Desktop and graphics: procfs rework and Emacs AI policy fights

Kernel developers debated a major procfs refactor aimed at cutting pinning attack surface and page-fault stalls, drawing sharp pushback on process. Emacs maintainers meanwhile contested permanent bans on LLM-generated code and related package submissions over long-term maintainability.

Procfs move toward kernfs model draws Linus complaint

Christian Brauner circulated a resend of a seven-patch series that would rework procfs along the lines of what kernfs already does, seeking to remove pinning attack surface. The discussion on dri-devel grew heated after Linus Torvalds objected bluntly to the volume of resends and the resulting mailbox spam. Readers tracking core VFS and security interfaces should note the scale of the proposed change and the process friction around it.

Emacs-devel disputes permanent ban on LLM contributions

An emacs-devel thread examined whether to bar LLM-generated contributions outright, focusing on maintainability burdens and broader community risks rather than copyright alone. Roughly twenty participants argued the issue across dozens of messages. The outcome matters for anyone contributing to or relying on Emacs packages, because it would set lasting policy on acceptable sources of code.

RFC series refactors procfs after kernfs pattern

Christian Brauner posted an RFC seven-patch series to move procfs toward the kernfs approach specifically to eliminate page-fault stalls when entries are removed. Linus Torvalds replied with an alternative approach. The technical debate affects how /proc behaves under concurrent removal and is relevant to anyone working on kernel interfaces that expose process state.

ELPA submission of vibe-spell sparks GPL and maintenance debate

A submission of vibe-spell, described as an LLM-translated Hunspell, to ELPA prompted heated discussion on emacs-devel about GPL source requirements and ongoing maintenance policy. Fourteen participants exchanged views on whether such packages belong in the archive. The thread highlights governance questions that will shape what external or machine-assisted packages Emacs can accept.

UBWC buffer layout defined for drm/msm and media

A seven-patch series documents the UBWC buffer layout for Qualcomm formats and adds validation that previously accepted undersized buffers. The work spans drm/msm and media. Developers targeting QCOM hardware gain clearer layout rules and stronger checks against malformed buffers.

Qualcomm DSP accelerator driver reaches v4

Version 4 of a twenty-five-patch series proposes an accel/qda DRM accelerator driver intended to replace fastrpc for Qualcomm DSPs. Maintainers flagged ABI and code issues during review. The series is of interest to anyone integrating DSP offload on Qualcomm platforms through the DRM accel subsystem.

Proposal to consolidate DRM trees for linux-next

An RFC floated replacing most DRM subsystem trees with a single drm-for-linux-next branch (also referred to as drm-tip) to reduce conflicts in linux-next. Eight participants discussed the consolidation. The change would simplify integration for DRM contributors if adopted.

DRM_MODE_ATOMIC_RESET flag added in v8

A seven-patch v8 series introduces the DRM_MODE_ATOMIC_RESET flag together with helpers that atomically reset all KMS object states to defaults. The work remains narrowly technical between two participants. It gives compositor and driver authors a cleaner path to restore baseline KMS state.