| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Date: Mon, 7 Feb 2022 17:54:29 +0200
From: Maxim Levitsky <mlevitsk@...hat.com>
To: kvm@...r.kernel.org
Cc: Tony Luck <tony.luck@...el.com>,
"Chang S. Bae" <chang.seok.bae@...el.com>,
Thomas Gleixner <tglx@...utronix.de>,
Wanpeng Li <wanpengli@...cent.com>,
Ingo Molnar <mingo@...hat.com>,
Vitaly Kuznetsov <vkuznets@...hat.com>,
Pawan Gupta <pawan.kumar.gupta@...ux.intel.com>,
Dave Hansen <dave.hansen@...ux.intel.com>,
Paolo Bonzini <pbonzini@...hat.com>,
linux-kernel@...r.kernel.org,
Rodrigo Vivi <rodrigo.vivi@...el.com>,
"H. Peter Anvin" <hpa@...or.com>,
intel-gvt-dev@...ts.freedesktop.org,
Joonas Lahtinen <joonas.lahtinen@...ux.intel.com>,
Joerg Roedel <joro@...tes.org>,
Sean Christopherson <seanjc@...gle.com>,
David Airlie <airlied@...ux.ie>,
Zhi Wang <zhi.a.wang@...el.com>,
Brijesh Singh <brijesh.singh@....com>,
Jim Mattson <jmattson@...gle.com>, x86@...nel.org,
Daniel Vetter <daniel@...ll.ch>,
Borislav Petkov <bp@...en8.de>,
Zhenyu Wang <zhenyuw@...ux.intel.com>,
Kan Liang <kan.liang@...ux.intel.com>,
Jani Nikula <jani.nikula@...ux.intel.com>,
Maxim Levitsky <mlevitsk@...hat.com>
Subject: [PATCH RESEND 12/30] KVM: x86: SVM: allow AVIC to co-exist with a nested guest running
Inhibit the AVIC of the vCPU that is running nested for the duration of the
nested run, so that all interrupts arriving from both its vCPU siblings
and from KVM are delivered using normal IPIs and cause that vCPU to vmexit.
Note that unlike normal AVIC inhibition, there is no need to
update the AVIC mmio memslot, because the nested guest uses its
own set of paging tables.
That also means that AVIC doesn't need to be inhibited VM wide.
Note that in the theory when a nested guest doesn't intercept
physical interrupts, we could continue using AVIC to deliver them
to it but don't bother doing so for now. Plus when nested AVIC
is implemented, the nested guest will likely use it, which will
not allow this optimization to be used
(can't use real AVIC to support both L1 and L2 at the same time)
Signed-off-by: Maxim Levitsky <mlevitsk@...hat.com>
---
arch/x86/include/asm/kvm-x86-ops.h | 1 +
arch/x86/include/asm/kvm_host.h | 8 +++++++-
arch/x86/kvm/svm/avic.c | 7 ++++++-
arch/x86/kvm/svm/nested.c | 15 ++++++++++-----
arch/x86/kvm/svm/svm.c | 31 +++++++++++++++++++-----------
arch/x86/kvm/svm/svm.h | 1 +
arch/x86/kvm/x86.c | 18 +++++++++++++++--
7 files changed, 61 insertions(+), 20 deletions(-)
diff --git a/arch/x86/include/asm/kvm-x86-ops.h b/arch/x86/include/asm/kvm-x86-ops.h
index 9e37dc3d88636..c0d8f351dcbc0 100644
--- a/arch/x86/include/asm/kvm-x86-ops.h
+++ b/arch/x86/include/asm/kvm-x86-ops.h
