There are cases folks want to disable spinlock optimization for
debug/test purpose. Xen and hyperv already have parameters "xen_nopvspin"
and "hv_nopvspin" to support that, but kvm doesn't.
The first patch adds that feature to KVM guest with "nopvspin".
For compatibility reason original parameters "xen_nopvspin" and
"hv_nopvspin" are retained and marked obsolete.
v6:
PATCH1: add Reviewed-by [Vitaly Kuznetsov]
PATCH2: change 'pv' to 'PV', add Reviewed-by [Vitaly Kuznetsov]
PATCH3: refactor 'if' branch in kvm_spinlock_init() [Vitaly Kuznetsov]
v5:
PATCH1: new patch to revert a currently unnecessory commit,
code is simpler a bit after that change. [Boris Ostrovsky]
PATCH3: fold 'if' statement,add comments on virt_spin_lock_key,
reorder with PATCH2 to better reflect dependency
PATCH4: fold 'if' statement, add Reviewed-by [Boris Ostrovsky]
PATCH5: add Reviewed-by [Michael Kelley]
v4:
PATCH1: use variable name nopvspin instead of pvspin and
defined it as __initdata, changed print message,
updated patch description [Sean Christopherson]
PATCH2: remove Suggested-by, use "kvm-guest:" prefix [Sean Christopherson]
PATCH3: make variable nopvsin and xen_pvspin coexist
remove Reviewed-by due to code change [Sean Christopherson]
PATCH4: make variable nopvsin and hv_pvspin coexist [Sean Christopherson]
v3:
PATCH2: Fix indentation
v2:
PATCH1: pick the print code change into separate PATCH2,
updated patch description [Vitaly Kuznetsov]
PATCH2: new patch with print code change [Vitaly Kuznetsov]
PATCH3: add Reviewed-by [Juergen Gross]
Zhenzhong Duan (5):
Revert "KVM: X86: Fix setup the virt_spin_lock_key before static key
get initialized"
x86/kvm: Change print code to use pr_*() format
x86/kvm: Add "nopvspin" parameter to disable PV spinlocks
xen: Mark "xen_nopvspin" parameter obsolete
x86/hyperv: Mark "hv_nopvspin" parameter obsolete
Documentation/admin-guide/kernel-parameters.txt | 14 ++++-
arch/x86/hyperv/hv_spinlock.c | 4 ++
arch/x86/include/asm/qspinlock.h | 1 +
arch/x86/kernel/kvm.c | 76 ++++++++++++++++---------
arch/x86/xen/spinlock.c | 4 +-
kernel/locking/qspinlock.c | 7 +++
6 files changed, 75 insertions(+), 31 deletions(-)
--
1.8.3.1
pr_*() is preferred than printk(KERN_* ...), after change all the print
in arch/x86/kernel/kvm.c will have "kvm_guest: xxx" style.
No functional change.
