KVM: LAPIC: Per vCPU control over kvm_can_post_timer_interrupt
PATCH 1/1 KVM: LAPIC: Per vCPU control over kvm_can_post_timer_interrupt
From: Aili Yao <[email protected]>
When we isolate some pyhiscal cores, We may not use them for kvm guests, We may use them for other purposes like DPDK, or we can make some kvm guests isolated and some not, the global judgement pi_inject_timer is not enough; We may make wrong decisions:
In such a scenario, the guests without isolated cores will not be permitted to use vmx preemption timer, and tscdeadline fastpath also be disabled, both will lead to performance penalty.
So check whether the vcpu->cpu is isolated, if not, don't post timer interrupt.
--- arch/x86/kvm/lapic.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)
arch/x86/kvm/lapic.c +3 -1
@@ -34,6 +34,7 @@#include <asm/delay.h> #include <linux/atomic.h> #include <linux/jump_label.h> +#include <linux/sched/isolation.h> #include "kvm_cache_regs.h" #include "irq.h" #include "ioapic.h"
@@ -113,7 +114,8 @@ static inline u32 kvm_x2apic_id(struct kvm_lapic *apic)static bool kvm_can_post_timer_interrupt (struct kvm_vcpu *vcpu) { -return pi_inject_timer &&kvm_vcpu_apicv_active (vcpu); +return pi_inject_timer &&kvm_vcpu_apicv_active (vcpu) && + !housekeeping_cpu (vcpu->cpu,HK_FLAG_TIMER ); }bool kvm_can_use_hv_timer (struct kvm_vcpu *vcpu)
Signature · 1 line
2.25.1
Sean Christopherson
maintainerQuoted 32 lines · Aili Yao
From: Aili Yao <[email protected]>
When we isolate some pyhiscal cores, We may not use them for kvm guests, We may use them for other purposes like DPDK, or we can make some kvm guests isolated and some not, the global judgement
pi_inject_timeris not enough; We may make wrong decisions:In such a scenario, the guests without isolated cores will not be permitted to use vmx preemption timer, and tscdeadline fastpath also be disabled, both will lead to performance penalty.
So check whether the
vcpu->cpuis isolated, if not, don't post timer interrupt.Signed-off-by Aili Yao <[email protected]>--- arch/x86/kvm/lapic.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)arch/x86/kvm/lapic.c +1
-0@@ -34,6 +34,7 @@#include <asm/delay.h> #include <linux/atomic.h> #include <linux/jump_label.h> +#include <linux/sched/isolation.h> #include "kvm_cache_regs.h" #include "irq.h" #include "ioapic.h"
arch/x86/kvm/lapic.c static inline u32 kvm_x2apic_id(struct kvm_lapic *apic) · line 118
Quoted 6 lines
arch/x86/kvm/lapic.c +2
-1@@ -113,7 +114,8 @@ static inline u32 kvm_x2apic_id(struct kvm_lapic *apic)static bool kvm_can_post_timer_interrupt (struct kvm_vcpu *vcpu) { -return pi_inject_timer &&kvm_vcpu_apicv_active (vcpu); +return pi_inject_timer &&kvm_vcpu_apicv_active (vcpu) && + !housekeeping_cpu (vcpu->cpu,HK_FLAG_TIMER );
I don't think this is safe, vcpu->cpu will be -1 if the vCPU isn't scheduled in. This also doesn't play nice with the admin forcing pi_inject_timer=1. Not saying there's a reasonable use case for doing that, but it's supported today and this would break that behavior. It would also lead to weird behavior if a vCPU were migrated on/off a housekeeping vCPU. Again, probably not a reasonable use case, but I don't see anything that would outright prevent that behavior.
The existing behavior also feels a bit unsafe as pi_inject_timer is writable while KVM is running, though I supposed that's orthogonal to this discussion.
Rather than check vcpu->cpu, is there an existing vCPU flag that can be queried, e.g. KVM_HINTS_REALTIME?
Quoted 6 lines
arch/x86/kvm/lapic.c +0
-0}bool kvm_can_use_hv_timer (struct kvm_vcpu *vcpu)Signature · 1 line
2.25.1
Quoted 48 lines · Sean Christopherson
Quoted 32 lines · Aili Yao
From: Aili Yao <[email protected]>
When we isolate some pyhiscal cores, We may not use them for kvm guests, We may use them for other purposes like DPDK, or we can make some kvm guests isolated and some not, the global judgement
pi_inject_timeris not enough; We may make wrong decisions:In such a scenario, the guests without isolated cores will not be permitted to use vmx preemption timer, and tscdeadline fastpath also be disabled, both will lead to performance penalty.
