diff mbox series

[v7,10/39] KVM: x86: hyper-v: Don't use sparse_set_to_vcpu_mask() in kvm_hv_send_ipi()

Message ID 20220613133922.2875594-11-vkuznets@redhat.com (mailing list archive)
State New, archived
Headers show
Series KVM: x86: hyper-v: Fine-grained TLB flush + L2 TLB flush features | expand

Commit Message

Vitaly Kuznetsov June 13, 2022, 1:38 p.m. UTC
Get rid of on-stack allocation of vcpu_mask and optimize kvm_hv_send_ipi()
for a smaller number of vCPUs in the request. When Hyper-V TLB flush
is in  use, HvSendSyntheticClusterIpi{,Ex} calls are not commonly used to
send IPIs to a large number of vCPUs (and are rarely used in general).

Introduce hv_is_vp_in_sparse_set() to directly check if the specified
VP_ID is present in sparse vCPU set.

Reviewed-by: Maxim Levitsky <mlevitsk@redhat.com>
Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
---
 arch/x86/kvm/hyperv.c | 37 ++++++++++++++++++++++++++-----------
 1 file changed, 26 insertions(+), 11 deletions(-)

Comments

Paolo Bonzini June 21, 2022, 12:58 p.m. UTC | #1
On 6/13/22 15:38, Vitaly Kuznetsov wrote:
> Get rid of on-stack allocation of vcpu_mask and optimize kvm_hv_send_ipi()
> for a smaller number of vCPUs in the request. When Hyper-V TLB flush
> is in  use, HvSendSyntheticClusterIpi{,Ex} calls are not commonly used to
> send IPIs to a large number of vCPUs (and are rarely used in general).
> 
> Introduce hv_is_vp_in_sparse_set() to directly check if the specified
> VP_ID is present in sparse vCPU set.
> 
> Reviewed-by: Maxim Levitsky <mlevitsk@redhat.com>
> Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
> ---
>   arch/x86/kvm/hyperv.c | 37 ++++++++++++++++++++++++++-----------
>   1 file changed, 26 insertions(+), 11 deletions(-)

I'm a bit confused by this patch being in this series.

Just to be clear, PV IPI does *not* support the VP_ID, right?  And since 
patch 12 only affects the sparse banks, the patch does not have any 
other dependency.  Is this correct?

Paolo

> diff --git a/arch/x86/kvm/hyperv.c b/arch/x86/kvm/hyperv.c
> index f41153c71beb..269a5fcca31b 100644
> --- a/arch/x86/kvm/hyperv.c
> +++ b/arch/x86/kvm/hyperv.c
> @@ -1747,6 +1747,25 @@ static void sparse_set_to_vcpu_mask(struct kvm *kvm, u64 *sparse_banks,
>   	}
>   }
>   
> +static bool hv_is_vp_in_sparse_set(u32 vp_id, u64 valid_bank_mask, u64 sparse_banks[])
> +{
> +	int bank, sbank = 0;
> +
> +	if (!test_bit(vp_id / HV_VCPUS_PER_SPARSE_BANK,
> +		      (unsigned long *)&valid_bank_mask))
> +		return false;
> +
> +	for_each_set_bit(bank, (unsigned long *)&valid_bank_mask,
> +			 KVM_HV_MAX_SPARSE_VCPU_SET_BITS) {
> +		if (bank == vp_id / HV_VCPUS_PER_SPARSE_BANK)
> +			break;
> +		sbank++;
> +	}
> +
> +	return test_bit(vp_id % HV_VCPUS_PER_SPARSE_BANK,
> +			(unsigned long *)&sparse_banks[sbank]);
> +}
> +
>   struct kvm_hv_hcall {
>   	u64 param;
>   	u64 ingpa;
> @@ -2029,8 +2048,8 @@ static u64 kvm_hv_flush_tlb(struct kvm_vcpu *vcpu, struct kvm_hv_hcall *hc)
>   		((u64)hc->rep_cnt << HV_HYPERCALL_REP_COMP_OFFSET);
>   }
>   
> -static void kvm_send_ipi_to_many(struct kvm *kvm, u32 vector,
> -				 unsigned long *vcpu_bitmap)
> +static void kvm_hv_send_ipi_to_many(struct kvm *kvm, u32 vector,
> +				    u64 *sparse_banks, u64 valid_bank_mask)
>   {
>   	struct kvm_lapic_irq irq = {
>   		.delivery_mode = APIC_DM_FIXED,
> @@ -2040,7 +2059,10 @@ static void kvm_send_ipi_to_many(struct kvm *kvm, u32 vector,
>   	unsigned long i;
>   
>   	kvm_for_each_vcpu(i, vcpu, kvm) {
> -		if (vcpu_bitmap && !test_bit(i, vcpu_bitmap))
> +		if (sparse_banks &&
> +		    !hv_is_vp_in_sparse_set(kvm_hv_get_vpindex(vcpu),
> +					    valid_bank_mask,
> +					    sparse_banks))
>   			continue;
>   
>   		/* We fail only when APIC is disabled */
> @@ -2053,7 +2075,6 @@ static u64 kvm_hv_send_ipi(struct kvm_vcpu *vcpu, struct kvm_hv_hcall *hc)
>   	struct kvm *kvm = vcpu->kvm;
>   	struct hv_send_ipi_ex send_ipi_ex;
>   	struct hv_send_ipi send_ipi;
> -	DECLARE_BITMAP(vcpu_mask, KVM_MAX_VCPUS);
>   	u64 valid_bank_mask;
>   	u64 sparse_banks[KVM_HV_MAX_SPARSE_VCPU_SET_BITS];
>   	u32 vector;
> @@ -2115,13 +2136,7 @@ static u64 kvm_hv_send_ipi(struct kvm_vcpu *vcpu, struct kvm_hv_hcall *hc)
>   	if ((vector < HV_IPI_LOW_VECTOR) || (vector > HV_IPI_HIGH_VECTOR))
>   		return HV_STATUS_INVALID_HYPERCALL_INPUT;
>   
> -	if (all_cpus) {
> -		kvm_send_ipi_to_many(kvm, vector, NULL);
> -	} else {
> -		sparse_set_to_vcpu_mask(kvm, sparse_banks, valid_bank_mask, vcpu_mask);
> -
> -		kvm_send_ipi_to_many(kvm, vector, vcpu_mask);
> -	}
> +	kvm_hv_send_ipi_to_many(kvm, vector, all_cpus ? NULL : sparse_banks, valid_bank_mask);
>   
>   ret_success:
>   	return HV_STATUS_SUCCESS;
Vitaly Kuznetsov June 21, 2022, 1:17 p.m. UTC | #2
Paolo Bonzini <pbonzini@redhat.com> writes:

