[PATCH v3 0/1] KVM: s390: diag9c forwarding

[PATCH v3 0/1] KVM: s390: diag9c forwarding

[Pierre Morel]

This patch will forward the yieldto hypercall (diag9c) if in the host
the target CPU is also not running. As we do not yet have performance
data (and recommendations) the default is turned off, but this can
be changed during runtime.


Pierre Morel (1):
  s390:kvm: diag9c forwarding

 Documentation/virt/kvm/s390-diag.rst | 33 ++++++++++++++++++++++++++++
 arch/s390/include/asm/kvm_host.h     |  1 +
 arch/s390/include/asm/smp.h          |  1 +
 arch/s390/kernel/smp.c               |  1 +
 arch/s390/kvm/diag.c                 | 31 +++++++++++++++++++++++---
 arch/s390/kvm/kvm-s390.c             |  6 +++++
 arch/s390/kvm/kvm-s390.h             |  8 +++++++
 7 files changed, 78 insertions(+), 3 deletions(-)

-- 
2.17.1

Changelog:

From v2:

- update overrun comparison, avoid comparison on unsigned
  (Pierre)

From v1:

- more precise comments
  (Connie)

- Documentation
  (Janosch)

[PATCH v3 1/1] s390:kvm: diag9c forwarding

[Pierre Morel]

When we receive intercept a DIAG_9C from the guest we verify
that the target real CPU associated with the virtual CPU
designated by the guest is running and if not we forward the
DIAG_9C to the target real CPU.

To avoid a diag9c storm we allow a maximal rate of diag9c forwarding.

The rate is calculated as a count per second defined as a
new parameter of the s390 kvm module: diag9c_forwarding_hz .

The default value is to not forward diag9c.

Signed-off-by: Pierre Morel <pmorel@linux.ibm.com>
---
 Documentation/virt/kvm/s390-diag.rst | 33 ++++++++++++++++++++++++++++
 arch/s390/include/asm/kvm_host.h     |  1 +
 arch/s390/include/asm/smp.h          |  1 +
 arch/s390/kernel/smp.c               |  1 +
 arch/s390/kvm/diag.c                 | 31 +++++++++++++++++++++++---
 arch/s390/kvm/kvm-s390.c             |  6 +++++
 arch/s390/kvm/kvm-s390.h             |  8 +++++++
 7 files changed, 78 insertions(+), 3 deletions(-)

diff --git a/Documentation/virt/kvm/s390-diag.rst b/Documentation/virt/kvm/s390-diag.rst
index eaac4864d3d6..a6371bc4ea90 100644
--- a/Documentation/virt/kvm/s390-diag.rst
+++ b/Documentation/virt/kvm/s390-diag.rst
@@ -84,3 +84,36 @@ If the function code specifies 0x501, breakpoint functions may be performed.
 This function code is handled by userspace.
 