So check whether the
vcpu->cpuis isolated, if not, don't post timer interrupt.Signed-off-by Aili Yao <[email protected]>--- arch/x86/kvm/lapic.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)arch/x86/kvm/lapic.c +1
-0@@ -34,6 +34,7 @@#include <asm/delay.h> #include <linux/atomic.h> #include <linux/jump_label.h> +#include <linux/sched/isolation.h> #include "kvm_cache_regs.h" #include "irq.h" #include "ioapic.h" arch/x86/kvm/lapic.c static inline u32 kvm_x2apic_id(struct kvm_lapic *apic) · line 118
Quoted 6 lines
arch/x86/kvm/lapic.c +2
-1@@ -113,7 +114,8 @@ static inline u32 kvm_x2apic_id(struct kvm_lapic *apic)static bool kvm_can_post_timer_interrupt (struct kvm_vcpu *vcpu) { -return pi_inject_timer &&kvm_vcpu_apicv_active (vcpu); +return pi_inject_timer &&kvm_vcpu_apicv_active (vcpu) && + !housekeeping_cpu (vcpu->cpu,HK_FLAG_TIMER );I don't think this is safe,
vcpu->cpuwill be -1 if the vCPU isn't scheduled in. This also doesn't play nice with the admin forcing pi_inject_timer=1. Not saying there's a reasonable use case for doing that, but it's supported today and this would break that behavior. It would also lead to weird behavior if a vCPU were migrated on/off a housekeeping vCPU. Again, probably not a reasonable use case, but I don't see anything that would outright prevent that behavior.
Quoted 5 lines
The existing behavior also feels a bit unsafe as
pi_inject_timeris writable while KVM is running, though I supposed that's orthogonal to this discussion.Rather than check
vcpu->cpu, is there an existing vCPU flag that can be queried, e.g.KVM_HINTS_REALTIME?
How about something like below:
From 67f605120e212384cb3d5788ba8c83f15659503b Mon Sep 17 00:00:00 2001
From: Wanpeng Li <[email protected]>
Date: Tue, 23 Nov 2021 10:36:10 +0800
Subject: [PATCH] KVM: LAPIC: To keep the vCPUs in non-root mode for timer-pi
From: Wanpeng Li <[email protected]>
As commit 0c5f81dad46 (KVM: LAPIC: Inject timer interrupt via posted interrupt) mentioned that the host admin should well tune the guest setup, so that vCPUs are placed on isolated pCPUs, and with several pCPUs surplus for *busy* housekeeping.
It is better to disable mwait/hlt/pause vmexits to keep the vCPUs in non-root mode. However, we may isolate pCPUs for other purpose like DPDK or we can make some guests isolated and others not, Let's add the checking kvm_mwait_in_guest() to kvm_can_post_timer_interrupt() since we can't benefit from timer posted-interrupt
w/o keeping the vCPUs in non-root mode.
--- arch/x86/kvm/lapic.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-)
arch/x86/kvm/lapic.c +1 -1
@@ -113,14 +113,13 @@ static inline u32 kvm_x2apic_id(struct kvm_lapic *apic)static bool kvm_can_post_timer_interrupt (struct kvm_vcpu *vcpu) { -return pi_inject_timer &&kvm_vcpu_apicv_active (vcpu); + return pi_inject_timer && kvm_mwait_in_guest(vcpu->kvm) &&
kvm_vcpu_apicv_active(vcpu);
}
bool kvm_can_use_hv_timer(struct kvm_vcpu *vcpu)
{
return kvm_x86_ops.set_hv_timer
- && !(kvm_mwait_in_guest(vcpu->kvm) ||
- kvm_can_post_timer_interrupt(vcpu));
+ && !kvm_mwait_in_guest(vcpu->kvm);
}
EXPORT_SYMBOL_GPL(kvm_can_use_hv_timer);
Quoted 38 lines
Quoted 36 lines · Sean Christopherson
Quoted 34 lines · Aili Yao
From: Aili Yao <[email protected]>
When we isolate some pyhiscal cores, We may not use them for kvm guests, We may use them for other purposes like DPDK, or we can make some kvm guests isolated and some not, the global judgement
pi_inject_timeris not enough; We may make wrong decisions:In such a scenario, the guests without isolated cores will not be permitted to use vmx preemption timer, and tscdeadline fastpath also be disabled, both will lead to performance penalty.
So check whether the
vcpu->cpuis isolated, if not, don't post timer interrupt.Signed-off-by Aili Yao <[email protected]>--- arch/x86/kvm/lapic.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)arch/x86/kvm/lapic.c index renamed from arch/x86/kvm/lapic.c +0
-0
759952dd1222..72dde5532101100644arch/x86/kvm/lapic.c +1
-0@@ -34,6 +34,7 @@#include <asm/delay.h> #include <linux/atomic.h> #include <linux/jump_label.h> +#include <linux/sched/isolation.h> #include "kvm_cache_regs.h" #include "irq.h" #include "ioapic.h" @@ -113,7 +114,8 @@ static inline u32 kvm_x2apic_id(struct kvm_lapic *apic)
Quoted 6 lines
Quoted 6 lines
arch/x86/kvm/lapic.c static inline u32 kvm_x2apic_id(struct kvm_lapic · line 118
Quoted 4 lines
arch/x86/kvm/lapic.c +2
-1static bool kvm_can_post_timer_interrupt (struct kvm_vcpu *vcpu) { -return pi_inject_timer &&kvm_vcpu_apicv_active (vcpu); +return pi_inject_timer &&kvm_vcpu_apicv_active (vcpu) && + !housekeeping_cpu (vcpu->cpu,HK_FLAG_TIMER );I don't think this is safe,
vcpu->cpuwill be -1 if the vCPU isn't scheduled in.
Yes, vcpu->cpu is -1 before vcpu create, but in my environments, it didn't trigger this issue. I need to dig more, Thanks!
Maybe I need one valid check here.
