Even if the CPU-Idle framework has improved significantly over the last years, with important consolidation efforts, the framework doesn’t scale for multi-cluster SMP systems and heterogeneous systems like big.LITTLE. Therefore Linaro have been working on a solution that unifies idle management for all kind of devices, including CPUs. This sessions provides the audience an update of this work
The Per Entity Load Tracking (PELT) is a key stone in tasks placement of the scheduler but suffers of some weakness when it’s not just bugs. During the last LPC, it has been decided to fix all pending issues of PELT before starting to consider another load tracking mechanism for scheduler and/or EAS. This session will show the improvement reached since the last connect and the LPC as well as the next ones. We will also looks at the RT class which lacks a good load tracking.
The time_t type in 32-bit Linux overflows in January 2038, which breaks all existing application binaries. We'll talk about how we are addressing this in the kernel, in glibc and in distributions, what problem are still unsolved, and the various ways in which even 64-bit systems are still affected.
Sensory swarms of an IoT can be wirelessly interconnected to interact with the edge of a cloud, and offer an unprecedented ability to monitor and act on a range of evolving physical quantities. The Swarm leverages the paradigm of independent, cross-niche and heterogeneous devices that can cooperate with each other in order to execute tasks synergistically. The Swarm architecture is device-oriented, focused on machine-to-machine communications. The Heterogeneous Broker is responsible for dynamically recruiting resources from the cloud; allowing information aggregation to make or aid decisions; and then to dynamically recruit actuation resources. In this talk we will describe our experience implementing and deploying the heterogeneous broker developed by USP in partnership with UC Berkeley. An special focus will be given to the 96Boards program.
The 2016.12 release includes sufficient support that all teams should now be able to port all their test jobs to LAVA V2. Some changes to the test definitions may be useful to make better use of V2 and to increase the portability of the test definitions in line with QA recommendations.
This presentation will showcase the current examples of how to migrate V1 MultiNode Android test jobs to V2 LXC and how to use V2 Secondary Media to support deployments on Juno. Details will also be included about the LAVA team can help test writers migrate their test jobs as the number of V2 devices increases and the number of V1 devices decreases. Details of how the Cambridge lab will migrate will also be covered as some device-types cannot support V1 and V2 simultaneously. The timetable of how V1 support will be removed during 2017 will be outlined.