Systemd/src/core/unit.h

677 lines
23 KiB
C
Raw Normal View History

#pragma once
2010-01-26 21:39:06 +01:00
/***
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/>.
***/
2010-01-26 21:39:06 +01:00
#include <stdbool.h>
#include <stdlib.h>
#include <unistd.h>
2010-01-26 21:39:06 +01:00
typedef struct Unit Unit;
2010-01-26 21:39:06 +01:00
typedef struct UnitVTable UnitVTable;
typedef struct UnitRef UnitRef;
typedef struct UnitStatusMessageFormats UnitStatusMessageFormats;
2010-01-26 21:39:06 +01:00
#include "condition.h"
#include "failure-action.h"
2011-07-31 18:28:02 +02:00
#include "install.h"
#include "list.h"
#include "unit-name.h"
2010-01-26 21:39:06 +01:00
typedef enum KillOperation {
KILL_TERMINATE,
KILL_TERMINATE_AND_LOG,
KILL_KILL,
KILL_ABORT,
_KILL_OPERATION_MAX,
_KILL_OPERATION_INVALID = -1
} KillOperation;
2010-01-26 21:39:06 +01:00
static inline bool UNIT_IS_ACTIVE_OR_RELOADING(UnitActiveState t) {
return t == UNIT_ACTIVE || t == UNIT_RELOADING;
2010-01-26 21:39:06 +01:00
}
static inline bool UNIT_IS_ACTIVE_OR_ACTIVATING(UnitActiveState t) {
return t == UNIT_ACTIVE || t == UNIT_ACTIVATING || t == UNIT_RELOADING;
2010-01-26 21:39:06 +01:00
}
static inline bool UNIT_IS_INACTIVE_OR_DEACTIVATING(UnitActiveState t) {
return t == UNIT_INACTIVE || t == UNIT_FAILED || t == UNIT_DEACTIVATING;
}
static inline bool UNIT_IS_INACTIVE_OR_FAILED(UnitActiveState t) {
return t == UNIT_INACTIVE || t == UNIT_FAILED;
2010-01-26 21:39:06 +01:00
}
2010-01-29 06:04:08 +01:00
#include "job.h"
struct UnitRef {
/* Keeps tracks of references to a unit. This is useful so
* that we can merge two units if necessary and correct all
* references to them */
Unit* unit;
LIST_FIELDS(UnitRef, refs);
};
struct Unit {
2010-01-26 21:39:06 +01:00
Manager *manager;
2010-04-06 02:43:58 +02:00
2010-01-26 21:39:06 +01:00
UnitType type;
UnitLoadState load_state;
2010-04-06 02:43:58 +02:00
Unit *merged_into;
2010-01-26 21:39:06 +01:00
char *id; /* One name is special because we use it for identification. Points to an entry in the names set */
2010-04-15 03:11:11 +02:00
char *instance;
2010-01-26 21:39:06 +01:00
Set *names;
Set *dependencies[_UNIT_DEPENDENCY_MAX];
char **requires_mounts_for;
2010-01-26 21:39:06 +01:00
char *description;
char **documentation;
char *fragment_path; /* if loaded from a config file this is the primary path to it */
char *source_path; /* if converted, the source file */
char **dropin_paths;
usec_t fragment_mtime;
usec_t source_mtime;
usec_t dropin_mtime;
2010-01-26 21:39:06 +01:00
/* If this is a transient unit we are currently writing, this is where we are writing it to */
FILE *transient_file;
core: add NOP jobs, job type collapsing Two of our current job types are special: JOB_TRY_RESTART, JOB_RELOAD_OR_START. They differ from other job types by being sensitive to the unit active state. They perform some action when the unit is active and some other action otherwise. This raises a question: when exactly should the unit state be checked to make the decision? Currently the unit state is checked when the job becomes runnable. It's more sensible to check the state immediately when the job is added by the user. When the user types "systemctl try-restart foo.service", he really intends to restart the service if it's running right now. If it isn't running right now, the restart is pointless. Consider the example (from Bugzilla[1]): sleep.service takes some time to start. hello.service has After=sleep.service. Both services get started. Two jobs will appear: hello.service/start waiting sleep.service/start running Then someone runs "systemctl try-restart hello.service". Currently the try-restart operation will block and wait for sleep.service/start to complete. The correct result is to complete the try-restart operation immediately with success, because hello.service is not running. The two original jobs must not be disturbed by this. To fix this we introduce two new concepts: - a new job type: JOB_NOP A JOB_NOP job does not do anything to the unit. It does not pull in any dependencies. It is always immediately runnable. When installed to a unit, it sits in a special slot (u->nop_job) where it never conflicts with the installed job (u->job) of a different type. It never merges with jobs of other types, but it can merge into an already installed JOB_NOP job. - "collapsing" of job types When a job of one of the two special types is added, the state of the unit is checked immediately and the job type changes: JOB_TRY_RESTART -> JOB_RESTART or JOB_NOP JOB_RELOAD_OR_START -> JOB_RELOAD or JOB_START Should a job type JOB_RELOAD_OR_START appear later during job merging, it collapses immediately afterwards. Collapsing actually makes some things simpler, because there are now fewer job types that are allowed in the transaction. [1] Fixes: https://bugzilla.redhat.com/show_bug.cgi?id=753586
2012-04-25 11:58:27 +02:00
/* If there is something to do with this unit, then this is the installed job for it */
2010-01-26 21:39:06 +01:00
Job *job;
