[PATCH v7 14/20] KVM: arm64: Implement lazy PMU context swaps
Colton Lewis
coltonlewis at google.com
Mon May 4 14:18:07 PDT 2026
Since many guests will never touch the PMU, they need not pay the cost
of context swapping those registers.
Use an enum to implement a simple state machine for PMU register
access. The PMU is either free or guest owned. We only need to context
swap if the PMU registers are guest owned. The PMU initially starts as
free and only transitions to guest owned if a guest has touched the
PMU registers.
Signed-off-by: Colton Lewis <coltonlewis at google.com>
---
arch/arm64/include/asm/kvm_host.h | 1 +
arch/arm64/include/asm/kvm_types.h | 6 +++++-
arch/arm64/kvm/debug.c | 2 +-
arch/arm64/kvm/pmu-direct.c | 25 +++++++++++++++++++++++--
arch/arm64/kvm/sys_regs.c | 11 +++++++++++
include/kvm/arm_pmu.h | 5 +++++
6 files changed, 46 insertions(+), 4 deletions(-)
diff --git a/arch/arm64/include/asm/kvm_host.h b/arch/arm64/include/asm/kvm_host.h
index 1f789ba589d56..891433fe304ac 100644
--- a/arch/arm64/include/asm/kvm_host.h
+++ b/arch/arm64/include/asm/kvm_host.h
@@ -1404,6 +1404,7 @@ static inline bool kvm_system_needs_idmapped_vectors(void)
return cpus_have_final_cap(ARM64_SPECTRE_V3A);
}
+void kvm_arm_setup_mdcr_el2(struct kvm_vcpu *vcpu);
void kvm_init_host_debug_data(void);
void kvm_debug_init_vhe(void);
void kvm_vcpu_load_debug(struct kvm_vcpu *vcpu);
diff --git a/arch/arm64/include/asm/kvm_types.h b/arch/arm64/include/asm/kvm_types.h
index 9a126b9e2d7c9..4e39cbc80aa0b 100644
--- a/arch/arm64/include/asm/kvm_types.h
+++ b/arch/arm64/include/asm/kvm_types.h
@@ -4,5 +4,9 @@
#define KVM_ARCH_NR_OBJS_PER_MEMORY_CACHE 40
-#endif /* _ASM_ARM64_KVM_TYPES_H */
+enum vcpu_pmu_register_access {
+ VCPU_PMU_ACCESS_FREE,
+ VCPU_PMU_ACCESS_GUEST_OWNED,
+};
+#endif /* _ASM_ARM64_KVM_TYPES_H */
diff --git a/arch/arm64/kvm/debug.c b/arch/arm64/kvm/debug.c
index 0ab89c91e19cb..c2cf6b308ec60 100644
--- a/arch/arm64/kvm/debug.c
+++ b/arch/arm64/kvm/debug.c
@@ -34,7 +34,7 @@ static int cpu_has_spe(u64 dfr0)
* - Self-hosted Trace Filter controls (MDCR_EL2_TTRF)
* - Self-hosted Trace (MDCR_EL2_TTRF/MDCR_EL2_E2TB)
*/
-static void kvm_arm_setup_mdcr_el2(struct kvm_vcpu *vcpu)
+void kvm_arm_setup_mdcr_el2(struct kvm_vcpu *vcpu)
{
int hpmn = kvm_pmu_hpmn(vcpu);
diff --git a/arch/arm64/kvm/pmu-direct.c b/arch/arm64/kvm/pmu-direct.c
index 14cc419dbafad..881cea5117515 100644
--- a/arch/arm64/kvm/pmu-direct.c
+++ b/arch/arm64/kvm/pmu-direct.c
@@ -72,10 +72,29 @@ bool kvm_vcpu_pmu_use_fgt(struct kvm_vcpu *vcpu)
u8 hpmn = vcpu->kvm->arch.nr_pmu_counters;
return kvm_vcpu_pmu_is_partitioned(vcpu) &&
+ vcpu->arch.pmu.access == VCPU_PMU_ACCESS_GUEST_OWNED &&
cpus_have_final_cap(ARM64_HAS_FGT) &&
(hpmn != 0 || cpus_have_final_cap(ARM64_HAS_HPMN0));
}
+/**
+ * kvm_pmu_set_guest_owned()
+ * @vcpu: Pointer to vcpu struct
+ *
+ * Reconfigure the guest for physical access of PMU hardware if
+ * allowed. This means reconfiguring mdcr_el2 and loading the vCPU
+ * state onto hardware.