Quoted 6 lines
Quoted 5 lines
This also doesn't play nice with the admin forcing pi_inject_timer=1. Not saying there's a reasonable use case for doing that, but it's supported today and this would break that behavior. It would also lead to weird behavior if a vCPU were migrated on/off a housekeeping vCPU. Again, probably not a reasonable use case, but I don't see anything
that would outright prevent that behavior.
Yes, this is not one common operation, But I did do test some scenarios:
1. isolated cpu --> housekeeping cpu;
isolated guest timer is in housekeeping CPU, for migration, kvm_can_post_timer_interrupt
will return false, so the timer may be migrated to vcpu->cpu;
This seems works in my test;2. isolated --> isolated
Isolated guest timer is in housekeeping cpu, for migration,kvm_can_post_timer_interrupt return true, timer is not migrated
3. housekeeping CPU --> isolated CPU
non-isolated CPU timer is usually in vcpu->cpu, for migration to isolated, kvm_can_post_timer_interrupt
will be true, the timer remain on the same CPU;
This seems works in my test;
4. housekeeping CPU --> housekeeping CPU
timer migrated;It seems this is not an affecting problem;
Quoted 47 lines
Quoted 3 lines
The existing behavior also feels a bit unsafe as
pi_inject_timeris writable while KVM is running, though I supposed that's orthogonal to thisdiscussion.
Quoted 3 lines
Rather than check
vcpu->cpu, is there an existing vCPU flag that can be queried, e.g.KVM_HINTS_REALTIME?How about something like below:
From
67f605120e212384cb3d5788ba8c83f15659503bMon Sep 17 00:00:00 2001
From: Wanpeng Li <[email protected]>
Date: Tue, 23 Nov 2021 10:36:10 +0800
Subject: [PATCH] KVM: LAPIC: To keep the vCPUs in non-root mode for timer- piFrom: Wanpeng Li <[email protected]>
As commit
0c5f81dad46(KVM: LAPIC: Inject timer interrupt via posted interrupt) mentioned that the host admin should well tune the guest setup, so that vCPUs are placed on isolated pCPUs, and with several pCPUs surplus for
*busy* housekeeping.
It is better to disable mwait/hlt/pause vmexits to keep the vCPUs in non-root mode. However, we may isolate pCPUs for other purpose like DPDK or we can make some guests isolated and others not, Let's add the checkingkvm_mwait_in_guest()tokvm_can_post_timer_interrupt()since we can't benefit from timer posted-interrupt w/o keeping the vCPUs in non-root mode.Reported-by Aili Yao <[email protected]>Signed-off-by Wanpeng Li <[email protected]>--- arch/x86/kvm/lapic.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-)arch/x86/kvm/lapic.c index renamed from arch/x86/kvm/lapic.c +0
-0
759952dd1222..8257566d44c7100644arch/x86/kvm/lapic.c +1
-1@@ -113,14 +113,13 @@ static inline u32 kvm_x2apic_id(struct kvm_lapic *apic)static bool kvm_can_post_timer_interrupt (struct kvm_vcpu *vcpu) { -return pi_inject_timer &&kvm_vcpu_apicv_active (vcpu); + return pi_inject_timer && kvm_mwait_in_guest(vcpu->kvm) &&
arch/x86/kvm/lapic.c static inline u32 kvm_x2apic_id(struct kvm_lapic · line 124
Quoted 10 lines
arch/x86/kvm/lapic.c +1
-2kvm_vcpu_apicv_active (vcpu); }bool kvm_can_use_hv_timer (struct kvm_vcpu *vcpu) {return kvm_x86_ops.set_hv_timer - && !(kvm_mwait_in_guest (vcpu->kvm) || -kvm_can_post_timer_interrupt (vcpu)); + && !kvm_mwait_in_guest (vcpu->kvm); }EXPORT_SYMBOL_GPL (kvm_can_use_hv_timer);
This method seems more quick and safe, but I have one question: Does this kvm_mwait_in_guest
can guarantee the CPU isolated, in some production environments and usually, MWAIT feature is disabled in host and even guests with isolated CPUs. And also we can set guests kvm_mwait_in_guest true with CPUs just pinned, not isolated.
Thanks
Quoted 123 lines · yaoaili [么爱利]
Quoted 39 lines
Quoted 36 lines · Sean Christopherson
Quoted 34 lines · Aili Yao
From: Aili Yao <[email protected]>
When we isolate some pyhiscal cores, We may not use them for kvm guests, We may use them for other purposes like DPDK, or we can make some kvm guests isolated and some not, the global judgement
pi_inject_timeris not enough; We may make wrong decisions:In such a scenario, the guests without isolated cores will not be permitted to use vmx preemption timer, and tscdeadline fastpath also be disabled, both will lead to performance penalty.
