/* Slab rebalancer thread.
* Does not use spinlocks since it is not timing sensitive. Burn less CPU and
* go to sleep if locks are contended
*/
static void *slab_maintenance_thread(void *arg) {
int was_busy = 0;
int src, dest;
while (do_run_slab_thread) {
if (slab_rebalance_signal == 1) {
if (slab_rebalance_start() < 0) {
/* Handle errors with more specifity as required. */
slab_rebalance_signal = 0;
}
} else if (slab_rebalance_signal && slab_rebal.slab_start != NULL) {
/* If we have a decision to continue, continue it */
was_busy = slab_rebalance_move();
} else if (settings.slab_automove && slab_automove_decision(&src, &dest) == 1) {
/* Blind to the return codes. It will retry on its own */
slabs_reassign(src, dest);
}
if (slab_rebal.done) {
slab_rebalance_finish();
}
/* Sleep a bit if no work to do, or waiting on busy objects */
if (was_busy || !slab_rebalance_signal)
sleep(1);
}
return NULL;
}
/* Loop up to N times:
* If too many items are in HOT_LRU, push to COLD_LRU
* If too many items are in WARM_LRU, push to COLD_LRU
* If too many items are in COLD_LRU, poke COLD_LRU tail
* 1000 loops with 1ms min sleep gives us under 1m items shifted/sec. The
* locks can't handle much more than that. Leaving a TODO for how to
* autoadjust in the future.
*/
static int lru_maintainer_juggle(const int slabs_clsid) {
int i;
int did_moves = 0;
bool mem_limit_reached = false;
uint64_t total_bytes = 0;
unsigned int chunks_perslab = 0;
unsigned int chunks_free = 0;
/* TODO: if free_chunks below high watermark, increase aggressiveness */
chunks_free = slabs_available_chunks(slabs_clsid, &mem_limit_reached,
&total_bytes, &chunks_perslab);
if (settings.expirezero_does_not_evict)
total_bytes -= noexp_lru_size(slabs_clsid);
/* If slab automove is enabled on any level, and we have more than 2 pages
* worth of chunks free in this class, ask (gently) to reassign a page
* from this class back into the global pool (0)
*/
if (settings.slab_automove > 0 && chunks_free > (chunks_perslab * 2.5)) {
slabs_reassign(slabs_clsid, SLAB_GLOBAL_PAGE_POOL);
}
/* Juggle HOT/WARM up to N times */
for (i = 0; i < 1000; i++) {
int do_more = 0;
if (lru_pull_tail(slabs_clsid, HOT_LRU, total_bytes, LRU_PULL_CRAWL_BLOCKS) ||
lru_pull_tail(slabs_clsid, WARM_LRU, total_bytes, LRU_PULL_CRAWL_BLOCKS)) {
do_more++;
}
do_more += lru_pull_tail(slabs_clsid, COLD_LRU, total_bytes, LRU_PULL_CRAWL_BLOCKS);
if (do_more == 0)
break;
did_moves++;
}
return did_moves;
}
/* Slab rebalancer thread.
* Does not use spinlocks since it is not timing sensitive. Burn less CPU and
* go to sleep if locks are contended
*/
static void *slab_maintenance_thread(void *arg) {
int src, dest;
while (do_run_slab_thread) {
if (settings.slab_automove == 1) {
if (slab_automove_decision(&src, &dest) == 1) {
/* Blind to the return codes. It will retry on its own */
slabs_reassign(src, dest);
}
sleep(1);
} else {
/* Don't wake as often if we're not enabled.
* This is lazier than setting up a condition right now. */
sleep(5);
}
}
return NULL;
}
/* Slab rebalancer thread.
