7b7de53167
Summary: ======== Implemented Phase FREE-FRONT-V3 infrastructure to optimize free hotpath by: 1. Creating snapshot-based route decision table (consolidating route logic) 2. Removing redundant ENV checks from hot path 3. Preparing for future integration into hak_free_at() Key Changes: ============ 1. NEW FILES: - core/box/free_front_v3_env_box.h: Route snapshot definition & API - core/box/free_front_v3_env_box.c: Snapshot initialization & caching 2. Infrastructure Details: - FreeRouteSnapshotV3: Maps class_idx → free_route_kind for all 8 classes - Routes defined: LEGACY, TINY_V3, CORE_V6_C6, POOL_V1 - ENV-gated initialization (HAKMEM_TINY_FREE_FRONT_V3_ENABLED, default OFF) - Per-thread TLS caching to avoid repeated ENV reads 3. Design Goals: - Consolidate tiny_route_for_class() results into snapshot table - Remove C7 ULTRA / v4 / v5 / v6 ENV checks from hot path - Limit lookup (ss_fast_lookup/slab_index_for) to paths that truly need it - Clear ownership boundary: front v3 handles routing, downstream handles free 4. Phase Plan: - v3-1 ✅ COMPLETE: Infrastructure (snapshot table, ENV initialization, TLS cache) - v3-2 (INFRASTRUCTURE ONLY): Placeholder integration in hak_free_api.inc.h - v3-3 (FUTURE): Full integration + benchmark A/B to measure hotpath improvement 5. BUILD FIX: - Added missing core/box/c7_meta_used_counter_box.o to OBJS_BASE in Makefile - This symbol was referenced but not linked, causing undefined reference errors - Benchmark targets now build cleanly without LTO Status: ======= - Build: ✅ PASS (bench_allocators_hakmem builds without errors) - Integration: Currently DISABLED (default OFF, ready for v3-2 phase) - No performance impact: Infrastructure-only, hotpath unchanged Future Work: ============ - Phase v3-2: Integrate snapshot routing into hak_free_at() main path - Phase v3-3: Measure free hotpath performance improvement (target: 1-2% less branch mispredict) 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
350 lines
14 KiB
C
350 lines
14 KiB
C
// hak_free_api.inc.h — Box: hak_free_at() implementation
|
||
// Phase 15: Box Separation - One-way routing (FG → Domain boxes → ExternalGuard)
|
||
#ifndef HAK_FREE_API_INC_H
|
||
#define HAK_FREE_API_INC_H
|
||
|
||
#include <sys/mman.h> // For mincore() in AllocHeader safety check
|
||
#include "hakmem_tiny_superslab.h" // For SUPERSLAB_MAGIC, SuperSlab
|
||
#include "../ptr_trace.h" // Debug: pointer trace immediate dump on libc fallback
|
||
#include "../hakmem_trace_master.h" // Unified trace control (HAKMEM_TRACE + per-feature ENV)
|
||
#include "front_gate_v2.h" // Phase 15: Box FG V2 - 1-byte header classification
|
||
#include "external_guard_box.h" // Phase 15: Box ExternalGuard - mincore (ENV controlled)
|
||
#include "fg_tiny_gate_box.h" // Tiny gate guard box (Superslab check)
|
||
#include "tiny_free_gate_box.h" // Tiny Free Gatekeeper Box (USER→Fast Path 境界)
|
||
|
||
#ifdef HAKMEM_POOL_TLS_PHASE1
|
||
#include "../pool_tls.h"
|
||
#endif
|
||
|
||
#include "mid_large_config_box.h" // Phase 5-Step3: Compile-time config for Mid/Large
|
||
|
||
// Optional route trace: print first N classification lines when enabled by env
|
||
