13:31 <Inga> Hi folks! I have this weird issue with resuming from lid close, and while I think that this is probably not a river issue, I have no idea what can cause it (wlroots? elogind?), so maybe you have any ideas?

13:31 <Inga> I have a pretty simple Alpine setup with elogind, greetd (+agreety) and river (and I disabled suspend-on-lid-close in both elogind and acpi configs, because it was causing audio problems for me).

13:31 <Inga> So the laptop does not go to suspend every time I close the lid, but when I open the lid, I get a black screen. And the only way I found to recover from it is to switch to another terminal (with agetty login prompt, but I can't see it because the screen is black regardless of the terminal), move the mouse around, then switch back to the terminal

13:31 <Inga> where river is running, move the mouse around, and a couple of seconds the screen will slowly fade in from black (and all the other terminals become visible too). Simply moving the mouse around in the terminal where river is running is not enough, I have to switch to another terminal and then back to river.

13:31 <Inga> The weird thing is that when I first log into river, I also get that fade-in effect (as in: I enter my username and password into terminal-based agreety prompt, press enter, the text on the screen disappears (or the screen turns black, which would look the same), and over the next ~half-second, river home screen fades in). And I'm not sure if it's

13:31 <Inga> supposed to do this, but I _think_ that it used to be instant on my previous PC, without any pretty fade-in effects?

13:33 <leon-p> we definitely don't have any fade effects

13:33 <Inga> Oh, and I don't have any lockscreens or anything installed

13:33 <leon-p> (although they would be pretty)

13:33 <leon-p> I wonder if closing the laptop lid disables the output somehow...

13:33 <leon-p> a river log would be helpful probably

13:34 <ifreund> I'd be curious if it happens with seatd rather than elogind

13:35 <ifreund> also, someone just opened an issue that sounds similar, also using elogind

13:36 <Inga> For logs, do I just run `river -log-level debug &>/tmp/river.log`? Sorry, I'm pretty new to all this Linux stuff :)

13:36 <ifreund> I wonder if there was an elogind update or something that changed things

13:37 <ifreund> yeah, anything that collects stderr should be fine

13:38 <Inga> Actually I've had this happening to me for over a week, what a coincidence that someone just opened a similar issue just as I decided to ask this in the chat :)

13:38 <Inga> (I even checked the issue tracker yesterday...)

13:40 <Inga> Just checked and I don't think it was a (recent) elogind update specifically, because Alpine elogind package was last built in August 2023

13:40 <Inga> will check logs and seatd now

13:40 <leon-p> the fade-in is suspicious IMO. I would be really surprised if that's caused by elogind. Fade-ins the big desktops use are done differently

13:41 <Inga> oh and one more thing, even without lid close, I'm not only getting fade-in on startup; when switching to another tty and then back to river, I also get that half-second fade-in before seeing river

13:43 <leon-p> maybe that's a new kernel KMS setting or something? haven't kept up with that in a while

13:48 <tiosgz> well fwiw i've had a short-lived black screen when switching tty for quite a while; never much of an issue

13:49 <tiosgz> (re-checked just now, it happens only when going back to river. and it doesn't happen on river startup)

13:49 <tiosgz> so it's possible that this is unrelated /shrug

13:53 <Inga> leon-p, switching between text ttys is instantaneous, the fade-in only happens when switching into the tty where river is running

13:54 <ifreund> the tty switching code is all in wlroots, libseat, and either seatd or elogind

13:54 <Inga> just checked, and I'm getting the same problem with lid open on seatd as well. to make things clear, the way I started river with seatd was: log in on text terminal using agetty, start seatd service (without stopping logind service), set XDG_RUNTIME_DIR manually, launch river the same way agreety usually does it.

13:54 <Inga> I also got a lot of logs, and some of them might be relevant. where do I post them?

13:56 <ifreund> any kind of pastebin works, I often use 0x0.st

13:59 <Inga> river logs for getting into river from greetd+agreety (with elogind): https://0x0.st/s/dXMCb5L14c8ygiDm2Vbipw/Hn_M.elogind

13:59 <Inga> last entry before the lid close there is at 2024-02-17 14:43:42.599

14:00 <Inga> river logs for getting into river from agetty (with seatd service started, with XDG_RUNTIME_DIR set manually): https://0x0.st/s/i2lr1cfgd7r0v025y1Kyag/Hn_1.seatd

14:00 <Inga> last entry before the lid close there is at 2024-02-17 14:49:55.081

14:14 <Inga> also I just checked, and it's only the screen that's blank, everything else is definitely still working:

14:14 <Inga> * launch gedit;

14:14 <Inga> * close the lid;

14:14 <Inga> * open the lid;

14:14 <Inga> * type some text while the screen is blank;

14:14 <Inga> * do the manipulations needed to get the screen not blank (via switching to another tty, moving the mouse around, switching back, moving the mouse around);

14:14 <Inga> * observe that the text typed while the screen was blank was actually processed by gedit

14:19 <LarstiQ> does it happen if you skip seatd/elogind/agetty, just straight river launching after logging into console?

