android文件系统挂载分析(1)---正常开机挂载
未完，更新中 。。。
"android"系列分为三部分：
1.正常开机挂载
2.encryption
3.dm-verity
我们知道android有很多分区，如"system","userdata","cache"，他们是何时挂载的？如何挂载的？这个系列的文章进行分析。这里介绍第一部分，android手机正常开机各分区的挂载。这里我们以mtk平台进行分析，高通与mtk差别不是很大。
我们知道kernel起来以后执行的第一个文件是init进程，init进程会根据init.rc的规则启动进程或者服务。init.rc通过"import /init.${ro.hardware}.rc"语句导入平台的规则。
device/mediatek/mt6797/init.mt6797.rc on fs write /proc/bootprof "INIT:Mount_START" mount_all /fstab.mt6797 chown system system /mobile_info chmod 0771 /mobile_info exec /system/bin/tune2fs -O has_journal -u 10010 -r 4096 /dev/block/platform/mtk-msdc.0/11230000.msdc0/by-name/userdata write /proc/bootprof "INIT:Mount_END"
mount_all是一条命令，/fstab.mt6797是传入的参数
system/core/init/keywords.h ..... KEYWORD(mount_all, COMMAND, 1, do_mount_all) .....
从上面我们可以看出，mount_all命令对应的是do_mount_all函数，/fstab.mt6797是do_mount_all函数的传入参数
system/core/init/builtins.cpp int do_mount_all(int nargs, char **args) { pid_t pid; int ret = -1; int child_ret = -1; int status; struct fstab *fstab; if (nargs != 2) { return -1; } /* * Call fs_mgr_mount_all() to mount all filesystems. We fork(2) and //使用fs_mgr_mount_all()函数去挂载所有的文件系统，我们使用fork()函数分配一个新的进程，在子进程中做挂载的事情,这样即使挂载出现问题，也能保护init主进程。 * do the call in the child to provide protection to the main init * process if anything goes wrong (crash or memory leak), and wait for * the child to finish in the parent. */ pid = fork(); if (pid > 0) { //父进程，等待子进程（pid=0）返回 /* Parent. Wait for the child to return */ int wp_ret = TEMP_FAILURE_RETRY(waitpid(pid, &status, 0)); if (wp_ret < 0) { /* Unexpected error code. We will continue anyway. */ NOTICE("waitpid failed rc=%d: %s\n", wp_ret, strerror(errno)); } if (WIFEXITED(status)) { ret = WEXITSTATUS(status); } else { ret = -1; } } else if (pid == 0) { //子进程 /* child, call fs_mgr_mount_all() */ klog_set_level(6); /* So we can see what fs_mgr_mount_all() does */ //修改kernel log的等级，让我们可以看到fs_mgr_mount_all函数的log fstab = fs_mgr_read_fstab(args[1]); //args[1]是传入的参数/fstab.mt6797,是一个文件。 加载分区挂载文件的内容到fstab结构体中。 child_ret = fs_mgr_mount_all(fstab); //挂载分区******************* fs_mgr_free_fstab(fstab); if (child_ret == -1) { ERROR("fs_mgr_mount_all returned an error\n"); } _exit(child_ret); } else { /* fork failed, return an error */ return -1; } if (ret == FS_MGR_MNTALL_DEV_NEEDS_ENCRYPTION) { property_set("vold.decrypt", "trigger_encryption"); } else if (ret == FS_MGR_MNTALL_DEV_MIGHT_BE_ENCRYPTED) { property_set("ro.crypto.state", "encrypted"); property_set("ro.crypto.type", "block"); property_set("vold.decrypt", "trigger_default_encryption"); } else if (ret == FS_MGR_MNTALL_DEV_NOT_ENCRYPTED) { property_set("ro.crypto.state", "unencrypted"); /* If fs_mgr determined this is an unencrypted device, then trigger * that action. */ action_for_each_trigger("nonencrypted", action_add_queue_tail); } else if (ret == FS_MGR_MNTALL_DEV_NEEDS_RECOVERY) { /* Setup a wipe via recovery, and reboot into recovery */ ERROR("fs_mgr_mount_all suggested recovery, so wiping data via recovery.