Linux内核文件系统-页高速缓存

Linux内核的VFS是非常经典的抽象，不仅抽象出了flesystem，super_block，inode，dentry，file等结构，而且还提供了像页高速缓存层的通用接口，当然，你可以自己选择是否使用或者自己定制使用方式。本文主要根据自己阅读Linux Kernel 3.19.3系统调用read相关的源码来追踪页高速缓存在整个流程中的痕迹，以常规文件的页高速缓存为例，了解页高速缓存的实现过程，不过于追究具体bio请求的底层细节。另外，在写操作的过程中，页高速缓存的处理流程有所不同(回写)，涉及的东西更多，本文主要关注读操作。Linux VFS相关的重要数据结构及概念可以参考Document目录下的vfs.txt。

---

1.与页高速缓存相关的重要数据结构

除了前述基本数据结构以外，struct address_space 和 struct address_space_operations也在页高速缓存中起着极其重要的作用。

• address_space结构通常被struct page的一个字段指向，主要存放已缓存页面的相关信息，便于快速查找对应文件的缓存页面，具体查找过程是通过radix tree结构的相关操作实现的。
• address_space_operations结构定义了具体读写页面等操作的钩子，比如生成并发送bio请求，我们可以定制相应的函数实现自己的读写逻辑。

//include/linux/fs.h
struct address_space {
    //指向文件的inode，可能为NULL
    struct inode        *host;  
    //存放装有缓存数据的页面
    struct radix_tree_root  page_tree;  
    spinlock_t      tree_lock;  
    atomic_t        i_mmap_writable;
    struct rb_root      i_mmap; 
    struct list_head    i_mmap_nonlinear;
    struct rw_semaphore i_mmap_rwsem;
    //已缓存页的数量
    unsigned long       nrpages;    
    unsigned long       nrshadows;  
    pgoff_t         writeback_index;
    //address_space相关操作，定义了具体读写页面的钩子
    const struct address_space_operations *a_ops;   
    unsigned long       flags;  
    struct backing_dev_info *backing_dev_info; 
    spinlock_t      private_lock;   
    struct list_head    private_list;   
    void            *private_data;
} __attribute__((aligned(sizeof(long))));

//include/linux/fs.h 
struct address_space_operations {
    //具体写页面的操作
    int (*writepage)(struct page *page, struct writeback_control *wbc);
    //具体读页面的操作
    int (*readpage)(struct file *, struct page *);
    int (*writepages)(struct address_space *, struct writeback_control *);
    //标记页面脏
    int (*set_page_dirty)(struct page *page);
    int (*readpages)(struct file *filp, struct address_space *mapping, struct list_head *pages, unsigned nr_pages);
    int (*write_begin)(struct file *, struct address_space  *mapping, loff_t pos, unsigned len, unsigned flags, struct page **pagep, void **fsdata);
    int (*write_end)(struct file *, struct address_space *mapping, loff_t pos, unsigned len, unsigned copied, struct page *page, void *fsdata);
    sector_t (*bmap)(struct address_space *, sector_t);
    void (*invalidatepage) (struct page *, unsigned int, unsigned int);
    int (*releasepage) (struct page *, gfp_t);
    void (*freepage)(struct page *);
    ssize_t (*direct_IO)(int, struct kiocb *, struct iov_iter *iter, loff_t offset);
    int (*get_xip_mem)(struct address_space *, pgoff_t, int, void **, unsigned long *);

    int (*migratepage) (struct address_space *, struct page *, struct page *, enum migrate_mode);
    int (*launder_page) (struct page *);
    int (*is_partially_uptodate) (struct page *, unsigned long, unsigned long);
    void (*is_dirty_writeback) (struct page *, bool *, bool *);
    int (*error_remove_page)(struct address_space *, struct page *);
    /* swapfile support */
    int (*swap_activate)(struct swap_info_struct *sis, struct file *file, sector_t *span);
    void (*swap_deactivate)(struct file *file);
};

---

2.系统调用read流程与页高速缓存相关代码分析

关于挂载和打开文件的操作，不赘述(涉及的细节也很多…)，(极其)简陋地理解，挂载返回挂载点的root dentry，并且读取磁盘数据生成了super_block链接到全局超级块链表中，这样，当前进程就可以通过root dentry找到其inode，从而找到并生成其子树的dentry和inode信息，从而实现查找路径的逻辑。打开文件简单理解就是分配fd，通过dentry将file结构与对应inode挂接，最后安装到进程的打开文件数组中，这里假设已经成功打开文件，返回了fd，我们从系统调用read开始。