14:21 <Inga> LarstiQ, river does not start for me when it cannot create a seat. just to make things clear, agetty is not related to seatd or logind, agetty is the default out-of-the-box login prompt on minimal Alpine, I won't be able to log into console without it :)

14:21 <LarstiQ> check

14:23 <Inga> here is what I'm getting when I try to launch river with seatd service stopped, after logging into console from minimal login prompt (without involving elogind): https://0x0.st/s/wZ5qBIRDfhCyUivb-xPN0Q/Hn_D.noseat

14:24 <novakane> yeah you can't launch river without elogind or seatd, that's normal

14:25 <Inga> yes I thought so, that's why I was surprised by that question :)

14:26 <LarstiQ> I'm just flailing around trying to rule things out, if more knowledgeable people are present, follow their lead :)

14:34 <Inga> one more thing I've just realized: since this is an OLED screen, it could be that what I see as black screen or fade-in is actually brightness set to zero or gradually increasing to the level that was set before the lid close / before switching to text terminal.

14:34 <Inga> but once again, no idea what would be responsible for that.

14:37 <novakane> I don't really see anything obvious in the logs, but then I'm on mobile right now so it's pretty shitty to look at logs

14:39 <ifreund> Inga: to be clear, this is a regression that happened in the past few weeks right? What version of river are you using?

14:40 <novakane> is it a true black screen, or something dark blue, because depending on how powerful the laptop is sometimes you see the default river background colour

14:40 <LarstiQ> how quick is the fadein?

14:41 <justine> :close

14:43 <Inga> ifreund not sure if this is a regression. I had another laptop before, used river there for 1.5 years, didn't have any problems. Now I got a new laptop (and no longer have the previous one, because I changed jobs and both are work-provided), and have this problem on it from day one. The system configurations were pretty similar on the old one and

14:43 <Inga> the new one, but then IIRC the last time I updated the old one was in December. so it can be something that's different between system configurations, could be hardware difference, or could be something that changed in the last couple of months. my river version is 0.2.6.

14:44 <ifreund> Inga: would you be able to test river's master branch? It uses wlroots 0.17.1 rather than 0.16 which could make a difference

14:45 <Inga> novakane looks like true black screen to me, and also it stays black when I switch to another tty (even though I should see a standard login prompt there, in white text on black background).

14:45 <Inga> but one thing with this screen that I've just realized is that when I set backlight to 0, I get absolutely black screen, as opposed to all the previous laptops that I've had where the screen only became very dim with backlight set to 0.

14:45 <ifreund> actually, testing with the tinywl example compositor in the wlroots repo would be interesting too

14:45 <Inga> LarstiQ, around half-second, plus a couple of seconds before fade-in starts happening (after lid open)

14:45 <ifreund> (it has hardcoded Alt+ESC to exit)

14:49 <Inga> ifreund, I will try to build river from main at some point later. I'm not sure where to get tinywl binaries for my system though, or how to build it.

15:01 <tiosgz> Inga: hmm, if there's actual fade-in, is it possible the laptop does it? (ig there could be a bios setting or something like that)

15:02 <ifreund> Inga: I don't think anyone distributes tinywl binaries, to build it one would clone the wlroots git repository and run make in the tinywl directory

15:02 <ifreund> https://gitlab.freedesktop.org/wlroots/wlroots/-/tree/master/tinywl?ref_type=heads

15:03 <ifreund> thanks for your help debugging

15:05 <Inga> tiosgz, I'd suspect that it was a hardware issue, if not for the fact that I'm getting the same fade-in when I launch river, or when I switch from working, visible river to another tty and then back

15:06 <ifreund> I think it's most likely an issue lower than river in the software stack, at a guess either wlroots or the linux kernel, but it's hard to say for sure

15:07 <ifreund> if it's reproducible with tinywl as well we can rule out river and move this to the wlroots issue tracker

15:10 <Inga> just installed sway and tried to launch it (with seatd), and while I'm not getting that fade-in effect on startup with sway, I'm still getting black screen on lid close. so yes, the issue is definitely not in river

15:33 <Inga> Got a suspicion, rebooted without logging in, the lid close issue is still there (except I couldn't get the screen back on).

15:33 <Inga> Disabled greetd and elogind services, rebooted without logging in, the lid close issue is still there. What the f...

15:33 <Inga> I guess this is a kernel driver bug then, and I should be grateful that with river I can at least recover from it.

15:33 <Inga> Thank you all for the support anyway! :)

16:19 <ifreund> no problem! hopefully it gets fixed upsteam soon