Android 12 init(1) 启动流程分析

文章托管在gitee上 Android Notes , 同步csdn
本文基于Android12 分析

概述

init是 Android 启动的第一个用户空间进程,它的地位非常重要,它fork产生系统的一些关键进程(如zygote,surfaceflinger进程),而zygote进一步fork产生system_server和其他应用进程,通过这套逻辑构建了Android的进程层次结构体系。init进程的功能包含但不限于以下:

  • 挂载系统分区和加载一些内核模块
  • 加载sepolicy 及使能 selinux
  • 支持属性服务
  • 启动脚本rc文件解析
  • 执行事件触发器和属性改变的事件
  • 子进程死亡监听,回收僵尸进程
  • 非oneshot服务保活

通过ps命令看看init进程信息

# ps -A|grep init                                                                                                           
root             1     0 10847128  4020 do_epoll_wait       0 S init  # 这个是 init 进程
root           166     1 10817360  1916 do_sys_poll         0 S init  # 这个是 subcontext 进程

在启动内核的start_kernel函数流程中,会调用run_init_process函数执行init程序,来启动init进程

run_init_process

在Android中执行的init是/init

/// @kernel_common/init/main.c
static int run_init_process(const char *init_filename)
{
	argv_init[0] = init_filename;
	pr_info("Run %s as init process\n", init_filename);
	return do_execve(getname_kernel(init_filename),
		(const char __user *const __user *)argv_init,
		(const char __user *const __user *)envp_init);
}

/init 实际上是一个软链接,指向的是/system/bin/init

# ls /init -lZ                                                                                           
lrwxr-x--- 1 root shell u:object_r:init_exec:s0  16 2021-12-20 15:52 /init -> /system/bin/init

接下来,进入init的main函数。

main

main执行分为几个阶段:

  • FirstStage 挂载一些基础文件系统和加载内核模块等
  • selinux_setup 执行selinux的初始化
  • SecondStage 挂载其他文件系统,启动属性服务,执行boot流程等,主要逻辑都在这里实现
/// @system/core/init/main.cpp
int main(int argc, char** argv) {
#if __has_feature(address_sanitizer)
    __asan_set_error_report_callback(AsanReportCallback);
#endif
    // Boost prio which will be restored later
    setpriority(PRIO_PROCESS, 0, -20);
    if (!strcmp(basename(argv[0]), "ueventd")) { // 处理uventd启动,共用一个main
        return ueventd_main(argc, argv);
    }

    if (argc > 1) {
        if (!strcmp(argv[1], "subcontext")) { // subcontext 子进程入口,用于执行来自init的某些任务
            android::base::InitLogging(argv, &android::base::KernelLogger);
            const BuiltinFunctionMap& function_map = GetBuiltinFunctionMap();

            return SubcontextMain(argc, argv, &function_map);
        }

        if (!strcmp(argv[1], "selinux_setup")) {// selinux初始化阶段
            return SetupSelinux(argv);
        }

        if (!strcmp(argv[1], "second_stage")) {// 启动第二阶段
            return SecondStageMain(argc, argv);
        }
    }

    return FirstStageMain(argc, argv); // 启动第一阶段
}

FirstStageMain

第一阶段初始化

/// @system/core/init/first_stage_init.cpp
int FirstStageMain(int argc, char** argv) {
    if (REBOOT_BOOTLOADER_ON_PANIC) {// 设置panic处理器
        InstallRebootSignalHandlers();
    }

    boot_clock::time_point start_time = boot_clock::now();

    std::vector errors;
#define CHECKCALL(x) \
    if ((x) != 0) errors.emplace_back(#x " failed", errno);

    // Clear the umask.
    umask(0);