* Does not use spinlocks since it is not timing sensitive. Burn less CPU and
* go to sleep if locks are contended
*/
static void *slab_maintenance_thread(void *arg)
{
int src, dest; //src是失败次数很少，或者很平均的那个slab-class id，dest是那些分配失败次数骤增的slab-class id
while (do_run_slab_thread)
{
if (settings.slab_automove == 1)//如果是前面的激烈模式,slab_automove=2,就不用该线程去选择移动的slab，每次失败主动会去移动那些slab
{
if (slab_automove_decision(&src, &dest) == 1)//这里是具体的选择src和dest的处理函数
{
/* Blind to the return codes. It will retry on its own */
slabs_reassign(src, dest);
}
sleep(1);
}
else
{
/* Don't wake as often if we're not enabled.
* This is lazier than setting up a condition right now. */
sleep(5);
}
}
return NULL;
}
/*@null@*/
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];
//计算这个item的空间
size_t ntotal = item_make_header(nkey + 1, flags, nbytes, suffix, &nsuffix);
if (settings.use_cas) {
ntotal += sizeof(uint64_t);
}
//根据大小判断从属于哪个slab
unsigned int id = slabs_clsid(ntotal);
if (id == 0)
return 0;
mutex_lock(&cache_lock);
/* do a quick check if we have any expired items in the tail.. */
//在LRU中尝试5次还没合适的空间，则执行申请空间的操作
int tries = 5;
int tried_alloc = 0;
item *search;
void *hold_lock = NULL;
rel_time_t oldest_live = settings.oldest_live;
search = tails[id];
/* We walk up *only* for locked items. Never searching for expired.
* Waste of CPU for almost all deployments */
for (; tries > 0 && search != NULL; tries--, search=search->prev) {
uint32_t hv = hash(ITEM_key(search), search->nkey, 0);
/* Attempt to hash item lock the "search" item. If locked, no
* other callers can incr the refcount
*/
/* FIXME: I think we need to mask the hv here for comparison? */
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 */
if (search->time + 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 */
// search指向的item过期了，则直接复用这块内存
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;
slabs_adjust_mem_requested(it->slabs_clsid, ITEM_ntotal(it), ntotal);
do_item_unlink_nolock(it, hv);
/* Initialize the item block: */
it->slabs_clsid = 0;
}
//此刻，过期失效的item没有找到，申请内存又失败了。看来只能使用
//LRU淘汰一个item(即使这个item并没有过期失效)
else if ((it = slabs_alloc(ntotal, id)) == NULL) {
tried_alloc = 1;
if (settings.evict_to_free == 0) {
itemstats[id].outofmemory++;
} else {
itemstats[id].evicted++;
itemstats[id].evicted_time = current_time - search->time;
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);
do_item_unlink_nolock(it, hv);
/* Initialize the item block: */
it->slabs_clsid = 0;
/* If we've just evicted an item, and the automover is set to
* angry bird mode, attempt to rip memory into this slab class.
* TODO: Move valid object detection into a function, and on a
* "successful" memory pull, look behind and see if the next alloc
* would be an eviction. Then kick off the slab mover before the
* eviction happens.
*/
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;
}
//从slab分配器中申请内存
if (!tried_alloc && (tries == 0 || search == NULL))
it = slabs_alloc(ntotal, id);
if (it == NULL) {
itemstats[id].outofmemory++;
mutex_unlock(&cache_lock);
return NULL;
}
assert(it->slabs_clsid == 0);
assert(it != heads[id]);
/* 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;
}
/*@null@*/
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);
if (settings.use_cas) {
ntotal += sizeof(uint64_t);
}
unsigned int id = slabs_clsid(ntotal);
if (id == 0)
return 0;
mutex_lock(&cache_lock);
/* do a quick check if we have any expired items in the tail.. */
int tries = 5;
/* Avoid hangs if a slab has nothing but refcounted stuff in it. */
int tries_lrutail_reflocked = 1000;
int tried_alloc = 0;
item *search;
item *next_it;
void *hold_lock = NULL;
rel_time_t oldest_live = settings.oldest_live;
search = tails[id];
/* We walk up *only* for locked items. Never searching for expired.
