Re: [RFC 03/12] KVM/VMX: Use the new host mapping API for virtual_apic_page

[Jim Mattson]

On Fri, Feb 9, 2018 at 4:15 AM, KarimAllah Ahmed  wrote:

> Can you elaborate a bit? I do not really understand what is the concern.

I can't find the posting that gave me this impression. The only thing
I can find is the following in Documentation/vm/highmem.txt:

 (*) kmap().  This permits a short duration mapping of a single page.  It needs
 global synchronization, but is amortized somewhat.  It is also prone to
 deadlocks when using in a nested fashion, and so it is not recommended for
 new code.

What constitutes a "short duration," and does this new usage qualify?

Re: [RFC 03/12] KVM/VMX: Use the new host mapping API for virtual_apic_page

[Jim Mattson]

On Fri, Feb 9, 2018 at 4:15 AM, KarimAllah Ahmed  wrote:

> Can you elaborate a bit? I do not really understand what is the concern.

I can't find the posting that gave me this impression. The only thing
I can find is the following in Documentation/vm/highmem.txt:

 (*) kmap().  This permits a short duration mapping of a single page.  It needs
 global synchronization, but is amortized somewhat.  It is also prone to
 deadlocks when using in a nested fashion, and so it is not recommended for
 new code.

What constitutes a "short duration," and does this new usage qualify?

Re: [RFC 03/12] KVM/VMX: Use the new host mapping API for virtual_apic_page

[KarimAllah Ahmed]

On 02/05/2018 11:26 PM, Jim Mattson wrote:

On Mon, Feb 5, 2018 at 10:48 AM KarimAllah Ahmed  wrote:


@@ -5264,9 +5264,8 @@ static void

vmx_complete_nested_posted_interrupt(struct kvm_vcpu *vcpu)


  max_irr = find_last_bit((unsigned long

*)vmx->nested.pi_desc->pir, 256);

  if (max_irr != 256) {
-   vapic_page = kmap(vmx->nested.virtual_apic_page);
+   vapic_page = vmx->nested.virtual_apic_mapping.kaddr;
  __kvm_apic_update_irr(vmx->nested.pi_desc->pir,

vapic_page);

-   kunmap(vmx->nested.virtual_apic_page);


Your kernel mapping now survives exits to userspace. I didn't think that
was kosher, but I'd be happy to hear that I'm wrong.


Can you elaborate a bit? I do not really understand what is the concern.


@@ -7502,10 +7501,8 @@ static void free_nested(struct vcpu_vmx *vmx)
  /* Unpin physical memory we referred to in the vmcs02 */
  if (vmx->nested.apic_access_mapping.pfn)


kvm_release_host_mapping(>nested.apic_access_mapping, true);

-   if (vmx->nested.virtual_apic_page) {
-   kvm_release_page_dirty(vmx->nested.virtual_apic_page);
-   vmx->nested.virtual_apic_page = NULL;
-   }
+   if (vmx->nested.virtual_apic_mapping.pfn)
+

kvm_release_host_mapping(>nested.virtual_apic_mapping, true);

Is the host PFN guaranteed to be non-zero?


Yeah, I should not have assumed that PFN zero is invalid. Will fix.
Amazon Development Center Germany GmbH
Berlin - Dresden - Aachen
main office: Krausenstr. 38, 10117 Berlin
Geschaeftsfuehrer: Dr. Ralf Herbrich, Christian Schlaeger
Ust-ID: DE289237879
Eingetragen am Amtsgericht Charlottenburg HRB 149173 B

Re: [RFC 03/12] KVM/VMX: Use the new host mapping API for virtual_apic_page

[KarimAllah Ahmed]

On 02/05/2018 11:26 PM, Jim Mattson wrote:

On Mon, Feb 5, 2018 at 10:48 AM KarimAllah Ahmed  wrote:


@@ -5264,9 +5264,8 @@ static void

vmx_complete_nested_posted_interrupt(struct kvm_vcpu *vcpu)


  max_irr = find_last_bit((unsigned long

*)vmx->nested.pi_desc->pir, 256);

  if (max_irr != 256) {
-   vapic_page = kmap(vmx->nested.virtual_apic_page);
+   vapic_page = vmx->nested.virtual_apic_mapping.kaddr;
  __kvm_apic_update_irr(vmx->nested.pi_desc->pir,

vapic_page);

-   kunmap(vmx->nested.virtual_apic_page);


Your kernel mapping now survives exits to userspace. I didn't think that
was kosher, but I'd be happy to hear that I'm wrong.


