Files
hakmem/core/tiny_refill.h

65 lines
2.3 KiB
C
Raw Normal View History

// tiny_refill.h - Refill Boundary box (inline helpers)
#pragma once
#include <stdatomic.h>
#include "hakmem_tiny_superslab.h"
#include "hakmem_trace_master.h" // Phase 4c: Master trace control
#include "slab_handle.h"
#include "tiny_sticky.h"
CRITICAL FIX: TLS 未初期化による 4T SEGV を完全解消 **問題:** - Larson 4T で 100% SEGV (1T は 2.09M ops/s で完走) - System/mimalloc は 4T で 33.52M ops/s 正常動作 - SS OFF + Remote OFF でも 4T で SEGV **根本原因: (Task agent ultrathink 調査結果)** ``` CRASH: mov (%r15),%r13 R15 = 0x6261 ← ASCII "ba" (ゴミ値、未初期化TLS) ``` Worker スレッドの TLS 変数が未初期化: - `__thread void* g_tls_sll_head[TINY_NUM_CLASSES];` ← 初期化なし - pthread_create() で生成されたスレッドでゼロ初期化されない - NULL チェックが通過 (0x6261 != NULL) → dereference → SEGV **修正内容:** 全 TLS 配列に明示的初期化子 `= {0}` を追加: 1. **core/hakmem_tiny.c:** - `g_tls_sll_head[TINY_NUM_CLASSES] = {0}` - `g_tls_sll_count[TINY_NUM_CLASSES] = {0}` - `g_tls_live_ss[TINY_NUM_CLASSES] = {0}` - `g_tls_bcur[TINY_NUM_CLASSES] = {0}` - `g_tls_bend[TINY_NUM_CLASSES] = {0}` 2. **core/tiny_fastcache.c:** - `g_tiny_fast_cache[TINY_FAST_CLASS_COUNT] = {0}` - `g_tiny_fast_count[TINY_FAST_CLASS_COUNT] = {0}` - `g_tiny_fast_free_head[TINY_FAST_CLASS_COUNT] = {0}` - `g_tiny_fast_free_count[TINY_FAST_CLASS_COUNT] = {0}` 3. **core/hakmem_tiny_magazine.c:** - `g_tls_mags[TINY_NUM_CLASSES] = {0}` 4. **core/tiny_sticky.c:** - `g_tls_sticky_ss[TINY_NUM_CLASSES][TINY_STICKY_RING] = {0}` - `g_tls_sticky_idx[TINY_NUM_CLASSES][TINY_STICKY_RING] = {0}` - `g_tls_sticky_pos[TINY_NUM_CLASSES] = {0}` **効果:** ``` Before: 1T: 2.09M ✅ | 4T: SEGV 💀 After: 1T: 2.41M ✅ | 4T: 4.19M ✅ (+15% 1T, SEGV解消) ``` **テスト:** ```bash # 1 thread: 完走 ./larson_hakmem 2 8 128 1024 1 12345 1 → Throughput = 2,407,597 ops/s ✅ # 4 threads: 完走(以前は SEGV) ./larson_hakmem 2 8 128 1024 1 12345 4 → Throughput = 4,192,155 ops/s ✅ ``` **調査協力:** Task agent (ultrathink mode) による完璧な根本原因特定 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com>
2025-11-07 01:27:04 +09:00
#include "tiny_ready.h"
#include "box/mailbox_box.h"
#include "tiny_remote_bg.h" // Background remote-drain step (best-effort)
#include "tiny_ready_bg.h" // Ready aggregator (mailbox→ready hint)
#include "tiny_route.h" // Route Fingerprint (Box boundary tracing)
#include <stdio.h>
#include <stdlib.h>
// External helpers from main TU
static inline uint32_t tiny_self_u32(void);
static inline void tiny_tls_bind_slab(TinyTLSSlab* tls, SuperSlab* ss, int slab_idx);
// Forward decls in main TU
static inline uintptr_t hot_slot_pop(int class_idx);
static inline uintptr_t bench_pub_pop(int class_idx);
static inline SuperSlab* slab_entry_ss(uintptr_t ent);
static inline int slab_entry_idx(uintptr_t ent);
// Mailbox/Ready consumption always allowed (ENV gate removed)
static inline int tiny_mail_ready_allowed(void) { return 1; }
