Re: [PATCH 1/6] KVM: Check for pending events before attempting injection

[Gleb Natapov <gleb@redhat.com>]

On Tue, Jul 27, 2010 at 04:19:35PM +0300, Avi Kivity wrote:
> Instead of blindly attempting to inject an event before each guest entry,
> check for a possible event first in vcpu->requests.  Sites that can trigger
> event injection are modified to set KVM_REQ_EVENT:
> 
> - interrupt, nmi window opening
> - ppr updates
> - i8259 output changes
> - local apic irr changes
> - rflags updates
> - gif flag set
> - event set on exit
> 
What about userspace irq chip? Does it work with this patch? I don't see
that you set KVM_REQ_EVENT on ioctl(KVM_INTERRUPT) for instance and
vcpu->run->request_interrupt_window should be probably checked out of
if (KVM_REQ_EVEN). It looks like with this approach we scatter irq
injection logic all over the code instead of having it in one place.

> This improves non-injecting entry performance, and sets the stage for
> non-atomic injection.
> 
> Signed-off-by: Avi Kivity <avi@redhat.com>
> ---
>  arch/x86/kvm/i8259.c     |    1 +
>  arch/x86/kvm/lapic.c     |   12 ++++++++++--
>  arch/x86/kvm/svm.c       |    8 +++++++-
>  arch/x86/kvm/vmx.c       |    6 ++++++
>  arch/x86/kvm/x86.c       |   35 ++++++++++++++++++++++++++---------
>  include/linux/kvm_host.h |    1 +
>  6 files changed, 51 insertions(+), 12 deletions(-)
> 
> diff --git a/arch/x86/kvm/i8259.c b/arch/x86/kvm/i8259.c
> index 8d10c06..9f7ab44 100644
> --- a/arch/x86/kvm/i8259.c
> +++ b/arch/x86/kvm/i8259.c
> @@ -64,6 +64,7 @@ static void pic_unlock(struct kvm_pic *s)
>  		if (!found)
>  			found = s->kvm->bsp_vcpu;
>  
> +		kvm_make_request(KVM_REQ_EVENT, found);
>  		kvm_vcpu_kick(found);
>  	}
>  }
> diff --git a/arch/x86/kvm/lapic.c b/arch/x86/kvm/lapic.c
> index 77d8c0f..e83d203 100644
> --- a/arch/x86/kvm/lapic.c
> +++ b/arch/x86/kvm/lapic.c
> @@ -259,9 +259,10 @@ static inline int apic_find_highest_isr(struct kvm_lapic *apic)
>  
>  static void apic_update_ppr(struct kvm_lapic *apic)
>  {
> -	u32 tpr, isrv, ppr;
> +	u32 tpr, isrv, ppr, old_ppr;
>  	int isr;
>  
> +	old_ppr = apic_get_reg(apic, APIC_PROCPRI);
>  	tpr = apic_get_reg(apic, APIC_TASKPRI);
>  	isr = apic_find_highest_isr(apic);
>  	isrv = (isr != -1) ? isr : 0;
> @@ -274,7 +275,10 @@ static void apic_update_ppr(struct kvm_lapic *apic)
>  	apic_debug("vlapic %p, ppr 0x%x, isr 0x%x, isrv 0x%x",
>  		   apic, ppr, isr, isrv);
>  
> -	apic_set_reg(apic, APIC_PROCPRI, ppr);
> +	if (old_ppr != ppr) {
> +		apic_set_reg(apic, APIC_PROCPRI, ppr);
> +		kvm_make_request(KVM_REQ_EVENT, apic->vcpu);
> +	}
>  }
>  
>  static void apic_set_tpr(struct kvm_lapic *apic, u32 tpr)
> @@ -391,6 +395,7 @@ static int __apic_accept_irq(struct kvm_lapic *apic, int delivery_mode,
