Linaro Connect Bangkok 2019
| title | author | length |
|---|---|---|
| Welcome to BKK19 | Joe Bates | 01m |
Welcome to BKK19 states that Linaro Connect meets every six months in a citywide event and that over 300 engineers working on Arm open source converge for a… | ||
| Enable debug tools on Golang for arm platform (MSan & TSan) | Fangming Fang | 31m |
Ports MemorySanitizer and ThreadSanitizer into Go on arm64 by adding required compiler-rt-compatible memory layout, shadow-memory mapping and small… | ||
| Latest storage status on Arm | Jun He | 24m |
Status update lists concrete wins: SPDK enabled on aarch64 with both 4KB and 64KB kernel page sizes and upstream patches accepted in Ceph, SPDK and DPDK; ISA-L… | ||
| Security WG Lightning talks | Joakim Bech, Jerome Forissier, Jens Wiklander | 26m |
Documentation moves all resources into a single searchable site, Risk I/O findings produce a completed patch series and coordinated disclosures, and OP-TEE 3.5… | ||
| LEDGE Overview and Goals | François-Frédéric Ozog | 27m |
LEDGE SIG produces a concrete reference platform that binds edge runtimes (EdgeX, TIBCO, SAP, Azure IoT) to SoC hardware so one software stack scales from tiny… | ||
| Building the Linux kernel with Clang | Nicholas Desaulniers, Tri Vo | 29m |
Linaro and Google engineers describe getting mainline Linux to build and boot with Clang and LLVM tools, shipping Pixel 2 and Pixel 3 kernels built with LTO… | ||
| Networking in Zephyr LTS and beyond | Paul Sokolovsky | 24m |
Abstract
Zephyr 1.14 is a long-expected LTS (Long-Term Support) release for Zephyr RTOS (https://www.zephyrproject.org/), a Linux Foundation project . A lot of… | ||
| EAS Unit Testing | Steve Muckle | 22m |
C-based LTP tests validate concrete EAS and scheduler behaviors by enabling ftrace and parsing it with a custom trace parser; merged into an AOSP LTP fork… | ||
| LCG Lightning Talks | Show Liu, Yongqin Liu, Sam Protsenko, Sumit Semwal, John Stultz | 34m |
Cut-down allocation API called DMA heaps rework is proposed with an RFC posted and aims to move allocation out of ion into per-heap DMA-buf ops; cache-op… | ||
| Arm-SVE enabled post-K processor | Mitsuhisa Sato | 28m |
A64FX with Arm‑SVE CPU targets energy‑efficient exascale by pairing A64FX with Arm‑SVE cores (48 compute + 2 OS threads) and on‑package 32 GB HBM2 per chip… | ||
| JVM on ARM. From sensor to cloud | Dmitry Chuyko | 25m |
Demo builds a secure ARM gateway that reads a hardware temperature sensor on a Raspberry Pi, sends data to Kafka, processes it in the cloud and renders it in a… | ||
| The value of being a 96Boards Partner | Attila Ambrus | 01m |
Attila Ambrus, from Arrow, explains the value gained by being a 96Boards Partner. | ||
| What is unique about Linaro Connect? | Attila Ambrus | 01m |
Attila Ambrus, from Arrow, explains what is unique about Linaro Connect. | ||
| Why Sponsor Linaro Connect? | Attila Ambrus | 01m |
Attila Ambrus from Arrow explains why you should sponsor Linaro Connect.
