[v5,3/5] x86/kvm: Add "nopvspin" parameter to disable PV spinlocks
diff mbox series

Message ID 1570439071-9814-4-git-send-email-zhenzhong.duan@oracle.com
State New
Headers show
Series
  • Add a unified parameter "nopvspin"
Related show

Commit Message

Zhenzhong Duan Oct. 7, 2019, 9:04 a.m. UTC
There are cases where a guest tries to switch spinlocks to bare metal
behavior (e.g. by setting "xen_nopvspin" on XEN platform and
"hv_nopvspin" on HYPER_V).

That feature is missed on KVM, add a new parameter "nopvspin" to disable
PV spinlocks for KVM guest.

The new 'nopvspin' parameter will also replace Xen and Hyper-V specific
parameters in future patches.

Define variable nopvsin as global because it will be used in future
patches as above.

Signed-off-by: Zhenzhong Duan <zhenzhong.duan@oracle.com>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krcmar <rkrcmar@redhat.com>
Cc: Sean Christopherson <sean.j.christopherson@intel.com>
Cc: Vitaly Kuznetsov <vkuznets@redhat.com>
Cc: Wanpeng Li <wanpengli@tencent.com>
Cc: Jim Mattson <jmattson@google.com>
Cc: Joerg Roedel <joro@8bytes.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Will Deacon <will@kernel.org>
---
 Documentation/admin-guide/kernel-parameters.txt |  5 +++++
 arch/x86/include/asm/qspinlock.h                |  1 +
 arch/x86/kernel/kvm.c                           | 21 +++++++++++++++++----
 kernel/locking/qspinlock.c                      |  7 +++++++
 4 files changed, 30 insertions(+), 4 deletions(-)

Comments

Vitaly Kuznetsov Oct. 13, 2019, 9:02 a.m. UTC | #1
Zhenzhong Duan <zhenzhong.duan@oracle.com> writes:

> There are cases where a guest tries to switch spinlocks to bare metal
> behavior (e.g. by setting "xen_nopvspin" on XEN platform and
> "hv_nopvspin" on HYPER_V).
>
> That feature is missed on KVM, add a new parameter "nopvspin" to disable
> PV spinlocks for KVM guest.
>
> The new 'nopvspin' parameter will also replace Xen and Hyper-V specific
> parameters in future patches.
>
> Define variable nopvsin as global because it will be used in future
> patches as above.
>
> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@oracle.com>
> Cc: Jonathan Corbet <corbet@lwn.net>
> Cc: Thomas Gleixner <tglx@linutronix.de>
> Cc: Ingo Molnar <mingo@redhat.com>
> Cc: Borislav Petkov <bp@alien8.de>
> Cc: "H. Peter Anvin" <hpa@zytor.com>
> Cc: Paolo Bonzini <pbonzini@redhat.com>
> Cc: Radim Krcmar <rkrcmar@redhat.com>
> Cc: Sean Christopherson <sean.j.christopherson@intel.com>
> Cc: Vitaly Kuznetsov <vkuznets@redhat.com>
> Cc: Wanpeng Li <wanpengli@tencent.com>
> Cc: Jim Mattson <jmattson@google.com>
> Cc: Joerg Roedel <joro@8bytes.org>
> Cc: Peter Zijlstra <peterz@infradead.org>
> Cc: Will Deacon <will@kernel.org>
> ---
>  Documentation/admin-guide/kernel-parameters.txt |  5 +++++
>  arch/x86/include/asm/qspinlock.h                |  1 +
>  arch/x86/kernel/kvm.c                           | 21 +++++++++++++++++----
>  kernel/locking/qspinlock.c                      |  7 +++++++
>  4 files changed, 30 insertions(+), 4 deletions(-)
>
> diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
> index c7ac2f3..89d77ea 100644
> --- a/Documentation/admin-guide/kernel-parameters.txt
> +++ b/Documentation/admin-guide/kernel-parameters.txt
> @@ -5330,6 +5330,11 @@
>  			as generic guest with no PV drivers. Currently support
>  			XEN HVM, KVM, HYPER_V and VMWARE guest.
>  
> +	nopvspin	[X86,KVM]
> +			Disables the qspinlock slow path using PV optimizations
> +			which allow the hypervisor to 'idle' the guest on lock
> +			contention.
> +
>  	xirc2ps_cs=	[NET,PCMCIA]
>  			Format:
>  			<irq>,<irq_mask>,<io>,<full_duplex>,<do_sound>,<lockup_hack>[,<irq2>[,<irq3>[,<irq4>]]]
> diff --git a/arch/x86/include/asm/qspinlock.h b/arch/x86/include/asm/qspinlock.h
> index 444d6fd..d86ab94 100644
> --- a/arch/x86/include/asm/qspinlock.h
> +++ b/arch/x86/include/asm/qspinlock.h
> @@ -32,6 +32,7 @@ static __always_inline u32 queued_fetch_set_pending_acquire(struct qspinlock *lo
>  extern void __pv_init_lock_hash(void);
>  extern void __pv_queued_spin_lock_slowpath(struct qspinlock *lock, u32 val);
>  extern void __raw_callee_save___pv_queued_spin_unlock(struct qspinlock *lock);
> +extern bool nopvspin;
>  
>  #define	queued_spin_unlock queued_spin_unlock
>  /**
> diff --git a/arch/x86/kernel/kvm.c b/arch/x86/kernel/kvm.c
> index ef836d6..6e14bd4 100644
> --- a/arch/x86/kernel/kvm.c
> +++ b/arch/x86/kernel/kvm.c
> @@ -825,18 +825,31 @@ __visible bool __kvm_vcpu_is_preempted(long cpu)
>   */
>  void __init kvm_spinlock_init(void)
>  {
> -	/* Does host kernel support KVM_FEATURE_PV_UNHALT? */
> -	if (!kvm_para_has_feature(KVM_FEATURE_PV_UNHALT))
> +	/*
> +	 * Disable PV qspinlocks if host kernel doesn't support
> +	 * KVM_FEATURE_PV_UNHALT feature or there is only 1 vCPU.
> +	 * virt_spin_lock_key is enabled to avoid lock holder
> +	 * preemption issue.
> +	 */
> +	if (!kvm_para_has_feature(KVM_FEATURE_PV_UNHALT) ||
> +	    num_possible_cpus() == 1) {
> +		pr_info("PV spinlocks disabled\n");

