(trying a different email for Frederic as I got a delivery failure)
Hi Loic,
Using out-of-band gpio/irq which are retrieved from the ACPI device.
wakeup can be initiated by host or controller depending which one wants
to transfer hci data.
Thanks for this explanation.
I'm going to retry on my T100 and let you know, but used to work.
Did you have chance to retry on a T100. Is it a T100TA model by the way, or
a T100 Chi (to make the distinction between the 2 bugzilla kernel entries)?
- What is the state of your interrupt counter ("host_wake") in
/proc/interrupts when you try to generate host wake-up events (using
mouse...).
Moving the mouse or clicking on it doesn't change the host_wake interrupt
counter in /proc/interrupts after the autosleep period. If I manually
trigger a new scan, then the counter starts increasing, but it freezes again
after the next autosleep.
- Could you confirm you don't meet the issue if you disable runtime-pm
via sysfs. You should find the device (BCM43*) as a child node of your
tty in /sys/class/tty*. (echo on > power/control).
I can confirm that with the following command, the mouse keeps working and
doesn't suddenly stop after a sleep period.
root@thinkpad8:/sys/class/tty/ttyS1/device/BCM2E55:00# echo on >
power/control
I remember Frederic had some gpio/irq polarity issues:> 5cebdfea32b89911d4540440c1c2854a1a3d591e
Bluetooth: hci_bcm: Fix IRQ polarity for T100
Maybe the ACPI table is not correct on some devices.
I had seen that patch indeed. I'm attaching the ACPI table I had shared on
the list a long time ago for the ThinkPad Tablet 8. Is there anything that
looks suspicious in there and that you could link to a similar gpio/irq
polarity issue maybe?
If you think there is a good chance that this could be the issue as well, I
will try to build a similar patch for this ThinkPad product and share my
results.
Regards,
Jérôme