summaryrefslogtreecommitdiff
path: root/src/core
diff options
context:
space:
mode:
Diffstat (limited to 'src/core')
-rw-r--r--src/core/automount.c6
-rw-r--r--src/core/cgroup.c11
-rw-r--r--src/core/cgroup.h1
-rw-r--r--src/core/dbus-cgroup.c35
-rw-r--r--src/core/dbus-execute.c25
-rw-r--r--src/core/dbus-scope.c2
-rw-r--r--src/core/execute.c46
-rw-r--r--src/core/execute.h10
-rw-r--r--src/core/killall.c7
-rw-r--r--src/core/load-fragment-gperf.gperf.m49
-rw-r--r--src/core/load-fragment.c27
-rw-r--r--src/core/machine-id-setup.c226
-rw-r--r--src/core/machine-id-setup.h2
-rw-r--r--src/core/macros.systemd.in8
-rw-r--r--src/core/main.c32
-rw-r--r--src/core/manager.c8
-rw-r--r--src/core/mount-setup.c14
-rw-r--r--src/core/namespace.c51
-rw-r--r--src/core/namespace.h6
-rw-r--r--src/core/scope.c16
-rw-r--r--src/core/scope.h3
-rw-r--r--src/core/selinux-access.c2
-rw-r--r--src/core/selinux-setup.c2
-rw-r--r--src/core/service.c2
-rw-r--r--src/core/shutdown.c20
-rw-r--r--src/core/system.conf2
-rw-r--r--src/core/transaction.c7
-rw-r--r--src/core/unit.c104
-rw-r--r--src/core/unit.h1
29 files changed, 365 insertions, 320 deletions
diff --git a/src/core/automount.c b/src/core/automount.c
index 85803a9c4a..4e9891569c 100644
--- a/src/core/automount.c
+++ b/src/core/automount.c
@@ -98,9 +98,6 @@ static void unmount_autofs(Automount *a) {
if (a->pipe_fd < 0)
return;
- automount_send_ready(a, a->tokens, -EHOSTDOWN);
- automount_send_ready(a, a->expire_tokens, -EHOSTDOWN);
-
a->pipe_event_source = sd_event_source_unref(a->pipe_event_source);
a->pipe_fd = safe_close(a->pipe_fd);
@@ -109,6 +106,9 @@ static void unmount_autofs(Automount *a) {
if (a->where &&
(UNIT(a)->manager->exit_code != MANAGER_RELOAD &&
UNIT(a)->manager->exit_code != MANAGER_REEXECUTE)) {
+ automount_send_ready(a, a->tokens, -EHOSTDOWN);
+ automount_send_ready(a, a->expire_tokens, -EHOSTDOWN);
+
r = repeat_unmount(a->where, MNT_DETACH);
if (r < 0)
log_error_errno(r, "Failed to unmount: %m");
diff --git a/src/core/cgroup.c b/src/core/cgroup.c
index 6e36e6b340..c19e43f571 100644
--- a/src/core/cgroup.c
+++ b/src/core/cgroup.c
@@ -800,7 +800,10 @@ static void cgroup_context_apply(Unit *u, CGroupMask mask, ManagerState state) {
"/dev/random\0" "rwm\0"
"/dev/urandom\0" "rwm\0"
"/dev/tty\0" "rwm\0"
- "/dev/pts/ptmx\0" "rw\0"; /* /dev/pts/ptmx may not be duplicated, but accessed */
+ "/dev/pts/ptmx\0" "rw\0" /* /dev/pts/ptmx may not be duplicated, but accessed */
+ /* Allow /run/systemd/inaccessible/{chr,blk} devices for mapping InaccessiblePaths */
+ "/run/systemd/inaccessible/chr\0" "rwm\0"
+ "/run/systemd/inaccessible/blk\0" "rwm\0";
const char *x, *y;
@@ -1136,7 +1139,7 @@ int unit_watch_cgroup(Unit *u) {
/* Only applies to the unified hierarchy */
r = cg_unified();
if (r < 0)
- return log_unit_error_errno(u, r, "Failed detect wether the unified hierarchy is used: %m");
+ return log_unit_error_errno(u, r, "Failed detect whether the unified hierarchy is used: %m");
if (r == 0)
return 0;
@@ -1658,7 +1661,7 @@ int manager_setup_cgroup(Manager *m) {
/* 3. Install agent */
if (unified) {
- /* In the unified hierarchy we can can get
+ /* In the unified hierarchy we can get
* cgroup empty notifications via inotify. */
m->cgroup_inotify_event_source = sd_event_source_unref(m->cgroup_inotify_event_source);
@@ -1705,7 +1708,7 @@ int manager_setup_cgroup(Manager *m) {
/* also, move all other userspace processes remaining
* in the root cgroup into that scope. */
- r = cg_migrate(SYSTEMD_CGROUP_CONTROLLER, m->cgroup_root, SYSTEMD_CGROUP_CONTROLLER, scope_path, false);
+ r = cg_migrate(SYSTEMD_CGROUP_CONTROLLER, m->cgroup_root, SYSTEMD_CGROUP_CONTROLLER, scope_path, 0);
if (r < 0)
log_warning_errno(r, "Couldn't move remaining userspace processes, ignoring: %m");
diff --git a/src/core/cgroup.h b/src/core/cgroup.h
index f21409bd5d..a57403e79f 100644
--- a/src/core/cgroup.h
+++ b/src/core/cgroup.h
@@ -119,7 +119,6 @@ struct CGroupContext {
bool delegate;
};
-#include "cgroup-util.h"
#include "unit.h"
void cgroup_context_init(CGroupContext *c);
diff --git a/src/core/dbus-cgroup.c b/src/core/dbus-cgroup.c
index 27bbe2d26d..85b0c86a2f 100644
--- a/src/core/dbus-cgroup.c
+++ b/src/core/dbus-cgroup.c
@@ -856,7 +856,7 @@ int bus_cgroup_set_property(
return 1;
- } else if (STR_IN_SET(name, "MemoryLowByPhysicalMemory", "MemoryHighByPhysicalMemory", "MemoryMaxByPhysicalMemory")) {
+ } else if (STR_IN_SET(name, "MemoryLowScale", "MemoryHighScale", "MemoryMaxScale")) {
uint32_t raw;
uint64_t v;
@@ -872,7 +872,7 @@ int bus_cgroup_set_property(
const char *e;
/* Chop off suffix */
- assert_se(e = endswith(name, "ByPhysicalMemory"));
+ assert_se(e = endswith(name, "Scale"));
name = strndupa(name, e - name);
if (streq(name, "MemoryLow"))
@@ -883,7 +883,8 @@ int bus_cgroup_set_property(
c->memory_max = v;
unit_invalidate_cgroup(u, CGROUP_MASK_MEMORY);
- unit_write_drop_in_private_format(u, mode, name, "%s=%" PRIu32 "%%", name, (uint32_t) (DIV_ROUND_UP((uint64_t) raw * 100, (uint64_t) UINT32_MAX)));
+ unit_write_drop_in_private_format(u, mode, name, "%s=%" PRIu32 "%%", name,
+ (uint32_t) (DIV_ROUND_UP((uint64_t) raw * 100U, (uint64_t) UINT32_MAX)));
}
return 1;
@@ -909,7 +910,7 @@ int bus_cgroup_set_property(
return 1;
- } else if (streq(name, "MemoryLimitByPhysicalMemory")) {
+ } else if (streq(name, "MemoryLimitScale")) {
uint64_t limit;
uint32_t raw;
@@ -924,7 +925,8 @@ int bus_cgroup_set_property(
if (mode != UNIT_CHECK) {
c->memory_limit = limit;
unit_invalidate_cgroup(u, CGROUP_MASK_MEMORY);
- unit_write_drop_in_private_format(u, mode, "MemoryLimit", "MemoryLimit=%" PRIu32 "%%", (uint32_t) (DIV_ROUND_UP((uint64_t) raw * 100, (uint64_t) UINT32_MAX)));
+ unit_write_drop_in_private_format(u, mode, "MemoryLimit", "MemoryLimit=%" PRIu32 "%%",
+ (uint32_t) (DIV_ROUND_UP((uint64_t) raw * 100U, (uint64_t) UINT32_MAX)));
}
return 1;
@@ -960,6 +962,7 @@ int bus_cgroup_set_property(
while ((r = sd_bus_message_read(message, "(ss)", &path, &rwm)) > 0) {
if ((!startswith(path, "/dev/") &&
+ !startswith(path, "/run/systemd/inaccessible/") &&
!startswith(path, "block-") &&
!startswith(path, "char-")) ||
strpbrk(path, WHITESPACE))
@@ -1059,6 +1062,8 @@ int bus_cgroup_set_property(
r = sd_bus_message_read(message, "t", &limit);
if (r < 0)
return r;
+ if (limit <= 0)
+ return sd_bus_error_set_errnof(error, EINVAL, "%s= is too small", name);
if (mode != UNIT_CHECK) {
c->tasks_max = limit;
@@ -1071,6 +1076,26 @@ int bus_cgroup_set_property(
}
return 1;
+ } else if (streq(name, "TasksMaxScale")) {
+ uint64_t limit;
+ uint32_t raw;
+
+ r = sd_bus_message_read(message, "u", &raw);
+ if (r < 0)
+ return r;
+
+ limit = system_tasks_max_scale(raw, UINT32_MAX);
+ if (limit <= 0 || limit >= UINT64_MAX)
+ return sd_bus_error_set_errnof(error, EINVAL, "%s= is out of range", name);
+
+ if (mode != UNIT_CHECK) {
+ c->tasks_max = limit;
+ unit_invalidate_cgroup(u, CGROUP_MASK_PIDS);
+ unit_write_drop_in_private_format(u, mode, name, "TasksMax=%" PRIu32 "%%",
+ (uint32_t) (DIV_ROUND_UP((uint64_t) raw * 100U, (uint64_t) UINT32_MAX)));
+ }
+
+ return 1;
}
if (u->transient && u->load_state == UNIT_STUB) {
