Sunday, August 22, 2021

Recent and not so recent changes in OpenBSD that make life better (and may turn up elsewhere too)

Known to be "functional, free and secure by default", the OpenBSD operating system has played an important role in open source for more than a quarter century. It has also been fairly central to what I have done for the last two decades and some. What follows is my personal view of what life with OpenBSD has been like, with an emphasis on moments and developments that I feel made life, or at least my life, better.

I will assume that you know already that one of the signature features of the OpenBSD project is the continuous code audit and the sharp focus on secure and correct code. The audit by itself has produced a number of improvements, including a stream of bugfixes with bugs of a similar kind fixed in the whole tree and even the occasional subsystem rewrite. In addition, even for a free operating system project, life just happens. The world changes around us and drives the developers to take up fresh approaches to both new and well known problems and in the process develop code in ways that improves life for us all.

The Norwegian tech news site digi.no took this article in as a three part series, see parts 1, 2 and 3 if you want to read this in my native language.  

If you are not that familiar with OpenBSD the system or project, my "OpenBSD and you" talk, which I update occasionally, might be a not too bad place to start. But in this article I will focus on some specific moments when I felt that changes in OpenBSD made my life better. These are the things that made me start and go on advocating the system.

So who am I and what can I offer?

My name is Peter Hansteen. I have worked in information technology and information technology related things like documentation since the late 1980s. I am in the process of transitioning from a "Security Engineer" role into a "Cloud Expert" one, and across several other roles and titles I have always done a bit, or a lot of Unix system and network administration. At most times you will find me on The Other West Coast, specifically in Bergen, Norway.

Note: If you are more of a slides person than a fulltext person, you may be relieved to hear that you can find the slides for the talk this article is based on (and vice versa) here.

The installer was always good, got better

When I found OpenBSD more than twenty years ago, my main Unix exposure was from working with Linuxes and FreeBSD. What attracted me to OpenBSD and finally had me buy an OpenBSD 2.5 CD set was the strong focus on security and code correctness. When the CD set and the classic wireframe daemon T-shirt finally arrived in the mail, I set about at first to install it on whatever spare hardware I had lying around.

OpenBSD wireframe daemon head


If I remember correctly, the first machine I tried installing OpenBSD on was an 80386/33MHz with 8MB RAM and I think a 100MB IDE hard disk. Which I can report sounded pretty crappy even then, but the thing did work.

The initial install was fairly straightforward, and when I started poking around I found two things about myself and the new system: Everyting made sense, and everything I could think of had a readable man page. So the first change I am aware of that made the world better with OpenBSD was the decision to enforce the "No commit without documentation" rule, which came into being early in the project's life, probably roughly at the same time the OpenBSD developers gave us a real-time view of development via anonymous CVS. You can see things happening in almost real time.

It is worth mentioning that the installer has remained famously non-graphical, text only. The reason the installer remains text-only is that this is a major advantage that enables the developers and the users to handle the fairly diverse collection of hardware platforms that OpenBSD runs on with the same portable, familiar and compact code everywhere.

The installer was always scriptable and extensible, and over the years the installer has added automatic, repeatable and scriptable installs (dubbed autoinstall(8) which appeared in OpenBSD 5.5 in 2014) and the sysupgrade(8) extension (first found in OpenBSD 6.6 in 2019) that automates snapshot to snapshot or release to subsequent release upgrades for all not too hacked-up configurations. Each of these moments, or more specifically when the new code started appearing in snapshots, had me appreciate the OpenBSD system a bit more, and made me feel quality of life had improved.

Now something for your laptop - hardware support

Fast forward some twenty-plus years and the last article I published, and even got into Norwegian mainstream IT news site Digi.no, centers on a few moments involving new OpenBSD developments. It took some interaction with OpenBSD developers, but those interactions lead to my new laptop with an 11th generation Intel Core chipset working even better with OpenBSD. Yes, OpenBSD developers and a significant subset of their user base actually run OpenBSD on their laptops. I do use a Mac and a work-issued Thinkpad with Ubuntu Linux too, but life is not complete without an OpenBSD laptop.

Now to be honest, what I saw within the space of a few days was development that had me going from "Oh, sh*t, the SSD isn't recognized" -- the controller was set to a RAID-ish mode by default -- through this kernel panic:

OpenBSD 6.9-current panic message


-- to seeing it all fully supported.

The SSD problem turned out to be simple to fix: Simply find the "Advanced" BIOS option that turned the pseudo-raid feature off and let the operating system speak directly to the storage device.