Can you elaborate a bit? I do not really understand what is the concern.


@@ -7502,10 +7501,8 @@ static void free_nested(struct vcpu_vmx *vmx)
  /* Unpin physical memory we referred to in the vmcs02 */
  if (vmx->nested.apic_access_mapping.pfn)


kvm_release_host_mapping(>nested.apic_access_mapping, true);

-   if (vmx->nested.virtual_apic_page) {
-   kvm_release_page_dirty(vmx->nested.virtual_apic_page);
-   vmx->nested.virtual_apic_page = NULL;
-   }
+   if (vmx->nested.virtual_apic_mapping.pfn)
+

kvm_release_host_mapping(>nested.virtual_apic_mapping, true);

Is the host PFN guaranteed to be non-zero?


Yeah, I should not have assumed that PFN zero is invalid. Will fix.
Amazon Development Center Germany GmbH
Berlin - Dresden - Aachen
main office: Krausenstr. 38, 10117 Berlin
Geschaeftsfuehrer: Dr. Ralf Herbrich, Christian Schlaeger
Ust-ID: DE289237879
Eingetragen am Amtsgericht Charlottenburg HRB 149173 B

Re: [RFC 03/12] KVM/VMX: Use the new host mapping API for virtual_apic_page

[Jim Mattson]

On Mon, Feb 5, 2018 at 10:48 AM KarimAllah Ahmed  wrote:

> @@ -5264,9 +5264,8 @@ static void
vmx_complete_nested_posted_interrupt(struct kvm_vcpu *vcpu)

>  max_irr = find_last_bit((unsigned long
*)vmx->nested.pi_desc->pir, 256);
>  if (max_irr != 256) {
> -   vapic_page = kmap(vmx->nested.virtual_apic_page);
> +   vapic_page = vmx->nested.virtual_apic_mapping.kaddr;
>  __kvm_apic_update_irr(vmx->nested.pi_desc->pir,
vapic_page);
> -   kunmap(vmx->nested.virtual_apic_page);

Your kernel mapping now survives exits to userspace. I didn't think that
was kosher, but I'd be happy to hear that I'm wrong.

> @@ -7502,10 +7501,8 @@ static void free_nested(struct vcpu_vmx *vmx)
>  /* Unpin physical memory we referred to in the vmcs02 */
>  if (vmx->nested.apic_access_mapping.pfn)

kvm_release_host_mapping(>nested.apic_access_mapping, true);
> -   if (vmx->nested.virtual_apic_page) {
> -   kvm_release_page_dirty(vmx->nested.virtual_apic_page);
> -   vmx->nested.virtual_apic_page = NULL;
> -   }
> +   if (vmx->nested.virtual_apic_mapping.pfn)
> +
kvm_release_host_mapping(>nested.virtual_apic_mapping, true);

Is the host PFN guaranteed to be non-zero?

Re: [RFC 03/12] KVM/VMX: Use the new host mapping API for virtual_apic_page

[Jim Mattson]

On Mon, Feb 5, 2018 at 10:48 AM KarimAllah Ahmed  wrote:

> @@ -5264,9 +5264,8 @@ static void
vmx_complete_nested_posted_interrupt(struct kvm_vcpu *vcpu)

>  max_irr = find_last_bit((unsigned long
*)vmx->nested.pi_desc->pir, 256);
>  if (max_irr != 256) {
> -   vapic_page = kmap(vmx->nested.virtual_apic_page);
> +   vapic_page = vmx->nested.virtual_apic_mapping.kaddr;
>  __kvm_apic_update_irr(vmx->nested.pi_desc->pir,
vapic_page);
> -   kunmap(vmx->nested.virtual_apic_page);

Your kernel mapping now survives exits to userspace. I didn't think that
was kosher, but I'd be happy to hear that I'm wrong.

> @@ -7502,10 +7501,8 @@ static void free_nested(struct vcpu_vmx *vmx)
>  /* Unpin physical memory we referred to in the vmcs02 */
>  if (vmx->nested.apic_access_mapping.pfn)

kvm_release_host_mapping(>nested.apic_access_mapping, true);
> -   if (vmx->nested.virtual_apic_page) {
> -   kvm_release_page_dirty(vmx->nested.virtual_apic_page);
> -   vmx->nested.virtual_apic_page = NULL;
> -   }
> +   if (vmx->nested.virtual_apic_mapping.pfn)
> +
kvm_release_host_mapping(>nested.virtual_apic_mapping, true);

Is the host PFN guaranteed to be non-zero?