core/ngx_slab.c源文件分析
本章我们介绍一下Nginx中基于slab的内存分配机制。
1. 相关宏定义
/*
* Copyright (C) Igor Sysoev
* Copyright (C) Nginx, Inc.
*/
#include <ngx_config.h>
#include <ngx_core.h>
#define NGX_SLAB_PAGE_MASK 3
#define NGX_SLAB_PAGE 0
#define NGX_SLAB_BIG 1
#define NGX_SLAB_EXACT 2
#define NGX_SLAB_SMALL 3
#if (NGX_PTR_SIZE == 4)
#define NGX_SLAB_PAGE_FREE 0
#define NGX_SLAB_PAGE_BUSY 0xffffffff
#define NGX_SLAB_PAGE_START 0x80000000
#define NGX_SLAB_SHIFT_MASK 0x0000000f
#define NGX_SLAB_MAP_MASK 0xffff0000
#define NGX_SLAB_MAP_SHIFT 16
#define NGX_SLAB_BUSY 0xffffffff
#else /* (NGX_PTR_SIZE == 8) */
#define NGX_SLAB_PAGE_FREE 0
#define NGX_SLAB_PAGE_BUSY 0xffffffffffffffff
#define NGX_SLAB_PAGE_START 0x8000000000000000
#define NGX_SLAB_SHIFT_MASK 0x000000000000000f
#define NGX_SLAB_MAP_MASK 0xffffffff00000000
#define NGX_SLAB_MAP_SHIFT 32
#define NGX_SLAB_BUSY 0xffffffffffffffff
#endif
#if (NGX_DEBUG_MALLOC)
#define ngx_slab_junk(p, size) ngx_memset(p, 0xA5, size)
#elif (NGX_HAVE_DEBUG_MALLOC)
#define ngx_slab_junk(p, size) \
if (ngx_debug_malloc) ngx_memset(p, 0xA5, size)
#else
#define ngx_slab_junk(p, size)
#endif
1) 所分配的内存大小尺寸
这里用2位
来表示所分配的内存大小尺寸,2位最多可以表示4种尺寸:
-
NGX_SLAB_PAGE_MASK: 表示尺寸的掩码
-
NGX_SLAB_PAGE: 超大内存,大于等于ngx_slab_max_size
-
NGX_SLAB_BIG: 大块内存,大于ngx_slab_exact_size而小于ngx_slab_max_size
-
NGX_SLAB_EXACT: 中等内存,等于ngx_slab_exact_size
-
NGX_SLAB_SMALL: 小块内存,小于ngx_slab_exact_size
下面是exact size与max size的计算:
ngx_slab_max_size = ngx_pagesize / 2; ngx_slab_exact_size = ngx_pagesize / (8 * sizeof(uintptr_t));
2) 掩码设置
在当前我们使用的32bit Ubuntu环境中,NGX_PTR_SIZE
的值为4,因此:
-
NGX_SLAB_PAGE_FREE: 值为0, 用于指示当前page处于空闲状态
-
NGX_SLAB_PAGE_BUSY: 值为0xffffffff,用于指示当前page处于busy状态(针对大于ngx_slab_max_size的内存块)
-
NGX_SLAB_PAGE_START: 值为0x80000000,用于指示连续多个页面的开始页
-
NGX_SLAB_SHIFT_MASK: 值为0x0000000f, 当所分配的空间大于ngx_slab_exact_size且小于等于ngx_slab_max_size时,page->slab的高NGX_SLAB_MAP_MASK位用作位图,用于标识当前
页
中内存块的分配情况,低NGX_SLAB_SHIFT_MASK位用于标识当前页的内存块大小移位
的掩码。比如当前内存块大小为256,即1<<8
,那么page->slab的低NGX_SLAB_SHIFT_MASK位的值就是8。 -
NGX_SLAB_MAP_MASK: 值为0xffff0000, 当所分配的空间大于ngx_slab_exact_size且小于等于ngx_slab_max_size时, page->slab的高NGX_SLAB_MAP_MASK位用作位图
-
NGX_SLAB_MAP_SHIFT: 值为16,当所分配的空间大于ngx_slab_exact_size且小于等于ngx_slab_max_size时, page->slab的高16位作为位图
-
NGX_SLAB_BUSY: 值为0xffffffff, 用于表示当前slab处于busy状态(针对小于等于ngx_slab_max_size的内存块)
3) 调试信息
我们当前并不支持NGX_DEBUG_MALLOC
与NGX_HAVE_DEBUG_MALLOC
,因此执行的是如下宏:
#define ngx_slab_junk(p, size)
2. 相关静态函数声明及变量定义
//用于分配指定数量的页
static ngx_slab_page_t *ngx_slab_alloc_pages(ngx_slab_pool_t *pool,
ngx_uint_t pages);
//释放指定数量的页
static void ngx_slab_free_pages(ngx_slab_pool_t *pool, ngx_slab_page_t *page,
ngx_uint_t pages);
//打印slab相关的错误日志
static void ngx_slab_error(ngx_slab_pool_t *pool, ngx_uint_t level,
char *text);
static ngx_uint_t ngx_slab_max_size;
static ngx_uint_t ngx_slab_exact_size;
static ngx_uint_t ngx_slab_exact_shift;
3. 函数ngx_slab_init()
void
ngx_slab_init(ngx_slab_pool_t *pool)
{
u_char *p;
size_t size;
ngx_int_t m;
ngx_uint_t i, n, pages;
ngx_slab_page_t *slots;
/* STUB */
if (ngx_slab_max_size == 0) {
ngx_slab_max_size = ngx_pagesize / 2;
ngx_slab_exact_size = ngx_pagesize / (8 * sizeof(uintptr_t));
for (n = ngx_slab_exact_size; n >>= 1; ngx_slab_exact_shift++) {
/* void */
}
}
/**/
pool->min_size = 1 << pool->min_shift;
p = (u_char *) pool + sizeof(ngx_slab_pool_t);
size = pool->end - p;
ngx_slab_junk(p, size);
slots = (ngx_slab_page_t *) p;
n = ngx_pagesize_shift - pool->min_shift;
for (i = 0; i < n; i++) {
slots[i].slab = 0;
slots[i].next = &slots[i];
slots[i].prev = 0;
}
p += n * sizeof(ngx_slab_page_t);
pages = (ngx_uint_t) (size / (ngx_pagesize + sizeof(ngx_slab_page_t)));
ngx_memzero(p, pages * sizeof(ngx_slab_page_t));
pool->pages = (ngx_slab_page_t *) p;
