Linux 增加 SWAP 空间

一、需求

通过阿里云启动项目时，使用Vuepress build编译静态页面时内存需要800MB，导致内存不够，因此考虑使用swap方式，置换一些内存资源存放swap磁盘。

[root@xxx myblog]# npm run docs:dev

> myblog@1.0.0 docs:dev
> vuepress dev docs

wait Extracting site metadata...
tip Apply theme @vuepress/theme-default ...
warning Invalid value for "plugin": expected a String, Function or Object but got Array.
warning An error was encountered in plugin "@vuepress/back-to-top"
tip Apply plugin container (i.e. "vuepress-plugin-container") ...
tip Apply plugin @vuepress/register-components (i.e. "@vuepress/plugin-register-components") ...
tip Apply plugin @vuepress/active-header-links (i.e. "@vuepress/plugin-active-header-links") ...
tip Apply plugin @vuepress/search (i.e. "@vuepress/plugin-search") ...
tip Apply plugin @vuepress/nprogress (i.e. "@vuepress/plugin-nprogress") ...
tip Apply plugin copyright (i.e. "vuepress-plugin-copyright") ...
tip Apply plugin sitemap (i.e. "vuepress-plugin-sitemap") ...
tip Apply plugin baidu-autopush (i.e. "vuepress-plugin-baidu-autopush") ...
tip Apply plugin @vuepress/medium-zoom (i.e. "@vuepress/plugin-medium-zoom") ...
tip Apply plugin img-lazy (i.e. "vuepress-plugin-img-lazy") ...
tip Apply plugin @vssue/vssue (i.e. "@vssue/vuepress-plugin-vssue") ...
tip Apply plugin one-click-copy (i.e. "vuepress-plugin-one-click-copy") ...

● Client █████████████████████████ building (68%) 2689/2748 modules 59 active
 url-loader › docs/blogs/interview/image/java/link.png

ℹ ｢wds｣: Project is running at http://0.0.0.0:8080/
ℹ ｢wds｣: webpack output is served from /
ℹ ｢wds｣: Content not from webpack is served from /root/project/myblog/docs/.vuepress/public
ℹ ｢wds｣: 404s will fallback to /index.html
Language does not exist: c++
Language does not exist: c++
Language does not exist: c++
Language does not exist: c++
Killed

swap分区是Linux操作系统中的一种虚拟化内存技术，将硬盘空间作为内存使用。由于内存和磁盘的读写性能差异较大，Linux会在内存充裕时将空闲内存用于缓存磁盘数据，以提高I/O性能。相对的在内存紧张时Linux会将这些缓存回收，将脏页回写到磁盘中。而在进程的地址空间中，如heap，stack等匿名页，在磁盘上并没有对应的文件，但同样有回收到磁盘上以释放出空闲内存的需求。swap机制通过在磁盘上开辟专用的swap分区作为匿名页的backing storage，满足了这一需求。

在Linux上可以使用swapon -s命令查看当前系统上正在使用的交换空间有哪些，以及相关信息：

[root@xxx myblog]# swapon
NAME      TYPE SIZE   USED PRIO
/etc/swap file   2G 677.9M   -2

二、SWAP 创建

Linux支持两种形式的swap分区： 使用分区空间swap disk和使用分区文件swap file。前者是一个专用于做swap的块设备，作为裸设备提供给swap机制操作；后者则是存放在文件系统上的一个特定文件，其实现依赖于不同的文件系统，会有所区别。

分区文件swap file

【1】创建swap文件

[root@xxx myblog]# fallocate -l 2G /etc/swap #指定文件为2G

【2】设置该文件为swap文件

[root@xxx myblog]# mkswap /etc/swap
Setting up swapspace version 1,size = 2097148 KiB
no label,UUID=5b9e4232-dad5-4dbd-9805-f2296452e6f8

【3】启动swap文件

[root@xxx myblog]# swapon /etc/swap
swapon: /etc/swap: insecure permissions 0644,0600 suggested.