core: add NOP jobs, job type collapsing Two of our current job types are special: JOB_TRY_RESTART, JOB_RELOAD_OR_START. They differ from other job types by being sensitive to the unit active state. They perform some action when the unit is active and some other action otherwise. This raises a question: when exactly should the unit state be checked to make the decision? Currently the unit state is checked when the job becomes runnable. It's more sensible to check the state immediately when the job is added by the user. When the user types "systemctl try-restart foo.service", he really intends to restart the service if it's running right now. If it isn't running right now, the restart is pointless. Consider the example (from Bugzilla[1]): sleep.service takes some time to start. hello.service has After=sleep.service. Both services get started. Two jobs will appear: hello.service/start waiting sleep.service/start running Then someone runs "systemctl try-restart hello.service". Currently the try-restart operation will block and wait for sleep.service/start to complete. The correct result is to complete the try-restart operation immediately with success, because hello.service is not running. The two original jobs must not be disturbed by this. To fix this we introduce two new concepts: - a new job type: JOB_NOP A JOB_NOP job does not do anything to the unit. It does not pull in any dependencies. It is always immediately runnable. When installed to a unit, it sits in a special slot (u->nop_job) where it never conflicts with the installed job (u->job) of a different type. It never merges with jobs of other types, but it can merge into an already installed JOB_NOP job. - "collapsing" of job types When a job of one of the two special types is added, the state of the unit is checked immediately and the job type changes: JOB_TRY_RESTART -> JOB_RESTART or JOB_NOP JOB_RELOAD_OR_START -> JOB_RELOAD or JOB_START Should a job type JOB_RELOAD_OR_START appear later during job merging, it collapses immediately afterwards. Collapsing actually makes some things simpler, because there are now fewer job types that are allowed in the transaction. [1] Fixes: https://bugzilla.redhat.com/show_bug.cgi?id=753586
2012-04-25 11:58:27 +02:00
/* JOB_NOP jobs are special and can be installed without disturbing the real job. */
Job *nop_job;
/* The slot used for watching NameOwnerChanged signals */
sd_bus_slot *match_bus_slot;
/* References to this unit from clients */
sd_bus_track *bus_track;
char **deserialized_refs;
/* Job timeout and action to take */
usec_t job_timeout;
FailureAction job_timeout_action;
char *job_timeout_reboot_arg;
/* References to this */
LIST_HEAD(UnitRef, refs);
/* Conditions to check */
LIST_HEAD(Condition, conditions);
LIST_HEAD(Condition, asserts);
dual_timestamp condition_timestamp;
dual_timestamp assert_timestamp;
/* Updated whenever the low-level state changes */
dual_timestamp state_change_timestamp;
/* Updated whenever the (high-level) active state enters or leaves the active or inactive states */
dual_timestamp inactive_exit_timestamp;
dual_timestamp active_enter_timestamp;
dual_timestamp active_exit_timestamp;
dual_timestamp inactive_enter_timestamp;
2010-01-26 21:39:06 +01:00
UnitRef slice;
2010-01-29 06:04:08 +01:00
/* Per type list */
LIST_FIELDS(Unit, units_by_type);
/* All units which have requires_mounts_for set */
LIST_FIELDS(Unit, has_requires_mounts_for);
/* Load queue */
LIST_FIELDS(Unit, load_queue);
/* D-Bus queue */
LIST_FIELDS(Unit, dbus_queue);
2010-04-06 02:43:58 +02:00
/* Cleanup queue */
LIST_FIELDS(Unit, cleanup_queue);
/* GC queue */
LIST_FIELDS(Unit, gc_queue);
/* CGroup realize members queue */
LIST_FIELDS(Unit, cgroup_queue);
/* Units with the same CGroup netclass */
LIST_FIELDS(Unit, cgroup_netclass);
/* PIDs we keep an eye on. Note that a unit might have many
* more, but these are the ones we care enough about to
* process SIGCHLD for */
Set *pids;
/* Used in sigchld event invocation to avoid repeat events being invoked */
uint64_t sigchldgen;
/* Used during GC sweeps */
2010-04-23 18:47:49 +02:00
unsigned gc_marker;
/* Error code when we didn't manage to load the unit (negative) */
int load_error;
/* Put a ratelimit on unit starting */
RateLimit start_limit;
FailureAction start_limit_action;
char *reboot_arg;
/* Make sure we never enter endless loops with the check unneeded logic, or the BindsTo= logic */
RateLimit auto_stop_ratelimit;
/* Reference to a specific UID/GID */
uid_t ref_uid;
gid_t ref_gid;
/* Cached unit file state and preset */
2011-07-31 18:28:02 +02:00
UnitFileState unit_file_state;
int unit_file_preset;
2011-07-31 18:28:02 +02:00
/* Where the cpu.stat or cpuacct.usage was at the time the unit was started */
nsec_t cpu_usage_base;
nsec_t cpu_usage_last; /* the most recently read value */
/* Counterparts in the cgroup filesystem */
char *cgroup_path;
