深度解析Linux根文件系统的挂载过程

在前面的文章中介绍《Linux操作系统启动过程》,而Linux系统的根文件系统(root file system)的挂载过程则是其中一个重要环节,下面这部分内容来自于网络,经整理分享如下,希望能给这部份知识点比较迷茫的朋友一点帮助。

一、rootfs的种类

总的来说,rootfs分为两种:虚拟rootfs和真实rootfs。现在kernel的发展趋势是将更多的功能放到用户空间完成。以保持内核的精简。虚拟rootfs也是各linux发行厂商普遍采用的一种方式。可以将一部份的初始化工作放在虚拟的rootfs里完成。然后切换到真实的文件系统。
在虚拟rootfs的发展过程中。又有以下几个版本:

  • initramfs: Initramfs是在 kernel 2.5中引入的技术,实际上它的含义就是:在内核镜像中附加一个cpio包,这个cpio包中包含了一个小型的文件系统,当内核启动时,内核将这个cpio包解开,并且将其中包含的文件系统释放到rootfs中,内核中的一部分初始化代码会放到这个文件系统中,作为用户层进程来执行。这样带来的明显的好处是精简了内核的初始化代码,而且使得内核的初始化过程更容易定制。这种这种方式的rootfs是包含在kernel image之中的。

  • cpio-initrd: cpio格式的rootfs
  • image-initrd: 传统格式的rootfs
二、rootfs文件系统的挂载过程

这里说的rootfs不同于上面分析的rootfs。这里指的是系统初始化时的根结点。即/结点。它是其于内存的rootfs文件系统。这部份之前在>和文件系统中已经分析过。为了知识的连贯性这里再重复一次。

1
2
3
4
5
6
7
8
Start_kernel() mnt_init():
void __init mnt_init( void)
{
    ……
    ……
    init_rootfs();
    init_mount_tree();
}

Init_rootfs的代码如下:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
int __init init_rootfs( void)
{
     int err;

    err = bdi_init(&ramfs_backing_dev_info);
     if (err)
         return err;

    err = register_filesystem(&rootfs_fs_type);
     if (err)
        bdi_destroy(&ramfs_backing_dev_info);
 
     return err;
}

这个函数很简单,就是注册了rootfs的文件系统。
init_mount_tree()代码如下:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
static  void __init init_mount_tree( void)
{
     struct vfsmount *mnt;
     struct mnt_namespace *ns;
     struct path root;
 
    mnt = do_kern_mount( "rootfs", 0,  "rootfs", NULL);
     if (IS_ERR(mnt))
        panic( "Can't create rootfs");
    ns = kmalloc(sizeof(*ns), GFP_KERNEL);
     if (!ns)
        panic( "Can't allocate initial namespace");
    atomic_set(&ns->count, 1);
    INIT_LIST_HEAD(&ns->list);
    init_waitqueue_head(&ns->poll);
    ns->event = 0;
    list_add(&mnt->mnt_list, &ns->list);
    ns->root = mnt;
    mnt->mnt_ns = ns;
 
    init_task.nsproxy->mnt_ns = ns;
    get_mnt_ns(ns);

    root.mnt = ns->root;
    root.dentry = ns->root->mnt_root;

    set_fs_pwd(current->fs, &root);
    set_fs_root(current->fs, &root);
}

在这里,将rootfs文件系统挂载。它的挂载点默认为”/”.最后切换进程的根目录和当前目录为”/”.这也就是根目录的由来。不过这里只是初始化。等挂载完具体的文件系统之后,一般都会将根目录切换到具体的文件系统。所以在系统启动之后,用mount命令是看不到rootfs的挂载信息的.

