| sicelo | hey ho! that's a beautiful sight :) | 00:22 |
|---|---|---|
| kona | Freemangordon progress!!!!! | 01:08 |
| uvos | great | 11:03 |
| lel | IMbackK closed an issue: https://github.com/maemo-leste/bugtracker/issues/383 (Change geometry file for Droid4 keyboard a tiny bit.) | 12:59 |
| lel | IMbackK milestoned an issue: https://github.com/maemo-leste/bugtracker/issues/524 (PowerVR DDK 1.17 Glamor/Xorg) | 13:04 |
| mighty17[m] | hey tmlind tried out your wlroots branch (https://github.com/tmlind/wlroots/tree/pvr-gles2-v2) and it doesnt seem to be working | 15:51 |
| mighty17[m] | i disabled `HAS_GBM_BO_GET_FD_FOR_PLANE` as well as you told in the issue still i get | 15:51 |
| mighty17[m] | `(phoc:1764): phoc-wlroots-CRITICAL **: 19:12:00.449: [GLES2] OpenGLES 2.0 API generated error code 0x500(GL_INVALID_ENUM)` | 15:51 |
| uvos | it works fine for me | 17:05 |
| uvos | what are you using? | 17:05 |
| uvos | pvr wise | 17:06 |
| mighty17[m] | uvos ddk 1.17 (pmOS) | 18:48 |
| tmlind | mighty17[m]: so i have some pending kernel related omapdrm and pvr changes in my branch, do you have those applied? that's currently https://github.com/tmlind/linux_openpvrsgx/commits/droid4-pending-pvr-omapdrm-v5.15 | 19:37 |
| mighty17[m] | tmlind: nope, i just used the openpvrsgx branch and added my dts as a patch | 19:37 |
| mighty17[m] | do you mind if i use your kernel as a base for my pmOS kernel? | 19:38 |
| tmlind | mighty17[m]: ok may see if my branch has something that helps, most of those are for command mode lcd panels though | 19:38 |
| tmlind | mighty17[m]: no go for it, more testing :) | 19:38 |
| mighty17[m] | thanks! I'll try it out | 19:39 |
| mighty17[m] | also did you build wlroots on device? | 19:39 |
| tmlind | seems the remaining unsolved issue for wlroots is how to deal with finding the right /dev/dri render node and still have the pvr blobs work | 19:40 |
| tmlind | mighty17[m]: yeah i've been building patched mesa, wlroots and sway on my d4.. mesa takes hours | 19:40 |
| tmlind | good thermal and memory test though | 19:41 |
| mighty17[m] | just make a package for the patched mesa, like pmOS does it? | 19:42 |
| mighty17[m] | lol great test xD | 19:42 |
| tmlind | well i've been adding some debug statements.. | 19:42 |
| mighty17[m] | oh, well cant imagine how you bear with it building for hour | 19:43 |
| mighty17[m] | for pmOS i suppose i dont need any changes on wlroots? or `HAS_GBM_BO_GET_FD_FOR_PLANE` needs to be removed? | 19:44 |
| tmlind | i just hack for a little while then let it build, then check later. slow going but no rush, it's a hobby :) | 19:44 |
| mighty17[m] | still must be painful | 19:45 |
| tmlind | yeah i just changed the HAS_GBM_BO_GET_FD_FOR_PLANE line to #if 0 | 19:46 |
| tmlind | yeah it's a pain, i could build it on a much faster machine but been too lazy to switch to that so far | 19:47 |
| mighty17[m] | well you get more time to test :P | 19:48 |
| tmlind | building wlroots and sway is quite tolerable | 19:51 |
| mighty17[m] | my tab gets quite hot so compiling is a no go xD | 19:52 |
| tmlind | Wizzup, uvos: finally figured out why n900 no longer hits off mode during idle while my 37xx based board does :) | 20:23 |
| Wizzup | wow | 20:23 |
| Wizzup | that's great news | 20:23 |
| tmlind | so in omap3430_idle_driver, the nokia measured target_residency is so high current mainline kernels will never pick it | 20:24 |
| tmlind | as a test, i set .target_residency = 46057 also for core ret and core off states | 20:24 |
| tmlind | core_pwrdm (ON),OFF:1047,RET:0,INA:0,ON:1048,RET-LOGIC-OFF:0,RET-MEMBANK1-OFF:0,RET-MEMBANK2-OFF:0 | 20:25 |
| tmlind | not sure what the values should be now.. i can't currently measure power on n900 as it's in a rack | 20:26 |
| uvos | why is it switching to on or off but never ret | 20:26 |
