s390x: Improvements to SIGP handling [KVM]
v5 cover s390x: Improvements to SIGP handling [KVM]
Here is an update to the handling of SIGP between kernel and userspace.
As before, I'm looking at problems encountered when a SIGP order that is processed in the kernel (for example, SIGP SENSE) is run concurrently with another one is processed in userspace (for example, SIGP STOP). Being able to provide an honest answer in the SIGP SENSE as to whether the targeted VCPU is/not stopped is important to provide a consistent answer while a guest OS is bringing its configuration online.
This version was suggested by David Hildenbrand on v3 [1], where we examine the target vcpu for a pending STOP/RESTART IRQ while processing a SIGP order in-kernel, and returning the CC2 if one is in-flight.
Unlike v2-v4 of this RFC, this solution requires no changes to userspace to exploit a new interface, but a small change is made on the QEMU side to keep the sequence of events in checks.
Thoughts?
[1] https://lore.kernel.org/r/[email protected]/
Previous RFCs:
v1: https://lore.kernel.org/r/[email protected]/
v2: https://lore.kernel.org/r/[email protected]/
v3: https://lore.kernel.org/r/[email protected]/
v4: https://lore.kernel.org/r/[email protected]/
Eric Farman (1):
KVM: s390: Clarify SIGP orders versus STOP/RESTART
arch/s390/kvm/interrupt.c | 7 +++++++ arch/s390/kvm/kvm-s390.c | 9 +++++++-- arch/s390/kvm/kvm-s390.h | 1 + arch/s390/kvm/sigp.c | 28 ++++++++++++++++++++++++++++ 4 files changed, 43 insertions(+), 2 deletions(-)
Signature · 1 line
2.32.0
RFC PATCH v5 1/1 KVM: s390: Clarify SIGP orders versus STOP/RESTART
With KVM_CAP_S390_USER_SIGP, there are only five Signal Processor orders (CONDITIONAL EMERGENCY SIGNAL, EMERGENCY SIGNAL, EXTERNAL CALL, SENSE, and SENSE RUNNING STATUS) which are intended for frequent use and thus are processed in-kernel. The remainder are sent to userspace with the KVM_CAP_S390_USER_SIGP capability. Of those, three orders (RESTART, STOP, and STOP AND STORE STATUS) have the potential to inject work back into the kernel, and thus are asynchronous.
Let's look for those pending IRQs when processing one of the in-kernel SIGP orders, and return BUSY (CC2) if one is in process. This is in agreement with the Principles of Operation, which states that only one order can be "active" on a CPU at a time.
--- arch/s390/kvm/interrupt.c | 7 +++++++ arch/s390/kvm/kvm-s390.c | 9 +++++++-- arch/s390/kvm/kvm-s390.h | 1 + arch/s390/kvm/sigp.c | 28 ++++++++++++++++++++++++++++ 4 files changed, 43 insertions(+), 2 deletions(-)
arch/s390/kvm/interrupt.c +7 -0
@@ -2115,6 +2115,13 @@ int kvm_s390_is_stop_irq_pending(struct kvm_vcpu *vcpu)return test_bit (IRQ_PEND_SIGP_STOP , &li->pending_irqs); } +int kvm_s390_is_restart_irq_pending (struct kvm_vcpu *vcpu) +{ +struct kvm_s390_local_interrupt *li = &vcpu->arch.local_int; + +return test_bit (IRQ_PEND_RESTART , &li->pending_irqs); +} +void kvm_s390_clear_stop_irq (struct kvm_vcpu *vcpu) {struct kvm_s390_local_interrupt *li = &vcpu->arch.local_int;
arch/s390/kvm/kvm-s390.c +7 -2