Why don't we need static_branch_disable(&virt_spin_lock_key) here?

Also, as you're printing the exact reason for PV spinlocks disablement
in other cases, I'd suggest separating "no host support" and "single
CPU" cases.

>  		return;
> +	}
>  
>  	if (kvm_para_has_hint(KVM_HINTS_REALTIME)) {
> +		pr_info("PV spinlocks disabled with KVM_HINTS_REALTIME hints.\n");
>  		static_branch_disable(&virt_spin_lock_key);
>  		return;
>  	}
>  
> -	/* Don't use the pvqspinlock code if there is only 1 vCPU. */
> -	if (num_possible_cpus() == 1)
> +	if (nopvspin) {
> +		pr_info("PV spinlocks disabled forced by \"nopvspin\" parameter.\n");

Nit: to make it sound better a comma is missing between 'disabled' and
'forced', or

"PV spinlocks forcefully disabled by ..." if you prefer.

> +		static_branch_disable(&virt_spin_lock_key);
>  		return;
> +	}
> +
> +	pr_info("PV spinlocks enabled\n");
>  
>  	__pv_init_lock_hash();
>  	pv_ops.lock.queued_spin_lock_slowpath = __pv_queued_spin_lock_slowpath;
> diff --git a/kernel/locking/qspinlock.c b/kernel/locking/qspinlock.c
> index 2473f10..75193d6 100644
> --- a/kernel/locking/qspinlock.c
> +++ b/kernel/locking/qspinlock.c
> @@ -580,4 +580,11 @@ void queued_spin_lock_slowpath(struct qspinlock *lock, u32 val)
>  #include "qspinlock_paravirt.h"
>  #include "qspinlock.c"
>  
> +bool nopvspin __initdata;
> +static __init int parse_nopvspin(char *arg)
> +{
> +	nopvspin = true;
> +	return 0;
> +}
> +early_param("nopvspin", parse_nopvspin);
>  #endif
Zhenzhong Duan Oct. 14, 2019, 1:52 a.m. UTC | #2
On 2019/10/13 17:02, Vitaly Kuznetsov wrote:
> Zhenzhong Duan <zhenzhong.duan@oracle.com> writes:
...snip
> diff --git a/arch/x86/kernel/kvm.c b/arch/x86/kernel/kvm.c
> index ef836d6..6e14bd4 100644
> --- a/arch/x86/kernel/kvm.c
> +++ b/arch/x86/kernel/kvm.c
> @@ -825,18 +825,31 @@ __visible bool __kvm_vcpu_is_preempted(long cpu)
>    */
>   void __init kvm_spinlock_init(void)
>   {
> -	/* Does host kernel support KVM_FEATURE_PV_UNHALT? */
> -	if (!kvm_para_has_feature(KVM_FEATURE_PV_UNHALT))
> +	/*
> +	 * Disable PV qspinlocks if host kernel doesn't support
> +	 * KVM_FEATURE_PV_UNHALT feature or there is only 1 vCPU.
> +	 * virt_spin_lock_key is enabled to avoid lock holder
> +	 * preemption issue.
> +	 */
> +	if (!kvm_para_has_feature(KVM_FEATURE_PV_UNHALT) ||
> +	    num_possible_cpus() == 1) {
> +		pr_info("PV spinlocks disabled\n");
> Why don't we need static_branch_disable(&virt_spin_lock_key) here?