\n"); ret = wipe_data_via_recovery(); /* If reboot worked, there is no return. */ } else if (ret == FS_MGR_MNTALL_DEV_DEFAULT_FILE_ENCRYPTED) { if (e4crypt_install_keyring()) { return -1; } property_set("ro.crypto.state", "encrypted"); property_set("ro.crypto.type", "file"); // Although encrypted, we have device key, so we do not need to // do anything different from the nonencrypted case. action_for_each_trigger("nonencrypted", action_add_queue_tail); } else if (ret == FS_MGR_MNTALL_DEV_NON_DEFAULT_FILE_ENCRYPTED) { if (e4crypt_install_keyring()) { return -1; } property_set("ro.crypto.state", "encrypted"); property_set("ro.crypto.type", "file"); property_set("vold.decrypt", "trigger_restart_min_framework"); } else if (ret > 0) { ERROR("fs_mgr_mount_all returned unexpected error %d\n", ret); } /* else ... < 0: error */ return ret; }
fork（）函数通过系统调用创建一个与原来进程几乎完全相同的进程，也就是两个进程可以做完全相同的事，但如果初始参数或者传入的变量不同，两个进程也可以做不同的事。返回值小于0为error，返回值等于0为子进程，返回值大于0为父进程。
args[1]是传入的参数/fstab.mt6797,是一个文件，生成的位置在/out/target/product/xxx/root/fstab.mt6797,生成这个文件的源文件位于vendor/mediatek/proprietary/hardware/fstab/mt6797/，根据编译规则确定fstab.mt6797文件的内容。
在do_mount_all()函数中，比较重要的两个函数如下，我们分析一下这两个函数
stab = fs_mgr_read_fstab(args[1]);
child_ret = fs_mgr_mount_all(fstab);
首先我们看下fstab结构体和fstab.mt6797文件，fstab结构提要存储fstab.mt6797文件中的挂载信息，
struct fstab { int num_entries; struct fstab_rec *recs; char *fstab_filename; }; struct fstab_rec { char *blk_device; char *mount_point; char *fs_type; unsigned long flags; char *fs_options; int fs_mgr_flags; char *key_loc; char *verity_loc; long long length; char *label; int partnum; int swap_prio; unsigned int zram_size; };
out/target/product/xxx/root/fstab.mt6797 # 1 "vendor/mediatek/proprietary/hardware/fstab/mt6797/fstab.in" # 1 "<built-in>" # 1 "<命令行>" # 1 "vendor/mediatek/proprietary/hardware/fstab/mt6797/fstab.in" # 20 "vendor/mediatek/proprietary/hardware/fstab/mt6797/fstab.in" /dev/block/platform/mtk-msdc.0/11230000.msdc0/by-name/system /system ext4 ro wait,verify /dev/block/platform/mtk-msdc.0/11230000.msdc0/by-name/userdata /data ext4 noatime,nosuid,nodev,noauto_da_alloc,discard wait,check,resize,forceencrypt=/dev/block/platform/mtk-msdc.0/11230000.msdc0/by-name/metadata, /dev/block/platform/mtk-msdc.0/11230000.msdc0/by-name/cache /cache ext4 noatime,nosuid,nodev,noauto_da_alloc,discard wait,check /dev/block/platform/mtk-msdc.0/11230000.msdc0/by-name/protect1 /protect_f ext4 noatime,nosuid,nodev,noauto_da_alloc,commit=1,nodelalloc wait,check,formattable /dev/block/platform/mtk-msdc.0/11230000.msdc0/by-name/protect2 /protect_s ext4 noatime,nosuid,nodev,noauto_da_alloc,commit=1,nodelalloc wait,check,formattable /dev/block/platform/mtk-msdc.0/11230000.msdc0/by-name/nvdata /nvdata ext4 noatime,nosuid,nodev,noauto_da_alloc,discard wait,check,formattable /dev/block/platform/mtk-msdc.0/11230000.msdc0/by-name/nvcfg /nvcfg ext4 noatime,nosuid,nodev,noauto_da_alloc,commit=1,nodelalloc wait,check,formattable /dev/block/platform/mtk-msdc.0/11230000.msdc0/by-name/mobile_info /mobile_info ext4 noatime,nosuid,nodev,noauto_da_alloc,discard wait,check # 39 "vendor/mediatek/proprietary/hardware/fstab/mt6797/fstab.in" /devices/mtk-msdc.0/11230000.msdc0* auto vfat defaults voldmanaged=sdcard0:auto /devices/mtk-msdc.0/11240000.msdc1* auto auto defaults voldmanaged=sdcard1:auto,encryptable=userdata /devices/soc/11270000.usb3_xhci* auto vfat defaults voldmanaged=usbotg:auto
.................
