Files
hakmem/core/box/tiny_header_box.h
Moe Charm (CI) f7b18aaf13 Phase 5 E5-2: Header Write-Once (NEUTRAL, FROZEN)
Target: tiny_region_id_write_header (3.35% self%)
- Hypothesis: Headers redundant for reused blocks
- Strategy: Write headers ONCE at refill boundary, skip in hot alloc

Implementation:
- ENV gate: HAKMEM_TINY_HEADER_WRITE_ONCE=0/1 (default 0)
- core/box/tiny_header_write_once_env_box.h: ENV gate
- core/box/tiny_header_write_once_stats_box.h: Stats counters
- core/box/tiny_header_box.h: Added tiny_header_finalize_alloc()
- core/front/tiny_unified_cache.c: Prefill at 3 refill sites
- core/box/tiny_front_hot_box.h: Use finalize function

A/B Test Results (Mixed, 10-run, 20M iters):
- Baseline (WRITE_ONCE=0): 44.22M ops/s (mean), 44.53M ops/s (median)
- Optimized (WRITE_ONCE=1): 44.42M ops/s (mean), 44.36M ops/s (median)
- Improvement: +0.45% mean, -0.38% median

Decision: NEUTRAL (within ±1.0% threshold)
- Action: FREEZE as research box (default OFF, do not promote)

Root Cause Analysis:
- Header writes are NOT redundant - existing code writes only when needed
- Branch overhead (~4 cycles) cancels savings (~3-5 cycles)
- perf self% ≠ optimization ROI (3.35% target → +0.45% gain)

Key Lessons:
1. Verify assumptions before optimizing (inspect code paths)
2. Hot spot self% measures time IN function, not savings from REMOVING it
3. Branch overhead matters (even "simple" checks add cycles)

Positive Outcome:
- StdDev reduced 50% (0.96M → 0.48M) - more stable performance

Health Check: PASS (all profiles)

Next Candidates:
- free_tiny_fast_cold: 7.14% self%
- unified_cache_push: 3.39% self%
- hakmem_env_snapshot_enabled: 2.97% self%

Deliverables:
- docs/analysis/PHASE5_E5_2_HEADER_REFILL_ONCE_DESIGN.md
- docs/analysis/PHASE5_E5_2_HEADER_REFILL_ONCE_AB_TEST_RESULTS.md
- CURRENT_TASK.md (E5-2 complete, FROZEN)

