Data Structures
struct	cgroup

struct	cgroup_controller

Basic infrastructure
`struct cgroup` is the heart of `libcgroup` API. The structure is opaque to applications, all access to the structure is through appropriate functions. The most important information is that one `struct cgroup` can represent zero, one or more real control groups in kernel*. The `struct cgroup` is identified by name of the group, which must be set by cgroup_new_cgroup(). Multiple controllers (aka subsystems) can be attached to one `struct cgroup` using cgroup_add_controller(). These controllers can belong to different hierarchies. This approach is different to the one in the Linux kernel - a control group must be part of exactly one hierarchy there. In `libcgroup`, a group can be part of multiple hierarchies, as long as the group name is the same. Example: Let there be following control groups: cpu,cpuacct:/ cpu,cpuacct:/foo cpu,cpuacct:/bar freezer:/ freezer:/foo I.e. there is `cpu` and `cpuacct` controller mounted together in one hierarchy, with `foo` and `bar` groups. In addition, `freezer` is mounted as separate hierarchy, with only one `foo` group. Following code creates `struct cgroup` structure, which represents one group `cpu,cpuacct:/foo`: struct cgroup foo = cgroup_new_cgroup("foo"); cgroup_add_controller(foo, "cpu"); cgroup_add_controller struct cgroup_controller cgroup_add_controller(struct cgroup cgrp, const char name) Definition wrapper.c:61 cgroup_new_cgroup struct cgroup * cgroup_new_cgroup(const char name) Definition* wrapper.c:43 cgroup Definition libcgroup-internal.h:121 Now, you can call e.g. cgroup_delete_cgroup() and the group is deleted from the hierarchy. You can note that it's enough to add only one controller to the group to fully identify a group in `cpu,cpuacct` hierarchy. Following code creates `struct cgroup` structure, which represents two* groups, `cpu,cpuacct:/foo` and `freezer:/foo`: struct cgroup foo = cgroup_new_cgroup("foo"); cgroup_add_controller(foo, "cpu"); cgroup_add_controller(foo, "freezer"); Now, if you call e.g. cgroup_delete_cgroup(), the group gets deleted from both* hierarchies. Todo add some propaganda what's so great on this approach... I personally think it is broken and confusing (see TODOs below). Following functions are provided to create/destroy various libcgroup structures. Please note that none of these functions actually create or delete a cgroup in kernel!
struct cgroup *	cgroup_new_cgroup (const char *name)

struct cgroup_controller *	cgroup_add_controller (struct cgroup cgrp, const char name)

int	cgroup_add_all_controllers (struct cgroup *cgrp)

struct cgroup_controller *	cgroup_get_controller (struct cgroup cgrp, const char name)

void	cgroup_free (struct cgroup **cgrp)

void	cgroup_free_controllers (struct cgroup *cgrp)

#define	NO_PERMS (-1U)

#define	NO_UID_GID (-1U)

Group manipulation API
Using following functions you can create and remove control groups and change their parameters. Note All access to kernel is through previously mounted cgroup filesystems. `libcgroup` does not mount/unmount anything for you.
int	cgroup_create_cgroup (struct cgroup *cgrp, int ignore_ownership)

int	cgroup_create_cgroup_from_parent (struct cgroup *cgrp, int ignore_ownership)

int	cgroup_modify_cgroup (struct cgroup *cgrp)

int	cgroup_delete_cgroup (struct cgroup *cgrp, int ignore_migration)

int	cgroup_delete_cgroup_ext (struct cgroup *cgrp, int flags)

Other functions
Helper functions to manipulate with control groups.
int	cgroup_get_cgroup (struct cgroup *cgrp)

int	cgroup_copy_cgroup (struct cgroup dst, struct cgroup src)

int	cgroup_compare_cgroup (struct cgroup cgrp_a, struct cgroup cgrp_b)

int	cgroup_compare_controllers (struct cgroup_controller cgca, struct cgroup_controller cgcb)

int	cgroup_set_uid_gid (struct cgroup *cgrp, uid_t tasks_uid, gid_t tasks_gid, uid_t control_uid, gid_t control_gid)

int	cgroup_get_uid_gid (struct cgroup cgrp, uid_t tasks_uid, gid_t tasks_gid, uid_t control_uid, gid_t *control_gid)

void	cgroup_set_permissions (struct cgroup *cgrp, mode_t control_dperm, mode_t control_fperm, mode_t task_fperm)