For the rest there was a period of a couple of weeks I had to run with not yet commited patches in a home baked kernel built from checkouts from Jonathan Grey's git repo. When the code was committed to -current, I could resume my normal sysupgrade(8) routine, going from one development snapshot to the next.

The process, even with the need to build custom kernels for a while, was actually quite pleasant, and when the support code went into the main development branch, that too was a a moment when I felt my life had been improved by changes in OpenBSD. The hardware support is now in snapshots and will be in OpenBSD 7.0 which is set to be released approximately early November 2021.

Living the life dynamic

Now that we're talking about laptops, there is another recent development that makes your OpenBSD on laptop experience even better. Laptops and other equipment that uses dynamic network configuration became easier to operate with dhcpleased(8) now enabled by default in OpenBSD 6.9-current after it was first introduced in OpenBSD 6.9. That change marked the completion of a five year cycle of incremental development which involved writing several new daemons. Each of those programs was designed with the Unix philosophy that a program should do one thing and do it well.

The first piece of the puzzle was slaacd(8) - the stateless IPv6 address autoconfiguration daemon - which appeared in OpenBSD 6.2 to handle IPv6 automatic configuration, listening for route advertisements.

The corresponding router advertisement daemon rad(8) arrived in OpenBSD 6.4. That got most of the things involved in IPv6 autoconfiguration in order.

Next up was the arrival in OpenBSD 6.5 of unwind(8), a validating DNS resolver which learns which resolvers to query from DHCP and other sources.

To complete the set, OpenBSD 6.9 brought us resolvd(8) to manage and edit /etc/resolv.conf, updating the file with information learned from other sources, and dhcpleased(8) now serves as the client for IPv4 DHCP client information which is then fed into the configuration.

Combined with setting your laptop to join networks as they become available, moving between networks can now be an non-disruptive, even unremarkable event.

This all comes into place if you edit your /etc/hostname.$if for (for example hostname.iwx0) to something like

join adipose wpakey thedoctorknows
join tardis  wpakey biggerontheinside
join cybermen wpakey nowedont
inet autoconf
inet6 autoconf

you should expect minimal efforts needed when moving between those networks. As usual, as soon as a new feature is trusted, it is on by default in OpenBSD-current, and OpenBSD 7.0 will ship with this behavior enabled by default for interfaces set to autoconf for either inet or inet6.

But that is the modern day and for some in the future. OpenBSD on a laptop is a good experience. On the other hand, most of the world sees the BSDs and OpenBSD in particular as mainly server or even network device operating systems, despite the fairly high BSD code content in such things as Apple systems.

The thing that lured me in

But I hear you ask, what made me turn into an almost all-in OpenBSD user?

Back in 2001 I was still only experimenting with OpenBSD, but my experience with Linux and iptables had made me long for a switch to a saner firewall. I had done some small experiments with the IPF firewall that was in OpenBSD until the 2.9 release. Then, as some of us will remember, the it was discovered that IPF's license was in fact not free, so it needed to be replaced.

There was a distinct rush, not quite a stampede, to replace IPF over the months that followed. Fortunately, the new code that replaced the previous packet filter proved to perform better. The OpenBSD Packet filter, dubbed PF for short, had been born and made its debut in OpenBSD 3.0 in December 2001. The release had originally been planned for November, but was pushed out a month to hack the "working prototype" packet filter into something usable.

Almost needless to say, this turn of events finally pushed me to take the final steps to replace the Linux gateways I had in place with OpenBSD ones. I was pleasantly surprised to find that not only did they perform well, but they also came with complete and reasonably well documented tools so I could understand what was going on. That's how I got started on the process that lead to among other things writing The Book of PF and taking that text through three editions so far. But more about that later.

It is worth noting that the IPFilter copyright episode spurred the OpenBSD developers to perform a license audit of the entire source tree and ports in order to avoid similar situations in the future. This activity ran for some months and uncovered a number of potential problems. Theo de Raadt summed up the effort in a message to the openbsd-misc mailing list on February 20th, 2003.

What they found when they started looking was that there was a significant number of files that were in fact not under a free license, much like the entire IPF subsystem had been. Those needed to be replaced. Other parts had either no license or no copyright stated. In some cases the developers gave explicit permission to continuing use, but quite a few things needed to be rewritten with a free license so OpenBSD and other free software would be able to move forward without copyright problems.

I later heard in a rather informal setting that among the no copyright and/or no license cases, it was usually possible to track down the developers via version control system logs or mailing list archives. In a large number of those cases, the initial reaction was along the lines "Say what? Are people still using that?".

