Debug Counters Implementation - Clean History

Major Features:
- Debug counter infrastructure for Refill Stage tracking
- Free Pipeline counters (ss_local, ss_remote, tls_sll)
- Diagnostic counters for early return analysis
- Unified larson.sh benchmark runner with profiles
- Phase 6-3 regression analysis documentation

Bug Fixes:
- Fix SuperSlab disabled by default (HAKMEM_TINY_USE_SUPERSLAB)
- Fix profile variable naming consistency
- Add .gitignore patterns for large files

Performance:
- Phase 6-3: 4.79 M ops/s (has OOM risk)
- With SuperSlab: 3.13 M ops/s (+19% improvement)

This is a clean repository without large log files.

🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>

benchmarks/scripts/comprehensive/run_bench_suite.sh

@@ -0,0 +1,71 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+# Run a small suite across system/mimalloc/hakmem and save logs under docs/benchmarks/<timestamp>.
+# Focus: mimalloc-bench/bench/larson patterns that cover Tiny/Mid/Large/Big.
+# Optionally include a 'hakmem best' run (WRAP L1 + learning on + DYN auto).
+
+RUNTIME=${RUNTIME:-1}
+THREADS_CSV=${THREADS:-"1,4"}
+BEST=${BEST:-0}
+# Hard wall-clock timeout per run (external). Defaults to RUNTIME+3s.
+BENCH_TIMEOUT=${BENCH_TIMEOUT:-}
+KILL_GRACE=${KILL_GRACE:-2}
+if [[ -z "${BENCH_TIMEOUT}" ]]; then
+  # Add small cushion to allow larson to exit cleanly
+  BENCH_TIMEOUT=$(( RUNTIME + 3 ))
+fi
+
+ROOT_DIR="$(cd "$(dirname "$0")/.." && pwd)"
+OUTDIR="$ROOT_DIR/docs/benchmarks/$(date +%Y%m%d_%H%M%S)_SUITE"
+mkdir -p "$OUTDIR"
+
+LARSON="$ROOT_DIR/mimalloc-bench/bench/larson/larson"
+LIB_HAK="$ROOT_DIR/libhakmem.so"
+LIB_HAK_ABS=$(readlink -f "$LIB_HAK")
+LIB_MI="/lib/x86_64-linux-gnu/libmimalloc.so.2"
+
+echo "Suite: RUNTIME=${RUNTIME}s THREADS=${THREADS_CSV} BEST=${BEST}" | tee "$OUTDIR/summary.txt"
+
+run_case() {
+  local label="$1"; shift
+  local min="$1"; local max="$2"; shift 2
+  local threads_csv="$1"; shift
+  IFS=',' read -r -a TARR <<< "$threads_csv"
+  for t in "${TARR[@]}"; do
+    echo "== ${label} | ${t}T | system ==" | tee -a "$OUTDIR/summary.txt"
+    timeout -k "${KILL_GRACE}s" "${BENCH_TIMEOUT}s" \
+      "$LARSON" "$RUNTIME" "$min" "$max" 10000 1 12345 "$t" 2>&1 \
+      | tee "$OUTDIR/${label}_system_T${t}.log" | tail -n 3 | tee -a "$OUTDIR/summary.txt"
+    if [[ -f "$LIB_MI" ]]; then
+      echo "== ${label} | ${t}T | mimalloc ==" | tee -a "$OUTDIR/summary.txt"
+      timeout -k "${KILL_GRACE}s" "${BENCH_TIMEOUT}s" \
+        env LD_PRELOAD="$LIB_MI" "$LARSON" "$RUNTIME" "$min" "$max" 10000 1 12345 "$t" 2>&1 \
+        | tee "$OUTDIR/${label}_mimalloc_T${t}.log" | tail -n 3 | tee -a "$OUTDIR/summary.txt"
+    fi
+    if [[ -f "$LIB_HAK" ]]; then
+      echo "== ${label} | ${t}T | hakmem(default) ==" | tee -a "$OUTDIR/summary.txt"
+      timeout -k "${KILL_GRACE}s" "${BENCH_TIMEOUT}s" \
+        env LD_PRELOAD="$LIB_HAK_ABS" "$LARSON" "$RUNTIME" "$min" "$max" 10000 1 12345 "$t" 2>&1 \
+        | tee "$OUTDIR/${label}_hakmem_default_T${t}.log" | tail -n 3 | tee -a "$OUTDIR/summary.txt"
+      if [[ "$BEST" == "1" ]]; then
+        echo "== ${label} | ${t}T | hakmem(best) ==" | tee -a "$OUTDIR/summary.txt"
+        timeout -k "${KILL_GRACE}s" "${BENCH_TIMEOUT}s" \
+          env HAKMEM_WRAP_L2=1 HAKMEM_WRAP_L25=1 HAKMEM_LEARN=1 HAKMEM_DYN1_AUTO=1 HAKMEM_DYN2_AUTO=1 HAKMEM_HIST_SAMPLE=7 HAKMEM_WMAX_LEARN=1 HAKMEM_WMAX_DWELL_SEC=2 \
+          LD_PRELOAD="$LIB_HAK_ABS" "$LARSON" "$RUNTIME" "$min" "$max" 10000 1 12345 "$t" 2>&1 \
+          | tee "$OUTDIR/${label}_hakmem_best_T${t}.log" | tail -n 3 | tee -a "$OUTDIR/summary.txt"
+      fi
+    fi
+  done
+}
+
+# Tiny band (8–64B)
+run_case tiny 8 64 "$THREADS_CSV"
+# Mid band (2–32KiB)
+run_case mid 2048 32768 "$THREADS_CSV"
+# Large band (64KiB–1MiB)
+run_case large 65536 1048576 "$THREADS_CSV"
+# Big band (2–4MiB)
+run_case big 2097152 4194304 "$THREADS_CSV"
+
+echo "Saved suite: $OUTDIR" | tee -a "$OUTDIR/summary.txt"