job.c revision bcfce235a3d757f0ec3b0c01e8583382db1adfe2
/*-*- Mode: C; c-basic-offset: 8; indent-tabs-mode: nil -*-*/
/***
This file is part of systemd.
Copyright 2010 Lennart Poettering
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 <assert.h>
#include <errno.h>
#include "sd-id128.h"
#include "sd-messages.h"
#include "set.h"
#include "unit.h"
#include "macro.h"
#include "strv.h"
#include "load-fragment.h"
#include "load-dropin.h"
#include "log.h"
#include "dbus-job.h"
#include "special.h"
#include "async.h"
#include "virt.h"
#include "dbus-client-track.h"
Job *j;
/* used for deserialization */
if (!j)
return NULL;
j->type = _JOB_TYPE_INVALID;
return j;
}
Job *j;
j = job_new_raw(unit);
if (!j)
return NULL;
/* We don't link it here, that's what job_dependency() is for */
return j;
}
assert(j);
assert(!j->transaction_prev);
assert(!j->transaction_next);
assert(!j->subject_list);
assert(!j->object_list);
if (j->in_run_queue)
if (j->in_dbus_queue)
free(j);
}
void job_uninstall(Job *j) {
/* Detach from next 'bigger' objects */
/* daemon-reload should be transparent to job observers */
if (j->manager->n_reloading <= 0)
unit_add_to_gc_queue(j->unit);
j->installed = false;
}
static bool job_type_allows_late_merge(JobType t) {
/* Tells whether it is OK to merge a job of type 't' with an already
* running job.
* Reloads cannot be merged this way. Think of the sequence:
* 1. Reload of a daemon is in progress; the daemon has already loaded
* its config file, but hasn't completed the reload operation yet.
* 2. Edit foo's config file.
* 3. Trigger another reload to have the daemon use the new config.
* Should the second reload job be merged into the first one, the daemon
* would not know about the new config.
* JOB_RESTART jobs on the other hand can be merged, because they get
* patched into JOB_START after stopping the unit. So if we see a
* JOB_RESTART running, it means the unit hasn't stopped yet and at
* this time the merge is still allowed. */
return t != JOB_RELOAD;
}
else
}
if (uj) {
else {
/* not conflicting, i.e. mergeable */
"Merged into installed job %s/%s as %u",
return uj;
} else {
/* already running and not safe to merge into */
/* Patch uj to become a merged job and re-run it. */
/* XXX It should be safer to queue j to run after uj finishes, but it is
* not currently possible to have more than one installed job per unit. */
"Merged into running job, re-running: %s/%s as %u",
return uj;
}
}
}
/* Install the job */
*pj = j;
j->installed = true;
j->manager->n_installed_jobs ++;
"Installed new job %s/%s as %u",
return j;
}
int job_install_deserialized(Job *j) {
return -EINVAL;
}
if (*pj) {
"Unit %s already has a job installed. Not installing deserialized job.",
return -EEXIST;
}
*pj = j;
j->installed = true;
"Reinstalled deserialized job %s/%s as %u",
return 0;
}
JobDependency *l;
/* Adds a new job link, which encodes that the 'subject' job
* needs the 'object' job in some way. If 'subject' is NULL
* this means the 'anchor' job (i.e. the one the user
* explicitly asked for) is the requester. */
return NULL;
if (subject)
return l;
}
void job_dependency_free(JobDependency *l) {
assert(l);
if (l->subject)
free(l);
}
assert(j);
assert(f);
if (!prefix)
prefix = "";
fprintf(f,
"%s-> Job %u:\n"
"%s\tAction: %s -> %s\n"
"%s\tState: %s\n"
"%s\tForced: %s\n"
"%s\tIrreversible: %s\n",
}
/*
* Merging is commutative, so imagine the matrix as symmetric. We store only
* its lower triangle to avoid duplication. We don't store the main diagonal,
* because A merged with A is simply A.
