init/init.cpp

/*
 * Copyright (C) 2008 The Android Open Source Project
 *
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at
 *
 *      http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */

#include <ctype.h>
#include <dirent.h>
#include <errno.h>
#include <fcntl.h>
#include <inttypes.h>
#include <libgen.h>
#include <paths.h>
#include <signal.h>
#include <stdarg.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <sys/epoll.h>
#include <sys/mount.h>
#include <sys/socket.h>
#include <sys/stat.h>
#include <sys/sysmacros.h>
#include <sys/types.h>
#include <sys/un.h>
#include <sys/wait.h>
#include <unistd.h>

#include <selinux/selinux.h>
#include <selinux/label.h>
#include <selinux/android.h>

#include <android-base/file.h>
#include <android-base/properties.h>
#include <android-base/stringprintf.h>
#include <android-base/strings.h>
#include <android-base/unique_fd.h>
#include <libavb/libavb.h>
#include <private/android_filesystem_config.h>

#include <fstream>
#include <memory>
#include <set>
#include <vector>

#include "action.h"
#include "bootchart.h"
#include "devices.h"
#include "fs_mgr.h"
#include "import_parser.h"
#include "init.h"
#include "init_parser.h"
#include "keychords.h"
#include "log.h"
#include "property_service.h"
#include "service.h"
#include "signal_handler.h"
#include "ueventd.h"
#include "util.h"
#include "watchdogd.h"

using android::base::GetProperty;
using android::base::StringPrintf;

struct selabel_handle *sehandle;
struct selabel_handle *sehandle_prop;

static int property_triggers_enabled = 0;

static char qemu[32];

std::string default_console = "/dev/console";
static time_t process_needs_restart_at;

const char *ENV[32];

static int epoll_fd = -1;

static std::unique_ptr<Timer> waiting_for_prop(nullptr);
static std::string wait_prop_name;
static std::string wait_prop_value;

void register_epoll_handler(int fd, void (*fn)()) {
    epoll_event ev;
    ev.events = EPOLLIN;
    ev.data.ptr = reinterpret_cast<void*>(fn);
    if (epoll_ctl(epoll_fd, EPOLL_CTL_ADD, fd, &ev) == -1) {
        PLOG(ERROR) << "epoll_ctl failed";
    }
}

/* add_environment - add "key=value" to the current environment */
int add_environment(const char *key, const char *val)
{
    size_t n;
    size_t key_len = strlen(key);

    /* The last environment entry is reserved to terminate the list */
    for (n = 0; n < (arraysize(ENV) - 1); n++) {

        /* Delete any existing entry for this key */
        if (ENV[n] != NULL) {
            size_t entry_key_len = strcspn(ENV[n], "=");
            if ((entry_key_len == key_len) && (strncmp(ENV[n], key, entry_key_len) == 0)) {
                free((char*)ENV[n]);
                ENV[n] = NULL;
            }
        }

        /* Add entry if a free slot is available */
        if (ENV[n] == NULL) {
            char* entry;
            asprintf(&entry, "%s=%s", key, val);
            ENV[n] = entry;
            return 0;
        }
    }

    LOG(ERROR) << "No env. room to store: '" << key << "':'" << val << "'";

    return -1;
}

bool start_waiting_for_property(const char *name, const char *value)
{
    if (waiting_for_prop) {
        return false;
    }
    if (GetProperty(name, "") != value) {
        // Current property value is not equal to expected value
        wait_prop_name = name;
        wait_prop_value = value;
        waiting_for_prop.reset(new Timer());
    } else {
        LOG(INFO) << "start_waiting_for_property(\""
                  << name << "\", \"" << value << "\"): already set";
    }
    return true;
}

void property_changed(const char *name, const char *value)
{
    if (property_triggers_enabled)
        ActionManager::GetInstance().QueuePropertyTrigger(name, value);
    if (waiting_for_prop) {
        if (wait_prop_name == name && wait_prop_value == value) {
            wait_prop_name.clear();
            wait_prop_value.clear();
            LOG(INFO) << "Wait for property took " << *waiting_for_prop;
            waiting_for_prop.reset();
        }
    }
}

static void restart_processes()
{
    process_needs_restart_at = 0;
    ServiceManager::GetInstance().ForEachServiceWithFlags(SVC_RESTARTING, [](Service* s) {
        s->RestartIfNeeded(&process_needs_restart_at);
    });
}

void handle_control_message(const std::string& msg, const std::string& name) {
    Service* svc = ServiceManager::GetInstance().FindServiceByName(name);
    if (svc == nullptr) {
        LOG(ERROR) << "no such service '" << name << "'";
        return;
    }

    if (msg == "start") {
        svc->Start();
    } else if (msg == "stop") {
        svc->Stop();
    } else if (msg == "restart") {
        svc->Restart();
    } else {
        LOG(ERROR) << "unknown control msg '" << msg << "'";
    }
}

static int wait_for_coldboot_done_action(const std::vector<std::string>& args) {
    Timer t;

    LOG(VERBOSE) << "Waiting for " COLDBOOT_DONE "...";

    // Historically we had a 1s timeout here because we weren't otherwise
    // tracking boot time, and many OEMs made their sepolicy regular
    // expressions too expensive (http://b/19899875).

