<Wizzup> so we can also mirror to github from gitea :)
<Wizzup> btw, I also added some experimental 'log in with github' thing, but I might change how some of it works or what can be done with this
mkfx has joined #maemo-leste
ashley has joined #maemo-leste
ashley is now known as Guest1477
mkfx has left #maemo-leste [#maemo-leste]
Guest1477 has quit [Read error: Connection reset by peer]
ashley has joined #maemo-leste
ashley is now known as Guest6946
norly has quit [Quit: Leaving.]
norly has joined #maemo-leste
Guest6946 has quit [Quit: Whatever.]
ashley has joined #maemo-leste
ashley is now known as Guest7986
System_Error has quit [Remote host closed the connection]
System_Error has joined #maemo-leste
joerg has quit [Ping timeout: 252 seconds]
joerg has joined #maemo-leste
donihalim has quit [Changing host]
donihalim has joined #maemo-leste
pabs3 has quit [Read error: Connection reset by peer]
pabs3 has joined #maemo-leste
mkfx has joined #maemo-leste
<freemangordon> sicelo: does that mean I don;t have to do anything with upower?
<freemangordon> arno11: yes, will have a look
<freemangordon> but first lemme see what I can do with priorities
Chemical-X has joined #maemo-leste
arno11 has joined #maemo-leste
<arno11> ok cool
<mkf> lima works \o/
<freemangordon> :)
ungeskriptet has quit [Remote host closed the connection]
ungeskriptet has joined #maemo-leste
<mkf> okay i'll go for mainline now
<mkf> do mainline needs any patch other than realtek one?
<freemangordon> no, but see what I pasted
<freemangordon> like, if you want BT...
<freemangordon> etc
<mkf> applied those. do maemo have an ui for bluetooth btw?
<freemangordon> no (still)
<mkf> i've used blueman but not sure if that worked
<mkf> alright. :)
<mkf> can we bring sunxi builds back?
<freemangordon> sure
<freemangordon> we juts need a proper kernel tree
<freemangordon> and one that volunteers to maintain it :p
<mkf> well i'm willing to, at least for devices i have (a a23 and a a33)
<mkf> *a13 and a23
<freemangordon> cool, I can test on A33
<freemangordon> Wizzup: ^^^%
<mkf> some of elements of this image i'm using apperently are for n900
<freemangordon> well, it should not be that different
<freemangordon> but still, better have a dedicated image
<mkf> like it tells me i need to insert an sim card. i like to if i could. :)
<freemangordon> umm... when?
<mkf> in the tray?
<mkf> near battery
<freemangordon> well, that won;t change
<mkf> it also loads bunch of n900 stuff during boot
<mkf> which fail
<freemangordon> we install telephony stack no mater the device
<freemangordon> no, it tells you there is no sim card :p
<freemangordon> up to you if you want to inster one :D
<freemangordon> *insert
* freemangordon needs another coffee
<mkf> i suppose i can think of no sim icon as leste logo for this tablet :P
<mkf> is the telephony stack being installed on devices with no telephony is a policy or technical?
<freemangordon> I would say policy
<freemangordon> nothing stops you from attaching usb modem
<freemangordon> and it will simply work
<freemangordon> so yes, you *can* insert a SIM card
<mkf> hm, ok.
<mkf> can we at least put an option to hide it later on?
<freemangordon> what would be the trigger to hide?
<mkf> some option down into settings
<freemangordon> like, I am not against, but it should be consistent
<freemangordon> as it can;t be separate, that should be part of phone settings I presume
<freemangordon> and that's not trivial
<freemangordon> but yeah, why not
<mkf> fair enough
<freemangordon> I'll let you know when I am done with connui-cell (when phone setting reside), feel free to make a patch
<freemangordon> and PR
<freemangordon> s/when/where
<mkf> thanks.
arno11 has quit [Quit: leaving]
akossh has joined #maemo-leste
<sicelo> mkf: the device doesn't have a modem at all?