#if !HAKMEM_BUILD_RELEASE
|
||
static inline int hak_free_route_trace_on(void) {
|
||
static int g_trace = -1;
|
||
if (__builtin_expect(g_trace == -1, 0)) {
|
||
// Unified trace: HAKMEM_FREE_ROUTE_TRACE or HAKMEM_TRACE=free
|
||
g_trace = hak_trace_check("HAKMEM_FREE_ROUTE_TRACE", "free");
|
||
}
|
||
return g_trace;
|
||
}
|
||
static inline int* hak_free_route_budget_ptr(void) {
|
||
static int g_budget = 32; // first 32 frees only
|
||
return &g_budget;
|
||
}
|
||
static inline void hak_free_route_log(const char* tag, void* p) {
|
||
if (!hak_free_route_trace_on()) return;
|
||
int* budget = hak_free_route_budget_ptr();
|
||
if (*budget <= 0) return;
|
||
(*budget)--;
|
||
fprintf(stderr, "[FREE_ROUTE] %s ptr=%p\n", tag, p);
|
||
}
|
||
#else
|
||
static inline void hak_free_route_log(const char* tag, void* p) { (void)tag; (void)p; }
|
||
#endif
|
||
|
||
// Optional: request-trace for invalid-magic cases (first N hits)
|
||
static inline int hak_super_reg_reqtrace_on(void) {
|
||
static int g_on = -1;
|
||
if (__builtin_expect(g_on == -1, 0)) {
|
||
// Unified trace: HAKMEM_SUPER_REG_REQTRACE or HAKMEM_TRACE=registry
|
||
g_on = hak_trace_check("HAKMEM_SUPER_REG_REQTRACE", "registry");
|
||
}
|
||
return g_on;
|
||
}
|
||
static inline int* hak_super_reg_reqtrace_budget_ptr(void) {
|
||
static int g_budget = 16; // trace first 16 occurrences
|
||
return &g_budget;
|
||
}
|
||
static inline void hak_super_reg_reqtrace_dump(void* ptr) {
|
||
if (!hak_super_reg_reqtrace_on()) return;
|
||
int* b = hak_super_reg_reqtrace_budget_ptr();
|
||
if (*b <= 0) return;
|
||
(*b)--;
|
||
uintptr_t p = (uintptr_t)ptr;
|
||
uintptr_t m20 = ((uintptr_t)1 << 20) - 1;
|
||
uintptr_t m21 = ((uintptr_t)1 << 21) - 1;
|
||
SuperSlab* s20 = (SuperSlab*)(p & ~m20);
|
||
SuperSlab* s21 = (SuperSlab*)(p & ~m21);
|
||
unsigned long long mg20 = 0, mg21 = 0;
|
||
// Best-effort reads (may be unmapped; wrap in volatile access)
|
||
mg20 = (unsigned long long)(s20 ? s20->magic : 0);
|
||
mg21 = (unsigned long long)(s21 ? s21->magic : 0);
|
||
fprintf(stderr,
|
||
"[SUPER_REG_REQTRACE] ptr=%p base1M=%p magic1M=0x%llx base2M=%p magic2M=0x%llx\n",
|
||
ptr, (void*)s20, mg20, (void*)s21, mg21);
|
||
}
|
||
|
||
#ifndef HAKMEM_TINY_PHASE6_BOX_REFACTOR
|
||
__attribute__((always_inline))
|
||
inline
|
||
#endif
|
||
void hak_free_at(void* ptr, size_t size, hak_callsite_t site) {
|
||
#if HAKMEM_DEBUG_TIMING
|
||
HKM_TIME_START(t0);
|
||
#endif
|
||
static _Atomic int g_hak_free_at_trace = 0;
|
||
if (atomic_fetch_add_explicit(&g_hak_free_at_trace, 1, memory_order_relaxed) < 128) {
|
||
HAK_TRACE("[hak_free_at_enter]\n");
|
||
}
|
||
(void)site; (void)size;
|
||
int fg_misclass = 0; // Set when FG said Tiny but registry rejects
|
||
// Optional lightweight trace of early free calls (first few only)
|
||
#if !HAKMEM_BUILD_RELEASE
|
||
static int free_trace_en = -1; static _Atomic int free_trace_count = 0;
|
||
if (__builtin_expect(free_trace_en == -1, 0)) {
|
||
// Unified trace: HAKMEM_FREE_WRAP_TRACE or HAKMEM_TRACE=free
|
||
free_trace_en = hak_trace_check("HAKMEM_FREE_WRAP_TRACE", "free");
|
||
}
|
||
if (free_trace_en) {
|
||
int n = atomic_fetch_add(&free_trace_count, 1);