• 定义系统调用read

//定义系统调用read
//fs/read_write.c
SYSCALL_DEFINE3(read, unsigned int, fd, char __user *, buf, size_t, count)
{
    //根据fd number获得struct fd
    struct fd f = fdget_pos(fd);
    ssize_t ret = -EBADF;
    if (f.file) {
        //偏移位置
        loff_t pos = file_pos_read(f.file);
        //进入vfs_read
        //参数：file指针，用户空间buffer指针，长度，偏移位置
        //主要做一些验证工作，最后进入__vfs_read
        ret = vfs_read(f.file, buf, count, &pos);
        if (ret >= 0)
            file_pos_write(f.file, pos);
        fdput_pos(f);
    }
    return ret;
}

• 进入__vfs_read

//fs/read_write.c
ssize_t __vfs_read(struct file *file, char __user *buf, size_t count, loff_t *pos)
{
    ssize_t ret;
    //注意这，我们可以在file_operations中定义自己的read操作，使不使用页高速缓存可以自己控制
    if (file->f_op->read)
        ret = file->f_op->read(file, buf, count, pos);
    else if (file->f_op->aio_read)
        //会调用f_ops->read_iter
        ret = do_sync_read(file, buf, count, pos);
    else if (file->f_op->read_iter)
        //会调用f_ops->read_iter
        //这里ext2中又将read_iter直接与generic_file_read_iter挂接，使用内核自带的read操作，稍后会以ext2为例分析
        ret = new_sync_read(file, buf, count, pos);
    else
        ret = -EINVAL;
    return ret;
}

• 以ext2为例，进入ext2的file_operations->read

//fs/ext2/file.c
const struct file_operations ext2_file_operations = {
    .llseek     = generic_file_llseek,
    .read       = new_sync_read,  //重定向到read_iter此处即generic_file_read_iter
    .write      = new_sync_write,
    .read_iter  = generic_file_read_iter, //使用内核自带的通用读操作，这里会进入页高速缓冲的部分
    .write_iter = generic_file_write_iter,
    .unlocked_ioctl = ext2_ioctl,
#ifdef CONFIG_COMPAT
    .compat_ioctl   = ext2_compat_ioctl,
#endif
    .mmap       = generic_file_mmap,
    .open       = dquot_file_open,
    .release    = ext2_release_file,
    .fsync      = ext2_fsync,
    .splice_read    = generic_file_splice_read,
    .splice_write   = iter_file_splice_write,
};

• 进入generic_file_read_iter

ssize_t
generic_file_read_iter(struct kiocb *iocb, struct iov_iter *iter)
{
    struct file *file = iocb->ki_filp;
    ssize_t retval = 0;
    loff_t *ppos = &iocb->ki_pos;
    loff_t pos = *ppos;
    /* coalesce the iovecs and go direct-to-BIO for O_DIRECT */
    if (file->f_flags & O_DIRECT) {
        struct address_space *mapping = file->f_mapping;
        struct inode *inode = mapping->host;
        size_t count = iov_iter_count(iter);
        loff_t size;
        if (!count)
            goto out; /* skip atime */
        size = i_size_read(inode);
        //先写？
        retval = filemap_write_and_wait_range(mapping, pos,
                    pos + count - 1);
        if (!retval) {
            struct iov_iter data = *iter;
            retval = mapping->a_ops->direct_IO(READ, iocb, &data, pos);
        }
        if (retval > 0) {
            *ppos = pos + retval;
            iov_iter_advance(iter, retval);
        }

        /*
         * Btrfs can have a short DIO read if we encounter
         * compressed extents, so if there was an error, or if
         * we've already read everything we wanted to, or if
         * there was a short read because we hit EOF, go ahead
         * and return.  Otherwise fallthrough to buffered io for
         * the rest of the read.
         */
        if (retval < 0 || !iov_iter_count(iter) || *ppos >= size) {
            file_accessed(file);
            goto out;
        }
    }
    //进入真正read,在address_space的radix tree中查找
    //偏移的page，如果找到，直接copy到用户空间如果未找到，
    //则调用a_ops->readpage读取发起bio，分配cache page，
    //读入数据，加入radix,然后拷贝到用户空间，完成读取数据的过程.
    retval = do_generic_file_read(file, ppos, iter, retval);
out:
    return retval;
}
EXPORT_SYMBOL(generic_file_read_iter);