Group parameters
These are functions can read or modify parameter of a group. Note All these functions read/write parameters to `libcgorup` internal structures. Use cgroup_get_cgroup() to load parameters from kernel to these internal structures and cgroup_modify_cgroup() or cgroup_create_cgroup() to write changes to kernel.
int	cgroup_add_value_string (struct cgroup_controller controller, const char name, const char *value)

int	cgroup_add_value_int64 (struct cgroup_controller controller, const char name, int64_t value)

int	cgroup_add_value_uint64 (struct cgroup_controller controller, const char name, u_int64_t value)

int	cgroup_add_value_bool (struct cgroup_controller controller, const char name, bool value)

int	cgroup_get_value_string (struct cgroup_controller controller, const char name, char **value)

int	cgroup_get_value_int64 (struct cgroup_controller controller, const char name, int64_t *value)

int	cgroup_get_value_uint64 (struct cgroup_controller controller, const char name, u_int64_t *value)

int	cgroup_get_value_bool (struct cgroup_controller controller, const char name, bool *value)

int	cgroup_set_value_string (struct cgroup_controller controller, const char name, const char *value)

int	cgroup_set_value_int64 (struct cgroup_controller controller, const char name, int64_t value)

int	cgroup_set_value_uint64 (struct cgroup_controller controller, const char name, u_int64_t value)

int	cgroup_set_value_bool (struct cgroup_controller controller, const char name, bool value)

int	cgroup_get_value_name_count (struct cgroup_controller *controller)

char *	cgroup_get_value_name (struct cgroup_controller *controller, int index)

int	cgroup_get_procs (const char name, const char controller, pid_t *pids, int size)

int	cgroup_get_threads (const char name, const char controller, pid_t *pids, int size)

int	cg_chmod_recursive (struct cgroup *cgrp, mode_t dir_mode, int dirm_change, mode_t file_mode, int filem_change)

char *	cgroup_get_cgroup_name (struct cgroup *cgrp)

int	cgroup_convert_cgroup (struct cgroup const out_cgrp, enum cg_version_t out_version, const struct cgroup const in_cgrp, enum cg_version_t in_version)

int	cgroup_list_mount_points (const enum cg_version_t cgrp_version, char ***mount_paths)

int	cgroup_get_controller_version (const char const controller, enum cg_version_t const version)

enum cg_setup_mode_t	cgroup_setup_mode (void)

int	cgroup_get_controller_count (struct cgroup *cgrp)

struct cgroup_controller *	cgroup_get_controller_by_index (struct cgroup *cgrp, int index)

char *	cgroup_get_controller_name (struct cgroup_controller *controller)

bool	is_cgroup_mode_legacy (void)

bool	is_cgroup_mode_hybrid (void)

bool	is_cgroup_mode_unified (void)

Detailed Description

Macro Definition Documentation

◆ NO_PERMS

#define NO_PERMS (-1U)

Uninitialized file/directory permissions used for task/control files.

◆ NO_UID_GID

#define NO_UID_GID (-1U)

Uninitialized UID/GID used for task/control files.

Function Documentation

◆ cg_chmod_recursive()

int cg_chmod_recursive	(	struct cgroup *	cgrp,
		mode_t	dir_mode,
		int	dirm_change,
		mode_t	file_mode,
		int	filem_change )

Change permission of files and directories of given group

Parameters

cgrp	The cgroup which permissions should be changed
dir_mode	The permission mode of group directory
dirm_change	Denotes whether the directory change should be done
file_mode	The permission mode of group files
filem_change	Denotes whether the directory change should be done

◆ cgroup_add_all_controllers()

int cgroup_add_all_controllers ( struct cgroup * cgrp )

Attach all mounted controllers to given cgroup. This function just modifies internal libcgroup structure, not the kernel control group.

