Re: [PATCH v4 4/5] KVM: guest_memfd: Establish memslot<=>guest_memfd bindings *after* memslot is ready

[Sean Christopherson <seanjc@google.com>]

On Mon, Sep 21, 2026, sashiko-bot@kernel.org wrote:
> Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
> > diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
> > --- a/virt/kvm/kvm_main.c
> > +++ b/virt/kvm/kvm_main.c
> [ ... ]
> > @@ -1944,6 +1945,15 @@ static int kvm_set_memslot(struct kvm *kvm,
> >  	r = kvm_prepare_memory_region(kvm, old, new, change);
> >  	if (r)
> >  		goto err;
> >  
> > +	if (change == KVM_MR_CREATE && (new->flags & KVM_MEM_GUEST_MEMFD)) {
> > +		r = kvm_gmem_bind(kvm, new, gmem_fd, gmem_offset);
> 
> [Severity: High]
> Does this new initialization order break guest_memfd creation on arm64?
> 
> By moving the kvm_gmem_bind() call to execute after kvm_prepare_memory_region(),
> the KVM_MEMSLOT_GMEM_ONLY flag is not populated in time for the architecture
> specific hooks.
> 
> In virt/kvm/guest_memfd.c:kvm_gmem_bind(), the flag is assigned:
> 
>     if (kvm_gmem_supports_mmap(inode))
>         slot->flags |= KVM_MEMSLOT_GMEM_ONLY;
> 
> Because kvm_prepare_memory_region() now runs before this assignment, the
> architecture hook in arch/arm64/kvm/mmu.c:kvm_arch_prepare_memory_region()
> evaluates the flag before it is set:
> 
>     if (kvm_slot_has_gmem(new) && !kvm_memslot_is_gmem_only(new))
>         return -EINVAL;

Argh, all that time spent trying to avoid adding a "prepare" phase to gmem, only
for it to be all for naught.