@@ -4641,10 +4641,15 @@ int kvm_s390_vcpu_stop(struct kvm_vcpu *vcpu) } } -/* SIGP STOP and SIGP STOP AND STORE STATUS has been fully processed */ +/* +* Set the VCPU to STOPPED and THEN clear the interrupt flag, +* now that the SIGP STOP and SIGP STOP AND STORE STATUS orders +* have been fully processed. This will ensure that the VCPU +* is kept BUSY if another VCPU is inquiring with SIGP SENSE. +*/ +kvm_s390_set_cpuflags (vcpu,CPUSTAT_STOPPED );kvm_s390_clear_stop_irq (vcpu); -kvm_s390_set_cpuflags (vcpu,CPUSTAT_STOPPED );__disable_ibs_on_vcpu (vcpu);for (i =0 ; i < online_vcpus; i++) {
arch/s390/kvm/kvm-s390.h +1 -0
@@ -427,6 +427,7 @@ void kvm_s390_destroy_adapters(struct kvm *kvm);int kvm_s390_ext_call_pending (struct kvm_vcpu *vcpu);extern struct kvm_device_ops kvm_flic_ops;int kvm_s390_is_stop_irq_pending (struct kvm_vcpu *vcpu); +int kvm_s390_is_restart_irq_pending (struct kvm_vcpu *vcpu);void kvm_s390_clear_stop_irq (struct kvm_vcpu *vcpu);int kvm_s390_set_irq_state (struct kvm_vcpu *vcpu,void __user *buf,int len);
arch/s390/kvm/sigp.c +28 -0
@@ -276,6 +276,34 @@ static int handle_sigp_dst(struct kvm_vcpu *vcpu, u8 order_code,if (!dst_vcpu)return SIGP_CC_NOT_OPERATIONAL ; +/* +* SIGP RESTART, SIGP STOP, and SIGP STOP AND STORE STATUS orders +* are processed asynchronously. Until the affected VCPU finishes +* its work and calls back into KVM to clear the (RESTART or STOP) +* interrupt, we need to return any new non-reset orders "busy". +* +* This is important because a single VCPU could issue: +* 1) SIGP STOP $DESTINATION +* 2) SIGP SENSE $DESTINATION +* +* If the SIGP SENSE would not be rejected as "busy", it could +* return an incorrect answer as to whether the VCPU is STOPPED +* or OPERATING. +*/ +if (order_code !=SIGP_INITIAL_CPU_RESET && + order_code !=SIGP_CPU_RESET ) { +/* +* Lockless check. Both SIGP STOP and SIGP (RE)START +* properly synchronize everything while processing +* their orders, while the guest cannot observe a +* difference when issuing other orders from two +* different VCPUs. +*/ +if (kvm_s390_is_stop_irq_pending (dst_vcpu) || +kvm_s390_is_restart_irq_pending (dst_vcpu)) +return SIGP_CC_BUSY ; + } +switch (order_code) {case SIGP_SENSE : vcpu->stat.instruction_sigp_sense++;
Signature · 1 line
2.32.0
Re: [RFC PATCH v5 1/1] KVM: s390: Clarify SIGP orders versus STOP/RESTART
Am 13.12.21 um 22:05 schrieb Eric Farman:
Quoted 8 lines
With
KVM_CAP_S390_USER_SIGP, there are only five Signal Processor orders (CONDITIONAL EMERGENCY SIGNAL, EMERGENCY SIGNAL, EXTERNAL CALL, SENSE, and SENSE RUNNING STATUS) which are intended for frequent use and thus are processed in-kernel. The remainder are sent to userspace with theKVM_CAP_S390_USER_SIGPcapability. Of those, three orders (RESTART, STOP, and STOP AND STORE STATUS) have the potential to inject work back into the kernel, and thus are asynchronous.
Quoted 4 lines
Let's look for those pending IRQs when processing one of the in-kernel SIGP orders, and return BUSY (CC2) if one is in process. This is in agreement with the Principles of Operation, which states that only one order can be "active" on a CPU at a time.
As far as I understand this fixes a real bug with some test tools. Correct? Then a stable tag might be appropriate.
(Still have to review this)
How hard would it be to also build a kvm-unit test testcase?
