PyCyBase 0.1.10__cp315-cp315-win_amd64.whl

cbase/allocator_protocol/c_heap_allocator.h

@@ -0,0 +1,554 @@
+#ifndef AP_HEAP_C_HEAP_ALLOCATOR_H
+#define AP_HEAP_C_HEAP_ALLOCATOR_H
+
+#include <errno.h>
+#include <stddef.h>
+#include <stdint.h>
+#include <stdlib.h>
+#include <string.h>
+
+#ifdef _WIN32
+#include "cbase/nt/pthread_nt_compat.h"
+#include <intrin.h>
+#else
+#include <pthread.h>
+#endif
+
+// ========== Configuration ==========
+
+#ifndef AP_HEAP_AUTOPAGE_CAPACITY
+#define AP_HEAP_AUTOPAGE_CAPACITY (64 * 1024) /* 64 KiB */
+#endif
+
+#ifndef AP_HEAP_AUTOPAGE_CAPACITY_MAX
+#define AP_HEAP_AUTOPAGE_CAPACITY_MAX (16 * 1024 * 1024) /* 16 MiB */
+#endif
+
+#ifndef AP_HEAP_AUTOPAGE_ALIGNMENT
+#define AP_HEAP_AUTOPAGE_ALIGNMENT (4 * 1024) /* 4 KiB */
+#endif
+
+/* Size-binned free lists: exact 8-byte-granular bins for capacities up to
+ * AP_HEAP_EXACT_BIN_COUNT * 8 bytes, pow2-class bins above. */
+#ifndef AP_HEAP_EXACT_BIN_COUNT
+#define AP_HEAP_EXACT_BIN_COUNT 8192 /* capacity <= 64 KiB gets an exact bin */
+#endif
+
+#ifndef AP_HEAP_LARGE_BIN_COUNT
+#define AP_HEAP_LARGE_BIN_COUNT 9 /* pow2-class bins for capacity > 64 KiB */
+#endif
+
+#define AP_HEAP_BIN_COUNT (AP_HEAP_EXACT_BIN_COUNT + AP_HEAP_LARGE_BIN_COUNT + 1)
+
+/* Miss-path page scaling cap: page sizes double from the current page up
+ * to this cap; a request larger than the cap forces a bigger page.
+ * Used only when AP_HEAP_PAGE_FIT_TO_REQUEST is 0. */
+#ifndef AP_HEAP_PAGE_EXTEND_MAX
+#define AP_HEAP_PAGE_EXTEND_MAX (128 * 1024 * 1024) /* 128 MiB */
+#endif
+
+/* Miss-path page policy: 0 (default) = capped power-of-2 page scaling
+ * (see AP_HEAP_PAGE_EXTEND_MAX); 1 = fit the page to the request (floor
+ * = autopage_capacity) so rare bin misses never extend oversized pages. */
+#ifndef AP_HEAP_PAGE_FIT_TO_REQUEST
+#define AP_HEAP_PAGE_FIT_TO_REQUEST 0
+#endif
+
+/* On an exact-bin miss, probe up to this many LARGER exact bins before
+ * the page path.  A slightly larger block satisfies the request; on free
+ * it returns to its own bin (recycling unaffected), and the fragmentation
+ * is bounded by the probe range.  0 disables the probe. */
+#ifndef AP_HEAP_EXACT_BIN_PROBE_COUNT
+#define AP_HEAP_EXACT_BIN_PROBE_COUNT 2
+#endif
+
+// ========== Heap Allocator Structs ==========
+
+typedef struct heap_memory_block {
+    size_t                    capacity;
+    size_t                    size;
+    struct heap_memory_block* next_free;
+    struct heap_memory_block* next_allocated;
+    struct heap_page*         parent_page;
+    char                      buffer[];
+} heap_memory_block;
+
+typedef struct heap_page {
+    size_t                    capacity;
+    size_t                    occupied;
+    struct heap_allocator*    allocator;
+    struct heap_page*         prev;
+    struct heap_memory_block* allocated;
+    char                      buffer[];
+} heap_page;
+
+typedef struct heap_allocator {
+    pthread_mutex_t           lock;
+    size_t                    mapped_pages;
+    struct heap_page*         active_page;
+    size_t                    autopage_capacity;
+    size_t                    autopage_capacity_max;
+    size_t                    autopage_alignment;
+    struct heap_memory_block* bins[AP_HEAP_BIN_COUNT];
+} heap_allocator;
+
+// ========== Utility Functions ==========
+
+static inline size_t c_heap_page_roundup(heap_allocator* allocator, size_t size) {
+    return (size + allocator->autopage_alignment - 1) & ~(allocator->autopage_alignment - 1);
+}
+
+static inline size_t c_heap_block_roundup(size_t size) {
+    return (size + sizeof(void*) - 1) & ~(sizeof(void*) - 1);
+}
+
+/* ceil(log2(v)) for 0 < v <= UINT32_MAX (portable; intrinsics on MSVC/GCC-Clang).
