BUG hunting Enlightenment (E24+)

Enlightenment E23 not visibly resuming from suspend

If from the log-out options "suspend" is chosen, the screen blanks (goes black) and the pointer can be seen and moved but the Desktop does not come back in any way. Requiring a reboot from one of the TTYs.
Choosing "hibernation" works as is expected, rebooting into the Desktop as it was before.

yeah it looks like, so I added this post in a comment of that bug :slight_smile:

we should know which actions are took in E23 when doing this, you can also try pm-suspend and similar commands just to verify that it correctly works in your hardware, and then we may know if E is doing something wrong with them

Yes, I'll look into that.
I hoped setting "intel_iommu=off" as kernel parameter would repair it but alas no. Only the errors at boot have gone.
It is hard to find what exact command E23 uses to suspend from the menu. It's clearly different to that used in E16.

I'm wondering if E23s expected usage of "acpid" has anything to do with it. On login I always get this pop-up:
Elive_Screenshot_2019-11-06_09%3A02%3A12__607x348

Enabling acpid gives some different results but all in all, hard to pinpoint what exactly is going wrong. Especially because the "power management" and "screen settings" in the E23 settings panel isn't very comprehensible. :face_with_head_bandage:

Question

Is there a reason for acpid.service to be disabled on Elive?

"sudo pm-suspend" turns the screen black, disables my Fn buttons and wakes using Power button or opening laptop lid.

Tried "xset dpms force suspend" and that works exactly as it should.

Doing more tests I found that the problem is NOT a wake-up problem.

  • It's the backlight not turning on.
    In principle the laptop is awake and fully functional. :thinking:

Edit

Taking a screenshot with PrtScrn renders a black screen too ...... So it's more than just the LCD backlight being off.

try also to switch to Software-based rendering instead of accelerated one (should be more stable)

yeah this popup basically says that, acpi is not running (lol), if you run acpi daemon it will not show anymore (and thus, sometimes is needed to have acpi, for specific hadware computers)

nope, it should be a default setting set by systemd / debian-buster, maybe it loads it in a smart way ? (when required, so by some harware specific computes that may require it to be running?)

that sounds like E doesn't knows that needs to restore the desktop visuals (if it does), maybe from the dim settings... try to check all those power-management / screen / suspend settings from E to see if any of them does a strange thing on suspend (like dimming-out the screen, unable to dim-in later)

The one thing that does work when waking up from hibernation (which only shows the pointer on a black screen) is to go to tty01 and then back to tty07 .... practically restarting the Xserver.
Restarting from hibernation rattles the kernel too, with "confused and dazed" reports.

Alas suspend is still is dead in the water, really nothing works. :face_with_head_bandage:

That normally happens when there's a bug in the drivers that doens't quite wake up, switching to VT will probably switch the graphical driver too (kernel drm video VS xorg video driver), or it may be also due to the desktop not restoring correctly (but on such case it should affect everybody else)

I assume it only happens on that specific computer and not other ones ? (didn't tried it)

Nope, the 3 machines (that have E23 installed) I have access to have the same suspend problem on E23.

As they are all fairly modern laptops with touchscreens (2 with detachable keyboards) I suspect that the issue has to do with USB being disconnected.
Some of the internal hardware emulates USB and uses a USB driver.

It's easy to try:
Just click "system", "suspend" and see if it will come alive again. I only get to see a pointer on a black screen.

I need to try it in a real machine (a testings one) :slight_smile:

BTW I forgot to say, in PHAB is better to assign the reported bugs to someone (active), like for example raster, otherwise its probably that the report will stay for long time without anybody seeing it

1 Like

I must say I'm pleased to see that so far my reports have seen reactions. So far so good.

Admittedly I have not submitted this one yet because it is so frigging hard to replicate or even to be sure when it happens and exactly what setting is the cause. :scream:

2 Likes

Hi!