pool->free.prev = 0;
pool->free.next = (ngx_slab_page_t *) p;
pool->pages->slab = pages;
pool->pages->next = &pool->free;
pool->pages->prev = (uintptr_t) &pool->free;
pool->start = (u_char *)
ngx_align_ptr((uintptr_t) p + pages * sizeof(ngx_slab_page_t),
ngx_pagesize);
m = pages - (pool->end - pool->start) / ngx_pagesize;
if (m > 0) {
pages -= m;
pool->pages->slab = pages;
}
pool->last = pool->pages + pages;
pool->log_nomem = 1;
pool->log_ctx = &pool->zero;
pool->zero = '\0';
}
本函数用于初始化整个ngx_slab_pool_t结构,下面我们简要分析一下函数的实现:
void
ngx_slab_init(ngx_slab_pool_t *pool)
{
//1) 计算ngx_slab_max_size,ngx_slab_exact_size, ngx_slab_exact_shift
// 从前面我们知道ngx_pagesize=4096, 因此
ngx_slab_max_size = 2048;
ngx_slab_exact_size = 128;
ngx_slab_exact_shift = 7;
//2) 计算得到min_size大小及n的值
pool->min_size = 8; //(pool->min_shift值为3)
ngx_pagesize_shift = 12;
n = 9;
//3) 初始化slots
for (i = 0; i < n; i++) {
slots[i].slab = 0;
slots[i].next = &slots[i];
slots[i].prev = 0;
}
//4) 计算剩余空间最多能分配多少个页(每一页大小为ngx_pagesize,另外每一页还对应一个ngx_slab_page_t管理结构)
pages = (ngx_uint_t) (size / (ngx_pagesize + sizeof(ngx_slab_page_t)));
//5) 初始化pool->pages及pool->free链表
pool->pages->slab = pages; //slab用于记录空闲页数
//6) 计算实际空闲页数
m = pages - (pool->end - pool->start) / ngx_pagesize;
if (m > 0) {
pages -= m;
pool->pages->slab = pages;
}
pool->last = pool->pages + pages;
//7)将pool->zero的值置为'\0'
//pool->zero = '\0';
}
下面我们画出初始化之后的示意图:
4. 函数ngx_slab_alloc()
void *
ngx_slab_alloc(ngx_slab_pool_t *pool, size_t size)
{
void *p;
ngx_shmtx_lock(&pool->mutex);
p = ngx_slab_alloc_locked(pool, size);
ngx_shmtx_unlock(&pool->mutex);
return p;
}
本函数首先使用pool->mutex
加锁,然后再调用ngx_slab_alloc_locked()函数分配指定大小的空间。
5. 函数ngx_slab_alloc_locked()
void *
ngx_slab_alloc_locked(ngx_slab_pool_t *pool, size_t size)
{
size_t s;
uintptr_t p, n, m, mask, *bitmap;
ngx_uint_t i, slot, shift, map;
ngx_slab_page_t *page, *prev, *slots;
if (size > ngx_slab_max_size) {
ngx_log_debug1(NGX_LOG_DEBUG_ALLOC, ngx_cycle->log, 0,
"slab alloc: %uz", size);
page = ngx_slab_alloc_pages(pool, (size >> ngx_pagesize_shift)
+ ((size % ngx_pagesize) ? 1 : 0));
if (page) {
p = (page - pool->pages) << ngx_pagesize_shift;
p += (uintptr_t) pool->start;
} else {
p = 0;
}
goto done;
}
if (size > pool->min_size) {
shift = 1;
for (s = size - 1; s >>= 1; shift++) { /* void */ }
slot = shift - pool->min_shift;
} else {
size = pool->min_size;
shift = pool->min_shift;
slot = 0;
}
ngx_log_debug2(NGX_LOG_DEBUG_ALLOC, ngx_cycle->log, 0,
"slab alloc: %uz slot: %ui", size, slot);
slots = (ngx_slab_page_t *) ((u_char *) pool + sizeof(ngx_slab_pool_t));
page = slots[slot].next;
if (page->next != page) {
if (shift < ngx_slab_exact_shift) {
do {
p = (page - pool->pages) << ngx_pagesize_shift;
bitmap = (uintptr_t *) (pool->start + p);
map = (1 << (ngx_pagesize_shift - shift))
/ (sizeof(uintptr_t) * 8);
for (n = 0; n < map; n++) {
if (bitmap[n] != NGX_SLAB_BUSY) {
for (m = 1, i = 0; m; m <<= 1, i++) {
if ((bitmap[n] & m)) {
continue;
}
bitmap[n] |= m;
i = ((n * sizeof(uintptr_t) * 8) << shift)
+ (i << shift);
if (bitmap[n] == NGX_SLAB_BUSY) {
for (n = n + 1; n < map; n++) {
if (bitmap[n] != NGX_SLAB_BUSY) {
p = (uintptr_t) bitmap + i;
goto done;
}
}
prev = (ngx_slab_page_t *)
(page->prev & ~NGX_SLAB_PAGE_MASK);
prev->next = page->next;
page->next->prev = page->prev;
page->next = NULL;
page->prev = NGX_SLAB_SMALL;
}
p = (uintptr_t) bitmap + i;
goto done;
}
}
}
page = page->next;
} while (page);
} else if (shift == ngx_slab_exact_shift) {
do {
if (page->slab != NGX_SLAB_BUSY) {
for (m = 1, i = 0; m; m <<= 1, i++) {
if ((page->slab & m)) {
continue;
}
page->slab |= m;
if (page->slab == NGX_SLAB_BUSY) {
prev = (ngx_slab_page_t *)
(page->prev & ~NGX_SLAB_PAGE_MASK);
