Systemd/src/core/path.c
Lennart Poettering 4b58153dd2 core: add "invocation ID" concept to service manager
This adds a new invocation ID concept to the service manager. The invocation ID
identifies each runtime cycle of a unit uniquely. A new randomized 128bit ID is
generated each time a unit moves from and inactive to an activating or active
state.

The primary usecase for this concept is to connect the runtime data PID 1
maintains about a service with the offline data the journal stores about it.
Previously we'd use the unit name plus start/stop times, which however is
highly racy since the journal will generally process log data after the service
already ended.

The "invocation ID" kinda matches the "boot ID" concept of the Linux kernel,
except that it applies to an individual unit instead of the whole system.

The invocation ID is passed to the activated processes as environment variable.
It is additionally stored as extended attribute on the cgroup of the unit. The
latter is used by journald to automatically retrieve it for each log logged
message and attach it to the log entry. The environment variable is very easily
accessible, even for unprivileged services. OTOH the extended attribute is only
accessible to privileged processes (this is because cgroupfs only supports the
"trusted." xattr namespace, not "user."). The environment variable may be
altered by services, the extended attribute may not be, hence is the better
choice for the journal.

Note that reading the invocation ID off the extended attribute from journald is
racy, similar to the way reading the unit name for a logging process is.

This patch adds APIs to read the invocation ID to sd-id128:
sd_id128_get_invocation() may be used in a similar fashion to
sd_id128_get_boot().

PID1's own logging is updated to always include the invocation ID when it logs
information about a unit.

A new bus call GetUnitByInvocationID() is added that allows retrieving a bus
path to a unit by its invocation ID. The bus path is built using the invocation
ID, thus providing a path for referring to a unit that is valid only for the
current runtime cycleof it.

Outlook for the future: should the kernel eventually allow passing of cgroup
information along AF_UNIX/SOCK_DGRAM messages via a unique cgroup id, then we
can alter the invocation ID to be generated as hash from that rather than
entirely randomly. This way we can derive the invocation race-freely from the
messages.
2016-10-07 20:14:38 +02:00