    // Now we're tracking boot time, just log the time taken to a system
    // property. We still panic if it takes more than a minute though,
    // because any build that slow isn't likely to boot at all, and we'd
    // rather any test lab devices fail back to the bootloader.
    if (wait_for_file(COLDBOOT_DONE, 60s) < 0) {
        LOG(ERROR) << "Timed out waiting for " COLDBOOT_DONE;
        panic();
    }

    property_set("ro.boottime.init.cold_boot_wait", std::to_string(t.duration_ms()).c_str());
    return 0;
}

/*
 * Writes 512 bytes of output from Hardware RNG (/dev/hw_random, backed
 * by Linux kernel's hw_random framework) into Linux RNG's via /dev/urandom.
 * Does nothing if Hardware RNG is not present.
 *
 * Since we don't yet trust the quality of Hardware RNG, these bytes are not
 * mixed into the primary pool of Linux RNG and the entropy estimate is left
 * unmodified.
 *
 * If the HW RNG device /dev/hw_random is present, we require that at least
 * 512 bytes read from it are written into Linux RNG. QA is expected to catch
 * devices/configurations where these I/O operations are blocking for a long
 * time. We do not reboot or halt on failures, as this is a best-effort
 * attempt.
 */
static int mix_hwrng_into_linux_rng_action(const std::vector<std::string>& args)
{
    int result = -1;
    int hwrandom_fd = -1;
    int urandom_fd = -1;
    char buf[512];
    ssize_t chunk_size;
    size_t total_bytes_written = 0;

    hwrandom_fd = TEMP_FAILURE_RETRY(
            open("/dev/hw_random", O_RDONLY | O_NOFOLLOW | O_CLOEXEC));
    if (hwrandom_fd == -1) {
        if (errno == ENOENT) {
            LOG(ERROR) << "/dev/hw_random not found";
            // It's not an error to not have a Hardware RNG.
            result = 0;
        } else {
            PLOG(ERROR) << "Failed to open /dev/hw_random";
        }
        goto ret;
    }

    urandom_fd = TEMP_FAILURE_RETRY(
            open("/dev/urandom", O_WRONLY | O_NOFOLLOW | O_CLOEXEC));
    if (urandom_fd == -1) {
        PLOG(ERROR) << "Failed to open /dev/urandom";
        goto ret;
    }

    while (total_bytes_written < sizeof(buf)) {
        chunk_size = TEMP_FAILURE_RETRY(
                read(hwrandom_fd, buf, sizeof(buf) - total_bytes_written));
        if (chunk_size == -1) {
            PLOG(ERROR) << "Failed to read from /dev/hw_random";
            goto ret;
        } else if (chunk_size == 0) {
            LOG(ERROR) << "Failed to read from /dev/hw_random: EOF";
            goto ret;
        }

        chunk_size = TEMP_FAILURE_RETRY(write(urandom_fd, buf, chunk_size));
        if (chunk_size == -1) {
            PLOG(ERROR) << "Failed to write to /dev/urandom";
            goto ret;
        }
        total_bytes_written += chunk_size;
    }

    LOG(INFO) << "Mixed " << total_bytes_written << " bytes from /dev/hw_random into /dev/urandom";
    result = 0;

ret:
    if (hwrandom_fd != -1) {
        close(hwrandom_fd);
    }
    if (urandom_fd != -1) {
        close(urandom_fd);
    }
    return result;
}

static void security_failure() {
    LOG(ERROR) << "Security failure...";
    panic();
}

static bool set_highest_available_option_value(std::string path, int min, int max)
{
    std::ifstream inf(path, std::fstream::in);
    if (!inf) {
        LOG(ERROR) << "Cannot open for reading: " << path;
        return false;
    }

    int current = max;
    while (current >= min) {
        // try to write out new value
        std::string str_val = std::to_string(current);
        std::ofstream of(path, std::fstream::out);
        if (!of) {
            LOG(ERROR) << "Cannot open for writing: " << path;
            return false;
        }
        of << str_val << std::endl;
        of.close();

        // check to make sure it was recorded
        inf.seekg(0);
        std::string str_rec;
        inf >> str_rec;
        if (str_val.compare(str_rec) == 0) {
            break;
        }
        current--;
    }
    inf.close();

    if (current < min) {
        LOG(ERROR) << "Unable to set minimum option value " << min << " in " << path;
        return false;
    }
    return true;
}

#define MMAP_RND_PATH "/proc/sys/vm/mmap_rnd_bits"
#define MMAP_RND_COMPAT_PATH "/proc/sys/vm/mmap_rnd_compat_bits"

/* __attribute__((unused)) due to lack of mips support: see mips block
 * in set_mmap_rnd_bits_action */
static bool __attribute__((unused)) set_mmap_rnd_bits_min(int start, int min, bool compat) {
    std::string path;
    if (compat) {
        path = MMAP_RND_COMPAT_PATH;
    } else {
        path = MMAP_RND_PATH;
    }

    return set_highest_available_option_value(path, min, start);
}

/*
 * Set /proc/sys/vm/mmap_rnd_bits and potentially
 * /proc/sys/vm/mmap_rnd_compat_bits to the maximum supported values.
 * Returns -1 if unable to set these to an acceptable value.
 *
 * To support this sysctl, the following upstream commits are needed:
 *
 * d07e22597d1d mm: mmap: add new /proc tunable for mmap_base ASLR
 * e0c25d958f78 arm: mm: support ARCH_MMAP_RND_BITS
 * 8f0d3aa9de57 arm64: mm: support ARCH_MMAP_RND_BITS
 * 9e08f57d684a x86: mm: support ARCH_MMAP_RND_BITS
 * ec9ee4acd97c drivers: char: random: add get_random_long()
 * 5ef11c35ce86 mm: ASLR: use get_random_long()
 */
static int set_mmap_rnd_bits_action(const std::vector<std::string>& args)
{
    int ret = -1;

