Adrian, On Tue, 2026-03-03 at 10:07 -0500, Adrian Vovk wrote: > Lol.
Maybe you to... It wasn't funny to read through it, nor the inaccuracies! > And to correct the record: no AI was used in writing any of my > messages. I do not use LLMs to author things. Here I didn't use them > to research things either. I was just over-explaining to get the > point across, because explaining it succinctly didn't work. And look, > it worked! No it did not, there were two minor things sd_login_monitor and sd_get_seats, among so many excessive words. Some stuff was flat out wrong, you can claim it as you, I say AI! The wording makes it clear. If it was you, as pointed out, several things were really wrong. I think you may need to spend some time with PAM to understand how that works. Plus literally saying GDM code is not multi-seat that has it in the code and commit comments... hello... You never explained anything succinctly, as shown below. > > I assumed that you've at the very least read the man pages for the > library you're using. Or even just looked at what functions are > available in the library you're using. That's step one, before coming > to the mailing list. I am looking at the logind code, and others, and YES I looked at man pam_systemd, which I am using, and other man pages. I quoted a section from pam_systemd a while back... I said early on I read it all for years, like almost 10 and it hasn't changed. Have YOU read the man pages? like on PAM... What library is it you think I am using? > But anyway, I told you about both sd_login_monitor and sd_get_seats > in my second email in this thread. That wasn't enough to help you at > the time, apparently. I must have missed those amongst all the rest. Just the same having mentioned those, it should have been clear the approach/implementation was off. This is the email, you said WAY to mnuch, that part was burried deep, mentioning inotify I may or may not need to bother with, and could have been said much sooner and succinctly. https://marc.info/?l=systemd-devel&m=177226643032211&w=2 Not to mention all the various ways mentioned, when all that needed to be said, succinctly... "For a DM to use logind, it should use sd_login_monitor and sd_get_seats, enough said." >From there I could have taken it, and not get lost in the weeds with useful info. A single sentence, instead of explaining things long understood like what is a seat, etc. > > FYI, seems all seats would be created at boot not just seat0. Guess > AI > > missed that. From there, seats could be added or removed, but at > > the > > time seat0 was created, hardware is likely probed in a loop so > > would > > create all seats. > > > > I didn't miss anything here. What you're saying is just inaccurate. So only seat0 is created at boot and other hardware devices already attached like seat1 are ignored till sometime later. Makes sense. Smart way to do things, why bother being aware of all seats on boot. > For one, your display manager will probably be running before all the > seats are finished probing, because GPUs take a long time to probe > and systemd is doing everything asynchronously, in parallel. seat0 is > special because it always exists, so logind will emit the seat added > event when it starts, before the display manager is running. That depends on the order of the init system and dependencies, that is a pretty incorrect assumption. FYI, I am openrc, please stop with assuptions! Stick to objective, answer specific narrow questions, no need to bring up PAM, etc. Some of us don't try to parallel things that obviously depend on others and trust me, my openrc non-parallel boot way faster than parallel systemd, per all the stuff systemd wants to start etc. I am not trying to get into a systemd debate, this is simply about logind, which elogind can be used just the same without any systemd. When you don't think narrowly. > For two, seats can appear or disappear at any time due to device > hotplug, like someone plugging in a GPU, or a USB dock, or something > like that. Who is hot plugging seats? Who is even multi-seating? Again, have you seen a single computer with multiple monitors, keyboards, and mice and different people or at least different seats? That whole setup makes little sense, especially with computers being so cheap, or maybe it will make more sense as prices skyrocket. > > FYI, pam is not part of the greeter/client UI, PAM is ONLY invoked as > > part of actual login. The pam env is not constructed until after a > > successful login. > > > > So you're not using pam for auth. Cool, then it's even easier for you > because you don't need to juggle long running PAM worker processes. I was working wiht pam before you got into all this buddy... https://github.com/wltjr/entrance/blob/master/src/daemon/entrance_pam.c I keep trying to tell you that > Anyway I'm glad it has been solved and this thread can end. Already stated, closed, and thanked, but had to safe face and try to claim no AI usage. Even AI would sayd that write up is AI, someone will try... Just like me, what we dont know in tech shows to those who do!!! Much I do not know that I am search of learning and knowing... -- Sincerely, William L. Thomson Jr.