For more information please visit: https://connect.linaro.org/sponsors/ | ||
| Arm Neoverse software ecosystem and the open source strategy | Kevin Ryan | 12m |
Arm announces the Neoverse roadmap plus N1 and E1 platforms and provides a full stack: reference architectures, compiler and IP optimizations, foundry and EDA… | ||
| PMWG: CPU Idle governor and irq prediction | Daniel Lezcano, Vincent Guittot | 56m |
Implementation of an idle governor that predicts the next wake source and subtracts wake-up overhead shows that exit latencies are highly variable up to 2–3 ms… | ||
| FPGAs for Highest Performance Inference | Craig Abramson | 25m |
FPGAs claim lower end-to-end latency by avoiding GPU-style batching and operate well at batch=1, letting pipelines start producing outputs immediately… | ||
| Empower the China Arm developers via Linaro Developer Cloud | Pengcheng Laboratory (PCL) | 07m |
A developer cloud project builds a public, free ARM-focused sandbox that combines an ARM server fleet, Kylin and Red Hat OS images, and tooling from Huawei and… | ||
| Design your own custom co-processors and acceleration hardw | Kevin Keryk | 51m |
Course outlines how to prototype custom accelerators on an Ultra96 board using SDSoC and the PYNQ-like Python framework, with concrete results: a… | ||
| Realtime Edge AI SoC for Computing Vision Applications | Darren Tsao | 31m |
Introduction to Sophon Edge ai chip solution with int8 TPU accelerator, which can enable high performance deep learning on edge devices in low power, real-time… | ||
| SynQuacer use case update | Shuichi Yamane | 19m |
SynQuacer gains a new DDT driver that consolidates external networking onto single SFP+ via PCIe, making one CPU card act as a network host to distribute… | ||
| Experiences and lessons we learned using kselftest and pote | Anders Roxell | 23m |
Linaro runs kselftest against LTS kernels on real hardware, builds test images and sends results via an email report wrapper around Squad; they add config… | ||
| Low Power AI Edge Hardware | Bin Lei | 20m |
Company ships four chips in two years and a tiny domain‑specific NPU matrix with 28,000 memory cores and dataflow architecture instead of a CPU instruction… | ||
| Inferencing at the edge and Fragmentation Challenges | Mark Charlebois | 27m |
Argues that edge inferencing suffers from massive hardware and software fragmentation and quantifies the space by contrasting device classes from… | ||
| End2End: From Silicon to System Delivery | Larry Wikelius | 31m |
Marvell delivers ThunderX2 end-to-end: silicon with 8 memory channels per SoC and out‑of‑order cores supporting up to four threads per core, a CRB reference… | ||
| Fedora IoT on 96Boards | Sahaj Sarup, Robert Wolff | 28m |
Fedora IoT boots 96Boards hardware using upstream kernels and many upstream device trees, producing one image that runs across dozens of SoCs. It uses… | ||
| Securing your next 96Boards design using Xilinx Zynq MPSoC | Kevin Keryk | 25m |
Practical recipe for protecting Ultra96 designs using Xilinx Zynq UltraScale+ MPSoC components: authenticate and decrypt PL bitstreams via the CSU/FSBL chain… | ||
| SMP Development on ARM Board | Satish Kumar | 24m |
Abstract
This session describes about SMP Development on ARM boards
Board used: Xilinx ZED Board (zynq ZC702) , ARM Cortex A9 Dual Core.
Kernel: zynq-linux… | ||
| Common Accelerator Framework Warpdrive update | Zhangfei Gao | 33m |
WarpDrive is a user‑space accelerator framework that bypasses kernel data paths with a kernel component and a user library, targets Hisilicon Kunpeng 920 and… | ||
| AOSP Device Upstreaming and Development with X15 | Praneeth Bajjuri, Mykhailo Sopiha | 23m |
TI BeagleBoard X15 is presented as an AOSP dev platform with concrete upstream results and remaining work: X15 mainlined in U-Boot and kernel with Android… | ||
| Open Source Software Stack for Heterogenous embedded devices | Tomas Evensen | 25m |
Heterogeneous Xilinx SoCs (Versal ACAP, Zynq MPSoC) require concrete runtime pieces: an OpenAMP clean‑room rpmsg/remoteproc implementation that lets non‑Linux… | ||
