Re: [PATCH v3 04/18] KVM: arm64: Move host page ownership tracking to the hyp vmemmap

[Quentin Perret <qperret@google.com>]

On Tuesday 17 Dec 2024 at 11:03:08 (+0000), Marc Zyngier wrote:
> On Mon, 16 Dec 2024 17:57:49 +0000,
> Quentin Perret <qperret@google.com> wrote:
> > 
> > We currently store part of the page-tracking state in PTE software bits
> > for the host, guests and the hypervisor. This is sub-optimal when e.g.
> > sharing pages as this forces to break block mappings purely to support
> > this software tracking. This causes an unnecessarily fragmented stage-2
> > page-table for the host in particular when it shares pages with Secure,
> > which can lead to measurable regressions. Moreover, having this state
> > stored in the page-table forces us to do multiple costly walks on the
> > page transition path, hence causing overhead.
> > 
> > In order to work around these problems, move the host-side page-tracking
> > logic from SW bits in its stage-2 PTEs to the hypervisor's vmemmap.
> > 
> > Signed-off-by: Quentin Perret <qperret@google.com>
> > ---
> >  arch/arm64/kvm/hyp/include/nvhe/memory.h |   6 +-
> >  arch/arm64/kvm/hyp/nvhe/mem_protect.c    | 100 ++++++++++++++++-------
> >  arch/arm64/kvm/hyp/nvhe/setup.c          |   7 +-
> >  3 files changed, 77 insertions(+), 36 deletions(-)
> > 
> > diff --git a/arch/arm64/kvm/hyp/include/nvhe/memory.h b/arch/arm64/kvm/hyp/include/nvhe/memory.h
> > index 45b8d1840aa4..8bd9a539f260 100644
> > --- a/arch/arm64/kvm/hyp/include/nvhe/memory.h
> > +++ b/arch/arm64/kvm/hyp/include/nvhe/memory.h
> > @@ -8,7 +8,7 @@
> >  #include <linux/types.h>
> >  
> >  /*
> > - * SW bits 0-1 are reserved to track the memory ownership state of each page:
> > + * Bits 0-1 are reserved to track the memory ownership state of each page:
> >   *   00: The page is owned exclusively by the page-table owner.
> >   *   01: The page is owned by the page-table owner, but is shared
> >   *       with another entity.
> > @@ -43,7 +43,9 @@ static inline enum pkvm_page_state pkvm_getstate(enum kvm_pgtable_prot prot)
> >  struct hyp_page {
> >  	u16 refcount;
> >  	u8 order;
> > -	u8 reserved;
> > +
> > +	/* Host (non-meta) state. Guarded by the host stage-2 lock. */
> > +	enum pkvm_page_state host_state : 8;
> 
> An enum as a bitfield? Crazy! :)

Hehe, it works so why not :)

> You probably want an assert somewhere that ensures that hyp_page is a
> 32bit quantity, just to make sure (and avoid hard to track bugs).

Sounds like a good idea, I'll stick a BUILD_BUG_ON() somewhere.