Quoted 101 lines
Suggested-by David Hildenbrand <[email protected]>Signed-off-by Eric Farman <[email protected]>--- arch/s390/kvm/interrupt.c | 7 +++++++ arch/s390/kvm/kvm-s390.c | 9 +++++++-- arch/s390/kvm/kvm-s390.h | 1 + arch/s390/kvm/sigp.c | 28 ++++++++++++++++++++++++++++ 4 files changed, 43 insertions(+), 2 deletions(-)arch/s390/kvm/interrupt.c +7
-0@@ -2115,6 +2115,13 @@ int kvm_s390_is_stop_irq_pending(struct kvm_vcpu *vcpu)return test_bit (IRQ_PEND_SIGP_STOP , &li->pending_irqs); } +int kvm_s390_is_restart_irq_pending (struct kvm_vcpu *vcpu) +{ +struct kvm_s390_local_interrupt *li = &vcpu->arch.local_int; + +return test_bit (IRQ_PEND_RESTART , &li->pending_irqs); +} +void kvm_s390_clear_stop_irq (struct kvm_vcpu *vcpu) {struct kvm_s390_local_interrupt *li = &vcpu->arch.local_int;arch/s390/kvm/kvm-s390.c +7
-2@@ -4641,10 +4641,15 @@ int kvm_s390_vcpu_stop(struct kvm_vcpu *vcpu) } } -/* SIGP STOP and SIGP STOP AND STORE STATUS has been fully processed */ +/* +* Set the VCPU to STOPPED and THEN clear the interrupt flag, +* now that the SIGP STOP and SIGP STOP AND STORE STATUS orders +* have been fully processed. This will ensure that the VCPU +* is kept BUSY if another VCPU is inquiring with SIGP SENSE. +*/ +kvm_s390_set_cpuflags (vcpu,CPUSTAT_STOPPED );kvm_s390_clear_stop_irq (vcpu); -kvm_s390_set_cpuflags (vcpu,CPUSTAT_STOPPED );__disable_ibs_on_vcpu (vcpu);for (i =0 ; i < online_vcpus; i++) {arch/s390/kvm/kvm-s390.h +1
-0@@ -427,6 +427,7 @@ void kvm_s390_destroy_adapters(struct kvm *kvm);int kvm_s390_ext_call_pending (struct kvm_vcpu *vcpu);extern struct kvm_device_ops kvm_flic_ops;int kvm_s390_is_stop_irq_pending (struct kvm_vcpu *vcpu); +int kvm_s390_is_restart_irq_pending (struct kvm_vcpu *vcpu);void kvm_s390_clear_stop_irq (struct kvm_vcpu *vcpu);int kvm_s390_set_irq_state (struct kvm_vcpu *vcpu,void __user *buf,int len);arch/s390/kvm/sigp.c +28
-0@@ -276,6 +276,34 @@ static int handle_sigp_dst(struct kvm_vcpu *vcpu, u8 order_code,if (!dst_vcpu)return SIGP_CC_NOT_OPERATIONAL ; +/* +* SIGP RESTART, SIGP STOP, and SIGP STOP AND STORE STATUS orders +* are processed asynchronously. Until the affected VCPU finishes +* its work and calls back into KVM to clear the (RESTART or STOP) +* interrupt, we need to return any new non-reset orders "busy". +* +* This is important because a single VCPU could issue: +* 1) SIGP STOP $DESTINATION +* 2) SIGP SENSE $DESTINATION +* +* If the SIGP SENSE would not be rejected as "busy", it could +* return an incorrect answer as to whether the VCPU is STOPPED +* or OPERATING. +*/ +if (order_code !=SIGP_INITIAL_CPU_RESET && + order_code !=SIGP_CPU_RESET ) { +/* +* Lockless check. Both SIGP STOP and SIGP (RE)START +* properly synchronize everything while processing +* their orders, while the guest cannot observe a +* difference when issuing other orders from two +* different VCPUs. +*/ +if (kvm_s390_is_stop_irq_pending (dst_vcpu) || +kvm_s390_is_restart_irq_pending (dst_vcpu)) +return SIGP_CC_BUSY ; + } +switch (order_code) {case SIGP_SENSE : vcpu->stat.instruction_sigp_sense++;
Re: [RFC PATCH v5 1/1] KVM: s390: Clarify SIGP orders versus STOP/RESTART
Quoted 20 lines · Christian Borntraeger
Am 13.12.21 um 22:05 schrieb Eric Farman:
Quoted 11 lines
With
KVM_CAP_S390_USER_SIGP, there are only five Signal Processor orders (CONDITIONAL EMERGENCY SIGNAL, EMERGENCY SIGNAL, EXTERNAL CALL,
SENSE, and SENSE RUNNING STATUS) which are intended for frequent use
and thus are processed in-kernel. The remainder are sent to userspace
with theKVM_CAP_S390_USER_SIGPcapability. Of those, three orders (RESTART, STOP, and STOP AND STORE STATUS) have the potential to inject work back into the kernel, and thus are asynchronous.Quoted 6 lines
Let's look for those pending IRQs when processing one of the in- kernel
SIGP orders, and return BUSY (CC2) if one is in process. This is in agreement with the Principles of Operation, which states that only one
order can be "active" on a CPU at a time.
