Read-only Rootfs: Theory and Practice for Embedded Systems

Explore the theory and practice of implementing a read-only rootfs in embedded Linux systems through this informative conference talk. Discover how configuring a read-only rootfs enhances system robustness, reduces flash storage wear, and simplifies system image updates and factory resets. Learn to identify components requiring state storage, distinguish between volatile and non-volatile states, and implement various techniques for mapping writes onto appropriate storage types. Follow along with a live demonstration using a standard Yocto Project build to understand the necessary changes for achieving a real-world embedded system with read-only rootfs. Delve into the implications for software image updates and gain insights from Chris Simmonds, an experienced embedded Linux consultant and trainer. Cover topics such as mount points, memory areas, stateless systems, journaling, path names, first boot challenges, log files, random seed generation, and Android implementations. Conclude with a Q&A session addressing advantages, disadvantages, symlinks, mounts, systemd, and volatile storage in the context of read-only rootfs systems.

Introduction
Readonly rootfs
Why readonly
Image updates
Mount rootfs
Memory areas
Stateless
Enlightenment
Example
Blockdump
Journaling
Verta
Path names
Problems on first boot
Log files
Random seed
Dropout
First boot
Android
Conclusion
QA
Advantages and Disadvantages
symlinks
mount
systemd
volatilepast