Signed-off-by: Zhenzhong Duan <[email protected]>
Reviewed-by: Vitaly Kuznetsov <[email protected]>
Cc: Paolo Bonzini <[email protected]>
Cc: Radim Krcmar <[email protected]>
Cc: Sean Christopherson <[email protected]>
Cc: Vitaly Kuznetsov <[email protected]>
Cc: Wanpeng Li <[email protected]>
Cc: Jim Mattson <[email protected]>
Cc: Joerg Roedel <[email protected]>
Cc: Thomas Gleixner <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Borislav Petkov <[email protected]>
Cc: "H. Peter Anvin" <[email protected]>
---
arch/x86/kernel/kvm.c | 30 ++++++++++++++++--------------
1 file changed, 16 insertions(+), 14 deletions(-)
diff --git a/arch/x86/kernel/kvm.c b/arch/x86/kernel/kvm.c
index 3bc6a266..249f14a 100644
--- a/arch/x86/kernel/kvm.c
+++ b/arch/x86/kernel/kvm.c
@@ -7,6 +7,8 @@
* Authors: Anthony Liguori <[email protected]>
*/
+#define pr_fmt(fmt) "kvm_guest: " fmt
+
#include <linux/context_tracking.h>
#include <linux/init.h>
#include <linux/kernel.h>
@@ -286,8 +288,8 @@ static void kvm_register_steal_time(void)
return;
wrmsrl(MSR_KVM_STEAL_TIME, (slow_virt_to_phys(st) | KVM_MSR_ENABLED));
- pr_info("kvm-stealtime: cpu %d, msr %llx\n",
- cpu, (unsigned long long) slow_virt_to_phys(st));
+ pr_info("stealtime: cpu %d, msr %llx\n", cpu,
+ (unsigned long long) slow_virt_to_phys(st));
}
static DEFINE_PER_CPU_DECRYPTED(unsigned long, kvm_apic_eoi) = KVM_PV_EOI_DISABLED;
@@ -321,8 +323,7 @@ static void kvm_guest_cpu_init(void)
wrmsrl(MSR_KVM_ASYNC_PF_EN, pa);
__this_cpu_write(apf_reason.enabled, 1);
- printk(KERN_INFO"KVM setup async PF for cpu %d\n",
- smp_processor_id());
+ pr_info("setup async PF for cpu %d\n", smp_processor_id());
}
if (kvm_para_has_feature(KVM_FEATURE_PV_EOI)) {
@@ -347,8 +348,7 @@ static void kvm_pv_disable_apf(void)
wrmsrl(MSR_KVM_ASYNC_PF_EN, 0);
__this_cpu_write(apf_reason.enabled, 0);
- printk(KERN_INFO"Unregister pv shared memory for cpu %d\n",
- smp_processor_id());
+ pr_info("Unregister PV shared memory for cpu %d\n", smp_processor_id());
}
static void kvm_pv_guest_cpu_reboot(void *unused)
@@ -469,7 +469,8 @@ static void __send_ipi_mask(const struct cpumask *mask, int vector)
} else {
ret = kvm_hypercall4(KVM_HC_SEND_IPI, (unsigned long)ipi_bitmap,
(unsigned long)(ipi_bitmap >> BITS_PER_LONG), min, icr);
- WARN_ONCE(ret < 0, "KVM: failed to send PV IPI: %ld", ret);
+ WARN_ONCE(ret < 0, "kvm_guest: failed to send PV IPI: %ld",
+ ret);
min = max = apic_id;
ipi_bitmap = 0;
}
@@ -479,7 +480,8 @@ static void __send_ipi_mask(const struct cpumask *mask, int vector)
if (ipi_bitmap) {
ret = kvm_hypercall4(KVM_HC_SEND_IPI, (unsigned long)ipi_bitmap,
(unsigned long)(ipi_bitmap >> BITS_PER_LONG), min, icr);
- WARN_ONCE(ret < 0, "KVM: failed to send PV IPI: %ld", ret);
+ WARN_ONCE(ret < 0, "kvm_guest: failed to send PV IPI: %ld",
+ ret);
}
local_irq_restore(flags);
@@ -509,7 +511,7 @@ static void kvm_setup_pv_ipi(void)
{
apic->send_IPI_mask = kvm_send_ipi_mask;