* Waste of CPU for almost all deployments */
for (; tries > 0 && search != NULL; tries--, search=next_it) {
/* we might relink search mid-loop, so search->prev isn't reliable */
next_it = search->prev;
if (search->nbytes == 0 && search->nkey == 0 && search->it_flags == 1) {
/* We are a crawler, ignore it. */
tries++;
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
*/
/* Don't accidentally grab ourselves, or bail if we can't quicklock */
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) {
/* Avoid pathological case with ref'ed items in tail */
do_item_update_nolock(search);
tries_lrutail_reflocked--;
tries++;
refcount_decr(&search->refcount);
itemstats[id].lrutail_reflocked++;
/* 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 */
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);
if (tries_lrutail_reflocked < 1)
break;
continue;
}
/* Expired or flushed */
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;
slabs_adjust_mem_requested(it->slabs_clsid, ITEM_ntotal(it), ntotal);
do_item_unlink_nolock(it, hv);
/* Initialize the item block: */
it->slabs_clsid = 0;
} else if ((it = slabs_alloc(ntotal, id)) == NULL) {
tried_alloc = 1;
if (settings.evict_to_free == 0) {
itemstats[id].outofmemory++;
} else {
itemstats[id].evicted++;
itemstats[id].evicted_time = current_time - search->time;
if (search->exptime != 0)
itemstats[id].evicted_nonzero++;
if ((search->it_flags & ITEM_FETCHED) == 0) {
itemstats[id].evicted_unfetched++;
}
shadow_item* new_shadow_it = create_shadow_item(search);
hv = hash(new_shadow_it->key, new_shadow_it->nkey);
shadow_assoc_insert(new_shadow_it, hv);
insert_shadowq_item(new_shadow_it,new_shadow_it->slabs_clsid);
it = search;
slabs_adjust_mem_requested(it->slabs_clsid, ITEM_ntotal(it), ntotal);
do_item_unlink_nolock(it, hv);
/* Initialize the item block: */
it->slabs_clsid = 0;
/* If we've just evicted an item, and the automover is set to
* angry bird mode, attempt to rip memory into this slab class.
* TODO: Move valid object detection into a function, and on a
* "successful" memory pull, look behind and see if the next alloc
* would be an eviction. Then kick off the slab mover before the
* eviction happens.
*/
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;
}
if (!tried_alloc && (tries == 0 || search == NULL))
it = slabs_alloc(ntotal, id);
if (it == NULL) {
itemstats[id].outofmemory++;
mutex_unlock(&cache_lock);
return NULL;
}
assert(it->slabs_clsid == 0);
assert(it != heads[id]);
/* 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;
}
/* Returns number of items remove, expired, or evicted.
* Callable from worker threads or the LRU maintainer thread */
static int lru_pull_tail(const int orig_id, const int cur_lru,
const uint64_t total_bytes, uint8_t flags) {
item *it = NULL;
int id = orig_id;
int removed = 0;
if (id == 0)
return 0;
int tries = 5;
item *search;
item *next_it;
void *hold_lock = NULL;
unsigned int move_to_lru = 0;
uint64_t limit = 0;
id |= cur_lru;
pthread_mutex_lock(&lru_locks[id]);
search = tails[id];
/* We walk up *only* for locked items, and if bottom is expired. */
for (; tries > 0 && search != NULL; tries--, search=next_it) {
/* we might relink search mid-loop, so search->prev isn't reliable */
next_it = search->prev;
if (search->nbytes == 0 && search->nkey == 0 && search->it_flags == 1) {
/* We are a crawler, ignore it. */
if (flags & LRU_PULL_CRAWL_BLOCKS) {
pthread_mutex_unlock(&lru_locks[id]);
return 0;
}
tries++;
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. Also skip ourselves. */
if ((hold_lock = item_trylock(hv)) == NULL)
continue;
/* Now see if the item is refcount locked */
if (refcount_incr(&search->refcount) != 2) {
/* Note pathological case with ref'ed items in tail.