【4】使swap文件永久生效

vim /etc/fstab

【5】在fstab末尾添加如下内容

/etc/swap  swap   swap  defaults  0 0

【6】更改swap配置

vim /etc/sysctl.conf

【7】添加如下内容：值越大表示越倾向于使用swap空间

vm.swappiness=30

【8】重启生效

init 6

分区空间swap disk

【1】创建分区：并设置为swap格式

fdisk /dev/sdb

参数	说明
n	创建分区
p	创建主分区
1	创建分区1
两次回车	起始扇区和Last扇区选择默认
t	转换分区格式
82	转换为swap空间
p	查看已创建的分区结果
w	保存退出
【2】格式化为swap空间

mkswap /dev/sdb1

【3】启用swap

swapon /dev/sdb1

【4】编辑配置文件，设为开机自动挂载

vim /etc/fstab

【5】fstab中添加如下内容：

/dev/sdb1  swap   swap  defaults  0 0

【6】设置自动启用所有swap空间

swapon -a

【7】重启验证

init 6

可通过swapon和swapoff命令开启或关闭对应的swap分区。通过cat /proc/swaps或swapon -s可以查看使用中的swap分区的状态。

[root@xxx myblog]# swapon -s
Filename                                Type            Size    Used    Priority
/etc/swap                               file    2097148 678772  -2

移除交换(Swap)文件

通过以下命令来移除交换Swap文件，或者通过命令删除/etc/fstab中的交换文件

[root@xxx]# sudo swapoff -v /swapfile

三、整Swappiness值

Swappiness是Linux内核的一个属性，用于定义交换空间的使用频率。如您所知，RAM比硬盘驱动器快。因此，每次您需要使用交换时，您都会注意到某些进程和应用程序运行速度会变慢。但是，您可以调整系统以使用比交换更多的RAM。这有助于提高整体系统性能。通常，默认的swappiness值为30。此值越小，将使用的RAM越多。

要验证swappiness值，请运行以下命令：

[root@xxxx ~]# cat /proc/sys/vm/swappiness
30

如果想要修改swappiness的值，可以编辑/etc/sysctl.conf文件。并添加以下以下内容。

vm.swappiness=20

为了应用更改，则需要重新启动系统。这样Linux内核将使用更多的RAM和更少的交换，但是当你的RAM内存严重满时它仍然会交换。通常，当您的RAM超过4Gb时，建议使用此设置。

四、页面回收机制

Linux触发页面回收有三种情况：
【1】直接回收：alloc_pages()分配物理页，内存紧缺时，会陷入回收机制，同步触发；
【2】周期性回收：当系统内存触发低水位时，唤醒kswapd线程，异步回收内存；
【3】slab收割机制：当内存紧缺时，直接回收，周期性回收，都会调用slab收割机回收，不过这里是内核的内存分配；

kswapd_wait等待队列： 等待队列用于使进程等待某一事件发生，而无需频繁轮询，进程在等待期间睡眠。在某事件发生时，由内核自动唤醒。

setup_arch()-->
	paging_init()-->
	bootmem_init()->
	zone_sizes_init()-->
	free_area_init_node()-->
	free_area_init_core()

kswapd_wait等待队列在free_area_init_core中进行初始化，每个内存节点一个。kswapd内核线程在kswapd_wait等待队列上等待TASK_INTERRUPTIBLE事件发生。

static void __paginginit free_area_init_core(struct pglist_data *pgdat,
        unsigned long node_start_pfn, unsigned long node_end_pfn,
        unsigned long *zones_size, unsigned long *zholes_size)
{
...
    init_waitqueue_head(&pgdat->kswapd_wait);
    init_waitqueue_head(&pgdat->pfmemalloc_wait);
    pgdat_page_ext_init(pgdat);

...
}

kswapd内核线程： kswapd内核线程负责在内存不足的情况下进行页面回收，为每NUMA内存节点创建一个kswap%d的内核线程。其中kswapd函数是内核线程kswapd的入口。