 This diagnose function code has no subfunctions and uses no parameters.
+
+
+DIAGNOSE function code 'X'9C - Voluntary Time Slice Yield
+---------------------------------------------------------
+
+General register 1 contains the target CPU address.
+
+In a guest of a hypervisor like LPAR, KVM or z/VM using shared host CPUs,
+DIAGNOSE with function code 'X'9C may improve system performance by
+yielding the host CPU on which the guest CPU is running to be assigned
+to another guest CPU, preferably the logical CPU containing the specified
+target CPU.
+
+
+DIAG 'X'9C forwarding
++++++++++++++++++++++
+
+Under KVM, the guest operating system may send a DIAGNOSE code 'X'9C to
+the host when it fails to acquire a spinlock for a virtual CPU
+and detects that the host CPU on which the virtual guest CPU owner is
+assigned to is not running to try to get this host CPU running and
+consequently the guest virtual CPU running and freeing the lock.
+
+However, on the logical partition the real CPU on which the previously
+targeted host CPU is assign may itself not be running.
+By forwarding the DIAGNOSE code 'X'9C, initially sent by the guest,
+from the host to LPAR hypervisor, this one will hopefully schedule
+the host CPU which will let KVM run the target guest CPU.
+
+diag9c_forwarding_hz
+    KVM kernel parameter allowing to specify the maximum number of DIAGNOSE
+    'X'9C forwarding per second in the purpose of avoiding a DIAGNOSE 'X'9C
+    forwarding storm.
diff --git a/arch/s390/include/asm/kvm_host.h b/arch/s390/include/asm/kvm_host.h
index 527776a1f076..cb19508c22fb 100644
--- a/arch/s390/include/asm/kvm_host.h
+++ b/arch/s390/include/asm/kvm_host.h
@@ -456,6 +456,7 @@ struct kvm_vcpu_stat {
 	u64 diagnose_44;
 	u64 diagnose_9c;
 	u64 diagnose_9c_ignored;
+	u64 diagnose_9c_forward;
 	u64 diagnose_258;
 	u64 diagnose_308;
 	u64 diagnose_500;
diff --git a/arch/s390/include/asm/smp.h b/arch/s390/include/asm/smp.h
index 01e360004481..e317fd4866c1 100644
--- a/arch/s390/include/asm/smp.h
+++ b/arch/s390/include/asm/smp.h
@@ -63,5 +63,6 @@ extern void __noreturn cpu_die(void);
 extern void __cpu_die(unsigned int cpu);
 extern int __cpu_disable(void);
 extern void schedule_mcck_handler(void);
+void notrace smp_yield_cpu(int cpu);
 
 #endif /* __ASM_SMP_H */
diff --git a/arch/s390/kernel/smp.c b/arch/s390/kernel/smp.c
index 27c763014114..15e207a671fd 100644
--- a/arch/s390/kernel/smp.c
+++ b/arch/s390/kernel/smp.c
@@ -422,6 +422,7 @@ void notrace smp_yield_cpu(int cpu)
 	asm volatile("diag %0,0,0x9c"
 		     : : "d" (pcpu_devices[cpu].address));
 }
+EXPORT_SYMBOL(smp_yield_cpu);
 
 /*
  * Send cpus emergency shutdown signal. This gives the cpus the
diff --git a/arch/s390/kvm/diag.c b/arch/s390/kvm/diag.c
index 5b8ec1c447e1..02c146f9e5cd 100644
--- a/arch/s390/kvm/diag.c
+++ b/arch/s390/kvm/diag.c
@@ -150,6 +150,19 @@ static int __diag_time_slice_end(struct kvm_vcpu *vcpu)
 	return 0;
 }
 
+static int forward_cnt;
+static unsigned long cur_slice;
+
+static int diag9c_forwarding_overrun(void)
+{
+	/* Reset the count on a new slice */
+	if (time_after(jiffies, cur_slice)) {
+		cur_slice = jiffies;
+		forward_cnt = diag9c_forwarding_hz / HZ;
+	}
+	return forward_cnt-- <= 0 ? 1 : 0;
+}
+
 static int __diag_time_slice_end_directed(struct kvm_vcpu *vcpu)
 {
 	struct kvm_vcpu *tcpu;
@@ -167,9 +180,21 @@ static int __diag_time_slice_end_directed(struct kvm_vcpu *vcpu)
 	if (!tcpu)
 		goto no_yield;
 
-	/* target already running */
-	if (READ_ONCE(tcpu->cpu) >= 0)
-		goto no_yield;
+	/* target guest VCPU already running */
+	if (READ_ONCE(tcpu->cpu) >= 0) {
+		if (!diag9c_forwarding_hz || diag9c_forwarding_overrun())
+			goto no_yield;
+
+		/* target host CPU already running */
+		if (!vcpu_is_preempted(tcpu->cpu))
+			goto no_yield;
+		smp_yield_cpu(tcpu->cpu);
+		VCPU_EVENT(vcpu, 5,
+			   "diag time slice end directed to %d: yield forwarded",
+			   tid);
+		vcpu->stat.diagnose_9c_forward++;
+		return 0;
+	}
 
