Job Management

Overview

Function

A job is a set of commands in the .cfg file of the init module. A maximum of 4096 jobs can be added. Jobs can be configured in the .cfg file. Generally, jobs are executed during initialization to serve the normal startup of services or the initialization of specific basic functions.

Basic Concepts

A job can be configured in the init.cfg file or the custom .cfg file of the module. The parser of the init process aggregates commands of the jobs with the same name into one job. For jobs with the same name, the init process only ensures that the commands in the init.cfg file are executed in preference. It does not guarantee the execution sequence of commands in other .cfg files.

  • Common job

    A job executed in a fixed phase during init process startup, for example, pre-init, init, or post-init.

    • pre-init: pre-initialization phase. Key services on which other services depend, such as ueventd, watchdog, and hilogd, are started in this phase. The mounting of data partitions is also done in this phase.
    • init: main phase of the initialization process. In this phase, a large number of commands are executed, and services in the boot group (first group) are started concurrently by the init group. Some important services related to system functions are also started in this phase.
    • post-init: post-initialization phase. In this phase, the trigger command is used to invoke the execution of other phases, which can be regarded as independent phases, or the post-init phase as a whole. In this phase, a large number of commands are executed, and the normal group (the second group) is started by the init group. Most services configured in the .cfg files are started in this phase.
  • Custom job (for standard system or higher)

    A job triggered based on specific rules. You can add commands to the job as required and run the trigger command to invoke the execution of commands in the job.

  • Conditional job (for standard system or higher)

    A job triggered based on specific conditions. You can add conditions to a job so that the job is executed when the conditions are met.

    A condition is a combination of system parameter values and supports operations such as && and ||, for example, "condition": "sys.usb.config = none && sys.usb.configfs = 0". When defining commands for a job, you can add attributes in the format of param:xxx to form different commands.

Constraints

With the system parameter module, the standard system is able to support basic, conditional, and custom jobs. The small system supports only basic jobs.

How to Develop

Use Cases

A job is a command set, where you can manage the commands to be executed. A maximum of 4096 commands can be added to a command set. During the init startup process, the execution of jobs helps ensure normal running of services.

