You are here: Home / RTLWS 1999-2017 / RTLWS Submitted Papers / 
2022-11-27 - 00:49

Dates and Events:

OSADL Articles:

2022-01-13 12:00

Phase #3 of OSADL project on OPC UA PubSub over TSN successfully completed

Another important milestone on the way to interoperable Open Source real-time Ethernet has been reached


2021-02-09 12:00

Open Source OPC UA PubSub over TSN project phase #3 launched

Letter of Intent with call for participation is now available


2016-11-12 12:00

Raspberry Pi and real-time Linux

Let's have a look at the OSADL QA Farm data



Real Time Linux Workshops

1999 - 2000 - 2001 - 2002 - 2003 - 2004 - 2005 - 2006 - 2007 - 2008 - 2009 - 2010 - 2011 - 2012 - 2013 - 2014 - 2015

Eleventh Real-Time Linux Workshop on September 28 to 30, in Dresden, Germany

Announcement - Hotels - Agenda - Paper Abstracts - Presentations - Registration - Abstract Submission - Xenomai User Meeting - Sponsors

Papers

Timing analysis of Linux CAN drivers

Michal Sojka, Czech Technical University
Pavel Pisa, Czech Technical University

CAN-bus (Controller-Area Network) is widely used communication bus in industrial and automotive systems. Besides that it simplifies wirings, CAN-bus offers deterministic medium access algorithm for which it is very suitable for certain type of time-critical applications. Typical CAN-based network consists of multiple microcontrollers and one or a few more complex nodes which implement higher-level control algorithms. These more complex nodes can run Linux OS so there is a need for CAN-bus drivers. There exist several projects which offer CAN drivers for Linux. Probably the most known driver is Socket-CAN, which is already partially included in the mainline kernel. Another option is LinCAN driver. The basic difference between these two drivers is that Socket-CAN is built around standard Linux networking infrastructure, whereas LinCAN is a character device driver with its own queuing infrastructure.

We have developed a tool for measuring delays on CAN-bus and use it in several tests which measure delays under different system loads and other conditions.

The results of this work are useful for the following reasons:

  1. Users can approximately know timing properties of a particular CAN-on-Linux solutions.
  2. Developed tools can be used for benchmarking other hardware to find suitability for intended applications.
  3. Driver developers can profile and improve their drivers and find factors contributing to worst-case delays in RX/TX paths.
  4. Overhead caused by the use of generic Linux networking infrastructure in Socket-CAN implementation was quantified. Our finding is that the difference is not negligible, but it is still suitable for many applications.