diff --git a/src/core/dbus-execute.c b/src/core/dbus-execute.c
index 644b9561b5..307c3d8e7a 100644
--- a/src/core/dbus-execute.c
+++ b/src/core/dbus-execute.c
@@ -695,9 +695,12 @@ const sd_bus_vtable bus_exec_vtable[] = {
SD_BUS_PROPERTY("Group", "s", NULL, offsetof(ExecContext, group), SD_BUS_VTABLE_PROPERTY_CONST),
SD_BUS_PROPERTY("SupplementaryGroups", "as", NULL, offsetof(ExecContext, supplementary_groups), SD_BUS_VTABLE_PROPERTY_CONST),
SD_BUS_PROPERTY("PAMName", "s", NULL, offsetof(ExecContext, pam_name), SD_BUS_VTABLE_PROPERTY_CONST),
- SD_BUS_PROPERTY("ReadWriteDirectories", "as", NULL, offsetof(ExecContext, read_write_dirs), SD_BUS_VTABLE_PROPERTY_CONST),
- SD_BUS_PROPERTY("ReadOnlyDirectories", "as", NULL, offsetof(ExecContext, read_only_dirs), SD_BUS_VTABLE_PROPERTY_CONST),
- SD_BUS_PROPERTY("InaccessibleDirectories", "as", NULL, offsetof(ExecContext, inaccessible_dirs), SD_BUS_VTABLE_PROPERTY_CONST),
+ SD_BUS_PROPERTY("ReadWriteDirectories", "as", NULL, offsetof(ExecContext, read_write_paths), SD_BUS_VTABLE_PROPERTY_CONST|SD_BUS_VTABLE_HIDDEN),
+ SD_BUS_PROPERTY("ReadOnlyDirectories", "as", NULL, offsetof(ExecContext, read_only_paths), SD_BUS_VTABLE_PROPERTY_CONST|SD_BUS_VTABLE_HIDDEN),
+ SD_BUS_PROPERTY("InaccessibleDirectories", "as", NULL, offsetof(ExecContext, inaccessible_paths), SD_BUS_VTABLE_PROPERTY_CONST|SD_BUS_VTABLE_HIDDEN),
+ SD_BUS_PROPERTY("ReadWritePaths", "as", NULL, offsetof(ExecContext, read_write_paths), SD_BUS_VTABLE_PROPERTY_CONST),
+ SD_BUS_PROPERTY("ReadOnlyPaths", "as", NULL, offsetof(ExecContext, read_only_paths), SD_BUS_VTABLE_PROPERTY_CONST),
+ SD_BUS_PROPERTY("InaccessiblePaths", "as", NULL, offsetof(ExecContext, inaccessible_paths), SD_BUS_VTABLE_PROPERTY_CONST),
SD_BUS_PROPERTY("MountFlags", "t", bus_property_get_ulong, offsetof(ExecContext, mount_flags), SD_BUS_VTABLE_PROPERTY_CONST),
SD_BUS_PROPERTY("PrivateTmp", "b", bus_property_get_bool, offsetof(ExecContext, private_tmp), SD_BUS_VTABLE_PROPERTY_CONST),
SD_BUS_PROPERTY("PrivateNetwork", "b", bus_property_get_bool, offsetof(ExecContext, private_network), SD_BUS_VTABLE_PROPERTY_CONST),
@@ -1323,8 +1326,8 @@ int bus_exec_context_set_transient_property(
return 1;
- } else if (STR_IN_SET(name, "ReadWriteDirectories", "ReadOnlyDirectories", "InaccessibleDirectories")) {
-
+ } else if (STR_IN_SET(name, "ReadWriteDirectories", "ReadOnlyDirectories", "InaccessibleDirectories",
+ "ReadWritePaths", "ReadOnlyPaths", "InaccessiblePaths")) {
_cleanup_strv_free_ char **l = NULL;
char ***dirs;
char **p;
@@ -1346,12 +1349,12 @@ int bus_exec_context_set_transient_property(
if (mode != UNIT_CHECK) {
_cleanup_free_ char *joined = NULL;
- if (streq(name, "ReadWriteDirectories"))
- dirs = &c->read_write_dirs;
- else if (streq(name, "ReadOnlyDirectories"))
- dirs = &c->read_only_dirs;
- else /* "InaccessibleDirectories" */
- dirs = &c->inaccessible_dirs;
+ if (STR_IN_SET(name, "ReadWriteDirectories", "ReadWritePaths"))
+ dirs = &c->read_write_paths;
+ else if (STR_IN_SET(name, "ReadOnlyDirectories", "ReadOnlyPaths"))
+ dirs = &c->read_only_paths;
+ else /* "InaccessiblePaths" */
+ dirs = &c->inaccessible_paths;
if (strv_length(l) == 0) {
*dirs = strv_free(*dirs);
diff --git a/src/core/dbus-scope.c b/src/core/dbus-scope.c
index f557eedfc3..1abaf9f658 100644
--- a/src/core/dbus-scope.c
+++ b/src/core/dbus-scope.c
@@ -225,5 +225,5 @@ int bus_scope_send_request_stop(Scope *s) {
if (r < 0)
return r;
- return sd_bus_send_to(UNIT(s)->manager->api_bus, m, /* s->controller */ NULL, NULL);
+ return sd_bus_send_to(UNIT(s)->manager->api_bus, m, s->controller, NULL);
}
diff --git a/src/core/execute.c b/src/core/execute.c
index 8c487b371f..7c178b97c3 100644
--- a/src/core/execute.c
+++ b/src/core/execute.c
@@ -290,10 +290,10 @@ static int connect_journal_socket(int fd, uid_t uid, gid_t gid) {
}
static int connect_logger_as(
+ Unit *unit,
const ExecContext *context,
ExecOutput output,
const char *ident,
- const char *unit_id,
int nfd,
uid_t uid,
gid_t gid) {
@@ -329,7 +329,7 @@ static int connect_logger_as(
"%i\n"
"%i\n",
context->syslog_identifier ? context->syslog_identifier : ident,
- unit_id,
+ unit->id,
context->syslog_priority,
!!context->syslog_level_prefix,
output == EXEC_OUTPUT_SYSLOG || output == EXEC_OUTPUT_SYSLOG_AND_CONSOLE,
@@ -544,7 +544,7 @@ static int setup_output(
case EXEC_OUTPUT_KMSG_AND_CONSOLE:
case EXEC_OUTPUT_JOURNAL:
case EXEC_OUTPUT_JOURNAL_AND_CONSOLE:
- r = connect_logger_as(context, o, ident, unit->id, fileno, uid, gid);
+ r = connect_logger_as(unit, context, o, ident, fileno, uid, gid);
if (r < 0) {
log_unit_error_errno(unit, r, "Failed to connect %s to the journal socket, ignoring: %m", fileno == STDOUT_FILENO ? "stdout" : "stderr");
r = open_null_as(O_WRONLY, fileno);
@@ -1507,9 +1507,9 @@ static bool exec_needs_mount_namespace(
assert(context);
assert(params);
- if (!strv_isempty(context->read_write_dirs) ||
- !strv_isempty(context->read_only_dirs) ||
- !strv_isempty(context->inaccessible_dirs))
+ if (!strv_isempty(context->read_write_paths) ||
+ !strv_isempty(context->read_only_paths) ||
+ !strv_isempty(context->inaccessible_paths))
return true;
if (context->mount_flags != 0)
@@ -1933,9 +1933,9 @@ static int exec_child(
r = setup_namespace(
params->apply_chroot ? context->root_directory : NULL,
- context->read_write_dirs,
- context->read_only_dirs,
- context->inaccessible_dirs,
+ context->read_write_paths,
+ context->read_only_paths,
+ context->inaccessible_paths,
tmp,
var,
context->private_devices,
@@ -2324,9 +2324,9 @@ void exec_context_done(ExecContext *c) {
c->pam_name = mfree(c->pam_name);
- c->read_only_dirs = strv_free(c->read_only_dirs);
- c->read_write_dirs = strv_free(c->read_write_dirs);
- c->inaccessible_dirs = strv_free(c->inaccessible_dirs);
+ c->read_only_paths = strv_free(c->read_only_paths);
+ c->read_write_paths = strv_free(c->read_write_paths);
+ c->inaccessible_paths = strv_free(c->inaccessible_paths);
if (c->cpuset)
CPU_FREE(c->cpuset);
@@ -2732,21 +2732,21 @@ void exec_context_dump(ExecContext *c, FILE* f, const char *prefix) {
if (c->pam_name)
fprintf(f, "%sPAMName: %s\n", prefix, c->pam_name);
- if (strv_length(c->read_write_dirs) > 0) {
- fprintf(f, "%sReadWriteDirs:", prefix);
- strv_fprintf(f, c->read_write_dirs);
+ if (strv_length(c->read_write_paths) > 0) {
+ fprintf(f, "%sReadWritePaths:", prefix);
+ strv_fprintf(f, c->read_write_paths);
fputs("\n", f);
}
- if (strv_length(c->read_only_dirs) > 0) {
- fprintf(f, "%sReadOnlyDirs:", prefix);
- strv_fprintf(f, c->read_only_dirs);
+ if (strv_length(c->read_only_paths) > 0) {
+ fprintf(f, "%sReadOnlyPaths:", prefix);
+ strv_fprintf(f, c->read_only_paths);
fputs("\n", f);
}
- if (strv_length(c->inaccessible_dirs) > 0) {
- fprintf(f, "%sInaccessibleDirs:", prefix);
- strv_fprintf(f, c->inaccessible_dirs);
+ if (strv_length(c->inaccessible_paths) > 0) {
+ fprintf(f, "%sInaccessiblePaths:", prefix);
+ strv_fprintf(f, c->inaccessible_paths);
fputs("\n", f);
}
@@ -2827,7 +2827,7 @@ void exec_context_dump(ExecContext *c, FILE* f, const char *prefix) {
bool exec_context_maintains_privileges(ExecContext *c) {
assert(c);
- /* Returns true if the process forked off would run run under
