* Mark Kettenis <[email protected]> [2017-07-06 15:31:34 +0200]:
Date: Thu, 6 Jul 2017 00:01:41 -0700
From: Pratik Vyas <[email protected]>
>Synopsis: suspend / resume on skylake machine doesn't work
>Category: system (inteldrm)
>Environment:
System : OpenBSD 6.1
Details : OpenBSD 6.1-current (GENERIC.MP) #87: Wed Jul 5 09:31:24
MDT 2017
[email protected]:/usr/src/sys/arch/amd64/compile/GENERIC.MP
Architecture: OpenBSD.amd64
Machine : amd64
>Description:
Tried to suspend a skylake machine (Intel NUC6i5SYB) from X using zzz.
Suspend worked as screen went blank and the power LED went from solid
blue to blinking orange. Pressing enter on the keyboard didn't resume
the machine. Pressed power button and video came back on but no X (only
the console). Keyboard typed wrong letters. Tried unplugging and
replugging my usb hub with keyboard but still same result. Shutdown by
pressing the power button worked.
In your case it is probably related to the usb keyboard issue. X will
try to re-open your usb keyboard/mouse but if those devices have not
been attached yet or aren't fully functional, the VT switch could
fail. Does suspend/resume work correctly if you switch to the console
before suspending?
Hi Mark,
Thank you for your reply. I also upgraded to the latest snapshot today
and tried several times.
(OpenBSD 6.1-current (GENERIC.MP) #91: Thu Jul 6 08:53:20 MDT 2017).
Sounds like the machine resumed just fine, but the final VT switch
back into X didn't happen. This happens if X doesn't acknowledge the
VT switch in a timely fashion. The whole procedure is a bit racy and
I've seen it happen on many machines. Usually you can just switch
back into X using the keyboard.
Yes, in one of the many tries, I had the keyboard function correctly and
was able to switch the VT console and resume work in X just fine.
Once time, with X on, I had a panic. Was dropped to ddb but couldn't
type anything. I think this might not be realted to X though.
kernel protection fault trap, code=0
Stopped at acpitz_setfan+0xc5 movq 0(%rax),%rsi
ddb{0}>
If the machine has a BIOS knob to select USB 2.0 (EHCI) instead of USB
3.0, it'd be interesting see if that makes a difference.
Without X on, I had success in resume (using the power button)
everytime. Couldn't find a knob in the BIOS to switch to USB 2.0.
So this doesn't look like an issue with the inteldrm(4) update.
I guess, that's a fair conclusion :)
--
Pratik