Quoted 2 lines
As far as I understand this fixes a real bug with some test tools. Correct?
Correct.
Quoted 1 line
Then a stable tag might be appropriate.
Agreed.
Quoted 3 lines
(Still have to review this)
How hard would it be to also build a kvm-unit test testcase?
I don't think it's too hard, and something I'd like to see done rather than the setup I'm using. It's on my list for after the holidays.
Quoted 111 lines
Quoted 111 lines
Suggested-by David Hildenbrand <[email protected]>Signed-off-by Eric Farman <[email protected]>--- arch/s390/kvm/interrupt.c | 7 +++++++ arch/s390/kvm/kvm-s390.c | 9 +++++++-- arch/s390/kvm/kvm-s390.h | 1 + arch/s390/kvm/sigp.c | 28 ++++++++++++++++++++++++++++ 4 files changed, 43 insertions(+), 2 deletions(-)arch/s390/kvm/interrupt.c +7
-0@@ -2115,6 +2115,13 @@ int kvm_s390_is_stop_irq_pending(struct kvm_vcpu *vcpu)return test_bit (IRQ_PEND_SIGP_STOP , &li->pending_irqs); } +int kvm_s390_is_restart_irq_pending (struct kvm_vcpu *vcpu) +{ +struct kvm_s390_local_interrupt *li = &vcpu->arch.local_int; + +return test_bit (IRQ_PEND_RESTART , &li->pending_irqs); +} +void kvm_s390_clear_stop_irq (struct kvm_vcpu *vcpu) {struct kvm_s390_local_interrupt *li = &vcpu->arch.local_int;arch/s390/kvm/kvm-s390.c +0
-1@@ -4641,10 +4641,15 @@ int kvm_s390_vcpu_stop(struct kvm_vcpu *vcpu) } } -/* SIGP STOP and SIGP STOP AND STORE STATUS has been fully processed */
arch/s390/kvm/kvm-s390.c +7
-1+/* +* Set the VCPU to STOPPED and THEN clear the interrupt flag, +* now that the SIGP STOP and SIGP STOP AND STORE STATUS orders +* have been fully processed. This will ensure that the VCPU +* is kept BUSY if another VCPU is inquiring with SIGP SENSE. +*/ +kvm_s390_set_cpuflags (vcpu,CPUSTAT_STOPPED );kvm_s390_clear_stop_irq (vcpu); -kvm_s390_set_cpuflags (vcpu,CPUSTAT_STOPPED );__disable_ibs_on_vcpu (vcpu);for (i =0 ; i < online_vcpus; i++) {arch/s390/kvm/kvm-s390.h +1
-0@@ -427,6 +427,7 @@ void kvm_s390_destroy_adapters(struct kvm *kvm);int kvm_s390_ext_call_pending (struct kvm_vcpu *vcpu);extern struct kvm_device_ops kvm_flic_ops;int kvm_s390_is_stop_irq_pending (struct kvm_vcpu *vcpu); +int kvm_s390_is_restart_irq_pending (struct kvm_vcpu *vcpu);void kvm_s390_clear_stop_irq (struct kvm_vcpu *vcpu); int kvm_s390_set_irq_state(struct kvm_vcpu *vcpu,void
__user*buf, int len);arch/s390/kvm/sigp.c +2
-0@@ -276,6 +276,34 @@ static int handle_sigp_dst(struct kvm_vcpu *vcpu, u8 order_code, if (!dst_vcpu)return SIGP_CC_NOT_OPERATIONAL ; +/* +* SIGP RESTART, SIGP STOP, and SIGP STOP AND STORE STATUS orders
arch/s390/kvm/sigp.c +1
-0+* are processed asynchronously. Until the affected VCPU finishes
arch/s390/kvm/sigp.c +1
-0+* its work and calls back into KVM to clear the (RESTART or STOP)
arch/s390/kvm/sigp.c +1
-0+* interrupt, we need to return any new non-reset orders "busy".
