Skip site navigation (1) Skip section navigation (2)


The fourth quarter of 2015 saw a great deal of activity for FreeBSD. This is now the third quarter running for which I can say that this is the largest report yet published! Many thanks to everyone who proactively submitted topics and entries — it is great to have more complete coverage of ongoing development for the community to learn about in these reports.

An experimental new Triage Team was formed this quarter to create a new way for community members to participate, and to improve issue management and productivity in general. Making more effective use of automation and tooling can help to increase developer productivity and the quality of FreeBSD, just as the adoption of Jenkins and continual integration tooling catches regressions quickly and maintains the high standards for the system.

Efforts to bring our BSD high standards to new architectures continue, with impressive work on arm64 leading to its promotion to Tier-2 status and a flurry of work bringing up the new RISC-V hardware architecture. Software architecture is also under active development, including system startup and service management. A handful of potential init system replacements are mentioned in this report: launchd, relaunchd, and nosh. Architectural changes originating both from academic research (multipath TCP) and from the realities of industry (sendfile(2) improvements) are also under way. It is heartening to see how FreeBSD provides a welcoming platform for contributions from both research and industry.

To all the readers, whether from academia or industry, hobbyist or professional: I hope you are as excited as I am to read about all of the progress and projects covered in this report, and the future of FreeBSD!

—Ben Kaduk

The deadline for submissions covering the period from January to March 2016 is April 7, 2016.

FreeBSD Team Reports




Userland Programs




    FreeBSD Team Reports

    FreeBSD Release Engineering Team

    FreeBSD 10.3-RELEASE schedule URL:
    FreeBSD Development Snapshots URL:

    Contact: FreeBSD Release Engineering Team <>

    The FreeBSD Release Engineering Team is responsible for setting and publishing release schedules for official project releases of FreeBSD, announcing code freezes, and maintaining the respective branches, among other things.

    During the last quarter of 2015, the Release Engineering team added support for three additional FreeBSD/arm systems: BANANAPI, CUBIEBOARD, and CUBIEBOARD2.

    In addition to regular development snapshot builds for FreeBSD 11.0-CURRENT and FreeBSD 10.2-STABLE, several changes and enhancements were made to the release build code. Of note, the release build code no longer produces MD5 checksums, in favor of SHA512.

    Toward the end of the year, focus was primarily centered on the upcoming FreeBSD 10.3 release cycle, which will begin in January 2016.

    As always, help testing development snapshot builds is crucial to producing quality releases, and we encourage testing development snapshots whenever possible.

    This project was sponsored by The FreeBSD Foundation.

    Issue Tracking (Bugzilla)

    Bugzilla Home Page URL:

    Contact: Bugmeisters <>
    Contact: Kubilay Kocak <>
    Contact: Mahdi Mokhtari <>

    The bugmeister team has gained a new member, Mahdi Mokhtari ( Mahdi has been contributing to the FreeBSD Project for just over one month. After getting started by creating ports for Chef-Server and MySQL 5.7 (with Bernard Spil's help), an introduction to Kubilay Kocak led to guidance on appropriate projects, such as Bugzilla development to help Bugmeister, the Bugzilla Triage team, Developers, and the community by making issue tracking better. This is how things are going so far:

    Issue Tracking can be either "Defect Tracking for Systems" or "Bug-Tracking for Systems". System Defect Tracking is to allow individual or groups of developers to keep track of outstanding issues in their product effectively. We use Bugzilla to manage issues for the FreeBSD project.

    We are pleased to announce some developments on our issue management systems:

    • We have made improvements to the AutoAssigner module (not yet deployed) that was previously developed by Marcus von Appen to assign port bugs to their maintainers by default, such as:
      • Improvements and bugfixes to port detection in the Summary: field of issues, for automatic assignment to their maintainers in a better way.
      • Refactoring code to make future development easier and faster in a more modular way.
    • We have developed a new module (FBSDAttachment), which automates setting maintainer-approval flag values on attachments under most conditions. This will improve time to resolution, consistency of triage, and reduce manual effort by triagers and maintainers.
    • We reported and upstreamed a number of bugs in Bugzilla, working with the upstream Bugzilla developers.

    Open tasks:

    1. Major improvements to templates for usability and simplicity.

    2. Further improvements to automation (for example, additional processing of commit logs).

    The FreeBSD Core Team

    Contact: FreeBSD Core Team <>

    Two major issues have occupied much of core's attention during the last quarter: the reorganisation of the Security Team and the question of whether to import GPLv3 licensed code into the source repository.

    1. The idea of reorganizing the Security team was first proposed to Core during a meeting at BSDCan this year by Gleb Smirnoff — core member and newly-appointed deputy Security Officer (SO). The "Security Team", which previously could contain several people (a varying number over time, but more than two) has been refashioned into just two roles: Security Officer and Deputy Security Officer. Accordingly, the role of the SO team has been redefined to be the controller of the distribution of security sensitive information into and within the project: they are responsible for interfacing with external bodies and individuals reporting security problems to the project, and connecting those reports to the appropriate individuals within the project with the technical expertise to address the identified concerns. These changes will improve the project's responsiveness to security alerts, help maintain security on privileged information received in confidence before general publication and, not least, reduce the work load on the security officer. The SO team will continue to benefit from liasons with the Core, Cluster Administration, and Release Engineering teams, and will be assisted by a secretary; they will also be able to obtain input and assistance in drafting security advisories from former and potential future (Deputy) Security Officers.

      Core would particularly like to thank the former members of the Security Team group for their past contributions, now that the Security Team role has been merged into the Security Officer's responsibilities.

    2. The other large question concerning Core is how to provide a modern toolchain for all supported achitectures. Tier 1 architectures are required to ship with a toolchain unencumbered by onerous license terms. This is currently provided for i386 and amd64 by the LLVM suite, including the Clang compiler, LLD and LLDB. However LLVM support for other (Tier 2 or below) architectures is not yet of sufficient quality to be viable, and the older but pre-existing GPLv2 toolchain cannot support some of the interesting new architectures such as arm64 and RISC-V. Pragmatically, in order for the project to support these architectures, until LLVM support arrives we must turn to the GNU project's GPLv3 licenced toolchain.

      The argument here is whether to import GPLv3 licensed code into the FreeBSD src repository with all of the obligations on patent terms and source code redistribution that would entail, not only for the FreeBSD project itself but for numerous downstream consumers of FreeBSD code. Not having a toolchain readily available is a big impediment to working on a new architecture.

      One potential solution is to create a range of "GPLv3 toolchain" base-system packages out of a completely separate source code repository, for instance within the FreeBSD area on Github. These would be distributed equivalently to the other base system binary packages when that mechanism is introduced.

      Core recognises that this is a decision with wide-ranging consequences and will be producing a position paper for circulation amongst all interested parties in order to judge community opinion on the matter. Core welcomes feedback from all interested parties on the subject.

    Beyond these two big questions, Core has handled a number of other items:

    • Core approved the formation of a wiki-admin team to take over managing the Wiki, to curate the Wiki content and work on navigation and organization of existing technical content and to evaluate new Wiki software with the aim of opening up the Wiki to contributions from the public.
    • An external review board has been assembled to look at the Code of Conduct, including a mixture of project members and experts from external groups. The review process is getting under way and Core is awaiting their report.
    • The standard documentation license was found to be unfit for its purpose, and the doceng group had temporarily reverted to the previous license while a new replacement was drafted. This new license is now the default for new documentation submissions. However, one factor emerging from this review was the difficulty of maintaining correct authorial attributions for sections of documentation, some of which may only be a few words long. Unlike source code, blocks of documentation are frequently moved around within individual files, or even between files. Consequently, Core would like to introduce a Voluntary Contribution Agreement along the lines of the one operated by the Apache Foundation. With this, copyrights are signed over to the FreeBSD Foundation, with individual contributions being recognised by recording names in a general "Authors" file. This will be another alternative alongside the existing copyright mechanisms used in the project. Core is interested to hear any opinions on the subject.
    • Core approved the formation of a new "dev-announce" mailing list, which all FreeBSD committers should be members of. This will be a low-traffic moderated list to contain important announcements, heads-ups, warnings of code freezes, changes in policy and notifications of events that affect the project as a whole.
    • Around eight years ago, an attempt was made to import the OpenBSD sensors framework. This was rejected at the time as potentially blocking the development of a better designed framework. However, no such development has occurred in the intervening time whilst the sensors framework has been in use successfully by both OpenBSD and FreeNAS. Despite some concerns about the efficiency of the framework and potential impacts on power consumption and hence battery lifetime, core is minded to approve the import, but wants to consult with interested developers first.
    • Core is exploring the legal ramifications for the project of the "Right to Be Forgotten" established by the European Court of Justice.
    • Core is also seeking an alternative means for holding their regular monthly conference calls. The current, paid-for, service has less than satisfactory sound quality and reliability, and Core would like to switch to a free video conferencing solution.

    This quarter also saw a particularly large influx of new commit bit requests, with on occasion, four votes running simultaneously. Please welcome Kurt Lidl, Svatopluk Kraus, Michal Meloun, Jonathan Looney (Juniper), Daisuke Aoyama, Phil Shafer (Juniper), Ravi Pokala (Panasas), Anish Gupta and Mark Bloch (Mellanox) to the ranks of src committers. In addition, core was delighted to restore commit privileges for Eric Melville after a hiatus of many years.

    No commit bits were taken in during the quarter. A non-committer account was approved for Kevin Bowling of LimeLight Networks. Kevin will be doing systems administration work with clusteradm, with particular interest in the parts of the cluster that are now hosted in LLNW's facilities. Deb Goodkin of the FreeBSD Foundation was added to the developers mailing list: she was one of the few members of the Foundation Board not already on the list, and having awareness of what is going on in the developer community will help her to support the project more effectively.

    The FreeBSD Issue Triage Team

    Contact: Bugmeister <>
    Contact: Kubilay Kocak <>
    Contact: Vladimir Krstulja <>
    Contact: Rodrigo N. Hernandez <>

    By the end of the Q4 2015 period, Kubilay Kocak (koobs@) started an initiative to form an experimental Bugzilla Triage Team. The main goals of the team are to increase community involvement (addition/training of new triagers) and enhance current procedures and tools, among others. This experiment was started with the participation of Vladimir (blackflow on irc/freenode) and Rodrigo (DanDare on irc/freenode), who approached koobs@ with a desire to contribute and get more involved with the FreeBSD Project. This experimental pilot project has the task of setting up procedures for enhanced Issue (Problem Report) management that include better classification and prioritization, eventually leading to faster resolution of issues.

    We are now happy to report on the progress of this experimental team:

    • The #FreeBSD-bugs IRC channel has been set up on Freenode and we are successfully using it to exchange information about triage processes, ask for help, propose changes and discuss related topics.
    • We have identified the primary role of an Issue Triage Team to be that of classification of problem reports of all kinds (currently limited mostly to ports and obvious src issues) and facilitation of issue assignment, which is making sure that the reported issues are explained well, contain all the appropriate information (or as much of it as possible), and are brought to attention of the people who can act upon them.
    • Vladimir and Rodrigo are successfully training in bug triage as well as porting processes (Vladimir is also taking maintainership of some ports).
    • This experiment is benefiting from the introduction of newcomers to issue tracking. It naturally resulted in a entire review of the tracking process from its very elementary aspects. This "fresh eyes" participation spotted minor details during the process, giving the opportunity to scrutinize actual procedures on a number of smaller points, followed by proposals on how to improve the overall Issue Tracking and Management. The new ideas include both organizational and technical ideas and solutions, such as new or modified keywords or flags for better classification, the triage workflow, and Bugzilla technical improvements, among others.
    • An important goal is producing documentation about best practices for using Bugzilla and issue management workflow. This documentation should be aimed not only at people directly engaged in issue triage tasks, but also at general users. Another relevant point is that feedback from the triage team can be used to improve Bugzilla in terms of adjusting existing features to best fit FreeBSD's needs, and the development of new features (please see Mahdi "Magic" Mokhtari's report on "Bugzilla improvements").
    • We are still collating ideas in preparation of setting up a Wiki namespace for the overall topic of issue management, containing information for all the parties involved in issue tracking: from users (reporters) to maintainers and committers. The unorganized brainstorming document is linked in this report.

