Talks
Global shifts and geopolitics in recent times have played a significant role in policy decisions. What do they mean for our globally collaborative worlds of AI and open source software? Amanda will take the audience on the journey of geopolitics and polic
There has been a tremendous amount of change in the Arm ecosystem that is transforming the industry. In this opening keynote to Linaro Connect 2025, Linaro CTO Grant Likely will look at where we are today and how we use all this technology to boost the ne
In this keynote John Thompson will talk about changes in the IoT landscape and how Arm is responding to these changes.
In this keynote Arm will talk about their priorities this coming year.
In this presentation, Marcus and Pedro will share the latest for Arm powered Windows devices and recent projects where community collaboration has played a pivotal role in unblocking critical components for the Windows software ecosystem for Arm. Addition
Open Source is pervasive and has come to dominate the embedded market, as well as laptops, servers and clouds. The embedded space however includes products like automobiles, medical devices, planes, etc. where safety is a consideration. Over
SystemReady has steadily evolved, delivering fundamental capabilities that enable Linux distributions to boot with minimal effort on compliant systems. This foundation has made such systems not only easily deployable but also maintainable and extensible w
Running AI workloads on ARM can be challenging due to the need for specialized hardware configurations, runtime optimizations, and dependency management. However, with the rise of tools like RamaLama, krunkit, libkrun, podman-machine, llama.cpp and vllm,
The TuxSuite ecosystem has evolved. This talk will highlight what the TuxSuite Ecosystem is and focuses on some of the exciting new feature additions to the project. This talk will focus on korg-gcc for optimized compilation of the Linux Kernel and TuxLAV
Technical presentation - 30 minutes (including q&a)
Wednesday, 10:30 - 10:55
As ARM64 adoption grows across PCs and embedded devices, developers need robust tooling to build, optimize, and debug applications efficiently. This session explores the latest advancements in developer tools, compilers, debuggers, emulation strategies, a
Technical presentation - 30 minutes (including q&a)
Wednesday, 11:00 - 11:25
The released Snapdragon X Elite laptops are featuring a Video Processing Unit (VPU) to offload video encoding and decoding from the main CPU's. While this a common hardware building block in embedded systems, the use case in a generic laptop device for en
## Abstract: The Linux kernel runs on devices from phones to supercomputers. Frequent updates and new Arm hardware make it essential to ensure that changes do not break anything in Arm Ecosystem. This is where Linux Kernel Functional Testing (LKFT) plays
KleidiAI is an optimization library designed to accelerate Machine Learning(ML) workloads for best-in-class performance on Arm CPUs. Since its inception in May 2024, it has experienced an eventful few months with integrations into multiple ML frameworks.
EFI sits at the heart of the Arm SystemReady-IR compliance program and it's the primary reason OS booting has consistent behavior across a large deployment of heterogeneous platforms. On top of that we can benefit from EFI and make our testing easier and
The Trusted Firmware open source community project is now the mature and established collaborative platform around secure software development for Armv8-A, Armv9-A and Armv8-M based products. The session will cover recent project milestones, introduce new
In this talk, we’ll explore a practical, repeatable process for tracking and evaluating Linux Kernel performance on ARM64-based systems in various environments. We will discuss the motivations behind continuous kernel performance evaluation, common chal
Technical presentation - 30 minutes (including q&a)
Wednesday, 11:30 - 11:55
Recently, more ARM64 laptops have become available, with more powerful CPUs and better GPUs than before, making them a viable alternative to mainstream x86 devices for daily use. However, most commercial games still do not provide native ARM binaries, so
Generative AI (GenAI) is rapidly reshaping our daily life, and entire industries. Yet their increasing computational demands often hinder cost-effective deployment. This talk presents an end-to-end solution for accelerating GenAI workloads on Arm® Neover
Technical presentation - 30 minutes (including q&a)
Wednesday, 12:00 - 12:25
Arm Base System Architecture (BSA) and the SBBR recipe in the Arm Base Boot Requirements (BBR) built the hardware and firmware foundation for the generic OSes to install and boot on the compliant systems with no need for customization. However, there are
Our initiative aims to support the kernel community by enabling the KernelCI workflow on Qualcomm SoCs, thereby enhancing the stability of the Linux kernel. We have created a highly flexible environment to validate open-source Linux device drivers and IO
Introduces Linaro Trusted Substrate (TS) Arm SystemReady/UEFI firmware and Trusted Reference Stack (TRS) kernel, initramfs and rootfs prototype for secure boot with TPM for secure storage. Discusses upstreaming status, problems and solutions. Linaro has i
The evolution of on-device AI is transforming how developers build and deploy machine learning models, enabling real-time performance, enhanced privacy, and power-efficient execution. Windows on Snapdragon (WoS) provides a robust AI development ecosystem,
The introduction of the Linaro Automation Appliance has led to an opportunity to make use of it to enable running Das U-Boot testing on hardware. As this was breaking new ground there were of course issues to resolve on the way.
