从ext2文件系统上读出超级块

简介: 概述            本篇博客中,我们将仔细分析如何从格式化为ext2文件系统的磁盘中读取超级块并填充内存超级块结构,每次将一个格式化了ext2文件系统的磁盘(分区)挂载到挂载点的时候会调用该方法,该方法在操作系统中的实现主要是函数ext2_fill_super。

概述

 
         本篇博客中,我们将仔细分析 如何从格式化为ext2文件系统的磁盘中读取超级块并填充内存超级块结构,每次将一个 格式化了ext2文件系统的磁盘(分区)挂载到挂载点的时候会调用该方法,该方法在操作系统中的实现主要是函数ext2_fill_super。
 

实现

 
        在ext2系列之前的博客中我们描述了ext2的磁盘划分,所以读取超级块的过程也就显得比较简单,只是在读取完成后可能需要进行一些列的检查等。废话不多说,我们直接来看该函数的实现。我们分为几段来阐述其实现机理。
第一阶段:从磁盘读出超级块
 
 1 static int ext2_fill_super(struct super_block *sb, void *data, int silent)
 2 {
 3     struct buffer_head * bh;
 4     struct ext2_sb_info * sbi;
 5     struct ext2_super_block * es;
 6     struct inode *root;
 7     unsigned long block;
 8     unsigned long sb_block = get_sb_block(&data);
 9     unsigned long logic_sb_block;
10     unsigned long offset = 0;
11     unsigned long def_mount_opts;
12     long ret = -EINVAL;
13     //default block size is 1024B
14     int blocksize = BLOCK_SIZE;
15     int db_count;
16     int i, j;
17     __le32 features;
18     int err;
19 
20     //allocate memory ext2_super_block in memory
21     sbi = kzalloc(sizeof(*sbi), GFP_KERNEL);
22     if (!sbi)
23         return -ENOMEM;
24 
25     sbi->s_blockgroup_lock =
26         kzalloc(sizeof(struct blockgroup_lock), GFP_KERNEL);
27     if (!sbi->s_blockgroup_lock) {
28         kfree(sbi);
29         return -ENOMEM;
30     }
31     //sb is vfs super_block
32     //sb->s_fs_info is specific file system super block 
33     sb->s_fs_info = sbi;
34     sbi->s_sb_block = sb_block;
35 
36     spin_lock_init(&sbi->s_lock);
37 
38     /*
39      * See what the current blocksize for the device is, and
40      * use that as the blocksize.  Otherwise (or if the blocksize
41      * is smaller than the default) use the default.
42      * This is important for devices that have a hardware
43      * sectorsize that is larger than the default.
44      */
45      //the block size can't be smaller than BLOCK_SIZE=1024B
46      //and block size must be smaller than PAGE_SIZE = 4096B now
47     blocksize = sb_min_blocksize(sb, BLOCK_SIZE);
48     if (!blocksize) {
49         ext2_msg(sb, KERN_ERR, "error: unable to set blocksize");
50         goto failed_sbi;
51     }
52 
53     /*
54      * If the superblock doesn't start on a hardware sector boundary,
55      * calculate the offset.  
56      */
57      //blocksize may bigger than BLOCK_SIZE=1024B
58      //because we read blocksize bytes data from disk
59      //Block 0 is 1024B and super_block is also 1024B 
