humanbeta | I tested Motorola Altrix Lapdock with Pinephone Keyboard..it works..have to push two times button of PP Keyboard and write: xrandr --output HDMI-1 --mode 1366x768 | 11:14 |
---|---|---|
humanbeta | needs quite many adapters to get connected. | 11:18 |
humanbeta | brightness controls and volume controls (inverted) works. | 11:35 |
Wizzup | cool | 11:35 |
Wizzup | arno11: cmt_pulse starting later, is that because otherwise pulse is not ready yet, or? | 11:37 |
arno11 | Wizzup: hi, nope cmt_pulse works fine excepting the alerting tone | 11:49 |
humanbeta | here is command needed to touchpad scrolling screen to see all pixels from right side of PP with Motorola Altrix: xrandr --fb 1440x768 --output HDMI-1 --mode 1366x768 --panning 1440x0 | 11:57 |
arno11 | Wizzup: starting cmt_pulse a bit later makes things working fine (like starting cmt_pulse from the user session) | 11:58 |
humanbeta | *Motorola Altrix Lapdock | 11:59 |
arno11 | now we just need to find a clean way to set RT prio for PA and sphone (cmt_pulse doesn't need any particular prio now) | 12:00 |
Wizzup | arno11: what I mean is that it doesn't work when started earlier? | 12:03 |
Wizzup | arno11: also, does sphone really need RT? not just PA? | 12:03 |
uvos__ | sphone in a call dose exactly nothing, it sits in select | 12:06 |
Wizzup | yeah | 12:06 |
uvos__ | rt prevent sphone from being swapped out | 12:06 |
uvos__ | swapped out sphone might be a problem when a call comes in | 12:06 |
uvos__ | and when you want to stop a call in extream circumstances | 12:06 |
uvos__ | but this is better solved by a cgroup | 12:07 |
uvos__ | that mlocks | 12:07 |
arno11 | Wizzup: as i said it works fine if it starts earlier | 12:08 |
Wizzup | arno11: ah ok, because your last commit makes it start later | 12:10 |
arno11 | yes ofc, to make the alerting tone working again | 12:11 |
Wizzup | uvos__: I see | 12:11 |
Wizzup | arno11: ok | 12:11 |
arno11 | for sphone, i really don't know | 12:11 |
Wizzup | it might be the swap thing | 12:11 |
arno11 | yes indeed | 12:12 |
uvos__ | Wizzup: oh btw, your sphone.ini.d file for sphone | 12:15 |
uvos__ | i dont like that i lacks a number prefix, sphone sorts the files based natsort haveing a name like 50-* makes it obvious what overrides what | 12:16 |
Wizzup | ok, easy fix | 12:16 |
uvos__ | yeah just mentioning | 12:16 |
Wizzup | re droid3 | 12:18 |
Wizzup | I don't know what to do next other than analyse the regs and make an empty dts, but I will probably go back to tp first since I feel I'm not effective in this area | 12:19 |
uvos__ | Wizzup: ok | 12:20 |
uvos__ | ill diff the cpcap regs soon | 12:21 |
uvos__ | looking for sutch issues is never very effective | 12:21 |
uvos__ | lots of blind poking around with a stick | 12:22 |
uvos__ | i want jtag :( | 12:22 |
Wizzup | yeah, but I also forgot what we did like in 2020 or 2021 wrt sgx so that doesn't help either :p | 12:22 |
Wizzup | ok @ regs, I will keep the devices close by so I can do/run other stuff as needed | 12:23 |
arno11 | Wizzup: ah, looking @my notes, sphone rt prio was a workaround when N900 was using 24bpp...because most of the time sphone window was too slow to appear during an incoming call. | 12:30 |
arno11 | now it seems fine with stock freq and 16bpp (and no sphone rt prio) | 12:31 |
arno11 | so we just need to focus on PA | 12:33 |
arno11 | btw giving high rt prio to PA makes yt-dlp working amazingly well to stream videos on n900 | 12:36 |
arno11 | the trick is to get the real google video link through yt-dlp with --get-url option (it takes around 20sec) and then start the stream from smplayer. | 12:39 |
arno11 | it works with live streaming as well (if 240-360p is available ofc) | 12:40 |
Wizzup | we shouldn't need too many of these rt tricks, maybe we should figure out what it really does for you | 12:41 |
arno11 | there is just one trick now: PA | 12:43 |
arno11 | and that's not really a trick with a low power device imo | 12:44 |
Wizzup | I understand, we need to up PA scheulding prio | 12:46 |
Wizzup | oh, ok, I misread what you wrote earlier | 12:47 |
arno11 | :) no probs | 12:47 |
Wizzup | doesn't pa have a realtime flag/option? | 12:47 |
Wizzup | realtime-scheduling in daemon.conf? | 12:47 |
Wizzup | or, how do you set ti? | 12:47 |
Wizzup | it* | 12:48 |
arno11 | the rt flag in daemon.conf doesn't work | 12:48 |
arno11 | only chrt works without rt kernel iirc | 12:49 |
Wizzup | ok | 12:49 |
arno11 | bbl | 13:02 |
uvos__ | cgroups would work | 13:25 |