core: unified cgroup hierarchy support This patch set adds full support the new unified cgroup hierarchy logic of modern kernels. A new kernel command line option "systemd.unified_cgroup_hierarchy=1" is added. If specified the unified hierarchy is mounted to /sys/fs/cgroup instead of a tmpfs. No further hierarchies are mounted. The kernel command line option defaults to off. We can turn it on by default as soon as the kernel's APIs regarding this are stabilized (but even then downstream distros might want to turn this off, as this will break any tools that access cgroupfs directly). It is possibly to choose for each boot individually whether the unified or the legacy hierarchy is used. nspawn will by default provide the legacy hierarchy to containers if the host is using it, and the unified otherwise. However it is possible to run containers with the unified hierarchy on a legacy host and vice versa, by setting the $UNIFIED_CGROUP_HIERARCHY environment variable for nspawn to 1 or 0, respectively. The unified hierarchy provides reliable cgroup empty notifications for the first time, via inotify. To make use of this we maintain one manager-wide inotify fd, and each cgroup to it. This patch also removes cg_delete() which is unused now. On kernel 4.2 only the "memory" controller is compatible with the unified hierarchy, hence that's the only controller systemd exposes when booted in unified heirarchy mode. This introduces a new enum for enumerating supported controllers, plus a related enum for the mask bits mapping to it. The core is changed to make use of this everywhere. This moves PID 1 into a new "init.scope" implicit scope unit in the root slice. This is necessary since on the unified hierarchy cgroups may either contain subgroups or processes but not both. PID 1 hence has to move out of the root cgroup (strictly speaking the root cgroup is the only one where processes and subgroups are still allowed, but in order to support containers nicey, we move PID 1 into the new scope in all cases.) This new unit is also used on legacy hierarchy setups. It's actually pretty useful on all systems, as it can then be used to filter journal messages coming from PID 1, and so on. The root slice ("-.slice") is now implicitly created and started (and does not require a unit file on disk anymore), since that's where "init.scope" is located and the slice needs to be started before the scope can. To check whether we are in unified or legacy hierarchy mode we use statfs() on /sys/fs/cgroup. If the .f_type field reports tmpfs we are in legacy mode, if it reports cgroupfs we are in unified mode. This patch set carefuly makes sure that cgls and cgtop continue to work as desired. When invoking nspawn as a service it will implicitly create two subcgroups in the cgroup it is using, one to move the nspawn process into, the other to move the actual container processes into. This is done because of the requirement that cgroups may either contain processes or other subgroups.
2015-09-01 19:22:36 +02:00
CGroupMask cgroup_realized_mask;
CGroupMask cgroup_enabled_mask;
core: unified cgroup hierarchy support This patch set adds full support the new unified cgroup hierarchy logic of modern kernels. A new kernel command line option "systemd.unified_cgroup_hierarchy=1" is added. If specified the unified hierarchy is mounted to /sys/fs/cgroup instead of a tmpfs. No further hierarchies are mounted. The kernel command line option defaults to off. We can turn it on by default as soon as the kernel's APIs regarding this are stabilized (but even then downstream distros might want to turn this off, as this will break any tools that access cgroupfs directly). It is possibly to choose for each boot individually whether the unified or the legacy hierarchy is used. nspawn will by default provide the legacy hierarchy to containers if the host is using it, and the unified otherwise. However it is possible to run containers with the unified hierarchy on a legacy host and vice versa, by setting the $UNIFIED_CGROUP_HIERARCHY environment variable for nspawn to 1 or 0, respectively. The unified hierarchy provides reliable cgroup empty notifications for the first time, via inotify. To make use of this we maintain one manager-wide inotify fd, and each cgroup to it. This patch also removes cg_delete() which is unused now. On kernel 4.2 only the "memory" controller is compatible with the unified hierarchy, hence that's the only controller systemd exposes when booted in unified heirarchy mode. This introduces a new enum for enumerating supported controllers, plus a related enum for the mask bits mapping to it. The core is changed to make use of this everywhere. This moves PID 1 into a new "init.scope" implicit scope unit in the root slice. This is necessary since on the unified hierarchy cgroups may either contain subgroups or processes but not both. PID 1 hence has to move out of the root cgroup (strictly speaking the root cgroup is the only one where processes and subgroups are still allowed, but in order to support containers nicey, we move PID 1 into the new scope in all cases.) This new unit is also used on legacy hierarchy setups. It's actually pretty useful on all systems, as it can then be used to filter journal messages coming from PID 1, and so on. The root slice ("-.slice") is now implicitly created and started (and does not require a unit file on disk anymore), since that's where "init.scope" is located and the slice needs to be started before the scope can. To check whether we are in unified or legacy hierarchy mode we use statfs() on /sys/fs/cgroup. If the .f_type field reports tmpfs we are in legacy mode, if it reports cgroupfs we are in unified mode. This patch set carefuly makes sure that cgls and cgtop continue to work as desired. When invoking nspawn as a service it will implicitly create two subcgroups in the cgroup it is using, one to move the nspawn process into, the other to move the actual container processes into. This is done because of the requirement that cgroups may either contain processes or other subgroups.