prev->next = page->next;
page->next->prev = page->prev;
page->next = NULL;
page->prev = NGX_SLAB_EXACT;
}
p = (page - pool->pages) << ngx_pagesize_shift;
p += i << shift;
p += (uintptr_t) pool->start;
goto done;
}
}
page = page->next;
} while (page);
} else { /* shift > ngx_slab_exact_shift */
n = ngx_pagesize_shift - (page->slab & NGX_SLAB_SHIFT_MASK);
n = 1 << n;
n = ((uintptr_t) 1 << n) - 1;
mask = n << NGX_SLAB_MAP_SHIFT;
do {
if ((page->slab & NGX_SLAB_MAP_MASK) != mask) {
for (m = (uintptr_t) 1 << NGX_SLAB_MAP_SHIFT, i = 0;
m & mask;
m <<= 1, i++)
{
if ((page->slab & m)) {
continue;
}
page->slab |= m;
if ((page->slab & NGX_SLAB_MAP_MASK) == mask) {
prev = (ngx_slab_page_t *)
(page->prev & ~NGX_SLAB_PAGE_MASK);
prev->next = page->next;
page->next->prev = page->prev;
page->next = NULL;
page->prev = NGX_SLAB_BIG;
}
p = (page - pool->pages) << ngx_pagesize_shift;
p += i << shift;
p += (uintptr_t) pool->start;
goto done;
}
}
page = page->next;
} while (page);
}
}
page = ngx_slab_alloc_pages(pool, 1);
if (page) {
if (shift < ngx_slab_exact_shift) {
p = (page - pool->pages) << ngx_pagesize_shift;
bitmap = (uintptr_t *) (pool->start + p);
s = 1 << shift;
n = (1 << (ngx_pagesize_shift - shift)) / 8 / s;
if (n == 0) {
n = 1;
}
bitmap[0] = (2 << n) - 1;
map = (1 << (ngx_pagesize_shift - shift)) / (sizeof(uintptr_t) * 8);
for (i = 1; i < map; i++) {
bitmap[i] = 0;
}
page->slab = shift;
page->next = &slots[slot];
page->prev = (uintptr_t) &slots[slot] | NGX_SLAB_SMALL;
slots[slot].next = page;
p = ((page - pool->pages) << ngx_pagesize_shift) + s * n;
p += (uintptr_t) pool->start;
goto done;
} else if (shift == ngx_slab_exact_shift) {
page->slab = 1;
page->next = &slots[slot];
page->prev = (uintptr_t) &slots[slot] | NGX_SLAB_EXACT;
slots[slot].next = page;
p = (page - pool->pages) << ngx_pagesize_shift;
p += (uintptr_t) pool->start;
goto done;
} else { /* shift > ngx_slab_exact_shift */
page->slab = ((uintptr_t) 1 << NGX_SLAB_MAP_SHIFT) | shift;
page->next = &slots[slot];
page->prev = (uintptr_t) &slots[slot] | NGX_SLAB_BIG;
slots[slot].next = page;
p = (page - pool->pages) << ngx_pagesize_shift;
p += (uintptr_t) pool->start;
goto done;
}
}
p = 0;
done:
ngx_log_debug1(NGX_LOG_DEBUG_ALLOC, ngx_cycle->log, 0,
"slab alloc: %p", (void *) p);
return (void *) p;
}
本函数用于分配指定大小的内存空间。这里slab将不等长的内存大小划分为4个大类:
-
小块内存(NGX_SLAB_SMALL): 内存大小 < ngx_slab_exact_size
-
中等内存(NGX_SLAB_EXACT): 内存大小 == ngx_slab_exact_size
-
大块内存(NGX_SLAB_BIG): ngx_slab_exact_size < 内存大小 <= ngx_slab_max_size
-
超大内存(NGX_SLAB_PAGE): ngx_slab_max_size < 内存大小
ngx_slab_exact_size = ngx_pagesize / (8 * sizeof(uintptr_t));
ngx_slab_exact_size表示uintptr_t
slab,此种情况下的slab用作bitmap,表示一页中的内存块使用状况。slab中的所有位(8 * sizeof(uintptr_t))刚好可以一一对应于一页找那个的大小为ngx_slab_exact_size的每一块内存。
下面我们详细分析一下函数的执行流程:
void *
ngx_slab_alloc_locked(ngx_slab_pool_t *pool, size_t size)
{
//1) 判断当前所请求分配的空间size是否大于ngx_slab_max_size(当前值为2048),如果大于,那么直接调用
//ngx_slab_alloc_pages()函数来进行分配
if (size > ngx_slab_max_size) {
ngx_log_debug1(NGX_LOG_DEBUG_ALLOC, ngx_cycle->log, 0,
"slab alloc: %uz", size);
page = ngx_slab_alloc_pages(pool, (size >> ngx_pagesize_shift)
+ ((size % ngx_pagesize) ? 1 : 0));
if (page) {
//1.1) 由返回page在页数组中的偏移量,计算出实际数组地址的偏移量。然后再加上真实可用内存的起始地址pool->start,
//即可算出本次所分配内存的起始地址
p = (page - pool->pages) << ngx_pagesize_shift;
p += (uintptr_t) pool->start;
} else {
p = 0;
}
goto done;
}
//2) 判断从哪一个slot来分配空间(当前pool->min_size=8)
// 2.1) 如果size > pool->min_size的话,则找到对应的slot;
// 2.2) 否则,按最小8字节的方式来分配,因此slot=0
//3) 计算出slots数组的起始位置
slots = (ngx_slab_page_t *) ((u_char *) pool + sizeof(ngx_slab_pool_t));
page = slots[slot].next;
//4) 如果page->next != page,说明指定slot下的page仍有空闲内存块
if (page->next != page) {