+ /* Returns true if the process forked off would run under
* an unchanged UID or as root. */
if (!c->user)
@@ -3062,7 +3062,7 @@ int exec_runtime_make(ExecRuntime **rt, ExecContext *c, const char *id) {
return r;
if (c->private_network && (*rt)->netns_storage_socket[0] < 0) {
- if (socketpair(AF_UNIX, SOCK_DGRAM, 0, (*rt)->netns_storage_socket) < 0)
+ if (socketpair(AF_UNIX, SOCK_DGRAM|SOCK_CLOEXEC, 0, (*rt)->netns_storage_socket) < 0)
return -errno;
}
diff --git a/src/core/execute.h b/src/core/execute.h
index 210eea0e82..189c4d0999 100644
--- a/src/core/execute.h
+++ b/src/core/execute.h
@@ -30,6 +30,7 @@ typedef struct ExecParameters ExecParameters;
#include <stdio.h>
#include <sys/capability.h>
+#include "cgroup-util.h"
#include "fdset.h"
#include "list.h"
#include "missing.h"
@@ -130,7 +131,7 @@ struct ExecContext {
bool ignore_sigpipe;
- /* Since resolving these names might might involve socket
+ /* Since resolving these names might involve socket
* connections and we don't want to deadlock ourselves these
* names are resolved on execution only and in the child
* process. */
@@ -152,7 +153,7 @@ struct ExecContext {
bool smack_process_label_ignore;
char *smack_process_label;
- char **read_write_dirs, **read_only_dirs, **inaccessible_dirs;
+ char **read_write_paths, **read_only_paths, **inaccessible_paths;
unsigned long mount_flags;
uint64_t capability_bounding_set;
@@ -203,9 +204,6 @@ struct ExecContext {
bool no_new_privileges_set:1;
};
-#include "cgroup-util.h"
-#include "cgroup.h"
-
struct ExecParameters {
char **argv;
char **environment;
@@ -236,6 +234,8 @@ struct ExecParameters {
int stderr_fd;
};
+#include "unit.h"
+
int exec_spawn(Unit *unit,
ExecCommand *command,
const ExecContext *context,
diff --git a/src/core/killall.c b/src/core/killall.c
index 09378f7085..a8b814e868 100644
--- a/src/core/killall.c
+++ b/src/core/killall.c
@@ -23,6 +23,7 @@
#include <unistd.h>
#include "alloc-util.h"
+#include "def.h"
#include "fd-util.h"
#include "formats-util.h"
#include "killall.h"
@@ -33,8 +34,6 @@
#include "terminal-util.h"
#include "util.h"
-#define TIMEOUT_USEC (10 * USEC_PER_SEC)
-
static bool ignore_proc(pid_t pid, bool warn_rootfs) {
_cleanup_fclose_ FILE *f = NULL;
char c;
@@ -80,7 +79,7 @@ static bool ignore_proc(pid_t pid, bool warn_rootfs) {
get_process_comm(pid, &comm);
if (r)
- log_notice("Process " PID_FMT " (%s) has been been marked to be excluded from killing. It is "
+ log_notice("Process " PID_FMT " (%s) has been marked to be excluded from killing. It is "
"running from the root file system, and thus likely to block re-mounting of the "
"root file system to read-only. Please consider moving it into an initrd file "
"system instead.", pid, strna(comm));
@@ -99,7 +98,7 @@ static void wait_for_children(Set *pids, sigset_t *mask) {
if (set_isempty(pids))
return;
- until = now(CLOCK_MONOTONIC) + TIMEOUT_USEC;
+ until = now(CLOCK_MONOTONIC) + DEFAULT_TIMEOUT_USEC;
for (;;) {
struct timespec ts;
int k;
diff --git a/src/core/load-fragment-gperf.gperf.m4 b/src/core/load-fragment-gperf.gperf.m4
index fe1006830b..6a5c16a000 100644
--- a/src/core/load-fragment-gperf.gperf.m4
+++ b/src/core/load-fragment-gperf.gperf.m4
@@ -80,9 +80,12 @@ $1.LimitMSGQUEUE, config_parse_limit, RLIMIT_MSGQ
$1.LimitNICE, config_parse_limit, RLIMIT_NICE, offsetof($1, exec_context.rlimit)
$1.LimitRTPRIO, config_parse_limit, RLIMIT_RTPRIO, offsetof($1, exec_context.rlimit)
$1.LimitRTTIME, config_parse_limit, RLIMIT_RTTIME, offsetof($1, exec_context.rlimit)
-$1.ReadWriteDirectories, config_parse_namespace_path_strv, 0, offsetof($1, exec_context.read_write_dirs)
-$1.ReadOnlyDirectories, config_parse_namespace_path_strv, 0, offsetof($1, exec_context.read_only_dirs)
-$1.InaccessibleDirectories, config_parse_namespace_path_strv, 0, offsetof($1, exec_context.inaccessible_dirs)
+$1.ReadWriteDirectories, config_parse_namespace_path_strv, 0, offsetof($1, exec_context.read_write_paths)
+$1.ReadOnlyDirectories, config_parse_namespace_path_strv, 0, offsetof($1, exec_context.read_only_paths)
+$1.InaccessibleDirectories, config_parse_namespace_path_strv, 0, offsetof($1, exec_context.inaccessible_paths)
+$1.ReadWritePaths, config_parse_namespace_path_strv, 0, offsetof($1, exec_context.read_write_paths)
+$1.ReadOnlyPaths, config_parse_namespace_path_strv, 0, offsetof($1, exec_context.read_only_paths)
+$1.InaccessiblePaths, config_parse_namespace_path_strv, 0, offsetof($1, exec_context.inaccessible_paths)
$1.PrivateTmp, config_parse_bool, 0, offsetof($1, exec_context.private_tmp)
$1.PrivateNetwork, config_parse_bool, 0, offsetof($1, exec_context.private_network)
$1.PrivateDevices, config_parse_bool, 0, offsetof($1, exec_context.private_devices)
diff --git a/src/core/load-fragment.c b/src/core/load-fragment.c
index 8295cf45a6..a36953f766 100644
--- a/src/core/load-fragment.c
+++ b/src/core/load-fragment.c
@@ -620,7 +620,7 @@ int config_parse_exec(
ignore = true;
else if (*f == '@' && !separate_argv0)
separate_argv0 = true;
- else if (*f == '!' && !privileged)
+ else if (*f == '+' && !privileged)
privileged = true;
else
break;
@@ -2429,7 +2429,7 @@ static int syscall_filter_parse_one(
int id;
id = seccomp_syscall_resolve_name(t);
- if (id < 0) {
+ if (id == __NR_SCMP_ERROR) {
if (warn)
log_syntax(unit, LOG_ERR, filename, line, 0, "Failed to parse system call, ignoring: %s", t);
return 0;
@@ -2823,8 +2823,8 @@ int config_parse_memory_limit(
} else
bytes = physical_memory_scale(r, 100U);
- if (bytes < 1) {
- log_syntax(unit, LOG_ERR, filename, line, 0, "Memory limit '%s' too small. Ignoring.", rvalue);
+ if (bytes <= 0 || bytes >= UINT64_MAX) {
+ log_syntax(unit, LOG_ERR, filename, line, 0, "Memory limit '%s' out of range. Ignoring.", rvalue);
return 0;
}
}
@@ -2861,9 +2861,18 @@ int config_parse_tasks_max(
return 0;
}
- r = safe_atou64(rvalue, &u);
- if (r < 0 || u < 1) {
- log_syntax(unit, LOG_ERR, filename, line, r, "Maximum tasks value '%s' invalid. Ignoring.", rvalue);
+ r = parse_percent(rvalue);
+ if (r < 0) {
+ r = safe_atou64(rvalue, &u);
+ if (r < 0) {
+ log_syntax(unit, LOG_ERR, filename, line, r, "Maximum tasks value '%s' invalid. Ignoring.", rvalue);
+ return 0;
+ }
+ } else
+ u = system_tasks_max_scale(r, 100U);
+
+ if (u <= 0 || u >= UINT64_MAX) {
+ log_syntax(unit, LOG_ERR, filename, line, 0, "Maximum tasks value '%s' out of range. Ignoring.", rvalue);
return 0;
}
@@ -3594,7 +3603,7 @@ int config_parse_protect_home(
assert(data);
/* Our enum shall be a superset of booleans, hence first try
- * to parse as as boolean, and then as enum */
+ * to parse as boolean, and then as enum */
k = parse_boolean(rvalue);
if (k > 0)
@@ -3637,7 +3646,7 @@ int config_parse_protect_system(
assert(data);
/* Our enum shall be a superset of booleans, hence first try
- * to parse as as boolean, and then as enum */
+ * to parse as boolean, and then as enum */
k = parse_boolean(rvalue);
if (k > 0)
diff --git a/src/core/machine-id-setup.c b/src/core/machine-id-setup.c
index 0145fe2894..76dfcfa6d7 100644
--- a/src/core/machine-id-setup.c
+++ b/src/core/machine-id-setup.c
@@ -17,11 +17,8 @@
along with systemd; If not, see <http://www.gnu.org/licenses/>.