 	if (kvm_vcpu_yield_to(tcpu) <= 0)
 		goto no_yield;
diff --git a/arch/s390/kvm/kvm-s390.c b/arch/s390/kvm/kvm-s390.c
index 053ef36784e4..c23e22610a89 100644
--- a/arch/s390/kvm/kvm-s390.c
+++ b/arch/s390/kvm/kvm-s390.c
@@ -157,6 +157,7 @@ struct kvm_stats_debugfs_item debugfs_entries[] = {
 	VCPU_STAT("instruction_diag_44", diagnose_44),
 	VCPU_STAT("instruction_diag_9c", diagnose_9c),
 	VCPU_STAT("diag_9c_ignored", diagnose_9c_ignored),
+	VCPU_STAT("diag_9c_forward", diagnose_9c_forward),
 	VCPU_STAT("instruction_diag_258", diagnose_258),
 	VCPU_STAT("instruction_diag_308", diagnose_308),
 	VCPU_STAT("instruction_diag_500", diagnose_500),
@@ -190,6 +191,11 @@ static bool use_gisa  = true;
 module_param(use_gisa, bool, 0644);
 MODULE_PARM_DESC(use_gisa, "Use the GISA if the host supports it.");
 
+/* maximum diag9c forwarding per second */
+unsigned int diag9c_forwarding_hz;
+module_param(diag9c_forwarding_hz, uint, 0644);
+MODULE_PARM_DESC(diag9c_forwarding_hz, "Maximum diag9c forwarding per second");
+
 /*
  * For now we handle at most 16 double words as this is what the s390 base
  * kernel handles and stores in the prefix page. If we ever need to go beyond
diff --git a/arch/s390/kvm/kvm-s390.h b/arch/s390/kvm/kvm-s390.h
index 79dcd647b378..9fad25109b0d 100644
--- a/arch/s390/kvm/kvm-s390.h
+++ b/arch/s390/kvm/kvm-s390.h
@@ -471,4 +471,12 @@ void kvm_s390_reinject_machine_check(struct kvm_vcpu *vcpu,
  * @kvm: the KVM guest
  */
 void kvm_s390_vcpu_crypto_reset_all(struct kvm *kvm);
+
+/**
+ * diag9c_forwarding_hz
+ *
+ * Set the maximum number of diag9c forwarding per second
+ */
+extern unsigned int diag9c_forwarding_hz;
+
 #endif
-- 
2.17.1

Re: [PATCH v3 1/1] s390:kvm: diag9c forwarding

[Cornelia Huck]

On Mon, 15 Feb 2021 17:06:50 +0100
Pierre Morel <pmorel@linux.ibm.com> wrote:

Make $SUBJECT

"KVM: s390: diag9c (directed yield) forwarding" ?

> When we receive intercept a DIAG_9C from the guest we verify

Either 'receive' or 'intercept', I guess :)