if (shift < ngx_slab_exact_shift) {
/*
* 4.1) 对应于分配小块内存
*
* 当从一个页中分配小于ngx_slab_exact_shift(ngx_slab_exact_size=128)的内存块时,无法用uintptr_t slab来标识
* 一页内所有内存块的使用情况,因此,这里不用page->slab来标识该页内所有内存块的使用情况,而是使用页数据空间的开始
* 几个uintptr_t空间来标识
*/
do{
//4.1.1) 计算对应page的实际地址
p = (page - pool->pages) << ngx_pagesize_shift;
bitmap = (uintptr_t *) (pool->start + p);
//4.1.2) 通过(1<<(ngx_pagesize_shift - shift))可以算出一页中可以存放多少内存块,然后再除以sizeof(uintptr_t) *8
// 则可以计算出需要多少个uintptr_t空间来作为整个页的位图
map = (1 << (ngx_pagesize_shift - shift))
/ (sizeof(uintptr_t) * 8);
//4.1.3) 遍历每一个位图,找出其中的可用内存块
for (n = 0; n < map; n++) {
if (bitmap[n] != NGX_SLAB_BUSY) {
//找出了其中一个bitmap具有可用内存空间
for (m = 1, i = 0; m; m <<= 1, i++) {
//a) 找出可用内存块
//b) 将对应bitmap位设为1,并计算出对应的内存块在页内的偏移地址。
bitmap[n] |= m;
i = ((n * sizeof(uintptr_t) * 8) << shift)
+ (i << shift);
//c) 若bitmap[n] == NGX_SLAB_BUSY,那么说明当前bitmap[n]所表示的一系列内存块已经没有空闲块了,此时
// 需要检查整个page页是否仍有空闲内存块,如果没有则说明当前页为全满页,要脱离对应链表
if (bitmap[n] == NGX_SLAB_BUSY) {
//获得当前页的前一页prev,将当前页脱离链表后会设置page->prev=NGX_SLAB_SMALL;
prev = (ngx_slab_page_t *)
(page->prev & ~NGX_SLAB_PAGE_MASK);
prev->next = page->next;
page->next->prev = page->prev;
page->next = NULL;
page->prev = NGX_SLAB_SMALL;
}
}
}
}
}while(page);
}else if (shift == ngx_slab_exact_shift) {
/*
* 4.2) 对应分配中等内存大小的情况
*
* 此种情况用page->slab作为位图就刚好能够表示已页中的所有内存块
*/
do{
if(page->slab != NGX_SLAB_BUSY)
{
//a) 找出可用内存块
//b) 将对应bitmap位设为1,并计算出对应的内存块在页内的偏移地址。
//c) 将全满也脱离链表
//d) 计算分配到的内存首地址
p = (page - pool->pages) << ngx_pagesize_shift;
p += i << shift;
p += (uintptr_t) pool->start;
}
}while(page);
}else{
/*
* 4.3) 大块内存(ngx_slab_exact_size < 内存块大小 <= ngx_slab_max_size)的情况
* 当前Ubuntu 32bit系统,ngx_slab_exact_size值为128, ngx_slab_max_size值为2048
*
* 当需要分配的空间大于ngx_slab_exact_size=128时,我们可以用一个uintptr_t的位(共32位)来表示这些空间
* 所以我们依然采用跟 '等于ngx_slab_exact_size' 时类似的情况,用page->slab来标识该page内所有内存块的使用情况
* 此时的page->slab同时存储bitmap及表示内存大小的shift,高位为bitmap
* 这里会有的内存块大小依次为: 256bytes、 512bytes、 1024bytes、 2048bytes
* 对应的shift依次为: 8、 9、 10、 11
* 那么采用page->slab的高16位来表示这些空间的占用情况,而最低NGX_SLAB_SHIFT_MASK位用来标识此页分配大小,即保存移位数
* 例如:
* 比如我们分配256,当分配该页的第一块空间时,此时的page->slab位图情况是:0x00010008
* 那分配该页的第二块空间时,page->slab的值就是: 0x00030008。当page->slab的值为0xffff0008时,就表示此页已经分配完毕
*
* #define NGX_SLAB_SHIFT_MASK 0x0000000f
* page->slab & NGX_SLAB_SHIFT_MASK 即得到最低4位的值,其实就是当前页的分配大小的移位数
* 这里用最低4位就足够了,因为shift最大为11(表示内存块大小为2048 bytes)
* ngx_pagesize_shift - (page->slab & NGX_SLAB_SHIFT_MASK);就是在一页中标记这些块所需的移位数,也就是块数对应的移位数
* 例如:
* 当页内所能分配的内存块大小是256bytes时,此时,page->slab & NGX_SLAB_SHIFT_MASK = 8
* 因此,n=ngx_pagesize_shift - (page->slab & NGX_SLAB_SHIFT_MASK) = 12 - 8 = 4
* 即4096bytes 可以分配为16个256bytes, 因此n = 1<<n = 16
*
* 其实,是对n = 2^ngx_pagesize_shift/2^(page->slab & NGX_SLAB_SHIFT_MASK) = 2^(ngx_pagesize_shift - (page->slab & NGX_SLAB_SHIFT_MASK))公式的简化
*/
n = ngx_pagesize_shift - (page->slab & NGX_SLAB_SHIFT_MASK);
n = 1 << n;
//得到表示这些块数都用完的bitmap,用现在是低16位的(当前Ubuntu 32bit操作系统)
n = ((uintptr_t) 1 << n) - 1;
//将低16位转换成高16位,因为我们是用高16位来表示空间地址的占用情况的,#define NGX_SLAB_MAP_SHIFT 16
do{
if ((page->slab & NGX_SLAB_MAP_MASK) != mask) {
//4.3.1) 表示当前页为非全满页,仍有空闲
//遍历整个位图
for (m = (uintptr_t) 1 << NGX_SLAB_MAP_SHIFT, i = 0; m & mask;m <<= 1, i++){
//a) 将对应bitmap位设为1,并计算出对应的内存块在页内的偏移地址。
//b) 将全满也脱离链表
//c) 计算分配到的内存首地址
p = (page - pool->pages) << ngx_pagesize_shift;
p += i << shift;
p += (uintptr_t) pool->start;
}
}
}while(page);
}
}
/*
* 5) 在 小块内存、中等内存、大块内存 等三种情况下(不包括超大页面的情况),
* 如果当前slab对应的page中没有空间可分配了,则重新从空闲page中分配一个页
* /
//6) 将分配的新页加入到对应的链表
if(page)
{
if (shift < ngx_slab_exact_shift) {
/*
* 6.1) 处理小页情况
*/
//6.1.1) 计算位图的位置
p = (page - pool->pages) << ngx_pagesize_shift;
bitmap = (uintptr_t *) (pool->start + p);
/*
* 6.1.2) 这里shift代表要分配多大内存块的移位数,因此s为需要分配内存块的大小;
* n用于表示需要用多少块大小为s的内存块来作为位图空间