So check whether the
vcpu->cpuis isolated, if not, don't post timer interrupt.Signed-off-by Aili Yao <[email protected]>--- arch/x86/kvm/lapic.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)arch/x86/kvm/lapic.c index renamed from arch/x86/kvm/lapic.c +0
-0
759952dd1222..72dde5532101100644arch/x86/kvm/lapic.c +1
-0@@ -34,6 +34,7 @@#include <asm/delay.h> #include <linux/atomic.h> #include <linux/jump_label.h> +#include <linux/sched/isolation.h> #include "kvm_cache_regs.h" #include "irq.h" #include "ioapic.h" @@ -113,7 +114,8 @@ static inline u32 kvm_x2apic_id(struct kvm_lapic *apic)Quoted 6 lines
Quoted 6 lines
arch/x86/kvm/lapic.c static inline u32 kvm_x2apic_id(struct kvm_lapic · line 118
Quoted 4 lines
arch/x86/kvm/lapic.c +2
-1static bool kvm_can_post_timer_interrupt (struct kvm_vcpu *vcpu) { -return pi_inject_timer &&kvm_vcpu_apicv_active (vcpu); +return pi_inject_timer &&kvm_vcpu_apicv_active (vcpu) && + !housekeeping_cpu (vcpu->cpu,HK_FLAG_TIMER );I don't think this is safe,
vcpu->cpuwill be -1 if the vCPU isn't scheduled in.Yes,
vcpu->cpuis -1 before vcpu create, but in my environments, it didn't trigger this issue. I need to dig more, Thanks!
Maybe I need one valid check here.Quoted 6 lines
Quoted 5 lines
This also doesn't play nice with the admin forcing pi_inject_timer=1. Not saying there's a reasonable use case for doing that, but it's supported today and this would break that behavior. It would also lead to weird behavior if a vCPU were migrated on/off a housekeeping vCPU. Again, probably not a reasonable use case, but I don't see anything
that would outright prevent that behavior.
Yes, this is not one common operation, But I did do test some scenarios:
1. isolated cpu --> housekeeping cpu; isolated guest timer is in housekeeping CPU, for migration, kvm_can_post_timer_interrupt will return false, so the timer may be migrated to vcpu->cpu; This seems works in my test;2. isolated --> isolated
Isolated guest timer is in housekeeping cpu, for migration,kvm_can_post_timer_interrupt return true, timer is not migrated
3. housekeeping CPU --> isolated CPUnon-isolated CPU timer is usually in vcpu->cpu, for migration to isolated, kvm_can_post_timer_interrupt will be true, the timer remain on the same CPU; This seems works in my test; 4. housekeeping CPU --> housekeeping CPU timer migrated;It seems this is not an affecting problem;
Quoted 50 lines
Quoted 3 lines
The existing behavior also feels a bit unsafe as
pi_inject_timeris writable while KVM is running, though I supposed that's orthogonal to thisdiscussion.
Quoted 3 lines
Rather than check
vcpu->cpu, is there an existing vCPU flag that can be queried, e.g.KVM_HINTS_REALTIME?How about something like below:
From
67f605120e212384cb3d5788ba8c83f15659503bMon Sep 17 00:00:00 2001
From: Wanpeng Li <[email protected]>
Date: Tue, 23 Nov 2021 10:36:10 +0800
Subject: [PATCH] KVM: LAPIC: To keep the vCPUs in non-root mode for timer- piFrom: Wanpeng Li <[email protected]>
As commit
0c5f81dad46(KVM: LAPIC: Inject timer interrupt via posted interrupt) mentioned that the host admin should well tune the guest setup, so that vCPUs are placed on isolated pCPUs, and with several pCPUs surplus for
*busy* housekeeping.
It is better to disable mwait/hlt/pause vmexits to keep the vCPUs in non-root mode. However, we may isolate pCPUs for other purpose like DPDK or we can make some guests isolated and others not, Let's add the checkingkvm_mwait_in_guest()tokvm_can_post_timer_interrupt()since we can't benefit from timer posted-interrupt w/o keeping the vCPUs in non-root mode.Reported-by Aili Yao <[email protected]>Signed-off-by Wanpeng Li <[email protected]>--- arch/x86/kvm/lapic.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-)arch/x86/kvm/lapic.c index renamed from arch/x86/kvm/lapic.c +0
-0
759952dd1222..8257566d44c7100644arch/x86/kvm/lapic.c +1
-1@@ -113,14 +113,13 @@ static inline u32 kvm_x2apic_id(struct kvm_lapic *apic)static bool kvm_can_post_timer_interrupt (struct kvm_vcpu *vcpu) { -return pi_inject_timer &&kvm_vcpu_apicv_active (vcpu); +return pi_inject_timer &&kvm_mwait_in_guest (vcpu->kvm) &&kvm_vcpu_apicv_active (vcpu); }
Quoted 11 lines
arch/x86/kvm/lapic.c static inline u32 kvm_x2apic_id(struct kvm_lapic · line 124
Quoted 7 lines
arch/x86/kvm/lapic.c +1
-2bool kvm_can_use_hv_timer (struct kvm_vcpu *vcpu) {return kvm_x86_ops.set_hv_timer - && !(kvm_mwait_in_guest (vcpu->kvm) || -kvm_can_post_timer_interrupt (vcpu)); + && !kvm_mwait_in_guest (vcpu->kvm); }EXPORT_SYMBOL_GPL (kvm_can_use_hv_timer);This method seems more quick and safe, but I have one question: Does this
kvm_mwait_in_guest
can guarantee the CPU isolated, in some production environments and usually, MWAIT feature is disabled in host and even guests with isolated CPUs. And also we can set guestskvm_mwait_in_guesttrue with CPUs just pinned, not isolated.