Thanks for review.

I have a brief explanation in above comment area.

Boris also raised the same question in v4 and see my detailed explanation

in https://lkml.org/lkml/2019/10/6/39

>
> Also, as you're printing the exact reason for PV spinlocks disablement
> in other cases, I'd suggest separating "no host support" and "single
> CPU" cases.

Will do after reaching a consensus on your first question.

>
>>   		return;
>> +	}
>>   
>>   	if (kvm_para_has_hint(KVM_HINTS_REALTIME)) {
>> +		pr_info("PV spinlocks disabled with KVM_HINTS_REALTIME hints.\n");
>>   		static_branch_disable(&virt_spin_lock_key);
>>   		return;
>>   	}
>>   
>> -	/* Don't use the pvqspinlock code if there is only 1 vCPU. */
>> -	if (num_possible_cpus() == 1)
>> +	if (nopvspin) {
>> +		pr_info("PV spinlocks disabled forced by \"nopvspin\" parameter.\n");
> Nit: to make it sound better a comma is missing between 'disabled' and
> 'forced', or
>
> "PV spinlocks forcefully disabled by ..." if you prefer.

Will do.

Zhenzhong
Vitaly Kuznetsov Oct. 14, 2019, 9:18 a.m. UTC | #3
Zhenzhong Duan <zhenzhong.duan@oracle.com> writes:

> On 2019/10/13 17:02, Vitaly Kuznetsov wrote:
>> Zhenzhong Duan <zhenzhong.duan@oracle.com> writes:
> ...snip
>> diff --git a/arch/x86/kernel/kvm.c b/arch/x86/kernel/kvm.c
>> index ef836d6..6e14bd4 100644
>> --- a/arch/x86/kernel/kvm.c
>> +++ b/arch/x86/kernel/kvm.c
>> @@ -825,18 +825,31 @@ __visible bool __kvm_vcpu_is_preempted(long cpu)
>>    */
>>   void __init kvm_spinlock_init(void)
>>   {
>> -	/* Does host kernel support KVM_FEATURE_PV_UNHALT? */
>> -	if (!kvm_para_has_feature(KVM_FEATURE_PV_UNHALT))
>> +	/*
>> +	 * Disable PV qspinlocks if host kernel doesn't support
>> +	 * KVM_FEATURE_PV_UNHALT feature or there is only 1 vCPU.
>> +	 * virt_spin_lock_key is enabled to avoid lock holder
>> +	 * preemption issue.
>> +	 */
>> +	if (!kvm_para_has_feature(KVM_FEATURE_PV_UNHALT) ||
>> +	    num_possible_cpus() == 1) {
>> +		pr_info("PV spinlocks disabled\n");
>> Why don't we need static_branch_disable(&virt_spin_lock_key) here?
>
> Thanks for review.
>
> I have a brief explanation in above comment area.
>
> Boris also raised the same question in v4 and see my detailed explanation
>
> in https://lkml.org/lkml/2019/10/6/39
>
>>
>> Also, as you're printing the exact reason for PV spinlocks disablement
>> in other cases, I'd suggest separating "no host support" and "single
>> CPU" cases.
>
> Will do after reaching a consensus on your first question.

Oh, sorry I missed v4 discussion. As I'm not the first to ask why we
don't do static_branch_disable(&virt_spin_lock_key) here I suggest we do
the followin:

- Split !kvm_para_has_feature(KVM_FEATURE_PV_UNHALT) and
  num_possible_cpus() == 1 cases
- Do static_branch_disable(&virt_spin_lock_key) for UP case (just for
  consistency).
- Add a comment why we don't do that for
  !kvm_para_has_feature(KVM_FEATURE_PV_UNHALT) case (basically, what you
  replied to Boris)

This will also allow us to print the exact reason.

