USB host controller drivers (xHCI, EHCI, OHCI, UHCI, and platform HCDs)
The set of low-level controllers that drive the USB ports on virtually every Linux machine, covering the modern xHCI standard used for USB 3.x and most current hardware as well as the older EHCI (USB 2.0), OHCI and UHCI (USB 1.1) controllers still found on legacy chipsets, plus various SoC-integrated platform variants. Without these, no USB keyboard, storage device, webcam, or hub would work.
recommendation
It should stay because this is the core collection of USB host controller drivers that every modern Linux system relies on to talk to USB devices. Active 2025–2026 fixes to xHCI debugfs and sideband code show the subsystem is being maintained, not retired, and xHCI in particular is the current standard shipped on essentially all new PCs, servers, and embedded boards. Only individual obsolete leaf drivers are being trimmed; the subsystem as a whole is fundamental infrastructure.
repository signals
sources
- docs.kernel.org
Kernel documentation describes Linux host-controller drivers as current core USB infrastructure and notes modern PCs typically expose multiple USB 3.x and legacy USB buses.
- en.wikipedia.org
xHCI is the current unified USB host-controller standard, intended to replace older UHCI/OHCI/EHCI families and used for modern USB 3.x-era hardware.
- spinics.net
There was active 2026 stable-fix traffic for drivers/usb/host/xhci-debugfs.c, showing ongoing upstream maintenance rather than deprecation.
- spinics.net
A 2026 patch series updated xhci-sideband logic in drivers/usb/host, again indicating active feature/bug-fix work instead of subsystem removal.
codex reasoning notes (technical)
This directory is an active subsystem directory containing many real USB host-controller drivers, not a helper library. Shell inspection of the tree showed current xHCI/UHCI/OHCI/EHCI and platform HCD sources; a local `git log` also showed fresh 2025-2026 fixes plus one targeted removal of an obsolete leaf driver, not a subsystem-wide retirement. URLs were obtained via web search (`docs.kernel.org`, Wikipedia, and spinics/lore mirrors). No credible subsystem-level removal/deprecation discussion was found; evidence instead shows broad current deployment on modern systems and continuing upstream maintenance, so keep is the defensible outcome.