On the eeepc 1001px, 3.7.x updated (not formely 3.8.2 at now), I got the same acpid warning at start, even after re-install it via synaptics...
When wanted to change launch icons in the l-bar, it (the window for changing contents) freezes all the system except mouse movements (need to change session and log in & reboot)...

Salutations.

See if you get some reports on what's going on with "tail -f /var/log/syslog" in a terminal ..... or switch to tty12 with Ctrl, Alt, F12 and see it there.

About ACPID is made on purpose, seems like E warns that is not running but it is installed (I assume that systemD decides if run it or not if is not needed), this message was made in order to help users to know that there's a cause of some things may be NOT working as expected, you can manually run or check it as:

sudo systemctl status acpid
sudo systemctl start acpid
sudo systemctl status acpid

But before we decide what to do with this message, we need first to know what to do with the acpid daemon

  • needs to be run in every computer?
  • in some ?
  • not needed at all?

this needs some research first

@triantares talking back about suspension / hibernate, it works well on my hardware on E16 (not tried E23), but please try too from the commands that are defaults on previous elive, maybe they are more compatible / working!

sudo hebernate --suspend
sudo hebernate --hibernate

1 Like

Hibernate is sometimes a problem because it in fact turns off the machine and boots back in the the remembered state which isn't always optimal. :face_with_head_bandage: A simple reboot is often faster.

Of late suspend works fine on E23 and E16.
Albeit that on e16 my touchpad sometimes doesn't come back alive, requiring me to unload and reload the "psmouse" module. :slightly_frowning_face: The touch-screen does not have this problem.

Using E23 the backlight sometimes does not want to come back on when for instance; I leave my laptop long enough for the screen to blank and then close the lid.
When opening "laptop-mode tools" signals "lid_open" but it doesn't do anything, leaving a reboot as the easiest option. Apparently a double suspend action confuses the scripts.

1 Like

well, that's the good thing about it :slight_smile:
it takes a little longer to shutdown, but it recovers your desktop as the same state as it was before to turn it off (because is a real turn-off, unlike suspension, which consumes battery or requires AC power for workstations)

on which scenarios this is a good thing? well, for example I recently used Picard (music tagger) with 18k songs, this takes some hours to process, you don't want to close the application to continue another day restarting it again from zero. Another scenario is when you have your desktop full of applications and terminals opened in specific places to work on different things, to make the desktop again in the same order and apps opened will require time. Another scenario is when you have some specific work opened and you cannot "remember tomorrow what you where doing", so just "snapshot" your computer on that state is the best

Mmmh, E16 is more plain (doesn't do anything by itself about suspension), but E23 probably does, so as you said maybe is a double suspend action, try to unload any module in e23 that can be related to that (you must try from a different user created maybe, for those E tests), or you can do a fast test by unloading ALL the modules directly :slight_smile:

2 Likes

True, but until now my daughter at Uni has developed a deep mistrust to hibernate/suspend due to the E16 keyboard/mouse issue (requiring a restart) and the black screen issue.
She understandably doesn't want to lose work moving from lecture to lecture....but maybe that'll change now.
She's moved over to E23 definitely now because it handles suspend better.

There are other positive things in the lock-screen/suspend/lid-close combos on E23, so actually not too happy to turn them all off. :face_with_head_bandage:

Coming wednesday I'll have my hands on another X1 carbon to do some extra tests w.o risking my work-horse. :happybounce:

We just suspend the laptop, after all the power consumption on modern devices is so damn slow.
I remember using stable elive that whenever i closed the screen it hibernated. That was quite useful since i barely use my battery on the laptop (after ten years it lasts 1.30h only. Even if it is quite a lot compared to the age is still too little ahahah)

reading the code, seems like E23 uses systemd for suspend (and we use mostly pm-suspend for e16), plus some extra E features (screen controls, acpi maybe, etc)

I did enable "acpid" to be rid of that warning screen but from what I can see it doesn't get called or used in any way.

Is this thread also handling bugs for E23 as well? Otherwise looks like there will be a duplication or division of attention to and labour with this other thread E23 installation / Testing . No ?