## Summary
Implemented Phase 12 Shared SuperSlab Pool (mimalloc-style) to address
SuperSlab allocation churn (877 SuperSlabs → 100-200 target).
## Implementation (ChatGPT + Claude)
1. **Metadata changes** (superslab_types.h):
- Added class_idx to TinySlabMeta (per-slab dynamic class)
- Removed size_class from SuperSlab (no longer per-SuperSlab)
- Changed owner_tid (16-bit) → owner_tid_low (8-bit)
2. **Shared Pool** (hakmem_shared_pool.{h,c}):
- Global pool shared by all size classes
- shared_pool_acquire_slab() - Get free slab for class_idx
- shared_pool_release_slab() - Return slab when empty
- Per-class hints for fast path optimization
3. **Integration** (23 files modified):
- Updated all ss->size_class → meta->class_idx
- Updated all meta->owner_tid → meta->owner_tid_low
- superslab_refill() now uses shared pool
- Free path releases empty slabs back to pool
4. **Build system** (Makefile):
- Added hakmem_shared_pool.o to OBJS_BASE and TINY_BENCH_OBJS_BASE
## Status: ⚠️ Build OK, Runtime CRASH
**Build**: ✅ SUCCESS
- All 23 files compile without errors
- Only warnings: superslab_allocate type mismatch (legacy code)
**Runtime**: ❌ SEGFAULT
- Crash location: sll_refill_small_from_ss()
- Exit code: 139 (SIGSEGV)
- Test case: ./bench_random_mixed_hakmem 1000 256 42
## Known Issues
1. **SEGFAULT in refill path** - Likely shared_pool_acquire_slab() issue
2. **Legacy superslab_allocate()** still exists (type mismatch warning)
3. **Remaining TODOs** from design doc:
- SuperSlab physical layout integration
- slab_handle.h cleanup
- Remove old per-class head implementation
## Next Steps
1. Debug SEGFAULT (gdb backtrace shows sll_refill_small_from_ss)
2. Fix shared_pool_acquire_slab() or superslab_init_slab()
3. Basic functionality test (1K → 100K iterations)
4. Measure SuperSlab count reduction (877 → 100-200)
5. Performance benchmark (+650-860% expected)
## Files Changed (25 files)
core/box/free_local_box.c
core/box/free_remote_box.c
core/box/front_gate_classifier.c
core/hakmem_super_registry.c
core/hakmem_tiny.c
core/hakmem_tiny_bg_spill.c
core/hakmem_tiny_free.inc
core/hakmem_tiny_lifecycle.inc
core/hakmem_tiny_magazine.c
core/hakmem_tiny_query.c
core/hakmem_tiny_refill.inc.h
core/hakmem_tiny_superslab.c
core/hakmem_tiny_superslab.h
core/hakmem_tiny_tls_ops.h
core/slab_handle.h
core/superslab/superslab_inline.h
core/superslab/superslab_types.h
core/tiny_debug.h
core/tiny_free_fast.inc.h
core/tiny_free_magazine.inc.h
core/tiny_remote.c
core/tiny_superslab_alloc.inc.h
core/tiny_superslab_free.inc.h
Makefile
## New Files (3 files)
PHASE12_SHARED_SUPERSLAB_POOL_DESIGN.md
core/hakmem_shared_pool.c
core/hakmem_shared_pool.h
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Co-Authored-By: ChatGPT <chatgpt@openai.com>
277 lines
9.7 KiB
C
277 lines
9.7 KiB
C
#ifndef HAKMEM_TINY_REFILL_P0_INC_H
|
|
#define HAKMEM_TINY_REFILL_P0_INC_H
|
|
|
|
// hakmem_tiny_refill_p0.inc.h
|
|
// P0: Batch refill implementation (sll_refill_batch_from_ss only).
|
|
// Phase 12: DO NOT alias or redefine sll_refill_small_from_ss here.
|
|
// NOTE: This file is active only when HAKMEM_TINY_P0_BATCH_REFILL=1.