***/
-#include <errno.h>
#include <fcntl.h>
#include <sched.h>
-#include <stdio.h>
-#include <string.h>
#include <sys/mount.h>
#include <unistd.h>
@@ -29,10 +26,8 @@
#include "alloc-util.h"
#include "fd-util.h"
-#include "fileio.h"
#include "fs-util.h"
-#include "hexdecoct.h"
-#include "io-util.h"
+#include "id128-util.h"
#include "log.h"
#include "machine-id-setup.h"
#include "macro.h"
@@ -46,101 +41,23 @@
#include "util.h"
#include "virt.h"
-static int shorten_uuid(char destination[34], const char source[36]) {
- unsigned i, j;
-
- assert(destination);
- assert(source);
-
- /* Converts a UUID into a machine ID, by lowercasing it and
- * removing dashes. Validates everything. */
-
- for (i = 0, j = 0; i < 36 && j < 32; i++) {
- int t;
-
- t = unhexchar(source[i]);
- if (t < 0)
- continue;
-
- destination[j++] = hexchar(t);
- }
-
- if (i != 36 || j != 32)
- return -EINVAL;
-
- destination[32] = '\n';
- destination[33] = 0;
- return 0;
-}
-
-static int read_machine_id(int fd, char id[34]) {
- char id_to_validate[34];
- int r;
-
- assert(fd >= 0);
- assert(id);
-
- /* Reads a machine ID from a file, validates it, and returns
- * it. The returned ID ends in a newline. */
-
- r = loop_read_exact(fd, id_to_validate, 33, false);
- if (r < 0)
- return r;
-
- if (id_to_validate[32] != '\n')
- return -EINVAL;
-
- id_to_validate[32] = 0;
-
- if (!id128_is_valid(id_to_validate))
- return -EINVAL;
-
- memcpy(id, id_to_validate, 32);
- id[32] = '\n';
- id[33] = 0;
- return 0;
-}
-
-static int write_machine_id(int fd, const char id[34]) {
- int r;
-
- assert(fd >= 0);
- assert(id);
-
- if (lseek(fd, 0, SEEK_SET) < 0)
- return -errno;
-
- r = loop_write(fd, id, 33, false);
- if (r < 0)
- return r;
-
- if (fsync(fd) < 0)
- return -errno;
-
- return 0;
-}
-
-static int generate_machine_id(char id[34], const char *root) {
- int fd, r;
- unsigned char *p;
- sd_id128_t buf;
- char *q;
+static int generate_machine_id(const char *root, sd_id128_t *ret) {
const char *dbus_machine_id;
+ _cleanup_close_ int fd = -1;
+ int r;
- assert(id);
-
- dbus_machine_id = prefix_roota(root, "/var/lib/dbus/machine-id");
+ assert(ret);
/* First, try reading the D-Bus machine id, unless it is a symlink */
+ dbus_machine_id = prefix_roota(root, "/var/lib/dbus/machine-id");
fd = open(dbus_machine_id, O_RDONLY|O_CLOEXEC|O_NOCTTY|O_NOFOLLOW);
if (fd >= 0) {
- r = read_machine_id(fd, id);
- safe_close(fd);
-
- if (r >= 0) {
+ if (id128_read_fd(fd, ID128_PLAIN, ret) >= 0) {
log_info("Initializing machine ID from D-Bus machine ID.");
return 0;
}
+
+ fd = safe_close(fd);
}
if (isempty(root)) {
@@ -151,13 +68,10 @@ static int generate_machine_id(char id[34], const char *root) {
if (detect_container() > 0) {
_cleanup_free_ char *e = NULL;
- r = getenv_for_pid(1, "container_uuid", &e);
- if (r > 0) {
- r = shorten_uuid(id, e);
- if (r >= 0) {
- log_info("Initializing machine ID from container UUID.");
- return 0;
- }
+ if (getenv_for_pid(1, "container_uuid", &e) > 0 &&
+ sd_id128_from_string(e, ret) >= 0) {
+ log_info("Initializing machine ID from container UUID.");
+ return 0;
}
} else if (detect_vm() == VIRTUALIZATION_KVM) {
@@ -166,51 +80,29 @@ static int generate_machine_id(char id[34], const char *root) {
* running in qemu/kvm and a machine ID was passed in
* via -uuid on the qemu/kvm command line */
- char uuid[36];
-
- fd = open("/sys/class/dmi/id/product_uuid", O_RDONLY|O_CLOEXEC|O_NOCTTY|O_NOFOLLOW);
- if (fd >= 0) {
- r = loop_read_exact(fd, uuid, 36, false);
- safe_close(fd);
-
- if (r >= 0) {
- r = shorten_uuid(id, uuid);
- if (r >= 0) {
- log_info("Initializing machine ID from KVM UUID.");
- return 0;
- }
- }
+ if (id128_read("/sys/class/dmi/id/product_uuid", ID128_UUID, ret) >= 0) {
+ log_info("Initializing machine ID from KVM UUID.");
+ return 0;
}
}
}
/* If that didn't work, generate a random machine id */
- r = sd_id128_randomize(&buf);
+ r = sd_id128_randomize(ret);
if (r < 0)
- return log_error_errno(r, "Failed to open /dev/urandom: %m");
-
- for (p = buf.bytes, q = id; p < buf.bytes + sizeof(buf); p++, q += 2) {
- q[0] = hexchar(*p >> 4);
- q[1] = hexchar(*p & 15);
- }
-
- id[32] = '\n';
- id[33] = 0;
+ return log_error_errno(r, "Failed to generate randomized : %m");
log_info("Initializing machine ID from random generator.");
-
return 0;
}
-int machine_id_setup(const char *root, sd_id128_t machine_id) {
+int machine_id_setup(const char *root, sd_id128_t machine_id, sd_id128_t *ret) {
const char *etc_machine_id, *run_machine_id;
_cleanup_close_ int fd = -1;
- bool writable = true;
- char id[34]; /* 32 + \n + \0 */
+ bool writable;
int r;
etc_machine_id = prefix_roota(root, "/etc/machine-id");
- run_machine_id = prefix_roota(root, "/run/machine-id");
RUN_WITH_UMASK(0000) {
/* We create this 0444, to indicate that this isn't really
@@ -218,7 +110,7 @@ int machine_id_setup(const char *root, sd_id128_t machine_id) {
* will be owned by root it doesn't matter much, but maybe
* people look. */
- mkdir_parents(etc_machine_id, 0755);
+ (void) mkdir_parents(etc_machine_id, 0755);
fd = open(etc_machine_id, O_RDWR|O_CREAT|O_CLOEXEC|O_NOCTTY, 0444);
if (fd < 0) {
int old_errno = errno;
@@ -239,41 +131,41 @@ int machine_id_setup(const char *root, sd_id128_t machine_id) {
}
writable = false;
- }
+ } else
+ writable = true;
}
- /* A machine id argument overrides all other machined-ids */
- if (!sd_id128_is_null(machine_id)) {
- sd_id128_to_string(machine_id, id);
- id[32] = '\n';
- id[33] = 0;
- } else {
- if (read_machine_id(fd, id) >= 0)
- return 0;
+ /* A we got a valid machine ID argument, that's what counts */
+ if (sd_id128_is_null(machine_id)) {
- /* Hmm, so, the id currently stored is not useful, then let's
- * generate one */
+ /* Try to read any existing machine ID */
+ if (id128_read_fd(fd, ID128_PLAIN, ret) >= 0)
+ return 0;
- r = generate_machine_id(id, root);
+ /* Hmm, so, the id currently stored is not useful, then let's generate one */
+ r = generate_machine_id(root, &machine_id);
if (r < 0)
return r;
+
+ if (lseek(fd, 0, SEEK_SET) == (off_t) -1)
+ return log_error_errno(errno, "Failed to seek: %m");
}
if (writable)
- if (write_machine_id(fd, id) >= 0)
- return 0;
+ if (id128_write_fd(fd, ID128_PLAIN, machine_id, true) >= 0)
+ goto finish;
fd = safe_close(fd);