apic->send_IPI_mask_allbutself = kvm_send_ipi_mask_allbutself;
- pr_info("KVM setup pv IPIs\n");
+ pr_info("setup PV IPIs\n");
}
static void kvm_smp_send_call_func_ipi(const struct cpumask *mask)
@@ -631,11 +633,11 @@ static void __init kvm_guest_init(void)
!kvm_para_has_hint(KVM_HINTS_REALTIME) &&
kvm_para_has_feature(KVM_FEATURE_STEAL_TIME)) {
smp_ops.send_call_func_ipi = kvm_smp_send_call_func_ipi;
- pr_info("KVM setup pv sched yield\n");
+ pr_info("setup PV sched yield\n");
}
if (cpuhp_setup_state_nocalls(CPUHP_AP_ONLINE_DYN, "x86/kvm:online",
kvm_cpu_online, kvm_cpu_down_prepare) < 0)
- pr_err("kvm_guest: Failed to install cpu hotplug callbacks\n");
+ pr_err("failed to install cpu hotplug callbacks\n");
#else
sev_map_percpu_data();
kvm_guest_cpu_init();
@@ -738,7 +740,7 @@ static __init int kvm_setup_pv_tlb_flush(void)
zalloc_cpumask_var_node(per_cpu_ptr(&__pv_tlb_mask, cpu),
GFP_KERNEL, cpu_to_node(cpu));
}
- pr_info("KVM setup pv remote TLB flush\n");
+ pr_info("setup PV remote TLB flush\n");
}
return 0;
@@ -866,8 +868,8 @@ static void kvm_enable_host_haltpoll(void *i)
void arch_haltpoll_enable(unsigned int cpu)
{
if (!kvm_para_has_feature(KVM_FEATURE_POLL_CONTROL)) {
- pr_err_once("kvm: host does not support poll control\n");
- pr_err_once("kvm: host upgrade recommended\n");
+ pr_err_once("host does not support poll control\n");
+ pr_err_once("host upgrade recommended\n");
return;
}
--
1.8.3.1
Map "xen_nopvspin" to "nopvspin", fix stale description of "xen_nopvspin"
as we use qspinlock now.
Signed-off-by: Zhenzhong Duan <[email protected]>
Reviewed-by: Boris Ostrovsky <[email protected]>
Cc: Jonathan Corbet <[email protected]>
Cc: Boris Ostrovsky <[email protected]>
Cc: Juergen Gross <[email protected]>
Cc: Stefano Stabellini <[email protected]>
Cc: Thomas Gleixner <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Borislav Petkov <[email protected]>
Cc: "H. Peter Anvin" <[email protected]>
---
Documentation/admin-guide/kernel-parameters.txt | 7 ++++---
arch/x86/xen/spinlock.c | 4 ++--
2 files changed, 6 insertions(+), 5 deletions(-)
diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
index bd49ed2..85059dd 100644
--- a/Documentation/admin-guide/kernel-parameters.txt
+++ b/Documentation/admin-guide/kernel-parameters.txt
@@ -5307,8 +5307,9 @@
panic() code such as dumping handler.
xen_nopvspin [X86,XEN]
- Disables the ticketlock slowpath using Xen PV
- optimizations.
+ Disables the qspinlock slowpath using Xen PV optimizations.
+ This parameter is obsoleted by "nopvspin" parameter, which
+ has equivalent effect for XEN platform.
xen_nopv [X86]
Disables the PV optimizations forcing the HVM guest to
@@ -5334,7 +5335,7 @@
as generic guest with no PV drivers. Currently support
XEN HVM, KVM, HYPER_V and VMWARE guest.
- nopvspin [X86,KVM]
+ nopvspin [X86,XEN,KVM]
Disables the qspinlock slow path using PV optimizations
which allow the hypervisor to 'idle' the guest on lock
contention.