三、虚拟文件系统的挂载

根目录已经挂上去了,可以挂载具体的文件系统了.
在start_kernel()àrest_init()àkernel_init():

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
static int __init kernel_init( void * unused)
{
    ……
    ……
    do_basic_setup();
     if (!ramdisk_execute_command)
        ramdisk_execute_command =  "/init";

     if (sys_access(( const char __user *) ramdisk_execute_command, 0) != 0) {
        ramdisk_execute_command = NULL;
        prepare_namespace();
    }
 
     /*
     * Ok, we have completed the initial bootup, and
     * we're essentially up and running. Get rid of the
     * initmem segments and start the user-mode stuff..
     */
    init_post();
     return 0;
}

do_basic_setup()是一个很关键的函数,所有直接编译在kernel中的模块都是由它启动的。代码片段如下:

1
2
3
4
5
6
7
8
9
static void __init do_basic_setup(void)
{
     /* drivers will send hotplug events */
    init_workqueues();
    usermodehelper_init();
    driver_init();
    init_irq_proc();
    do_initcalls();
}

Do_initcalls()用来启动所有在__initcall_start和__initcall_end段的函数,而静态编译进内核的modules也会将其入口放置在这段区间里。
跟根文件系统相关的初始化函数都会由rootfs_initcall()所引用。注意到有以下初始化函数:

1rootfs_initcall(populate_rootfs);

也就是说会在系统初始化的时候会调用populate_rootfs进行初始化。代码如下:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
static int __init populate_rootfs(void)
{
     char *err = unpack_to_rootfs(__initramfs_start,
                    __initramfs_end - __initramfs_start, 0);
     if (err)
        panic(err);
     if (initrd_start) {
#ifdef CONFIG_BLK_DEV_RAM
         int fd;
        printk(KERN_INFO  "checking if image is initramfs...");
        err = unpack_to_rootfs(( char *)initrd_start,
                            initrd_end - initrd_start, 1);
         if (!err) {
            printk( " it is\n");
            unpack_to_rootfs((char *)initrd_start,
                            initrd_end - initrd_start, 0);
            free_initrd();
             return 0;
        }
        printk( "it isn't (%s); looks like an initrd\n", err);
        fd = sys_open( "/initrd.image", O_WRONLY|O_CREAT, 0700);
         if (fd >= 0) {
            sys_write(fd, ( char *)initrd_start,
                              initrd_end - initrd_start);
            sys_close(fd);
            free_initrd();
        }
#else
        printk(KERN_INFO  "Unpacking initramfs...");
        err = unpack_to_rootfs((char *)initrd_start,
                              initrd_end - initrd_start, 0);
         if (err)
            panic(err);
        printk( " done\n");
        free_initrd();
#endif
    }
     return 0;
}

unpack_to_rootfs:顾名思义就是解压包,并将其释放至rootfs。它实际上有两个功能,一个是释放包,一个是查看包,看其是否属于cpio结构的包。功能选择是根据最后的一个参数来区分的.
在这个函数里,对应我们之前分析的三种虚拟根文件系统的情况。一种是跟kernel融为一体的initramfs.在编译kernel的时候,通过链接脚本将其存放在__initramfs_start至__initramfs_end的区域。这种情况下,直接调用unpack_to_rootfs将其释放到根目录.如果不是属于这种形式的。也就是__initramfs_start和__initramfs_end的值相等,长度为零。不会做任何处理。退出.

对应后两种情况。从代码中看到,必须要配制CONFIG_BLK_DEV_RAM才会支持image-initrd。否则全当成cpio-initrd的形式处理。
对于是cpio-initrd的情况。直接将其释放到根目录。对于是image-initrd的情况。将其释放到/initrd.image.最后将initrd内存区域归入伙伴系统。这段内存就可以由操作系统来做其它的用途了。
接下来,内核对这几种情况又是怎么处理的呢?不要着急。往下看:

回到kernel_init()这个函数:

1
2
3
4
5
6
7
8
9
10
11
12
13
14 
15
16
17
18
19
20
21
22
23
24
25
26
27
static int __init kernel_init(void * unused)
{
    …….
    …….
    do_basic_setup();

     /*
     * check if there is an early userspace init.  If yes, let it do all
     * the work
     */

     if (!ramdisk_execute_command)
        ramdisk_execute_command =  "/init";
     if (sys_access(( const char __user *) ramdisk_execute_command, 0) != 0) {
        ramdisk_execute_command = NULL;
        prepare_namespace();
    }

