[v2,2/2] kvm: nVMX: interrupt-window exiting should wake L2 from HLT
diff mbox series

Message ID 20181126192331.140309-1-jmattson@google.com
State New
Headers show
Series
  • [v2,1/2] kvm: nVMX: NMI-window exiting should wake L2 from HLT
Related show

Commit Message

Jim Mattson Nov. 26, 2018, 7:23 p.m. UTC
According to the SDM, "interrupt-window exiting" VM-exits wake a
logical processor from the same inactive states as would an external
interrupt. Specifically, they wake a logical processor from the states
entered using the HLT and MWAIT instructions.

Fixes: 6dfacadd5858 ("KVM: nVMX: Add support for activity state HLT")
Signed-off-by: Jim Mattson <jmattson@google.com>
Reviewed-by: Peter Shier <pshier@google.com>
---
 arch/x86/kvm/vmx.c | 17 ++++++++++++++---
 1 file changed, 14 insertions(+), 3 deletions(-)

Comments

Sean Christopherson Nov. 26, 2018, 8:18 p.m. UTC | #1
On Mon, Nov 26, 2018 at 11:23:31AM -0800, Jim Mattson wrote:
> According to the SDM, "interrupt-window exiting" VM-exits wake a
> logical processor from the same inactive states as would an external
> interrupt. Specifically, they wake a logical processor from the states
> entered using the HLT and MWAIT instructions.
> 
> Fixes: 6dfacadd5858 ("KVM: nVMX: Add support for activity state HLT")
> Signed-off-by: Jim Mattson <jmattson@google.com>
> Reviewed-by: Peter Shier <pshier@google.com>
> ---
>  arch/x86/kvm/vmx.c | 17 ++++++++++++++---
>  1 file changed, 14 insertions(+), 3 deletions(-)
> 
> diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
> index 358e96f0c21d..a29921b216f5 100644
> --- a/arch/x86/kvm/vmx.c
> +++ b/arch/x86/kvm/vmx.c
> @@ -13498,6 +13498,16 @@ static bool nested_vmx_nmi_window_exit(struct vmcs12 *vmcs12)
>  		  (GUEST_INTR_STATE_NMI | GUEST_INTR_STATE_MOV_SS));
>  }
>  
> +static bool nested_vmx_intr_window_exit(struct vmcs12 *vmcs12)
> +{
> +	return (vmcs12->cpu_based_vm_exec_control &
> +		CPU_BASED_VIRTUAL_INTR_PENDING) &&
> +		vmcs12->guest_activity_state != GUEST_ACTIVITY_WAIT_SIPI &&

Technically this also needs to weed out GUEST_ACTIVITY_SHUTDOWN as
INTR doesn't break shutdown.  That being said, I wonder if it makes
sense to remove the activity state and guest_interruptibility_info
checks altogether from both the NMI and INTR helpers.  The current use
case is only for guest_activity_state==HLT, which makes both checks
redundant since guest_activity_state obviously can't be two things at
once and STI/MOVSS blocking are only legal when going to ACTIVE state.
Event blocking and prioritization is so complex that it seems unlikely
that a generic kitchen-sink helper will be useful, e.g.
handle_invalid_guest_state() already has this code and more by way of
vmx_interrupt_allowed().

At that point it's probably easiest to just open code the checks, e.g.:

	exec_controls = vmcs12->cpu_based_vm_exec_control;
	if ((vmcs12->guest_activity_state == GUEST_ACTIVITY_HLT) &&
	    !(vmcs12->vm_entry_intr_info_field & INTR_INFO_VALID_MASK) &&
	    !(exec_controls & CPU_BASED_VIRTUAL_NMI_PENDING) &&
	    !((exec_controls & CPU_BASED_VIRTUAL_INTR_PENDING) &&
	      (vmcs12->guest_rflags & X86_EFLAGS_IF))) {
		vmx->nested.nested_run_pending = 0;
		return kvm_vcpu_halt(vcpu);
	}
	return 1;



