Linux页高速缓存缓存回写机制，基于2.6版本内核

注：2.6内核版本使用下列方式，旧版本内核可以参考

内核版本3.2页回写机制可参考：

https://blog.csdn.net/SweeNeil/article/details/84993387

之后版本的页回写机制还有待考证。

本文转自：

http://oenhan.com/linux-cache-writeback

在做进程安全监控的时候，拍脑袋决定的，如果发现一个进程在D状态时，即TASK_UNINTERRUPTIBLE（不可中断的睡眠状态），时间超过了8min，就将系统panic掉。恰好DB组做日志时，将整个log缓存到内存中，最后刷磁盘，结果系统就D状态了很长时间，自然panic了，中间涉及到Linux的缓存写回刷磁盘的一些机制和调优方法，写一下总结。

目前机制需要将脏页刷回到磁盘一般是以下情况：

1. 脏页缓存占用的内存太多，内存空间不足；
2. 脏页已经更改了很长时间，时间上已经到了临界值，需要及时刷新保持内存和磁盘上数据一致性；
3. 外界命令强制刷新脏页到磁盘
4. write写磁盘时检查状态刷新

内核使用pdflush线程刷新脏页到磁盘，pdflush线程个数在2和8之间，可以通过/proc/sys/vm/nr_pdflush_threads文件直接查看，具体策略机制参看源码函数__pdflush。

一、内核其他模块强制刷新

先说一下第一种和第三种情况：当内存空间不足或外界强制刷新的时候，脏页的刷新是通过调用wakeup_pdflush函数实现的，调用其函数的有do_sync、free_more_memory、try_to_free_pages。wakeup_pdflush的功能是通过background_writeout的函数实现的：

static void background_writeout(unsigned long _min_pages)
{
     long min_pages = _min_pages;
     struct writeback_control wbc = {
         .bdi = NULL,
         .sync_mode = WB_SYNC_NONE,
         .older_than_this = NULL,
         .nr_to_write = 0,
         .nonblocking = 1,
    };

    for ( ; ; ) {
         struct writeback_state wbs;
         long background_thresh;
         long dirty_thresh;

         get_dirty_limits(&wbs, &background_thresh, &dirty_thresh, NULL);
         if (wbs.nr_dirty + wbs.nr_unstable < background_thresh
             && min_pages <= 0)
             break;
         wbc.encountered_congestion = 0;
         wbc.nr_to_write = MAX_WRITEBACK_PAGES;
         wbc.pages_skipped = 0;
         writeback_inodes(&wbc);
         min_pages -= MAX_WRITEBACK_PAGES - wbc.nr_to_write;
         if (wbc.nr_to_write > 0 || wbc.pages_skipped > 0) {
              /* Wrote less than expected */
              blk_congestion_wait(WRITE, HZ/10);
              if (!wbc.encountered_congestion)
                  break;
         }
   }
}

background_writeout进到一个死循环里面，通过get_dirty_limits获取脏页开始刷新的临界值background_thresh，即为dirty_background_ratio的总内存页数百分比，可以通过proc接口/proc/sys/vm/dirty_background_ratio调整，一般默认为10。当脏页超过临界值时，调用writeback_inodes写MAX_WRITEBACK_PAGES（1024）个页，直到脏页比例低于临界值。

二、内核定时器启动刷新

内核在启动的时候在page_writeback_init初始化wb_timer定时器，超时时间是dirty_writeback_centisecs，单位是0.01秒，可以通过/proc/sys/vm/dirty_writeback_centisecs调节。wb_timer的触发函数是wb_timer_fn，最终是通过wb_kupdate实现。

static void wb_kupdate(unsigned long arg)
{
    sync_supers();
    get_writeback_state(&wbs);
    oldest_jif = jiffies - (dirty_expire_centisecs * HZ) / 100;
    start_jif = jiffies;
    next_jif = start_jif + (dirty_writeback_centisecs * HZ) / 100;
    nr_to_write = wbs.nr_dirty + wbs.nr_unstable +
    (inodes_stat.nr_inodes - inodes_stat.nr_unused);
    while (nr_to_write > 0) {
        wbc.encountered_congestion = 0;
        wbc.nr_to_write = MAX_WRITEBACK_PAGES;
        writeback_inodes(&wbc);
        if (wbc.nr_to_write > 0) {
            if (wbc.encountered_congestion)
                blk_congestion_wait(WRITE, HZ/10);
            else
                break; /* All the old data is written */
        }
        nr_to_write -= MAX_WRITEBACK_PAGES - wbc.nr_to_write;
    }
    if (time_before(next_jif, jiffies + HZ))
    next_jif = jiffies + HZ;
    if (dirty_writeback_centisecs)
    mod_timer(&wb_timer, next_jif);
 }