我们先分析fstab结构体存放的挂载信息，通过fs_mgr_read_fstab实现
system/core/fs_mgr_fstab.c struct fstab *fs_mgr_read_fstab(const char *fstab_path) //从上面可以知道fstab_path为/fstab.mt6797 { FILE *fstab_file; int cnt, entries; ssize_t len; size_t alloc_len = 0; char *line = NULL; const char *delim = " \t"; char *save_ptr, *p; struct fstab *fstab = NULL; struct fs_mgr_flag_values flag_vals; #define FS_OPTIONS_LEN 1024 char tmp_fs_options[FS_OPTIONS_LEN]; fstab_file = fopen(fstab_path, "r"); //打开文件 if (!fstab_file) { ERROR("Cannot open file %s\n", fstab_path); return 0; } entries = 0; while ((len = getline(&line, &alloc_len, fstab_file)) != -1) { //一行一行的读取文件内容，line是指向存放该行字符的指针 第一次读取文件内容，填充fstab结构体的内容 /* if the last character is a newline, shorten the string by 1 byte */ if (line[len - 1] == '\n') { //如果最后一行是新行，缩短一字节的字符串 line[len - 1] = '\0'; } /* Skip any leading whitespace */ p = line; while (isspace(*p)) { p++; } /* ignore comments or empty lines */ //忽略注释和空格开始的行 if (*p == '#' || *p == '\0') continue; entries++; //有用信息的行数 } if (!entries) { ERROR("No entries found in fstab\n"); goto err; } /* Allocate and init the fstab structure */ fstab = calloc(1, sizeof(struct fstab)); //给fstab结构体分配内存 fstab->num_entries = entries; //fstab->num_entries 存放可用信息的总行数 fstab->fstab_filename = strdup(fstab_path); // fstab->fstab_filename 存放"fstab.mt6797"名称 fstab->recs = calloc(fstab->num_entries, sizeof(struct fstab_rec)); //给fstab->recs结构体分配内存 fseek(fstab_file, 0, SEEK_SET); cnt = 0; while ((len = getline(&line, &alloc_len, fstab_file)) != -1) { //第一次读取文件内容，填充结构体fstab->recs的内容 /* if the last character is a newline, shorten the string by 1 byte */ if (line[len - 1] == '\n') { line[len - 1] = '\0'; } /* Skip any leading whitespace */ p = line; while (isspace(*p)) { p++; } /* ignore comments or empty lines */ if (*p == '#' || *p == '\0') continue; /* If a non-comment entry is greater than the size we allocated, give an * error and quit. This can happen in the unlikely case the file changes * between the two reads. */ if (cnt >= entries) { ERROR("Tried to process more entries than counted\n"); break; } //下面以/dev/block/platform/mtk-msdc.0/11230000.msdc0/by-name/userdata /data ext4 noatime,nosuid,nodev,noauto_da_alloc,discard wait,check,resize,forceencrypt=/dev/block/platform/mtk-msdc.0/11230000.msdc0/by-name/metadata,为例 if (!(p = strtok_r(line, delim, &save_ptr))) { //strtok_r字符串分割函数，line表示要分割的字符串，delim要分割的标志，p存放分割后的字符串 ERROR("Error parsing mount source\n"); goto err; } fstab->recs[cnt].blk_device = strdup(p); //fstab->recs[cnt].blk_device 存放文件系统绝对路径 /dev/block/platform/mtk-msdc.0/11230000.msdc0/by-name/userdata if (!(p = strtok_r(NULL, delim, &save_ptr))) { ERROR("Error parsing mount_point\n"); goto err; } fstab->recs[cnt].mount_point = strdup(p); //fstab->recs[cnt].mount_point 挂载点 /data if (!(p = strtok_r(NULL, delim, &save_ptr))) { ERROR("Error parsing fs_type\n"); goto err; } fstab->recs[cnt].fs_type = strdup(p); //fstab->recs[cnt].fs_type 文件系统类型 ext4 if (!(p = strtok_r(NULL, delim, &save_ptr))) { //此时的p存放的参数 noatime,nosuid,nodev,noauto_da_alloc,discard ERROR("Error parsing mount_flags\n"); goto err; } tmp_fs_options[0] = '\0'; fstab->recs[cnt].flags = parse_flags(p, mount_flags, NULL, //fstab->recs[cnt].flags 表示这行有无参数 tmp_fs_options, FS_OPTIONS_LEN);
