X-Git-Url: http://www.chiark.greenend.org.uk/ucgi/~ianmdlvl/git?p=elogind.git;a=blobdiff_plain;f=src%2Fcore%2Fmain.c;h=bd7fc468b4109921dbfa3b3b41b143d11004ef72;hp=be080d6737f748c963637c0b409714768924127a;hb=e62d8c394474;hpb=c93ff2e913a6da516c83c0d3f8df73004acc6c8b diff --git a/src/core/main.c b/src/core/main.c index be080d673..bd7fc468b 100644 --- a/src/core/main.c +++ b/src/core/main.c @@ -50,36 +50,43 @@ #include "watchdog.h" #include "path-util.h" #include "switch-root.h" +#include "capability.h" +#include "killall.h" +#include "env-util.h" +#include "hwclock.h" +#include "sd-daemon.h" #include "mount-setup.h" #include "loopback-setup.h" +#ifdef HAVE_KMOD #include "kmod-setup.h" +#endif #include "hostname-setup.h" #include "machine-id-setup.h" #include "locale-setup.h" -#include "hwclock.h" #include "selinux-setup.h" #include "ima-setup.h" +#include "fileio.h" +#include "smack-setup.h" static enum { ACTION_RUN, ACTION_HELP, + ACTION_VERSION, ACTION_TEST, ACTION_DUMP_CONFIGURATION_ITEMS, ACTION_DONE } arg_action = ACTION_RUN; static char *arg_default_unit = NULL; -static ManagerRunningAs arg_running_as = _MANAGER_RUNNING_AS_INVALID; +static SystemdRunningAs arg_running_as = _SYSTEMD_RUNNING_AS_INVALID; static bool arg_dump_core = true; static bool arg_crash_shell = false; static int arg_crash_chvt = -1; static bool arg_confirm_spawn = false; static bool arg_show_status = true; -#ifdef HAVE_SYSV_COMPAT -static bool arg_sysv_console = true; -#endif +static bool arg_switched_root = false; static char **arg_default_controllers = NULL; static char ***arg_join_controllers = NULL; static ExecOutput arg_default_std_output = EXEC_OUTPUT_JOURNAL; @@ -87,6 +94,8 @@ static ExecOutput arg_default_std_error = EXEC_OUTPUT_INHERIT; static usec_t arg_runtime_watchdog = 0; static usec_t arg_shutdown_watchdog = 10 * USEC_PER_MINUTE; static struct rlimit *arg_default_rlimit[RLIMIT_NLIMITS] = {}; +static uint64_t arg_capability_bounding_set_drop = 0; +static nsec_t arg_timer_slack_nsec = (nsec_t) -1; static FILE* serialization = NULL; @@ -107,8 +116,10 @@ _noreturn_ static void crash(int sig) { sa.sa_flags = SA_NOCLDSTOP|SA_RESTART; assert_se(sigaction(SIGCHLD, &sa, NULL) == 0); - if ((pid = fork()) < 0) - log_error("Caught <%s>, cannot fork for core dump: %s", signal_to_string(sig), strerror(errno)); + pid = fork(); + if (pid < 0) + log_error("Caught <%s>, cannot fork for core dump: %s", + signal_to_string(sig), strerror(errno)); else if (pid == 0) { struct rlimit rl; @@ -138,12 +149,17 @@ _noreturn_ static void crash(int sig) { int r; /* Order things nicely. */ - if ((r = wait_for_terminate(pid, &status)) < 0) - log_error("Caught <%s>, waitpid() failed: %s", signal_to_string(sig), strerror(-r)); + r = wait_for_terminate(pid, &status); + if (r < 0) + log_error("Caught <%s>, waitpid() failed: %s", + signal_to_string(sig), strerror(-r)); else if (status.si_code != CLD_DUMPED) - log_error("Caught <%s>, core dump failed.", signal_to_string(sig)); + log_error("Caught <%s>, core dump failed.", + signal_to_string(sig)); else - log_error("Caught <%s>, dumped core as pid %lu.", signal_to_string(sig), (unsigned long) pid); + log_error("Caught <%s>, dumped core as pid %lu.", + signal_to_string(sig), + (unsigned long) pid); } } @@ -163,23 +179,19 @@ _noreturn_ static void crash(int sig) { sa.sa_flags = SA_NOCLDSTOP|SA_NOCLDWAIT|SA_RESTART; assert_se(sigaction(SIGCHLD, &sa, NULL) == 0); - if ((pid = fork()) < 0) - log_error("Failed to fork off crash shell: %s", strerror(errno)); + pid = fork(); + if (pid < 0) + log_error("Failed to fork off crash shell: %m"); else if (pid == 0) { - int fd, r; - - if ((fd = acquire_terminal("/dev/console", false, true, true)) < 0) - log_error("Failed to acquire terminal: %s", strerror(-fd)); - else if ((r = make_stdio(fd)) < 0) - log_error("Failed to duplicate terminal fd: %s", strerror(-r)); - + make_console_stdio(); execl("/bin/sh", "/bin/sh", NULL); - log_error("execl() failed: %s", strerror(errno)); + log_error("execl() failed: %m"); _exit(1); } - log_info("Successfully spawned crash shell as pid %lu.", (unsigned long) pid); + log_info("Successfully spawned crash shell as pid %lu.", + (unsigned long) pid); } log_info("Freezing execution."); @@ -229,11 +241,13 @@ static int set_default_unit(const char *u) { assert(u); - if (!