| tmlind | uvos: well right now they core of and ret have the same values, cpuidle picks off as the lower state | 20:27 |
| uvos | ok | 20:27 |
| uvos | anyhow sadly this dosent help us atm since the leste kernel dosent hit ret either. | 20:27 |
| uvos | ddk1.17 things... | 20:27 |
| tmlind | that thing again :) | 20:27 |
| tmlind | roughly i recall the core off latency is up to 150ms on n900 | 20:28 |
| uvos | thats pretty long (comeing from a uc background here) | 20:28 |
| uvos | (might be normal for this kind of processor idk) | 20:29 |
| tmlind | yeah, on 37xx it's way faster some tens of ms i recall, maybe 30 ms i forgot | 20:29 |
| uvos | that sounds more sane | 20:30 |
| tmlind | the exit_latency numbers measured by nokia should be very accurate | 20:30 |
| tmlind | there was some workaround issue on 34xx causing super long wake up latency | 20:30 |
| tmlind | we really should have omap37xx_idle_driver there too with values picked from the nokia n9 kernel if anybody wants to patch that in | 20:31 |
| tmlind | well we should rename omap3_idle_driver to omap37xx_idle_driver, and compare the current omap3_idle_driver to the nokia n9 kernel values | 20:33 |
| tmlind | hmm i wonder if the large target_residency values cause some timer calculation overflow error in cpuidle? | 20:36 |
| _inky | i know someone ported modrana, but can can can i tell you, i miss maemo mapper so much! | 20:37 |
| tmlind | _inky: heh, i mostly use foxtrotgps, have not figured out how to configure custom map repos though.. i just hardcoded some in | 20:39 |
| uvos | navit also works pretty well | 20:40 |
| tmlind | uvos: yup trying with target_residency of 50000 and 60000 makes core ret work | 20:40 |
| tmlind | core_pwrdm (ON),OFF:19,RET:6,INA:0,ON:26,RET-LOGIC-OFF:0,RET-MEMBANK1-OFF:0,RET-MEMBANK2-OFF:0 | 20:40 |
| uvos | great | 20:40 |
| uvos | probubly finishing the linux port on this would be best https://github.com/organicmaps/organicmaps | 20:41 |
| _inky | tmlind: let me see, i didn't know it's in the repos. do you remember maemo-mapper? | 20:41 |
| uvos | it compiles and works on my desktop but its quite rough | 20:41 |
| uvos | (especcaly the build system needs work) | 20:41 |
| uvos | otherwise it looks like an awsome modern osm navigator | 20:42 |
| tmlind | _inky: oh yeah | 20:42 |
| tmlind | hmms so we have new u64 target_residency_ns and old unsigned int target_residency.. i also recall some issue with those from years ago | 20:44 |
| _inky | uvos: never heard of it. i'll try when it'll be in the repos. | 20:50 |
| Wizzup | _inky: maemo mapper? | 20:53 |
| Wizzup | _inky: you could probably port it? | 20:53 |
| Wizzup | _inky: this? http://maemo.org/packages/view/maemo-mapper/ | 20:53 |
| Wizzup | _inky: probably we can do an initial port in a matter of hours | 20:54 |
| Wizzup | you probably could too, all the deps should just be there | 20:55 |
| _inky | i am concerned about gps, i have no idea how the app connects to gps. | 20:58 |
| _inky | but i may try, thank you for encouraging. | 20:58 |
| Wizzup | _inky: we already ported liblocation | 21:03 |
| Wizzup | it is API compatible, so your build should just work | 21:03 |
| _inky | yes, so there is gpsd and liblocation. | 21:03 |
| _inky | aaa, liblocation? i was using it in other my app! | 21:03 |
| _inky | it was a location sharing via xmpp. | 21:04 |
| _inky | http://norayr.am/meridian23/meridian23.gif | 21:04 |
| Wizzup | cool :) | 21:06 |
| uvos | _inky: porting maemo mapper would be neat but if you need something right now that works dose navigation and vector maps navit is in the repos and works fine and is even touch frendly (atho clunky) | 21:19 |
| _inky | uvos, i never was able to use navit on fremantle, don't remember why. but i'll try again. i remember i was in love with maemo-mapper for some reason. | 21:27 |
| uvos | on leste it works fine i use it regularly | 21:27 |