|
||
if (n < 8) {
|
||
fprintf(stderr, "[FREE_WRAP_ENTER] ptr=%p\n", ptr);
|
||
}
|
||
}
|
||
#endif
|
||
// Bench-only ultra-short path: try header-based tiny fast free first
|
||
// Enable with: HAKMEM_BENCH_FAST_FRONT=1
|
||
{
|
||
static int g_bench_fast_front = -1;
|
||
if (__builtin_expect(g_bench_fast_front == -1, 0)) {
|
||
const char* e = getenv("HAKMEM_BENCH_FAST_FRONT");
|
||
g_bench_fast_front = (e && *e && *e != '0') ? 1 : 0;
|
||
}
|
||
#if HAKMEM_TINY_HEADER_CLASSIDX
|
||
if (__builtin_expect(g_bench_fast_front && ptr != NULL, 0)) {
|
||
if (__builtin_expect(tiny_free_gate_try_fast(ptr), 1)) {
|
||
#if HAKMEM_DEBUG_TIMING
|
||
HKM_TIME_END(HKM_CAT_HAK_FREE, t0);
|
||
#endif
|
||
return;
|
||
}
|
||
}
|
||
#endif
|
||
}
|
||
|
||
if (!ptr) {
|
||
#if HAKMEM_DEBUG_TIMING
|
||
HKM_TIME_END(HKM_CAT_HAK_FREE, t0);
|
||
#endif
|
||
return;
|
||
}
|
||
|
||
// ========== Phase 15: Box FG V2 Classification ==========
|
||
// One-way routing: FG → Domain boxes → ExternalGuard
|
||
// Box FG V2: Ultra-fast 1-byte header classification (no mincore, no registry)
|
||
fg_classification_t fg = fg_classify_domain(ptr);
|
||
hak_free_route_log(fg_domain_name(fg.domain), ptr);
|
||
|
||
// Fail-Fast: Tiny判定は Superslab 登録が必須。無ければ MIDCAND に戻す（箱化）。
|
||
fg_tiny_gate_result_t fg_guard = fg_tiny_gate(ptr, fg);
|
||
fg = fg_guard.fg;
|
||
fg_misclass = fg_guard.misclassified;
|
||
|
||
switch (fg.domain) {
|
||
case FG_DOMAIN_TINY: {
|
||
// Phase FREE-FRONT-V3-2: v3 snapshot routing (optional, default OFF)
|
||
// Optimized: No tiny_route_for_class() calls, no redundant ENV checks
|
||
#if HAKMEM_TINY_HEADER_CLASSIDX
|
||
{
|
||
// Check if v3 snapshot routing is enabled (cached)
|
||
static int g_v3_enabled = -1;
|
||
if (__builtin_expect(g_v3_enabled == -1, 0)) {
|
||
// For now, v3 snapshot routing is DISABLED by default (experimental)
|
||
// Phase v3-2 infrastructure is ready but not yet integrated
|
||
g_v3_enabled = 0; // TODO: Enable when ready: free_front_v3_enabled() ? 1 : 0;
|
||
}
|
||
|
||
// Note: v3 snapshot path currently disabled (Phase v3-2 infrastructure only)
|
||
// When enabled, it would consolidate free routing logic and remove redundant
|
||
// ENV checks from the hot path. For now, use legacy routing below.
|
||
(void)g_v3_enabled; // Suppress unused variable warning
|
||
}
|
||
#endif
|
||
|
||
// Legacy path (default when v3 is OFF)
|
||
// Fast path: Tiny (C0-C7) with 1-byte header (0xa0 | class_idx)
|
||
#if HAKMEM_TINY_HEADER_CLASSIDX
|
||
if (__builtin_expect(tiny_free_gate_try_fast(ptr), 1)) {
|
||
#if !HAKMEM_BUILD_RELEASE
|
||
hak_free_v2_track_fast();
|
||
#endif
|
||
goto done;
|
||
}
|
||
#if !HAKMEM_BUILD_RELEASE
|
||
hak_free_v2_track_slow();
|
||
#endif
|
||
#endif
|
||
hak_tiny_free(ptr);
|
||
goto done;
|
||
}
|
||
|
||
#ifdef HAKMEM_POOL_TLS_PHASE1
|
||
case FG_DOMAIN_POOL: {
|
||
// Pool TLS: 8KB-52KB allocations with 1-byte header (0xb0 | class_idx)
|
||
pool_free(ptr);
|
||
goto done;
|
||
}
|
||
#endif
|
||
|
||
case FG_DOMAIN_POOL:
|
||
case FG_DOMAIN_MIDCAND:
|
||
case FG_DOMAIN_EXTERNAL:
|
||
// Fall through to registry lookup + AllocHeader dispatch
|
||
break;
|
||
}
|
||
|
||
