Disable broken xhci device before suspend and avoid freeze.

system-sleep-xhci.sh

#!/bin/sh
# 
# This script should prevent the following suspend errors
#  which freezes the Dell Inspiron laptop. 
#
# Put it in /usr/lib/systemd/system-sleep/xhci.sh
#
# The PCI 00:14.0 device is the usb xhci controller.
#
#    kernel: [67445.560610] pci_pm_suspend(): hcd_pci_suspend+0x0/0x30 returns -16
#    kernel: [67445.560619] dpm_run_callback(): pci_pm_suspend+0x0/0x150 returns -16
#    kernel: [67445.560624] PM: Device 0000:00:14.0 failed to suspend async: error -16
#    kernel: [67445.886961] PM: Some devices failed to suspend, or early wake event detected

if [ "${1}" == "pre" ]; then
  # Do the thing you want before suspend here, e.g.:
  echo "Disable broken xhci module before suspending at $(date)..." > /tmp/systemd_suspend_test
  grep XHC.*enable /proc/acpi/wakeup && echo XHC > /proc/acpi/wakeup
elif [ "${1}" == "post" ]; then
  # Do the thing you want after resume here, e.g.:
  echo "Enable broken xhci module at wakeup from $(date)" >> /tmp/systemd_suspend_test
  grep XHC.*disable /proc/acpi/wakeup && echo XHC > /proc/acpi/wakeup
fi

thanks for this. had similar problem on Fedora 35 (Gnome) Dell Latitude 5480 laptop

Recently started encountering issues with suspend on my Dell Precision 7710
I tried the above script but to no avail

I found this ancient link https://bugs.launchpad.net/ubuntu/+source/pm-utils/+bug/562484/comments/3 which deals with the issue by unloading the module itself.

I modified the above script as follows and now suspend works fine

  # Do the thing you want before suspend here, e.g.:
  echo "Unload xhci_pci module before suspending at $(date)..." > /tmp/systemd_suspend_test
  modprobe -r xhci_pci
elif [ "${1}" == "post" ]; then
  # Do the thing you want after resume here, e.g.:
  echo "Load xhci_pci module at wakeup from $(date)" >> /tmp/systemd_suspend_test
  modprobe xhci_pci
fi

One thing: it should be bash in shebang not sh.

One thing: it should be bash in shebang not sh.

why would it need to be bash? it's a perfectly valid posix sh script

interesting... with sh I've got:

./xhci: 15: [: pre: unexpected operator
./xhci: 19: [: pre: unexpected operator

It's debian, so sh is dash.

$ ls -la /bin/sh
lrwxrwxrwx 1 root root 4 Jun 21 13:28 /bin/sh -> dash

When I change to bash it works perfectly fine.

ah, i was wrong; the test utility specification doesn't require == to be supported, only = or -eq

for maximum compatibility, the script could be edited to use = instead