/*
 * 一个pglist_data，对应一个内存节点，是最顶层的内存管理数据结构
 * 主要包括三部分：
 * 1.描述zone
 * 2.描述内存节点的信息；
 * 3.和页面回收相关；
 */
typedef struct pglist_data {
	int node_id;
	wait_queue_head_t kswapd_wait;
	struct task_struct *kswapd; /* Protected by
					   mem_hotplug_begin/end() */
	int kswapd_order;
	enum zone_type kswapd_highest_zoneidx;

	struct lruvec		__lruvec;  ///lru链表向量(包括所有，5种lru链表)

} pg_data_t;

wakeup_kswapd唤醒kswaped内核线程： 分配内存路径上的唤醒函数wakeup_kswapd把kswapd_order和kswapd_highest_zoneidx作为参数传递给kswaped内核线程；

alloc_page()->
	__alloc_pages_nodemask()->
	__alloc_pages_slowpth()->
	wake_all_kswapds()->
	wakeup_kswapd()


void wakeup_kswapd(struct zone *zone, gfp_t gfp_flags, int order,
		   enum zone_type highest_zoneidx)
{
	pg_data_t *pgdat;
	enum zone_type curr_idx;

	if (!managed_zone(zone))
		return;

	if (!cpuset_zone_allowed(zone, gfp_flags))
		return;

	pgdat = zone->zone_pgdat;
	///准备本内存节点的kswapd_order和kswapd_highest_zoneidx
	curr_idx = READ_ONCE(pgdat->kswapd_highest_zoneidx);

	if (curr_idx == MAX_NR_ZONES || curr_idx < highest_zoneidx)
		WRITE_ONCE(pgdat->kswapd_highest_zoneidx, highest_zoneidx);

	if (READ_ONCE(pgdat->kswapd_order) < order)
		WRITE_ONCE(pgdat->kswapd_order, order);

	if (!waitqueue_active(&pgdat->kswapd_wait))
		return;

	/* Hopeless node,leave it to direct reclaim if possible */
	if (pgdat->kswapd_failures >= MAX_RECLAIM_RETRIES ||
		(pgdat_balanced(pgdat, order, highest_zoneidx) &&
		 !pgdat_watermark_boosted(pgdat, highest_zoneidx))) {
		/*
		 * There may be plenty of free memory available,but it's too
		 * fragmented for high-order allocations.  Wake up kcompactd
		 * and rely on compaction_suitable() to determine if it's
		 * needed.	If it fails,it will defer subsequent attempts to
		 * ratelimit its work.
		 */
		if (!(gfp_flags & __GFP_DIRECT_RECLAIM))
			wakeup_kcompactd(pgdat, highest_zoneidx);
		return;
	}

	trace_mm_vmscan_wakeup_kswapd(pgdat->node_id, highest_zoneidx,
					  gfp_flags);
		 
	///唤醒kswapd_wait队列
	wake_up_interruptible(&pgdat->kswapd_wait);
}

回收函数kswapd

static int kswapd(void *p)
{

...
	 ///PF_MEMALLOC允许使用系统预留内存，即不考虑水位
	tsk->flags |= PF_MEMALLOC | PF_SWAPWRITE | PF_KSWAPD;
	for ( ; ; ) {
		bool ret;

	///回收页面数量，2的order次幂
		alloc_order = reclaim_order = READ_ONCE(pgdat->kswapd_order);
	
	///classzone_idx内核线程扫描和回收的最高zone
		highest_zoneidx = kswapd_highest_zoneidx(pgdat,
							highest_zoneidx);

kswapd_try_sleep:
		///睡眠，等待wakeup_kswapd唤醒
		kswapd_try_to_sleep(pgdat, alloc_order, reclaim_order,
					highest_zoneidx);
...
		reclaim_order = balance_pgdat(pgdat,
						highest_zoneidx);
		if (reclaim_order < alloc_order)
			goto kswapd_try_sleep;
	}

	tsk->flags &= ~(PF_MEMALLOC | PF_SWAPWRITE | PF_KSWAPD);

	return 0;
}

kswapd内核线程扫描过程： kswapd()->balance_pgdat()