SSH, open and better

PF was written from scratch to replace a subsystem that it turned out was illegal to use in an open source context. But it was not the first time the OpenBSD project had performed a nonlibreectomy, that is, taken on the task of replacing code for license reasons.

A few years earlier it had become clear that the original developer of the secure shell system ssh had commercial ambitions and the license for the software had changed in a proprietary direction. After a bit of deliberation on how to resolve the situation, the OpenBSD developers started digging around for earlier versions of the code that had been published with an acceptable license. Then they forked their version from the last version they found that still had free license. Next came an intensive period of re-introducing the features that were missing in the old code.

The result was introduced as OpenSSH in OpenBSD 2.6 in 1999. Over the next few years OpenSSH grew a portable version that started grabbing market share rapidly. The last I heard OpenSSH's market share is somewhere in the high nineties percent.

With a state of the art secure shell subsystem in place and growing all sorts of useful features, the time finally came to end unencrypted shell login sessions on OpenBSD. OpenBSD's telnetd was moved to the CVS attic in time for OpenBSD 3.8, which was released November 2005.

One other notable thing about OpenSSH is that it was the first daemon to be properly privilege separated, a model practice that debuted with the overhauled OpenSSH in OpenBSD 3.2 in March 2002. Since then privilege separation has been put in place in all daemons where it made sense to do so, and it is now a signature part of the secure by default stance of all newer OpenBSD daemons.

And yes, that packet filter

I mentioned PF, the OpenBSD packet filter, earlier. I must confess that PF has been an important part of my life in various context since the early noughties. Over the years, things I have written have contributed to creating the popular but actually wrong perception that OpenBSD was primarily a firewall operating system. There are a lot of useful and fun features that turned up in or in connection with PF over the years and were pioneered by OpenBSD. Some features were ported to or imitated in other systems, while others remain stubbornly OpenBSD only.

So I will touch on some of my favorite PF and PF-attached features, in quasi-random but almost chronological order.

Beating up spammers with OpenBSD spamd(8) since OpenBSD 3.3

When I started playing with OpenBSD in general and PF in particular way back when, I was already responsible for the SMTP mail service for my colleagues. My gateways by then ran OpenBSD, while the mail server rosalita, named after a Springsteen song, was not too badly specced server running FreeBSD with exim as the mail transfer agent that fed the incoming messages to spamassassin and clamav for content filtering before handing off to user mailboxes.

So when it dawned on me that I could set up spamd(8) the spam deferral daemon on the internet-facing gateway and save load on the poor suffering rosalita that was running hot with content filtering, I was quick to implement a setup that sucked in well known block lists.

Going grey, then trapping

The effect was obvious and immediate, the mail server's fans grew noticeably quieter. When greylisting was introduced in spamd soon after, I implemented that too, and witnessed yet another drop in pitch and intensity of the sound from rosalita's fans. Then a couple of releases later greytrapping -- the practice of adding IP addresses of incoming SMTP connections to blocklists if the attempted delivery is aimed at a known-bad address in the target domain -- was introduced, and that sounded like enough fun that I just went ahead and did it.

The idea of detecting spam senders by the bogus addresses they were already trying to deliver to just sounded too good to not try. And we knew that getting started would be pretty easy too. We had seen rejects for addresses that had never existed in our domains in our mail server logs for quite a while, so it was simply a matter of harvesting from a fairly bountiful source and adding stuff that we were sure would never ever be actually deliverable here to the spamtrap list. I think the first setup had only a couple of hundred entries in it, but I did not note the exact number at the time.

By July 2007 I had decided to publish both the list of spamtrap addresses and an hourly dump of the greytrapped addresses. Both remain free to download. The list of spamtraps, harvested from various log sources, by now numbers just over 270,000 imaginary friends, while the number of trapped hosts is typically in the 3000 to 5000 range. We occasionally see the list swell to 20,000 or more when high volume campaigns run with bad address lists fed to them. I am pretty sure it went over 100,000 at one point.

It's fun to watch, and it looks like a significant subset of the spamtraps have made it into the address lists of active spam operations. I frankly never thought I would still be collecting spam traps from logs all these years later. Yes, it all sounds a bit absurd, but it is effective for keeping our mailboxes largely spam free, even though it feels at times like running a weird found object-ish art project. Anyway, a summary of the lists we publish can be found in this article.

