CIS490/collectors/perf_qemu.py
Max Gorog dac03d2eff perf: emit per-episode lifecycle events; emit row even with empty agg
Validation on k-gamingcom (commit ac7b85f) showed perf enabled in
production but rows_perf=0 on every episode. Without lifecycle events
the failure mode is indistinguishable from "perf wasn't enabled" — §1
silent-downgrade. The events now surface the actual cause:

  - perf_unavailable     — binary missing OR launch failed (with reason)
  - perf_started         — perf is running (pid, events, interval)
  - perf_first_row       — first row written; counters_populated tells
                           whether any event was actually counted
  - perf_finished        — final tally (intervals_seen,
                           intervals_with_values)
  - perf_no_counters     — perf was alive but every interval came back
                           <not counted> (likely paranoid > 2 or PID
                           ownership mismatch)

`_flush()` now writes a row whenever an interval is observed, even
when every event was <not counted>. The all-None row is honest data
("perf observed this interval and counted nothing"), and the rows
become a count of observed intervals rather than a count of
successful measurements — distinct from rows_proc / rows_qmp which
do count successful measurements. Trainers filter on
`cycles is not None` etc. when they need only populated rows.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-03 18:08:42 -05:00

264 lines
9.3 KiB
Python

"""Source 3 (oracle): ``perf stat -p <qemu_pid>`` sampler.
Spawns ``perf stat`` in interval-JSON mode against the qemu pid and
aggregates the per-event counter values into per-interval telemetry
rows. Unlike the /proc and QMP collectors, perf needs CAP_SYS_ADMIN
or ``kernel.perf_event_paranoid <= 1`` to read counters for a process
the collector doesn't own — typically true on a lab host running
QEMU under the cis490 service user.
Source 3 is **oracle-only** — perf counters are not available on a
deployed device. Every row carries ``available_in_deployment: false``.
The events we ask for are the small canonical set named in
docs/data-model.md:
cycles, instructions, cache-references, cache-misses,
branches, branch-misses, page-faults, context-switches
Anything perf can't enable on the host (e.g. cache-misses without
hardware support) is silently dropped from the row.
"""
from __future__ import annotations
import json
import logging
import shutil
import subprocess
import threading
import time
from pathlib import Path
log = logging.getLogger("cis490.collectors.perf_qemu")
SOURCE = "host_perf"
AVAILABLE_IN_DEPLOYMENT = False
DEFAULT_EVENTS = (
"cycles",
"instructions",
"cache-references",
"cache-misses",
"branches",
"branch-misses",
"page-faults",
"context-switches",
)
def perf_available() -> bool:
return shutil.which("perf") is not None
def _coerce_int(s: str | int | None) -> int | None:
if s is None:
return None
if isinstance(s, int):
return s
s = s.strip()
if not s or s in ("<not counted>", "<not supported>"):
return None
# perf prints comma-separated thousands by default; we asked -j so
# we usually get plain numbers, but guard for both shapes.
s = s.replace(",", "")
try:
return int(s)
except ValueError:
try:
return int(float(s))
except ValueError:
return None
def _build_row(t_mono_origin_ns: int, interval_s: float, agg: dict[str, int]) -> dict:
cycles = agg.get("cycles")
insns = agg.get("instructions")
cache_refs = agg.get("cache-references")
cache_miss = agg.get("cache-misses")
ipc = (insns / cycles) if (cycles and insns) else None
miss_rate = (cache_miss / cache_refs) if (cache_refs and cache_miss is not None) else None
return {
"t_mono_ns": time.monotonic_ns() - t_mono_origin_ns,
"t_wall_ns": time.time_ns(),
"source": SOURCE,
"available_in_deployment": AVAILABLE_IN_DEPLOYMENT,
"interval_s": interval_s,
"cycles": cycles,
"instructions": insns,
"cache_references": cache_refs,
"cache_misses": cache_miss,
"branches": agg.get("branches"),
"branch_misses": agg.get("branch-misses"),
"page_faults": agg.get("page-faults"),
"context_switches": agg.get("context-switches"),
"ipc": ipc,
"cache_miss_rate": miss_rate,
}
def parse_perf_event_line(line: str) -> dict | None:
"""Parse one ``perf stat -j`` event line. Returns None for blanks
or status messages perf occasionally interleaves on stderr-ish
paths but stdout-on-error in practice."""
line = line.strip()
if not line.startswith("{"):
return None
try:
return json.loads(line)
except json.JSONDecodeError:
return None
def run_loop(
pid: int,
output_path: Path,
t_mono_origin_ns: int,