(c = strdup(u))) + c = strdup(u); + if (!c) return -ENOMEM; free(arg_default_unit); arg_default_unit = c; + return 0; } @@ -255,10 +269,17 @@ static int parse_proc_cmdline_word(const char *word) { assert(word); - if (startswith(word, "systemd.unit=")) - return set_default_unit(word + 13); + if (startswith(word, "systemd.unit=")) { + + if (!in_initrd()) + return set_default_unit(word + 13); + + } else if (startswith(word, "rd.systemd.unit=")) { - else if (startswith(word, "systemd.log_target=")) { + if (in_initrd()) + return set_default_unit(word + 16); + + } else if (startswith(word, "systemd.log_target=")) { if (log_set_target_from_string(word + 19) < 0) log_warning("Failed to parse log target %s. Ignoring.", word + 19); @@ -332,7 +353,8 @@ static int parse_proc_cmdline_word(const char *word) { else arg_default_std_error = r; } else if (startswith(word, "systemd.setenv=")) { - char *cenv, *eq; + _cleanup_free_ char *cenv = NULL; + char *eq; int r; cenv = strdup(word + 15); @@ -341,56 +363,62 @@ static int parse_proc_cmdline_word(const char *word) { eq = strchr(cenv, '='); if (!eq) { - r = unsetenv(cenv); - if (r < 0) - log_warning("unsetenv failed %s. Ignoring.", strerror(errno)); + if (!env_name_is_valid(cenv)) + log_warning("Environment variable name '%s' is not valid. Ignoring.", cenv); + else { + r = unsetenv(cenv); + if (r < 0) + log_warning("Unsetting environment variable '%s' failed, ignoring: %m", cenv); + } } else { - *eq = 0; - r = setenv(cenv, eq + 1, 1); - if (r < 0) - log_warning("setenv failed %s. Ignoring.", strerror(errno)); + if (!env_assignment_is_valid(cenv)) + log_warning("Environment variable assignment '%s' is not valid. Ignoring.", cenv); + else { + *eq = 0; + r = setenv(cenv, eq + 1, 1); + if (r < 0) + log_warning("Setting environment variable '%s=%s' failed, ignoring: %m", cenv, eq + 1); + } } - free(cenv); -#ifdef HAVE_SYSV_COMPAT - } else if (startswith(word, "systemd.sysv_console=")) { - int r; - - if ((r = parse_boolean(word + 21)) < 0) - log_warning("Failed to parse SysV console switch %s. Ignoring.", word + 20); - else - arg_sysv_console = r; -#endif - } else if (startswith(word, "systemd.")) { - - log_warning("Unknown kernel switch %s. Ignoring.", word); + } else if (startswith(word, "systemd.") || + (in_initrd() && startswith(word, "rd.systemd."))) { + + const char *c; + + /* Ignore systemd.journald.xyz and friends */ + c = word; + if (startswith(c, "rd.")) + c += 3; + if (startswith(c, "systemd.")) + c += 8; + if (c[strcspn(c, ".=")] != '.') { + + log_warning("Unknown kernel switch %s. Ignoring.", word); + + log_info("Supported kernel switches:\n" + "systemd.unit=UNIT Default unit to start\n" + "rd.systemd.unit=UNIT Default unit to start when run in initrd\n" + "systemd.dump_core=0|1 Dump core on crash\n" + "systemd.crash_shell=0|1 Run shell on crash\n" + "systemd.crash_chvt=N Change to VT #N on crash\n" + "systemd.confirm_spawn=0|1 Confirm every process spawn\n" + "systemd.show_status=0|1 Show status updates on the console during bootup\n" + "systemd.log_target=console|kmsg|journal|journal-or-kmsg|syslog|syslog-or-kmsg|null\n" + " Log target\n" + "systemd.log_level=LEVEL Log level\n" + "systemd.log_color=0|1 Highlight important log messages\n" + "systemd.log_location=0|1 Include code location in log messages\n" + "systemd.default_standard_output=null|tty|syslog|syslog+console|kmsg|kmsg+console|journal|journal+console\n" + " Set default log output for services\n" + "systemd.default_standard_error=null|tty|syslog|syslog+console|kmsg|kmsg+console|journal|journal+console\n" + " Set default log error output for services\n" + "systemd.setenv=ASSIGNMENT Set an environment variable for all spawned processes\n"); + } - log_info("Supported kernel switches:\n" - "systemd.unit=UNIT Default unit to start\n" - "systemd.dump_core=0|1 Dump core on crash\n" - "systemd.crash_shell=0|1 Run shell on crash\n" - "systemd.crash_chvt=N Change to VT #N on crash\n" - "systemd.confirm_spawn=0|1 Confirm every process spawn\n" - "systemd.show_status=0|1 Show status updates on the console during bootup\n" -#ifdef HAVE_SYSV_COMPAT - "systemd.sysv_console=0|1 Connect output of SysV scripts to console\n" -#endif - "systemd.log_target=console|kmsg|journal|journal-or-kmsg|syslog|syslog-or-kmsg|null\n" - " Log