The brutes, the password gropers and the state tracking options

If you run an SSH service or really any kind of listening service with the option to log in, you will see some number of failed authentication attempts that generate noise in the logs. The password guessing, or as some of us say, password groping, turned out to be annoying enough that OpenBSD 3.6-current and later OpenBSD 3.7 introduced a set of features to use data that would anyway be available in the state table, to track the state of active connections, and to act on limits you define such as number of connections from a single host over a set number of seconds.

The action could be to add the source IP that tripped the limit to a table. Additional rules could then subject the members of that table to special treatment. Since that time, my internet-facing rule sets have tended to include variations on

table <bruteforce> persist
block quick from <bruteforce>
pass inet proto tcp from any to $localnet port $tcp_services \
        flags S/SA keep state \
	(max-src-conn 100, max-src-conn-rate 15/5, \
         overload <bruteforce> flush global)

which means that any host that tries more than 100 simultaneous connections or more than 15 new connections over 5 seconds are added to the table and blocked, with any existing connections terminated.

It is a good practice to let table entries in such setups expire eventually. At first I followed the spamd(8) defaults' example and set expiry at 24 hours, but with password gropers like those caught by this rule being what they are, I switched a few years ago to at four weeks at first, then upped again a few months later to six weeks. Groperbots tend to stay broken for that long. And since they target any service you may be running, state tracking options with overload tables can be useful in a lot of non-SSH contexts as well.

It is also worth noting that state tracking actions are useful for essentially all services. The article Forcing the password gropers through a smaller hole with OpenBSD's PF queues has a few suggestions on how to handle noise sources with various other services.

One final point I would like to make about the state tracking and actions is that much like the greytrapping feature of spamd, this feature gives you the tools to build a configuration that adapts to network conditions and learns from the traffic it sees. 

While this does not rise to the level of being an actual Artificial Intelligence or AI, this has enough buzzwordability potential that I remain to this day extremely puzzled that none of the other big names at least imitated those features in their own products and marketed for all it would be worth. 

I certainly know what I would have done in their position. But then I am more engineer than marketer and in the contexts where I call the shots, the best option is just to keep running OpenBSD.

NAT's guts ripped out

When the OpenBSD 4.7 release cycle came around, Henning Brauer had been hard at work for a while maintaining a diff of several thousand lines -- which when applied actually shrunk the code -- that contained a total rewrite of the IPv4 network address translation code.

Previous PF versions had 'nat on interface' and 'rdr on interface' rules, while the new code introduced nat-to and rdr-to as options on pass or match rules.

The match keyword had been introduced in the previous release to act on packets and connections without affecting pass or block state, such as applying specific options or adding tags for processing later in the rule set. Now with the major NAT rewrite in place, it became even clearer why match was in fact a useful keyword and feature.

The NAT rewrite added a lot of flexibility to how you can write PF rule sets, and of course for my own part that rewrite made it necessary to write the second edition of The Book of PF, timed to hit bookshelves as close as possible to the OpenBSD 4.7 release. And yes, the rewrite improved the performance too.

We went to modern queueing

OpenBSD has had traffic shaping available in the ALTQ subsystem since the very early days. ALTQ was rolled into PF at some point, but the code was still marked experimental 15 years after it was written, and most people who tried to use it in anger at the time found the syntax inelegant at best, infuriating or worse at most times.

So Henning Brauer took a keen interest in the problem, and reached the conclusion that all the various traffic shaping algorithms were not in fact needed. They could all except one be reduced to mere configuration options, either as setting priorities on pass or match rules or as variations of the theme of the mother algorithm Hierarchical Fair Service Curve (HFSC for short).

Soon after, another not-small diff was making the rounds. The patch was applied early in the OpenBSD 5.5 cycle, and for the lifetime of that release older ALTQ setups were possible side by side with the new queueing system.

The feedback I get is that the saner syntax in the new queueing system lead to more users taking up traffic shaping. Here is the queue setup that I came up with for one of my sites:

queue rootq on $ext_if bandwidth 20M
        queue main parent rootq bandwidth 20479K min 1M \
                                    max 20479K qlimit 100
             queue qdef parent main bandwidth 9600K min 6000K \ 
                                    max 18M default
             queue qweb parent main bandwidth 9600K min 6000K \
                                    max 18M
             queue qpri parent main bandwidth 700K min 100K \
                                    max 1200K
             queue qdns parent main bandwidth 200K min 12K \
                                    burst 600K for 3000ms
        queue spamd parent rootq bandwidth 1K min 0K max 1K \
                                    qlimit 300

while tying the queues into the subsequent rules with a set of match rules just following that block.