60      //if blocksize is not 1024B,it must be bigger than 1024B,for example,if blocksize is 2048B
61      //we must read block 0(first 2048B on disk),then we read offset 1024~2047 as super block
62     if (blocksize != BLOCK_SIZE) {
63         logic_sb_block = (sb_block*BLOCK_SIZE) / blocksize;
64         offset = (sb_block*BLOCK_SIZE) % blocksize;
65     } else {
66         logic_sb_block = sb_block;
67     }
68     //read block @logic_sb_block containg super block
69     if (!(bh = sb_bread(sb, logic_sb_block))) {
70         ext2_msg(sb, KERN_ERR, "error: unable to read superblock");
71         goto failed_sbi;
72     }
73     /*
74      * Note: s_es must be initialized as soon as possible because
75      *       some ext2 macro-instructions depend on its value
76      */
77     es = (struct ext2_super_block *) (((char *)bh->b_data) + offset);
78     //sbi is ext2_super_block in memory while sbi->s_es is ext2_super_block on disk
79     sbi->s_es = es;
第一部分的函数是核心,它负责从ext2的磁盘(分区)上读出超级块,那么这里的问题就产生了:
  1. 超级块的起始位置在哪?
  2. 超级块的大小是多少?
  3. 在实现中我们自己定义的块大小(默认1024)与磁盘设备的块大小如果不一致怎么办?
所以我们看上面的很多代码其实都是在处理这个问题。让我们一一来解答。
首先,超级块位于磁盘(分区)的第二个1024位置上,因为第一个1024字节默认作为引导块,文件系统并不使用。
其次,ext2的超级块大小也为1024字节,这在ext2超级块数据结构的定义中可以看出。
最后,因为读之前我们默认磁盘块大小是1024字节,但磁盘设备定义的块大小可能不同,有可能是2048,4096等等,而我们读磁盘数据的时候是以逻辑块为单位读取的(虽然最终的物理读取是以扇区为单位的),因此,我们必须确定到底块大小是多少,如果决定块大小是1024,那我们只需读出第二个磁盘块即可读出超级块,而如果块大小是2048,那我们读出第一个磁盘块,然后再取1024~2047这一段,下图比较清晰地阐述了这个过程:
另外,我们读出超级块是要缓存在内存中的,而内存中的超级块结构需要在磁盘超级块结构上增添一些管理成员。ext2内存超级块结构为struct ext2_sb_info。
ext2_fill_super所展示的第一段代码所做工作主要有:
  1. 分配ext2内存超级块结构struct ext2_sb_info,如果分配内存失败,则直接返回-ENOMEM;
  2. 确定逻辑磁盘块大小,比较默认逻辑块大小和真实逻辑块大小(根据磁盘设备的一些信息确定),将最大者设置为逻辑块大小,但注意:该最大者必须是2的次幂且不可大于4096
  3. 从磁盘上读出超级块根据2中计算的块大小确定超级块所在逻辑块号和块内偏移,读出超级块,存储在1中分配的内存超级块结构中sbi->s_es = es。
第二阶段:根据磁盘超级块初始化内存超级块的成员
        上文描述的第一阶段从磁盘上读出了超级块内容,接下来我们就要根据磁盘上的超级块结构来初始化内存超级块结构,在这个过程中可能还伴随着磁盘超级块内容的检查,确认其是否已经损坏等。
 1 sb->s_magic = le16_to_cpu(es->s_magic);
 2 
 3     if (sb->s_magic != EXT2_SUPER_MAGIC)
 4         goto cantfind_ext2;
 5 
 6     /* Set defaults before we parse the mount options */
 7     /* 接下来这段根据磁盘超级块
 8     ** 结构来设置内存超级块结构的部分选项
 9     ** 相比较而言这些选项的重要性没那么高
10     */
11     def_mount_opts = le32_to_cpu(es->s_default_mount_opts);
12     if (def_mount_opts & EXT2_DEFM_DEBUG)
13         set_opt(sbi->s_mount_opt, DEBUG);
14     if (def_mount_opts & EXT2_DEFM_BSDGROUPS)
15         set_opt(sbi->s_mount_opt, GRPID);
16     if (def_mount_opts & EXT2_DEFM_UID16)
