Open Source Summit North America 2025

Embedded Linux devices operate in constrained environments with limited storage, bandwidth, and connectivity. Traditional core dumps can be quite large, making it impractical for some of the more constrained embedded systems. Over the past year, we’ve tackled this challenge head-on—optimizing Linux core dumps directly on the device to reduce size, protect privacy, and enable better debugging for IoT developers.

What We’ll Cover:
Inside ELF Core Dumps – A look at the ELF structure and how it applies to core dumps.

On-Device Optimization – How we reduced core dump size by capturing only the first N bytes of each stack, minimizing storage and bandwidth impact.

Privacy-Preserving Debugging – How our custom built (in Rust!) on-device stack unwinder hooks into the core handler, and reduces a coredump to a set of PCs per frame to save space and prevent potential PII from leaking.

Scaling to Millions of Coredumps – Lessons learned from parsing an unprecedented volume of core dumps with Rust.

There are numerous tools to measure boot-time performance of Linux. However, there is no standard regression test of boot performance for Linux. This is due to a number of factors, including disparities in system performance, different requirements for quickly-needed functionality, and differences in bootloader, kernel and user-space configuration. In this session Tim will present a boot-time regression test that utilizes a collection of reference value data files for different platforms, kernel versions and configurations. A meta-data matching system is used to select an appropriate reference data file. Boot time data (including initcall durations, and the durations of pre-selected boot operations) is compared with reference values, in order to report regressions in boot-time duration for specific elements of the boot sequence. The upstream status of this effort, along with the test and supporting tools, as well as issues found with this approach, will be discussed.

We will describe in this session a variety of strategies and
techniques that were used to optimize the boot up time of the Linux kernel, initramfs, and systemd for Red Hat In Vehicle Operating System. The unreported time before the kernel timer is initialized will be described, and how to measure this on an aarch64 system. Some of the techniques include changes that are specific to the realtime (RT) kernel, such as temporarily tuning RCU on boot up. We’ll also cover how a new project called autoinit was used in the initramfs with systemd to replace udev. Some kernel patch sets that have been merged upstream that help the overall kernel boot speed will also be described.

The automotive world is rapidly moving towards software defined vehicles & zonal architecture. Ethernet is the backbone of zonal architecture. The bandwidth and response time requirements of ethernet are higher than ever. The network stack with Linux makes it a compelling choice of OS for Gateway/ADAS SOCs.

With growing number of ECUs in the car, the gateway SOC needs to be ethernet ready very early to support FOTA and telematics applications. All the ECUs sitting behind the gateway need ethernet functionality for communication. With ROM code, boot loader kernel & the user space application taking time to initialize, the goal of early ethernet is a tricky one.

The presentation aims to cover the optimizations done at each phase of the Linux boot to achieve early ethernet. Top level optimizations:

1. Streamlined the boot flow from bootloader to get to Linux kernel faster.
2. Opening the network device early with driver optimization allows the hardware to be initialized early
3. Configuring majority of network using networkd scripts allows the interfaces to be ready early

Although the demonstration is done using TI's DRA821 Gateway SOC, this is applicable to any SOC using Linux.

Modern Heterogeneous SoCs designed for automotive use cases integrate multiple remote processors (rprocs) for Real-Time Operations, like Cortex-R5F for handling Capture and Display, alongside a general-purpose processor like Cortex-A to run HLOS like Linux. Automotive requirements are Time-Critical, Safety-Oriented and often demand for the Real-Time rprocs to be booted early for use cases like Early Chime, Early Rear View Camera (RVC). This presentation discusses about the “Early Boot + Late Attach” feature in remoteproc framework, where-in the rprocs would be brought up early in the boot flow by bootloader, and later Linux would just “attach” into the pre-running rprocs to establish IPC & suspend/resume. It also explains how developers can adopt this approach and shares debugging lessons. The talk also addresses how the framework handles key challenges with this approach, like identifying a pre-running rproc and allocating the resources required by it (memory carveouts, mappings, trace buffers etc.) by parsing the pre-loaded firmware, rather than re-initializing the core like in normal boot flow. A live case study will be presented on Early RVC application showcasing this feature.

In this BOF, attendees will discuss the current status of Linux Boot, and boot-time reduction efforts for Linux systems. This will be an open discussion about the currently available tools, instrumentation, techniques, patches, documentation and online resources for developers working on Linux boot-time reduction.

If you are working on Linux boot-time, or plan to sometime in the near future, please join this discussion. You can describe your requirements, or tell us about your current Linux boot-time reduction activities or projects.