*/
s = 1 << shift;
n = (1 << (ngx_pagesize_shift - shift)) / 8 / s;
//6.1.3) 前面n块内存作为位图被占用,另外还有一块内存由本次分配出去,因此位图中要(n+1)位来表示这写内存被占用
bitmap[0] = (2 << n) - 1;
map = (1 << (ngx_pagesize_shift - shift)) / (sizeof(uintptr_t) * 8);
for (i = 1; i < map; i++) {
bitmap[i] = 0;
}
//6.1.4) 将当前page插入到对应的slot。因为这里是新分配的一页,因此对应slot链表肯定为空,否则不会运行到此处
page->slab = shift;
page->next = &slots[slot];
page->prev = (uintptr_t) &slots[slot] | NGX_SLAB_SMALL;
slots[slot].next = page;
//6.1.5) 返回实际分配的空间的地址
}else if(shift == ngx_slab_exact_shift){
/*
* 6.2) 处理中等大小内存的情况。
* 此种情况直接用page->slab作为位图,很容易进行处理
*/
}else{
/*
* 6.3) 用于处理大页
* page->slab的高位用于存储位图,低NGX_SLAB_SHIFT_MASK位用于存储当前页中内存块的大小
*/
page->slab = ((uintptr_t) 1 << NGX_SLAB_MAP_SHIFT) | shift;
}
}
done:
return p;
}
6. 函数ngx_slab_calloc()
void *
ngx_slab_calloc(ngx_slab_pool_t *pool, size_t size)
{
void *p;
ngx_shmtx_lock(&pool->mutex);
p = ngx_slab_calloc_locked(pool, size);
ngx_shmtx_unlock(&pool->mutex);
return p;
}
与ngx_slab_alloc()函数类似,不过这里会将分配的内存清0。
7. 函数ngx_slab_calloc_locked()
void *
ngx_slab_calloc_locked(ngx_slab_pool_t *pool, size_t size)
{
void *p;
p = ngx_slab_alloc_locked(pool, size);
if (p) {
ngx_memzero(p, size);
}
return p;
}
与ngx_slab_alloc_locked()函数类似,不过这里会将分配的内存清0.
8. 函数ngx_slab_free()
void
ngx_slab_free(ngx_slab_pool_t *pool, void *p)
{
ngx_shmtx_lock(&pool->mutex);
ngx_slab_free_locked(pool, p);
ngx_shmtx_unlock(&pool->mutex);
}
本函数首先使用pool->mutex加锁,然后再调用ngx_slab_free_locked()函数来释放对应的内存。
9. 函数ngx_slab_free_locked()
void
ngx_slab_free_locked(ngx_slab_pool_t *pool, void *p)
{
size_t size;
uintptr_t slab, m, *bitmap;
ngx_uint_t n, type, slot, shift, map;
ngx_slab_page_t *slots, *page;
ngx_log_debug1(NGX_LOG_DEBUG_ALLOC, ngx_cycle->log, 0, "slab free: %p", p);
if ((u_char *) p < pool->start || (u_char *) p > pool->end) {
ngx_slab_error(pool, NGX_LOG_ALERT, "ngx_slab_free(): outside of pool");
goto fail;
}
n = ((u_char *) p - pool->start) >> ngx_pagesize_shift;
page = &pool->pages[n];
slab = page->slab;
type = page->prev & NGX_SLAB_PAGE_MASK;
switch (type) {
case NGX_SLAB_SMALL:
shift = slab & NGX_SLAB_SHIFT_MASK;
size = 1 << shift;
if ((uintptr_t) p & (size - 1)) {
goto wrong_chunk;
}
n = ((uintptr_t) p & (ngx_pagesize - 1)) >> shift;
m = (uintptr_t) 1 << (n & (sizeof(uintptr_t) * 8 - 1));
n /= (sizeof(uintptr_t) * 8);
bitmap = (uintptr_t *)
((uintptr_t) p & ~((uintptr_t) ngx_pagesize - 1));
if (bitmap[n] & m) {
if (page->next == NULL) {
slots = (ngx_slab_page_t *)
((u_char *) pool + sizeof(ngx_slab_pool_t));
slot = shift - pool->min_shift;
page->next = slots[slot].next;
slots[slot].next = page;
page->prev = (uintptr_t) &slots[slot] | NGX_SLAB_SMALL;
page->next->prev = (uintptr_t) page | NGX_SLAB_SMALL;
}
bitmap[n] &= ~m;
n = (1 << (ngx_pagesize_shift - shift)) / 8 / (1 << shift);
if (n == 0) {
n = 1;
}
if (bitmap[0] & ~(((uintptr_t) 1 << n) - 1)) {
goto done;
}
map = (1 << (ngx_pagesize_shift - shift)) / (sizeof(uintptr_t) * 8);
for (n = 1; n < map; n++) {
if (bitmap[n]) {
goto done;
}
}
ngx_slab_free_pages(pool, page, 1);
goto done;
}
goto chunk_already_free;
case NGX_SLAB_EXACT:
m = (uintptr_t) 1 <<
(((uintptr_t) p & (ngx_pagesize - 1)) >> ngx_slab_exact_shift);
size = ngx_slab_exact_size;
if ((uintptr_t) p & (size - 1)) {
goto wrong_chunk;
}
if (slab & m) {
if (slab == NGX_SLAB_BUSY) {
slots = (ngx_slab_page_t *)
((u_char *) pool + sizeof(ngx_slab_pool_t));
slot = ngx_slab_exact_shift - pool->min_shift;
page->next = slots[slot].next;
slots[slot].next = page;
page->prev = (uintptr_t) &slots[slot] | NGX_SLAB_EXACT;
page->next->prev = (uintptr_t) page | NGX_SLAB_EXACT;
}
page->slab &= ~m;
if (page->slab) {
goto done;
}
ngx_slab_free_pages(pool, page, 1);
goto done;
}