> +		(vmcs12->guest_rflags & X86_EFLAGS_IF) &&
> +		!(vmcs12->guest_interruptibility_info &
> +		  (GUEST_INTR_STATE_STI | GUEST_INTR_STATE_MOV_SS));
> +}
> +
>  /*
>   * nested_vmx_run() handles a nested entry, i.e., a VMLAUNCH or VMRESUME on L1
>   * for running an L2 nested guest.
> @@ -13590,12 +13600,13 @@ static int nested_vmx_run(struct kvm_vcpu *vcpu, bool launch)
>  
>  	/*
>  	 * If we're entering a halted L2 vcpu and the L2 vcpu won't be
> -	 * awakened by event injection or by an NMI-window VM-exit,
> -	 * halt the vcpu.
> +	 * awakened by event injection or by an NMI-window VM-exit or
> +	 * by an interrupt-window VM-exit, halt the vcpu.
>  	 */
>  	if ((vmcs12->guest_activity_state == GUEST_ACTIVITY_HLT) &&
>  	    !(vmcs12->vm_entry_intr_info_field & INTR_INFO_VALID_MASK) &&
> -	    !nested_vmx_nmi_window_exit(vmcs12)) {
> +	    !nested_vmx_nmi_window_exit(vmcs12) &&
> +	    !nested_vmx_intr_window_exit(vmcs12)) {
>  		vmx->nested.nested_run_pending = 0;
>  		return kvm_vcpu_halt(vcpu);
>  	}
> -- 
> 2.20.0.rc0.387.gc7a69e6b6c-goog
>
Jim Mattson Nov. 26, 2018, 8:34 p.m. UTC | #2
On Mon, Nov 26, 2018 at 12:18 PM Sean Christopherson
<sean.j.christopherson@intel.com> wrote:
>
> On Mon, Nov 26, 2018 at 11:23:31AM -0800, Jim Mattson wrote:
> > According to the SDM, "interrupt-window exiting" VM-exits wake a
> > logical processor from the same inactive states as would an external
> > interrupt. Specifically, they wake a logical processor from the states
> > entered using the HLT and MWAIT instructions.
> >
> > Fixes: 6dfacadd5858 ("KVM: nVMX: Add support for activity state HLT")
> > Signed-off-by: Jim Mattson <jmattson@google.com>
> > Reviewed-by: Peter Shier <pshier@google.com>
> > ---
> >  arch/x86/kvm/vmx.c | 17 ++++++++++++++---
> >  1 file changed, 14 insertions(+), 3 deletions(-)
> >
> > diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
> > index 358e96f0c21d..a29921b216f5 100644
> > --- a/arch/x86/kvm/vmx.c
> > +++ b/arch/x86/kvm/vmx.c
> > @@ -13498,6 +13498,16 @@ static bool nested_vmx_nmi_window_exit(struct vmcs12 *vmcs12)
> >                 (GUEST_INTR_STATE_NMI | GUEST_INTR_STATE_MOV_SS));
> >  }
> >
> > +static bool nested_vmx_intr_window_exit(struct vmcs12 *vmcs12)
> > +{
> > +     return (vmcs12->cpu_based_vm_exec_control &
> > +             CPU_BASED_VIRTUAL_INTR_PENDING) &&
> > +             vmcs12->guest_activity_state != GUEST_ACTIVITY_WAIT_SIPI &&
>
> Technically this also needs to weed out GUEST_ACTIVITY_SHUTDOWN as
> INTR doesn't break shutdown.  That being said, I wonder if it makes
> sense to remove the activity state and guest_interruptibility_info
> checks altogether from both the NMI and INTR helpers.  The current use
> case is only for guest_activity_state==HLT, which makes both checks
> redundant since guest_activity_state obviously can't be two things at
> once and STI/MOVSS blocking are only legal when going to ACTIVE state.
> Event blocking and prioritization is so complex that it seems unlikely
> that a generic kitchen-sink helper will be useful, e.g.
> handle_invalid_guest_state() already has this code and more by way of
> vmx_interrupt_allowed().
>
> At that point it's probably easiest to just open code the checks, e.g.:
>
>         exec_controls = vmcs12->cpu_based_vm_exec_control;
>         if ((vmcs12->guest_activity_state == GUEST_ACTIVITY_HLT) &&
>             !(vmcs12->vm_entry_intr_info_field & INTR_INFO_VALID_MASK) &&
>             !(exec_controls & CPU_BASED_VIRTUAL_NMI_PENDING) &&
>             !((exec_controls & CPU_BASED_VIRTUAL_INTR_PENDING) &&
>               (vmcs12->guest_rflags & X86_EFLAGS_IF))) {
>                 vmx->nested.nested_run_pending = 0;
>                 return kvm_vcpu_halt(vcpu);
>         }
>         return 1;