2015-09-01 19:22:36 +02:00
CGroupMask cgroup_subtree_mask;
CGroupMask cgroup_members_mask;
int cgroup_inotify_wd;
/* How to start OnFailure units */
JobMode on_failure_job_mode;
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.
2016-08-30 23:18:46 +02:00
/* The current invocation ID */
sd_id128_t invocation_id;
char invocation_id_string[SD_ID128_STRING_MAX]; /* useful when logging */
/* Garbage collect us we nobody wants or requires us anymore */
bool stop_when_unneeded;
/* Create default dependencies */
bool default_dependencies;
/* Refuse manual starting, allow starting only indirectly via dependency. */
bool refuse_manual_start;
/* Don't allow the user to stop this unit manually, allow stopping only indirectly via dependency. */
bool refuse_manual_stop;
2010-08-30 22:45:46 +02:00
/* Allow isolation requests */
bool allow_isolate;
/* Ignore this unit when isolating */
bool ignore_on_isolate;
2012-07-15 10:41:40 +02:00
/* Did the last condition check succeed? */
bool condition_result;
bool assert_result;
/* Is this a transient unit? */
bool transient;
bool in_load_queue:1;
bool in_dbus_queue:1;
bool in_cleanup_queue:1;
bool in_gc_queue:1;
bool in_cgroup_queue:1;
bool sent_dbus_new_signal:1;
bool no_gc:1;
bool in_audit:1;
bool cgroup_realized:1;
bool cgroup_members_mask_valid:1;
bool cgroup_subtree_mask_valid:1;
bool start_limit_hit:1;
/* Did we already invoke unit_coldplug() for this unit? */
2015-05-19 16:03:01 +02:00
bool coldplugged:1;
/* For transient units: whether to add a bus track reference after creating the unit */
bool bus_track_add:1;
2010-01-26 21:39:06 +01:00
};
struct UnitStatusMessageFormats {
const char *starting_stopping[2];
const char *finished_start_job[_JOB_RESULT_MAX];
const char *finished_stop_job[_JOB_RESULT_MAX];
};
typedef enum UnitSetPropertiesMode {
UNIT_CHECK = 0,
UNIT_RUNTIME = 1,
UNIT_PERSISTENT = 2,
} UnitSetPropertiesMode;
#include "automount.h"
#include "busname.h"
2010-01-26 21:39:06 +01:00
#include "device.h"
#include "path.h"
#include "scope.h"
#include "slice.h"
#include "socket.h"
#include "swap.h"
#include "target.h"
#include "timer.h"
2010-01-26 21:39:06 +01:00
struct UnitVTable {
/* How much memory does an object of this unit type need */
size_t object_size;
/* If greater than 0, the offset into the object where
* ExecContext is found, if the unit type has that */
size_t exec_context_offset;
/* If greater than 0, the offset into the object where
* CGroupContext is found, if the unit type has that */
size_t cgroup_context_offset;
/* If greater than 0, the offset into the object where
* KillContext is found, if the unit type has that */
size_t kill_context_offset;
/* If greater than 0, the offset into the object where the
* pointer to ExecRuntime is found, if the unit type has
* that */
size_t exec_runtime_offset;
/* If greater than 0, the offset into the object where the pointer to DynamicCreds is found, if the unit type
* has that. */
size_t dynamic_creds_offset;
/* The name of the configuration file section with the private settings of this unit */
const char *private_section;
/* Config file sections this unit type understands, separated
* by NUL chars */
const char *sections;
/* This should reset all type-specific variables. This should
2010-04-21 03:27:44 +02:00
* not allocate memory, and is called with zero-initialized
* data. It should hence only initialize variables that need
* to be set != 0. */
void (*init)(Unit *u);
2010-04-21 03:27:44 +02:00
/* This should free all type-specific variables. It should be
* idempotent. */
void (*done)(Unit *u);
/* Actually load data from disk. This may fail, and should set
* load_state to UNIT_LOADED, UNIT_MERGED or leave it at
* UNIT_STUB if no configuration could be found. */
int (*load)(Unit *u);
/* If a lot of units got created via enumerate(), this is
* where to actually set the state and call unit_notify(). */
int (*coldplug)(Unit *u);
2010-01-26 21:39:06 +01:00
void (*dump)(Unit *u, FILE *f, const char *prefix);
int (*start)(Unit *u);