    /* values are arch-dependent */
#if defined(__aarch64__)
    /* arm64 supports 18 - 33 bits depending on pagesize and VA_SIZE */
    if (set_mmap_rnd_bits_min(33, 24, false)
            && set_mmap_rnd_bits_min(16, 16, true)) {
        ret = 0;
    }
#elif defined(__x86_64__)
    /* x86_64 supports 28 - 32 bits */
    if (set_mmap_rnd_bits_min(32, 32, false)
            && set_mmap_rnd_bits_min(16, 16, true)) {
        ret = 0;
    }
#elif defined(__arm__) || defined(__i386__)
    /* check to see if we're running on 64-bit kernel */
    bool h64 = !access(MMAP_RND_COMPAT_PATH, F_OK);
    /* supported 32-bit architecture must have 16 bits set */
    if (set_mmap_rnd_bits_min(16, 16, h64)) {
        ret = 0;
    }
#elif defined(__mips__) || defined(__mips64__)
    // TODO: add mips support b/27788820
    ret = 0;
#else
    LOG(ERROR) << "Unknown architecture";
#endif

    if (ret == -1) {
        LOG(ERROR) << "Unable to set adequate mmap entropy value!";
        security_failure();
    }
    return ret;
}

#define KPTR_RESTRICT_PATH "/proc/sys/kernel/kptr_restrict"
#define KPTR_RESTRICT_MINVALUE 2
#define KPTR_RESTRICT_MAXVALUE 4

/* Set kptr_restrict to the highest available level.
 *
 * Aborts if unable to set this to an acceptable value.
 */
static int set_kptr_restrict_action(const std::vector<std::string>& args)
{
    std::string path = KPTR_RESTRICT_PATH;

    if (!set_highest_available_option_value(path, KPTR_RESTRICT_MINVALUE, KPTR_RESTRICT_MAXVALUE)) {
        LOG(ERROR) << "Unable to set adequate kptr_restrict value!";
        security_failure();
    }
    return 0;
}

static int keychord_init_action(const std::vector<std::string>& args)
{
    keychord_init();
    return 0;
}

static int console_init_action(const std::vector<std::string>& args)
{
    std::string console = GetProperty("ro.boot.console", "");
    if (!console.empty()) {
        default_console = "/dev/" + console;
    }
    return 0;
}

static void import_kernel_nv(const std::string& key, const std::string& value, bool for_emulator) {
    if (key.empty()) return;

    if (for_emulator) {
        // In the emulator, export any kernel option with the "ro.kernel." prefix.
        property_set(StringPrintf("ro.kernel.%s", key.c_str()).c_str(), value.c_str());
        return;
    }

    if (key == "qemu") {
        strlcpy(qemu, value.c_str(), sizeof(qemu));
    } else if (android::base::StartsWith(key, "androidboot.")) {
        property_set(StringPrintf("ro.boot.%s", key.c_str() + 12).c_str(), value.c_str());
    }
}

static void export_oem_lock_status() {
    if (!android::base::GetBoolProperty("ro.oem_unlock_supported", false)) {
        return;
    }

    std::string value = GetProperty("ro.boot.verifiedbootstate", "");

    if (!value.empty()) {
        property_set("ro.boot.flash.locked", value == "orange" ? "0" : "1");
    }
}

static void export_kernel_boot_props() {
    struct {
        const char *src_prop;
        const char *dst_prop;
        const char *default_value;
    } prop_map[] = {
        { "ro.boot.serialno",   "ro.serialno",   "", },
        { "ro.boot.mode",       "ro.bootmode",   "unknown", },
        { "ro.boot.baseband",   "ro.baseband",   "unknown", },
        { "ro.boot.bootloader", "ro.bootloader", "unknown", },
        { "ro.boot.hardware",   "ro.hardware",   "unknown", },
        { "ro.boot.revision",   "ro.revision",   "0", },
    };
    for (size_t i = 0; i < arraysize(prop_map); i++) {
        std::string value = GetProperty(prop_map[i].src_prop, "");
        property_set(prop_map[i].dst_prop, (!value.empty()) ? value.c_str() : prop_map[i].default_value);
    }
}

static constexpr char android_dt_dir[] = "/proc/device-tree/firmware/android";

static bool is_dt_compatible() {
    std::string dt_value;
    std::string file_name = StringPrintf("%s/compatible", android_dt_dir);

    if (android::base::ReadFileToString(file_name, &dt_value)) {
        // trim the trailing '\0' out, otherwise the comparison
        // will produce false-negatives.
        dt_value.resize(dt_value.size() - 1);
        if (dt_value == "android,firmware") {
            return true;
        }
    }

    return false;
}

static bool is_dt_fstab_compatible() {
    std::string dt_value;
    std::string file_name = StringPrintf("%s/%s/compatible", android_dt_dir, "fstab");

    if (android::base::ReadFileToString(file_name, &dt_value)) {
        dt_value.resize(dt_value.size() - 1);
        if (dt_value == "android,fstab") {
            return true;
        }
    }

    return false;
}

static void process_kernel_dt() {
    if (!is_dt_compatible()) return;

    std::unique_ptr<DIR, int(*)(DIR*)>dir(opendir(android_dt_dir), closedir);
    if (!dir) return;

    std::string dt_file;
    struct dirent *dp;
    while ((dp = readdir(dir.get())) != NULL) {
        if (dp->d_type != DT_REG || !strcmp(dp->d_name, "compatible") || !strcmp(dp->d_name, "name")) {
            continue;
        }

        std::string file_name = StringPrintf("%s/%s", android_dt_dir, dp->d_name);

        android::base::ReadFileToString(file_name, &dt_file);
        std::replace(dt_file.begin(), dt_file.end(), ',', '.');

        std::string property_name = StringPrintf("ro.boot.%s", dp->d_name);
        property_set(property_name.c_str(), dt_file.c_str());
    }
}

static void process_kernel_cmdline() {
    // The first pass does the common stuff, and finds if we are in qemu.
    // The second pass is only necessary for qemu to export all kernel params
    // as properties.
    import_kernel_cmdline(false, import_kernel_nv);
    if (qemu[0]) import_kernel_cmdline(true, import_kernel_nv);
}