Operating System software is often closely tied to hardware, firmware, and the deployed distribution and build infrastructure. Even minor changes in the boot chain can necessitate corresponding kernel adjustments. Similarly, small modifications in hardwar
In past talks I've often talked about how to write this simplest possible static check. This talk will be the opposite. The Smatch check for locking bugs is the oldest, largest and most complicated check. I have re-written this check since the last Linaro
Description * Over the last 18 months, Qualcomm support in U-Boot via mach-snapdragon has gone from being near-unused, relying on downstream DT and hard-coded board specific configuration, to being one of the most modern and generic architectures in U-Boo
Panel discussion on Windows on Arm with Microsoft, Arm, Qualcomm, Blender
Representatives from both Linaro and Schneider Electric reflect upon the relationship built over the past six years, as Schneider Electric increasingly moves towards Linux and open-source. The engagement with Linaro began with LTS engineering, and has evo
Technical presentation - 30 minutes (including q&a)
Wednesday, 14:00 - 14:25
When migrating large-scale x86 applications to Arm, organizations often face the challenge of achieving optimal performance without a complete architectural rewrite. This talk demystifies the performance tuning process for x86-to-Arm migrations, offering
Over the last 2 years, Qualcomm and Linaro have been collaborating to enable support for open source and industry standard boot firmware for Qualcomms IoT & robotics platforms. This has led to the revival of Snapdragon support in U-Boot (now the most
The Linux Kernel, along with its platform configuration, is today limited to support a fixed number of system-wide low power-states. Suspend-To-Ram and Suspend-To-Idle are two examples of these low power-states. Moving forward, we have observed new use-ca
Technical presentation - 30 minutes (including q&a)
Wednesday, 14:30 - 14:55
Following OSPM'25 discussion, we will present the rework of the scheduler related to EAS that is under review on the mailing list. This includes the new push callback mechanism that adds more opportunity to migrate tasks across CPUs and the rework of ener
LLVM has become the dominant compiler for Windows on ARM, powering major open-source projects and surpassing competition in performance and adoption. But what’s still missing? This talk will explore the current state of LLVM on Windows on ARM, highlight
Technical presentation - 30 minutes (including q&a)
Wednesday, 15:00 - 15:25
PROPOSALS FOR LINARO CONNECT 2025 (LISBON) Topic: Presenting SIMD programming model, comparing ARM SIMD(NEON & SVEx) vs X86 SIMD(SSEx, AVXx) as one of the most important and efficient software performance optimization approaches. Title: SIMD programmi
A Trusted Platform Module (TPM) is a fundamental component of modern trusted computing systems, providing a secure vault for cryptographic operations that ensure system authenticity and integrity. Arm provides various security solutions for the ecosystem
Technical presentation - 30 minutes (including q&a)
Wednesday, 15:30 - 15:55
“Thread Affinity†controls which cores are favoured when distributing threads and processes. Getting the thread affinity right is key for getting the most out an MPI/OpenMP job, and bad affinity can often silently degrade the performance. For existing
In this talk, we will go through the use of readily available Large Language Models (LLMs) to facilitate engineering workflow when running Continuous Integration (CI) loops and analyzing regressions. For this presentation, Anthropic’s Claude Sonnet 3.5
This talk presents ongoing work at Arm to implement an fTPM SP in Trusted Services. The first functional version of the SP is already public. It implements the DEN0138 TPM CRB over FF-A specification as the communication layer. It contains the ms-20-tpm-r
Technical presentation - 30 minutes (including q&a)
