[PATCH] kvm: add halt-polling cpu usage stats

[PATCH] kvm: add halt-polling cpu usage stats

[Jon Cargille]

From: David Matlack <dmatlack@google.com>

Two new stats for exposing halt-polling cpu usage:
halt_poll_success_ns
halt_poll_fail_ns

Thus sum of these 2 stats is the total cpu time spent polling. "success"
means the VCPU polled until a virtual interrupt was delivered. "fail"
means the VCPU had to schedule out (either because the maximum poll time
was reached or it needed to yield the CPU).

To avoid touching every arch's kvm_vcpu_stat struct, only update and
export halt-polling cpu usage stats if we're on x86.

Exporting cpu usage as a u64 and in nanoseconds means we will overflow at
~500 years, which seems reasonably large.

Signed-off-by: David Matlack <dmatlack@google.com>
Signed-off-by: Jon Cargille <jcargill@google.com>
Reviewed-by: Jim Mattson <jmattson@google.com>

---
 arch/x86/include/asm/kvm_host.h |  2 ++
 arch/x86/kvm/x86.c              |  2 ++
 virt/kvm/kvm_main.c             | 20 +++++++++++++++++---
 3 files changed, 21 insertions(+), 3 deletions(-)

diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
index a239a297be33..3287159ab15b 100644
--- a/arch/x86/include/asm/kvm_host.h
+++ b/arch/x86/include/asm/kvm_host.h
@@ -1032,6 +1032,8 @@ struct kvm_vcpu_stat {
 	u64 irq_injections;
 	u64 nmi_injections;
 	u64 req_event;
+	u64 halt_poll_success_ns;
+	u64 halt_poll_fail_ns;
 };
 
 struct x86_instruction_info;
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 8c0b77ac8dc6..9736d91ce877 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -217,6 +217,8 @@ struct kvm_stats_debugfs_item debugfs_entries[] = {
 	VCPU_STAT("nmi_injections", nmi_injections),
 	VCPU_STAT("req_event", req_event),
 	VCPU_STAT("l1d_flush", l1d_flush),
+	VCPU_STAT( "halt_poll_success_ns", halt_poll_success_ns),
+	VCPU_STAT( "halt_poll_fail_ns", halt_poll_fail_ns),
 	VM_STAT("mmu_shadow_zapped", mmu_shadow_zapped),
 	VM_STAT("mmu_pte_write", mmu_pte_write),
 	VM_STAT("mmu_pte_updated", mmu_pte_updated),
diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
index 33e1eee96f75..348b4a6bde53 100644
--- a/virt/kvm/kvm_main.c
+++ b/virt/kvm/kvm_main.c
@@ -2664,19 +2664,30 @@ static int kvm_vcpu_check_block(struct kvm_vcpu *vcpu)
 	return ret;
 }
 
+static inline void
+update_halt_poll_stats(struct kvm_vcpu *vcpu, u64 poll_ns, bool waited)
+{
+#ifdef CONFIG_X86
+	if (waited)
+		vcpu->stat.halt_poll_fail_ns += poll_ns;
+	else
+		vcpu->stat.halt_poll_success_ns += poll_ns;
+#endif
+}
+
 /*
  * The vCPU has executed a HLT instruction with in-kernel mode enabled.
  */
 void kvm_vcpu_block(struct kvm_vcpu *vcpu)
 {
-	ktime_t start, cur;
+	ktime_t start, cur, poll_end;
 	DECLARE_SWAITQUEUE(wait);
 	bool waited = false;
 	u64 block_ns;
 
 	kvm_arch_vcpu_blocking(vcpu);
 
-	start = cur = ktime_get();
+	start = cur = poll_end = ktime_get();
 	if (vcpu->halt_poll_ns && !kvm_arch_no_poll(vcpu)) {
 		ktime_t stop = ktime_add_ns(ktime_get(), vcpu->halt_poll_ns);
 
@@ -2692,7 +2703,7 @@ void kvm_vcpu_block(struct kvm_vcpu *vcpu)
 					++vcpu->stat.halt_poll_invalid;
 				goto out;
 			}
-			cur = ktime_get();
+			poll_end = cur = ktime_get();
 		} while (single_task_running() && ktime_before(cur, stop));
 	}
 
@@ -2712,6 +2723,9 @@ void kvm_vcpu_block(struct kvm_vcpu *vcpu)
 	kvm_arch_vcpu_unblocking(vcpu);
 	block_ns = ktime_to_ns(cur) - ktime_to_ns(start);
 
+	update_halt_poll_stats(
+		vcpu, ktime_to_ns(ktime_sub(poll_end, start)), waited);
+
 	if (!kvm_arch_no_poll(vcpu)) {
 		if (!vcpu_valid_wakeup(vcpu)) {
 			shrink_halt_poll_ns(vcpu);
-- 
2.26.2.303.gf8c07b1a785-goog

Re: [PATCH] kvm: add halt-polling cpu usage stats

[Paolo Bonzini]

On 08/05/20 20:22, Jon Cargille wrote:
> From: David Matlack <dmatlack@google.com>
> 
> Two new stats for exposing halt-polling cpu usage:
> halt_poll_success_ns
> halt_poll_fail_ns
> 
> Thus sum of these 2 stats is the total cpu time spent polling. "success"
> means the VCPU polled until a virtual interrupt was delivered. "fail"
> means the VCPU had to schedule out (either because the maximum poll time
> was reached or it needed to yield the CPU).
> 
> To avoid touching every arch's kvm_vcpu_stat struct, only update and
> export halt-polling cpu usage stats if we're on x86.

