Memcached源码分析之items.c

时间：2016-09-07 19:19:46 阅读：235 评论：0 收藏：0 [点我收藏+]

标签：

#include "memcached.h"
#include <sys/stat.h>
#include <sys/socket.h>
#include <sys/signal.h>
#include <sys/resource.h>
#include <fcntl.h>
#include <netinet/in.h>
#include <errno.h>
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#include <time.h>
#include <assert.h>
#include <unistd.h>
static void item_link_q(item *it);
static void item_unlink_q(item *it);
#define LARGEST_ID POWER_LARGEST
typedef struct {
uint64_t evicted;
uint64_t evicted_nonzero;
rel_time_t evicted_time;
uint64_t reclaimed;
uint64_t outofmemory;
uint64_t tailrepairs;
uint64_t expired_unfetched;
uint64_t evicted_unfetched;
uint64_t crawler_reclaimed;
} itemstats_t;
static item *heads[LARGEST_ID]; //各个slabclass的LRU队列头指针数组
static item *tails[LARGEST_ID]; //各个slabclass的LRU队列尾指针数组
static crawler crawlers[LARGEST_ID]; //各个slabclass的item爬虫数组
static itemstats_t itemstats[LARGEST_ID]; //各个slabclass的item统计数组
static unsigned int sizes[LARGEST_ID]; //各个slabclass的chunk大小数组
static int crawler_count = 0;
static volatile int do_run_lru_crawler_thread = 0;
static int lru_crawler_initialized = 0;
static pthread_mutex_t lru_crawler_lock = PTHREAD_MUTEX_INITIALIZER;
static pthread_cond_t lru_crawler_cond = PTHREAD_COND_INITIALIZER;
//重置统计
void item_stats_reset(void) {
mutex_lock(&cache_lock);
memset(itemstats, 0, sizeof(itemstats));
mutex_unlock(&cache_lock);
}
/* Get the next CAS id for a new item. */
uint64_t get_cas_id(void) {
static uint64_t cas_id = 0;
return ++cas_id;
}
/* Enable this for reference-count debugging. */
#if 0
# define DEBUG_REFCNT(it,op) \
fprintf(stderr, "item %x refcnt(%c) %d %c%c%c\n", \
it, op, it->refcount, \
(it->it_flags & ITEM_LINKED) ? ‘L‘ : ‘ ‘, \
(it->it_flags & ITEM_SLABBED) ? ‘S‘ : ‘ ‘)
#else
# define DEBUG_REFCNT(it,op) while(0)
#endif
/**
算出item总大小
*/
static size_t item_make_header(const uint8_t nkey, const int flags, const int nbytes,
char *suffix, uint8_t *nsuffix) {
/* suffix is defined at 40 chars elsewhere.. */
*nsuffix = (uint8_t) snprintf(suffix, 40, " %d %d\r\n", flags, nbytes - 2);
return sizeof(item) + nkey + *nsuffix + nbytes;
}
/**
item分配
把这个函数弄清楚，基本就把memcached内存管理机制大体弄清楚了。
*/
item *do_item_alloc(char *key, const size_t nkey, const int flags,
const rel_time_t exptime, const int nbytes,
const uint32_t cur_hv) {
uint8_t nsuffix;
item *it = NULL;
char suffix[40];
size_t ntotal = item_make_header(nkey + 1, flags, nbytes, suffix, &nsuffix); //item总大小
if (settings.use_cas) {
ntotal += sizeof(uint64_t); //如果有用到cas 那么item大小还要加上unit64_t的size
}
unsigned int id = slabs_clsid(ntotal); //根据item大小，找到适合的slabclass
if (id == 0)
return 0;
mutex_lock(&cache_lock); //cache锁
/* do a quick check if we have any expired items in the tail.. */
/* 准备分配新的item了，随便快速瞄一下lru链表末尾有没有过期item，有的话就用过期的空间 */
int tries = 5;
int tried_alloc = 0;
item *search;
void *hold_lock = NULL;
rel_time_t oldest_live = settings.oldest_live;
search = tails[id]; //这个tails是一个全局变量，tails[xx]是id为xx的slabclass lru链表的尾部
/* We walk up *only* for locked items. Never searching for expired.