@@ -125,6 +125,7 @@ KVM_X86_OP_NULL(migrate_timers)
KVM_X86_OP(msr_filter_changed)
KVM_X86_OP_NULL(complete_emulated_msr)
KVM_X86_OP(vcpu_deliver_sipi_vector)
+KVM_X86_OP_NULL(vcpu_has_apicv_inhibit_condition);
#undef KVM_X86_OP
#undef KVM_X86_OP_NULL
diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
index c371ee7e45f78..256539c0481c5 100644
--- a/arch/x86/include/asm/kvm_host.h
+++ b/arch/x86/include/asm/kvm_host.h
@@ -1039,7 +1039,6 @@ struct kvm_x86_msr_filter {
#define APICV_INHIBIT_REASON_DISABLE 0
#define APICV_INHIBIT_REASON_HYPERV 1
-#define APICV_INHIBIT_REASON_NESTED 2
#define APICV_INHIBIT_REASON_IRQWIN 3
#define APICV_INHIBIT_REASON_PIT_REINJ 4
#define APICV_INHIBIT_REASON_X2APIC 5
@@ -1494,6 +1493,12 @@ struct kvm_x86_ops {
int (*complete_emulated_msr)(struct kvm_vcpu *vcpu, int err);
void (*vcpu_deliver_sipi_vector)(struct kvm_vcpu *vcpu, u8 vector);
+
+ /*
+ * Returns true if for some reason APICv (e.g guest mode)
+ * must be inhibited on this vCPU
+ */
+ bool (*vcpu_has_apicv_inhibit_condition)(struct kvm_vcpu *vcpu);
};
struct kvm_x86_nested_ops {
@@ -1784,6 +1789,7 @@ gpa_t kvm_mmu_gva_to_gpa_system(struct kvm_vcpu *vcpu, gva_t gva,
bool kvm_apicv_activated(struct kvm *kvm);
void kvm_vcpu_update_apicv(struct kvm_vcpu *vcpu);
+bool vcpu_has_apicv_inhibit_condition(struct kvm_vcpu *vcpu);
void kvm_request_apicv_update(struct kvm *kvm, bool activate,
unsigned long bit);
diff --git a/arch/x86/kvm/svm/avic.c b/arch/x86/kvm/svm/avic.c
index c6072245f7fbb..8f23e7d239097 100644
--- a/arch/x86/kvm/svm/avic.c
+++ b/arch/x86/kvm/svm/avic.c
@@ -677,6 +677,12 @@ bool avic_dy_apicv_has_pending_interrupt(struct kvm_vcpu *vcpu)
return false;
}
+bool avic_has_vcpu_inhibit_condition(struct kvm_vcpu *vcpu)
+{
+ return is_guest_mode(vcpu);
+}
+
+
static void svm_ir_list_del(struct vcpu_svm *svm, struct amd_iommu_pi_data *pi)
{
unsigned long flags;
@@ -888,7 +894,6 @@ bool avic_check_apicv_inhibit_reasons(ulong bit)
ulong supported = BIT(APICV_INHIBIT_REASON_DISABLE) |
BIT(APICV_INHIBIT_REASON_ABSENT) |
BIT(APICV_INHIBIT_REASON_HYPERV) |
- BIT(APICV_INHIBIT_REASON_NESTED) |
BIT(APICV_INHIBIT_REASON_IRQWIN) |
BIT(APICV_INHIBIT_REASON_PIT_REINJ) |
BIT(APICV_INHIBIT_REASON_X2APIC) |
diff --git a/arch/x86/kvm/svm/nested.c b/arch/x86/kvm/svm/nested.c
index 39d280e7e80ef..ac9159b0618c7 100644
--- a/arch/x86/kvm/svm/nested.c
+++ b/arch/x86/kvm/svm/nested.c
@@ -551,11 +551,6 @@ static void nested_vmcb02_prepare_control(struct vcpu_svm *svm)
* exit_int_info, exit_int_info_err, next_rip, insn_len, insn_bytes.
*/
- /*
- * Also covers avic_vapic_bar, avic_backing_page, avic_logical_id,
- * avic_physical_id.