| TPM in TEE | Joakim Bech | 46m |
Claim: a full TPM 2.0 stack runs inside a TEE as a firmware TPM in TEE and speaks the standard TCG APIs. Implementation: Microsoft’s fTPM reference compiles… | ||
| Improvement and enhancement of LLVM for HPC | Masaki Arai | 19m |
Describes concrete LLVM changes for HPC: implements a modified register allocation strategy with virtual-register splitting and a virtual-register spill… | ||
| Cross compilation with clang and LLVM tools. | Peter Smith | 45m |
Clang and LLVM aim to be drop‑in for GCC but require explicit plumbing: use clang with --gcc-toolchain and --sysroot to point at a GCC cross toolchain, rebuild… | ||
| OpenAMP Libmetal Shared Memory Cross OS Interface | Wendy Liang | 31m |
OpenAMP extends by adding a libmetal-backed shared memory API and a Linux backend that aims for zero copy large‑data transfer instead of current rpmsg limits… | ||
| Why the Next Generation of HPC systems Needs Open Source Driven Standardization | Yan Fisher | 26m |
Panel argues that ARM64 is ready for primetime with ACPI/UEFI enablement, off‑the‑shelf systems and cloud instances (AWS A1) now available. Work focuses on… | ||
| The Practice of Cloud Native NFV Infrastructure Networking on Arm Platform | Trevor Tao | 28m |
Shows a concrete Arm-first stack: Contiv/VPP images and Kubernetes templates with VPP data plane giving microsecond ping times versus Flannel’s seconds-level… | ||
| Introduction to OpenAMP | Edward Mooring | 27m |
OpenAMP provides an open source framework that standardizes lifecycle, firmware loading and IPC between heterogeneous cores, implementing remoteproc/rpmsg and… | ||
| XDP for TSN | Ivan Khoronzhuk, Ilias Apalodimas | 28m |
Measured XDP/AF_XDP bypass reduces packet path latency from prior ~60–70 µs to under 10 µs end-to-user‑space; a 10,000‑packet test at 1,000 PPS shows ~3 µs… | ||
| LAVA Users Forum | Steve McIntyre | 41m |
Community LAVA updates include a move to a community-led GitLab with official releases, Debian packages and Docker images and docs at lava.software.org; CI… | ||
| Device power management and idle | Ulf Hansson | 59m |
Compares system-wide sleep (suspend-to-ram, suspend-to-idle, suspend-to-disk) and runtime PM using Linux PM core, dev_pm_ops callbacks and the runtime PM API… | ||
| KernelCI New Generation | Charles Daniel De Oliveira | 30m |
KernelCI New Generation reindexes KernelCI data into an ELK stack to make logs and test results searchable and visual, ingesting builders, LAVA boot JSON and… | ||
| Linux Kernel Power Management - 101 | Thara Gopinath | 47m |
Practical overview of kernel power and thermal features with concrete APIs and behaviors: suspend states (freeze/standby/mem/disk) and how wake works via… | ||
| Fantastic tracepoints and where to find them | Daniel Thompson | 31m |
Shows how to use only shell tools (cat, echo, trace_pipe) then add perf and BPF to inspect kernel activity, noting ~1,600–2,000 static tracepoints available on… | ||
| Arm NN and the plug-in framework | Narumol Prangnawarat, Matteo Martincigh, Les Bell | 52m |
Step‑by‑step guide to adding a hardware backend to Arm NN: use the Graph Builder API to build a network, call the optimizer which assigns layers to backends… | ||
| Vanguard Astra - Petascale ARM Platform for U.S. DOE/ASC Supercomputing | Kevin pedretti, Andrew Younge | 31m |
Astra is a 2.3 petaflop ARM prototype at Sandia with about 2,600 nodes (dual‑socket Cavium ThunderX2, 28 cores @2.0 GHz, eight DDR4 channels per socket)… | ||
| Retrospective on upstreaming VPN in AOSP | Sam Protsenko | 24m |
AOSP VPN moves from a downstream driver into an upstream kernel VPN driver and the retrospective shows what actually changed: one regression appears during… | ||
| OP-TEE: Shared memory between TAs | Jens Wiklander | 21m |
OP-TEE adds on-demand shared mapping of ELF read-only code pages so multiple TA instances or libraries reuse the same physical pages, cutting memory use; it… | ||
| Boot Requirements and Strategies for the Edge | François-Frédéric Ozog | 22m |
