Introduction
This report covers FreeBSD related projects for the period between October and December, and is the fourth of four planned reports for 2020.
This quarter had quite a lot of work done, including but certainly not limited to, in areas relating to everything from multiple architectures such as x86, aarch64, riscv, and ppc64 for both base and ports, over kernel changes such as vectored aio, routing lookups and multipathing, an alternative random(4) implementation, zstd integration for kernel dumps, log compression, zfs and preparations for pkg(8), along with wifi changes, changes to the toolchain like the new elfctl utility, and all the way to big changes like the git migration and moving the documentation from DocBook to Hugo/AsciiDoctor, as well as many other things too numerous to mention in an introduction.
This report with 42 entries, which don't hold the answer to life, the universe and everything, couldn't have happened without all the people doing the work also writing an entry for the report, so the quarterly team would like to thank them, as otherwise, we wouldn't have anything to do.
Please note that the deadline for submissions covering the period between January and March is March 31st.
We hope you'll enjoy reading as much as we enjoyed compiling it.
Daniel Ebdrup Jensen, on behalf of the quarterly team.
FreeBSD Team Reports
- FreeBSD Foundation
- FreeBSD Release Engineering Team
- Cluster Administration Team
- Continuous Integration
- Ports Collection
- Office Hours
Projects
- GPL in Base
- Git Migration Working Group
- Linux compatibility layer update
- LLDB Debugger Improvements
- Upstreaming NetApp Changes
- NFS over TLS implementation
- OpenBSM Synchronisation
- Tool Chain
Kernel
- ENA FreeBSD Driver Update
- Intel wireless update
- Fenestras X random(4)
- pf performance improvement
- IP Routing lookup improvements
- Scalable routing multipath support
- Thunderbolt3/USB4 stack
- Vectored AIO
- ZSTD Compression in ZFS
Architectures
Userland Programs
Ports
- KDE on FreeBSD
- FreeBSD Office team
- Ports On Non-x86 Architectures
- Python 2.7 removal from Ports
- Xfce on FreeBSD
Documentation
Miscellaneous
Third-Party Projects
- FreeBSD Aarch64 under VMWare ESXi-ARM Fling
- Bastille
- CheriBSD
- Embedded Lab Project
- FreshPorts
- helloSystem
- K8S-bhyve
- Puppet
FreeBSD Team Reports
Entries from the various official and semi-official teams, as found in the Administration Page.
FreeBSD Foundation
Contact: Deb Goodkin <[email protected]>
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 software development projects, conferences and developer summits, and provide travel grants to FreeBSD contributors. The Foundation purchases and supports hardware to improve and maintain FreeBSD infrastructure and provides resources to improve security, quality assurance, and release engineering efforts; publishes marketing material to promote, educate, and advocate for the FreeBSD Project; facilitates collaboration between commercial vendors and FreeBSD developers; and finally, 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:
COVID-19 Impact to the Foundation
Like most organizations, we transitioned all of our staff to work from home. We also put a temporary ban on travel for staff members, which didn't affect our output too much, since most conferences went virtual. We continued supporting the community and Project, even though some of our work and responses may have been delayed because of changes in some of our priorities and the impact of limited childcare for a few of our staff members.
Partnerships and Commercial User Support
We help facilitate collaboration between commercial users and FreeBSD developers. We also meet with companies to discuss their needs and bring that information back to the Project. Not surprisingly, the stay at home orders, combined with our company ban on travel during Q4 made in-person meetings non-existent. However, the team was able to continue meeting with our partners and commercial users virtually. These meetings help us understand some of the applications where FreeBSD is used.
An event we help plan and organize, that helps with vendor/developer engagement, is the annual Bay Area Vendor Summit. We weren't going to let a pandemic stop us from holding this invaluable yearly event, so we went virtual! From the feedback we received from the vendor community on how we should run this, so it would be beneficial for them, we decided to hold this over 3 half days in November. One unexpected result was that more commercial users from around the world attended. Since a Vendor/Developer Summit is typically invitation only, we opened this up to FreeBSD contributors from around the world to watch the livestream. Because of the success and excitement of this event, we are planning to hold another one around June or July.
Fundraising Efforts
We want to take a moment to say thank you to all the individuals and corporations that stepped up to help fund our efforts last year. As of this writing, we raised $1,235,926, and will have the final tally by mid-January. The companies that gave generous financial contributions include Arm, NetApp, Netflix, Juniper Networks, Beckhoff, VMware, Stormshield, Tarsnap, and Google. We also want to say thank you to the Koum Family Foundation for awarding us a large grant, and to "the employees of Ngnix" who also made generous financial contributions.
We truly appreciate these large contributions, which makes the most impact on how much we can contribute back to the Project. However, it's the individual donations that have the most meaning to us. Those are the folks who are giving because they trust we will invest their personal donations, whether large or small, into improving the operating system and Project. As stewards of your donations, we want to thank you for your trust in us and your commitment to making FreeBSD the best platform for products, education, research, computing, and more.
You'll find out how we used your donations for Q4 in our report, as well as in individual reports throughout this status report.
Though we know this is a Q4 status report, we are excited about our plans for 2021, including growing our software development team! We'll be posting two job descriptions for a Senior Software Developer and Project Coordinator soon.
Please consider making a donation to help us continue and increase our support for FreeBSD in 2021: https://www.FreeBSDfoundation.org/donate/.
We also have the Partnership Program, to provide more benefits for our larger commercial donors. Find out more information at https://www.FreeBSDfoundation.org/FreeBSD-foundation-partnership-program/ and share with your companies!
OS Improvements
The Foundation provided many project grants over the last quarter, and you can read about OpenZFS Zstd support, Linuxulator application compatibility improvements, LLDB target support, test lab infrastructure, and WiFi projects in other entries in this quarterly report.
The Foundation hired six co-op students from the University of Waterloo during the 2020 fall term, as well as one intern. Former co-op student Tiger returned, and new students Yang and Zac joined us for the first time.
Tiger worked on improvements to the code-coverage guided kernel fuzzing tool Syzkaller, adding new system call definitions so that Syzkaller can expand the code it tests. A number of FreeBSD kernel bug fixes have already resulted from this work. Tiger also contributed a number of improvements to the ELF Tool Chain set of binary utilities, and worked on tooling to run tests from other tool suites against ELF Tool Chain.
Zac worked on an improvement to the pkg package management tool, investigating and upstreaming patches for FreeBSD support in FreePBX, and investigating compiler support for addressing the stack clash vulnerability.
Yang investigated and fixed a compilation bug with the kernel's Skein-1024 assembly implementation (used by ZFS), and then a number of projects related to Capsicum: applying Capsicum to sort(1), implementing a Capsicum service to execute utilities, and finally working with developers of the Game of Trees (got) version control system to adapt it for Capsicum support.
Our intern Ka Ho focused on improving the desktop experience of the FreeBSD. He fixed and improved many items of OBS (Open Broadcaster Software) on FreeBSD, worked on FreeBSD native audio support on Firefox, adding a facility that user-space audio programs could make use of to enumerate a list of audio devices. He also ported the fcitx5 input method framework.
The five Foundation staff members continued contributions in 2020 in both ongoing operational tasks (including the Git working group and security team) and software development for a number of projects.
Staff members responded to reported security vulnerabilities and release errata, prepared patches, and participated in the security advisory process. We also worked on proactive security vulnerability mitigations. Syzkaller also provided many reports of kernel issues that resulted in Foundation-sponsored bug fixes. We worked on several issues relating to FreeBSD/arm64 to move it along the path of being a Tier-1 architecture.
We participated in code reviews and supported community members in integrating changes into FreeBSD, and triaged incoming bug reports.
We contributed enhancements to many kernel and userland subsystems, including the x86 pmap layer, ELF run-time linker and kernel loader, the Capsicum sandboxing framework and Casper services, the threading library, some RISC-V changes, the build system, tool chain and freebsd-update, network stack stability improvements, machine-dependent optimizations, new kernel interfaces, DTrace bug fixes, documentation improvements, and others.
Continuous Integration and Quality Assurance
The Foundation provides a full-time staff member and funds projects on improving continuous integration, automated testing, and overall quality assurance efforts for the FreeBSD Project.
During the fourth quarter of 2020, Foundation staff continued improving and monitoring the Project's CI infrastructure, and working with experts to fix the failing builds and the regressions found by tests. The work was focused on pre-commit tests and development of the CI staging environment. The other main working item is working on the VCS migration to change the src and doc source from Subversion to Git. There are also many work-in-progress tasks like analysis and improve the tests of non-x86 platforms.
See the FreeBSD CI section of this report for completed work items and detailed information.
Supporting FreeBSD Infrastructure
The Foundation provides hardware and support to improve the FreeBSD infrastructure. Last quarter, we continued supporting FreeBSD hardware located around the world. We coordinated efforts between the new NYI Chicago facility and clusteradm to start working on getting the facility prepared for some of the new FreeBSD hardware we are planning on purchasing. NYI generously provides this for free to the Project. We also worked on connecting with the new owners of the NYI Bridgewater site, where most of the existing FreeBSD infrastructure is located.
