目录

一    EXT2文件系统结构概览    2

1.1 EXT2文件系统结构框图    2

1.2 EXT2重要数据结构    3

二    块缓存    6

三    EXT2文件系统挂载    7

3.1 注册ext2文件系统类型    7

3.2 ext2文件系统挂载    7

3.3文件系统操作    9

EXT2文件系统结构概览

1.1 EXT2文件系统结构框图

如果文件比较小，其数据块少于12个，其数据块索引就放在inode->i_blocks中，如果文件比较大，操作12个数据块就需要分配间接块来保存数据块索引

1.2 EXT2重要数据结构

super_block是VFS中的标准结构，通过成员s_fs_info与特定文件系统相连

结构体ext2_sb_info包含特定文件系统的所有信息，包含超级块，组描述符等等：

ext2_super_block保存在磁盘上的原始超级块

结构体ext2_inode_info链接VFS inode和原始inode

结构体ext2_inode是存放于磁盘上的原始inode

结构体ext2_dir_entry_2也是磁盘上的一个结构，它表示目录下面的一个目录项。也就是目录的内容（子目录或者文件），目录也有一个inode，它也有数据块，其数据块上的存放的每一项都是用ext2_dir_entry_2来表示，例如：

chenying@chenying:~/workspace/kernel_4.12/linux-4.12.3/mm$ ls ~/workspace/

1496324869gf_common.h androidJ6 aosp dumpe2fs.txt gf_common.h kernel_4.12 log readme

用命令dumpe2fs可以dump出文件系统的信息：

chenying@chenying:~/workspace/kernel_4.12/linux-4.12.3/fs$ sudo dumpe2fs -h /dev/sda1

dumpe2fs 1.43.3 (04-Sep-2016)

Filesystem volume name:

Last mounted on: /

Filesystem UUID: 22af4caf-a05b-4d8f-8004-30d531867b55

Filesystem magic number: 0xEF53

Filesystem revision #: 1 (dynamic)

Filesystem features: has_journal ext_attr resize_inode dir_index filetype needs_recovery extent 64bit flex_bg sparse_super large_file huge_file dir_nlink extra_isize metadata_csum

Filesystem flags: signed_directory_hash

Default mount options: user_xattr acl

Filesystem state: clean

Errors behavior: Continue

Filesystem OS type: Linux

Inode count: 16252928

Block count: 65011456

Reserved block count: 3250572

Free blocks: 23809796

Free inodes: 14213383

First block: 0

Block size: 4096

Fragment size: 4096

Group descriptor size: 64

Reserved GDT blocks: 1024

Blocks per group: 32768

Fragments per group: 32768

Inodes per group: 8192

Inode blocks per group: 512

Flex block group size: 16

Filesystem created: Fri Feb 17 21:00:51 2017

Last mount time: Sat Aug 12 10:26:31 2017

Last write time: Sat Aug 12 10:26:26 2017

Mount count: 32

块缓存

在深入ext2实现逻辑之前我们先插一节块缓存，下面先看快缓存结构：

块缓存主要用在两个地方，页缓存和块设备原始数据读取（独立块缓存），例如超级块，组描述符块等等。在页缓存中块缓存依附于页，页释放之后块缓存就释放。独立块缓存由一个lru缓存来管理，这个时候页依附于块缓存，块缓存释放页就释放。

struct bh_lru {
struct buffer_head *bhs[BH_LRU_SIZE];
};

这两种块缓存都是用下面函数创建，他们的不同在于管理的视角不同。

struct buffer_head *alloc_page_buffers(struct page *page, unsigned long size,
int retry)

EXT2文件系统挂载

3.1 注册ext2文件系统类型

静态定义EXT2文件系统类型ext2_fs_type，并通过register_filesystem将其添加到全局链表file_systems上

static struct file_system_type ext2_fs_type = {
.owner      = THIS_MODULE,
.name       = "ext2",
.mount      = ext2_mount,
.kill_sb    = kill_block_super,
.fs_flags   = FS_REQUIRES_DEV,
};
MODULE_ALIAS_FS("ext2");
static int __init init_ext2_fs(void)
{
int err;
err = init_inodecache();
if (err)
return err;
err = register_filesystem(&ext2_fs_type);//将ext2_fs_type挂到全局链表file_systems上
if (err)
goto out;
return 0;
out:
destroy_inodecache();
return err;
}