uvos__ | idk how to manage cgroups without the advantage of systemd however | 13:25 |
Wizzup | openrc can do cgroups for sure | 13:25 |
uvos__ | a rt and a mlock cgroup would be great | 13:25 |
Wizzup | https://wiki.gentoo.org/wiki/OpenRC/CGroups | 13:26 |
uvos__ | Wizzup: sure but openrc dosent start processies in user contexts | 13:26 |
uvos__ | Wizzup: systemd-user is another full systemd for eatch session that allows one to do these kinds of things properly | 13:26 |
Wizzup | isn't pa still started through /etc/init.d ? | 13:26 |
uvos__ | pa sure | 13:26 |
uvos__ | but the stuff we want to mlock no | 13:26 |
uvos__ | actually pa no | 13:27 |
uvos__ | we switched to session pa a while back | 13:27 |
Wizzup | it looks like with cgclassify one can move processes to a cgroup | 13:28 |
Wizzup | or through procfs or so | 13:28 |
uvos__ | i gues we have to add this to dsme, which itself is really just a hack around not having a real session level init itself anyhow | 13:28 |
Wizzup | more than anything it just predates a lot of this stuff :) | 13:29 |
uvos__ | but then it wont work for dbus activation | 13:29 |
uvos__ | hmm | 13:29 |
freemangordon | dsme already moves processes it starts to elogind session if one exists (and is defined through env var) | 13:34 |
freemangordon | moving process to elogind session is basically assigning it to a cgroup | 13:34 |
freemangordon | so at least XSession dsme started processes already belong to /sys/fs/cgroup/elogind/c1/ | 13:35 |
freemangordon | uvos__: but, my cgroups-fu is not good enough so I have no idea what will happen if we create subdir of c1 dir with mlock and whatever cgroups settings we want for some pids | 13:36 |
uvos__ | freemangordon: thatl work but not for processies dsme dosent start directly like dbus activates ones | 13:37 |
Wizzup | from a practical pov, do we need to mlock some of the dbus activated ones? | 13:38 |
freemangordon | do we care? I mean - do you have a particular process in mind that woudl suffer from that issue? | 13:38 |
freemangordon | Wizzup: exactly | 13:38 |
freemangordon | uvos__: we can always pre-start dbus process | 13:39 |
freemangordon | if we want it mlocked or something | 13:39 |
freemangordon | uvos__: on fremantle, those thing are done by ohm | 13:40 |
freemangordon | so, we can choose to have ohm in leste as well | 13:40 |
uvos__ | i gues, but n900 would probubly be thankfull if not everyhing where started on boot directly since it gets hit pretty roughly on boot | 13:40 |
freemangordon | right, but thats another story, like, we shall reconsider what needs to be started in xsession.d and what in xsession.post | 13:41 |
freemangordon | assigning pids to cgroup is orthogonal to that, no? | 13:42 |
freemangordon | also, I think elogind supports some kind of "process started" agent | 13:42 |
freemangordon | so we may hook on that | 13:43 |
uvos__ | no idea, maybe | 13:44 |
freemangordon | still, what will happen if we move a pid from c1 to a subdir, do you know? | 13:44 |
freemangordon | or, it has to be in both dirs? | 13:44 |
freemangordon | I don;t really have details on how cgroups work in that regard | 13:45 |
uvos__ | im pretty fuzzy on this since systemd deals with the details with this on other systems | 13:46 |
uvos__ | but i seam to remember the hierachy gettig removed in cgroups2 | 13:46 |
uvos__ | so that may be legacy in the first place | 13:46 |
uvos__ | but i would have to look it up | 13:47 |
freemangordon | please do what you have some spare time | 13:47 |
freemangordon | *when you* | 13:47 |
freemangordon | as without those details we won;t be able to fix it properly | 13:47 |
freemangordon | I'll experiment a bit on leste as well in the meanwhile | 13:48 |
uvos__ | check | 13:48 |
freemangordon | hmm, we cannot mlock through cgroups | 13:57 |
freemangordon | we can only set swappiness to 0 | 13:57 |
freemangordon | I have to check what ohm does on fremantle | 14:00 |
uvos__ | cgroups has unevictable which is mlock | 14:22 |
uvos__ | besides some difference in behavirou around non-anynomes mmaps | 14:23 |
diejuse | testing testing | 17:30 |
Wizzup | works | 17:49 |
diejuse | perfect. It illll | 17:54 |
diejuse | From Maemo Leste, proot, Z Flip5, Pidgin app | 17:55 |
diejuse | and okuda theme hehe | 17:58 |
arno11 | diejuse: Leste proot on Z Flip5 ? amazing !!! | 17:59 |
diejuse | yes! | 18:00 |
freemangordon | video or didn't happen :p | 18:01 |
bencoh | using halium ? | 18:04 |