- /* Hmm, we couldn't write it? So let's write it to
- * /run/machine-id as a replacement */
+ /* Hmm, we couldn't write it? So let's write it to /run/machine-id as a replacement */
- RUN_WITH_UMASK(0022) {
- r = write_string_file(run_machine_id, id, WRITE_STRING_FILE_CREATE);
- if (r < 0) {
- (void) unlink(run_machine_id);
- return log_error_errno(r, "Cannot write %s: %m", run_machine_id);
- }
+ run_machine_id = prefix_roota(root, "/run/machine-id");
+
+ RUN_WITH_UMASK(0022)
+ r = id128_write(run_machine_id, ID128_PLAIN, machine_id, false);
+ if (r < 0) {
+ (void) unlink(run_machine_id);
+ return log_error_errno(r, "Cannot write %s: %m", run_machine_id);
}
/* And now, let's mount it over */
@@ -286,7 +178,11 @@ int machine_id_setup(const char *root, sd_id128_t machine_id) {
/* Mark the mount read-only */
if (mount(NULL, etc_machine_id, NULL, MS_BIND|MS_RDONLY|MS_REMOUNT, NULL) < 0)
- log_warning_errno(errno, "Failed to make transient %s read-only: %m", etc_machine_id);
+ log_warning_errno(errno, "Failed to make transient %s read-only, ignoring: %m", etc_machine_id);
+
+finish:
+ if (ret)
+ *ret = machine_id;
return 0;
}
@@ -294,16 +190,20 @@ int machine_id_setup(const char *root, sd_id128_t machine_id) {
int machine_id_commit(const char *root) {
_cleanup_close_ int fd = -1, initial_mntns_fd = -1;
const char *etc_machine_id;
- char id[34]; /* 32 + \n + \0 */
+ sd_id128_t id;
int r;
+ /* Replaces a tmpfs bind mount of /etc/machine-id by a proper file, atomically. For this, the umount is removed
+ * in a mount namespace, a new file is created at the right place. Afterwards the mount is also removed in the
+ * original mount namespace, thus revealing the file that was just created. */
+
etc_machine_id = prefix_roota(root, "/etc/machine-id");
r = path_is_mount_point(etc_machine_id, 0);
if (r < 0)
return log_error_errno(r, "Failed to determine whether %s is a mount point: %m", etc_machine_id);
if (r == 0) {
- log_debug("%s is is not a mount point. Nothing to do.", etc_machine_id);
+ log_debug("%s is not a mount point. Nothing to do.", etc_machine_id);
return 0;
}
@@ -312,10 +212,6 @@ int machine_id_commit(const char *root) {
if (fd < 0)
return log_error_errno(errno, "Cannot open %s: %m", etc_machine_id);
- r = read_machine_id(fd, id);
- if (r < 0)
- return log_error_errno(r, "We didn't find a valid machine ID in %s.", etc_machine_id);
-
r = fd_is_temporary_fs(fd);
if (r < 0)
return log_error_errno(r, "Failed to determine whether %s is on a temporary file system: %m", etc_machine_id);
@@ -324,6 +220,10 @@ int machine_id_commit(const char *root) {
return -EROFS;
}
+ r = id128_read_fd(fd, ID128_PLAIN, &id);
+ if (r < 0)
+ return log_error_errno(r, "We didn't find a valid machine ID in %s.", etc_machine_id);
+
fd = safe_close(fd);
/* Store current mount namespace */
@@ -342,15 +242,9 @@ int machine_id_commit(const char *root) {
return log_error_errno(errno, "Failed to unmount transient %s file in our private namespace: %m", etc_machine_id);
/* Update a persistent version of etc_machine_id */
- fd = open(etc_machine_id, O_RDWR|O_CREAT|O_CLOEXEC|O_NOCTTY, 0444);
- if (fd < 0)
- return log_error_errno(errno, "Cannot open for writing %s. This is mandatory to get a persistent machine-id: %m", etc_machine_id);
-
- r = write_machine_id(fd, id);
+ r = id128_write(etc_machine_id, ID128_PLAIN, id, true);
if (r < 0)
- return log_error_errno(r, "Cannot write %s: %m", etc_machine_id);
-
- fd = safe_close(fd);
+ return log_error_errno(r, "Cannot write %s. This is mandatory to get a persistent machine ID: %m", etc_machine_id);
/* Return to initial namespace and proceed a lazy tmpfs unmount */
r = namespace_enter(-1, initial_mntns_fd, -1, -1, -1);
diff --git a/src/core/machine-id-setup.h b/src/core/machine-id-setup.h
index a7e7678ed9..29f4620646 100644
--- a/src/core/machine-id-setup.h
+++ b/src/core/machine-id-setup.h
@@ -20,4 +20,4 @@
***/
int machine_id_commit(const char *root);
-int machine_id_setup(const char *root, sd_id128_t machine_id);
+int machine_id_setup(const char *root, sd_id128_t requested, sd_id128_t *ret);
diff --git a/src/core/macros.systemd.in b/src/core/macros.systemd.in
index 2cace3d3ba..6e8a3b3e3d 100644
--- a/src/core/macros.systemd.in
+++ b/src/core/macros.systemd.in
@@ -29,6 +29,8 @@
%_sysusersdir @sysusersdir@
%_sysctldir @sysctldir@
%_binfmtdir @binfmtdir@
+%_systemdgeneratordir @systemgeneratordir@
+%_systemdusergeneratordir @usergeneratordir@
%systemd_requires \
Requires(post): systemd \
@@ -36,6 +38,12 @@ Requires(preun): systemd \
Requires(postun): systemd \
%{nil}
+%systemd_ordering \
+OrderWithRequires(post): systemd \
+OrderWithRequires(preun): systemd \
+OrderWithRequires(postun): systemd \
+%{nil}
+
%systemd_post() \
if [ $1 -eq 1 ] ; then \
# Initial installation \
diff --git a/src/core/main.c b/src/core/main.c
index 3d74ef1adf..f2adca7d2b 100644
--- a/src/core/main.c
+++ b/src/core/main.c
@@ -127,7 +127,7 @@ static bool arg_default_io_accounting = false;
static bool arg_default_blockio_accounting = false;
static bool arg_default_memory_accounting = false;
static bool arg_default_tasks_accounting = true;
-static uint64_t arg_default_tasks_max = UINT64_C(512);
+static uint64_t arg_default_tasks_max = UINT64_MAX;
static sd_id128_t arg_machine_id = {};
noreturn static void freeze_or_reboot(void) {
@@ -291,14 +291,16 @@ static int parse_crash_chvt(const char *value) {
}
static int set_machine_id(const char *m) {
+ sd_id128_t t;
assert(m);
- if (sd_id128_from_string(m, &arg_machine_id) < 0)
+ if (sd_id128_from_string(m, &t) < 0)
return -EINVAL;
- if (sd_id128_is_null(arg_machine_id))
+ if (sd_id128_is_null(t))
return -EINVAL;
+ arg_machine_id = t;
return 0;
}
@@ -1298,6 +1300,11 @@ static int fixup_environment(void) {
_cleanup_free_ char *term = NULL;
int r;
+ /* We expect the environment to be set correctly
+ * if run inside a container. */
+ if (detect_container() > 0)
+ return 0;
+
/* 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
@@ -1314,7 +1321,7 @@ static int fixup_environment(void) {
if (r == 0) {
term = strdup(default_term_for_tty("/dev/console") + 5);
if (!term)
- return -errno;
+ return -ENOMEM;
}
if (setenv("TERM", term, 1) < 0)
@@ -1444,7 +1451,7 @@ int main(int argc, char *argv[]) {
/*
* Do a dummy very first call to seal the kernel's time warp magic.
