Anatomy of Linux process management

Creation, management, scheduling, and destruction

The creation and management of user-space processes in Linux® have many principles in common with UNIX® but also include several unique optimizations specific to Linux. Here, review the life cycle of Linux processes and explore the kernel internals for user process creation, memory management, scheduling, and death.

M. Tim Jones, Consultant Engineer, Emulex Corp.

M. Tim JonesM. Tim Jones is an embedded firmware architect and the author of Artificial Intelligence: A Systems Approach, GNU/Linux Application Programming (now in its second edition), AI Application Programming (in its second edition), and BSD Sockets Programming from a Multilanguage Perspective. His engineering background ranges from the development of kernels for geosynchronous spacecraft to embedded systems architecture and networking protocols development. Tim is a Consultant Engineer for Emulex Corp. in Longmont, Colorado.



20 December 2008

Also available in Russian Japanese Portuguese

Linux is a very dynamic system with constantly changing computing needs. The representation of the computational needs of Linux centers around the common abstraction of the process. Processes can be short-lived (a command executed from the command line) or long-lived (a network service). For this reason, the general management of processes and their scheduling is very important.

From user-space, processes are represented by process identifiers (PIDs). From the user's perspective, a PID is a numeric value that uniquely identifies the process. A PID doesn't change during the life of a process, but PIDs can be reused after a process dies, so it's not always ideal to cache them.

In user-space, you can create processes in any of several ways. You can execute a program (which results in the creation of a new process) or, within a program, you can invoke a fork or exec system call. The fork call results in the creation of a child process, while an exec call replaces the current process context with the new program. I discuss each of these methods to understand how they work.

For this article, I build the description of processes by first showing the kernel representation of processes and how they're managed in the kernel, then review the various means by which processes are created and scheduled on one or more processors, and finally, what happens if they die.

Process representation

Within the Linux kernel, a process is represented by a rather large structure called task_struct. This structure contains all of the necessary data to represent the process, along with a plethora of other data for accounting and to maintain relationships with other processes (parents and children). A full description of the task_struct is beyond the scope of this article, but a portion of task_struct is shown in Listing 1. This code contains the specific elements this article explores. Note that task_struct resides in ./linux/include/linux/sched.h.

Listing 1. A small portion of task_struct
struct task_struct {

	volatile long state;
	void *stack;
	unsigned int flags;

	int prio, static_prio;

	struct list_head tasks;

	struct mm_struct *mm, *active_mm;

	pid_t pid;
	pid_t tgid;

	struct task_struct *real_parent;

	char comm[TASK_COMM_LEN];

	struct thread_struct thread;

	struct files_struct *files;

	...

};

In Listing 1, you can see several items that you'd expect, such as the state of execution, a stack, a set of flags, the parent process, the thread of execution (of which there can be many), and open files. I explore these later in the article but will introduce a few here. The state variable is a set of bits that indicate the state of the task. The most common states indicate that the process is running or in a run queue about to be running (TASK_RUNNING), sleeping (TASK_INTERRUPTIBLE), sleeping but unable to be woken up (TASK_UNINTERRUPTIBLE), stopped (TASK_STOPPED), or a few others. A complete list of these flags is available in ./linux/include/linux/sched.h.

The flags word defines a large number of indicators, indicating everything from whether the process is being created (PF_STARTING) or exiting (PF_EXITING), or even if the process is currently allocating memory (PF_MEMALLOC). The name of the executable (excluding the path) occupies the comm (command) field.

Each process is also given a priority (called static_prio), but the actual priority of the process is determined dynamically based on loading and other factors. The lower the priority value, the higher its actual priority.

The tasks field provides the linked-list capability. It contains a prev pointer (pointing to the previous task) and a next pointer (pointing to the next task).

The process's address space is represented by the mm and active_mm fields. The mm represents the process's memory descriptors, while the active_mm is the previous process's memory descriptors (an optimization to improve context switch times).

Finally, the thread_struct identifies the stored state of the process. This element depends on the particular architecture on which Linux is running, but you can see an example of this in ./linux/include/asm-i386/processor.h. In this structure, you'll find the storage for the process when it is switched from the executing context (hardware registers, program counter, and so on).


Process management

Maximum processes

Although processes are dynamically allocated within Linux, certain maximums are observed. The maximum is represented in the kernel by a symbol called max_threads, which can be found in ./linux/kernel/fork.c). You can change this value from user-space through the proc file system at /proc/sys/kernel/threads-max.

