Additional investigation since my previous report:
I recovered the complete kernel journal for the affected 2026-08-14
boot.
At 19:19:35 the Intel xHCI controller 0000:00:14.0 emitted seven messages of
the form:
Event dma ... for ep 5 status 1 not part of TD ...
followed immediately by:
xhci_hcd 0000:00:14.0: WARNING: Host System Error
In the same second DMAR reported:
DMAR: [DMA Read NO_PASID] Request device [00:14.0]
fault addr 0x400ff78e000
[fault reason 0x04] Access beyond MGAW
USB-attached storage immediately began returning:
hostbyte=DID_ERROR driverbyte=DRIVER_OK
and USB fallout followed.
This matches the observed failure: external USB keyboard and mouse stopped
producing input, physically disconnected devices remained stale in USB
enumeration, and the external display froze.
At 19:38:35 I manually unbound xHCI 0000:00:14.0. Rebinding it at 19:38:38
recreated the buses and USB devices began enumerating normally. This restored
the external USB keyboard/mouse and the frozen external display.
The laptop's built-in keyboard is not USB; it is an AT Raw Set 2 keyboard on
i8042/serio0. It remained nonfunctional after the xHCI reset. An atkbd
unbind/rebind re-registered it but did not restore usable input. A full reboot
restored it.
There were also repeated correctable PCIe AER Data Link Layer errors on NVMe
10000:e1:00.0, including BadTLP/BadDLLP, clustered shortly before the xHCI
failure. I am not claiming those caused the xHCI failure, but they may be
relevant to a broader PCIe/IOMMU/platform problem.
I have since upgraded the ASUS G814JVR firmware from BIOS 320 to BIOS
322.
During that work I independently confirmed multiple firmware ACPI namespace
defects, including a DSDT <-> INTEL xh_rplsb SSDT collision which reproduces
the boot-time:
\_SB.PC00.XHCI.RHUB.HSxx._UPC/_PLD
AE_ALREADY_EXISTS
errors.
The same ACPI defects remain with BIOS 322, and the implicated XHCI SSDT is
byte-for-byte identical between BIOS 320 and 322. Upstream ACPICA 20260408
also reproduces the firmware namespace collisions.
I have filed that ACPI/firmware issue separately as:
Bug #[2163555]
I am cross-referencing it here because it concerns the same 00:14.0/XHCI
platform, but I am NOT claiming that the ACPI namespace defect has been proven
to cause this runtime xHCI Host System Error.
The key runtime signature in this bug remains:
xHCI Event DMA/TD mismatch
->
xHCI Host System Error
->
DMAR DMA-read fault for device 00:14.0, "Access beyond MGAW"
->
USB/storage/input failure
Does this signature suggest an xHCI controller/driver issue, an IOMMU/DMA
mapping problem, or a broader platform/firmware failure?
System is currently Ubuntu 26.04 LTS with
7.0.0-29-generic / Ubuntu 7.0.0-29.29-generic.
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/2163551
Title:
Intel Raptor Lake xHCI stalls with stale USB devices; USB input and
external display freeze, i8042 keyboard also stops responding
To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2163551/+subscriptions
--
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
