Re: [RFC v2] [x86]: abort secondary cpu bringup gracefully

[Igor Mammedov <imammedo@redhat.com>]

On 05/25/2012 08:11 PM, Rob Landley wrote:
> On 05/24/2012 10:41 AM, Igor Mammedov wrote:
>> ping for reviewers.
>>
>> Please review patch.
>
> I can't hugely comment on the guts of what the patch is doing, but:
>
>>> --- a/arch/x86/kernel/smpboot.c
>>> +++ b/arch/x86/kernel/smpboot.c
>>> @@ -136,6 +136,10 @@ EXPORT_PER_CPU_SYMBOL(cpu_info);
>>>
>>>    atomic_t init_deasserted;
>>>
>>> +#ifdef CONFIG_HOTPLUG_CPU
>>> +static void remove_siblinginfo(int cpu);
>>> +#endif
>>> +
>
> #ifdefs should almost never be in C code, they should be in header
> files. You can stub out functions with empty inline versions.  For a
> random example, see kernel/smpboot.h

If it were not just local function, then I'd do so. But since this
function is used only in this file and defined under the same #ifdefs
after smp_callin(), it could stay there and not pollute headers
with non public function declaration.

I made remove_siblinginfo() a forward declaration  before smp_callin(),
because of I didn't have much justification to move ~20 lines function
to be available a bit earlier that it is now.

However thanks for mentioning stubs. I should add stub for building
kernel without CONFIG_HOTPLUG_CPU, to prevent build breakage.

I'll fix and repost patch.

-- 
-----
  Igor