goto chunk_already_free;
case NGX_SLAB_BIG:
shift = slab & NGX_SLAB_SHIFT_MASK;
size = 1 << shift;
if ((uintptr_t) p & (size - 1)) {
goto wrong_chunk;
}
m = (uintptr_t) 1 << ((((uintptr_t) p & (ngx_pagesize - 1)) >> shift)
+ NGX_SLAB_MAP_SHIFT);
if (slab & m) {
if (page->next == NULL) {
slots = (ngx_slab_page_t *)
((u_char *) pool + sizeof(ngx_slab_pool_t));
slot = shift - pool->min_shift;
page->next = slots[slot].next;
slots[slot].next = page;
page->prev = (uintptr_t) &slots[slot] | NGX_SLAB_BIG;
page->next->prev = (uintptr_t) page | NGX_SLAB_BIG;
}
page->slab &= ~m;
if (page->slab & NGX_SLAB_MAP_MASK) {
goto done;
}
ngx_slab_free_pages(pool, page, 1);
goto done;
}
goto chunk_already_free;
case NGX_SLAB_PAGE:
if ((uintptr_t) p & (ngx_pagesize - 1)) {
goto wrong_chunk;
}
if (slab == NGX_SLAB_PAGE_FREE) {
ngx_slab_error(pool, NGX_LOG_ALERT,
"ngx_slab_free(): page is already free");
goto fail;
}
if (slab == NGX_SLAB_PAGE_BUSY) {
ngx_slab_error(pool, NGX_LOG_ALERT,
"ngx_slab_free(): pointer to wrong page");
goto fail;
}
n = ((u_char *) p - pool->start) >> ngx_pagesize_shift;
size = slab & ~NGX_SLAB_PAGE_START;
ngx_slab_free_pages(pool, &pool->pages[n], size);
ngx_slab_junk(p, size << ngx_pagesize_shift);
return;
}
/* not reached */
return;
done:
ngx_slab_junk(p, size);
return;
wrong_chunk:
ngx_slab_error(pool, NGX_LOG_ALERT,
"ngx_slab_free(): pointer to wrong chunk");
goto fail;
chunk_already_free:
ngx_slab_error(pool, NGX_LOG_ALERT,
"ngx_slab_free(): chunk is already free");
fail:
return;
}
本函数用于将对应的空间释放回内存池。下面我们简要分析一下函数的执行流程:
void
ngx_slab_free_locked(ngx_slab_pool_t *pool, void *p)
{
//1) 判定对应的内存p是否属于内存池所管理
if ((u_char *) p < pool->start || (u_char *) p > pool->end) {}
//2) 计算空间p所属的page以及page的类型
n = ((u_char *) p - pool->start) >> ngx_pagesize_shift;
page = &pool->pages[n];
slab = page->slab;
type = page->prev & NGX_SLAB_PAGE_MASK;
//3) 根据不同类型的内存,做不同的处理
switch(type){
case NGX_SLAB_SMALL:
/*
* 3.1) 小块内存
*
* 对于小块内存无法用uintptr_t slab;来标识一页内所有内存块的使用情况,因此,这里不用
* page->slab来标识该页内所有内存块的使用情况,而是使用页数据空间的开始几个uintptr_t
* 空间来表示了
*
* 3.1.1) 在 '小内存块' 中,page->slab存放的是等长内存块的大小(用位偏移的方式存储)
* 因此,size为当前页小块内存的大小
*/
shift = slab & NGX_SLAB_SHIFT_MASK;
size = 1 << shift;
//3.1.2) 因为从pool中分配的空间首先是ngx_pagesize对齐的,自然也是这里的size对齐的,因此 (uintptr_t)p & (size -1)的值应该为0
if ((uintptr_t) p & (size - 1)) {
goto wrong_chunk;
}
/*
* 3.1.3) 这里很巧妙:由于前面对页进行了内存对齐处理,因此下面的式子可直接
*
* 求出 p 对应的slot块的位置,即 p 对应的小块内存位于page中的第几个块:
* 首先((uintptr_t) p & (ngx_pagesize -1))可以计算出页内的偏移地址p_offset,然后将页内偏移地址p_offset除以每一块的大小即可
* 求出是属于page内的第几块
*/
n = ((uintptr_t) p & (ngx_pagesize - 1)) >> shift;
//3.1.4) m是计算page中第n块内存对应于uintptr_t中的第几位
m = (uintptr_t) 1 << (n & (sizeof(uintptr_t) * 8 - 1));
//3.1.5) 用于计算page中第n块内存对应于位图中的第几个uintptr_t
n /= (sizeof(uintptr_t) * 8);
//3.1.6) 用于计算位图的起始位置
bitmap = (uintptr_t *)
((uintptr_t) p & ~((uintptr_t) ngx_pagesize - 1));
/*
* 3.1.7) bitmap[n] & m 值不等于0,表明了此处了内存处于busy状态,即表示分配了出去。这种情况才有释放内存一说,
* 否则直接 go chunk_already_free,表明此块内存已经处于free状态,不需要释放
*/
if(bitmap[n] & m){
/*
* a) 如果页面的当前状态是全部已使用(全满页,全满页不在任何链表中;全满页释放一个内存块后变为半满页),
* 则把它重新加入到slot链表的表头
*/
if(page->next == NULL){}
//b) 将对应的内存块的位图置为空闲状态
bitmap[n] &= ~m;
/*
* c) 如下是计算要多少个内存块来存放位图
* (1<<(ngx_pagesize_shift -shift))用于计算一页内总共有多少块内存块,再除以8用于计算要多少个字节来作为位图表示这些内存块;
* 最后再除以(1<<shift)用于计算要用多少块内存块来作为位图
*/
n = (1 << (ngx_pagesize_shift - shift)) / 8 / (1 << shift);
if (n == 0) {
n = 1;
}
/*
* d) 从上面步骤c)可知道,要用n个内存块来存放位图。因此这里我们要判定bitmap[0]中除用来存放位图的内存块外,是否处于busy状态
* ((uintptr_t)1 << n) -1)用于表示存储位图的内存块的掩码,再取反,则可以得出前32个内存块中除用作位图的内存块外的掩码,然后
* 在按位与上bitmap[0],则可以知道前32块内存块中实际用于存储数据的内存块是否使用。
* 如果仍处于busy状态,那么直接goto done;
*/
if (bitmap[0] & ~(((uintptr_t) 1 << n) - 1)) {
goto done;
}
//e) 计算有多少个uintptr_t对象才能表示整个位图