    CHECKCALL(clearenv());
    CHECKCALL(setenv("PATH", _PATH_DEFPATH, 1));
    // 挂载一些基础文件系统
    // Get the basic filesystem setup we need put together in the initramdisk
    // on / and then we'll let the rc file figure out the rest.
    CHECKCALL(mount("tmpfs", "/dev", "tmpfs", MS_NOSUID, "mode=0755"));
    CHECKCALL(mkdir("/dev/pts", 0755));
    CHECKCALL(mkdir("/dev/socket", 0755));
    CHECKCALL(mkdir("/dev/dm-user", 0755));
    CHECKCALL(mount("devpts", "/dev/pts", "devpts", 0, NULL));
#define MAKE_STR(x) __STRING(x)
    // /proc 伪文件系统,记录进程、线程相关实时状态
    CHECKCALL(mount("proc", "/proc", "proc", 0, "hidepid=2,gid=" MAKE_STR(AID_READPROC)));
#undef MAKE_STR
    // Don't expose the raw commandline to unprivileged processes.
    CHECKCALL(chmod("/proc/cmdline", 0440)); // 只读
    std::string cmdline;
    android::base::ReadFileToString("/proc/cmdline", &cmdline);
    // Don't expose the raw bootconfig to unprivileged processes.
    chmod("/proc/bootconfig", 0440);
    std::string bootconfig;
    android::base::ReadFileToString("/proc/bootconfig", &bootconfig);

InstallRebootSignalHandlers

init信号处理器,调试版本当init crash,默认重启到 bootLoader

void InstallRebootSignalHandlers() {
    // Instead of panic'ing the kernel as is the default behavior when init crashes,
    // we prefer to reboot to bootloader on development builds, as this will prevent
    // boot looping bad configurations and allow both developers and test farms to easily
    // recover.
    struct sigaction action;
    memset(&action, 0, sizeof(action));
    sigfillset(&action.sa_mask);
    action.sa_handler = [](int signal) {
        // These signal handlers are also caught for processes forked from init, however we do not
        // want them to trigger reboot, so we directly call _exit() for children processes here.
        if (getpid() != 1) { // 非init直接退出
            _exit(signal);
        }

        // Calling DoReboot() or LOG(FATAL) is not a good option as this is a signal handler.
        // RebootSystem uses syscall() which isn't actually async-signal-safe, but our only option
        // and probably good enough given this is already an error case and only enabled for
        // development builds.
        InitFatalReboot(signal); // 执行重启操作
    };
    action.sa_flags = SA_RESTART;
    // 设置信号处理器
    sigaction(SIGABRT, &action, nullptr);
    sigaction(SIGBUS, &action, nullptr);
    sigaction(SIGFPE, &action, nullptr);
    sigaction(SIGILL, &action, nullptr);
    sigaction(SIGSEGV, &action, nullptr);
#if defined(SIGSTKFLT)
    sigaction(SIGSTKFLT, &action, nullptr);
#endif
    sigaction(SIGSYS, &action, nullptr);
    sigaction(SIGTRAP, &action, nullptr);
}

InitFatalReboot

默认执行重启的 init_fatal_reboot_target 的值是 bootloader

/// @system/core/init/reboot_utils.cpp
static std::string init_fatal_reboot_target = "bootloader";

void __attribute__((noreturn)) InitFatalReboot(int signal_number) {
    auto pid = fork();

    if (pid == -1) {
        // Couldn't fork, don't even try to backtrace, just reboot.
        RebootSystem(ANDROID_RB_RESTART2, init_fatal_reboot_target);
    } else if (pid == 0) { // 子进程确保能重启
        // Fork a child for safety, since we always want to shut down if something goes wrong, but
        // its worth trying to get the backtrace, even in the signal handler, since typically it
        // does work despite not being async-signal-safe.
        sleep(5);
        RebootSystem(ANDROID_RB_RESTART2, init_fatal_reboot_target);
    }

    // 先尝试获取 backtrace ,然后执行重启,
    // In the parent, let's try to get a backtrace then shutdown.
    LOG(ERROR) 

DoFirstStageMount

这里探究一下,在这个first stage挂载了那些分区

/// @system/core/init/first_stage_mount.cpp
// Mounts partitions specified by fstab in device tree.
bool DoFirstStageMount(bool create_devices) {
    // Skips first stage mount if we're in recovery mode.
    if (IsRecoveryMode()) {
        LOG(INFO) 

FirstStageMount::DoFirstStageMount

FirstStageMount::MountPartitions

bool FirstStageMount::MountPartitions() {
	 // 挂载 /system
    if (!TrySwitchSystemAsRoot()) return false;
		// 移除不需要挂载的分区
    if (!SkipMountingPartitions(&fstab_, true /* verbose */)) return false;