target\n" - "systemd.log_level=LEVEL Log level\n" - "systemd.log_color=0|1 Highlight important log messages\n" - "systemd.log_location=0|1 Include code location in log messages\n" - "systemd.default_standard_output=null|tty|syslog|syslog+console|kmsg|kmsg+console|journal|journal+console\n" - " Set default log output for services\n" - "systemd.default_standard_error=null|tty|syslog|syslog+console|kmsg|kmsg+console|journal|journal+console\n" - " Set default log error output for services\n"); - - } else if (streq(word, "quiet")) { + } else if (streq(word, "quiet")) arg_show_status = false; -#ifdef HAVE_SYSV_COMPAT - arg_sysv_console = false; -#endif - } else { + else if (!in_initrd()) { unsigned i; /* SysV compatibility */ @@ -500,14 +528,14 @@ static int config_parse_cpu_affinity2( unsigned cpu; if (!(t = strndup(w, l))) - return -ENOMEM; + return log_oom(); r = safe_atou(t, &cpu); free(t); if (!c) if (!(c = cpu_set_malloc(&ncpus))) - return -ENOMEM; + return log_oom(); if (r < 0 || cpu >= ncpus) { log_error("[%s:%u] Failed to parse CPU affinity: %s", filename, line, rvalue); @@ -573,7 +601,7 @@ static int config_parse_join_controllers( s = strndup(w, length); if (!s) - return -ENOMEM; + return log_oom(); l = strv_split(s, ","); free(s); @@ -589,7 +617,7 @@ static int config_parse_join_controllers( arg_join_controllers = new(char**, 2); if (!arg_join_controllers) { strv_free(l); - return -ENOMEM; + return log_oom(); } arg_join_controllers[0] = l; @@ -603,7 +631,7 @@ static int config_parse_join_controllers( t = new0(char**, n+2); if (!t) { strv_free(l); - return -ENOMEM; + return log_oom(); } n = 0; @@ -617,7 +645,7 @@ static int config_parse_join_controllers( if (!c) { strv_free(l); strv_free_free(t); - return -ENOMEM; + return log_oom(); } strv_free(l); @@ -629,7 +657,7 @@ static int config_parse_join_controllers( if (!c) { strv_free(l); strv_free_free(t); - return -ENOMEM; + return log_oom(); } t[n++] = c; @@ -656,9 +684,6 @@ static int parse_config_file(void) { { "Manager", "DumpCore", config_parse_bool, 0, &arg_dump_core }, { "Manager", "CrashShell", config_parse_bool, 0, &arg_crash_shell }, { "Manager", "ShowStatus", config_parse_bool, 0, &arg_show_status }, -#ifdef HAVE_SYSV_COMPAT - { "Manager", "SysVConsole", config_parse_bool, 0, &arg_sysv_console }, -#endif { "Manager", "CrashChVT", config_parse_int, 0, &arg_crash_chvt }, { "Manager", "CPUAffinity", config_parse_cpu_affinity2, 0, NULL }, { "Manager", "DefaultControllers", config_parse_strv, 0, &arg_default_controllers }, @@ -667,6 +692,8 @@ static int parse_config_file(void) { { "Manager", "JoinControllers", config_parse_join_controllers, 0, &arg_join_controllers }, { "Manager", "RuntimeWatchdogSec", config_parse_usec, 0, &arg_runtime_watchdog }, { "Manager", "ShutdownWatchdogSec", config_parse_usec, 0, &arg_shutdown_watchdog }, + { "Manager", "CapabilityBoundingSet", config_parse_bounding_set, 0, &arg_capability_bounding_set_drop }, + { "Manager", "TimerSlackNSec", config_parse_nsec, 0, &arg_timer_slack_nsec }, { "Manager", "DefaultLimitCPU", config_parse_limit, 0, &arg_default_rlimit[RLIMIT_CPU]}, { "Manager", "DefaultLimitFSIZE", config_parse_limit, 0, &arg_default_rlimit[RLIMIT_FSIZE]}, { "Manager", "DefaultLimitDATA", config_parse_limit, 0, &arg_default_rlimit[RLIMIT_DATA]}, @@ -690,7 +717,7 @@ static int parse_config_file(void) { const char *fn; int r; - fn = arg_running_as == MANAGER_SYSTEM ? SYSTEM_CONFIG_FILE : USER_CONFIG_FILE; + fn = arg_running_as == SYSTEMD_SYSTEM ? SYSTEM_CONFIG_FILE : USER_CONFIG_FILE; f = fopen(fn, "re"); if (!f) { if (errno == ENOENT) @@ -733,10 +760,13 @@ static int parse_proc_cmdline(void) { } r = parse_proc_cmdline_word(word); - free(word); - - if (r < 0) + if (r < 0) { + log_error("Failed on cmdline argument %s: %s", word, strerror(-r)); + free(word); goto finish; + } + + free(word); } r = 0; @@ -757,12 +787,12 @@ static int parse_argv(int argc, char *argv[]) { ARG_SYSTEM, ARG_USER, ARG_TEST, + ARG_VERSION, ARG_DUMP_CONFIGURATION_ITEMS, ARG_DUMP_CORE, ARG_CRASH_SHELL, ARG_CONFIRM_SPAWN, ARG_SHOW_STATUS, - ARG_SYSV_CONSOLE, ARG_DESERIALIZE, ARG_SWITCHED_ROOT, ARG_INTROSPECT, @@ -780,14 +810,12 @@ static int parse_argv(int argc, char *argv[]) { { "user", no_argument, NULL, ARG_USER }, { "test", no_argument, NULL, ARG_TEST }, { "help", no_argument, NULL, 'h' }, + { "version", no_argument, NULL, ARG_VERSION }, { "dump-configuration-items", no_argument, NULL, ARG_DUMP_CONFIGURATION_ITEMS }, { "dump-core", optional_argument, NULL, ARG_DUMP_CORE }, { "crash-shell", optional_argument, NULL, ARG_CRASH_SHELL }, { "confirm-spawn", optional_argument, NULL, ARG_CONFIRM_SPAWN }, { "show-status", optional_argument, NULL, ARG_SHOW_STATUS }, -#ifdef HAVE_SYSV_COMPAT - { "sysv-console", optional_argument, NULL, ARG_SYSV_CONSOLE }, -#endif { "deserialize", required_argument, NULL, ARG_DESERIALIZE }, { "switched-root", no_argument, NULL, ARG_SWITCHED_ROOT }, { "introspect", optional_argument, NULL, ARG_INTROSPECT }, @@ -877,17 +905,21 @@ static int parse_argv(int argc, char *argv[]) { break; case ARG_SYSTEM: - arg_running_as = MANAGER_SYSTEM; + arg_running_as = SYSTEMD_SYSTEM; break; case ARG_USER: - arg_running_as = MANAGER_USER; + arg_running_as = SYSTEMD_USER; break; case ARG_TEST: arg_action = ACTION_TEST; break; + case ARG_VERSION: + arg_action = ACTION_VERSION; + break; + case ARG_DUMP_CONFIGURATION_ITEMS: arg_action = ACTION_DUMP_CONFIGURATION_ITEMS; break; @@ -928,29 +960,22 @@ static int parse_argv(int argc, char *argv[]) { arg_show_status = r; break; -#ifdef HAVE_SYSV_COMPAT - case ARG_SYSV_CONSOLE: - r = optarg ? parse_boolean(optarg) : 1; - if (r < 0) { - log_error("Failed to parse SysV console boolean %s.", optarg); - return r; - } - arg_sysv_console = r; - break; -#endif - case ARG_DESERIALIZE: { int fd; FILE *f; - if ((r = safe_atoi(optarg, &fd)) < 0 || fd < 0) { + r = safe_atoi(optarg, &fd); + if (r < 0 || fd < 0) { log_error("Failed to parse deserialize option %s.", optarg); - return r; + return r < 0 ? r : -EINVAL; } - if (!(f = fdopen(fd, "r"))) { + fd_cloexec(fd, true); + + f = fdopen(fd, "r"); + if (!f) { log_error("Failed to open serialization fd: %m"); - return r; + return -errno; } if (serialization) @@ -962,7 +987,7 @@ static int parse_argv(int argc, char *argv[]) { } case ARG_SWITCHED_ROOT: - /* Nothing special yet */ + arg_switched_root = true; break; case ARG_INTROSPECT: { @@ -1032,8 +1057,10 @@ static int parse_argv(int argc, char *argv[]) { * instead. */ for (a = argv; a < argv + argc; a++) - if ((r = parse_proc_cmdline_word(*a)) < 0) + if ((r = parse_proc_cmdline_word(*a)) < 0) { + log_error("Failed on cmdline argument %s: %s", *a, strerror(-r)); return r; + } } return 0; @@ -1054,9 +1081,6 @@ static int help(void) { " --crash-shell[=0|1] Run shell on crash\n" " --confirm-spawn[=0|1] Ask for confirmation when spawning processes\n" " --show-status[=0|1] Show status updates on the console during bootup\n" -#ifdef HAVE_SYSV_COMPAT - " --sysv-console[=0|1] Connect output of SysV scripts to console\n" -#endif " --log-target=TARGET Set log target (console, journal, syslog, kmsg, journal-or-kmsg, syslog-or-kmsg, null)\n" " --log-level=LEVEL Set log level (debug, info, notice, warning, err, crit, alert, emerg)\n" " --log-color[=0|1] Highlight important log messages\n" @@ -1068,7 +1092,14 @@ static int help(void) { return 0; } -static int prepare_reexecute(Manager *m, FILE **_f, FDSet **_fds) { +static int version(void) { + puts(PACKAGE_STRING); + puts(SYSTEMD_FEATURES); + + return 0; +} + +static int prepare_reexecute(Manager *m, FILE **_f, FDSet **_fds, bool serialize_jobs) { FILE *f = NULL; FDSet *fds = NULL; int r; @@ -1080,18 +1111,21 @@ static int prepare_reexecute(Manager *m, FILE **_f, FDSet **_fds) { /* Make sure nothing is really destructed when we shut down */ m->n_reloading ++; - if ((r = manager_open_serialization(m, &f)) < 0) { + r = manager_open_serialization(m, &f); + if (r < 0) { log_error("Failed to create serialization file: %s", strerror(-r)); goto fail; } - if (!