| uvos | you have to configure the xml file a bit to get resonable ui | 21:28 |
| uvos | otherise it works out of the box | 21:28 |
| uvos | you need to donwload maps for it too ofc get them here | 21:30 |
| uvos | http://maps3.navit-project.org/ | 21:30 |
| _inky | thank you so much for doing it. | 21:32 |
| _inky | i am opening the link now. | 21:33 |
| uvos | i dident do anything :P its in the debian/devuan repos | 21:33 |
| _inky | downloadable maps are cool for offline work. (: | 21:33 |
| _inky | oh. | 21:33 |
| uvos | but i think i should publish my config | 21:33 |
| _inky | but what does it use? liblocation? gpsd? | 21:33 |
| uvos | gpsd | 21:33 |
| _inky | it works on droid with gpsd? | 21:33 |
| _inky | cool. | 21:33 |
| uvos | yeah | 21:33 |
| tmlind | so for the n900 off idle issue, i think the bug is kernel is too busy with timers to hardly ever hit the 484ms target residency. somebody should measure longer term idle power consumption though with lower values, i suspect it still saves power below 484ms sleep lengths | 21:35 |
| uvos | tmlind: btw i noticed that ret wakeups on d4 can be reduced in frequency a bit by messing wihth the sched | 21:37 |
| uvos | might help here too | 21:37 |
| uvos | *the sched values | 21:37 |
| tmlind | uvos: yeah cool i saw that | 21:39 |
| uvos | but yeah not way the kernel is ever going to sleep for half a second | 21:39 |
| uvos | at least in its current state | 21:39 |
| tmlind | nope.. n9 often slept for about 8 seconds between debug led blinks :( | 21:39 |
| uvos | might make sense to trace the timers a bit | 21:40 |
| tmlind | yeah | 21:40 |
| tmlind | so n900 should be idle and limited to a single cpuidle state at a time while measuring | 21:42 |
| tmlind | some kind of sample integration is needed to get let's say 10 minute total power consumption | 21:43 |
| tmlind | well i guess those cheapo usb charge monitor dongles will show the cumulative power consumption | 21:44 |
| uvos | the n900 battery chip should be enough no | 21:44 |
| uvos | it has a charge counter | 21:44 |
| tmlind | oh ok yeah that should do | 21:44 |
| Wizzup | I have a n900 on lab psu, so could use that too for measuring | 21:47 |
| tmlind | Wizzup: ok, not seeing a charge counter after bq2415x_charger, don't have a battery connected if that matters | 21:48 |
| tmlind | hmm or was it the twl madc that's needed? | 21:49 |
| tmlind | rx51_battery module using twl madc i guess | 21:51 |
| _inky | can you remind me, so we supposed to have maemo/beowulf branch, and the builder bilds code from that branch? | 21:51 |
| _inky | or from master? | 21:51 |
| Wizzup | inky_: tag master, maemo/beowulf for debian/ dir | 21:54 |
| _inky | so maemo/beowulf branch contains debian dir, ok, i understand that. | 21:55 |
| _inky | but tag should be done in master? | 21:55 |
| _inky | ok. | 21:55 |
| _inky | and the builder will take debian dir from maemo branch, and tag from master? | 21:55 |
| _inky | ok. | 21:55 |
| tmlind | not seeing a charge counter with cat /sys/class/power_supply/*/uevent, should not be hard to add it based on the madc data. although reading might affect the idle states.. | 21:58 |
| uvos | someone had a userspace script to dump that | 21:58 |
| uvos | i have been trying to find it the last 5min :P | 21:58 |
| tmlind | i don't think madc has a charge counter, but maybe it's sampling time can be set long enough to not require much polling | 21:58 |
| uvos | ok i was pretty sure it did | 21:59 |
| uvos | but maybe the script was simply calculating behind my back | 21:59 |
| tmlind | oh ok | 21:59 |
| tmlind | i sort of recall madc being just a few raw adc inputs | 21:59 |
| uvos | i might be wrong | 22:00 |
| uvos | im just basing this off me running this script and viewing its output once | 22:00 |
| tmlind | the bq chip might have one though | 22:00 |
