lel | levomer opened an issue: https://github.com/maemo-leste/bugtracker/issues/584 (droid4:Only speaker output works for audio during calls) | 09:12 |
---|---|---|
Wizzup | someone who actually manages to get further in call audio than me :-D | 09:13 |
Wizzup | uvos: https://github.com/maemo-leste/bugtracker/issues/521 - you wrote you use it quite often, in car or on foot? | 12:25 |
Wizzup | freemangordon: can we close https://github.com/maemo-leste/bugtracker/issues/207 ? | 14:11 |
lel | MerlijnWajer closed an issue: https://github.com/maemo-leste/bugtracker/issues/207 (Import modest/tinymail (mail client)) | 14:14 |
lel | MerlijnWajer edited a pull request: https://github.com/maemo-leste/libicd-network-ipv4/pull/3 (etc/50_ipv4_network_setup: use resolvconf) | 14:49 |
lel | MerlijnWajer edited a pull request: https://github.com/maemo-leste/libicd-network-ipv4/pull/3 (etc/50_ipv4_network_setup: use resolvconf) | 14:49 |
Wizzup | uvos: parazyd: should we add the profiles cpa to the hildon-meta ? | 15:11 |
Wizzup | same for modest | 15:11 |
freemangordon | Wizzup: we need fixed libglib | 16:43 |
freemangordon | which reminds me to ping the upstream | 16:43 |
Wizzup | freemangordon: ok, well, we can build that now | 16:46 |
freemangordon | oh, we also need the thumbnailer | 16:48 |
freemangordon | ummm.... | 16:48 |
freemangordon | what was the name? | 16:49 |
freemangordon | hildon-thumbnail | 16:49 |
Wizzup | it's probably in repodiff | 16:51 |
Wizzup | freemangordon: what should I write about osso-abook in our news update | 19:00 |
freemangordon | not sure... maybe something between the lines "soon to be ready" | 19:01 |
Wizzup | any screenshot or something to share? | 19:02 |
Wizzup | https://maemo-leste.github.io/images/VirtualBox_leste-beowulf_17_09_2020_15_06_07.png we shared this last time | 19:02 |
sunshavi | Wizzup: is that an abook pic? | 19:06 |
Wizzup | sunshavi: from our previous update, yes | 19:09 |
freemangordon | Wizzup: lemme check if I keep some other pic | 19:22 |
freemangordon | Wizzup: trying to send you a file via irc :) | 19:25 |
freemangordon | LMK if it doesn't work | 19:25 |
Wizzup | freemangordon: ah let me check | 19:26 |
Wizzup | I don't think it works | 19:27 |
freemangordon | ok, will put it on my web server | 19:27 |
Wizzup | meanwhile I'm extending the infobox a bit to automatically add categories for some packages https://leste.maemo.org/Category:Packages | 19:28 |
freemangordon | Wizzup: I have 2 screenshots of modest, do you want them as well? | 19:29 |
Wizzup | sure, I did tweet some which I planned to use | 19:31 |
Wizzup | but why not | 19:31 |
Wizzup | I'll see what I include | 19:31 |
freemangordon | /var/www/leste/screenshots_nov_2021 | 19:31 |
freemangordon | oops | 19:31 |
freemangordon | http://46.249.74.23/leste/screenshots_nov_2021/ | 19:32 |
freemangordon | this | 19:32 |
Wizzup | great, ty | 19:32 |
buZz | 'are sent to Nokia' | 19:39 |
buZz | :D | 19:39 |
freemangordon | well, yeah | 19:40 |
freemangordon | this has to be fixed | 19:40 |
Wizzup | this is the case in various places | 19:41 |
Wizzup | (not the sending, but the nokia references) | 19:41 |
buZz | hmhm, yeah like in the appmanager | 19:41 |
Wizzup | right | 19:43 |
Wizzup | actually this could just be fixed in weblate now | 19:43 |
sunshavi | Wizzup: thx for the answer | 20:22 |
sunshavi | nice pictures | 20:22 |
tmlind | freemangordon: so i tested your patched mesa finally, glmark2-es2-wayland still causes sgx lockup oops | 20:22 |
