Re: [PATCH v2 3/5] x86/time: Initialise time earlier during start_secondary()

[Andrew Cooper <andrew.cooper3@citrix.com>]

On 04/03/14 09:02, Jan Beulich wrote:
>>>> On 28.02.14 at 19:57, Andrew Cooper <andrew.cooper3@citrix.com> wrote:
>> --- a/xen/arch/x86/smpboot.c
>> +++ b/xen/arch/x86/smpboot.c
>> @@ -347,6 +347,8 @@ void start_secondary(void *unused)
>>  
>>      percpu_traps_init();
>>  
>> +    init_percpu_time();
>> +
>>      cpu_init();
>>  
>>      smp_callin();
> I consider it rather risky to put this before the call to cpu_init().
> While currently it doesn't seem like the function depends on any
> of the CPU feature flags, I could easily see this to become
> necessary.
>
> Jan
>

The thing is that cpu_init() has quite a few printk()s in, mainly on
error paths, which use a timestamp of all 0's as the TSC scale and delta
have not yet been copied from the bsp.

Any potential new functionality in the future could easily define
init_percpu_time_late() or similar.

~Andrew