/*****************************************************************************
  * 回收页面的主函数:
  *
  * highmem->normal->dma,从高端往低端方向,查找处于不平衡状态，
  * 即free_pages <= high_wmark_pagesend_zone的zone
  * 
  * 
  ****************************************************************************/
static int balance_pgdat(pg_data_t *pgdat, int highest_zoneidx)
{
	///用于内存碎片化
	unsigned long nr_boost_reclaim;
...
	nr_boost_reclaim = 0;
	for (i = 0; i <= highest_zoneidx; i++) {
		zone = pgdat->node_zones + i;
		if (!managed_zone(zone))
			continue;

		nr_boost_reclaim += zone->watermark_boost;
		zone_boosts[i] = zone->watermark_boost;
	}
	boosted = nr_boost_reclaim;

restart:
	sc.priority = DEF_PRIORITY;
	do {
		...
		 ///检查这个节点中是否有合格的zone，其水位高于高水位且能分配2的sc.order次幂个连续的物理页面
		balanced = pgdat_balanced(pgdat, sc.order, highest_zoneidx);

		///若所有zone都不合格，关闭nr_boost_reclaim，重新检查一次
		if (!balanced && nr_boost_reclaim) {
			nr_boost_reclaim = 0;
			goto restart;
		}

		 //若符合条件，不需要回收，直接跳出
		if (!nr_boost_reclaim && balanced)
			goto out;

...
		///老化匿名页面的活跃链表
		age_active_anon(pgdat, &sc);

...
		 ///真正扫描和页回收函数，扫描的参数和结果存放在struct scan_control中，
		 ///返回true表明回收了所需要的页面，不需要再提高扫描优先级
		if (kswapd_shrink_node(pgdat, &sc))
			raise_priority = false;

...
		///加大扫描粒度
		if (raise_priority || !nr_reclaimed)
			sc.priority--;
	} while (sc.priority >= 1);

...

out:
	/* If reclaim was boosted,account for the reclaim done in this pass */

///若设置了nr_boost_reclaim，唤醒kcompacted线程
	if (boosted) {
		...
		wakeup_kcompactd(pgdat, pageblock_order, highest_zoneidx);
	}

	...
	return sc.order;
}

对活跃链表中页面的老化：kswapd()->balance_pgdat()->age_active_anon()

///老化匿名页面的活跃链表
static void age_active_anon(struct pglist_data *pgdat,
				struct scan_control *sc)
{
	struct mem_cgroup *memcg;
	struct lruvec *lruvec;

	if (!total_swap_pages)
		return;

	lruvec = mem_cgroup_lruvec(NULL, pgdat);
	if (!inactive_is_low(lruvec, LRU_INACTIVE_ANON))
		return;

	memcg = mem_cgroup_iter(NULL, NULL, NULL);
	do {
		lruvec = mem_cgroup_lruvec(memcg, pgdat);
		shrink_active_list(SWAP_CLUSTER_MAX, lruvec,
				   sc, LRU_ACTIVE_ANON);
		memcg = mem_cgroup_iter(NULL, memcg, NULL);
	} while (memcg);
}

执行回收：kswapd()->balance_pgdat()->kswapd_shrink_node()->shrink_node()->shrink_node_memcgs()

static void shrink_node_memcgs(pg_data_t *pgdat, struct scan_control *sc)
{
	struct mem_cgroup *target_memcg = sc->target_mem_cgroup;
	struct mem_cgroup *memcg;

	memcg = mem_cgroup_iter(target_memcg, NULL);
	do {
		///获取LRU链表的集合
		struct lruvec *lruvec = mem_cgroup_lruvec(memcg, pgdat);
		unsigned long reclaimed;
		unsigned long scanned;

		/*
		 * This loop can become CPU-bound when target memcgs
		 * aren't eligible for reclaim - either because they
		 * don't have any reclaimable pages,or because their
		 * memory is explicitly protected. Avoid soft lockups.
		 */
		cond_resched();

		mem_cgroup_calculate_protection(target_memcg, memcg);

		if (mem_cgroup_below_min(memcg)) {
			/*
			 * Hard protection.
			 * If there is no reclaimable memory,OOM.
			 */
			continue;
		} else if (mem_cgroup_below_low(memcg)) {
			/*
			 * Soft protection.
			 * Respect the protection only as long as
			 * there is an unprotected supply
			 * of reclaimable memory from other cgroups.
			 */
			if (!sc->memcg_low_reclaim) {
				sc->memcg_low_skipped = 1;
				continue;
			}
			memcg_memory_event(memcg, MEMCG_LOW);
		}

		reclaimed = sc->nr_reclaimed;
		scanned = sc->nr_scanned;