* Waste of CPU for almost all deployments */
//从LRU链表尾部（就是最久没使用过的item）开始往前找
for (; tries > 0 && search != NULL; tries--, search=search->prev) {
if (search->nbytes == 0 && search->nkey == 0 && search->it_flags == 1) {
/* We are a crawler, ignore it. */
/*
这里注释意思是说我们现在是以爬虫的身份来爬出过期的空间，
像爬到这种很怪的item，就别管了，不是爬虫要做的事，不要就行了。
*/
tries++;
continue;
}
/**
你会看到很多地方有这个hv，简单说下，其实它是对item的一个hash，得到hv值，这个hv主要有两个
作用：
1）用于hash表保存item，通过hv计算出哈希表中的桶号
2）用于item lock表中锁住item，通过hv计算出应该用item lock表中哪个锁对当前item进行加锁
这两者都涉及到一个粒度问题，不可能保证每个不一样的key的hv不会相同，所有hash方法都可能
出现冲突。
所以hash表中用链表的方式处理冲突的item，而item lock表中会多个item共享一个锁，或者说
多个桶共享一个锁。
*/
uint32_t hv = hash(ITEM_key(search), search->nkey);
/**
尝试去锁住当前item。
*/
if (hv == cur_hv || (hold_lock = item_trylock(hv)) == NULL)
continue;
/* Now see if the item is refcount locked */
if (refcount_incr(&search->refcount) != 2) {
refcount_decr(&search->refcount);
/* Old rare bug could cause a refcount leak. We haven‘t seen
* it in years, but we leave this code in to prevent failures
* just in case
没看懂这里的意思.....
*/
if (settings.tail_repair_time &&
search->time + settings.tail_repair_time < current_time) {
itemstats[id].tailrepairs++;
search->refcount = 1;
do_item_unlink_nolock(search, hv);
}
if (hold_lock)
item_trylock_unlock(hold_lock);
continue;
}
/* Expired or flushed */
//超时了...
if ((search->exptime != 0 && search->exptime < current_time)
|| (search->time <= oldest_live && oldest_live <= current_time)) {
itemstats[id].reclaimed++;
if ((search->it_flags & ITEM_FETCHED) == 0) {
itemstats[id].expired_unfetched++;
}
it = search; //拿下空间
slabs_adjust_mem_requested(it->slabs_clsid, ITEM_ntotal(it), ntotal); //更新统计数据
/**
什么是link，在这简单说下，就是把item加到哈希表和LRU链表的过程。详见items::do_item_link函数这里把item旧的link取消掉，当前函数do_item_alloc的工作只是拿空间，而往后可知道拿到item空间后会对这块item进行“link”工作，而这里这块item空间是旧的item超时然后拿来用的，所以先把它unlink掉
*/
do_item_unlink_nolock(it, hv);
/* Initialize the item block: */
it->slabs_clsid = 0;
} else if ((it = slabs_alloc(ntotal, id)) == NULL) {/*如果没有找到超时的item，则
调用slabs_alloc分配空间，详见slabs_alloc
如果slabs_alloc分配空间失败，即返回NULL，则往下走，下面的代码是
把LRU列表最后一个给淘汰，即使item没有过期。
这里一般是可用内存已经满了，需要按LRU进行淘汰的时候。
//************mark: $1**************
*/
tried_alloc = 1; //标记一下，表示有进入此分支，表示有尝试过调用slabs_alloc去分配新的空间。
//记下被淘汰item的信息，像我们使用memcached经常会查看的evicted_time就是在这里赋值啦！
if (settings.evict_to_free == 0) {
itemstats[id].outofmemory++;
} else {
itemstats[id].evicted++;
itemstats[id].evicted_time = current_time - search->time; //被淘汰的item距离上次使用多长时间了
if (search->exptime != 0)
itemstats[id].evicted_nonzero++;
if ((search->it_flags & ITEM_FETCHED) == 0) {
itemstats[id].evicted_unfetched++;
}
it = search;
slabs_adjust_mem_requested(it->slabs_clsid, ITEM_ntotal(it), ntotal);//更新统计数据
do_item_unlink_nolock(it, hv); //从哈希表和LRU链表中删掉
/* Initialize the item block: */
it->slabs_clsid = 0;
/*
如果当前slabclass有item被淘汰掉了，说明可用内存都满了，再也没有
slab可分配了，
而如果 slab_automove=2 (默认是1)，这样会导致angry模式，
就是只要分配失败了，就马上进行slab重分配：把别的slabclass空间牺牲
掉一些，马上给现在的slabclass分配空间，而不会合理地根据淘汰统计
数据来分析要怎么重分配（slab_automove = 1则会）。
*/
if (settings.slab_automove == 2)
slabs_reassign(-1, id);
}
}
refcount_decr(&search->refcount);
/* If hash values were equal, we don‘t grab a second lock */
if (hold_lock)
item_trylock_unlock(hold_lock);
break;
}
/**