+ * 32-bit is sufficient: the bin structure caps block capacities well below
+ * 4 GiB, so a narrower parameter type truthfully advertises the domain. */
+static inline size_t c_heap_block_ceil_log2(uint32_t v) {
+    size_t floor;
+#if defined(_MSC_VER)
+    unsigned long idx = 0;
+    _BitScanReverse(&idx, (unsigned long) v);
+    floor = (size_t) idx;
+#else
+    floor = (size_t) (31u - __builtin_clz((unsigned int) v));
+#endif
+    return floor + (v != ((size_t) 1u << floor));
+}
+
+/* Size bin for a block capacity (always a multiple of 8).
+ * Exact bins: index = capacity / 8 (1..AP_HEAP_EXACT_BIN_COUNT), so any
+ * block in bin(cap_net) has capacity == cap_net and always fits.
+ * Larger capacities: pow2-class bins keyed by ceil(log2(capacity))
+ * (indices AP_HEAP_EXACT_BIN_COUNT+1..); the last bin is the overflow
+ * class.  In-bin first-fit is used there (blocks can be up to one class
+ * smaller than the request). */
+static inline size_t c_heap_block_bin(size_t capacity) {
+    if (capacity <= (size_t) AP_HEAP_EXACT_BIN_COUNT * 8u) {
+        return capacity >> 3;
+    }
+    size_t idx = (size_t) AP_HEAP_EXACT_BIN_COUNT + (c_heap_block_ceil_log2(capacity) - 16u);
+    if (idx > (size_t) AP_HEAP_EXACT_BIN_COUNT + AP_HEAP_LARGE_BIN_COUNT) {
+        idx = (size_t) AP_HEAP_EXACT_BIN_COUNT + AP_HEAP_LARGE_BIN_COUNT;
+    }
+    return idx;
+}
+
+static inline void c_heap_page_reclaim(heap_allocator* allocator, heap_page* page) {
+    if (!allocator || !page) {
+        errno = EINVAL;
+        return;
+    }
+
+    heap_memory_block** prevp = &page->allocated;
+    while (*prevp) {
+        heap_memory_block* block = *prevp;
+        if (block->size != 0) {
+            break;
+        }
+
+        *prevp = block->next_allocated;
+        block->next_allocated = NULL;
+
+        heap_memory_block** free_prev = &allocator->bins[c_heap_block_bin(block->capacity)];
+        while (*free_prev && *free_prev != block) {
+            free_prev = &(*free_prev)->next_free;
+        }
+        if (*free_prev == block) {
+            *free_prev = block->next_free;
+        }
+        block->next_free = NULL;
+
+        size_t cap_total = block->capacity + sizeof(heap_memory_block);
+        if (page->occupied >= cap_total) {
+            page->occupied -= cap_total;
+        }
+    }
+}
+
+// ========== Public API Functions ==========
+
+static inline heap_page* c_heap_allocator_extend(heap_allocator* allocator, size_t capacity, pthread_mutex_t* lock) {
+    if (!allocator) {
+        errno = EINVAL;
+        return NULL;
+    }
+
+    uint8_t locked = 0;
+
+    if (lock) {
+        int ret = pthread_mutex_lock(lock);
+        if (ret != 0) {
+            errno = ret;
+            return NULL;
+        }
+        locked = 1;
+    }
+
+    if (capacity == 0) {
+        if (!allocator->active_page) {
+            capacity = allocator->autopage_capacity;
+        }
+        else {
+            size_t prev_cap = allocator->active_page->capacity;
+            capacity = prev_cap * 2;
+            if (capacity < allocator->autopage_capacity) {
+                capacity = allocator->autopage_capacity;
+            }
+            else if (capacity > allocator->autopage_capacity_max) {
+                capacity = allocator->autopage_capacity_max;
+            }
+        }
+    }
+
+    size_t     total_capacity = c_heap_page_roundup(allocator, capacity);
+
+    heap_page* page = (heap_page*) calloc(1, total_capacity);
+
+    if (!page) {
+        if (locked) pthread_mutex_unlock(lock);
+        return NULL;
+    }
+
+    page->capacity = total_capacity;
+    page->occupied = sizeof(heap_page);
+    page->allocator = allocator;
+    page->allocated = NULL;
+    page->prev = allocator->active_page;
+    allocator->active_page = page;
+    allocator->mapped_pages++;
+
+    if (locked) pthread_mutex_unlock(lock);
+    return page;
+}
+
+static inline heap_allocator* c_heap_allocator_new() {
+    heap_allocator* allocator = (heap_allocator*) calloc(1, sizeof(heap_allocator));
+    if (!allocator) {