int (*stop)(Unit *u);
int (*reload)(Unit *u);
int (*kill)(Unit *u, KillWho w, int signo, sd_bus_error *error);
2010-10-22 16:11:50 +02:00
2010-01-26 21:39:06 +01:00
bool (*can_reload)(Unit *u);
2010-04-21 03:27:44 +02:00
/* Write all data that cannot be restored from other sources
* away using unit_serialize_item() */
int (*serialize)(Unit *u, FILE *f, FDSet *fds);
/* Restore one item from the serialization */
int (*deserialize_item)(Unit *u, const char *key, const char *data, FDSet *fds);
/* Try to match up fds with what we need for this unit */
void (*distribute_fds)(Unit *u, FDSet *fds);
2010-01-26 21:39:06 +01:00
/* Boils down the more complex internal state of this unit to
* a simpler one that the engine can understand */
UnitActiveState (*active_state)(Unit *u);
/* Returns the substate specific to this unit type as
* string. This is purely information so that we can give the
* user a more fine grained explanation in which actual state a
* unit is in. */
const char* (*sub_state_to_string)(Unit *u);
/* Return true when there is reason to keep this entry around
* even nothing references it and it isn't active in any
* way */
bool (*check_gc)(Unit *u);
/* When the unit is not running and no job for it queued we
* shall release its runtime resources */
void (*release_resources)(Unit *u);
/* Invoked on every child that died */
2010-01-26 21:39:06 +01:00
void (*sigchld_event)(Unit *u, pid_t pid, int code, int status);
/* Reset failed state if we are in failed state */
void (*reset_failed)(Unit *u);
/* Called whenever any of the cgroups this unit watches for
* ran empty */
void (*notify_cgroup_empty)(Unit *u);
2010-03-31 16:29:55 +02:00
/* Called whenever a process of this unit sends us a message */
void (*notify_message)(Unit *u, pid_t pid, char **tags, FDSet *fds);
/* Called whenever a name this Unit registered for comes or goes away. */
void (*bus_name_owner_change)(Unit *u, const char *name, const char *old_owner, const char *new_owner);
/* Called for each property that is being set */
int (*bus_set_property)(Unit *u, const char *name, sd_bus_message *message, UnitSetPropertiesMode mode, sd_bus_error *error);
/* Called after at least one property got changed to apply the necessary change */
int (*bus_commit_properties)(Unit *u);
/* Return the unit this unit is following */
Unit *(*following)(Unit *u);
/* Return the set of units that are following each other */
int (*following_set)(Unit *u, Set **s);
/* Invoked each time a unit this unit is triggering changes
* state or gains/loses a job */
void (*trigger_notify)(Unit *u, Unit *trigger);
/* Called whenever CLOCK_REALTIME made a jump */
void (*time_change)(Unit *u);
/* Returns the next timeout of a unit */
int (*get_timeout)(Unit *u, usec_t *timeout);
/* Returns the main PID if there is any defined, or 0. */
pid_t (*main_pid)(Unit *u);
/* Returns the main PID if there is any defined, or 0. */
pid_t (*control_pid)(Unit *u);
2010-01-29 03:18:09 +01:00
/* This is called for each unit type and should be used to
* enumerate existing devices and load them. However,
* everything that is loaded here should still stay in
* inactive state. It is the job of the coldplug() call above
* to put the units into the initial state. */
void (*enumerate)(Manager *m);
2010-01-29 03:18:09 +01:00
/* Type specific cleanups. */
void (*shutdown)(Manager *m);
/* If this function is set and return false all jobs for units
* of this type will immediately fail. */
bool (*supported)(void);
/* The bus vtable */
const sd_bus_vtable *bus_vtable;
/* The strings to print in status messages */
UnitStatusMessageFormats status_message_formats;
/* True if transient units of this type are OK */
bool can_transient:1;
2010-01-26 21:39:06 +01:00
};
extern const UnitVTable * const unit_vtable[_UNIT_TYPE_MAX];
#define UNIT_VTABLE(u) unit_vtable[(u)->type]
2010-01-26 21:39:06 +01:00
/* For casting a unit into the various unit types */
#define DEFINE_CAST(UPPERCASE, MixedCase) \
static inline MixedCase* UPPERCASE(Unit *u) { \