|
|
|
|
#if HAKMEM_TINY_P0_BATCH_REFILL
|
|
|
|
#include "tiny_box_geometry.h" // Box 3: Geometry & Capacity Calculator
|
|
#include "tiny_refill_opt.h"
|
|
#include "tiny_fc_api.h"
|
|
#include "superslab/superslab_inline.h" // For _ss_remote_drain_to_freelist_unsafe()
|
|
#include "box/integrity_box.h" // Box I: Integrity verification (Priority ALPHA)
|
|
#include "box/tiny_next_ptr_box.h" // Box API: Next pointer read/write
|
|
|
|
// Debug counters (compile-time gated)
|
|
#if HAKMEM_DEBUG_COUNTERS
|
|
extern unsigned long long g_rf_hit_slab[];
|
|
extern unsigned long long g_rf_early_no_ss[];
|
|
extern unsigned long long g_rf_early_no_meta[];
|
|
extern unsigned long long g_rf_early_no_room[];
|
|
extern unsigned long long g_rf_early_want_zero[];
|
|
#endif
|
|
|
|
// Optional P0 diagnostic logging helper
|
|
static inline int p0_should_log(void) {
|
|
static int en = -1;
|
|
if (__builtin_expect(en == -1, 0)) {
|
|
const char* e = getenv("HAKMEM_TINY_P0_LOG");
|
|
en = (e && *e && *e != '0') ? 1 : 0;
|
|
}
|
|
return en;
|
|
}
|
|
|
|
// P0 batch refill entry point
|
|
static inline int sll_refill_batch_from_ss(int class_idx, int max_take) {
|
|
// Phase E1-CORRECT: C7 now has headers, can use P0 batch refill
|
|
|
|
// Runtime A/B kill switch (defensive). Set HAKMEM_TINY_P0_DISABLE=1 to bypass P0 path.
|
|
do {
|
|
static int g_p0_disable = -1;
|
|
if (__builtin_expect(g_p0_disable == -1, 0)) {
|
|
const char* e = getenv("HAKMEM_TINY_P0_DISABLE");
|
|
g_p0_disable = (e && *e && *e != '0') ? 1 : 0;
|
|
}
|
|
if (__builtin_expect(g_p0_disable, 0)) {
|
|
return 0;
|
|
}
|
|
} while (0);
|
|
|
|
if (!g_use_superslab || max_take <= 0) {
|
|
#if HAKMEM_DEBUG_COUNTERS
|
|
if (!g_use_superslab) g_rf_early_no_ss[class_idx]++;
|
|
#endif
|
|
return 0;
|
|
}
|
|
|
|
TinyTLSSlab* tls = &g_tls_slabs[class_idx];
|
|
uint32_t active_before = 0;
|
|
if (tls->ss) {
|
|
active_before = atomic_load_explicit(&tls->ss->total_active_blocks, memory_order_relaxed);
|
|
}
|
|
|
|
if (!tls->ss) {
|
|
if (!superslab_refill(class_idx)) {
|
|
return 0;
|
|
}
|
|
}
|
|
|
|
TinySlabMeta* meta = tls->meta;
|
|
if (!meta) {
|
|
#if HAKMEM_DEBUG_COUNTERS
|
|
g_rf_early_no_meta[class_idx]++;
|
|
#endif
|
|
return 0;
|
|
}
|
|
|
|
#if HAKMEM_INTEGRITY_LEVEL >= 4
|
|
uint8_t* initial_slab_base =
|
|
tls->slab_base ? tls->slab_base : tiny_slab_base_for(tls->ss, tls->slab_idx);
|
|
SlabMetadataState meta_initial =
|
|
integrity_capture_slab_metadata(meta, initial_slab_base, class_idx);
|
|
INTEGRITY_CHECK_SLAB_METADATA(meta_initial, "P0 refill entry");
|
|
#endif
|
|
|
|
// Optional: Direct-FC fast path (kept as-is from original P0, no aliasing)
|
|
do {
|
|
static int g_direct_fc = -1;
|
|
static int g_direct_fc_c7 = -1;
|
|
if (__builtin_expect(g_direct_fc == -1, 0)) {
|
|