static int property_enable_triggers_action(const std::vector<std::string>& args)
{
    /* Enable property triggers. */
    property_triggers_enabled = 1;
    return 0;
}

static int queue_property_triggers_action(const std::vector<std::string>& args)
{
    ActionManager::GetInstance().QueueBuiltinAction(property_enable_triggers_action, "enable_property_trigger");
    ActionManager::GetInstance().QueueAllPropertyTriggers();
    return 0;
}

static void selinux_init_all_handles(void)
{
    sehandle = selinux_android_file_context_handle();
    selinux_android_set_sehandle(sehandle);
    sehandle_prop = selinux_android_prop_context_handle();
}

enum selinux_enforcing_status { SELINUX_PERMISSIVE, SELINUX_ENFORCING };

static selinux_enforcing_status selinux_status_from_cmdline() {
    selinux_enforcing_status status = SELINUX_ENFORCING;

    import_kernel_cmdline(false, [&](const std::string& key, const std::string& value, bool in_qemu) {
        if (key == "androidboot.selinux" && value == "permissive") {
            status = SELINUX_PERMISSIVE;
        }
    });

    return status;
}

static bool selinux_is_enforcing(void)
{
    if (ALLOW_PERMISSIVE_SELINUX) {
        return selinux_status_from_cmdline() == SELINUX_ENFORCING;
    }
    return true;
}

static int audit_callback(void *data, security_class_t /*cls*/, char *buf, size_t len) {

    property_audit_data *d = reinterpret_cast<property_audit_data*>(data);

    if (!d || !d->name || !d->cr) {
        LOG(ERROR) << "audit_callback invoked with null data arguments!";
        return 0;
    }

    snprintf(buf, len, "property=%s pid=%d uid=%d gid=%d", d->name,
            d->cr->pid, d->cr->uid, d->cr->gid);
    return 0;
}

/*
 * Forks, executes the provided program in the child, and waits for the completion in the parent.
 * Child's stderr is captured and logged using LOG(ERROR).
 *
 * Returns true if the child exited with status code 0, returns false otherwise.
 */
static bool fork_execve_and_wait_for_completion(const char* filename, char* const argv[],
                                                char* const envp[]) {
    // Create a pipe used for redirecting child process's output.
    // * pipe_fds[0] is the FD the parent will use for reading.
    // * pipe_fds[1] is the FD the child will use for writing.
    int pipe_fds[2];
    if (pipe(pipe_fds) == -1) {
        PLOG(ERROR) << "Failed to create pipe";
        return false;
    }

    pid_t child_pid = fork();
    if (child_pid == -1) {
        PLOG(ERROR) << "Failed to fork for " << filename;
        return false;
    }

    if (child_pid == 0) {
        // fork succeeded -- this is executing in the child process

        // Close the pipe FD not used by this process
        TEMP_FAILURE_RETRY(close(pipe_fds[0]));

        // Redirect stderr to the pipe FD provided by the parent
        if (TEMP_FAILURE_RETRY(dup2(pipe_fds[1], STDERR_FILENO)) == -1) {
            PLOG(ERROR) << "Failed to redirect stderr of " << filename;
            _exit(127);
            return false;
        }
        TEMP_FAILURE_RETRY(close(pipe_fds[1]));

        if (execve(filename, argv, envp) == -1) {
            PLOG(ERROR) << "Failed to execve " << filename;
            return false;
        }
        // Unreachable because execve will have succeeded and replaced this code
        // with child process's code.
        _exit(127);
        return false;
    } else {
        // fork succeeded -- this is executing in the original/parent process

        // Close the pipe FD not used by this process
        TEMP_FAILURE_RETRY(close(pipe_fds[1]));

        // Log the redirected output of the child process.
        // It's unfortunate that there's no standard way to obtain an istream for a file descriptor.
        // As a result, we're buffering all output and logging it in one go at the end of the
        // invocation, instead of logging it as it comes in.
        const int child_out_fd = pipe_fds[0];
        std::string child_output;
        if (!android::base::ReadFdToString(child_out_fd, &child_output)) {
            PLOG(ERROR) << "Failed to capture full output of " << filename;
        }
        TEMP_FAILURE_RETRY(close(child_out_fd));
        if (!child_output.empty()) {
            // Log captured output, line by line, because LOG expects to be invoked for each line
            std::istringstream in(child_output);
            std::string line;
            while (std::getline(in, line)) {
                LOG(ERROR) << filename << ": " << line;
            }
        }

        // Wait for child to terminate
        int status;
        if (TEMP_FAILURE_RETRY(waitpid(child_pid, &status, 0)) != child_pid) {
            PLOG(ERROR) << "Failed to wait for " << filename;
            return false;
        }

        if (WIFEXITED(status)) {
            int status_code = WEXITSTATUS(status);
            if (status_code == 0) {
                return true;
            } else {
                LOG(ERROR) << filename << " exited with status " << status_code;
            }
        } else if (WIFSIGNALED(status)) {
            LOG(ERROR) << filename << " killed by signal " << WTERMSIG(status);
        } else if (WIFSTOPPED(status)) {
            LOG(ERROR) << filename << " stopped by signal " << WSTOPSIG(status);
        } else {
            LOG(ERROR) << "waitpid for " << filename << " returned unexpected status: " << status;
        }

        return false;
    }
}

static bool read_first_line(const char* file, std::string* line) {
    line->clear();

    std::string contents;
    if (!android::base::ReadFileToString(file, &contents, true /* follow symlinks */)) {
        return false;
    }
    std::istringstream in(contents);
    std::getline(in, *line);
    return true;
}

static bool selinux_find_precompiled_split_policy(std::string* file) {
    file->clear();

