Re: [PATCH 8/9] KVM, VMX: Add support for guest/host-only profiling

[Gleb Natapov <gleb@redhat.com>]

On Tue, Oct 04, 2011 at 01:24:18PM +0200, Avi Kivity wrote:
> On 10/04/2011 01:17 PM, Gleb Natapov wrote:
> >>
> >>  I mean, CONFIG_KVM&&  !CONFIG_PERF_EVENTS is an unlikely
> >>  combination.  If you're using kvm, you usually want PERF_EVENTS.
> >>
> >Who knows. Think about someone building appliance with embedded KVM and
> >trying to achieve minimal code footprint.
> 
> Saving a few dozen bytes, then launching a 1GB guest?
> 
> >It is much easier to add ifdefs
> >at the development stage then trying to figure out later what can be
> >ifdeffed. If we will do:
> >  if (!(cnt = perf_guest_get_msrs_count())
> >	return;
> >
> >at the beginning of atomic_switch_perf_msrs() then compiler can
> >eliminate dead code in case of !CONFIG_PERF_EVENTS since
> >perf_guest_get_msrs_count() will become 0, but this will add two
> >function calls on vmentry in CONFIG_PERF_EVENTS case.
> >
> 
> Then move it to the beginning:
> 
>    nr_msrs = perf_get_guest_msrs(&msr_buffer);
>    for (i = 0; i < nr_msrs; ++i)
>           add_atomic_switch_msr((*msr_buffer)[i], nr);
> 
> the compiler will kill the loop if nr_msrs is statically 0.
> 
Good idea. Just get rid of perf_guest_get_msrs_count().

--
			Gleb.