如果上面的for循环里面没有找到空间，并且没有进入过else if ((it = slabs_alloc(ntotal, id)) == NULL)这个分支没有
尝试调slabs_alloc分配空间（有这种可能性），那么，下面这行代码就是再尝试分配。
你会觉得上面那个循环写得特纠结，逻辑不清，估计你也看醉了。其实整个分配原则是这样子：
1）先从LRU链表找下看看有没有恰好过期的空间，有的话就用这个空间。
2）如果没有过期的空间，就分配新的空间。
3）如果分配新的空间失败，那么往往是内存都用光了，则从LRU链表中把最旧的即使没过期的item淘汰掉，空间分给新的item用。
问题是：这个从“LRU链表找到的item”是一个不确定的东西，有可能这个item数据异常，有可能这个item由于与别的item共用锁的桶号
这个桶被锁住了，所以总之各种原因这个item此刻不一定可用，因此用了一个循环尝试找几次（上面是5）。
所以逻辑是：
1）我先找5次LRU看看有没有可用的过期的item，有就用它。（for循环5次）
2）5次没有找到可用的过期的item，那我分配新的。
3）分配新的不成功，那我再找5次看看有没有可用的虽然没过期的item，淘汰它，把空间给新的item用。（for循环5次）
那么这里有个问题，如果代码要写得逻辑清晰一点，我得写两个for循环，一个是为了第2）步前“找可用的过期的”item，
一个是第2）步不成功后“找可用的用来淘汰的”空间。而且有重复的逻辑“找到可用的”，所以memcached作者就合在一起了，
然后只能把第2）步也塞到for循环里面，确实挺尴尬的。。。估计memcached作者也写得很纠结。。。
所以就很有可能出现5次都没找到可用的空间，都没进入过elseif那个分支就被continue掉了，为了记下有没有进过elseif
分支就挫挫地用一个tried_alloc变量来做记号。。
*/
if (!tried_alloc && (tries == 0 || search == NULL))
it = slabs_alloc(ntotal, id);
if (it == NULL) {
itemstats[id].outofmemory++;
mutex_unlock(&cache_lock);
return NULL; //没错！会有分配新空间不成功，而且尝试5次淘汰旧的item也没成功的时候，只能返回NULL。。
}
assert(it->slabs_clsid == 0);
assert(it != heads[id]);
//来到这里，说明item分配成功，下面主要是一些初始化工作。
/* Item initialization can happen outside of the lock; the item‘s already
* been removed from the slab LRU.
*/
it->refcount = 1; /* the caller will have a reference */
mutex_unlock(&cache_lock);
it->next = it->prev = it->h_next = 0;
it->slabs_clsid = id;
DEBUG_REFCNT(it, ‘*‘);
it->it_flags = settings.use_cas ? ITEM_CAS : 0;
it->nkey = nkey;
it->nbytes = nbytes;
memcpy(ITEM_key(it), key, nkey);
it->exptime = exptime;
memcpy(ITEM_suffix(it), suffix, (size_t)nsuffix);
it->nsuffix = nsuffix;
return it;
}
/**
把这块item free掉，以供再利用，注意这里的free不是指把内存空间释放哦，
而是把这块item 变为“空闲”
*/
void item_free(item *it) {
size_t ntotal = ITEM_ntotal(it);
unsigned int clsid;
assert((it->it_flags & ITEM_LINKED) == 0);
assert(it != heads[it->slabs_clsid]);
assert(it != tails[it->slabs_clsid]);
assert(it->refcount == 0);
/* so slab size changer can tell later if item is already free or not */
clsid = it->slabs_clsid;
it->slabs_clsid = 0; //在这把free掉的item 的slabs_clsid设为0
DEBUG_REFCNT(it, ‘F‘);
slabs_free(it, ntotal, clsid);
}
/**
* 检查item大小
*/
bool item_size_ok(const size_t nkey, const int flags, const int nbytes) {
char prefix[40];
uint8_t nsuffix;
size_t ntotal = item_make_header(nkey + 1, flags, nbytes,
prefix, &nsuffix);
if (settings.use_cas) {
ntotal += sizeof(uint64_t);
}
return slabs_clsid(ntotal) != 0;
}
/**
把item插入相应的slabclass lru链表中而已
*/
static void item_link_q(item *it) { /* item is the new head */
item **head, **tail;
assert(it->slabs_clsid < LARGEST_ID);
assert((it->it_flags & ITEM_SLABBED) == 0);
head = &heads[it->slabs_clsid];
tail = &tails[it->slabs_clsid];
assert(it != *head);
assert((*head && *tail) || (*head == 0 && *tail == 0));
it->prev = 0;
it->next = *head;
if (it->next) it->next->prev = it;
*head = it;
if (*tail == 0) *tail = it;
sizes[it->slabs_clsid]++;
return;
}
/**
把item从相应的slabclass lru链表中删掉而已，下面就是经典的删除链表逻辑代码了
*/
static void item_unlink_q(item *it) {
item **head, **tail;
assert(it->slabs_clsid < LARGEST_ID);
head = &heads[it->slabs_clsid];
tail = &tails[it->slabs_clsid];
if (*head == it) {
assert(it->prev == 0);
*head = it->next;
}
if (*tail == it) {
assert(it->next == 0);
*tail = it->prev;