     /*
     * Ok, we have completed the initial bootup, and
     * we're essentially up and running. Get rid of the
     * initmem segments and start the user-mode stuff..
     */
    init_post();
     return 0;
}

ramdisk_execute_command:在kernel解析引导参数的时候使用。如果用户指定了init文件路径,即使用了“init=”,就会将这个参数值存放到这里。
如果没有指定init文件路径。默认为/init
对应于前面一段的分析,我们知道,对于initramdisk和cpio-initrd的情况,都会将虚拟根文件系统释放到根目录。如果这些虚拟文件系统里有/init这个文件。就会转入到init_post()。
Init_post()代码如下:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
static int noinline init_post(void)
{
    free_initmem();
    unlock_kernel();
    mark_rodata_ro();
    system_state = SYSTEM_RUNNING;
    numa_default_policy();
 
     if (sys_open(( const char __user *)  "/dev/console", O_RDWR, 0)
        printk(KERN_WARNING  "Warning: unable to open an initial console.\n");
    ( void) sys_dup(0);
    ( void) sys_dup(0);
 
     if (ramdisk_execute_command) {
        run_init_process(ramdisk_execute_command);
        printk(KERN_WARNING  "Failed to execute %s\n",
        ramdisk_execute_command);
    }
 
     /*
     * We try each of these until one succeeds.
     *
     * The Bourne shell can be used instead of init if we are
     * trying to recover a really broken machine.
     */
     if (execute_command) {
        run_init_process(execute_command);
        printk(KERN_WARNING  "Failed to execute %s.  Attempting "
                                        "defaults...\n", execute_command);
    }
    run_init_process( " /sbin/init ");
    run_init_process( " /etc/init ");
    run_init_process( "/bin/init");
    run_init_process( "/bin/sh");

    panic( " No init found.  Try passing init= option to kernel. ");
}

从代码中可以看中,会依次执行指定的init文件,如果失败,就会执行/sbin/init, /etc/init, /bin/init, /bin/sh
注意的是,run_init_process在调用相应程序运行的时候,用的是kernel_execve。也就是说调用进程会替换当前进程。只要上述任意一个文件调用成功,就不会返回到这个函数。如果上面几个文件都无法执行。打印出没有找到init文件的错误。
对于image-hdr或者是虚拟文件系统中没有包含 /init的情况,会由prepare_namespace()处理。代码如下:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45 
46
47
48
49
50
void __init prepare_namespace( void)
{
     int is_floppy;

     if (root_delay) {
        printk(KERN_INFO  "Waiting %dsec before mounting root device...\n",
        root_delay);
        ssleep(root_delay);
    }
 
     /* wait for the known devices to complete their probing */
     while (driver_probe_done() != 0)
        msleep(100);
     //mtd的处理
    md_run_setup();

     if (saved_root_name[0]) {
        root_device_name = saved_root_name;
         if (! strncmp(root_device_name,  "mtd", 3)) {
            mount_block_root(root_device_name, root_mountflags);
             goto out;
        }
        ROOT_DEV = name_to_dev_t(root_device_name);
         if ( strncmp(root_device_name,  "/dev/", 5) == 0)
            root_device_name += 5;
    }

     if (initrd_load())
         goto out;

     /* wait for any asynchronous scanning to complete */
     if ((ROOT_DEV == 0) && root_wait) {
        printk(KERN_INFO  "Waiting for root device %s...\n",
        saved_root_name);
         while (driver_probe_done() != 0 ||
                   (ROOT_DEV = name_to_dev_t(saved_root_name)) == 0)
            msleep(100);
    }

    is_floppy = MAJOR(ROOT_DEV) == FLOPPY_MAJOR;

     if (is_floppy && rd_doload && rd_load_disk(0))
    ROOT_DEV = Root_RAM0;

    mount_root();
out:
    sys_mount( ".""/", NULL, MS_MOVE, NULL);
    sys_chroot( ".");
}

这里有几个比较有意思的处理,首先用户可以用root=来指定根文件系统。它的值保存在saved_root_name中。如果用户指定了以mtd开始的字串做为它的根文件系统。就会直接去挂载。这个文件是mtdblock的设备文件。
否则将设备结点文件转换为ROOT_DEV即设备节点号
然后,转向initrd_load()执行initrd预处理后,再将具体的根文件系统挂载。
注意到,在这个函数末尾。会调用sys_mount()来移动当前文件系统挂载点到”/”目录下。然后将根目录切换到当前目录。这样,根文件系统的挂载点就成为了我们在用户空间所看到的”/”了.
对于其它根文件系统的情况,会先经过initrd的处理。即