Parameters

cgrp

Returns: zero or error number

◆ cgroup_add_controller()

struct cgroup_controller * cgroup_add_controller	(	struct cgroup *	cgrp,
		const char *	name )

Attach new controller to cgroup. This function just modifies internal libcgroup structure, not the kernel control group.

Parameters

cgrp
name	The name of the controller, e.g. "freezer".

Returns: Created controller or NULL on error.

◆ cgroup_add_value_bool()

int cgroup_add_value_bool	(	struct cgroup_controller *	controller,
		const char *	name,
		bool	value )

Add parameter and its value to internal libcgroup structures. Use cgroup_modify_cgroup() or cgroup_create_cgroup() to write it to kernel.

Parameters

controller
name	The name of the parameter.
value

◆ cgroup_add_value_int64()

int cgroup_add_value_int64	(	struct cgroup_controller *	controller,
		const char *	name,
		int64_t	value )

Add parameter and its value to internal libcgroup structures. Use cgroup_modify_cgroup() or cgroup_create_cgroup() to write it to kernel. Content of the value is copied to internal structures and is not needed after return from the function.

Parameters

controller
name	The name of the parameter.
value

◆ cgroup_add_value_string()

int cgroup_add_value_string	(	struct cgroup_controller *	controller,
		const char *	name,
		const char *	value )

Add parameter and its value to internal libcgroup structures. Use cgroup_modify_cgroup() or cgroup_create_cgroup() to write it to kernel.

Parameters

controller
name	The name of the parameter.
value

◆ cgroup_add_value_uint64()

int cgroup_add_value_uint64	(	struct cgroup_controller *	controller,
		const char *	name,
		u_int64_t	value )

Add parameter and its value to internal libcgroup structures. Use cgroup_modify_cgroup() or cgroup_create_cgroup() to write it to kernel.

Parameters

controller
name	The name of the parameter.
value

◆ cgroup_compare_cgroup()

int cgroup_compare_cgroup	(	struct cgroup *	cgrp_a,
		struct cgroup *	cgrp_b )

Compare names, owners, controllers, parameters and values of two groups.

Parameters

cgrp_a
cgrp_b

Return values

0	if the groups are the same.
ECGROUPNOTEQUAL	if the groups are not the same.
ECGCONTROLLERNOTEQUAL	if the only difference are controllers, parameters or their values.

◆ cgroup_compare_controllers()

int cgroup_compare_controllers	(	struct cgroup_controller *	cgca,
		struct cgroup_controller *	cgcb )

Compare names, parameters and values of two controllers.

Parameters

cgca
cgcb

Return values

0	if the controllers are the same.
ECGCONTROLLERNOTEQUAL	if the controllers are not equal.

◆ cgroup_copy_cgroup()

int cgroup_copy_cgroup	(	struct cgroup *	dst,
		struct cgroup *	src )

Copy all controllers, their parameters and values. Group name, permissions and ownerships are not copied. All existing controllers in the source group are discarded.

Parameters

dst	Destination group.
src	Source group.

@dst: Destination control group @src: Source from which values will be copied to dst

Create a duplicate copy of src in dst. This will be useful for those who that intend to create new instances based on an existing control group

◆ cgroup_create_cgroup()

int cgroup_create_cgroup	(	struct cgroup *	cgrp,
		int	ignore_ownership )

Physically create a control group in kernel. The group is created in all hierarchies, which cover controllers added by cgroup_add_controller(). All parameters set by cgroup_add_value_* functions are written. The created groups has owner which was set by cgroup_set_uid_gid() and permissions set by cgroup_set_permissions.

Parameters

cgrp
ignore_ownership	When nozero, all errors are ignored when setting owner of the group and/or its tasks file.

Todo: what is ignore_ownership good for?

Return values

ECGROUPNOTEQUAL if not all specified controller parameters were successfully set.

cgroup_create_cgroup creates a new control group. struct cgroup *cgrp: The control group to be created

returns 0 on success. We recommend calling cg_delete_cgroup if this routine fails. That should do the cleanup operation. If ECGCANTSETVALUE is returned, the group was created successfully but not all controller parameters were successfully set.

