2018-04-18 2:08 GMT+08:00 Michael S. Tsirkin <mst@xxxxxxxxxx>:
> On Tue, Apr 17, 2018 at 01:24:15AM -0700, Wanpeng Li wrote:
>> From: Wanpeng Li <wanpengli@xxxxxxxxxxx>
>>
>> This patch adds support for KVM_CAP_X86_DISABLE_EXITS. Provides userspace with
>> per-VM capability(KVM_CAP_X86_DISABLE_EXITS) to not intercept MWAIT/HLT/PAUSE
>> in order that to improve latency in some workloads.
>>
>> Cc: Paolo Bonzini <pbonzini@xxxxxxxxxx>
>> Cc: Radim Krčmář <rkrcmar@xxxxxxxxxx>
>> Cc: Eduardo Habkost <ehabkost@xxxxxxxxxx>
>> Signed-off-by: Wanpeng Li <wanpengli@xxxxxxxxxxx>
>> ---
>>
>> linux-headers/linux/kvm.h | 6 +++++-
>> target/i386/cpu.h | 2 ++
>> target/i386/kvm.c | 16 ++++++++++++++++
>> 3 files changed, 23 insertions(+), 1 deletion(-)
>>
>> diff --git a/linux-headers/linux/kvm.h b/linux-headers/linux/kvm.h
>> index a167be8..857df15 100644
>> --- a/linux-headers/linux/kvm.h
>> +++ b/linux-headers/linux/kvm.h
>> @@ -925,7 +925,7 @@ struct kvm_ppc_resize_hpt {
>> #define KVM_CAP_S390_GS 140
>> #define KVM_CAP_S390_AIS 141
>> #define KVM_CAP_SPAPR_TCE_VFIO 142
>> -#define KVM_CAP_X86_GUEST_MWAIT 143
>> +#define KVM_CAP_X86_DISABLE_EXITS 143
>> #define KVM_CAP_ARM_USER_IRQ 144
>> #define KVM_CAP_S390_CMMA_MIGRATION 145
>> #define KVM_CAP_PPC_FWNMI 146
>> @@ -1508,6 +1508,10 @@ struct kvm_assigned_msix_entry {
>> #define KVM_X2APIC_API_USE_32BIT_IDS (1ULL << 0)
>> #define KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK (1ULL << 1)
>>
>> +#define KVM_X86_DISABLE_EXITS_MWAIT (1 << 0)
>> +#define KVM_X86_DISABLE_EXITS_HLT (1 << 1)
>> +#define KVM_X86_DISABLE_EXITS_PAUSE (1 << 2)
>> +
>> /* Available with KVM_CAP_ARM_USER_IRQ */
>>
>> /* Bits for run->s.regs.device_irq_level */
>> diff --git a/target/i386/cpu.h b/target/i386/cpu.h
>> index 1b219fa..965de1b 100644
>> --- a/target/i386/cpu.h
>> +++ b/target/i386/cpu.h
>> @@ -685,6 +685,8 @@ typedef uint32_t FeatureWordArray[FEATURE_WORDS];
>> #define CPUID_7_0_EDX_AVX512_4FMAPS (1U << 3) /* AVX512 Multiply Accumulation Single Precision */
>> #define CPUID_7_0_EDX_SPEC_CTRL (1U << 26) /* Speculation Control */
>>
>> +#define KVM_PV_UNHALT (1U << 7)
>> +
>
> Why don't we use KVM_FEATURE_PV_UNHALT from kvm_para.h?
>
>> #define KVM_HINTS_DEDICATED (1U << 0)
>>
>
> BTW I wonder whether we should switch to a value from
> kvm_para.h? I'll send a version to do it, pls take a look.
Yeah, your patchset looks good.
Regards,
Wanpeng Li
>
>
>> #define CPUID_8000_0008_EBX_IBPB (1U << 12) /* Indirect Branch Prediction Barrier */
>> diff --git a/target/i386/kvm.c b/target/i386/kvm.c
>> index 6c49954..3e99830 100644
>> --- a/target/i386/kvm.c
>> +++ b/target/i386/kvm.c
>> @@ -1029,6 +1029,22 @@ int kvm_arch_init_vcpu(CPUState *cs)
>> }
>> }
>>
>> + if (env->features[FEAT_KVM_HINTS] & KVM_HINTS_DEDICATED) {
>> + int disable_exits = kvm_check_extension(cs->kvm_state, KVM_CAP_X86_DISABLE_EXITS);
>> +
>> + if (disable_exits) {
>> + disable_exits &= (KVM_X86_DISABLE_EXITS_MWAIT |
>> + KVM_X86_DISABLE_EXITS_HLT |
>> + KVM_X86_DISABLE_EXITS_PAUSE);
>> + if (env->user_features[FEAT_KVM] & KVM_PV_UNHALT) {
>> + disable_exits &= ~KVM_X86_DISABLE_EXITS_HLT;
>> + }
>> + }
>> + if (kvm_vm_enable_cap(cs->kvm_state, KVM_CAP_X86_DISABLE_EXITS, 0, disable_exits)) {
>> + error_report("kvm: DISABLE EXITS not supported");
>> + }
>> + }
>> +
>> qemu_add_vm_change_state_handler(cpu_update_state, env);
>>
>> c = cpuid_find_entry(&cpuid_data.cpuid, 1, 0);
>> --
>> 2.7.4