const char* e = getenv("HAKMEM_TINY_P0_DIRECT_FC");
|
|
g_direct_fc = (e && *e && *e == '0') ? 0 : 1;
|
|
}
|
|
if (__builtin_expect(g_direct_fc_c7 == -1, 0)) {
|
|
const char* e7 = getenv("HAKMEM_TINY_P0_DIRECT_FC_C7");
|
|
g_direct_fc_c7 = (e7 && *e7) ? ((*e7 == '0') ? 0 : 1) : 0;
|
|
}
|
|
if (__builtin_expect((g_direct_fc && class_idx == 5) ||
|
|
(g_direct_fc_c7 && class_idx == 7), 0)) {
|
|
int room = tiny_fc_room(class_idx);
|
|
if (room <= 0) return 0;
|
|
|
|
uint32_t rmt = atomic_load_explicit(
|
|
&tls->ss->remote_counts[tls->slab_idx], memory_order_relaxed);
|
|
static int g_drain_th = -1;
|
|
if (__builtin_expect(g_drain_th == -1, 0)) {
|
|
const char* e = getenv("HAKMEM_TINY_P0_DRAIN_THRESH");
|
|
int v = (e && *e) ? atoi(e) : 64;
|
|
g_drain_th = (v < 0) ? 0 : v;
|
|
}
|
|
if (rmt >= (uint32_t)g_drain_th) {
|
|
static int no_drain = -1;
|
|
if (__builtin_expect(no_drain == -1, 0)) {
|
|
const char* e = getenv("HAKMEM_TINY_P0_NO_DRAIN");
|
|
no_drain = (e && *e && *e != '0') ? 1 : 0;
|
|
}
|
|
if (!no_drain) {
|
|
_ss_remote_drain_to_freelist_unsafe(
|
|
tls->ss, tls->slab_idx, tls->meta);
|
|
}
|
|
}
|
|
|
|
void* out[128];
|
|
int produced = 0;
|
|
TinySlabMeta* m = tls->meta;
|
|
size_t bs = tiny_stride_for_class(class_idx);
|
|
uint8_t* base = tls->slab_base
|
|
? tls->slab_base
|
|
: tiny_slab_base_for_geometry(tls->ss, tls->slab_idx);
|
|
while (produced < room) {
|
|
if (m->freelist) {
|
|
void* p = m->freelist;
|
|
m->freelist = tiny_next_read(class_idx, p);
|
|
m->used++;
|
|
out[produced++] = p;
|
|
} else if (m->carved < m->capacity) {
|
|
void* p = (void*)(base + ((size_t)m->carved * bs));
|
|
m->carved++;
|
|
m->used++;
|
|
out[produced++] = p;
|
|
} else {
|
|
if (!superslab_refill(class_idx)) break;
|
|
tls = &g_tls_slabs[class_idx];
|
|
m = tls->meta;
|
|
base = tls->slab_base
|
|
? tls->slab_base
|
|
: tiny_slab_base_for(tls->ss, tls->slab_idx);
|
|
}
|
|
}
|
|
if (produced > 0) {
|
|
ss_active_add(tls->ss, (uint32_t)produced);
|
|
(void)tiny_fc_push_bulk(class_idx, out, produced);
|
|
return produced;
|
|
}
|
|
// fallthrough to regular path
|
|
}
|
|
} while (0);
|
|
|
|
uint32_t sll_cap = sll_cap_for_class(class_idx, (uint32_t)TINY_TLS_MAG_CAP);
|
|
int room = (int)sll_cap - (int)g_tls_sll_count[class_idx];
|
|
if (room <= 0) {
|
|
#if HAKMEM_DEBUG_COUNTERS
|
|
g_rf_early_no_room[class_idx]++;
|
|
#endif
|
|
return 0;
|
|
}
|
|
|
|
uint32_t want = (uint32_t)max_take;
|
|
if (want > (uint32_t)room) want = (uint32_t)room;
|
|
if (want == 0) {
|
|
#if HAKMEM_DEBUG_COUNTERS
|
|
g_rf_early_want_zero[class_idx]++;
|
|
#endif
|
|
return 0;
|
|
}
|
|
|
|
size_t bs = tiny_stride_for_class(class_idx);
|
|
int total_taken = 0;
|
|
|
|
while (want > 0) {
|
|
uintptr_t ss_base = 0;
|
|
uintptr_t ss_limit = 0;
|
|
