KVM: Update gfn_to_pfn_cache khva when it moves within the same page
authorDavid Woodhouse <dwmw@amazon.co.uk>
Sat, 19 Nov 2022 09:25:39 +0000 (09:25 +0000)
committerPaolo Bonzini <pbonzini@redhat.com>
Wed, 23 Nov 2022 23:58:46 +0000 (18:58 -0500)
In the case where a GPC is refreshed to a different location within the
same page, we didn't bother to update it. Mostly we don't need to, but
since the ->khva field also includes the offset within the page, that
does have to be updated.

Fixes: 3ba2c95ea180 ("KVM: Do not incorporate page offset into gfn=>pfn cache user address")
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Reviewed-by: Paul Durrant <paul@xen.org>
Reviewed-by: Sean Christopherson <seanjc@google.com>
Cc: stable@kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
virt/kvm/pfncache.c

index 346e47f15572484b91d7b72be62a65cef1481936..7c248193ca26e2c3b3f14947b4514a5171897866 100644 (file)
@@ -297,7 +297,12 @@ int kvm_gfn_to_pfn_cache_refresh(struct kvm *kvm, struct gfn_to_pfn_cache *gpc,
        if (!gpc->valid || old_uhva != gpc->uhva) {
                ret = hva_to_pfn_retry(kvm, gpc);
        } else {
-               /* If the HVA→PFN mapping was already valid, don't unmap it. */
+               /*
+                * If the HVA→PFN mapping was already valid, don't unmap it.
+                * But do update gpc->khva because the offset within the page
+                * may have changed.
+                */
+               gpc->khva = old_khva + page_offset;
                old_pfn = KVM_PFN_ERR_FAULT;
                old_khva = NULL;
                ret = 0;