3.2 ext2文件系统挂载

sget：分配super_block并且将super_block添加到全局链表super_blocks和file_system_type ->fs_supers

fill_super：函数指针，这里指向ext2_fill_super，用于从文件系统中读取super_block，下面细讲：

static int ext2_fill_super(struct super_block *sb, void *data, int silent)
{
struct buffer_head * bh;
struct ext2_sb_info * sbi;
struct ext2_super_block * es;
struct inode *root;
unsigned long sb_block = get_sb_block(&data);
unsigned long logic_sb_block;
unsigned long offset = 0;
int blocksize = BLOCK_SIZE;
int db_count;
sbi = kzalloc(sizeof(*sbi), GFP_KERNEL); //分配ext2_sb_info结构
if (!sbi)
goto failed;
sb->s_fs_info = sbi; //VFS中的super_block通过sb->s_fs_info与ext2_sb_info相连接
sbi->s_sb_block = sb_block;
blocksize = sb_min_blocksize(sb, BLOCK_SIZE);
......
if (!(bh = sb_bread(sb, logic_sb_block))) { //从磁盘中读取原始的超级块结构ext2_super_block
ext2_msg(sb, KERN_ERR, "error: unable to read superblock");
goto failed_sbi;
}
es = (struct ext2_super_block *) (((char *)bh->b_data) + offset);
sbi->s_es = es;
......
sb->s_magic = le16_to_cpu(es->s_magic);
blocksize = BLOCK_SIZE <s_es->s_log_block_size);
/*如果超级块的实际块大小与假设的大小不一致就重新读取超级块，因为超级块占用一个块大小，函数sb_bread也是从指定块号读取一个块大小，如果实际块与假设的块大小不一致就重新读取一个准确的块
大小*/
if (sb->s_blocksize != blocksize) { 
brelse(bh);
if (!sb_set_blocksize(sb, blocksize)) {
ext2_msg(sb, KERN_ERR,
"error: bad blocksize %d", blocksize);
goto failed_sbi;
}
logic_sb_block = (sb_block*BLOCK_SIZE) / blocksize;
offset = (sb_block*BLOCK_SIZE) % blocksize;
bh = sb_bread(sb, logic_sb_block);
es = (struct ext2_super_block *) (((char *)bh->b_data) + offset);
sbi->s_es = es;
}
......
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);
sbi->s_inodes_per_block = sb->s_blocksize / EXT2_INODE_SIZE(sb);
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; //让s_sbh指向原始超级块数据
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));
......
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);
......
for (i = 0; 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 brelse (sbi->s_group_desc[j]);
ext2_msg(sb, KERN_ERR,
"error: unable to read group descriptors");
goto failed_mount_group_desc;
}
}
sbi->s_gdb_count = db_count; //设置组描述符所占用的块数
......
/*初始化预分配窗口*/
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);
......
sb->s_op = &ext2_sops; //设置super_operations
......
root = ext2_iget(sb, EXT2_ROOT_INO); 
if (IS_ERR(root)) {
ret = PTR_ERR(root);
goto failed_mount3;
}
sb->s_root = d_make_root(root); //创建根目录的dentry
......
ext2_write_super(sb);
......


3.3文件系统操作

inode包含了文件操作的全部信息，文件打开时候的file结构初始化信息页都是来源于inode，下面是inode创建时的主要逻辑：

struct inode *ext2_iget (struct super_block *sb, unsigned long ino)
{   
struct ext2_inode_info *ei;
struct buffer_head * bh;
struct ext2_inode *raw_inode;
struct inode *inode;
inode = iget_locked(sb, ino); //创建VFS inode和ext2_inode_info
ei = EXT2_I(inode);
ei->i_block_alloc_info = NULL;
raw_inode = ext2_get_inode(inode->i_sb, ino, &bh); //到inode块表中去读取原始inode
......
if (S_ISREG(inode->i_mode)) {
inode->i_op = &ext2_file_inode_operations;
if (test_opt(inode->i_sb, NOBH)) {
inode->i_mapping->a_ops = &ext2_nobh_aops;
inode->i_fop = &ext2_file_operations;
} else {
inode->i_mapping->a_ops = &ext2_aops;  //页缓存操作函数集
inode->i_fop = &ext2_file_operations;  //设置file_operations
}
} else if (S_ISDIR(inode->i_mode)) {
inode->i_op = &ext2_dir_inode_operations;
......
}

