perf/x86: KVM: Disable vPMU on hybrid CPUs
Subsystems (4)
Sean Christopherson
maintainerv2 cover perf/x86: KVM: Disable vPMU on hybrid CPUs
Disable vPMU support in KVM when running on hybrid CPUs to avoid inducing #GPs and other issues in guests. This is intended to be a stopgap to prevent unwitting KVM users from shooting themselves in the foot until KVM KVM gets proper enabling for hybrid CPUs.
Effectively squash exporting PMU capabilities from perf until it too gets proper enabling.
v2:
- Disable vPMU on hybrid CPUs in KVM _and_ in perf. [Like]
- Use X86_FEATURE_HYBRID_CPU instead of is_hybrid().
- Tweak comments/changelogs to more clearly state that there are options beyond pinning vCPUs (though they still need KVM support). [Like]
v1: https://lore.kernel.org/all/[email protected]
Sean Christopherson (2):
KVM: x86/pmu: Disable vPMU support on hybrid CPUs (host PMUs) perf/x86: Refuse to export capabilities for hybrid PMUs
arch/x86/events/core.c | 14 ++++++++------ arch/x86/kvm/pmu.h | 26 +++++++++++++++++++------- 2 files changed, 27 insertions(+), 13 deletions(-)
base-commit: 7cb79f433e75b05d1635aefaa851cfcd1cb7dc4f
Signature · 1 line
2.39.1.519.gcb327c4b5f-goog
Sean Christopherson
maintainerPATCH v2 1/2 KVM: x86/pmu: Disable vPMU support on hybrid CPUs (host PMUs)
Disable KVM support for virtualizing PMUs on hosts with hybrid PMUs until KVM gains a sane way to enumeration the hybrid vPMU to userspace and/or gains a mechanism to let userspace opt-in to the dangers of exposing a hybrid vPMU to KVM guests. Virtualizing a hybrid PMU, or at least part of a hybrid PMU, is possible, but it requires careful, deliberate configuration from userspace.
E.g. to expose full functionality, vCPUs need to be pinned to pCPUs to prevent migrating a vCPU between a big core and a little core, userspace must enumerate a reasonable topology to the guest, and guest CPUID must be curated per vCPU to enumerate accurate vPMU capabilities.
The last point is especially problematic, as KVM doesn't control which pCPU it runs on when enumerating KVM's vPMU capabilities to userspace, i.e. userspace can't rely on KVM_GET_SUPPORTED_CPUID in it's current form.
Alternatively, userspace could enable vPMU support by enumerating the set of features that are common and coherent across all cores, e.g. by filtering PMU events and restricting guest capabilities. But again, that requires userspace to take action far beyond reflecting KVM's supported feature set into the guest.
For now, simply disable vPMU support on hybrid CPUs to avoid inducing seemingly random #GPs in guests, and punt support for hybrid CPUs to a future enabling effort.
--- arch/x86/kvm/pmu.h | 26 +++++++++++++++++++------- 1 file changed, 19 insertions(+), 7 deletions(-)
arch/x86/kvm/pmu.h +19 -7
@@ -165,15 +165,27 @@ static inline void kvm_init_pmu_capability(void) {bool is_intel = boot_cpu_data.x86_vendor ==X86_VENDOR_INTEL ; -perf_get_x86_pmu_capability (&kvm_pmu_cap); - -/* -* For Intel, only support guest architectural pmu -* on a host with architectural pmu. -*/ -if ((is_intel && !kvm_pmu_cap.version) || !kvm_pmu_cap.num_counters_gp) +/* +* Hybrid PMUs don't play nice with virtualization without careful +* configuration by userspace, and KVM's APIs for reporting supported +* vPMU features do not account for hybrid PMUs. Disable vPMU support +* for hybrid PMUs until KVM gains a way to let userspace opt-in. +*/ +if (cpu_feature_enabled (X86_FEATURE_HYBRID_CPU )) enable_pmu = false; +if (enable_pmu) { +perf_get_x86_pmu_capability (&kvm_pmu_cap); + +/* +* For Intel, only support guest architectural pmu +* on a host with architectural pmu. +*/ +if ((is_intel && !kvm_pmu_cap.version) || + !kvm_pmu_cap.num_counters_gp) + enable_pmu = false; + } +if (!enable_pmu) {memset (&kvm_pmu_cap,0 ,sizeof (kvm_pmu_cap));return ;
Signature · 1 line
2.39.1.519.gcb327c4b5f-goog
Sean Christopherson
maintainerPATCH v2 2/2 perf/x86: Refuse to export capabilities for hybrid PMUs
In reply to Sean ChristophersonNow that KVM disables vPMU support on hybrid CPUs, WARN and return zeros if perf_get_x86_pmu_capability() is invoked on a hybrid CPU. The helper doesn't provide an accurate accounting of the PMU capabilities for hybrid CPUs and needs to be enhanced if KVM, or anything else outside of perf, wants to act on the PMU capabilities.
--- arch/x86/events/core.c | 14 ++++++++------ 1 file changed, 8 insertions(+), 6 deletions(-)
arch/x86/events/core.c +8 -6
@@ -2974,17 +2974,19 @@ unsigned long perf_misc_flags(struct pt_regs *regs)void perf_get_x86_pmu_capability (struct x86_pmu_capability *cap) { -if (!x86_pmu_initialized ()) { +/* This API doesn't currently support enumerating hybrid PMUs. */ +if (WARN_ON_ONCE (cpu_feature_enabled (X86_FEATURE_HYBRID_CPU )) || + !x86_pmu_initialized ()) {memset (cap,0 ,sizeof (*cap));return ; } +/* +* Note, hybrid CPU models get tracked as having hybrid PMUs even when +* all E-cores are disabled via BIOS. When E-cores are disabled, the +* base PMU holds the correct number of counters for P-cores. +*/ cap->version = x86_pmu.version; -/* -* KVM doesn't support the hybrid PMU yet. -* Return the common value in global x86_pmu, -* which available for all cores. -*/ cap->num_counters_gp = x86_pmu.num_counters; cap->num_counters_fixed = x86_pmu.num_counters_fixed; cap->bit_width_gp = x86_pmu.cntval_bits;
Signature · 1 line
2.39.1.519.gcb327c4b5f-goog
Queued, thanks.
Paolo