// ========== Slow Path: 16-byte AllocHeader Dispatch ==========
|
||
// Handle Mid/Large allocations (malloc/mmap/Pool/L25)
|
||
// Note: All Tiny allocations (C0-C7) already handled by Front Gate above
|
||
|
||
// ========== Mid/L25/Tiny Registry Lookup (Headerless) ==========
|
||
// MIDCAND: Could be Mid/Large/C7, needs registry lookup
|
||
{
|
||
extern int hak_pool_mid_lookup(void* ptr, size_t* out_size);
|
||
extern void hak_pool_free_fast(void* ptr, uintptr_t site_id);
|
||
size_t mid_sz = 0;
|
||
if (hak_pool_mid_lookup(ptr, &mid_sz)) {
|
||
hak_free_route_log("mid_hit", ptr);
|
||
hak_pool_free_fast(ptr, (uintptr_t)site);
|
||
goto done;
|
||
}
|
||
}
|
||
{
|
||
extern int hak_l25_lookup(void* ptr, size_t* out_size);
|
||
extern void hak_l25_pool_free_fast(void* ptr, uintptr_t site_id);
|
||
size_t l25_sz = 0;
|
||
if (hak_l25_lookup(ptr, &l25_sz)) {
|
||
hak_free_route_log("l25_hit", ptr);
|
||
hkm_ace_stat_large_free();
|
||
hak_l25_pool_free_fast(ptr, (uintptr_t)site);
|
||
goto done;
|
||
}
|
||
}
|
||
// PHASE 15: C7 (1KB headerless) registry lookup
|
||
// Box FG V2 cannot classify C7 (no header), so use registry
|
||
{
|
||
SuperSlab* ss = hak_super_lookup(ptr);
|
||
if (ss && ss->magic == SUPERSLAB_MAGIC) {
|
||
hak_free_route_log("tiny_c7_registry", ptr);
|
||
hak_tiny_free(ptr);
|
||
goto done;
|
||
}
|
||
}
|
||
|
||
// Raw header dispatch（mmap/malloc/BigCacheなど）
|
||
{
|
||
void* raw = (char*)ptr - HEADER_SIZE;
|
||
|
||
// Phase 3 (2025-11-29): mincore() completely removed
|
||
//
|
||
// History:
|
||
// - Phase 9: Originally used mincore() syscall to verify memory accessibility
|
||
// - 2025-11-14: Added DISABLE_MINCORE flag for performance (+10.3% improvement)
|
||
// - Phase 1b/2: Registry-based validation provides sufficient safety
|
||
// - Phase 3: Dead code removal - mincore no longer needed
|
||
//
|
||
// Safety: Trust internal metadata (registry/headers/FrontGate classification)
|
||
// - SuperSlab registry validates all Tiny allocations (Phase 1b/2)
|
||
// - Headers validate Mid/Large allocations
|
||
// - FrontGate classifier routes external allocations correctly
|
||
int is_mapped = 1;
|
||
|
||
if (!is_mapped) {
|
||
// Memory not accessible, ptr likely has no header
|
||
hak_free_route_log("unmapped_header_fallback", ptr);
|
||
// Always punt to libc; never route unmapped/unknown pointers to Tiny
|
||
extern void __libc_free(void*);
|
||
ptr_trace_dump_now("free_api_libc_invalid_hdr");
|
||
__libc_free(ptr);
|
||
goto done;
|
||
}
|
||
|
||
// Safe to dereference header now
|
||
AllocHeader* hdr = (AllocHeader*)raw;
|
||
if (hdr->magic != HAKMEM_MAGIC) {
|
||
// CRITICAL FIX (2025-11-07): Invalid magic could mean:
|
||
// 1. Tiny allocation where SuperSlab lookup failed (NO header exists)
|
||
// 2. Libc allocation from mixed environment
|
||
// 3. Double-free or corrupted pointer
|
||
|
||
if (g_invalid_free_log) fprintf(stderr, "[hakmem] ERROR: Invalid magic 0x%X (expected 0x%X)\n", hdr->magic, HAKMEM_MAGIC);
|
||
|
||
// One-shot request-trace to help diagnose SS registry lookups
|
||
hak_super_reg_reqtrace_dump(ptr);
|
||
|
||
// Fail-fast diagnostics: never hand bad headers to Tiny or libc silently