+ *
+ */
+void kvm_pmu_set_guest_owned(struct kvm_vcpu *vcpu)
+{
+ if (kvm_vcpu_pmu_is_partitioned(vcpu) &&
+ vcpu->arch.pmu.access == VCPU_PMU_ACCESS_FREE) {
+ vcpu->arch.pmu.access = VCPU_PMU_ACCESS_GUEST_OWNED;
+ kvm_arm_setup_mdcr_el2(vcpu);
+ }
+}
+
/**
* kvm_pmu_hpmn() - Calculate HPMN field value
* @vcpu: Pointer to struct kvm_vcpu
@@ -278,7 +297,8 @@ void kvm_pmu_load(struct kvm_vcpu *vcpu)
* If we aren't guest-owned then we know the guest isn't using
* the PMU anyway, so no need to bother with the swap.
*/
- if (!kvm_vcpu_pmu_is_partitioned(vcpu))
+ if (!kvm_vcpu_pmu_is_partitioned(vcpu) ||
+ vcpu->arch.pmu.access != VCPU_PMU_ACCESS_GUEST_OWNED)
return;
preempt_disable();
@@ -353,7 +373,8 @@ void kvm_pmu_put(struct kvm_vcpu *vcpu)
* accessing the PMU anyway, so no need to bother with the
* swap.
*/
- if (!kvm_vcpu_pmu_is_partitioned(vcpu))
+ if (!kvm_vcpu_pmu_is_partitioned(vcpu) ||
+ vcpu->arch.pmu.access != VCPU_PMU_ACCESS_GUEST_OWNED)
return;
preempt_disable();
diff --git a/arch/arm64/kvm/sys_regs.c b/arch/arm64/kvm/sys_regs.c
index cc3d1804ab200..e4840d93a769f 100644
--- a/arch/arm64/kvm/sys_regs.c
+++ b/arch/arm64/kvm/sys_regs.c
@@ -1035,6 +1035,7 @@ static bool pmu_access_event_counter_el0_disabled(struct kvm_vcpu *vcpu)
static void pmu_write_pmcr(struct kvm_vcpu *vcpu, u64 val)
{
if (kvm_vcpu_pmu_is_partitioned(vcpu)) {
+ kvm_pmu_set_guest_owned(vcpu);
write_sysreg(val, pmcr_el0);
return;
}
@@ -1086,6 +1087,7 @@ static bool access_pmcr(struct kvm_vcpu *vcpu, struct sys_reg_params *p,
static void pmu_write_pmselr(struct kvm_vcpu *vcpu, u64 val)
{
if (kvm_vcpu_pmu_is_partitioned(vcpu)) {
+ kvm_pmu_set_guest_owned(vcpu);
write_sysreg(val, pmselr_el0);
return;
}
@@ -1193,6 +1195,8 @@ static void pmu_write_evcntr(struct kvm_vcpu *vcpu, u64 val, u64 idx)
return;
}
+ kvm_pmu_set_guest_owned(vcpu);
+
if (idx == ARMV8_PMU_CYCLE_IDX) {
write_sysreg(val, pmccntr_el0);
return;
@@ -1281,6 +1285,7 @@ static void pmu_write_evtyper(struct kvm_vcpu *vcpu, u64 val, u64 idx)
u64 mask;
if (kvm_vcpu_pmu_is_partitioned(vcpu)) {
+ kvm_pmu_set_guest_owned(vcpu);
mask = kvm_pmu_evtyper_mask(vcpu->kvm);
__vcpu_assign_sys_reg(vcpu, PMEVTYPER0_EL0 + idx, val & mask);
return;
