Comment 5 for bug 1603659

No, /etc/rc.local was untouched. The file was empty besides of an "exit 0".

/etc$ sudo fgrep -ril powertop *
/etc$ sudo fgrep -ril autotune *
both give an empty result.

I installed TLP again, which automatially removed package
"laptop-mode-tools".

Using powertop, I've set back the two USB devices to "Bad" again, just to see what TLP
does to it.

Before I started TLP, the output of "tlp-stat -u" was:

Bus 002 Device 004 ID 046a:0023 control = auto, autosuspend_delay_ms =
2000 -- Cherry GmbH CyMotion Master Linux Keyboard G230 (usbhid)
Bus 002 Device 003 ID 046d:c24c control = auto, autosuspend_delay_ms =
2000 -- Logitech, Inc. G400s Optical Mouse (usbhid)

Note that 2000 ms of autosuspend_delay is pretty much the time I
experienced for the devices to become unresponsive.

Then I started TLP with a

sudo tlp start

Now that TLP was started, the output of "tlp-stat -u" was:

TLP started in AC mode.
Bus 002 Device 004 ID 046a:0023 control = on, autosuspend_delay_ms =
2000 -- Cherry GmbH CyMotion Master Linux Keyboard G230 (usbhid)
Bus 002 Device 003 ID 046d:c24c control = on, autosuspend_delay_ms =
2000 -- Logitech, Inc. G400s Optical Mouse (usbhid)

Basically that means that TLP disables autosuspend in AC mode.
Both devices are working great, like the way they did with the
workaround using powertop after every reboot. With TLP enabled, everything still works after a reboot. I also can type in my login password without loss of the first characters.

Same for battery mode:
sudo tlp start
TLP started in battery mode.
tlp-stat -u
Bus 002 Device 004 ID 046a:0023 control = on, autosuspend_delay_ms = 2000 -- Cherry GmbH CyMotion Master Linux Keyboard G230 (usbhid)
Bus 002 Device 003 ID 046d:c24c control = on, autosuspend_delay_ms = 2000 -- Logitech, Inc. G400s Optical Mouse (usbhid)

Now that my computer had autosuspend enabled after a reboot (before I reinstalled TLP), which
package was responsible for that? laptop-mode-tools?

If laptop-mode-tools enabled autosuspend for input devices, even in AC mode, that should be avoided with the behaviour I experienced. That said, I think this issue is not related to the linux kernel at all, but to one of the power saving tools.