Re: [PATCH 2/2] x86/HVM: Use fixed TSC value when saving or restoring domain

[Boris Ostrovsky <boris.ostrovsky@oracle.com>]

On 03/31/2014 05:51 AM, Jan Beulich wrote:
>
>> --- a/xen/arch/x86/hvm/save.c
>> +++ b/xen/arch/x86/hvm/save.c
>> @@ -24,7 +24,7 @@
>>   #include <asm/hvm/support.h>
>>   #include <public/hvm/save.h>
>>   
>> -void arch_hvm_save(struct domain *d, struct hvm_save_header *hdr)
>> +void arch_hvm_save(struct domain *dom, struct hvm_save_header *hdr)
> The change is unmotivated (afaict) and inconsistent with most other
> code - we conventionally use "d" for struct domain * variables. Please
> drop the change here and use "d" throughout the rest of the patch,
> at once resulting in less churn and hence making it easier to review.

The reason for this change is subsequent
     rdtscll(dom->arch.chkpt_tsc);

which will not work as
     rdtscll(d->arch.chkpt_tsc);

The alternatives as I see are
* Declare a temporary variable for use with rdtscll
* Change rdtscll's definition to use something else instead of variable 
'd'. Say, eax and edx (hopefully this won't clash with something else)


>
>> index 95b4b91..032eb23 100644
>> --- a/xen/include/xen/time.h
>> +++ b/xen/include/xen/time.h
>> @@ -32,7 +32,8 @@ struct vcpu;
>>   typedef s64 s_time_t;
>>   #define PRI_stime PRId64
>>   
>> -s_time_t get_s_time(void);
>> +s_time_t get_s_time_fixed(u64 at_tick);
>> +#define get_s_time() get_s_time_fixed(0)
> get_s_time(), through NOW(), has quite many users, so I'm not certain
> the code bloat resulting from this is desirable. I'd suggest the function
> to remain such; the compiler will be able to make it a mov+jmp.

Sorry, not sure I understand what you are asking for.

There shouldn't be much of code size increase since get_s_time() 
currently (and get_s_time_fixed() after this patch is applied) are not 
inlines. The only increase is due to routine itself getting very 
slightly larger.

But I suspect you meant something else.

-boris