+        return NULL;
+    }
+
+    if (pthread_mutex_init(&allocator->lock, NULL) != 0) {
+        free(allocator);
+        return NULL;
+    }
+
+    allocator->mapped_pages = 0;
+    allocator->active_page = NULL;
+    allocator->autopage_capacity = AP_HEAP_AUTOPAGE_CAPACITY;
+    allocator->autopage_capacity_max = AP_HEAP_AUTOPAGE_CAPACITY_MAX;
+    allocator->autopage_alignment = AP_HEAP_AUTOPAGE_ALIGNMENT;
+
+    return allocator;
+}
+
+static inline void c_heap_allocator_free(heap_allocator* allocator) {
+    if (!allocator) {
+        return;
+    }
+
+    heap_page* page = allocator->active_page;
+    while (page) {
+        heap_page* prev = page->prev;
+        free(page);
+        page = prev;
+    }
+
+    pthread_mutex_destroy(&allocator->lock);
+    free(allocator);
+}
+
+static inline void* c_heap_calloc(heap_allocator* allocator, size_t size, pthread_mutex_t* lock) {
+    if (!allocator || size == 0) {
+        errno = EINVAL;
+        return NULL;
+    }
+
+    size_t           cap_net = c_heap_block_roundup(size);
+    size_t           overhead = sizeof(heap_memory_block);
+    size_t           cap_total = cap_net + overhead;
+
+    uint8_t          locked = 0;
+    pthread_mutex_t* builtin_lock = &allocator->lock;
+    pthread_mutex_t* child_lock = &allocator->lock;
+    if (lock) {
+        if (lock == builtin_lock) {
+            child_lock = NULL;
+        }
+        int ret = pthread_mutex_lock(lock);
+        if (ret != 0) {
+            errno = ret;
+            return NULL;
+        }
+        locked = 1;
+    }
+    else {
+        child_lock = NULL;
+    }
+
+    heap_page* page = allocator->active_page;
+    if (!page) {
+        size_t target_cap = allocator->autopage_capacity;
+        while (target_cap < cap_total + sizeof(heap_page)) {
+            target_cap *= 2;
+        }
+
+        page = c_heap_allocator_extend(allocator, target_cap, child_lock);
+        if (!page) {
+            if (locked) pthread_mutex_unlock(lock);
+            return NULL;
+        }
+    }
+
+    if (page->occupied + cap_total > page->capacity) {
+        size_t target_cap = page->capacity;
+
+        if (target_cap < allocator->autopage_capacity) {
+            target_cap = allocator->autopage_capacity;
+        }
+        else if (target_cap < allocator->autopage_capacity_max) {
+            target_cap *= 2;
+        }
+
+        while (target_cap < cap_total + sizeof(heap_page)) {
+            target_cap *= 2;
+        }
+        page = c_heap_allocator_extend(allocator, target_cap, child_lock);
+        if (!page) {
+            if (locked) pthread_mutex_unlock(lock);
+            return NULL;
+        }
+    }
+
+    size_t             offset = page->occupied;
+    heap_memory_block* block = (heap_memory_block*) ((char*) page + offset);
+    block->capacity = cap_net;
+    block->size = size;
+    block->next_free = NULL;
+
+    block->parent_page = page;
+    block->next_allocated = page->allocated;
+    page->allocated = block;
+    page->occupied += cap_total;
+
+    if (locked) pthread_mutex_unlock(lock);
+
+    memset(block + 1, 0, cap_net);
+    return (void*) block->buffer;
+}
+
+static inline void* c_heap_request(heap_allocator* allocator, size_t size, int scan_all_pages, pthread_mutex_t* lock) {
+    if (!allocator || size == 0) {
+        errno = EINVAL;
+        return NULL;
+    }
+
+    size_t           cap_net = c_heap_block_roundup(size);
+    size_t           overhead = sizeof(heap_memory_block);
+    size_t           cap_total = cap_net + overhead;
+
+    uint8_t          locked = 0;
+    pthread_mutex_t* builtin_lock = &allocator->lock;
+    pthread_mutex_t* child_lock = &allocator->lock;
+    if (lock) {
+        if (lock == builtin_lock) {
+            child_lock = NULL;
+        }
+        int ret = pthread_mutex_lock(lock);
+        if (ret != 0) {
+            errno = ret;
+            return NULL;
+        }
+        locked = 1;
+    }
+    else {
+        child_lock = NULL;
+    }
+
+    /* Step 1: size-binned free-list reuse (see c_heap_block_bin).
