The evolution of the Linux kernel necessitates the shedding of legacy technical debt—the "ballast" that inevitably hampers the advancement of modern systems. Maintaining support for decade-old hardware has become a liability, bloating the codebase and stifling the pace of innovation. The developer community is now embarking on a comprehensive overhaul of the ARM subsystem, aiming to decisively close the chapter on obsolete platforms. This move signals a strategic shift toward a more rigorous approach to hardware support lifecycles.
Sustaining the viability of one of the world's most complex software projects requires a delicate equilibrium between backward compatibility and architectural evolution. Arnd Bergmann, who oversees the ARM subsystem of the Linux kernel, has initiated a new phase of code cleanup aimed at purging support for platforms that have long since ceased to be relevant.
The deprecation lifecycle will be implemented in stages: targeted platforms will first be marked as "deprecated," serving as a final warning to any remaining users, before their code is entirely excised from the kernel tree. This revision is slated for completion by early 2027. Notably, this is not the first attempt at such an "audit"—similar initiatives were launched in 2024 but were subsequently paused. This time, developers intend to see the process through to completion, aligning the final phase with the release of a Long-Term Support (LTS) kernel version to ensure a stable transition for the enterprise sector.
The list of candidates for removal encompasses a wide array of legacy technologies. Primarily targeted are platforms that were never migrated to the Device Tree—the modern hardware description standard that eliminated the need for cumbersome static files for every individual board. Additionally, the kernel will drop support for early ARMv6 processors (including the ARM1136r0), OMAP24xx and i.MX31 series Systems-on-Chip (SoCs), the iWMMXt extension, and the legacy OABI interface.
Particular focus has been placed on the Cortex-M3, M4, and M7 microcontrollers, whose support in the mainline Linux kernel will also be terminated. Furthermore, the big-endian mode for ARMv7 will be reclassified as BROKEN. In kernel development terminology, this signifies that while the code may remain nominally present, active testing of the configuration will cease, and its stability will no longer be guaranteed.
The logic behind this decision is driven less by the chronological age of the hardware than by the absence of an active community surrounding it. When a platform no longer has users or developers willing to test new kernel versions and patch bugs, it effectively becomes "dead code." Maintaining such systems unnecessarily inflates the codebase and creates friction during the modernization of contemporary kernel subsystems, as every change must be vetted for compatibility with this legacy baggage.
For the vast majority of modern users, these changes will be entirely transparent. The impact is limited to highly specialized embedded devices and legacy boards that have either been discontinued or are running on heavily modified, "frozen" kernel forks. The modern ARM ecosystem—ranging from current single-board computers and smartphones to high-performance server processors—remains completely unaffected, as it is built upon entirely different implementation standards.
The Evolutionary Trajectory of the Perfscale System