    static constexpr const char precompiled_sepolicy[] = "/vendor/etc/selinux/precompiled_sepolicy";
    if (access(precompiled_sepolicy, R_OK) == -1) {
        return false;
    }
    std::string actual_plat_id;
    if (!read_first_line("/system/etc/selinux/plat_and_mapping_sepolicy.cil.sha256",
                         &actual_plat_id)) {
        PLOG(INFO) << "Failed to read "
                      "/system/etc/selinux/plat_and_mapping_sepolicy.cil.sha256";
        return false;
    }
    std::string precompiled_plat_id;
    if (!read_first_line("/vendor/etc/selinux/precompiled_sepolicy.plat_and_mapping.sha256",
                         &precompiled_plat_id)) {
        PLOG(INFO) << "Failed to read "
                      "/vendor/etc/selinux/"
                      "precompiled_sepolicy.plat_and_mapping.sha256";
        return false;
    }
    if ((actual_plat_id.empty()) || (actual_plat_id != precompiled_plat_id)) {
        return false;
    }

    *file = precompiled_sepolicy;
    return true;
}

static constexpr const char plat_policy_cil_file[] = "/system/etc/selinux/plat_sepolicy.cil";

static bool selinux_is_split_policy_device() { return access(plat_policy_cil_file, R_OK) != -1; }

/*
 * Loads SELinux policy split across platform/system and non-platform/vendor files.
 *
 * Returns true upon success, false otherwise (failure cause is logged).
 */
static bool selinux_load_split_policy() {
    // IMPLEMENTATION NOTE: Split policy consists of three CIL files:
    // * platform -- policy needed due to logic contained in the system image,
    // * non-platform -- policy needed due to logic contained in the vendor image,
    // * mapping -- mapping policy which helps preserve forward-compatibility of non-platform policy
    //   with newer versions of platform policy.
    //
    // secilc is invoked to compile the above three policy files into a single monolithic policy
    // file. This file is then loaded into the kernel.

    // Load precompiled policy from vendor image, if a matching policy is found there. The policy
    // must match the platform policy on the system image.
    std::string precompiled_sepolicy_file;
    if (selinux_find_precompiled_split_policy(&precompiled_sepolicy_file)) {
        android::base::unique_fd fd(
            open(precompiled_sepolicy_file.c_str(), O_RDONLY | O_CLOEXEC | O_BINARY));
        if (fd != -1) {
            if (selinux_android_load_policy_from_fd(fd, precompiled_sepolicy_file.c_str()) < 0) {
                LOG(ERROR) << "Failed to load SELinux policy from " << precompiled_sepolicy_file;
                return false;
            }
            return true;
        }
    }
    // No suitable precompiled policy could be loaded

    LOG(INFO) << "Compiling SELinux policy";

    // Determine the highest policy language version supported by the kernel
    set_selinuxmnt("/sys/fs/selinux");
    int max_policy_version = security_policyvers();
    if (max_policy_version == -1) {
        PLOG(ERROR) << "Failed to determine highest policy version supported by kernel";
        return false;
    }

    // We store the output of the compilation on /dev because this is the most convenient tmpfs
    // storage mount available this early in the boot sequence.
    char compiled_sepolicy[] = "/dev/sepolicy.XXXXXX";
    android::base::unique_fd compiled_sepolicy_fd(mkostemp(compiled_sepolicy, O_CLOEXEC));
    if (compiled_sepolicy_fd < 0) {
        PLOG(ERROR) << "Failed to create temporary file " << compiled_sepolicy;
        return false;
    }

    // clang-format off
    const char* compile_args[] = {
        "/system/bin/secilc",
        plat_policy_cil_file,
        "-M", "true",
        // Target the highest policy language version supported by the kernel
        "-c", std::to_string(max_policy_version).c_str(),
        "/system/etc/selinux/mapping_sepolicy.cil",
        "/vendor/etc/selinux/nonplat_sepolicy.cil",
        "-o", compiled_sepolicy,
        // We don't care about file_contexts output by the compiler
        "-f", "/sys/fs/selinux/null",  // /dev/null is not yet available
        nullptr};
    // clang-format on

    if (!fork_execve_and_wait_for_completion(compile_args[0], (char**)compile_args, (char**)ENV)) {
        unlink(compiled_sepolicy);
        return false;
    }
    unlink(compiled_sepolicy);

    LOG(INFO) << "Loading compiled SELinux policy";
    if (selinux_android_load_policy_from_fd(compiled_sepolicy_fd, compiled_sepolicy) < 0) {
        LOG(ERROR) << "Failed to load SELinux policy from " << compiled_sepolicy;
        return false;
    }

    return true;
}

/*
 * Loads SELinux policy from a monolithic file.
 *
 * Returns true upon success, false otherwise (failure cause is logged).
 */
static bool selinux_load_monolithic_policy() {
    LOG(VERBOSE) << "Loading SELinux policy from monolithic file";
    if (selinux_android_load_policy() < 0) {
        PLOG(ERROR) << "Failed to load monolithic SELinux policy";
        return false;
    }
    return true;
}

/*
 * Loads SELinux policy into the kernel.
 *
 * Returns true upon success, false otherwise (failure cause is logged).
 */
static bool selinux_load_policy() {
    return selinux_is_split_policy_device() ? selinux_load_split_policy()
                                            : selinux_load_monolithic_policy();
}

static void selinux_initialize(bool in_kernel_domain) {
    Timer t;

    selinux_callback cb;
    cb.func_log = selinux_klog_callback;
    selinux_set_callback(SELINUX_CB_LOG, cb);
    cb.func_audit = audit_callback;
    selinux_set_callback(SELINUX_CB_AUDIT, cb);

    if (in_kernel_domain) {
        LOG(INFO) << "Loading SELinux policy";
        if (!selinux_load_policy()) {
            panic();
        }

        bool kernel_enforcing = (security_getenforce() == 1);
        bool is_enforcing = selinux_is_enforcing();
        if (kernel_enforcing != is_enforcing) {
            if (security_setenforce(is_enforcing)) {
                PLOG(ERROR) << "security_setenforce(%s) failed" << (is_enforcing ? "true" : "false");
                security_failure();
            }
        }

        if (!write_file("/sys/fs/selinux/checkreqprot", "0")) {
            security_failure();
        }