}
assert(it->next != it);
assert(it->prev != it);
if (it->next) it->next->prev = it->prev;
if (it->prev) it->prev->next = it->next;
sizes[it->slabs_clsid]--;
return;
}
/**
把item "link"起来，主要包括：
1）改变一些统计数据
2）把item加到哈希表
3）把item插入到相应的slabclass lru链表中
*/
int do_item_link(item *it, const uint32_t hv) {
MEMCACHED_ITEM_LINK(ITEM_key(it), it->nkey, it->nbytes);
assert((it->it_flags & (ITEM_LINKED|ITEM_SLABBED)) == 0);
mutex_lock(&cache_lock);
it->it_flags |= ITEM_LINKED;
it->time = current_time;
STATS_LOCK();
stats.curr_bytes += ITEM_ntotal(it);
stats.curr_items += 1;
stats.total_items += 1;
STATS_UNLOCK();
/* Allocate a new CAS ID on link. */
ITEM_set_cas(it, (settings.use_cas) ? get_cas_id() : 0);
assoc_insert(it, hv); //插入哈希表
item_link_q(it); //加入LRU链表
refcount_incr(&it->refcount);
mutex_unlock(&cache_lock);
return 1;
}
/**
就是和do_item_link反过来的一些操作
*/
void do_item_unlink(item *it, const uint32_t hv) {
MEMCACHED_ITEM_UNLINK(ITEM_key(it), it->nkey, it->nbytes);
mutex_lock(&cache_lock);
if ((it->it_flags & ITEM_LINKED) != 0) {
it->it_flags &= ~ITEM_LINKED;
STATS_LOCK();
stats.curr_bytes -= ITEM_ntotal(it);
stats.curr_items -= 1;
STATS_UNLOCK();
assoc_delete(ITEM_key(it), it->nkey, hv);
item_unlink_q(it);
do_item_remove(it);
}
mutex_unlock(&cache_lock);
}
/* FIXME: Is it necessary to keep this copy/pasted code? */
void do_item_unlink_nolock(item *it, const uint32_t hv) {
MEMCACHED_ITEM_UNLINK(ITEM_key(it), it->nkey, it->nbytes);
if ((it->it_flags & ITEM_LINKED) != 0) {
it->it_flags &= ~ITEM_LINKED;
STATS_LOCK();
stats.curr_bytes -= ITEM_ntotal(it);
stats.curr_items -= 1;
STATS_UNLOCK();
assoc_delete(ITEM_key(it), it->nkey, hv);
item_unlink_q(it);
do_item_remove(it);
}
}
/**
指向item的指针不用的时候都会调用此函数
*/
void do_item_remove(item *it) {
MEMCACHED_ITEM_REMOVE(ITEM_key(it), it->nkey, it->nbytes);
assert((it->it_flags & ITEM_SLABBED) == 0);
assert(it->refcount > 0);
if (refcount_decr(&it->refcount) == 0) { //引用计数减1，当引用计数为0时，才真正把item free掉。
item_free(it);
}
}
/**
主要作用是重置在最近使用链表中的位置，更新最近使用时间
*/
void do_item_update(item *it) {
MEMCACHED_ITEM_UPDATE(ITEM_key(it), it->nkey, it->nbytes);
if (it->time < current_time - ITEM_UPDATE_INTERVAL) {
assert((it->it_flags & ITEM_SLABBED) == 0);
mutex_lock(&cache_lock);
if ((it->it_flags & ITEM_LINKED) != 0) {
item_unlink_q(it);
it->time = current_time;
item_link_q(it);
}
mutex_unlock(&cache_lock);
}
}
int do_item_replace(item *it, item *new_it, const uint32_t hv) {
MEMCACHED_ITEM_REPLACE(ITEM_key(it), it->nkey, it->nbytes,
ITEM_key(new_it), new_it->nkey, new_it->nbytes);
assert((it->it_flags & ITEM_SLABBED) == 0);
do_item_unlink(it, hv);
return do_item_link(new_it, hv);
}
void item_stats_evictions(uint64_t *evicted) {
int i;
mutex_lock(&cache_lock);
for (i = 0; i < LARGEST_ID; i++) {
evicted[i] = itemstats[i].evicted;
}
mutex_unlock(&cache_lock);
}
void do_item_stats_totals(ADD_STAT add_stats, void *c) {
itemstats_t totals;
memset(&totals, 0, sizeof(itemstats_t));
int i;
for (i = 0; i < LARGEST_ID; i++) {
totals.expired_unfetched += itemstats[i].expired_unfetched;
totals.evicted_unfetched += itemstats[i].evicted_unfetched;
totals.evicted += itemstats[i].evicted;
totals.reclaimed += itemstats[i].reclaimed;
totals.crawler_reclaimed += itemstats[i].crawler_reclaimed;