uvos | Wizzup: "hm I think having the battery guard without charge mode is quite tricky" ??? | 20:46 |
uvos | makes no sense to me battery guard IS charge mode | 20:46 |
uvos | Wizzup: "uvos: I *think* it resets my device at like 35% battery" it powers the device off if it stops charging for more than 5 seconds | 20:47 |
uvos | Wizzup: this never happens to me on my main charger but when connected to my laptop cpcap can jus stop charging for no reason | 20:47 |
uvos | this causes a reboot because then the device will shut off as its not charging anymore | 20:48 |
uvos | and then after some time mbm will trigger a boot again because the device is charging again. | 20:49 |
uvos | this really is just a bug in the mapphone kernel unrelated to charge-mode | 20:49 |
uvos | Wizzup: "hm now it says 2% charging, makes no sense" if the device is uncallibrated (very likely when entering charge-mode on d4) it just uses what the kernel reports as a voltage and current and estimates the charge state | 20:51 |
uvos | unfortionatly this is wildy unaccurate | 20:51 |
uvos | again this is more down to the fact that battery callibration dosent really work right on d4 (we dont save the charge counter becasue we lack nvram) | 20:51 |
Wizzup | hi | 20:53 |
Wizzup | what I meant was that my device was powered off when it wasn't really depleted, at least percentage wise, it wasn't being charged at the time, but yeah, later on I did try to charge it and I thought that was happening | 20:53 |
uvos | Wizzup: cant parse | 20:54 |
uvos | a more detailed desctiption of behavior please | 20:54 |
uvos | freemangordon: so i dont see what the problem is with modesetting with your glamor replacement. | 20:54 |
tmlind | uvos: would be nice to save the charge_full value into some cpcap scratch register for reboot, probably also a timestamp must be saved | 20:54 |
uvos | tmlind: yeah | 20:54 |
uvos | freemangordon: it tears, thats more or less unavoidable in x | 20:55 |
uvos | freemangordon: its slow in rotation | 20:55 |
uvos | freemangordon: well thats not really its department drm should expose accelerated rotation and modesetting should be using it | 20:55 |
uvos | freemangordon: if its not working figureing out why would make more sense than to hack pvr2d acclerated rotation into a ddx | 20:55 |
uvos | freemangordon: maybe look at wlroots it rotates fast | 20:56 |
uvos | Wizzup: regaring the sphone problem | 20:58 |
tmlind | uvos: i recall also with weston there is not a huge difference in fps in landscape compared to portrait mode, only tested looking at the es2gears output so not accurate | 20:58 |
uvos | Wizzup: i just tryied repoducing that | 20:58 |
uvos | tmlind: its "fine" for me in sway | 20:58 |
uvos | tmlind: its a bit slower but not massively slow | 20:58 |
tmlind | agreed, yeah it's just fine for me too | 20:59 |
uvos | es2gears is a great test for this btw | 20:59 |
uvos | it dose little work besides fliping the buffers | 20:59 |
tmlind | ok | 20:59 |
uvos | so its a test of buffer flipping perfomance :) | 20:59 |
uvos | Wizzup: so cant repoduce the sphone problem atm | 21:00 |
Wizzup | ok | 21:00 |
uvos | Wizzup: i called myself | 21:00 |
Wizzup | sorry, I will try to catch up momentarily | 21:00 |
uvos | Wizzup: and hung up on the remote side | 21:00 |
uvos | and the d4 stoped rinnging as expected | 21:00 |
Wizzup | sphone also said it was 'not responding' | 21:00 |
