Renesas uPD720202 XHCI floods klog with warnings

[Phil R]

I have a 2 machine vision cameras each connected to a PCIe card based on the Renesas uPD720202 USB controller. During data transfer the following warning is spammed to klog:

Nov 23 11:30:30 $hostname kernel: xhci_hcd 0000:08:00.0: WARN Successful completion on short TX: needs XHCI_TRUST_TX_LENGTH quirk?
Nov 23 11:30:30 $hostname kernel: xhci_hcd 0000:07:00.0: WARN Successful completion on short TX: needs XHCI_TRUST_TX_LENGTH quirk?
Nov 23 11:30:30 $hostname kernel: xhci_hcd 0000:08:00.0: WARN Successful completion on short TX: needs XHCI_TRUST_TX_LENGTH quirk?
Nov 23 11:30:30 $hostname kernel: xhci_hcd 0000:07:00.0: WARN Successful completion on short TX: needs XHCI_TRUST_TX_LENGTH quirk?
Nov 23 11:30:30 $hostname kernel: xhci_hcd 0000:08:00.0: WARN Successful completion on short TX: needs XHCI_TRUST_TX_LENGTH quirk?
Nov 23 11:30:30 $hostname kernel: xhci_hcd 0000:07:00.0: WARN Successful completion on short TX: needs XHCI_TRUST_TX_LENGTH quirk?
Nov 23 11:30:30 $hostname kernel: xhci_hcd 0000:08:00.0: WARN Successful completion on short TX: needs XHCI_TRUST_TX_LENGTH quirk?
Nov 23 11:30:30 $hostname kernel: xhci_hcd 0000:07:00.0: WARN Successful completion on short TX: needs XHCI_TRUST_TX_LENGTH quirk?
Nov 23 11:30:30 $hostname kernel: xhci_hcd 0000:08:00.0: WARN Successful completion on short TX: needs XHCI_TRUST_TX_LENGTH quirk?
Nov 23 11:30:30 $hostname kernel: xhci_hcd 0000:07:00.0: WARN Successful completion on short TX: needs XHCI_TRUST_TX_LENGTH quirk?
Nov 23 11:30:35 $hostname kernel: handle_tx_event: 992 callbacks suppressed

Based on the frequency of messages I conclude that this happens each time a frame is transferred. (2 cameras each at 100 fps).

I'm reporting this against linux-lowlatency-hwe-16.04-edge (4.13.0-19-lowlatency) but I have observed this in several kernel versions. I find references to it online dating back to 2014, and a proposed patch available at https://patchwork.kernel.org/patch/9907581/

It seems likely that the underlying issue is the same as in https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1710548

(I got a few timeout from launchpad.net while submitting this. Let me know if the apport files didn't come through)

ProblemType: Bug
DistroRelease: Ubuntu 16.04
Package: linux-image-4.13.0-19-lowlatency 4.13.0-19.22~16.04.1
ProcVersionSignature: Ubuntu 4.13.0-19.22~16.04.1-lowlatency 4.13.13
Uname: Linux 4.13.0-19-lowlatency x86_64
NonfreeKernelModules: nvidia_uvm nvidia_drm nvidia_modeset nvidia
ApportVersion: 2.20.1-0ubuntu2.10
Architecture: amd64
Date: Mon Dec 11 13:38:23 2017
SourcePackage: linux-hwe-edge
UpgradeStatus: No upgrade log present (probably fresh install)

[Jeffrey Miller (mysterylectricity)]

To see the latest regarding this bug please see bug 1773704. Someone tried to fix this bug but introduced a typo. If you can build and install your own kernel, as I've done to address this bug, you'll be all set. If you can install the proposed test kernel (I tried and failed) you may be able to convince salisbury to commit the fix.