+     * Exact bins always fit (capacity == cap_net) — O(1) head pop.
+     * Pow2-class bins use in-bin first-fit: bins are short and same-
+     * shaped, so the head is the common hit; larger-capacity blocks in
+     * the bin also satisfy smaller requests. */
+    size_t             bin = c_heap_block_bin(cap_net);
+    heap_memory_block* free_blk;
+    if (bin <= (size_t) AP_HEAP_EXACT_BIN_COUNT) {
+        size_t found = bin;
+        free_blk = allocator->bins[bin];
+        if (!free_blk && AP_HEAP_EXACT_BIN_PROBE_COUNT > 0) {
+            size_t probe_end = bin + (size_t) AP_HEAP_EXACT_BIN_PROBE_COUNT;
+            if (probe_end > (size_t) AP_HEAP_EXACT_BIN_COUNT) {
+                probe_end = (size_t) AP_HEAP_EXACT_BIN_COUNT;
+            }
+            for (size_t b = bin + 1; b <= probe_end; b++) {
+                free_blk = allocator->bins[b];
+                if (free_blk) {
+                    found = b;
+                    break;
+                }
+            }
+        }
+        if (free_blk) {
+            allocator->bins[found] = free_blk->next_free;
+            free_blk->next_free = NULL;
+            free_blk->size = size;
+            if (locked) pthread_mutex_unlock(lock);
+            memset(free_blk + 1, 0, cap_net);
+            return (void*) free_blk->buffer;
+        }
+    }
+    else {
+        heap_memory_block** prevp = &allocator->bins[bin];
+        free_blk = *prevp;
+        while (free_blk) {
+            if (free_blk->capacity >= cap_net) {
+                *prevp = free_blk->next_free;
+                free_blk->next_free = NULL;
+                free_blk->size = size;
+                if (locked) pthread_mutex_unlock(lock);
+                memset(free_blk + 1, 0, cap_net);
+                return (void*) free_blk->buffer;
+            }
+            prevp = &free_blk->next_free;
+            free_blk = free_blk->next_free;
+        }
+    }
+
+    heap_page* target_page = NULL;
+    if (scan_all_pages) {
+        heap_page* iter = allocator->active_page;
+        while (iter) {
+            if (iter->occupied + cap_total <= iter->capacity) {
+                target_page = iter;
+                break;
+            }
+            iter = iter->prev;
+        }
+    }
+    else if (allocator->active_page) {
+        heap_page* meta = allocator->active_page;
+        if (meta && meta->occupied + cap_total <= meta->capacity) {
+            target_page = allocator->active_page;
+        }
+    }
+
+    if (!target_page) {
+        size_t target_cap;
+#if AP_HEAP_PAGE_FIT_TO_REQUEST
+        /* Fit the page to the request (floor = autopage_capacity): rare
+         * bin misses must not extend oversized pages. */
+        target_cap = allocator->autopage_capacity;
+#else
+        /* Page scaling with a cap: double from the current page size up to
+         * AP_HEAP_PAGE_EXTEND_MAX.  A page larger than the cap (e.g. from
+         * a huge request) scales back down to the cap.  Requests larger
+         * than the cap force a bigger page via the fit loop below. */
+        heap_page* current = allocator->active_page;
+        if (!current) {
+            target_cap = allocator->autopage_capacity;
+        }
+        else {
+            target_cap = current->capacity;
+            if (target_cap < allocator->autopage_capacity) {
+                target_cap = allocator->autopage_capacity;