Wednesday, 15:30 - 15:55
We will provide an overview of ACPI support on Radxa Orion O6 board in this session. The Radxa Orion O6 is a mini-ITX motherboard built around the CIX CD8180 12-core Armv9 SoC. We support a fully functional Linux including GPU, DPU, VPU, NPU, ISP, USB, PC
A session to share various small updates on technical work done by the LCG Android team. Some of the topics included will be:
Qualcomm aims to enable remote processors on the SA8775p SoC running Linux at EL2 (e.g., KVM). To support this configuration, the existing Qualcomm remoteproc PAS driver needs revision. The primary challenge lies in the differences in IOMMU translation ha
Bjorn will talk about the importance of convenience for developers and testers to run the upstream Linux kernel, followed by a review of past/existing solutions applied by the members of the Qualcomm upstream community, and a walk through of recent develo
This BoF aims at
High level topics for discussion are:
This BoF aims at providing a platform to discuss the state of Samsung Exynos and Google Tensor SoCs and their platforms in the Mainline Linux Kernel. Focusing mainly on. Where we are, Current Status of SoC drivers. Ongoing efforts on new Drivers/Framework
The Embedded Base Boot Requirements (EBBR) specification defines requirements for embedded systems to enable inter-operability between SoCs, hardware platforms, firmware implementations, and operating system distributions. This specification is key for Sy
Qualcomm is dedicated to shaping the future of computing through open-source innovation and community collaboration. In this keynote, we will delve into Qualcomm’s contributions to the Linux kernel, open-source AI frameworks, and edge computing, which a
Introduction: Brief overview of the global energy transition and its importance. Portugal's Commitment: Highlight Portugal's ambitious renewable energy targets and policies and successful history. Renewable Energy Sources: Discuss the key renewable energy
The FIDO Alliance’s FIDO Device Onboard (FDO) specification is an automatic onboarding protocol for edge nodes and IoT devices. It installs secrets and configuration data to securely connect devices with cloud and edge platforms. FDO's key benefits incl
Embedded computing typically serves a market characterized as low-volume, high-mix. Traditionally, the BIOS abstraction layer in x86 products has minimized the software support burden for embedded computer manufacturers.
Over the past 25 years, this abstra
This presentation will give an overview and status of the work on Virtio-msg. Virtio-msg is a new transport layer that allows Virtio to be used in new areas such as: Linux + RTOS on co-processor, Linux to secure world TEE , and across PCIe between differe
Ceph SPDK NVMe-oF Gateway is a SAN-like network block disk solution. It can be used in virtualization, cloud computing, data lakes, AI, etc. Compared to the Ceph-iSCSI Gateway, it supports the modern storage network NVMe-oF protocol, which features multi-
For Linux systems featuring without a BIOS that configures the PCI(e) bus (common for most of the ARM platforms), it is paramount to have a PCI(e) controller driver in the kernel to configure the bus and enumerate the endpoint devices. Such controller dri
In this session, we will talk about the actual process of enabling page-agnostic AOSP with the Linaro devboards. We will also cover some of the common potential problems, the issues we faced and how they were resolved. We also tested some of the popular a
Cuttlefish is a virtual device platform meant for Android development and supported by Google since 2017. I'll cover the state of the platform and what the team is working on this year. Amongst these projects is a generic Android Boot solution (GBL) we've
Introduce the features and benefits of NFSv4.1, NFSv4.2. The evolution of NFS will continue to play a critical role in cloud, HPC, and hybrid storage architectures.