arch/s390/kvm/sigp.c +23
-0+* +* This is important because a single VCPU could issue: +* 1) SIGP STOP $DESTINATION +* 2) SIGP SENSE $DESTINATION +* +* If the SIGP SENSE would not be rejected as "busy", it could +* return an incorrect answer as to whether the VCPU is STOPPED +* or OPERATING. +*/ +if (order_code !=SIGP_INITIAL_CPU_RESET && + order_code !=SIGP_CPU_RESET ) { +/* +* Lockless check. Both SIGP STOP and SIGP (RE)START +* properly synchronize everything while processing +* their orders, while the guest cannot observe a +* difference when issuing other orders from two +* different VCPUs. +*/ +if (kvm_s390_is_stop_irq_pending (dst_vcpu) || +kvm_s390_is_restart_irq_pending (dst_vcpu)) +return SIGP_CC_BUSY ; + } +switch (order_code) {case SIGP_SENSE : vcpu->stat.instruction_sigp_sense++;
Re: [RFC PATCH v5 1/1] KVM: s390: Clarify SIGP orders versus STOP/RESTART
In reply to Eric FarmanQuoted 8 lines · Eric Farman
With
KVM_CAP_S390_USER_SIGP, there are only five Signal Processor orders (CONDITIONAL EMERGENCY SIGNAL, EMERGENCY SIGNAL, EXTERNAL CALL, SENSE, and SENSE RUNNING STATUS) which are intended for frequent use and thus are processed in-kernel. The remainder are sent to userspace with theKVM_CAP_S390_USER_SIGPcapability. Of those, three orders (RESTART, STOP, and STOP AND STORE STATUS) have the potential to inject work back into the kernel, and thus are asynchronous.
Quoted 8 lines
Let's look for those pending IRQs when processing one of the in-kernel SIGP orders, and return BUSY (CC2) if one is in process. This is in agreement with the Principles of Operation, which states that only one order can be "active" on a CPU at a time.
Suggested-by David Hildenbrand <[email protected]>Signed-off-by Eric Farman <[email protected]>---
In general, LGTM. As raised, with SIGP RESTART there are other cases we could fix in the kernel, but they are of very low priority IMHO.
Signature · 3 lines
Thanks, David / dhildenb
Re: [RFC PATCH v5 1/1] KVM: s390: Clarify SIGP orders versus STOP/RESTART
Am 15.12.21 um 14:24 schrieb David Hildenbrand:
Quoted 14 lines · Eric Farman
Quoted 13 lines
With
KVM_CAP_S390_USER_SIGP, there are only five Signal Processor orders (CONDITIONAL EMERGENCY SIGNAL, EMERGENCY SIGNAL, EXTERNAL CALL, SENSE, and SENSE RUNNING STATUS) which are intended for frequent use and thus are processed in-kernel. The remainder are sent to userspace with theKVM_CAP_S390_USER_SIGPcapability. Of those, three orders (RESTART, STOP, and STOP AND STORE STATUS) have the potential to inject work back into the kernel, and thus are asynchronous.Let's look for those pending IRQs when processing one of the in-kernel SIGP orders, and return BUSY (CC2) if one is in process. This is in agreement with the Principles of Operation, which states that only one order can be "active" on a CPU at a time.
Quoted 6 lines
Quoted 3 lines
Suggested-by David Hildenbrand <[email protected]>Signed-off-by Eric Farman <[email protected]>---
In general, LGTM. As raised, with SIGP RESTART there are other cases we could fix in the kernel, but they are of very low priority IMHO.
Does that qualify as an RB, assuming that we can fix the other cases later on?
Re: [RFC PATCH v5 1/1] KVM: s390: Clarify SIGP orders versus STOP/RESTART
Quoted 17 lines · Christian Borntraeger
Am 15.12.21 um 14:24 schrieb David Hildenbrand:
Quoted 14 lines · Eric Farman
Quoted 13 lines
With
KVM_CAP_S390_USER_SIGP, there are only five Signal Processor orders (CONDITIONAL EMERGENCY SIGNAL, EMERGENCY SIGNAL, EXTERNAL CALL, SENSE, and SENSE RUNNING STATUS) which are intended for frequent use and thus are processed in-kernel. The remainder are sent to userspace with theKVM_CAP_S390_USER_SIGPcapability. Of those, three orders (RESTART, STOP, and STOP AND STORE STATUS) have the potential to inject work back into the kernel, and thus are asynchronous.Let's look for those pending IRQs when processing one of the in-kernel SIGP orders, and return BUSY (CC2) if one is in process. This is in agreement with the Principles of Operation, which states that only one order can be "active" on a CPU at a time.
Quoted 9 lines
Quoted 6 lines
Quoted 3 lines
Suggested-by David Hildenbrand <[email protected]>Signed-off-by Eric Farman <[email protected]>---
In general, LGTM. As raised, with SIGP RESTART there are other cases we could fix in the kernel, but they are of very low priority IMHO.