Parameters

Table 1 Command set description

Command Format and Example Description Supported System Type
mkdir mkdir target folder [mode] [owner] [group]
Example:
mkdir /storage/myDirectory
mkdir /storage/myDirectory 0755 root root
Creates a folder. mkdir and the target folder must be separated by only one space. Folders cannot be created recursively. Small and standard systems
chmod chmod permission target
Example:
chmod 0600 /storage/myFile.txt
chmod 0750 /storage/myDir
Modifies the permission, which must be in the 0xxx format. chmod, permission, and target must be separated by only one space. Small and standard systems
chown chown uid gid target
Example:
chown 900 800 /storage/myDir
chown 100 100 /storage/myFile.txt
Modifies the owner group. chown, uid, gid, and target must be separated by only one space. Small and standard systems
mount mount fileSystemType src dst flags data
Example:
mount vfat /dev/mmcblk0 /sdc rw,umask=000
mount jffs2 /dev/mtdblock3 /storage nosuid
Mounts devices. Every two parameters must be separated by only one space.
For details about flags, see the mountFlagMap[] array of the mountFlagMap function in base/startup/init/services/init/init_common_cmds.c. The data field is optional.
Small and standard systems
start start serviceName
Example:
start foundationstart
Starts services. serviceName must be contained in the services array. Small and standard systems
export export key value
Example:
export TEST /data/test
Sets environment variables. key and value respectively indicate the environment variable and its value. Small and standard systems
rm rm filename
Example:
rm /data/testfile
Deletes a file. filename indicates the absolute file path. Small and standard systems
rmdir rmdir dirname
Example:
rmdir /data/testdir
Deletes a directory. dirname indicates the absolute path of the directory. Small and standard systems
write write filename value
Example:
write /data/testfile 0
Writes a file. filename and value respectively indicate the absolute file path and the string to write. Standard system
stop stop servicename
Example:
stop console
Stops a service. servicename indicates the name of the service to stop. Small and standard systems
copy copy oldfile newfile
Example:
copy /data/old /data/new
Copies a file. oldfile and newfile respectively indicate the old and new absolute file paths. Small and standard systems
reset reset servicename
Example:
reset console
Resets a service. servicename indicates the name of the service to reset. If the service has not been started, this command will start the service. If the service is running, the command will stop the service and then restart it. Small and standard systems
reboot reboot subsystem
Example:
reboot updater
Restarts the system. subsystem is optional. If it is not specified, the device enters the current system upon restarting. If it is specified, the device enters the corresponding subsystem upon restarting. For example, if you run reboot updater, the device enters the updater subsystem upon restarting. Small and standard systems
sleep sleep time
Example:
sleep 5
Enters the sleep mode. time indicates the sleep time.
To avoid impact on services, exercise caution when running this command.
Small and standard systems
domainname domainname name
Example:
domainname localdomain
Sets the domain name. Small and standard systems
hostname hostname name
Example:
hostname localhost
Sets the host name. Small and standard systems
wait wait filepath
Example:
wait /data/testfile or wait /data/testfile 5
Waits for commands. The waiting time must not exceed 5 seconds. Small and standard systems
setrlimit setrlimit resource curValue maxValue
Example:
setrlimit RLIMIT_CPU 10 100
Sets resource usage restrictions.
For details, see the resource[] array of the DoSetrlimit function in base/startup/init/services/init/init_common_cmds.c.
Small and standard systems
exec exec Path of the executable file Parameters passed by the executable file
Example:
exec /system/bin/udevadm trigger
Runs an executable file. The command must not contain more than 10 parameters. Small and standard systems
syncexec syncexec Path of the executable file Parameters passed by the executable file
Example:
syncexec /system/bin/udevadm trigger
Runs an executable file synchronously. The **wait** function will be called to wait for the child process to end. The command must not contain more than 10 parameters. Standard system
mknode mknod name { b | c } Major Minor
Example:
mknod path b 0644 1 9
Creates an index node corresponding to a directory entry and a special file. Standard system
makedev makedev major minor
Example:
makedev -v update
Creates a static device node, which is usually in the /dev directory. Standard system
symlink symlink target link_name
Example
symlink /proc/self/fd/0 /dev/stdin
Creates a symbolic link. Standard system
trigger trigger jobName
Example
trigger early-fs
Triggers a job. Standard system
insmod insmod [-f] [options]
Example
insmod xxx.ko
Loads a kernel module file. Standard system
setparam setparam paramname paramvalue
Example:
setparam sys.usb.config hdc
Sets system parameters. Standard system
load_persist_params load persist params
Example:
load_persist_params 
Loads persist parameters. There must be one and only one space after the load_persist_params command. Standard system
load_param load params
Example:
load_param /data/test.normal.para
Loads the parameters from a file to the memory. Standard system
load_access_token_id load_access_token_id  Writes the access token to the data/service/el0/access_token/nativetoken.json file. There is one and only one space after load_access_token_id. Standard system
ifup ifup NIC
Example:
ifup eth0
Activates the specified NIC. Standard system
mount_fstab mount_fstab fstab.test
Example:
mount_fstab /vendor/etc/fstab.test
Mounts partitions based on the fstab file. Standard system
umount_fstab umount_fstab fstab.test
Example:
umount_fstab /vendor/etc/fstab.test
Unmounts partitions based on the fstab file. Standard system
restorecon restorecon file or dir
Example:
restorecon /file
Reloads the SELinux context. Standard system
stopAllServices stopAllServices [bool]
Example:
stopAllServices false
stopAllServices
Stops all services. Standard system
umount umount path
Example:
umount /vendor
Unmounts a mounted device. Standard system
sync sync  Writes data to the disk synchronously. There is only one and only one space after sync. Standard system
timer_start timer_start serviceName
Example:
timer_start console
Starts the service timer. Standard system
timer_stop timer_stop serviceName
Example:
timer_stop console
Stops the service timer. Standard system
init_global_key init_global_key path
Example:
init_global_key /data
Initializes the encryption key of the data partition file. Standard system
init_main_user init_main_user Encrypts the main user directory. Standard system
mkswap mkswap file
Example:
mkswap /swapfile1
Creates a swap partition on a file or device. Standard system
swapon swapon file
Example:
swapon /swapfile1
Activates the swap space. Standard system
mksandbox mksandbox fileName
Example:
mksandbox system
Creates a sandbox. Standard system
loadcfg loadcfg filePath
Example:
loadcfg /patch/fstab.cfg
Loads other .cfg files. The maximum size of the target file (only /patch/fstab.cfg supported currently) is 50 KB. Each line in the /patch/fstab.cfg file is a command. The command types and formats must comply with their respective requirements mentioned in this table. A maximum of 20 commands are allowed. Small system

Available APIs

Job management is a part of the init startup process. It is a process-based function that completely serves the init startup process. It does not provide any functional APIs for other modules. It works in a way similar to command group management and does not provide help for other types of management. The following describes the job management APIs.