    Since the Issue Triage Team is very young, we expect more information be available and more actions to be reported in the next status report.

    Open tasks:

    1. Set up the Wiki namespace and organize the brainstorming document into a meaningful set of documents.

    2. We are actively recruiting to grow our FreeBSD Triage Team. If you are interested in participating and contributing to one of the most important community-facing areas of the FreeBSD project, join #freebsd-bugs on the freenode IRC and let us know!

      Experience with issue tracking is desirable, but not required. No prior internal project knowledge or technical skills are required, just bring your communication skills and awesome attitude. Training is provided.


    CAM I/O Scheduler

    BSDCan Paper URL:
    Phabricator Review URL:

    Contact: Warner Losh <>

    Reviews have begun on the CAM I/O scheduler that I wrote for Netflix. It is anticipated that this process will be done in time for the FreeBSD 11 branch.

    Details about this work can be found in the linked BSDcan paper from last year.

    Briefly, the scheduler allows one to differentiate I/O types and limit I/O based on the type and characteristics of the I/Os (including the latency of recent requests relative to historical averages). This is most useful when tuning system loads to SSD performance. Both a simple default scheduler, the same that we use today in FreeBSD, as well as a scheduler that can be well-tuned for system loads related to video streaming will be included.

    This project was sponsored by Netflix, Inc.

    Encrypted Kernel Crash Dumps

    Technical Details URL:
    Patch Review URL:

    Contact: Konrad Witaszczyk <>

    Kernel crash dumps contain information about currently running processes. This can include sensitive data, for example passwords kept in memory by a browser when a kernel panic occurred. An entity that can read data from a dump device or a crash directory can also extract this information from a core dump. To prevent this situation, the core dump should be encrypted before it is stored on the dump device.

    This project allows a kernel to encrypt a core dump during a panic. A user can configure the kernel for encrypted dumps and save the core dump after reboot using the existing tools, dumpon(8) and savecore(8). A new tool decryptcore(8) was added to decrypt the core files.

    A patch has been uploaded to Phabricator for review. The patch is currently being updated to address the review comments, and should be committed as soon as it is accepted. For more technical details, please visit the FreeBSD-security mailing list archive or see the Phabricator review.

    Jenkins Continuous Integration for FreeBSD

    The Jenkins CI Server in the FreeBSD Cluster URL:
    Portest Script URL:
    Jenkins Workflow Plugin URL:
    Cloudbees URL:
    Jenkins Phabricator Plugin URL:
    Phabricator Plugin Fixes URL:
    Durable Task Plugin Fixes URL:
    Clang Scanbuild Plugin Fixes URL:
    Multiple SCMs Plugin Fixes URL:
    SCM Sync Configuration Plugin Fixes URL:
    Porting Jobs to the Workflow Plugin URL:
    Akuma Fixes for FreeBSD URL:
    Kyua Fix for Invalid Characters URL:

    Contact: Craig Rodrigues <>
    Contact: Jenkins Administrators <>
    Contact: FreeBSD Testing <>

    The Jenkins Continuous Integration and Testing project has been helping to improve the quality of FreeBSD. Since the last status report, we have quickly found commits that caused build breakage or test failures. FreeBSD developers saw these problems and quickly fixed them. Some of the highlights include:

    • Ricky Gallagher wrote a script named portest, which can take a patch to the FreeBSD ports tree as input, and can generate a sequence of commands to check out the ports tree from Subversion, apply the patch, and then invoke poudriere to build the affected part of the ports tree. Ricky consulted with Torsten Zühlsdorff during its development. This script will be used later to test changes to the ports tree.
    • Craig Rodrigues converted some Jenkins builds to use the Workflow plugin. Workflow is a plugin written by Jesse Glick and other developers at Cloudbees, the main company providing commercial support for Jenkins. With this plugin, a Jenkins job can be written in a Domain Specific Language (DSL) which is written in the Groovy scripting language. Workflow scripts are meant to provide sophisticated access to Jenkins functionality, in a simple scripting language. As Jenkins jobs get more complicated and have more interdependencies, using a DSL is easier for maintainability instead of creating Jenkins jobs via menus.

      Craig Rodrigues worked with Jesse Glick to identify and fix a problem with the Durable Task plugin used by the workflow plugin. This problem seemed to show up mostly on non-Linux platforms such as OS X and FreeBSD.

    • Eitan Adler worked with Craig Rodrigues to test a Jenkins plugin written by Aiden Scandella at Uber which integrates Phabricator and Jenkins. With this plugin, if someone submits a code review with Phabricator's Differential tool, a Jenkins build with this code change will be triggered. The Phabricator code review would then be updated with the result of the build.

      Eitan Adler and Craig Rodrigues had some initial success testing this plugin using the FreeBSD docs repository, but this plugin still has a lot of hardcoded dependencies specific to Uber's environment which make it difficult to use out-of-the-box for FreeBSD. Alexander Yerenkow submitted some patches upstream to fix some of these problems, but this plugin still needs more work. Craig Rodrigues thinks that it might be better to write a workflow script to call Phabricator commands directly.

    • Craig Rodrigues pushed fixes upstream to several plugins including:

      • SCM Sync configuration plugin
      • NodeLabel parameter plugin
      • Subversion plugin
      • Multiple SCMs plugin
      • Clang Scanbuild plugin

      Craig Rodrigues was granted commit access to the SCM Sync configuration plugin, Multiple SCMs plugin, and Clang Scanbuild plugin.

    • Li-Wen Hsu set up multiple builds using jails on machines located at NYI and administered by the FreeBSD Cluster Administrators. One of these builds targets 64-bit ARM.
    • Michael Zhilin fixed the Akuma library for FreeBSD. The Akuma library is used by Jenkins to determine what command-line arguments were passed to a running process. To fix it, Michael invoked an FreeBSD-specific sysctl() with KERN_PROC_ARGS to determine the arguments for a running pid. This fix allows a running Jenkins instance to restart itself after new plugins are installed.
    • Julio Merino accepted a fix for Kyua from Craig Rodrigues to fix writing out XML characters to test report files.

    Open tasks:

    1. Work more on using the workflow plugin for various builds.

    2. Set up a build to test bmake's meta-mode.

    3. Finish off integration with Phabricator.

    4. People interested in helping out should join the list.

    Mellanox iSCSI Extensions for RDMA (iSER) Support

    GitHub repository URL:

    Contact: Max Gurtovoy <>
    Contact: Sagi Grimberg <>

    Building on the new in-kernel iSCSI initiator stack released in FreeBSD 10.0 and the recently added iSCSI offload interface, Mellanox Technologies has developed iSCSI extensions for RDMA (iSER) initiator support to enable efficient data movement using the hardware offload capabilities of Mellanox's 10, 40, 56, and 100 Gigabit Infiniband (IB)/Ethernet adapters.

    Remote Direct Memory Access (RDMA) has been shown to have great value for storage applications. RDMA infrastructure provides benefits such as zero-copy, CPU offload, reliable transport, fabric consolidation, and many more. The iSER protocol eliminates some of the bottlenecks in the traditional iSCSI/TCP stack, provides low latency and high throughput, and is well suited for latency aware workloads.

    This work includes a new ICL module that implements the iSER initiator. The iSCSI stack is slightly modified to support some extra features such as asynchronous IO completions, unmapped data buffers, and data-transfer offloads. The user will be able to choose iSER as the iSCSI transport with iscsictl.

    The project is in the process of being merged to FreeBSD 11-CURRENT and is expected to ship with FreeBSD 11.0.

    This project was sponsored by Mellanox Technologies.

    MIPS: Ralink/Mediatek Support

    Github Branch With Work in Progress URL:

    Contact: Stanislav Galabov <>

    This project is aimed at adding FreeBSD support for Ralink/Mediatek's family of WiFi router system-on-chip (SoC) devices based on MIPS processors. These SoCs are commonly found in embedded network devices such as WiFi routers. Having support for these SoCs would allow FreeBSD to run on a number of additional low-cost devices, which could help spread FreeBSD's popularity in the embedded systems world.

    The project currently aims to support the following Ralink/Mediatek chipsets: RT3050, RT3052, RT3350, RT3352, RT3662, RT3883, RT5350, RT6855, RT6856, MT7620, MT7621, MT7628 and MT7688. The following functionality (where applicable) is currently planned to be supported: Interrupt controller, UART, GPIO, USB, PCI/PCIe, Ethernet, and SPI.

    This project was sponsored by Smartcom - Bulgaria AD.

    Open tasks:

    1. Help with adding WiFi driver support (possibly to ral(4)) for the above SoCs would be greatly appreciated.

    2. Help with refactoring if_rt(4) to be usable on all of the above SoCs would be appreciated.

    3. Help wth testing target boards (e.g., WiFi routers) would be appreciated.

    Multipath TCP for FreeBSD

    MPTCP for FreeBSD Repository URL:
    MPTCP for FreeBSD Project Website URL:

    Contact: Nigel Williams <>

    Multipath TCP (MPTCP) is an extension to TCP that allows for the use of multiple network interfaces on a standard TCP session. The addition of new addresses and scheduling of data across these occurs transparently from the perspective of the TCP application.

    The goal of this project is to deliver an MPTCP kernel patch that interoperates with the reference MPTCP implementation, along with additional enhancements to aid network research.

    A v0.51 release has been tagged in our repository, with some minor improvements over v0.5.

    We have now removed much of the MPTCP code that was inside the functions tcp_do_segment, tcp_output, and other code used for standard TCP connections. The goal of this is to restrict the added MPTCP code to just MPTCP connections, leaving regular TCP connections using the existing code.

    We are currently in the process of implementing a subflow socket buffer upcall and event processing. These will handle changes in subflow socket state, MP-signalling, and incoming data segments.

    This also requires some re-working of the MP option processing, particularly how incoming DSN maps are parsed and stored for use during MP-layer reassembly.

    We are also looking at how our changes might take advantage of the new TCP stack modularisation enhancements to create subflow-specific TCP functions.

    This project was sponsored by The Cisco University Research Program Fund at Community Foundation Silicon Valley, and The FreeBSD Foundation.

    Open tasks:

    1. Complete the implementations of subflow event processing and new option parsing.

    2. Update documentation and task lists.


    OpenBSM: Open Source Basic Security Module (BSM) Audit Implementation URL:
    OpenBSM on GitHub URL:
    FreeBSD Audit Handbook Chapter URL:

    Contact: Christian Brueffer <>
    Contact: Robert Watson <>
    Contact: TrustedBSD audit mailing list <>

    OpenBSM is a BSD-licensed implementation of Sun's Basic Security Module (BSM) API and file format. It is the user-space side of the CAPP Audit implementations in FreeBSD and Mac OS X. Additionally, the audit trail processing tools are expected to work on Linux.

    Progress has been slow but steady this quarter, culminating in OpenBSM 1.2 alpha 4, the first release in three years. It features various bug fixes and documentation improvements; the complete list of changes is documented in the NEWS file on GitHub. The release was imported into FreeBSD head and merged to FreeBSD 10-STABLE. As such, it will be part of FreeBSD 10.3-RELEASE.

    Open tasks:

    1. Test the new release on different versions of FreeBSD, Mac OS X, and Linux. In particular, testing on Mac OS X 10.9 (Mavericks) and newer would be greatly appreciated.

    2. Fix problems that have been reported via GitHub and the FreeBSD bug tracker.

    3. Implement features mentioned in the TODO list on GitHub.

    Raspberry Pi: VideoCore Userland Application Packaging

    Contact: Mikaël Urankar <>
    Contact: Oleksandr Tymoshenko <>

    The Raspberry Pi SoC consists of two parts: ARM and GPU (VideoCore). Many interesting features like OpenGL, video playback, and HDMI controls are implemented on the VideoCore side and can be accessed from the OS through libraries provided by Broadcom (userland repo). These libraries were ported to FreeBSD some time ago, so Mikaël created the port misc/raspberrypi-userland for them. He also created a port for omxplayer (a low-level video player that utilizes VideoCore APIs) and is working on a port for Kodi (formerly XBMC), a more user-firendly media player software with Raspberry Pi support.

    RCTL Disk IO Limits

    Contact: Edward Tomasz Napierala <>

    An important missing piece of the RCTL resource limits mechanism was the ability to limit disk throughput. This project aims to fill that hole by making it possible to add RCTL rules for read bytes per second (BPS), write BPS, read I/O operations per second (IOPS), and write IOPS. It also adds a new throttling mechanism to delay process execution when a limit is reached.

