Yesterday, I observed a (presumably new) sporadic failure of the trackpoint of my x260 laptop when using Sculpt 22.10:
After removing the laptop from the docking station, I was not able to used the trackpoint on the undocked laptop. However, the keyboard did work. I forcibly restarted the ps2 driver by setting a version attribute on the driver's <start> node in /config/drivers. After the restart, the trackpoint worked.
Two of three boots in the evening showed the same behavior directly at boot time. The forced restart of the ps2 driver solved the problem.
I could not spot any ps2-related error in /report/log. However, I noticed that the ps2 driver prints it diagnostic output very early, while the acpi driver is still discovering the hardware. So there could be a potential race between the acpi/pci discovery and ps2 with respect to shared interrupts. When inspecting /report/drivers/devices, no irq is annotated as shared though. The IRQ number 12 notably appears at both ps2 and the PCI device 04:00.0 (wifi). But the former is a legacy interrupt whereas the latter is an MSI. So this should not be a problem.
To investigate whether the order of acpi discovery and ps2 startup is the culprit, @skalk suggested to add an artificial delay in the acpi driver to magnify the effect.
Yesterday, I observed a (presumably new) sporadic failure of the trackpoint of my x260 laptop when using Sculpt 22.10:
version
attribute on the driver's<start>
node in/config/drivers
. After the restart, the trackpoint worked.I could not spot any ps2-related error in
/report/log
. However, I noticed that the ps2 driver prints it diagnostic output very early, while the acpi driver is still discovering the hardware. So there could be a potential race between the acpi/pci discovery and ps2 with respect to shared interrupts. When inspecting/report/drivers/devices
, noirq
is annotated asshared
though. The IRQ number 12 notably appears at both ps2 and the PCI device 04:00.0 (wifi). But the former is a legacy interrupt whereas the latter is an MSI. So this should not be a problem.To investigate whether the order of acpi discovery and ps2 startup is the culprit, @skalk suggested to add an artificial delay in the acpi driver to magnify the effect.