- */
- WARN_ON(kvm_apicv_activated(svm->vcpu.kvm));
/* Copied from vmcb01. msrpm_base can be overwritten later. */
svm->vmcb->control.nested_ctl = svm->vmcb01.ptr->control.nested_ctl;
@@ -659,6 +654,9 @@ int enter_svm_guest_mode(struct kvm_vcpu *vcpu, u64 vmcb12_gpa,
svm_set_gif(svm, true);
+ if (kvm_vcpu_apicv_active(vcpu))
+ kvm_make_request(KVM_REQ_APICV_UPDATE, vcpu);
+
return 0;
}
@@ -923,6 +921,13 @@ int nested_svm_vmexit(struct vcpu_svm *svm)
if (unlikely(svm->vmcb->save.rflags & X86_EFLAGS_TF))
kvm_queue_exception(&(svm->vcpu), DB_VECTOR);
+ /*
+ * Un-inhibit the AVIC right away, so that other vCPUs can start
+ * to benefit from VM-exit less IPI right away
+ */
+ if (kvm_apicv_activated(vcpu->kvm))
+ kvm_vcpu_update_apicv(vcpu);
+
return 0;
}
diff --git a/arch/x86/kvm/svm/svm.c b/arch/x86/kvm/svm/svm.c
index 18d4d87e12e15..85035324ed762 100644
--- a/arch/x86/kvm/svm/svm.c
+++ b/arch/x86/kvm/svm/svm.c
@@ -1392,7 +1392,8 @@ static void svm_set_vintr(struct vcpu_svm *svm)
/*
* The following fields are ignored when AVIC is enabled
*/
- WARN_ON(kvm_apicv_activated(svm->vcpu.kvm));
+ if (!is_guest_mode(&svm->vcpu))
+ WARN_ON(kvm_apicv_activated(svm->vcpu.kvm));
svm_set_intercept(svm, INTERCEPT_VINTR);
@@ -2898,10 +2899,16 @@ static int interrupt_window_interception(struct kvm_vcpu *vcpu)
svm_clear_vintr(to_svm(vcpu));
/*
- * For AVIC, the only reason to end up here is ExtINTs.
+ * If not running nested, for AVIC, the only reason to end up here is ExtINTs.
* In this case AVIC was temporarily disabled for
* requesting the IRQ window and we have to re-enable it.
+ *
+ * If running nested, still uninhibit the AVIC in case irq window
+ * was requested when it was not running nested.
+ * All vCPUs which run nested will have their AVIC still
+ * inhibited due to AVIC inhibition override for that.
*/
+
kvm_request_apicv_update(vcpu->kvm, true, APICV_INHIBIT_REASON_IRQWIN);
++vcpu->stat.irq_window_exits;
@@ -3451,8 +3458,16 @@ static void svm_enable_irq_window(struct kvm_vcpu *vcpu)
* unless we have pending ExtINT since it cannot be injected
* via AVIC. In such case, we need to temporarily disable AVIC,
* and fallback to injecting IRQ via V_IRQ.
+ *
+ * If running nested, this vCPU will use separate page tables
+ * which don't have L1's AVIC mapped, and the AVIC is
+ * already inhibited thus there is no need for global
+ * AVIC inhibition.
*/
- kvm_request_apicv_update(vcpu->kvm, false, APICV_INHIBIT_REASON_IRQWIN);
+
+ if (!is_guest_mode(vcpu))
+ kvm_request_apicv_update(vcpu->kvm, false, APICV_INHIBIT_REASON_IRQWIN);
+
svm_set_vintr(svm);
}
}
@@ -3927,14 +3942,6 @@ static void svm_vcpu_after_set_cpuid(struct kvm_vcpu *vcpu)
if (guest_cpuid_has(vcpu, X86_FEATURE_X2APIC))
kvm_request_apicv_update(vcpu->kvm, false,
APICV_INHIBIT_REASON_X2APIC);
-
- /*
- * Currently, AVIC does not work with nested virtualization.