> that the target real CPU associated with the virtual CPU
> designated by the guest is running and if not we forward the
> DIAG_9C to the target real CPU.
> 
> To avoid a diag9c storm we allow a maximal rate of diag9c forwarding.
> 
> The rate is calculated as a count per second defined as a
> new parameter of the s390 kvm module: diag9c_forwarding_hz .
> 
> The default value is to not forward diag9c.
> 
> Signed-off-by: Pierre Morel <pmorel@linux.ibm.com>
> ---
>  Documentation/virt/kvm/s390-diag.rst | 33 ++++++++++++++++++++++++++++
>  arch/s390/include/asm/kvm_host.h     |  1 +
>  arch/s390/include/asm/smp.h          |  1 +
>  arch/s390/kernel/smp.c               |  1 +
>  arch/s390/kvm/diag.c                 | 31 +++++++++++++++++++++++---
>  arch/s390/kvm/kvm-s390.c             |  6 +++++
>  arch/s390/kvm/kvm-s390.h             |  8 +++++++
>  7 files changed, 78 insertions(+), 3 deletions(-)
> 
> diff --git a/Documentation/virt/kvm/s390-diag.rst b/Documentation/virt/kvm/s390-diag.rst
> index eaac4864d3d6..a6371bc4ea90 100644
> --- a/Documentation/virt/kvm/s390-diag.rst
> +++ b/Documentation/virt/kvm/s390-diag.rst
> @@ -84,3 +84,36 @@ If the function code specifies 0x501, breakpoint functions may be performed.
>  This function code is handled by userspace.
>  
>  This diagnose function code has no subfunctions and uses no parameters.
> +
> +
> +DIAGNOSE function code 'X'9C - Voluntary Time Slice Yield
> +---------------------------------------------------------
> +
> +General register 1 contains the target CPU address.
> +
> +In a guest of a hypervisor like LPAR, KVM or z/VM using shared host CPUs,
> +DIAGNOSE with function code 'X'9C may improve system performance by
> +yielding the host CPU on which the guest CPU is running to be assigned
> +to another guest CPU, preferably the logical CPU containing the specified
> +target CPU.
> +
> +
> +DIAG 'X'9C forwarding
> ++++++++++++++++++++++
> +
> +Under KVM, the guest operating system may send a DIAGNOSE code 'X'9C to
> +the host when it fails to acquire a spinlock for a virtual CPU
> +and detects that the host CPU on which the virtual guest CPU owner is
> +assigned to is not running to try to get this host CPU running and
> +consequently the guest virtual CPU running and freeing the lock.

What about:

"The guest may send a DIAGNOSE 0x9c in order to yield to a certain
other vcpu. An example is a Linux guest that tries to yield to the vcpu
that is currently holding a spinlock, but not running."

> +
> +However, on the logical partition the real CPU on which the previously
> +targeted host CPU is assign may itself not be running.

"However, on the host the real cpu backing the vcpu may itself not be
running."

> +By forwarding the DIAGNOSE code 'X'9C, initially sent by the guest,
> +from the host to LPAR hypervisor, this one will hopefully schedule
> +the host CPU which will let KVM run the target guest CPU.

"Forwarding the DIAGNOSE 0x9c initially sent by the guest to yield to
the backing cpu will hopefully cause that cpu, and thus subsequently
the guest's vcpu, to be scheduled."

[I don't think we should explicitly talk about LPAR here, as the same
should apply if we are running second-or-deeper level, right?]

> +
> +diag9c_forwarding_hz
> +    KVM kernel parameter allowing to specify the maximum number of DIAGNOSE
> +    'X'9C forwarding per second in the purpose of avoiding a DIAGNOSE 'X'9C
> +    forwarding storm.

I think 0x9c is the more common way to write the hex code.

Also,

"A value of 0 turns the forwarding off" ?

(...)

> diff --git a/arch/s390/kernel/smp.c b/arch/s390/kernel/smp.c
> index 27c763014114..15e207a671fd 100644
> --- a/arch/s390/kernel/smp.c
> +++ b/arch/s390/kernel/smp.c
> @@ -422,6 +422,7 @@ void notrace smp_yield_cpu(int cpu)
>  	asm volatile("diag %0,0,0x9c"
>  		     : : "d" (pcpu_devices[cpu].address));
>  }
> +EXPORT_SYMBOL(smp_yield_cpu);

EXPORT_SYMBOL_GPL?

>  
>  /*
>   * Send cpus emergency shutdown signal. This gives the cpus the

(...)

> @@ -190,6 +191,11 @@ static bool use_gisa  = true;
>  module_param(use_gisa, bool, 0644);
>  MODULE_PARM_DESC(use_gisa, "Use the GISA if the host supports it.");
>  
> +/* maximum diag9c forwarding per second */
> +unsigned int diag9c_forwarding_hz;
> +module_param(diag9c_forwarding_hz, uint, 0644);
> +MODULE_PARM_DESC(diag9c_forwarding_hz, "Maximum diag9c forwarding per second");

Maybe also add "(0 to turn off forwarding)" here?

> +
>  /*
>   * For now we handle at most 16 double words as this is what the s390 base
>   * kernel handles and stores in the prefix page. If we ever need to go beyond

(...)