diff --git a/arch/x86/xen/spinlock.c b/arch/x86/xen/spinlock.c
index 6deb490..799f4eb 100644
--- a/arch/x86/xen/spinlock.c
+++ b/arch/x86/xen/spinlock.c
@@ -114,9 +114,8 @@ void xen_uninit_lock_cpu(int cpu)
*/
void __init xen_init_spinlocks(void)
{
-
/* Don't need to use pvqspinlock code if there is only 1 vCPU. */
- if (num_possible_cpus() == 1)
+ if (num_possible_cpus() == 1 || nopvspin)
xen_pvspin = false;
if (!xen_pvspin) {
@@ -137,6 +136,7 @@ void __init xen_init_spinlocks(void)
static __init int xen_parse_nopvspin(char *arg)
{
+ pr_notice("\"xen_nopvspin\" is deprecated, please use \"nopvspin\" instead\n");
xen_pvspin = false;
return 0;
}
--
1.8.3.1
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 <[email protected]>
Cc: Jonathan Corbet <[email protected]>
Cc: Thomas Gleixner <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Borislav Petkov <[email protected]>
Cc: "H. Peter Anvin" <[email protected]>
Cc: Paolo Bonzini <[email protected]>
Cc: Radim Krcmar <[email protected]>
Cc: Sean Christopherson <[email protected]>
Cc: Vitaly Kuznetsov <[email protected]>
Cc: Wanpeng Li <[email protected]>
Cc: Jim Mattson <[email protected]>
Cc: Joerg Roedel <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Will Deacon <[email protected]>
---
Documentation/admin-guide/kernel-parameters.txt | 5 ++++
arch/x86/include/asm/qspinlock.h | 1 +
arch/x86/kernel/kvm.c | 34 ++++++++++++++++++++++---
kernel/locking/qspinlock.c | 7 +++++
4 files changed, 43 insertions(+), 4 deletions(-)
diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
index a84a83f..bd49ed2 100644
--- a/Documentation/admin-guide/kernel-parameters.txt
+++ b/Documentation/admin-guide/kernel-parameters.txt
@@ -5334,6 +5334,11 @@
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 249f14a..e9c76d8 100644
--- a/arch/x86/kernel/kvm.c
+++ b/arch/x86/kernel/kvm.c
@@ -825,18 +825,44 @@ __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))
+ /*
+ * PV spinlocks is disabled if no host side support, then native
+ * qspinlock will be used. As native qspinlock is a fair lock, there is
+ * lock holder preemption issue using it in a guest, imaging one pCPU
+ * running 10 vCPUs of same guest contending same lock.
+ *
+ * virt_spin_lock() is introduced as an optimization for that scenario
+ * which is enabled by virt_spin_lock_key key. To use that optimization,
+ * virt_spin_lock_key isn't disabled here.
+ */
+ if (!kvm_para_has_feature(KVM_FEATURE_PV_UNHALT)) {
+ pr_info("PV spinlocks disabled, no host support.\n");
return;
+ }
+ /*
+ * Disable PV qspinlock and use native qspinlock when dedicated pCPUs
+ * are available.
+ */
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;
+ }
+
+ if (num_possible_cpus() == 1) {
+ pr_info("PV spinlocks disabled, single CPU.\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
--
1.8.3.1
Map "hv_nopvspin" to "nopvspin".
Signed-off-by: Zhenzhong Duan <[email protected]>
Reviewed-by: Michael Kelley <[email protected]>
Cc: Jonathan Corbet <[email protected]>
Cc: "K. Y. Srinivasan" <[email protected]>
Cc: Haiyang Zhang <[email protected]>
Cc: Stephen Hemminger <[email protected]>
Cc: Sasha Levin <[email protected]>
Cc: Thomas Gleixner <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Borislav Petkov <[email protected]>
Cc: "H. Peter Anvin" <[email protected]>
---
Documentation/admin-guide/kernel-parameters.txt | 6 +++++-
arch/x86/hyperv/hv_spinlock.c | 4 ++++
2 files changed, 9 insertions(+), 1 deletion(-)
diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
index 85059dd..78648bb 100644
--- a/Documentation/admin-guide/kernel-parameters.txt
+++ b/Documentation/admin-guide/kernel-parameters.txt
@@ -1436,6 +1436,10 @@
hv_nopvspin [X86,HYPER_V] Disables the paravirt spinlock optimizations
which allow the hypervisor to 'idle' the
guest on lock contention.
+ This parameter is obsoleted by "nopvspin"
+ parameter, which has equivalent effect for
+ HYPER_V platform.
+
keep_bootcon [KNL]
Do not unregister boot console at start. This is only
@@ -5335,7 +5339,7 @@
as generic guest with no PV drivers. Currently support
XEN HVM, KVM, HYPER_V and VMWARE guest.