Goal is a concrete reference: implement UEFI secure boot, add a UEFI random number protocol, and use UEFI capsule updates for variable changes so runtime… | ||
| Bootloader testing in LAVA | Rémi Duraffort, Loic Poulain | 31m |
LAVA now supports Interactive testing that sends typed commands to a bootloader console, matches expected success and failure strings, and records named… | ||
| What is this Fuego thing and where is it going? | Tim Bird | 28m |
Fuego is a board-focused automated test framework that runs Jenkins + Docker on a host to build (cross‑compile) tests from source, deploy binaries to targets… | ||
| Living on Channel ZGC | Stuart Monteith | 24m |
Experimental ZGC is implemented for OpenJDK on AArch64 as a region-based concurrent copying collector supporting multi-terabyte virtual heaps and aiming for… | ||
| Using DTB overlays in OP-TEE | Bryan O'Donoghue | 20m |
Upstream OP-TEE adds an in-memory device tree overlay mechanism so a board port supplies an overlay that U-Boot merges into the kernel DTB at boot, avoiding… | ||
| 96Boards Mezzanine Community Update | Sahaj Sarup | 26m |
96Boards Mezzanine community maintains open-source templates on GitHub, runs bi-weekly design calls and a staging folder for project files, and partners with… | ||
| Bigtop 101 | Guodong Xu | 23m |
Bigtop 1.3.0 builds Apache Hadoop ecosystem packages into RPMs and DEBs inside a containerized, reproducible build image using Docker (plus docker-compose and… | ||
| Indoor Position with IVY5661 96Boards | Bo Dong | 19m |
IVY5661 is a $39 96Boards IoT board with a UniSoC UWP5661 SoC running Zephyr; it can act as AP or station and uses Wi‑Fi RTT to measure ranges to three or more… | ||
| Trusted Firmware Project update | Matteo Carlini | 41m |
Trusted Firmware reports concrete progress: TFA 2.1 is released with upstream support for new ARM cores and mitigation flags, a public test suite, and more… | ||
| Reliability, Availability, and Serviceability (RAS) on ARM64 status | Wei Fu | 24m |
Status update on ARM64 RAS and firmware integration: ARMv8 RAS extensions plus SDEI, APEI and a secure-partition prototype provide a firmware-first error flow… | ||
| Android on Docker Containers in a Nutshell | Khasim Syed Mohammed | 32m |
Abstract
Evaluating Android applications with out downloading them on the device, dedicating the display on handheld to UI thereby moving the real time data… | ||
| 96Boards CE Software Compliance - How and Why it is needed? | Manivannan Sadhasivam | 28m |
New 96Boards CE software compliance defines concrete requirements and a quarterly upstreaming roadmap: Q1 basic SOC patches, Q2 onboard storage and distro… | ||
| gVisor Container on Arm64: Let‘s Talk about Our Progress! | Bin Lu, Haibo Xu | 27m |
gVisor runs a user‑space kernel on Arm64 using the ptrace platform on Arm64 to intercept syscalls and provide isolation, with the runsc OCI runtime wired into… | ||
| Autoware Introduction | Theodore Grey, Servando German Serrano | 26m |
Autoware.AI on ROS1 provides modules for 3D point cloud localization, sensor fusion detection, prediction and planning and runs on over 30 vehicles in 20… | ||
| Opening Keynote at Linaro Connect Bangkok 2019 | Li Gong | 22m |
Argues that Linaro drives real engineering work: consolidating CoreSight into the Linux kernel, adding better Cortex-A support to QEMU, and hosting mnn hosted… | ||
| Linux Code of Conduct | Mishi Choudhary | 26m |
Mature FOSS projects adopt written codes of conduct with defined scope, a published issue‑resolution process, named contacts for reports and frictionless… | ||
| 96problems | John Stultz | 38m |
Describes 96Boards as fragmented: inconsistent USB‑C, M.2 placement and missing analog audio, a spec frozen since 2015, and upstreaming that typically takes… | ||
| Works on Arm cluster BOF | Ed Vielmetti | 29m |
Packet runs a WorksOnARM cluster offering bare metal arm64 nodes, weekly video office hours, a newsletter and Slack/IRC access; projects run native Debian and… | ||
| Arm Developer - what we learned about you, and the changes we're making | Chris Royston | 18m |