uvos | oh ok | 21:00 |
uvos | hmm | 21:00 |
Wizzup | yeah | 21:00 |
Wizzup | it's ok, I'll debug later | 21:00 |
uvos | sure | 21:01 |
Wizzup | I was outside in the park showing my parents the device, hehe | 21:01 |
Wizzup | so wasn't really able to debug | 21:01 |
uvos | upps :P | 21:01 |
uvos | ok can you still repo? | 21:01 |
Wizzup | I can try later tonight | 21:01 |
Wizzup | let me read up | 21:01 |
uvos | ok ill do a relese of the sphone version in the repo atm | 21:01 |
uvos | since thats what i use (and fixes the ui issues) | 21:02 |
Wizzup | cool | 21:02 |
Wizzup | looked like pavel was interested too | 21:02 |
Wizzup | and a user filed a bug report about the headphones not working for calls :P | 21:02 |
uvos | heh | 21:02 |
Wizzup | when/if should we use sphone to non-devel? maybe just not yet, right | 21:03 |
uvos | not yet | 21:03 |
Wizzup | ok | 21:03 |
Wizzup | I started porting yappari, it'll be quite some work | 21:03 |
Wizzup | but made a start at least | 21:03 |
uvos | not even considering how green my new sphone backend code is, calling needs to work right on some defice first | 21:03 |
uvos | kernel wise | 21:03 |
Wizzup | right | 21:03 |
Wizzup | we actually don't have -devel documented anywhere, which is perhaps a bit silly | 21:04 |
uvos | that is silly | 21:04 |
Wizzup | yeah, ok | 21:06 |
freemangordon | tmlind: does glmark2-es2-wayland oopses the kernel for you? | 21:18 |
freemangordon | uvos: keep in mind I am using kernel with latest tmlind's patches | 21:18 |
freemangordon | not that I tried WL though | 21:19 |
freemangordon | uvos: how did you test modesetting performance? | 21:20 |
freemangordon | I mean - do you use pathced glamor? | 21:20 |
freemangordon | *patched | 21:20 |
uvos | freemangordon: ? | 21:20 |
uvos | freemangordon: i did not test modesetting performance | 21:20 |
freemangordon | "(21,54,14) uvos: freemangordon: so i dont see what the problem is with modesetting with your glamor replacement." | 21:21 |
freemangordon | there are lots of problems | 21:21 |
uvos | right that was more of a question | 21:21 |
freemangordon | ah, ok | 21:21 |
uvos | what is the problem rn? | 21:21 |
uvos | sry that was unclearly worded | 21:21 |
uvos | so you reported rotation is slow and tearing | 21:22 |
uvos | what else? | 21:22 |
uvos | thats what i wanted to say | 21:22 |
freemangordon | not only the rotation, on n900 rendering is slow | 21:22 |
freemangordon | even in landscape | 21:22 |
freemangordon | with compositing that is | 21:22 |
uvos | freemangordon: ok | 21:23 |
freemangordon | this can be fixed | 21:23 |
freemangordon | or at least optimized | 21:23 |
uvos | ok and rotation needs to be solved in omapdrm if possible | 21:23 |
uvos | omap4 should expose tiler as acclerated rotation | 21:23 |
uvos | and on ompa3 im not sure | 21:23 |
freemangordon | but, if bad rotated performance can't be solved, I don;t see a point in continuing in that direction | 21:24 |
freemangordon | n900 does not have a tiler | 21:24 |
freemangordon | omap3 that is | 21:24 |
freemangordon | also, 16bpp does not work in modesetting | 21:24 |
uvos | why not? | 21:24 |
freemangordon | no idea | 21:24 |
uvos | it should work according to docs | 21:24 |
uvos | 16bpp is unusable anyhow | 21:24 |
freemangordon | well, es2gears work, but not hildon-desktop | 21:24 |
uvos | since its really slow with clients | 21:25 |
uvos | some clients dont work too | 21:25 |
freemangordon | not maemo clients | 21:25 |