        // init's first stage can't set properties, so pass the time to the second stage.
        setenv("INIT_SELINUX_TOOK", std::to_string(t.duration_ms()).c_str(), 1);
    } else {
        selinux_init_all_handles();
    }
}

// The files and directories that were created before initial sepolicy load
// need to have their security context restored to the proper value.
// This must happen before /dev is populated by ueventd.
static void selinux_restore_context() {
    LOG(INFO) << "Running restorecon...";
    restorecon("/dev");
    restorecon("/dev/kmsg");
    restorecon("/dev/socket");
    restorecon("/dev/random");
    restorecon("/dev/urandom");
    restorecon("/dev/__properties__");

    restorecon("/file_contexts.bin");
    restorecon("/plat_file_contexts");
    restorecon("/nonplat_file_contexts");
    restorecon("/plat_property_contexts");
    restorecon("/nonplat_property_contexts");
    restorecon("/plat_seapp_contexts");
    restorecon("/nonplat_seapp_contexts");
    restorecon("/plat_service_contexts");
    restorecon("/nonplat_service_contexts");
    restorecon("/sepolicy");
    restorecon("/vndservice_contexts");

    restorecon("/sys", SELINUX_ANDROID_RESTORECON_RECURSE);
    restorecon("/dev/block", SELINUX_ANDROID_RESTORECON_RECURSE);
    restorecon("/dev/device-mapper");
}

// Set the UDC controller for the ConfigFS USB Gadgets.
// Read the UDC controller in use from "/sys/class/udc".
// In case of multiple UDC controllers select the first one.
static void set_usb_controller() {
    std::unique_ptr<DIR, decltype(&closedir)>dir(opendir("/sys/class/udc"), closedir);
    if (!dir) return;

    dirent* dp;
    while ((dp = readdir(dir.get())) != nullptr) {
        if (dp->d_name[0] == '.') continue;

        property_set("sys.usb.controller", dp->d_name);
        break;
    }
}

static bool early_mount_one(struct fstab_rec* rec) {
    if (rec && fs_mgr_is_verified(rec)) {
        // setup verity and create the dm-XX block device
        // needed to mount this partition
        int ret = fs_mgr_setup_verity(rec, false);
        if (ret == FS_MGR_SETUP_VERITY_FAIL) {
            PLOG(ERROR) << "early_mount: Failed to setup verity for '" << rec->mount_point << "'";
            return false;
        }

        // The exact block device name is added as a mount source by
        // fs_mgr_setup_verity() in ->blk_device as "/dev/block/dm-XX"
        // We create that device by running coldboot on /sys/block/dm-XX
        std::string dm_device(basename(rec->blk_device));
        std::string syspath = StringPrintf("/sys/block/%s", dm_device.c_str());
        device_init(syspath.c_str(), [&](uevent* uevent) -> coldboot_action_t {
            if (uevent->device_name && !strcmp(dm_device.c_str(), uevent->device_name)) {
                LOG(VERBOSE) << "early_mount: creating dm-verity device : " << dm_device;
                return COLDBOOT_STOP;
            }
            return COLDBOOT_CONTINUE;
        });
    }

    if (rec && fs_mgr_do_mount_one(rec)) {
        PLOG(ERROR) << "early_mount: Failed to mount '" << rec->mount_point << "'";
        return false;
    }

    return true;
}

// Creates devices with uevent->partition_name matching one in the in/out
// partition_names. Note that the partition_names MUST have A/B suffix
// when A/B is used. Found partitions will then be removed from the
// partition_names for caller to check which devices are NOT created.
static void early_device_init(std::set<std::string>* partition_names) {
    if (partition_names->empty()) {
        return;
    }
    device_init(nullptr, [=](uevent* uevent) -> coldboot_action_t {
        if (!strncmp(uevent->subsystem, "firmware", 8)) {
            return COLDBOOT_CONTINUE;
        }

        // we need platform devices to create symlinks
        if (!strncmp(uevent->subsystem, "platform", 8)) {
            return COLDBOOT_CREATE;
        }

        // Ignore everything that is not a block device
        if (strncmp(uevent->subsystem, "block", 5)) {
            return COLDBOOT_CONTINUE;
        }

        if (uevent->partition_name) {
            // match partition names to create device nodes for partitions
            // both partition_names and uevent->partition_name have A/B suffix when A/B is used
            auto iter = partition_names->find(uevent->partition_name);
            if (iter != partition_names->end()) {
                LOG(VERBOSE) << "early_mount: found partition: " << *iter;
                partition_names->erase(iter);
                if (partition_names->empty()) {
                    return COLDBOOT_STOP;  // found all partitions, stop coldboot
                } else {
                    return COLDBOOT_CREATE;  // create this device and continue to find others
                }
            }
        }
        // Not found a partition or find an unneeded partition, continue to find others
        return COLDBOOT_CONTINUE;
    });
}

static bool get_early_partitions(const std::vector<fstab_rec*>& early_fstab_recs,
                                 std::set<std::string>* out_partitions, bool* out_need_verity) {
    std::string meta_partition;
    out_partitions->clear();
    *out_need_verity = false;

    for (auto fstab_rec : early_fstab_recs) {
        // don't allow verifyatboot for early mounted partitions
        if (fs_mgr_is_verifyatboot(fstab_rec)) {
            LOG(ERROR) << "early_mount: partitions can't be verified at boot";
            return false;
        }
        // check for verified partitions
        if (fs_mgr_is_verified(fstab_rec)) {
            *out_need_verity = true;
        }
        // check if verity metadata is on a separate partition and get partition
        // name from the end of the ->verity_loc path. verity state is not partition
        // specific, so there must be only 1 additional partition that carries
        // verity state.
        if (fstab_rec->verity_loc) {
            if (!meta_partition.empty()) {
                LOG(ERROR) << "early_mount: more than one meta partition found: " << meta_partition
                           << ", " << basename(fstab_rec->verity_loc);
                return false;
            } else {
                meta_partition = basename(fstab_rec->verity_loc);
            }
        }
    }

