Copyright (c) 2011 Alexander Motin <mav@FreeBSD.org> All rights reserved. Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met: 1. Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer. 2. Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the docum...
NAMEtimecounters - kernel time counters subsystem
SYNOPSISThe kernel uses several types of time-related devices, such as: real time clocks, time counters and event timers. Real time clocks are responsible for tracking real world time, mostly when the system is down. Time counters are responsible for tracking purposes, when the system is running. Event timers are responsible for generating interrupts at a specified time or periodically, to run different time-based events. This page is about the second.
DESCRIPTIONTime counters are the lowest level of time tracking in the kernel. They provide monotonically increasing timestamps with known width and update frequency. They can overflow, drift, etc and so in raw form can be used only in very limited performance-critical places like the process scheduler.
More usable time is created by scaling the values read from the selected time counter and combining it with some offset, regularly updated by Fn tc_windup on Fn hardclock invocation.
Different platforms provide different kinds of timer hardware. The goal of the time counters subsystem is to provide a unified way to access that hardware.
Each driver implementing time counters registers them with the subsystem. It is possible to see the list of present time counters, via the kern.timecounter sysctl(8) variable:
kern.timecounter.choice: TSC-low(-100) HPET(950) i8254(0) ACPI-fast(900) dummy(-1000000) kern.timecounter.tc.ACPI-fast.mask: 16777215 kern.timecounter.tc.ACPI-fast.counter: 13467909 kern.timecounter.tc.ACPI-fast.frequency: 3579545 kern.timecounter.tc.ACPI-fast.quality: 900 kern.timecounter.tc.i8254.mask: 65535 kern.timecounter.tc.i8254.counter: 62692 kern.timecounter.tc.i8254.frequency: 1193182 kern.timecounter.tc.i8254.quality: 0 kern.timecounter.tc.HPET.mask: 4294967295 kern.timecounter.tc.HPET.counter: 3013495652 kern.timecounter.tc.HPET.frequency: 14318180 kern.timecounter.tc.HPET.quality: 950 kern.timecounter.tc.TSC-low.mask: 4294967295 kern.timecounter.tc.TSC-low.counter: 4067509463 kern.timecounter.tc.TSC-low.frequency: 11458556 kern.timecounter.tc.TSC-low.quality: -100
The output nodes are defined as follows:
- kern.timecounter.tc. X .mask
- is a bitmask, defining valid counter bits,
- kern.timecounter.tc. X .counter
- is a present counter value,
- kern.timecounter.tc. X .frequency
- is a counter update frequency,
- kern.timecounter.tc. X .quality
- is an integral value, defining the quality of this time counter compared to others. A negative value means this time counter is broken and should not be used.
The time management code of the kernel automatically switches to a higher-quality time counter when it registers, unless the kern.timecounter.hardware sysctl has been used to choose a specific device.
There is no way to unregister a time counter once it has registered with the kernel. If a dynamically loaded module contains a time counter you will not be able to unload that module, even if the time counter it contains is not the one currently in use.