}
APPEND_STAT("expired_unfetched", "%llu",
(unsigned long long)totals.expired_unfetched);
APPEND_STAT("evicted_unfetched", "%llu",
(unsigned long long)totals.evicted_unfetched);
APPEND_STAT("evictions", "%llu",
(unsigned long long)totals.evicted);
APPEND_STAT("reclaimed", "%llu",
(unsigned long long)totals.reclaimed);
APPEND_STAT("crawler_reclaimed", "%llu",
(unsigned long long)totals.crawler_reclaimed);
}
void do_item_stats(ADD_STAT add_stats, void *c) {
int i;
for (i = 0; i < LARGEST_ID; i++) {
if (tails[i] != NULL) {
const char *fmt = "items:%d:%s";
char key_str[STAT_KEY_LEN];
char val_str[STAT_VAL_LEN];
int klen = 0, vlen = 0;
if (tails[i] == NULL) {
/* We removed all of the items in this slab class */
continue;
}
APPEND_NUM_FMT_STAT(fmt, i, "number", "%u", sizes[i]);
APPEND_NUM_FMT_STAT(fmt, i, "age", "%u", current_time - tails[i]->time);
APPEND_NUM_FMT_STAT(fmt, i, "evicted",
"%llu", (unsigned long long)itemstats[i].evicted);
APPEND_NUM_FMT_STAT(fmt, i, "evicted_nonzero",
"%llu", (unsigned long long)itemstats[i].evicted_nonzero);
APPEND_NUM_FMT_STAT(fmt, i, "evicted_time",
"%u", itemstats[i].evicted_time);
APPEND_NUM_FMT_STAT(fmt, i, "outofmemory",
"%llu", (unsigned long long)itemstats[i].outofmemory);
APPEND_NUM_FMT_STAT(fmt, i, "tailrepairs",
"%llu", (unsigned long long)itemstats[i].tailrepairs);
APPEND_NUM_FMT_STAT(fmt, i, "reclaimed",
"%llu", (unsigned long long)itemstats[i].reclaimed);
APPEND_NUM_FMT_STAT(fmt, i, "expired_unfetched",
"%llu", (unsigned long long)itemstats[i].expired_unfetched);
APPEND_NUM_FMT_STAT(fmt, i, "evicted_unfetched",
"%llu", (unsigned long long)itemstats[i].evicted_unfetched);
APPEND_NUM_FMT_STAT(fmt, i, "crawler_reclaimed",
"%llu", (unsigned long long)itemstats[i].crawler_reclaimed);
}
}
/* getting here means both ascii and binary terminators fit */
add_stats(NULL, 0, NULL, 0, c);
}
void do_item_stats_sizes(ADD_STAT add_stats, void *c) {
/* max 1MB object, divided into 32 bytes size buckets */
const int num_buckets = 32768;
unsigned int *histogram = calloc(num_buckets, sizeof(int));
if (histogram != NULL) {
int i;
/* build the histogram */
for (i = 0; i < LARGEST_ID; i++) {
item *iter = heads[i];
while (iter) {
int ntotal = ITEM_ntotal(iter);
int bucket = ntotal / 32;
if ((ntotal % 32) != 0) bucket++;
if (bucket < num_buckets) histogram[bucket]++;
iter = iter->next;
}
}
/* write the buffer */
for (i = 0; i < num_buckets; i++) {
if (histogram[i] != 0) {
char key[8];
snprintf(key, sizeof(key), "%d", i * 32);
APPEND_STAT(key, "%u", histogram[i]);
}
}
free(histogram);
}
add_stats(NULL, 0, NULL, 0, c);
}
//读取item数据
item *do_item_get(const char *key, const size_t nkey, const uint32_t hv) {
//mutex_lock(&cache_lock);
item *it = assoc_find(key, nkey, hv);
if (it != NULL) {
refcount_incr(&it->refcount);
if (slab_rebalance_signal &&
((void *)it >= slab_rebal.slab_start && (void *)it < slab_rebal.slab_end)) {
do_item_unlink_nolock(it, hv);
do_item_remove(it);
it = NULL;
}
}
//mutex_unlock(&cache_lock);
int was_found = 0;
if (settings.verbose > 2) {
int ii;
if (it == NULL) {
fprintf(stderr, "> NOT FOUND ");
} else {
fprintf(stderr, "> FOUND KEY ");
was_found++;
}
for (ii = 0; ii < nkey; ++ii) {
fprintf(stderr, "%c", key[ii]);