* Can still unlink the item, but it won't be reusable yet */
itemstats[id].lrutail_reflocked++;
/* In case of refcount leaks, enable for quick workaround. */
/* WARNING: This can cause terrible corruption */
if (settings.tail_repair_time &&
search->time + settings.tail_repair_time < current_time) {
itemstats[id].tailrepairs++;
search->refcount = 1;
/* This will call item_remove -> item_free since refcnt is 1 */
do_item_unlink_nolock(search, hv);
item_trylock_unlock(hold_lock);
continue;
}
}
/* Expired or flushed */
if ((search->exptime != 0 && search->exptime < current_time)
|| item_is_flushed(search)) {
itemstats[id].reclaimed++;
if ((search->it_flags & ITEM_FETCHED) == 0) {
itemstats[id].expired_unfetched++;
}
/* refcnt 2 -> 1 */
do_item_unlink_nolock(search, hv);
/* refcnt 1 -> 0 -> item_free */
do_item_remove(search);
item_trylock_unlock(hold_lock);
removed++;
/* If all we're finding are expired, can keep going */
continue;
}
/* If we're HOT_LRU or WARM_LRU and over size limit, send to COLD_LRU.
* If we're COLD_LRU, send to WARM_LRU unless we need to evict
*/
switch (cur_lru) {
case HOT_LRU:
limit = total_bytes * settings.hot_lru_pct / 100;
case WARM_LRU:
if (limit == 0)
limit = total_bytes * settings.warm_lru_pct / 100;
if (sizes_bytes[id] > limit) {
itemstats[id].moves_to_cold++;
move_to_lru = COLD_LRU;
do_item_unlink_q(search);
it = search;
removed++;
break;
} else if ((search->it_flags & ITEM_ACTIVE) != 0) {
/* Only allow ACTIVE relinking if we're not too large. */
itemstats[id].moves_within_lru++;
search->it_flags &= ~ITEM_ACTIVE;
do_item_update_nolock(search);
do_item_remove(search);
item_trylock_unlock(hold_lock);
} else {
/* Don't want to move to COLD, not active, bail out */
it = search;
}
break;
case COLD_LRU:
it = search; /* No matter what, we're stopping */
if (flags & LRU_PULL_EVICT) {
if (settings.evict_to_free == 0) {
/* Don't think we need a counter for this. It'll OOM. */
break;
}
itemstats[id].evicted++;
itemstats[id].evicted_time = current_time - search->time;
if (search->exptime != 0)
itemstats[id].evicted_nonzero++;
if ((search->it_flags & ITEM_FETCHED) == 0) {
itemstats[id].evicted_unfetched++;
}
LOGGER_LOG(NULL, LOG_EVICTIONS, LOGGER_EVICTION, search);
do_item_unlink_nolock(search, hv);
removed++;
if (settings.slab_automove == 2) {
slabs_reassign(-1, orig_id);
}
} else if ((search->it_flags & ITEM_ACTIVE) != 0
&& settings.lru_maintainer_thread) {
itemstats[id].moves_to_warm++;
search->it_flags &= ~ITEM_ACTIVE;
move_to_lru = WARM_LRU;
do_item_unlink_q(search);
removed++;
}
break;
}
if (it != NULL)
break;
}
pthread_mutex_unlock(&lru_locks[id]);
if (it != NULL) {
if (move_to_lru) {
it->slabs_clsid = ITEM_clsid(it);
it->slabs_clsid |= move_to_lru;
item_link_q(it);
}
do_item_remove(it);
item_trylock_unlock(hold_lock);
}
return removed;
}
//分配一个item，这个函数包含了memcached具体item分配的逻辑
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的大小，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.. */
int tries = 5;
int tried_alloc = 0;
item *search;
void *hold_lock = NULL;