(fds = fdset_new())) { + fds = fdset_new(); + if (!fds) { r = -ENOMEM; log_error("Failed to allocate fd set: %s", strerror(-r)); goto fail; } - if ((r = manager_serialize(m, f, fds)) < 0) { + r = manager_serialize(m, f, fds, serialize_jobs); + if (r < 0) { log_error("Failed to serialize state: %s", strerror(-r)); goto fail; } @@ -1101,12 +1135,14 @@ static int prepare_reexecute(Manager *m, FILE **_f, FDSet **_fds) { goto fail; } - if ((r = fd_cloexec(fileno(f), false)) < 0) { + r = fd_cloexec(fileno(f), false); + if (r < 0) { log_error("Failed to disable O_CLOEXEC for serialization: %s", strerror(-r)); goto fail; } - if ((r = fdset_cloexec(fds, false)) < 0) { + r = fdset_cloexec(fds, false); + if (r < 0) { log_error("Failed to disable O_CLOEXEC for serialization fds: %s", strerror(-r)); goto fail; } @@ -1125,6 +1161,42 @@ fail: return r; } +static int bump_rlimit_nofile(struct rlimit *saved_rlimit) { + struct rlimit nl; + int r; + + assert(saved_rlimit); + + /* Save the original RLIMIT_NOFILE so that we can reset it + * later when transitioning from the initrd to the main + * systemd or suchlike. */ + if (getrlimit(RLIMIT_NOFILE, saved_rlimit) < 0) { + log_error("Reading RLIMIT_NOFILE failed: %m"); + return -errno; + } + + /* Make sure forked processes get the default kernel setting */ + if (!arg_default_rlimit[RLIMIT_NOFILE]) { + struct rlimit *rl; + + rl = newdup(struct rlimit, saved_rlimit, 1); + if (!rl) + return log_oom(); + + arg_default_rlimit[RLIMIT_NOFILE] = rl; + } + + /* Bump up the resource limit for ourselves substantially */ + nl.rlim_cur = nl.rlim_max = 64*1024; + r = setrlimit_closest(RLIMIT_NOFILE, &nl); + if (r < 0) { + log_error("Setting RLIMIT_NOFILE failed: %s", strerror(-r)); + return r; + } + + return 0; +} + static struct dual_timestamp* parse_initrd_timestamp(struct dual_timestamp *t) { const char *e; unsigned long long a, b; @@ -1191,6 +1263,28 @@ static void test_cgroups(void) { sleep(10); } +static int initialize_join_controllers(void) { + /* By default, mount "cpu" + "cpuacct" together, and "net_cls" + * + "net_prio". We'd like to add "cpuset" to the mix, but + * "cpuset" does't really work for groups with no initialized + * attributes. */ + + arg_join_controllers = new(char**, 3); + if (!arg_join_controllers) + return -ENOMEM; + + arg_join_controllers[0] = strv_new("cpu", "cpuacct", NULL); + if (!arg_join_controllers[0]) + return -ENOMEM; + + arg_join_controllers[1] = strv_new("net_cls", "net_prio", NULL); + if (!arg_join_controllers[1]) + return -ENOMEM; + + arg_join_controllers[2] = NULL; + return 0; +} + int main(int argc, char *argv[]) { Manager *m = NULL; int r, retval = EXIT_FAILURE; @@ -1205,7 +1299,9 @@ int main(int argc, char *argv[]) { int j; bool loaded_policy = false; bool arm_reboot_watchdog = false; + bool queue_default_job = false; char *switch_root_dir = NULL, *switch_root_init = NULL; + static struct rlimit saved_rlimit_nofile = { 0, 0 }; #ifdef HAVE_SYSV_COMPAT if (getpid() != 1 && strstr(program_invocation_short_name, "init")) { @@ -1222,19 +1318,13 @@ int main(int argc, char *argv[]) { /* Determine if this is a reexecution or normal bootup. We do * the full command line parsing much later, so let's just * have a quick peek here. */ - for (j = 1; j < argc; j++) - if (streq(argv[j], "--deserialize")) { - skip_setup = true; - break; - } + if (strv_find(argv+1, "--deserialize")) + skip_setup = true; /* If we have switched root, do all the special setup * things */ - for (j = 1; j < argc; j++) - if (streq(argv[j], "--switched-root")) { - skip_setup = false; - break; - } + if (strv_find(argv+1, "--switched-root")) + skip_setup = false; /* If we get started via the /sbin/init symlink then we are called 'init'. After a subsequent reexecution we are then @@ -1247,10 +1337,15 @@ int main(int argc, char *argv[]) { saved_argc = argc; log_show_color(isatty(STDERR_FILENO) > 0); - log_show_location(false); - log_set_max_level(LOG_INFO); - if (getpid() == 1) { + if (getpid() == 1 && detect_container(NULL) <= 0) { + + /* Running outside of a container as PID 1 */ + arg_running_as = SYSTEMD_SYSTEM; + make_null_stdio(); + log_set_target(LOG_TARGET_KMSG); + log_open(); + if (in_initrd()) { char *rd_timestamp = NULL; @@ -1264,56 +1359,87 @@ int main(int argc, char *argv[]) { } } - arg_running_as = MANAGER_SYSTEM; - log_set_target(detect_container(NULL) > 0 ? LOG_TARGET_JOURNAL : LOG_TARGET_JOURNAL_OR_KMSG); - if (!skip_setup) { + mount_setup_early(); if (selinux_setup(&loaded_policy) < 0) goto finish; if (ima_setup() < 0) goto finish; + if (smack_setup() < 0) + goto finish; } - log_open(); - if (label_init(NULL) < 0) goto finish; - if (!skip_setup) + if (!skip_setup) { if (hwclock_is_localtime() > 0) { int min; - r = hwclock_apply_localtime_delta(&min); + /* The