Sounds good to me. Let's call that v3.
Radim Krčmář Dec. 20, 2018, 8:19 p.m. UTC | #3
2018-11-26 12:34-0800, Jim Mattson:
> On Mon, Nov 26, 2018 at 12:18 PM Sean Christopherson
> <sean.j.christopherson@intel.com> wrote:
> >
> > On Mon, Nov 26, 2018 at 11:23:31AM -0800, Jim Mattson wrote:
> > > According to the SDM, "interrupt-window exiting" VM-exits wake a
> > > logical processor from the same inactive states as would an external
> > > interrupt. Specifically, they wake a logical processor from the states
> > > entered using the HLT and MWAIT instructions.
> > >
> > > Fixes: 6dfacadd5858 ("KVM: nVMX: Add support for activity state HLT")
> > > Signed-off-by: Jim Mattson <jmattson@google.com>
> > > Reviewed-by: Peter Shier <pshier@google.com>
> > > ---
> > >  arch/x86/kvm/vmx.c | 17 ++++++++++++++---
> > >  1 file changed, 14 insertions(+), 3 deletions(-)
> > >
> > > diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
> > > index 358e96f0c21d..a29921b216f5 100644
> > > --- a/arch/x86/kvm/vmx.c
> > > +++ b/arch/x86/kvm/vmx.c
> > > @@ -13498,6 +13498,16 @@ static bool nested_vmx_nmi_window_exit(struct vmcs12 *vmcs12)
> > >                 (GUEST_INTR_STATE_NMI | GUEST_INTR_STATE_MOV_SS));
> > >  }
> > >
> > > +static bool nested_vmx_intr_window_exit(struct vmcs12 *vmcs12)
> > > +{
> > > +     return (vmcs12->cpu_based_vm_exec_control &
> > > +             CPU_BASED_VIRTUAL_INTR_PENDING) &&
> > > +             vmcs12->guest_activity_state != GUEST_ACTIVITY_WAIT_SIPI &&
> >
> > Technically this also needs to weed out GUEST_ACTIVITY_SHUTDOWN as
> > INTR doesn't break shutdown.  That being said, I wonder if it makes
> > sense to remove the activity state and guest_interruptibility_info
> > checks altogether from both the NMI and INTR helpers.  The current use
> > case is only for guest_activity_state==HLT, which makes both checks
> > redundant since guest_activity_state obviously can't be two things at
> > once and STI/MOVSS blocking are only legal when going to ACTIVE state.
> > Event blocking and prioritization is so complex that it seems unlikely
> > that a generic kitchen-sink helper will be useful, e.g.
> > handle_invalid_guest_state() already has this code and more by way of
> > vmx_interrupt_allowed().
> >
> > At that point it's probably easiest to just open code the checks, e.g.:
> >
> >         exec_controls = vmcs12->cpu_based_vm_exec_control;
> >         if ((vmcs12->guest_activity_state == GUEST_ACTIVITY_HLT) &&
> >             !(vmcs12->vm_entry_intr_info_field & INTR_INFO_VALID_MASK) &&
> >             !(exec_controls & CPU_BASED_VIRTUAL_NMI_PENDING) &&
> >             !((exec_controls & CPU_BASED_VIRTUAL_INTR_PENDING) &&
> >               (vmcs12->guest_rflags & X86_EFLAGS_IF))) {
> >                 vmx->nested.nested_run_pending = 0;
> >                 return kvm_vcpu_halt(vcpu);
> >         }
> >         return 1;

I have queued this patch.  Let me know if you'd like to make any
amendments.  (I deliberately made the lines over 80 characters.)

---8<---
From: Jim Mattson <jmattson@google.com>
Subject: kvm: nVMX: NMI-window and interrupt-window exiting should

According to the SDM, "NMI-window exiting" VM-exits wake a logical
processor from the same inactive states as would an NMI and
"interrupt-window exiting" VM-exits wake a logical processor from the
same inactive states as would an external interrupt. Specifically, they
wake a logical processor from the shutdown state and from the states
entered using the HLT and MWAIT instructions.