- nopvspin [X86,XEN,KVM]
+ nopvspin [X86,XEN,KVM,HYPER_V]
Disables the qspinlock slow path using PV optimizations
which allow the hypervisor to 'idle' the guest on lock
contention.
diff --git a/arch/x86/hyperv/hv_spinlock.c b/arch/x86/hyperv/hv_spinlock.c
index 07f21a0..47c7d6c 100644
--- a/arch/x86/hyperv/hv_spinlock.c
+++ b/arch/x86/hyperv/hv_spinlock.c
@@ -64,6 +64,9 @@ __visible bool hv_vcpu_is_preempted(int vcpu)
void __init hv_init_spinlocks(void)
{
+ if (nopvspin)
+ hv_pvspin = false;
+
if (!hv_pvspin || !apic ||
!(ms_hyperv.hints & HV_X64_CLUSTER_IPI_RECOMMENDED) ||
!(ms_hyperv.features & HV_X64_MSR_GUEST_IDLE_AVAILABLE)) {
@@ -82,6 +85,7 @@ void __init hv_init_spinlocks(void)
static __init int hv_parse_nopvspin(char *arg)
{
+ pr_notice("\"hv_nopvspin\" is deprecated, please use \"nopvspin\" instead\n");
hv_pvspin = false;
return 0;
}
--
1.8.3.1
This reverts commit 34226b6b70980a8f81fff3c09a2c889f77edeeff.
Commit 8990cac6e5ea ("x86/jump_label: Initialize static branching
early") adds jump_label_init() call in setup_arch() to make static
keys initialized early, so we could use the original simpler code
again.
The similar change for XEN is in commit 090d54bcbc54 ("Revert
"x86/paravirt: Set up the virt_spin_lock_key after static keys get
initialized"")
Signed-off-by: Zhenzhong Duan <[email protected]>
Reviewed-by: Vitaly Kuznetsov <[email protected]>
Cc: Paolo Bonzini <[email protected]>
Cc: Radim Krcmar <[email protected]>
Cc: Sean Christopherson <[email protected]>
Cc: Vitaly Kuznetsov <[email protected]>
Cc: Wanpeng Li <[email protected]>
Cc: Jim Mattson <[email protected]>
Cc: Joerg Roedel <[email protected]>
Cc: Thomas Gleixner <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Borislav Petkov <[email protected]>
Cc: "H. Peter Anvin" <[email protected]>
---
arch/x86/kernel/kvm.c | 12 +++---------
1 file changed, 3 insertions(+), 9 deletions(-)
diff --git a/arch/x86/kernel/kvm.c b/arch/x86/kernel/kvm.c
index e820568..3bc6a266 100644
--- a/arch/x86/kernel/kvm.c
+++ b/arch/x86/kernel/kvm.c
@@ -527,13 +527,6 @@ static void kvm_smp_send_call_func_ipi(const struct cpumask *mask)
}
}
-static void __init kvm_smp_prepare_cpus(unsigned int max_cpus)
-{
- native_smp_prepare_cpus(max_cpus);
- if (kvm_para_has_hint(KVM_HINTS_REALTIME))
- static_branch_disable(&virt_spin_lock_key);
-}
-
static void __init kvm_smp_prepare_boot_cpu(void)
{
/*
@@ -633,7 +626,6 @@ static void __init kvm_guest_init(void)
apic_set_eoi_write(kvm_guest_apic_eoi_write);
#ifdef CONFIG_SMP
- smp_ops.smp_prepare_cpus = kvm_smp_prepare_cpus;
smp_ops.smp_prepare_boot_cpu = kvm_smp_prepare_boot_cpu;
if (kvm_para_has_feature(KVM_FEATURE_PV_SCHED_YIELD) &&
!kvm_para_has_hint(KVM_HINTS_REALTIME) &&
@@ -835,8 +827,10 @@ void __init kvm_spinlock_init(void)
if (!kvm_para_has_feature(KVM_FEATURE_PV_UNHALT))
return;
- if (kvm_para_has_hint(KVM_HINTS_REALTIME))
+ if (kvm_para_has_hint(KVM_HINTS_REALTIME)) {
+ 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)
--
1.8.3.1
Hi vitaly
This patch is based on your suggestion on v5, appreciate your further
review:) Thanks
Zhenzhong
On 2019/10/15 9:19, Zhenzhong Duan wrote:
> 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 <[email protected]>
> Cc: Jonathan Corbet <[email protected]>
> Cc: Thomas Gleixner <[email protected]>
> Cc: Ingo Molnar <[email protected]>
> Cc: Borislav Petkov <[email protected]>
> Cc: "H. Peter Anvin" <[email protected]>
> Cc: Paolo Bonzini <[email protected]>
> Cc: Radim Krcmar <[email protected]>
> Cc: Sean Christopherson <[email protected]>
