diff options
Diffstat (limited to 'src')
-rw-r--r-- | src/core/dbus-execute.c | 5 | ||||
-rw-r--r-- | src/core/execute.c | 103 | ||||
-rw-r--r-- | src/core/execute.h | 4 | ||||
-rw-r--r-- | src/core/load-fragment-gperf.gperf.m4 | 2 | ||||
-rw-r--r-- | src/core/main.c | 56 | ||||
-rw-r--r-- | src/core/service.c | 56 | ||||
-rw-r--r-- | src/core/service.h | 2 | ||||
-rw-r--r-- | src/core/unit.c | 1 | ||||
-rw-r--r-- | src/fstab-generator/fstab-generator.c | 6 | ||||
-rw-r--r-- | src/machine/machinectl.c | 6 | ||||
-rw-r--r-- | src/shared/bus-unit-util.c | 3 | ||||
-rw-r--r-- | src/shared/fdset.c (renamed from src/basic/fdset.c) | 0 | ||||
-rw-r--r-- | src/shared/fdset.h (renamed from src/basic/fdset.h) | 0 | ||||
-rw-r--r-- | src/systemctl/systemctl.c | 2 | ||||
-rw-r--r-- | src/systemd/sd-daemon.h | 5 |
15 files changed, 223 insertions, 28 deletions
diff --git a/src/core/dbus-execute.c b/src/core/dbus-execute.c index 4c88c41127..644b9561b5 100644 --- a/src/core/dbus-execute.c +++ b/src/core/dbus-execute.c @@ -720,6 +720,7 @@ const sd_bus_vtable bus_exec_vtable[] = { SD_BUS_PROPERTY("RuntimeDirectoryMode", "u", bus_property_get_mode, offsetof(ExecContext, runtime_directory_mode), SD_BUS_VTABLE_PROPERTY_CONST), SD_BUS_PROPERTY("RuntimeDirectory", "as", NULL, offsetof(ExecContext, runtime_directory), SD_BUS_VTABLE_PROPERTY_CONST), SD_BUS_PROPERTY("MemoryDenyWriteExecute", "b", bus_property_get_bool, offsetof(ExecContext, memory_deny_write_execute), SD_BUS_VTABLE_PROPERTY_CONST), + SD_BUS_PROPERTY("RestrictRealtime", "b", bus_property_get_bool, offsetof(ExecContext, restrict_realtime), SD_BUS_VTABLE_PROPERTY_CONST), SD_BUS_VTABLE_END }; @@ -1057,7 +1058,7 @@ int bus_exec_context_set_transient_property( } else if (STR_IN_SET(name, "IgnoreSIGPIPE", "TTYVHangup", "TTYReset", "PrivateTmp", "PrivateDevices", "PrivateNetwork", - "NoNewPrivileges", "SyslogLevelPrefix", "MemoryDenyWriteExecute")) { + "NoNewPrivileges", "SyslogLevelPrefix", "MemoryDenyWriteExecute", "RestrictRealtime")) { int b; r = sd_bus_message_read(message, "b", &b); @@ -1083,6 +1084,8 @@ int bus_exec_context_set_transient_property( c->syslog_level_prefix = b; else if (streq(name, "MemoryDenyWriteExecute")) c->memory_deny_write_execute = b; + else if (streq(name, "RestrictRealtime")) + c->restrict_realtime = b; unit_write_drop_in_private_format(u, mode, name, "%s=%s", name, yes_no(b)); } diff --git a/src/core/execute.c b/src/core/execute.c index 3c3369373f..8cb18dbd5b 100644 --- a/src/core/execute.c +++ b/src/core/execute.c @@ -1237,7 +1237,7 @@ static int apply_memory_deny_write_execute(const ExecContext *c) { r = seccomp_rule_add( seccomp, - SCMP_ACT_KILL, + SCMP_ACT_ERRNO(EPERM), SCMP_SYS(mmap), 1, SCMP_A2(SCMP_CMP_MASKED_EQ, PROT_EXEC|PROT_WRITE, PROT_EXEC|PROT_WRITE)); @@ -1246,7 +1246,7 @@ static int apply_memory_deny_write_execute(const ExecContext *c) { r = seccomp_rule_add( seccomp, - SCMP_ACT_KILL, + SCMP_ACT_ERRNO(EPERM), SCMP_SYS(mprotect), 1, SCMP_A2(SCMP_CMP_MASKED_EQ, PROT_EXEC, PROT_EXEC)); @@ -1264,6 +1264,76 @@ finish: return r; } +static int apply_restrict_realtime(const ExecContext *c) { + static const int permitted_policies[] = { + SCHED_OTHER, + SCHED_BATCH, + SCHED_IDLE, + }; + + scmp_filter_ctx *seccomp; + unsigned i; + int r, p, max_policy = 0; + + assert(c); + + seccomp = seccomp_init(SCMP_ACT_ALLOW); + if (!seccomp) + return -ENOMEM; + + /* Determine the highest policy constant we want to allow */ + for (i = 0; i < ELEMENTSOF(permitted_policies); i++) + if (permitted_policies[i] > max_policy) + max_policy = permitted_policies[i]; + + /* Go through all policies with lower values than that, and block them -- unless they appear in the + * whitelist. */ + for (p = 0; p < max_policy; p++) { + bool good = false; + + /* Check if this is in the whitelist. */ + for (i = 0; i < ELEMENTSOF(permitted_policies); i++) + if (permitted_policies[i] == p) { + good = true; + break; + } + + if (good) + continue; + + /* Deny this policy */ + r = seccomp_rule_add( + seccomp, + SCMP_ACT_ERRNO(EPERM), + SCMP_SYS(sched_setscheduler), + 1, + SCMP_A1(SCMP_CMP_EQ, p)); + if (r < 0) + goto finish; + } + + /* Blacklist all other policies, i.e. the ones with higher values. Note that all comparisons are unsigned here, + * hence no need no check for < 0 values. */ + r = seccomp_rule_add( + seccomp, + SCMP_ACT_ERRNO(EPERM), + SCMP_SYS(sched_setscheduler), + 1, + SCMP_A1(SCMP_CMP_GT, max_policy)); + if (r < 0) + goto finish; + + r = seccomp_attr_set(seccomp, SCMP_FLTATR_CTL_NNP, 0); + if (r < 0) + goto finish; + + r = seccomp_load(seccomp); + +finish: + seccomp_release(seccomp); + return r; +} + #endif static void do_idle_pipe_dance(int idle_pipe[4]) { @@ -1951,10 +2021,20 @@ static int exec_child( int secure_bits = context->secure_bits; for (i = 0; i < _RLIMIT_MAX; i++) { + if (!context->rlimit[i]) continue; - if (setrlimit_closest(i, context->rlimit[i]) < 0) { + r = setrlimit_closest(i, context->rlimit[i]); + if (r < 0) { + *exit_status = EXIT_LIMITS; + return r; + } + } + + /* Set the RTPRIO resource limit to 0, but only if nothing else was explicitly requested. */ + if (context->restrict_realtime && !context->rlimit[RLIMIT_RTPRIO]) { + if (setrlimit(RLIMIT_RTPRIO, &RLIMIT_MAKE_CONST(0)) < 0) { *exit_status = EXIT_LIMITS; return -errno; } @@ -2015,7 +2095,7 @@ static int exec_child( } if (context->no_new_privileges || - (!have_effective_cap(CAP_SYS_ADMIN) && (use_address_families || use_syscall_filter))) + (!have_effective_cap(CAP_SYS_ADMIN) && (use_address_families || context->memory_deny_write_execute || context->restrict_realtime || use_syscall_filter))) if (prctl(PR_SET_NO_NEW_PRIVS, 1, 0, 0, 0) < 0) { *exit_status = EXIT_NO_NEW_PRIVILEGES; return -errno; @@ -2037,6 +2117,15 @@ static int exec_child( return r; } } + + if (context->restrict_realtime) { + r = apply_restrict_realtime(context); + if (r < 0) { + *exit_status = EXIT_SECCOMP; + return r; + } + } + if (use_syscall_filter) { r = apply_seccomp(context); if (r < 0) { @@ -2472,7 +2561,8 @@ void exec_context_dump(ExecContext *c, FILE* f, const char *prefix) { "%sProtectHome: %s\n" "%sProtectSystem: %s\n" "%sIgnoreSIGPIPE: %s\n" - "%sMemoryDenyWriteExecute: %s\n", + "%sMemoryDenyWriteExecute: %s\n" + "%sRestrictRealtime: %s\n", prefix, c->umask, prefix, c->working_directory ? c->working_directory : "/", prefix, c->root_directory ? c->root_directory : "/", @@ -2483,7 +2573,8 @@ void exec_context_dump(ExecContext *c, FILE* f, const char *prefix) { prefix, protect_home_to_string(c->protect_home), prefix, protect_system_to_string(c->protect_system), prefix, yes_no(c->ignore_sigpipe), - prefix, yes_no(c->memory_deny_write_execute)); + prefix, yes_no(c->memory_deny_write_execute), + prefix, yes_no(c->restrict_realtime)); STRV_FOREACH(e, c->environment) fprintf(f, "%sEnvironment: %s\n", prefix, *e); diff --git a/src/core/execute.h b/src/core/execute.h index cd1f7b36f6..210eea0e82 100644 --- a/src/core/execute.h +++ b/src/core/execute.h @@ -193,12 +193,14 @@ struct ExecContext { char **runtime_directory; mode_t runtime_directory_mode; + bool memory_deny_write_execute; + bool restrict_realtime; + bool oom_score_adjust_set:1; bool nice_set:1; bool ioprio_set:1; bool cpu_sched_set:1; bool no_new_privileges_set:1; - bool memory_deny_write_execute; }; #include "cgroup-util.h" diff --git a/src/core/load-fragment-gperf.gperf.m4 b/src/core/load-fragment-gperf.gperf.m4 index eb58586523..fe1006830b 100644 --- a/src/core/load-fragment-gperf.gperf.m4 +++ b/src/core/load-fragment-gperf.gperf.m4 @@ -56,11 +56,13 @@ m4_ifdef(`HAVE_SECCOMP', $1.SystemCallArchitectures, config_parse_syscall_archs, 0, offsetof($1, exec_context.syscall_archs) $1.SystemCallErrorNumber, config_parse_syscall_errno, 0, offsetof($1, exec_context) $1.MemoryDenyWriteExecute, config_parse_bool, 0, offsetof($1, exec_context.memory_deny_write_execute) +$1.RestrictRealtime, config_parse_bool, 0, offsetof($1, exec_context.restrict_realtime) $1.RestrictAddressFamilies, config_parse_address_families, 0, offsetof($1, exec_context)', `$1.SystemCallFilter, config_parse_warn_compat, DISABLED_CONFIGURATION, 0 $1.SystemCallArchitectures, config_parse_warn_compat, DISABLED_CONFIGURATION, 0 $1.SystemCallErrorNumber, config_parse_warn_compat, DISABLED_CONFIGURATION, 0 $1.MemoryDenyWriteExecute, config_parse_warn_compat, DISABLED_CONFIGURATION, 0 +$1.RestrictRealtime, config_parse_warn_compat, DISABLED_CONFIGURATION, 0 $1.RestrictAddressFamilies, config_parse_warn_compat, DISABLED_CONFIGURATION, 0') $1.LimitCPU, config_parse_limit, RLIMIT_CPU, offsetof($1, exec_context.rlimit) $1.LimitFSIZE, config_parse_limit, RLIMIT_FSIZE, offsetof($1, exec_context.rlimit) diff --git a/src/core/main.c b/src/core/main.c index 2785a3aa0b..237c9c9ebe 100644 --- a/src/core/main.c +++ b/src/core/main.c @@ -1294,6 +1294,35 @@ static int bump_unix_max_dgram_qlen(void) { return 1; } +static int fixup_environment(void) { + _cleanup_free_ char *term = NULL; + int r; + + /* When started as PID1, the kernel uses /dev/console + * for our stdios and uses TERM=linux whatever the + * backend device used by the console. We try to make + * a better guess here since some consoles might not + * have support for color mode for example. + * + * However if TERM was configured through the kernel + * command line then leave it alone. */ + + r = get_proc_cmdline_key("TERM=", &term); + if (r < 0) + return r; + + if (r == 0) { + term = strdup(default_term_for_tty("/dev/console") + 5); + if (!term) + return -errno; + } + + if (setenv("TERM", term, 1) < 0) + return -errno; + + return 0; +} + int main(int argc, char *argv[]) { Manager *m = NULL; int r, retval = EXIT_FAILURE; @@ -1353,7 +1382,6 @@ int main(int argc, char *argv[]) { saved_argv = argv; saved_argc = argc; - log_show_color(colors_enabled()); log_set_upgrade_syslog_to_journal(true); /* Disable the umask logic */ @@ -1364,7 +1392,6 @@ int main(int argc, char *argv[]) { /* Running outside of a container as PID 1 */ arg_system = true; - make_null_stdio(); log_set_target(LOG_TARGET_KMSG); log_open(); @@ -1480,6 +1507,21 @@ int main(int argc, char *argv[]) { (void) write_string_file("/proc/sys/kernel/core_pattern", "|/bin/false", 0); } + /* We expect the environment to be set correctly if run inside a + * container. */ + if (arg_system && detect_container() <= 0) { + if (fixup_environment() < 0) { + error_message = "Failed to fix up PID1 environment"; + goto finish; + } + + /* Try to figure out if we can use colors with the console. No + * need to do that for user instances since they never log + * into the console. */ + log_show_color(colors_enabled()); + make_null_stdio(); + } + /* Initialize default unit */ r = free_and_strdup(&arg_default_unit, SPECIAL_DEFAULT_TARGET); if (r < 0) { @@ -1967,6 +2009,9 @@ finish: log_error_errno(r, "Failed to switch root, trying to continue: %m"); } + /* Reopen the console */ + (void) make_console_stdio(); + args_size = MAX(6, argc+1); args = newa(const char*, args_size); @@ -1992,10 +2037,6 @@ finish: args[i++] = sfd; args[i++] = NULL; - /* do not pass along the environment we inherit from the kernel or initrd */ - if (switch_root_dir) - (void) clearenv(); - assert(i <= args_size); /* @@ -2018,9 +2059,6 @@ finish: arg_serialization = safe_fclose(arg_serialization); fds = fdset_free(fds); - /* Reopen the console */ - (void) make_console_stdio(); - for (j = 1, i = 1; j < (unsigned) argc; j++) args[i++] = argv[j]; args[i++] = NULL; diff --git a/src/core/service.c b/src/core/service.c index 78c33b1530..13de671700 100644 --- a/src/core/service.c +++ b/src/core/service.c @@ -200,16 +200,27 @@ static void service_stop_watchdog(Service *s) { s->watchdog_timestamp = DUAL_TIMESTAMP_NULL; } +static usec_t service_get_watchdog_usec(Service *s) { + assert(s); + + if (s->watchdog_override_enable) + return s->watchdog_override_usec; + else + return s->watchdog_usec; +} + static void service_start_watchdog(Service *s) { int r; + usec_t watchdog_usec; assert(s); - if (s->watchdog_usec <= 0) + watchdog_usec = service_get_watchdog_usec(s); + if (watchdog_usec == 0 || watchdog_usec == USEC_INFINITY) return; if (s->watchdog_event_source) { - r = sd_event_source_set_time(s->watchdog_event_source, usec_add(s->watchdog_timestamp.monotonic, s->watchdog_usec)); + r = sd_event_source_set_time(s->watchdog_event_source, usec_add(s->watchdog_timestamp.monotonic, watchdog_usec)); if (r < 0) { log_unit_warning_errno(UNIT(s), r, "Failed to reset watchdog timer: %m"); return; @@ -221,7 +232,7 @@ static void service_start_watchdog(Service *s) { UNIT(s)->manager->event, &s->watchdog_event_source, CLOCK_MONOTONIC, - usec_add(s->watchdog_timestamp.monotonic, s->watchdog_usec), 0, + usec_add(s->watchdog_timestamp.monotonic, watchdog_usec), 0, service_dispatch_watchdog, s); if (r < 0) { log_unit_warning_errno(UNIT(s), r, "Failed to add watchdog timer: %m"); @@ -246,6 +257,17 @@ static void service_reset_watchdog(Service *s) { service_start_watchdog(s); } +static void service_reset_watchdog_timeout(Service *s, usec_t watchdog_override_usec) { + assert(s); + + s->watchdog_override_enable = true; + s->watchdog_override_usec = watchdog_override_usec; + service_reset_watchdog(s); + + log_unit_debug(UNIT(s), "watchdog_usec="USEC_FMT, s->watchdog_usec); + log_unit_debug(UNIT(s), "watchdog_override_usec="USEC_FMT, s->watchdog_override_usec); +} + static void service_fd_store_unlink(ServiceFDStore *fs) { if (!fs) @@ -1992,6 +2014,9 @@ static int service_start(Unit *u) { s->notify_state = NOTIFY_UNKNOWN; + s->watchdog_override_enable = false; + s->watchdog_override_usec = 0; + service_enter_start_pre(s); return 1; } @@ -2123,6 +2148,9 @@ static int service_serialize(Unit *u, FILE *f, FDSet *fds) { unit_serialize_item(u, f, "forbid-restart", yes_no(s->forbid_restart)); + if (s->watchdog_override_enable) + unit_serialize_item_format(u, f, "watchdog-override-usec", USEC_FMT, s->watchdog_override_usec); + return 0; } @@ -2317,6 +2345,14 @@ static int service_deserialize_item(Unit *u, const char *key, const char *value, s->stderr_fd = fdset_remove(fds, fd); s->exec_context.stdio_as_fds = true; } + } else if (streq(key, "watchdog-override-usec")) { + usec_t watchdog_override_usec; + if (timestamp_deserialize(value, &watchdog_override_usec) < 0) + log_unit_debug(u, "Failed to parse watchdog_override_usec value: %s", value); + else { + s->watchdog_override_enable = true; + s->watchdog_override_usec = watchdog_override_usec; + } } else log_unit_debug(u, "Unknown serialization key: %s", key); @@ -2895,12 +2931,15 @@ static int service_dispatch_timer(sd_event_source *source, usec_t usec, void *us static int service_dispatch_watchdog(sd_event_source *source, usec_t usec, void *userdata) { Service *s = SERVICE(userdata); char t[FORMAT_TIMESPAN_MAX]; + usec_t watchdog_usec; assert(s); assert(source == s->watchdog_event_source); + watchdog_usec = service_get_watchdog_usec(s); + log_unit_error(UNIT(s), "Watchdog timeout (limit %s)!", - format_timespan(t, sizeof(t), s->watchdog_usec, 1)); + format_timespan(t, sizeof(t), watchdog_usec, 1)); service_enter_signal(s, SERVICE_STOP_SIGABRT, SERVICE_FAILURE_WATCHDOG); @@ -3037,6 +3076,15 @@ static void service_notify_message(Unit *u, pid_t pid, char **tags, FDSet *fds) service_add_fd_store_set(s, fds, name); } + e = strv_find_startswith(tags, "WATCHDOG_USEC="); + if (e) { + usec_t watchdog_override_usec; + if (safe_atou64(e, &watchdog_override_usec) < 0) + log_unit_warning(u, "Failed to parse WATCHDOG_USEC=%s", e); + else + service_reset_watchdog_timeout(s, watchdog_override_usec); + } + /* Notify clients about changed status or main pid */ if (notify_dbus) unit_add_to_dbus_queue(u); diff --git a/src/core/service.h b/src/core/service.h index 4af3d40439..cfef375b03 100644 --- a/src/core/service.h +++ b/src/core/service.h @@ -120,6 +120,8 @@ struct Service { dual_timestamp watchdog_timestamp; usec_t watchdog_usec; + usec_t watchdog_override_usec; + bool watchdog_override_enable; sd_event_source *watchdog_event_source; ExecCommand* exec_command[_SERVICE_EXEC_COMMAND_MAX]; diff --git a/src/core/unit.c b/src/core/unit.c index 