Some of the purchases we made for the Project last quarter to support infrastructure includes:
5 application servers to run tasks like bugzilla, wiki, website, cgi, Phabricator, host git, etc.
1 server to replace the old pkg server, which will provide a lot more IOPS to avoid the slowdowns seen during peak times of the day where the disks simply cannot keep up with the request volume.
1 server for exp-runs and to make them faster.
1 server to build packages more frequently.
FreeBSD Advocacy and Education
A large part of our efforts are dedicated to advocating for the Project. This includes promoting work being done by others with FreeBSD; producing advocacy literature to teach people about FreeBSD and help make the path to starting using FreeBSD or contributing to the Project easier; and attending and getting other FreeBSD contributors to volunteer to run FreeBSD events, staff FreeBSD tables, and give FreeBSD presentations.
The FreeBSD Foundation sponsors many conferences, events, and summits around the globe. These events can be BSD-related, open source, or technology events geared towards underrepresented groups. We support the FreeBSD-focused events to help provide a venue for sharing knowledge, to work together on projects, and to facilitate collaboration between developers and commercial users. This all helps provide a healthy ecosystem. We support the non-FreeBSD events to promote and raise awareness of FreeBSD, to increase the use of FreeBSD in different applications, and to recruit more contributors to the Project.
While we were still unable to attend in-person meetings due to COVID-19, we were able to attend virtual events at new venues and facilitate the first online FreeBSD Vendor Summit. In addition to attending and planning virtual events, we are continually working on new training initiatives and updating our selection of how-to guides to facilitate getting more folks to try out FreeBSD.
Check out some of the advocacy and education work we did last quarter:
Continued our FreeBSD Fridays series of 101 classes. Topics included an Introduction to Capsicum, Introduction to Bhyve, Introduction to DTrace, and more. Videos of the past sessions can be found here. We'll be back with new sessions in early 2021.
Gave a FreeBSD talk at the nerdear.la conference on October 20th.
Participated in the podcast: What the Dev: A Dive into the FreeBSD Foundation on its 20th Anniversary
Promoted the Foundation's 20th Anniversary in the FossBytes article: 20 Years of The FreeBSD Foundation
Continued to promote the FreeBSD Office Hours series. Videos from the one hour sessions can be found on the Project's YouTube Channel. See the Office Hours section of this report for more information.
Added two new How-To Guides: Contributing FreeBSD Documentation and How to Submit a Bug Report.
Worked with the organizing committee to host the November 2020 Vendor Summit
Promoted the use of FreeBSD in regards to CHERI and ARM's Morello Processor
Authored a Beginners Guide to FreeBSD for Fosslife.
Sponsored All Things Open as a Media Sponsor.
Sponsored OpenZFS Developers Summit at the Bronze level.
Applied for a virtual stand at FOSDEM 2021.
Committed to attend the online Apricot 2021.
Keep up to date with our latest work in our newsletters: https://www.freebsdfoundation.org/news-and-events/newsletter/
Netflix provided an update on how and why they use FreeBSD in our latest Contributor Case Study.
We help educate the world about FreeBSD by publishing the professionally produced FreeBSD Journal. As we mentioned previously, the FreeBSD Journal is now a free publication. Find out more and access the latest issues at https://www.FreeBSDfoundation.org/journal/
You can find out more about events we attended and upcoming events at https://www.FreeBSDfoundation.org/news-and-events/.
Legal/FreeBSD IP
The Foundation owns the FreeBSD trademarks, and it is our responsibility to protect them. We also provide legal support for the core team to investigate questions that arise.
Go to http://www.FreeBSDfoundation.org to find out how we support FreeBSD and how we can help you!
FreeBSD Release Engineering Team
Contact: FreeBSD Release Engineering Team <[email protected]>
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 fourth quarter of 2020, the Release Engineering Team completed work on 12.2-RELEASE, the third release from the stable/12 branch, released on October 27. Thank you to all involved for the hard work that went into this release.
Additionally throughout the quarter, several development snapshots builds were released for the head, stable/12, and stable/11 branches. Development snapshot builds for 13.0-CURRENT have recently been built from the Git tree within the project, while further snapshot builds for 12.x and 11.x will continue to be built from Subversion. As we approach the end of 2020, continued preparations are being put in place for the upcoming 13.0 release, which will be the first release from Git.
Much of this work was sponsored by Rubicon Communications, LLC (netgate.com) and the FreeBSD Foundation.
Cluster Administration Team
| Links | |
| Cluster Administration Team members | URL: https://www.freebsd.org/administration.html#t-clusteradm |
Contact: Cluster Administration Team <[email protected]>
The FreeBSD Cluster Administration Team consists of the people responsible for administering the machines that the Project relies on for its distributed work and communications to be synchronised. In this quarter, the team has worked on the following:
Finished setting up the Malaysia mirror site, generously hosted by the Malaysian Research & Education Network. Traffic from Oceania and parts of Asia is now going to this mirror instead of farther away sites like Japan and California.
Upgraded the package building machines to a version of head from mid-October 2020.
Upgraded developer machines in the cluster (freefall, ref* and universe*) to a version of head from mid-October 2020.
Installed eight new x86 servers in our New Jersey site: five application servers, two package builders and one mirror server.
The new mirror server is in production (pkg0.nyi.freebsd.org).
The two package builders are in production.
Two of the application servers have been put into production as the Git source of truth and the cgit web frontend, respectively.
Installed two new aarch64 servers in our New Jersey site. Both are now building aarch64 packages.
Fixed package mirror synchronisation for powerpc64 packages.
Rebuilt the ZFS pool on the UK mirror server (pkg0.bme.freebsd.org) for better I/O parallelism. This should improve download performance especially at peak times.
Ongoing systems administration work:
Accounts management for committers.
Backups of critical infrastructure.
Keeping up with security updates in 3rd party software.
Hardware refreshing for web services, backup version control system in NYI
Upgrading production machines in the FreeBSD cluster to 12.2
Most machines have been upgraded as of mid-December 2020
Remaining machines will be decommissioned / repurposed after services migrate to newer hardware
Supporting Git migration and infrastructure setup
powerpc pkgbuilder/ref/universal machines
Preparations for a new mirror site in Australia, to be hosted by IX Australia.
Setup Brazil (BRA) mirror.
Review the service jails and service administrators operation.
Searching for more providers that can fit the requirements for a generic mirrored layout or a tiny mirror.
Continuous Integration
Contact: Jenkins Admin <[email protected]>
Contact: Li-Wen Hsu <[email protected]>
Contact: freebsd-testing Mailing List
Contact: IRC #freebsd-ci channel on EFNet
The FreeBSD CI team maintains the continuous integration system of the FreeBSD project. The CI system firstly checks the committed changes can be successfully built, then performs various tests and analysis over the newly built results. The artifacts from those builds are archived in the artifact server for further testing and debugging needs. The CI team members examine the failing builds and unstable tests and work with the experts in that area to fix the code or adjust test infrastructure. The details of these efforts are available in the weekly CI reports.
During the fourth quarter of 2020, we continued working with the contributors and developers in the project to fulfil their testing needs and also keep collaborating with external projects and companies to improve their products and FreeBSD.
Important changes:
doc jobs were changed to use git to follow VCS migration:
https://ci.freebsd.org/job/FreeBSD-doc-main-igor/ Thanks Brandon Bergren (bdragon@)
head and stable/12 build environment have been upgraded to 12.2-RELEASE
Follow VCS migration, change src jobs to use Git - PRs are available Thanks Brandon Bergren (bdragon@)
Collecting and sorting CI tasks and ideas here
Testing and merging pull requests in the the FreeBSD-ci repo
Designing and implementing pre-commit CI building and testing
Reducing the procedures of CI/test environment setting up for contributors and developers
Setting up the CI stage environment and putting the experimental jobs on it
Setting up public network access for the VM guest running tests
Implementing automatic tests on bare metal hardware
Adding drm ports building tests against -CURRENT
Planning to run ztest and network stack tests
Adding more external toolchain related jobs
Improving the hardware lab to be more mature and adding more hardware
Helping more software get FreeBSD support in their CI pipeline Wiki pages: 3rdPartySoftwareCI, HostedCI
Working with hosted CI providers to have better FreeBSD support
The build and test results will be sent to the dev-ci mailing list soon. Feedback and help with analysis is very appreciated!
Sponsor: The FreeBSD Foundation
Ports Collection
Contact: Ren� Ladan <[email protected]>
Contact: FreeBSD Ports Management Team <[email protected]>
The Ports Management Team is responsible for overseeing the overall direction of the Ports Tree, building packages, and personnel matters. Below is what happened in the last quarter.
For the last quarter the dashboard looks like:
41500 ports (including flavors)
2516 open PRs of which 625 are unassigned
8715 commits to the HEAD branch by 164 committers
420 commits to the 2020Q4 branch by 59 committers
were slightly fewer commits by the same number of people. The number of ports again grew steadily, this time by almost 4 percent.
During the last quarter, we welcomed Juray Lutter (otis@) as a new ports committer and said goodbye to cpm, jadawin, knu, araujo, mmokhi and scottl.
