tomoaki/hakmem
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
131cdb7b885494a4a7a80e430d9284c5505c2b91
hakmem/core/box
History
Moe Charm (CI) eae0435c03 Adaptive CAS: Single-threaded fast path optimization 
PROBLEM:
- Atomic freelist (Phase 1) introduced 3-5x overhead in hot path
- CAS loop overhead: 16-27 cycles vs 4-6 cycles (non-atomic)
- Single-threaded workloads pay MT safety cost unnecessarily

SOLUTION:
- Runtime thread detection with g_hakmem_active_threads counter
- Single-threaded (1T): Skip CAS, use relaxed load/store (fast)
- Multi-threaded (2+T): Full CAS loop for MT safety

IMPLEMENTATION:
1. core/hakmem_tiny.c:240 - Added g_hakmem_active_threads atomic counter
2. core/hakmem_tiny.c:248 - Added hakmem_thread_register() for per-thread init
3. core/hakmem_tiny.h:160-163 - Exported thread counter and registration API
4. core/box/hak_alloc_api.inc.h:34 - Call hakmem_thread_register() on first alloc
5. core/box/slab_freelist_atomic.h:58-68 - Adaptive CAS in pop_lockfree()
6. core/box/slab_freelist_atomic.h:118-126 - Adaptive CAS in push_lockfree()

DESIGN:
- Thread counter: Incremented on first allocation per thread
- Fast path check: if (num_threads <= 1) → relaxed ops
- Slow path: Full CAS loop (existing Phase 1 implementation)
- Zero overhead when truly single-threaded

PERFORMANCE:
Random Mixed 256B (Single-threaded):
  Before (Phase 1): 16.7M ops/s
  After:            14.9M ops/s (-11%, thread counter overhead)

Larson (Single-threaded):
  Before: 47.9M ops/s
  After:  47.9M ops/s (no change, already fast)

Larson (Multi-threaded 8T):
  Before: 48.8M ops/s
  After:  48.3M ops/s (-1%, within noise)

MT STABILITY:
  1T: 47.9M ops/s ✅
  8T: 48.3M ops/s ✅ (zero crashes, stable)

NOTES:
- Expected Larson improvement (0.80M → 1.80M) not observed
- Larson was already fast (47.9M) in Phase 1
- Possible Task investigation used different benchmark
- Adaptive CAS implementation verified and working correctly