    The project is at the late implementation stage. The major piece of work left apart from testing is to integrate it with ZFS. The project is expected to ship with FreeBSD 11.0.

    This project was sponsored by The FreeBSD Foundation.

    Root Remount

    Commit to Head URL:
    reboot(8) Manual Page Changes URL:

    Contact: Edward Tomasz Napierala <>

    One of the long-missing features of FreeBSD was the ability to boot up with a temporary rootfs, configure the kernel to be able to access the real rootfs, and then replace the temporary root with the real one. In Linux, this functionality is known as pivot_root. The reroot projects provides similar functionality in a different, slightly more user-friendly way: rerooting. Simply put, from the user point of view it looks like the system performs a partial shutdown, killing all processes and unmounting the rootfs, and then partial bringup, mounting the new rootfs, running init, and running the startup scripts as usual.

    The project is finished. All the relevant code has been committed to FreeBSD 11-CURRENT and is expected to ship with FreeBSD 11.0.

    This project was sponsored by The FreeBSD Foundation.

    Routing Stack Update

    Initial Proposal URL:

    Contact: Alexander Chernikov <>

    The projects/routing Subversion branch is a FreeBSD routing system rework aimed at providing performance, scalability and the ability to add advanced features to the routing stack.

    The current packet output path suffers from excessive locking. Acquiring and releasing four distinct contested locks is required to convert a packet to a frame suitable to put on the wire. The first project goal is to reduce the number of locks needed to just two rmlock(9)s for the output path, which permits close-to-linear scaling.

    Since September, one of the locks (used to protect link-level entries) has been completely eliminated from the packet data path. A new routing API was introduced, featuring better scalability and hiding routing internals. Most of the consumers of the old routing API were converted to use the new API.

    The Graphics Stack on FreeBSD

    Graphics Stack Roadmap and Supported Hardware Matrix URL:
    Ports Development Tree on GitHub URL:

    Contact: FreeBSD Graphics team <>

    Several important ports were updated: Mesa to 11.0.8, the X.Org server to 1.17.4, libdrm to 2.4.65, as well as many applications and libraries. The latest release of the X.Org server, 1.18, is being tested in our Ports development tree.

    On the kernel side, the i915 update is almost ready to land. There are a couple known regressions for currently supported GPUs that we want to fix before committing.

    We started a discussion on the FreeBSD-x11@ mailing list to organize future contributions to the kernel drivers. We have already received some valuable comments. We are confident that future updates will happen at a faster pace, thanks to several motivated people!

    FOSDEM is held in Brussels on the 30th and 31st of January. We will attend this conference. It will be a perfect time to see people again from FreeBSD and from the XDC. On Sunday, we will give a talk about how to contribute to the Graphics Stack.

    Our blog is currently down because the service was discontinued. We hope to get a dump of our data to put it back online elsewhere. Unfortunately, there is no ETA for this item.

    Open tasks:

    1. See the "Graphics" wiki page for up-to-date information.

    The nosh Project

    Introduction URL:
    FreeBSD binary packages URL:
    Installation How-To URL:
    Roadmap URL:
    Commands URL:
    A Slightly Outdated User Guide URL:
    The Supervision Mailing List URL:

    Contact: Jonathan de Boyne Pollard <J.deBoynePollard-newsgroups@NTLWorld.COM>

    The nosh project is a suite of system-level utilities for initializing, running, and shutting down BSD systems, and for managing daemons, terminals, and logging. It supersedes BSD init and the NetBSD rc.d system, drawing inspiration from Solaris SMF for named milestones, daemontools-encore for service control/status mechanisms, UCSPI, and IBM AIX for separated service and system management. It comprises a range of compatibility mechanisms, including shims for familiar commands from other systems, and an automatic import mechanism that takes existing configuration data from /etc/fstab, /etc/rc.conf{,.local}, /etc/ttys, and elsewhere, applying them to its native service definitions and creating additional native services. It is portable (including to Linux) and composable, it provides a migration path from the world of systemd Linux, and it does not require new kernel APIs. It provides clean service environments, orderings and dependencies between services, parallelized startup and shutdown (including fsck), strictly size-capped and autorotated logging, the service manager as a "subreaper", and uses kevent(2) for event-driven parallelism.

    Since the last status report, in October 2015, the project has seen: the complete replacement of its event-handling subsystem on Linux; the introduction of tools for exporting cyclog/multilog logs via RFC 5426 to remote log handlers (such as logstash); and the switching of the user-mode virtual terminal subsystem on BSD to using USB devices directly, a more powerful device interface than sysmouse et al. because it permits directly positioning touch devices for mice and other things (thus permitting "mouse integration" under VirtualBox for those who run PC-BSD/FreeBSD on VirtualBox virtual machines), but sysmouse et al. can still be used if desired.

    In version 1.24, released shortly before publication of this report, there are extensive additions for supporting a purely-ZFS system with an empty /etc/fstab (as the PC-BSD 10.2 system installer creates), and the ability to convert systemd unit files' process priority settings to BSD's rtprio/idprio.

    Version 1.24 also sees a large chunk taken out of the remainder of the on-going project to create enough native service bundles and ancillary utilities to entirely supplant the rc.d system. The progress of this project has been open from the start, and can be followed on the nosh roadmap web page. As of version 1.24, there are a mere 27 items remaining out of the original target list of 157, with a 28th and a 29th (from PC-BSD 10.2) added. Items crossed off by version 1.24 include (amongst others) mfs support for /tmp, static ARP and networking, persistent entropy for the randomness subsystem, pefs, and hald.

    The remaining items in the task list are mostly aimed at making the overall system integration cleaner and friendlier to modern systems. We are also interested in receiving suggestions, bug reports, and other feedback from users. Try following the how-to guide and see how things go!

    Open tasks:

    1. Add kernel support for passing a -b option to PID 1, and support for a boot_bare variable in the loader, to allow "emergency" (where no shell dotfiles are loaded) and "rescue" mode bootstraps, akin to Linux. (History: the -b mechanism and idea date back to version 2.57d of Miquel van Smoorenburg's System 5 init clone, dated 1995-12-03, and was already known as "emergency boot" by 1997.)

    2. Add support to FreeBSD's fsck(8) for outputting machine-readable progress reports to a designated file descriptor, so that nosh can provide progress bars for multiple fscks running in parallel. nosh already provides this functionality on Linux, where fsck(8) does provide machine-readable output.

    3. Identify when the configuration import system needs to be triggered, such as when bsdconfig alters configuration files, and create the necessary hooks to import external configuration changes into nosh.

    UEFI Boot and Framebuffer Support

    Contact: Ed Maste <>

    A number of UEFI bug fixes were committed over the last quarter, further improving compatibility with different UEFI implementations. Specifically: on some implementations, FreeBSD failed to boot with an "ExitBootServices() returned 0x8000000000000002" error. This has been fixed with a retry loop (as required by UEFI) in r292515 and r292338.

    UEFI improvements from other developers have recently been committed or are in progress. These include support for environment variables set on the EFI loader command line, improved text console mode setting, support for nvram variables, and root-on-ZFS support.

    This project was sponsored by The FreeBSD Foundation.

    Open tasks:

    1. Test FreeBSD-CURRENT snapshots on a variety of UEFI implementations.

    2. Merge UEFI changes to stable/10 for FreeBSD 10.3-RELEASE.


    Chelsio iSCSI Offload Driver (Initiator and Target)

    Commit Adding Hardware Acceleration Support URL:

    Contact: Navdeep Parhar <>

    A new driver, cxgbei, enabling hardware accelerated iSCSI with Chelsio's T5- and T4-based offload-capable cards, has been committed to head. Both Initiator and Target are supported. The wire traffic is standard iSCSI (SCSI over TCP as per RFC 3720, etc.) so an Initiator/Target using this driver will interoperate with all other standards-compliant implementations.

    Hardware assistance provided by the T5 and T4 ASICs includes:

    • Complete TCP processing.
    • iSCSI PDU identification and extraction from the byte oriented TCP stream.
    • Header and/or data digest generation and verification.
    • Zero copy support for both transmit and receive.

    This project was sponsored by Chelsio Communications.

    Open tasks:

    1. The cxgbei(4) man page is missing but will be committed shortly.

    2. The driver is in advanced stage QA and will see some bugfixes and performance enhancements in the very near future. MFC is possible as soon as the QA cycle completes.

    FreeBSD Integration Services (BIS)

    FreeBSD Virtual Machines on Microsoft Hyper-V URL:
    Linux and FreeBSD Virtual Machines on Hyper-V URL:

    Contact: Dexuan Cui <>
    Contact: Hongjiang Zhang <>

    When FreeBSD virtual machines (VMs) run on Hyper-V, using Hyper-V synthetic devices is recommended to get the best network and storage performance and make full use of all the benefits that Hyper-V provides. The collection of drivers that are required to run Hyper-V synthetic devices in FreeBSD are known as FreeBSD Integration Services (BIS). Some of the BIS drivers (like network and storage drivers) have existed in FreeBSD 9.x and 10.x for years, but there are still some performance and stability issues and bugs. Compared with Windows and Linux VMs, the current BIS lacks some important features, such as virtual Receive Side Scaling (vRSS) support in the Hyper-V network driver and support for UEFI VM (boot from UEFI), among others.

    We are now working more on the issues and performance tuning to make FreeBSD VMs run better on Hyper-V and the Hyper-V based cloud platform Azure.

    Our work during 2015Q4 is documented below:

    • Optimizing the VMBus driver and Hyper-V network driver for performance:
      • Sent out patches to enable INTR_MPSAFE for the interrupt handling thread, speed up relid-to-channel lookup in the thread by map table, and optimize the VMBus ringbuffer writable notification to the host.
      • Developing a patch to enable the virtual Receive Side Scaling (vRSS) for Hyper-V network device driver. This will greatly improve the network performance for SMP virtual machine (VM).
      • Sent out a patch to enable the Hyper-V timer, which will improve the accuracy of timekeeping when FreeBSD VMs run on Hyper-V.
    • Fixing bugs and cleaning up the code:
      • Fixed a bug in checksum offloading (PR 203630 — [Hyper-V] [nat] [tcp] 10.2 NAT bug in TCP stack or hyperv netsvc driver) in the Hyper-V network driver, making FreeBSD VM based NAT gateways work more reliably.
      • Fixed a serialization issue in the initialization of VMBus devices, fixing PR 205156 ([Hyper-V] NICs' (hn0, hn1) MAC addresses can appear in an uncertain way across reboot).
      • Fixed a KVP (Key-Value Pair) issue (retrieving a key's value can hang for an uncertain period of time).
      • Added ioctl support for SIOCGIFMEDIA for the Hyper-V network driver, fixing PR 187006 ([Hyper-V] dynamic address (DHCP) obtaining does not work on HYPER-V OS 2012 R2).
      • Sent out patches to add an interrupt counter for Hyper-V VMBus interrupts (so the user can easily get statistical information about VMBus interrupts), and fix the KVP daemon's poll timeout (so the daemon will avoid unnecessary polling every 100 milliseconds.
      • Identified a TSC calibration issue: the i8254 PIT timer emulation of Hyper-V is not fully reliable, so the Hyper-V time counter should be used to calibrate the TSC. A patch was drafted. With the patch, it looks the warning kernel message (e.g., "calcru: runtime went backwards from 46204978 usec to 23362331 usec for pid 0 (kernel)") will go away, and the time-based tracing of Dtrace will be more accurate.
    • We plan to add support for UEFI VMs (Hyper-V Generation-2 VMs). Currently some issues and to-do items were identified. For example, we cannot use the i8254 PIT to calibrate the TSC because the i8254 PIT does not exist in a UEFI VM, and we need to add support for the Hyper-V synthetic keyboard/mouse/framebuffer device.
    • We are working on a disk detection issue: when a FreeBSD VM runs on a Windows Server 2016 Technical Preview host, the VM will detect 16 disks when only one disk is configured for the VM. VMs running on these hosts can fail to boot. A workaround patch was created and we are trying to make a formal fix.
    • We are tidying up some internal BIS test cases and plan to publish them on github.

    This project was sponsored by Microsoft.