Does that qualify as an RB, assuming that we can fix the other cases later on?
Certainly an Acked-by: David Hildenbrand <[email protected]> , to fully review I need some more time (maybe tomorrow)
Signature · 3 lines
Thanks, David / dhildenb
Re: [RFC PATCH v5 1/1] KVM: s390: Clarify SIGP orders versus STOP/RESTART
Quoted 34 lines · David Hildenbrand
Quoted 33 lines · Christian Borntraeger
Am 15.12.21 um 14:24 schrieb David Hildenbrand:
Quoted 30 lines · Eric Farman
Quoted 22 lines
With
KVM_CAP_S390_USER_SIGP, there are only five Signal Processor
orders (CONDITIONAL EMERGENCY SIGNAL, EMERGENCY SIGNAL, EXTERNAL CALL,
SENSE, and SENSE RUNNING STATUS) which are intended for frequent use
and thus are processed in-kernel. The remainder are sent to userspace
with theKVM_CAP_S390_USER_SIGPcapability. Of those, three orders
(RESTART, STOP, and STOP AND STORE STATUS) have the potential to
inject work back into the kernel, and thus are asynchronous.Let's look for those pending IRQs when processing one of the in-kernel
SIGP orders, and return BUSY (CC2) if one is in process. This is in
agreement with the Principles of Operation, which states that only one
order can be "active" on a CPU at a time.Quoted 3 lines
Suggested-by David Hildenbrand <[email protected]>Signed-off-by Eric Farman <[email protected]>---
In general, LGTM. As raised, with SIGP RESTART there are other cases we
could fix in the kernel, but they are of very low priority IMHO.
Quoted 8 lines
Quoted 3 lines
Does that qualify as an RB, assuming that we can fix the other cases later on?
Certainly an Acked-by: David Hildenbrand <[email protected]> , to fully
review I need some more time (maybe tomorrow)
I have the list you'd provided for the other orders on my todo list; wasn't having a lot of success building a testcase that would prove/disprove the situation we were in, besides the orders described here.
(I had previously mentioned the CPU RESET orders, but I'm worried that that was an error in my setup based on debugging I have been doing lately.)
Re: [RFC PATCH v5 1/1] KVM: s390: Clarify SIGP orders versus STOP/RESTART
In reply to Eric FarmanAm 13.12.21 um 22:05 schrieb Eric Farman:
Quoted 100 lines
With
KVM_CAP_S390_USER_SIGP, there are only five Signal Processor orders (CONDITIONAL EMERGENCY SIGNAL, EMERGENCY SIGNAL, EXTERNAL CALL, SENSE, and SENSE RUNNING STATUS) which are intended for frequent use and thus are processed in-kernel. The remainder are sent to userspace with theKVM_CAP_S390_USER_SIGPcapability. Of those, three orders (RESTART, STOP, and STOP AND STORE STATUS) have the potential to inject work back into the kernel, and thus are asynchronous.Let's look for those pending IRQs when processing one of the in-kernel SIGP orders, and return BUSY (CC2) if one is in process. This is in agreement with the Principles of Operation, which states that only one order can be "active" on a CPU at a time.