rel_time_t oldest_live = settings.oldest_live;
search = tails[id]; //全局变量，tails[x]是id为x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 */
//首先从lru链表尾部查找有没有过期的item，tries = 5，最多循环5次
//注意这里是最多查找5次，只要找到一个没有被其他地方引用的item，那么就不再继续查找，如果这个item过期，就使用这个item的空间，否则创建新的slab
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. */
//这里只是搜索过期的item，对于异常的item，直接忽略继续查找
tries++;
continue;
}
//计算item的hash值，hv有两个作用：1.用于hash表保存item 2.用于item lock表中锁住item，通过hv计算出item_lock中哪个锁对当前item加锁
//不同item的hash值可能相同，hash表中用链表的方式解决冲突；item lock中多个item共享一个锁
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
*/
/* Don't accidentally grab ourselves, or bail if we can't quicklock */
//锁住当前item
if (hv == cur_hv || (hold_lock = item_trylock(hv)) == NULL)
continue;
/* Now see if the item is refcount locked */
//检查这个指向的这个item是否被其他地方引用，如果是的话，继续向前查找
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 */
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 */
//如果找到过期的item
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;
//更新统计数据
slabs_adjust_mem_requested(it->slabs_clsid, ITEM_ntotal(it), ntotal);
//把旧的item从hash表和LRU链表中移除
do_item_unlink_nolock(it, hv);
/* Initialize the item block: */
it->slabs_clsid = 0;
}
//如果没有找到过期的item，则调用slabs_alloc分配空间
//如果slabs_alloc返回null，表示分配失败，内存空间已满
//需要按LRU进行淘汰
else if ((it = slabs_alloc(ntotal, id)) == NULL) {
tried_alloc = 1; //标记一下，表示有尝试调用slabs_alloc分配空间
//记录被淘汰item的信息, 使用memcached经常会查看的evicted_time就是在这里赋值的
if (settings.evict_to_free == 0) {
itemstats[id].outofmemory++;
} else {
itemstats[id].evicted++;
itemstats[id].evicted_time = current_time - search->time; //被淘汰item距离上次使用的时间
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hash表和LRU链表中移除
/* Initialize the item block: */
it->slabs_clsid = 0;
/* If we've just evicted an item, and the automover is set to
* angry bird mode, attempt to rip memory into this slab class.
* TODO: Move valid object detection into a function, and on a
* "successful" memory pull, look behind and see if the next alloc
* would be an eviction. Then kick off the slab mover before the
* eviction happens.
*/
//默认情况下，slab_automove=1，会合理地更具淘汰统计数据来分析怎么进行slabclass空间的分配
//如果slab_automove=2，只要分配失败了，马上进行slabclass空间的重分配
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;
}
//查找5次过期的item都失败，并且也没有淘汰可用且没有过期的item
//分配新的内存空间
if (!tried_alloc && (tries == 0 || search == NULL))
it = slabs_alloc(ntotal, id);
//分配失败，返回null
if (it == NULL) {
itemstats[id].outofmemory++;
mutex_unlock(&cache_lock);
return NULL;
}
assert(it->slabs_clsid == 0);
assert(it != heads[id]);
/* Item initialization can happen outside of the lock; the item's already
* been removed from the slab LRU.
*/
//item内存空间分配成功，做一些初始化工作
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;
}
/*@null@*/
item *do_item_alloc(char *key, const size_t nkey, const int flags, const rel_time_t exptime, const int nbytes) {
uint8_t nsuffix;
item *it = NULL;
char suffix[40];
size_t ntotal = item_make_header(nkey + 1, flags, nbytes, suffix, &nsuffix);
if (settings.use_cas) {
ntotal += sizeof(uint64_t);
}
