Follow-up — 2026-07-04: mitigation held for 5 days; an accidental revert reproduced the freeze
After my last comment I disabled PCI runtime power management on the xHCI controllers by setting each xhci_hcd controller's power/control to on (i.e. runtime PM off — the controllers never runtime-suspend, so there is no suspend→resume cycle to fail). With that mitigation in effect the system ran stable for ~5 days, including its normal workload — the longest clean stretch since this bug was filed. I accidentally reverted the mitigation today. I was moving some files and so systemd lost its target to apply the fix on boot, and next boot the unit failed and the controllers fell back to power/control=auto (runtime PM re-enabled). Shortly after that revert, a USB hotplug (attaching an Android device for adb) — a controller resume/enumerate event on a CPU-side xHCI controller [1022:15b6]/[1022:15b7] — triggered the hard freeze again. The right global fix is not simply abandoning power management but this does localize the fault to xHCI runtime suspend -> resume on these specific AMD Raphael/Granite Ridge CPU-root-complex controllers, which fail to come back after "Controller not ready at resume -19" The right fix is probably a PCI/xHCI quirk for these device IDs (disabling runtime suspend or D3cold for the affected controllers) similar to other existing XHCI_* quirks for other vendors, not sure. Happy to test whatever would help narrow it if that's helpful. -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. https://bugs.launchpad.net/bugs/2158539 Title: Hard freezes on 7.0.0-22 and -27 with X670E/Ryzen; AMD xHCI controller not ready at resume (-19) To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2158539/+subscriptions -- ubuntu-bugs mailing list [email protected] https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