17         set_opt(sbi->s_mount_opt, NO_UID32);
18 #ifdef CONFIG_EXT2_FS_XATTR
19     if (def_mount_opts & EXT2_DEFM_XATTR_USER)
20         set_opt(sbi->s_mount_opt, XATTR_USER);
21 #endif
22 #ifdef CONFIG_EXT2_FS_POSIX_ACL
23     if (def_mount_opts & EXT2_DEFM_ACL)
24         set_opt(sbi->s_mount_opt, POSIX_ACL);
25 #endif
26     /* 这个选项决定了挂载出错时的处理方法
27     ** 如PANIC即指示出错就奔溃...
28     */
29     if (le16_to_cpu(sbi->s_es->s_errors) == EXT2_ERRORS_PANIC)
30         set_opt(sbi->s_mount_opt, ERRORS_PANIC);
31     else if (le16_to_cpu(sbi->s_es->s_errors) == EXT2_ERRORS_CONTINUE)
32         set_opt(sbi->s_mount_opt, ERRORS_CONT);
33     else
34         set_opt(sbi->s_mount_opt, ERRORS_RO);
35 
36     sbi->s_resuid = le16_to_cpu(es->s_def_resuid);
37     sbi->s_resgid = le16_to_cpu(es->s_def_resgid);
38     
39     set_opt(sbi->s_mount_opt, RESERVATION);
40 
41     if (!parse_options((char *) data, sb))
42         goto failed_mount;
43 
44     sb->s_flags = (sb->s_flags & ~MS_POSIXACL) |
45         ((EXT2_SB(sb)->s_mount_opt & EXT2_MOUNT_POSIX_ACL) ?
46          MS_POSIXACL : 0);
47 
48     ext2_xip_verify_sb(sb); /* see if bdev supports xip, unset
49                     EXT2_MOUNT_XIP if not */
50 
51     if (le32_to_cpu(es->s_rev_level) == EXT2_GOOD_OLD_REV &&
52         (EXT2_HAS_COMPAT_FEATURE(sb, ~0U) ||
53          EXT2_HAS_RO_COMPAT_FEATURE(sb, ~0U) ||
54          EXT2_HAS_INCOMPAT_FEATURE(sb, ~0U)))
55         ext2_msg(sb, KERN_WARNING,
56             "warning: feature flags set on rev 0 fs, "
57             "running e2fsck is recommended");
58     /*
59      * Check feature flags regardless of the revision level, since we
60      * previously didn't change the revision level when setting the flags,
61      * so there is a chance incompat flags are set on a rev 0 filesystem.
62      */
63     features = EXT2_HAS_INCOMPAT_FEATURE(sb, ~EXT2_FEATURE_INCOMPAT_SUPP);
64     if (features) {
65         ext2_msg(sb, KERN_ERR,    "error: couldn't mount because of "
66                "unsupported optional features (%x)",
67             le32_to_cpu(features));
68         goto failed_mount;
69     }
70     if (!(sb->s_flags & MS_RDONLY) &&
71         (features = EXT2_HAS_RO_COMPAT_FEATURE(sb, ~EXT2_FEATURE_RO_COMPAT_SUPP))){
72         ext2_msg(sb, KERN_ERR, "error: couldn't mount RDWR because of "
73                "unsupported optional features (%x)",
74                le32_to_cpu(features));
75         goto failed_mount;
76     }
相对来说,这部分的代码重要性没那么高,我们无需花费太多的精力,简单阅读下注释即可。
 