Suggested-by David Hildenbrand <[email protected]>Signed-off-by Eric Farman <[email protected]>--- arch/s390/kvm/interrupt.c | 7 +++++++ arch/s390/kvm/kvm-s390.c | 9 +++++++-- arch/s390/kvm/kvm-s390.h | 1 + arch/s390/kvm/sigp.c | 28 ++++++++++++++++++++++++++++ 4 files changed, 43 insertions(+), 2 deletions(-)arch/s390/kvm/interrupt.c +7
-0@@ -2115,6 +2115,13 @@ int kvm_s390_is_stop_irq_pending(struct kvm_vcpu *vcpu)return test_bit (IRQ_PEND_SIGP_STOP , &li->pending_irqs); } +int kvm_s390_is_restart_irq_pending (struct kvm_vcpu *vcpu) +{ +struct kvm_s390_local_interrupt *li = &vcpu->arch.local_int; + +return test_bit (IRQ_PEND_RESTART , &li->pending_irqs); +} +void kvm_s390_clear_stop_irq (struct kvm_vcpu *vcpu) {struct kvm_s390_local_interrupt *li = &vcpu->arch.local_int;arch/s390/kvm/kvm-s390.c +7
-2@@ -4641,10 +4641,15 @@ int kvm_s390_vcpu_stop(struct kvm_vcpu *vcpu) } } -/* SIGP STOP and SIGP STOP AND STORE STATUS has been fully processed */ +/* +* Set the VCPU to STOPPED and THEN clear the interrupt flag, +* now that the SIGP STOP and SIGP STOP AND STORE STATUS orders +* have been fully processed. This will ensure that the VCPU +* is kept BUSY if another VCPU is inquiring with SIGP SENSE. +*/ +kvm_s390_set_cpuflags (vcpu,CPUSTAT_STOPPED );kvm_s390_clear_stop_irq (vcpu); -kvm_s390_set_cpuflags (vcpu,CPUSTAT_STOPPED );__disable_ibs_on_vcpu (vcpu);for (i =0 ; i < online_vcpus; i++) {arch/s390/kvm/kvm-s390.h +1
-0@@ -427,6 +427,7 @@ void kvm_s390_destroy_adapters(struct kvm *kvm);int kvm_s390_ext_call_pending (struct kvm_vcpu *vcpu);extern struct kvm_device_ops kvm_flic_ops;int kvm_s390_is_stop_irq_pending (struct kvm_vcpu *vcpu); +int kvm_s390_is_restart_irq_pending (struct kvm_vcpu *vcpu);void kvm_s390_clear_stop_irq (struct kvm_vcpu *vcpu);int kvm_s390_set_irq_state (struct kvm_vcpu *vcpu,void __user *buf,int len);arch/s390/kvm/sigp.c +18
-0@@ -276,6 +276,34 @@ static int handle_sigp_dst(struct kvm_vcpu *vcpu, u8 order_code,if (!dst_vcpu)return SIGP_CC_NOT_OPERATIONAL ; +/* +* SIGP RESTART, SIGP STOP, and SIGP STOP AND STORE STATUS orders +* are processed asynchronously. Until the affected VCPU finishes +* its work and calls back into KVM to clear the (RESTART or STOP) +* interrupt, we need to return any new non-reset orders "busy". +* +* This is important because a single VCPU could issue: +* 1) SIGP STOP $DESTINATION +* 2) SIGP SENSE $DESTINATION +* +* If the SIGP SENSE would not be rejected as "busy", it could +* return an incorrect answer as to whether the VCPU is STOPPED +* or OPERATING. +*/ +if (order_code !=SIGP_INITIAL_CPU_RESET && + order_code !=SIGP_CPU_RESET ) { +/* +* Lockless check. Both SIGP STOP and SIGP (RE)START
arch/s390/kvm/sigp.c static int handle_sigp_dst(struct kvm_vcpu *vcpu, u8 order_code, · line 306
Quoted 10 lines
arch/s390/kvm/sigp.c +10
-0+* properly synchronize everything while processing +* their orders, while the guest cannot observe a +* difference when issuing other orders from two +* different VCPUs. +*/ +if (kvm_s390_is_stop_irq_pending (dst_vcpu) || +kvm_s390_is_restart_irq_pending (dst_vcpu)) +return SIGP_CC_BUSY ; + } +
Kind of similar to v1 but much simpler
took me a while to review, but I think you are right, no need for taking the lock.
Quoted 4 lines
arch/s390/kvm/sigp.c +0
-0switch (order_code) {case SIGP_SENSE : vcpu->stat.instruction_sigp_sense++;
Re: [RFC PATCH v5 1/1] KVM: s390: Clarify SIGP orders versus STOP/RESTART
In reply to Eric FarmanAm 13.12.21 um 22:05 schrieb Eric Farman:
Quoted 8 lines
With
KVM_CAP_S390_USER_SIGP, there are only five Signal Processor orders (CONDITIONAL EMERGENCY SIGNAL, EMERGENCY SIGNAL, EXTERNAL CALL, SENSE, and SENSE RUNNING STATUS) which are intended for frequent use and thus are processed in-kernel. The remainder are sent to userspace with theKVM_CAP_S390_USER_SIGPcapability. Of those, three orders (RESTART, STOP, and STOP AND STORE STATUS) have the potential to inject work back into the kernel, and thus are asynchronous.
Quoted 7 lines
Let's look for those pending IRQs when processing one of the in-kernel SIGP orders, and return BUSY (CC2) if one is in process. This is in agreement with the Principles of Operation, which states that only one order can be "active" on a CPU at a time.