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

Co-Authored-By: Claude <noreply@anthropic.com>
2025-11-22 03:30:47 +09:00
..
ace_pool_connector.c
Tiny: fix header/stride mismatch and harden refill paths
2025-11-09 18:55:50 +09:00
ace_pool_connector.h
Tiny: fix header/stride mismatch and harden refill paths
2025-11-09 18:55:50 +09:00
adopt_gate_box.h
CRITICAL FIX: TLS 未初期化による 4T SEGV を完全解消
2025-11-07 01:27:04 +09:00
bench_fast_box.c
Phase 3d-B: TLS Cache Merge - Unified g_tls_sll[] structure (+12-18% expected)
2025-11-20 07:32:30 +09:00
bench_fast_box.h
Phase 20-2: BenchFast mode - Structural bottleneck analysis (+4.5% ceiling)
2025-11-16 06:36:02 +09:00
capacity_box.c
Phase 3d-B: TLS Cache Merge - Unified g_tls_sll[] structure (+12-18% expected)
2025-11-20 07:32:30 +09:00
capacity_box.d
Phase E3-FINAL: Fix Box API offset bugs - ALL classes now use correct offsets
2025-11-13 06:50:20 +09:00
capacity_box.h
Box API Phase 1-3: Capacity Manager, Carve-Push, Prewarm 実装
2025-11-13 01:45:30 +09:00
carve_push_box.c
Phase 3d-B: TLS Cache Merge - Unified g_tls_sll[] structure (+12-18% expected)
2025-11-20 07:32:30 +09:00
carve_push_box.d
Code Cleanup: Remove false positives, redundant validations, and reduce verbose logging
2025-11-21 23:00:24 +09:00
carve_push_box.h
Box API Phase 1-3: Capacity Manager, Carve-Push, Prewarm 実装
2025-11-13 01:45:30 +09:00
external_guard_box.h
Phase 3d-A: SlabMeta Box boundary - Encapsulate SuperSlab metadata access
2025-11-20 02:01:52 +09:00
free_local_box.c
Code Cleanup: Remove false positives, redundant validations, and reduce verbose logging
2025-11-21 23:00:24 +09:00
free_local_box.d
Code Cleanup: Remove false positives, redundant validations, and reduce verbose logging
2025-11-21 23:00:24 +09:00
free_local_box.h
CRITICAL FIX: TLS 未初期化による 4T SEGV を完全解消
2025-11-07 01:27:04 +09:00
free_publish_box.c
CRITICAL FIX: TLS 未初期化による 4T SEGV を完全解消
2025-11-07 01:27:04 +09:00
free_publish_box.d
Phase12 debug: restore SUPERSLAB constants/APIs, implement Box2 drain boundary, fix tiny_fast_pop to return BASE, honor TLS SLL toggle in alloc/free fast paths, add fail-fast stubs, and quiet capacity sentinel. Update CURRENT_TASK with A/B results (SLL-off stable; SLL-on crash).
2025-11-14 01:02:00 +09:00
free_publish_box.h
CRITICAL FIX: TLS 未初期化による 4T SEGV を完全解消
2025-11-07 01:27:04 +09:00
free_remote_box.c
Code Cleanup: Remove false positives, redundant validations, and reduce verbose logging
2025-11-21 23:00:24 +09:00
free_remote_box.d
Code Cleanup: Remove false positives, redundant validations, and reduce verbose logging
2025-11-21 23:00:24 +09:00
free_remote_box.h
CRITICAL FIX: TLS 未初期化による 4T SEGV を完全解消
2025-11-07 01:27:04 +09:00
front_gate_box.c
Phase 3d-B: TLS Cache Merge - Unified g_tls_sll[] structure (+12-18% expected)
2025-11-20 07:32:30 +09:00
front_gate_box.d
Code Cleanup: Remove false positives, redundant validations, and reduce verbose logging
2025-11-21 23:00:24 +09:00
front_gate_box.h
CRITICAL FIX: TLS 未初期化による 4T SEGV を完全解消
2025-11-07 01:27:04 +09:00
front_gate_classifier.c
Phase 23 Unified Cache + PageFaultTelemetry generalization: Mid/VM page-fault bottleneck identified
2025-11-17 02:47:58 +09:00
front_gate_classifier.d