You won't benefit from timer posted-interrupt if mwait is not exposed to the guest since you can't keep CPU in non-root mode. kvm_mwait_in_guest() will not guarantee the CPU is isolated, but what's still bothering?
Wanpeng
Quoted 147 lines · On Tue, 23 Nov 2021 14:24:19 +0800 Wanpeng Li <[email protected]> wrote:
Quoted 130 lines · yaoaili [么爱利]
Quoted 41 lines
Quoted 38 lines · Sean Christopherson
Quoted 36 lines · Aili Yao
From: Aili Yao <[email protected]>
When we isolate some pyhiscal cores, We may not use them for kvm guests, We may use them for other purposes like DPDK, or we can make some kvm guests isolated and some not, the global judgement
pi_inject_timeris not enough; We may make wrong decisions:In such a scenario, the guests without isolated cores will not be permitted to use vmx preemption timer, and tscdeadline fastpath also be disabled, both will lead to performance penalty.
So check whether the
vcpu->cpuis isolated, if not, don't post timer interrupt.Signed-off-by Aili Yao <[email protected]>--- arch/x86/kvm/lapic.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)arch/x86/kvm/lapic.c index renamed from arch/x86/kvm/lapic.c +0
-0
759952dd1222..72dde5532101100644arch/x86/kvm/lapic.c +1
-0@@ -34,6 +34,7 @@#include <asm/delay.h> #include <linux/atomic.h> #include <linux/jump_label.h> +#include <linux/sched/isolation.h> #include "kvm_cache_regs.h" #include "irq.h" #include "ioapic.h" @@ -113,7 +114,8 @@ static inline u32 kvm_x2apic_id(struct kvm_lapic *apic)Quoted 8 lines
Quoted 8 lines
arch/x86/kvm/lapic.c static inline u32 kvm_x2apic_id(struct · line 119
Quoted 5 lines
arch/x86/kvm/lapic.c +2
-1static bool kvm_can_post_timer_interrupt (struct kvm_vcpu *vcpu) { -return pi_inject_timer &&kvm_vcpu_apicv_active (vcpu); +return pi_inject_timer &&kvm_vcpu_apicv_active (vcpu) && + !housekeeping_cpu (vcpu->cpu,HK_FLAG_TIMER );I don't think this is safe,
vcpu->cpuwill be -1 if the vCPU isn't scheduled in.Yes,
vcpu->cpuis -1 before vcpu create, but in my environments, it didn't trigger this issue. I need to dig more, Thanks! Maybe I need one valid check here.Quoted 7 lines
Quoted 6 lines
This also doesn't play nice with the admin forcing pi_inject_timer=1. Not saying there's a reasonable use case for doing that, but it's supported today and this would break that behavior. It would also lead to weird behavior if a vCPU were migrated on/off a housekeeping vCPU. Again, probably not a reasonable use case, but I don't see anything
that would outright prevent that behavior.
Yes, this is not one common operation, But I did do test some
scenarios: 1. isolated cpu --> housekeeping cpu;
isolated guest timer is in housekeeping CPU, for migration,kvm_can_post_timer_interruptwill return false, so the timer may be migrated tovcpu->cpu; This seems works in my test;
2. isolated --> isolated
Isolated guest timer is in housekeeping cpu, for
migration,kvm_can_post_timer_interrupt return true, timer is not migrated 3. housekeeping CPU --> isolated CPUnon-isolated CPU timer is usually in vcpu->cpu, for migration to isolated, kvm_can_post_timer_interrupt will be true, the timer remain on the same CPU; This seems works in my test; 4. housekeeping CPU --> housekeeping CPU timer migrated;It seems this is not an affecting problem;
Quoted 52 lines
Quoted 4 lines
The existing behavior also feels a bit unsafe as
pi_inject_timeris writable while KVM is running, though I supposed that's orthogonal to thisdiscussion.
Quoted 3 lines
Rather than check
vcpu->cpu, is there an existing vCPU flag that can be queried, e.g.KVM_HINTS_REALTIME?How about something like below:
From
67f605120e212384cb3d5788ba8c83f15659503bMon Sep 17 00:00:00 2001
From: Wanpeng Li <[email protected]>
Date: Tue, 23 Nov 2021 10:36:10 +0800
Subject: [PATCH] KVM: LAPIC: To keep the vCPUs in non-root mode for timer- piFrom: Wanpeng Li <[email protected]>
As commit
0c5f81dad46(KVM: LAPIC: Inject timer interrupt via posted interrupt) mentioned that the host admin should well tune the guest setup, so that vCPUs are placed on isolated pCPUs, and with several pCPUs surplus for
*busy* housekeeping.
