[v1] KVM: s390: VSIE: sort out virtual/physical address in pin_guest_page

[v1] KVM: s390: VSIE: sort out virtual/physical address in pin_guest_page

[Nico Boehr]

pin_guest_page() used page_to_virt() to calculate the hpa of the pinned
page. This currently works, because virtual and physical addresses are
the same. Use page_to_phys() instead to resolve the virtual-real address
confusion.

One caller of pin_guest_page() actually expected the hpa to be a hva, so
add the missing phys_to_virt() conversion here.

Signed-off-by: Nico Boehr <nrb@linux.ibm.com>
---
 arch/s390/kvm/vsie.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/arch/s390/kvm/vsie.c b/arch/s390/kvm/vsie.c
index 94138f8f0c1c..c6a10ff46d58 100644
--- a/arch/s390/kvm/vsie.c
+++ b/arch/s390/kvm/vsie.c
@@ -654,7 +654,7 @@ static int pin_guest_page(struct kvm *kvm, gpa_t gpa, hpa_t *hpa)
 	page = gfn_to_page(kvm, gpa_to_gfn(gpa));
 	if (is_error_page(page))
 		return -EINVAL;
-	*hpa = (hpa_t) page_to_virt(page) + (gpa & ~PAGE_MASK);
+	*hpa = (hpa_t)page_to_phys(page) + (gpa & ~PAGE_MASK);
 	return 0;
 }
 
@@ -869,7 +869,7 @@ static int pin_scb(struct kvm_vcpu *vcpu, struct vsie_page *vsie_page,
 		WARN_ON_ONCE(rc);
 		return 1;
 	}
-	vsie_page->scb_o = (struct kvm_s390_sie_block *) hpa;
+	vsie_page->scb_o = (struct kvm_s390_sie_block *)phys_to_virt(hpa);
 	return 0;
 }
 
-- 
2.37.3

Re: [v1] KVM: s390: VSIE: sort out virtual/physical address in pin_guest_page

[Christian Borntraeger]

Am 24.10.22 um 18:02 schrieb Nico Boehr:
> pin_guest_page() used page_to_virt() to calculate the hpa of the pinned
> page. This currently works, because virtual and physical addresses are
> the same. Use page_to_phys() instead to resolve the virtual-real address
> confusion.
> 
> One caller of pin_guest_page() actually expected the hpa to be a hva, so
> add the missing phys_to_virt() conversion here.
> 
> Signed-off-by: Nico Boehr <nrb@linux.ibm.com>
> ---
>   arch/s390/kvm/vsie.c | 4 ++--
>   1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/arch/s390/kvm/vsie.c b/arch/s390/kvm/vsie.c
> index 94138f8f0c1c..c6a10ff46d58 100644
> --- a/arch/s390/kvm/vsie.c
> +++ b/arch/s390/kvm/vsie.c
> @@ -654,7 +654,7 @@ static int pin_guest_page(struct kvm *kvm, gpa_t gpa, hpa_t *hpa)
>   	page = gfn_to_page(kvm, gpa_to_gfn(gpa));
>   	if (is_error_page(page))
>   		return -EINVAL;
> -	*hpa = (hpa_t) page_to_virt(page) + (gpa & ~PAGE_MASK);
> +	*hpa = (hpa_t)page_to_phys(page) + (gpa & ~PAGE_MASK);
>   	return 0;
>   }
>   
> @@ -869,7 +869,7 @@ static int pin_scb(struct kvm_vcpu *vcpu, struct vsie_page *vsie_page,
>   		WARN_ON_ONCE(rc);
>   		return 1;
>   	}
> -	vsie_page->scb_o = (struct kvm_s390_sie_block *) hpa;
> +	vsie_page->scb_o = (struct kvm_s390_sie_block *)phys_to_virt(hpa);

Do we still need the cast here? phys_to_virt should return a void * and the assignment should succeed.


>   	return 0;
>   }
>   

Re: [v1] KVM: s390: VSIE: sort out virtual/physical address in pin_guest_page

[Nico Boehr]

Quoting Christian Borntraeger (2022-10-25 08:37:21)
[...]
> > diff --git a/arch/s390/kvm/vsie.c b/arch/s390/kvm/vsie.c
> > index 94138f8f0c1c..c6a10ff46d58 100644
[...]
> > @@ -869,7 +869,7 @@ static int pin_scb(struct kvm_vcpu *vcpu, struct vsie_page *vsie_page,
> >               WARN_ON_ONCE(rc);
> >               return 1;
> >       }
> > -     vsie_page->scb_o = (struct kvm_s390_sie_block *) hpa;
> > +     vsie_page->scb_o = (struct kvm_s390_sie_block *)phys_to_virt(hpa);
> 
> Do we still need the cast here? phys_to_virt should return a void * and the assignment should succeed.

Yes, right. Fixed in v2.