Traditionally merges to the quarterly ports branches, which are more conservative versions of the HEAD tree, required approval of either the Ports Security Team (ports-secteam@) or portgmr@. There were already a number of blanket approvals for tested commits, ranging from fixing typing mistakes to upgrading web browsers to their latest version. As of last December, all ports committers are free to merge on their own, lessening the burden on ports-secteam@.
Patent limitations have been disconnected from the license framework, given that patents are a complex topic with implications varying from one jurisdiction to another.
The last quarter saw a number of updates to default versions of ports:
librsvg2: "rust" on supported platforms, "legacy" otherwise
Mono: 5.10
FPC switched to 3.2.0
GCC switched to 10 for powerpc64le
Lazarus switched to 2.0.10
Ruby switched to 2.7.X
Samba switched to 4.12
that for instance help the user to learn about a certain topic or help facilitating examinations.
The @shell and @sample keywords have been rewritten in Lua which makes root-dir compliant (see pkg -r) and ensures they are Capsicum-sandboxed.
The last quarter also saw updates to several user-facing ports:
Firefox 84.0.1
Firefox-esr 78.6.0
Chromium 87.0.4280.88
Ruby 2.7.2
Qt5 5.15.2
XFce 4.16
various ports upgrades
changing sys/cdefs.h in base
adding "set pipefail" to most framework scripts to catch errors earlier
changing the default locale to C.UTF-8 in base
using bsdgrep as /usr/bin/grep
Office Hours
Contact: Allan Jude <[email protected]>
Contact: Ed Maste <[email protected]>
During the final quarter of 2020 three office hours sessions were held.
The first was hosted by the core team in a time slot conducive to Asia and Australia, covering topics including the transition to git, recruiting for project teams, and core's todo list.
The second was hosted by the git transition team, and answered attendee questions about the transition to git and how it would impact the project's workflows.
The third session was hosted by bhyve maintainers Peter Grehan and John Baldwin to present recent development efforts and answer questions about bhyve.
The project is looking for volunteers to host future office hours sessions, as well as taking topic suggestions. We also hope to improve the system to allow people to submit questions ahead of time, so that we can take maximum advantage of subject matter experts when we have them for these calls.
You can find the schedule for future office hours, and videos of past office hours on the FreeBSD Wiki
Sponsor: ScaleEngine Inc.
Projects
Projects that span multiple categories, from the kernel and userspace to the Ports Collection or external projects.
GPL in Base
| Links | |
| GPL Software in the Base System | URL: https://wiki.freebsd.org/GPLinBase |
Contact: Ed Maste <[email protected]>
Contact: Kyle Evans <[email protected]>
Contact: Baptiste Daroussin <[email protected]>
A long-standing goal of the FreeBSD project is for the base system to migrate to modern, copyfree or more permissively licensed components. In this quarter, the following components have been successfully removed or replaced:
gdb (removed in favor of lldb in base or devel/gdb in ports)
gnugrep (replaced with bsdgrep)
libgnuregex (removed)
may be listed on the above-linked wiki page. Interested parties are welcome to evaluate the options to restart the discussion:
dialog
gcov (kernel)
Note that partial completion likely means that a component is partially compatible, but could use evaluation and patches to bring parity with the component that it is replacing.
diff3 (Contact bapt@ if interested)
Git Migration Working Group
Contact: Li-Wen Hsu <[email protected]>
Contact: Warner Losh <[email protected]>
Contact: Ed Maste <[email protected]>
Contact: Ulrich Sp�rlein <[email protected]>
The Git working group largely completed the migration of the doc and src (base system) trees from Subversion to Git in December 2020. We are currently working on some minor outstanding issues and preparing for the ports tree migration.
We set up new hosts to serve as the Git repositories and mirrors, and developed commit hooks for restrictions on commits to various branches, generation of commit mail, and similar needs.
The doc tree migration occurred on December 8th and 9th. After the conversion some minor changes to the documentation build infrastructure were necessary.
The src tree migration occurred between December 20th and 23rd for the main branch; some additional tasks occurred over the next week or so. These included enabling the stable branches, vendor (contrib) code updates, and the git->svn gateway. We are translating stable branch commits to Subversion for the stable/11 and stable/12 branches and associated release branches. This allows FreeBSD users who follow stable branches or releases to continue using existing processes and tooling.
An experimental Git conversion of the ports tree is available at the link above. There are some unique challenges in the ports tree (that do not impact the doc or src repos in the same way), so additional work is ongoing. The window for migrating the ports tree is immediately prior to a quarterly branch, so we anticipate a migration at the end of March 2021. Over the next few months testing of the experimental ports repo is very welcome.
Process documentation for developer and user interaction with FreeBSD's repositories is currently available in Warner's GitHub repository at the link above. It will be moved to the FreeBSD developer's handbook and/or other suitable locations following the documentation project's asciidoc conversion.
The working group is experimenting with two permissively-licensed tools that are compatible with Git servers or repositories. Game of Trees is a version control system that is compatible with Git repositories. It is being developed by Stefan Sperling along with some OpenBSD developers and other contributions.
John Mehr's gitup is a minimal, dependency-free program that clones and synchronizes a local tree with a remote repository. It is intended for use cases that would otherwise be served by tools like portsnap.
Sponsor: The FreeBSD Foundation (in part)
Linux compatibility layer update
Contact: Edward Tomasz Napierala <[email protected]>
Linuxulator improvements have been ongoing for the last two years, with support from the FreeBSD foundation over a few distinct project grants as well as contributions from the community. The goal of this project is to improve FreeBSD's ability to execute unmodified Linux binaries. Current status is being tracked at Linux app status Wiki page. The work has now shifted from command-line apps to desktop applications.
There wasn't much Foundation-sponsored work done during this quarter, apart from extending fuse(4) to make it possible to run Linux FUSE servers, which is one of the things required to run AppImages. The Foundation-sponsored effort will continue into the first quarter of 2021 in order to make sure the 13.0-RELEASE ships with Linuxulator in a good shape.
There was a very significant contribution from Conrad Meyer in the form
of SO_PASSCRED setsockopt(2) support, PR_SETDUMPABLE and PR_GETDUMPABLE
prctl(2) flags, and also CLONE_FS and CLONE_FILES handling. This,
along with some more cleanups and improvements, leads to working Linux
Chromium; it has been tested with Netflix and Spotify clients. It still
requires three flags (--no-sandbox --no-zygote --in-process-gpu)
to be passed on the command line to work around missing functionality, though. Also,
the name_to_handle_at(2) and open_by_handle_at(2) syscalls are now supported.
There are also much better debug messages for unrecognized socket options.
Sponsor: The FreeBSD Foundation
LLDB Debugger Improvements
Contact: Kamil Rytarowski <[email protected]>
Contact: Michał G�rny <[email protected]>
The LLDB project builds on libraries provided by LLVM and Clang to provide a great modern debugger. It uses the Clang ASTs and the expression parser, LLVM JIT, LLVM disassembler, etc so that it provides an experience that “just works”. It is also blazing fast and more permissively licensed than GDB, the GNU Debugger.
LLDB is the default debugger in Xcode on macOS and supports debugging C, Objective-C, and C++ on the desktop and iOS devices and the simulator.
FreeBSD includes LLDB in the base system. At present, it has some limitations in comparison with the GNU GDB debugger, and does not yet provide a complete replacement. It used to rely on an obsolete plugin model in LLDB that was a growing technical debt. This project aimed to bring LLDB closer to a fully featured replacement for GDB, and therefore for FreeBSD to feature a modern debugger for software developers.
The legacy monolithic target support executed the application being debugged in the same process space as the debugger. The modern LLDB plugin approach, used on other supported targets, executes the target process under a separate lldb-server process. This improves reliability and simplifies the process / thread model in LLDB itself. In addition, remote and local debugging is now performed using the same approach.
After the migration to the new process model on 32 and 64-bit x86 CPUs, the project focused on reviewing the results of LLDB’s test suite and fixing tests as time permits.
During the Moritz Systems work, the FreeBSD Project gained numerous important improvements: in the kernel, userland base libraries (the dynamic loader) and the LLVM toolchain FreeBSD support.
The introduced changes are expected to be shipped with LLDB 12.0, and where applicable in FreeBSD 13.0.
The overall experience of FreeBSD/LLDB developers and advanced users on this rock solid Operating System reached the state known from other environments. Furthermore, the FreeBSD-focused work also resulted in generic improvements, enhancing the LLDB support for Linux and NetBSD.
Sponsor: The FreeBSD Foundation
Upstreaming NetApp Changes
| Links | |
| Klara Inc. | URL: https://klarasystems.com/freebsd-development/ |
Contact: Alexander Sideropoulos <[email protected]>
Contact: Allan Jude <[email protected]>
NetApp has started an effort to upstream bug fixes and other improvements from the ONTAP code line into FreeBSD. These changes benefit the FreeBSD community by providing many fixes that NetApp has made over the past few years, while allowing NetApp to reduce the number of customizations needed when bringing in the latest FreeBSD changes back into the ONTAP tree.
NetApp has partnered with Klara to facilitate this project, to help identify interesting and useful changes to send upstream, to rework and generalize those changes as required to make them suitable for upstreaming, and to shepherd them through the FreeBSD code review process.