Fix C7 TLS SLL header restoration regression + Document Larson MT race condition
2025-11-22 02:15:34 +09:00
front_gate_classifier.h
Phase 23 Unified Cache + PageFaultTelemetry generalization: Mid/VM page-fault bottleneck identified
2025-11-17 02:47:58 +09:00
front_gate_v2.h
Phase 15: Box Separation (partial) - Box headers completed, routing deferred
2025-11-15 22:08:51 +09:00
front_metrics_box.c
Phase 19 & 20-1: Frontend optimization + TLS cache prewarm (+16.2% total)
2025-11-16 05:48:59 +09:00
front_metrics_box.d
Phase 21-1-B: Ring cache Alloc/Free 統合 - C2/C3 hot path integration
2025-11-16 07:51:37 +09:00
front_metrics_box.h
Phase 19 & 20-1: Frontend optimization + TLS cache prewarm (+16.2% total)
2025-11-16 05:48:59 +09:00
hak_alloc_api.inc.h
Adaptive CAS: Single-threaded fast path optimization
2025-11-22 03:30:47 +09:00
hak_core_init.inc.h
Phase 23 Unified Cache + PageFaultTelemetry generalization: Mid/VM page-fault bottleneck identified
2025-11-17 02:47:58 +09:00
hak_exit_debug.inc.h
CRITICAL FIX: TLS 未初期化による 4T SEGV を完全解消
2025-11-07 01:27:04 +09:00
hak_free_api.inc.h
Phase 15: Box BenchMeta separation + ExternalGuard debug + investigation report
2025-11-15 23:00:21 +09:00
hak_kpi_util.inc.h
CRITICAL FIX: TLS 未初期化による 4T SEGV を完全解消
2025-11-07 01:27:04 +09:00
hak_wrappers.inc.h
Phase 3d-B: TLS Cache Merge - Unified g_tls_sll[] structure (+12-18% expected)
2025-11-20 07:32:30 +09:00
integrity_box.c
Phase 3d-B: TLS Cache Merge - Unified g_tls_sll[] structure (+12-18% expected)
2025-11-20 07:32:30 +09:00
integrity_box.d
Add Box I (Integrity), Box E (Expansion), and comprehensive P0 debugging infrastructure
2025-11-12 02:45:00 +09:00
integrity_box.h
Add Box I (Integrity), Box E (Expansion), and comprehensive P0 debugging infrastructure
2025-11-12 02:45:00 +09:00
mailbox_box.c
CRITICAL FIX: TLS 未初期化による 4T SEGV を完全解消
2025-11-07 01:27:04 +09:00
mailbox_box.d
Phase12 debug: restore SUPERSLAB constants/APIs, implement Box2 drain boundary, fix tiny_fast_pop to return BASE, honor TLS SLL toggle in alloc/free fast paths, add fail-fast stubs, and quiet capacity sentinel. Update CURRENT_TASK with A/B results (SLL-off stable; SLL-on crash).
2025-11-14 01:02:00 +09:00
mailbox_box.h
CRITICAL FIX: TLS 未初期化による 4T SEGV を完全解消
2025-11-07 01:27:04 +09:00
pagefault_telemetry_box.c
Phase 23 Unified Cache + PageFaultTelemetry generalization: Mid/VM page-fault bottleneck identified
2025-11-17 02:47:58 +09:00
pagefault_telemetry_box.d
Phase 23 Unified Cache + PageFaultTelemetry generalization: Mid/VM page-fault bottleneck identified
2025-11-17 02:47:58 +09:00
pagefault_telemetry_box.h
Phase 23 Unified Cache + PageFaultTelemetry generalization: Mid/VM page-fault bottleneck identified
2025-11-17 02:47:58 +09:00
pool_api.inc.h
Phase 23 Unified Cache + PageFaultTelemetry generalization: Mid/VM page-fault bottleneck identified
2025-11-17 02:47:58 +09:00
pool_core_api.inc.h
CRITICAL FIX: TLS 未初期化による 4T SEGV を完全解消
2025-11-07 01:27:04 +09:00
pool_init_api.inc.h
release: silence runtime logs and stabilize benches
2025-11-11 01:47:06 +09:00
pool_mf2_adoption.inc.h
Phase 1: Box Theory refactoring + include reduction
2025-11-06 21:54:12 +09:00
pool_mf2_core.inc.h
CRITICAL FIX: TLS 未初期化による 4T SEGV を完全解消
2025-11-07 01:27:04 +09:00