    FreeBSD Xen

    FreeBSD PVH DomU Wiki Page URL:
    FreeBSD PVH Dom0 Wiki Page URL:
    FreeBSD/Xen HVMlite Implementation URL:;a=shortlog;h=refs/heads/new_entry_point_v5

    Contact: Roger Pau Monné <>
    Contact: Wei Liu <>

    Xen is a hypervisor using a microkernel design, providing services that allow multiple computer operating systems to execute on the same computer hardware concurrently. Xen support for FreeBSD on x86 as a guest was introduced in version 8, and ARM support is currently being worked on. Support for running FreeBSD as an amd64 Xen host (Dom0) is available in head.

    The x86 work done during this quarter has been focused on rewriting the PVH implementation inside of Xen, into what is now being called HVMlite to differentiate it with the previous PVH implementation. The Xen side of patches have already been committed to the Xen source tree, and will be available in Xen 4.7, the next version. Work has also begun on implementing HVMlite Dom0 support, although no patches have yet been published.

    HVMlite support for FreeBSD has not yet been committed, although an initial implementation is available in a personal git repository. The plan is to completely replace PVH with HVMlite on FreeBSD as soon as HVMlite supports Dom0 mode.

    Apart from this, Wei Liu is working on improving netfront performance on FreeBSD. Initial patches have been posted to the FreeBSD review system.

    The x86 unmapped bounce buffer code has also been improved, and unmapped IO support has been added to the blkfront driver.

    This project was sponsored by Citrix Systems R&D.

    Open tasks:

    1. Finish HVMlite Dom0 support inside of Xen.

    2. Deprecate and remove PVH support from Xen.

    3. Remove PVH support from FreeBSD and switch to HVMlite.

    4. Generalize the event channel code so it can be used on ARM.

    5. Improve the performance of the various backends (netback, blkback).

    Improvements to the QLogic HBA Driver

    Contact: Alexander Motin <>

    The QLogic HBA driver, isp(4), received a substantial set of changes. The primary goal was to make the Fibre Channel target role work well with CTL, but many other things were also fixed/improved:

    • Added support for modern 16Gbps 26xx FC cards.
    • The firmware in ispfw(4) were updated to the latest versions.
    • Target role support was fixed and tested for all FC cards from ancient 1Gbps 22xx to modern 16Gbps 26xx.
    • Port database handling was unified for target and initiator roles, allowing an HBA port to play both roles at the same time.
    • The maximal number of ports was increased from 256 to 1024.
    • Multi-ID (NPIV) functionality was fixed/implemented, allowing 24xx and above cards to provide up to 255 virtual FC ports per physical port.
    • Added support for 8-byte LUNs for 24xx and above cards.

    The code is committed to FreeBSD head and stable/10 branches.

    This project was sponsored by iXsystems, Inc..

    Open tasks:

    1. NVRAM data reading is hackish and requires rework.

    2. FCoE support for 26xx cards was not tested yet.

    iMX.6 Video Output Support

    Commit Adding Basic Video Support URL:

    Contact: Oleksandr Tymoshenko <>

    iMX.6 is a family of SoC used in multiple hobbyist ARM boards such as the Hummingboard, RIoTboard, and Cubox. Most of these products have HDMI output, but until recently, FreeBSD did not benefit from it. As of r292574, there is basic video output support so you can use the console on iMX6-based boards and probably run Xorg (not yet tested).

    Due to the lack of some kernel functionality (see open tasks), the only supported mode is 1024x768.

    Open tasks:

    1. Proper pixel clock initialization (relies on a clock framework).

    2. More flexible video output path (support multiple IPUs and DIs).

    ioat(4) Driver Enhancements

    Wikipedia on I/OAT URL:
    Last quarter's ioat(4) report URL:

    Contact: Conrad Meyer <>

    I/OAT DMA engines are bulk memory operation offload engines built into some Intel Server/Storage platform CPUs.

    Several enhancements were made to the driver. It now avoids memory allocation in locked paths, which should avoid deadlocking in memory pressure scenarios. Support for Broadwell-EP devices has been added. The "blockfill" operation and a non-contiguous 8 KB copy operation have been added to the API. The driver can recover from various programming errors by resetting the hardware.

    This project was sponsored by EMC / Isilon Storage Division.

    Open tasks:

    1. XOR and other advanced ("RAID") operation support.

    Kernel Vnode Cache Tuning

    MFC to stable/10 URL:

    Contact: Kirk McKusick <>
    Contact: Bruce Evans <>
    Contact: Konstantin Belousov <>
    Contact: Peter Holm <>
    Contact: Mateusz Guzik <>

    This completed project includes changes to better manage the vnode freelist and to streamline the allocation and freeing of vnodes.

    Vnode cache recycling was reworked to meet free and unused vnode targets. Free vnodes are rarely completely free; rather, they are just ones that are cheap to recycle. Usually they are for files which have been stat'd but not read; these usually have inode and namecache data attached to them. The free vnode target is the preferred minimum size of a sub-cache consisting mostly of such files. The system balances the size of this sub-cache with its complement to try to prevent either from thrashing while the other is relatively inactive. The targets express a preference for the best balance.

    "Above" this target there are 2 further targets (watermarks) related to the recyling of free vnodes. In the best-operating case, the cache is exactly full, the free list has size between vlowat and vhiwat above the free target, and recycling from the free list and normal use maintains this state. Sometimes the free list is below vlowat or even empty, but this state is even better for immediate use, provided the cache is not full. Otherwise, vnlru_proc() runs to reclaim enough vnodes (usually non-free ones) to reach one of these states. The watermarks are currently hard-coded as 4% and 9% of the available space. These, and the default of 25% for wantfreevnodes, are too large if the memory size is large. For example, 9% of 75% of MAXVNODES is more than 566000 vnodes to reclaim whenever vnlru_proc() becomes active.

    The vfs.vlru_alloc_cache_src sysctl is removed. The new code frees namecache sources as the last chance to satisfy the highest watermark, instead of selecting source vnodes randomly. This provides good enough behavior to keep vn_fullpath() working in most situations. Filesystem layouts with deep trees, where the removed knob was required, are thus handled automatically.

    As the kernel allocates and frees vnodes, it fully initializes them on every allocation and fully releases them on every free. These are not trivial costs: it starts by zeroing a large structure, then initializes a mutex, a lock manager lock, an rw lock, four lists, and six pointers. Looking at vfs.vnodes_created, these operations are being done millions of times an hour on a busy machine.

    As a performance optimization, this code update uses the uma_init and uma_fini routines to do these initializations and cleanups only as the vnodes enter and leave the vnode zone. With this change, the initializations are done kern.maxvnodes times at system startup, and then only rarely again. The frees are done only if the vnode zone shrinks, which never happens in practice. For those curious about the avoided work, look at the vnode_init() and vnode_fini() functions in sys/kern/vfs_subr.c to see the code that has been removed from the main vnode allocation/free path.

    Mellanox Drivers

    Hardware Information URL:
    Commit Adding the Driver URL:

    Contact: Hans Petter Selasky <>

    The Mellanox FreeBSD team is proud to announce support for the ConnectX-4 series of network cards in FreeBSD 11-current and FreeBSD 10-stable. These devices deliver top performance, with up to 100GBit/s of raw transfer capacity, and support both Ethernet and Infiniband. Currently, the Ethernet driver is ready for use and the Infiniband support for ConnectX-4 is making good progress. We hope that it will be complete before FreeBSD 11.0 is released. For more technical information, refer to the mlx5en(4) manual page in 11-current. The new driver for ConnectX-4 cards is called mlx5 and is put under /sys/dev and not under /sys/ofed as was done for the previous mlx4 driver. The mlx5en(4) kernel module is compiled by default in GENERIC kernels.

    This project was sponsored by Mellanox Technologies.

    Minimal Kernel with PNP-Based Autoloading

    Blog Post URL:

    Contact: Warner Losh <>

    Work on automatically loading modules based on the plug-and-play data from devices that are scanned and found to not already have a driver attached is in progress. Digging this information out from kernel modules, as well as tagging relevant bits of driver tables, has been committed. PC Card, USB, and some PCI devices now have these markings. This data is stored in a file that the kernel, boot loader, and userland processes all can access.

    When complete, a user will be able to run a minimal kernel (currently checked in as the MINIMAL config). Devices necessary for booting will be loaded by loader(8). Other devices may be loaded there, or early in the boot (depending on which gives better performance). Users will still be able to run more monolithic; configurations, as well as limit which kernel modules are available as can be done today, though without the convenience that automatic loading will provide. This work remains ongoing.

    Open tasks:

    1. Go through all the simplebus drivers and add plug-and-play information there. Some additional minor simplebus functionality is needed. There is some work in progress for this.

    2. Go through all the PCI drivers and add plug-and-play information to them. Unlike PC Card or USB, the PCI bus does not have a stylized table of PCI IDs, so each driver invents its own method, meaning that the semi-mechanical conversion that was done with PC Card and USB will not be possible. Instead, customized code for each driver will be needed. Since a large number of drivers have their own device tables, the work will be primarily writing a description of the current table style.

    3. Run-time parsing and loading is still needed.

    MMC Stack Under CAM Framework

    Project Information URL:
    Source Code URL:
    Patch for Review URL:

    Contact: Ilya Bakulin <>

    The goal of this project is to reimplement the existing MMC/SD stack using the CAM framework. This will permit utilizing the well-tested CAM locking model and debug features. It will also be possible to process interrupts generated by the inserted card, which is a prerequisite for implementing the SDIO interface.

    The first version of the code was uploaded to Phabricator for review. The new stack is able to attach to the SD card and bring it to an operational state so it is possible to read and write to the card.

    The only supported SD controller driver is ti_sdhci, which is used on the BeagleBone Black. Modifying other SDHCI-compliant drivers should not be difficult.

    Open tasks:

    1. Rework bus/target/LUN enumeration and the locking model. I do not really understand the CAM locking and am likely to do it incorrectly.

    2. Modify the SDHCI driver on at least one x86 platform. This will make development and collaboration easier.

    3. Begin implementing SDIO-specific bits.

    ntb_hw(4)/if_ntb(4) Driver Synced up to Linux

    Jon Mason's NTB wiki URL:
    Intel NTB whitepaper URL:

    Contact: Conrad Meyer <>

    ntb_hw(4) is now up-to-date with the Linux NTB driver as of the work-in-progress 4.4 kernel (and actually, contains some fixes that haven't landed in the mainline Linux tree yet but will land in 4.5). Only Back-to-back ("B2B") configurations are supported at this time. Going forward, newer hardware may only support the B2B configuration.

    if_ntb(4) is mostly up-to-date with the Linux NTB netdevice driver. Notably absent is support for changing the MTU at runtime.

    This project was sponsored by EMC / Isilon Storage Division.

    Open tasks:

    1. Improving if_ntb(4) to avoid using the entire Base Address Register (BAR) when very large BAR sizes are configured (e.g., 512 GB).

    2. Improving pmap_mapdev(9) to somehow allocate only superpage mappings for large BARs, on platforms that support superpages. (NTB BARs can be as large as 512 GB.)

    Out of Memory Handler Rewrite

    Contact: Konstantin Belousov <>

    The Out of Memory (OOM) code is intended to handle the situation where the system needs free memory to make progress, but no memory can be reused. Most often, the situation is that to free memory, the system needs more free memory. Consider a case where the system needs to page-out dirty pages, but needs to allocate structures to track the writes. OOM "solves" the problem by killing some selection of user processes. In other words, it trades away system deadlock by suffering a partial loss of user data. The assumption is that it is better to kill a process and recover data in other processes than to lose everything.

    Free memory in the FreeBSD Virtual Memory (VM) system appears from two sources. One is the voluntary reclamation of pages used by a process, for example unmapping private anonymous regions, or the last unlink of an otherwise unreferenced file with cached pages. Another source is the pagedaemon, which forcefully frees pages which carry data, of course after the data is moved to some other storage, like swap or file blocks. OOM is triggered when the pagedaemon definitely cannot free memory to satisfy the requests.

    The old criteria to trigger the OOM action was a combination of low free swap space and a low count of free pages (the latter is expressed precisely with the paging targets constants, but this is not relevant to the discussion). That test is mostly incorrect. For example, a low free page state might be caused by a greedy consumer allocating all pages freed by the page daemon in the current pass, but this does not preclude the page daemon from producing more pages. Also, since page-outs are asynchronous, the previous page daemon pass might not immmediately produce any free pages, but they would appear some short time later.