从上面分析可知,挂载的信息存储在fstab_ 里面,它是在FirstStageMount::Create函数中读取的

Result FirstStageMount::Create() {
    auto fstab = ReadFirstStageFstab(); // 此处读取 fstab
    if (!fstab.ok()) {
        return fstab.error();
    }

    if (IsDtVbmetaCompatible(*fstab)) { // 根据 compatible 创建不同对象
        return std::make_unique(std::move(*fstab));
    } else {
        return std::make_unique(std::move(*fstab));
    }
}

ReadFirstStageFstab

/// @system/core/init/first_stage_mount.cpp
static Result ReadFirstStageFstab() {
    Fstab fstab;
    if (!ReadFstabFromDt(&fstab)) { // 首先读取device tree, 默认值 /proc/device-tree/firmware/android/fstab
        if (ReadDefaultFstab(&fstab)) { // 没有读到,再读默认Fstab
            fstab.erase(std::remove_if(fstab.begin(), fstab.end(),
                                       [](const auto& entry) {
                                           return !entry.fs_mgr_flags.first_stage_mount;
                                       }),
                        fstab.end());
        } else {
            return Error() 

ReadFstabFromDt

/// @system/core/fs_mgr/fs_mgr_fstab.cpp
std::string ReadFstabFromDt() {
    if (!is_dt_compatible() || !IsDtFstabCompatible()) {
        return {};
    }
		// 默认值 /proc/device-tree/firmware/android/fstab
    std::string fstabdir_name = get_android_dt_dir() + "/fstab";
    std::unique_ptr fstabdir(opendir(fstabdir_name.c_str()), closedir);
    if (!fstabdir) return {};

    dirent* dp;
    // Each element in fstab_dt_entries is .
    std::vector fstab_dt_entries;
    while ((dp = readdir(fstabdir.get())) != NULL) { // 读取 fstab 信息
        // skip over name, compatible and .
        if (dp->d_type != DT_DIR || dp->d_name[0] == '.') continue;

        // create        \n
				...
			}
}

ReadDefaultFstab

/// @system/core/fs_mgr/fs_mgr_fstab.cpp
// Loads the fstab file and combines with fstab entries passed in from device tree.
bool ReadDefaultFstab(Fstab* fstab) {
    fstab->clear();
    ReadFstabFromDt(fstab, false /* verbose */); // 重新从 device tree 读取一次 ??

    std::string default_fstab_path;
    // Use different fstab paths for normal boot and recovery boot, respectively
    if (access("/system/bin/recovery", F_OK) == 0) { // recovery模式
        default_fstab_path = "/etc/recovery.fstab";
    } else {  // normal boot
        default_fstab_path = GetFstabPath(); // 获取 fstab 文件路径
    }

    Fstab default_fstab;
		// 从 fstab 文件读取 fstab信息
    if (!default_fstab_path.empty() & ReadFstabFromFile(default_fstab_path, &default_fstab)) {
        for (auto&& entry : default_fstab) {
            fstab->emplace_back(std::move(entry));
        }
    } else {
        LINFO 

看看GetFstabPath实现,决定从哪读取fstab

// Return the path to the fstab file.  There may be multiple fstab files; the
// one that is returned will be the first that exists of fstab.,
// fstab., and fstab..  The fstab is searched for
// in /odm/etc/ and /vendor/etc/, as well as in the locations where it may be in
// the first stage ramdisk during early boot.  Previously, the first stage
// ramdisk's copy of the fstab had to be located in the root directory, but now
// the system/etc directory is supported too and is the preferred location.
std::string GetFstabPath() {
    for (const char* prop : {"fstab_suffix", "hardware", "hardware.platform"}) {
        std::string suffix;
				// 从 ro.boot.(prop值)或 kernel cmdline 等处读取文件名后缀,
				// 从我的模拟器测试获取 ro.boot.hardware 为 ranchu
        if (!fs_mgr_get_boot_config(prop, &suffix)) continue;
				// 遍历访问 prefix + suffix 路径的文件是否存在, 比如 /vendor/etc/fstab.ranchu
        for (const char* prefix : {// late-boot/post-boot locations
                                   "/odm/etc/fstab.", "/vendor/etc/fstab.",
                                   // early boot locations
                                   "/system/etc/fstab.", "/first_stage_ramdisk/system/etc/fstab.",
                                   "/fstab.", "/first_stage_ramdisk/fstab."}) {
            std::string fstab_path = prefix + suffix;
            if (access(fstab_path.c_str(), F_OK) == 0) {
                return fstab_path;
            }
        }
    }