As Web3 adoption accelerates, robust security measures are essential to safeguarding user assets and data. The Apache Teaclave TrustZone SDK, the recommended Rust SDK for OP-TEE, provides a memory-safe framework for developing Trusted Applications. Beyond
libkrun is a dynamic library providing Virtualization-based process isolation capabilities or, in fewer words, a VMM in library form. Written in Rust and designed for minimal boot time and small footprint, it has evolved since its initial inception as a c
The Arm64 Linux Kernel is constantly evolving to align with the Arm architecture enablement , spanning across 3 spheres : performance, compute and security. At Arm, we drive this kernel enablement story to facilitate our architecture adoption. The shared
UADK (Unified Acceleration Development Kit) was initially designed to support hardware accelerators. However, it has now expanded to include CPU-based accelerators like CE (Cryptographic Extensions) and SVE (Scalable Vector Extensions) to optimize perform
Generic Boot Loader (GBL) is a UEFI app under development at Google, aimed at reducing the bootloader changes Android vendors have to implement based on changing needs. We have been involved with enabling and validating GBL on our devboards using uBoot as
The talk will talk bout Virtio-msg protocol, its kernel implementation and how it works with FFA to interface pVM with host and trustzone via pKVM.
This presentation will cover how we at AMD in collaboration with Linaro, ARM and others made use of the emerging virtio-msg transport to enable virtio devices between Xen guests running in different and heterogeneous CPU clusters (x86 and ARM). We will co
Over the last few years, 64-bit Linux has made it from Servers, PCs and high-end embedded machines lower down in the market, everywhere including the smallest embedded Linux targets. This gives new challenges for users that rely on existing 32-bit hardwar
UADK(User-space Acceleration Development Kit) is an open-source framework designed to offload compute-intensive cryptographic algorithms to hardware accelerators. By integrating UADK with ARM cVMs, we enable hardware-accelerated cryptographic operations t
Feedback Driven Optimizations are a set of toolchain techniques that use traces from program execution to improve the performance of output binaries. In this talk, we will discuss the support for AutoFDO available for native code (C/C++) in Android usersp
Attestation is a process designed to establish trust in an entity by authenticating and verifying that the device has booted into a known, secure state. Building trust is a critical requirement in various use cases, such as device onboarding and the deplo
Maintaining an Android phone over many OS versions is still difficult. At Fairphone we want to support our phones for up to 10 years but it’s not a simple task given the dependencies on all the suppliers and technical difficulties connected to that. Weâ
Today, Xen is the leading open-source hypervisor in the automotive industry. Over the past decade, the Xen community has worked tirelessly to develop key features that now form a top-tier automotive solution. Capabilities such as Dom0less, Cache Coloring
This presentation will provide details of the Linux kernel driver for the new GIC architecture (GICv5) developed by Arm[1].