During the fourth quarter, Klara has made 40 upstream fixes in the FreeBSD kernel in various subsystems including geom, dev, amd64, net, kern, netinet, and several other areas of the tree on behalf of NetApp.
NetApp intends to continue to sponsor this effort throughout 2021.
Sponsor: NetApp
NFS over TLS implementation
Contact: Rick Macklem <[email protected]>
In an effort to improve NFS security, an Internet Draft titled "Towards Remote Procedure Call Encryption By Default" specifies use of TLS 1.3 to encrypt all data traffic on a Sun RPC connection used for NFS.
Although NFS has been able to use sec=krb5p to encrypt data on the wire, this requires a Kerberos environment and, as such, has not been widely adopted. It also required that encryption/decryption be done in software, since only the RPC message NFS arguments are encrypted. Since Kernel TLS is capable of using hardware assist to improve performance and does not require Kerberos, NFS over TLS may be more widely adopted, once implementations are available.
The coding for this project has now been completed. All required changes to the NFS and kernel RPC code have been committed to the head/current kernel and will be in FreeBSD13. The daemons can now be built from a port that depends upon the security/openssl-devel port of Openssl3 that includes patches for support of ktls. The port for the daemons is called sysutils/nfs-over-tls and should be committed to the ports framework soon. In the meantime, the port can easily be fetched, as described in https://people.freebsd.org/~rmacklem/nfs-over-tls-setup.txt.
To support clients such as laptops, the daemons that perform the TLS handshake may optionally handle client X.509 certificates from a site local CA. There are now exports(5) options to require client(s) to provide a valid X.509 certificate. The case where a "user" name is stored in the certificate and is used to map all RPC credentials to that user is probably in violation of the Internet Draft. This is only enabled when the "-u" command line option is provided to rpc.tlsservd(8).
The code is now available for testing. See: https://people.freebsd.org/~rmacklem/nfs-over-tls-setup.txt
Setting up system(s) for testing still requires building a custom kernel with "options KERN_TLS" from recent head/FreeBSD13 sources plus installing the port for the daemons, as explained by the above document.
The main limitation in the current implementation is that it uses TLS1.2 and not TLS1.3, as required by the Internet Draft. This should change once the KERN_TLS rx patch includes TLS1.3 support.
Third party testing would be appreciated.
OpenBSM Synchronisation
Contact: Gordon Bergling <[email protected]>
OpenBSM is a crucial part of FreeBSD, which provides auditing features for the operating system. OpenBSM is incorporated into FreeBSD and macOS. Both Apple and FreeBSD have currently made changes to the OpenBSM framework, which weren't upstreamed. This small project aims to consolidate these changes and upstream them to the OpenBSM github repository, so that both development efforts can be merged to FreeBSD later on. The tricky part of this project is the manual comparison, since Apple doesn't provide any changelogs.
I am currently working on on the macOS Catalina sources and hopefully Apple will release the sources of macOS Big Sur in time for FreeBSD 13.
Tool Chain
| Links | |
| ELF Tool Chain homepage | URL: https://sourceforge.net/p/elftoolchain |
Contact: Dimitry Andric <[email protected]>
Contact: Ed Maste <[email protected]>
In October Clang/LLVM was updated to 11.0.0, followed by a number of bug fixes
from upstream, including improvements for a number of Tier-2 architectures.
We also enabled the -fstack-clash-protection flag to enable compiler
mitigation for the "stack clash" vulnerability and are coordinating with
upstream.
Upstream LLDB support for FreeBSD improved substantially over the last quarter, as detailed elsewhere in this report. These improvements will make it into the FreeBSD base system early in 2021 when LLVM is next updated to 12.0. As also mentioned elsewhere, we removed the obsolete copy of GDB 6.1.1.
The ELF Tool Chain received a number of bug fixes, as well as support for
readelf -z (handling compressed ELF debug sections) and an improvement
to addr2line to report based on labels when other debug information is not
available. We are working to upstream these changes to the ELF Tool Chain
project.
There are a number of open issues and opportunities for improvements in various ELF Tool Chain components. Contributions in these areas are very welcome,
Sponsor: The FreeBSD Foundation (in part)
Kernel
Updates to kernel subsystems/features, driver support, filesystems, and more.
ENA FreeBSD Driver Update
| Links | |
| ENA README | URL: https://github.com/amzn/amzn-drivers/blob/master/kernel/fbsd/ena/README |
Contact: Michal Krawczyk <[email protected]>
Contact: Artur Rojek <[email protected]>
Contact: Marcin Wojtas <[email protected]>
ENA (Elastic Network Adapter) is the smart NIC available in the virtualized environment of Amazon Web Services (AWS). The ENA driver supports multiple transmit and receive queues and can handle up to 100 Gb/s of network traffic, depending on the instance type on which it is used.
Completed since the last update:
MFC of the ENA v2.3.0 driver to the FreeBSD 11-STABLE branch
MFC of the ENA v2.3.0 driver to the upcoming FreeBSD 12-STABLE branch
Add feature that allows reading extra ENI (Elastic Network Interface) metrics about exceeding BW/pps limits
Add SPDX license tag to the ENA driver files
Add Rx offsets (hardware feature) support for the ENA driver
Fix completion descriptors alignment for the ENA device - on some of the platforms ENA needs alignment to 4k
Introduce full kernel RSS API support.
Allow reconfiguration of the RSS indirection table and hash key
Prototype the driver port to the iflib framework
Intel wireless update
| Links | |
| The freebsd-wireless mailing list | URL: https://lists.freebsd.org/mailman/listinfo/freebsd-wireless |
Contact: Bjoern A. Zeeb <[email protected]>
The Intel Wireless driver update project aims to bring support for newer chipsets and also get station side to 11ac in a first step.
During the last months connection code between net80211 and the Linux driver KPI was implemented and scanning is working. Currently the focus is on sending and driving one state machine from the other and syncing state between net80211 and the Linux compat code.
In addition the driver and firmware was updated from upstream sources to include support for the AX210 hardware generation, which was already tested to attach.
The hope is that by the time the status report gets published authentication and association are working and basic data packet passing will work soon.
Sponsor: The FreeBSD Foundation
Fenestras X random(4)
Contact: Conrad Meyer <[email protected]>
Contact: FreeBSD CSPRNG group <[email protected]>
Since FreeBSD 11, the default random(4) implementation is based on the
Fortuna (2003) design by Ferguson and Schneier. At a high level, Fortuna
accumulates entropy into a series of pools, and reseeds a single generator from
some of these pools according to some criteria.
In 2019, Ferguson (at Microsoft) published a whitepaper on the design of the
Windows 10 system random number generator. Fenestras X is a random(4)
implementation based on the published Windows 10 design.
The Fenestras X / Windows 10 design is similar to Fortuna, so it is probably most interesting to describe their differences:
Fenestras X has per-CPU generators seeded from a root generator. Fortuna only has the root generator. This change eliminates lock contention between
random(4)readers running on multiple cores.Generators in Fenestras X form a tree from the root RNG. When read, generators efficiently check if their parent generator has been seeded with newer entropy. If so, child generators reseed themselves before serving the read operation. This is integrated with
arc4random(9), as well as userspacearc4random(3).Fenestras X generators are buffered. Requests smaller than some arbitrary threshold (currently 128 bytes) are served from the buffer. Bytes read from the buffer are securely erased when they are consumed. The buffer is refreshed if the request consumes more bytes than were available in the buffer. This amortizes the cost of rekeying and generating output from a cryptographic CTR-mode cipher, which is especially slow with AES.
should read Ferguson's whitepaper. It is short and accessible. For more
information on the FreeBSD implementation, please see the SVN commit messages
— especially r366620.
The Fenestras X implementation is available in CURRENT, but disabled by
default. (The default remains Fortuna.) At this time, you must set the
RANDOM_FENESTRASX option in your custom kernel configuration and rebuild your
kernel to use the new design. There are no known bugs or weaknesses relative
to the Fortuna implementation.
Future work and call to action:
Additional design review, implementation review, and testing is welcome.
Additional entropy sources: we could use implementations of some of the sources described in the whitepaper, in both Fortuna and Fenestras X. In particular, we're missing a jitter entropy source.
pf performance improvement
Contact: Kristof Provost <[email protected]>
The performance of pf was not as good as it could be. Some investigation with the invaluable hwpmc tooling eventually pointed to very poor cache behaviour. The longest_lat_cache.miss event was very informative.
This turned out to be due to pf doing packet and byte counting in states, rules and interfaces.
The pf code took the very straightforward approach of having a simple uint64_t variable and incrementing it for every packet. The downside of this is that when multiple cores do it simultaneously the CPU ends up having to write this to memory very often, slowing packet processing down greatly. Happily the counter(9) framework is designed for this exact situation.
One additional complication is that pf uses the same structure definitions for its internal data as it uses for configuration from user space. To avoid breaking user space these data structures have been decoupled. That is, where pf_rule used to be used both to set rules via the ioctl() interface and to evaluate rules while processing packets we now only use pf_rule for configuration. The new pf_krule structure is used when evaluating packets. This allows us to change the pf_krule structure, to change uint64_t to counter_u64_t, without affecting user space.