> On 6/13/22 15:38, Vitaly Kuznetsov wrote:
>> Get rid of on-stack allocation of vcpu_mask and optimize kvm_hv_send_ipi()
>> for a smaller number of vCPUs in the request. When Hyper-V TLB flush
>> is in  use, HvSendSyntheticClusterIpi{,Ex} calls are not commonly used to
>> send IPIs to a large number of vCPUs (and are rarely used in general).
>> 
>> Introduce hv_is_vp_in_sparse_set() to directly check if the specified
>> VP_ID is present in sparse vCPU set.
>> 
>> Reviewed-by: Maxim Levitsky <mlevitsk@redhat.com>
>> Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
>> ---
>>   arch/x86/kvm/hyperv.c | 37 ++++++++++++++++++++++++++-----------
>>   1 file changed, 26 insertions(+), 11 deletions(-)
>
> I'm a bit confused by this patch being in this series.
>
> Just to be clear, PV IPI does *not* support the VP_ID, right? 

Hm, with Hyper-V PV IPI hypercall vCPUs are also addressed by their
VP_IDs, not by their APIC ids so similar to Hyper-V PV TLB flush we need
to convert the supplied set (either flat u64 bitmask of VP_IDs for
non-EX hypercall or a sparse set for -EX).

> And since patch 12 only affects the sparse banks, the patch does not
> have any other dependency.  Is this correct?
>

This patch introduces hv_is_vp_in_sparse_set() with its first user:
kvm_hv_send_ipi_to_many(). Later in the series, hv_is_vp_in_sparse_set()
gets a second user: kvm_hv_flush_tlb() for 'nested' case where we
actually have to traverse all vCPUs checking *nested* VP_IDs (as KVM
doesn't currently keep a convenient L2-VP_ID-to-kvm-vcpu-id mapping.

