November 18, 2022
During the third quarter of 2022, 300 src, 36 ports, and 13 doc tree commits were made that identified The FreeBSD Foundation as a sponsor. It’s challenging to concisely summarize all this work. It varies from complex new features to various bug fixes spanning the src tree. A few highlights include adding ZFS support to makefs, adding experimental 16k page support on arm64, removing the default pager, and fixing race conditions in the event timer.
The FreeBSD Foundation also contracts companies and individuals to work on projects that benefit FreeBSD. Here is a summary of the projects that are in progress.
FreeBSD as a Tier I cloud-init Platform
cloud-init is the standard way of provisioning servers in the cloud. Unfortunately, cloud-init support for operating systems other than GNU/Linux is rather poor, and the lack of cloud-init support on FreeBSD is a hindrance to cloud providers who want to offer FreeBSD as a Tier 1 platform. To remedy the situation, this project aims to bring FreeBSD cloud-init support on par with GNU/Linux support. The broader plan is to lift support across all BSDs.
The project deliverables include completing an extraction of certain networking classes, implementing ifconfig and login.conf parsers, implementing IPv6 configuration, creating devd.conf rules for Azure, and FreeBSD Handbook documentation about productionizing FreeBSD.
On the way there, any BSD-related bugs found in modules and documentation will also be fixed.
People interested in helping with the project can test new features and fixes via the net/cloud-init-devel port, which will be updated on a weekly basis.
Intel wireless towards 11ac
The ongoing project aims to support the latest Intel wireless chipsets on FreeBSD using LinuxKPI compat code backed by native net80211 and kernel code. In addition, work is on the way to support 11n and 11ac standards in the LinuxKPI 802.11 compat code and fill gaps for mostly 11ac in the native net80211 wireless stack.
For the Intel iwlwifi wireless driver there were no major updates in the last months. We updated the firmware to the latest publicly available version and fixed some of the most visible bugs. Work is also on the way to support the D3 power saving code.
LinuxKPI compat code also got some improvements and fixes which at times were only observable on certain generations of iwlwifi chipsets.
Changes in net80211 and LinuxKPI compat code for 11n and 11ac have little public visibility so far in order to not break basic support. Updates to constants based on newer 802.11 standards and other changes without user-visible effect were merged, and functional changes will follow in the coming months, initially hidden behind compile-time or runtime options.
Improvements and updates were largely merged back to stable/13 for the benefit of the users tracking this branch and to help with more testing.
For the latest state of the development, please follow the freebsd-wireless mailing list and check the wiki pages.
LLDB multiprocess debugging support
According to the upstream description, “LLDB is a next generation, high-performance debugger. It is built as a set of reusable components which highly leverage existing libraries in the larger LLVM Project, such as the Clang expression parser and LLVM disassembler.”
FreeBSD includes LLDB in the base system. The previous Foundation-sponsored projects improved LLDB, to make it a credible debugger for the base system, although it still has a few limitations compared to the contemporary versions of GNU GDB. This project started in April 2022. It aims to implement full support for debugging multiple processes simultaneously.
At the start of the project, LLDB featured very limited support for multiprocess debugging. Currently, the server is already able to monitor multiple processes using the multiprocess extension to the GDB Remote Serial Protocol. The work on implementing the client-side counterpart for this protocol is ongoing.
Once the project is finished, LLDB will be able to trace an arbitrary number of forked processes simultaneously (equivalent to GDB’s detach-on-fork off).
OpenStack on FreeBSD
OpenStack is an open-source cloud operating system for different types of resources like virtual and bare-metal machines. Users can spawn FreeBSD instances on the open cloud platform, but it is not currently possible to run OpenStack control plane on FreeBSD hosts. The goal of this project is to port key OpenStack components so that FreeBSD can function as an OpenStack host.
Academic and industrial research groups have been evaluating CHERI-enabled Morello boards since mid-2022. A resource orchestration platform like OpenStack can improve the speed and cost of provisioning, managing, and recycling those boards.
Starting in January 2022, Chih-Hsin Chang has been working to port several OpenStack components to run on FreeBSD, including:
- Keystone (identity service)
- Glance (image service)
- Placement (resource tracking and inventory service)
- Neutron (networking service)
- Nova (compute service)
Some of the items are still under heavy development. For instance, due to the design of Neutron, the DHCP servers are placed inside Linux network namespaces. It is necessary to find an alternative, e.g. vnet, on FreeBSD and adapt it. Most of the time the porting strategy is to make as small of an impact as possible by working around obstacles. But something like oslo.privsep deserves a true porting. oslo.privsep is rooted in Linux capabilities to do the privilege separation work. Right now we just bypassed any Linux capabilities-related operation inside oslo.privsep. So there is plenty of hackish spots in the source code and configurations currently. All of these along with the building and installation steps will be collected in the project repositories.
Snapshots on Filesystems Using Journaled Soft Updates
This project will make UFS/FFS filesystem snapshots available when running with journaled soft updates.
The UFS/FFS filesystem has the ability to take snapshots. Because the taking of snapshots was added after soft updates were written they were fully integrated with soft updates. When journaled soft updates were added in 2010, they were never integrated with snapshots. So snapshots cannot be used on filesystems running with journaled soft updates.
Snapshots became less important with the support for ZFS on FreeBSD since ZFS can take snapshots quickly and easily. However there remain two instances where UFS snapshots are still important. The first is that they allow reliable dumps of live filesystems which avoids possibly hours of down time. The second is that they allow the running of background fsck. Similar to the need for scrub in ZFS, fsck needs to be run periodically to find undetected disk failures. Snapshots allow fsck to be run on live filesystems rather than needing to schedule down time to run it.
This project has two milestones:
- enable snapshots when running with journaled soft updates and ensure that they can be used for doing background dumps on a live filesystem. This milestone was completed on November 12.
- extend fsck_ffs to be able to do a background check using a snapshot on a filesystem running with journaled soft updates. This milestone is expected by Q3 of 2023.
Bhyve Issue Support
The Foundation contracted John Baldwin to dedicate time to Bhyve as issues arise, especially security issues. Here is a summary of his 2022q3 work on that contract.
- bb31aee bhyve virtio-scsi: Avoid out of bounds accesses to guest requests.
- 62806a7 bhyve virtio-scsi: Tidy warning and debug prints.
- 7afe342 bhyve e1000: Sanitize transmit ring indices.
- c94f30e bhyve: Validate host PAs used to map passthrough BARs.
- 16bedf5 pci: Add helper routines to iterate over a device’s BARs.
- baf753c bhyve: Support other schemes for naming pass-through devices.
- fa46f37 bhyve e1000: Skip packets with a small header.
- e7439f6 bhyve xhci: Cache the value of MaxPStreams when initializing an endpoint.
At the end of the quarter, the Foundation contracted Mitchell Horne to add and improve support for RISC-V hardware. Mitchell will also perform general maintenance such as fixing bugs, handling reports, providing review for new code changes, and improving source code legibility and documentation.