Files
hakmem/core/box/tiny_env_box.c
Moe Charm (CI) deecda7336 Phase 3 C2: Slab Metadata Cache Optimization (3 patches) - NEUTRAL
Patch 1: Policy Hot Cache
- Add TinyPolicyHot struct (route_kind[8] cached in TLS)
- Eliminate policy_snapshot() calls (~2 memory ops saved)
- Safety: disabled when learner v7 active
- Files: tiny_metadata_cache_env_box.h, tiny_metadata_cache_hot_box.{h,c}
- Integration: malloc_tiny_fast.h route selection

Patch 2: First Page Inline Cache
- Cache current slab page pointer in TLS per-class
- Avoid superslab metadata lookup (1-2 memory ops)
- Fast-path in tiny_legacy_fallback_free_base()
- Files: tiny_first_page_cache.h, tiny_unified_cache.c
- Integration: tiny_legacy_fallback_box.h

Patch 3: Bounds Check Compile-out
- Hardcode unified_cache capacity as MACRO constant
- Eliminate modulo operation (constant fold)
- Macros: TINY_UNIFIED_CACHE_CAPACITY_POW2=11, CAPACITY=2048, MASK=2047
- File: tiny_unified_cache.h

A/B Test Results (Mixed, 10-run):
- Baseline (C2=0): 40.43M ops/s (avg), 40.72M ops/s (median)
- Optimized (C2=1): 40.25M ops/s (avg), 40.29M ops/s (median)
- Improvement: -0.45% (avg), -1.06% (median)
- DECISION: NEUTRAL (within ±1.0% threshold)
- Action: Keep as research box (ENV gate OFF by default)

Cumulative Gain (Phase 2-3):
- B3 (Routing shape): +2.89%
- B4 (Wrapper split): +1.47%
- C3 (Static routing): +2.20%
- C2 (Metadata cache): -0.45%
- Total: ~6.1% (from baseline 37.5M → 39.8M ops/s)

🤖 Generated with Claude Code

Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
2025-12-13 19:19:42 +09:00

74 lines
3.7 KiB
C

#include "tiny_env_box.h"
#include <stdlib.h>
#include <stdatomic.h>
tiny_env_cfg_t g_tiny_env = {0};
static int env_flag(const char* name, int def_val) {
const char* e = getenv(name);
if (!e || *e == '\0') return def_val;
return (*e != '0');
}
static int env_int(const char* name, int def_val) {
const char* e = getenv(name);
if (!e || *e == '\0') return def_val;
return atoi(e);
}
void tiny_env_init_once(void) {
static _Atomic int init_state = 0; // 0: uninit, 1: inited
int expected = 0;
if (!atomic_compare_exchange_strong_explicit(&init_state, &expected, 1,
memory_order_acq_rel, memory_order_relaxed)) {
return; // already initialized
}
g_tiny_env.tiny_tls_sll = env_flag("HAKMEM_TINY_TLS_SLL", 1);
g_tiny_env.sll_multiplier = env_int("HAKMEM_SLL_MULTIPLIER", 2);
g_tiny_env.tiny_no_front_cache = env_flag("HAKMEM_TINY_NO_FRONT_CACHE", 0);
g_tiny_env.tiny_no_quick = env_flag("HAKMEM_TINY_NO_QUICK", 0);
g_tiny_env.tiny_prefetch = env_flag("HAKMEM_TINY_PREFETCH", 0);
g_tiny_env.front_direct = env_flag("HAKMEM_TINY_FRONT_DIRECT", 0);
g_tiny_env.use_class_map = env_flag("HAKMEM_TINY_NO_CLASS_MAP", 0) ? 0 : 1;
g_tiny_env.route_enable = env_flag("HAKMEM_ROUTE", 0);
g_tiny_env.route_sample_lg = env_int("HAKMEM_ROUTE_SAMPLE_LG", 10);
g_tiny_env.larson_fix = env_flag("HAKMEM_TINY_LARSON_FIX", 0);
g_tiny_env.active_track = env_flag("HAKMEM_TINY_ACTIVE_TRACK", 0);
g_tiny_env.drain_to_sll = env_int("HAKMEM_TINY_DRAIN_TO_SLL", 0);
g_tiny_env.free_to_ss = env_flag("HAKMEM_TINY_FREE_TO_SS", 0);
g_tiny_env.route_free = env_flag("HAKMEM_TINY_ROUTE_FREE", 0);
g_tiny_env.sll_diag = env_flag("HAKMEM_TINY_SLL_DIAG", 0);
g_tiny_env.sll_safeheader = env_flag("HAKMEM_TINY_SLL_SAFEHEADER", 0);
g_tiny_env.sll_ring = env_flag("HAKMEM_TINY_SLL_RING", 0);
g_tiny_env.free_fast = env_flag("HAKMEM_TINY_FREE_FAST", 1);
g_tiny_env.sll_canary_fast = env_flag("HAKMEM_TINY_SLL_CANARY_FAST", 0);
g_tiny_env.ss_free_debug = env_flag("HAKMEM_SS_FREE_DEBUG", 0);
g_tiny_env.ss_adopt = env_flag("HAKMEM_TINY_SS_ADOPT", 1);
g_tiny_env.disable_remote = env_flag("HAKMEM_TINY_DISABLE_REMOTE", 0);
g_tiny_env.freelist_mask = env_flag("HAKMEM_TINY_FREELIST_MASK", 0);
g_tiny_env.alloc_1024_metric = env_flag("HAKMEM_TINY_ALLOC_1024_METRIC", 0);
g_tiny_env.tiny_profile = env_flag("HAKMEM_TINY_PROFILE", 0);
g_tiny_env.tiny_fast_stats = env_flag("HAKMEM_TINY_FAST_STATS", 0);
g_tiny_env.heap_v2_stats = env_flag("HAKMEM_TINY_HEAP_V2_STATS", 0);
g_tiny_env.front_slim = env_flag("HAKMEM_TINY_FRONT_SLIM", 0);
{
int pct = env_int("HAKMEM_SFC_CASCADE_PCT", 50);
if (pct < 0) pct = 0;
if (pct > 100) pct = 100;
g_tiny_env.sfc_cascade_pct = pct;
}
g_tiny_env.sfc_cascade = env_flag("HAKMEM_TINY_SFC_CASCADE", 0);
g_tiny_env.alloc_remote_relax = env_flag("HAKMEM_TINY_ALLOC_REMOTE_RELAX", 0);
g_tiny_env.ss_empty_reuse = env_flag("HAKMEM_SS_EMPTY_REUSE", 1);
g_tiny_env.ss_empty_scan_limit = env_int("HAKMEM_SS_EMPTY_SCAN_LIMIT", 32);
g_tiny_env.ss_acquire_debug = env_flag("HAKMEM_SS_ACQUIRE_DEBUG", 0);
g_tiny_env.tension_drain_enable = env_flag("HAKMEM_TINY_TENSION_DRAIN_ENABLE", 1);
g_tiny_env.tension_drain_threshold = env_int("HAKMEM_TINY_TENSION_DRAIN_THRESHOLD", 1024);
g_tiny_env.alloc_route_shape = env_flag("HAKMEM_TINY_ALLOC_ROUTE_SHAPE", 0);
g_tiny_env.tiny_metadata_cache = env_flag("HAKMEM_TINY_METADATA_CACHE", 0);
g_tiny_env.inited = 1;
}