[1] https://developer.arm.com/documentation/aes0070
At Linaro Connect 2024, we announced that our next generation of Arm-branded toolchains would be fully open source. This is now available at https://github.com/arm/arm-toolchain, and our first releases will be available shortly after the OSS LLVM 20 relea
Given the shared nature of AWS instances, one might think it would be difficult to obtain accurate and reproducible benchmark results on them, due to interference from other instances running on the same physical machine. But in practice, how much variati
The presentation provides an overview of the audio subsystem on Qualcomm ARM devices, exploring the essential components required for hassle-free sound functionality on older and modern Qualcomm sound architectures -- Elite and Audioreach. The presentatio
Attested TLS is an essential ingredient of any confidential computing solution. In this talk, we explore the design space and the corresponding typical security and performance tradeoffs of attested TLS protocols. We also summarize the standardization eff
In this talk, I will describe the history and issues of providing a CSRNG in userland on Linux, from the initial proposal on glibc to provide arc4random (and the pitfalls to support old kernels), to the first short-lived implementation based on BSD, to th
Device Assignment (DA) is the method which allows a device to be uniquely assigned to an individual Realm, and to allow the Realm to attest the device before granting it access to the Realm’s contents. This session will cover the overview of Device assi
Technical presentation - 30 minutes (including q&a)
Regulatory licensing and compliance
Thursday, 14:30 - 14:55
In this talk, presented by a former Google Android Product Manager in charge of EU Digital Markets Act (now CEO of source.dev), we will explore key upcoming regulations related to device software longevity and security. We’ll examine how these regulator
Flang, the Fortran compiler frontend in the LLVM Project, is under active development. However, the Flang/LLVM community lacked comprehensive Fortran test-suites, unlike those available for C/C++. This prevented rapid development of Flang without regressi
Device assignment in Confidential Virtual Machines (CVMs) enables applications such as confidential machine learning and enhances CVM I/O performance. Current development relies on TDISP, which lacks support for devices available on the market. In this ta
AArch64's Scalable Vector Extension (SVE) and Scalable Matrix Extension (SME) have registers whose size are given by the SVE vector length — a parameter that can be changed by the running application at its convenience. GDB already supports this when de
The Runtime Security Engine (RSE) has emerged as a crucial component in modern security architectures, serving as a hardware root-of-trust and an isolated attestation enclave for A-profile compute subsystems. As computing environments evolve to support Co
Over the years Qualcomm camera subsystem IP evolved from a relatively simple ISP controller to a number of separately managed interconnected hardware components. The constantly rising complexity of Qualcomm camera IP entails reworks in the current CAMSS d
Technical presentation - 30 minutes (including q&a)
Regulatory licensing and compliance
Thursday, 15:30 - 15:55
OEMs are facing the prospect of mandatory compliance with cybersecurity regulations such as the EU Cyber Resilience Act (CRA). To comply, product manufacturers will need to pay more attention to security practices related to software and hardware supply c
Technical presentation - 30 minutes (including q&a)
Regulatory licensing and compliance
Thursday, 15:30 - 15:55
Bringing an existing codebase into MISRA compliance is known to be a challenging, risky, and time-consuming task. Yet, it is a necessity when a product needs functional safety certifications. Such an endeavor requires facing multiple tradeoffs and, conseq
With AI models handling sensitive data, security must extend beyond data protection at rest and in transit to also protect data during execution. Arm Confidential Compute Architecture (CCA) introduces Realm Virtual Machines (Realm VMs) that provides a tru
Opening up MIPI bayer only camera stacks with libcamera and GPU acceleration to create a standard open ISP stack which can be reused across x86, aarch64 and others.
BoF aims at providing platform to discuss state of Qualcomm SoCs and Platforms support in Mainline Linux Kernel. Focusing mainly on. -> Where we are, Current Status of Qualcomm drivers. -> Ongoing efforts on new Drivers/Frameworks. -> Pending pro
An open Birds of Feather session for all things AOSP and Android. Some of the initial list of topics are:
This is a discussion rather than a technical presentation. There's an ongoing effort in the kernel to add support for secure or restricted memory for various use cases, such as Secure Video Playback, Trusted UI, and Secure Video Recording. I've posted a p
The Cyber Resilience Act is a new EU law that requires all hardware and software products to comply. One requirement is the availability of free updates and fixing all security vulnerabilities starting in 2027.In this talk, Marta will explore the CRA's re
With Arm9 came CCA Realms, and with CCA, Confidential Computing comes to the Arm ecosystem. CCA allows Trusted Execution Environments (TEEs), providing hardware-based isolation for applications and, with attestation services, further security assura
In addition to improving all-scenario capabilities, the latest version of openEuler further enhances support for AI technologies, such as GMEM, hybrid inference, openEuler Intelligence, and AI development desktop: DevStation. When all of these technologie
Kernel debug mechanisms are traditionally reliant on a software handler, that can inspect or help inspecting a faulting kernel. It could be it a panic trigger or a specific event, the ability to start a failsafe kernel, or a working userspace. An embedded
Last year I presented about major changes coming to the TianoCore community, especially a migration to github. This year we'll present on how that went, future plans, and other changes in the community - including a shakeup of edk2-platforms.