*
* If the resulting type is collapsed immediately afterwards (to get rid of
* the JOB_RELOAD_OR_START, which lies outside the lookup function's domain),
* the following properties hold:
*
* Merging is associative! A merged with B merged with C is the same as
* A merged with C merged with B.
*
* Mergeability is transitive! If A can be merged with B and B with C then
* A also with C.
*
* Also, if A merged with B cannot be merged with C, then either A or B cannot
* be merged with C either.
*/
static const JobType job_merging_table[] = {
/* What \ With * JOB_START JOB_VERIFY_ACTIVE JOB_STOP JOB_RELOAD */
/*********************************************************************************/
/*JOB_START */
/*JOB_VERIFY_ACTIVE */ JOB_START,
/*JOB_STOP */ -1, -1,
};
assert_cc(ELEMENTSOF(job_merging_table) == _JOB_TYPE_MAX_MERGING * (_JOB_TYPE_MAX_MERGING - 1) / 2);
assert(a >= 0 && a < _JOB_TYPE_MAX_MERGING);
assert(b >= 0 && b < _JOB_TYPE_MAX_MERGING);
if (a == b)
return a;
if (a < b) {
a = b;
b = tmp;
}
}
switch (a) {
case JOB_START:
return
b == UNIT_ACTIVE ||
b == UNIT_RELOADING;
case JOB_STOP:
return
b == UNIT_INACTIVE ||
b == UNIT_FAILED;
case JOB_VERIFY_ACTIVE:
return
b == UNIT_ACTIVE ||
b == UNIT_RELOADING;
case JOB_RELOAD:
return
b == UNIT_RELOADING;
case JOB_RESTART:
return
b == UNIT_ACTIVATING;
default:
assert_not_reached("Invalid job type");
}
}
switch (*t) {
case JOB_TRY_RESTART:
s = unit_active_state(u);
if (UNIT_IS_INACTIVE_OR_DEACTIVATING(s))
*t = JOB_NOP;
else
*t = JOB_RESTART;
break;
case JOB_RELOAD_OR_START:
s = unit_active_state(u);
if (UNIT_IS_INACTIVE_OR_DEACTIVATING(s))
*t = JOB_START;
else
*t = JOB_RELOAD;
break;
default:
;
}
}
JobType t = job_type_lookup_merge(*a, b);
if (t < 0)
return -EEXIST;
*a = t;
job_type_collapse(a, u);
return 0;
}
static bool job_is_runnable(Job *j) {
Iterator i;
assert(j);
/* Checks whether there is any job running for the units this
* job needs to be running after (in the case of a 'positive'
* job type) or before (in the case of a 'negative' job
* type. */
/* Note that unit types have a say in what is runnable,
* too. For example, if they return -EAGAIN from
* unit_start() they can indicate they are not
* runnable yet. */
/* First check if there is an override */
if (j->ignore_order)
return true;
return true;
j->type == JOB_VERIFY_ACTIVE ||
j->type == JOB_RELOAD) {
/* Immediate result is that the job is or might be
* started. In this case lets wait for the
* dependencies, regardless whether they are
* starting or stopping something. */
return false;
}
/* Also, if something else is being stopped and we should
* change state after it, then lets wait. */
return false;
/* This means that for a service a and a service b where b
* shall be started after a:
*
* start a + start b → 1st step start a, 2nd step start b
* start a + stop b → 1st step stop b, 2nd step start a
* stop a + start b → 1st step stop a, 2nd step start b
* stop a + stop b → 1st step stop b, 2nd step stop a
*
* This has the side effect that restarts are properly
* synchronized too. */
return true;
}
"Converting job %s/%s -> %s/%s",
}
int job_run_and_invalidate(Job *j) {
int r;
assert(j);
assert(j->in_run_queue);