Table 2 Description of job parsing APIs

API Input Parameter Return Value Description Supported System Type
void ParseAllJobs(const cJSON *fileRoot) const cJSON *fileRoot void Provides the general entry for parsing jobs. Small and standard systems
static void ParseJob(const cJSON  *jobItem, Job *resJob) const cJSON *jobItem, Job *resJob void Checks whether a job exists and parses cmds in it. Small system
int GetCmdLinesFromJson(const cJSON *root, CmdLines **cmdLines) const cJSON *root, CmdLines **cmdLines int Parses cmds in the job. This API is used for the small system. It does not apply to the standard system because the trigger command and condition attribute are involved. Small and standard systems
int ParseTriggerConfig(const cJSON *fileRoot, int (*checkJobValid)(const char *jobName)) const cJSON *fileRoot, int (*checkJobValid)(const char *jobName) int Parses the trigger command in the job. Standard system
static int ParseTrigger_(const TriggerWorkSpace *workSpace, const cJSON *triggerItem, int (*checkJobValid)(const char *jobName)) const TriggerWorkSpace *workSpace, const cJSON *triggerItem, int (*checkJobValid)(const char *jobName) int Obtains the job name, condition attribute, and cmds command group. Jobs are stored in a hash table, and commands are stored in a queue structure. Standard system

Table 3 Description of the job triggering APIs

API Input Parameter Return Value Description Supported System Type
void PostTrigger(EventType type, const char *content, uint32_t contentLen) EventType type, const char *content, uint32_t contentLen void Verifies the validity of the job name and sends a job triggering event. Standard system
static void SendTriggerEvent(int type, const char *content, uint32_t contentLen) int type, const char *content, uint32_t contentLen void Performs functions such as system control and starting or stopping of services based on system parameters. Standard system
static void DoTriggerCmd(const struct CmdArgs *ctx) const struct CmdArgs *ctx void Executes the trigger command. Standard system
void DoTriggerExec(const char *triggerName) const char *triggerName void Finds a command group based on the job name and pushes the commands in the command group to the execution queue. This API is available only for the standard system. Standard system
void DoJob(const char *jobName) const char *jobName void Matches a job based on the job name and invokes DoCmdByIndex to execute the commands in the job. Small system
void DoCmdByIndex(int index, const char *cmdContent) int index, const char *cmdContent void Combines parameters and commands. Small and standard systems

Development Example

The following is the template for configuring jobs in the .cfg file. You can use it to verify the job management function.

{
    "jobs" : [{     // Basic job
            "name" : "stage1",
            "cmds" : [
                "start service1",
                "mkdir dir1"
            ]
        }, {        // Conditional job
            "name" : "param:test.name1=0",
            "condition" : "test.name1=0",
            "cmds" : [
                "chmod 0755 dir1",
                "chown root root dir1"
            ]
        }, {        // Custom job
            "name" : "param:test.name2=*",
            "condition" : "test.name2=*",
            "cmds" : [
                "chmod 0644 dir1",
                "chown system system dir1"
            ]
        }
    ]
}

The differences in job configuration are described as follows:

  1. name and cmds are mandatory for a job, and cmds must contain commands supported by the system.

  2. condition is an optional attribute of a job. It indicates that the job is triggered only when the specified condition is met; that is, the job will not be invoked in a specific phase by the code or the trigger command.

  3. The job name must comply with the specified rules. For a job whose condition is a system parameter, its name is prefixed with param:.

  4. Commands in a renamed job can be executed only after being triggered by the trigger command in other executable job command groups. By default, the trigger command is executed in the post-init phase.

  5. An existing job name can be used in different files. Jobs with the same name are regarded as the same job. When jobs with the same name are executed, the commands in these jobs are executed together.

  6. For a conditional job, a condition is usually a system parameter. You can set a specific value so that the job is triggered when the condition is met. You can also set the value to an asterisk (*) so that the job is triggered whenever the condition is met, regardless of the parameter value.

  7. For the small system, the commands in a job cannot be triggered by the trigger command in the post-init phase.