SYNOPSIS

Kernel uses several types of time-related devices, such as: real time clocks, time counters and event timers. Real time clocks responsible for tracking real world time, mostly when system is down. Time counters are responsible for generation of monotonically increasing timestamps for precise uptime tracking purposes, when system is running. Event timers are responsible for generating interrupts at specified time or periodically, to run different time-based events. This page is about the last.

DESCRIPTION

Kernel uses time-based events for many different purposes: scheduling, statistics, time keeping, profiling and many other things, based on callout(9) mechanism. These purposes now grouped into three main callbacks:

Fn hardclock

callout(9) and timekeeping events entry. Called with frequency defined by hz variable, usually 1000Hz.

Fn statclock

statistics and scheduler events entry. Called with frequency about 128Hz.

Fn profclock

profiler events entry. When enabled, called with frequency about 8KHz.

Different platforms provide different kinds of timer hardware. The goal of the event timers subsystem is to provide unified way to control that hardware, and to use it, supplying kernel with all required time-based events.

Each driver implementing event timers, registers them at the subsystem. It is possible to see the list of present event timers, like this, via kern.eventtimer sysctl:

kern.eventtimer.choice: HPET(550) LAPIC(400) i8254(100) RTC(0)
kern.eventtimer.et.LAPIC.flags: 15
kern.eventtimer.et.LAPIC.frequency: 0
kern.eventtimer.et.LAPIC.quality: 400
kern.eventtimer.et.i8254.flags: 1
kern.eventtimer.et.i8254.frequency: 1193182
kern.eventtimer.et.i8254.quality: 100
kern.eventtimer.et.RTC.flags: 17
kern.eventtimer.et.RTC.frequency: 32768
kern.eventtimer.et.RTC.quality: 0
kern.eventtimer.et.HPET.flags: 7
kern.eventtimer.et.HPET.frequency: 14318180
kern.eventtimer.et.HPET.quality: 550

where:

kern.eventtimer.et. X .flags

is a bitmask, defining event timer capabilities:

1

periodic mode supported,

2

one-shot mode supported,

4

timer is per-CPU,

8

timer may stop when CPU goes to sleep state,

16

timer supports only power-of-2 divisors.

kern.eventtimer.et. X .frequency

is a timer base frequency,

kern.eventtimer.et. X .quality

is an integral value, defining how good is this timer, comparing to others.

Timers management code of the kernel chooses one timer from that list. Current choice can be read and affected via kern.eventtimer.timer tunable/sysctl. Several other tunables/sysctls are affecting how exactly this timer is used:

kern.eventtimer.periodic

allows to choose periodic and one-shot operation mode. In periodic mode, periodic interrupts from timer hardware are taken as the only source of time for time events. One-shot mode instead uses currently selected time counter to precisely schedule all needed events and programs event timer to generate interrupt exactly in specified time. Default value depends of chosen timer capabilities, but one-shot mode is preferred, until other is forced by user or hardware.

kern.eventtimer.singlemul

in periodic mode specifies how much times higher timer frequency should be, to not strictly alias Fn hardclock and Fn statclock events. Default values are 1, 2 or 4, depending on configured HZ value.

kern.eventtimer.idletick

makes each CPU to receive every timer interrupt independently of whether they busy or not. By default this options is disabled. If chosen timer is per-CPU and runs in periodic mode, this option has no effect - all interrupts are always generating.