Our BoF session proposes topics as informal meeting during the conference. The topic lead (submitter) will drive the conversations on the area of interest described in each BoF.
The attendees group together based on a shared interest and carry out discussions without any pre-planned agenda.
For many years developers have leveraged gdb or crash to look at kernel crash dumps on linux. Although those tools have served us well, it can sometimes be difficult to navigate the crash dump to find the information you really need. In this talk, we would like to present some new tools that make it easier to debug kernel crash dumps and enhance kernel developer's ability to root causes...
Wayland is getting close to being ready for day 2 day generic desktop use, close but there still are many small issues to tackle, see e.g. :
The purpose of this microconference is to get people together to discuss the various open issues, try to come up with solutions for some of them and possibly...
For long time, The kernel have contained two mechanisms with similar packet filtering functionality: tc filter (with chains) and iptables/nftables.
As eBPF is starting to take over, once again we seem to have two mechanisms with similar functionality: BPFilter and the newly suggested OVS-eBPF datapath (on top on tc).
As we move to using eBPF, I'd like to discuss the possibility of uniting...
Testing the upstream kernel is not an easy task. The burden is
still largely put on developers, although several projects are
now covering parts of it such as 0-day, LKFT, CKI, Coccinelle,
syzkaller and kernelci.org. While they all tend to have their
own speciality, they also face a lot of similar challenges.
This BoF is to give an opportunity to exchange ideas and bring
What I'd like to get to is to discuss that buffered IO basically sucks for databases with high throughput, and direct IO sucks for databases that aren't individually well tuned, and is not adaptive to memory pressure at all.
Buffered IO is slow, until recently only synchronous, has double buffering issues and writeback is hard to control.
Direct IO requires that the application's equivalent...
Modern server and compute intesive systems are naturally built around several top performance CPUs with large amount of cores and equipped by shared memory that spans a number of NUMA domains. Compute intensive workloads usually implement highly parallel CPU bound cyclic codes performing mathematics calculations that reference data located in the shared memory. Performance observability and...
Follow up on the tracing microconference
Topics to be discussed:
- Perf related events
- Histogram sql syntaxes
It looks like there may well be enough critical folks present to have a good BOF about safety and linux. Topics can include safety processes and methodologies, tooling to support analysis, security update concerns, etc. Basically, if you're interested in using Linux in safety critical systems come join, and we'll see where the conversation goes.
This BoF session aims to bring together Linux kernel developers who have an interest in formal methods (or formal methods experts with an interest in kernel development). Topics for discussion:
- A poll of formal methods currently used in the context of the Linux kernel: SPIN, TLA+, CBMC, herd, plain English etc.
- High level design specification vs. low level algorithm modelling. What...
Discussion of using Persistent Memory as first- (or second-) class memory.
Google has a successful prototype of a software-managed "Transparent" mode for 3dXPoint / AEP memory, but we're working on re-designing this into something that is more supportable and at least partially upstreamable.
We want to open a discussion of how we can represent this "swap"-like use of AEP sensibly.
Code review is a collaborative activity involving sentiments and emotions that can affect developers' productivity, creativity, and contribution satisfaction. Discussions in a code review environment in open source could get spirited at times as people from diverse backgrounds and interests are part of it. As a consequence, open source communities have become introspective and started to think...
The csky architecture officially merged the main line in linux-4.20. Before that, eight architectures have just been removed from the main line. Many people ask what is the meaning of csky upstream? Also includes our colleagues. Here, we will give some examples to introduce the progress of the csky architecture in the past six months and the value and significance of linux-csky. This is an...
For some time now, special camera setups exist having features which are challenging for I2C address layouts as we know them in Linux: a) a high-speed serial link which can embed I2C communication (e.g. GMSL or FPD-Link III) and b) the ability to reprogram the client addresses of the I2C devices on the camera.
The use case for these cameras is to run multiple of them in parallel, and not just...
Application-specific accelerators are going to start showing up in larger numbers in the times ahead. Today there's often no suitable subsystem for them to aggregate into, and the first of them have landed under drivers/misc for the time being.
The goal of this BoF is to introduce and discuss the ground rules for a new drivers/accel subsystem, how it fits in with other subsystems, and...
Discussion around topics related
to PCI specifications and microconference follow up
- Root complex integrated endpoints
- Native host controllers link management
- VFIO/IOMMU/PCI follow up
This session will focus on answering questions on the internals and the usage of Linux-kernel RCU. However, questions regarding details of the RCU-related patches in the -rt patchset will be deferred to other venues, given that this topic consumed the entire time in the 2018 informal RCU BoF session.
This is not intended to be a tutorial on RCU basics, though a separate session on this topic...
When multiple instances of workloads are consolidated in same host it is
good practice to partition them for best performance. For e.g give a NUMA
node parition to each instance. Currently Linux kernel provides two
interfaces to hard parition: sched_setaffinity system call or cpuset.cpus
cgroup. But this doesn't allow one instance to burst out of its partition
and use available CPUs from other...
Today every modern multimedia supported SoC’s comprises of variety of display controller interfaces bounded with LCD panels or bridges and a GPU, for providing feasible display acceleration.
The Linux kernel handle all these display controller interfaces with associated panels, bridges via DRM subsystem, but it becomes a daunting task for many of the display users to make use of this DRM...