Re: [patch 03/18] x86: pvclock: remove pvclock_shadow_time

[Avi Kivity <avi@redhat.com>]

On 10/30/2012 11:23 AM, Avi Kivity wrote:
> On 10/24/2012 03:13 PM, Marcelo Tosatti wrote:
>> Originally from Jeremy Fitzhardinge.
>> 
>> We can copy the information directly from "struct pvclock_vcpu_time_info", 
>> remove pvclock_shadow_time.
>> 
>> Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
>> 
>>  
>>  unsigned long pvclock_tsc_khz(struct pvclock_vcpu_time_info *src)
>> @@ -90,21 +54,20 @@ void pvclock_resume(void)
>>  
>>  cycle_t pvclock_clocksource_read(struct pvclock_vcpu_time_info *src)
>>  {
>> -	struct pvclock_shadow_time shadow;
>>  	unsigned version;
>>  	cycle_t ret, offset;
>>  	u64 last;
>>  
>>  	do {
>> -		version = pvclock_get_time_values(&shadow, src);
>> +		version = src->version;
>>  		rdtsc_barrier();
>> -		offset = pvclock_get_nsec_offset(&shadow);
>> -		ret = shadow.system_timestamp + offset;
>> +		offset = pvclock_get_nsec_offset(src);
>> +		ret = src->system_time + offset;
>>  		rdtsc_barrier();
>> -	} while (version != src->version);
>> +	} while ((src->version & 1) || version != src->version);
>>  
>>  	if ((valid_flags & PVCLOCK_TSC_STABLE_BIT) &&
>> -		(shadow.flags & PVCLOCK_TSC_STABLE_BIT))
>> +		(src->flags & PVCLOCK_TSC_STABLE_BIT))
>>  		return ret;
>>  
> 
> You're now reading PVCLOCK_TSC_STABLE outside the critical section.  We
> could have live migrated to a tsc-unstable host and had this flag cleared.
> 

I see it's fixed in patch 5.  So there's a window of two patches where
this is broken.


-- 
error compiling committee.c: too many arguments to function