    More seriously, low swap space does not necessarily indicate that we are in trouble: lots of pages might not require swap allocations to be freed, like clean pages or pages backed by files. The last notion is serious, since swap-less systems were considered as having full swap.

    Instead of trying to deduce the deadlock from looking at the current VM state, the new OOM handler tracks the history of page daemon passes. Only when several consecutive passes failed to meet the paging target is an OOM kill considered necessary. The count of consequent failed passes was selected empirically, by testing on small (32M) and large (512G) machines. Auto-tuning of the counter is possible, but requires some more architectural changes to the I/O subsystem.

    Another issue was identified with the algorithm which selects a victim process for OOM kill. It compared the counts of pages mapping entries (PTEs) installed into the machine paging structures. For different reasons, the machine-dependent VM code (pmap) may remove the pte for a memory-resident page. Under some circumstances related to other measures to prevent low memory deadlock, very large processes which consume all system memory could have few or no ptes. The old OOM selector ignored the process which caused the deadlock, killing unrelated processes.

    A new function, vm_pageout_oom_pagecount(), was written which applies a reasonable heuristic to estimate the number of pages freed by killing the given process. This eliminates the effect of selecting small unrelated processes for OOM kill.

    The rewrite was committed to head in r290917 and r290920.

    This project was sponsored by The FreeBSD Foundation.

    sendfile(2) Improvements

    Commit to Head URL:
    Slides URL:
    Presentation (in Russian) URL:

    Contact: Gleb Smirnoff <>

    The sendfile(2) system call was introduced in 1998 as an alternative to a traditional read(2)/write(2) loop, speeding up server performance by a factor of ten at the time. Since it was adopted by all major operating systems, it is now used by any serious web server software. Wherever there is high traffic, there is sendfile(2) under the hood.

    Now, with FreeBSD 11, we are making the next revolutinary step in serving traffic. sendfile(2) no longer blocks waiting on disk I/O. Instead, it immediately returns control to the application, performing the necessary I/O in the background. The original sendfile(2) waited for the disk read operation to complete and then put the data that was read into the socket, then returned to userspace. If a web server served thousands of clients with thousands of requests, it was forced to spawn extra contexts from which to run sendfile(2) to avoid stalls. Alternatively, it could use special tricks like the SF_NODISKIO flag that forces sendfile(2) to serve only content that is cached in memory. Now, these tricks are in the past, and a web server can simply use sendfile(2) as it would use write(2), without any extra care. The new sendfile cuts out the overhead of extra contexts, short writes, and extra syscalls to prepopulate the cache, bringing performance to a new level.

    The new syscall is built on top of two newly-introduced kernel features. The first is an asynchronous VM pager interface and the corresponding VOP_GETPAGES_ASYNC() file system method for UFS. The second is the concept of "not ready" data in sockets. When sendfile(2) is called, first VOP_GETPAGES_ASYNC() is called, which dispatches I/O requests for completion. Buffers with pages to be populated are put into the socket buffer, but flagged as not-yet-ready. Control immediately returns to the application. When the I/O is finished, the buffers are marked as ready, and the socket is activated to continue transmission.

    Additional features of the new sendfile are new flags that provide the application with extra control over the transmitted content. Now it is possible to prevent caching of content in memory, which is useful when it is known that the content is unlikely to be reused any time soon. In such cases, it is better to let the associated storage be freed, rather than putting the data in cache. It is also possible to specify a readahead with every syscall, if the application can predict client behavior.

    The new sendfile(2) is a drop-in replacement, API and ABI compatible with the old one. Applications do not even need to recompile to benefit from the new implementation.

    This work is a joint effort between two companies: NGINX, Inc., and Netflix. There were many people involved in the project. At its initial stage, before code was written, the idea of such an asynchronous drop-in replacement was discussed amongst Gleb Smirnoff, Scott Long, Konstantin Belousov, Adrian Chadd, and Igor Sysoev. The initial prototype was coded by Gleb under the supervision of Kostik on the VM parts of the patch, and under constant pressure from Igor, who demanded that nginx be capable of running with the new sendfile(2) with no modifications. The prototype demonstrated good performance and stability and quickly went into Netflix production in late 2014. During 2015, the code matured and continued serving production traffic at Netflix. Scott Long, Randall R. Stewart, Maksim Yevmenkin, and Andrew Gallatin added their contributions to the code.

    Now we are releasing the code behind our success to the FreeBSD community, making it available to all FreeBSD users worldwide!

    This project was sponsored by Netflix, and NGINX, Inc..

    Open tasks:

    1. SSL_sendfile() — an extension to the new sendfile(2) that allows uploading session keys to the kernel, and then using sendfile(2) on an SSL-enabled socket.

    sysctl Enhancements

    Wikipedia Entry on C99 Fixed-Width Integer Types URL:
    sysctl(8) -t Submission PR URL:

    Contact: Conrad Meyer <>
    Contact: Ravi Pokala <>
    Contact: Marcelo Araujo <>

    Support was added for fixed-width sysctls (signed and unsigned 8-bit, 16-bit, 32-bit, and 64-bit integers). The new KPIs are documented in the sysctl(9) manual page. The sysctl(8) command line tool supports all of the new types.

    sysctl(8) gained the -t flag, which prints sysctl type information (the original patch was submitted by Yoshihiro Ota). This support includes the newly added fixed-width types.

    This project was sponsored by EMC / Isilon Storage Division.

    Touchscreen Support for Raspberry Pi and Beaglebone Black

    Beaglebone Black with 4DCAPE-43T Demo URL:
    Input Stack Plans URL:
    evdev Port URL:

    Contact: Oleksandr Tymoshenko <>

    There are two working proof-of-concept drivers for the AM335x touchscreen and for the official Raspberry Pi's touchscreen LCD.

    Proper touchscreen support would consist of a userland event reading API, a kernel event reporting API, and kernel hardware drivers for specific devices. There is an ongoing effort to port the Linux evdev API to FreeBSD so applications that use libraries like libinput or tslib could be used without any major changes. Since it is not yet complete, I created a naive evdev-like API for both kernel and tslib and was able to run a demo on a Beaglebone Black with 4DCAPE-43T.

    Once evdev makes it into the tree, both hardware drivers can be modified to include "report events" portions and committed.


    armv6 Hard Float Default ABI

    Blog Entry URL:

    Contact: Warner Losh <>

    Work on moving armv6 from a "soft float" ABI (but still using hardware floating point) to a fully "hardware float" API moves forward. The ability to have both soft and hard ABI libraries on the same system is now functional. All armv6 and armv7 systems we support have hardware floating point capabilities. We currently use the floating-point hardware, but with a slightly un-optimal ABI, for compatibility with older versions of FreeBSD. The ABI differences are only at the userspace level — the kernel does not care what floating-point ABI is used, and both types of binaries can run at the same time.

    The run-time linker now knows if a binary uses the hardware float ABI or the software float ABI by examining some fields in the ELF header. The linker uses different paths and config files for hard versus soft binaries. The rc system has been enhanced to load the software float paths. ldconfig now understands soft libraries in much the same way that it understands 32-bit libraries on 64-bit systems. No additional kernel support was necessary for this, apart from a minor patch to pass the ELF header information to the binary, which has been in the tree since last summer.

    The experimental armv6hf MACHINE_ARCH will be retired after a transition period. It will cease to mean anything different from armv6 after the build system changes go in. Support for building soft-float ABI libraries will remain in the tree, to support the WITH_LIBSOFT build option.

    Open tasks:

    1. Complete documentation needs to be written.

    2. Hooks into the FreeBSD build system to generate soft float and transition to hard float after a flag day need to be polished up and committed.

    3. A number of different upgrade/coexistence scenarios need to be tested, and a full package run needs to be done to assess the latest state of the ports tree. This work should be completed by the end of January.

    FreeBSD on Marvell Armada38x

    Contact: Marcin Wojtas <>
    Contact: Michal Stanek <>
    Contact: Bartosz Szczepanek <>
    Contact: Jan Dabros <>

    FreeBSD has been ported to run on the Marvell Armada38x platform. This SoC family boasts single/dual high-performance ARM Cortex-A9 CPUs.

    The multi-user SMP system is fully working and has been tested on Marvell DB-88F6288-GP and SolidRun ClearFog development boards.

    The root filesystem can be hosted on a USB 3.0/2.0 drive or via NFS using a PCIe network card. Experimental support is available for on-chip Gigabit Ethernet (NETA).

    Additional features:

    • GIC+MPIC cascaded interrupts courtesy of INTRNG
    • CESA dual-channel cryptographic engine
    • USB 3.0 and 2.0
    • PCIe 2.0
    • I2C
    • GPIO
    • Watchdog
    • RTC

    The port is under community review and will be integrated into head soon.

    This project was sponsored by Stormshield, and Semihalf.

    Open tasks:

    1. Optimize performance of NETA and prepare for submission.

    FreeBSD on Newer ARM Boards

    FreeBSD on Odroid-C1 URL:
    Commit Adding Glue Driver URL:

    Contact: John Wehle <>
    Contact: Ganbold Tsagaankhuu <>

    We made the changes required to support the Amlogic Meson Ethernet controller on the Hardkernel ODROID-C1 board, which has an Amlogic aml8726-m8b SoC. The main effort needed was to write a glue driver for the Ethernet controller — the Amlogic Meson Ethernet controller is compatible with Synopsys DesignWare 10/100/1000 Ethernet MAC (if_dwc).

    FreeBSD on SoftIron Overdrive 3000

    SoftIron Website URL:

    Contact: Andrew Turner <>

    The SoftIron Overdrive 3000 is an ARMv8 based server with an 8-core AMD Opteron A1100 processor. The Overdrive 3000 has two 10Gbase-T Ethernet ports, two PCI Express ports, and eight SATA ports. FreeBSD has been updated to be able to boot on this hardware.

    Support for the SATA device was added to the ahci(4) driver. Unlike on x86, this is a Memory Mapped (mmio) device, and not on the PCI bus. To support this, a new ahci mmio driver attachment has been added.

    The generic PCIe driver has been updated to improve interrupt handling. This includes supporting the interrupt-map devicetree property, and supporting MSI and MSI-X interrupts on arm64.

    Support for MSI and MSI-X interrupts has been added to the ARM Generic Interrupt Controller v2 (gicv2) driver. This allows devices to use these interrupts. This has been tested with a collection of PCIe NIC hardware.

    This project was sponsored by SoftIron Inc..

    Open tasks:

    1. Write a driver for the 10Gbase-T NIC.


    FreeBSD arm64 Wiki Entry URL:

    Contact: Andrew Turner <>
    Contact: Konstantin Belousov <>
    Contact: Ed Maste <>
    Contact: Ed Schouten <>

    Support was added for kernel modules. This included adding the needed relocation types to the in-kernel relocator, and updating the build logic to build modules for arm64. CTF data is currently not generated for modules due to a linker bug.

    Shared page support was added. This allows gettimeofday(2) to be implemented in userland by directly accessing the timer register. This reduces the overhead of these calls as we no longer need to call into the kernel. This also moves the signal trampoline code away from the stack, allowing for the stack to become non-executable.

    CloudABI support for arm64 was added. This included moving the machine-independent code into a separate file to be shared among all architectures. An issue in the arm64 kernel was found and fixed thanks to the CloudABI test suite.

    Self-hosted poudriere package builds have been tested. These complement the previous build strategy of using qemu usermode emulation. With this combination of self-hosted and qemu usermode building, many ports that used to be broken on arm64 have been fixed, resulting in over 17,000 ports building for the architecture.

    The machine-dependent portion of kernel support for single-stepping userland binaries has been started. This will allow debuggers like lldb to step through an application while debugging.

    Many small fixes have been made to FreeBSD/arm64. These include fixing stack tracing through exceptions, printing more information about "data abort" kernel panics, cleaning up the atomic functions, supporting multi-pass driver attachment, fixing userland stack alignment, cleaning up early page table creation, fixing asynchronous software trap handling, and enabling interrupts in exception handlers.

    This project was sponsored by The FreeBSD Foundation, and ABT Systems Ltd.


