Devlink Health Reporting & Recovery System Queued For Linux 5.1 Kernel

Within the networking subsystem of the Linux kernel one of the changes we are most looking forward to hopefully seeing for Linux 5.1 would be the long-awaited WireGuard, but another interesting feature was queued this past week into net-next.

Networking maintainer David Miller has queued the Devlink Health Reporting & Recovery System into the net-next tree ahead of the Linux 5.1 merge window. This infrastructure allows for real-time alerting of when "something bad" happens to a PCI device in order to generalize the debug information, self-healing, and other information. Eran Ben Elisha of the patch series explains, "The main idea is to unify and centralize driver health reports in the generic devlink instance and allow the user to set different attributes of the health reporting and recovery procedures."

This Devlink Health Reporting & Recovery System could be for problems ranging from transmission errors to PCI errors, firmware problems, and more. At this stage the only driver implementing the necessary hooks for this new devlink infrastructure is the Mellanox mlx5e driver, but surely with time more drivers will likely be onboard for this improved infrastructure for error handling / recovery.

More details on this pending work via this net-next commit.