[PATCH] kvm: nVMX: Remove unnecessary sync_roots from handle_invept

[PATCH] kvm: nVMX: Remove unnecessary sync_roots from handle_invept

[Jim Mattson]

When L0 is executing handle_invept(), the TDP MMU is active. Emulating
an L1 INVEPT does require synchronizing the appropriate shadow EPT
root(s), but a call to kvm_mmu_sync_roots in this context won't do
that. Similarly, the hardware TLB and paging-structure-cache entries
associated with the appropriate shadow EPT root(s) must be flushed,
but requesting a TLB_FLUSH from this context won't do that either.

How did this ever work? KVM always does a sync_roots and TLB flush (in
the correct context) when transitioning from L1 to L2. That isn't the
best choice for nested VM performance, but it effectively papers over
the mistakes here.

Remove the unnecessary operations and leave a comment to try to do
better in the future.

Reported-by: Junaid Shahid <junaids@google.com>
Fixes: bfd0a56b90005f ("nEPT: Nested INVEPT")
Cc: Xiao Guangrong <xiaoguangrong@linux.vnet.ibm.com>
Cc: Nadav Har'El <nyh@il.ibm.com>
Cc: Jun Nakajima <jun.nakajima@intel.com>
Cc: Xinhao Xu <xinhao.xu@intel.com>
Cc: Yang Zhang <yang.z.zhang@Intel.com>
Cc: Gleb Natapov <gleb@redhat.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by Peter Shier <pshier@google.com>
Reviewed-by: Junaid Shahid <junaids@google.com>
Signed-off-by: Jim Mattson <jmattson@google.com>
---
 arch/x86/kvm/vmx/nested.c | 8 +++-----
 1 file changed, 3 insertions(+), 5 deletions(-)

diff --git a/arch/x86/kvm/vmx/nested.c b/arch/x86/kvm/vmx/nested.c
index 1032f068f0b9..35621e73e726 100644
--- a/arch/x86/kvm/vmx/nested.c
+++ b/arch/x86/kvm/vmx/nested.c
@@ -4670,13 +4670,11 @@ static int handle_invept(struct kvm_vcpu *vcpu)
 
 	switch (type) {
 	case VMX_EPT_EXTENT_GLOBAL:
+	case VMX_EPT_EXTENT_CONTEXT:
 	/*
-	 * TODO: track mappings and invalidate
-	 * single context requests appropriately
+	 * TODO: Sync the necessary shadow EPT roots here, rather than
+	 * at the next emulated VM-entry.
 	 */
-	case VMX_EPT_EXTENT_CONTEXT:
-		kvm_mmu_sync_roots(vcpu);
-		kvm_make_request(KVM_REQ_TLB_FLUSH, vcpu);
 		break;
 	default:
 		BUG_ON(1);
-- 
2.22.0.rc2.383.gf4fbbf30c2-goog

Re: [PATCH] kvm: nVMX: Remove unnecessary sync_roots from handle_invept

[Jim Mattson]

