powerpc/kvm: Add vpa latency counters to kvm_vcpu_arch

Commit e1f288d2f9 ("KVM: PPC: Book3S HV nestedv2: Add support
for reading VPA counters for pseries guests") introduced support for new
Virtual Process Area(VPA) based software counters. These counters are
useful when observing context switch latency of L1 <-> L2. It also
added access to counters in lppaca, which is good enough to understand
latency details per-cpu level. But to extend and aggregate
per-process level(qemu) or per-pid/tid level(vcpu), these
counters also needs to be added as part of kvm_vcpu_arch struct.
Additional code added to update these new kvm_vcpu_arch variables
in do_trace_nested_cs_time function.

Signed-off-by: Kajol Jain <kjain@linux.ibm.com>
Co-developed-by: Madhavan Srinivasan <maddy@linux.ibm.com>
Signed-off-by: Madhavan Srinivasan <maddy@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://patch.msgid.link/20241118114114.208964-3-kjain@linux.ibm.com
This commit is contained in:
Kajol Jain 2024-11-18 17:11:13 +05:30 committed by Michael Ellerman
parent 4ae0b32ece
commit 5f0b48c6a1
2 changed files with 8 additions and 0 deletions

View File

@ -871,6 +871,11 @@ struct kvm_vcpu_arch {
struct kvmhv_tb_accumulator cede_time; /* time napping inside guest */
#endif
#endif /* CONFIG_KVM_BOOK3S_HV_EXIT_TIMING */
#ifdef CONFIG_KVM_BOOK3S_HV_POSSIBLE
u64 l1_to_l2_cs;
u64 l2_to_l1_cs;
u64 l2_runtime_agg;
#endif
};
#define VCPU_FPR(vcpu, i) (vcpu)->arch.fp.fpr[i][TS_FPROFFSET]

View File

@ -4167,6 +4167,9 @@ static void do_trace_nested_cs_time(struct kvm_vcpu *vcpu)
*l1_to_l2_cs_ptr = l1_to_l2_ns;
*l2_to_l1_cs_ptr = l2_to_l1_ns;
*l2_runtime_agg_ptr = l2_runtime_ns;
vcpu->arch.l1_to_l2_cs = l1_to_l2_ns;
vcpu->arch.l2_to_l1_cs = l2_to_l1_ns;
vcpu->arch.l2_runtime_agg = l2_runtime_ns;
}
u64 kvmhv_get_l1_to_l2_cs_time(void)