<freemangordon> no
<freemangordon> this is allwinner tablet
<sicelo> we can automatically hide the icon if ofono did not find a modem. we do same for no battery, ensuring.g
<sicelo> e.g. vm
<freemangordon> no, that's not correct
<freemangordon> remember d4 modem dropping off usb
ikmaak has quit [Read error: Connection reset by peer]
ikmaak has joined #maemo-leste
<sicelo> I think that's exactly where this would be useful. currently, we still have misleading icon after modem drops. with this change, icon would disappear, so you know something is amiss
<freemangordon> hmm
<freemangordon> ok, will think about it when I am back on connui-cell
Twig has joined #maemo-leste
<sicelo> re: upower, I'll continue working on it. I'm mostly unavailable this week, so if timeframe becomes a concern, I would share my work so far so you can continue with it
<freemangordon> ok
<freemangordon> not that I have spare time, but yeah
<freemangordon> thanks
freemangordon has quit [Quit: Leaving.]
freemangordon has joined #maemo-leste
<freemangordon> Wizzup: so, we either do https://github.com/maemo-leste/maemo-system-services/blob/master/debian/maemo-system-services.xorg.init#L17 or I implement similar script that does inotifywait on proc and sets nicenes based on some rules
<freemangordon> I prefer the latter
<freemangordon> because I don;t really want to add sudoers file to each user daemon that needs lower nice level
<freemangordon> neither I like the idea to allow user to change niceness < 0
<freemangordon> please comment why are you against having another daemon (besides obvious - one more thing to maintain)
freemangordon has quit [Quit: Leaving.]
freemangordon has joined #maemo-leste
<freemangordon> Wizzup: see what AI generated for that purpose (didn't test, but final script should be similar if not the same) https://paste.debian.net/1364730/
<freemangordon> I can write that in C in an hour if having resident script is a concern
<mkf> is there a better on screen keyboard?
<mkf> default one is really not suitable for usage on the terminal
<freemangordon> which terminal? osso-xterm?
<freemangordon> what's wrong with it?
<mkf> you can type stuff alright, but you can't use direction keys, combos (ctrl-d) or delete stuff that are already written
<mkf> yes, osso-xterm
<freemangordon> also, if you have lots of console stuff, you'd better connect through ssh
<freemangordon> hmm, you can typ ctrl-d
<freemangordon> press ctrl button on bottom
<mkf> you can use combos, but i assume the way osso-xterm works is menant for n900 and not really for touch only devices
<freemangordon> right
<freemangordon> but you'd better connect usb kbd
<mkf> i do that, but i wonder if there is a touch way :)
<freemangordon> not atm
<freemangordon> inky has some magic kbd, but we don't have maintainer for it
<freemangordon> so it is not in the repos
<mkf> can i install it?
<mkf> ah.
<freemangordon> no idea
<mkf> thank you.
<Wizzup> morn
<freemangordon> Wizzup: AI did it even faster :) https://paste.debian.net/1364733/
Guest7986 is now known as ashley
<Wizzup> mkf: would like a sunxi maintainer very much, I have a bunch of olinuxino lime2 (a2) devices and olinuxino a64
ashley has quit [Changing host]
ashley has joined #maemo-leste
<mkf> Wizzup: i have a13, a23 and a h618/h616
<freemangordon> PP is same, more or less, no?
<freemangordon> so you can adopt it as well :)
<mkf> i suppose, but i can't find one to test it. :)
<Wizzup> freemangordon: not too happy with 'while true' though, but ok @ renice
<Wizzup> freemangordon: yes, one more thing to maintain is what I don't like about daemon
<Wizzup> and things that have to monitor the system state, and change it every now and then
<Wizzup> I really prefer things to be explicit
<freemangordon> Wizzup: are you talking about bash script or C code
<freemangordon> see C code
<freemangordon> what do you mean "explicit"
<Wizzup> in the sense that it's clear from the X script that it runs at a certain nice level
<Wizzup> not having to grep the whole code base to find the one daemon that changes the value later
<freemangordon> I understand
<Wizzup> and this daemon will add wakeups for every process we ever start
<Wizzup> surely that's not the way to go
<freemangordon> well, I don;t think this is really an issue (wakeups) as this is what happens in fremantle
<Wizzup> I suspect netlink can let you monitor proc events
<Wizzup> this is just ai generated garbage imo
<freemangordon> why it is better than inotify?
<Wizzup> because netlink was designed for this
<Wizzup> are you sure that inotify works on proc?
<freemangordon> it works on sys, why not on proc?