}
}
if (it != NULL) {
if (settings.oldest_live != 0 && settings.oldest_live <= current_time &&
it->time <= settings.oldest_live) {
do_item_unlink(it, hv);
do_item_remove(it);
it = NULL;
if (was_found) {
fprintf(stderr, " -nuked by flush");
}
} else if (it->exptime != 0 && it->exptime <= current_time) {
do_item_unlink(it, hv);
do_item_remove(it);
it = NULL;
if (was_found) {
fprintf(stderr, " -nuked by expire");
}
} else {
it->it_flags |= ITEM_FETCHED;
DEBUG_REFCNT(it, ‘+‘);
}
}
if (settings.verbose > 2)
fprintf(stderr, "\n");
return it;
}
item *do_item_touch(const char *key, size_t nkey, uint32_t exptime,
const uint32_t hv) {
item *it = do_item_get(key, nkey, hv);
if (it != NULL) {
it->exptime = exptime;
}
return it;
}
/* expires items that are more recent than the oldest_live setting. */
void do_item_flush_expired(void) {
int i;
item *iter, *next;
if (settings.oldest_live == 0)
return;
for (i = 0; i < LARGEST_ID; i++) {
for (iter = heads[i]; iter != NULL; iter = next) {
/* iter->time of 0 are magic objects. */
if (iter->time != 0 && iter->time >= settings.oldest_live) {
next = iter->next;
if ((iter->it_flags & ITEM_SLABBED) == 0) {
do_item_unlink_nolock(iter, hash(ITEM_key(iter), iter->nkey));
}
} else {
/* We‘ve hit the first old item. Continue to the next queue. */
break;
}
}
}
}
static void crawler_link_q(item *it) { /* item is the new tail */
item **head, **tail;
assert(it->slabs_clsid < LARGEST_ID);
assert(it->it_flags == 1);
assert(it->nbytes == 0);
head = &heads[it->slabs_clsid];
tail = &tails[it->slabs_clsid];
assert(*tail != 0);
assert(it != *tail);
assert((*head && *tail) || (*head == 0 && *tail == 0));
it->prev = *tail;
it->next = 0;
if (it->prev) {
assert(it->prev->next == 0);
it->prev->next = it;
}
*tail = it;
if (*head == 0) *head = it;
return;
}
static void crawler_unlink_q(item *it) {
item **head, **tail;
assert(it->slabs_clsid < LARGEST_ID);
head = &heads[it->slabs_clsid];
tail = &tails[it->slabs_clsid];
if (*head == it) {
assert(it->prev == 0);
*head = it->next;
}
if (*tail == it) {
assert(it->next == 0);
*tail = it->prev;
}
assert(it->next != it);
assert(it->prev != it);
if (it->next) it->next->prev = it->prev;
if (it->prev) it->prev->next = it->next;
return;
}
static item *crawler_crawl_q(item *it) {
item **head, **tail;
assert(it->it_flags == 1);
assert(it->nbytes == 0);
assert(it->slabs_clsid < LARGEST_ID);
head = &heads[it->slabs_clsid];
tail = &tails[it->slabs_clsid];
/* We‘ve hit the head, pop off */
if (it->prev == 0) {
assert(*head == it);
if (it->next) {
*head = it->next;
assert(it->next->prev == it);
it->next->prev = 0;
}
return NULL; /* Done */
}
assert(it->prev != it);
if (it->prev) {
if (*head == it->prev) {
*head = it;
}
if (*tail == it) {
*tail = it->prev;
}
assert(it->next != it);
if (it->next) {
assert(it->prev->next == it);
it->prev->next = it->next;
it->next->prev = it->prev;
} else {
it->prev->next = 0;
}
it->next = it->prev;
it->prev = it->next->prev;
it->next->prev = it;
if (it->prev) {
it->prev->next = it;
}
}
assert(it->next != it);
assert(it->prev != it);
return it->next; /* success */
}
/* I pulled this out to make the main thread clearer, but it reaches into the
* main thread‘s values too much. Should rethink again.