    return "";
}

查看 /vendor/etc/fstab.ranchu , 看其中相关分区信息, 比如 /system、/data

$ cat /vendor/etc/fstab.ranchu
# Android fstab file.
#      
system   /system     ext4    ro,barrier=1     wait,logical,avb=vbmeta,first_stage_mount
vendor   /vendor     ext4    ro,barrier=1     wait,logical,first_stage_mount
product  /product    ext4    ro,barrier=1     wait,logical,first_stage_mount
system_ext  /system_ext  ext4   ro,barrier=1   wait,logical,first_stage_mount
/dev/block/vdc   /data     ext4      noatime,nosuid,nodev,nomblk_io_submit,errors=panic   wait,check,quota,fileencryption=aes-256-xts:aes-256-cts,reservedsize=128M,fsverity,keydirectory=/metadata/vold/metadata_encryption,latemount
/dev/block/pci/pci0000:00/0000:00:06.0/by-name/metadata    /metadata    ext4    noatime,nosuid,nodev    wait,formattable,first_stage_mount
/devices/\*\/block/vdf auto   auto      defaults    voldmanaged=sdcard:auto,encryptable=userdata
dev/block/zram0 none swap  defaults zramsize=75%

SetupSelinux

初始化 selinux 阶段

/// @system/core/init/selinux.cpp
// The SELinux setup process is carefully orchestrated around snapuserd. Policy
// must be loaded off dynamic partitions, and during an OTA, those partitions
// cannot be read without snapuserd. But, with kernel-privileged snapuserd
// running, loading the policy will immediately trigger audits.
//
// We use a five-step process to address this:
//  (1) Read the policy into a string, with snapuserd running.
//  (2) Rewrite the snapshot device-mapper tables, to generate new dm-user
//      devices and to flush I/O.
//  (3) Kill snapuserd, which no longer has any dm-user devices to attach to.
//  (4) Load the sepolicy and issue critical restorecons in /dev, carefully
//      avoiding anything that would read from /system.
//  (5) Re-launch snapuserd and attach it to the dm-user devices from step (2).
//
// After this sequence, it is safe to enable enforcing mode and continue booting.
int SetupSelinux(char** argv) {
    SetStdioToDevNull(argv);
    InitKernelLogging(argv);

    if (REBOOT_BOOTLOADER_ON_PANIC) { // panic 重启到 BootLoader
        InstallRebootSignalHandlers();
    }

    boot_clock::time_point start_time = boot_clock::now();

    MountMissingSystemPartitions();

    SelinuxSetupKernelLogging();

    LOG(INFO) StartTransition();
    }

    LoadSelinuxPolicy(policy); // 加载 selinux policy

    if (snapuserd_helper) { // resume snapused
        // Before enforcing, finish the pending snapuserd transition.
        snapuserd_helper->FinishTransition();
        snapuserd_helper = nullptr;
    }

    SelinuxSetEnforcement(); // 设置 selinux policy 启动状态, 写 /sys/fs/selinux/enforce

    // We're in the kernel domain and want to transition to the init domain.  File systems that
    // store SELabels in their xattrs, such as ext4 do not need an explicit restorecon here,
    // but other file systems do.  In particular, this is needed for ramdisks such as the
    // recovery image for A/B devices.
    if (selinux_android_restorecon("/system/bin/init", 0) == -1) {
        PLOG(FATAL) 

SecondStageMain

第二阶段执行

/// system/core/init/init.cpp
int SecondStageMain(int argc, char** argv) {
    if (REBOOT_BOOTLOADER_ON_PANIC) {
        InstallRebootSignalHandlers();// 设置Signal处理器
    }

    boot_clock::time_point start_time = boot_clock::now();
    // shutdown 处理函数
    trigger_shutdown = [](const std::string& command) { shutdown_state.TriggerShutdown(command); };