    Project Wiki URL:

    Contact: Ruslan Bukin <>
    Contact: Ed Maste <>
    Contact: Arun Thomas <>

    We have begun work on support for the RISC-V architecture.

    RISC-V is a new ISA designed to support computer architecture research and education that is now set to become a standard open architecture for industry implementations.

    A minimal set of changes needed to compile the kernel toolchain has been committed, along with machine headers, run-time linker (rtld-elf) support, and libc/libstand.

    All development has been happening in a separate branch, with a goal of moving development to head in a few weeks.

    At present, FreeBSD/RISC-V boots to multiuser in the Spike simulator.

    This project was sponsored by DARPA, AFRL, and HEIF5.

    Open tasks:

    1. We plan to commit the rest of userspace (i.e., libc), kernel support, etc., in a few weeks.

    Improvements for ARMv6/v7 Support

    Contact: Dominik Ermel <>
    Contact: Wojciech Macek <>
    Contact: Zbigniew Bodek <>

    Numerous improvements for the ARMv6/v7 kernel and tools have been developed by the Semihalf team. Those include:

    • Fixes for KGDB support.
    • Support for branch instructions in ptrace single stepping.
    • Fixes for kernel minidumps.
    • Improvements for LIBUSBBOOT.
    • Support for Exynos EHCI in the loader.
    • A fix for instruction single stepping in DDB.
    • Support for hardware watchpoints, including watchpoints on SMP systems.
    • Single stepping using the ARM Debug Architecture.
    • Support for gzip-compressed kernel modules in kldload.
    • Backport of the new pmap VM code to FreeBSD 10-STABLE (not yet sent to upstream).

    Most of the introduced changes have been committed to head and more are on the way.

    This project was sponsored by Juniper Networks Inc., and Semihalf.

    Open tasks:

    1. Finish upstreaming the hardware watchpoints support.

    Userland Programs

    Base System Build Improvements

    FreeBSD-Arch Post Describing Plans URL:
    BSDCan 2014 META_MODE Presentation URL:

    Contact: Bryan Drewery <>

    Bryan Drewery (bdrewery@) has been working to improve the build framework as well as buildworld build times. The build system has been largely untouched by large-scale changes for many years. Most of the effort has been on improving the recent META_MODE merge that was presented at BSDCan 2014. This is a new build system that is not currently enabled by default but brings many benefits. Beyond that, some highlights of the work changing buildworld are:

    • WITH_FAST_DEPEND, which avoids calling "mkdep" during the make depend phase and instead generates dependency files during compilation. The old scheme was pre-processing all source files twice. The new version saves 16-35% in build times.
    • WITH_CCACHE_BUILD adds built-in ccache support, avoiding many of the historical pitfalls of changing CC in make.conf to use ccache.
    • Many improvements for parallelization of the build.
    • LIBADD improvements to ensure proper usage of this tool to replace duplicate LDADD and DPADD statements. Further work is under way to reduce overlinking.
    • A lot of cleanup of improper framework usage.
    • Ensuring that installing files from the build tree fails if the destination directory is missing, rather than installing a file as the directory name.

    This project was sponsored by EMC / Isilon Storage Division.

    Open tasks:

    1. See the FreeBSD-arch mail for more information on planned work.

    ELF Tool Chain Tools

    ELF Tool Chain Website URL:

    Contact: Ed Maste <>

    The ELF Tool Chain project provides BSD-licensed implementations of compilation tools and libraries for building and analyzing ELF objects. The project began as part of FreeBSD but later became an independent project in order to encourage wider participation from others in the open-source developer community.

    In the last quarter of 2015 the ELF Tool Chain tools were updated to a snapshot of upstream Subversion revision 3272. Improvements include better input file validation, RISC-V support, support for Xen ELF notes, additional MIPS and ARM relocations, better performance, and bug fixes.

    The ELF Tool Chain project is planning a new release in the first quarter of 2016, which will facilitate wider testing and use by projects in addition to FreeBSD.

    This project was sponsored by The FreeBSD Foundation.

    Open tasks:

    1. Add missing functionality (PE/COFF support) to elfcopy and migrate the base system build.

    2. Fix issues found by fuzzing inputs to the tools.

    3. Add automatic support for separate debug files.

    The LLDB Debugger

    FreeBSD LLDB Wiki Page URL:

    Contact: Ed Maste <>

    LLDB is the debugger from the LLVM family of projects. Originally developed for Mac OS X, it now also supports FreeBSD, NetBSD, Linux, Android, and Windows. It builds on existing components in the larger LLVM project, for example using Clang's expression parser and LLVM's disassembler.

    LLDB in the FreeBSD base system was upgraded to version 3.7.0 as part of the Clang and LLVM upgrade, and it will similarly be upgraded again to 3.8.0 for FreeBSD 11.0-RELEASE.

    LLDB is now enabled by default on the amd64 and arm64 platforms. It is now a functional basic debugger on arm64, after a number of fixes were made in the last quarter to both LLDB and the FreeBSD kernel.

    This project was sponsored by The FreeBSD Foundation.

    Open tasks:

    1. Rework the LLDB build to use LLVM and Clang shared libraries.

    2. Port a remote debugging stub to FreeBSD.

    3. Add support for local and core file kernel debugging.

    4. Improve support on architectures other than amd64 and arm64.

    Updates to GDB

    New 1:1-Only Thread Target for FreeBSD URL:

    Contact: John Baldwin <>

    The KGDB option is now on by default in the devel/gdb port.

    Changes to support cross-debugging of crashdumps in libkvm were committed to head in r291406.

    A new thread target for FreeBSD that is suitable for merging upstream has been written and lightly tested. However, it is not yet available as an option in the port. This thread target uses ptrace(2) directly rather than libthread_db and as such supports threads on all ABIs (such as FreeBSD/i386 binaries on FreeBSD/amd64 and possibly Linux binaries, though that is not yet tested). It also requires less-invasive changes in the MD targets in GDB compared to the libthread_db-based target.

    Open tasks:

    1. Add a port option for the new 1:1-only thread target.

    2. Test the new 1:1-only thread target.

    3. Figure out why the powerpc kgdb targets are not able to unwind the stack past the initial frame.

    4. Add support for more platforms (arm, mips, aarch64) to upstream gdb for both userland and kgdb.

    5. Add support for debugging powerpc vector registers.


    Bringing GitLab into the Ports Collection

    PR for the New Port URL:
    Installation Guide URL:
    Upstream GitLab website URL:

    Contact: Torsten Zühlsdorff <>

    GitLab is a web-based Git repository manager with many features that is used by more than 100,000 organizations including NASA and Alibaba. It also is a very long-standing entry on the "Wanted Ports" list of the FreeBSD Wiki.

    In the last quarter, there was steady progress in the project itself and the porting. The current release of GitLab 8.3 is now based on Rails 4.2, which obsoletes the need for around 50 new ports. Now there are only 5 dependencies left to be committed!

    While the new version of GitLab 8.3 eases the porting, there are big changes since the last working port of GitLab 7.14. Nonetheless, it could be expected to see the next working port in the first quarter of 2016.

    This project was sponsored by anyMOTION GRAPHICS GmbH, Düsseldorf, Germany.

    Open tasks:

    1. Update the patches from GitLab 7.14 to 8.3.

    2. Update the documentation.

    3. Provide an updated patch.

    GNOME on FreeBSD

    FreeBSD Gnome Website URL:
    Devel Repository URL:
    Upstream Build Bot URL:
    USE_GNOME Porter's Handbook Chapter URL:

    Contact: FreeBSD GNOME Team <>

    The FreeBSD GNOME Team maintains the GNOME, MATE, and CINNAMON desktop environments and graphical user interfaces for FreeBSD. GNOME 3 is part of the GNU Project. MATE is a fork of the GNOME 2 desktop. CINNAMON is a desktop environment using GNOME 3 technologies but with a GNOME 2 look and feel.

    This quarter, due to limited available time there was not much progress. This began to change in December, when work started on porting MATE 1.12 and CINNAMON 2.8 to FreeBSD.

    Open tasks:

    1. The FreeBSD GNOME website is stale. Work is under way to improve it.

    2. Continue working on investigating the issues blocking GNOME 3.18.

    IPv6 Promotion Campaign

    Wiki Page URL:

    Contact: Torsten Zühlsdorff <>

    There are more and more machines on the internet that only support IPv6. I manage some of them, and was regularly hit by missing IPv6 support when fetching the distfiles needed for building ports.

    I did some research into the impact of missing IPv6 support on the ports tree. The results are that 10,308 of 25,522 ports are not fetchable when using IPv6. This renders, through dependencies, a total of 17,715 ports unbuildable from IPv6-only systems. All you can do then is wait and hope that caches the distfile. But this will take some time, which might not be a luxury available when a piece of software in use is hit by a security issue.

    Based on the research, a promotion campaign for IPv6 was started. Some volunteers will contact the relevant system administrators and try to convince them to support IPv6. This will start in January 2016 and will hopefully create some progress soon.

    KDE on FreeBSD

    KDE on FreeBSD Website URL:
    Experimental KDE Ports Staging Area URL:
    KDE on FreeBSD Wiki URL:
    KDE/FreeBSD Mailing List URL:
    Development Repository for Integrating KDE Frameworks 5 and Plasma 5 URL:

    Contact: KDE on FreeBSD team <>

    The KDE on FreeBSD team focuses on packaging and making sure that the experience of KDE and Qt on FreeBSD is as good as possible.

    The team kept busy during the last quarter of 2015. Quite a few big updates were committed to the ports tree, and a few more are being worked on in our experimental repository.

    As in previous quarters, we would like to thank several people who have contributed with machines, patches, and general help. Tobias Berner, Guido Falsi (madpilot@), Adriaan de Groot, Ralf Nolden, Steve Wills (swills@), and Josh Paetzel (jpaetzel@) have been essential to our work.

    The following big updates landed in the ports tree this quarter. In many cases, we have also contributed patches to the upstream projects.

    • CMake 3.4.0 and 3.4.1
    • Calligra 2.9.1, the latest release of the integrated work applications suite. Calligra had last been updated in the ports tree at the end of 2013!
    • PyQt4 4.11.4, QScintilla2 2.9.1 and SIP 4.17.
    • PyQt5 5.5.1. Thanks to the work spearheaded by Guido Falsi and Tobias Berner in the previous quarter, the PyQt5 ports have finally been committed to the ports tree. Not only was this long-awaited on its own, it allows other ports to be updated to their latest versions.
    • QtCreator 3.5.1 and 3.6.0.
    • A couple of Qt5 packaging bugs were fixed: it should now be more straightforward to use the Qt5 ports to build software outside the ports tree, and it is now possible to build ports that require a C++11 compiler and Qt5 on FreeBSD 9.x.

    Work on updating the Qt5 ports to their latest version, as well as porting KDE Frameworks 5 and Plasma 5 to FreeBSD, is well under way in our experimental area51 repository. At the moment, it contains Qt5 5.5.1, KDE Frameworks 5.17.0, Plasma 5.5.1 and KDE Applications 15.12.0.

    Users interested in testing those ports are encouraged to follow the instructions in our website and report their results to our mailing list. Qt5 5.5.1 is in our "qt-5.5" branch, and Plasma 5 and the rest is in the "plasma5" branch (which also contains Qt 5.5.1).

    Open tasks:

    1. Commit the Qt5 5.5.1 update.

    2. Land the KDE Frameworks 5 and Plasma 5 ports in the tree.

    3. Investigate what needs to be done to make QtWebEngine, the Chromium-based replacement for QtWebKit, work on FreeBSD.

    Linux Kernel as a Library Added to the Ports Collection

    Upstream LKL Github repository URL:

    Contact: Conrad Meyer <>

    LKL ("Linux Kernel as a Library") is a special "architecture" of the full Linux kernel that builds as a userspace library on various platforms, including FreeBSD. One application of such a library is using Linux filesystem drivers to implement a FUSE backend.

    fusefs-lkl's lklfuse binary is such a FUSE filesystem. It can mount ext4/3/2, XFS, and BTRFS read-write, using the native drivers from Linux.

    sysutils/fusefs-lkl can now be installed either from packages or ports, providing access to these filesystems on FreeBSD via FUSE.