文件数据读取

文件数据块分散在磁盘上，要对数据进行读写操作就先要找到文件数据块的块号，函数do_mpage_readpage的工作就是根据文件数据位置偏移找到对应的数据块块号。map_bh用于读取inode的映射块。

static struct bio *
do_mpage_readpage(struct bio *bio, struct page *page, unsigned nr_pages,sector_t *last_block_in_bio, struct buffer_head *map_bh,unsigned long *first_logical_block, get_block_t get_block,gfp_t gfp)
{struct inode *inode = page->mapping->host;const unsigned blkbits = inode->i_blkbits;const unsigned blocks_per_page = PAGE_SIZE >> blkbits;const unsigned blocksize = 1 <index <<(PAGE_SHIFT - blkbits);last_block = block_in_file + nr_pages * blocks_per_page;last_block_in_file = (i_size_read(inode) + blocksize - 1) >> blkbits;if (last_block > last_block_in_file)last_block = last_block_in_file;page_block = 0;nblocks = map_bh->b_size >> blkbits;if (buffer_mapped(map_bh) && block_in_file > *first_logical_block &&block_in_file <(*first_logical_block + nblocks)) { //如果前一次循环已经读取了映射块，就在其中查找映射关系unsigned map_offset = block_in_file - *first_logical_block;unsigned last = nblocks - map_offset;for (relative_block = 0; ; relative_block++) {if (relative_block == last) {clear_buffer_mapped(map_bh);break;}if (page_block == blocks_per_page)break;blocks[page_block] = map_bh->b_blocknr + map_offset +relative_block;page_block++;block_in_file++;}bdev = map_bh->b_bdev;}map_bh->b_page = page;while (page_block b_state = 0;map_bh->b_size = 0;if (block_in_file b_size = (last_block-block_in_file) <b_size >> blkbits;for (relative_block = 0; ; relative_block++) {if (relative_block == nblocks) {clear_buffer_mapped(map_bh);break;} else if (page_block == blocks_per_page)break;blocks[page_block] = map_bh->b_blocknr+relative_block;page_block++;block_in_file++;}bdev = map_bh->b_bdev;}......
alloc_new:if (bio == NULL) {if (first_hole == blocks_per_page) {if (!bdev_read_page(bdev, blocks[0] <<(blkbits - 9),page))goto out;}bio = mpage_alloc(bdev, blocks[0] <<(blkbits - 9), min_t(int, nr_pages, BIO_MAX_PAGES), gfp); //根据前面找到的数据块编号初始化bioif (bio == NULL)goto confused;}length = first_hole <out:return bio;......

块映射

前面提到的函数指针get_block指向函数ext2_get_block，其实现逻辑如下：

inode预分配窗口的初始化：

void ext2_init_block_alloc_info(struct inode *inode)
{
struct ext2_inode_info *ei = EXT2_I(inode);
struct ext2_block_alloc_info *block_i;
struct super_block *sb = inode->i_sb;
block_i = kmalloc(sizeof(*block_i), GFP_NOFS);
if (block_i) {
struct ext2_reserve_window_node *rsv = &block_i->rsv_window_node;
rsv->rsv_start = EXT2_RESERVE_WINDOW_NOT_ALLOCATED;
rsv->rsv_end = EXT2_RESERVE_WINDOW_NOT_ALLOCATED; //标识预分配窗口为空
if (!test_opt(sb, RESERVATION))
rsv->rsv_goal_size = 0;
else
rsv->rsv_goal_size = EXT2_DEFAULT_RESERVE_BLOCKS; //默认预分配窗口大小为8
rsv->rsv_alloc_hit = 0;
block_i->last_alloc_logical_block = 0;
block_i->last_alloc_physical_block = 0;
}
ei->i_block_alloc_info = block_i;
}

下面正式讲解块的分配：

ext2_fsblk_t ext2_new_blocks(struct inode *inode, ext2_fsblk_t goal,
unsigned long *count, int *errp)
{
......
struct ext2_super_block *es;
struct ext2_sb_info *sbi;
struct ext2_reserve_window_node *my_rsv = NULL;
struct ext2_block_alloc_info *block_i;
unsigned short windowsz = 0;
unsigned long ngroups;
unsigned long num = *count;
sb = inode->i_sb;
sbi = EXT2_SB(sb);
block_i = EXT2_I(inode)->i_block_alloc_info;
if (block_i) {
windowsz = block_i->rsv_window_node.rsv_goal_size;
if (windowsz > 0)
my_rsv = &block_i->rsv_window_node;
}
group_no = (goal - le32_to_cpu(es->s_first_data_block)) /
EXT2_BLOCKS_PER_GROUP(sb);  // 计算goal所在的块组
goal_group = group_no;
retry_alloc:
gdp = ext2_get_group_desc(sb, group_no, &gdp_bh); //获取组描述符
if (!gdp)
goto io_error;
free_blocks = le16_to_cpu(gdp->bg_free_blocks_count);
if (free_blocks > 0) {
grp_target_blk = ((goal - le32_to_cpu(es->s_first_data_block)) %
EXT2_BLOCKS_PER_GROUP(sb));
bitmap_bh = read_block_bitmap(sb, group_no); //读取块组数据块位图
if (!bitmap_bh)
goto io_error;
grp_alloc_blk = ext2_try_to_allocate_with_rsv(sb, group_no,
bitmap_bh, grp_target_blk,
my_rsv, &num); //分配数据块，并实现数据块预分配
if (grp_alloc_blk >= 0)
goto allocated;
}
ngroups = EXT2_SB(sb)->s_groups_count;
for (bgi = 0; bgi ......
}
if (my_rsv) {
my_rsv = NULL;
windowsz = 0;
group_no = goal_group;
goto retry_alloc;
}
allocated:
ret_block = grp_alloc_blk + ext2_group_first_block_no(sb, group_no);
......
return ret_block; //返回分配到的块组的块号
......