> Paolo
>
>> diff --git a/arch/x86/kvm/hyperv.c b/arch/x86/kvm/hyperv.c
>> index f41153c71beb..269a5fcca31b 100644
>> --- a/arch/x86/kvm/hyperv.c
>> +++ b/arch/x86/kvm/hyperv.c
>> @@ -1747,6 +1747,25 @@ static void sparse_set_to_vcpu_mask(struct kvm *kvm, u64 *sparse_banks,
>>   	}
>>   }
>>   
>> +static bool hv_is_vp_in_sparse_set(u32 vp_id, u64 valid_bank_mask, u64 sparse_banks[])
>> +{
>> +	int bank, sbank = 0;
>> +
>> +	if (!test_bit(vp_id / HV_VCPUS_PER_SPARSE_BANK,
>> +		      (unsigned long *)&valid_bank_mask))
>> +		return false;
>> +
>> +	for_each_set_bit(bank, (unsigned long *)&valid_bank_mask,
>> +			 KVM_HV_MAX_SPARSE_VCPU_SET_BITS) {
>> +		if (bank == vp_id / HV_VCPUS_PER_SPARSE_BANK)
>> +			break;
>> +		sbank++;
>> +	}
>> +
>> +	return test_bit(vp_id % HV_VCPUS_PER_SPARSE_BANK,
>> +			(unsigned long *)&sparse_banks[sbank]);
>> +}
>> +
>>   struct kvm_hv_hcall {
>>   	u64 param;
>>   	u64 ingpa;
>> @@ -2029,8 +2048,8 @@ static u64 kvm_hv_flush_tlb(struct kvm_vcpu *vcpu, struct kvm_hv_hcall *hc)
>>   		((u64)hc->rep_cnt << HV_HYPERCALL_REP_COMP_OFFSET);
>>   }
>>   
>> -static void kvm_send_ipi_to_many(struct kvm *kvm, u32 vector,
>> -				 unsigned long *vcpu_bitmap)
>> +static void kvm_hv_send_ipi_to_many(struct kvm *kvm, u32 vector,
>> +				    u64 *sparse_banks, u64 valid_bank_mask)
>>   {
>>   	struct kvm_lapic_irq irq = {
>>   		.delivery_mode = APIC_DM_FIXED,
>> @@ -2040,7 +2059,10 @@ static void kvm_send_ipi_to_many(struct kvm *kvm, u32 vector,
>>   	unsigned long i;
>>   
>>   	kvm_for_each_vcpu(i, vcpu, kvm) {
>> -		if (vcpu_bitmap && !test_bit(i, vcpu_bitmap))
>> +		if (sparse_banks &&
>> +		    !hv_is_vp_in_sparse_set(kvm_hv_get_vpindex(vcpu),
>> +					    valid_bank_mask,
>> +					    sparse_banks))
>>   			continue;
>>   
>>   		/* We fail only when APIC is disabled */
>> @@ -2053,7 +2075,6 @@ static u64 kvm_hv_send_ipi(struct kvm_vcpu *vcpu, struct kvm_hv_hcall *hc)
>>   	struct kvm *kvm = vcpu->kvm;
>>   	struct hv_send_ipi_ex send_ipi_ex;
>>   	struct hv_send_ipi send_ipi;
>> -	DECLARE_BITMAP(vcpu_mask, KVM_MAX_VCPUS);
>>   	u64 valid_bank_mask;
>>   	u64 sparse_banks[KVM_HV_MAX_SPARSE_VCPU_SET_BITS];
>>   	u32 vector;
>> @@ -2115,13 +2136,7 @@ static u64 kvm_hv_send_ipi(struct kvm_vcpu *vcpu, struct kvm_hv_hcall *hc)
>>   	if ((vector < HV_IPI_LOW_VECTOR) || (vector > HV_IPI_HIGH_VECTOR))
>>   		return HV_STATUS_INVALID_HYPERCALL_INPUT;
>>   
>> -	if (all_cpus) {
>> -		kvm_send_ipi_to_many(kvm, vector, NULL);
>> -	} else {
>> -		sparse_set_to_vcpu_mask(kvm, sparse_banks, valid_bank_mask, vcpu_mask);
>> -
>> -		kvm_send_ipi_to_many(kvm, vector, vcpu_mask);
>> -	}
>> +	kvm_hv_send_ipi_to_many(kvm, vector, all_cpus ? NULL : sparse_banks, valid_bank_mask);
>>   
>>   ret_success:
>>   	return HV_STATUS_SUCCESS;
>
Paolo Bonzini June 22, 2022, 1:59 p.m. UTC | #3
On 6/21/22 15:17, Vitaly Kuznetsov wrote:
>>
>> Just to be clear, PV IPI does*not*  support the VP_ID, right?
> Hm, with Hyper-V PV IPI hypercall vCPUs are also addressed by their
> VP_IDs, not by their APIC ids so similar to Hyper-V PV TLB flush we need
> to convert the supplied set (either flat u64 bitmask of VP_IDs for
> non-EX hypercall or a sparse set for -EX).
> 

So this means the series needs a v8, right?