I fixed all the other architectures and queued it, thanks.

Paolo

> 
> Exporting cpu usage as a u64 and in nanoseconds means we will overflow at
> ~500 years, which seems reasonably large.
> 
> Signed-off-by: David Matlack <dmatlack@google.com>
> Signed-off-by: Jon Cargille <jcargill@google.com>
> Reviewed-by: Jim Mattson <jmattson@google.com>
> 
> ---
>  arch/x86/include/asm/kvm_host.h |  2 ++
>  arch/x86/kvm/x86.c              |  2 ++
>  virt/kvm/kvm_main.c             | 20 +++++++++++++++++---
>  3 files changed, 21 insertions(+), 3 deletions(-)
> 
> diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
> index a239a297be33..3287159ab15b 100644
> --- a/arch/x86/include/asm/kvm_host.h
> +++ b/arch/x86/include/asm/kvm_host.h
> @@ -1032,6 +1032,8 @@ struct kvm_vcpu_stat {
>  	u64 irq_injections;
>  	u64 nmi_injections;
>  	u64 req_event;
> +	u64 halt_poll_success_ns;
> +	u64 halt_poll_fail_ns;
>  };
>  
>  struct x86_instruction_info;
> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> index 8c0b77ac8dc6..9736d91ce877 100644
> --- a/arch/x86/kvm/x86.c
> +++ b/arch/x86/kvm/x86.c
> @@ -217,6 +217,8 @@ struct kvm_stats_debugfs_item debugfs_entries[] = {
>  	VCPU_STAT("nmi_injections", nmi_injections),
>  	VCPU_STAT("req_event", req_event),
>  	VCPU_STAT("l1d_flush", l1d_flush),
> +	VCPU_STAT( "halt_poll_success_ns", halt_poll_success_ns),
> +	VCPU_STAT( "halt_poll_fail_ns", halt_poll_fail_ns),
>  	VM_STAT("mmu_shadow_zapped", mmu_shadow_zapped),
>  	VM_STAT("mmu_pte_write", mmu_pte_write),
>  	VM_STAT("mmu_pte_updated", mmu_pte_updated),
> diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
> index 33e1eee96f75..348b4a6bde53 100644
> --- a/virt/kvm/kvm_main.c
> +++ b/virt/kvm/kvm_main.c
> @@ -2664,19 +2664,30 @@ static int kvm_vcpu_check_block(struct kvm_vcpu *vcpu)
>  	return ret;
>  }
>  
> +static inline void
> +update_halt_poll_stats(struct kvm_vcpu *vcpu, u64 poll_ns, bool waited)
> +{
> +#ifdef CONFIG_X86
> +	if (waited)
> +		vcpu->stat.halt_poll_fail_ns += poll_ns;
> +	else
> +		vcpu->stat.halt_poll_success_ns += poll_ns;
> +#endif
> +}
> +
>  /*
>   * The vCPU has executed a HLT instruction with in-kernel mode enabled.
>   */
>  void kvm_vcpu_block(struct kvm_vcpu *vcpu)
>  {
> -	ktime_t start, cur;
> +	ktime_t start, cur, poll_end;
>  	DECLARE_SWAITQUEUE(wait);
>  	bool waited = false;
>  	u64 block_ns;
>  
>  	kvm_arch_vcpu_blocking(vcpu);
>  
> -	start = cur = ktime_get();
> +	start = cur = poll_end = ktime_get();
>  	if (vcpu->halt_poll_ns && !kvm_arch_no_poll(vcpu)) {
>  		ktime_t stop = ktime_add_ns(ktime_get(), vcpu->halt_poll_ns);
>  
> @@ -2692,7 +2703,7 @@ void kvm_vcpu_block(struct kvm_vcpu *vcpu)
>  					++vcpu->stat.halt_poll_invalid;
>  				goto out;
>  			}
> -			cur = ktime_get();
> +			poll_end = cur = ktime_get();
>  		} while (single_task_running() && ktime_before(cur, stop));
>  	}
>  
> @@ -2712,6 +2723,9 @@ void kvm_vcpu_block(struct kvm_vcpu *vcpu)
>  	kvm_arch_vcpu_unblocking(vcpu);
>  	block_ns = ktime_to_ns(cur) - ktime_to_ns(start);
>  
> +	update_halt_poll_stats(
> +		vcpu, ktime_to_ns(ktime_sub(poll_end, start)), waited);
> +
>  	if (!kvm_arch_no_poll(vcpu)) {
>  		if (!vcpu_valid_wakeup(vcpu)) {
>  			shrink_halt_poll_ns(vcpu);
>