>
>>
>>>   		return;
>>> +	}
>>>   
>>>   	if (kvm_para_has_hint(KVM_HINTS_REALTIME)) {
>>> +		pr_info("PV spinlocks disabled with KVM_HINTS_REALTIME hints.\n");
>>>   		static_branch_disable(&virt_spin_lock_key);
>>>   		return;
>>>   	}
>>>   
>>> -	/* Don't use the pvqspinlock code if there is only 1 vCPU. */
>>> -	if (num_possible_cpus() == 1)
>>> +	if (nopvspin) {
>>> +		pr_info("PV spinlocks disabled forced by \"nopvspin\" parameter.\n");
>> Nit: to make it sound better a comma is missing between 'disabled' and
>> 'forced', or
>>
>> "PV spinlocks forcefully disabled by ..." if you prefer.
>
> Will do.
>
> Zhenzhong
>
>

Patch
diff mbox series

diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
index c7ac2f3..89d77ea 100644
--- a/Documentation/admin-guide/kernel-parameters.txt
+++ b/Documentation/admin-guide/kernel-parameters.txt
@@ -5330,6 +5330,11 @@ 
 			as generic guest with no PV drivers. Currently support
 			XEN HVM, KVM, HYPER_V and VMWARE guest.
 
+	nopvspin	[X86,KVM]
+			Disables the qspinlock slow path using PV optimizations
+			which allow the hypervisor to 'idle' the guest on lock
+			contention.
+
 	xirc2ps_cs=	[NET,PCMCIA]
 			Format:
 			<irq>,<irq_mask>,<io>,<full_duplex>,<do_sound>,<lockup_hack>[,<irq2>[,<irq3>[,<irq4>]]]
diff --git a/arch/x86/include/asm/qspinlock.h b/arch/x86/include/asm/qspinlock.h
index 444d6fd..d86ab94 100644
--- a/arch/x86/include/asm/qspinlock.h
+++ b/arch/x86/include/asm/qspinlock.h
@@ -32,6 +32,7 @@  static __always_inline u32 queued_fetch_set_pending_acquire(struct qspinlock *lo
 extern void __pv_init_lock_hash(void);
 extern void __pv_queued_spin_lock_slowpath(struct qspinlock *lock, u32 val);
 extern void __raw_callee_save___pv_queued_spin_unlock(struct qspinlock *lock);
+extern bool nopvspin;
 
 #define	queued_spin_unlock queued_spin_unlock
 /**
diff --git a/arch/x86/kernel/kvm.c b/arch/x86/kernel/kvm.c
index ef836d6..6e14bd4 100644
--- a/arch/x86/kernel/kvm.c
+++ b/arch/x86/kernel/kvm.c
@@ -825,18 +825,31 @@  __visible bool __kvm_vcpu_is_preempted(long cpu)
  */
 void __init kvm_spinlock_init(void)
 {
-	/* Does host kernel support KVM_FEATURE_PV_UNHALT? */
-	if (!kvm_para_has_feature(KVM_FEATURE_PV_UNHALT))
+	/*
+	 * Disable PV qspinlocks if host kernel doesn't support
+	 * KVM_FEATURE_PV_UNHALT feature or there is only 1 vCPU.
+	 * virt_spin_lock_key is enabled to avoid lock holder
+	 * preemption issue.
+	 */
+	if (!kvm_para_has_feature(KVM_FEATURE_PV_UNHALT) ||
+	    num_possible_cpus() == 1) {
+		pr_info("PV spinlocks disabled\n");
 		return;
+	}
 
 	if (kvm_para_has_hint(KVM_HINTS_REALTIME)) {
+		pr_info("PV spinlocks disabled with KVM_HINTS_REALTIME hints.\n");
 		static_branch_disable(&virt_spin_lock_key);
 		return;
 	}
 
-	/* Don't use the pvqspinlock code if there is only 1 vCPU. */
-	if (num_possible_cpus() == 1)
+	if (nopvspin) {
+		pr_info("PV spinlocks disabled forced by \"nopvspin\" parameter.\n");
+		static_branch_disable(&virt_spin_lock_key);
 		return;
+	}
+
+	pr_info("PV spinlocks enabled\n");
 
 	__pv_init_lock_hash();
 	pv_ops.lock.queued_spin_lock_slowpath = __pv_queued_spin_lock_slowpath;
diff --git a/kernel/locking/qspinlock.c b/kernel/locking/qspinlock.c
index 2473f10..75193d6 100644
--- a/kernel/locking/qspinlock.c
+++ b/kernel/locking/qspinlock.c
@@ -580,4 +580,11 @@  void queued_spin_lock_slowpath(struct qspinlock *lock, u32 val)
 #include "qspinlock_paravirt.h"
 #include "qspinlock.c"
 
+bool nopvspin __initdata;
+static __init int parse_nopvspin(char *arg)
+{
+	nopvspin = true;
+	return 0;
+}
+early_param("nopvspin", parse_nopvspin);
 #endif