// Registry scan window (ENV: HAKMEM_TINY_REG_SCAN_MAX, default 256)
static inline int tiny_reg_scan_max(void) {
static int v = -1;
if (__builtin_expect(v == -1, 0)) {
const char* s = getenv("HAKMEM_TINY_REG_SCAN_MAX");
int defv = 256; // conservative default
if (s && *s) {
int parsed = atoi(s);
v = (parsed > 0) ? parsed : defv;
} else {
v = defv;
}
}
return v;
}
// Opportunistic background remote-drain knobs (ENV removed; fixed defaults)
static inline int tiny_bg_remote_tryrate(void) { return 16; }
static inline int tiny_bg_remote_budget_default(void) { return 2; }
CRITICAL FIX: TLS 未初期化による 4T SEGV を完全解消 **問題:** - Larson 4T で 100% SEGV (1T は 2.09M ops/s で完走) - System/mimalloc は 4T で 33.52M ops/s 正常動作 - SS OFF + Remote OFF でも 4T で SEGV **根本原因: (Task agent ultrathink 調査結果)** ``` CRASH: mov (%r15),%r13 R15 = 0x6261 ← ASCII "ba" (ゴミ値、未初期化TLS) ``` Worker スレッドの TLS 変数が未初期化: - `__thread void* g_tls_sll_head[TINY_NUM_CLASSES];` ← 初期化なし - pthread_create() で生成されたスレッドでゼロ初期化されない - NULL チェックが通過 (0x6261 != NULL) → dereference → SEGV **修正内容:** 全 TLS 配列に明示的初期化子 `= {0}` を追加: 1. **core/hakmem_tiny.c:** - `g_tls_sll_head[TINY_NUM_CLASSES] = {0}` - `g_tls_sll_count[TINY_NUM_CLASSES] = {0}` - `g_tls_live_ss[TINY_NUM_CLASSES] = {0}` - `g_tls_bcur[TINY_NUM_CLASSES] = {0}` - `g_tls_bend[TINY_NUM_CLASSES] = {0}` 2. **core/tiny_fastcache.c:** - `g_tiny_fast_cache[TINY_FAST_CLASS_COUNT] = {0}` - `g_tiny_fast_count[TINY_FAST_CLASS_COUNT] = {0}` - `g_tiny_fast_free_head[TINY_FAST_CLASS_COUNT] = {0}` - `g_tiny_fast_free_count[TINY_FAST_CLASS_COUNT] = {0}` 3. **core/hakmem_tiny_magazine.c:** - `g_tls_mags[TINY_NUM_CLASSES] = {0}` 4. **core/tiny_sticky.c:** - `g_tls_sticky_ss[TINY_NUM_CLASSES][TINY_STICKY_RING] = {0}` - `g_tls_sticky_idx[TINY_NUM_CLASSES][TINY_STICKY_RING] = {0}` - `g_tls_sticky_pos[TINY_NUM_CLASSES] = {0}` **効果:** ``` Before: 1T: 2.09M ✅ | 4T: SEGV 💀 After: 1T: 2.41M ✅ | 4T: 4.19M ✅ (+15% 1T, SEGV解消) ``` **テスト:** ```bash # 1 thread: 完走 ./larson_hakmem 2 8 128 1024 1 12345 1 → Throughput = 2,407,597 ops/s ✅ # 4 threads: 完走(以前は SEGV) ./larson_hakmem 2 8 128 1024 1 12345 4 → Throughput = 4,192,155 ops/s ✅ ``` **調査協力:** Task agent (ultrathink mode) による完璧な根本原因特定 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com>
2025-11-07 01:27:04 +09:00
// Mid-size simple refill (ENV: HAKMEM_TINY_MID_REFILL_SIMPLE)
static inline int tiny_mid_refill_simple_enabled(void) {
static int v = -1;
if (__builtin_expect(v == -1, 0)) {
const char* s = getenv("HAKMEM_TINY_MID_REFILL_SIMPLE");
v = (s && *s && *s != '0') ? 1 : 0;
}
return v;
}
#include "box/tiny_adopt_refill_box.h" // adopt/refill 境界の集約Step 5
// Try a quick adopt from sticky/hot/bench/mailbox (single pass)
static inline SuperSlab* tiny_refill_try_fast(int class_idx, TinyTLSSlab* tls) {
return tiny_adopt_refill_box(class_idx, tls);
}