map = (1 << (ngx_pagesize_shift - shift)) / (sizeof(uintptr_t) * 8);
//f) 判断剩余的位图是否处于busy状态
//g) 到此步骤时,表明当前整个页都处于空闲状态,此时需要调用ngx_slab_free_pages()来释放整个页。
ngx_slab_free_pages(pool, page, 1);
}
goto chunk_already_free;
case NGX_SLAB_EXACT:
/*
* 用于释放中等大小内存块。
*
* 此种情况下,page->slab作为当前页的使用情况的位图
*/
/*
* a)
* ((uintptr_t) p & (ngx_pagesize -1))可以计算出当前内存块的页内偏移地址
* (((uintptr_t) p & (ngx_pagesize - 1)) >> ngx_slab_exact_shift)可以计算出当前内存块是属于页内的第几块内存块
* 再接着左移,则可以计算出该块内存块对应与位图(page->slab)中的哪一位
*/
m = (uintptr_t) 1 <<
(((uintptr_t) p & (ngx_pagesize - 1)) >> ngx_slab_exact_shift);
//b) 因为从pool中分配的空间首先是ngx_pagesize对齐的,自然也是这里的size对齐的,因此 (uintptr_t)p & (size -1)的值应该为0
if ((uintptr_t) p & (size - 1)) {
goto wrong_chunk;
}
/*
* c) slab & m 值不等于0,表明了此处了内存处于busy状态,即表示分配了出去。这种情况才有释放内存一说,
* 否则直接 go chunk_already_free,表明此块内存已经处于free状态,不需要释放
*/
if (slab & m) {
if(slab == NGX_SLAB_BUSY)
{
//d) 表明当前整个page中的内存块都处于busy状态。此种情况即使释放了当前块,整个page仍处于半满页状态,需要
//加入到对应slot链表的表头
}
//e) 将当前内存块为空闲状态
page->slab &= ~m;
//f) 说明当前page页并不是空闲页,直接goto done;
if(page->slab){
}
//g) 整个page处于空闲状态,需要释放整个page
ngx_slab_free_pages(pool, page, 1);
}
case NGX_SLAB_BIG:
/*
* 用于释放大内存页的情况
*/
//a) slab的低NGX_SLAB_SHIFT_MASK位用于表示当前页内存块大小的移位,然后就可以算出当前页内存块的大小
shift = slab & NGX_SLAB_SHIFT_MASK;
size = 1 << shift;
//b) 因为从pool中分配的空间首先是ngx_pagesize对齐的,自然也是这里的size对齐的,因此 (uintptr_t)p & (size -1)的值应该为0
if ((uintptr_t) p & (size - 1)) {
goto wrong_chunk;
}
/*
* c) 用于计算当前内存块的位图
* ((uintptr_t) p & (ngx_pagesize - 1))用于计算页内偏移地址
* (((uintptr_t) p & (ngx_pagesize - 1)) >> shift) 用于计算出当前是属于页内的第几个内存块
* ((((uintptr_t) p & (ngx_pagesize - 1)) >> shift)+ NGX_SLAB_MAP_SHIFT) 用于计算出在位图的第几位
*/
m = (uintptr_t) 1 << ((((uintptr_t) p & (ngx_pagesize - 1)) >> shift)
+ NGX_SLAB_MAP_SHIFT);
/*
* d) slab & m 值不等于0,表明了此处了内存处于busy状态,即表示分配了出去。这种情况才有释放内存一说,
* 否则直接 go chunk_already_free,表明此块内存已经处于free状态,不需要释放
*/
if (slab & m) {
/*
* e) 如果页面的当前状态是全部已使用(全满页,全满页不在任何链表中;全满页释放一个内存块后变为半满页),
* 则把它重新加入到slot链表的表头
*/
if(page->next == NULL){}
//f) 将当前内存块置为空闲状态
page->slab &= ~m;
//g) 说明当前page页并不是空闲页,直接goto done;
if (page->slab & NGX_SLAB_MAP_MASK) {
goto done;
}
//h) 整个page处于空闲状态,需要释放整个page
ngx_slab_free_pages(pool, page, 1);
}
case NGX_SLAB_PAGE:
/*
* 释放超大内存
*/
//a) 因为从pool中分配的空间首先是ngx_pagesize对齐的,自然也是这里的size对齐的,因此 (uintptr_t)p & (size -1)的值应该为0
if ((uintptr_t) p & (ngx_pagesize - 1)) {
goto wrong_chunk;
}
//b) 当前页的slab处于free状态,表明不需要进行释放
if (slab == NGX_SLAB_PAGE_FREE) {}
//c) 对于按页分配的内存,第一页的page->slab的值为 pages | NGX_SLAB_PAGE_START;即NGX_SLAB_PAGE_START标识再按位或上连续的页数
//而后续页才会设置为NGX_SLAB_PAGE_BUSY
if (slab == NGX_SLAB_PAGE_BUSY) {}
//d) 用于计算页偏移,及当前内存块所包含的页数
n = ((u_char *) p - pool->start) >> ngx_pagesize_shift;
size = slab & ~NGX_SLAB_PAGE_START;
//e) 用于释放对应的页
ngx_slab_free_pages(pool, &pool->pages[n], size);
}
}
10. 函数ngx_slab_alloc_pages()
static ngx_slab_page_t *
static ngx_slab_page_t *
ngx_slab_alloc_pages(ngx_slab_pool_t *pool, ngx_uint_t pages)
{
ngx_slab_page_t *page, *p;
for (page = pool->free.next; page != &pool->free; page = page->next) {
if (page->slab >= pages) {
if (page->slab > pages) {
page[page->slab - 1].prev = (uintptr_t) &page[pages];
page[pages].slab = page->slab - pages;
page[pages].next = page->next;
page[pages].prev = page->prev;
p = (ngx_slab_page_t *) page->prev;
p->next = &page[pages];
page->next->prev = (uintptr_t) &page[pages];
} else {
p = (ngx_slab_page_t *) page->prev;
p->next = page->next;
page->next->prev = page->prev;
}
page->slab = pages | NGX_SLAB_PAGE_START;
page->next = NULL;
page->prev = NGX_SLAB_PAGE;
if (--pages == 0) {
return page;
}
for (p = page + 1; pages; pages--) {
p->slab = NGX_SLAB_PAGE_BUSY;
p->next = NULL;
p->prev = NGX_SLAB_PAGE;
p++;
}
return page;
}
}
if (pool->log_nomem) {
ngx_slab_error(pool, NGX_LOG_CRIT,