    LXQt on FreeBSD

    FreeBSD LXQt Project URL:
    LXQt Devel Repository URL:

    Contact: Olivier Duchateau <>

    LXQt is the Qt port of and the upcoming version of LXDE, the Lightweight Desktop Environment. It is the product of the merge between the LXDE-Qt and the Razor-qt projects.

    The porting effort remains very much a work in progress: it needs some components of Plasma 5, the new major KDE workspace.

    Currently, only the 0.10 branch is functional. See our wiki page for a complete list of applications.

    We also sent updates for some components of LXDE, required for the LXQt desktop:

    • x11/menu-cache 1.0.1
    • x11/lxmenu-data 0.1.4

    Binary packages are available (only for test purposes) which are regularly tested with the KDE development repository.

    Open tasks:

    1. Port libsysstat to BSD systems.

    2. Fix some issues that need to be resolved, especially the shutdown and reboot commands.

    New Tools to Enhance the Porting Experience

    pytoport: Generate FreeBSD Ports from Python modules on PyPI URL:
    bandar: Create Development Overlays for the Ports Tree URL:
    skog: Generate Visual Dependency Trees for FreeBSD Ports URL:
    spdx-lookup: SPDX License List Query Tool URL:

    Contact: Brendan Molloy <>

    When I starting working on ports for FreeBSD in the last couple of weeks, I found that my workflow was not as efficient as it could be using just the available tools, so I made a few that could be useful to the development community at large. All of these have been or will soon be added to the Ports tree, so you can play with them today!

    pytoport is a command-line application that generates a skeleton port for a given PyPI package name. It attempts to generate the correct dependencies, makes a good attempt at guessing the license using spdx-lookup, and generates a pkg-descr. This made generating the fifteen or so ports I was working on a complete breeze.

    While doing this, however, I noticed that some ports were bringing in dependencies that I did not expect, and I needed some way to visualise this. skog builds a dependency tree from the depends lists output by the Ports framework, and displays it on the command line (with extra shiny output if you are using UTF-8). No more pesky example and documentation dependencies being dragged in when you clearly toggled that OPTION as far off as it would go.

    While doing all of this, I found it cumbersome to be copying ports back and forth between my small development tree living in git and the larger upstream SVN tree I was using in poudriere. I built a tool called bandar that takes advantage of the FUSE version of unionfs to easily overlay my dev tree on the upstream tree, run lint checks, poudriere, and generate archives with ease.

    I am very impressed with how easy it was to build more tooling for FreeBSD. I hope some of these tools will be of some use to you, and as always, I'd love to hear your feedback!

    Open tasks:

    1. Improve skog to support searching a tree for a certain port.

    2. Get the bandar port completed.

    3. Continue to improve pytoport, adding trove support and better dependency handling.

    4. Patches welcome for all of the above!

    Node.js Modules

    Node.js Modules Repository URL:

    Contact: Olivier Duchateau <>

    Node.js is a platform built on Chrome's JavaScript runtime for easily building fast, scalable network applications. It uses an event-driven, non-blocking I/O model that makes it lightweight and efficient — perfect for data-intensive real-time applications that run across distributed devices.

    The goal of this project is to make it easy to install the modules available in the npm package registry.

    Currently, the repository contains slightly fewer than 300 new ports, in particular:

    • Socket.IO, a library for realtime web applications
    • Jison, a JavaSript parser generator

    We have improved the USES framework:

    • Users can define which version of Node.js will be installed through /etc/make.conf.
    • node-gyp is now well-integrated into the USES framework, via the build argument.
    • The pkg-plist is now automatically generated to make portlint happy.

    Each port is up-to-date.

    Open tasks:

    1. Update the pre-draft documentation.

    2. Bring in grunt.js (and modules), the JavaScript task runner.

    Ports Collection

    Ports Collection Landing Page URL:
    Contributor's Guide URL:
    Ports Monitoring Service URL:
    Ports Management Team Website URL:
    Portmgr on Facebook URL:

    Contact: Frederic Culot <>
    Contact: Frederic Culot <>
    Contact: FreeBSD Ports Management Team <>

    As of the end of the fourth quarter, the ports tree holds a bit more than 25,000 ports, and the PR count is around 2,000. The activity on the ports tree remains steady, with about 7,000 commits performed by almost 120 active committers.

    On the problem reports front, figures show an encouraging trend, with a significant increase in the number of PRs fixed during Q4. Indeed, almost 1,800 reports were fixed, which makes an increase of about 20% compared to Q3.

    In Q4, eight commit bits were taken in for safekeeping, following an inactivity period of more than 18 months (lioux, lippe, simon, jhay, max, sumikawa, alexey, sperber). Three new developers were granted a ports commit bit (Kenji Takefu, Carlos Puga Medina, and Ian Lepore), and one returning committer (miwi) had his commit bit reinstated.

    Also related to the management of ports commit bits, nox's grants were revoked, since the FreeBSD developers learned that Juergen Lock had passed away.

    On the management side, no changes were made to the portmgr team during Q4.

    On QA side 33 exp-runs were performed to validate sensitive updates or cleanups. Amongst those noticeable changes are the update to GCC 4.9, CMake to 3.4.1, PostgreSQL to 9.4, and ruby-gems to 2.5.0. Some infrastructure changes included the usage of a WRKSRC different from WRKDIR when NO_WRKSUBDIR is set, the removal of from, and the move of QT_NONSTANDARD to

    Open tasks:

    1. We would like to remind everyone that the ports tree is built and run by volunteers, and any help is greatly appreciated. While Q4 saw a significant increase in the number of problem reports fixed, we encourage all ports committers to have a look at the issues reported by our users and try to fix as many as possible. Many thanks to all who made a contribution during Q4, and keep up the good work in 2016!

    Supporting Variants in the Ports Framework

    Poudriere PoC with Variants URL:
    Ports Makefile PoC with Examples URL:

    Contact: Brendan Molloy <>

    I recently became involved with FreeBSD (as in, the last 2-3 weeks), and found myself quickly involved with Ports development. What struck me immediately was the difficulty in providing a Python package that was depended upon by multiple versions of Python. As it turns out, poudriere can currently only generate one package per port, meaning that a Python version-neutral (compatible with 2.x and 3.x) port cannot simultaneously be packaged for each variant at the same time.

    I discussed the issue with Kubilay Kocak, who suggested that I look into implementing a "variants protocol" within the Ports framework and the necessary changes to poudriere to allow a port to generate more than one package.

    Support for variants is strongly needed in Ports and provides significant benefits.

    • It would allow Python and other languages to provide packages for dependencies for multiple language versions from the same port.
    • It alleviates the need for so-called "slave ports", as a single port could now have multiple generated packages from a single port.
    • It would have a very small impact on the greater Ports ecosystem: adding only two new variables, VARIANT and VARIANTS.
    • It would provide a more consistent approach between different packaging teams for handling variations.

    For a simple example, editors/vim-lite could be folded into the editors/vim port, while still generating a vim and vim-lite package. For Python, VARIANTS can be derived from the already used USES flags and generate compatible packages. py27-foobar and py34-foobar could now be consistently generated by poudriere without issue.

    Fortunately, this is not a wishful thinking piece. I dug in my heels and have implemented a proof-of-concept implementation of variants in the Ports framework, including the necessary modifications to poudriere in order to support it. It was mildly upsettling to find that poudriere is mostly written in Bourne shell scripts, but I pressed on nonetheless.

    I started with the prototype made by Baptiste Daroussin as a base, and built from there. The poudriere PoC aims to limit changes as much as possible to merely adding support for the new variants flags, while also at the request of Kubilay Kocak making the logging output more package-centric (as opposed to port-centric) as a result of these changes.

    This is a work in progress, and I would love to hear your feedback. I have enjoyed my first few weeks working on FreeBSD, and I hope to stay here for quite some time.

    Open tasks:

    1. Any constructive feedback on the implementation would be very welcome!

    2. Hopefully the code will be of sufficient quality to be considered for formal review in the coming months.

    Xfce on FreeBSD

    FreeBSD Xfce Project URL:
    FreeBSD Xfce Repository URL:

    Contact: FreeBSD Xfce Team <>

    Xfce is a free software desktop environment for Unix and Unix-like platforms, such as FreeBSD. It aims to be fast and lightweight, while still being visually appealing and easy to use.

    During this quarter, the team has kept these applications up-to-date:

    • audio/xfce4-pulseaudio-plugin 0.2.4
    • multimedia/xfce4-parole 0.8.1
    • x11/xfce4-whiskermenu-plugin 1.5.2

    We also follow the unstable releases (available in our experimental repository) of:

    • x11/xfce4-dashboard 0.5.4

    Open tasks:

    1. Propose a patch to upstream to fix Xfdashboard with our version of OpenGL (it currently coredumps).


    "FreeBSD Mastery: Specialty Filesystems" Early Access Version Now Available

    Book site URL:
    Early access version URL:

    Contact: Michael Lucas <>

    FreeBSD Mastery: Specialty Filesystems is now in copyediting. The ebook should be available by the end of January at all major vendors, and the print in February.

    The book covers everything from removable media, to FUSE, NFSv4 ACLs, iSCSI, CIFS, and more.

    If you act really quickly, you can get the electronic early access version at a 10% discount. You will get the final ebook when it comes out as well. (This offer evaporates when the final version comes out.)

    style(9) Enhanced to Allow C99 bool

    Bruce's Email Requesting bool be Added to style(9) URL:
    Differential Revision for the Change URL:

    Contact: Bruce Evans <>
    Contact: Conrad Meyer <>

    Use of bool is now allowed. It was allowed previously, as well, but now it is really allowed. Party like it's 1999!

    This project was sponsored by EMC / Isilon Storage Division.

    Open tasks:

    1. Specify style(9)'s opinion on iso646.h.

    2. Fix intmax_t to be 128-bit on platforms where __int128_t is used.



    HardenedBSD Website URL:
    Introducing HardenedBSD's New Binary Updater URL:
    secadm Beta Published URL:
    New Package Building Server URL:
    secadm URL:
    HardenedBSD Haswell Support URL:
    Nightly Builds for HardenedBSD Haswell Support URL:

    Contact: Shawn Webb <>
    Contact: Oliver Pinter <>

    HardenedBSD has been hard at work improving the performance and stability of our security enhancements. Security flags are now per-thread instead of per-process, removing some locking overhead. ASLR for mmap(MAP_32BIT) requests has been refactored, but lib32 is now disabled by default.

    We have developed a new binary update utility, hbsd-update, akin to freebsd-update. In addition to normal OS installs, it can also update jails and ZFS Boot Environments (ZFS BEs). Updates are signed using X.509 certificates.

    secadm 0.3-beta has landed. It has been rewritten from scratch to be more efficient. As part of the rewrite, the rule syntax has changed and users must update their rulesets as described in the README.

    Thanks to generous donations of a server from G2, Inc and hosting from Automated Tendencies, we can now do full package builds in just 35 hours, down from 75 hours. This machine will also provide weekly binary updates for the kernel and base system.

    Owing partly to the needs of the developers, we have an experimental branch that includes the work Jean-Sébastien Pédron has under way for Haswell graphics support, on top of FreeBSD 11-current. Binary updates are also provided for this branch.

    Unfortunately, in order to focus our efforts on improving HardenedBSD, we have had to pull back from submitting our ASLR patches to FreeBSD. The past two years' efforts to address comments on the submission have taken their toll, and the effort is no longer sustainable. We are proud to be based on FreeBSD and believe that the whole community could benefit from the security technologies we are developing. We hope that someone else will be able to step forward and finish off the task of integrating ASLR into FreeBSD.

    This project was sponsored by Automated Tendencies, G2, Inc, and SoldierX.

    NanoBSD Modernization

    Contact: Warner Losh <>

    This quarter's NanoBSD updates target three main areas. First, building a NanoBSD image required root privileges. Second, building for embedded platforms required detailed knowledge of the format required to boot. Third, the exact image sizes needed to be known to produce an image.

    When NanoBSD was written, FreeBSD's build system required root privileges for the install step and onward. NanoBSD added to this by creating a md(4) device in which to construct the image. Some configurations of NanoBSD added further to this by creating a chroot in which to cleanly build packages. NanoBSD solves the first problem using the new NO_ROOT build option to create a meta file. NanoBSD also augments this record as files are created and removed. The meta file is then fed into makefs(8) to create a UFS image with the proper permissions. The UFS image, and sometimes a DOS FAT partition, are then passed to mkimg(1) to create the final SD image. The mtree manipulation has been written as a separate script to allow it to move into the base system where it could assist with other build orchestration tools (though the move has not happened yet).

