Avi Kivity wrote:
> On 03/08/2010 03:53 PM, Alexander Graf wrote:
>>
>>> So do it the other way around. Always load the registers (of course,
>>> do nothing if already loaded) and then access them in just one way. I
>>> assume during emulation the registers will always be loaded?
>>>
>> During emulation we're always in VCPU_RUN, so the vcpu is loaded.
>>
>> Do you mean something like:
>>
>> read_register(num) {
>> vcpu_load();
>> read register from PACA(num);
>> vcpu_put();
>> }
>>
>> ? Does vcpu_load incur overhead when it doesnt' need to do anything?
>>
>
> If the vcpu is always loaded, this would be redundant, no?
>
> The situation is that a piece of data is in one of two places.
> Instead of checking and loading it from either, force it to the place
> where it normally is, and load it from there.
>
> So instead of
>
> if (x)
> y = p1;
> else
> y = p2;
>
> in a zillion places, just do
>
> force_to_p2(); // the common case anyway
> y = p2;
>
> which results in cleaner code. Assuming that you have a common case
> of course.
We're looking at two different ifs here.
1) GPR Inside the PACA or not (volatile vs non-volatile)
This is constant. Volatile registers go to the PACA; non-volatiles go to
the vcpu struct.
2) GPR actually loaded in the PACA
When we're in vcpu_load context the registers in the PACA, when not
they're in the vcpu struct
If you have a really easy and fast way to assure that we're always
inside a vcpu_load context, all is great. I could probably even just put
in a BUG_ON(not in vcpu_load context) and make the callers safe. But
some check needs to be done.
Alex
--
To unsubscribe from this list: send the line "unsubscribe kvm" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