581962eba6..0a1a5321df 100644 --- a/src/core/unit.c +++ b/src/core/unit.c @@ -3364,6 +3364,7 @@ int unit_write_drop_in(Unit *u, UnitSetPropertiesMode mode, const char *name, co /* When this is a transient unit file in creation, then let's not create a new drop-in but instead * write to the transient unit file. */ fputs(data, u->transient_file); + fputc('\n', u->transient_file); return 0; } diff --git a/src/fstab-generator/fstab-generator.c b/src/fstab-generator/fstab-generator.c index fc7cf39847..880dc1bd2a 100644 --- a/src/fstab-generator/fstab-generator.c +++ b/src/fstab-generator/fstab-generator.c @@ -502,6 +502,12 @@ static int add_sysroot_mount(void) { return 0; } + if (streq(arg_root_what, "/dev/nfs")) { + /* This is handled by the kernel or the initrd */ + log_debug("Skipping root directory handling, as /dev/nfs was requested."); + return 0; + } + what = fstab_node_to_udev_node(arg_root_what); if (!what) return log_oom(); diff --git a/src/machine/machinectl.c b/src/machine/machinectl.c index 583d2a21e7..5ca557abbf 100644 --- a/src/machine/machinectl.c +++ b/src/machine/machinectl.c @@ -1535,11 +1535,7 @@ static int make_service_name(const char *name, char **ret) { return -EINVAL; } - e = unit_name_escape(name); - if (!e) - return log_oom(); - - r = unit_name_build("systemd-nspawn", e, ".service", ret); + r = unit_name_build("systemd-nspawn", name, ".service", ret); if (r < 0) return log_error_errno(r, "Failed to build unit name: %m"); diff --git a/src/shared/bus-unit-util.c b/src/shared/bus-unit-util.c index 6fc201b885..04471e2373 100644 --- a/src/shared/bus-unit-util.c +++ b/src/shared/bus-unit-util.c @@ -1123,7 +1123,8 @@ static int dump_processes( assert(n == cg->n_children); qsort_safe(children, n, sizeof(struct CGroupInfo*), cgroup_info_compare_func); - n_columns = MAX(LESS_BY(n_columns, 2U), 20U); + if (n_columns != 0) + n_columns = MAX(LESS_BY(n_columns, 2U), 20U); for (i = 0; i < n; i++) { _cleanup_free_ char *pp = NULL; diff --git a/src/basic/fdset.c b/src/shared/fdset.c index 527f27bc67..527f27bc67 100644 --- a/src/basic/fdset.c +++ b/src/shared/fdset.c diff --git a/src/basic/fdset.h b/src/shared/fdset.h index 16efe5bdf2..16efe5bdf2 100644 --- a/src/basic/fdset.h +++ b/src/shared/fdset.h diff --git a/src/systemctl/systemctl.c b/src/systemctl/systemctl.c index 0dfdae4538..38b5a7e082 100644 --- a/src/systemctl/systemctl.c +++ b/src/systemctl/systemctl.c @@ -6093,7 +6093,7 @@ static int get_file_to_edit( return log_oom(); if (arg_runtime) { - run = strjoin(paths->runtime_config, name, NULL); + run = strjoin(paths->runtime_config, "/", name, NULL); if (!run) return log_oom(); } diff --git a/src/systemd/sd-daemon.h b/src/systemd/sd-daemon.h index e6787b0a64..740b176903 100644 --- a/src/systemd/sd-daemon.h +++ b/src/systemd/sd-daemon.h @@ -196,6 +196,11 @@ int sd_is_mq(int fd, const char *path); invocation. This variable is only supported with sd_pid_notify_with_fds(). + WATCHDOG_USEC=... + Reset watchdog_usec value during runtime. + To reset watchdog_usec value, start the service again. + Example: "WATCHDOG_USEC=20000000" + Daemons can choose to send additional variables. However, it is recommended to prefix variable names not listed above with X_. |