*
- * Do not call this this from inside the initrd. The initrd might not
+ * Do not call 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.
@@ -1508,13 +1515,10 @@ int main(int argc, char *argv[]) {
}
if (arg_system) {
- /* We expect the environment to be set correctly
- * if run inside a container. */
- if (detect_container() <= 0)
- if (fixup_environment() < 0) {
- error_message = "Failed to fix up PID1 environment";
- goto finish;
- }
+ 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
@@ -1556,6 +1560,8 @@ int main(int argc, char *argv[]) {
(void) reset_all_signal_handlers();
(void) ignore_signals(SIGNALS_IGNORE, -1);
+ arg_default_tasks_max = system_tasks_max_scale(15U, 100U); /* 15% the system PIDs equals 4915 by default. */
+
if (parse_config_file() < 0) {
error_message = "Failed to parse config file";
goto finish;
@@ -1717,7 +1723,7 @@ int main(int argc, char *argv[]) {
status_welcome();
hostname_setup();
- machine_id_setup(NULL, arg_machine_id);
+ machine_id_setup(NULL, arg_machine_id, NULL);
loopback_setup();
bump_unix_max_dgram_qlen();
diff --git a/src/core/manager.c b/src/core/manager.c
index 902c2a0a27..4d84a0b37e 100644
--- a/src/core/manager.c
+++ b/src/core/manager.c
@@ -64,7 +64,6 @@
#include "manager.h"
#include "missing.h"
#include "mkdir.h"
-#include "mkdir.h"
#include "parse-util.h"
#include "path-lookup.h"
#include "path-util.h"
@@ -570,7 +569,7 @@ int manager_new(UnitFileScope scope, bool test_run, Manager **_m) {
m->exit_code = _MANAGER_EXIT_CODE_INVALID;
m->default_timer_accuracy_usec = USEC_PER_MINUTE;
m->default_tasks_accounting = true;
- m->default_tasks_max = UINT64_C(512);
+ m->default_tasks_max = UINT64_MAX;
#ifdef ENABLE_EFI
if (MANAGER_IS_SYSTEM(m) && detect_container() <= 0)
@@ -1729,7 +1728,10 @@ static void invoke_sigchld_event(Manager *m, Unit *u, const siginfo_t *si) {
unit_unwatch_pid(u, si->si_pid);
if (UNIT_VTABLE(u)->sigchld_event) {
- if (set_size(u->pids) <= 1 || iteration != u->sigchldgen) {
+ if (set_size(u->pids) <= 1 ||
+ iteration != u->sigchldgen ||
+ unit_main_pid(u) == si->si_pid ||
+ unit_control_pid(u) == si->si_pid) {
UNIT_VTABLE(u)->sigchld_event(u, si->si_pid, si->si_code, si->si_status);
u->sigchldgen = iteration;
} else
diff --git a/src/core/mount-setup.c b/src/core/mount-setup.c
index f9c9b4a91f..5d8ab0ec70 100644
--- a/src/core/mount-setup.c
+++ b/src/core/mount-setup.c
@@ -28,6 +28,7 @@
#include "cgroup-util.h"
#include "dev-setup.h"
#include "efivars.h"
+#include "fs-util.h"
#include "label.h"
#include "log.h"
#include "macro.h"
@@ -403,9 +404,16 @@ int mount_setup(bool loaded_policy) {
* really needs to stay for good, otherwise software that
* copied sd-daemon.c into their sources will misdetect
* systemd. */
- mkdir_label("/run/systemd", 0755);
- mkdir_label("/run/systemd/system", 0755);
- mkdir_label("/run/systemd/inaccessible", 0000);
+ (void) mkdir_label("/run/systemd", 0755);
+ (void) mkdir_label("/run/systemd/system", 0755);
+ (void) mkdir_label("/run/systemd/inaccessible", 0000);
+ /* Set up inaccessible items */
+ (void) mknod("/run/systemd/inaccessible/reg", S_IFREG | 0000, 0);
+ (void) mkdir_label("/run/systemd/inaccessible/dir", 0000);
+ (void) mknod("/run/systemd/inaccessible/chr", S_IFCHR | 0000, makedev(0, 0));
+ (void) mknod("/run/systemd/inaccessible/blk", S_IFBLK | 0000, makedev(0, 0));
+ (void) mkfifo("/run/systemd/inaccessible/fifo", 0000);
+ (void) mknod("/run/systemd/inaccessible/sock", S_IFSOCK | 0000, 0);
return 0;
}
diff --git a/src/core/namespace.c b/src/core/namespace.c
index 203d122810..52a2505d94 100644
--- a/src/core/namespace.c
+++ b/src/core/namespace.c
@@ -278,6 +278,7 @@ static int apply_mount(
const char *what;
int r;
+ struct stat target;
assert(m);
@@ -287,12 +288,21 @@ static int apply_mount(
/* First, get rid of everything that is below if there
* is anything... Then, overmount it with an
- * inaccessible directory. */
+ * inaccessible path. */
umount_recursive(m->path, 0);
- what = "/run/systemd/inaccessible";
- break;
+ if (lstat(m->path, &target) < 0) {
+ if (m->ignore && errno == ENOENT)
+ return 0;
+ return -errno;
+ }
+ what = mode_to_inaccessible_node(target.st_mode);
+ if (!what) {
+ log_debug("File type not supported for inaccessible mounts. Note that symlinks are not allowed");
+ return -ELOOP;
+ }
+ break;
case READONLY:
case READWRITE:
/* Nothing to mount here, we just later toggle the
@@ -317,12 +327,14 @@ static int apply_mount(
assert(what);
r = mount(what, m->path, NULL, MS_BIND|MS_REC, NULL);
- if (r >= 0)
+ if (r >= 0) {
log_debug("Successfully mounted %s to %s", what, m->path);
- else if (m->ignore && errno == ENOENT)
- return 0;
-
- return r;
+ return r;
+ } else {
+ if (m->ignore && errno == ENOENT)
+ return 0;
+ return log_debug_errno(errno, "Failed to mount %s to %s: %m", what, m->path);
+ }
}
static int make_read_only(BindMount *m) {
@@ -335,7 +347,8 @@ static int make_read_only(BindMount *m) {
else if (IN_SET(m->mode, READWRITE, PRIVATE_TMP, PRIVATE_VAR_TMP, PRIVATE_DEV)) {
r = bind_remount_recursive(m->path, false);
if (r == 0 && m->mode == PRIVATE_DEV) /* can be readonly but the submounts can't*/
- r = mount(NULL, m->path, NULL, MS_REMOUNT|DEV_MOUNT_OPTIONS|MS_RDONLY, NULL);
+ if (mount(NULL, m->path, NULL, MS_REMOUNT|DEV_MOUNT_OPTIONS|MS_RDONLY, NULL) < 0)
+ r = -errno;
} else
r = 0;
@@ -347,9 +360,9 @@ static int make_read_only(BindMount *m) {
int setup_namespace(
const char* root_directory,
- char** read_write_dirs,
- char** read_only_dirs,
- char** inaccessible_dirs,
+ char** read_write_paths,
+ char** read_only_paths,
+ char** inaccessible_paths,
const char* tmp_dir,
const char* var_tmp_dir,
bool private_dev,
@@ -368,9 +381,9 @@ int setup_namespace(
return -errno;
n = !!tmp_dir + !!var_tmp_dir +
- strv_length(read_write_dirs) +
- strv_length(read_only_dirs) +
- strv_length(inaccessible_dirs) +
+ strv_length(read_write_paths) +
+ strv_length(read_only_paths) +
+ strv_length(inaccessible_paths) +
private_dev +
(protect_home != PROTECT_HOME_NO ? 3 : 0) +
(protect_system != PROTECT_SYSTEM_NO ? 2 : 0) +
@@ -378,15 +391,15 @@ int setup_namespace(
if (n > 0) {
m = mounts = (BindMount *) alloca0(n * sizeof(BindMount));
- r = append_mounts(&m, read_write_dirs, READWRITE);
+ r = append_mounts(&m, read_write_paths, READWRITE);
if (r < 0)
return r;
- r = append_mounts(&m, read_only_dirs, READONLY);
+ r = append_mounts(&m, read_only_paths, READONLY);
if (r < 0)
return r;
- r = append_mounts(&m, inaccessible_dirs, INACCESSIBLE);
+ r = append_mounts(&m, inaccessible_paths, INACCESSIBLE);
if (r < 0)
return r;
@@ -629,7 +642,7 @@ int setup_netns(int netns_storage_socket[2]) {
}
fail:
- lockf(netns_storage_socket[0], F_ULOCK, 0);
+ (void) lockf(netns_storage_socket[0], F_ULOCK, 0);
return r;
}
diff --git a/src/core/namespace.h b/src/core/namespace.h
