2017-11-29 16:48 GMT+08:00 Paolo Bonzini <pbonzini@xxxxxxxxxx>:
> On 29/11/2017 07:07, Wanpeng Li wrote:
>> From: Wanpeng Li <wanpeng.li@xxxxxxxxxxx>
>>
>> MSR_IA32_DEBUGCTLMSR is zeroed on VMEXIT, so it is saved/restored
>> each time during world switch. Jim from Google pointed out that
>> when running schbench in L2, vmx_vcpu_run will occupy 4% cpu time,
>> and the 25% of vmx_vcpu_run cpu time is occupied by get_debugctlmsr().
>> This patch caches the host IA32_DEBUGCTL MSR and saves/restores
>> the host IA32_DEBUGCTL msr when guest/host switches to avoid to
>> save/restore each time during world switch.
>>
>> Suggested-by: Jim Mattson <jmattson@xxxxxxxxxx>
>> Cc: Jim Mattson <jmattson@xxxxxxxxxx>
>> Cc: Paolo Bonzini <pbonzini@xxxxxxxxxx>
>> Cc: Radim Krčmář <rkrcmar@xxxxxxxxxx>
>> Signed-off-by: Wanpeng Li <wanpeng.li@xxxxxxxxxxx>
>
> The update_debugctlmsr should stay in vmx_vcpu_run so that tracing
> features work correctly. However, the get_debugctlmsr indeed can be
The tracing can't run except vCPU is schedule out, so why
update_debugctlmsr should stay in vmx_vcpu_run?
Regards,
Wanpeng Li
> moved to vmx_vcpu_load.
>
>> ---
>> arch/x86/include/asm/kvm_host.h | 1 +
>> arch/x86/kvm/vmx.c | 11 +++++------
>> 2 files changed, 6 insertions(+), 6 deletions(-)
>>
>> diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
>> index 63d34bc..c904250 100644
>> --- a/arch/x86/include/asm/kvm_host.h
>> +++ b/arch/x86/include/asm/kvm_host.h
>> @@ -619,6 +619,7 @@ struct kvm_vcpu_arch {
>> unsigned long dr7;
>> unsigned long eff_db[KVM_NR_DB_REGS];
>> unsigned long guest_debug_dr7;
>> + unsigned long debugctlmsr;
>
> Please rename to host_debugctlmsr and place it in struct vcpu_vmx.
>
> Thanks,
>
> Paolo
>
>> u64 msr_platform_info;
>> u64 msr_misc_features_enables;
>>
>> diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
>> index 8c7e816..b167bba 100644
>> --- a/arch/x86/kvm/vmx.c
>> +++ b/arch/x86/kvm/vmx.c
>> @@ -2326,6 +2326,7 @@ static void vmx_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
>>
>> vmx_vcpu_pi_load(vcpu, cpu);
>> vmx->host_pkru = read_pkru();
>> + vcpu->arch.debugctlmsr = get_debugctlmsr();
>> }
>>
>> static void vmx_vcpu_pi_put(struct kvm_vcpu *vcpu)
>> @@ -2347,6 +2348,9 @@ static void vmx_vcpu_put(struct kvm_vcpu *vcpu)
>> vmx_vcpu_pi_put(vcpu);
>>
>> __vmx_load_host_state(to_vmx(vcpu));
>> + /* MSR_IA32_DEBUGCTLMSR is zeroed on vmexit. Restore it if needed */
>> + if (vcpu->arch.debugctlmsr)
>> + update_debugctlmsr(vcpu->arch.debugctlmsr);
>> }
>>
>> static bool emulation_required(struct kvm_vcpu *vcpu)
>> @@ -9346,7 +9350,7 @@ static void vmx_arm_hv_timer(struct kvm_vcpu *vcpu)
>> static void __noclone vmx_vcpu_run(struct kvm_vcpu *vcpu)
>> {
>> struct vcpu_vmx *vmx = to_vmx(vcpu);
>> - unsigned long debugctlmsr, cr3, cr4;
>> + unsigned long cr3, cr4;
>>
>> /* Record the guest's net vcpu time for enforced NMI injections. */
>> if (unlikely(!enable_vnmi &&
>> @@ -9399,7 +9403,6 @@ static void __noclone vmx_vcpu_run(struct kvm_vcpu *vcpu)
>> __write_pkru(vcpu->arch.pkru);
>>
>> atomic_switch_perf_msrs(vmx);
>> - debugctlmsr = get_debugctlmsr();
>>
>> vmx_arm_hv_timer(vcpu);
>>
>> @@ -9509,10 +9512,6 @@ static void __noclone vmx_vcpu_run(struct kvm_vcpu *vcpu)
>> #endif
>> );
>>
>> - /* MSR_IA32_DEBUGCTLMSR is zeroed on vmexit. Restore it if needed */
>> - if (debugctlmsr)
>> - update_debugctlmsr(debugctlmsr);
>> -
>> #ifndef CONFIG_X86_64
>> /*
>> * The sysexit path does not restore ds/es, so we must set them to
>>
>