		///扫描回收lru链表
		shrink_lruvec(lruvec, sc);

		///扫描回收slab链表
		shrink_slab(sc->gfp_mask, pgdat->node_id,
			    sc->priority);

		/* Record the group's reclaim efficiency */
		vmpressure(sc->gfp_mask, false,
			   sc->nr_scanned - scanned,
			   sc->nr_reclaimed - reclaimed);

	} while ((memcg = mem_cgroup_iter(target_memcg, NULL)));
}

回收函数shrink_lruvec()

static void shrink_lruvec(struct lruvec *lruvec, struct scan_control *sc)
{
	unsigned long nr[NR_LRU_LISTS];
	unsigned long targets[NR_LRU_LISTS];
	unsigned long nr_to_scan;
	enum lru_list lru;
	unsigned long nr_reclaimed = 0;
	unsigned long nr_to_reclaim = sc->nr_to_reclaim;
	struct blk_plug plug;
	bool scan_adjusted;

	///计算每个链表应该扫描的页面数量，结果放在nr[]
	get_scan_count(lruvec, sc, nr);

	  ///全局回收，优化当内存紧缺时，触发直接回收
	scan_adjusted = (!cgroup_reclaim(sc) && !current_is_kswapd() &&
			 sc->priority == DEF_PRIORITY);


	///遍历所有链表，回收页面
	///主要处理不活跃匿名页面，活跃文件映射页面和不活跃文件映射页面
	while (nr[LRU_INACTIVE_ANON] || nr[LRU_ACTIVE_FILE] ||
					nr[LRU_INACTIVE_FILE]) {
		unsigned long nr_anon, nr_file, percentage;
		unsigned long nr_scanned;

		for_each_evictable_lru(lru) {
			if (nr[lru]) {
				nr_to_scan = min(nr[lru], SWAP_CLUSTER_MAX);
				nr[lru] -= nr_to_scan;
			
				//扫描链表，回收页面，返回成功回收的页面数量
				nr_reclaimed += shrink_list(lru, nr_to_scan,
								lruvec, sc);
			}
		}

		cond_resched();

		///没完成回收任务，或设置了scan_adjusted，继续进行页面扫描
		if (nr_reclaimed < nr_to_reclaim || scan_adjusted)
			continue;

		...
		scan_adjusted = true;
	}
	blk_finish_plug(&plug);
	sc->nr_reclaimed += nr_reclaimed;

	 ///老化活跃链表
	 ///如果不活跃链表页面数量太少，从活跃链表迁移一部分页面到不活跃链表
	if (total_swap_pages && inactive_is_low(lruvec, LRU_INACTIVE_ANON))
		shrink_active_list(SWAP_CLUSTER_MAX, LRU_ACTIVE_ANON);
}

shrink_lruvec()->shrink_list()

static unsigned long shrink_list(enum lru_list lru, unsigned long nr_to_scan,
					 struct lruvec *lruvec, struct scan_control *sc)
	{
		if (is_active_lru(lru)) {
			///扫描活跃的文件映射链表
			if (sc->may_deactivate & (1 << is_file_lru(lru)))
				shrink_active_list(nr_to_scan, lru);
			else
				sc->skipped_deactivate = 1;
			return 0;
		}
	
	///扫描不活跃链表
		return shrink_inactive_list(nr_to_scan, lru);
	}

扫描活跃链表函数shrink_active_list()实现:

/*************************************************************************************
 * func:扫描活跃链表，包括匿名页或文件映射页面，
 *      把最近没访问的页面，从活跃链表尾部移到不活跃链表头部
 * nr_to_scan: 待扫描页面的数量
 * lruvec：LRU链表集合
 * sc：页面扫描控制参数
*  lru： 待扫描的LRU链表类型
*************************************************************************************/
static void shrink_active_list(unsigned long nr_to_scan,
			       struct lruvec *lruvec,
			       struct scan_control *sc,
			       enum lru_list lru)
{
	unsigned long nr_taken;
	unsigned long nr_scanned;
	unsigned long vm_flags;
	///定义三个临时链表
	LIST_HEAD(l_hold);	/* The pages which were snipped off */
	LIST_HEAD(l_active);
	LIST_HEAD(l_inactive);
	struct page *page;
	unsigned nr_deactivate, nr_activate;
	unsigned nr_rotated = 0;

	///判断是否为文件映射链表
	int file = is_file_lru(lru);

	///获取内存节点
	struct pglist_data *pgdat = lruvec_pgdat(lruvec);

	lru_add_drain();

	spin_lock_irq(&lruvec->lru_lock);

	///将页面批量迁移到临时链表l_hold中
	nr_taken = isolate_lru_pages(nr_to_scan, &l_hold,
				     &nr_scanned, lru);


	///增加内存节点NR_ISOLATED_ANON计数
	__mod_node_page_state(pgdat, NR_ISOLATED_ANON + file, nr_taken);

	if (!cgroup_reclaim(sc))
		__count_vm_events(PGREFILL, nr_scanned);
	__count_memcg_events(lruvec_memcg(lruvec), PGREFILL, nr_scanned);

	spin_unlock_irq(&lruvec->lru_lock);

	///扫描临时链表l_hold，有些页面放到不活跃链表，有些会放回到活跃链表
	while (!list_empty(&l_hold)) {
		cond_resched();
		page = lru_to_page(&l_hold);
		list_del(&page->lru);

		///如果不能回收，放入不能回收链表
		if (unlikely(!page_evictable(page))) {
			putback_lru_page(page);
			continue;
		}

		if (unlikely(buffer_heads_over_limit)) {
			if (page_has_private(page) && trylock_page(page)) {
				if (page_has_private(page))
					try_to_release_page(page, 0);
				unlock_page(page);
			}
		}
		
		///page_referenced()返回该页面最近访问，应用pte个数，若返回0，表示最近没访问
		if (page_referenced(page, 0, sc->target_mem_cgroup,
				    &vm_flags)) {
			/*
			 * Identify referenced,file-backed active pages and
			 * give them one more trip around the active list. So
			 * that executable code get better chances to stay in
			 * memory under moderate memory pressure.  Anon pages
			 * are not likely to be evicted by use-once streaming
			 * IO,plus JVM can create lots of anon VM_EXEC pages,* so we ignore them here.
			 */
			if ((vm_flags & VM_EXEC) && page_is_file_lru(page)) {
				nr_rotated += thp_nr_pages(page);
				///放回活跃链表
				list_add(&page->lru, &l_active); 
				continue;
			}
		}

		ClearPageActive(page);	/* we are de-activating */
		SetPageWorkingset(page);
		///加入不活跃链表
		list_add(&page->lru, &l_inactive);
	}

	/*
	 * Move pages back to the lru list.
	 */
	spin_lock_irq(&lruvec->lru_lock);

	///将l_active,l_inactive分别加入到相应的链表
	nr_activate = move_pages_to_lru(lruvec, &l_active);
	nr_deactivate = move_pages_to_lru(lruvec, &l_inactive);
	/* Keep all free pages in l_active list */
	list_splice(&l_inactive, &l_active);

	__count_vm_events(PGDEACTIVATE, nr_deactivate);
	__count_memcg_events(lruvec_memcg(lruvec), PGDEACTIVATE, nr_deactivate);

	__mod_node_page_state(pgdat, -nr_taken);
	spin_unlock_irq(&lruvec->lru_lock);

	mem_cgroup_uncharge_list(&l_active);
	free_unref_page_list(&l_active);
	trace_mm_vmscan_lru_shrink_active(pgdat->node_id, nr_taken, nr_activate,
			nr_deactivate, nr_rotated, sc->priority, file);
}

扫描不活跃链表shrink_inactive_list()实现:

///扫描不活跃LRU链表，尝试回收页面，返回已经回收的页面数量
static noinline_for_stack unsigned long
shrink_inactive_list(unsigned long nr_to_scan, struct lruvec *lruvec,
		     struct scan_control *sc, enum lru_list lru)
{
	LIST_HEAD(page_list);
	unsigned long nr_scanned;
	unsigned int nr_reclaimed = 0;
	unsigned long nr_taken;
	struct reclaim_stat stat;
	bool file = is_file_lru(lru);
	enum vm_event_item item;
	struct pglist_data *pgdat = lruvec_pgdat(lruvec);
	bool stalled = false;

	while (unlikely(too_many_isolated(pgdat, file, sc))) {
		if (stalled)
			return 0;

		/* wait a bit for the reclaimer. */
		///太多进程在直接回收页面，睡眠，避免内存抖动
		msleep(100);  
		stalled = true;

		/* We are about to die and free our memory. Return now. */
		if (fatal_signal_pending(current))
			return SWAP_CLUSTER_MAX;
	}

	lru_add_drain();

	spin_lock_irq(&lruvec->lru_lock);

///分离页面到临时页表
	nr_taken = isolate_lru_pages(nr_to_scan, &page_list, lru);

	///增加内存节点NR_ISOLATED_ANON计数
	__mod_node_page_state(pgdat, nr_taken);
	item = current_is_kswapd() ? PGSCAN_KSWAPD : PGSCAN_DIRECT;
	if (!cgroup_reclaim(sc))
		__count_vm_events(item, item, nr_scanned);
	__count_vm_events(PGSCAN_ANON + file, nr_scanned);

	spin_unlock_irq(&lruvec->lru_lock);

	if (nr_taken == 0)
		return 0;

	///执行回收页面，返回nr_reclaimed个
	nr_reclaimed = shrink_page_list(&page_list, pgdat, &stat, false);

	spin_lock_irq(&lruvec->lru_lock);

	///page_list链表剩余页面迁回不活跃链表
	move_pages_to_lru(lruvec, &page_list);
	
	///减少NR_ISOLATED_ANON计数
	__mod_node_page_state(pgdat, -nr_taken);
	item = current_is_kswapd() ? PGSTEAL_KSWAPD : PGSTEAL_DIRECT;
	if (!cgroup_reclaim(sc))
		__count_vm_events(item, nr_reclaimed);
	__count_memcg_events(lruvec_memcg(lruvec), nr_reclaimed);
	__count_vm_events(PGSTEAL_ANON + file, nr_reclaimed);
	spin_unlock_irq(&lruvec->lru_lock);

	lru_note_cost(lruvec, stat.nr_pageout);
	mem_cgroup_uncharge_list(&page_list);
	free_unref_page_list(&page_list);

	/*
	 * If dirty pages are scanned that are not queued for IO,it
	 * implies that flushers are not doing their job. This can
	 * happen when memory pressure pushes dirty pages to the end of
	 * the LRU before the dirty limits are breached and the dirty
	 * data has expired. It can also happen when the proportion of
	 * dirty pages grows not through writes but through memory
	 * pressure reclaiming all the clean cache. And in some cases,* the flushers simply cannot keep up with the allocation
	 * rate. Nudge the flusher threads in case they are asleep.
	 */
	if (stat.nr_unqueued_dirty == nr_taken)
		wakeup_flusher_threads(WB_REASON_VMSCAN);

	sc->nr.dirty += stat.nr_dirty;
	sc->nr.congested += stat.nr_congested;
	sc->nr.unqueued_dirty += stat.nr_unqueued_dirty;
	sc->nr.writeback += stat.nr_writeback;
	sc->nr.immediate += stat.nr_immediate;
	sc->nr.taken += nr_taken;
	if (file)
		sc->nr.file_taken += nr_taken;

	trace_mm_vmscan_lru_shrink_inactive(pgdat->node_id,
			nr_scanned, nr_reclaimed, file);
	return nr_reclaimed;
}

原文地址：https://blog.csdn.net/zhengzhaoyang122/article/details/135889064