interval_ms: int,
stop_event: threading.Event,
*,
events: tuple[str, ...] = DEFAULT_EVENTS,
emit_event: "callable | None" = None,
) -> int:
"""Spawn perf stat -j against ``pid`` and stream rows until stop.
Returns the number of rows written.
When ``emit_event`` is provided, perf lifecycle markers
(perf_unavailable / perf_started / perf_first_row / perf_no_counters)
are written into the orchestrator's events.jsonl. Without this, a
silent perf produces only a `rows_perf=0` count in meta.json which
is indistinguishable from "perf was never enabled" — the §1
silent-downgrade pattern."""
if not perf_available():
log.warning("perf binary not on PATH — perf collector disabled")
if emit_event is not None:
emit_event("perf_unavailable", reason="binary_not_on_path")
return 0
# perf stat writes its output (including -j JSON) to stderr by
# default when -p / --pid is in use; only when perf forks the
# workload itself does it go to stdout. --log-fd 1 forces output
# onto fd 1 so we can stream it through proc.stdout. Without this
# the collector silently writes 0 rows on every episode.
cmd = [
"perf", "stat",
"-p", str(pid),
"-I", str(interval_ms),
"-j",
"--log-fd", "1",
"-e", ",".join(events),
]
log.info("starting perf: %s", " ".join(cmd))
try:
proc = subprocess.Popen(
cmd,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
bufsize=1,
text=True,
)
except (FileNotFoundError, PermissionError) as e:
log.warning("perf launch failed: %s", e)
if emit_event is not None:
emit_event("perf_unavailable", reason=f"launch_failed: {e}")
return 0
if emit_event is not None:
emit_event("perf_started", pid=pid, events=list(events),
interval_ms=interval_ms)
rows = 0
output_path.parent.mkdir(parents=True, exist_ok=True)
cur_interval: float | None = None
agg: dict[str, int] = {}
intervals_seen = 0
intervals_with_values = 0
def _flush() -> None:
nonlocal rows, intervals_seen, intervals_with_values
if cur_interval is None:
return
intervals_seen += 1
# Always emit a row when an interval is observed, even if every
# event came back <not counted>. agg-empty means perf saw the
# interval but couldn't measure anything (paranoid level too
# high, qemu PID owned by different user, hardware counters
# unavailable, etc). Emitting None-valued rows distinguishes
# "perf was running but counters were unavailable" from "perf
# never ran at all" — which is the §1 visibility requirement
# the silent return-without-write was violating.
if agg:
intervals_with_values += 1
row = _build_row(t_mono_origin_ns, cur_interval, agg)
out_f.write(json.dumps(row) + "\n")
rows += 1
if (rows == 1) and emit_event is not None:
emit_event("perf_first_row",
counters_populated=bool(agg),
counters=list(agg.keys()))
try:
with output_path.open("a", buffering=1) as out_f:
# perf interleaves events and writes to stdout in -j mode.
# We read line by line until the process exits (which
# happens when we kill it on stop, or when the target pid
# disappears and perf's internal -p polling notices).
assert proc.stdout is not None
for line in proc.stdout:
if stop_event.is_set():
break
evt = parse_perf_event_line(line)
if evt is None:
continue
interval = evt.get("interval")
event_name = evt.get("event")
value = _coerce_int(evt.get("counter-value"))
if interval is None or event_name is None:
continue
# perf annotates event names with the privilege scope it
# was actually able to measure (e.g. "cycles:u" when only
# userspace is permitted under perf_event_paranoid=2).
# Strip the suffix so _build_row's plain-name lookups
# ("cycles", "instructions", ...) hit.
event_name = event_name.split(":", 1)[0]
# perf emits one JSON per (event, interval); a new
# interval value means we should flush the previous row.
if cur_interval is not None and interval != cur_interval:
_flush()
agg = {}
cur_interval = interval
if value is not None:
agg[event_name] = value
# End of stream — flush the last partial row.
_flush()
if emit_event is not None:
emit_event(
"perf_finished",
rows=rows,
intervals_seen=intervals_seen,
intervals_with_values=intervals_with_values,
)
if intervals_seen > 0 and intervals_with_values == 0:
# Loud signal: perf was running, intervals ticked, but
# every event came back <not counted>. Likely cause:
# perf_event_paranoid > 2, or qemu PID owned by a
# different user than the perf collector.
emit_event(
"perf_no_counters",
intervals_seen=intervals_seen,
likely_cause=("perf_event_paranoid > 2 or "
"qemu PID owned by different user"),
)
finally:
if proc.poll() is None:
proc.terminate()
try:
proc.wait(timeout=3.0)
except subprocess.TimeoutExpired:
proc.kill()
proc.wait(timeout=2.0)
return rows