diejuse | I will make and upload a video when all problems are solved. osso apps not working for example | 18:04 |
bencoh | which makes me think ... maybe I should try using proot for the crossbuilder instead of lxc | 18:05 |
bencoh | it might be simpler for everyone | 18:05 |
freemangordon | I guess maemo-launcher is not started | 18:05 |
bencoh | diejuse: how does the graphic stack work with proot on android? | 18:06 |
diejuse | i am shopping. latter I will tell | 18:07 |
bencoh | :) | 18:07 |
diejuse | bencoh When I chrooted Maemo with my Unihertz Titan smartphone I used XSDL Xserver app as X Window System / X11 server for Android. Now I am using Termux-X11 which works better. It is more efficient and uses less battery. | 22:19 |
diejuse | battery power | 22:19 |
diejuse96 | freemangordon The problem is "/etc/init.d/dbus start". I get "[FAIL] dbus is not running ... failed!". And errors like "Failed to connect to socket /tmp-dbus-xxxxxxx: No such file or directory" from hildon-home, hildon-desktop and hildon-input-method | 22:26 |
diejuse96 | But that happens on my Z Flip5 (Android 14) and Surface Duo (Android 12) but not on older phones like Blackberry Key 1 (Android 7) and ZTE Axon M (Android 7). Apparently the higher the version of Android, the more secure it is and the more difficult it is to root and mount directories such as /proc and /sys. | 22:29 |
diejuse96 | The fact is that on the most modern mobile phones I cannot see the time in the status bar, nor open other apps such as X terminal, Setting, Clock, PDF reader, App manager. But I can use LXTerminal, Abiword, Thunar... The main screen backgrounds and window management work well. | 22:33 |
freemangordon | diejuse96: without session bus... | 22:46 |
freemangordon | why would dbus not start? | 22:47 |
diejuse96 | I'm investigating but I don't know. It throws an error message: "/etc/init.d/dbus: 50: [: -lt: unexpected operator" and then "[ok] Starting system message bus: dbus" but it is not real because with "/etc/init .d/dbus status" get FAIL. | 22:53 |
freemangordon | ah, startup script fails | 22:58 |
freemangordon | hmm, /etc/init.d/dbus is not provided by us, it is dbus package that provides it, afaik | 22:59 |
freemangordon | yeah, that's upstream provided | 22:59 |
freemangordon | seems shell interpreter is buggy or something | 23:00 |
diejuse96 | In any case, I modified the shebang of the script to bash and adapted line 50, ensuring that there were no errors. But it still doesn't work. Therefore I think the script is not the problem. As I said before, on my older phones with Android 7 there is no problem with the script. | 23:00 |
freemangordon | dif you try to run dbus by hane, to see if it starts? | 23:00 |
freemangordon | *hand | 23:01 |
freemangordon | I mean dbus the binary | 23:01 |
diejuse96 | The truth is that I don't know what the problem is. Well, in summary, I think the problem is proot with Android 14 is more difficult for security reasons than with Android 7. | 23:01 |
freemangordon | I see | 23:01 |
freemangordon | well, I can;t help here, my android knowledge is close to zero | 23:02 |
diejuse96 | What is your daily use smartphone? I'm curious, | 23:08 |
Wizzup | nokia n900 ;) | 23:08 |
diejuse | The best. It is the phone that I remember best and that is why I try to emulate its system on modern cell phones. Unfortunately the battery had a very short life. | 23:14 |
diejuse | *has | 23:15 |
diejuse | What steps do I do for vertical rotation? | 23:19 |
Wizzup | diejuse: xrandr might be enough | 23:43 |
Wizzup | but there is also a dbus message to change rotation | 23:43 |
Wizzup | or a signal, rather | 23:44 |
diejuse | I understand. Is it possible to open an application like osso-xterm without using dbus, how can I do it with xfce4-terminal? | 23:45 |
Wizzup | I am not sure if I fully understand the question | 23:46 |
diejuse | I mean that, using Maemo proot on modern Android smartphones, I can open Debian applications but not Maemo's own applications such as osso-xterm, Settings, Clock, PDF Reader... | 23:48 |
diejuse | The reason is that with Maemo's own applications I get the error "Could not connect to System D-Bus. Failed to connect to socket /run/dbus/system_bus_socket: No such file or directory". I don't have that error when I open Debian applications. | 23:49 |
Wizzup | right, mameo-launcher might not be running | 23:50 |
Wizzup | or at least applications can't talk to it over dbus | 23:50 |
Wizzup | that is definitely worth fixing | 23:50 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!