<freemangordon> lemme try it
<Wizzup> inotif doesn't work on everything
<freemangordon> I know
<freemangordon> but ok, disregard the implementation
<freemangordon> I am talking about th eprinciple
<freemangordon> keep in mind we have to renice things lige gst-renededr, system-ui etc
<freemangordon> *like
<freemangordon> also, dbus-daemon runs with nice -5 on fremantle
<freemangordon> how shale we do that on leste?
<freemangordon> argh
<freemangordon> *shall
<Wizzup> find where dbus-daemon is started and add 'nice -5' in front of it would be my preferences
<Wizzup> or set it in the init script that starts it
<freemangordon> init script comes with dbus
<freemangordon> so we shall fork it
<Wizzup> in openrc you can set init script variables using /etc/conf.d/
<Wizzup> don't we start dbus from our init script btw?
<freemangordon> no
<freemangordon> system bus is started from its own script
<freemangordon> and user is started by xorg or something
<freemangordon> IIRC
<Wizzup> what is /etc/sysconfig for btw
<Wizzup> hm, looks like /etc/conf.d is better
<freemangordon> no idea
<freemangordon> what is it for?
<freemangordon> also, how do we assign niceness on processes started by xsession?
<freemangordon> with sudoer?
<Wizzup> let me search for a bit, I am sure there are other ways to do this than reinventing the wheel
<freemangordon> ok
<freemangordon> Wizzup: also, while searching, think about processes started by maemo-launcher
<Wizzup> such as?
<freemangordon> /usr/bin/hildon-home
<freemangordon> @exe in /proc points to /usr/bin/maemo-launcher
<Wizzup> posts I find online are suggesting that cgroups are -the- way to do this nowadays, though
<Wizzup> do we know we still need the nice changes?
<freemangordon> yes, we do
<freemangordon> changing cgroups does not affect niceness
<Wizzup> but the scheduler does
<freemangordon> sorry, can;t parse
<freemangordon> they chagne scheduler?
<Wizzup> cpu (since Linux 2.6.24; CONFIG_CGROUP_SCHED)
<Wizzup> Cgroups can be guaranteed a minimum number of "CPU shares"
<Wizzup> when a system is busy. This does not limit a cgroup's CPU
<Wizzup> see Documentation/scheduler/sched-design-CFS.rst (or
<Wizzup> usage if the CPUs are not busy. For further information,
<freemangordon> is that what you mean
<Wizzup> Documentation/scheduler/sched-design-CFS.txt in Linux 5.2
<Wizzup> and earlier).
<freemangordon> yes, but this is *within* process group
<freemangordon> not system wide, IIUC
<Wizzup> let me check
<freemangordon> also, I don;t want a minimum number of cpu shares, I want *all* the shares when systemui or mce (for example) needs them
<freemangordon> we will use cgroups, that's for sure
<freemangordon> but it is orthogonal to niceness
<Wizzup> looks like you can make a realtime cgroup
<freemangordon> only one?
<Wizzup> This uses the cgroup virtual file system and “<cgroup>/cpu.rt_runtime_us” to control the CPU time reserved for each control group.
<Wizzup> but, this will -reserve- cpu and not allow any other process to use it
<Wizzup> maybe not ideal
<freemangordon> maybe? :)
<Wizzup> in any case, from what I can see/tell...
<freemangordon> again, how do we deal with maemo-launcher pids?
<Wizzup> ananicy-cpp does what you want
<freemangordon> yeah, and it pulls averything and the kitches sink, iiuc. and it is not in debian
<Wizzup> and it uses netlink
<freemangordon> ok
<Wizzup> and bpf
<mkf> yay kitchen sink
<Wizzup> I mean if it uses 40MB of ram then yeah let's not use it
<freemangordon> but if we are to use code that's not in debian, I would prefer this to be our code
<Wizzup> but this seems to be what people using when searching for 'nice daemon'
<Wizzup> it literally does everything you want btw :)
<Wizzup> cgroups, oom score, niceness
<freemangordon> this does not affect niceness
<freemangordon> sorry
<freemangordon> wrong paste
<freemangordon> well, better take ohm from fremantle
<Wizzup> how is that better?
<freemangordon> because it was doen with mobile in mind
<freemangordon> and already cove maemo usecases
<freemangordon> like maemo-launcher :)
<freemangordon> lemme check that anynice
<Wizzup> does it support cgroup v2?