◆ cgroup_create_cgroup_from_parent()

int cgroup_create_cgroup_from_parent	(	struct cgroup *	cgrp,
		int	ignore_ownership )

Physically create new control group in kernel, with all parameters and values copied from its parent group. The group is created in all hierarchies, where the parent group exists. I.e. following code creates subgroup in all hierarchies, because all of them have root (=parent) group.

struct cgroup *foo = cgroup_new_cgroup("foo");

cgroup_create_cgroup_from_parent(foo, 0);

cgroup_create_cgroup_from_parent

int cgroup_create_cgroup_from_parent(struct cgroup *cgrp, int ignore_ownership)

Definition api.c:3321

Todo: what is this good for? Why the list of controllers added by cgroup_add_controller() is not used, like in cgroup_create_cgroup()? I can't create subgroup of root group in just one hierarchy with this function!

Parameters

cgrp	The cgroup to create. Only it's name is used, everything else is discarded.
ignore_ownership	When nozero, all errors are ignored when setting owner of the group and/or its tasks file.

Todo: what is ignore_ownership good for?

Return values

ECGROUPNOTEQUAL if not all inherited controller parameters were successfully set (this is expected).

@cgrp: cgroup data structure to be filled with parent values and then passed down for creation @ignore_ownership: Ignore doing a chown on the newly created cgroup

Returns: 0 on success, > 0 on failure. If ECGCANTSETVALUE is returned, the group was created successfully, but not all controller parameters were copied from the parent successfully; unfortunately, this is expected...

◆ cgroup_delete_cgroup()

int cgroup_delete_cgroup	(	struct cgroup *	cgrp,
		int	ignore_migration )

Physically remove a control group from kernel. The group is removed from all hierarchies, which cover controllers added by cgroup_add_controller() or cgroup_get_cgroup(). All tasks inside the group are automatically moved to parent group.

The group being removed must be empty, i.e. without subgroups. Use cgroup_delete_cgroup_ext() for recursive delete.

Parameters

cgrp
ignore_migration	When nozero, all errors are ignored when migrating tasks from the group to the parent group.

Todo: what is ignore_migration good for? rmdir() will fail if tasks were not moved.

cgroup_delete cgroup deletes a control group. struct cgroup *cgrp takes the group which is to be deleted.

returns 0 on success.

◆ cgroup_delete_cgroup_ext()

int cgroup_delete_cgroup_ext	(	struct cgroup *	cgrp,
		int	flags )

Physically remove a control group from kernel. All tasks are automatically moved to parent group. If CGFLAG_DELETE_IGNORE_MIGRATION flag is used, the errors that occurred during the task movement are ignored. CGFLAG_DELETE_RECURSIVE flag specifies that all subgroups should be removed too. If root group is being removed with this flag specified, all subgroups are removed but the root group itself is left undeleted.

See also: cgroup_delete_flag.

Parameters

cgrp
flags	Combination of CGFLAG_DELETE_* flags, which indicate what and how to delete.

◆ cgroup_free()

void cgroup_free ( struct cgroup ** cgrp )

Free internal cgroup structure. This function frees also all controllers attached to the cgroup, including all parameters and their values.

Parameters

cgrp

◆ cgroup_free_controllers()

void cgroup_free_controllers ( struct cgroup * cgrp )

Free internal list of controllers from the group.

Todo: should this function be public???

Parameters

cgrp

◆ cgroup_get_cgroup()

int cgroup_get_cgroup ( struct cgroup * cgrp )

Read all information regarding the group from kernel. Based on name of the group, list of controllers and all parameters and their values are read from all hierarchies, where a group with given name exists. All existing controllers are replaced. I.e. following code will fill root with controllers from all hierarchies, because the root group is available in all of them.

struct cgroup *root = cgroup_new_cgroup("/");

cgroup_get_cgroup(root);

cgroup_get_cgroup

int cgroup_get_cgroup(struct cgroup *cgrp)

Definition api.c:3855

Todo: what is this function good for? Why is not considered only the list of controllers attached by cgroup_add_controller()? What owners will return cgroup_get_uid_gid() if the group is in multiple hierarchies, each with different owner of tasks file?

Parameters

cgrp	The cgroup to load. Only it's name is used, everything else is replaced.