if (_unlikely_(!u || u->type != UNIT_##UPPERCASE)) \
2010-01-26 21:39:06 +01:00
return NULL; \
\
return (MixedCase*) u; \
}
/* For casting the various unit types into a unit */
#define UNIT(u) (&(u)->meta)
2010-01-26 21:39:06 +01:00
#define UNIT_HAS_EXEC_CONTEXT(u) (UNIT_VTABLE(u)->exec_context_offset > 0)
#define UNIT_HAS_CGROUP_CONTEXT(u) (UNIT_VTABLE(u)->cgroup_context_offset > 0)
#define UNIT_HAS_KILL_CONTEXT(u) (UNIT_VTABLE(u)->kill_context_offset > 0)
#define UNIT_TRIGGER(u) ((Unit*) set_first((u)->dependencies[UNIT_TRIGGERS]))
2010-01-26 21:39:06 +01:00
DEFINE_CAST(SERVICE, Service);
DEFINE_CAST(SOCKET, Socket);
DEFINE_CAST(BUSNAME, BusName);
2010-01-26 21:39:06 +01:00
DEFINE_CAST(TARGET, Target);
DEFINE_CAST(DEVICE, Device);
DEFINE_CAST(MOUNT, Mount);
DEFINE_CAST(AUTOMOUNT, Automount);
2010-05-09 18:44:11 +02:00
DEFINE_CAST(SWAP, Swap);
DEFINE_CAST(TIMER, Timer);
DEFINE_CAST(PATH, Path);
DEFINE_CAST(SLICE, Slice);
DEFINE_CAST(SCOPE, Scope);
2010-01-26 21:39:06 +01:00
Unit *unit_new(Manager *m, size_t size);
2010-01-26 21:39:06 +01:00
void unit_free(Unit *u);
int unit_add_name(Unit *u, const char *name);
2010-04-15 03:11:11 +02:00
int unit_add_dependency(Unit *u, UnitDependency d, Unit *other, bool add_reference);
int unit_add_two_dependencies(Unit *u, UnitDependency d, UnitDependency e, Unit *other, bool add_reference);
int unit_add_dependency_by_name(Unit *u, UnitDependency d, const char *name, const char *filename, bool add_reference);
int unit_add_two_dependencies_by_name(Unit *u, UnitDependency d, UnitDependency e, const char *name, const char *path, bool add_reference);
2010-04-06 02:43:58 +02:00
int unit_add_exec_dependencies(Unit *u, ExecContext *c);
int unit_choose_id(Unit *u, const char *name);
2010-01-29 03:18:09 +01:00
int unit_set_description(Unit *u, const char *description);
2010-01-26 21:39:06 +01:00
bool unit_check_gc(Unit *u);
2010-01-26 21:39:06 +01:00
void unit_add_to_load_queue(Unit *u);
void unit_add_to_dbus_queue(Unit *u);
2010-04-06 02:43:58 +02:00
void unit_add_to_cleanup_queue(Unit *u);
void unit_add_to_gc_queue(Unit *u);
2010-01-26 21:39:06 +01:00
int unit_merge(Unit *u, Unit *other);
2010-04-06 02:43:58 +02:00
int unit_merge_by_name(Unit *u, const char *other);
Unit *unit_follow_merge(Unit *u) _pure_;
2010-01-26 21:39:06 +01:00
int unit_load_fragment_and_dropin(Unit *u);
int unit_load_fragment_and_dropin_optional(Unit *u);
2010-01-26 21:39:06 +01:00
int unit_load(Unit *unit);
int unit_set_slice(Unit *u, Unit *slice);
int unit_set_default_slice(Unit *u);
const char *unit_description(Unit *u) _pure_;
2010-01-26 21:39:06 +01:00
bool unit_has_name(Unit *u, const char *name);
2010-01-26 21:39:06 +01:00
UnitActiveState unit_active_state(Unit *u);
const char* unit_sub_state_to_string(Unit *u);
2010-01-26 21:39:06 +01:00
void unit_dump(Unit *u, FILE *f, const char *prefix);
bool unit_can_reload(Unit *u) _pure_;
bool unit_can_start(Unit *u) _pure_;
bool unit_can_isolate(Unit *u) _pure_;
2010-01-26 21:39:06 +01:00
int unit_start(Unit *u);
int unit_stop(Unit *u);
int unit_reload(Unit *u);
int unit_kill(Unit *u, KillWho w, int signo, sd_bus_error *error);
int unit_kill_common(Unit *u, KillWho who, int signo, pid_t main_pid, pid_t control_pid, sd_bus_error *error);
2010-10-22 16:11:50 +02:00
void unit_notify(Unit *u, UnitActiveState os, UnitActiveState ns, bool reload_success);
2010-01-26 21:39:06 +01:00
int unit_watch_pid(Unit *u, pid_t pid);
void unit_unwatch_pid(Unit *u, pid_t pid);
void unit_unwatch_all_pids(Unit *u);
void unit_tidy_watch_pids(Unit *u, pid_t except1, pid_t except2);
2010-01-26 21:39:06 +01:00
int unit_install_bus_match(Unit *u, sd_bus *bus, const char *name);
int unit_watch_bus_name(Unit *u, const char *name);
void unit_unwatch_bus_name(Unit *u, const char *name);
2010-01-26 21:39:06 +01:00
bool unit_job_is_applicable(Unit *u, JobType j);
2010-01-27 00:15:56 +01:00
int set_unit_path(const char *p);
2010-04-08 00:52:14 +02:00
char *unit_dbus_path(Unit *u);
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.
