XFS Could Gain a Self-Healing Feature in Linux Kernel 7.0

The XFS filesystem may receive groundbreaking self-healing capabilities in the Linux kernel 7.0 release cycle, based on a pull request from XFS maintainer Darrick J. Wong.

The proposed feature, titled “xfs: autonomous self-healing of filesystems,” targets the 7.0 merge window and introduces a real-time filesystem problem reporting system that enables userspace applications to perform automatic repairs. This implementation leverages new VFS error-reporting infrastructure developed by Amutable’s CTO Christian Brauner, which is also slated for Linux kernel 7.0.

At its core, the enhancement adds kernel functionality that broadcasts health events whenever XFS detects issues including metadata corruption, file I/O failures, media verification problems, or major state changes like shutdowns and unmounts.

Instead of relying solely on kernel logs, these events are delivered through a dedicated anonymous file descriptor. Userspace applications holding CAP_SYS_ADMIN privileges can access this descriptor and retrieve event information as C structures.

Events are queued internally with safeguards to prevent resource depletion. The architecture enables reporting of multiple health issues simultaneously without interfering with regular filesystem operations.

The patchset further introduces a media verification ioctl. When media checks detect problems, the findings are channeled into the same health monitoring framework, ensuring uniform reporting of integrity concerns.

For userspace management, a new daemon named xfs_healer will process these health events and trigger automatic repairs. The daemon integrates with systemd for service management and supports automatic activation via fanotify. It only prevents unmounting when actively performing repairs.

If merged, this development would fundamentally transform XFS error handling during runtime. Historically, repairs were primarily reactive processes requiring manual intervention with utilities like xfs_repair. This new framework enables continuous monitoring and potentially autonomous resolution of certain issues.

Currently, these modifications are proposed for Linux kernel 7.0 and await mainline integration.