The RT patchset has come to EoL and an alternative to meet the determinism and latency requirements for networking workloads is required. Compare the two solutions and the impacts migrating to Core isolation may impose.
This session is a continuation of the Advanced Toolchain Usage Part 1 & 2 presentations given at LCU14. Parts 3 and 4 will cover a variety of topics, such as: Linker tips and tricks, adding symbol versioning interfaces to a system library, debugging the dynamic linker, debugging applications that use malloc, gcc attributes, manually constructing a backtrace on arm & Aarch64, how to add lightweight debugging to your program, how to use a signal handler appropriately, and TLS Models on Aarch64 and when to use them.
After an entire year of full time involvement of Linaro in the CoreSight area it is time to go back and look at what was accomplished, where things are, what worked well and where we need to improve. The presentation will start with a summary of what we accomplished and list features that were successfully pushed upstreamed. From our original planning some activities were carried out as expected, some had to be curbed for various reasons and other were added out of necessity.
The second part will concentrate on the remaining work that needs to be done and lay out a rough estimate of the order in which it will be addressed. ARMv8, power management, trace decoding and the integration of various CoreSight functionality with other kernel sub-systems are very high on our list of things to so. We hope to receive feedback from the audience on the soundness of our approach and definitely open to opinion or suggestions that can help make CoreSight a better sub-system usable by everyone.
Performing kernel validation on hundreds of platforms for every merge on every upstream tree is a very challenging endeavor. Currently, much of this testing is performed by the maintainers, and/or chip vendors. The end result is that the validation efforts become fragmented in various locations, and the results are presented in different manners making it difficult to compare results. The goal of the kernelci.org project is to provide a uniform interface for these aggregated results. In this session the team will provide brief overview of the system architecture, and a walk through of the dashboard. A demo will provide users an example of how they may interact with the system. The second half of the session will be a dedicated discussions on next steps.
One of the problems we experience quite a lot is to do with people setting inordinately long timeouts on their hacking sessions, which leaves the board unavailable for queued jobs.
This session is to explore some of the possible solutions, in LAVA, the Lab and the general v.l.o community of users. Some possible scenarios are:
- you have to specify in the job that if you leave the ssh session, you want to keep hacking up for later
- if a hacking session detects that there are no users ssh