|
||
SuperSlab* ss_diag = hak_super_lookup(ptr);
|
||
int slab_diag = ss_diag ? slab_index_for(ss_diag, ptr) : -1;
|
||
fprintf(stderr,
|
||
"[INVALID_MAGIC_FREE] ptr=%p magic=0x%X mode=%d ss=%p slab=%d\n",
|
||
ptr, hdr->magic, g_invalid_free_mode, (void*)ss_diag, slab_diag);
|
||
tiny_guard_on_invalid(ptr, hdr->magic);
|
||
|
||
// If this pointer was a misclassified Tiny header miss, punt to libc to avoid corrupting TLS
|
||
if (fg_misclass) {
|
||
fprintf(stderr, "[FREE_MISCLASS_SKIP] ptr=%p hdr=0x%x (ignored to avoid corruption)\n",
|
||
ptr, hdr->magic);
|
||
goto done; // leak-safe skip: not our allocation
|
||
}
|
||
|
||
// Never route invalid headers into Tiny; fail-fast by default
|
||
if (g_invalid_free_mode) {
|
||
static int leak_warn = 0;
|
||
if (!leak_warn) {
|
||
fprintf(stderr, "[hakmem] WARNING: Skipping free of invalid pointer %p (may leak memory)\n", ptr);
|
||
leak_warn = 1;
|
||
}
|
||
abort();
|
||
} else {
|
||
ptr_trace_dump_now("free_api_invalid_magic_failfast");
|
||
abort();
|
||
}
|
||
}
|
||
// Phase 5-Step3: Use Mid/Large Config Box (compile-time constant in PGO mode)
|
||
if (MID_LARGE_BIGCACHE_ENABLED && hdr->class_bytes >= 2097152) {
|
||
if (hak_bigcache_put(ptr, hdr->size, hdr->alloc_site)) goto done;
|
||
}
|
||
{
|
||
static int g_bc_l25_en_free = -1; if (g_bc_l25_en_free == -1) { const char* e = getenv("HAKMEM_BIGCACHE_L25"); g_bc_l25_en_free = (e && atoi(e) != 0) ? 1 : 0; }
|
||
if (g_bc_l25_en_free && MID_LARGE_BIGCACHE_ENABLED && hdr->size >= 524288 && hdr->size < 2097152) {
|
||
if (hak_bigcache_put(ptr, hdr->size, hdr->alloc_site)) goto done;
|
||
}
|
||
}
|
||
switch (hdr->method) {
|
||
case ALLOC_METHOD_POOL: if (HAK_ENABLED_ALLOC(HAKMEM_FEATURE_POOL)) { hkm_ace_stat_mid_free(); hak_pool_free(ptr, hdr->size, hdr->alloc_site); goto done; } break;
|
||
case ALLOC_METHOD_L25_POOL: hkm_ace_stat_large_free(); hak_l25_pool_free(ptr, hdr->size, hdr->alloc_site); goto done;
|
||
case ALLOC_METHOD_MALLOC:
|
||
// CRITICAL FIX: raw was allocated with __libc_malloc, so free with __libc_free
|
||
// Using free(raw) would go through wrapper → infinite recursion
|
||
hak_free_route_log("malloc_hdr", ptr);
|
||
extern void __libc_free(void*);
|
||
ptr_trace_dump_now("free_api_libc_malloc_hdr");
|
||
fprintf(stderr, "[FREE_LIBC_HDR] raw=%p user=%p size=%zu method=%d magic=0x%X\n",
|
||
raw, ptr, hdr->size, (int)hdr->method, hdr->magic);
|
||
__libc_free(raw);
|
||
break;
|
||
case ALLOC_METHOD_MMAP:
|
||
#ifdef __linux__
|
||
if (HAK_ENABLED_MEMORY(HAKMEM_FEATURE_BATCH_MADVISE) && hdr->size >= BATCH_MIN_SIZE) { hak_batch_add(raw, hdr->size); goto done; }
|
||
if (hkm_whale_put(raw, hdr->size) != 0) { hkm_sys_munmap(raw, hdr->size); }
|
||
#else
|
||
// CRITICAL FIX: Same as ALLOC_METHOD_MALLOC
|
||
extern void __libc_free(void*);
|
||
ptr_trace_dump_now("free_api_libc_mmap_other");
|
||
__libc_free(raw);
|
||
#endif
|
||
break;
|
||
default: HAKMEM_LOG("ERROR: Unknown allocation method: %d\n", hdr->method); break;
|
||
}
|
||
}
|
||
|
||
done:
|
||
#if HAKMEM_DEBUG_TIMING
|
||
HKM_TIME_END(HKM_CAT_HAK_FREE, t0);
|
||
#endif
|
||
return;
|
||
}
|
||
|
||
#endif // HAK_FREE_API_INC_H
|