• 进入do_generic_file_read
  这个函数基本是整个页高速缓存的核心了，在具体的bio操作请求操作之前判断是否存在缓存页面，如果存在拷贝数据到用户空间，否则分配新页面，调用具体文件系统address_space_operations->readpage读取块数据到页面中，并且加入到radix tree中。

static ssize_t do_generic_file_read(struct file *filp, loff_t *ppos,struct iov_iter *iter, ssize_t written)
{
    /* 省略部分 */
    for (;;) {
        struct page *page;
        pgoff_t end_index;
        loff_t isize;
        unsigned long nr, ret;
        //读页面的过程中可能重新调度
        cond_resched();
find_page:
        //redix tree中查找 
        page = find_get_page(mapping, index);
        //没找到
        if (!page) {
            //先读到页缓存
            //分配list page_pool
            //调用a_ops->readpages or a_ops->readpage读取数据
            //a_ops->readpage负责提交bio
            page_cache_sync_readahead(mapping,
                    ra, filp,
                    index, last_index - index);
            //再找
            page = find_get_page(mapping, index);
            //还是没找到...
            if (unlikely(page == NULL))
                //去分配页面再读
                goto no_cached_page;
        }
        //readahead related 
        if (PageReadahead(page)) {
            page_cache_async_readahead(mapping,
                    ra, filp, page,
                    index, last_index - index);
        }
        //不是最新
        if (!PageUptodate(page)) {
            if (inode->i_blkbits == PAGE_CACHE_SHIFT ||
                    !mapping->a_ops->is_partially_uptodate)
                goto page_not_up_to_date;
            if (!trylock_page(page))
                goto page_not_up_to_date;

            if (!page->mapping)
                goto page_not_up_to_date_locked;
            if (!mapping->a_ops->is_partially_uptodate(page,
                            offset, iter->count))
                goto page_not_up_to_date_locked;
            unlock_page(page);
        }
page_ok: //好，拿到的cached page正常了

         /* 省略其他检查部分 */

        //到这，从磁盘中读取块到page cache或者本身page cache存在，一切正常，拷贝到用户空间
        ret = copy_page_to_iter(page, offset, nr, iter);
        offset += ret;
        index += offset >> PAGE_CACHE_SHIFT;
        offset &= ~PAGE_CACHE_MASK;
        prev_offset = offset;

        //释放页面
        page_cache_release(page);
        written += ret;
        if (!iov_iter_count(iter))
            goto out;
        if (ret < nr) {
            error = -EFAULT;
            goto out;
        }
        //继续
        continue;

page_not_up_to_date:
        /* Get exclusive access to the page ... */
        error = lock_page_killable(page);
        if (unlikely(error))
            goto readpage_error;

page_not_up_to_date_locked:
        /* Did it get truncated before we got the lock? */
        if (!page->mapping) {
            unlock_page(page);
            page_cache_release(page);
            continue;
        }
        /* Did somebody else fill it already? */
        if (PageUptodate(page)) {
            unlock_page(page);
            goto page_ok;
        }

readpage: //为了no_cached_page
        /*
         * A previous I/O error may have been due to temporary
         * failures, eg. multipath errors.
         * PG_error will be set again if readpage fails.
         */
        ClearPageError(page);
        /* Start the actual read. The read will unlock the page. */
        //还是调用a_ops->readpage 
        error = mapping->a_ops->readpage(filp, page);
        if (unlikely(error)) {
            if (error == AOP_TRUNCATED_PAGE) {
                page_cache_release(page);
                error = 0;
                goto find_page;
            }
            goto readpage_error;
        }
        if (!PageUptodate(page)) {
            error = lock_page_killable(page);
            if (unlikely(error))
                goto readpage_error;
            if (!PageUptodate(page)) {
                if (page->mapping == NULL) {
                    /*
                     * invalidate_mapping_pages got it
                     */
                    unlock_page(page);
                    page_cache_release(page);
                    goto find_page;
                }
                unlock_page(page);
                shrink_readahead_size_eio(filp, ra);
                error = -EIO;
                goto readpage_error;
            }
            unlock_page(page);
        }
        //page ok
        goto page_ok;

readpage_error:
        /* UHHUH! A synchronous read error occurred. Report it */
        page_cache_release(page);
        goto out;

no_cached_page:
        /*
         * Ok, it wasn't cached, so we need to create a new
         * page..
         */
        //从冷页面链表中拿一个page
        page = page_cache_alloc_cold(mapping);
        if (!page) {
            error = -ENOMEM;
            goto out;
        }
        //加入cache
        error = add_to_page_cache_lru(page, mapping,
                        index, GFP_KERNEL);
        if (error) {
            page_cache_release(page);
            if (error == -EEXIST) {
                error = 0;
                goto find_page;
            }
            goto out;
        }
        goto readpage;
    }
/* 省略部分 */

---

ref: Linux Kernel 3.19.3 source code