freemangordon | they will have to be fixed | 21:25 |
uvos | no its just over with that | 21:25 |
freemangordon | we gain nothing from 24 bpp | 21:25 |
freemangordon | says who? | 21:25 |
freemangordon | uvos: please, lets not go into that | 21:25 |
freemangordon | we have low resources devices which will benefit a lot of reducing the needed bus bandwidth twice | 21:26 |
freemangordon | *lot from | 21:26 |
freemangordon | esp n900 | 21:26 |
freemangordon | but, 16bpp is a small problem | 21:27 |
freemangordon | rotation is major one IMO | 21:27 |
freemangordon | in general - MS/glamor is created with desktop/opengl in mind | 21:27 |
freemangordon | at least that's my understanding | 21:27 |
freemangordon | uvos: oh, BTW, did you pull the latest pvr mesa? | 21:28 |
uvos | freemangordon: not sure what is latest | 21:28 |
uvos | why? | 21:28 |
freemangordon | because I pushed a fix for WL segfaulting | 21:28 |
uvos | oh ok | 21:28 |
freemangordon | (hopefully a fix) | 21:28 |
uvos | no dident try that | 21:28 |
uvos | yet that is | 21:28 |
freemangordon | https://github.com/freemangordon/mesa/commit/5bd40a453e1484efea842be9a3f388adb5fac93c | 21:29 |
freemangordon | hmm? | 21:29 |
freemangordon | ah, you didn't try it yet | 21:29 |
uvos | right | 21:30 |
freemangordon | uvos: so, on DDX - MS/glamor is not made with mobile in mind | 21:30 |
uvos | i dont see what special things we need on "mobile" right now | 21:30 |
freemangordon | it has lots of features I guess, but it requires resources we don;t have on mobile | 21:30 |
freemangordon | performance | 21:30 |
uvos | besides the work you already did | 21:30 |
freemangordon | rotation support | 21:30 |
uvos | it has rotation support | 21:30 |
freemangordon | low memory footprint | 21:31 |
uvos | supporing rotation is kernels job | 21:31 |
uvos | and ms suppots those interfcaces | 21:31 |
freemangordon | also, even in native orientation it still does some copy, somewhere | 21:31 |
uvos | not sure what low memory footprint means in this context, non glamor modesetting has exactly one buffer | 21:31 |
freemangordon | not really, it has shadow buffer as well | 21:32 |
uvos | it dosent have to | 21:32 |
freemangordon | which is used for rotation | 21:32 |
freemangordon | without it it cannot rotate | 21:32 |
uvos | it can with drm accelerated rotation | 21:32 |
freemangordon | ok, please enable it | 21:32 |
uvos | freemangordon: sure but thats more of a pvr problem than a mobile vs desktop problem | 21:32 |
freemangordon | but, we have PVR | 21:32 |
uvos | yeah sure | 21:33 |
uvos | freemangordon: i think tiler would a possiblity | 21:33 |
uvos | on n900 yeah | 21:33 |
freemangordon | so, we have to deal with it | 21:33 |
uvos | no idea | 21:33 |
freemangordon | mhm | 21:33 |
freemangordon | also, given we use gles, which can rotate as we want, doing SW rotation is not the best solution | 21:33 |
uvos | so what do you want to do? | 21:34 |
uvos | wirte a new ddx that uses gles? | 21:34 |
freemangordon | write ddx from scratch | 21:34 |
freemangordon | no | 21:34 |
freemangordon | that uses mmap-ed bo's | 21:34 |
uvos | and rotates via gles | 21:34 |
freemangordon | similar to modesetting, but without overhead it has | 21:34 |
uvos | im unsure where modesetting has avoidable overhead.... | 21:35 |
freemangordon | modesetting/glamor actually hack xorg interfaces | 21:35 |
freemangordon | there is not EXA there | 21:35 |
uvos | exa is depricated | 21:35 |
