[PATCH v23 00/14] KVM: arm64: CCA: Add basic plumbing for Realms

[Marc Zyngier]

On Fri, 09 Oct 2026 15:20:44 +0100,
Fuad Tabba <fuad.tabba at linux.dev> wrote:
> 
> Hi Marc,
> 
> On Fri, 09 Oct 2026 14:28:03 +0100, Marc Zyngier <maz at kernel.org> wrote:
> [...]
> > Of course, mmu.c is where the crap happens, so anyone with any interest
> > in this stuff being halfway sane, please review the conflict resolution.
> 
> I went through the mmu.c resolution and it looks good to me. Unmap and
> aging are still no-ops for protected VMs, their abort handler is
> unchanged, and the default op keeps the tracked unmap for nested MMUs.
> Two minor things you might want to tidy up.
> 
> kvm_vm_stage2_unmap_range() repeats the lockdep_assert_held_write()
> and WARN_ON(size & ~PAGE_MASK) that kvm_stage2_unmap_range() already
> does before calling the op.
> 
> The VM_WARN_ON_ONCE() in kvm_vm_mem_abort() reads the ESR from the
> vcpu again, where next had moved these checks to s2fd->esr. It's the
> same result today, since the only caller handles a real abort, but
> this (and dropping the vcpu local) would keep it in line with the
> rest:
> 
>         VM_WARN_ON_ONCE(kvm_s2_fault_is_perm(s2fd) &&
>                         !esr_abt_is_write_fault(s2fd->esr) &&
>                         !kvm_s2_fault_is_exec(s2fd));

Thanks for looking. I've now applied this to the merge resolution.

Cheers,

	M.

-- 
Without deviation from the norm, progress is not possible.