Olivier Cochard-Labb� tested the full set of changes, and found (depending on hardware) substantial improvements in throughput:
Sponsor: Orange Business Services
IP Routing lookup improvements
| Links | |
| Add modular routing lookup framework. | URL: https://reviews.freebsd.org/D27401 |
Contact: Alexander Chernikov <[email protected]>
This work adds a fib lookup framework, allowing to attach custom IP lookup algorithms to any routing table on the fly. It allows to use more performant and efficient lookup algorithms, dynamically selected based on the number of routes in the routing table. Finally, it provides an implementation of modified DIR-24-8 for IPv4/IPv6, speeding IP lookups for the large-fib use case.
This work is a part of a larger effort to modernise the routing subsystem.
Background
FreeBSD runs diverse workloads on both low-end and high-end devices, resulting in different networking/memory requirements for each case. Small boxes with a couple of routes are different from routers with full-view. IPv4 lookups are different from IPv6 ones. Conditions can change dynamically: one may easily reconfigure a system to receive full view instead of a default route.
Currently, FreeBSD uses radix (compressed binary tree) to perform all unicast route manipulations, including routing lookups. Radix implementation requires storing key length in each item, allowing to use sockaddrs, transparently supporting virtually any address family. This flexibility comes at a cost: radix is relatively slow, cache-unfriendly and adds locking to the hot path. Finally, radix is closely coupled to the rest of the system, making it hard to switch to something else.
Implementation overview
Overview
Modular fib IP lookup framework has been designed to address flexibility and performance requirements.
It keeps system radix as the "control plane" source of truth, simplifying actual algorithms implementation. It allows dynamic load new algorithms as the kernel modules and abstracts most OS-specific details, reducing algorithm "glue" code. It automatically adapts to the current system state by picking the best matching algorithm for the routing table on-the-fly.
The following algorithms are provided by default.
IPv4:
bsearch4 (lockless binary search in a specially-crafted IP array), tailored for small-fib (less than 16 routes)
radix4_lockless (lockless immutable radix, re-created on every routing table change), tailored for small-fib (less than 1000 routes)
radix4 (base system radix backend)
dpdk_lpm4 (DPDK DIR24-8-based lookups), lockless datastructure optimised for large-fib ( D27412 )
radix6_lockless: lockless immutable radix, re-created on every routing table change, tailored for small-fib (less than 1000 routes)
radix6: wrapper around existing system radix
dpdk_lpm6: DPDK DIR24-8-based lookups, lockless datastructure optimised for large-fib ( D27412 )
Performance changes
Micro benchmarks (i7-7660U, single-core lookups, 2048 destinations, benchmark code in D27604).
IPv4:
8 routes: radix4: ~20mpps, radix4_lockless: ~25mpps, bsearch4: ~69mpps, dpdk_lpm4: ~67 mpps
700k routes: radix4_lockless: 3.3mpps, dpdk_lpm4: 46mpps
8 routes: radix6_lockless: ~20mpps, dpdk_lpm6: ~70mpps
100k routes: radix6_lockless: ~14mpps, dpdk_lpm6: ~57mpps
+10-15% IPv4: small-fib, bsearch4
+25% IPv4: full-view, dpdk_lpm4
+20% IPv6: full-view, dpdk_lpm6
Status
Modular longest-prefix-match lookup algorithms (D27401) [ DONE ]
Scalable routing multipath support
| Links | |
| Implementation of scalable multipath | URL: https://reviews.freebsd.org/D24141#change-ZOjdMqgDgUr7 |
| Introduce scalable route multipath | URL: https://reviews.freebsd.org/D26449 |
Contact: Alexander Chernikov <[email protected]>
This work targets implementing scalable routing multipath support and enabling it by default. It closes the long-standing feature gap with other modern networking OSes.
This work is a part of on-going efforts to modernize the routing subsystem.
Background
Initial FreeBSD multipath implementation, RADIX_MPATH, was added back in 2008. It was based on the radix changes and represented multipath routes as a linked-list of chained paths. It was not fully finished and tested, resulting in many crash reports.
Implementation overview
Multipath-related change changes are based on the introduction of the concept of next hops. Nexthops are separate data structures, containing the necessary information to perform packet forwarding. They are shared among the routes, providing more pre-computed cache-efficient data while requiring less memory. Interested readers can find a more detailed description in D24141. They can find another overview in Nexthop objects talk describing Linux kernel implementation.
Multipath implementation extends the nexthop concept further by introducing nexthop groups. Nexthop group is simply an array of nexthops, compiled according to each nexthop relative weight.
Each route has a pointer to either nexthops or a nexthop group, decoupling lookup algorithm from the routing stack internals. Both nexthops and nexthop groups are immutable and use epoch(9)-backed reclamation.
Status
Nexthop objects (D24232) [ DONE ]
Introduction of nexthop objects [ DONE ]
Conversion of old KPI users to the new one [ DONE ]
Conversion of route caching to nexthop caching [ DONE ]
Conversion of struct
rtentryfield access to nhop field access [ DONE ]Eliminating old lookup KPI and hiding struct rtentry [ DONE ]
Multipath routing (D26449) [ DONE ]
Switch control plane customers to use (rtentry, nexthop) pairs instead of rtentry to allow multipath changes happen transparently [ DONE ]
Introduce nexthop group objects [ DONE ]
Add multipath support for the rib (routing information base) manipulation functions [ DONE ]
Add flowid generation for outbound traffic to enable load balancing [ DONE ]
Routing daemon support
Add net/bird support for multipath routing [ NOT STARTED ]
Add explicit nexthop/nexthop groups control via rtsock [ IN PROGRESS ]
Work with FRR developers to add nexthop-based route control [ NOT STARTED ]
Thunderbolt3/USB4 stack
Contact: Scott Long <[email protected]>
This project implements a driver stack for Thunderbolt3 and USB4. These technologies differ radically from USB3 and prior, and require completely new drivers for the host interface adapter and topology as well as configuration management layers. At their most fundamental level, a TBT3/USB4 topology appears as PCI bridges and buses, and attached devices appear as either PCI devices, USB3 devices, or DisplayPort devices. Early TBT3 controllers don't even appear in the system topology unless a TBT3 device is plugged in. These early TBT3 systems also implement a security policy meant to protect against unauthorised or malicious devices, though that scheme has been proven to not be effective and has been removed from later TBT3 and USB4 implementations. Besides security control, the TBT3/USB4 stack controls power management and topology hotplug.
The FreeBSD driver currently supports Alpine Ridge and Ice Lake TBT3 controllers, and can perform basic security validation and topology awareness. USB4 support as well as full connection manager and power management support is still being worked on. The current driver will be committed to FreeBSD in early January 2021.
Though this work is not sponsored, it has been done with the encouragement and support of the FreeBSD Foundation and Netgate.
Vectored AIO
Contact: Alan Somers <[email protected]>
POSIX AIO is a facility for asynchronous I/O to files and devices. FreeBSD's
implementation is efficient, especially when writing to disk files. But a
long-standing defect in the standard API is a lack of vectored functions. That
is, there is no asynchronous equivalent of pwritev(2) and preadv(2). A
common workaround is to use lio_listio(2) instead. However, that has several
drawbacks. It's more effort for the programmer, it might return early with
only a subset of the operations completed, it requires more total syscalls, and
there is no guarantee that the operations will complete in-order.
This quarter I added two new syscalls: aio_writev(2) and aio_readv(2).
They work just like their non-vectored counterparts, but they take an array of
iovec elements, just like pwritev and preadv. You can't use them in
combination with lio_listio, but that could be added in the future.
ZSTD Compression in ZFS
Contact: Allan Jude <[email protected]>
Zstandard (ZSTD) is a modern high-performance compression algorithm designed to provide the compression ratios of gzip while offering much better performance. ZSTD has been adopted in FreeBSD for a number of other uses, including compressing kernel crash dumps, as a replacement for gzip or bzip for compressing log files, and for future versions of pkg(8).
This effort to complete the integration of ZSTD into ZFS is funded by the FreeBSD Foundation.
During the four quarter the final tasks in the project to integrate ZSTD into OpenZFS were completed.
Completed milestones in this project:
Integrated ZSTD in the FreeBSD boot loader (Warner Losh [email protected])
Added a section to the FreeBSD Handbook ZFS chapter explaining ZSTD
Wrote a FreeBSD Journal Article explaining considerations when selecting a suitable compression level
Monitored for bug reports after the changes were integrated into -CURRENT
compressed with zstd or zstd-fast. For comparison, a standard installation of FreeBSD 13 (without debug symbols) uncompressed is 1175 MB, and when compressed with LZ4, is only 570 MB (2.15x) but when compressed with ZSTD's default level of 3 is only 417 MB (3.00x), and with the maximum level, 19, only 374 MB (3.36x).
Sponsor: The FreeBSD Foundation
Architectures
Updating platform-specific features and bringing in support for new hardware platforms.
arm64 platform updatesq
Contact: Mitchell Horne <[email protected]>
In the interest of seeing the arm64 architecture promoted to Tier-1 status, an
effort was undertaken to test building and serving of release and patch-level
updates via freebsd-update(1). The conclusion of this investigation is that
the process works with very few changes required; a small tweak is needed for
the update build scripts, and a minor bugfix in the bsdiff(1) utility was
committed. The hope is that the project can begin providing security updates for
the platform with the release of FreeBSD 13.0, removing the requirement that
users compile these updates from source.