uvos | those interfaces | 21:35 |
freemangordon | and what is the next shiny? | 21:35 |
uvos | glamor :P | 21:35 |
uvos | all the modern ddxes use it | 21:35 |
uvos | ie radion and intel and sutch | 21:35 |
freemangordon | which gives something like 15fps on n900 | 21:36 |
uvos | im not saying use it | 21:36 |
freemangordon | with compositing | 21:36 |
uvos | im saying exa is depricated | 21:36 |
freemangordon | well, xorg is deprecated :p | 21:36 |
uvos | well not officaly | 21:36 |
uvos | but yeah in a way | 21:36 |
freemangordon | anyways, what I think: | 21:37 |
freemangordon | 1. crate DDX that operates on gbm backed primary framebuffer | 21:37 |
uvos | btw: im unsure where modesetting has avoidable overhead... | 21:37 |
uvos | i dont see it really | 21:37 |
uvos | takling modesetting only here (ie dri2 noAcell path) | 21:37 |
freemangordon | uvos: I am not sure how easy it would be to fix the rotation issue there | 21:38 |
uvos | ok sure but thats a different problem | 21:38 |
uvos | anyhow | 21:38 |
uvos | do explain | 21:38 |
uvos | [21:37] <freemangordon> anyways, what I think: | 21:38 |
freemangordon | 1.... | 21:38 |
freemangordon | this is more or less what omap driver does ATM | 21:39 |
freemangordon | all the pixmaps shall be backed by BOs and dri3 rendering will be done by memcpy | 21:39 |
freemangordon | as you expect, the performance will be awful | 21:40 |
freemangordon | but.. | 21:40 |
freemangordon | once we have that as a template, I can start optimizing that | 21:40 |
freemangordon | like, using gles for copy operations | 21:41 |
freemangordon | or, using pvr 2d engine to accelerate almost everything | 21:41 |
freemangordon | what I lack here is how to use double-buffering for display framebuffer to avoid tearing | 21:42 |
freemangordon | I lack the theory there | 21:42 |
freemangordon | shall I use single buffer and draw on vsync only? | 21:43 |
freemangordon | or, shall I use double-buffering and track the damage? | 21:43 |
uvos | the latter, the former would be way to slow | 21:43 |
uvos | on d4 you could use the display itself as a frond buffer so to speak | 21:44 |
freemangordon | the same on n900 | 21:44 |
uvos | and just update it whenever your done | 21:44 |
freemangordon | what do you mean "update'? | 21:44 |
freemangordon | ah | 21:44 |
freemangordon | right | 21:44 |
freemangordon | well, yeah, wont; work on n900 | 21:44 |
uvos | yeah i know | 21:44 |
uvos | really in modern x | 21:45 |
uvos | the compositor is expected to take care of this | 21:45 |
uvos | *compositing window manager rather | 21:45 |
freemangordon | yeah | 21:45 |
uvos | so the wm ends up implmenting double buffering | 21:45 |
uvos | essentaly | 21:46 |
freemangordon | but still, blitting on the framebuffer shall be done on vsync only | 21:46 |
freemangordon | on the primary framebuffer | 21:46 |
uvos | yeah the wm dose this by its own volition iiuc | 21:46 |
freemangordon | anyway, have to run now | 21:46 |
freemangordon | ttyl | 21:46 |
uvos | ttyl | 21:46 |
uvos | Wizzup: regarding liblocation i have been using gpsrecorder quite a bit for hikeing/cycleing | 21:54 |
uvos | Wizzup: i did use it in car once too | 21:54 |
uvos | Wizzup: but otherwise have not used it while driveing | 21:54 |
uvos | Wizzup: i also ocasionally use navit for navigation | 21:55 |
uvos | Wizzup: but that uses gpsd directly | 21:55 |
Wizzup | ok | 21:56 |
Wizzup | uvos: so it sounds like h-d might need work too if the wm plays a part in this | 21:57 |
