Intel i915GM not coming back up from suspend or hibernate

[eccerr0r]

This is an old machine (laptop/lvds panel) using the i915 chipset using the intel_drv driver.

Graphics seems to work fine, but if I suspend or hibernate the machine, and try to resume, the machine comes back up, but with a black screen.

The only indication is that in Xorg.0.log reporting that the crtc had some invalid state and was stopped. The panel's backlight appears to be off.

The key observations:
- I can control-alt-Fx to a different VT and the machine is definitely alive and usable.
- After ctrl-alt-Fx and logging in, I can restart lightdm/X11 and graphics mode comes up again!
Unfortunately this is a very unusable workaround, but at least it shows the machine is mostly functional.

This is using kernel 4.14.65, i686 32-bit. Both suspend and hibernate have the same behavior. Using intel_drv xf86-video-intel-2.99.917_p20180214-r1 I believe. Not sure if I can use the modesetting driver on this machine.

I also vaguely recall this machine successfully suspending and resuming in the past, over 5 years ago...
_________________
Intel Core i7 2700K/Radeon R7 250/24GB DDR3/256GB SSD
What am I supposed watching?

[ShadowCat8]

Greetings,

I have to ask; What happens if you do the "Ctrl+Alt+F[1-6]", and after you are at the console, then hit "Alt+F7"? (I know you have probably already tried that, but you didn't mention it, so I am curious what the results on that were/are since going to the console should restart the backlight.)

And, have you checked the kernel configs for any of the different weird laptop settings that may apply to your laptop?

HTH... Let us know.
_________________
________________________

"As far as the laws of mathematics refer to reality, they are not
certain, and as far as they are certain, they do not refer to reality."

-- Albert Einstein

[eccerr0r]

The screen goes blank once more, backlight turns off (as far as I can tell).

I can switch back and forth from the dead X11 console and the text consoles at will as far as I can tell. I haven't done it continually to see if it will trigger a crash (if it would prove anything or not), but didn't try it as I wanted to save my backlight's cathodes from turning on and off too often to keep them from wearing out as this machine is already very old. However it appears to be more than just the backlight that's off, shining a bright light into the display shows the screen truly blank/off.

It's a good question what kernel configs would affect suspend/resume, after all, I can get back to running X11 if I kill the X server and starting it back up, so it seems there's a bug restoring state -- unless there's something else I've missed...

The question that burns is if anyone still is using a i915 machine and has suspend/resume working. Last I tried it within the past year, my NM10 (i945GM) chipset laptop works fine, but not the i915 machine.

Incidentally and off topic, I feel this broken suspend-resume Celeron-1.5-i915GM/HDD is faster than my Atom-1.6-i945GM/NM10/SSD when not doing disk intensive stuff. But the Atom suspends/resumes just fine...
_________________
Intel Core i7 2700K/Radeon R7 250/24GB DDR3/256GB SSD
What am I supposed watching?

[khayyam]

[eccerr0r]

Then quite possibly there's a regression somewhere as I distinctly recall the in-kernel suspend/resume routines working just fine with an older version of the kernel .

As far as I know, the current kernel suspend/hibernate routines in 4.14.65-gentoo fail at restoring the CRTCs, along with 4.9.34-gentoo IIRC. However there may be something that works with xf86-video-intel-2.99.917_p20180214-r1 which may explain why the KMS fbcon resumes just fine and switching between VT's after a resume doesn't seem to budge the X11 session... so I suppose the version of xf86-video-intel would be interesting to know too. Then again ideally, all mode setting should have been done in kernel along with VT switches...
_________________
Intel Core i7 2700K/Radeon R7 250/24GB DDR3/256GB SSD
What am I supposed watching?

[khayyam]