Added this quarter was arm64 support for the new ossl(4) crypto driver. This
driver provides acceleration of SHA-1 and SHA-2 cryptographic operations by
leveraging OpenSSL's assembly routines. These routines will detect and use
optimized instructions, as supported by the CPU. This support benefits userland
applications via the cryptodev(4) device, and in-kernel consumers of the
crypto(9) interface, such as the IPSec Authentication Header protocol and
kernel TLS.
Finally, work was done to add the necessary machine-dependent bits for the
kernel's gdb(4) interface. This enables remote debugging of the kernel with
gdb(1) over a serial line.
Sponsor: The FreeBSD Foundation
FreeBSD/RISC-V Project
| Links | |
| Wiki | URL: https://wiki.freebsd.org/riscv |
Contact: Mitchell Horne <[email protected]>
Contact: freebsd-riscv Mailing List
Contact: IRC #freebsd-riscv on freenode
The FreeBSD/RISC-V project is providing support for running FreeBSD on the RISC-V Instruction Set Architecture.
This quarter saw a number of improvements and bugfixes from committers and contributors alike. A few small items from this quarter:
Added riscv64 LINT kernel config plus CI job (
FreeBSD-head-riscv64-LINT)Switched
emulators/riscv-isa-simto official upstream and updated to 2020-11-02 snapshotCreated
sysutils/u-boot-sifive-fu540, a u-boot port for the HiFive UnleashedImproved SBI extension support
runtime issues with large dependencies devel/python-setuptools and
devel/glib20 were fixed, enabling several thousand skipped ports. There is
some in-progress work to address failures in other significant ports, such as
devel/nspr and databases/sqlite3. By addressing some of these small-effort
issues, some 15000+ ports can now be built for the platform with
qemu-user-static.
Finally, December saw the arrival of the first riscv64 weekly development
snapshots. This includes the usual memstick installer, a virtual machine image,
and a generic SD card image. There are still some minor tweaks to be made, but
this marks a significant step forward for the platform, and lowers the barrier
of entry for running a FreeBSD/RISC-V system. This also means that FreeBSD 13
will likely be the first downloadable release for the architecture. For those
interested in trying out the VM image for themselves, see
Userland Programs
Changes affecting the base system and programs in it.
Dual-stack ping command
Contact: Alan Somers <[email protected]>
The venerable ping command has until now only supported IPv4. A separate utility, ping6, was originally written by WIDE as a research tool to develop IPv6. As a research tool, it didn't need IPv4 support, but since then, it's been put to practical use by countless developers and sysadmins everywhere.
The ping/ping6 split has been a perennial complaint. It's annoying that two separate commands are needed, even though they do almost exactly the same thing. This quarter, I merged J�n Sučan's GSoC work, which merged the two commands. Now ping can handle either protocol, based on the -4 and -6 switches, or based on the format of the target. A ping6 hard link is provided for backwards compatibility.
Sponsor: Google Summer of Code
Ports
Changes affecting the Ports Collection, whether sweeping changes that touch most of the tree, or individual ports themselves.
KDE on FreeBSD
| Links | |
| KDE FreeBSD | URL: https://freebsd.kde.org/ |
| KDE Community FreeBSD | URL: https://community.kde.org/FreeBSD |
Contact: Adriaan de Groot <[email protected]>
The KDE on FreeBSD project aims to package all of the software produced by the KDE Community for the FreeBSD ports tree. The software includes a full desktop environment called KDE Plasma, graphics applications, instant-messengers, a video-editing suite, as well as a tea timer and hundreds of other applications that can be used on any FreeBSD machine.
The KDE team (kde@) is part of desktop@ and x11@ as well, building the software stack to make FreeBSD beautiful and usable as a daily-driver graphics-based desktop machine.
This quarter the kde@ team:
Landed the October, November and December updates to KDE Applications and to KDE Plasma
Landed all of the bi-weekly KDE Frameworks releases
Updated Qt to 5.12.2, including Qt5 WebEngine
Followed up with two cmake patch releases
Followed up one ninja patch release
updates and a new Matrix client from the KDE community as well, which we typically fail to administer and write about so this report is fairly short with mostly big-ticket items. We had fun, we chased the things that are most useful to us, and got through the year. Here's to next year with actually seeing FreeBSD people again.
I (adridg@) would like to especially thank Kai Knoblich (kai@) for chasing WebEngine: that's a huge and painful codebase to deal with, and here we are, all up-to-date.
FreeBSD Office team
| Links | |
| The FreeBSD Office project | URL: https://wiki.freebsd.org/Office |
Contact: FreeBSD Office team ML <[email protected]>
Contact: Dima Panov <[email protected]>
Contact: Li-Wen Hsu <[email protected]>
The FreeBSD Office team works on a number of office-related software suites
and tools such as OpenOffice and LibreOffice.
Work during this quarter focused on providing the latest stable release of LibreOffice suite and companion apps to all FreeBSD users.
Latest and quarterly ports branches got a series of updates of the LibreOffice suite from 7.0.1 thru 7.0.4 releases, compilation patches for all Tier 1 architectures, and updates of all companion libraries. Some of our local and critical to build patches were sent to and accepted by upstream.
Meanwhile, our WIP repository was moved to new home under official github.org/freebsd resources.
The WIP repository also has a major update with development versions of the LibreOffice suite, version 7.1.0.0.beta1 for now. Release will be planned in March, 2021.
We are looking for people to help the project.
All unstable work with LibreOffice snapshots is staged in our WIP repository.
The open bugs list
contains all filed issues which need some attention.
Patches, comments and objections are always welcome in the mailing list and bugzilla.
Ports On Non-x86 Architectures
Contact: Mark Linimon <[email protected]>
It has been some time since the last report on the status of FreeBSD ports on non-x86 architectures.
Traditionally, we have referred to these as "tier-2 architectures". However, aarch64 and powerpc64 have aspirations for tier-1. Also, although riscv64 is currently tier-3, it has aspirations for tier-2.
The big news is that, thanks to the FreeBSD Foundation (and the assistance of Philip Paeps), FreeBSD now has two new aarch64 boxes, which have replaced the previous, badly-aging, alternatives. For the first time since August, we once again have up-to-date aarch64 packages.
Thanks to the above, and the work of Emmanuel Vadot and others, some bitrot in aarch64 ports has been reversed.
Piotr Kubaj (pkubaj@) continues QA on powerpc64 (big-endian) ports. Almost everything that is buildable now does so. The Linux ports and some of the graphics drivers are excluded. Otherwise, powerpc64 is up to parity with amd64.
Piotr has also begun the task of bringing powerpc64le (little-endian) up to parity with powerpc64. Although several of the powerpc64 src committers (and your author) have a fondness for big-endian, the fact is that our most feasible path to getting graphics capability anywhere near parity with x86 is via the little-endian choice.
Mark Linimon (linimon@) has begun his own test-builds of ports on riscv64 just to ascertain overall buildability. Surprisingly, many ports do indeed build. Thanks to contributions from several people already working on riscv64, including John Baldwin (jhb@) with an LLVM fix, we are now able to build around 20,000 packages. NB: these packages are unofficial and not guaranteed.
The work of Kyle Evans (kevans@) on chasing bitrot in qemu has been key to work on both aarch64 and riscv64. All users are encouraged to update to the latest version.
Unfortunately mips/mips64 are badly in need of work. The fact that devel/libffi does not build on mips64 blocks nearly half the ports tree.
We need users of riscv64 to actually test the packages that have been built (so far, they have only been tested for buildability). Contact linimon@ if you are interested.
If anyone is still using mips/mips64 for other than the most trivial tasks, we would welcome patches.
Python 2.7 removal from Ports
Contact: Ren� Ladan <[email protected]>
Contact: FreeBSD Ports Management Team <[email protected]>
As of January 2020, Python 2.7 reached its end-of-life after several years of extensions. Portmgr subsequently started the project of phasing Python 2.7 out of the Ports Tree by tagging lang/python27 for expiration on 2020-12-31. Last year, some 740 ports were removed from the Ports Tree as they were incompatible with Python 3, mostly because these ports were either unmaintained or abandoned upstream.
During this process, there were several instances of an upstream still being active but where the upstream have not had the resources yet to upgrade their software to Python 3. A noticeable example of this is www/chromium and derived software, such as devel/electron7 and www/qt5-webengine. Portmgr is currently looking into ideas on how to minimize the impact of Python 2.7 on the Ports Tree while keeping Chromium and KDE 5 functional. As various software packages on the FreeBSD cluster itself also use Python 2.7, portmgr started coordinating with affected parties on upgrade plans. Currently there are 40 ports left that directly depend on Python 2.7 to build or run, and an unknown number of indirect ports. All those ports should eventually be upgraded to Python 3 or be removed too, ideally some time this year.
Portmgr is currently cleaning up (unused) Python 2.7 code from ports which do not need Python 2.7. New ports should not be using Python 2.7 anymore, i.e. they should not have USES=python but instead something like USES=python:3.6+.