if (tls->ss && tls->slab_idx >= 0) {
|
|
uint8_t* slab_base =
|
|
tiny_slab_base_for_geometry(tls->ss, tls->slab_idx);
|
|
ss_base = (uintptr_t)slab_base;
|
|
ss_limit = ss_base + tiny_usable_bytes_for_slab(tls->slab_idx);
|
|
}
|
|
|
|
if (tls->ss && tls->slab_idx >= 0) {
|
|
uint32_t remote_count = atomic_load_explicit(
|
|
&tls->ss->remote_counts[tls->slab_idx], memory_order_relaxed);
|
|
if (remote_count > 0) {
|
|
static int no_drain = -1;
|
|
if (__builtin_expect(no_drain == -1, 0)) {
|
|
const char* e = getenv("HAKMEM_TINY_P0_NO_DRAIN");
|
|
no_drain = (e && *e && *e != '0') ? 1 : 0;
|
|
}
|
|
if (!no_drain) {
|
|
_ss_remote_drain_to_freelist_unsafe(tls->ss, tls->slab_idx, meta);
|
|
}
|
|
}
|
|
}
|
|
|
|
TinyRefillChain chain;
|
|
uint32_t from_freelist = trc_pop_from_freelist(
|
|
meta, class_idx, ss_base, ss_limit, bs, want, &chain);
|
|
if (from_freelist > 0) {
|
|
trc_splice_to_sll(
|
|
class_idx, &chain,
|
|
&g_tls_sll_head[class_idx],
|
|
&g_tls_sll_count[class_idx]);
|
|
ss_active_add(tls->ss, from_freelist);
|
|
meta->used = (uint16_t)((uint32_t)meta->used + from_freelist);
|
|
#if HAKMEM_DEBUG_COUNTERS
|
|
extern unsigned long long g_rf_freelist_items[];
|
|
g_rf_freelist_items[class_idx] += from_freelist;
|
|
#endif
|
|
total_taken += from_freelist;
|
|
want -= from_freelist;
|
|
if (want == 0) break;
|
|
}
|
|
|
|
if (meta->carved >= meta->capacity) {
|
|
if (!superslab_refill(class_idx)) break;
|
|
tls = &g_tls_slabs[class_idx];
|
|
meta = tls->meta;
|
|
if (!meta) break;
|
|
continue;
|
|
}
|
|
|
|
uint32_t available = meta->capacity - meta->carved;
|
|
uint32_t batch = want;
|
|
if (batch > available) batch = available;
|
|
if (batch == 0) break;
|
|
|
|
uint8_t* slab_base = tls->slab_base
|
|
? tls->slab_base
|
|
: tiny_slab_base_for(tls->ss, tls->slab_idx);
|
|
|
|
TinyRefillChain carve;
|
|
trc_linear_carve(slab_base, bs, meta, batch, class_idx, &carve);
|
|
trc_splice_to_sll(
|
|
class_idx, &carve,
|
|
&g_tls_sll_head[class_idx],
|
|
&g_tls_sll_count[class_idx]);
|
|
ss_active_add(tls->ss, batch);
|
|
#if HAKMEM_DEBUG_COUNTERS
|
|
extern unsigned long long g_rf_carve_items[];
|
|
g_rf_carve_items[class_idx] += batch;
|
|
#endif
|
|
total_taken += batch;
|
|
want -= batch;
|
|
}
|
|
|
|
#if HAKMEM_DEBUG_COUNTERS
|
|
g_rf_hit_slab[class_idx]++;
|
|
#endif
|
|
|
|
if (tls->ss && p0_should_log()) {
|
|
uint32_t active_after = atomic_load_explicit(
|
|
&tls->ss->total_active_blocks, memory_order_relaxed);
|
|
int32_t delta =
|
|
(int32_t)active_after - (int32_t)active_before;
|
|
fprintf(stderr,
|
|
"[P0_COUNTER] cls=%d slab=%d taken=%d active_delta=%d\n",
|
|
class_idx, tls->slab_idx, total_taken, delta);
|
|
}
|
|
|
|
return total_taken;
|
|
}
|
|
|
|
#endif // HAKMEM_TINY_P0_BATCH_REFILL
|
|
#endif // HAKMEM_TINY_REFILL_P0_INC_H
|