This is what triggered the need to write the third edition of The Book of PF. The book includes descriptions of both the new and the old system as well as tips on how to make a smooth transition. The ALTQ code was removed from OpenBSD during the OpenBSD 5.6 cycle, but continues to live on in some form in FreeBSD and NetBSD.

And yes, if you think my queues setup punishes spammers a bit more in addtion to being subjected to spamd(8), you're right.

pflow(4) offers network insights lite

Everybody who has been tasked with looking after a network has at some point been at least a little curious about what actually moves around there. At times we will see situations where it is essential for troubleshooting purposes to see the traffic flows with data about endpoints, packets and bytes transferred, protocol and so forth.

If you do not need to see the data itself, but rather the metadata, the NetFlow standard and its close cousin IPFIX offers just that. Netflow tools existed as packages on OpenBSD already, but from OpenBSD 4.5 PF has the pflow state tracking option, paired with the pflow(4) virtual network interface which together offer a full netflow sensor package.

Set up one or more pflow interfaces to send data to one or more collectors, and add the pflow option to specific rules or as a state default and you have started your collecting. You can even have metadata for traffic matching specific rules going to separate pflow devices and collectors.

My field notes in Yes, You Too Can Be An Evil Network Overlord - On The Cheap With OpenBSD, pflow And nfsen offers some practical examples and insights, including how we used a pflow setup to track down a noisy machine on a somewhat critical network as well as some pointers to valueable further reading.

LibreSSL, the great deobfuscation

People tell me they think that the reason LibreSSL was created was the Heartbleed bug, but no, actually not, just damn close.

The LibreSSL project was in fact started a few weeks before heartbleed became common knowledge. LibreSSL is the result of a group of OpenBSD developers taking the existing OpenSSL code and starting to fix it.

This time it was not a matter of a bad license. No, this was the result of the number of OpenBSD developers who took a look at the OpenSSL code that had been part of the OpenBSD base system since quite early on, and turned away in disgust and with symptoms of physical pain, reached a critical mass of sorts. I had heard OpenBSD developers complain about the absolute horror of the OpenSSL code for at least ten years. The code quality was just that bad.

What happened next was that a group of hardened OpenBSD developers grabbed the OpenSSL code and started two activities in parallel. One was looking in the OpenSSL request tracker for bugs that had not been addressed. The other was reformatting the OpenSSL code into something resembling the OpenBSD style of readable and maintainable C.

With the code in more readable form, discovering what it did became easier. In addition to a few obvious eye-stinging bugs the LibreSSL developers found a number of oddities, including, but not limited to

  • Code was never deleted even when it became irrelevant or obsolete

  • OpenSSL did not use the system memory allocation system, but rather opted for their own which never actually deallocated memory, but rather used LIFO recycling, and could easily be made to put private info into logs

  • all written in "OpenSSL C", which according to beck@ is a dialect of the "worst common denominator"

It is worth digging out the various articles and presentations made by LibreSSL developers over the years, with specific emphasis on Bob Beck's BSDCan talk on the first 30 days of LibreSSL (available on youtube), which is the original source of the term code flensing.

Since the OpenBSD 5.6 release in 2014, LibreSSL has been the default TLS library in OpenBSD. LibreSSL has been ported elsewhere based on the -portable variant.

For my own part I can only attest to not ever running into a TLS problem that was LibreSSL's fault. It probably still has bugs, but it is a lot more of a healthy choice than its predecessor.

This was my list of life improving OpenBSD events - I'd love to hear yours

As I warned earlier, this has been about my personal list of OpenBSD events that I remember fondly.

I am sure your list is at least a little different. I am sure there are things from the innovations page that I have simply forgotten about.

Each release comes with a detailed list of changes, such as this one for OpenBSD 6.9, and the page has pointers back to the equivalent pages for previous releases.

I would love to hear about your favorite OpenBSD moments.


More items for your OpenBSD reading

www.openbsd.org is the official OpenBSD web site. If you want to donate, go to the donations page and find the most appropriate option. Corporate entities may prefer to donate via The OpenBSD Foundation, which is a Canadian non-profit corporation.

undeadly.org is the OpenBSD Journal news site.

bsdly.blogspot.com My rant^H^H^H^Hblog posts

https://flak.tedunangst.com/ Ted Unangst (tedu@) on developments

Michael W Lucas: Absolute OpenBSD, 2nd edition

Peter N. M. Hansteen: The Book of PF, 3rd edition

Henning Brauer: OpenBSD sucks (… least)