    // includes those early mount partitions and meta_partition (if any)
    // note that fstab_rec->blk_device has A/B suffix updated by fs_mgr when A/B is used
    for (auto fstab_rec : early_fstab_recs) {
        out_partitions->emplace(basename(fstab_rec->blk_device));
    }

    if (!meta_partition.empty()) {
        out_partitions->emplace(std::move(meta_partition));
    }

    return true;
}

/* Early mount vendor and ODM partitions. The fstab is read from device-tree. */
static bool early_mount() {
    // skip early mount if we're in recovery mode
    if (access("/sbin/recovery", F_OK) == 0) {
        LOG(INFO) << "Early mount skipped (recovery mode)";
        return true;
    }

    // first check if device tree fstab entries are compatible
    if (!is_dt_fstab_compatible()) {
        LOG(INFO) << "Early mount skipped (missing/incompatible fstab in device tree)";
        return true;
    }

    std::unique_ptr<fstab, decltype(&fs_mgr_free_fstab)> tab(
        fs_mgr_read_fstab_dt(), fs_mgr_free_fstab);
    if (!tab) {
        LOG(ERROR) << "Early mount failed to read fstab from device tree";
        return false;
    }

    // find out fstab records for odm, system and vendor
    std::vector<fstab_rec*> early_fstab_recs;
    for (auto mount_point : {"/odm", "/system", "/vendor"}) {
        fstab_rec* fstab_rec = fs_mgr_get_entry_for_mount_point(tab.get(), mount_point);
        if (fstab_rec != nullptr) {
            early_fstab_recs.push_back(fstab_rec);
        }
    }

    // nothing to early mount
    if (early_fstab_recs.empty()) return true;

    bool need_verity;
    std::set<std::string> partition_names;
    // partition_names MUST have A/B suffix when A/B is used
    if (!get_early_partitions(early_fstab_recs, &partition_names, &need_verity)) {
        return false;
    }

    bool success = false;
    // create the devices we need..
    early_device_init(&partition_names);

    // early_device_init will remove found partitions from partition_names
    // So if the partition_names is not empty here, means some partitions
    // are not found
    if (!partition_names.empty()) {
        LOG(ERROR) << "early_mount: partition(s) not found: "
                   << android::base::Join(partition_names, ", ");
        goto done;
    }

    if (need_verity) {
        // create /dev/device mapper
        device_init("/sys/devices/virtual/misc/device-mapper",
                    [&](uevent* uevent) -> coldboot_action_t { return COLDBOOT_STOP; });
    }

    for (auto fstab_rec : early_fstab_recs) {
        if (!early_mount_one(fstab_rec)) goto done;
    }
    success = true;

done:
    device_close();
    return success;
}

static void install_reboot_signal_handlers() {
    // 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) {
        // panic() reboots to bootloader
        panic();
    };
    action.sa_flags = SA_RESTART;
    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);
}

int main(int argc, char** argv) {
    if (!strcmp(basename(argv[0]), "ueventd")) {
        return ueventd_main(argc, argv);
    }

    if (!strcmp(basename(argv[0]), "watchdogd")) {
        return watchdogd_main(argc, argv);
    }

    if (REBOOT_BOOTLOADER_ON_PANIC) {
        install_reboot_signal_handlers();
    }

    add_environment("PATH", _PATH_DEFPATH);

    bool is_first_stage = (getenv("INIT_SECOND_STAGE") == nullptr);

    if (is_first_stage) {
        boot_clock::time_point start_time = boot_clock::now();

        // Clear the umask.
        umask(0);

        // 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.
        mount("tmpfs", "/dev", "tmpfs", MS_NOSUID, "mode=0755");
        mkdir("/dev/pts", 0755);
        mkdir("/dev/socket", 0755);
        mount("devpts", "/dev/pts", "devpts", 0, NULL);
        #define MAKE_STR(x) __STRING(x)
        mount("proc", "/proc", "proc", 0, "hidepid=2,gid=" MAKE_STR(AID_READPROC));
        // Don't expose the raw commandline to unprivileged processes.
        chmod("/proc/cmdline", 0440);
        gid_t groups[] = { AID_READPROC };
        setgroups(arraysize(groups), groups);
        mount("sysfs", "/sys", "sysfs", 0, NULL);
        mount("selinuxfs", "/sys/fs/selinux", "selinuxfs", 0, NULL);
        mknod("/dev/kmsg", S_IFCHR | 0600, makedev(1, 11));
        mknod("/dev/random", S_IFCHR | 0666, makedev(1, 8));
        mknod("/dev/urandom", S_IFCHR | 0666, makedev(1, 9));

        // Now that tmpfs is mounted on /dev and we have /dev/kmsg, we can actually
        // talk to the outside world...
        InitKernelLogging(argv);

        LOG(INFO) << "init first stage started!";

        if (!early_mount()) {
            LOG(ERROR) << "Failed to mount required partitions early ...";
            panic();
        }

        // Set up SELinux, loading the SELinux policy.
        selinux_initialize(true);