Re: [PATCH v3 1/1] s390:kvm: diag9c forwarding

[Pierre Morel]

On 2/19/21 5:09 PM, Cornelia Huck wrote:
> On Mon, 15 Feb 2021 17:06:50 +0100
> Pierre Morel <pmorel@linux.ibm.com> wrote:
> 
> Make $SUBJECT
> 
> "KVM: s390: diag9c (directed yield) forwarding" ?
> 
>> When we receive intercept a DIAG_9C from the guest we verify
> 
> Either 'receive' or 'intercept', I guess :)

right.

...snip...

>> +DIAG 'X'9C forwarding
>> ++++++++++++++++++++++
>> +
>> +Under KVM, the guest operating system may send a DIAGNOSE code 'X'9C to
>> +the host when it fails to acquire a spinlock for a virtual CPU
>> +and detects that the host CPU on which the virtual guest CPU owner is
>> +assigned to is not running to try to get this host CPU running and
>> +consequently the guest virtual CPU running and freeing the lock.
> 
> What about:
> 
> "The guest may send a DIAGNOSE 0x9c in order to yield to a certain
> other vcpu. An example is a Linux guest that tries to yield to the vcpu
> that is currently holding a spinlock, but not running."

Yes, thanks.

> 
>> +
>> +However, on the logical partition the real CPU on which the previously
>> +targeted host CPU is assign may itself not be running.
> 
> "However, on the host the real cpu backing the vcpu may itself not be
> running."

Yes, better too, thanks.

> 
>> +By forwarding the DIAGNOSE code 'X'9C, initially sent by the guest,
>> +from the host to LPAR hypervisor, this one will hopefully schedule
>> +the host CPU which will let KVM run the target guest CPU.
> 
> "Forwarding the DIAGNOSE 0x9c initially sent by the guest to yield to
> the backing cpu will hopefully cause that cpu, and thus subsequently
> the guest's vcpu, to be scheduled."

yes.

> 
> [I don't think we should explicitly talk about LPAR here, as the same
> should apply if we are running second-or-deeper level, right?]

yes right.

> 
>> +
>> +diag9c_forwarding_hz
>> +    KVM kernel parameter allowing to specify the maximum number of DIAGNOSE
>> +    'X'9C forwarding per second in the purpose of avoiding a DIAGNOSE 'X'9C
>> +    forwarding storm.
> 
> I think 0x9c is the more common way to write the hex code.

yes it is the purpose was to keep it consistent with the existing 
documentation

> 
> Also,
> 
> "A value of 0 turns the forwarding off" ?

yes, I can add this explicitly.

> 
> (...)
> 
>> diff --git a/arch/s390/kernel/smp.c b/arch/s390/kernel/smp.c
>> index 27c763014114..15e207a671fd 100644
>> --- a/arch/s390/kernel/smp.c
>> +++ b/arch/s390/kernel/smp.c
>> @@ -422,6 +422,7 @@ void notrace smp_yield_cpu(int cpu)
>>   	asm volatile("diag %0,0,0x9c"
>>   		     : : "d" (pcpu_devices[cpu].address));
>>   }
>> +EXPORT_SYMBOL(smp_yield_cpu);
> 
> EXPORT_SYMBOL_GPL?

OK, clear.

> 
>>   
>>   /*
>>    * Send cpus emergency shutdown signal. This gives the cpus the
> 
> (...)
> 
>> @@ -190,6 +191,11 @@ static bool use_gisa  = true;
>>   module_param(use_gisa, bool, 0644);
>>   MODULE_PARM_DESC(use_gisa, "Use the GISA if the host supports it.");
>>   
>> +/* maximum diag9c forwarding per second */
>> +unsigned int diag9c_forwarding_hz;
>> +module_param(diag9c_forwarding_hz, uint, 0644);
>> +MODULE_PARM_DESC(diag9c_forwarding_hz, "Maximum diag9c forwarding per second");
> 
> Maybe also add "(0 to turn off forwarding)" here?

OK, I will add it.

Thanks for the comments,
Regards,
Pierre


-- 
Pierre Morel
IBM Lab Boeblingen