> Cc: Vitaly Kuznetsov <[email protected]>
> Cc: Wanpeng Li <[email protected]>
> Cc: Jim Mattson <[email protected]>
> Cc: Joerg Roedel <[email protected]>
> Cc: Peter Zijlstra <[email protected]>
> Cc: Will Deacon <[email protected]>
> ---
> Documentation/admin-guide/kernel-parameters.txt | 5 ++++
> arch/x86/include/asm/qspinlock.h | 1 +
> arch/x86/kernel/kvm.c | 34 ++++++++++++++++++++++---
> kernel/locking/qspinlock.c | 7 +++++
> 4 files changed, 43 insertions(+), 4 deletions(-)
>
> diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
> index a84a83f..bd49ed2 100644
> --- a/Documentation/admin-guide/kernel-parameters.txt
> +++ b/Documentation/admin-guide/kernel-parameters.txt
> @@ -5334,6 +5334,11 @@
> 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 249f14a..e9c76d8 100644
> --- a/arch/x86/kernel/kvm.c
> +++ b/arch/x86/kernel/kvm.c
> @@ -825,18 +825,44 @@ __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))
> + /*
> + * PV spinlocks is disabled if no host side support, then native
> + * qspinlock will be used. As native qspinlock is a fair lock, there is
> + * lock holder preemption issue using it in a guest, imaging one pCPU
> + * running 10 vCPUs of same guest contending same lock.
> + *
> + * virt_spin_lock() is introduced as an optimization for that scenario
> + * which is enabled by virt_spin_lock_key key. To use that optimization,
> + * virt_spin_lock_key isn't disabled here.
> + */
> + if (!kvm_para_has_feature(KVM_FEATURE_PV_UNHALT)) {
> + pr_info("PV spinlocks disabled, no host support.\n");
> return;
> + }
>
> + /*
> + * Disable PV qspinlock and use native qspinlock when dedicated pCPUs
> + * are available.
> + */
> 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;
> + }
> +
> + if (num_possible_cpus() == 1) {
> + pr_info("PV spinlocks disabled, single CPU.\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
Zhenzhong Duan <[email protected]> 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 <[email protected]>
> Cc: Jonathan Corbet <[email protected]>
> Cc: Thomas Gleixner <[email protected]>
> Cc: Ingo Molnar <[email protected]>
> Cc: Borislav Petkov <[email protected]>
> Cc: "H. Peter Anvin" <[email protected]>
> Cc: Paolo Bonzini <[email protected]>
> Cc: Radim Krcmar <[email protected]>
> Cc: Sean Christopherson <[email protected]>
> Cc: Vitaly Kuznetsov <[email protected]>
> Cc: Wanpeng Li <[email protected]>
> Cc: Jim Mattson <[email protected]>
> Cc: Joerg Roedel <[email protected]>
> Cc: Peter Zijlstra <[email protected]>
> Cc: Will Deacon <[email protected]>
> ---
> Documentation/admin-guide/kernel-parameters.txt | 5 ++++
> arch/x86/include/asm/qspinlock.h | 1 +
> arch/x86/kernel/kvm.c | 34 ++++++++++++++++++++++---
> kernel/locking/qspinlock.c | 7 +++++
> 4 files changed, 43 insertions(+), 4 deletions(-)
>
> diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
> index a84a83f..bd49ed2 100644
> --- a/Documentation/admin-guide/kernel-parameters.txt
> +++ b/Documentation/admin-guide/kernel-parameters.txt
> @@ -5334,6 +5334,11 @@
> 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 249f14a..e9c76d8 100644
> --- a/arch/x86/kernel/kvm.c
> +++ b/arch/x86/kernel/kvm.c
> @@ -825,18 +825,44 @@ __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))
> + /*
> + * PV spinlocks is disabled if no host side support, then native
> + * qspinlock will be used. As native qspinlock is a fair lock, there is
> + * lock holder preemption issue using it in a guest, imaging one pCPU
> + * running 10 vCPUs of same guest contending same lock.