    SetStdioToDevNull(argv);
    InitKernelLogging(argv);
    LOG(INFO) 

接下来看一些主要流程

PropertyInit

void PropertyInit() {
    selinux_callback cb;
    cb.func_audit = PropertyAuditCallback;
    selinux_set_callback(SELINUX_CB_AUDIT, cb);

    mkdir("/dev/__properties__", S_IRWXU | S_IXGRP | S_IXOTH);
    CreateSerializedPropertyInfo(); // 创建property se contexts
    if (__system_property_area_init()) { // 将 /dev/__properties__/properties_serial 映射到内存, 创建 ContextNodes,对每个node打开映射
        LOG(FATAL) 

StartPropertyService

启动系统服务,建立与init之间通信socket,以及设置属性监听

/// @system/core/init/property_service.cpp
void StartPropertyService(int* epoll_socket) {
    InitPropertySet("ro.property_service.version", "2");

    int sockets[2];
    // 创建 socket 对,用于init和属性服务间通信
    if (socketpair(AF_UNIX, SOCK_SEQPACKET | SOCK_CLOEXEC, 0, sockets) != 0) {
        PLOG(FATAL) 

LoadBootScripts

加载并解析 init rc 脚本

static void LoadBootScripts(ActionManager& action_manager, ServiceList& service_list) {
    Parser parser = CreateParser(action_manager, service_list);

    std::string bootscript = GetProperty("ro.boot.init_rc", "");
    if (bootscript.empty()) {
        parser.ParseConfig("/system/etc/init/hw/init.rc"); // 首先解析 init.rc
        if (!parser.ParseConfig("/system/etc/init")) { // 解析 /system/etc/init 目录
            late_import_paths.emplace_back("/system/etc/init"); // 解析失败延时解析
        }
        // late_import is available only in Q and earlier release. As we don't
        // have system_ext in those versions, skip late_import for system_ext.
        parser.ParseConfig("/system_ext/etc/init"); // 解析 /system_ext/etc/init 目录
        if (!parser.ParseConfig("/vendor/etc/init")) { // 解析 /vendor/etc/init 目录
            late_import_paths.emplace_back("/vendor/etc/init");
        }
        if (!parser.ParseConfig("/odm/etc/init")) { // 解析 /odm/etc/init 目录
            late_import_paths.emplace_back("/odm/etc/init");
        }
        if (!parser.ParseConfig("/product/etc/init")) { // 解析 /product/etc/init 目录
            late_import_paths.emplace_back("/product/etc/init");
        }
    } else {
        parser.ParseConfig(bootscript);
    }
}

添加内置动作和事件触发器

  • 内置动作(Builtin Action)
    只在代码里面调用QueueBuiltinAction的action,其他action在rc里使用 on 声明。action通常需要一些事件来触发
  • 事件触发器(Trigger)
    调用QueueEventTrigger插入事件触发器
// 添加相关action,会同时添加到 事件队列 和 action队列
am.QueueBuiltinAction(SetupCgroupsAction, "SetupCgroups");
am.QueueBuiltinAction(SetKptrRestrictAction, "SetKptrRestrict");
am.QueueBuiltinAction(TestPerfEventSelinuxAction, "TestPerfEventSelinux");
am.QueueEventTrigger("early-init"); // 触发 early-init

// Queue an action that waits for coldboot done so we know ueventd has set up all of /dev...
am.QueueBuiltinAction(wait_for_coldboot_done_action, "wait_for_coldboot_done");
// ... so that we can start queuing up actions that require stuff from /dev.
am.QueueBuiltinAction(SetMmapRndBitsAction, "SetMmapRndBits");
Keychords keychords;
am.QueueBuiltinAction(...,"KeychordInit");

// Trigger all the boot actions to get us started.
am.QueueEventTrigger("init"); // 触发 init

