Re: [kvm-devel] [ RfC / patch ] kvmclock fixes

[Gerd H. <kr...@re...>]

Jeremy Fitzhardinge wrote:
> Gerd Hoffmann wrote:
>> +cycle_t pvclock_clocksource_read(struct kvm_vcpu_time_info *src)
>> +{
>> +	struct pvclock_shadow_time *shadow = &get_cpu_var(shadow_time);
>> +	cycle_t ret;
>> +
>> +	pvclock_get_time_values(shadow, src);
>> +	ret = shadow->system_timestamp + pvclock_get_nsec_offset(shadow);
>>   
> 
> You need to put this in a loop in case the system clock parameters 
> change between the pvclock_get_time_values() and pvclock_get_nsec_offset().

Fixed, new patch attached.

> How does kvm deal with suspend/resume with respect to time?  Is the 
> "system" timestamp guaranteed to remain monotonic?  For Xen, I think 
> we'll need to maintain an offset between the initial system timestamp 
> and whatever it is after resuming.

Havn't looked at it yet.

cheers,
  Gerd

-- 
http://kraxel.fedorapeople.org/xenner/