"ngx_slab_alloc() failed: no memory");
}
return NULL;
}
本函数用于从pool中分配出指定的 n 页。下面我们简单分析一下函数的执行流程:
static ngx_slab_page_t *
ngx_slab_alloc_pages(ngx_slab_pool_t *pool, ngx_uint_t pages)
{
//遍历整个空闲链表
for (page = pool->free.next; page != &pool->free; page = page->next) {
//找出连续空闲页大于指定大小的内存块
if (page->slab >= pages) {
if (page->slab > pages){
//说明分配之后,还有剩余的空闲页,此时需要重新设置链表
}else{
//说明刚好够分配,此种情况只需要将这一系列页从free链表摘除即可
}
//将新分配出的这些page设置如下标识
page->slab = pages | NGX_SLAB_PAGE_START;
page->next = NULL;
page->prev = NGX_SLAB_PAGE;
//若请求分配的页数是1页,则直接返回
//否则,从第2页起将page->slab都设置为NGX_SLAB_PAGE_BUSY
for (p = page + 1; pages; pages--) {
p->slab = NGX_SLAB_PAGE_BUSY;
p->next = NULL;
p->prev = NGX_SLAB_PAGE;
p++;
}
}
}
}
11. 函数ngx_slab_free_pages()
static void
ngx_slab_free_pages(ngx_slab_pool_t *pool, ngx_slab_page_t *page,
ngx_uint_t pages)
{
ngx_uint_t type;
ngx_slab_page_t *prev, *join;
page->slab = pages--;
if (pages) {
ngx_memzero(&page[1], pages * sizeof(ngx_slab_page_t));
}
if (page->next) {
prev = (ngx_slab_page_t *) (page->prev & ~NGX_SLAB_PAGE_MASK);
prev->next = page->next;
page->next->prev = page->prev;
}
join = page + page->slab;
if (join < pool->last) {
type = join->prev & NGX_SLAB_PAGE_MASK;
if (type == NGX_SLAB_PAGE) {
if (join->next != NULL) {
pages += join->slab;
page->slab += join->slab;
prev = (ngx_slab_page_t *) (join->prev & ~NGX_SLAB_PAGE_MASK);
prev->next = join->next;
join->next->prev = join->prev;
join->slab = NGX_SLAB_PAGE_FREE;
join->next = NULL;
join->prev = NGX_SLAB_PAGE;
}
}
}
if (page > pool->pages) {
join = page - 1;
type = join->prev & NGX_SLAB_PAGE_MASK;
if (type == NGX_SLAB_PAGE) {
if (join->slab == NGX_SLAB_PAGE_FREE) {
join = (ngx_slab_page_t *) (join->prev & ~NGX_SLAB_PAGE_MASK);
}
if (join->next != NULL) {
pages += join->slab;
join->slab += page->slab;
prev = (ngx_slab_page_t *) (join->prev & ~NGX_SLAB_PAGE_MASK);
prev->next = join->next;
join->next->prev = join->prev;
page->slab = NGX_SLAB_PAGE_FREE;
page->next = NULL;
page->prev = NGX_SLAB_PAGE;
page = join;
}
}
}
if (pages) {
page[pages].prev = (uintptr_t) page;
}
page->prev = (uintptr_t) &pool->free;
page->next = pool->free.next;
page->next->prev = (uintptr_t) page;
pool->free.next = page;
}
此函数用于释放指定页数的page。下面我们简要分析一下函数的执行流程:
static void
ngx_slab_free_pages(ngx_slab_pool_t *pool, ngx_slab_page_t *page,
ngx_uint_t pages)
{
/*
* 将page->slab设置为指定的页数,
*/
page->slab = pages--;
//1) 如果释放的页数大于1,那么会将从第1叶开始的后续所有的页清0
if (pages) {
ngx_memzero(&page[1], pages * sizeof(ngx_slab_page_t));
}
//2) 如果page->next不为空,表示当前page还没有脱离原来的slot链表(因为对于全满页,早已脱离了链表)
//3) 检查当前所释放page(s)是否能够和后续页进行合并
if (join < pool->last) {
type = join->prev & NGX_SLAB_PAGE_MASK;
/*
* 3.1) 表明此处是一个完整的页,并不是slot中对应的页(因为只有全满页 和 全空页 的type值会是NGX_SLAB_PAGE)
*/
if(type == NGX_SLAB_PAGE){
if (join->next != NULL) {
/*
* join->next不为NULL,表明不是全满页,因为全满页已经脱离了任何的链表,其next一定为NULL
*
* 将join脱离链表,并加入到当前释放的page中
*/
}
}
}
//4) 检查当前所释放的page(s)是否能够和前面的页进行合并
if (page > pool->pages) {
join = page - 1;
type = join->prev & NGX_SLAB_PAGE_MASK;
/*
* 4.1) 表明此处是一个完整的页,并不是slot中对应的页(因为只有全满页 和 全空页 的type值会是NGX_SLAB_PAGE)
*/
if (type == NGX_SLAB_PAGE) {
//a) 找出该空闲内存页(组)的首页
if (join->slab == NGX_SLAB_PAGE_FREE) {
join = (ngx_slab_page_t *) (join->prev & ~NGX_SLAB_PAGE_MASK);
}
if (join->next != NULL){
/*
* join->next不为NULL,表明不是全满页,因为全满页已经脱离了任何的链表,其next一定为NULL
*
* 将join脱离链表,并加入到当前释放的page中
*/
}
}
}
//5)将所释放的内存页组的最后一页的prev指向该组的第一页,这样方便后续合并
if (pages) {
page[pages].prev = (uintptr_t) page;
}
//6) 将page加入到free链表的首部
}
下面给出一张释放pages的示意图:
12. 函数ngx_slab_error()
static void
ngx_slab_error(ngx_slab_pool_t *pool, ngx_uint_t level, char *text)
{
ngx_log_error(level, ngx_cycle->log, 0, "%s%s", text, pool->log_ctx);
}
用于打印slab内存分配相关的错误消息
[参看]