> + *
> + * virt_spin_lock() is introduced as an optimization for that scenario
> + * which is enabled by virt_spin_lock_key key. To use that optimization,
> + * virt_spin_lock_key isn't disabled here.
> + */
My take (if I properly understood what you say) would be:
"In case host doesn't support KVM_FEATURE_PV_UNHALT there is still an
advantage of keeping virt_spin_lock_key enabled: virt_spin_lock() is
preferred over native qspinlock when vCPU is preempted."
> + if (!kvm_para_has_feature(KVM_FEATURE_PV_UNHALT)) {
> + pr_info("PV spinlocks disabled, no host support.\n");
> return;
> + }
>
> + /*
> + * Disable PV qspinlock and use native qspinlock when dedicated pCPUs
> + * are available.
> + */
> 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;
> + }
> +
> + if (num_possible_cpus() == 1) {
> + pr_info("PV spinlocks disabled, single CPU.\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;
You could've replaced this 'static_branch_disable(); return;' pattern
with a goto to the end of the function to save a few lines but this
looks good anyways.
Reviewed-by: Vitaly Kuznetsov <[email protected]>
> + }
> +
> + 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
--
Vitaly
On 2019/10/21 19:14, Vitaly Kuznetsov wrote:
>> index 249f14a..e9c76d8 100644
>> --- a/arch/x86/kernel/kvm.c
>> +++ b/arch/x86/kernel/kvm.c
>> @@ -825,18 +825,44 @@ __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))
>> + /*
>> + * PV spinlocks is disabled if no host side support, then native
>> + * qspinlock will be used. As native qspinlock is a fair lock, there is
>> + * lock holder preemption issue using it in a guest, imaging one pCPU
>> + * running 10 vCPUs of same guest contending same lock.
>> + *
>> + * virt_spin_lock() is introduced as an optimization for that scenario
>> + * which is enabled by virt_spin_lock_key key. To use that optimization,
>> + * virt_spin_lock_key isn't disabled here.
>> + */
> My take (if I properly understood what you say) would be:
>
> "In case host doesn't support KVM_FEATURE_PV_UNHALT there is still an
> advantage of keeping virt_spin_lock_key enabled: virt_spin_lock() is
> preferred over native qspinlock when vCPU is preempted."
Yes, that's what I mean, maybe I didn't explain clearly due to my pool
english,
I'll use your explanation instead.
>
>> + if (!kvm_para_has_feature(KVM_FEATURE_PV_UNHALT)) {
>> + pr_info("PV spinlocks disabled, no host support.\n");
>> return;
>> + }
>>
>> + /*
>> + * Disable PV qspinlock and use native qspinlock when dedicated pCPUs
>> + * are available.
>> + */
>> 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;
>> + }
>> +
>> + if (num_possible_cpus() == 1) {
>> + pr_info("PV spinlocks disabled, single CPU.\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;
> You could've replaced this 'static_branch_disable(); return;' pattern
> with a goto to the end of the function to save a few lines but this
> looks good anyways.
>
> Reviewed-by: Vitaly Kuznetsov<[email protected]>
Ok, will do, thanks for review.
Zhenzhong