        // We're in the kernel domain, so re-exec init to transition to the init domain now
        // that the SELinux policy has been loaded.
        if (restorecon("/init") == -1) {
            PLOG(ERROR) << "restorecon failed";
            security_failure();
        }

        setenv("INIT_SECOND_STAGE", "true", 1);

        static constexpr uint32_t kNanosecondsPerMillisecond = 1e6;
        uint64_t start_ms = start_time.time_since_epoch().count() / kNanosecondsPerMillisecond;
        setenv("INIT_STARTED_AT", StringPrintf("%" PRIu64, start_ms).c_str(), 1);

        char* path = argv[0];
        char* args[] = { path, nullptr };
        execv(path, args);

        // execv() only returns if an error happened, in which case we
        // panic and never fall through this conditional.
        PLOG(ERROR) << "execv(\"" << path << "\") failed";
        security_failure();
    }

    // At this point we're in the second stage of init.
    InitKernelLogging(argv);
    LOG(INFO) << "init second stage started!";

    // Indicate that booting is in progress to background fw loaders, etc.
    close(open("/dev/.booting", O_WRONLY | O_CREAT | O_CLOEXEC, 0000));

    property_init();

    // If arguments are passed both on the command line and in DT,
    // properties set in DT always have priority over the command-line ones.
    process_kernel_dt();
    process_kernel_cmdline();

    // Propagate the kernel variables to internal variables
    // used by init as well as the current required properties.
    export_kernel_boot_props();

    // Make the time that init started available for bootstat to log.
    property_set("ro.boottime.init", getenv("INIT_STARTED_AT"));
    property_set("ro.boottime.init.selinux", getenv("INIT_SELINUX_TOOK"));

    // Set libavb version for Framework-only OTA match in Treble build.
    property_set("ro.boot.init.avb_version", std::to_string(AVB_MAJOR_VERSION).c_str());

    // Clean up our environment.
    unsetenv("INIT_SECOND_STAGE");
    unsetenv("INIT_STARTED_AT");
    unsetenv("INIT_SELINUX_TOOK");

    // Now set up SELinux for second stage.
    selinux_initialize(false);
    selinux_restore_context();

    epoll_fd = epoll_create1(EPOLL_CLOEXEC);
    if (epoll_fd == -1) {
        PLOG(ERROR) << "epoll_create1 failed";
        exit(1);
    }

    signal_handler_init();

    property_load_boot_defaults();
    export_oem_lock_status();
    start_property_service();
    set_usb_controller();

    const BuiltinFunctionMap function_map;
    Action::set_function_map(&function_map);

    Parser& parser = Parser::GetInstance();
    parser.AddSectionParser("service",std::make_unique<ServiceParser>());
    parser.AddSectionParser("on", std::make_unique<ActionParser>());
    parser.AddSectionParser("import", std::make_unique<ImportParser>());
    std::string bootscript = GetProperty("ro.boot.init_rc", "");
    if (bootscript.empty()) {
        parser.ParseConfig("/init.rc");
        parser.set_is_system_etc_init_loaded(
                parser.ParseConfig("/system/etc/init"));
        parser.set_is_vendor_etc_init_loaded(
                parser.ParseConfig("/vendor/etc/init"));
        parser.set_is_odm_etc_init_loaded(parser.ParseConfig("/odm/etc/init"));
    } else {
        parser.ParseConfig(bootscript);
        parser.set_is_system_etc_init_loaded(true);
        parser.set_is_vendor_etc_init_loaded(true);
        parser.set_is_odm_etc_init_loaded(true);
    }

    // Turning this on and letting the INFO logging be discarded adds 0.2s to
    // Nexus 9 boot time, so it's disabled by default.
    if (false) parser.DumpState();

    ActionManager& am = ActionManager::GetInstance();

    am.QueueEventTrigger("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(mix_hwrng_into_linux_rng_action, "mix_hwrng_into_linux_rng");
    am.QueueBuiltinAction(set_mmap_rnd_bits_action, "set_mmap_rnd_bits");
    am.QueueBuiltinAction(set_kptr_restrict_action, "set_kptr_restrict");
    am.QueueBuiltinAction(keychord_init_action, "keychord_init");
    am.QueueBuiltinAction(console_init_action, "console_init");

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

    // Repeat mix_hwrng_into_linux_rng in case /dev/hw_random or /dev/random
    // wasn't ready immediately after wait_for_coldboot_done
    am.QueueBuiltinAction(mix_hwrng_into_linux_rng_action, "mix_hwrng_into_linux_rng");

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

    // Run all property triggers based on current state of the properties.
    am.QueueBuiltinAction(queue_property_triggers_action, "queue_property_triggers");

    while (true) {
        // By default, sleep until something happens.
        int epoll_timeout_ms = -1;

        if (!(waiting_for_prop || ServiceManager::GetInstance().IsWaitingForExec())) {
            am.ExecuteOneCommand();
        }
        if (!(waiting_for_prop || ServiceManager::GetInstance().IsWaitingForExec())) {
            restart_processes();

            // If there's a process that needs restarting, wake up in time for that.
            if (process_needs_restart_at != 0) {
                epoll_timeout_ms = (process_needs_restart_at - time(nullptr)) * 1000;
                if (epoll_timeout_ms < 0) epoll_timeout_ms = 0;
            }

            // If there's more work to do, wake up again immediately.
            if (am.HasMoreCommands()) epoll_timeout_ms = 0;
        }

        epoll_event ev;
        int nr = TEMP_FAILURE_RETRY(epoll_wait(epoll_fd, &ev, 1, epoll_timeout_ms));
        if (nr == -1) {
            PLOG(ERROR) << "epoll_wait failed";
        } else if (nr == 1) {
            ((void (*)()) ev.data.ptr)();
        }
    }

    return 0;
}