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
int __init initrd_load( void)
{
     if (mount_initrd) {
        create_dev( "/dev/ram", Root_RAM0);
         /*
         * Load the initrd data into /dev/ram0. Execute it as initrd
         * unless /dev/ram0 is supposed to be our actual root device,
         * in that case the ram disk is just set up here, and gets
         * mounted in the normal path.
         */
         if (rd_load_image( "/initrd.image") && ROOT_DEV != Root_RAM0) {
            sys_unlink( "/initrd.image");
            handle_initrd();
             return 1;
        }
    }
    sys_unlink( "/initrd.image");
     return 0;
}

建立一个ROOT_RAM)的设备节点,并将/initrd/.image释放到这个节点中,/initrd.image的内容,就是我们之前分析的image-initrd。
如果根文件设备号不是ROOT_RAM0( 用户指定的根文件系统不是/dev/ram0就会转入到handle_initrd()
如果当前根文件系统是/dev/ram0.将其直接挂载就好了。

handle_initrd()代码如下:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
static void __init handle_initrd( void)
{
     int error;
     int pid;
    real_root_dev = new_encode_dev(ROOT_DEV);
    create_dev( "/dev/root.old", Root_RAM0);
 
     /* mount initrd on rootfs' /root */
    mount_block_root( "/dev/root.old", root_mountflags & ~MS_RDONLY);
    sys_mkdir( "/old", 0700);
    root_fd = sys_open( "/", 0, 0);
    old_fd = sys_open( "/old", 0, 0);
 
     /* move initrd over / and chdir/chroot in initrd root */
    sys_chdir( "/root");
    sys_mount( ".""/", NULL, MS_MOVE, NULL);
    sys_chroot( ".");
 
     /*
     * In case that a resume from disk is carried out by linuxrc or one of
     * its children, we need to tell the freezer not to wait for us.
     */
    current->flags |= PF_FREEZER_SKIP;
    pid = kernel_thread(do_linuxrc,  "/linuxrc", SIGCHLD);
     if (pid > 0)
         while (pid != sys_wait4(-1, NULL, 0, NULL))
    yield();

    current->flags &= ~PF_FREEZER_SKIP;

     /* move initrd to rootfs' /old */
    sys_fchdir(old_fd);
    sys_mount( "/"".", NULL, MS_MOVE, NULL);
     /* switch root and cwd back to / of rootfs */
    sys_fchdir(root_fd);
    sys_chroot( ".");
    sys_close(old_fd);
    sys_close(root_fd);
     if (new_decode_dev(real_root_dev) == Root_RAM0) {
        sys_chdir( "/old");
         return;
    }

    ROOT_DEV = new_decode_dev(real_root_dev);
    mount_root();
    printk(KERN_NOTICE  "Trying to move old root to /initrd ... ");
    error = sys_mount( "/old""/root/initrd", NULL, MS_MOVE, NULL);
     if (!error)
        printk( "okay\n");
     else {
         int fd = sys_open( "/dev/root.old", O_RDWR, 0);
         if (error == -ENOENT)
            printk( "/initrd does not exist. Ignored.\n");
         else
            printk( "failed\n");
        printk(KERN_NOTICE  "Unmounting old root\n");
        sys_umount( "/old", MNT_DETACH);
        printk(KERN_NOTICE  "Trying to free ramdisk memory ... ");
         if (fd < 0) {
            error = fd;
        }  else {
            error = sys_ioctl(fd, BLKFLSBUF, 0);
            sys_close(fd);
        }
        printk(!error ?  "okay\n" :  "failed\n");
    }
}

先将/dev/ram0挂载,而后执行/linuxrc.等其执行完后。切换根目录,再挂载具体的根文件系统.
到这里。文件系统挂载的全部内容就分析完了.

四、小结

在本小节里。分析了根文件系统的挂载流程。并对几个虚拟根文件系统的情况做了详细的分析。理解这部份,对我们构建linux嵌入式开发系统是很有帮助的.

五、参考资料

IBM技术论坛的附根文件系统挂载流程图:

 

评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值