第三阶段:根据磁盘超级块初始化内存超级块的成员(续)
 
        第二阶段初始化内存超级块的只是一些比较简单的选项,到了这个阶段,初始化的东西就比较重要了,它关乎着文件系统的正确性。因此我们作比较详细的分析。
/*
    ** 超级块中可能记录着逻辑块大小,因此我们必须
    ** 以此为准
    */
    blocksize = BLOCK_SIZE << le32_to_cpu(sbi->s_es->s_log_block_size);

    if (ext2_use_xip(sb) && blocksize != PAGE_SIZE) {
        if (!silent)
            ext2_msg(sb, KERN_ERR,
                "error: unsupported blocksize for xip");
        goto failed_mount;
    }

    /* If the blocksize doesn't match, re-read the thing.. */
    /* 如果块大小和我们之前确定的不太一样
    ** 我们有必要重新读一次超级块
    ** 因为之前读的可能并不准确
    */
    if (sb->s_blocksize != blocksize) {
        brelse(bh);

        if (!sb_set_blocksize(sb, blocksize)) {
            ext2_msg(sb, KERN_ERR, "error: blocksize is too small");
            goto failed_sbi;
        }

        logic_sb_block = (sb_block*BLOCK_SIZE) / blocksize;
        offset = (sb_block*BLOCK_SIZE) % blocksize;
        bh = sb_bread(sb, logic_sb_block);
        if(!bh) {
            ext2_msg(sb, KERN_ERR, "error: couldn't read"
                "superblock on 2nd try");
            goto failed_sbi;
        }
        es = (struct ext2_super_block *) (((char *)bh->b_data) + offset);
        sbi->s_es = es;
        if (es->s_magic != cpu_to_le16(EXT2_SUPER_MAGIC)) {
            ext2_msg(sb, KERN_ERR, "error: magic mismatch");
            goto failed_mount;
        }
    }

    /* 计算ext2最大可支持文件的大小*/
    sb->s_maxbytes = ext2_max_size(sb->s_blocksize_bits);

    if (le32_to_cpu(es->s_rev_level) == EXT2_GOOD_OLD_REV) {
        sbi->s_inode_size = EXT2_GOOD_OLD_INODE_SIZE;
        sbi->s_first_ino = EXT2_GOOD_OLD_FIRST_INO;
    } else {
        sbi->s_inode_size = le16_to_cpu(es->s_inode_size);
        sbi->s_first_ino = le32_to_cpu(es->s_first_ino);
        if ((sbi->s_inode_size < EXT2_GOOD_OLD_INODE_SIZE) ||
            !is_power_of_2(sbi->s_inode_size) ||
            (sbi->s_inode_size > blocksize)) {
            ext2_msg(sb, KERN_ERR,
                "error: unsupported inode size: %d",
                sbi->s_inode_size);
            goto failed_mount;
        }
    }

    /*  对于逻辑块较大的ext2文件系统,为了
    **  减少块内碎片问题,设置了fragment,
    **  即每个磁盘块内可再细分成多个fragment
    **  这个思想源自FFS,对于1024大小的磁盘块
    **  也就没有必要再划分fragment了
    **  因为最小的fragment大小就是1024字节
    */
    sbi->s_frag_size = EXT2_MIN_FRAG_SIZE <<
                   le32_to_cpu(es->s_log_frag_size);
    if (sbi->s_frag_size == 0)
        goto cantfind_ext2;
    /* 初始化一些静态信息*/
    sbi->s_frags_per_block = sb->s_blocksize / sbi->s_frag_size;

    sbi->s_blocks_per_group = le32_to_cpu(es->s_blocks_per_group);
    sbi->s_frags_per_group = le32_to_cpu(es->s_frags_per_group);
    sbi->s_inodes_per_group = le32_to_cpu(es->s_inodes_per_group);

    if (EXT2_INODE_SIZE(sb) == 0)
        goto cantfind_ext2;
    sbi->s_inodes_per_block = sb->s_blocksize / EXT2_INODE_SIZE(sb);
    if (sbi->s_inodes_per_block == 0 || sbi->s_inodes_per_group == 0)
        goto cantfind_ext2;
    sbi->s_itb_per_group = sbi->s_inodes_per_group /
                    sbi->s_inodes_per_block;
    sbi->s_desc_per_block = sb->s_blocksize /
                    sizeof (struct ext2_group_desc);
    sbi->s_sbh = bh;
    sbi->s_mount_state = le16_to_cpu(es->s_state);
    sbi->s_addr_per_block_bits =
        ilog2 (EXT2_ADDR_PER_BLOCK(sb));
    sbi->s_desc_per_block_bits =
        ilog2 (EXT2_DESC_PER_BLOCK(sb));

    if (sb->s_magic != EXT2_SUPER_MAGIC)
        goto cantfind_ext2;

    if (sb->s_blocksize != bh->b_size) {
        if (!silent)
            ext2_msg(sb, KERN_ERR, "error: unsupported blocksize");
        goto failed_mount;
    }

