This is a discussion on various methods put forward to reduce the size of Linux kernel and user space binaries to make them suitable for small IoT applications, ranging from low hanging fruits to more daring approaches. Results from on-going work will also be presented.
**Title: TianoCore – Open Source UEFI Community Update**
The TianoCore project hosts EDK II, an open source implementation of the Unified Extensible Firmware Interface (UEFI). EDK II has become the defacto UEFI implementation for ARM and Intel platforms, expanding standards based firmware across multiple architectures. This keynote will provide an update on the current status of the TianoCore project, plans for future improvements, and a discussion of why firmware is critical in today’s digital ecosystem.
Brian Richardson is an Intel technical evangelist who has spent most of his career as a “BIOS guy” working on the firmware that quietly boots billions of computers. Brian has focused on the industry transition to the Unified Extensible Firmware Interface (UEFI), demystifying how firmware works and simplifying firmware development tools. Brian has presented at LinuxCon, UEFI Plugfests, and Intel Developer Forum. He is a blogger for the Intel Software Evangelists project, former writer forlinux.com, and (apropos of nothing) executive producer for DragonConTV.
The Android Open Source Project is one community which is strategic to Linaro and it’s members. The purpose of this mini conference is to gather fellow Android engineers together from the community, member companies, and Linaro to discuss engineering activities and improve collaboration across different groups.
Within this mini conference we encourage discussion and presentations to advance engineering topics, forge consensus and educate each other.
The tentative agenda for this mini conference includes :
- Quick introduction
- Filesystems - Between requirements for encryption and standing concerns about degrading performance as an Android file system age, let’s have some discussion involving current data, known issues and towards improvements in this area for Android.
- HAL consolidation - Review current status and discuss next steps to work on.
One build for many devices: device/build configuration. Next features and platforms to add. Gaps in HiKey support vs. AOSP build.
- Graphics - YUV support in mesa and hwc.
- WiFi and sensor HAL status and next steps
- New developments with AOSP + the Kernel - With regards to the Google Common Kernel tree and upstream Linux kernel activities related to Android, there are a few topics up for discussion:
- - Updates on HiKey in AOSP
- - EAS in common.git & integration with AOSP userspace
- - New Sync API in 4.6+ kernels, and how it will affects graphics drivers
- AOSP transition to clang - As everyone knows GCC in AOSP has been deprecated. Let’s cover current status, issues and next steps. Let’s also discuss the elephant in the room, building the kernel with clang.
- Out of tree AOSP User space Patches - This is a discussion with the goal of organized action to see forward progress on AOSP user space patches that aren’t in AOSP for whatever reason.
- Android is used in some environments where booting can be frequent and affect the product experience. Do you want to wait for a minute while your car boots? We’ll spend time brainstorming on improving Android boot time.
OpenDataPlane provides a portable framework for data plane acceleration that is the basis for higher-level functions such as the full IP protocol stack provided by OpenFastPath (OFP). ODP+OFP in turn can be used to offer acceleration to applications like the open source NGiNX web server. This talk discusses experiences using these tools with a focus on the performance and scaling benefits of using ODP and OFP.
Tutorial material to support getting started/evaluating LAVA v2. An end-to-end tutorial including physical bootloader device with a stand-alone installation of LAVA. Covering - device requirements, device configuration for 32- and 64-bit platforms with QEMU, ARMv7 and ARMv8 targets. An important part would be having links between the v2 documentation and the tutorial material.
Today the remoteproc & rpmsg code available in mainline serves as a base for numerous out-of-tree implementations, ranging from bug fixes to larger feature additions. As we're discussing how to bring these additions towards mainline a common set of topics shows up between the various trees. This talk serves to give an insight into these discussions, ongoing work and connect people with interest in these subsystems.
Abstract Xen on ARM: The Xen port is exploiting this set of new hardware capabilities to run guest VMs in the most efficient way possible while keeping ARM specific changes to Xen and Linux to a minimum. ARM virtualization is set to be increasingly relevant for the embedded industry in the coming years.
Whilst Xen is best known as the technology powering the biggest clouds in the industry, it also a great fit for automotive deployments and mobile devices that can fit in your pocket. The talk will give concrete examples of the ways Xen can add value to your platforms, not only by providing an excellent general purpose virtualization solution, but also by providing simple, yet effective ways to partition the platform into different security domains.
This presentation will include a brief overview of the Xen on ARM architecture, covering the key design principles employed. The techniques pioneered during the ARM port that allowed the Xen community to remove many legacy components from the Xen code base, streamlining both the ARM and x86 implementations. The talk will conclude explaining how to port Xen to any new ARM boards with the least amount of effort.
Abstract LuaJIT: Lua is a scripting language commonly embedded by web
front-ends. Enabling Lua JIT compilation can reduce CPU usage when
handling huge amounts of network traffic. This year Linaro (and
others) started to work on porting LuaJIT to AArch64. Though the work
is not finished we have made good progress. This presentation will
briefly introduce LuaJIT, discuss the technical challenges of porting
to AArch64, and address the progress of the porting effort and the
Continuing the discussion from Linaro Connect Bangkok 2016, we discuss further experiments with Window Assisted Load Tracking (WALT, formerly WinLT) on other architectures such as x86 and with different workloads including laptop/desktop and server usecases. We propose that WALT is more accurate in tracking cpu/task utilization than PELT (which we believe is better at load rather than util tracking) and that better utilization estimation schemes can be built on top of WALT. Experiments use the new schedutil governor. Examination of individual use cases in detail as well as how to handle migration and potential caveats of a windowing scheme are discussed.