上面这句注释作者是说，他把用爬虫处理过期的item的工作放到另一个专门的线程里去做
是为了让主线程干净一点，但是这线程的工作涉及到太多主线程的东西了，得重新想想..
这个函数的作用是“评估”一下这个item是否应该free掉。其实主要就是看下有没有过期啦~
当然用户设置的settings.oldest_live参数也加入到考虑中
*/
static void item_crawler_evaluate(item *search, uint32_t hv, int i) {
rel_time_t oldest_live = settings.oldest_live;
if ((search->exptime != 0 && search->exptime < current_time)
|| (search->time <= oldest_live && oldest_live <= current_time)) {
itemstats[i].crawler_reclaimed++;
if (settings.verbose > 1) {
int ii;
char *key = ITEM_key(search);
fprintf(stderr, "LRU crawler found an expired item (flags: %d, slab: %d): ",
search->it_flags, search->slabs_clsid);
for (ii = 0; ii < search->nkey; ++ii) {
fprintf(stderr, "%c", key[ii]);
}
fprintf(stderr, "\n");
}
if ((search->it_flags & ITEM_FETCHED) == 0) {
itemstats[i].expired_unfetched++;
}
do_item_unlink_nolock(search, hv);
do_item_remove(search);
assert(search->slabs_clsid == 0);
} else {
refcount_decr(&search->refcount);
}
}
/**
item爬虫线程入口，负责从lru链表中把过期的item free掉
*/
static void *item_crawler_thread(void *arg) {
int i;
pthread_mutex_lock(&lru_crawler_lock);
if (settings.verbose > 2)
fprintf(stderr, "Starting LRU crawler background thread\n");
while (do_run_lru_crawler_thread) {
pthread_cond_wait(&lru_crawler_cond, &lru_crawler_lock);
while (crawler_count) {
item *search = NULL;
void *hold_lock = NULL;
for (i = 0; i < LARGEST_ID; i++) {
if (crawlers[i].it_flags != 1) {
continue;
}
pthread_mutex_lock(&cache_lock);
search = crawler_crawl_q((item *)&crawlers[i]);
if (search == NULL ||
(crawlers[i].remaining && --crawlers[i].remaining < 1)) {
if (settings.verbose > 2)
fprintf(stderr, "Nothing left to crawl for %d\n", i);
crawlers[i].it_flags = 0;
crawler_count--;
crawler_unlink_q((item *)&crawlers[i]);
pthread_mutex_unlock(&cache_lock);
continue;
}
uint32_t hv = hash(ITEM_key(search), search->nkey);
/* Attempt to hash item lock the "search" item. If locked, no
* other callers can incr the refcount
*/
if ((hold_lock = item_trylock(hv)) == NULL) {
pthread_mutex_unlock(&cache_lock);
continue;
}
/* Now see if the item is refcount locked */
if (refcount_incr(&search->refcount) != 2) {
refcount_decr(&search->refcount);
if (hold_lock)
item_trylock_unlock(hold_lock);
pthread_mutex_unlock(&cache_lock);
continue;
}
item_crawler_evaluate(search, hv, i);
if (hold_lock)
item_trylock_unlock(hold_lock);
pthread_mutex_unlock(&cache_lock);
if (settings.lru_crawler_sleep)
usleep(settings.lru_crawler_sleep);
}
}
if (settings.verbose > 2)
fprintf(stderr, "LRU crawler thread sleeping\n");
STATS_LOCK();
stats.lru_crawler_running = false;
STATS_UNLOCK();
}
pthread_mutex_unlock(&lru_crawler_lock);
if (settings.verbose > 2)
fprintf(stderr, "LRU crawler thread stopping\n");
return NULL;
}
static pthread_t item_crawler_tid;
//停止item爬虫线程