It is better to disable mwait/hlt/pause vmexits to keep the vCPUs in non-root mode. However, we may isolate pCPUs for other purpose like DPDK or we can make some guests isolated and others not, Let's add the checkingkvm_mwait_in_guest()tokvm_can_post_timer_interrupt()since we can't benefit from timer posted-interrupt w/o keeping the vCPUs in non-root mode.Reported-by Aili Yao <[email protected]>Signed-off-by Wanpeng Li <[email protected]>--- arch/x86/kvm/lapic.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-)arch/x86/kvm/lapic.c index renamed from arch/x86/kvm/lapic.c +0
-0
759952dd1222..8257566d44c7100644arch/x86/kvm/lapic.c +1
-1@@ -113,14 +113,13 @@ static inline u32 kvm_x2apic_id(struct kvm_lapic *apic)static bool kvm_can_post_timer_interrupt (struct kvm_vcpu *vcpu) { -return pi_inject_timer &&kvm_vcpu_apicv_active (vcpu); +return pi_inject_timer &&kvm_mwait_in_guest (vcpu->kvm) &&kvm_vcpu_apicv_active (vcpu); }Quoted 14 lines
arch/x86/kvm/lapic.c static inline u32 kvm_x2apic_id(struct · line 125
Quoted 7 lines
arch/x86/kvm/lapic.c +1
-2bool kvm_can_use_hv_timer (struct kvm_vcpu *vcpu) {return kvm_x86_ops.set_hv_timer - && !(kvm_mwait_in_guest (vcpu->kvm) || -kvm_can_post_timer_interrupt (vcpu)); + && !kvm_mwait_in_guest (vcpu->kvm); }EXPORT_SYMBOL_GPL (kvm_can_use_hv_timer);This method seems more quick and safe, but I have one question:
Does thiskvm_mwait_in_guestcan guarantee the CPU isolated, in some production environments and usually, MWAIT feature is disabled in host and even guests with isolated CPUs. And also we can set guestskvm_mwait_in_guesttrue with CPUs just pinned, not isolated.
Quoted 4 lines
You won't benefit from timer posted-interrupt if mwait is not exposed to the guest since you can't keep CPU in non-root mode.
kvm_mwait_in_guest()will not guarantee the CPU is isolated, but what's still bothering?
Sorry, Did I miss some thing?
What in my mind: MWait may be disabled in bios, so host will use halt instruction as one replacement for idle operation, in such a configuration, Mwait in guest will also be disabled even if you try to set kvm_mwait_in_guest true; As a result, halt,pause may not exit the guest, so the post interrupt still counts?
For current code, We can migrate guest between isolated and housekeeping or we can change the cpu pinning on the fly, we allow this even the operation is not usually used, right?
Thanks!
Aili Yao
Quoted 156 lines · Aili Yao
Quoted 147 lines · On Tue, 23 Nov 2021 14:24:19 +0800 Wanpeng Li <[email protected]> wrote:
Quoted 130 lines · yaoaili [么爱利]
Quoted 41 lines
Quoted 38 lines · Sean Christopherson
Quoted 36 lines · Aili Yao
From: Aili Yao <[email protected]>
When we isolate some pyhiscal cores, We may not use them for kvm guests, We may use them for other purposes like DPDK, or we can make some kvm guests isolated and some not, the global judgement
pi_inject_timeris not enough; We may make wrong decisions:In such a scenario, the guests without isolated cores will not be permitted to use vmx preemption timer, and tscdeadline fastpath also be disabled, both will lead to performance penalty.
So check whether the
vcpu->cpuis isolated, if not, don't post timer interrupt.Signed-off-by Aili Yao <[email protected]>--- arch/x86/kvm/lapic.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)arch/x86/kvm/lapic.c index renamed from arch/x86/kvm/lapic.c +0
-0
759952dd1222..72dde5532101100644arch/x86/kvm/lapic.c +1
-0@@ -34,6 +34,7 @@#include <asm/delay.h> #include <linux/atomic.h> #include <linux/jump_label.h> +#include <linux/sched/isolation.h> #include "kvm_cache_regs.h" #include "irq.h" #include "ioapic.h" @@ -113,7 +114,8 @@ static inline u32 kvm_x2apic_id(struct kvm_lapic *apic)Quoted 8 lines
Quoted 8 lines
arch/x86/kvm/lapic.c static inline u32 kvm_x2apic_id(struct · line 119
Quoted 5 lines
arch/x86/kvm/lapic.c +2
-1static bool kvm_can_post_timer_interrupt (struct kvm_vcpu *vcpu) { -return pi_inject_timer &&kvm_vcpu_apicv_active (vcpu); +return pi_inject_timer &&kvm_vcpu_apicv_active (vcpu) && + !housekeeping_cpu (vcpu->cpu,HK_FLAG_TIMER );I don't think this is safe,
vcpu->cpuwill be -1 if the vCPU isn't scheduled in.Yes,
vcpu->cpuis -1 before vcpu create, but in my environments, it didn't trigger this issue. I need to dig more, Thanks! Maybe I need one valid check here.Quoted 7 lines
Quoted 6 lines
This also doesn't play nice with the admin forcing pi_inject_timer=1. Not saying there's a reasonable use case for doing that, but it's supported today and this would break that behavior. It would also lead to weird behavior if a vCPU were migrated on/off a housekeeping vCPU. Again, probably not a reasonable use case, but I don't see anything
that would outright prevent that behavior.
Yes, this is not one common operation, But I did do test some
scenarios: 1. isolated cpu --> housekeeping cpu;
isolated guest timer is in housekeeping CPU, for migration,kvm_can_post_timer_interruptwill return false, so the timer may be migrated tovcpu->cpu; This seems works in my test;
2. isolated --> isolated
Isolated guest timer is in housekeeping cpu, for
migration,kvm_can_post_timer_interrupt return true, timer is not migrated 3. housekeeping CPU --> isolated CPUnon-isolated CPU timer is usually in vcpu->cpu, for migration to isolated, kvm_can_post_timer_interrupt will be true, the timer remain on the same CPU; This seems works in my test; 4. housekeeping CPU --> housekeeping CPU timer migrated;It seems this is not an affecting problem;
Quoted 52 lines
Quoted 4 lines
The existing behavior also feels a bit unsafe as
pi_inject_timeris writable while KVM is running, though I supposed that's orthogonal to thisdiscussion.