// Don't mount filesystems or start core system services in charger mode.
std::string bootmode = GetProperty("ro.bootmode", "");
if (bootmode == "charger") { // 充电模式
		am.QueueEventTrigger("charger");
} else { // 正常模式, 触发 late-init
		am.QueueEventTrigger("late-init");
}

// Run all property triggers based on current state of the properties.
// 添加属性触发器。在queue_property_triggers_action中添加的trigger执行之后开始处理属性变化事件, 同时将已匹配的属性事件触发
am.QueueBuiltinAction(queue_property_triggers_action, "queue_property_triggers");

以上操作实际上只是向事件队列和action集合添加,而没有真正的去执行,真正触发执行是在主循环中,通过调用 ActionManager#ExecuteOneCommand。

SecondStageMain 循环处理事件

如下是 init 主循环,负责处理相关事件。

int SecondStageMain(int argc, char** argv) {
...
// Restore prio before main loop
setpriority(PRIO_PROCESS, 0, 0);
while (true) {
    // By default, sleep until something happens. 计算epool超时
    auto epoll_timeout = std::optional{};

    auto shutdown_command = shutdown_state.CheckShutdown();
    if (shutdown_command) { // 处理关机请求
        LOG(INFO) 

内置action和触发器执行

当初次进入会直接调用 ActionManager::ExecuteOneCommand,去执行之前的事件,因此会依次执行

  • 触发SetKptrRestrict, 调用 SetupCgroupsAction

  • 触发SetKptrRestrict, 调用 SetKptrRestrictAction

  • 触发 early-init

    • 启动 ueventd
  • 触发wait_for_coldboot_done,调用 wait_for_coldboot_done_action

    • 等待ro.cold_boot_done=true,即ueventd执行完
  • 触发SetMmapRndBits,调用 SetMmapRndBitsAction

  • 触发KeychordInit

  • 触发init

    • 启动logd、servicemanager、hwservicemanager、vndservicemanager
  • 触发late-init / charger(充电模式下),下面都是在 late-init 情况下触发

    • trigger early-fs
    • trigger fs
    • trigger post-fs
    • trigger late-fs
    • trigger post-fs-data
    • trigger load_bpf_programs
    • trigger zygote-start # 触发启动zygote 框架
    • trigger firmware_mounts_complete
    • trigger early-boot
    • trigger boot
  • 触发queue_property_triggers, 调用queue_property_triggers_action

late-init

# Mount filesystems and start core system services.
on late-init
    trigger early-fs		// 启动 vold

    # Mount fstab in init.{$device}.rc by mount_all command. Optional parameter
    # '--early' can be specified to skip entries with 'latemount'.
    # /system and /vendor must be mounted by the end of the fs stage,
    # while /data is optional.
    trigger fs		// 如 mount_all /vendor/etc/fstab.ranchu --early
    trigger post-fs	 // 创建和挂载一些目录 如 /mnt/user/0 -> /storage

    # Mount fstab in init.{$device}.rc by mount_all with '--late' parameter
    # to only mount entries with 'latemount'. This is needed if '--early' is
    # specified in the previous mount_all command on the fs stage.
    # With /system mounted and properties form /system + /factory available,
    # some services can be started.
    trigger late-fs  // 如 mount_all /vendor/etc/fstab.ranchu --late , /data配置的latemount

    # Now we can mount /data. File encryption requires keymaster to decrypt
    # /data, which in turn can only be loaded when system properties are present.
    trigger post-fs-data	// 挂载 /data,创建一些主要目录

    # Should be before netd, but after apex, properties and logging is available.
    trigger load_bpf_programs

    # Now we can start zygote for devices with file based encryption
    trigger zygote-start	//启动zygote和相关服务

    # Remove a file to wake up anything waiting for firmware.
    trigger firmware_mounts_complete

    trigger early-boot
    trigger boot 	// 启动 hal、core 类别的 service,也就是 native daemons

trigger 会触发调用do_trigger,向事件队列添加相关触发器,之后会依次取出相关事件执行对应的action

/// @system/core/init/builtins.cpp
static Result do_trigger(const BuiltinArguments& args) {
    ActionManager::GetInstance().QueueEventTrigger(args[1]);
    return {};
}