Paolo
Vitaly Kuznetsov June 22, 2022, 2:28 p.m. UTC | #4
Paolo Bonzini <pbonzini@redhat.com> writes:

> On 6/21/22 15:17, Vitaly Kuznetsov wrote:
>>>
>>> Just to be clear, PV IPI does*not*  support the VP_ID, right?
>> Hm, with Hyper-V PV IPI hypercall vCPUs are also addressed by their
>> VP_IDs, not by their APIC ids so similar to Hyper-V PV TLB flush we need
>> to convert the supplied set (either flat u64 bitmask of VP_IDs for
>> non-EX hypercall or a sparse set for -EX).
>> 
>
> So this means the series needs a v8, right?
>

No, I was just trying to explaini what the patch is doing in the series,
it looks good to me (but I'm biased, of course).
diff mbox series

Patch

diff --git a/arch/x86/kvm/hyperv.c b/arch/x86/kvm/hyperv.c
index f41153c71beb..269a5fcca31b 100644
--- a/arch/x86/kvm/hyperv.c
+++ b/arch/x86/kvm/hyperv.c
@@ -1747,6 +1747,25 @@  static void sparse_set_to_vcpu_mask(struct kvm *kvm, u64 *sparse_banks,
 	}
 }
 
+static bool hv_is_vp_in_sparse_set(u32 vp_id, u64 valid_bank_mask, u64 sparse_banks[])
+{
+	int bank, sbank = 0;
+
+	if (!test_bit(vp_id / HV_VCPUS_PER_SPARSE_BANK,
+		      (unsigned long *)&valid_bank_mask))
+		return false;
+
+	for_each_set_bit(bank, (unsigned long *)&valid_bank_mask,
+			 KVM_HV_MAX_SPARSE_VCPU_SET_BITS) {
+		if (bank == vp_id / HV_VCPUS_PER_SPARSE_BANK)
+			break;
+		sbank++;
+	}
+
+	return test_bit(vp_id % HV_VCPUS_PER_SPARSE_BANK,
+			(unsigned long *)&sparse_banks[sbank]);
+}
+
 struct kvm_hv_hcall {
 	u64 param;
 	u64 ingpa;
@@ -2029,8 +2048,8 @@  static u64 kvm_hv_flush_tlb(struct kvm_vcpu *vcpu, struct kvm_hv_hcall *hc)
 		((u64)hc->rep_cnt << HV_HYPERCALL_REP_COMP_OFFSET);
 }
 
-static void kvm_send_ipi_to_many(struct kvm *kvm, u32 vector,
-				 unsigned long *vcpu_bitmap)
+static void kvm_hv_send_ipi_to_many(struct kvm *kvm, u32 vector,
+				    u64 *sparse_banks, u64 valid_bank_mask)
 {
 	struct kvm_lapic_irq irq = {
 		.delivery_mode = APIC_DM_FIXED,
@@ -2040,7 +2059,10 @@  static void kvm_send_ipi_to_many(struct kvm *kvm, u32 vector,
 	unsigned long i;
 
 	kvm_for_each_vcpu(i, vcpu, kvm) {
-		if (vcpu_bitmap && !test_bit(i, vcpu_bitmap))
+		if (sparse_banks &&
+		    !hv_is_vp_in_sparse_set(kvm_hv_get_vpindex(vcpu),
+					    valid_bank_mask,
+					    sparse_banks))
 			continue;
 
 		/* We fail only when APIC is disabled */
@@ -2053,7 +2075,6 @@  static u64 kvm_hv_send_ipi(struct kvm_vcpu *vcpu, struct kvm_hv_hcall *hc)
 	struct kvm *kvm = vcpu->kvm;
 	struct hv_send_ipi_ex send_ipi_ex;
 	struct hv_send_ipi send_ipi;
-	DECLARE_BITMAP(vcpu_mask, KVM_MAX_VCPUS);
 	u64 valid_bank_mask;
 	u64 sparse_banks[KVM_HV_MAX_SPARSE_VCPU_SET_BITS];
 	u32 vector;
@@ -2115,13 +2136,7 @@  static u64 kvm_hv_send_ipi(struct kvm_vcpu *vcpu, struct kvm_hv_hcall *hc)
 	if ((vector < HV_IPI_LOW_VECTOR) || (vector > HV_IPI_HIGH_VECTOR))
 		return HV_STATUS_INVALID_HYPERCALL_INPUT;
 
-	if (all_cpus) {
-		kvm_send_ipi_to_many(kvm, vector, NULL);
-	} else {
-		sparse_set_to_vcpu_mask(kvm, sparse_banks, valid_bank_mask, vcpu_mask);
-
-		kvm_send_ipi_to_many(kvm, vector, vcpu_mask);
-	}
+	kvm_hv_send_ipi_to_many(kvm, vector, all_cpus ? NULL : sparse_banks, valid_bank_mask);
 
 ret_success:
 	return HV_STATUS_SUCCESS;