Now, let's explore how you manage processes within Linux. In most cases, processes are dynamically created and represented by a dynamically allocated task_struct. One exception is the init process itself, which always exists and is represented by a statically allocated task_struct. You can see an example of this in ./linux/arch/i386/kernel/init_task.c.

All processes in Linux are collected in two different ways. The first is a hash table, which is hashed by the PID value; the second is a circular doubly linked list. The circular list is ideal for iterating through the task list. As the list is circular, there's no head or tail; but as the init_task always exists, you can use it as an anchor point to iterate further. Let's look at an example of this to walk through the current set of tasks.

The task list is not accessible from user-space, but you can easily solve that problem by inserting code into the kernel in the form of a module. A very simple program is shown in Listing 2 that iterates the task list and provides a small amount of information about each task (name, pid, and parent name). Note here that the module uses printk to emit the output. To view the output, you need to view the /var/log/messages file with the cat utility (or tail -f /var/log/messages in real time). The next_task function is a macro in sched.h that simplifies the iteration of the task list (returns a task_struct reference of the next task).

Listing 2. Simple kernel module to emit task information (procsview.c)
#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/sched.h>

int init_module( void )
{
  /* Set up the anchor point */
  struct task_struct *task = &init_task;

  /* Walk through the task list, until we hit the init_task again */
  do {

    printk( KERN_INFO "*** %s [%d] parent %s\n",
		task->comm, task->pid, task->parent->comm );

  } while ( (task = next_task(task)) != &init_task );

  return 0;

}

void cleanup_module( void )
{
  return;
}

You can compile this module with the Makefile shown in Listing 3. When compiled, you can insert the kernel object with insmod procsview.ko and remove it with rmmod procsview.

Listing 3. Makefile to build the kernel module
obj-m += procsview.o

KDIR := /lib/modules/$(shell uname -r)/build
PWD := $(shell pwd)

default:
	$(MAKE) -C $(KDIR) SUBDIRS=$(PWD) modules

After insertion, /var/log/messages displays output as shown below. You can see here the idle task (called swapper) and the init task (pid 1).

Nov 12 22:19:51 mtj-desktop kernel: [8503.873310] *** swapper [0] parent swapper
Nov 12 22:19:51 mtj-desktop kernel: [8503.904182] *** init [1] parent swapper
Nov 12 22:19:51 mtj-desktop kernel: [8503.904215] *** kthreadd [2] parent swapper
Nov 12 22:19:51 mtj-desktop kernel: [8503.904233] *** migration/0 [3] parent kthreadd
...

Note that it's also possible to identify the currently running task. Linux maintains a symbol called current that is the currently running process (of type task_struct). If at the end of init_module you add the line:

printk( KERN_INFO, "Current task is %s [%d], current->comm, current->pid );

you would see:

Nov 12 22:48:45 mtj-desktop kernel: [10233.323662] Current task is insmod [6538]

Note that the current task is insmod, because the init_module function executes within the context of the execution of the insmod command. The current symbol actually refers to a function (get_current) and can be found in an arch-specific header (for example, ./linux/include/asm-i386/current.h).


Process creation

System call functions

You've probably seen a pattern with the system calls. In many cases, system calls are named sys_* and provide some of the initial functionality to implement the call (such as error checking or user-space activities). The real work is often delegated to another function called do_*.

So, let's walk through the creation of a process from user-space. The underlying mechanism is the same for user-space tasks and kernel tasks, as both eventually rely on a function called do_fork to create the new process. In the case of creating a kernel thread, the kernel calls a function called kernel_thread (see ./linux/arch/i386/kernel/process.c), which performs some initialization, then calls do_fork.

A similar action occurs for user-space process creation. In user-space, a program calls fork, which results in a system call to the kernel function called sys_fork (see ./linux/arch/i386/kernel/process.c). The function relationships are shown graphically in Figure 1.

Figure 1. Function hierarchy for process creation
Function hierarchy for process creation

From Figure 1, you can see that do_fork provides the basis for process creation. You can find the do_fork function in ./linux/kernel/fork.c (along with the partner function, copy_process).

The do_fork function begins with a call to alloc_pidmap, which allocates a new PID. Next, do_fork checks to see whether the debugger is tracing the parent process. If it is, the CLONE_PTRACE flag is set in the clone_flags in preparation for forking. The do_fork function then continues with a call to copy_process, passing the flags, stack, registers, parent process, and newly allocated PID.

The copy_process function is where the new process is created as a copy of the parent. This function performs all actions except for starting the process, which is handled later. The first step in copy_process is validation of the CLONE flags to ensure that they're consistent. If they're not, an EINVAL error is returned. Next, the Linux Security Module (LSM) is consulted to see whether the current task may create a new task. To learn more about LSMs in the context of Security-Enhanced Linux (SELinux), check out the Resources section.