<Wizzup> or will be have to reinvent
<Wizzup> will we*
<freemangordon> no idea
<Wizzup> please check it out, I think it seems to do what we want
<freemangordon> ok, on it
<Wizzup> and it already has a config spec and everything
<freemangordon> oh
<freemangordon> it is "Ananicy rewritten in C++ for much lower CPU and memory usage."
<Wizzup> yes
<Wizzup> >I mostly used Ananicy on older computers to improve interactivity. However, having Ananicy use megabytes of RAM and a decent amount of CPU time troubled me. Thus I decided to rewrite it in C++, using an event based approach. RAM usage is much lower (only a few thousands of bytes !), and CPU usage is almost always zero thanks to its event-based implementation.
<inky> > inky has some magic kbd, but we don't have maintainer for it
<inky> i will package it.
<mkf> yay
<inky> and write a wiki page
<freemangordon> inky: and who will support the code?
<Wizzup> mkf: in daedalus there is vkb for qt5 and gtk2/gtk3 btw
<inky> mkf do you need one now?
<freemangordon> Wizzup: he does not like HIM vkb :)
<inky> it is a very small codebase, i can, and the author is my friend.
<Wizzup> freemangordon: btw I think we'd want to not build with ebpf support
<freemangordon> no idea what it is
<mkf> inky: if that works better than him in osso-xterm sure, that'd be awesome
<freemangordon> anyway, lemme pull/build and see
<inky> mkf just ping me and i'll guide u with the waay of building the kbd and mapping it to one of the hardware keys to hide it and show it.
<mkf> awesome, when?
<freemangordon> Wizzup: https://paste.debian.net/1364738/
<freemangordon> I hate packages pulling source code from random repos
<inky> these are screenshots https://norayr.am/weblog/2024/06/27/23926270/ (was excited when i first built it and stanted usitf, i see to life without it)
<inky> get this repo https://repo.coolbug.org/bw/coolkbd and i'll come back in five minutes.
<mkf> looks ok.
<mkf> hopefully it'll fit in my screen.
<mkf> freemangordon: should i change the governers btw? i recall n900 had problems with it in leste (being laggy on ondemand)
<freemangordon> afaik ondemand should be ok
<mkf> and do audio buttons appear to be reverse for you in your q8?
<freemangordon> could be, but didn't noticed
<freemangordon> Wizzup: and of course it is infected:
<freemangordon> /home/user/git/ananicy-cpp/src/platform/systemd/service.cpp:3:10: fatal error: systemd/sd-daemon.h: No such file or directory
<mkf> can you please check later on?
<mkf> idk if thats because of maemo or because of something in dtb
<Wizzup> freemangordon: you can disable systemd support
<Wizzup> and I can make an openrc init script np
<freemangordon> lemme try
<Wizzup> freemangordon: and yeah it's annoying that it pulls code, but at least it is versioned and statically linked
<Wizzup> bbiab
<freemangordon> Wizzup: can't we use auditd or stap that are already in the repos?
<inky> mkf i am back and i am near keyboard.
<inky> you got the repo, cd maemo, and run build.sh
<inky> you most probably want to use so called 'docked' version. once you build it, you'll see three versions: docked, bottom and top kbds.
<inky> you also want to have .xbindkeysrc, this is mine for bionic: https://bpa.st/QGJA
<inky> and you want to have xbindkeys package.
<inky> and this is the sw.sh script referenced in .xbindkeysrc: https://bpa.st/L7WA
<inky> when you press power button on bionic, it shows the keyboard, then kills it.
<inky> and this is .xbindkeysrc one i made for pinephone, very simple one: https://bpa.st/I5TQ
<inky> (file for bionic maps to its power button, file for pinephone: to its. you have to decide on the key and find its keycode by using xev program, also i see bionic's file has unnecessary for you things, key shortcuts to switch monitors, i use it on lapdock sometimes)
<inky> anyway, first build it, you'll see you can run ./coolkbd.dock
<inky> and enjoy. even without fancy stuff i just described.
<freemangordon> Wizzup: hmm, this thing seems to do polling, see check_freq
leste has joined #maemo-leste
<freemangordon> yep, it does wake-up every check_freq seconds
<freemangordon> default is 60
<freemangordon> Wizzup: anyway, this is not going to work on devuan/openrc without patching
<freemangordon> it does not even parse /proc/self/mounts properly
arno11 has joined #maemo-leste
<freemangordon> this mistakenly uses 1st token (device) to detect cgroups type, instead of 3rd (fs type)
<Wizzup> freemangordon: re: auditd, yes, we should start using it also when we get apparmor profiles
<Wizzup> it uses netlink but does polling? that'd be insane
<Wizzup> 60 seconds wake up is ok
<Wizzup> freemangordon: what is the problem on openrc/devuan btw?