unsigned int id = slabs_clsid(ntotal);
if (id == 0)
return 0;
mutex_lock(&cache_lock);
/* do a quick check if we have any expired items in the tail.. */
item *search;
rel_time_t oldest_live = settings.oldest_live;
search = tails[id];
if (search != NULL && (refcount_incr(&search->refcount) == 2)) {
if ((search->exptime != 0 && search->exptime < current_time)
|| (search->time <= oldest_live && oldest_live <= current_time)) { // dead by flush
STATS_LOCK();
stats.reclaimed++;
STATS_UNLOCK();
itemstats[id].reclaimed++;
if ((search->it_flags & ITEM_FETCHED) == 0) {
STATS_LOCK();
stats.expired_unfetched++;
STATS_UNLOCK();
itemstats[id].expired_unfetched++;
}
it = search;
slabs_adjust_mem_requested(it->slabs_clsid, ITEM_ntotal(it), ntotal);
do_item_unlink_nolock(it, hash(ITEM_key(it), it->nkey, 0));
/* Initialize the item block: */
it->slabs_clsid = 0;
} else if ((it = slabs_alloc(ntotal, id)) == NULL) {
if (settings.evict_to_free == 0) {
itemstats[id].outofmemory++;
mutex_unlock(&cache_lock);
return NULL;
}
itemstats[id].evicted++;
itemstats[id].evicted_time = current_time - search->time;
if (search->exptime != 0)
itemstats[id].evicted_nonzero++;
if ((search->it_flags & ITEM_FETCHED) == 0) {
STATS_LOCK();
stats.evicted_unfetched++;
STATS_UNLOCK();
itemstats[id].evicted_unfetched++;
}
STATS_LOCK();
stats.evictions++;
STATS_UNLOCK();
it = search;
slabs_adjust_mem_requested(it->slabs_clsid, ITEM_ntotal(it), ntotal);
do_item_unlink_nolock(it, hash(ITEM_key(it), it->nkey, 0));
/* Initialize the item block: */
it->slabs_clsid = 0;
/* If we've just evicted an item, and the automover is set to
* angry bird mode, attempt to rip memory into this slab class.
* TODO: Move valid object detection into a function, and on a
* "successful" memory pull, look behind and see if the next alloc
* would be an eviction. Then kick off the slab mover before the
* eviction happens.
*/
if (settings.slab_automove == 2)
slabs_reassign(-1, id, 1);
} else {
refcount_decr(&search->refcount);
}
} else {
/* If the LRU is empty or locked, attempt to allocate memory */
it = slabs_alloc(ntotal, id);
if (search != NULL)
refcount_decr(&search->refcount);
}
if (it == NULL) {
itemstats[id].outofmemory++;
/* Last ditch effort. There was a very rare bug which caused
* refcount leaks. We leave this just in case they ever happen again.
* We can reasonably assume no item can stay locked for more than
* three hours, so if we find one in the tail which is that old,
* free it anyway.
*/
if (search != NULL &&
search->refcount != 2 &&
search->time + TAIL_REPAIR_TIME < current_time) {
itemstats[id].tailrepairs++;
search->refcount = 1;
do_item_unlink_nolock(search, hash(ITEM_key(search), search->nkey, 0));
}
mutex_unlock(&cache_lock);
return NULL;
}
assert(it->slabs_clsid == 0);
assert(it != heads[id]);
/* 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;
}
void process_command(conn *c, char *command) {
int comm = 0;
int incr = 0;
/*
* for commands set/add/replace, we build an item and read the data
* directly into it, then continue in nread_complete().