Quoted 3 lines
Rather than check
vcpu->cpu, is there an existing vCPU flag that can be queried, e.g.KVM_HINTS_REALTIME?How about something like below:
From
67f605120e212384cb3d5788ba8c83f15659503bMon Sep 17 00:00:00 2001
From: Wanpeng Li <[email protected]>
Date: Tue, 23 Nov 2021 10:36:10 +0800
Subject: [PATCH] KVM: LAPIC: To keep the vCPUs in non-root mode for timer- piFrom: Wanpeng Li <[email protected]>
As commit
0c5f81dad46(KVM: LAPIC: Inject timer interrupt via posted interrupt) mentioned that the host admin should well tune the guest setup, so that vCPUs are placed on isolated pCPUs, and with several pCPUs surplus for
*busy* housekeeping.
It is better to disable mwait/hlt/pause vmexits to keep the vCPUs in non-root mode. However, we may isolate pCPUs for other purpose like DPDK or we can make some guests isolated and others not, Let's add the checkingkvm_mwait_in_guest()tokvm_can_post_timer_interrupt()since we can't benefit from timer posted-interrupt w/o keeping the vCPUs in non-root mode.Reported-by Aili Yao <[email protected]>Signed-off-by Wanpeng Li <[email protected]>--- arch/x86/kvm/lapic.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-)arch/x86/kvm/lapic.c index renamed from arch/x86/kvm/lapic.c +0
-0
759952dd1222..8257566d44c7100644arch/x86/kvm/lapic.c +1
-1@@ -113,14 +113,13 @@ static inline u32 kvm_x2apic_id(struct kvm_lapic *apic)static bool kvm_can_post_timer_interrupt (struct kvm_vcpu *vcpu) { -return pi_inject_timer &&kvm_vcpu_apicv_active (vcpu); +return pi_inject_timer &&kvm_mwait_in_guest (vcpu->kvm) &&kvm_vcpu_apicv_active (vcpu); }Quoted 14 lines
arch/x86/kvm/lapic.c static inline u32 kvm_x2apic_id(struct · line 125
Quoted 7 lines
arch/x86/kvm/lapic.c +1
-2bool kvm_can_use_hv_timer (struct kvm_vcpu *vcpu) {return kvm_x86_ops.set_hv_timer - && !(kvm_mwait_in_guest (vcpu->kvm) || -kvm_can_post_timer_interrupt (vcpu)); + && !kvm_mwait_in_guest (vcpu->kvm); }EXPORT_SYMBOL_GPL (kvm_can_use_hv_timer);This method seems more quick and safe, but I have one question:
Does thiskvm_mwait_in_guestcan guarantee the CPU isolated, in some production environments and usually, MWAIT feature is disabled in host and even guests with isolated CPUs. And also we can set guestskvm_mwait_in_guesttrue with CPUs just pinned, not isolated.Quoted 4 lines
You won't benefit from timer posted-interrupt if mwait is not exposed to the guest since you can't keep CPU in non-root mode.
kvm_mwait_in_guest()will not guarantee the CPU is isolated, but what's still bothering?
Quoted 7 lines
Sorry, Did I miss some thing?
What in my mind: MWait may be disabled in bios, so host will use halt instruction as one replacement for idle operation, in such a configuration, Mwait in guest will also be disabled even if you try to set
kvm_mwait_in_guesttrue; As a result, halt,pause may not exit the guest, so the post interrupt still counts?
I prefer to expose mwait/hlt/pause to the guest simultaneously, you don't need the ultra schedule latency/performance if you aren't exposing mwait. Then why do you care about latency from the timer?
Quoted 4 lines
For current code, We can migrate guest between isolated and housekeeping or we can change the cpu pinning on the fly, we allow this even the operation is not usually used, right?
My patch will not prevent using vmx preemption timer or tscdeadline fastpath.
Wanpeng
Quoted 34 lines · On Mon, 22 Nov 2021 19:13:02 +0000 Sean Christopherson <[email protected]>…
Quoted 33 lines · Aili Yao
From: Aili Yao <[email protected]>
When we isolate some pyhiscal cores, We may not use them for kvm guests, We may use them for other purposes like DPDK, or we can make some kvm guests isolated and some not, the global judgement
pi_inject_timeris not enough; We may make wrong decisions:In such a scenario, the guests without isolated cores will not be permitted to use vmx preemption timer, and tscdeadline fastpath also be disabled, both will lead to performance penalty.