j->in_run_queue = false;
if (j->state != JOB_WAITING)
return 0;
if (!job_is_runnable(j))
return -EAGAIN;
j->state = JOB_RUNNING;
m->n_running_jobs++;
/* While we execute this operation the job might go away (for
* example: because it is replaced by a new, conflicting
* job.) To make sure we don't access a freed job later on we
* store the id here, so that we can verify the job is still
* valid. */
switch (j->type) {
case JOB_START:
r = unit_start(j->unit);
/* If this unit cannot be started, then simply wait */
if (r == -EBADR)
r = 0;
break;
case JOB_VERIFY_ACTIVE: {
if (UNIT_IS_ACTIVE_OR_RELOADING(t))
r = -EALREADY;
else if (t == UNIT_ACTIVATING)
r = -EAGAIN;
else
r = -EBADR;
break;
}
case JOB_STOP:
case JOB_RESTART:
/* If this unit cannot stopped, then simply wait. */
if (r == -EBADR)
r = 0;
break;
case JOB_RELOAD:
r = unit_reload(j->unit);
break;
case JOB_NOP:
r = -EALREADY;
break;
default:
assert_not_reached("Unknown job type");
}
j = manager_get_job(m, id);
if (j) {
if (r == -EALREADY)
r = job_finish_and_invalidate(j, JOB_DONE, true);
else if (r == -EBADR)
r = job_finish_and_invalidate(j, JOB_SKIPPED, true);
else if (r == -ENOEXEC)
r = job_finish_and_invalidate(j, JOB_INVALID, true);
else if (r == -EAGAIN) {
j->state = JOB_WAITING;
m->n_running_jobs--;
} else if (r < 0)
r = job_finish_and_invalidate(j, JOB_FAILED, true);
}
return r;
}
const UnitStatusMessageFormats *format_table;
assert(u);
assert(t >= 0);
assert(t < _JOB_TYPE_MAX);
if (!format_table)
return NULL;
if (t == JOB_START)
else if (t == JOB_STOP || t == JOB_RESTART)
return NULL;
}
_pure_ static const char *job_get_status_message_format_try_harder(Unit *u, JobType t, JobResult result) {
const char *format;
assert(u);
assert(t >= 0);
assert(t < _JOB_TYPE_MAX);
if (format)
return format;
/* Return generic strings */
if (t == JOB_START) {
return "Started %s.";
else if (result == JOB_FAILED)
return "Failed to start %s.";
else if (result == JOB_DEPENDENCY)
return "Dependency failed for %s.";
else if (result == JOB_TIMEOUT)
return "Timed out starting %s.";
} else if (t == JOB_STOP || t == JOB_RESTART) {
return "Stopped %s.";
else if (result == JOB_FAILED)
return "Stopped (with error) %s.";
else if (result == JOB_TIMEOUT)
return "Timed out stoppping %s.";
} else if (t == JOB_RELOAD) {
return "Reloaded %s.";
else if (result == JOB_FAILED)
return "Reload failed for %s.";
else if (result == JOB_TIMEOUT)
return "Timed out reloading %s.";
}
return NULL;
}
const char *format;
assert(u);
assert(t >= 0);
assert(t < _JOB_TYPE_MAX);
if (t == JOB_START) {
if (!format)
return;
switch (result) {
case JOB_DONE:
if (u->condition_result)
break;
case JOB_FAILED:
manager_flip_auto_status(u->manager, true);
break;
case JOB_DEPENDENCY:
manager_flip_auto_status(u->manager, true);
break;
case JOB_TIMEOUT:
manager_flip_auto_status(u->manager, true);
break;
default:
;
}
} else if (t == JOB_STOP || t == JOB_RESTART) {
if (!format)
return;
switch (result) {
case JOB_TIMEOUT:
manager_flip_auto_status(u->manager, true);
break;
case JOB_DONE:
case JOB_FAILED:
break;
default:
;
}
} else if (t == JOB_VERIFY_ACTIVE) {
/* When verify-active detects the unit is inactive, report it.
