Linux Nice Command and CPU Scheduling Priority

The nice command in Linux is a utility used to influence the scheduling priority of processes managed by the operating system's kernel. This article explores how niceness values translate into scheduling weights, how the kernel's scheduler allocates CPU time based on these values, and the rules governing priority adjustments for both new and running processes.

Understanding the Nice Value Scale

In Linux, process priority is measured on a scale from -20 to 19, known as the "nice" value:

• -20: Highest scheduling priority (the process is "least nice" to other processes and requests maximum CPU time).
• 0: Default priority assigned to standard user processes.
• 19: Lowest scheduling priority (the process is "nicest" and readily yields CPU time to others).

How the Scheduler Interprets Niceness

Modern Linux systems primarily use the Completely Fair Scheduler (CFS) for standard non-real-time tasks. The CFS does not assign fixed time slices based on priority; instead, it aims to allocate CPU proportions fairly using a metric called vruntime (virtual runtime).

1. Weight Mapping: The kernel maps each nice value to an internal weight. These weights follow a geometric progression where each step in the nice scale alters CPU allocation by roughly 10%.
2. Virtual Runtime Tracking: As a process runs, CFS tracks its vruntime. The rate at which vruntime increases is inversely proportional to the task's weight:
   • A process with a negative nice value (high priority/high weight) accumulates vruntime slowly.
   • A process with a positive nice value (low priority/low weight) accumulates vruntime quickly.
3. Task Selection: The scheduler always chooses the runnable task with the lowest vruntime to execute next. Because higher-priority tasks accumulate vruntime at a slower rate, they are selected more frequently and receive a larger share of overall CPU cycles.

Using the nice and renice Commands

The nice command sets the priority when starting a new process, while renice adjusts an already running task.

• Starting a low-priority task:

  nice -n 10 python script.py

• Starting a high-priority task:

  sudo nice -n -5 python script.py

• Adjusting an existing process:

  renice -n 15 -p 1234

Permissions and Security

To prevent resource abuse, the operating system enforces permission constraints on nice values:

• Unprivileged Users: Can only set positive nice values (0 to 19) and can only increase the niceness of their own processes (lowering their priority). Unprivileged users cannot lower a nice value once it has been raised.
• Root/Superuser: Can assign any value from -20 to 19, allowing processes to receive boosted priority over standard system tasks.