Analytics show a surprisingly large Asia audience and 60% software audience while site performance and documentation lag; most users have more than ten years’… | ||
| XDP Offload for OPC UA | Maxim Uvarov | 27m |
Experimental setup uses XDP sockets merged into the mainline kernel to deliver zero copy packets into userspace where a lightweight LWIP network stack runs… | ||
| Memory Recycling for Network Interfaces | Ilias Apalodimas | 20m |
Proposes a page_pool-based memory recycling API that gives drivers DMA mappings automatically and unifies XDP and skb paths, implemented as skb free… | ||
| Baselining AArch64 Java AOT Performance | Gary Morrison | 56m |
Comparison uses jaotc (Graal) AOT front-ending OpenJDK HotSpot across microbenchmarks and SpecJVM startup test, measuring 100 loop warmups and 15s runs… | ||
| Scheduling in CI/CD systems | Ed Vielmetti | 34m |
WorksOnARM/Packet runs CI across heterogeneous ARM servers and finds the key problems are build velocity, reproducibility, and missing native toolchains… | ||
| Automating test results analysis using neural networks | Carlos Hernández, Jonah Willis | 30m |
Neural network classifies embedded functional test results into pass, known bug, setup issue, test issue or new regression by combining 1024 tokens from tf‑idf… | ||
| Toolchain Strategy for Arm Server CPUs | Siddhesh Poyarekar | 20m |
Hardware vendors must time toolchain work to upstream schedules: GCC has an annual cycle with a stage one freeze in November and a stage three freeze in… | ||
| Autoware on ARM Improvements | Theodore Grey, Servando German Serrano | 17m |
Code coverage rises from 0.7% to 2.7% lines and 1.1% to 4.5% functions after adding unit tests and gating PRs; CI moves from Travis to GitLab CI via mirrored… | ||
| Libcamera: Making Complex Cameras Easy | Kieran Bingham | 25m |
An open C++ camera stack claims to map complex media-controller + V4L2 topologies into a single camera object, already capturing frames from UVC and Intel IPU3… | ||
| BFQ I/O scheduler: more throughput, control and efficiency | Paolo Valente | 49m |
BFQ claims up to five‑times higher throughput in hard server scenarios and reports a ~150% throughput uplift and ~80% faster app start in recent changes. It… | ||
| Xen Dom0-less | Stefano Stabellini | 32m |
Claim: Xen’s Dom0-less feature boots Xen plus multiple guest kernels and initrds in parallel so a small critical guest goes from ~10s to boot critical app in… | ||
| TF-A: Dynamic Configuration and PIE support | Soby Mathew | 44m |
Trusted Firmware-A adds runtime device-style config files (example TB_FW_config) passed as extra boot arguments between BL stages so stages can enable/disable… | ||
| Designing a next generation ARM Developer Platform | David Tischler, Carl Perry, Sahaj Sarup, Ed Vielmetti | 24m |
A proposal for an ARM developer desktop called ADUC aims to sit between Raspberry Pi and servers by using a 96Boards compute module on a Mini‑ITX carrier with… | ||
| Software architecture for High-end IoT devices | Tushar Khandelwal, Sudipto Paul | 30m |
Corstone‑700 combines Cortex‑A and Cortex‑M cores with a hardware firewall IP and interrupt router to sandbox address spaces and route interrupts, plus an MHU… | ||
| VMs in a container-centric world | Alex Bennée | 27m |
Containers escape by targeting kernel feature edges — one public case overwrote the host runC binary — so projects combine VMs and containers to regain… | ||
| Opensource GPU Drivers BoF | Tomeu Vizoso, Rob Herring | 48m |
Status update on mainline ARM GPU support: vc4/v3d run on products (v3d uses an MMU; vc4 remains CMA and shows OOM pains), Lima and Panfrost kernel drivers are… | ||
| Cluster Idling | Ulf Hansson, Vincent Guittot | 56m |
Describes an extension to CPU idle using the generic PM domain and a new genpd governor that chooses cluster-wide idle states by aggregating per‑CPU timers and… | ||
| Enable deepest suspend power state on S2I | Chunyan Zhang, Vincent Guittot | 28m |