*/
if (settings.verbose > 1)
fprintf(stderr, "<%d %s\n", c->sfd, command);
/* All incoming commands will require a response, so we cork at the beginning,
and uncork at the very end (usually by means of out_string) */
set_cork(c, 1);
if ((strncmp(command, "add ", 4) == 0 && (comm = NREAD_ADD)) ||
(strncmp(command, "set ", 4) == 0 && (comm = NREAD_SET)) ||
(strncmp(command, "replace ", 8) == 0 && (comm = NREAD_REPLACE))) {
char key[251];
int flags;
time_t expire;
int len, res;
item *it;
res = sscanf(command, "%*s %250s %u %ld %d\n", key, &flags, &expire, &len);
if (res!=4 || strlen(key)==0 ) {
out_string(c, "CLIENT_ERROR bad command line format");
return;
}
expire = realtime(expire);
it = item_alloc(key, flags, expire, len+2);
if (it == 0) {
out_string(c, "SERVER_ERROR out of memory");
/* swallow the data line */
c->write_and_go = conn_swallow;
c->sbytes = len+2;
return;
}
c->item_comm = comm;
c->item = it;
c->rcurr = ITEM_data(it);
c->rlbytes = it->nbytes;
c->state = conn_nread;
return;
}
if ((strncmp(command, "incr ", 5) == 0 && (incr = 1)) ||
(strncmp(command, "decr ", 5) == 0)) {
char temp[32];
unsigned int value;
item *it;
unsigned int delta;
char key[251];
int res;
char *ptr;
time_t now = time(0);
res = sscanf(command, "%*s %250s %u\n", key, &delta);
if (res!=2 || strlen(key)==0 ) {
out_string(c, "CLIENT_ERROR bad command line format");
return;
}
it = assoc_find(key);
if (it && (it->it_flags & ITEM_DELETED)) {
it = 0;
}
if (it && it->exptime && it->exptime < now) {
item_unlink(it);
it = 0;
}
if (!it) {
out_string(c, "NOT_FOUND");
return;
}
ptr = ITEM_data(it);
while (*ptr && (*ptr<'0' && *ptr>'9')) ptr++;
value = atoi(ptr);
if (incr)
value+=delta;
else {
if (delta >= value) value = 0;
else value-=delta;
}
sprintf(temp, "%u", value);
res = strlen(temp);
if (res + 2 > it->nbytes) { /* need to realloc */
item *new_it;
new_it = item_alloc(ITEM_key(it), it->flags, it->exptime, res + 2 );
if (new_it == 0) {
out_string(c, "SERVER_ERROR out of memory");
return;
}
memcpy(ITEM_data(new_it), temp, res);
memcpy(ITEM_data(new_it) + res, "\r\n", 2);
item_replace(it, new_it);
} else { /* replace in-place */
memcpy(ITEM_data(it), temp, res);
memset(ITEM_data(it) + res, ' ', it->nbytes-res-2);
}
out_string(c, temp);
return;
}
if (strncmp(command, "get ", 4) == 0) {
char *start = command + 4;
char key[251];
int next;
int i = 0;
item *it;
time_t now = time(0);
while(sscanf(start, " %250s%n", key, &next) >= 1) {
start+=next;
stats.get_cmds++;
it = assoc_find(key);
if (it && (it->it_flags & ITEM_DELETED)) {
it = 0;
}
if (settings.oldest_live && it &&
it->time <= settings.oldest_live) {
item_unlink(it);
it = 0;
}
if (it && it->exptime && it->exptime < now) {
item_unlink(it);
it = 0;
}
if (it) {
if (i >= c->isize) {
item **new_list = realloc(c->ilist, sizeof(item *)*c->isize*2);
if (new_list) {
c->isize *= 2;
c->ilist = new_list;
} else break;
}
stats.get_hits++;
it->refcount++;
item_update(it);
*(c->ilist + i) = it;
i++;
} else stats.get_misses++;
}
c->icurr = c->ilist;
c->ileft = i;
if (c->ileft) {
c->ipart = 0;
c->state = conn_mwrite;
c->ibytes = 0;
return;
} else {
out_string(c, "END");
return;
}
}
if (strncmp(command, "delete ", 7) == 0) {
char key[251];
item *it;
int res;
time_t exptime = 0;
res = sscanf(command, "%*s %250s %ld", key, &exptime);
it = assoc_find(key);
if (!it) {
out_string(c, "NOT_FOUND");
return;
}
if (exptime == 0) {
item_unlink(it);
out_string(c, "DELETED");
return;
}
if (delcurr >= deltotal) {
item **new_delete = realloc(todelete, sizeof(item *) * deltotal * 2);
if (new_delete) {
todelete = new_delete;
deltotal *= 2;
} else {
/*
* can't delete it immediately, user wants a delay,