* Most likely a DEPEND warning from a requisiting unit will
* occur next and it's nice to see what was requisited. */
if (result == JOB_SKIPPED)
}
}
const char *format;
assert(u);
assert(t >= 0);
assert(t < _JOB_TYPE_MAX);
/* Skip this if it goes to the console. since we already print
* to the console anyway... */
if (log_on_console())
return;
if (!format)
return;
if (t == JOB_START) {
u->id,
"MESSAGE=%s", buf,
NULL);
} else if (t == JOB_STOP)
u->id,
"MESSAGE=%s", buf,
NULL);
else if (t == JOB_RELOAD)
u->id,
"MESSAGE=%s", buf,
NULL);
}
Unit *u;
JobType t;
Iterator i;
assert(j);
u = j->unit;
t = j->type;
if (j->state == JOB_RUNNING)
j->manager->n_running_jobs--;
job_print_status_message(u, t, result);
job_log_status_message(u, t, result);
/* Patch restart jobs so that they become normal start jobs */
job_change_type(j, JOB_START);
j->state = JOB_WAITING;
goto finish;
}
j->manager->n_failed_jobs ++;
job_uninstall(j);
job_free(j);
/* Fail depending jobs on failure */
if (t == JOB_START ||
t == JOB_VERIFY_ACTIVE) {
} else if (t == JOB_STOP) {
}
}
/* Trigger OnFailure dependencies that are not generated by
* the unit itself. We don't treat JOB_CANCELED as failure in
* this context. And JOB_FAILURE is already handled by the
* unit itself. */
u->id,
"JOB_TYPE=%s", job_type_to_string(t),
"Job %s/%s failed with result '%s'.",
u->id,
NULL);
}
/* Try to start the next jobs that can be started */
return 0;
}
assert(j);
assert(s == j->timer_event_source);
job_finish_and_invalidate(j, JOB_TIMEOUT, true);
return 0;
}
int job_start_timer(Job *j) {
int r;
if (j->timer_event_source)
return 0;
if (j->unit->job_timeout <= 0)
return 0;
r = sd_event_add_monotonic(j->manager->event, &j->timer_event_source, j->begin_usec + j->unit->job_timeout, 0, job_dispatch_timer, j);
if (r < 0)
return r;
return 0;
}
void job_add_to_run_queue(Job *j) {
assert(j);
if (j->in_run_queue)
return;
j->in_run_queue = true;
}
void job_add_to_dbus_queue(Job *j) {
assert(j);
if (j->in_dbus_queue)
return;
/* We don't check if anybody is subscribed here, since this
* job might just have been created and not yet assigned to a
* connection/client. */
j->in_dbus_queue = true;
}
char *job_dbus_path(Job *j) {
char *p;
assert(j);
return NULL;
return p;
}
if (j->begin_usec > 0)
/* End marker */
fputc('\n', f);
return 0;
}
assert(j);
for (;;) {
size_t k;
if (feof(f))
return 0;
return -errno;
}
/* End marker */
if (l[0] == 0)
return 0;
k = strcspn(l, "=");
if (l[k] == '=') {
l[k] = 0;
v = l+k+1;
} else
v = l+k;
if (streq(l, "job-id")) {
if (safe_atou32(v, &j->id) < 0)
log_debug("Failed to parse job id value %s", v);
} else if (streq(l, "job-type")) {
JobType t;
t = job_type_from_string(v);
if (t < 0)
log_debug("Failed to parse job type %s", v);
else if (t >= _JOB_TYPE_MAX_IN_TRANSACTION)
log_debug("Cannot deserialize job of type %s", v);
else
j->type = t;
} else if (streq(l, "job-state")) {
JobState s;
s = job_state_from_string(v);
if (s < 0)
log_debug("Failed to parse job state %s", v);
else