Goal is to make suspend-to-idle reach the same low power as suspend-to-RAM by adding a deepest C-state and using cluster idling with correct power-domain… | ||
| AArch64 and Apache Bigtop: empowering big data everywhere | Jun He, Evans Ye, Ganesh Raju | 30m |
Concrete progress enabling ARM servers: AArch64 support in Apache Bigtop is upstreamed and officially added (picked-up server support in v1.3.0), with the… | ||
| Arm Server Manageability and OpenBMC | Dong Wei | 28m |
Arm is developing a concrete manageability spec called SBMG server base manageability guide with subteams defining remote debug via BMC, platform monitoring… | ||
| IOT CUBE BOX – New ST NBIOT IOT Platform | Guy Zohar | 24m |
Abstract
IOT CUBE BOX – Shiratech First Indoor/Outdoor out of the Box IOT solution Based on 96Board IOT Extended addition
based on New ST CPU + Quectel BG96… | ||
| New Technologies in the Arm Architecture | Nigel Stephens | 52m |
Announces two committed Arm architecture features: SVE2 (adds NEON-style DSP/media ops and extra instructions) and TME (hardware transactional memory with… | ||
| PMWG Non upstreamable part of EAS | Chris Redpath, Vincent Guittot | 41m |
Approximately 2,000 lines of Android-only EAS code exist, about 1,500 of which implement sketching. The plan moves placement and frequency decisions onto… | ||
| Linux sensor device driver programming with 96Boards | Koki Okada | 09m |
Overview of concrete tooling and results: describes UPM userspace drivers (C++, Python, Node.js) that talk to hardware over SPI/I2C/GPIO, contrasts using the… | ||
| Challenging idle period predictions - Where are we ? | Daniel Lezcano | 54m |
Interrupt-driven idle-state selection uses a per-CPU per-interrupt circular timing buffer (upstream irq_timings) that records recent timestamps, reduces values… | ||
| DRM HW Composer for Beagle X15 Board | Mykhailo Sopiha | 11m |
Adapts AOSP DRM Hardware Composer to Beagle X15 (4.14 kernel) by adding a new platform section, headers for the board buffer structure and small platform files… | ||
| The Linux Kernel Scheduler - Overview | Viresh Kumar | 41m |
Abstract
This session will cover the basics of the Linux kernel scheduler. It has already been presented twice at previous connects.
Viresh Kumar
Working in… | ||
| Empowering High Performance User Space Network Stack on Arm | Trevor Tao | 25m |
F-Stack is ported to Arm as a FreeBSD-based user‑space TCP/IP stack with F-Stack default TCP buffer 16KB vs kernel 4KB, uses DPDK for kernel bypass, and runs… | ||
| SUSE Linux from A to Z: Arm from big to LITTLE | Andreas Färber | 18m |
SUSE Linux Enterprise Server for Arm is built natively and tested end to end using Open Build Service to produce ISOs and Open QA to run automated test suites… | ||
| The value of being a 96Boards Partner | Kevin Keryk | 01m |
Partnership gives direct engineering collaboration inside the 96Boards ecosystem and access to the Xilinx SoC on the Ultra96 platform, plus coordinated work… | ||
| LAVA community enabled testing | Rémi Duraffort | 22m |
LAVA Federation runs daily tests by spinning up the latest server and dispatcher Docker images and connecting them to community labs so CI can borrow real… | ||
| Trusted Firmware M – what’s cooking? | Miklos Balint | 23m |
Trusted Firmware-M is an open source PSA implementation for Arm M that ships two partition models: a shared-memory function-call model and an IPC/thread model… | ||
| Introduction to Linaro | Li Gong | 01m |
Linaro positions itself as a central role in the ARM ecosystem, coordinating upstream contributions and cross-company engineering so vendors avoid duplicated… | ||
| Debugging with OP-TEE | Sumit Garg | 25m |
Overview of concrete debugging work for OP-TEE: describes existing techniques (secure UART abort dumps, symbolize.py to turn TA addresses into symbols, GProf… | ||
| Arm on Arm Panel Discussion 2019 | Larry Wikelius, Grant Likely, Jacob Smith, David Rusling | 46m |
Panel documents that real Arm servers and cloud offerings now exist: vendors ship systems with Ampere and ThunderX2 silicon, a Lenovo-Ampere box launches with… | ||
| nothing matches | ||