2016-08-30 23:18:46 +02:00
char *unit_dbus_path_invocation_id(Unit *u);
2010-04-08 00:52:14 +02:00
int unit_load_related_unit(Unit *u, const char *type, Unit **_found);
bool unit_can_serialize(Unit *u) _pure_;
int unit_serialize(Unit *u, FILE *f, FDSet *fds, bool serialize_jobs);
2010-04-21 03:27:44 +02:00
int unit_deserialize(Unit *u, FILE *f, FDSet *fds);
int unit_serialize_item(Unit *u, FILE *f, const char *key, const char *value);
int unit_serialize_item_escaped(Unit *u, FILE *f, const char *key, const char *value);
int unit_serialize_item_fd(Unit *u, FILE *f, FDSet *fds, const char *key, int fd);
void unit_serialize_item_format(Unit *u, FILE *f, const char *key, const char *value, ...) _printf_(4,5);
int unit_add_node_link(Unit *u, const char *what, bool wants, UnitDependency d);
int unit_coldplug(Unit *u);
void unit_status_printf(Unit *u, const char *status, const char *unit_status_msg_format) _printf_(3, 0);
void unit_status_emit_starting_stopping_reloading(Unit *u, JobType t);
bool unit_need_daemon_reload(Unit *u);
void unit_reset_failed(Unit *u);
Unit *unit_following(Unit *u);
int unit_following_set(Unit *u, Set **s);
const char *unit_slice_name(Unit *u);
bool unit_stop_pending(Unit *u) _pure_;
bool unit_inactive_or_pending(Unit *u) _pure_;
bool unit_active_or_pending(Unit *u);
2010-09-01 03:35:04 +02:00
int unit_add_default_target_dependency(Unit *u, Unit *target);
void unit_start_on_failure(Unit *u);
void unit_trigger_notify(Unit *u);
2011-07-31 18:28:02 +02:00
UnitFileState unit_get_unit_file_state(Unit *u);
int unit_get_unit_file_preset(Unit *u);
2011-07-31 18:28:02 +02:00
Unit* unit_ref_set(UnitRef *ref, Unit *u);
void unit_ref_unset(UnitRef *ref);
#define UNIT_DEREF(ref) ((ref).unit)
#define UNIT_ISSET(ref) (!!(ref).unit)
int unit_patch_contexts(Unit *u);
ExecContext *unit_get_exec_context(Unit *u) _pure_;
KillContext *unit_get_kill_context(Unit *u) _pure_;
CGroupContext *unit_get_cgroup_context(Unit *u) _pure_;
ExecRuntime *unit_get_exec_runtime(Unit *u) _pure_;
int unit_setup_exec_runtime(Unit *u);
int unit_setup_dynamic_creds(Unit *u);
int unit_write_drop_in(Unit *u, UnitSetPropertiesMode mode, const char *name, const char *data);
int unit_write_drop_in_format(Unit *u, UnitSetPropertiesMode mode, const char *name, const char *format, ...) _printf_(4,5);
int unit_write_drop_in_private(Unit *u, UnitSetPropertiesMode mode, const char *name, const char *data);
int unit_write_drop_in_private_format(Unit *u, UnitSetPropertiesMode mode, const char *name, const char *format, ...) _printf_(4,5);
int unit_kill_context(Unit *u, KillContext *c, KillOperation k, pid_t main_pid, pid_t control_pid, bool main_pid_alien);
int unit_make_transient(Unit *u);
int unit_require_mounts_for(Unit *u, const char *path);
bool unit_type_supported(UnitType t);
bool unit_is_pristine(Unit *u);
pid_t unit_control_pid(Unit *u);
pid_t unit_main_pid(Unit *u);
static inline bool unit_supported(Unit *u) {
return unit_type_supported(u->type);
}
void unit_warn_if_dir_nonempty(Unit *u, const char* where);
int unit_fail_if_symlink(Unit *u, const char* where);
int unit_start_limit_test(Unit *u);
void unit_unref_uid(Unit *u, bool destroy_now);
int unit_ref_uid(Unit *u, uid_t uid, bool clean_ipc);
void unit_unref_gid(Unit *u, bool destroy_now);
int unit_ref_gid(Unit *u, gid_t gid, bool clean_ipc);
int unit_ref_uid_gid(Unit *u, uid_t uid, gid_t gid);
void unit_unref_uid_gid(Unit *u, bool destroy_now);
void unit_notify_user_lookup(Unit *u, uid_t uid, gid_t gid);
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.
2016-08-30 23:18:46 +02:00
int unit_set_invocation_id(Unit *u, sd_id128_t id);
int unit_acquire_invocation_id(Unit *u);
/* Macros which append UNIT= or USER_UNIT= to the message */
core,network: major per-object logging rework This changes log_unit_info() (and friends) to take a real Unit* object insted of just a unit name as parameter. The call will now prefix all logged messages with the unit name, thus allowing the unit name to be dropped from the various passed romat strings, simplifying invocations drastically, and unifying log output across messages. Also, UNIT= vs. USER_UNIT= is now derived from the Manager object attached to the Unit object, instead of getpid(). This has the benefit of correcting the field for --test runs. Also contains a couple of other logging improvements: - Drops a couple of strerror() invocations in favour of using %m. - Not only .mount units now warn if a symlinks exist for the mount point already, .automount units do that too, now. - A few invocations of log_struct() that didn't actually pass any additional structured data have been replaced by simpler invocations of log_unit_info() and friends. - For structured data a new LOG_UNIT_MESSAGE() macro has been added, that works like LOG_MESSAGE() but prefixes the message with the unit name. Similar, there's now LOG_LINK_MESSAGE() and LOG_NETDEV_MESSAGE(). - For structured data new LOG_UNIT_ID(), LOG_LINK_INTERFACE(), LOG_NETDEV_INTERFACE() macros have been added that generate the necessary per object fields. The old log_unit_struct() call has been removed in favour of these new macros used in raw log_struct() invocations. In addition to removing one more function call this allows generated structured log messages that contain two object fields, as necessary for example for network interfaces that are joined into another network interface, and whose messages shall be indexed by both. - The LOG_ERRNO() macro has been removed, in favour of log_struct_errno(). The latter has the benefit of ensuring that %m in format strings is properly resolved to the specified error number. - A number of logging messages have been converted to use log_unit_info() instead of log_info() - The client code in sysv-generator no longer #includes core code from src/core/. - log_unit_full_errno() has been removed, log_unit_full() instead takes an errno now, too. - log_unit_info(), log_link_info(), log_netdev_info() and friends, now avoid double evaluation of their parameters
2015-05-11 20:38:21 +02:00
#define log_unit_full(unit, level, error, ...) \
({ \
const Unit *_u = (unit); \
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.
