Re: [PATCH v11 14/18] KVM: arm64: Handle guest_memfd-backed guest page faults

[James Houghton <jthoughton@google.com>]

On Thu, Jun 5, 2025 at 8:38 AM Fuad Tabba <tabba@google.com> wrote:
>
> Add arm64 support for handling guest page faults on guest_memfd backed
> memslots. Until guest_memfd supports huge pages, the fault granule is
> restricted to PAGE_SIZE.
>
> Signed-off-by: Fuad Tabba <tabba@google.com>

Hi Fuad, sorry for not getting back to you on v10. I like this patch
much better than the v9 version, thank you! Some small notes below.

> ---
>  arch/arm64/kvm/mmu.c | 93 ++++++++++++++++++++++++++++++++++++++++++--
>  1 file changed, 90 insertions(+), 3 deletions(-)
>
> diff --git a/arch/arm64/kvm/mmu.c b/arch/arm64/kvm/mmu.c
> index ce80be116a30..f14925fe6144 100644
> --- a/arch/arm64/kvm/mmu.c
> +++ b/arch/arm64/kvm/mmu.c
> @@ -1508,6 +1508,89 @@ static void adjust_nested_fault_perms(struct kvm_s2_trans *nested,
>         *prot |= kvm_encode_nested_level(nested);
>  }
>
> +#define KVM_PGTABLE_WALK_MEMABORT_FLAGS (KVM_PGTABLE_WALK_HANDLE_FAULT | KVM_PGTABLE_WALK_SHARED)
> +
> +static int gmem_abort(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa,
> +                     struct kvm_s2_trans *nested,
> +                     struct kvm_memory_slot *memslot, bool is_perm)
> +{
> +       bool logging, write_fault, exec_fault, writable;
> +       enum kvm_pgtable_walk_flags flags = KVM_PGTABLE_WALK_MEMABORT_FLAGS;
> +       enum kvm_pgtable_prot prot = KVM_PGTABLE_PROT_R;
> +       struct kvm_pgtable *pgt = vcpu->arch.hw_mmu->pgt;
> +       struct page *page;
> +       struct kvm *kvm = vcpu->kvm;
> +       void *memcache;
> +       kvm_pfn_t pfn;
> +       gfn_t gfn;
> +       int ret;
> +
> +       ret = prepare_mmu_memcache(vcpu, !is_perm, &memcache);
> +       if (ret)
> +               return ret;
> +
> +       if (nested)
> +               gfn = kvm_s2_trans_output(nested) >> PAGE_SHIFT;
> +       else
> +               gfn = fault_ipa >> PAGE_SHIFT;
> +
> +       logging = memslot_is_logging(memslot);

AFAICT, `logging` will always be `false` for now, so we can simplify
this function quite a bit. And IMHO, it *should* be simplified, as it
cannot be tested.

> +       write_fault = kvm_is_write_fault(vcpu);
> +       exec_fault = kvm_vcpu_trap_is_exec_fault(vcpu);
> +
> +       if (write_fault && exec_fault) {
> +               kvm_err("Simultaneous write and execution fault\n");
> +               return -EFAULT;
> +       }
> +
> +       if (is_perm && !write_fault && !exec_fault) {
> +               kvm_err("Unexpected L2 read permission error\n");
> +               return -EFAULT;
> +       }

I think, ideally, these above checks should be put into a separate
function and shared with user_mem_abort(). (The VM_BUG_ON(write_fault
&& exec_fault) that user_mem_abort() does seems fine to me, I don't see a
real need to change it to -EFAULT.)

> +
> +       ret = kvm_gmem_get_pfn(kvm, memslot, gfn, &pfn, &page, NULL);
> +       if (ret) {
> +               kvm_prepare_memory_fault_exit(vcpu, fault_ipa, PAGE_SIZE,
> +                                             write_fault, exec_fault, false);
> +               return ret;
> +       }
> +
> +       writable = !(memslot->flags & KVM_MEM_READONLY) &&
> +                  (!logging || write_fault);
> +
> +       if (nested)
> +               adjust_nested_fault_perms(nested, &prot, &writable);
> +
> +       if (writable)
> +               prot |= KVM_PGTABLE_PROT_W;
> +
> +       if (exec_fault ||
> +           (cpus_have_final_cap(ARM64_HAS_CACHE_DIC) &&
> +            (!nested || kvm_s2_trans_executable(nested))))
> +               prot |= KVM_PGTABLE_PROT_X;
> +
> +       kvm_fault_lock(kvm);
> +       if (is_perm) {
> +               /*
> +                * Drop the SW bits in favour of those stored in the
> +                * PTE, which will be preserved.
> +                */
> +               prot &= ~KVM_NV_GUEST_MAP_SZ;
> +               ret = KVM_PGT_FN(kvm_pgtable_stage2_relax_perms)(pgt, fault_ipa, prot, flags);

I think you should drop this `is_perm` path, as it is an optimization
for dirty logging, which we don't currently do. :)

When we want to add dirty logging support, we probably ought to move
this mapping code (the lines kvm_fault_lock() and kvm_fault_unlock())
into its own function and share it with user_mem_abort().

> +       } else {
> +               ret = KVM_PGT_FN(kvm_pgtable_stage2_map)(pgt, fault_ipa, PAGE_SIZE,
> +                                            __pfn_to_phys(pfn), prot,
> +                                            memcache, flags);
> +       }
> +       kvm_release_faultin_page(kvm, page, !!ret, writable);
> +       kvm_fault_unlock(kvm);
> +
> +       if (writable && !ret)
> +               mark_page_dirty_in_slot(kvm, memslot, gfn);
> +
> +       return ret != -EAGAIN ? ret : 0;
> +}
> +
>  static int user_mem_abort(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa,
>                           struct kvm_s2_trans *nested,
>                           struct kvm_memory_slot *memslot, unsigned long hva,
> @@ -1532,7 +1615,7 @@ static int user_mem_abort(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa,
>         enum kvm_pgtable_prot prot = KVM_PGTABLE_PROT_R;
>         struct kvm_pgtable *pgt;
>         struct page *page;
> -       enum kvm_pgtable_walk_flags flags = KVM_PGTABLE_WALK_HANDLE_FAULT | KVM_PGTABLE_WALK_SHARED;
> +       enum kvm_pgtable_walk_flags flags = KVM_PGTABLE_WALK_MEMABORT_FLAGS;
>
>         if (fault_is_perm)
>                 fault_granule = kvm_vcpu_trap_get_perm_fault_granule(vcpu);
> @@ -1959,8 +2042,12 @@ int kvm_handle_guest_abort(struct kvm_vcpu *vcpu)
>                 goto out_unlock;
>         }
>
> -       ret = user_mem_abort(vcpu, fault_ipa, nested, memslot, hva,
> -                            esr_fsc_is_permission_fault(esr));
> +       if (kvm_slot_has_gmem(memslot))
> +               ret = gmem_abort(vcpu, fault_ipa, nested, memslot,
> +                                esr_fsc_is_permission_fault(esr));
> +       else
> +               ret = user_mem_abort(vcpu, fault_ipa, nested, memslot, hva,
> +                                    esr_fsc_is_permission_fault(esr));

I like this split! Thank you!


>         if (ret == 0)
>                 ret = 1;
>  out:
> --
> 2.49.0.1266.g31b7d2e469-goog
>