Fixes: 6dfacadd5858 ("KVM: nVMX: Add support for activity state HLT")
Signed-off-by: Jim Mattson <jmattson@google.com>
Reviewed-by: Peter Shier <pshier@google.com>
Suggested-by: Sean Christopherson <sean.j.christopherson@intel.com>
[Squashed comments of two Jim's patches and used the simplified code
 hunk provided by Sean. - Radim]
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
---
 arch/x86/kvm/vmx/nested.c | 10 +++++++---
 1 file changed, 7 insertions(+), 3 deletions(-)

diff --git a/arch/x86/kvm/vmx/nested.c b/arch/x86/kvm/vmx/nested.c
index 3f019aa63341..adc8493132ee 100644
--- a/arch/x86/kvm/vmx/nested.c
+++ b/arch/x86/kvm/vmx/nested.c
@@ -3156,11 +3156,15 @@ static int nested_vmx_run(struct kvm_vcpu *vcpu, bool launch)
 	nested_cache_shadow_vmcs12(vcpu, vmcs12);
 
 	/*
-	 * If we're entering a halted L2 vcpu and the L2 vcpu won't be woken
-	 * by event injection, halt vcpu.
+	 * If we're entering a halted L2 vcpu and the L2 vcpu won't be
+	 * awakened by event injection or by an NMI-window VM-exit or
+	 * by an interrupt-window VM-exit, halt the vcpu.
 	 */
 	if ((vmcs12->guest_activity_state == GUEST_ACTIVITY_HLT) &&
-	    !(vmcs12->vm_entry_intr_info_field & INTR_INFO_VALID_MASK)) {
+	    !(vmcs12->vm_entry_intr_info_field & INTR_INFO_VALID_MASK) &&
+	    !(vmcs12->cpu_based_vm_exec_control & CPU_BASED_VIRTUAL_NMI_PENDING) &&
+	    !((vmcs12->cpu_based_vm_exec_control & CPU_BASED_VIRTUAL_INTR_PENDING) &&
+	      (vmcs12->guest_rflags & X86_EFLAGS_IF))) {
 		vmx->nested.nested_run_pending = 0;
 		return kvm_vcpu_halt(vcpu);
 	}

Patch
diff mbox series

diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
index 358e96f0c21d..a29921b216f5 100644
--- a/arch/x86/kvm/vmx.c
+++ b/arch/x86/kvm/vmx.c
@@ -13498,6 +13498,16 @@  static bool nested_vmx_nmi_window_exit(struct vmcs12 *vmcs12)
 		  (GUEST_INTR_STATE_NMI | GUEST_INTR_STATE_MOV_SS));
 }
 
+static bool nested_vmx_intr_window_exit(struct vmcs12 *vmcs12)
+{
+	return (vmcs12->cpu_based_vm_exec_control &
+		CPU_BASED_VIRTUAL_INTR_PENDING) &&
+		vmcs12->guest_activity_state != GUEST_ACTIVITY_WAIT_SIPI &&
+		(vmcs12->guest_rflags & X86_EFLAGS_IF) &&
+		!(vmcs12->guest_interruptibility_info &
+		  (GUEST_INTR_STATE_STI | GUEST_INTR_STATE_MOV_SS));
+}
+
 /*
  * nested_vmx_run() handles a nested entry, i.e., a VMLAUNCH or VMRESUME on L1
  * for running an L2 nested guest.
@@ -13590,12 +13600,13 @@  static int nested_vmx_run(struct kvm_vcpu *vcpu, bool launch)
 
 	/*
 	 * If we're entering a halted L2 vcpu and the L2 vcpu won't be
-	 * awakened by event injection or by an NMI-window VM-exit,
-	 * halt the vcpu.
+	 * awakened by event injection or by an NMI-window VM-exit or
+	 * by an interrupt-window VM-exit, halt the vcpu.
 	 */
 	if ((vmcs12->guest_activity_state == GUEST_ACTIVITY_HLT) &&
 	    !(vmcs12->vm_entry_intr_info_field & INTR_INFO_VALID_MASK) &&
-	    !nested_vmx_nmi_window_exit(vmcs12)) {
+	    !nested_vmx_nmi_window_exit(vmcs12) &&
+	    !nested_vmx_intr_window_exit(vmcs12)) {
 		vmx->nested.nested_run_pending = 0;
 		return kvm_vcpu_halt(vcpu);
 	}