>  			break;
>  		}
>  
> +		kvm_make_request(KVM_REQ_EVENT, vcpu);
>  		kvm_vcpu_kick(vcpu);
>  		break;
>  
> @@ -416,6 +421,7 @@ static int __apic_accept_irq(struct kvm_lapic *apic, int delivery_mode,
>  				       "INIT on a runnable vcpu %d\n",
>  				       vcpu->vcpu_id);
>  			vcpu->arch.mp_state = KVM_MP_STATE_INIT_RECEIVED;
> +			kvm_make_request(KVM_REQ_EVENT, vcpu);
>  			kvm_vcpu_kick(vcpu);
>  		} else {
>  			apic_debug("Ignoring de-assert INIT to vcpu %d\n",
> @@ -430,6 +436,7 @@ static int __apic_accept_irq(struct kvm_lapic *apic, int delivery_mode,
>  			result = 1;
>  			vcpu->arch.sipi_vector = vector;
>  			vcpu->arch.mp_state = KVM_MP_STATE_SIPI_RECEIVED;
> +			kvm_make_request(KVM_REQ_EVENT, vcpu);
>  			kvm_vcpu_kick(vcpu);
>  		}
>  		break;
> @@ -475,6 +482,7 @@ static void apic_set_eoi(struct kvm_lapic *apic)
>  		trigger_mode = IOAPIC_EDGE_TRIG;
>  	if (!(apic_get_reg(apic, APIC_SPIV) & APIC_SPIV_DIRECTED_EOI))
>  		kvm_ioapic_update_eoi(apic->vcpu->kvm, vector, trigger_mode);
> +	kvm_make_request(KVM_REQ_EVENT, apic->vcpu);
>  }
>  
>  static void apic_send_ipi(struct kvm_lapic *apic)
> diff --git a/arch/x86/kvm/svm.c b/arch/x86/kvm/svm.c
> index 56c9b6b..a51e067 100644
> --- a/arch/x86/kvm/svm.c
> +++ b/arch/x86/kvm/svm.c
> @@ -2258,6 +2258,7 @@ static int stgi_interception(struct vcpu_svm *svm)
>  
>  	svm->next_rip = kvm_rip_read(&svm->vcpu) + 3;
>  	skip_emulated_instruction(&svm->vcpu);
> +	kvm_make_request(KVM_REQ_EVENT, &svm->vcpu);
>  
>  	enable_gif(svm);
>  
> @@ -2644,6 +2645,7 @@ static int interrupt_window_interception(struct vcpu_svm *svm)
>  {
>  	struct kvm_run *kvm_run = svm->vcpu.run;
>  
> +	kvm_make_request(KVM_REQ_EVENT, &svm->vcpu);
>  	svm_clear_vintr(svm);
>  	svm->vmcb->control.int_ctl &= ~V_IRQ_MASK;
>  	/*
> @@ -3089,8 +3091,10 @@ static void svm_complete_interrupts(struct vcpu_svm *svm)
>  
>  	svm->int3_injected = 0;
>  
> -	if (svm->vcpu.arch.hflags & HF_IRET_MASK)
> +	if (svm->vcpu.arch.hflags & HF_IRET_MASK) {
>  		svm->vcpu.arch.hflags &= ~(HF_NMI_MASK | HF_IRET_MASK);
> +		kvm_make_request(KVM_REQ_EVENT, &svm->vcpu);
> +	}
>  
>  	svm->vcpu.arch.nmi_injected = false;
>  	kvm_clear_exception_queue(&svm->vcpu);
> @@ -3099,6 +3103,8 @@ static void svm_complete_interrupts(struct vcpu_svm *svm)
>  	if (!(exitintinfo & SVM_EXITINTINFO_VALID))
>  		return;
>  
> +	kvm_make_request(KVM_REQ_EVENT, &svm->vcpu);
> +
>  	vector = exitintinfo & SVM_EXITINTINFO_VEC_MASK;
>  	type = exitintinfo & SVM_EXITINTINFO_TYPE_MASK;
>  
> diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
> index 2fdcc98..d8edfe3 100644
> --- a/arch/x86/kvm/vmx.c
> +++ b/arch/x86/kvm/vmx.c
> @@ -3348,6 +3348,7 @@ static int handle_wrmsr(struct kvm_vcpu *vcpu)
>  
