Files
hakmem/core/hakmem_tiny_hotmag.inc.h

150 lines
5.2 KiB
C
Raw Permalink Normal View History

// Inline helpers for Tiny HotMag (small TLS magazine for hot classes)
// This header is textually included from hakmem_tiny.c after the following
// symbols are defined:
// - static int g_hotmag_enable;
// - static __thread TinyHotMag g_tls_hot_mag[TINY_NUM_CLASSES];
// - tiny_mag_init_if_needed(int)
// - g_tls_sll_head[], g_tls_sll_count[], g_tls_mags[]
Phase E3-FINAL: Fix Box API offset bugs - ALL classes now use correct offsets ## Root Cause Analysis (GPT5) **Physical Layout Constraints**: - Class 0: 8B = [1B header][7B payload] → offset 1 = 9B needed = ❌ IMPOSSIBLE - Class 1-6: >=16B = [1B header][15B+ payload] → offset 1 = ✅ POSSIBLE - Class 7: 1KB → offset 0 (compatibility) **Correct Specification**: - HAKMEM_TINY_HEADER_CLASSIDX != 0: - Class 0, 7: next at offset 0 (overwrites header when on freelist) - Class 1-6: next at offset 1 (after header) - HAKMEM_TINY_HEADER_CLASSIDX == 0: - All classes: next at offset 0 **Previous Bug**: - Attempted "ALL classes offset 1" unification - Class 0 with offset 1 caused immediate SEGV (9B > 8B block size) - Mixed 2-arg/3-arg API caused confusion ## Fixes Applied ### 1. Restored 3-Argument Box API (core/box/tiny_next_ptr_box.h) ```c // Correct signatures void tiny_next_write(int class_idx, void* base, void* next_value) void* tiny_next_read(int class_idx, const void* base) // Correct offset calculation size_t offset = (class_idx == 0 || class_idx == 7) ? 0 : 1; ``` ### 2. Updated 123+ Call Sites Across 34 Files - hakmem_tiny_hot_pop_v4.inc.h (4 locations) - hakmem_tiny_fastcache.inc.h (3 locations) - hakmem_tiny_tls_list.h (12 locations) - superslab_inline.h (5 locations) - tiny_fastcache.h (3 locations) - ptr_trace.h (macro definitions) - tls_sll_box.h (2 locations) - + 27 additional files Pattern: `tiny_next_read(base)` → `tiny_next_read(class_idx, base)` Pattern: `tiny_next_write(base, next)` → `tiny_next_write(class_idx, base, next)` ### 3. Added Sentinel Detection Guards - tiny_fast_push(): Block nodes with sentinel in ptr or ptr->next - tls_list_push(): Block nodes with sentinel in ptr or ptr->next - Defense-in-depth against remote free sentinel leakage ## Verification (GPT5 Report) **Test Command**: `./out/release/bench_random_mixed_hakmem --iterations=70000` **Results**: - ✅ Main loop completed successfully - ✅ Drain phase completed successfully - ✅ NO SEGV (previous crash at iteration 66151 is FIXED) - ℹ️ Final log: "tiny_alloc(1024) failed" is normal fallback to Mid/ACE layers **Analysis**: - Class 0 immediate SEGV: ✅ RESOLVED (correct offset 0 now used) - 66K iteration crash: ✅ RESOLVED (offset consistency fixed) - Box API conflicts: ✅ RESOLVED (unified 3-arg API) ## Technical Details ### Offset Logic Justification ``` Class 0: 8B block → next pointer (8B) fits ONLY at offset 0 Class 1: 16B block → next pointer (8B) fits at offset 1 (after 1B header) Class 2: 32B block → next pointer (8B) fits at offset 1 ... Class 6: 512B block → next pointer (8B) fits at offset 1 Class 7: 1024B block → offset 0 for legacy compatibility ``` ### Files Modified (Summary) - Core API: `box/tiny_next_ptr_box.h` - Hot paths: `hakmem_tiny_hot_pop*.inc.h`, `tiny_fastcache.h` - TLS layers: `hakmem_tiny_tls_list.h`, `hakmem_tiny_tls_ops.h` - SuperSlab: `superslab_inline.h`, `tiny_superslab_*.inc.h` - Refill: `hakmem_tiny_refill.inc.h`, `tiny_refill_opt.h` - Free paths: `tiny_free_magazine.inc.h`, `tiny_superslab_free.inc.h` - Documentation: Multiple Phase E3 reports ## Remaining Work None for Box API offset bugs - all structural issues resolved. Future enhancements (non-critical): - Periodic `grep -R '*(void**)' core/` to detect direct pointer access violations - Enforce Box API usage via static analysis - Document offset rationale in architecture docs 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com>
