The Great Pruning: Linux Prepares to Sunset Legacy Arm Support

In a move that underscores the relentless evolution of the world’s most significant open-source project, the Linux kernel is preparing for a significant architectural cleanup. Kernel maintainers have recently moved to formalize the deprecation and eventual removal of a vast array of legacy 32-bit Arm platform code. This initiative, spearheaded by long-time kernel contributor Arnd Bergmann, aims to shed decades of technical debt, effectively closing the book on computing hardware that defined the transition from the late 1990s to the early 2000s.

The Main Facts: A Massive Cleanup Operation

At the heart of this development is a series of thirteen patches submitted to the Linux kernel mailing list. These patches represent the first major strike in a long-anticipated plan to excise obsolete platform-specific code from the arch/arm/ directory.

The scope of this "spring cleaning" is immense. Bergmann, who has been advocating for this transition for several years, noted that the full scope of the deprecated code encompasses approximately 300 individual patches. The current submission focuses specifically on the removal of arch/arm/mach-* files—the board-specific machine code—along with the corresponding Kconfig entries that allowed these platforms to be compiled into the kernel.

The proposed timeline suggests that these removals will take effect during the Linux 7.4 release cycle. Given the current cadence of kernel development, Linux 7.4 is expected to reach the public between late December 2026 and early January 2027. Once the infrastructure code is removed in 7.4, the specific CPU support for these architectures is slated for formal exclusion in the following release, Linux 7.5.

Chronology: A Long Road to Deprecation

The decision to drop support for these platforms is not a sudden whim; it is the culmination of a multi-year strategy to streamline the kernel’s footprint.

The Stalled Efforts of 2024–2025

The discussion regarding the removal of legacy Arm support has been active since at least 2024. An earlier proposal aimed to finalize these deletions following the release of a matching Long Term Support (LTS) kernel. However, as Bergmann admitted in his July 2026 mailing list thread, the project suffered from a significant period of stagnation. In an act of transparency, Bergmann apologized to the community for allowing the effort to remain dormant for nearly two years, a delay that ultimately allowed the code to persist well past its intended shelf life.

The Turning Point

By the time the September 2026 patches were submitted, the consensus within the kernel development community had shifted. With newer architectures like Apple’s M3 silicon and advanced Qualcomm and Samsung chipsets demanding more attention and optimized drivers, the overhead of maintaining hardware that is nearly thirty years old became increasingly difficult to justify. The current push is not merely a request for deletion; it is a methodical execution of a plan that has been vetted through multiple rounds of debate on the Linux Kernel Mailing List (LKML).

The Hardware Under the Knife

The list of platforms facing sunset is a veritable "who’s who" of computing history, representing a time when embedded systems were just beginning to find their footing. Among the systems set to lose mainline kernel support are:

• Early Embedded Pioneers: The SA1100 boards, which powered many of the earliest handheld devices, and the Acorn RiscPC, a staple of British computing history.
• System-on-Chip (SoC) Families: Legacy PXA board files, as well as the Orion, Dove, and MV78xx0 boards.
• Microcontroller Series: The STM32F4, F7, and H7 series, which have long served as workhorses for industrial automation and hobbyist electronics.
• Legacy Processors: Support for OMAP24xx, i.MX31, and various no-MMU i.MX chips, along with the Axxia, LPC18xx, Versatile MPS2, and AT91 SAMV7.

It is critical to clarify what this means for users: The hardware will not self-destruct. The devices will continue to function exactly as they do today. However, they will be relegated to older, "frozen" versions of the Linux kernel. Any attempt to boot a modern, mainline kernel from 7.4 or 7.5 onwards will be met with failure, as the necessary drivers and boot sequences will no longer exist within the source tree.

Linus Torvalds and the Philosophy of Deletion

The philosophy behind this removal is deeply rooted in the pragmatic approach of Linus Torvalds. Throughout his tenure as the project’s lead, Torvalds has maintained that the kernel cannot be a museum of dead technology.

His stance has been consistent: if code is no longer being actively tested, utilized, or maintained by a community, it becomes a liability. It creates "dead weight" that complicates refactoring and security auditing. A notable precedent occurred in 2022, when Torvalds pushed for the removal of the i486 CPU series, citing the lack of any rational reason for developers to continue optimizing for hardware that had been effectively obsolete for decades.

By early 2025, the removal of i386-era code was largely treated as a formality. The removal of the legacy Arm platforms follows this exact lineage of thinking. The kernel must be allowed to shed its past to effectively support the future.

Implications: The Shifting Kernel Landscape

The removal of these 300 patches carries significant implications for the Linux ecosystem.

The Cost of Maintenance

Maintaining support for obsolete hardware is not free. Every line of code in the kernel must be reviewed when global changes are made. By removing legacy platforms, maintainers can reduce the "surface area" of the kernel, allowing for faster compilation, smaller binary sizes, and a reduced risk of security vulnerabilities being introduced through obscure, rarely-tested code paths.

A Focus on Modern Innovation

While legacy support is being pruned, the kernel is rapidly expanding in other directions. The same release cycles that are sunsetting 1990s hardware are simultaneously introducing sophisticated support for cutting-edge silicon. The recent inclusion of Apple’s M3 Pro, Max, and Ultra, as well as Samsung’s Exynos 1580 and Qualcomm’s latest offerings, demonstrates that the kernel is prioritizing the hardware that defines the current—and future—computing landscape.

The Preservation Alternative

For those who rely on this legacy hardware—whether for industrial, educational, or historical reasons—the news is not entirely bleak. Because Linux is open-source, the code does not disappear from the world. It remains in the git history. Enterprises and enthusiasts who require support for these platforms can continue to maintain their own forks or contribute to long-term community-maintained "retro-kernel" projects.

Arnd Bergmann has deliberately left the door open for dissent, acknowledging that there are still niche users and embedded systems running these chips. If a compelling case can be made that these platforms are still being used in a critical capacity with active development, the removal patches could theoretically be delayed. However, as of now, the momentum is clearly in favor of modernization.

Conclusion: Looking Forward

The upcoming removal of legacy Arm support is a testament to the longevity of Linux. It is because Linux has been so successful for so long that we even have to deal with the problem of removing hardware that has been supported for over two decades.

As we approach the 7.4 release, the Linux kernel is proving once again that it is a living entity. It evolves by consuming new advancements and shedding the remnants of its past. For the vast majority of users, this change will be invisible, yet it is essential for the continued health and performance of the system. We are witnessing the end of an era for 32-bit Arm, but in doing so, we are clearing the path for the next generation of computing.