Re: [PATCH v12 04/11] kvm/x86: remove kvm memblock dependency

[Pavel Tatashin]

I think using __initdata during init_hypervisor_platform()  +
allocating during x86_init.hyper.guest_late_init() is a good approach.
My only concern, it would mean we need to init/uinit/init clock for
boot cpu. Does it mean the clock continuity is preserved during the
transition? I believe so, but needs to be verified.

Pavel
On Fri, Jul 6, 2018 at 6:53 AM Thomas Gleixner <tglx@xxxxxxxxxxxxx> wrote:
>
> On Fri, 6 Jul 2018, Thomas Gleixner wrote:
> > On Fri, 6 Jul 2018, Paolo Bonzini wrote:
> > > On 06/07/2018 11:45, Thomas Gleixner wrote:
> > > >> One possibility is to introduce another layer of indirection: in
> > > >> addition to the percpu pvclock data, add a percpu pointer to the pvclock
> > > >> data and initialize it to point to a page-aligned variable in BSS.  CPU0
> > > >> (used by vDSO) doesn't touch the pointer and keeps using the BSS
> > > >> variable, APs instead redirect the pointer to the percpu data.
> > > > Yeah, thought about that, but the extra indirection is ugly. Instead of
> > > > using per cpu data, I just can allocate the memory _after_ the allocators
> > > > are up and running and use a single page sized static __initdata for the
> > > > early boot.
> > >
> > > Either works for me.  Assembly-wise, the indirection should be more or
> > > less the same as what we have now; even more efficient because it
> > > accesses a percpu pointer instead of computing it based on
> > > smp_processor_id().
> >
> > Good point. Let me try that.
>
> Duh, that either requires a static key or a static PER_CPU pointer thingy,
> which then wastes 8 bytes per cpu instead of 64 byte if unused.
>
> Thanks,
>
>         tglx