int stop_item_crawler_thread(void) {
int ret;
pthread_mutex_lock(&lru_crawler_lock);
do_run_lru_crawler_thread = 0;
pthread_cond_signal(&lru_crawler_cond);
pthread_mutex_unlock(&lru_crawler_lock);
if ((ret = pthread_join(item_crawler_tid, NULL)) != 0) {
fprintf(stderr, "Failed to stop LRU crawler thread: %s\n", strerror(ret));
return -1;
}
settings.lru_crawler = false;
return 0;
}
/**
启动item 爬虫线程
*/
int start_item_crawler_thread(void) {
int ret;
if (settings.lru_crawler)
return -1;
pthread_mutex_lock(&lru_crawler_lock);
do_run_lru_crawler_thread = 1;
settings.lru_crawler = true;
if ((ret = pthread_create(&item_crawler_tid, NULL,
item_crawler_thread, NULL)) != 0) {
fprintf(stderr, "Can‘t create LRU crawler thread: %s\n",
strerror(ret));
pthread_mutex_unlock(&lru_crawler_lock);
return -1;
}
pthread_mutex_unlock(&lru_crawler_lock);
return 0;
}
enum crawler_result_type lru_crawler_crawl(char *slabs) {
char *b = NULL;
uint32_t sid = 0;
uint8_t tocrawl[POWER_LARGEST];
if (pthread_mutex_trylock(&lru_crawler_lock) != 0) {
return CRAWLER_RUNNING;
}
pthread_mutex_lock(&cache_lock);
if (strcmp(slabs, "all") == 0) {
for (sid = 0; sid < LARGEST_ID; sid++) {
tocrawl[sid] = 1;
}
} else {
for (char *p = strtok_r(slabs, ",", &b);
p != NULL;
p = strtok_r(NULL, ",", &b)) {
if (!safe_strtoul(p, &sid) || sid < POWER_SMALLEST
|| sid > POWER_LARGEST) {
pthread_mutex_unlock(&cache_lock);
pthread_mutex_unlock(&lru_crawler_lock);
return CRAWLER_BADCLASS;
}
tocrawl[sid] = 1;
}
}
for (sid = 0; sid < LARGEST_ID; sid++) {
if (tocrawl[sid] != 0 && tails[sid] != NULL) {
if (settings.verbose > 2)
fprintf(stderr, "Kicking LRU crawler off for slab %d\n", sid);
crawlers[sid].nbytes = 0;
crawlers[sid].nkey = 0;
crawlers[sid].it_flags = 1; /* For a crawler, this means enabled. */
crawlers[sid].next = 0;
crawlers[sid].prev = 0;
crawlers[sid].time = 0;
crawlers[sid].remaining = settings.lru_crawler_tocrawl;
crawlers[sid].slabs_clsid = sid;
crawler_link_q((item *)&crawlers[sid]);
crawler_count++;
}
}
pthread_mutex_unlock(&cache_lock);
pthread_cond_signal(&lru_crawler_cond);
STATS_LOCK();
stats.lru_crawler_running = true;
STATS_UNLOCK();
pthread_mutex_unlock(&lru_crawler_lock);
return CRAWLER_OK;
}
//初始化lru item爬虫线程
int init_lru_crawler(void) {
if (lru_crawler_initialized == 0) {
if (pthread_cond_init(&lru_crawler_cond, NULL) != 0) {
fprintf(stderr, "Can‘t initialize lru crawler condition\n");
return -1;
}
pthread_mutex_init(&lru_crawler_lock, NULL);
lru_crawler_initialized = 1;
}
return 0;
}

标签：

原文地址：http://www.cnblogs.com/guolanzhu/p/5850256.html

踩

(0)

评论一句话评论（0）

分享档案

更多>

2021年07月29日 (22)
2021年07月28日 (40)
2021年07月27日 (32)
2021年07月26日 (79)
2021年07月23日 (29)
2021年07月22日 (30)
2021年07月21日 (42)
2021年07月20日 (16)
2021年07月19日 (90)
2021年07月16日 (35)

周排行