2016-08-30 23:18:46 +02:00
_u ? log_object_internal(level, error, __FILE__, __LINE__, __func__, _u->manager->unit_log_field, _u->id, _u->manager->invocation_log_field, _u->invocation_id_string, ##__VA_ARGS__) : \
core,network: major per-object logging rework This changes log_unit_info() (and friends) to take a real Unit* object insted of just a unit name as parameter. The call will now prefix all logged messages with the unit name, thus allowing the unit name to be dropped from the various passed romat strings, simplifying invocations drastically, and unifying log output across messages. Also, UNIT= vs. USER_UNIT= is now derived from the Manager object attached to the Unit object, instead of getpid(). This has the benefit of correcting the field for --test runs. Also contains a couple of other logging improvements: - Drops a couple of strerror() invocations in favour of using %m. - Not only .mount units now warn if a symlinks exist for the mount point already, .automount units do that too, now. - A few invocations of log_struct() that didn't actually pass any additional structured data have been replaced by simpler invocations of log_unit_info() and friends. - For structured data a new LOG_UNIT_MESSAGE() macro has been added, that works like LOG_MESSAGE() but prefixes the message with the unit name. Similar, there's now LOG_LINK_MESSAGE() and LOG_NETDEV_MESSAGE(). - For structured data new LOG_UNIT_ID(), LOG_LINK_INTERFACE(), LOG_NETDEV_INTERFACE() macros have been added that generate the necessary per object fields. The old log_unit_struct() call has been removed in favour of these new macros used in raw log_struct() invocations. In addition to removing one more function call this allows generated structured log messages that contain two object fields, as necessary for example for network interfaces that are joined into another network interface, and whose messages shall be indexed by both. - The LOG_ERRNO() macro has been removed, in favour of log_struct_errno(). The latter has the benefit of ensuring that %m in format strings is properly resolved to the specified error number. - A number of logging messages have been converted to use log_unit_info() instead of log_info() - The client code in sysv-generator no longer #includes core code from src/core/. - log_unit_full_errno() has been removed, log_unit_full() instead takes an errno now, too. - log_unit_info(), log_link_info(), log_netdev_info() and friends, now avoid double evaluation of their parameters
2015-05-11 20:38:21 +02:00
log_internal(level, error, __FILE__, __LINE__, __func__, ##__VA_ARGS__); \
})
#define log_unit_debug(unit, ...) log_unit_full(unit, LOG_DEBUG, 0, ##__VA_ARGS__)
#define log_unit_info(unit, ...) log_unit_full(unit, LOG_INFO, 0, ##__VA_ARGS__)
#define log_unit_notice(unit, ...) log_unit_full(unit, LOG_NOTICE, 0, ##__VA_ARGS__)
#define log_unit_warning(unit, ...) log_unit_full(unit, LOG_WARNING, 0, ##__VA_ARGS__)
#define log_unit_error(unit, ...) log_unit_full(unit, LOG_ERR, 0, ##__VA_ARGS__)
#define log_unit_debug_errno(unit, error, ...) log_unit_full(unit, LOG_DEBUG, error, ##__VA_ARGS__)
#define log_unit_info_errno(unit, error, ...) log_unit_full(unit, LOG_INFO, error, ##__VA_ARGS__)
#define log_unit_notice_errno(unit, error, ...) log_unit_full(unit, LOG_NOTICE, error, ##__VA_ARGS__)
#define log_unit_warning_errno(unit, error, ...) log_unit_full(unit, LOG_WARNING, error, ##__VA_ARGS__)
#define log_unit_error_errno(unit, error, ...) log_unit_full(unit, LOG_ERR, error, ##__VA_ARGS__)
#define LOG_UNIT_MESSAGE(unit, fmt, ...) "MESSAGE=%s: " fmt, (unit)->id, ##__VA_ARGS__
#define LOG_UNIT_ID(unit) (unit)->manager->unit_log_format_string, (unit)->id