+            }
+            else if (target_cap > (size_t) AP_HEAP_PAGE_EXTEND_MAX) {
+                target_cap = (size_t) AP_HEAP_PAGE_EXTEND_MAX;
+            }
+            else if (target_cap < (size_t) AP_HEAP_PAGE_EXTEND_MAX) {
+                target_cap *= 2;
+                if (target_cap > (size_t) AP_HEAP_PAGE_EXTEND_MAX) {
+                    target_cap = (size_t) AP_HEAP_PAGE_EXTEND_MAX;
+                }
+            }
+        }
+#endif
+        while (target_cap < cap_total + sizeof(heap_page)) {
+            target_cap *= 2;
+        }
+
+        target_page = c_heap_allocator_extend(allocator, target_cap, child_lock);
+        if (!target_page) {
+            if (locked) pthread_mutex_unlock(lock);
+            return NULL;
+        }
+    }
+
+    size_t             offset = target_page->occupied;
+    heap_memory_block* block = (heap_memory_block*) ((char*) target_page + offset);
+    block->capacity = cap_net;
+    block->size = size;
+    block->next_free = NULL;
+
+    block->parent_page = target_page;
+    block->next_allocated = target_page->allocated;
+    target_page->allocated = block;
+
+    target_page->occupied += cap_total;
+
+    if (locked) pthread_mutex_unlock(lock);
+
+    memset(block + 1, 0, cap_net);
+    return (void*) block->buffer;
+}
+
+static inline void c_heap_free(void* ptr, pthread_mutex_t* lock) {
+    if (!ptr) {
+        errno = EINVAL;
+        return;
+    }
+
+    heap_memory_block* block = (heap_memory_block*) ((char*) ptr - sizeof(heap_memory_block));
+    heap_page*         page = block->parent_page;
+    if (!page || !page->allocator) {
+        errno = EINVAL;
+        return;
+    }
+
+    heap_allocator* allocator = page->allocator;
+
+    uint8_t         locked = 0;
+    if (lock) {
+        int ret = pthread_mutex_lock(lock);
+        if (ret != 0) {
+            errno = ret;
+            return;
+        }
+        locked = 1;
+    }
+
+    block->size = 0;
+    size_t bin = c_heap_block_bin(block->capacity);
+    block->next_free = allocator->bins[bin];
+    allocator->bins[bin] = block;
+
+    if (locked) pthread_mutex_unlock(lock);
+}
+
+static inline void c_heap_reclaim(heap_allocator* allocator, pthread_mutex_t* lock) {
+    if (!allocator) {
+        errno = EINVAL;
+        return;
+    }
+
+    uint8_t locked = 0;
+    if (lock) {
+        int ret = pthread_mutex_lock(lock);
+        if (ret != 0) {
+            errno = ret;
+            return;
+        }
+        locked = 1;
+    }
+
+    heap_page* page = allocator->active_page;
+    while (page) {
+        c_heap_page_reclaim(allocator, page);
+        page = page->prev;
+    }
+
+    if (locked) pthread_mutex_unlock(lock);
+}
+
+#endif /* AP_HEAP_C_HEAP_ALLOCATOR_H */

cbase/allocator_protocol/c_heap_allocator.pxd

@@ -0,0 +1,105 @@
+
+cdef extern from "cbase/allocator_protocol/c_heap_allocator.h":
+    const size_t AP_HEAP_AUTOPAGE_CAPACITY
+    const size_t AP_HEAP_AUTOPAGE_CAPACITY_MAX
+    const size_t AP_HEAP_AUTOPAGE_ALIGNMENT
+    const size_t AP_HEAP_EXACT_BIN_COUNT
+    const size_t AP_HEAP_LARGE_BIN_COUNT
+    const size_t AP_HEAP_BIN_COUNT
+    const size_t AP_HEAP_PAGE_EXTEND_MAX
+    const size_t AP_HEAP_PAGE_FIT_TO_REQUEST
+    const size_t AP_HEAP_EXACT_BIN_PROBE_COUNT
+
+    ctypedef struct pthread_mutex_t:
+        pass
+