>  static int handle_tpr_below_threshold(struct kvm_vcpu *vcpu)
>  {
> +	kvm_make_request(KVM_REQ_EVENT, vcpu);
>  	return 1;
>  }
>  
> @@ -3360,6 +3361,8 @@ static int handle_interrupt_window(struct kvm_vcpu *vcpu)
>  	cpu_based_vm_exec_control &= ~CPU_BASED_VIRTUAL_INTR_PENDING;
>  	vmcs_write32(CPU_BASED_VM_EXEC_CONTROL, cpu_based_vm_exec_control);
>  
> +	kvm_make_request(KVM_REQ_EVENT, vcpu);
> +
>  	++vcpu->stat.irq_window_exits;
>  
>  	/*
> @@ -3616,6 +3619,7 @@ static int handle_nmi_window(struct kvm_vcpu *vcpu)
>  	cpu_based_vm_exec_control &= ~CPU_BASED_VIRTUAL_NMI_PENDING;
>  	vmcs_write32(CPU_BASED_VM_EXEC_CONTROL, cpu_based_vm_exec_control);
>  	++vcpu->stat.nmi_window_exits;
> +	kvm_make_request(KVM_REQ_EVENT, vcpu);
>  
>  	return 1;
>  }
> @@ -3849,6 +3853,8 @@ static void vmx_complete_interrupts(struct vcpu_vmx *vmx)
>  	if (!idtv_info_valid)
>  		return;
>  
> +	kvm_make_request(KVM_REQ_EVENT, &vmx->vcpu);
> +
>  	vector = idt_vectoring_info & VECTORING_INFO_VECTOR_MASK;
>  	type = idt_vectoring_info & VECTORING_INFO_TYPE_MASK;
>  
> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> index 76fbc32..38e91b6 100644
> --- a/arch/x86/kvm/x86.c
> +++ b/arch/x86/kvm/x86.c
> @@ -282,6 +282,8 @@ static void kvm_multiple_exception(struct kvm_vcpu *vcpu,
>  	u32 prev_nr;
>  	int class1, class2;
>  
> +	kvm_make_request(KVM_REQ_EVENT, vcpu);
> +
>  	if (!vcpu->arch.exception.pending) {
>  	queue:
>  		vcpu->arch.exception.pending = true;
> @@ -337,6 +339,7 @@ void kvm_inject_page_fault(struct kvm_vcpu *vcpu, unsigned long addr,
>  
>  void kvm_inject_nmi(struct kvm_vcpu *vcpu)
>  {
> +	kvm_make_request(KVM_REQ_EVENT, vcpu);
>  	vcpu->arch.nmi_pending = 1;
>  }
>  EXPORT_SYMBOL_GPL(kvm_inject_nmi);
> @@ -2356,6 +2359,8 @@ static int kvm_vcpu_ioctl_x86_set_vcpu_events(struct kvm_vcpu *vcpu,
>  	if (events->flags & KVM_VCPUEVENT_VALID_SIPI_VECTOR)
>  		vcpu->arch.sipi_vector = events->sipi_vector;
>  
> +	kvm_make_request(KVM_REQ_EVENT, vcpu);
> +
>  	return 0;
>  }
>  
> @@ -4059,6 +4064,7 @@ restart:
>  
>  	toggle_interruptibility(vcpu, vcpu->arch.emulate_ctxt.interruptibility);
>  	kvm_x86_ops->set_rflags(vcpu, vcpu->arch.emulate_ctxt.eflags);
> +	kvm_make_request(KVM_REQ_EVENT, vcpu);
>  	memcpy(vcpu->arch.regs, c->regs, sizeof c->regs);
>  	kvm_rip_write(vcpu, vcpu->arch.emulate_ctxt.eip);
>  
> @@ -4731,17 +4737,19 @@ static int vcpu_enter_guest(struct kvm_vcpu *vcpu)
>  		goto out;
>  	}
>  
> -	inject_pending_event(vcpu);
> +	if (kvm_check_request(KVM_REQ_EVENT, vcpu)) {
> +		inject_pending_event(vcpu);
>  
> -	/* enable NMI/IRQ window open exits if needed */
> -	if (vcpu->arch.nmi_pending)
> -		kvm_x86_ops->enable_nmi_window(vcpu);
> -	else if (kvm_cpu_has_interrupt(vcpu) || req_int_win)