<freemangordon> well, I don;t get it why it does n seconds wait, but that' not a biggie, we can set to hight enough value
<freemangordon> however, see ^^^ link
<freemangordon> because of that issye, it does not detect cgroups2 on devuan
<freemangordon> *issue
<Wizzup> it seems to do some full scan every now and then in your link, maybe we can just disable that too
<freemangordon> it seems it does it only if some other option is set
<Wizzup> re: cgroups, shouldn't the proc mount path be init independenent?
<freemangordon> it is
<freemangordon> but parser is buggy
<Wizzup> so it doesn't understand openrc cgroup paths, is that it?
<Wizzup> ok
<freemangordon> it is not the path
<Wizzup> well, upstream is active, so we can also file some reports with them
<freemangordon> it does not use fs type, but device instead
<Wizzup> did you get a sense of the memory usage?
<freemangordon> which is 'non' in our case :)
<freemangordon> no, lemme check
<Wizzup> and btw, if you can find something that's already in debian that does the same I am all for it
<Wizzup> but I couldn't find anything like that
<freemangordon> I was not able too
<freemangordon> that was the first thing to do - look if we already have something
<freemangordon> the closest is auditd
<freemangordon> so RES on VM is 6280
<Wizzup> auditd is quite different though
<freemangordon> yeah
<Wizzup> and I don't think we can set niceness with auditd, it's about auditing
<Wizzup> but you mean linjux audit subsystem?
<Wizzup> linux*
<freemangordon> well, it seems it might be able to execute code on system call
<freemangordon> but anyway, this is not what it was created for
<Wizzup> yeah
<freemangordon> so, are we ok with 6MB on 64bits?
<Wizzup> netlink os bpf is the way here
<Wizzup> I think so, is none of that shared?
<freemangordon> 4736
<freemangordon> that's SHR
<freemangordon> well, not that lightweight, given that we have ~200 processes
<freemangordon> but thats a debug build, lemme check release
<Wizzup> I think it's pretty good for what it does
<Wizzup> right
<mkf> i wonder if why a shell script that runs after processes isn't a good idea
<freemangordon> Wizzup: same
<freemangordon> for release that is
<Wizzup> just a sh instance uses 3-4MB ram
<freemangordon> :nod:
<mkf> ash or bash?
<Wizzup> dash I think, bash is more
<freemangordon> well, I can write a C code that does the same
<Wizzup> freemangordon: I think this is acceptable, given that it should also solve our cgroups problems
<freemangordon> will take 100k I guess
<Wizzup> I doubt that
<Wizzup> glibc alone will probably take more :)
<Wizzup> did you measure the actual ram increase in say htop - was that also 3-4M?
<freemangordon> Wizzup: ok, so shall I start patching that and sending PRs?
<Wizzup> from my pov, yes, but you looked at it longer than I did, so it's also up to you
<freemangordon> yeah, whole 5 minutes longer :p
<Wizzup> but this seems like it will get us 99% of the way there pretty quickly and it's packages on arch, gentoo, nix, etc
<Wizzup> packaged*
<freemangordon> right
<Wizzup> I didn't look at the cgroups stuff that it does, but presumably it'll be ok?
<freemangordon> will ask sicelo do become debian maintainer of it :)
<freemangordon> how would I know? (cgroups)
<Wizzup> hm, it seems to want cgroups v1 from the readme
<freemangordon> it doesn;t evern identify them properly here
<freemangordon> nto, it says it supports v2, but partially
<freemangordon> anyway, lemme fir it first
<freemangordon> *fix
<Wizzup> ok
apac has quit [Ping timeout: 244 seconds]
<freemangordon> I've already fixed it
<freemangordon> but now, I still think we have maemo-launcher issue
<freemangordon> lemme check if it will work
<freemangordon> it does
<freemangordon> Wizzup: however, I think we have issue there
<freemangordon> it uses basename as key in rules file
<freemangordon> lemme see what will happen if I create a script with the same name and start it
<Wizzup> or just try to apply it to say osso-xterm ?