So while this all looks rather invasive, it is not feasible to keep Python 2.7 around for much longer. Over time security vulnerabilities might show up which will likely no longer be fixed, because the Python Software Foundation no longer supports Python 2.7. Other problems are that the software gets outdated over time and thereby loses its usefulness as part of a development kit.
Help needed:
Coordinate with postmaster on isolating or migrating away from mail/mailman
Coordinate with clusteradm (?) for upgrading svnweb and our wiki
Xfce on FreeBSD
| Links | |
| Xfce 4.16 Upstream Release Announcement | URL: https://xfce.org/about/news/?post=1608595200 |
| Xfce meta-port on FreshPorts | URL: https://www.freshports.org/x11-wm/xfce4 |
Contact: Xfce team <[email protected]>
Contact: Guido Falsi <[email protected]>
The FreeBSD Xfce team (xfce@) work to ensure the Xfce desktop environment is maintained and fully functional on FreeBSD.
This quarter the Xfce team are pleased to welcome Xfce 4.16 to the FreeBSD ports tree!
Some of the highlights of this Xfce 4.16 release:
The panel now supports dark mode (enabled by default) and an animated autohide transition
A new panel plugin dubbed "statustray" which combines both StatusNotifier and legacy Systray items
Fractional scaling support was added to the display dialog (helpful on HiDPI displays)
A new tab in the "About Xfce" dialog shows basic system information like CPU or GPU type
The settings manager has improved search and filter capabilities
All settings dialogs now use window decorations drawn by Gtk (client side decorations)
The "Mime Settings" and "Preferred Applications" dialogs were merged into the "Default Applications" dialog
The Thunar file manager now supports pause for copy/move operations, and queued file transfer
Generating thumbnails for .epub (e-book format) was added to tumbler
A new default wallpaper and icon theme
The application finder now allows for sorting applications by "frecency" - a combination of frequency and recency
Dropped GTK2 support from all components and plugins
Due to GTK2 and libxfce4gui support being removed, some panel plugins and libraries will be removed since they no longer work with Xfce 4.16:
deskutils/orage
deskutils/xfce4-volumed
print/xfce4-print
science/xfce4-equake-plugin
x11/xfce4-embed-plugin
x11/xfce4-quicklauncher-plugin
x11/xfce4-wmdock-plugin
x11-toolkits/libxfce4gui
cause files from the libexo package to be absent after upgrade. To avoid the issue, upgrade the libexo package by itself before the rest of the packages, as described in UPDATING entry 20210102.
Thanks also to riggs@, Olivier Duchateau [email protected], woodsb02@, Sergey Dyatko [email protected], and ehaupt@ for their help and contributions.
Documentation
Noteworthy changes in the documentation tree, in manpages, or in external books/documents.
FreeBSD Translations on Weblate
| Links | |
| Translate FreeBSD on Weblate wiki | URL: https://wiki.freebsd.org/DocTranslationOnWeblate |
| FreeBSD Weblate Instance | URL: https://translate-dev.freebsd.org/ |
Contact: Danilo G. Baio <[email protected]>
Contact: Edson Brandi <[email protected]>
In search of new contributors an article was published in the September/October 2020 issue of the FreeBSD Journal about How to Become a FreeBSD Translator.
During the whole year we received new contributors to the effort; numbers are still growing and we are receiving translations almost daily on our Weblate platform.
Q4 2020 Status
11 languages (1 new language)
116 registered users (69 new users since 2020q1)
Languages
Chinese (Simplified) (zh_CN)
Chinese (Traditional) (zh_TW)
Dutch (nl_NL) - Added
French (fr_FR)
German (de_DE)
Italian (it_IT)
Norwegian (nb_NO)
Persian (fa_IR)
Portuguese (pt_BR)
Spanish (es_ES)
Turkish (tr-TR)
And please, help promote this effort on your local user group, we always need more volunteers.
DOCNG on FreeBSD
Contact: Sergio Carlavilla <[email protected]>
The Doc New Generation project is finished. The switch-over date will be Saturday, January 23rd.
The objective of using Hugo and AsciiDoctor is to reduce the learning curve and let people to start quickly contributing to our documentation system. Other benefits of using Hugo is that we can use other technologies aside from AsciiDoctor, like MarkDown, RST, Pandoc, etc.
You can find a work in progress on updating the FreeBSD Documentation Project Primer to Hugo/AsciiDoctor.
Miscellaneous
Objects that defy categorization.
Prometheus NFS Exporter
| Links |
Contact: Alan Somers <[email protected]>
FreeBSD's nfsstat(8) utility provides a wealth of statistics, but I wanted to monitor them with Prometheus. Screen-scraping the --libxo output would've been possible, but some of the stats are preprocessed in a way that interferes with my Prometheus processing. So I wrote a separate utility that publishes the raw stats provided by the kernel. Along the way I found and fixed a few bugs in nfsstat, too. If anybody is interested, I can add a port for it.
Sponsor: Axcient
Third-Party Projects
Many projects build upon FreeBSD or incorporate components of FreeBSD into their project. As these projects may be of interest to the broader FreeBSD community, we sometimes include brief updates submitted by these projects in our quarterly report. The FreeBSD project makes no representation as to the accuracy or veracity of any claims in these submissions.
FreeBSD Aarch64 under VMWare ESXi-ARM Fling
Contact: Vincent Milum Jr <[email protected]>
VMWare is a company that produces a commercial hypervisor known as vSphere ESXi for AMD64 and i386. In early October, they released a tech demo hypervisor for ARM Aarch64 which runs on ARM ServerReady hardware as well as single board computers such as the Raspberry Pi 4b (4GB and 8GB models). This new hypervisor is known as VMWare ESXi-ARM Fling.
Since the release of ESXi-ARM Fling, work has been done on both the hypervisor as well as FreeBSD, to make the two more compatible with one another. Even though the work was initially done to make these two work better together, the work overall has been more general purpose for FreeBSD in support of both bare-metal Aarch64 installations as well as running FreeBSD under other hypervisors such as QEMU.
An example of others building off of this work is
When ESXi-ARM Fling first released, to get FreeBSD to work under it, the process required taking the Aarch64 premade VMDK file, uploading it to the hypervisor storage, and then running a series of CLI commands to convert the disk image to a supported file format. The initial work done was to get the FreeBSD Aarch64 ISO bootable and with the required drivers to complete the install process. With this, users can do fresh installs of FreeBSD Aarch64 using the same methods they would use for AMD64 or i386 under ESXi.
The CD-ROM driver's inclusion into FreeBSD 12 barely missed the cut-off date for 12.2-RELEASE. However, the very first 12.2-STABLE build published for Aarch64 includes the CD-ROM driver. FreeBSD 13-CURRENT also includes this driver. Due to this, only 12-STABLE and 13-CURRENT support fresh CD ISO installations.
The next step was getting the major pieces of virtual hardware working. This included adding more USB controllers, the vmxnet virtual network card, and pvscsi para-virtual SCSI drivers added to Aarch64 GENERIC.
There is a known bug in ESXi-ARM Fling's virtual UEFI that prevents booting from pvscsi, so for the time being the boot device must be on a virtual disk attached to the SATA controller inside the VM.
ESXi-ARM Fling uses a new virtual SVGA device which currently does not have working drivers on any platform, as the specifications are not finalized yet. Due to this, only efi-fb/scfb is available for console and Xorg for the time being.
The VMCI driver is currently not compiling at all. This driver has sections of x86 assembly code that will need to be converted over to ARM. This would be a great area for anyone to look into that is experienced with converting assembly language!
At ESXi-ARM Fling launch, there was a hypervisor bug preventing more than 1 vCPU from working inside FreeBSD. This has since been fixed, allowing up to 8 vCPUs. Going beyond this requires a a patch to FreeBSD, which was authored by VMWare developer Cypou.
Things that are currently fixed/working:
Booting from CD ISO image
Virtual USB 2.0 controller
Virtual USB 3.1 controller
Virtual USB Keyboard
Virtual USB Mouse
vmxnet3 Virtual Network Card
pvscsi Para-Virtual SCSI Storage Controller
open-vm-tools Guest Virtual Machine Tools
Xorg Enhanced Mouse Driver (untested)
Multi-Core CPU (up to 8 vCPUs inside guest)
Booting from pvscsi
Xorg SVGA Driver
vmci Virtual Machine Communication Interface
Multi-Core CPU (more than 8 vCPUs)
collection easier by having a high quality virtualization environment for development and testing. This environment has already been used to update the ZeroTier port and Facebook's RocksDB used in the MariaDB port.
FreeBSD now has a Discord chat! Discussion about FreeBSD on Aarch64 in general takes place in our #embedded channel. Despite the name, we discuss all levels of ARM development, from large servers, to virtualized environments, all the way down to single board computers.
Bastille
| Links | |
| Bastille GitHub | URL: https://github.com/bastillebsd/bastille |
| Bastille Templates | URL: https://gitlab.com/bastillebsd-templates |
| Bastille Website | URL: https://bastillebsd.org |
Contact: Christer Edwards <[email protected]>
Bastille is an open-source system for automating deployment and management of containerised applications on FreeBSD.