> -		kvm_x86_ops->enable_irq_window(vcpu);
> +		/* enable NMI/IRQ window open exits if needed */
> +		if (vcpu->arch.nmi_pending)
> +			kvm_x86_ops->enable_nmi_window(vcpu);
> +		else if (kvm_cpu_has_interrupt(vcpu) || req_int_win)
> +			kvm_x86_ops->enable_irq_window(vcpu);
>  
> -	if (kvm_lapic_enabled(vcpu)) {
> -		update_cr8_intercept(vcpu);
> -		kvm_lapic_sync_to_vapic(vcpu);
> +		if (kvm_lapic_enabled(vcpu)) {
> +			update_cr8_intercept(vcpu);
> +			kvm_lapic_sync_to_vapic(vcpu);
> +		}
>  	}
>  
>  	srcu_read_unlock(&vcpu->kvm->srcu, vcpu->srcu_idx);
> @@ -4980,6 +4988,8 @@ int kvm_arch_vcpu_ioctl_set_regs(struct kvm_vcpu *vcpu, struct kvm_regs *regs)
>  
>  	vcpu->arch.exception.pending = false;
>  
> +	kvm_make_request(KVM_REQ_EVENT, vcpu);
> +
>  	return 0;
>  }
>  
> @@ -5043,6 +5053,7 @@ int kvm_arch_vcpu_ioctl_set_mpstate(struct kvm_vcpu *vcpu,
>  				    struct kvm_mp_state *mp_state)
>  {
>  	vcpu->arch.mp_state = mp_state->mp_state;
> +	kvm_make_request(KVM_REQ_EVENT, vcpu);
>  	return 0;
>  }
>  
> @@ -5077,6 +5088,7 @@ int kvm_task_switch(struct kvm_vcpu *vcpu, u16 tss_selector, int reason,
>  	memcpy(vcpu->arch.regs, c->regs, sizeof c->regs);
>  	kvm_rip_write(vcpu, vcpu->arch.emulate_ctxt.eip);
>  	kvm_x86_ops->set_rflags(vcpu, vcpu->arch.emulate_ctxt.eflags);
> +	kvm_make_request(KVM_REQ_EVENT, vcpu);
>  	return EMULATE_DONE;
>  }
>  EXPORT_SYMBOL_GPL(kvm_task_switch);
> @@ -5147,6 +5159,8 @@ int kvm_arch_vcpu_ioctl_set_sregs(struct kvm_vcpu *vcpu,
>  	    !is_protmode(vcpu))
>  		vcpu->arch.mp_state = KVM_MP_STATE_RUNNABLE;
>  
> +	kvm_make_request(KVM_REQ_EVENT, vcpu);
> +
>  	return 0;
>  }
>  
> @@ -5375,6 +5389,8 @@ int kvm_arch_vcpu_reset(struct kvm_vcpu *vcpu)
>  	vcpu->arch.dr6 = DR6_FIXED_1;
>  	vcpu->arch.dr7 = DR7_FIXED_1;
>  
> +	kvm_make_request(KVM_REQ_EVENT, vcpu);
> +
>  	return kvm_x86_ops->vcpu_reset(vcpu);
>  }
>  
> @@ -5683,6 +5699,7 @@ void kvm_set_rflags(struct kvm_vcpu *vcpu, unsigned long rflags)
>  	    kvm_is_linear_rip(vcpu, vcpu->arch.singlestep_rip))
>  		rflags |= X86_EFLAGS_TF;
>  	kvm_x86_ops->set_rflags(vcpu, rflags);
> +	kvm_make_request(KVM_REQ_EVENT, vcpu);
>  }
>  EXPORT_SYMBOL_GPL(kvm_set_rflags);
>  
> diff --git a/include/linux/kvm_host.h b/include/linux/kvm_host.h
> index c13cc48..e41e66b 100644
> --- a/include/linux/kvm_host.h
> +++ b/include/linux/kvm_host.h
> @@ -39,6 +39,7 @@
>  #define KVM_REQ_KVMCLOCK_UPDATE    8
>  #define KVM_REQ_KICK               9
>  #define KVM_REQ_DEACTIVATE_FPU    10
> +#define KVM_REQ_EVENT             11
>  
>  #define KVM_USERSPACE_IRQ_SOURCE_ID	0
>  
> -- 
> 1.7.1
> 
> --
> To unsubscribe from this list: send the line "unsubscribe kvm" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

--
			Gleb.