◆ cgroup_get_cgroup_name()

char * cgroup_get_cgroup_name ( struct cgroup * cgrp )

Get the name of the cgroup from a given cgroup

Parameters

cgrp	The cgroup whose name is needed

◆ cgroup_get_controller()

struct cgroup_controller * cgroup_get_controller	(	struct cgroup *	cgrp,
		const char *	name )

Return appropriate controller from given group. The controller must be added before using cgroup_add_controller() or loaded from kernel using cgroup_get_cgroup().

Parameters

cgrp
name	The name of the controller, e.g. "freezer".

◆ cgroup_get_controller_by_index()

struct cgroup_controller * cgroup_get_controller_by_index	(	struct cgroup *	cgrp,
		int	index )

Return requested controller from given group

Parameters

cgrp
index	The index into the cgroup controller list

◆ cgroup_get_controller_count()

int cgroup_get_controller_count ( struct cgroup * cgrp )

Return the number of controllers for the specified cgroup in libcgroup internal structures.

Parameters

cgrp

Returns: Count of the controllers or -1 on error.

◆ cgroup_get_controller_name()

char * cgroup_get_controller_name ( struct cgroup_controller * controller )

Given a controller pointer, get the name of the controller

Parameters

controller

Returns: controller name string, NULL if there's an error

◆ cgroup_get_controller_version()

int cgroup_get_controller_version	(	const char *const	controller,
		enum cg_version_t *const	version )

Get the cgroup version of a controller. Version is set to CGROUP_UNK if the version cannot be determined.

Parameters

controller	The controller of interest
version	The version of the controller

◆ cgroup_get_procs()

int cgroup_get_procs	(	const char *	name,
		const char *	controller,
		pid_t **	pids,
		int *	size )

Get the list of process in a cgroup. This list is guaranteed to be sorted. It is not necessary that it is unique.

Parameters

name	The name of the cgroup
controller	The name of the controller
pids	The list of pids. Should be uninitialized when passed to the API. Should be freed by the caller using free.
size	The size of the pids array returned by the API.

◆ cgroup_get_threads()

int cgroup_get_threads	(	const char *	name,
		const char *	controller,
		pid_t **	pids,
		int *	size )

Get the list of threads in a cgroup. This list is guaranteed to be sorted. It is not necessary that it is unique.

Parameters

name	The name of the cgroup
controller	The name of the controller
pids	The list of pids. Should be uninitialized when passed to the API. Should be freed by the caller using free.
size	The size of the pids array returned by the API.

◆ cgroup_get_uid_gid()

int cgroup_get_uid_gid	(	struct cgroup *	cgrp,
		uid_t *	tasks_uid,
		gid_t *	tasks_gid,
		uid_t *	control_uid,
		gid_t *	control_gid )

Return owners of the group's tasks file and control files. The data is read from libcgroup internal cgroup structure, use cgroup_set_uid_gid() or cgroup_get_cgroup() to fill it.

◆ cgroup_get_value_bool()

int cgroup_get_value_bool	(	struct cgroup_controller *	controller,
		const char *	name,
		bool *	value )

Read a parameter value from libcgroup internal structures. Use cgroup_get_cgroup() to fill these structures with data from kernel.

Parameters

controller
name	The name of the parameter.
value

◆ cgroup_get_value_int64()

int cgroup_get_value_int64	(	struct cgroup_controller *	controller,
		const char *	name,
		int64_t *	value )

Read a parameter value from libcgroup internal structures. Use cgroup_get_cgroup() to fill these structures with data from kernel.

Parameters

controller
name	The name of the parameter.
value

◆ cgroup_get_value_name()

char * cgroup_get_value_name	(	struct cgroup_controller *	controller,
		int	index )

Return the name of parameter of controller at given index. The index goes from 0 to cgroup_get_value_name_count()-1. Use this function to list all parameter of the controller.

Note: The returned value is pointer to internal libcgroup structure, do not free it.

Parameters

controller
index	The index of the parameter.

Returns: Name of the parameter.

◆ cgroup_get_value_name_count()

int cgroup_get_value_name_count ( struct cgroup_controller * controller )

Return the number of variables for the specified controller in libcgroup internal structures. Use cgroup_get_cgroup() to fill these structures with data from kernel. Use this function together with cgroup_get_value_name() to list all parameters of a group.