first-time call to settimeofday() does a time warp in the kernel */ + r = hwclock_set_timezone(&min); if (r < 0) log_error("Failed to apply local time delta, ignoring: %s", strerror(-r)); else log_info("RTC configured in localtime, applying delta of %i minutes to system time.", min); + } else if (!in_initrd()) { + /* + * Do dummy first-time call to seal the kernel's time warp magic + * + * Do not call this this from inside the initrd. The initrd might not + * carry /etc/adjtime with LOCAL, but the real system could be set up + * that way. In such case, we need to delay the time-warp or the sealing + * until we reach the real system. + */ + hwclock_reset_timezone(); + + /* Tell the kernel our time zone */ + r = hwclock_set_timezone(NULL); + if (r < 0) + log_error("Failed to set the kernel's time zone, ignoring: %s", strerror(-r)); } + } + + /* Set the default for later on, but don't actually + * open the logs like this for now. Note that if we + * are transitioning from the initrd there might still + * be journal fd open, and we shouldn't attempt + * opening that before we parsed /proc/cmdline which + * might redirect output elsewhere. */ + log_set_target(LOG_TARGET_JOURNAL_OR_KMSG); + + } else if (getpid() == 1) { + + /* Running inside a container, as PID 1 */ + arg_running_as = SYSTEMD_SYSTEM; + log_set_target(LOG_TARGET_CONSOLE); + log_open(); + + /* For the later on, see above... */ + log_set_target(LOG_TARGET_JOURNAL); } else { - arg_running_as = MANAGER_USER; + + /* Running as user instance */ + arg_running_as = SYSTEMD_USER; log_set_target(LOG_TARGET_AUTO); log_open(); } /* Initialize default unit */ - if (set_default_unit(SPECIAL_DEFAULT_TARGET) < 0) - goto finish; - - /* By default, mount "cpu" and "cpuacct" together */ - arg_join_controllers = new(char**, 2); - if (!arg_join_controllers) + r = set_default_unit(SPECIAL_DEFAULT_TARGET); + if (r < 0) { + log_error("Failed to set default unit %s: %s", SPECIAL_DEFAULT_TARGET, strerror(-r)); goto finish; + } - arg_join_controllers[0] = strv_new("cpu", "cpuacct", NULL); - arg_join_controllers[1] = NULL; - - if (!arg_join_controllers[0]) + r = initialize_join_controllers(); + if (r < 0) goto finish; /* Mount /proc, /sys and friends, so that /proc/cmdline and * /proc/$PID/fd is available. */ - if (geteuid() == 0 && !getenv("SYSTEMD_SKIP_API_MOUNTS")) { + if (getpid() == 1) { r = mount_setup(loaded_policy); if (r < 0) goto finish; @@ -1328,7 +1454,7 @@ int main(int argc, char *argv[]) { if (parse_config_file() < 0) goto finish; - if (arg_running_as == MANAGER_SYSTEM) + if (arg_running_as == SYSTEMD_SYSTEM) if (parse_proc_cmdline() < 0) goto finish; @@ -1337,12 +1463,20 @@ int main(int argc, char *argv[]) { if (parse_argv(argc, argv) < 0) goto finish; - if (arg_action == ACTION_TEST && geteuid() == 0) { + if (arg_action == ACTION_TEST && + geteuid() == 0) { log_error("Don't run test mode as root."); goto finish; } - if (arg_running_as == MANAGER_SYSTEM && + if (arg_running_as == SYSTEMD_USER && + arg_action == ACTION_RUN && + sd_booted() <= 0) { + log_error("Trying to run as user instance, but the system has not been booted with systemd."); + goto finish; + } + + if (arg_running_as == SYSTEMD_SYSTEM && arg_action == ACTION_RUN && running_in_chroot() > 0) { log_error("Cannot be run in a chroot() environment."); @@ -1352,6 +1486,9 @@ int main(int argc, char *argv[]) { if (arg_action == ACTION_HELP) { retval = help(); goto finish; + } else if (arg_action == ACTION_VERSION) { + retval = version(); + goto finish; } else if (arg_action == ACTION_DUMP_CONFIGURATION_ITEMS) { unit_dump_config_items(stdout); retval = EXIT_SUCCESS; @@ -1367,16 +1504,15 @@ int main(int argc, char *argv[]) { log_close(); /* Remember open file descriptors for later deserialization */ - if (serialization) { - r = fdset_new_fill(&fds); - if (r < 0) { - log_error("Failed to allocate fd set: %s", strerror(-r)); - goto finish; - } + r = fdset_new_fill(&fds); + if (r < 0) { + log_error("Failed to allocate fd set: %s", strerror(-r)); + goto finish; + } else + fdset_cloexec(fds, true); + if (serialization) assert_se(fdset_remove(fds, fileno(serialization)) >= 0); - } else - close_all_fds(NULL, 0); /* Set up PATH unless it is already set */ setenv("PATH", @@ -1385,9 +1521,9 @@ int main(int argc, char *argv[]) { #else "/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin", #endif - arg_running_as == MANAGER_SYSTEM); + arg_running_as == SYSTEMD_SYSTEM); - if (arg_running_as == MANAGER_SYSTEM) { + if (arg_running_as == SYSTEMD_SYSTEM) { /* Parse the data passed to us. We leave this * variables set, but the manager later on will not * pass them on to our children. */ @@ -1411,6 +1547,12 @@ int main(int argc, char *argv[]) { unsetenv("USER"); unsetenv("LOGNAME"); + /* We suppress the socket activation env vars, as + * we'll try to match *any* open fd to units if + * possible. */ + unsetenv("LISTEN_FDS"); + unsetenv("LISTEN_PID"); + /* All other variables are left as is, so that clients * can still read them via /proc/1/environ */ } @@ -1418,7 +1560,7 @@ int main(int argc, char *argv[]) { /* Move out of the way, so that we won't block unmounts */ assert_se(chdir("/") == 0); - if (arg_running_as == MANAGER_SYSTEM) { + if (arg_running_as == SYSTEMD_SYSTEM) { /* Become a session leader if we aren't one yet. */ setsid(); @@ -1431,35 +1573,46 @@ int main(int argc, char *argv[]) { /* Reset the console, but only if this is really init and we * are freshly booted */ - if (arg_running_as == MANAGER_SYSTEM && arg_action == ACTION_RUN) { + if (arg_running_as == SYSTEMD_SYSTEM && arg_action == ACTION_RUN) console_setup(getpid() == 1 && !skip_setup); - make_null_stdio(); - } /* Open the logging devices, if possible and necessary */ log_open(); /* Make sure we leave a core dump without panicing the * kernel. */ - if (getpid() == 1) + if (getpid() == 1) { install_crash_handler(); - if (geteuid() == 0 && !getenv("SYSTEMD_SKIP_API_MOUNTS")) { r = mount_cgroup_controllers(arg_join_controllers); if (r < 0) goto finish; } - log_full(arg_running_as == MANAGER_SYSTEM ? LOG_INFO : LOG_DEBUG, - PACKAGE_STRING " running in %s mode. (" SYSTEMD_FEATURES "; " DISTRIBUTION ")", manager_running_as_to_string(arg_running_as)); + if (arg_running_as == SYSTEMD_SYSTEM) { + const char *virtualization = NULL; + + log_info(PACKAGE_STRING " running in system mode. (" SYSTEMD_FEATURES ")"); - if (arg_running_as == MANAGER_SYSTEM && !skip_setup) { + detect_virtualization(&virtualization); + if (virtualization) + log_info("Detected virtualization '%s'.", virtualization); + + if (in_initrd()) + log_info("Running in initial RAM disk."); + + } else + log_debug(PACKAGE_STRING " running in user mode. (" SYSTEMD_FEATURES ")"); + + if (arg_running_as == SYSTEMD_SYSTEM && !skip_setup) { locale_setup(); if (arg_show_status || plymouth_running()) status_welcome(); +#ifdef HAVE_KMOD kmod_setup(); +#endif hostname_setup(); machine_id_setup(); loopback_setup(); @@ -1469,9 +1622,38 @@ int main(int argc, char *argv[]) { test_cgroups(); } - if (arg_running_as == MANAGER_SYSTEM && arg_runtime_watchdog > 0) + if (arg_running_as == SYSTEMD_SYSTEM && arg_runtime_watchdog > 0) watchdog_set_timeout(&arg_runtime_watchdog); + if (arg_timer_slack_nsec != (nsec_t) -1) + if (prctl(PR_SET_TIMERSLACK, arg_timer_slack_nsec) < 0) + log_error("Failed to adjust timer slack: %m"); + + if (arg_capability_bounding_set_drop) { + r = capability_bounding_set_drop(arg_capability_bounding_set_drop, true); + if (r < 0) { + log_error("Failed to drop capability bounding set: %s", strerror(-r)); + goto finish; + } + r = capability_bounding_set_drop_usermode(arg_capability_bounding_set_drop); + if (r < 0) { + log_error("Failed to drop capability bounding set of usermode helpers: %s", strerror(-r)); + goto finish; + } + } + + if (arg_running_as == SYSTEMD_USER) { + /* Become reaper of our children */ + if (prctl(PR_SET_CHILD_SUBREAPER, 1) < 0) { + log_warning("Failed to make us a subreaper: %m"); + if (errno == EINVAL) + log_info("Perhaps the kernel version is too old (< 3.4?)"); + } + } + + if (arg_running_as == SYSTEMD_SYSTEM) + bump_rlimit_nofile(&saved_rlimit_nofile); + r = manager_new(arg_running_as, &m); if (r < 0) { log_error("Failed to allocate manager object: %s", strerror(-r)); @@ -1479,9 +1661,6 @@ int main(int argc, char *argv[]) { } m->confirm_spawn = arg_confirm_spawn; -#ifdef HAVE_SYSV_COMPAT - m->sysv_console = arg_sysv_console; -#endif m->default_std_output = arg_default_std_output; m->default_std_error = arg_default_std_error; m->runtime_watchdog = arg_runtime_watchdog; @@ -1497,24 +1676,25 @@ int main(int argc, char *argv[]) { manager_set_show_status(m, arg_show_status); + /* Remember whether we should queue the default job */ + queue_default_job = !serialization || arg_switched_root; + before_startup = now(CLOCK_MONOTONIC); r = manager_startup(m, serialization, fds); if (r < 0) log_error("Failed to fully start up daemon: %s", strerror(-r)); - if (fds) { - /* This will close all file descriptors that were opened, but - * not claimed by any unit. */ - - fdset_free(fds); - fds = NULL; - } + /* This will close all file descriptors that were opened, but + * not claimed by any unit. */ + fdset_free(fds); if (serialization) { fclose(serialization); serialization = NULL; - } else { + } + + if (queue_default_job) { DBusError error; Unit *target = NULL; Job *default_unit_job; @@ -1556,12 +1736,23 @@ int main(int argc, char *argv[]) { manager_dump_units(m, stdout, "\t"); } - r = manager_add_job(m, JOB_START, target, JOB_REPLACE, false, &error, &default_unit_job); - if (r < 0) { - log_error("Failed to start default target: %s", bus_error(&error, r)); + r = manager_add_job(m, JOB_START, target, JOB_ISOLATE, false, &error, &default_unit_job); + if (r == -EPERM) { + log_error("Default target could not be isolated, starting instead: %s", bus_error(&error, r)); + dbus_error_free(&error); + + r = manager_add_job(m, JOB_START, target, JOB_REPLACE, false, &error, &default_unit_job); + if (r < 0) { + log_error("Failed to start default target: %s", bus_error(&error, r)); + dbus_error_free(&error); + goto finish; + } + } else if (r < 0) { + log_error("Failed to isolate default target: %s", bus_error(&error, r)); dbus_error_free(&error); goto finish; } + m->default_unit_job_id = default_unit_job->id; after_startup = now(CLOCK_MONOTONIC); @@ -1600,7 +1791,7 @@ int main(int argc, char *argv[]) { case MANAGER_REEXECUTE: - if (prepare_reexecute(m, &serialization, &fds) < 0) + if (prepare_reexecute(m, &serialization, &fds, true) < 0) goto finish; reexecute = true; @@ -1614,7 +1805,7 @@ int main(int argc, char *argv[]) { m->switch_root = m->switch_root_init = NULL; if (!switch_root_init) - if (prepare_reexecute(m, &serialization, &fds) < 0) + if (prepare_reexecute(m, &serialization, &fds, false) < 0) goto finish; reexecute = true; @@ -1649,7 +1840,7 @@ finish: manager_free(m); for (j = 0; j < RLIMIT_NLIMITS; j++) - free (arg_default_rlimit[j]); + free(arg_default_rlimit[j]); free(arg_default_unit); strv_free(arg_default_controllers); @@ -1667,7 +1858,20 @@ finish: * rebooted while we do that */ watchdog_close(true); + /* Reset the RLIMIT_NOFILE to the kernel default, so + * that the new systemd can pass the kernel default to + * its child processes */ + if (saved_rlimit_nofile.rlim_cur > 0) + setrlimit(RLIMIT_NOFILE, &saved_rlimit_nofile); + if (switch_root_dir) { + /* Kill all remaining processes from the + * initrd, but don't wait for them, so that we + * can handle the SIGCHLD for them after + * deserializing. */ + broadcast_signal(SIGTERM, false); + + /* And switch root */ r = switch_root(switch_root_dir); if (r < 0) log_error("Failed to switch root, ignoring: %s", strerror(-r)); @@ -1694,7 +1898,7 @@ finish: args[i++] = SYSTEMD_BINARY_PATH; if (switch_root_dir) args[i++] = "--switched-root"; - args[i++] = arg_running_as == MANAGER_SYSTEM ? "--system" : "--user"; + args[i++] = arg_running_as == SYSTEMD_SYSTEM ? "--system" : "--user"; args[i++] = "--deserialize"; args[i++] = sfd; args[i++] = NULL; @@ -1719,6 +1923,9 @@ finish: fds = NULL; } + /* Reopen the console */ + make_console_stdio(); + for (j = 1, i = 1; j < argc; j++) args[i++] = argv[j]; args[i++] = NULL; @@ -1733,12 +1940,15 @@ finish: args[0] = "/sbin/init"; execv(args[0], (char* const*) args); - log_warning("Failed to execute /sbin/init, trying fallback: %m"); + if (errno == ENOENT) { + log_warning("No /sbin/init, trying fallback"); - args[0] = "/bin/sh"; - args[1] = NULL; - execv(args[0], (char* const*) args); - log_error("Failed to execute /bin/sh, giving up: %m"); + args[0] = "/bin/sh"; + args[1] = NULL; + execv(args[0], (char* const*) args); + log_error("Failed to execute /bin/sh, giving up: %m"); + } else + log_warning("Failed to execute /sbin/init, giving up: %m"); } if (serialization)