上面的代码没有拷贝全。内核首先将超级块信息刷新到文件系统上，然后获取oldest_jif作为wbc的参数只刷新已修改时间大于dirty_expire_centisecs的脏页，dirty_expire_centisecs参数可以通过/proc/sys/vm/dirty_expire_centisecs调整。

三、WRITE写文件刷新缓存

用户态使用WRITE函数写文件时也有可能要刷新脏页，generic_file_buffered_write函数会在将写的内存页标记为脏之后，根据条件刷新磁盘以平衡当前脏页比率，参看balance_dirty_pages_ratelimited函数：

void balance_dirty_pages_ratelimited(struct address_space *mapping)
{
    static DEFINE_PER_CPU(int, ratelimits) = 0;
    long ratelimit;

    ratelimit = ratelimit_pages;
    if (dirty_exceeded)
        ratelimit = 8;

    /*
     * Check the rate limiting. Also, we do not want to throttle real-time
     * tasks in balance_dirty_pages(). Period.
     */
    if (get_cpu_var(ratelimits)++ >= ratelimit) {
        __get_cpu_var(ratelimits) = 0;
        put_cpu_var(ratelimits);
        balance_dirty_pages(mapping);
        return;
    }
    put_cpu_var(ratelimits);
}

balance_dirty_pages_ratelimited函数通过ratelimit_pages调节刷新（调用balance_dirty_pages函数）的次数，每ratelimit_pages次调用才会刷新一次，具体刷新过程看balance_dirty_pages函数：

static void balance_dirty_pages(struct address_space *mapping)
{
    struct writeback_state wbs;
    long nr_reclaimable;
    long background_thresh;
    long dirty_thresh;
    unsigned long pages_written = 0;
    unsigned long write_chunk = sync_writeback_pages();

    struct backing_dev_info *bdi = mapping->backing_dev_info;

    for (;;) {
        struct writeback_control wbc = {
            .bdi        = bdi,
            .sync_mode  = WB_SYNC_NONE,
            .older_than_this = NULL,
            .nr_to_write    = write_chunk,
        };

        get_dirty_limits(&wbs, &background_thresh,
                    &dirty_thresh, mapping);
        nr_reclaimable = wbs.nr_dirty + wbs.nr_unstable;
        if (nr_reclaimable + wbs.nr_writeback <= dirty_thresh)
            break;

        if (!dirty_exceeded)
            dirty_exceeded = 1;

        /* Note: nr_reclaimable denotes nr_dirty + nr_unstable.
         * Unstable writes are a feature of certain networked
         * filesystems (i.e. NFS) in which data may have been
         * written to the server's write cache, but has not yet
         * been flushed to permanent storage.
         */
        if (nr_reclaimable) {
            writeback_inodes(&wbc);
            get_dirty_limits(&wbs, &background_thresh,
                    &dirty_thresh, mapping);
            nr_reclaimable = wbs.nr_dirty + wbs.nr_unstable;
            if (nr_reclaimable + wbs.nr_writeback <= dirty_thresh)
                break;
            pages_written += write_chunk - wbc.nr_to_write;
            if (pages_written >= write_chunk)
                break;      /* We've done our duty */
        }
        blk_congestion_wait(WRITE, HZ/10);
    }

    if (nr_reclaimable + wbs.nr_writeback <= dirty_thresh && dirty_exceeded)
        dirty_exceeded = 0;

    if (writeback_in_progress(bdi))
        return;     /* pdflush is already working this queue */

    /*
     * In laptop mode, we wait until hitting the higher threshold before
     * starting background writeout, and then write out all the way down
     * to the lower threshold.  So slow writers cause minimal disk activity.
     *
     * In normal mode, we start background writeout at the lower
     * background_thresh, to keep the amount of dirty memory low.
     */
    if ((laptop_mode && pages_written) ||
         (!laptop_mode && (nr_reclaimable > background_thresh)))
        pdflush_operation(background_writeout, 0);
}

函数走进一个死循环，通过get_dirty_limits获取dirty_background_ratio和dirty_ratio对应的内存页数值，当24行做判断，如果脏页大于dirty_thresh，则调用writeback_inodes开始刷缓存到磁盘，如果一次没有将脏页比率刷到dirty_ratio之下，则用blk_congestion_wait阻塞写，然后反复循环，直到比率降低到dirty_ratio；当比率低于dirty_ratio之后，但脏页比率大于dirty_background_ratio，则用pdflush_operation启用background_writeout，pdflush_operation是非阻塞函数，唤醒pdflush后直接返回，background_writeout在有pdflush调用。

如此可知：WRITE写的时候，缓存超过dirty_ratio，则会阻塞写操作，回刷脏页，直到缓存低于dirty_ratio；如果缓存高于background_writeout，则会在写操作时，唤醒pdflush进程刷脏页，不阻塞写操作。

四，问题总结

导致进程D状态大部分是因为第3种和第4种情况：有大量写操作，缓存由Linux系统管理，一旦脏页累计到一定程度，无论是继续写还是fsync刷新，都会使进程D住。