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
2025-12-14 06:22:25 +09:00

226 lines
7.7 KiB
C

// tiny_header_box.h - Header Box: Single Source of Truth for Header Operations
//
// Design Principles:
// 1. All header logic flows from tiny_nextptr.h specification
// 2. Encapsulates "which classes preserve headers" knowledge
// 3. Eliminates hardcoded class_idx checks (class_idx != 7, etc.)
// 4. Provides type-safe header read/write/validate operations
//
// Background:
// - C0 (8B): next_off=0 → header overwritten by next pointer
// - C1-C6 (16B-1024B): next_off=1 → header preserved in freelist
// - C7 (2048B): next_off=0 → header overwritten by next pointer
//
// Migration:
// ❌ FORBIDDEN: class_idx != 7, class_idx == 0 || class_idx == 7
// ❌ FORBIDDEN: *(uint8_t*)base = HEADER_MAGIC | ...
// ✅ USE: tiny_class_preserves_header(class_idx)
// ✅ USE: tiny_header_write_if_preserved(base, class_idx)
// ✅ USE: tiny_header_validate(base, class_idx, ...)
#ifndef TINY_HEADER_BOX_H
#define TINY_HEADER_BOX_H
#include <stdint.h>
#include <stdbool.h>
#include "../hakmem_build_flags.h"
#include "tiny_layout_box.h"
#include "../tiny_region_id.h"
// ============================================================================
// Core Predicate: Does this class preserve headers in freelist?
// ============================================================================
//
// This is the SINGLE SOURCE OF TRUTH for header preservation logic.
// All code must use this instead of hardcoded class_idx checks.
//
// Implementation:
// - Delegates to tiny_nextptr_offset() from tiny_layout_box.h
// - offset=0 → header overwritten by next pointer → false
// - offset!=0 → header preserved → true
//
// Returns:
// true - C1-C6: Header preserved at offset 0, next at offset 1
// false - C0, C7: Header overwritten by next pointer at offset 0
static inline bool tiny_class_preserves_header(int class_idx) {
#if HAKMEM_TINY_HEADER_CLASSIDX
// Delegate to tiny_layout_box.h specification (Single Source of Truth)
// next_off=0 → header overwritten (C0, C7)
// next_off=1 → header preserved (C1-C6)
return tiny_nextptr_offset(class_idx) != 0;
#else
// Headers disabled globally
(void)class_idx;
return false;
#endif
}
// ============================================================================
// Header Write (Conditional - for freelist/TLS SLL operations)
// ============================================================================
//
// Writes header ONLY if this class preserves headers.
// For C0/C7, writing header is pointless (next pointer will overwrite it).
//
// Use this when:
// - Pushing blocks to TLS SLL
// - Moving blocks from freelist to TLS SLL
// - Splicing chains into TLS SLL
//
// DO NOT use this for:
// - Allocation path (use tiny_header_write_for_alloc instead)
static inline void tiny_header_write_if_preserved(void* base, int class_idx) {
#if HAKMEM_TINY_HEADER_CLASSIDX
if (tiny_class_preserves_header(class_idx)) {
*(uint8_t*)base = HEADER_MAGIC | (class_idx & HEADER_CLASS_MASK);
}
#else
(void)base;
(void)class_idx;
#endif
}
// ============================================================================
// Header Validate (Conditional - for TLS SLL pop operations)
// ============================================================================
//
// Validates header ONLY if this class preserves headers.
// For C0/C7, validation is impossible (next pointer is stored at offset 0).
//
// Arguments:
// base - BASE pointer (not user pointer)
// class_idx - Expected class index
// out_got - [optional] Store actual header byte read
// out_expect- [optional] Store expected header byte
//
// Returns:
// true - Header valid OR class doesn't preserve headers (C0/C7)
// false - Header mismatch (corruption detected)
//
// Use this when:
// - Popping blocks from TLS SLL
// - Validating freelist integrity
static inline bool tiny_header_validate(const void* base, int class_idx,
uint8_t* out_got, uint8_t* out_expect) {
#if HAKMEM_TINY_HEADER_CLASSIDX
// C0/C7: Validation impossible (next pointer stored at offset 0)
if (!tiny_class_preserves_header(class_idx)) {
return true; // Always valid (no header to check)
}
// C1-C6: Validate header
uint8_t got = *(const uint8_t*)base;
uint8_t expect = HEADER_MAGIC | (class_idx & HEADER_CLASS_MASK);
if (out_got) *out_got = got;
if (out_expect) *out_expect = expect;
return got == expect;
#else
(void)base;
(void)class_idx;
(void)out_got;
(void)out_expect;
return true;
#endif
}
// ============================================================================
// Header Write (Unconditional - for allocation path)
// ============================================================================
//
// ALWAYS writes header, regardless of class.
// For C0/C7, header will be overwritten when block enters freelist,
// but must be valid when returned to user.
//
// Use this ONLY in allocation path:
// - HAK_RET_ALLOC_BLOCK macro
// - HAK_RET_ALLOC_BLOCK_TRACED macro
// - Before returning block to user
//
// DO NOT use this for:
// - Freelist operations (use tiny_header_write_if_preserved)
// - TLS SLL operations (use tiny_header_write_if_preserved)
static inline void tiny_header_write_for_alloc(void* base, int class_idx) {
#if HAKMEM_TINY_HEADER_CLASSIDX
*(uint8_t*)base = HEADER_MAGIC | (class_idx & HEADER_CLASS_MASK);
#else
(void)base;
(void)class_idx;
#endif
}
// ============================================================================
// Header Read (for diagnostics/debugging)
// ============================================================================
//
// Reads header byte without validation.
// Returns -1 if headers disabled or class doesn't preserve headers.
//
// Use this for:
// - Diagnostics
// - Debug logging
// - Corruption analysis
//
// DO NOT use this for:
// - Validation (use tiny_header_validate)
static inline int tiny_header_read(const void* base, int class_idx) {
#if HAKMEM_TINY_HEADER_CLASSIDX
if (!tiny_class_preserves_header(class_idx)) {
return -1; // No header to read
}
return (int)(*(const uint8_t*)base);
#else
(void)base;
(void)class_idx;
return -1;
#endif
}
// ============================================================================
// Header Finalize for Allocation (Phase 5 E5-2: Write-Once Optimization)
// ============================================================================
//
// Replaces direct calls to tiny_region_id_write_header() in allocation paths.
// Enables header write-once optimization:
// - C1-C6: Skip header write if already prefilled at refill boundary
// - C0, C7: Always write header (next pointer overwrites it anyway)
//
// Use this in allocation hot paths:
// - tiny_hot_alloc_fast()
// - unified_cache_pop()
// - All other allocation returns
//
// DO NOT use this for:
// - Freelist operations (use tiny_header_write_if_preserved)
// - Refill boundary (use direct write in unified_cache_refill)
// Forward declaration from tiny_region_id.h
void* tiny_region_id_write_header(void* base, int class_idx);
// Forward declaration from tiny_header_write_once_env_box.h
int tiny_header_write_once_enabled(void);
static inline void* tiny_header_finalize_alloc(void* base, int class_idx) {
#if HAKMEM_TINY_HEADER_CLASSIDX
// Write-once optimization: Skip header write for C1-C6 if already prefilled
if (tiny_header_write_once_enabled() && tiny_class_preserves_header(class_idx)) {
// Header already written at refill boundary → skip write, return USER pointer
return (void*)((uint8_t*)base + 1);
}
// Traditional path: C0, C7, or WRITE_ONCE=0
return tiny_region_id_write_header(base, class_idx);
#else
(void)class_idx;
return base;
#endif
}
#endif // TINY_HEADER_BOX_H