2025-11-13 06:50:20 +09:00
#include "box/tiny_next_ptr_box.h" // Box API: Next pointer read/write
static inline int hkm_is_hot_class(int class_idx) {
return class_idx >= 0 && class_idx <= 3 && g_hotmag_class_en[class_idx];
}
static inline uint16_t hotmag_effective_cap(int class_idx) {
if (!hkm_is_hot_class(class_idx)) return 0;
int cap = g_hotmag_cap_current[class_idx];
if (!g_hotmag_cap_locked[class_idx] || cap <= 0) {
cap = g_hotmag_cap_default;
}
if (cap < 16) cap = 16;
if (cap > 1024) cap = 1024;
return (uint16_t)cap;
}
static inline uint16_t hotmag_refill_target(int class_idx) {
if (!hkm_is_hot_class(class_idx)) return 0;
int refill = g_hotmag_refill_current[class_idx];
if (!g_hotmag_refill_locked[class_idx] && refill <= 0) {
refill = g_hotmag_refill_default;
}
if (refill < 0) refill = 0;
uint16_t cap = hotmag_effective_cap(class_idx);
if ((uint16_t)refill > cap) refill = cap;
return (uint16_t)refill;
}
static inline void hotmag_init_if_needed(int class_idx) {
if (!g_hotmag_enable || !hkm_is_hot_class(class_idx)) return;
TinyHotMag* hm = &g_tls_hot_mag[class_idx];
uint16_t desired = hotmag_effective_cap(class_idx);
if (hm->cap != desired) {
hm->cap = desired;
if (hm->top > hm->cap) hm->top = hm->cap;
}
}
static inline void* hotmag_pop(int class_idx) {
if (!g_hotmag_enable || !hkm_is_hot_class(class_idx)) return NULL;
TinyHotMag* hm = &g_tls_hot_mag[class_idx];
if (__builtin_expect(hm->top > 0, 1)) {
return hm->slots[--hm->top];
}
return NULL;
}
static inline int hotmag_push(int class_idx, void* ptr) {
if (!g_hotmag_enable || !hkm_is_hot_class(class_idx)) return 0;
TinyHotMag* hm = &g_tls_hot_mag[class_idx];
hotmag_init_if_needed(class_idx);
if (__builtin_expect(hm->top < hm->cap, 1)) {
hm->slots[hm->top++] = ptr;
return 1;
}
return 0;
}
static inline int hotmag_refill_from_sll(int class_idx, int max_take) {
if (!g_hotmag_enable || !hkm_is_hot_class(class_idx) || max_take <= 0) return 0;
TinyHotMag* hm = &g_tls_hot_mag[class_idx];
int room = (int)hm->cap - (int)hm->top;
if (room <= 0) return 0;
if (max_take > room) max_take = room;
int taken = 0;
while (taken < max_take) {
// CRITICAL: Use Box TLS-SLL API to avoid race condition (rbp=0xa0 SEGV)
void* head = NULL;
if (!tls_sll_pop(class_idx, &head)) break;
hm->slots[hm->top++] = head;
taken++;
}
return taken;
}
static inline int hotmag_refill_from_mag(int class_idx, int max_take) {
if (!g_hotmag_enable || !hkm_is_hot_class(class_idx) || max_take <= 0) return 0;
tiny_mag_init_if_needed(class_idx);
TinyTLSMag* mag = &g_tls_mags[class_idx];
TinyHotMag* hm = &g_tls_hot_mag[class_idx];
int room = (int)hm->cap - (int)hm->top;
if (room <= 0 || mag->top <= 0) return 0;
if (max_take > room) max_take = room;
int take = mag->top < max_take ? mag->top : max_take;
for (int i = 0; i < take; i++) {
hm->slots[hm->top++] = mag->items[--mag->top].ptr;
}
return take;
}
static inline int hotmag_try_refill(int class_idx, TinyHotMag* hm) {
if (!g_hotmag_enable || !hkm_is_hot_class(class_idx)) return 0;
uint16_t target = hotmag_refill_target(class_idx);
if (target == 0) return 0;
int room = (int)hm->cap - (int)hm->top;
if (room <= 0) return 0;
if (target > (uint16_t)room) target = (uint16_t)room;
int filled = 0;
if (g_tls_list_enable && target > 0) {
TinyTLSList* tls = &g_tls_lists[class_idx];
uint32_t seq = atomic_load_explicit(&g_tls_param_seq[class_idx], memory_order_relaxed);
if (__builtin_expect(seq != g_tls_param_seen[class_idx], 0)) {
tiny_tls_refresh_params(class_idx, tls);
}
void* chain_head = NULL;
void* chain_tail = NULL;
uint32_t taken = tls_list_bulk_take(tls, target, &chain_head, &chain_tail, class_idx);
if (taken > 0u) {
void* node = chain_head;
for (uint32_t i = 0; i < taken && node; i++) {