Parameters

controller

Returns: Count of the parameters or -1 on error.

◆ cgroup_get_value_string()

int cgroup_get_value_string	(	struct cgroup_controller *	controller,
		const char *	name,
		char **	value )

Read a parameter value from libcgroup internal structures. Use cgroup_get_cgroup() to fill these structures with data from kernel. It's up to the caller to free returned value.

This function works only for 'short' parameters. Use cgroup_read_stats_begin(), cgroup_read_stats_next() and cgroup_read_stats_end() to read stats parameter, which can be longer than libcgroup's internal buffers.

Todo: rephrase, it's too vague... How big is the buffer actually?

Parameters

controller
name	The name of the parameter.
value

◆ cgroup_get_value_uint64()

int cgroup_get_value_uint64	(	struct cgroup_controller *	controller,
		const char *	name,
		u_int64_t *	value )

Read a parameter value from libcgroup internal structures. Use cgroup_get_cgroup() to fill these structures with data from kernel.

Parameters

controller
name	The name of the parameter.
value

◆ cgroup_list_mount_points()

int cgroup_list_mount_points	(	const enum cg_version_t	cgrp_version,
		char ***	mount_paths )

List the mount paths, that matches the specified version

Parameters

cgrp_version	The cgroup type/version
mount_paths	Holds the list of mount paths

Returns: 0 success and list of mounts paths in mount_paths ECGOTHER on failure and mount_paths is NULL.

◆ cgroup_modify_cgroup()

int cgroup_modify_cgroup ( struct cgroup * cgrp )

Physically modify a control group in kernel. All parameters added by cgroup_add_value_ or cgroup_set_value_ are written. Currently it's not possible to change and owner of a group.

Parameters

cgrp

cgroup_modify_cgroup modifies the cgroup control files. struct cgroup *cgrp: The name will be the cgroup to be modified. The values will be the values to be modified, those not mentioned in the structure will not be modified.

The uids cannot be modified yet.

returns 0 on success.

◆ cgroup_new_cgroup()

struct cgroup * cgroup_new_cgroup ( const char * name )

Allocate new cgroup structure. This function itself does not create new control group in kernel, only new struct cgroup inside libcgroup!

Parameters

name	Path to the group, relative from root group. Use `"/"` or `"."` for the root group itself and `"/foo/bar/baz"` or `"foo/bar/baz"` for subgroups.

Todo: suggest one preferred way, either "/foo" or "foo".

Returns: Created group or NULL on error.

◆ cgroup_set_permissions()

void cgroup_set_permissions	(	struct cgroup *	cgrp,
		mode_t	control_dperm,
		mode_t	control_fperm,
		mode_t	task_fperm )

Stores given file permissions of the group's control and tasks files into the cgroup data structure. Use NO_PERMS if permissions shouldn't be changed or a value which applicable to chmod(2). Please note that the given permissions are masked with the file owner's permissions. For example if a control file has permissions 640 and control_fperm is 471 the result will be 460.

Parameters

cgrp
control_dperm	Directory permission for the group.
control_fperm	File permission for the control files.
task_fperm	File permissions for task file.

◆ cgroup_set_uid_gid()

int cgroup_set_uid_gid	(	struct cgroup *	cgrp,
		uid_t	tasks_uid,
		gid_t	tasks_gid,
		uid_t	control_uid,
		gid_t	control_gid )

Set owner of the group control files and the tasks file. This function modifies only libcgroup internal cgroup structure, use cgroup_create_cgroup() afterwards to create the group with given owners.

Parameters

cgrp
tasks_uid	UID of the owner of group's `tasks` file.
tasks_gid	GID of the owner of group's `tasks` file.
control_uid	UID of the owner of group's control files (i.e. parameters).
control_gid	GID of the owner of group's control files (i.e. parameters).

◆ cgroup_set_value_bool()

int cgroup_set_value_bool	(	struct cgroup_controller *	controller,
		const char *	name,
		bool	value )

Set a parameter value in libcgroup internal structures. Use cgroup_modify_cgroup() or cgroup_create_cgroup() to write it to kernel.