Next, the dup_task_struct function (found in ./linux/kernel/fork.c) is called, which allocates a new task_struct and copies the current process's descriptors into it. After a new thread stack is set up, some state information is initialized and control returns to copy_process. Back in copy_process, some housekeeping is performed in addition to several other limit and security checks, including a variety of initialization on your new task_struct. A sequence of copy functions is then invoked that copy individual aspects of the process, from copying open file descriptors (copy_files), copying signal information (copy_sighand and copy_signal), copying process memory (copy_mm), and finally copying the thread (copy_thread).

The new task is then assigned to a processor, with some additional checking based on the processors on which the process is allowed to execute (cpus_allowed). After the priority of the new process inherits the priority of the parent, a small amount additional housekeeping is performed, and control returns to do_fork. At this point, your new process exists but is not yet running. The do_fork function fixes this with a call to wake_up_new_task. This function, which you can find in ./linux/kernel/sched.c), initializes some of the scheduler housekeeping information, places the new process in a run queue, then wakes it up for execution. Finally, upon returning to do_fork, the PID value is returned to the caller and the process is complete.


Process scheduling

While a process exists in Linux, it can potentially be scheduled through the Linux scheduler. Although outside of the scope of this article, the Linux scheduler maintains a set of lists for each priority level on which task_struct references reside. Tasks are invoked through the schedule function (available in ./linux/kernel/sched.c), which determines the best process to run based on loading and prior process execution history. You can learn more about the Linux version 2.6 scheduler in Resources.


Process destruction

Process destruction can be driven by several events—from normal process termination, through a signal, or through a call to the exit function. However process exit is driven, the process ends through a call to the kernel function do_exit (available in ./linux/kernel/exit.c). This process is shown graphically in Figure 2.

Figure 2. Function hierarchy for process destruction
Function hierarchy for process destruction

The purpose behind do_exit is to remove all references to the current process from the operating system (for all resources that are not shared). The destruction process first indicates that the process is exiting by setting the PF_EXITING flag. Other aspects of the kernel use this indication to avoid manipulating this process while it's being removed. The cycle of detaching the process from the various resources that it attained during its life is performed through a series of calls, including exit_mm (to remove memory pages) to exit_keys (which disposes of per-thread session and process security keys). The do_exit function performs various accountings for the disposal of the process, then a series of notifications (for example, to signal the parent that the child is exiting) is performed through a call to exit_notify. Finally, the process state is changed to PF_DEAD, and the schedule function is called to select a new process to execute. Note that if signalling is required to the parent (or the process is being traced), the task will not completely disappear. If no signalling is necessary, a call to release_task will actually reclaim the memory that the process used.


Going further

Linux continues to evolve, and one area that will see further innovation and optimization is process management. While keeping true to UNIX principles, Linux continues to push the boundaries. New processor architectures, symmetrical multiprocessing (SMP), and virtualization will drive new advances in this area of the kernel. One example is the new O(1) scheduler introduced in Linux version 2.6, which provides scalability for systems with large numbers of tasks. Another is the updated threading model using the Native POSIX Thread Library (NPTL), which enables efficient threading beyond the prior LinuxThreads model. You can learn more about these innovations and what's ahead in Resources.

Resources

Learn

Get products and technologies

  • With IBM trial software, available for download directly from developerWorks, build your next development project on Linux.

Discuss

Comments

developerWorks: Sign in

Required fields are indicated with an asterisk (*).


Need an IBM ID?
Forgot your IBM ID?


Forgot your password?
Change your password

By clicking Submit, you agree to the developerWorks terms of use.

 


The first time you sign into developerWorks, a profile is created for you. Information in your profile (your name, country/region, and company name) is displayed to the public and will accompany any content you post, unless you opt to hide your company name. You may update your IBM account at any time.

All information submitted is secure.

Choose your display name



The first time you sign in to developerWorks, a profile is created for you, so you need to choose a display name. Your display name accompanies the content you post on developerWorks.

Please choose a display name between 3-31 characters. Your display name must be unique in the developerWorks community and should not be your email address for privacy reasons.

Required fields are indicated with an asterisk (*).

(Must be between 3 – 31 characters.)

By clicking Submit, you agree to the developerWorks terms of use.

 


All information submitted is secure.

Dig deeper into Linux on developerWorks


static.content.url=http://www.ibm.com/developerworks/js/artrating/
SITE_ID=1
Zone=Linux
ArticleID=360339
ArticleTitle=Anatomy of Linux process management
publish-date=12202008