/// system/core/init/action_manager.cpp
void ActionManager::QueueEventTrigger(const std::string& trigger) {
    auto lock = std::lock_guard{event_queue_lock_};
    event_queue_.emplace(trigger);
}

queue_property_triggers

这个触发器是在 late-init 触发器之后加入事件队列的,早于late-init的action中添加的触发器,比如early-fs。该触发器对应的action是queue_property_triggers_action

/// @system/core/init/init.cpp
static Result queue_property_triggers_action(const BuiltinArguments& args) {
		// 添加一个enable_property_trigger,将触发init使能处理属性事件。 从时序来看,将晚于 boot trigger 执行
		// late-init -> queue_property_triggers -> boot -> enable_property_trigger
    ActionManager::GetInstance().QueueBuiltinAction(property_enable_triggers_action, "enable_property_trigger");
    ActionManager::GetInstance().QueueAllPropertyActions(); // 将所有属性满足的action添加到队列
    return {};
}

static Result property_enable_triggers_action(const BuiltinArguments& args) {
    /* Enable property triggers. */
    property_triggers_enabled = 1;
    return {};
}

将所有属性匹配的action添加到队列。

/// @system/core/init/action_manager.cpp
void ActionManager::QueueAllPropertyActions() {
    QueuePropertyChange("", "");
}

// 比如当此时 persist.traced_perf.enable 的值已经为1 ,则会添加相关action到队列,最终会执行 start traced_perf
// init/traced_perf.rc
on property:persist.traced_perf.enable=1
    start traced_perf

trigger zygote-start

zygote-start触发器是用来启动zygote和相关进程的,整个action的执行会依赖加密状态来执行,这些encrypted状态是在执行 mount_all 操作中设置的。可以看到,依次启动了statsd、netd和zygote等进程,zygote的启动会建立系统服务system_server进程的创建。

trigger boot

触发boot事件

on boot
		...
		# Update dm-verity state and set partition.*.verified properties.
		verity_update_state

		# Start standard binderized HAL daemons
		class_start hal	// 启动类别为hal的服务(在rc中使用 class hal定义), 比如vendor.audio-hal

		class_start core	// 启动类别为core的服务,比如 surfaceflinger

总结

init是kernel启动的第一个用户空间进程(pid为1),它在经历FirstStage、selinux_setup和SecondStage后,进入loop循环等待事件发生,比如属性事件或者子进程死亡处理。流程大致如下(正常开机模式):

  • FirstStage 挂载一些基础文件系统和加载内核模块等
  • selinux_setup 执行selinux的初始化
  • SecondStage 挂载其他文件系统,启动属性服务,执行boot流程等,主要逻辑都在这里实现
    • PropertyInit - StartPropertyService 初始化和启动属性服务
    • LoadBootScripts 解析开机脚本 rc文件
    • early-init 早期init阶段,执行启动 ueventd
    • init init阶段,在此阶段会启动logd、servicemanager、hwservicemanager、vndservicemanager
    • late-init 末期init
      • early-fs 启动 vold
      • fs 使用mount_all挂载 init.{$device}.rc 中的fstab相关分区,使用 --early 参数
      • post-fs 创建和挂载一些目录 如 /mnt/user/0 -> /storage
      • late-fs 使用mount_all挂载 init.{$device}.rc 中的fstab相关分区,使用 --late 参数
      • post-fs-data 创建/data一些主要目录
      • load_bpf_programs
      • zygote-start 触发启动zygote 框架
      • firmware_mounts_complete
      • early-boot 在boot之前的一个事件
      • boot 启动核心native服务,如 surfaceflinger、audioserver
    • queue_property_triggers 添加property_triggers,早于early-fs 晚于late-init
      • enable_property_trigger 晚于boot trigger添加,在其之后执行。触发使能init处理属性事件
      • QueueAllPropertyActions 将所有属性匹配的action添加到队列
    • 进入loop循环等待事件发生并处理(主要以下几种)
      • 处理build-in action,在执行结束后被移除(oneshot)
      • 处理唤醒事件
      • 处理属性事件
      • 处理子进程死亡事件
作者:pecuyu 原文地址:https://blog.csdn.net/qq_28261343/article/details/128107048

%s 个评论

要回复文章请先登录注册