    /* 目前仅支持块大小和fragment size大小相同*/
    if (sb->s_blocksize != sbi->s_frag_size) {
        ext2_msg(sb, KERN_ERR,
            "error: fragsize %lu != blocksize %lu"
            "(not supported yet)",
            sbi->s_frag_size, sb->s_blocksize);
        goto failed_mount;
    }

    if (sbi->s_blocks_per_group > sb->s_blocksize * 8) {
        ext2_msg(sb, KERN_ERR,
            "error: #blocks per group too big: %lu",
            sbi->s_blocks_per_group);
        goto failed_mount;
    }
    if (sbi->s_frags_per_group > sb->s_blocksize * 8) {
        ext2_msg(sb, KERN_ERR,
            "error: #fragments per group too big: %lu",
            sbi->s_frags_per_group);
        goto failed_mount;
    }
    if (sbi->s_inodes_per_group > sb->s_blocksize * 8) {
        ext2_msg(sb, KERN_ERR,
            "error: #inodes per group too big: %lu",
            sbi->s_inodes_per_group);
        goto failed_mount;
    }

    if (EXT2_BLOCKS_PER_GROUP(sb) == 0)
        goto cantfind_ext2;
     sbi->s_groups_count = ((le32_to_cpu(es->s_blocks_count) -
                 le32_to_cpu(es->s_first_data_block) - 1)
                     / EXT2_BLOCKS_PER_GROUP(sb)) + 1;
    db_count = (sbi->s_groups_count + EXT2_DESC_PER_BLOCK(sb) - 1) /
           EXT2_DESC_PER_BLOCK(sb);
    sbi->s_group_desc = kmalloc (db_count * sizeof (struct buffer_head *), GFP_KERNEL);
    if (sbi->s_group_desc == NULL) {
        ext2_msg(sb, KERN_ERR, "error: not enough memory");
        goto failed_mount;
    }
    bgl_lock_init(sbi->s_blockgroup_lock);
    /* 这个数据结构干嘛的现在还不得而知*/
    sbi->s_debts = kcalloc(sbi->s_groups_count, sizeof(*sbi->s_debts), GFP_KERNEL);
    if (!sbi->s_debts) {
        ext2_msg(sb, KERN_ERR, "error: not enough memory");
        goto failed_mount_group_desc;
    }

    /* 读出块组描述符信息 */
    for (i = 0; i < db_count; i++) {
        block = descriptor_loc(sb, logic_sb_block, i);
        sbi->s_group_desc[i] = sb_bread(sb, block);
        if (!sbi->s_group_desc[i]) {
            for (j = 0; j < i; j++)
                brelse (sbi->s_group_desc[j]);
            ext2_msg(sb, KERN_ERR,
                "error: unable to read group descriptors");
            goto failed_mount_group_desc;
        }
    }
    if (!ext2_check_descriptors (sb)) {
        ext2_msg(sb, KERN_ERR, "group descriptors corrupted");
        goto failed_mount2;
    }
    sbi->s_gdb_count = db_count;
    get_random_bytes(&sbi->s_next_generation, sizeof(u32));

    spin_lock_init(&sbi->s_next_gen_lock);

    /* per fileystem reservation list head & lock */
    //init something for reservation windows of every file
    spin_lock_init(&sbi->s_rsv_window_lock);
    sbi->s_rsv_window_root = RB_ROOT;
    /*
     * Add a single, static dummy reservation to the start of the
     * reservation window list --- it gives us a placeholder for
     * append-at-start-of-list which makes the allocation logic
     * _much_ simpler.
     */
    /* 初始化内存超级块的预留窗口
    ** 所谓的预留窗口是在分配数据块的时候
    ** 每一次多分配一点,以提高文件数据存储
    ** 的连续性
    */
    sbi->s_rsv_window_head.rsv_start = EXT2_RESERVE_WINDOW_NOT_ALLOCATED;
    sbi->s_rsv_window_head.rsv_end = EXT2_RESERVE_WINDOW_NOT_ALLOCATED;
    sbi->s_rsv_window_head.rsv_alloc_hit = 0;
    sbi->s_rsv_window_head.rsv_goal_size = 0;
    ext2_rsv_window_add(sb, &sbi->s_rsv_window_head);