Suggested-by David Hildenbrand <[email protected]>Signed-off-by Eric Farman <[email protected]>
I have applied this and will queue for next (but with cc stable).
Quoted 99 lines
--- arch/s390/kvm/interrupt.c | 7 +++++++ arch/s390/kvm/kvm-s390.c | 9 +++++++-- arch/s390/kvm/kvm-s390.h | 1 + arch/s390/kvm/sigp.c | 28 ++++++++++++++++++++++++++++ 4 files changed, 43 insertions(+), 2 deletions(-)arch/s390/kvm/interrupt.c +7
-0@@ -2115,6 +2115,13 @@ int kvm_s390_is_stop_irq_pending(struct kvm_vcpu *vcpu)return test_bit (IRQ_PEND_SIGP_STOP , &li->pending_irqs); } +int kvm_s390_is_restart_irq_pending (struct kvm_vcpu *vcpu) +{ +struct kvm_s390_local_interrupt *li = &vcpu->arch.local_int; + +return test_bit (IRQ_PEND_RESTART , &li->pending_irqs); +} +void kvm_s390_clear_stop_irq (struct kvm_vcpu *vcpu) {struct kvm_s390_local_interrupt *li = &vcpu->arch.local_int;arch/s390/kvm/kvm-s390.c +7
-2@@ -4641,10 +4641,15 @@ int kvm_s390_vcpu_stop(struct kvm_vcpu *vcpu) } } -/* SIGP STOP and SIGP STOP AND STORE STATUS has been fully processed */ +/* +* Set the VCPU to STOPPED and THEN clear the interrupt flag, +* now that the SIGP STOP and SIGP STOP AND STORE STATUS orders +* have been fully processed. This will ensure that the VCPU +* is kept BUSY if another VCPU is inquiring with SIGP SENSE. +*/ +kvm_s390_set_cpuflags (vcpu,CPUSTAT_STOPPED );kvm_s390_clear_stop_irq (vcpu); -kvm_s390_set_cpuflags (vcpu,CPUSTAT_STOPPED );__disable_ibs_on_vcpu (vcpu);for (i =0 ; i < online_vcpus; i++) {arch/s390/kvm/kvm-s390.h +1
-0@@ -427,6 +427,7 @@ void kvm_s390_destroy_adapters(struct kvm *kvm);int kvm_s390_ext_call_pending (struct kvm_vcpu *vcpu);extern struct kvm_device_ops kvm_flic_ops;int kvm_s390_is_stop_irq_pending (struct kvm_vcpu *vcpu); +int kvm_s390_is_restart_irq_pending (struct kvm_vcpu *vcpu);void kvm_s390_clear_stop_irq (struct kvm_vcpu *vcpu);int kvm_s390_set_irq_state (struct kvm_vcpu *vcpu,void __user *buf,int len);arch/s390/kvm/sigp.c +28
-0@@ -276,6 +276,34 @@ static int handle_sigp_dst(struct kvm_vcpu *vcpu, u8 order_code,if (!dst_vcpu)return SIGP_CC_NOT_OPERATIONAL ; +/* +* SIGP RESTART, SIGP STOP, and SIGP STOP AND STORE STATUS orders +* are processed asynchronously. Until the affected VCPU finishes +* its work and calls back into KVM to clear the (RESTART or STOP) +* interrupt, we need to return any new non-reset orders "busy". +* +* This is important because a single VCPU could issue: +* 1) SIGP STOP $DESTINATION +* 2) SIGP SENSE $DESTINATION +* +* If the SIGP SENSE would not be rejected as "busy", it could +* return an incorrect answer as to whether the VCPU is STOPPED +* or OPERATING. +*/ +if (order_code !=SIGP_INITIAL_CPU_RESET && + order_code !=SIGP_CPU_RESET ) { +/* +* Lockless check. Both SIGP STOP and SIGP (RE)START +* properly synchronize everything while processing +* their orders, while the guest cannot observe a +* difference when issuing other orders from two +* different VCPUs. +*/ +if (kvm_s390_is_stop_irq_pending (dst_vcpu) || +kvm_s390_is_restart_irq_pending (dst_vcpu)) +return SIGP_CC_BUSY ; + } +switch (order_code) {case SIGP_SENSE : vcpu->stat.instruction_sigp_sense++;