tomoaki/hakmem
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Phase v6-6: Inline hot path optimization for SmallObject Core v6

Browse Source 
Optimize v6 alloc/free by eliminating redundant route checks and adding
inline hot path functions:

- smallobject_core_v6_box.h: Add inline hot path functions:
  - small_alloc_c6_hot_v6() / small_alloc_c5_hot_v6(): Direct TLS pop
  - small_free_c6_hot_v6() / small_free_c5_hot_v6(): Direct TLS push
  - No route check needed (caller already validated via switch case)

- smallobject_core_v6.c: Add cold path functions:
  - small_alloc_cold_v6(): Handle TLS refill from page
  - small_free_cold_v6(): Handle page freelist push (TLS full/cross-thread)

- malloc_tiny_fast.h: Update front gate to use inline hot path:
  - Alloc: hot path first, cold path fallback on TLS miss
  - Free: hot path first, cold path fallback on TLS full

Performance results:
- C5-heavy: v6 ON 42.2M ≈ baseline (parity restored)
- C6-heavy: v6 ON 34.5M ≈ baseline (parity restored)
- Mixed 16-1024B: ~26.5M (v3-only: ~28.1M, gap is routing overhead)

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

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
This commit is contained in:
Moe Charm (CI)
2025-12-11 15:59:29 +09:00
parent 1e04debb1b
commit e2ca52d59d
3 changed files with 190 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

29
core/front/malloc_tiny_fast.h
View File

@ -158,10 +158,20 @@ static inline void* malloc_tiny_fast(size_t size) {
switch (route) {
case TINY_ROUTE_SMALL_HEAP_V6: {
// Phase v6-1: C6-only Core v6 route stub (pool v1 fallback)
// Phase v6-6: Inline hot path (no route check, direct TLS pop)
SmallHeapCtxV6* ctx_v6 = small_heap_ctx_v6();
const SmallPolicySnapshotV6* snap_v6 = tiny_policy_snapshot_v6();
void* v6p = small_alloc_fast_v6(size, (uint32_t)class_idx, ctx_v6, snap_v6);
void* v6p = NULL;
if (class_idx == 6) {
v6p = small_alloc_c6_hot_v6(ctx_v6);
if (TINY_HOT_UNLIKELY(!v6p)) {
v6p = small_alloc_cold_v6(6, ctx_v6);
}
} else if (class_idx == 5) {
v6p = small_alloc_c5_hot_v6(ctx_v6);
if (TINY_HOT_UNLIKELY(!v6p)) {
v6p = small_alloc_cold_v6(5, ctx_v6);
}
}
if (TINY_HOT_LIKELY(v6p != NULL)) {
return v6p;
}
@ -374,10 +384,17 @@ static inline int free_tiny_fast(void* ptr) {
if (__builtin_expect(use_tiny_heap, 0)) {
switch (route) {
case TINY_ROUTE_SMALL_HEAP_V6: {
// Phase v6-1: C6-only Core v6 route stub
// Phase v6-6: Inline hot path (no route check, direct TLS push)
SmallHeapCtxV6* ctx_v6 = small_heap_ctx_v6();
const SmallPolicySnapshotV6* snap_v6 = tiny_policy_snapshot_v6();
small_free_fast_v6(base, (uint32_t)class_idx, ctx_v6, snap_v6);
int handled = 0;
if (class_idx == 6) {
handled = small_free_c6_hot_v6(ctx_v6, base);
} else if (class_idx == 5) {
handled = small_free_c5_hot_v6(ctx_v6, base);
}
if (!handled) {
small_free_cold_v6(base, (uint32_t)class_idx);
}
return 1;
}
case TINY_ROUTE_SMALL_HEAP_V5: {