789 lines
21 KiB
C

/***
This file is part of systemd.
Copyright 2010 Lennart Poettering
systemd is free software; you can redistribute it and/or modify it
under the terms of the GNU Lesser General Public License as published by
the Free Software Foundation; either version 2.1 of the License, or
(at your option) any later version.
systemd is distributed in the hope that it will be useful, but
WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public License
along with systemd; If not, see <http://www.gnu.org/licenses/>.
***/
#include <errno.h>
#include <sys/epoll.h>
#include <sys/inotify.h>
#include <unistd.h>
#include "bus-error.h"
#include "bus-util.h"
#include "dbus-path.h"
#include "fd-util.h"
#include "fs-util.h"
#include "glob-util.h"
#include "macro.h"
#include "mkdir.h"
#include "path.h"
#include "special.h"
#include "stat-util.h"
#include "string-table.h"
#include "string-util.h"
#include "unit-name.h"
#include "unit.h"
static const UnitActiveState state_translation_table[_PATH_STATE_MAX] = {
[PATH_DEAD] = UNIT_INACTIVE,
[PATH_WAITING] = UNIT_ACTIVE,
[PATH_RUNNING] = UNIT_ACTIVE,
[PATH_FAILED] = UNIT_FAILED
};
static int path_dispatch_io(sd_event_source *source, int fd, uint32_t revents, void *userdata);
int path_spec_watch(PathSpec *s, sd_event_io_handler_t handler) {
static const int flags_table[_PATH_TYPE_MAX] = {
[PATH_EXISTS] = IN_DELETE_SELF|IN_MOVE_SELF|IN_ATTRIB,
[PATH_EXISTS_GLOB] = IN_DELETE_SELF|IN_MOVE_SELF|IN_ATTRIB,
[PATH_CHANGED] = IN_DELETE_SELF|IN_MOVE_SELF|IN_ATTRIB|IN_CLOSE_WRITE|IN_CREATE|IN_DELETE|IN_MOVED_FROM|IN_MOVED_TO,
[PATH_MODIFIED] = IN_DELETE_SELF|IN_MOVE_SELF|IN_ATTRIB|IN_CLOSE_WRITE|IN_CREATE|IN_DELETE|IN_MOVED_FROM|IN_MOVED_TO|IN_MODIFY,
[PATH_DIRECTORY_NOT_EMPTY] = IN_DELETE_SELF|IN_MOVE_SELF|IN_ATTRIB|IN_CREATE|IN_MOVED_TO
};
bool exists = false;
char *slash, *oldslash = NULL;
int r;
assert(s);
assert(s->unit);
assert(handler);
path_spec_unwatch(s);
s->inotify_fd = inotify_init1(IN_NONBLOCK|IN_CLOEXEC);
if (s->inotify_fd < 0) {
r = -errno;
goto fail;
}
r = sd_event_add_io(s->unit->manager->event, &s->event_source, s->inotify_fd, EPOLLIN, handler, s);
if (r < 0)
goto fail;
(void) sd_event_source_set_description(s->event_source, "path");
/* This assumes the path was passed through path_kill_slashes()! */
for (slash = strchr(s->path, '/'); ; slash = strchr(slash+1, '/')) {
char *cut = NULL;
int flags;
char tmp;
if (slash) {
cut = slash + (slash == s->path);
tmp = *cut;
*cut = '\0';
flags = IN_MOVE_SELF | IN_DELETE_SELF | IN_ATTRIB | IN_CREATE | IN_MOVED_TO;
} else
flags = flags_table[s->type];
r = inotify_add_watch(s->inotify_fd, s->path, flags);
if (r < 0) {
if (errno == EACCES || errno == ENOENT) {
if (cut)
*cut = tmp;
break;
}
r = log_warning_errno(errno, "Failed to add watch on %s: %s", s->path, errno == ENOSPC ? "too many watches" : strerror(-r));
if (cut)
*cut = tmp;
goto fail;
} else {
exists = true;
/* Path exists, we don't need to watch parent too closely. */
if (oldslash) {
char *cut2 = oldslash + (oldslash == s->path);
char tmp2 = *cut2;
*cut2 = '\0';
(void) inotify_add_watch(s->inotify_fd, s->path, IN_MOVE_SELF);
/* Error is ignored, the worst can happen is we get spurious events. */
*cut2 = tmp2;
}
}
if (cut)
*cut = tmp;
if (slash)
oldslash = slash;
else {
/* whole path has been iterated over */
s->primary_wd = r;
break;
}
}
if (!exists) {
r = log_error_errno(errno, "Failed to add watch on any of the components of %s: %m", s->path);
/* either EACCESS or ENOENT */
goto fail;
}
return 0;
fail:
path_spec_unwatch(s);
return r;
}
void path_spec_unwatch(PathSpec *s) {
assert(s);
s->event_source = sd_event_source_unref(s->event_source);
s->inotify_fd = safe_close(s->inotify_fd);
}
int path_spec_fd_event(PathSpec *s, uint32_t revents) {
union inotify_event_buffer buffer;
struct inotify_event *e;