Wizzup | uvos: parazyd: I don't know what to write for fbkeyboard and charge-mode, should I just put it off until we get it in place? | 22:27 |
uvos | Wizzup: fbkeyboard is in place | 22:28 |
Wizzup | in the recovery mode? | 22:28 |
uvos | Wizzup: yes | 22:28 |
Wizzup | ok | 22:28 |
uvos | Wizzup: on bionic | 22:28 |
uvos | Wizzup: (only) | 22:28 |
uvos | Wizzup: since on d4 it makes not sense ofc | 22:28 |
uvos | Wizzup: mainly i took it from its original author | 22:28 |
Wizzup | ok | 22:28 |
uvos | Wizzup: and rewrote its event loop and ported it to libinput | 22:29 |
uvos | Wizzup: so now it can be callibrated | 22:29 |
uvos | Wizzup: (ts wise) | 22:29 |
uvos | it should be used on pp too | 22:30 |
uvos | in some capcacity | 22:30 |
Wizzup | uvos: I see it also on droid3, but I don't think ts events work | 22:30 |
Wizzup | yet another thing to look at I guess | 22:30 |
uvos | Wizzup: ok yeah since you took the bionic image you shal have it | 22:30 |
uvos | Wizzup: it uses libinput to choose a input device | 22:31 |
uvos | Wizzup: if you still have the main touchscreen disabled in libinput | 22:31 |
uvos | Wizzup: it wont work | 22:31 |
uvos | (disabled in favor of ts-buttons) | 22:31 |
uvos | it should be disabled on d3 anyhow | 22:31 |
uvos | regarding pp | 22:31 |
uvos | we just lack a method of entering a special mode for an emergency shell | 22:32 |
Wizzup | d3 does't do ts-buttons atm though | 22:33 |
Wizzup | at least in my setup | 22:33 |
uvos | right so you should remove the udev file that disables the main ts for libinput | 22:33 |
uvos | (i think you did this allready for x but not sure) | 22:33 |
uvos | https://github.com/maemo-leste/leste-config/blob/master/leste-config-mapphone/lib/udev/rules.d/85-input-devices.rules.leste | 22:34 |
uvos | ENV{LIBINPUT_IGNORE_DEVICE}="1" | 22:34 |
uvos | oh and fbkeyboard only works right if the framebuffer is in its natve orientation | 22:35 |
uvos | thats simply a current fbkeyboard limitation | 22:35 |
Wizzup | where did we fork fbkeyboard from? | 22:35 |
uvos | https://github.com/julianwi/fbkeyboard | 22:36 |
uvos | but its a 90% rewrite | 22:36 |
Wizzup | oh, ok | 22:37 |
uvos | Wizzup: https://maedevu.maemo.org/pkgweb/ | 22:54 |
uvos | dosent appear to work for me | 22:54 |
uvos | atm | 22:54 |
Wizzup | it loads for me | 22:55 |
uvos | same here | 22:55 |
uvos | but searching anything dosent work | 22:55 |
Wizzup | parazyd: ^ any idea? | 22:56 |
Wizzup | yeah 504 for me as well | 22:57 |
uvos | Wizzup: new sphone should be in repos shortly | 23:29 |
Wizzup | cool | 23:33 |
Wizzup | uvos: I guess I need to push my changes to the maemo-leste/clown-boot repo | 23:39 |
Wizzup | ok... almost done with the news post | 23:54 |
Wizzup | I guess tomorrow I get to do all the rebuilding and fixing whatever remains to be fixed | 23:54 |
Wizzup | wip: https://github.com/maemo-leste/maemo-leste.github.io/blob/august/content/maemo-leste-update-october-2021.rst | 23:56 |
uvos | power-generic module that allows mce to operate in adsence of dsme. | 23:57 |
uvos | mce is powerd by adsense(tm) now :P | 23:57 |
Wizzup | heh | 23:57 |
Wizzup | fixed | 23:57 |
parazyd | Search seems ok here | 23:59 |
parazyd | https://maedevu.maemo.org/pkgweb/search?q=hildon | 23:59 |
parazyd | Any specific bug? | 23:59 |
uvos | parazyd: just clicking on armhf dosent work for me | 23:59 |
uvos | for instance | 23:59 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!