pool_mf2_helpers.inc.h
Phase 1: Box Theory refactoring + include reduction
2025-11-06 21:54:12 +09:00
pool_mf2_types.inc.h
Phase 1: Box Theory refactoring + include reduction
2025-11-06 21:54:12 +09:00
pool_mid_desc.inc.h
CRITICAL FIX: TLS 未初期化による 4T SEGV を完全解消
2025-11-07 01:27:04 +09:00
pool_mid_tc.inc.h
CRITICAL FIX: TLS 未初期化による 4T SEGV を完全解消
2025-11-07 01:27:04 +09:00
pool_refill.inc.h
CRITICAL FIX: TLS 未初期化による 4T SEGV を完全解消
2025-11-07 01:27:04 +09:00
pool_stats.inc.h
CRITICAL FIX: TLS 未初期化による 4T SEGV を完全解消
2025-11-07 01:27:04 +09:00
pool_tls_core.inc.h
CRITICAL FIX: TLS 未初期化による 4T SEGV を完全解消
2025-11-07 01:27:04 +09:00
pool_tls_ring.inc.h
CRITICAL FIX: TLS 未初期化による 4T SEGV を完全解消
2025-11-07 01:27:04 +09:00
pool_tls_types.inc.h
Phase 1: Box Theory refactoring + include reduction
2025-11-06 21:54:12 +09:00
prewarm_box.c
Phase 3d-B: TLS Cache Merge - Unified g_tls_sll[] structure (+12-18% expected)
2025-11-20 07:32:30 +09:00
prewarm_box.d
Phase12 debug: restore SUPERSLAB constants/APIs, implement Box2 drain boundary, fix tiny_fast_pop to return BASE, honor TLS SLL toggle in alloc/free fast paths, add fail-fast stubs, and quiet capacity sentinel. Update CURRENT_TASK with A/B results (SLL-off stable; SLL-on crash).
2025-11-14 01:02:00 +09:00
prewarm_box.h
Box API Phase 1-3: Capacity Manager, Carve-Push, Prewarm 実装
2025-11-13 01:45:30 +09:00
ptr_conversion_box.h
Phase E3-FINAL: Fix Box API offset bugs - ALL classes now use correct offsets
2025-11-13 06:50:20 +09:00
slab_freelist_atomic.h
Adaptive CAS: Single-threaded fast path optimization
2025-11-22 03:30:47 +09:00
ss_ace_box.c
Phase 3d-B: TLS Cache Merge - Unified g_tls_sll[] structure (+12-18% expected)
2025-11-20 07:32:30 +09:00
ss_ace_box.h
Phase 3d-B: TLS Cache Merge - Unified g_tls_sll[] structure (+12-18% expected)
2025-11-20 07:32:30 +09:00
ss_allocation_box.c
Fix C0/C7 class confusion: Upgrade C7 stride to 2048B and fix meta->class_idx initialization
2025-11-21 13:44:05 +09:00
ss_allocation_box.h
Phase 3d-B: TLS Cache Merge - Unified g_tls_sll[] structure (+12-18% expected)
2025-11-20 07:32:30 +09:00
ss_cache_box.c
Phase 3d-B: TLS Cache Merge - Unified g_tls_sll[] structure (+12-18% expected)
2025-11-20 07:32:30 +09:00
ss_cache_box.h
Phase 3d-B: TLS Cache Merge - Unified g_tls_sll[] structure (+12-18% expected)
2025-11-20 07:32:30 +09:00
ss_hot_cold_box.h
Phase 12-1.1: EMPTY Slab Detection + Immediate Reuse (+13% improvement, 10.2M→11.5M ops/s)
2025-11-21 04:56:48 +09:00
ss_hot_prewarm_box.c
Phase 26: Front Gate Unification - Tiny allocator fast path (+12.9%)
2025-11-17 05:29:08 +09:00
ss_hot_prewarm_box.d
Phase 21-1-B: Ring cache Alloc/Free 統合 - C2/C3 hot path integration
2025-11-16 07:51:37 +09:00
ss_hot_prewarm_box.h
Phase 19 & 20-1: Frontend optimization + TLS cache prewarm (+16.2% total)
2025-11-16 05:48:59 +09:00
ss_legacy_backend_box.c
Code Cleanup: Remove false positives, redundant validations, and reduce verbose logging
2025-11-21 23:00:24 +09:00
ss_legacy_backend_box.h
Phase 3d-B: TLS Cache Merge - Unified g_tls_sll[] structure (+12-18% expected)
2025-11-20 07:32:30 +09:00
ss_os_acquire_box.c
Phase 3d-B: TLS Cache Merge - Unified g_tls_sll[] structure (+12-18% expected)
2025-11-20 07:32:30 +09:00