ssize_t l;
int r = 0;
if (revents != EPOLLIN) {
log_error("Got invalid poll event on inotify.");
return -EINVAL;
}
l = read(s->inotify_fd, &buffer, sizeof(buffer));
if (l < 0) {
if (errno == EAGAIN || errno == EINTR)
return 0;
return log_error_errno(errno, "Failed to read inotify event: %m");
}
FOREACH_INOTIFY_EVENT(e, buffer, l) {
if ((s->type == PATH_CHANGED || s->type == PATH_MODIFIED) &&
s->primary_wd == e->wd)
r = 1;
}
return r;
}
static bool path_spec_check_good(PathSpec *s, bool initial) {
bool good = false;
switch (s->type) {
case PATH_EXISTS:
good = access(s->path, F_OK) >= 0;
break;
case PATH_EXISTS_GLOB:
good = glob_exists(s->path) > 0;
break;
case PATH_DIRECTORY_NOT_EMPTY: {
int k;
k = dir_is_empty(s->path);
good = !(k == -ENOENT || k > 0);
break;
}
case PATH_CHANGED:
case PATH_MODIFIED: {
bool b;
b = access(s->path, F_OK) >= 0;
good = !initial && b != s->previous_exists;
s->previous_exists = b;
break;
}
default:
;
}
return good;
}
static void path_spec_mkdir(PathSpec *s, mode_t mode) {
int r;
if (s->type == PATH_EXISTS || s->type == PATH_EXISTS_GLOB)
return;
r = mkdir_p_label(s->path, mode);
if (r < 0)
log_warning_errno(r, "mkdir(%s) failed: %m", s->path);
}
static void path_spec_dump(PathSpec *s, FILE *f, const char *prefix) {
fprintf(f,
"%s%s: %s\n",
prefix,
path_type_to_string(s->type),
s->path);
}
void path_spec_done(PathSpec *s) {
assert(s);
assert(s->inotify_fd == -1);
free(s->path);
}
static void path_init(Unit *u) {
Path *p = PATH(u);
assert(u);
assert(u->load_state == UNIT_STUB);
p->directory_mode = 0755;
}
void path_free_specs(Path *p) {
PathSpec *s;
assert(p);
while ((s = p->specs)) {
path_spec_unwatch(s);
LIST_REMOVE(spec, p->specs, s);
path_spec_done(s);
free(s);
}
}
static void path_done(Unit *u) {
Path *p = PATH(u);
assert(p);
path_free_specs(p);
}
static int path_add_mount_links(Path *p) {
PathSpec *s;
int r;
assert(p);
LIST_FOREACH(spec, s, p->specs) {
r = unit_require_mounts_for(UNIT(p), s->path);
if (r < 0)
return r;
}
return 0;
}
static int path_verify(Path *p) {
assert(p);
if (UNIT(p)->load_state != UNIT_LOADED)
return 0;
if (!p->specs) {
log_unit_error(UNIT(p), "Path unit lacks path setting. Refusing.");
return -EINVAL;
}
return 0;
}
static int path_add_default_dependencies(Path *p) {
int r;
assert(p);
if (!UNIT(p)->default_dependencies)
return 0;
r = unit_add_dependency_by_name(UNIT(p), UNIT_BEFORE, SPECIAL_PATHS_TARGET, NULL, true);
if (r < 0)
return r;
if (MANAGER_IS_SYSTEM(UNIT(p)->manager)) {
r = unit_add_two_dependencies_by_name(UNIT(p), UNIT_AFTER, UNIT_REQUIRES, SPECIAL_SYSINIT_TARGET, NULL, true);
if (r < 0)
return r;
}
return unit_add_two_dependencies_by_name(UNIT(p), UNIT_BEFORE, UNIT_CONFLICTS, SPECIAL_SHUTDOWN_TARGET, NULL, true);
}
static int path_load(Unit *u) {
Path *p = PATH(u);
int r;
assert(u);
assert(u->load_state == UNIT_STUB);
r = unit_load_fragment_and_dropin(u);
if (r < 0)
return r;
if (u->load_state == UNIT_LOADED) {
if (set_isempty(u->dependencies[UNIT_TRIGGERS])) {
Unit *x;
r = unit_load_related_unit(u, ".service", &x);
if (r < 0)
return r;
r = unit_add_two_dependencies(u, UNIT_BEFORE, UNIT_TRIGGERS, x, true);
if (r < 0)
return r;
}
r = path_add_mount_links(p);
if (r < 0)
return r;
r = path_add_default_dependencies(p);
if (r < 0)
return r;
}
return path_verify(p);
}
static void path_dump(Unit *u, FILE *f, const char *prefix) {
Path *p = PATH(u);
Unit *trigger;
PathSpec *s;
assert(p);
assert(f);
trigger = UNIT_TRIGGER(u);
fprintf(f,
"%sPath State: %s\n"
"%sResult: %s\n"
"%sUnit: %s\n"
"%sMakeDirectory: %s\n"
"%sDirectoryMode: %04o\n",
prefix, path_state_to_string(p->state),
prefix, path_result_to_string(p->result),
prefix, trigger ? trigger->id : "n/a",
prefix, yes_no(p->make_directory),
prefix, p->directory_mode);
LIST_FOREACH(spec, s, p->specs)