On Thu, Jun 13, 2019 at 9:16 AM Jim Mattson <jmattson@google.com> wrote:
>
> When L0 is executing handle_invept(), the TDP MMU is active. Emulating
> an L1 INVEPT does require synchronizing the appropriate shadow EPT
> root(s), but a call to kvm_mmu_sync_roots in this context won't do
> that. Similarly, the hardware TLB and paging-structure-cache entries
> associated with the appropriate shadow EPT root(s) must be flushed,
> but requesting a TLB_FLUSH from this context won't do that either.
>
> How did this ever work? KVM always does a sync_roots and TLB flush (in
> the correct context) when transitioning from L1 to L2. That isn't the
> best choice for nested VM performance, but it effectively papers over
> the mistakes here.
>
> Remove the unnecessary operations and leave a comment to try to do
> better in the future.
>
> Reported-by: Junaid Shahid <junaids@google.com>
> Fixes: bfd0a56b90005f ("nEPT: Nested INVEPT")
> Cc: Xiao Guangrong <xiaoguangrong@linux.vnet.ibm.com>
> Cc: Nadav Har'El <nyh@il.ibm.com>
> Cc: Jun Nakajima <jun.nakajima@intel.com>
> Cc: Xinhao Xu <xinhao.xu@intel.com>
> Cc: Yang Zhang <yang.z.zhang@Intel.com>
> Cc: Gleb Natapov <gleb@redhat.com>
> Cc: Paolo Bonzini <pbonzini@redhat.com>
> Reviewed-by Peter Shier <pshier@google.com>
> Reviewed-by: Junaid Shahid <junaids@google.com>
> Signed-off-by: Jim Mattson <jmattson@google.com>
> ---
>  arch/x86/kvm/vmx/nested.c | 8 +++-----
>  1 file changed, 3 insertions(+), 5 deletions(-)
>
> diff --git a/arch/x86/kvm/vmx/nested.c b/arch/x86/kvm/vmx/nested.c
> index 1032f068f0b9..35621e73e726 100644
> --- a/arch/x86/kvm/vmx/nested.c
> +++ b/arch/x86/kvm/vmx/nested.c
> @@ -4670,13 +4670,11 @@ static int handle_invept(struct kvm_vcpu *vcpu)
>
>         switch (type) {
>         case VMX_EPT_EXTENT_GLOBAL:
> +       case VMX_EPT_EXTENT_CONTEXT:
>         /*
> -        * TODO: track mappings and invalidate
> -        * single context requests appropriately
> +        * TODO: Sync the necessary shadow EPT roots here, rather than
> +        * at the next emulated VM-entry.
>          */
> -       case VMX_EPT_EXTENT_CONTEXT:
> -               kvm_mmu_sync_roots(vcpu);
> -               kvm_make_request(KVM_REQ_TLB_FLUSH, vcpu);
>                 break;
>         default:
>                 BUG_ON(1);
> --
> 2.22.0.rc2.383.gf4fbbf30c2-goog
>

Ping.

Re: [PATCH] kvm: nVMX: Remove unnecessary sync_roots from handle_invept

[Paolo Bonzini]

On 13/06/19 18:16, Jim Mattson wrote:
> When L0 is executing handle_invept(), the TDP MMU is active. Emulating
> an L1 INVEPT does require synchronizing the appropriate shadow EPT
> root(s), but a call to kvm_mmu_sync_roots in this context won't do
> that. Similarly, the hardware TLB and paging-structure-cache entries
> associated with the appropriate shadow EPT root(s) must be flushed,
> but requesting a TLB_FLUSH from this context won't do that either.
> 
> How did this ever work? KVM always does a sync_roots and TLB flush (in
> the correct context) when transitioning from L1 to L2. That isn't the
> best choice for nested VM performance, but it effectively papers over
> the mistakes here.
> 
> Remove the unnecessary operations and leave a comment to try to do
> better in the future.
> 
> Reported-by: Junaid Shahid <junaids@google.com>
> Fixes: bfd0a56b90005f ("nEPT: Nested INVEPT")
> Cc: Xiao Guangrong <xiaoguangrong@linux.vnet.ibm.com>
> Cc: Nadav Har'El <nyh@il.ibm.com>
> Cc: Jun Nakajima <jun.nakajima@intel.com>
> Cc: Xinhao Xu <xinhao.xu@intel.com>
> Cc: Yang Zhang <yang.z.zhang@Intel.com>
> Cc: Gleb Natapov <gleb@redhat.com>
> Cc: Paolo Bonzini <pbonzini@redhat.com>
> Reviewed-by Peter Shier <pshier@google.com>
> Reviewed-by: Junaid Shahid <junaids@google.com>
> Signed-off-by: Jim Mattson <jmattson@google.com>
> ---
>  arch/x86/kvm/vmx/nested.c | 8 +++-----
>  1 file changed, 3 insertions(+), 5 deletions(-)

Queued, thanks.

Paolo