/*
static struct flag_list mount_flags[] = {
{ "noatime", MS_NOATIME },
{ "noexec", MS_NOEXEC },
{ "nosuid", MS_NOSUID },
{ "nodev", MS_NODEV },
{ "nodiratime", MS_NODIRATIME },
{ "ro", MS_RDONLY },
{ "rw", 0 },
{ "remount", MS_REMOUNT },
{ "bind", MS_BIND },
{ "rec", MS_REC },
{ "unbindable", MS_UNBINDABLE },
{ "private", MS_PRIVATE },
{ "slave", MS_SLAVE },
{ "shared", MS_SHARED },
{ "defaults", 0 },
{ 0, 0 },
};
*/
/* fs_options are optional */ if (tmp_fs_options[0]) { //是个flags list，读取noatime,nosuid,nodev,noauto_da_alloc,discard fstab->recs[cnt].fs_options = strdup(tmp_fs_options); } else { fstab->recs[cnt].fs_options = NULL; } if (!(p = strtok_r(NULL, delim, &save_ptr))) { //此时p存放剩下的参数wait,check,resize,forceencrypt=/dev/block/platform/mtk-msdc.0/11230000.msdc0/by-name/metadata ERROR("Error parsing fs_mgr_options\n"); goto err; } fstab->recs[cnt].fs_mgr_flags = parse_flags(p, fs_mgr_flags, &flag_vals, NULL, 0);
/*
static struct flag_list fs_mgr_flags[] = {
{ "wait", MF_WAIT },
{ "check", MF_CHECK },
{ "encryptable=",MF_CRYPT },
{ "forceencrypt=",MF_FORCECRYPT },
{ "fileencryption",MF_FILEENCRYPTION },
{ "nonremovable",MF_NONREMOVABLE },
{ "voldmanaged=",MF_VOLDMANAGED},
{ "length=", MF_LENGTH },
{ "recoveryonly",MF_RECOVERYONLY },
{ "swapprio=", MF_SWAPPRIO },
{ "zramsize=", MF_ZRAMSIZE },
{ "verify", MF_VERIFY },
{ "noemulatedsd", MF_NOEMULATEDSD },
{ "notrim", MF_NOTRIM },
{ "formattable", MF_FORMATTABLE },
#ifdef MTK_FSTAB_FLAGS
{ "resize", MF_RESIZE },
#endif
{ "defaults", 0 },
{ 0, 0 },
};
*/
fstab->recs[cnt].key_loc = flag_vals.key_loc; //fstab->recs[cnt].key_loc 存放"="后的内容 这里是 "/dev/block/platform/mtk-msdc.0/11230000.msdc0/by-name/metadata" fstab->recs[cnt].verity_loc = flag_vals.verity_loc; fstab->recs[cnt].length = flag_vals.part_length; fstab->recs[cnt].label = flag_vals.label; fstab->recs[cnt].partnum = flag_vals.partnum; fstab->recs[cnt].swap_prio = flag_vals.swap_prio; fstab->recs[cnt].zram_size = flag_vals.zram_size; cnt++; } fclose(fstab_file); free(line); return fstab; err: fclose(fstab_file); free(line); if (fstab) fs_mgr_free_fstab(fstab); return NULL; }
fstab.mt6797文件的内容已经读取到fstab结构提中，下面开始分析挂载函数fs_mgr_mount_all,传入的参数就是上面分析的fstab。
system/core/fs_mgr/fs_mgr.c int fs_mgr_mount_all(struct fstab *fstab) { int i = 0; int encryptable = FS_MGR_MNTALL_DEV_NOT_ENCRYPTED; //这个变量涉及到encryption加密，后面的文章会详细介绍这块 int error_count = 0; int mret = -1; int mount_errno = 0; int attempted_idx = -1; if (!fstab) { return -1; } for (i = 0; i < fstab->num_entries; i++) { /* Don't mount entries that are managed by vold */ if (fstab->recs[i].fs_mgr_flags & (MF_VOLDMANAGED | MF_RECOVERYONLY)) { continue; } /* Skip swap and raw partition entries such as boot, recovery, etc */ if (!strcmp(fstab->recs[i].fs_type, "swap") || !strcmp(fstab->recs[i].fs_type, "emmc") || !strcmp(fstab->recs[i].fs_type, "mtd")) { continue; } /* Translate LABEL= file system labels into block devices */ if (!strcmp(fstab->recs[i].fs_type, "ext2") || !strcmp(fstab->recs[i].fs_type, "ext3") || !strcmp(fstab->recs[i].fs_type, "ext4")) { int tret = translate_ext_labels(&fstab->recs[i]); if (tret < 0) { ERROR("Could not translate label to block device\n"); continue; } } ERROR("blk device name %s\n", fstab->recs[i].blk_device); #if defined(MTK_UBIFS_SUPPORT) || defined (MTK_FTL_SUPPORT) //这里支持UBIFS/FTL，这里没有使用 if (strcmp(fstab->recs[i].fs_type, "ubifs") == 0 && strncmp("ubi@", fstab->recs[i].blk_device, 4) == 0) { char tmp[25]; int n = ubi_attach_mtd(fstab->recs[i].blk_device + 4); if (n < 0) { ERROR("ubi_attach_mtd fail device name %s\n", fstab->recs[i].blk_device+4); return -1; } n = sprintf(tmp, "/dev/ubi%d_0", n); free(fstab->recs[i].blk_device); fstab->recs[i].blk_device = malloc(n+1); sprintf(fstab->recs[i].blk_device, "%s", tmp); ERROR("debug : ubifs blk_device %s", fstab->recs[i].blk_device); } else if (!strcmp(fstab->recs[i].fs_type, "rawfs") || !strcmp(fstab->recs[i].fs_type, "yaffs2")) { char tmp[25]; int n = mtd_name_to_number(fstab->recs[i].blk_device + 4); if (n < 0) { return -1; } n = sprintf(tmp, "/dev/block/mtdblock%d", n); free(fstab->recs[i].blk_device); fstab->recs[i].blk_device = malloc(n+1); sprintf(fstab->recs[i].blk_device, "%s", tmp); ERROR("debug : rawfs blk_device %s", fstab->recs[i].blk_device); } #ifdef MTK_FTL_SUPPORT else if (!strcmp(fstab->recs[i].fs_type, "ext4") && strstr(fstab->recs[i].blk_device, "ftl")) { char tmp[30]; int err = 0; int n = -1; int ubi_num = fstab->recs[i].blk_device[21] - '0'; ERROR("debug : mtk_ftl_blk %s ubi_num %d\n", fstab->recs[i].blk_device, ubi_num); if(strstr(fstab->recs[i].mount_point, "system")){ n = ubi_attach_mtd("system"); }else if(strstr(fstab->recs[i].mount_point, "data")){ n = ubi_attach_mtd("userdata"); }else if(strstr(fstab->recs[i].mount_point, "cache")){ n = ubi_attach_mtd("cache"); } if((n != ubi_num) && (n >= 0)) { ERROR("ubi number: %d == %d\n", n, ubi_num); ubi_num = n; } n = sprintf(tmp, "/dev/ubi%d_0", ubi_num); if (fstab->recs[i].fs_mgr_flags & MF_WAIT) { int ret = wait_for_file(tmp, WAIT_TIMEOUT); ERROR("wait_for_file(%s) ret = %d, errno = %s\n", fstab->recs[i].blk_device, ret, strerror(errno)); } err = ftl_attach_ubi(ubi_num); if (err < 0) { return -1; } } #endif #endif if (fstab->recs[i].fs_mgr_flags & MF_WAIT) { //当有"wait"关键字时，让系统sleep一会 wait_for_file(fstab->recs[i].blk_device, WAIT_TIMEOUT); } if ((fstab->recs[i].fs_mgr_flags & MF_VERIFY) && device_is_secure()) { //这个if涉及到system的挂载问题，系统默认是把system挂载到dm-01上，用户不可以remount，使能这个if，用户可以remount，这块会在后面的文章详细介绍 int rc = fs_mgr_setup_verity(&fstab->recs[i]); if (device_is_debuggable() && rc == FS_MGR_SETUP_VERITY_DISABLED) { INFO("Verity disabled"); } else if (rc != FS_MGR_SETUP_VERITY_SUCCESS) { ERROR("Could not set up verified partition, skipping!\n"); continue; } }