index b54b7b47d6..1aedf5f208 100644
--- a/src/core/namespace.h
+++ b/src/core/namespace.h
@@ -40,9 +40,9 @@ typedef enum ProtectSystem {
} ProtectSystem;
int setup_namespace(const char *chroot,
- char **read_write_dirs,
- char **read_only_dirs,
- char **inaccessible_dirs,
+ char **read_write_paths,
+ char **read_only_paths,
+ char **inaccessible_paths,
const char *tmp_dir,
const char *var_tmp_dir,
bool private_dev,
diff --git a/src/core/scope.c b/src/core/scope.c
index decd1a1f3f..b45e238974 100644
--- a/src/core/scope.c
+++ b/src/core/scope.c
@@ -240,7 +240,7 @@ static void scope_enter_signal(Scope *s, ScopeState state, ScopeResult f) {
/* If we have a controller set let's ask the controller nicely
* to terminate the scope, instead of us going directly into
- * SIGTERM beserk mode */
+ * SIGTERM berserk mode */
if (state == SCOPE_STOP_SIGTERM)
skip_signal = bus_scope_send_request_stop(s) > 0;
@@ -248,7 +248,9 @@ static void scope_enter_signal(Scope *s, ScopeState state, ScopeResult f) {
r = unit_kill_context(
UNIT(s),
&s->kill_context,
- state != SCOPE_STOP_SIGTERM ? KILL_KILL : KILL_TERMINATE,
+ state != SCOPE_STOP_SIGTERM ? KILL_KILL :
+ s->was_abandoned ? KILL_TERMINATE_AND_LOG :
+ KILL_TERMINATE,
-1, -1, false);
if (r < 0)
goto fail;
@@ -369,6 +371,7 @@ static int scope_serialize(Unit *u, FILE *f, FDSet *fds) {
assert(fds);
unit_serialize_item(u, f, "state", scope_state_to_string(s->state));
+ unit_serialize_item(u, f, "was-abandoned", yes_no(s->was_abandoned));
return 0;
}
@@ -389,6 +392,14 @@ static int scope_deserialize_item(Unit *u, const char *key, const char *value, F
else
s->deserialized_state = state;
+ } else if (streq(key, "was-abandoned")) {
+ int k;
+
+ k = parse_boolean(value);
+ if (k < 0)
+ log_unit_debug(u, "Failed to parse boolean value: %s", value);
+ else
+ s->was_abandoned = k;
} else
log_unit_debug(u, "Unknown serialization key: %s", key);
@@ -474,6 +485,7 @@ int scope_abandon(Scope *s) {
if (!IN_SET(s->state, SCOPE_RUNNING, SCOPE_ABANDONED))
return -ESTALE;
+ s->was_abandoned = true;
s->controller = mfree(s->controller);
/* The client is no longer watching the remaining processes,
diff --git a/src/core/scope.h b/src/core/scope.h
index 2dc86325c5..eaf8e8b447 100644
--- a/src/core/scope.h
+++ b/src/core/scope.h
@@ -21,7 +21,9 @@
typedef struct Scope Scope;
+#include "cgroup.h"
#include "kill.h"
+#include "unit.h"
typedef enum ScopeResult {
SCOPE_SUCCESS,
@@ -43,6 +45,7 @@ struct Scope {
usec_t timeout_stop_usec;
char *controller;
+ bool was_abandoned;
sd_event_source *timer_event_source;
};
diff --git a/src/core/selinux-access.c b/src/core/selinux-access.c
index cc287d602d..2b96a9551b 100644
--- a/src/core/selinux-access.c
+++ b/src/core/selinux-access.c
@@ -191,7 +191,7 @@ int mac_selinux_generic_access_check(
const char *tclass = NULL, *scon = NULL;
struct audit_info audit_info = {};
_cleanup_free_ char *cl = NULL;
- security_context_t fcon = NULL;
+ char *fcon = NULL;
char **cmdline = NULL;
int r = 0;
diff --git a/src/core/selinux-setup.c b/src/core/selinux-setup.c
index 4072df58e6..527aa8add0 100644
--- a/src/core/selinux-setup.c
+++ b/src/core/selinux-setup.c
@@ -44,7 +44,7 @@ int mac_selinux_setup(bool *loaded_policy) {
#ifdef HAVE_SELINUX
int enforce = 0;
usec_t before_load, after_load;
- security_context_t con;
+ char *con;
int r;
union selinux_callback cb;
bool initialized = false;
diff --git a/src/core/service.c b/src/core/service.c
index 13de671700..afb198507b 100644
--- a/src/core/service.c
+++ b/src/core/service.c
@@ -1674,7 +1674,7 @@ static void service_kill_control_processes(Service *s) {
return;
p = strjoina(UNIT(s)->cgroup_path, "/control");
- cg_kill_recursive(SYSTEMD_CGROUP_CONTROLLER, p, SIGKILL, true, true, true, NULL);
+ cg_kill_recursive(SYSTEMD_CGROUP_CONTROLLER, p, SIGKILL, CGROUP_SIGCONT|CGROUP_IGNORE_SELF|CGROUP_REMOVE, NULL, NULL, NULL);
}
static void service_enter_start(Service *s) {
diff --git a/src/core/shutdown.c b/src/core/shutdown.c
index e14755d84e..a795d875bb 100644
--- a/src/core/shutdown.c
+++ b/src/core/shutdown.c
@@ -157,7 +157,6 @@ static int switch_root_initramfs(void) {
return switch_root("/run/initramfs", "/oldroot", false, MS_BIND);
}
-
int main(int argc, char *argv[]) {
bool need_umount, need_swapoff, need_loop_detach, need_dm_detach;
bool in_container, use_watchdog = false;
@@ -203,20 +202,25 @@ int main(int argc, char *argv[]) {
}
(void) cg_get_root_path(&cgroup);
+ in_container = detect_container() > 0;
use_watchdog = !!getenv("WATCHDOG_USEC");
- /* lock us into memory */
+ /* Lock us into memory */
mlockall(MCL_CURRENT|MCL_FUTURE);
+ /* Synchronize everything that is not written to disk yet at this point already. This is a good idea so that
+ * slow IO is processed here already and the final process killing spree is not impacted by processes
+ * desperately trying to sync IO to disk within their timeout. */
+ if (!in_container)
+ sync();
+
log_info("Sending SIGTERM to remaining processes...");
broadcast_signal(SIGTERM, true, true);
log_info("Sending SIGKILL to remaining processes...");
broadcast_signal(SIGKILL, true, false);
- in_container = detect_container() > 0;
-
need_umount = !in_container;
need_swapoff = !in_container;
need_loop_detach = !in_container;
@@ -345,10 +349,10 @@ int main(int argc, char *argv[]) {
need_loop_detach ? " loop devices," : "",
need_dm_detach ? " DM devices," : "");
- /* The kernel will automaticall flush ATA disks and suchlike
- * on reboot(), but the file systems need to be synce'd
- * explicitly in advance. So let's do this here, but not
- * needlessly slow down containers. */
+ /* The kernel will automatically flush ATA disks and suchlike on reboot(), but the file systems need to be
+ * sync'ed explicitly in advance. So let's do this here, but not needlessly slow down containers. Note that we
+ * sync'ed things already once above, but we did some more work since then which might have caused IO, hence
+ * let's doit once more. */
if (!in_container)
sync();
diff --git a/src/core/system.conf b/src/core/system.conf
index db8b7acd78..c6bb050aac 100644
--- a/src/core/system.conf
+++ b/src/core/system.conf
@@ -42,7 +42,7 @@
#DefaultBlockIOAccounting=no
#DefaultMemoryAccounting=no
#DefaultTasksAccounting=yes
-#DefaultTasksMax=512
+#DefaultTasksMax=15%
#DefaultLimitCPU=
#DefaultLimitFSIZE=
#DefaultLimitDATA=
diff --git a/src/core/transaction.c b/src/core/transaction.c
index e06a48a2f1..8370b864fb 100644
--- a/src/core/transaction.c
+++ b/src/core/transaction.c
@@ -373,7 +373,7 @@ static int transaction_verify_order_one(Transaction *tr, Job *j, Job *from, unsi
delete = NULL;
for (k = from; k; k = ((k->generation == generation && k->marker != k) ? k->marker : NULL)) {
- /* logging for j not k here here to provide consistent narrative */
