xiaomi-sm8450/android_kernel_xiaomi_sm8450
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

KVM: X86: Improve latency for single target IPI fastpath

Browse Source 
IPI and Timer cause the main MSRs write vmexits in cloud environment
observation, let's optimize virtual IPI latency more aggressively to
inject target IPI as soon as possible.

Running kvm-unit-tests/vmexit.flat IPI testing on SKX server, disable
adaptive advance lapic timer and adaptive halt-polling to avoid the
interference, this patch can give another 7% improvement.

w/o fastpath   -> x86.c fastpath      4238 -> 3543  16.4%
x86.c fastpath -> vmx.c fastpath      3543 -> 3293     7%
w/o fastpath   -> vmx.c fastpath      4238 -> 3293  22.3%

Cc: Haiwei Li <lihaiwei@tencent.com>
Signed-off-by: Wanpeng Li <wanpengli@tencent.com>
Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Message-Id: <20200410174703.1138-3-sean.j.christopherson@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
This commit is contained in:
Wanpeng Li
2020-04-10 10:47:03 -07:00
committed by Paolo Bonzini
parent 873e1da169
commit a9ab13ff6e
4 changed files with 34 additions and 23 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
arch/x86/kvm/x86.c
View File

@@ -8179,7 +8179,7 @@ static int vcpu_enter_guest(struct kvm_vcpu *vcpu)
bool req_int_win =
dm_request_for_irq_injection(vcpu) &&
kvm_cpu_accept_dm_intr(vcpu);
enum exit_fastpath_completion exit_fastpath = EXIT_FASTPATH_NONE;
enum exit_fastpath_completion exit_fastpath;
bool req_immediate_exit = false;
@@ -8406,7 +8406,7 @@ static int vcpu_enter_guest(struct kvm_vcpu *vcpu)
vcpu->arch.switch_db_regs &= ~KVM_DEBUGREG_RELOAD;
}
kvm_x86_ops.run(vcpu);
exit_fastpath = kvm_x86_ops.run(vcpu);
/*
* Do this here before restoring debug registers on the host. And
@@ -8438,7 +8438,7 @@ static int vcpu_enter_guest(struct kvm_vcpu *vcpu)
vcpu->mode = OUTSIDE_GUEST_MODE;
smp_wmb();
kvm_x86_ops.handle_exit_irqoff(vcpu, &exit_fastpath);
kvm_x86_ops.handle_exit_irqoff(vcpu);
/*
* Consume any pending interrupts, including the possible source of