* but we ran out of memory for the delete queue
*/
out_string(c, "SERVER_ERROR out of memory");
return;
}
}
exptime = realtime(exptime);
it->refcount++;
/* use its expiration time as its deletion time now */
it->exptime = exptime;
it->it_flags |= ITEM_DELETED;
todelete[delcurr++] = it;
out_string(c, "DELETED");
return;
}
if (strncmp(command, "stats", 5) == 0) {
process_stat(c, command);
return;
}
if (strcmp(command, "flush_all") == 0) {
settings.oldest_live = time(0);
out_string(c, "OK");
return;
}
if (strcmp(command, "version") == 0) {
out_string(c, "VERSION " VERSION);
return;
}
if (strcmp(command, "quit") == 0) {
c->state = conn_closing;
return;
}
if (strncmp(command, "slabs reassign ", 15) == 0) {
int src, dst;
char *start = command+15;
if (sscanf(start, "%u %u\r\n", &src, &dst) == 2) {
int rv = slabs_reassign(src, dst);
if (rv == 1) {
out_string(c, "DONE");
return;
}
if (rv == 0) {
out_string(c, "CANT");
return;
}
if (rv == -1) {
out_string(c, "BUSY");
return;
}
}
out_string(c, "CLIENT_ERROR bogus command");
return;
}
out_string(c, "ERROR");
return;
}
/* 从 slab 系统分配一个空闲 item */
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);
if (settings.use_cas) {
ntotal += sizeof(uint64_t);
}
unsigned int id = slabs_clsid(ntotal);
if (id == 0)
return 0;
mutex_lock(&cache_lock);
/* do a quick check if we have any expired items in the tail.. */
int tries = 5;
int tried_alloc = 0;
item *search;
void *hold_lock = NULL;
rel_time_t oldest_live = settings.oldest_live;
search = tails[id];
/* We walk up *only* for locked items. Never searching for expired.
* Waste of CPU for almost all deployments */
for (; tries > 0 && search != NULL; tries--, search=search->prev)
{
uint32_t hv = hash(ITEM_key(search), search->nkey, 0);
/* Attempt to hash item lock the "search" item. If locked, no
* other callers can incr the refcount
*/
/* FIXME: I think we need to mask the hv here for comparison? */
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 */
if (search->time + 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;
}
/* 先检查 LRU 队列最后一个 item 是否超时, 超时的话就把这个 item 分配给用户 */
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;
slabs_adjust_mem_requested(it->slabs_clsid, ITEM_ntotal(it), ntotal);
/* 把这个 item 从 LRU 队列和哈希表中移除 */
do_item_unlink_nolock(it, hv);
/* Initialize the item block: */
it->slabs_clsid = 0;
}
/* 没有超时的item, 那就尝试从slabclass分配, 运气不好的话, 分配失败, 那就把 LRU 队列最后一个 item 剔除, 然后分配给用户 */
else if ((it = slabs_alloc(ntotal, id)) == NULL)
{
tried_alloc = 1;
if (settings.evict_to_free == 0)
{
itemstats[id].outofmemory++;//显示出的统计信息
}
else
{
itemstats[id].evicted++;//这个slab的分配失败次数加1，后面的分析统计信息的线程会用到这个统计信息
itemstats[id].evicted_time = current_time - search->time;//显示的统计信息
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);//不用请求新的item了，减少相关的统计信息
/* 把老的item从hash表和lru队列中删除 */
do_item_unlink_nolock(it, hv);
/* Initialize the item block: */
it->slabs_clsid = 0;
/* If we've just 回收 an item, and the automover is set to angry bird mode, attempt to rip memory into this
slab class. TODO: Move valid object detection into a function, and on a "successful" memory pull, look
behind and see if the next alloc would be an eviction. Then kick off the slab mover before the eviction
happens.可以看到如果slab_automove=2(默认是1)，这样会导致angry模式，就是只要分配失败了，马上就选择一个slab(旧的slagclass
释放的)，把这个slab移动到当前slab-class中（不会有通过统计信息有选择的移动slab）*/
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;
}
if (!tried_alloc && (tries == 0 || search == NULL)) it = slabs_alloc(ntotal, id);
if (it == NULL)
{
itemstats[id].outofmemory++;
mutex_unlock(&cache_lock);
return NULL;
}
assert(it->slabs_clsid == 0);
assert(it != heads[id]);
/* 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;
}