    err = percpu_counter_init(&sbi->s_freeblocks_counter,
                ext2_count_free_blocks(sb));
    if (!err) {
        err = percpu_counter_init(&sbi->s_freeinodes_counter,
                ext2_count_free_inodes(sb));
    }
    if (!err) {
        err = percpu_counter_init(&sbi->s_dirs_counter,
                ext2_count_dirs(sb));
    }
    if (err) {
        ext2_msg(sb, KERN_ERR, "error: insufficient memory");
        goto failed_mount3;
    }
    /*
     * set up enough so that it can read an inode
     */
    sb->s_op = &ext2_sops;
    sb->s_export_op = &ext2_export_ops;
    sb->s_xattr = ext2_xattr_handlers;

#ifdef CONFIG_QUOTA
    sb->dq_op = &dquot_operations;
    sb->s_qcop = &dquot_quotactl_ops;
#endif
这里面涉及到的细节问题比较多,但是都比较简单,主要是文件系统各种统计数据的计算等,这里不再赘述,请直接参考代码注释。
 
第四阶段:构造根目录
        当超级块完全读出并构造内存超级块以后,接下来就是构造根目录了,让我们直接看代码:
/* 读根目录的inode,inode号为默认值2
    ** 读出后保存在内存inode结构中
    */
    root = ext2_iget(sb, EXT2_ROOT_INO);
    if (IS_ERR(root)) {
        ret = PTR_ERR(root);
        goto failed_mount3;
    }
    if (!S_ISDIR(root->i_mode) || !root->i_blocks || !root->i_size) {
        iput(root);
        ext2_msg(sb, KERN_ERR, "error: corrupt root inode, run e2fsck");
        goto failed_mount3;
    }

    /* 分配根目录的内存目录项
    ** 因为根目录没有父目录这个概念
    ** 因此,没法从其父目录中读出其目录
    ** 只能在内存中构造一个
    */
    sb->s_root = d_alloc_root(root);
    if (!sb->s_root) {
        iput(root);
        ext2_msg(sb, KERN_ERR, "error: get root inode failed");
        ret = -ENOMEM;
        goto failed_mount3;
    }
    if (EXT2_HAS_COMPAT_FEATURE(sb, EXT3_FEATURE_COMPAT_HAS_JOURNAL))
        ext2_msg(sb, KERN_WARNING,
            "warning: mounting ext3 filesystem as ext2");
    if (ext2_setup_super (sb, es, sb->s_flags & MS_RDONLY))
        sb->s_flags |= MS_RDONLY;
    
    /* 在填充超级块时有可能会修改磁盘超级块
    ** 因此有必要作一次写回操作
    */
    ext2_write_super(sb);
    return 0;

到此,整个从磁盘读出超级块直至填充内存超级块结构的过程就结束了,整个流程虽然繁杂,但还算简单。

相关文章
|
2月前
|
存储 缓存 IDE
xv6(10) 文件系统:磁盘&缓冲区
文件系统:磁盘&缓冲区
58 0
|
11月前
7.1.5 EXT2/EXT3/EXT4 文件的存取与日志式文件系统的功能
7.1.5 EXT2/EXT3/EXT4 文件的存取与日志式文件系统的功能
80 0
系统诊断小技巧(9):如何从Ext3或者Ext4文件系统推断分区位置
扩容失败或者系统重启后分区丢失,需要文件系统推定分区位置。我们这里给出一个自动化的工具
2238 0
|
Oracle 关系型数据库 Unix