@@ -1350,6 +1355,8 @@ static int get_pmreg(struct kvm_vcpu *vcpu, const struct sys_reg_desc *r, u64 *v
static void pmu_write_pmcnten(struct kvm_vcpu *vcpu, u64 val, bool set)
{
if (kvm_vcpu_pmu_is_partitioned(vcpu)) {
+ kvm_pmu_set_guest_owned(vcpu);
+
if (set)
write_sysreg(val, pmcntenset_el0);
else
@@ -1398,6 +1405,8 @@ static bool access_pmcnten(struct kvm_vcpu *vcpu, struct sys_reg_params *p,
static void pmu_write_pminten(struct kvm_vcpu *vcpu, u64 val, bool set)
{
if (kvm_vcpu_pmu_is_partitioned(vcpu)) {
+ kvm_pmu_set_guest_owned(vcpu);
+
if (set)
write_sysreg(val, pmintenset_el1);
else
@@ -1453,6 +1462,8 @@ static bool access_pmovs(struct kvm_vcpu *vcpu, struct sys_reg_params *p,
return false;
if (p->is_write) {
+ kvm_pmu_set_guest_owned(vcpu);
+
if (r->CRm & 0x2)
/* accessing PMOVSSET_EL0 */
__vcpu_rmw_sys_reg(vcpu, PMOVSSET_EL0, |=, (p->regval & mask));
diff --git a/include/kvm/arm_pmu.h b/include/kvm/arm_pmu.h
index fa881dc5f5832..0de63cc48fef9 100644
--- a/include/kvm/arm_pmu.h
+++ b/include/kvm/arm_pmu.h
@@ -7,6 +7,7 @@
#ifndef __ASM_ARM_KVM_PMU_H
#define __ASM_ARM_KVM_PMU_H
+#include <linux/kvm_types.h>
#include <linux/perf_event.h>
#include <linux/perf/arm_pmuv3.h>
#include <linux/perf/arm_pmu.h>
@@ -40,6 +41,7 @@ struct kvm_pmu {
int irq_num;
bool created;
bool irq_level;
+ enum vcpu_pmu_register_access access;
};
struct arm_pmu_entry {
@@ -101,6 +103,8 @@ u64 kvm_pmu_guest_counter_mask(struct arm_pmu *pmu);
void kvm_pmu_load(struct kvm_vcpu *vcpu);
void kvm_pmu_put(struct kvm_vcpu *vcpu);
+void kvm_pmu_set_guest_owned(struct kvm_vcpu *vcpu);
+
#if !defined(__KVM_NVHE_HYPERVISOR__)
bool kvm_vcpu_pmu_is_partitioned(struct kvm_vcpu *vcpu);
bool kvm_vcpu_pmu_use_fgt(struct kvm_vcpu *vcpu);
@@ -173,6 +177,7 @@ static inline u8 kvm_pmu_hpmn(struct kvm_vcpu *vcpu)
}
static inline void kvm_pmu_load(struct kvm_vcpu *vcpu) {}
static inline void kvm_pmu_put(struct kvm_vcpu *vcpu) {}
+static inline void kvm_pmu_set_guest_owned(struct kvm_vcpu *vcpu) {}
static inline void kvm_pmu_set_counter_value(struct kvm_vcpu *vcpu,
u64 select_idx, u64 val) {}
static inline void kvm_pmu_set_counter_value_user(struct kvm_vcpu *vcpu,
--
2.54.0.545.g6539524ca2-goog
More information about the linux-arm-kernel
mailing list