<freemangordon> I already did to hildon-status-menu
<freemangordon> and it reniced both processes
<Wizzup> cool
<freemangordon> yeah, ok
_fab has joined #maemo-leste
<freemangordon> Wizzup: yeah, running another binary with the same name, resulted in ananicy identify it as the original one :)
<freemangordon> that's not very secure
<Wizzup> it can folow the exe symlink in proc
<freemangordon> sure, but then it will hit maemo-launcher :)
<Wizzup> how about absolute paths?
<freemangordon> hmm, seems it supports regexes
<freemangordon> lemme explore that
<mkf> inky: installed it
<mkf> it's a bit small in this device, and i'm afraid i dont have a way to pop it up
<inky> its easy
<mkf> this just has volume and power bottom
<inky> i increased the size
<inky> i should've gave u my sources
<inky> i would map to power
<inky> and ignore that power also brings upper menu
<inky> i do like that on pinephone and on bionic
<inky> what's your device?
<inky> let me find where was the size
<mkf> q8 tablet
<inky> i think in my last sources i use 50% of screen
<inky> wow, i dont know what is it, you were able to install/port leste on it yourself?
<inky> so i did
<inky> in coolkbd/maemo/config
<inky> config.h.dock
<inky> #define HEIGHT_SCALE 0.45
<inky> there u can define fonts and also
<inky> MIN_HEIGHT
<inky> MIN_WIDTH
<freemangordon> Wizzup: currently, it supports only "name" match, which is "basename()"
<inky> KEY_BORDER
<freemangordon> lemme see how hoard it to patch the full patch
<freemangordon> *path
<inky> mkf, please try and rebuild. (:
<inky> also you can configure which layouts do u need and rebuild by using only those u need
<inky> did you run build in maemo version ? in maemo directory?
<Wizzup> freemangordon: ok, ty
mkfx has left #maemo-leste [Error from remote client]
<mkf> ok let me see
<mkf> inky, yeah. with help of freemangordon, wizzup and sicelo. :D
<mkf> albeit previously a port was done for a device very simliar.
Anasko has quit [Ping timeout: 246 seconds]
<inky> good!
_fab has quit [Quit: _fab]
_fab_ has joined #maemo-leste
_fab_ has quit [Read error: Connection reset by peer]
arno11 has left #maemo-leste [#maemo-leste]
xmn has quit [Quit: Leaving]
arno11 has joined #maemo-leste
<arno11> freemangordon: i saw your mce and Xorg renice commits, cool. Xorg, mce and H-D have the biggest impact on leste n900. renice on other processes (like in fremantle) are quite useless atm
<arno11> and H-D renice is possible from the H-D fake job init script. i tried, it works
<arno11> so i mean, with just another commit for H-D, things work fine and it seems enough imo
Langoor has quit [Quit: No Ping reply in 180 seconds.]
pere has quit [Ping timeout: 252 seconds]
Langoor has joined #maemo-leste
<arno11> *enough for good perfs
<Wizzup> would anyone mind logging in to git.maemo.org using sign in with github and tell me what happens
<arno11> yep
<Wizzup> I guess it probably offers to make an account, and does that work?
<Wizzup> I might have to allow external registration first
<Wizzup> my hope/plan/idea is to allow users to report bugs with their github accounts, and 'we' can link github accounts to gitea, but most folks would get a gitea account
<donihalim> Wizzup: https://0x0.st/8jT8.png
<Wizzup> thx
<arno11> same for me
<Wizzup> maybe I should make the oath2 app not be owned by my user
<Wizzup> but rather by maemo leste organisation, if possible
<Wizzup> can one of you authorize being able to read the email and see what happens?
<Wizzup> I went through the same workflow but I linked it to my account
<Wizzup> I suspect you might not get through the registration yet
<arno11> sorry, can't paste a png but it asks for creating a new account or using an existing one
<arno11> ok
<arno11> i can't create a new account indeed
<arno11> registration is disabled
<Wizzup> ok, let me open it for external
<Wizzup> one sec
<Wizzup> thanks for thehewlp
<arno11> np
<Wizzup> does it work now?
<arno11> argh 505 internal server error
<Wizzup> ok
<Wizzup> might be because I restarted it mid-way, one sec
<arno11> yep
<arno11> it works again but still not able to register a new account
<Wizzup> ok, can you try once more? I just enabled general sign up now
<arno11> ok
<arno11> yeah seems to work now
<Wizzup> did you have to fill in some additional password to make an account, or just a username/email?
