic/io_apic.c
> ===================================================================
> --- linux-2.6.orig/arch/x86/kernel/apic/io_apic.c
> +++ linux-2.6/arch/x86/kernel/apic/io_apic.c
> @@ -4162,7 +4162,12 @@ fake_ioapic_page:
>  				alloc_bootmem_pages(PAGE_SIZE);
>  			ioapic_phys = __pa(ioapic_phys);
>  		}
> -		set_fixmap_nocache(idx, ioapic_phys);
> +		/*
> +		 * when acpi ioapic is used, mp_register_ioapic() map
> +		 * ioapic_phys already
> +		 */
> +		if (!acpi_ioapic)
> +			set_fixmap_nocache(idx, ioapic_phys);

The change is correct, but i'm not sure we want to do this - the 
acpi_ioapic flag might change its meaning (and then break this code 
subtly) and this is bootup code so doing the mapping twice should be 
no issue.

If it were some expensive initialization i'd agree, but here it's 
really just a couple of instructions and an INVLPG.

Hm?

	Ingo