The Locking Infrastructure in the FreeBSD kernel

The presentation at AsiaBSDCon 2009 is focused on explaining theoretical approaches and pratical aspect of the locking support in the FreeBSD kernel. The locking KPI as well as underlying mechanisms and interactions with the scheduler will be analized and discussed deeply. Typical locking strategies and some edge cases will also be shown along with a detailed explanation about how to debug locking problems (deadlocks and races primirally).

PDF version – Direct video 1 2

Isolating Cluster Users for Performance and Predictability

At The Aerospace Corporation, we run a large FreeBSD based computing cluster to support engineering applications. These applications come in all shapes, sizes, and qualities of implementation. To support them and our diverse userbase we have been searching for ways to isolate jobs from one another in ways that are more effective than Unix time sharing and more fine grained than allocating whole nodes to jobs. In this paper we discuss the problem space and our efforts so far. These efforts include implementation of partial file systems virtualization and CPU isolation using CPU sets.

This presentation was held at AsiaBSDCon 2009.

PDF SummaryDirect video link

Mohamad Dikshie Fauzie: FreeBSD and SOI-Asia Project

In this paper we describe usage of FreeBSD operating system for IPv6 Multicast routing platform in SOI-Asia Project. SOI-Asia project is platform to deliver realtime lecture via UniDirectional Link of satellite to several countries in Asia. Because of limited bandwidth in satellite, we use IPv6 multicast to deliver material of lecture and realtime video and audio lecture. We also describe human resources development of operational aspect of the project in several countries in Asia.

This presentation was done at AsiaBSDCon 2009

PDF summaryDirect link

Mail system for distributed network – A. Zakharchenko

Mail system for distributed network. Presentation held by Andrey Zakharchenko at AsiaBSDCon2009:

Sometimes it is necessary to organize a mail domain for large and geographically distributed network, which may consist of independent subnets with their own separate Internet connections (and some of those connections may be not very good, or not very fast, or not very cheap). But users of the network need mail system, and it should be fast, convenient, and reliable.

In this paper some ways of distributed mail domain implementation are discussed, and a new one is introduced. The new method allows us to implement distributed mail system fast and convenient for end users, convenient for administrators, using network traffic sparingly, and reliable enough (at least without single point of failure).

PDF of presentation

Direct link

Deprecating groff for BSD manual display

Kristaps Dzonsons’ presentation: Deprecating groff for BSD manual display, AsiaBSDCon2009:

There are few GPL-licensed utilities remaining in BSD base installations, most of them written in C++. groff, GNU’s roff text-processing language implementation, claims a significant share of this count. Why does groff still persist in base? Although its text-processing features have been mostly usurped by LaTeX, groff persists in order to render Unix manual pages. In this paper, we introduce mdocml, a compiler for mdoc documents. mdocml replaces a very specific function of groff — namely, that of rendering mdoc documents for one or more output devices. It’s our intention, in contributing this tool, to deprecate groff as the default utility for Unix manual page display; by doing so, and presuming that manual display is the primary usage of groff, we intend to decouple BSD base installations from another GPL and C++ tool.

PDF of presentation

Direct link

FreeBSD on high performance multi-core embedded PowerPC systems

Rafal Jaworowski’s presentation on FreeBSD on high performance multi-core embedded PowerPC systems, AsiaBSDCon2009:

This paper describes low level design and implementation of the FreeBSD operating system port for the MPC8572 system-on-chip device, a high-end member of the Freescale PowerQUICC III family. The primary focus of this work is describing how the multi-core operation was brought forward and full SMP capabilities achieved, but other major components developed in the course of this project, device drivers in particular, are also covered.

PDF of presentation

direct link

Released: pfSense 1.2.1

pfSense project logo

pfSense project logo

The pfSense project have announced the availability of version 1.2.1:

This is a strictly a maintenance release, meaning it contains only bug fixes in the pfSense code, no new features. Though we also upgraded the base OS from FreeBSD 6.2 to 7.0, which necessitated numerous changes in how things are configured. The change to FreeBSD 7.0 brings improved performance and more hardware support.

AnnouncementChangelog | Downloads

If you’re interested in pfSense or if you’re using it in a production environment, you may be interested in the Network Perimeter Redundancy with pfSense presentation that Chris Buechler will be doing at the DCBSDCon 2009.

What do you get when you cross an enterprise-class packet filtering subsystem with a graphical front-end for easy configuration and maintenance?  A throbbing headache for commercial vendors like SonicWALL, that’s what.

More details on the presentation can be found here.

To find the details of other 2009 FreeBSD related events and conferences, check my FreeBSD Events calendar.