j->state = s;
} else if (streq(l, "job-override")) {
int b;
b = parse_boolean(v);
if (b < 0)
log_debug("Failed to parse job override flag %s", v);
else
} else if (streq(l, "job-irreversible")) {
int b;
b = parse_boolean(v);
if (b < 0)
log_debug("Failed to parse job irreversible flag %s", v);
else
j->irreversible = j->irreversible || b;
} else if (streq(l, "job-sent-dbus-new-signal")) {
int b;
b = parse_boolean(v);
if (b < 0)
log_debug("Failed to parse job sent_dbus_new_signal flag %s", v);
else
j->sent_dbus_new_signal = j->sent_dbus_new_signal || b;
} else if (streq(l, "job-ignore-order")) {
int b;
b = parse_boolean(v);
if (b < 0)
log_debug("Failed to parse job ignore_order flag %s", v);
else
j->ignore_order = j->ignore_order || b;
} else if (streq(l, "job-begin")) {
unsigned long long ull;
log_debug("Failed to parse job-begin value %s", v);
else
j->begin_usec = ull;
} else {
log_debug("Unknown deserialization key '%s'", l);
}
}
}
int job_coldplug(Job *j) {
int r;
assert(j);
return 0;
if (j->timer_event_source)
r = sd_event_add_monotonic(j->manager->event, &j->timer_event_source, j->begin_usec + j->unit->job_timeout, 0, job_dispatch_timer, j);
if (r < 0)
return r;
}
void job_shutdown_magic(Job *j) {
assert(j);
/* The shutdown target gets some special treatment here: we
* tell the kernel to begin with flushing its disk caches, to
* optimize shutdown time a bit. Ideally we wouldn't hardcode
* this magic into PID 1. However all other processes aren't
* options either since they'd exit much sooner than PID 1 and
* asynchronous sync() would cause their exit to be
* delayed. */
return;
return;
return;
/* In case messages on console has been disabled on boot */
if (detect_container(NULL) > 0)
return;
}
int r = 0, q = 0;
assert(u);
if (j->timer_event_source) {
r = sd_event_source_get_time(j->timer_event_source, &x);
if (r < 0)
return r;
r = 1;
}
if (UNIT_VTABLE(u)->get_timeout) {
q = UNIT_VTABLE(u)->get_timeout(u, &y);
if (q < 0)
return q;
}
if (r == 0 && q == 0)
return 0;
return 1;
}
static const char* const job_state_table[_JOB_STATE_MAX] = {
[JOB_WAITING] = "waiting",
[JOB_RUNNING] = "running"
};
static const char* const job_type_table[_JOB_TYPE_MAX] = {
[JOB_START] = "start",
[JOB_VERIFY_ACTIVE] = "verify-active",
[JOB_STOP] = "stop",
[JOB_RELOAD] = "reload",
[JOB_RELOAD_OR_START] = "reload-or-start",
[JOB_RESTART] = "restart",
[JOB_TRY_RESTART] = "try-restart",
[JOB_NOP] = "nop",
};
static const char* const job_mode_table[_JOB_MODE_MAX] = {
[JOB_FAIL] = "fail",
[JOB_REPLACE] = "replace",
[JOB_REPLACE_IRREVERSIBLY] = "replace-irreversibly",
[JOB_ISOLATE] = "isolate",
[JOB_IGNORE_DEPENDENCIES] = "ignore-dependencies",
[JOB_IGNORE_REQUIREMENTS] = "ignore-requirements",
[JOB_FLUSH] = "flush",
};
static const char* const job_result_table[_JOB_RESULT_MAX] = {
[JOB_DONE] = "done",
[JOB_CANCELED] = "canceled",
[JOB_TIMEOUT] = "timeout",
[JOB_FAILED] = "failed",
[JOB_DEPENDENCY] = "dependency",
[JOB_SKIPPED] = "skipped",
[JOB_INVALID] = "invalid",
};