<mkf> what governer is the best suited for leste? linux 5.8 had performance but 6.12 has shedutil
<arno11> no additional password, just email/username
<mkf> i wonder if should change that
<Wizzup> arno11: cool, did you also make the account?
<Wizzup> https://git.maemo.org/merlijn/test-bugtracker can you file an issue here or something?
<arno11> let me try
<Wizzup> mkf: we use whatever the default is
<mkf> ok
<Wizzup> I think ondemand
<Wizzup> isn't schedutil one that allows userspace to schedule or something?
<Wizzup> (didn't look it up)
<Wizzup> arno11: looks like that worked
<arno11> yeah :)
<Wizzup> it also shows you as the second user on the gitea
<Wizzup> cool
<Wizzup> and just to confirm once more, you didn't have to set up some local password right?
<Wizzup> some online resource said that this was a bug in gitea, but it seems to no longer be the case
<arno11> no local password, just github credentials
<Wizzup> cool
<Wizzup> I'd like to disable local registration and only keep oauth2 unless we manually make accs, but I am not sure how doable that will be
<arno11> no idea
<Wizzup> ah
<Wizzup> let me try
<mkf> Wizzup: userspace is that one, shed uses scheduler to guess
<Wizzup> arno11: ok, cool, I think this is it
<arno11> cool
<Wizzup> this allows github users to register with oauth2, and regular registration is disabled
<mkf> what if someone doesnt have github
<Wizzup> yeah..hm
<Wizzup> argh :)
<Wizzup> I just don't want to deal with spam accounts
<Wizzup> my plan was to make accounts on demand, but make it easy for github users to file issues
<Wizzup> most of us here could just get a gitea account
<freemangordon> Wizzup: hmm, I don;t think that ananicy will do the job for cgroups, if we use it, in its current shape, it will move processes out of elogind cgroup
<freemangordon> which is the criteria if a process belongs to s session
<Wizzup> where would we want the processes to go to
<Wizzup> do we want them in the elogind cgroup?
<freemangordon> they are alredy there
<freemangordon> *already
<Wizzup> right, but how would we make additional cgroups?
<freemangordon> mkdir
<Wizzup> I mean what would we do different than ananicy
<freemangordon> sorry, I don;t follow
<Wizzup> can a process be in multiple cgroups
<freemangordon> not sure
<Wizzup> if not, how do we work around the problem with processing having to be in the elogind cgroup
<Wizzup> can we stack cgroups?
<freemangordon> I think if it is a subdir of elogind session dir, it still counts
<freemangordon> lemme check
<freemangordon> hmm, ok, it seems process can be in multiple cgroups
<freemangordon> lemme ask goggle/AI what is that supposed to mean
<mkf> freemangordon: given our kernel is mostly patch free, i wonder if we can use a config from "known-good" setup for mobile/desktop rather than building one by hand?
<mkf> idk if i should configure kernel by hand or there is a better way to do that :)
<freemangordon> no good known config will enable drivers from staging :)
<freemangordon> but yeah, if you find one...
<Wizzup> ideally we'd have a single 32 bit arm kernel
<Wizzup> so if the patches can go into our omap kernel...
<freemangordon> Wizzup: according to AI, a process can be in a different cgroup, as long as each cgroup manages different resource
<freemangordon> seems to be ok for our use case
<Wizzup> yeah, assuming elogind doesn't manage any
<freemangordon> mhm
arno11 has left #maemo-leste [#maemo-leste]
<Wizzup> ok, not the oauth2 uses the maemo-leste github orga
<Wizzup> now the*
lyubov has quit [Read error: Connection reset by peer]
lyubov has joined #maemo-leste
lyubov has quit [Read error: Connection reset by peer]
mkfx has joined #maemo-leste
<mkf> freemangordon: ok, i'm now in mainline linux
<mkf> applied your patch, icd still can't connect
<mkf> however with wpa_cli i can connect, sometimes.
lyubov has joined #maemo-leste
ashley has quit [Quit: Whatever.]
ashley has joined #maemo-leste
ashley is now known as Guest5331
Guest5331 has quit [Remote host closed the connection]
ashley has joined #maemo-leste
ashley is now known as Guest6923
pere has joined #maemo-leste
Guest6923 is now known as ashley
ashley has quit [Changing host]
ashley has joined #maemo-leste
pere has quit [Ping timeout: 260 seconds]
DPA has quit [Ping timeout: 265 seconds]
Livio has joined #maemo-leste
DPA has joined #maemo-leste
<freemangordon> mkf: weird
<freemangordon> ok, why the hell gitlabhp blocked my account?!?