+    ctypedef struct heap_memory_block:
+        size_t capacity
+        size_t size
+        heap_memory_block* next_free
+        heap_memory_block* next_allocated
+        heap_page* parent_page
+        char buffer[]
+
+    ctypedef struct heap_page:
+        size_t capacity
+        size_t occupied
+        heap_allocator* allocator
+        heap_page* prev
+        heap_memory_block* allocated
+        char buffer[]
+
+    ctypedef struct heap_allocator:
+        pthread_mutex_t lock
+        size_t mapped_pages
+        heap_page* active_page
+        size_t autopage_capacity
+        size_t autopage_capacity_max
+        size_t autopage_alignment
+        heap_memory_block* bins[AP_HEAP_BIN_COUNT]
+
+    int pthread_mutex_init(pthread_mutex_t* mutex, void* attr) noexcept nogil
+    int pthread_mutex_lock(pthread_mutex_t* mutex) noexcept nogil
+    int pthread_mutex_unlock(pthread_mutex_t* mutex) noexcept nogil
+    int pthread_mutex_destroy(pthread_mutex_t* mutex) noexcept nogil
+
+    size_t c_heap_page_roundup(heap_allocator* allocator, size_t size) noexcept nogil
+    size_t c_heap_block_roundup(size_t size) noexcept nogil
+    size_t c_heap_block_ceil_log2(unsigned int v) noexcept nogil
+    size_t c_heap_block_bin(size_t capacity) noexcept nogil
+    void c_heap_page_reclaim(heap_allocator* allocator, heap_page* page) noexcept nogil
+
+    heap_page* c_heap_allocator_extend(heap_allocator* allocator, size_t capacity, pthread_mutex_t* lock) noexcept nogil
+    heap_allocator* c_heap_allocator_new() noexcept nogil
+    void c_heap_allocator_free(heap_allocator* allocator) noexcept nogil
+    void* c_heap_calloc(heap_allocator* allocator, size_t size, pthread_mutex_t* lock) noexcept nogil
+    void* c_heap_request(heap_allocator* allocator, size_t size, int scan_all_pages, pthread_mutex_t* lock) noexcept nogil
+    void c_heap_free(void* ptr, pthread_mutex_t* lock) noexcept nogil
+    void c_heap_reclaim(heap_allocator* allocator, pthread_mutex_t* lock) noexcept nogil
+
+
+cdef class HeapMemoryPage:
+    cdef heap_page* page
+
+    @staticmethod
+    cdef inline HeapMemoryPage c_from_header(heap_page* header)
+
+    cpdef void reclaim(self)
+
+
+cdef class HeapMemoryBlock:
+    cdef heap_memory_block* block
+    cdef readonly bint owner
+
+    @staticmethod
+    cdef inline HeapMemoryBlock c_from_header(heap_memory_block* header, bint owner=?)
+
+
+cdef class HeapAllocator:
+    cdef heap_allocator* allocator
+    cdef readonly bint owner
+
+    @staticmethod
+    cdef HeapAllocator c_from_header(heap_allocator* header, bint owner=?)
+
+    cdef inline heap_page* c_extend(self, size_t capacity=?, pthread_mutex_t* lock=?)
+
+    cdef inline void* c_calloc(self, size_t size, pthread_mutex_t* lock=?)
+
+    cdef inline void* c_request(self, size_t size, int scan_all_pages=?, pthread_mutex_t* lock=?)
+
+    cdef inline void c_free(self, void* ptr, pthread_mutex_t* lock=?)
+
+    cpdef HeapMemoryPage extend(self, size_t capacity=?, bint with_lock=?)
+
+    cpdef HeapMemoryBlock calloc(self, size_t size, bint with_lock=?)
+
+    cpdef HeapMemoryBlock request(self, size_t size, bint with_lock=?, bint scan_all_pages=?)
+
+    cpdef void free(self, HeapMemoryBlock buffer, bint with_lock=?)
+
+    cpdef void reclaim(self, bint with_lock=?)
+
+
+cdef HeapAllocator ALLOCATOR
+cdef heap_allocator* C_ALLOCATOR