- * So, we disable AVIC when cpuid for SVM is set in the L1 guest.
- */
- if (nested && guest_cpuid_has(vcpu, X86_FEATURE_SVM))
- kvm_request_apicv_update(vcpu->kvm, false,
- APICV_INHIBIT_REASON_NESTED);
}
init_vmcb_after_set_cpuid(vcpu);
}
@@ -4657,6 +4664,7 @@ static struct kvm_x86_ops svm_x86_ops __initdata = {
.complete_emulated_msr = svm_complete_emulated_msr,
.vcpu_deliver_sipi_vector = svm_vcpu_deliver_sipi_vector,
+ .vcpu_has_apicv_inhibit_condition = avic_has_vcpu_inhibit_condition,
};
/*
@@ -4840,6 +4848,7 @@ static __init int svm_hardware_setup(void)
} else {
svm_x86_ops.vcpu_blocking = NULL;
svm_x86_ops.vcpu_unblocking = NULL;
+ svm_x86_ops.vcpu_has_apicv_inhibit_condition = NULL;
}
if (vls) {
diff --git a/arch/x86/kvm/svm/svm.h b/arch/x86/kvm/svm/svm.h
index 83f9f95eced3e..c02903641d13d 100644
--- a/arch/x86/kvm/svm/svm.h
+++ b/arch/x86/kvm/svm/svm.h
@@ -580,6 +580,7 @@ int avic_pi_update_irte(struct kvm *kvm, unsigned int host_irq,
void avic_vcpu_blocking(struct kvm_vcpu *vcpu);
void avic_vcpu_unblocking(struct kvm_vcpu *vcpu);
void avic_ring_doorbell(struct kvm_vcpu *vcpu);
+bool avic_has_vcpu_inhibit_condition(struct kvm_vcpu *vcpu);
/* sev.c */
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 8cb5390f75efe..63d84c373e465 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -9697,6 +9697,14 @@ void kvm_make_scan_ioapic_request(struct kvm *kvm)
kvm_make_all_cpus_request(kvm, KVM_REQ_SCAN_IOAPIC);
}
+bool vcpu_has_apicv_inhibit_condition(struct kvm_vcpu *vcpu)
+{
+ if (kvm_x86_ops.vcpu_has_apicv_inhibit_condition)
+ return static_call(kvm_x86_vcpu_has_apicv_inhibit_condition)(vcpu);
+ else
+ return false;
+}
+
void kvm_vcpu_update_apicv(struct kvm_vcpu *vcpu)
{
bool activate;
@@ -9706,7 +9714,9 @@ void kvm_vcpu_update_apicv(struct kvm_vcpu *vcpu)
down_read(&vcpu->kvm->arch.apicv_update_lock);
- activate = kvm_apicv_activated(vcpu->kvm);
+ activate = kvm_apicv_activated(vcpu->kvm) &&
+ !vcpu_has_apicv_inhibit_condition(vcpu);
+
if (vcpu->arch.apicv_active == activate)
goto out;
@@ -10110,7 +10120,11 @@ static int vcpu_enter_guest(struct kvm_vcpu *vcpu)
* per-VM state, and responsing vCPUs must wait for the update
* to complete before servicing KVM_REQ_APICV_UPDATE.
*/
- WARN_ON_ONCE(kvm_apicv_activated(vcpu->kvm) != kvm_vcpu_apicv_active(vcpu));
+ if (vcpu_has_apicv_inhibit_condition(vcpu))
+ WARN_ON(kvm_vcpu_apicv_active(vcpu));
+ else
+ WARN_ON_ONCE(kvm_apicv_activated(vcpu->kvm) != kvm_vcpu_apicv_active(vcpu));
+
exit_fastpath = static_call(kvm_x86_vcpu_run)(vcpu);
if (likely(exit_fastpath != EXIT_FASTPATH_REENTER_GUEST))
--
2.26.3
Powered by blists - more mailing lists