Bastille Templates automate container setup allowing you to easily reproduce containers as needed.
Bastille is available in ports as sysutils/bastille.
Q4 2020 Status
In Q4 2020 Bastille merged some exciting new features. Changes include:
full adoption of the previously experimental Bastillefile format
new
configsub-commanddefault templates included and applied by default
support for -CURRENT jails on -CURRENT hosts
support for 32bit containers on 64bit hosts
support in templates for dynamic arguments & defining variables
over two dozen bug fixes and general improvements
upstream was updated to 0.8.202010101 (latest).
ports (sysutils/bastille) was updated to 0.7.20200414.
CheriBSD
| Links |
Contact: Alex Richardson <[email protected]>
Contact: Andrew Turner <[email protected]>
Contact: Brooks Davis <[email protected]>
Contact: Edward Tomasz Napierala <[email protected]>
Contact: George Neville-Neil <[email protected]>
Contact: Jessica Clarke <[email protected]>
Contact: John Baldwin <[email protected]>
Contact: Robert Watson <[email protected]>
Contact: Ruslan Bukin <[email protected]>
CheriBSD extends FreeBSD to implement memory protection and software compartmentalization features supported by the CHERI instruction-set extensions. There are three architectural implementations of the CHERI protection model: CHERI-MIPS, CHERI-RISC-V, and Arm's forthcoming experimental Morello processor (due late 2021). CheriBSD is a research operating system with a stable baseline implementation into which various new research features have been, or are currently being, merged:
Arm Morello - In October, we released a developer preview of CheriBSD ported to Arm's Morello architecture. This release supports a dynamically linked runtime and is generally functional. It was cut from a development branch and work is in progress to merge the contents of this branch with the CheriBSD main line. We anticipate producing a new release from this branch in early 2021.
Kernel spatial memory safety (pure-capability kernel) - The current CheriBSD kernel is a hybrid C program where only pointers to userspace are CHERI capabilities. This ensures that the kernel follows the intent of the application runtime and cannot be used to defeat bounds on application pointers. We have developed and will soon merge a pure-capability kernel where all pointers in the kernel are appropriately bounded capabilities. This vastly reduces the opportunity for buffer overflows. This spatial memory safety lays the groundwork for future work such as device driver compartmentalization and kernel temporal safety.
Userspace heap temporal memory safety (Cornucopia) - CHERI capabilities provide the necessary features to enable robust and efficient revocation of freed pointers. With Cornucopia we have implemented a light-weight revocation framework providing protection from use-after-reallocation bugs with an average cost below 2%. We aim to bring these overheads down further over the next year and merge this functionality into the mainline CheriBSD.
Syncing with upstream FreeBSD - We spent considerable time this quarter catching up with FreeBSD-CURRENT. As of the beginning of December, we had caught up. Merges are currently paused while we work to land Morello and pure-capability kernel changes. In the interim, we have performed a test merge between our tree based on the legacy export of the FreeBSD tree to git and the new FreeBSD git repository. The process went smoothly and is expected to have few impacts.
We have been working on updating the arm64 bhyve from Politehnica University of Bucharest to have it committed to FreeBSD. We have been upstreaming initial changes to help support this.
Embedded Lab Project
| Links | |
| FreeBSD Embedded Lab Design | URL: https://www.funkthat.com/gitea/jmg/fbsdembdev |
| Lab API code | URL: https://www.funkthat.com/gitea/jmg/bitelab |
Contact: John-Mark Gurney <[email protected]>
The Embedded Lab Project's goal is to make SBCs and other devices more accessible to developers. Despite SBCs often being inexpensive, it is not inexpensive to maintain them, in terms of the cost of time to keep them up to date, infrastructure to support them, etc.
The goal of this project is to support and enhance the existing CI work but also make it easier for developers to test their code and changes on one, or many different boards.
Once the work is [mostly] complete, I will host a lab that will be freely available to everyone who has a FreeBSD.org account. Information about this will be sent once it is closer to launch.
The core part of the architecture is each time a board is reserved via the API, a new jail is created which contains the serial console tty, an interface for internet access, and an interface that is connected to the board's ethernet port (assuming it has one). This allows a clean system for each run, and allows complete control over the network interfaces to support netbooting and other development. The jail will have a basic set of FreeBSD packages installed that matches the board.
Part of the API will also allow power cycling the board to aid in debugging. This part is relatively extensible, so adding additional modules to provide additional support should not be difficult.
The API includes support for running interactive commands in the jail. This will make it easy to script control of the environment, such as directly running an expect script against the serial console, or even just running a script in the jail.
The work is progressing well, and currently a single board, a Pine64 A64-LTS, is integrated and working. Board reserves and releases are working, along with the ability to run commands in the jail via the API. Power control is functional, and is currently using a PoE smart switch to control power.
Work has stalled on being able to use the SDWire with an environment due to power issues. USB is not made for power isolation, which is causing issues w/ power control. The existing board, the A64-lTS, is using a USB serial console adapter that is opto-isolated, ensuring that there is no problems w/ power control. But there I have not found a solution for high speed USB. I believe that cutting the VBUS (power) line of a USB cable would allow fine grain power control, but tests have not been conducted yet.
Sponsor: The FreeBSD Foundation
FreshPorts
| Links | |
| FreshPorts | URL: http://freshports.org/ |
| FreshPorts blog | URL: http://news.freshports.org/ |
Contact: Dan Langille <[email protected]>
FreshPorts, and its sister site, FreshSource, have reported upon FreeBSD commits for 20 years. They cover all commits, not just ports.
FreshPorts tracks the commits and extracts data from the port Makefiles to create a database of information useful to both port developers and port users.
For example, https://www.freshports.org/security/acme.sh/ shows the history of this port, back to its creation in May 2017.
git
The work to become git-ready is mostly complete. Both src and doc commits are flowing into devgit.freshports.org. Some work is required on various issues, but nothing that stops the flow of commits into the database.
Help wanted
Amazon have donated enough to try FreshPorts on AWS. I need help with the following:
getting IPv6 working
working with RDS
Thank you
helloSystem
| Links | |
| Documentation | URL: https://hellosystem.github.io/docs/ |
Contact: Simon Peter <[email protected]>
Contact: #helloSystem on irc.freenode.net, mirrored to #helloSystem:matrix.org on Matrix
helloSystem is FreeBSD preconfigured as a desktop operating system with a focus on simplicity, elegance, and usability. Its design follows the “Less, but better” philosophy. It is intended as a system for “mere mortals”, welcoming to switchers from a world in which a global menu bar exists, the Command key is used rather than Control, and applications are contained in .app bundles.
helloSystem grew out of frustration with usability shortcomings of existing open source desktop environments. FreeBSD was chosen as the base because it offers one consistent base system rather than a fragmented landscape of distributions lacking a common platform.
helloSystem aims at providing a "it just works" out-of-the-box user experience in which a non-technical user can just use the system without ever opening the terminal, without having to configure anything, and without ever seeing white text on a black background scroll by during system boot. Technologies embraced include DNS-SD/Zeroconf (also known as Bonjour), IPP Everywhere (also known as AirPrint), eSCL (also known as AirScan), etc.
Prerelease installable Live ISO images are available.
Help is needed in a number of areas, especially:
FreeBSD/kernel: allowing to put the system into a read-only disk image with a writable overlay, e.g., using unionfs
Qt, Python: writing various easy-to use frontends for FreeBSD/OpenZFS functionality, e.g., Disk Utility.app
Testing and bugfixing
K8S-bhyve
| Links | |
| K8S-bhyve | URL: https://k8s-bhyve.convectix.com |
| K8S-bhyve | URL: https://github.com/k8s-bhyve |
| Kubernetes | URL: https://kubernetes.io/ |
Contact: Kirill Ponomarev <[email protected]>
Contact: Oleg Ginzburg <[email protected]>
K8S-bhyve is opensource project concentrating primarily on deploying and use of kubernetes on FreeBSD/bhyve in a more agile and more comfortable manner. We are going to provide distributed multi-DC environment or just stand-alone clusters with native PV/PVC support.
For 2020Q4 we made and published a k8s-bhyve image which you might install with ISO/memstick, as well as with bsdinstall.
Puppet
Contact: Puppet Team <[email protected]>
Since our last status report a few months ago, the FreeBSD ports tree has seen the addition of the Choria (sysutils/choria) orchestration tool, and the Puppet Platform 7 with the Puppet Agent (sysutils/puppet7), Puppet Server (sysutils/puppetserver7) and PuppetDB (databases/puppetdb7).
Older versions of Puppet (5 and 6) are still in the ports tree, allowing a smooth transition, but please note that Puppet 5 will reach EOL soon, and as it is not compatible with the recent ecosystem provided by FreeBSD (i.e. it is not compatible with the latest version of Ruby and depends on old FreeBSD primitives not available anymore), it is recommended to update at least to Puppet 6 as soon as possible.
Ports depending on Puppet (e.g. sysutils/rubygem-bolt) have been updated to add options allowing to choose which version of Puppet to depend on. For now, the default is Puppet 6, but we plan to switch the default to Puppet 7 in a few weeks, probably when Puppet 5 will have reached EOL.
News Home | Status Home