path_spec_dump(s, f, prefix);
}
static void path_unwatch(Path *p) {
PathSpec *s;
assert(p);
LIST_FOREACH(spec, s, p->specs)
path_spec_unwatch(s);
}
static int path_watch(Path *p) {
int r;
PathSpec *s;
assert(p);
LIST_FOREACH(spec, s, p->specs) {
r = path_spec_watch(s, path_dispatch_io);
if (r < 0)
return r;
}
return 0;
}
static void path_set_state(Path *p, PathState state) {
PathState old_state;
assert(p);
old_state = p->state;
p->state = state;
if (state != PATH_WAITING &&
(state != PATH_RUNNING || p->inotify_triggered))
path_unwatch(p);
if (state != old_state)
log_unit_debug(UNIT(p), "Changed %s -> %s", path_state_to_string(old_state), path_state_to_string(state));
unit_notify(UNIT(p), state_translation_table[old_state], state_translation_table[state], true);
}
static void path_enter_waiting(Path *p, bool initial, bool recheck);
static int path_coldplug(Unit *u) {
Path *p = PATH(u);
assert(p);
assert(p->state == PATH_DEAD);
if (p->deserialized_state != p->state) {
if (p->deserialized_state == PATH_WAITING ||
p->deserialized_state == PATH_RUNNING)
path_enter_waiting(p, true, true);
else
path_set_state(p, p->deserialized_state);
}
return 0;
}
static void path_enter_dead(Path *p, PathResult f) {
assert(p);
if (p->result == PATH_SUCCESS)
p->result = f;
path_set_state(p, p->result != PATH_SUCCESS ? PATH_FAILED : PATH_DEAD);
}
static void path_enter_running(Path *p) {
_cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL;
Unit *trigger;
int r;
assert(p);
/* Don't start job if we are supposed to go down */
if (unit_stop_pending(UNIT(p)))
return;
trigger = UNIT_TRIGGER(UNIT(p));
if (!trigger) {
log_unit_error(UNIT(p), "Unit to trigger vanished.");
path_enter_dead(p, PATH_FAILURE_RESOURCES);
return;
}
r = manager_add_job(UNIT(p)->manager, JOB_START, trigger, JOB_REPLACE, &error, NULL);
if (r < 0)
goto fail;
p->inotify_triggered = false;
r = path_watch(p);
if (r < 0)
goto fail;
path_set_state(p, PATH_RUNNING);
return;
fail:
log_unit_warning(UNIT(p), "Failed to queue unit startup job: %s", bus_error_message(&error, r));
path_enter_dead(p, PATH_FAILURE_RESOURCES);
}
static bool path_check_good(Path *p, bool initial) {
PathSpec *s;
bool good = false;
assert(p);
LIST_FOREACH(spec, s, p->specs) {
good = path_spec_check_good(s, initial);
if (good)
break;
}
return good;
}
static void path_enter_waiting(Path *p, bool initial, bool recheck) {
int r;
if (recheck)
if (path_check_good(p, initial)) {
log_unit_debug(UNIT(p), "Got triggered.");
path_enter_running(p);
return;
}
r = path_watch(p);
if (r < 0)
goto fail;
/* Hmm, so now we have created inotify watches, but the file
* might have appeared/been removed by now, so we must
* recheck */
if (recheck)
if (path_check_good(p, false)) {
log_unit_debug(UNIT(p), "Got triggered.");
path_enter_running(p);
return;
}
path_set_state(p, PATH_WAITING);
return;
fail:
log_unit_warning_errno(UNIT(p), r, "Failed to enter waiting state: %m");
path_enter_dead(p, PATH_FAILURE_RESOURCES);
}
static void path_mkdir(Path *p) {
PathSpec *s;
assert(p);
if (!p->make_directory)
return;
LIST_FOREACH(spec, s, p->specs)
path_spec_mkdir(s, p->directory_mode);
}
static int path_start(Unit *u) {
Path *p = PATH(u);
Unit *trigger;
int r;
assert(p);
assert(p->state == PATH_DEAD || p->state == PATH_FAILED);
trigger = UNIT_TRIGGER(u);
if (!trigger || trigger->load_state != UNIT_LOADED) {
log_unit_error(u, "Refusing to start, unit to trigger not loaded.");
return -ENOENT;
}
r = unit_start_limit_test(u);
if (r < 0) {
path_enter_dead(p, PATH_FAILURE_START_LIMIT_HIT);
return r;
}
r = unit_acquire_invocation_id(u);
if (r < 0)
return r;
path_mkdir(p);
p->result = PATH_SUCCESS;
path_enter_waiting(p, true, true);
return 1;
}
static int path_stop(Unit *u) {
Path *p = PATH(u);
assert(p);
assert(p->state == PATH_WAITING || p->state == PATH_RUNNING);