Linaro has been a long time contributor to openEuler. Come to this session to learn about the technical importance of openEuler in the market and the latest advancements Linaro contributed to openEuler in various areas such as Big Data & Data Science,
Join us as we bring together renowned experts in cybersecurity regulations, supply chain challenges and confidential computing technologies to navigate the dynamic interplay of these critical domains. Evolving digital landscapes and escalating threats imp
KubeVirt is an incubating project in CNCF providing Kubernetes Virtualization API and runtime in order to define and manage virtual machines. In this presentation, I will share how we have successfully built and maintained a strong relationship with the K
Compute eXpress Link (CXL) has evolved substantially over the last few years as the specifications and software support have matured. (http://computeexpresslink.com/) This has enabled new use cases but also brings new challenges for the software ecosystem
As computationally intensive workloads with high-density processing requirements, big data and data science applications have always been crucial deployment scenarios for ARM server architectures. These domains place exceptional demands on CPU computation
The Arm Confidential Compute Architecture (Arm CCA) builds on top of the Armv9-A Realm Management Extension (RME) by providing a reference security architecture and open-source implementation of hypervisor-based confidential computing. This talk describes
Trusted Firmware-A provides an open-source reference implementation of the secure software for Armv8-A and Armv9-A. It provides SoC developers and OEMs with a reference trusted code base complying with the relevant Arm specifications for A-class processor
Recently I began an earnest effort to bring Rust support to Zephyr. This work is now in a Zephyr module and included in the mainline support. This talk will describe where we're at with this, what can be done with it, and why someone might want to use Rus
The first part of this presentation will provide an update on the work done by Linaro, Arm and community members over the past year to support confidential computing on Arm platforms. From there we will outline our plans for the coming year in
When seeking to use Rust for embedded development, there are several options, including bare-metal, a few rust-specific rtos implementations, and several existing RTOSes. This talk discusses why Zephyr is a good fit for an embedded Rust, and especially ho
MPAM for ARM stands for Memory System Resource Partitioning and Monitoring, it's a ARMv8.4 feature which can partition the cache and memory bandwidth for VMs, containers. In this topic will have the introduction of how MPAM works on Kunpeng ARM server, an
Firmware handoff specification 1.0 is now live along with the standalone Transfer List Library on tf.org. In this session will discuss about - Importing libTL in different firmware projects - Collaboration on development of specification and library - Ado
Today’s edge and IoT embedded systems must support multi-domain software, secure deployment and runtime, and in the case of FPGA SoCs dynamic hardware reconfigurations. This presentation will outline the system architecture AMD is implementing to suppor
The TF-RMM is a firmware component that runs at Realm EL2 and forms part of a system which implements the Arm Confidential Compute Architecture (Arm CCA). This talk discuses the EL0 app framework which allows to deprivilege parts of RMM to run in a separa
In this presentation, we explore the integration of Arm's Ethos-U NPUs with Executorch to enable efficient edge AI solutions. We leverage the TOSA (Tensor Operator Set Architecture) framework to optimize neural network operations, ensuring seamless compat
Return Oriented Programming (ROP) is a powerful technique where attackers divert a program by modifying a function's return address. Then by returning through a chain of code snippets found in the binary ("gadgets") they can make programs behave malicious