+ /* logging for j not k here to provide consistent narrative */
log_unit_warning(j->unit,
"Found dependency on %s/%s",
k->unit->id, job_type_to_string(k->type));
@@ -392,7 +392,7 @@ static int transaction_verify_order_one(Transaction *tr, Job *j, Job *from, unsi
if (delete) {
const char *status;
- /* logging for j not k here here to provide consistent narrative */
+ /* logging for j not k here to provide consistent narrative */
log_unit_warning(j->unit,
"Breaking ordering cycle by deleting job %s/%s",
delete->unit->id, job_type_to_string(delete->type));
@@ -591,6 +591,9 @@ static int transaction_apply(Transaction *tr, Manager *m, JobMode mode) {
HASHMAP_FOREACH(j, m->jobs, i) {
assert(j->installed);
+ if (j->unit->ignore_on_isolate)
+ continue;
+
if (hashmap_get(tr->jobs, j->unit))
continue;
diff --git a/src/core/unit.c b/src/core/unit.c
index 5f06a7dfe7..4934a0e56f 100644
--- a/src/core/unit.c
+++ b/src/core/unit.c
@@ -1683,7 +1683,7 @@ static void unit_check_unneeded(Unit *u) {
if (unit_active_or_pending(other))
return;
- /* If stopping a unit fails continously we might enter a stop
+ /* If stopping a unit fails continuously we might enter a stop
* loop here, hence stop acting on the service being
* unnecessary after a while. */
if (!ratelimit_test(&u->auto_stop_ratelimit)) {
@@ -1728,7 +1728,7 @@ static void unit_check_binds_to(Unit *u) {
if (!stop)
return;
- /* If stopping a unit fails continously we might enter a stop
+ /* If stopping a unit fails continuously we might enter a stop
* loop here, hence stop acting on the service being
* unnecessary after a while. */
if (!ratelimit_test(&u->auto_stop_ratelimit)) {
@@ -3144,7 +3144,7 @@ int unit_kill_common(
if (!pid_set)
return -ENOMEM;
- q = cg_kill_recursive(SYSTEMD_CGROUP_CONTROLLER, u->cgroup_path, signo, false, false, false, pid_set);
+ q = cg_kill_recursive(SYSTEMD_CGROUP_CONTROLLER, u->cgroup_path, signo, 0, pid_set, NULL, NULL);
if (q < 0 && q != -EAGAIN && q != -ESRCH && q != -ENOENT)
r = q;
else
@@ -3512,6 +3512,43 @@ int unit_make_transient(Unit *u) {
return 0;
}
+static void log_kill(pid_t pid, int sig, void *userdata) {
+ _cleanup_free_ char *comm = NULL;
+
+ (void) get_process_comm(pid, &comm);
+
+ /* Don't log about processes marked with brackets, under the assumption that these are temporary processes
+ only, like for example systemd's own PAM stub process. */
+ if (comm && comm[0] == '(')
+ return;
+
+ log_unit_notice(userdata,
+ "Killing process " PID_FMT " (%s) with signal SIG%s.",
+ pid,
+ strna(comm),
+ signal_to_string(sig));
+}
+
+static int operation_to_signal(KillContext *c, KillOperation k) {
+ assert(c);
+
+ switch (k) {
+
+ case KILL_TERMINATE:
+ case KILL_TERMINATE_AND_LOG:
+ return c->kill_signal;
+
+ case KILL_KILL:
+ return SIGKILL;
+
+ case KILL_ABORT:
+ return SIGABRT;
+
+ default:
+ assert_not_reached("KillOperation unknown");
+ }
+}
+
int unit_kill_context(
Unit *u,
KillContext *c,
@@ -3520,58 +3557,63 @@ int unit_kill_context(
pid_t control_pid,
bool main_pid_alien) {
- bool wait_for_exit = false;
+ bool wait_for_exit = false, send_sighup;
+ cg_kill_log_func_t log_func;
int sig, r;
assert(u);
assert(c);
+ /* Kill the processes belonging to this unit, in preparation for shutting the unit down. Returns > 0 if we
+ * killed something worth waiting for, 0 otherwise. */
+
if (c->kill_mode == KILL_NONE)
return 0;
- switch (k) {
- case KILL_KILL:
- sig = SIGKILL;
- break;
- case KILL_ABORT:
- sig = SIGABRT;
- break;
- case KILL_TERMINATE:
- sig = c->kill_signal;
- break;
- default:
- assert_not_reached("KillOperation unknown");
- }
+ sig = operation_to_signal(c, k);
+
+ send_sighup =
+ c->send_sighup &&
+ IN_SET(k, KILL_TERMINATE, KILL_TERMINATE_AND_LOG) &&
+ sig != SIGHUP;
+
+ log_func =
+ k != KILL_TERMINATE ||
+ IN_SET(sig, SIGKILL, SIGABRT) ? log_kill : NULL;
if (main_pid > 0) {
- r = kill_and_sigcont(main_pid, sig);
+ if (log_func)
+ log_func(main_pid, sig, u);
+ r = kill_and_sigcont(main_pid, sig);
if (r < 0 && r != -ESRCH) {
_cleanup_free_ char *comm = NULL;
- get_process_comm(main_pid, &comm);
+ (void) get_process_comm(main_pid, &comm);
log_unit_warning_errno(u, r, "Failed to kill main process " PID_FMT " (%s), ignoring: %m", main_pid, strna(comm));
} else {
if (!main_pid_alien)
wait_for_exit = true;
- if (c->send_sighup && k == KILL_TERMINATE)
+ if (r != -ESRCH && send_sighup)
(void) kill(main_pid, SIGHUP);
}
}
if (control_pid > 0) {
- r = kill_and_sigcont(control_pid, sig);
+ if (log_func)
+ log_func(control_pid, sig, u);
+ r = kill_and_sigcont(control_pid, sig);
if (r < 0 && r != -ESRCH) {
_cleanup_free_ char *comm = NULL;
- get_process_comm(control_pid, &comm);
+ (void) get_process_comm(control_pid, &comm);
log_unit_warning_errno(u, r, "Failed to kill control process " PID_FMT " (%s), ignoring: %m", control_pid, strna(comm));
} else {
wait_for_exit = true;
- if (c->send_sighup && k == KILL_TERMINATE)
+ if (r != -ESRCH && send_sighup)
(void) kill(control_pid, SIGHUP);
}
}
@@ -3585,7 +3627,11 @@ int unit_kill_context(
if (!pid_set)
return -ENOMEM;
- r = cg_kill_recursive(SYSTEMD_CGROUP_CONTROLLER, u->cgroup_path, sig, true, k != KILL_TERMINATE, false, pid_set);
+ r = cg_kill_recursive(SYSTEMD_CGROUP_CONTROLLER, u->cgroup_path,
+ sig,
+ CGROUP_SIGCONT|CGROUP_IGNORE_SELF,
+ pid_set,
+ log_func, u);
if (r < 0) {
if (r != -EAGAIN && r != -ESRCH && r != -ENOENT)
log_unit_warning_errno(u, r, "Failed to kill control group %s, ignoring: %m", u->cgroup_path);
@@ -3610,14 +3656,18 @@ int unit_kill_context(
(detect_container() == 0 && !unit_cgroup_delegate(u)))
wait_for_exit = true;
- if (c->send_sighup && k != KILL_KILL) {
+ if (send_sighup) {
set_free(pid_set);
pid_set = unit_pid_set(main_pid, control_pid);
if (!pid_set)
return -ENOMEM;
- cg_kill_recursive(SYSTEMD_CGROUP_CONTROLLER, u->cgroup_path, SIGHUP, false, true, false, pid_set);
+ cg_kill_recursive(SYSTEMD_CGROUP_CONTROLLER, u->cgroup_path,
+ SIGHUP,
+ CGROUP_IGNORE_SELF,
+ pid_set,
+ NULL, NULL);
}
}
}
@@ -3790,7 +3840,7 @@ bool unit_is_pristine(Unit *u) {
/* Check if the unit already exists or is already around,
* in a number of different ways. Note that to cater for unit
* types such as slice, we are generally fine with units that
- * are marked UNIT_LOADED even even though nothing was
+ * are marked UNIT_LOADED even though nothing was
* actually loaded, as those unit types don't require a file
* on disk to validly load. */
diff --git a/src/core/unit.h b/src/core/unit.h
index c41011ed9d..1eabfa51e2 100644
--- a/src/core/unit.h
+++ b/src/core/unit.h
@@ -36,6 +36,7 @@ typedef struct UnitStatusMessageFormats UnitStatusMessageFormats;
typedef enum KillOperation {
KILL_TERMINATE,
+ KILL_TERMINATE_AND_LOG,
KILL_KILL,
KILL_ABORT,
_KILL_OPERATION_MAX,
generated by cgit v1.2.3-54-g00ecf (git 2.45.2) at 2025-04-05 19:01:08 +0000