So check whether the
vcpu->cpuis isolated, if not, don't post timer interrupt.Signed-off-by Aili Yao <[email protected]>--- arch/x86/kvm/lapic.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)arch/x86/kvm/lapic.c +1
-0@@ -34,6 +34,7 @@#include <asm/delay.h> #include <linux/atomic.h> #include <linux/jump_label.h> +#include <linux/sched/isolation.h> #include "kvm_cache_regs.h" #include "irq.h" #include "ioapic.h" @@ -113,7 +114,8 @@ static inline u32 kvm_x2apic_id(struct kvm_lapic *apic)
Quoted 8 lines
arch/x86/kvm/lapic.c static inline u32 kvm_x2apic_id(struct kvm_lapic *apic) · line 118
Quoted 5 lines
arch/x86/kvm/lapic.c +2
-1static bool kvm_can_post_timer_interrupt (struct kvm_vcpu *vcpu) { -return pi_inject_timer &&kvm_vcpu_apicv_active (vcpu); +return pi_inject_timer &&kvm_vcpu_apicv_active (vcpu) && + !housekeeping_cpu (vcpu->cpu,HK_FLAG_TIMER );I don't think this is safe,
vcpu->cpuwill be -1 if the vCPU isn't scheduled in.
I checked this, It seems we will set vcpu->cpu to a valid value when we create vcpu( kvm_vm_ioctl_create_vcpu()), only after that we can configure lapic through vcpu fd and start the timer, this may not be one real problem.
Currently, the patch seems work as expected in my test, maybe one possible candidate for the issue listed above.
Thanks
Quoted 21 lines
This also doesn't play nice with the admin forcing pi_inject_timer=1. Not saying there's a reasonable use case for doing that, but it's supported today and this would break that behavior. It would also lead to weird behavior if a vCPU were migrated on/off a housekeeping vCPU. Again, probably not a reasonable use case, but I don't see anything that would outright prevent that behavior.
The existing behavior also feels a bit unsafe as
pi_inject_timeris writable while KVM is running, though I supposed that's orthogonal to this discussion.Rather than check
vcpu->cpu, is there an existing vCPU flag that can be queried, e.g.KVM_HINTS_REALTIME?Quoted 6 lines
arch/x86/kvm/lapic.c +0
-0}bool kvm_can_use_hv_timer (struct kvm_vcpu *vcpu)Signature · 1 line
2.25.1
Quoted 46 lines · On Tue, 23 Nov 2021 16:18:34 +0800 Aili Yao <[email protected]> wrote:
Quoted 34 lines · On Mon, 22 Nov 2021 19:13:02 +0000 Sean Christopherson <[email protected]>…
Quoted 33 lines · Aili Yao
From: Aili Yao <[email protected]>
When we isolate some pyhiscal cores, We may not use them for kvm guests, We may use them for other purposes like DPDK, or we can make some kvm guests isolated and some not, the global judgement
pi_inject_timeris not enough; We may make wrong decisions:In such a scenario, the guests without isolated cores will not be permitted to use vmx preemption timer, and tscdeadline fastpath also be disabled, both will lead to performance penalty.
So check whether the
vcpu->cpuis isolated, if not, don't post timer interrupt.Signed-off-by Aili Yao <[email protected]>--- arch/x86/kvm/lapic.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)arch/x86/kvm/lapic.c +1
-0@@ -34,6 +34,7 @@#include <asm/delay.h> #include <linux/atomic.h> #include <linux/jump_label.h> +#include <linux/sched/isolation.h> #include "kvm_cache_regs.h" #include "irq.h" #include "ioapic.h" @@ -113,7 +114,8 @@ static inline u32 kvm_x2apic_id(struct kvm_lapic *apic)Quoted 8 lines
arch/x86/kvm/lapic.c static inline u32 kvm_x2apic_id(struct kvm_lapic *apic) · line 118
Quoted 5 lines
arch/x86/kvm/lapic.c +2
-1static bool kvm_can_post_timer_interrupt (struct kvm_vcpu *vcpu) { -return pi_inject_timer &&kvm_vcpu_apicv_active (vcpu); +return pi_inject_timer &&kvm_vcpu_apicv_active (vcpu) && + !housekeeping_cpu (vcpu->cpu,HK_FLAG_TIMER );I don't think this is safe,
vcpu->cpuwill be -1 if the vCPU isn't scheduled in.
Quoted 2 lines
I checked this, It seems we will set
vcpu->cputo a valid value when we createvcpu( kvm_vm_ioctl_create_vcpu()),
Really Sorry, My code base is too old; This vcpu->cpu assignment has been deleted in latest code, And this housekeeping_cpu() check will result problem.
Thanks!
Quoted 31 lines
only after that we can
configure lapic through vcpu fd and start the timer, this may not be one real problem.Currently, the patch seems work as expected in my test, maybe one possible candidate for the issue listed above.
Thanks
Quoted 21 lines
This also doesn't play nice with the admin forcing pi_inject_timer=1. Not saying there's a reasonable use case for doing that, but it's supported today and this would break that behavior. It would also lead to weird behavior if a vCPU were migrated on/off a housekeeping vCPU. Again, probably not a reasonable use case, but I don't see anything that would outright prevent that behavior.
The existing behavior also feels a bit unsafe as
pi_inject_timeris writable while KVM is running, though I supposed that's orthogonal to this discussion.Rather than check
vcpu->cpu, is there an existing vCPU flag that can be queried, e.g.KVM_HINTS_REALTIME?Quoted 6 lines
arch/x86/kvm/lapic.c +0
-0}bool kvm_can_use_hv_timer (struct kvm_vcpu *vcpu)Signature · 1 line
2.25.1