Re: [PATCH] KVM: vmx: speed up MSR bitmap merge

[Paolo Bonzini]

> > -	memset(msr_bitmap_l0, 0xff, PAGE_SIZE);
> > +	msr_bitmap_l1 = (unsigned long *)kmap(page);
> 
> Wouldn't it be easier to simply set everything to 0xff as before and
> then only handle the one special case where you don't do that? e.g. the
> complete else part would be gone.

It can change from one vmentry to the next, so I'm not sure how you'd
do that.

> > +	nested_vmx_disable_intercept_for_msr(
> > +		msr_bitmap_l1, msr_bitmap_l0,
> > +		APIC_BASE_MSR + (APIC_TASKPRI >> 4),
> > +		MSR_TYPE_W);
> 
> I'd vote for indenting the parameters properly (even though we exceed 80
> chars by 1 then :) )

Ok, will do.  Regarding the WARN, I'm undecided.  I like the idea of not
calling a function that works on MSR bitmaps unless you have one...

Paolo