[PATCH 2/3] KVM: arm64: Skip empty IPA ranges when destroying stage-2 page-tables

[Fuad Tabba]

Hi Suzuki,

On Tue, 03 Nov 2026 10:27:54 +0000, Suzuki K Poulose
<suzuki.poulose at arm.com> wrote:
[...]
> > After destroying each chunk, look up the entry at its start and, if it
> > is empty and larger than a chunk, skip its whole range. The lookup reads
>
> Could we not lookup first before we destroy, that way we can avoid
> destroy walks ?

I'd rather keep it after the walk. Looking first relies on no empty
table being left linked above KVM_PGTABLE_MIN_BLOCK_LEVEL: the lookup
would return one of its empty entries, and the loop would skip past
the table without freeing it.

I don't think anything leaves such a table behind today, but looking
after the walk doesn't depend on that, since the walk frees every
empty table it passes through. That's one extra walk per empty region.
I'll add to the comment why the lookup comes after the walk.

[...]
> > diff --git a/arch/arm64/kvm/mmu.c b/arch/arm64/kvm/mmu.c
[...]
> > @@ -1010,9 +1010,20 @@ static void stage2_destroy_range(struct kvm_pgtable *pgt, phys_addr_t addr,
[...]
> >                   next = stage2_range_addr_end(addr, end);
> > -                KVM_PGT_FN(kvm_pgtable_stage2_destroy_range)(pgt, addr,
> > -                                                             next - addr);
> > +                kvm_pgtable_stage2_destroy_range(pgt, addr, next - addr);
>
> minor nit: Given the kvm_pgtable_stage2_destroy_range() is the only
> called from here, and this caller would benefit from the walk
> information, would it make sense to return a hint back ? Not sure how
> much of that is possible. We do this in the RMM, always return a hint
> of the next possible address. Of course this is not that expensive
> as the CCA case and could be dealt in a separate series.

kvm_pgtable_stage2_destroy_range() is also called by
kvm_pgtable_stage2_destroy(), which EL2 uses to free a pKVM guest's
stage-2, so I'd rather not change it in a fix.

Cheers,
/fuad