下面函数是块分配的核心函数，首先查找一个可以容纳预分配窗口大小的空闲空间，然后将数据块位图上对应的位置设置为1，表示已分配。

static ext2_grpblk_t
ext2_try_to_allocate_with_rsv(struct super_block *sb, unsigned int group,struct buffer_head *bitmap_bh, ext2_grpblk_t grp_goal,struct ext2_reserve_window_node * my_rsv,unsigned long *count)
{ext2_fsblk_t group_first_block, group_last_block;ext2_grpblk_t ret = 0;
unsigned long num = *count;if (my_rsv == NULL) { //直接分配数据块不做预分配return ext2_try_to_allocate(sb, group, bitmap_bh,grp_goal, count, NULL);
}group_first_block = ext2_group_first_block_no(sb, group);group_last_block = group_first_block + (EXT2_BLOCKS_PER_GROUP(sb) - 1);while (1) {if (rsv_is_empty(&my_rsv->rsv_window) || (ret <0) ||!goal_in_my_reservation(&my_rsv->rsv_window,grp_goal, group, sb)) {//预分配窗口为空或者目标块不在my_rsv中if (my_rsv->rsv_goal_size <*count) // my_rsv->rsv_goal_size初始值为8my_rsv->rsv_goal_size = *count; ret = alloc_new_reservation(my_rsv, grp_goal, sb,group, bitmap_bh); //查找一个可以容纳预分配窗口大小的空闲空间if (ret <0)break; /* failed */if (!goal_in_my_reservation(&my_rsv->rsv_window,grp_goal, group, sb))grp_goal = -1; } else if (grp_goal >= 0) {int curr = my_rsv->rsv_end -(grp_goal + group_first_block) + 1;if (curr <*count)try_to_extend_reservation(my_rsv, sb,*count - curr);}if ((my_rsv->rsv_start > group_last_block) ||(my_rsv->rsv_end s_rsv_window_root, 1);BUG();}ret = ext2_try_to_allocate(sb, group, bitmap_bh, grp_goal,&num, &my_rsv->rsv_window);// 将预分配窗口中的数据块在位图上对应的位置设置为1if (ret >= 0) {my_rsv->rsv_alloc_hit += num; //统计预分配命中数*count = num; break; /* succeed */}num = *count; //返回分配到的块数}return ret;
}

下面函数是预分配的核心：

static int alloc_new_reservation(struct ext2_reserve_window_node *my_rsv,ext2_grpblk_t grp_goal, struct super_block *sb,unsigned int group, struct buffer_head *bitmap_bh)
{ struct ext2_reserve_window_node *search_head;ext2_fsblk_t group_first_block, group_end_block, start_block;ext2_grpblk_t first_free_block;struct rb_root *fs_rsv_root = &EXT2_SB(sb)->s_rsv_window_root;unsigned long size;int ret;spinlock_t *rsv_lock = &EXT2_SB(sb)->s_rsv_window_lock;group_first_block = ext2_group_first_block_no(sb, group);group_end_block = group_first_block + (EXT2_BLOCKS_PER_GROUP(sb) - 1);start_block = grp_goal + group_first_block; //搜索空间区间的起始位置size = my_rsv->rsv_goal_size;search_head = search_reserve_window(fs_rsv_root, start_block); //查找离start_block最近的预留窗口retry:ret = find_next_reservable_window(search_head, my_rsv, sb,start_block, group_end_block); //以search_head为起点查找一个可以容纳my_rsv且不与其他预留窗口重叠的空闲区间......first_free_block = bitmap_search_next_usable_block(my_rsv->rsv_start - group_first_block,bitmap_bh, group_end_block - group_first_block + 1); //在位表中查找块组中rsv_start往后第一个空闲块,因为不是所有块分配都是通过预分配，所以有些块可能已经分配了但是在预留窗口中找不到......start_block = first_free_block + group_first_block;if (start_block >= my_rsv->rsv_start && start_block <= my_rsv->rsv_end)//空闲块是否在my_rsv中return 0; /* success */search_head = my_rsv; //如果my_rsv区间中的块都被分配出去了就以my_rsv为起点重新搜索goto retry;
}