Parameters

controller
name	The name of the parameter.
value

◆ cgroup_set_value_int64()

int cgroup_set_value_int64	(	struct cgroup_controller *	controller,
		const char *	name,
		int64_t	value )

Set a parameter value in libcgroup internal structures. Use cgroup_modify_cgroup() or cgroup_create_cgroup() to write it to kernel. Content of the value is copied to internal structures and is not needed after return from the function.

Parameters

controller
name	The name of the parameter.
value

◆ cgroup_set_value_string()

int cgroup_set_value_string	(	struct cgroup_controller *	controller,
		const char *	name,
		const char *	value )

Set a parameter value in libcgroup internal structures. Use cgroup_modify_cgroup() or cgroup_create_cgroup() to write it to kernel.

Parameters

controller
name	The name of the parameter.
value

◆ cgroup_set_value_uint64()

int cgroup_set_value_uint64	(	struct cgroup_controller *	controller,
		const char *	name,
		u_int64_t	value )

Set a parameter value in libcgroup internal structures. Use cgroup_modify_cgroup() or cgroup_create_cgroup() to write it to kernel.

Parameters

controller
name	The name of the parameter.
value

◆ cgroup_setup_mode()

enum cg_setup_mode_t cgroup_setup_mode ( void )

Get the current group setup mode (legacy/unified/hybrid)

Returns: CGROUP_MODE_UNK on failure and setup mode on success

Finds the current cgroup setup mode (legacy/unified/hybrid). Returns unknown of failure and setup mode on success.

◆ is_cgroup_mode_hybrid()

bool is_cgroup_mode_hybrid ( void )

Return true if cgroup setup mode is cgroup v1/v2 (hybrid), else returns false.

◆ is_cgroup_mode_legacy()

bool is_cgroup_mode_legacy ( void )

Return true if cgroup setup mode is cgroup v1 (legacy), else returns false.

◆ is_cgroup_mode_unified()

bool is_cgroup_mode_unified ( void )

Return true if cgroup setup mode is cgroup v2 (unified), else returns false.

Data Structures

Basic infrastructure

Group manipulation API

Other functions

Group parameters

Detailed Description

Macro Definition Documentation

◆ NO_PERMS

◆ NO_UID_GID

Function Documentation

◆ cg_chmod_recursive()

◆ cgroup_add_all_controllers()

◆ cgroup_add_controller()

◆ cgroup_add_value_bool()

◆ cgroup_add_value_int64()

◆ cgroup_add_value_string()

◆ cgroup_add_value_uint64()

◆ cgroup_compare_cgroup()

◆ cgroup_compare_controllers()

◆ cgroup_copy_cgroup()

◆ cgroup_create_cgroup()

◆ cgroup_create_cgroup_from_parent()

◆ cgroup_delete_cgroup()

◆ cgroup_delete_cgroup_ext()

◆ cgroup_free()

◆ cgroup_free_controllers()

◆ cgroup_get_cgroup()

◆ cgroup_get_cgroup_name()

◆ cgroup_get_controller()

◆ cgroup_get_controller_by_index()

◆ cgroup_get_controller_count()

◆ cgroup_get_controller_name()

◆ cgroup_get_controller_version()

◆ cgroup_get_procs()

◆ cgroup_get_threads()

◆ cgroup_get_uid_gid()

◆ cgroup_get_value_bool()

◆ cgroup_get_value_int64()

◆ cgroup_get_value_name()

◆ cgroup_get_value_name_count()

◆ cgroup_get_value_string()

◆ cgroup_get_value_uint64()

◆ cgroup_list_mount_points()

◆ cgroup_modify_cgroup()

◆ cgroup_new_cgroup()

◆ cgroup_set_permissions()

◆ cgroup_set_uid_gid()

◆ cgroup_set_value_bool()

◆ cgroup_set_value_int64()

◆ cgroup_set_value_string()

◆ cgroup_set_value_uint64()

◆ cgroup_setup_mode()

◆ is_cgroup_mode_hybrid()

◆ is_cgroup_mode_legacy()

◆ is_cgroup_mode_unified()