Struct ProgRuntimeStats 

pub struct ProgRuntimeStats {
    pub name: String,
    pub cnt: u64,
    pub nsecs: u64,
    pub misses: u64,
}

Per-program runtime stats summed across all CPUs.

Mirrors the kernel’s struct bpf_prog_stats (include/linux/filter.h): cnt (invocations), nsecs (cumulative runtime), misses (recursion re-entries skipped via bpf_prog_inc_misses_counter, kernel/bpf/syscall.c). All three counters are u64 monotonics summed across the program’s per-CPU bpf_prog_stats slots.

Fields

name: String

Program name as registered with the kernel.

cnt: u64

Total invocation count across all CPUs.

nsecs: u64

Total CPU time in nanoseconds across all CPUs.

misses: u64

Total recursion misses across all CPUs. A miss is a re-entry attempt blocked by the program’s per-CPU recursion guard.

Implementations

impl ProgRuntimeStats

pub fn ns_per_call(&self) -> f64

Mean nanoseconds per invocation: nsecs / cnt. Returns 0.0 when cnt == 0 (program never ran or counter not running) so the result never propagates NaN / Infinity into downstream finite_or_zero filters. Method-only access (no stored shadow) — recomputed every call from the raw fields, matching the super::super::assert::CgroupStats::wake_latency_tail_ratio derived-ratio convention.

Unitless-from-bpftop’s perspective: bpftop-style triage reads “ns/call” as the primary cost-per-invocation metric; surfacing it here lets a failure-dump consumer compare two programs’ per-call cost without dividing the wire counters manually.

pub fn miss_rate(&self) -> f64

Fraction of invocation attempts blocked by the per-CPU recursion guard: misses / (cnt + misses). Returns 0.0 when both counters are zero (no signal); never produces NaN / Infinity even on a saturated cnt + misses overflow because saturating_add floors at u64::MAX and the resulting denominator is non-zero.

A non-trivial miss rate signals lock contention or a misconfigured recursion guard — bpftop-style triage flags any program with miss_rate > 0.01 as a hot recursion path. Method-only access (no stored shadow); the wire format carries cnt and misses separately so consumers who want the raw counts can recover them.

Trait Implementations

impl Clone for ProgRuntimeStats

fn clone(&self) -> ProgRuntimeStats

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for ProgRuntimeStats

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for ProgRuntimeStats

fn default() -> ProgRuntimeStats

Returns the “default value” for a type.

impl<'de> Deserialize<'de> for ProgRuntimeStats

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Display for ProgRuntimeStats

fn fmt(&self, f: &mut Formatter<'_>) -> Result

One-line summary used by super::dump::FailureDumpReport’s human-readable rendering: name + the three counter sums plus the bpftop-style derived metrics (ns/call, miss-rate fraction). Derived metrics elide when their guards fire (cnt==0 or cnt+misses==0) so a program that never ran renders without misleading “0.000 ns/call” noise.

impl Serialize for ProgRuntimeStats

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.