ss_os_acquire_box.h
Phase 3d-B: TLS Cache Merge - Unified g_tls_sll[] structure (+12-18% expected)
2025-11-20 07:32:30 +09:00
ss_slab_management_box.c
Phase 3d-B: TLS Cache Merge - Unified g_tls_sll[] structure (+12-18% expected)
2025-11-20 07:32:30 +09:00
ss_slab_management_box.h
Phase 3d-B: TLS Cache Merge - Unified g_tls_sll[] structure (+12-18% expected)
2025-11-20 07:32:30 +09:00
ss_slab_meta_box.h
Phase 1: Atomic Freelist Implementation - MT Safety Foundation
2025-11-22 02:46:57 +09:00
ss_stats_box.c
Phase 3d-B: TLS Cache Merge - Unified g_tls_sll[] structure (+12-18% expected)
2025-11-20 07:32:30 +09:00
ss_stats_box.h
Phase 3d-B: TLS Cache Merge - Unified g_tls_sll[] structure (+12-18% expected)
2025-11-20 07:32:30 +09:00
ss_unified_backend_box.c
Fix C0/C7 class confusion: Upgrade C7 stride to 2048B and fix meta->class_idx initialization
2025-11-21 13:44:05 +09:00
ss_unified_backend_box.h
Phase 3d-B: TLS Cache Merge - Unified g_tls_sll[] structure (+12-18% expected)
2025-11-20 07:32:30 +09:00
superslab_expansion_box.c
Fix C0/C7 class confusion: Upgrade C7 stride to 2048B and fix meta->class_idx initialization
2025-11-21 13:44:05 +09:00
superslab_expansion_box.d
Phase12 debug: restore SUPERSLAB constants/APIs, implement Box2 drain boundary, fix tiny_fast_pop to return BASE, honor TLS SLL toggle in alloc/free fast paths, add fail-fast stubs, and quiet capacity sentinel. Update CURRENT_TASK with A/B results (SLL-off stable; SLL-on crash).
2025-11-14 01:02:00 +09:00
superslab_expansion_box.h
Add Box I (Integrity), Box E (Expansion), and comprehensive P0 debugging infrastructure
2025-11-12 02:45:00 +09:00
tiny_near_empty_box.c
Phase 3d-B: TLS Cache Merge - Unified g_tls_sll[] structure (+12-18% expected)
2025-11-20 07:32:30 +09:00
tiny_near_empty_box.d
Phase 3d-B: TLS Cache Merge - Unified g_tls_sll[] structure (+12-18% expected)
2025-11-20 07:32:30 +09:00
tiny_near_empty_box.h
Phase 3d-B: TLS Cache Merge - Unified g_tls_sll[] structure (+12-18% expected)
2025-11-20 07:32:30 +09:00
tiny_next_ptr_box.h
Fix C0/C7 class confusion: Upgrade C7 stride to 2048B and fix meta->class_idx initialization
2025-11-21 13:44:05 +09:00
tiny_sizeclass_hist_box.c
Phase 3d-B: TLS Cache Merge - Unified g_tls_sll[] structure (+12-18% expected)
2025-11-20 07:32:30 +09:00
tiny_sizeclass_hist_box.d
Phase 3d-B: TLS Cache Merge - Unified g_tls_sll[] structure (+12-18% expected)
2025-11-20 07:32:30 +09:00
tiny_sizeclass_hist_box.h
Phase 3d-B: TLS Cache Merge - Unified g_tls_sll[] structure (+12-18% expected)
2025-11-20 07:32:30 +09:00
tls_sll_box.h
Fix C7 TLS SLL header restoration regression + Document Larson MT race condition
2025-11-22 02:15:34 +09:00
tls_sll_drain_box.h
Fix C7 TLS SLL corruption: Protect next pointer from user data overwrites
2025-11-21 23:42:43 +09:00
unified_batch_box.c
Phase 23 Unified Cache + PageFaultTelemetry generalization: Mid/VM page-fault bottleneck identified
2025-11-17 02:47:58 +09:00
unified_batch_box.d
Code Cleanup: Remove false positives, redundant validations, and reduce verbose logging
2025-11-21 23:00:24 +09:00
unified_batch_box.h
Phase 23 Unified Cache + PageFaultTelemetry generalization: Mid/VM page-fault bottleneck identified
2025-11-17 02:47:58 +09:00