path_enter_dead(p, PATH_SUCCESS);
return 1;
}
static int path_serialize(Unit *u, FILE *f, FDSet *fds) {
Path *p = PATH(u);
assert(u);
assert(f);
assert(fds);
unit_serialize_item(u, f, "state", path_state_to_string(p->state));
unit_serialize_item(u, f, "result", path_result_to_string(p->result));
return 0;
}
static int path_deserialize_item(Unit *u, const char *key, const char *value, FDSet *fds) {
Path *p = PATH(u);
assert(u);
assert(key);
assert(value);
assert(fds);
if (streq(key, "state")) {
PathState state;
state = path_state_from_string(value);
if (state < 0)
log_unit_debug(u, "Failed to parse state value: %s", value);
else
p->deserialized_state = state;
} else if (streq(key, "result")) {
PathResult f;
f = path_result_from_string(value);
if (f < 0)
log_unit_debug(u, "Failed to parse result value: %s", value);
else if (f != PATH_SUCCESS)
p->result = f;
} else
log_unit_debug(u, "Unknown serialization key: %s", key);
return 0;
}
_pure_ static UnitActiveState path_active_state(Unit *u) {
assert(u);
return state_translation_table[PATH(u)->state];
}
_pure_ static const char *path_sub_state_to_string(Unit *u) {
assert(u);
return path_state_to_string(PATH(u)->state);
}
static int path_dispatch_io(sd_event_source *source, int fd, uint32_t revents, void *userdata) {
PathSpec *s = userdata;
Path *p;
int changed;
assert(s);
assert(s->unit);
assert(fd >= 0);
p = PATH(s->unit);
if (p->state != PATH_WAITING &&
p->state != PATH_RUNNING)
return 0;
/* log_debug("inotify wakeup on %s.", u->id); */
LIST_FOREACH(spec, s, p->specs)
if (path_spec_owns_inotify_fd(s, fd))
break;
if (!s) {
log_error("Got event on unknown fd.");
goto fail;
}
changed = path_spec_fd_event(s, revents);
if (changed < 0)
goto fail;
/* If we are already running, then remember that one event was
* dispatched so that we restart the service only if something
* actually changed on disk */
p->inotify_triggered = true;
if (changed)
path_enter_running(p);
else
path_enter_waiting(p, false, true);
return 0;
fail:
path_enter_dead(p, PATH_FAILURE_RESOURCES);
return 0;
}
static void path_trigger_notify(Unit *u, Unit *other) {
Path *p = PATH(u);
assert(u);
assert(other);
/* Invoked whenever the unit we trigger changes state or gains
* or loses a job */
if (other->load_state != UNIT_LOADED)
return;
if (p->state == PATH_RUNNING &&
UNIT_IS_INACTIVE_OR_FAILED(unit_active_state(other))) {
log_unit_debug(UNIT(p), "Got notified about unit deactivation.");
/* Hmm, so inotify was triggered since the
* last activation, so I guess we need to
* recheck what is going on. */
path_enter_waiting(p, false, p->inotify_triggered);
}
}
static void path_reset_failed(Unit *u) {
Path *p = PATH(u);
assert(p);
if (p->state == PATH_FAILED)
path_set_state(p, PATH_DEAD);
p->result = PATH_SUCCESS;
}
static const char* const path_type_table[_PATH_TYPE_MAX] = {
[PATH_EXISTS] = "PathExists",
[PATH_EXISTS_GLOB] = "PathExistsGlob",
[PATH_DIRECTORY_NOT_EMPTY] = "DirectoryNotEmpty",
[PATH_CHANGED] = "PathChanged",
[PATH_MODIFIED] = "PathModified",
};
DEFINE_STRING_TABLE_LOOKUP(path_type, PathType);
static const char* const path_result_table[_PATH_RESULT_MAX] = {
[PATH_SUCCESS] = "success",
[PATH_FAILURE_RESOURCES] = "resources",
[PATH_FAILURE_START_LIMIT_HIT] = "start-limit-hit",
};
DEFINE_STRING_TABLE_LOOKUP(path_result, PathResult);
const UnitVTable path_vtable = {
.object_size = sizeof(Path),
.sections =
"Unit\0"
"Path\0"
"Install\0",
.init = path_init,
.done = path_done,
.load = path_load,
.coldplug = path_coldplug,
.dump = path_dump,
.start = path_start,
.stop = path_stop,
.serialize = path_serialize,
.deserialize_item = path_deserialize_item,
.active_state = path_active_state,
.sub_state_to_string = path_sub_state_to_string,
.trigger_notify = path_trigger_notify,
.reset_failed = path_reset_failed,
.bus_vtable = bus_path_vtable
};