Phase E3-FINAL: Fix Box API offset bugs - ALL classes now use correct offsets ## Root Cause Analysis (GPT5) **Physical Layout Constraints**: - Class 0: 8B = [1B header][7B payload] → offset 1 = 9B needed = ❌ IMPOSSIBLE - Class 1-6: >=16B = [1B header][15B+ payload] → offset 1 = ✅ POSSIBLE - Class 7: 1KB → offset 0 (compatibility) **Correct Specification**: - HAKMEM_TINY_HEADER_CLASSIDX != 0: - Class 0, 7: next at offset 0 (overwrites header when on freelist) - Class 1-6: next at offset 1 (after header) - HAKMEM_TINY_HEADER_CLASSIDX == 0: - All classes: next at offset 0 **Previous Bug**: - Attempted "ALL classes offset 1" unification - Class 0 with offset 1 caused immediate SEGV (9B > 8B block size) - Mixed 2-arg/3-arg API caused confusion ## Fixes Applied ### 1. Restored 3-Argument Box API (core/box/tiny_next_ptr_box.h) ```c // Correct signatures void tiny_next_write(int class_idx, void* base, void* next_value) void* tiny_next_read(int class_idx, const void* base) // Correct offset calculation size_t offset = (class_idx == 0 || class_idx == 7) ? 0 : 1; ``` ### 2. Updated 123+ Call Sites Across 34 Files - hakmem_tiny_hot_pop_v4.inc.h (4 locations) - hakmem_tiny_fastcache.inc.h (3 locations) - hakmem_tiny_tls_list.h (12 locations) - superslab_inline.h (5 locations) - tiny_fastcache.h (3 locations) - ptr_trace.h (macro definitions) - tls_sll_box.h (2 locations) - + 27 additional files Pattern: `tiny_next_read(base)` → `tiny_next_read(class_idx, base)` Pattern: `tiny_next_write(base, next)` → `tiny_next_write(class_idx, base, next)` ### 3. Added Sentinel Detection Guards - tiny_fast_push(): Block nodes with sentinel in ptr or ptr->next - tls_list_push(): Block nodes with sentinel in ptr or ptr->next - Defense-in-depth against remote free sentinel leakage ## Verification (GPT5 Report) **Test Command**: `./out/release/bench_random_mixed_hakmem --iterations=70000` **Results**: - ✅ Main loop completed successfully - ✅ Drain phase completed successfully - ✅ NO SEGV (previous crash at iteration 66151 is FIXED) - ℹ️ Final log: "tiny_alloc(1024) failed" is normal fallback to Mid/ACE layers **Analysis**: - Class 0 immediate SEGV: ✅ RESOLVED (correct offset 0 now used) - 66K iteration crash: ✅ RESOLVED (offset consistency fixed) - Box API conflicts: ✅ RESOLVED (unified 3-arg API) ## Technical Details ### Offset Logic Justification ``` Class 0: 8B block → next pointer (8B) fits ONLY at offset 0 Class 1: 16B block → next pointer (8B) fits at offset 1 (after 1B header) Class 2: 32B block → next pointer (8B) fits at offset 1 ... Class 6: 512B block → next pointer (8B) fits at offset 1 Class 7: 1024B block → offset 0 for legacy compatibility ``` ### Files Modified (Summary) - Core API: `box/tiny_next_ptr_box.h` - Hot paths: `hakmem_tiny_hot_pop*.inc.h`, `tiny_fastcache.h` - TLS layers: `hakmem_tiny_tls_list.h`, `hakmem_tiny_tls_ops.h` - SuperSlab: `superslab_inline.h`, `tiny_superslab_*.inc.h` - Refill: `hakmem_tiny_refill.inc.h`, `tiny_refill_opt.h` - Free paths: `tiny_free_magazine.inc.h`, `tiny_superslab_free.inc.h` - Documentation: Multiple Phase E3 reports ## Remaining Work None for Box API offset bugs - all structural issues resolved. Future enhancements (non-critical): - Periodic `grep -R '*(void**)' core/` to detect direct pointer access violations - Enforce Box API usage via static analysis - Document offset rationale in architecture docs 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com>
2025-11-13 06:50:20 +09:00
// Header-aware next from TLS list chain (Box API handles offset)
void* next = tiny_next_read(class_idx, node);
hm->slots[hm->top++] = node;
node = next;
}
filled += (int)taken;
room = (int)hm->cap - (int)hm->top;
target -= (uint16_t)taken;
}
}
if (target > 0 && g_tls_sll_enable && room > 0) {
int from_sll = hotmag_refill_from_sll(class_idx, target);
filled += from_sll;
room -= from_sll;
if (room < 0) room = 0;
if (target > from_sll) target -= (uint16_t)((from_sll < 0) ? 0 : from_sll);
else target = 0;
}
if (target > 0 && room > 0) {
int from_mag = hotmag_refill_from_mag(class_idx, target);
filled += from_mag;
}
return filled;
}