<mkf> any idea where to look to debug this issue?
<freemangordon> no, sorry
<Wizzup> freemangordon: which gitlab service
<freemangordon> gitlabhq
<Wizzup> strange
<freemangordon> Wizzup: from https://gitlab.com/ananicy-cpp/ananicy-cpp
<freemangordon> I twanted to sign-in to fork
<Wizzup> no idea, any reason provided
<freemangordon> no
<freemangordon> tried to login with my github account
<freemangordon> maybe I have created a real account there, can't remember
<freemangordon> tried to reset the password, no email
<freemangordon> "422: Sign-in using GitHub auth failed "
<Wizzup> :/
Livio has quit [Ping timeout: 264 seconds]
<freemangordon> ok, what now?
<freemangordon> like, I made some fixes
<freemangordon> when try to register, they say both username and email are already used
<freemangordon> but they don;t send password recovery link
<freemangordon> wtf?
<Wizzup> :/
<mkf> wifi seems to work without the patches. odd.
<freemangordon> yeah
<freemangordon> could be FW dependent
arno11 has joined #maemo-leste
<freemangordon> Wizzup: I registered another account, and made PRs
<freemangordon> now, what we do for leste?
<freemangordon> would you package that https://gitlab.com/freemangordon1/ananicy-cpp ?
<arno11> Wizzup: btw, i found a bug with calls: if sound is activated in profile, it breaks sound in call. otherwise no trouble in silent mode. probably something buggy with ringtone. (shm bytes, remixing, ucm2 are ok). will open an issue
<arno11> *on daedalus
<Wizzup> freemangordon: yes I can package it
<freemangordon> please do
<freemangordon> or rather
<freemangordon> wait to see how my PRs will get treated
<freemangordon> if upstream is unresponsive, or does not care, I would prefer to not use this code
<Wizzup> ok
<Wizzup> freemangordon: they might not want to take non-backwards compat changes
<Wizzup> or at least some way to toggle between then
<Wizzup> them
<freemangordon> both patches are backward-compat :)
<Wizzup> great :)
pere has joined #maemo-leste
<mkf> can i have a gittea account?
<Wizzup> yeah once the setup is done/migrated I'll start doing that
<Wizzup> right now it's still just a demo setup
<Wizzup> we haven't migrated anything yet from github, I might also just open the reg up completely
<Wizzup> for now we're still using github
<mkf> ok
<Wizzup> mkf: if you have a use for it now, I can give you one too, it's just just not useful yet at this point, I'm going to remove/delete all the repos that are on there currently before we switch
<arno11> Wizzup: again @renice, the 2 fmg's commits for mce and xorg + H-D renice give good perfs. you should try. not sure it is useful for you, devs, to spend time on cgroup ATM. but ofc that's just my POV.
<Wizzup> I think fmg already built the packages
<arno11> ah ok, so worth a try ofc
<mkf> idk a git repo would be nice. :)
<mkf> i was going to upload kernel somewhere
<mkf> and device trees
<arno11> Wizzup: btw HW accel seems not working on n900: xorg returns 'GLX: Initialized DRISWRAST GL provider for screen 0'
<arno11> that's maybe why Qt5 is buggy without raster
<sicelo> Wizzup: you will also need to watch out for AI/LLM scrapers. seems many forges are struggling with that these days. some have implements some strict login measures as a result
<Wizzup> sicelo: yeah we get that a lot with archive.org too
<freemangordon> arno11: GLX is GL, not GLES
<dsc_> compiling chromium, my pc starting to make strange noises...
<dsc_> freemangordon: since you do kernel stuff, do you have a super computer, or no
<dsc_> how many cores?
<arno11> freemangordon: oh yeah. so you mean that's just some generic stuff failing on boot. ok so the problem is eslsewhere
<arno11> *elsewhere
<Wizzup> dsc_: kernels compile way faster than chromium
<Wizzup> like 10x or something :)
<dsc_> :P
<arno11> freemangordon: indeed, es2_info seems to return proper output