//正常开机mount主要从是下面的代码 int last_idx_inspected; int top_idx = i; mret = mount_with_alternatives(fstab, i, &last_idx_inspected, &attempted_idx, encryptable); //函数挂载 i = last_idx_inspected; mount_errno = errno; /* Deal with encryptability. */ if (!mret) { int status = handle_encryptable(fstab, &fstab->recs[attempted_idx]); //处理加密 if (status == FS_MGR_MNTALL_FAIL) { /* Fatal error - no point continuing */ return status; } if (status != FS_MGR_MNTALL_DEV_NOT_ENCRYPTED) { if (encryptable != FS_MGR_MNTALL_DEV_NOT_ENCRYPTED) { // Log and continue ERROR("Only one encryptable/encrypted partition supported\n"); } encryptable = status; } /* Success! Go get the next one */ continue; } /* mount(2) returned an error, handle the encryptable/formattable case */ bool wiped = partition_wiped(fstab->recs[top_idx].blk_device); if (mret && mount_errno != EBUSY && mount_errno != EACCES && fs_mgr_is_formattable(&fstab->recs[top_idx]) && wiped) { /* top_idx and attempted_idx point at the same partition, but sometimes * at two different lines in the fstab. Use the top one for formatting * as that is the preferred one. */ ERROR("%s(): %s is wiped and %s %s is formattable. Format it.\n", __func__, fstab->recs[top_idx].blk_device, fstab->recs[top_idx].mount_point, fstab->recs[top_idx].fs_type); if (fs_mgr_is_encryptable(&fstab->recs[top_idx]) && strcmp(fstab->recs[top_idx].key_loc, KEY_IN_FOOTER)) { int fd = open(fstab->recs[top_idx].key_loc, O_WRONLY, 0644); if (fd >= 0) { INFO("%s(): also wipe %s\n", __func__, fstab->recs[top_idx].key_loc); wipe_block_device(fd, get_file_size(fd)); close(fd); } else { ERROR("%s(): %s wouldn't open (%s)\n", __func__, fstab->recs[top_idx].key_loc, strerror(errno)); } } if (fs_mgr_do_format(&fstab->recs[top_idx]) == 0) { /* Let's replay the mount actions. */ i = top_idx - 1; continue; } } if (mret && mount_errno != EBUSY && mount_errno != EACCES && fs_mgr_is_encryptable(&fstab->recs[attempted_idx])) { if (wiped) { ERROR("%s(): %s is wiped and %s %s is encryptable. Suggest recovery...\n", __func__, fstab->recs[attempted_idx].blk_device, fstab->recs[attempted_idx].mount_point, fstab->recs[attempted_idx].fs_type); encryptable = FS_MGR_MNTALL_DEV_NEEDS_RECOVERY; continue; } else { /* Need to mount a tmpfs at this mountpoint for now, and set * properties that vold will query later for decrypting */ ERROR("%s(): possibly an encryptable blkdev %s for mount %s type %s )\n", __func__, fstab->recs[attempted_idx].blk_device, fstab->recs[attempted_idx].mount_point, fstab->recs[attempted_idx].fs_type); if (fs_mgr_do_tmpfs_mount(fstab->recs[attempted_idx].mount_point) < 0) { ++error_count; continue; } } encryptable = FS_MGR_MNTALL_DEV_MIGHT_BE_ENCRYPTED; } else { ERROR("Failed to mount an un-encryptable or wiped partition on" "%s at %s options: %s error: %s\n", fstab->recs[attempted_idx].blk_device, fstab->recs[attempted_idx].mount_point, fstab->recs[attempted_idx].fs_options, strerror(mount_errno)); ++error_count; continue; } } if (error_count) { return -1; } else { return encryptable; } }