    The detailed knowledge of how to build each embedded image (as well as some of the base images for qemu) has always been hard to enshrine. Crochet puts this knowledge into its builds. The FreeBSD release system puts it into its system. NanoBSD, prior to the current work, provided no way to access its knowledge of how to build images. The current state of this project allows the user to set a simple image type and have NanoBSD deal with all of the details needed to create that image type. This includes using the u-boot ports and installing the right files into a FAT partition so that FreeBSD can boot with ubldr(8), creating the right boot1.elf file for powerpc64 qemu booting, or the more familiar (though needlessly complicated) x86 setup. Previous versions of NanoBSD required too much specialized knowledge from the user. This work aims to concentrate the knowledge into a set of simple scripts for any build orchestration system to use.

    Finally, NanoBSD images in the past have needed very specific knowledge of the target device. Part of this is a legacy of the BIOS state-of-the-art a decade ago, which required very careful matching of the image to the actual device in the deployed system. Although relevant at the time, such systems are now vanishingly rare. Support for them will be phased out (though given the flexibility of NanoBSD, it can be moved to the few remaining examples in the tree and also partially covered by the generic image scripts). Today, the typical use case is to create an SD or microSD card image, and have the image resize itself on boot. NanoBSD now supports that workflow.

    In addition to these items, a number of minor improvements have been made:

    • Support for CPUTYPE-specialized builds. This includes both NanoBSD support as well as important bug fixes in the base system.
    • Support for marking MBR partitions as active.
    • Support for more partition types.

    Open tasks:

    1. mkimg(8) needs to be augmented to create images for the i.MX6 and Allwinner (and others) SoCs. These SoCs require a boot image to be written after the MBR, but before the first partition starts.

    2. The chroot functionality of some NanoBSD configurations has not yet been migrated for non-privileged builds.

    3. The functionality to manipulate mtree(8) files should be moved into the base system for use by other build orchestration tools.

    4. The script to create a bootable image from one or more trees of files, as well as some creation of those trees, should be moved into the base system for use with other build orchestration tools.

    5. The growfs functionality works great for single images growing to the whole disk. However, NanoBSD would prefer that the boot FS/partition grow to approximately 1/2 the size of the media and another identical (or close) partition be created for the ping-ponging upgrades that NanoBSD is setup for. This needs to be implemented in the growfs rc.d(8) script.


    Development tree on GitHub URL:

    Contact: Mark Heily <>

    The relaunchd project provides a service management daemon that is similar to the original launchd introduced in Apple OS X.

    It is not limited to the original features of launchd, however: interesting work is being done to add support for launching programs in jails, passing socket descriptors from the host to a jail, and launching programs within a preconfigured capsicum(4) sandbox. Additionally, relaunchd uses UCL for its configuration files, so jobs can be defined in JSON or other formats supported by UCL.

    While there is still work to be done, most of the important features of the original launchd have been implemented, and relaunchd has been made available in the FreeBSD Ports Collection. It should still be considered experimental and not ready for production use, but everyone is welcome to try it, report issues, and contribute code or ideas for improvement.

    Open tasks:

    1. Add support for restarting jobs if they crash.

    2. Implement the cron(8) emulation feature.

    3. Add support for monitoring files and directories for changes and launching jobs when changes are detected.

    4. Finish things that are incomplete, such as support for jails and passing open socket descriptors to child processes.

    5. Improve the documentation and provide more examples of usage.

    System Initialization and Service Management

    A Comparison of init(8) and rc(8) Replacements URL:

    Contact: Mark Heily <>
    Contact: Jonathan de Boyne Pollard <J.deBoynePollard-newsgroups@NTLWorld.COM>
    Contact: Jordan Hubbard <>

    There are three active projects to provide an alternative to the traditional init(8) and rc(8) subsystems that manage the boot process and system services. There are a number of reasons driving the desire for change, including:

    • Faster boot times, made possible by launching services in parallel
    • Greater reliability, by ensuring that services are automatically restarted if they terminate unexpectedly
    • Simplified dependency management, using socket activation and similar techniques
    • The ability to launch services "on demand", and have them self-terminate when idle
    • Improved security, by removing the need to start common daemons as the root user

    Two of the projects, launchd and relaunchd, are based on the launchd(8) API introduced by Apple in Mac OS X. The NextBSD project has ported the original Apple source code by writing a Mach compatibility layer that allows launchd to run on FreeBSD. The relaunchd project started from scratch with the goal of creating a more modular, lightweight, and portable implementation of the launchd API. The third project, nosh, is a unique creation that borrows concepts from launchd, systemd, and several other Unix operating systems.

    While the FreeBSD Project has not made a decision to replace the current init(8) and rc(8) subsystems, the existence and active development of alternatives will continue to drive innovation in this space.

    Jordan Hubbard is the contact point for the NextBSD launchd, Jonathan de Boyne Pollard is the contact point for nosh, and Mark Heily is the contact point for relaunchd.

    The FreeBSD Foundation

    Foundation Website URL:
    FreeBSD Journal URL:

    Contact: Deb Goodkin <>

    The FreeBSD Foundation is a 501(c)(3) non-profit organization dedicated to supporting and promoting the FreeBSD Project and community worldwide. Funding comes from individual and corporate donations and is used to fund and manage development projects, conferences and developer summits, and provide travel grants to FreeBSD developers. The Foundation purchases hardware to improve and maintain FreeBSD infrastructure and publishes FreeBSD white papers and marketing material to promote, educate, and advocate for the FreeBSD Project. The Foundation also represents the FreeBSD Project in executing contracts, license agreements, and other legal arrangements that require a recognized legal entity.

    Here are some highlights of what we did to help FreeBSD last quarter:

    On the advocacy front, the Foundation attended and sponsored EuroBSDcon, which took place Oct 1-4 ( in Stockholm, Sweden. Two days prior, during the developer summit, Deb Goodkin ran a session on Recruiting to FreeBSD. The Foundation was also very active during the event itself; in addition to Deb, we had Dru Lavigne, Kirk McKusick, Erwin Lansing, Ed Maste, Hiroki Sato, Benedict Reuschling, and Edward Tomasz Napierała attend the conference. Deb and Ed gave a presentation on how the Foundation supports a BSD project. Kirk gave a presentation on "a Brief History of the BSD Fast File System," and he taught the two-day tutorial "Introduction to the FreeBSD Open-Source Operating System."

    Deb then attended the 2015 Grace Hopper Conference that was held in Houston, TX, October 14-16. The conference is for women in computing and most of the attendees were female computer science majors, female software developers, and college professors. The Foundation was proud to be a Silver Sponsor. The conference was very successful for us. Our presence allowed us to raise awareness of the Project, help recruit more women, and get more professors to include FreeBSD in their curriculum.

    George V. Neville-Neil traveled to Bangkok, Thailand to present talks on DTrace, FreeBSD, and teaching with DTrace. The talks were presented at Chulalongkorn University, which is the largest University in Thailand with the largest engineering school. The first talk was the practitioner's introduction to DTrace in which the technology, history and usage is explained without diving into all the kernel subsystems. The second was the sales pitch for teaching with Dtrace and with FreeBSD. The pitch was well received and there were some very good points made by the audience. The facts that the course materials are both open source and hosted on github were also well received.

    Kirk McKusick completed a 10-hour tutorial about FreeBSD for Pearson Education in their "Live Lesson" program. In particular, there is a great free snippet from that course comparing FreeBSD against Linux here: Find out more about the whole session at:*uS0&subid=&offerid=163217.1&type=10&tmpid=3559&

    Anne Dickison resumed the Faces of FreeBSD series with interviews featuring Michael Dexter and Erin Clark. She also continued to produce and distribute FreeBSD materials for conferences, as well as advocating for FreeBSD over our social channels.

    George V. Neville-Neil headed up the latest Silicon Valley Vendor and Developer Summit, November 2-3, at the NetApp campus in Sunnyvale, California. Topics of discussion ranged over new developments in persistent memory, the use of FreeBSD by a company that builds rackscale systems, developments in our compiler and tool suite, as well as others. Additional Foundation Board and Staff attending the summit included: Deb Goodkin, Glen Barber, Justin T. Gibbs, Kirk McKusick, Ed Maste, and Hiroki Sato. The complete schedule, and some of the slides, are available on the FreeBSD Wiki

    Notes from the always lively "Have/Need/Want session" are available at

    While in the Bay Area, some Foundation members visited commercial users of FreeBSD to help understand their needs, update them on the work the Foundation is doing, and facilitate collaboration between them and the Project.

    We were a sponsor of the 2015 OpenZFS Developer Summit, which took place October 19-20, in San Francisco, CA. Justin T. Gibbs and Kirk McKusick attended the conference.

    Justin T. Gibbs continued his semester long class teaching Intro to Computer Science using FreeBSD at a middle school.

    Ed Maste, Edward Tomasz Napierała, and Konstantin Belousov continue to make progress on Foundation funded development projects. More specifically:

    • Ed Worked on a number of items relating to the tool chain: LLD linker, ELF Tool Chain components, and LLDB debugger, and tested, integrated, and merged outstanding UEFI work.
    • Edward finished work on the reroot project as well as spending some time on a certificate-transparency port. He also implemented a prototype to support disk IO limit in RCTL.
    • Konstantin rewrote the out of memory killer logic, which, in particular, fixed FreeBSD operation on systems without swap, especially systems with very little memory. The latter are becoming more and more common with the popularity of embedded ARM platforms where FreeBSD runs, but it also affects large systems which are usually configured without swap. He also finalized and committed the shared page support for the ARMv7 and ARMv8 systems. This allows for a non-executable stack on ARMv7, and a much faster userspace gettimeofday(2) for both, similar to x86.

    Ed Maste presented a FreeBSD/arm64 talk and a hands-on demo at ARM Techcon, which took place November 10-12, 2015, in Santa Clara, CA.

    We continued publishing our monthly newsletters and acquiring new company testimonials about using FreeBSD, including from Verisign and Nginx.

    Anne Dickison, Dru Lavigne, and Glen Barber represented the Foundation at USENIX LISA '15, which took place November 3-8, in Washington D.C.. The Foundation had a booth in the Expo Hall and participated in a BoF. Besides connecting with current community members, we spoke with attendees who were interested in getting involved with the Project and helped set them on the correct path. We also took the opportunity to remind those who had not used FreeBSD in a while what they were missing. Glen also attended the USENIX Release Engineering Summit, which was co-located with LISA '15.

    We published the Sept/Oct and Nov/Dec issues of the FreeBSD Journal.

    George V. Neville-Neil and Robert Watson announced the release of their TeachBSD initiative: TeachBSD offers a set of open source reusable course materials designed to allow others to teach both university students and software practitioners FreeBSD operating system fundamentals. The Foundation is proud to have partly sponsored their efforts to teach the initial graduate level course on operating systems with tracing at the University of Cambridge.

    Deb Goodkin invited a representative from the Outreachy program to talk at the Ottawa FreeBSD Developer Summit about the program and how we can get involved.

    Deb also started discussions with CS professors from the University of Colorado, Boulder to offer some Intro to FreeBSD workshops.

    Glen Barber continued wearing many hats to support to the Project. For Release Engineering:

    • Added support for building BANANAPI, CUBIEBOARD, and CUBIEBOARD2 arm images.
    • Deprecated the use of MD5 checksums for verifying installation media downloaded from the FreeBSD Project mirrors.
    • Various miscellaneous updates and fixes to release build code.
    • Continued providing regular development snapshot builds.
    Under Systems Administration:
    • Assisted the Admins team with migrating various services to two new colocation facilities near Sunnyvale, generously provided by RootBSD and LimeLight Networks.
    • Moved email services for the Foundation to a new server.

    Ed Maste attended the Reproducible Builds World Summit, which took place in Athens, Greece, December 1-3, 2015.

    We wrapped up our 2015 fundraising efforts with our End-of-Year fundraising campaign by participating in #GivingTuesday, and continuing with weekly email and social media requests for support of the Foundation. Final fundraising numbers will be available in Q1 2016.

    News Home | Status Home