| uvos | the bq chip has one internaly for sure | 22:00 |
| uvos | but looking at the register datasheet you cant read it drectly | 22:01 |
| tmlind | so maybe the script you're talking about is using i2cget on the bq2415x chip? | 22:01 |
| tmlind | ah ok | 22:01 |
| uvos | ah its bq27200 | 22:03 |
| tmlind | oh | 22:03 |
| uvos | there is another bq27200 fule guage | 22:03 |
| tmlind | right, i forgot :0 | 22:03 |
| tmlind | :) | 22:03 |
| uvos | heh :P | 22:03 |
| uvos | yeah the script was just i2cget'ing from that | 22:03 |
| uvos | for sure | 22:03 |
| uvos | you wil lwant to read 0x15 and 0x14 | 22:06 |
| tmlind | looks like after modprobe of bq27xxx_battery_i2c i see power average but not charge counter, weird that it's not implemented if it can be just read | 22:07 |
| tmlind | here's another good project for somebody :) | 22:07 |
| uvos | er 0x21 -0x20 | 22:08 |
| uvos | its not totaly directly | 22:08 |
| tmlind | uvos: any idea what units the register shows? | 22:09 |
| uvos | Available Energy in mWh | 22:09 |
| tmlind | ok | 22:09 |
| uvos | effectively a charge counter | 22:09 |
| uvos | (just in reverse) | 22:09 |
| tmlind | yeah ok | 22:09 |
| uvos | you also have to devide by the shunt | 22:10 |
| uvos | its external to the chip | 22:10 |
| uvos | so you need to know what value nokia used | 22:11 |
| uvos | but its irellivant if you just want to compear values | 22:11 |
| uvos | between to operating states to figure out what uses more power | 22:11 |
| tmlind | hmm yeah, that should be a fun project to award to somebody as writing linux kernel drivers project :) | 22:12 |
| tmlind | zzz time here now, ttyl | 22:13 |
| freemangordon | kona: could you please package http://safr.mrrau.eu/maemo/maemo.muarf.org/apt-mirror/mirror/repository.maemo.org/pool/maemo5.0/nokia-binaries/4bc37c7c77ebe90177c050b805a8dc79/nokia-binaries/m/modest-providers-data/modest-providers-data_1.3-2010.06-1+0m5_i386.deb for leste? | 22:21 |
| freemangordon | modest needs this | 22:21 |
| _inky | do you remember what is gbp.conf? is it necessary in debian dir? | 22:25 |
| _inky | aaand can you remind me how was i building the XB-Maemo-Icon-26? | 22:28 |
| _inky | ah ok found | 22:28 |
| _inky | the icon part, not the gbp.conf part | 22:28 |
| freemangordon | gbp is needed by CI | 22:34 |
| freemangordon | like jenkins | 22:34 |
| _inky | ok! | 22:34 |
| freemangordon | gbp.conf that is | 22:34 |
| _inky | thanks! | 22:34 |
| freemangordon | otherwise gbp is git-build-package, IIRC | 22:34 |
| _inky | woow | 22:34 |
| sicelo | tmlind: thanks for these great news of 2021 (N900 off mode) :) | 22:35 |
| freemangordon | hmm, what did I miss? | 22:38 |
| uvos | nothing off mode dosent work because the kenrel has way to many fireing timers | 22:39 |
| uvos | you can make it enter the mode anyhow | 22:39 |
| uvos | but it wont under normal circumstances with the current latency figures | 22:40 |
| freemangordon | I see | 22:40 |
| freemangordon | (23,39,11) phoenix-bot: Project modest-binaries ยป architecture=amd64,label=amd64 build #1: SUCCESS in 7 min 59 sec: https://phoenix.maemo.org/job/modest-binaries/architecture=amd64,label=amd64/1/ | 22:41 |
| freemangordon | :) | 22:41 |
| uvos | great | 22:41 |
| uvos | what works? | 22:41 |
| freemangordon | it doens;t have abook integration enabled still, but it will help me a lot with abook REing | 22:41 |
| freemangordon | everything else | 22:41 |
| uvos | neat | 22:42 |
| freemangordon | well, I set-up my gmail accoun and it works fine | 22:42 |
| freemangordon | didn;t try to send mail though | 22:42 |
| freemangordon | uvos: any clue why hildon-desktop does not update applicatyions in the launcher when new one is installed? is that common issue? | 22:43 |
| uvos | hmm | 22:44 |
| uvos | works for me | 22:44 |
| _inky | if i remove 'end' at the end of uuencoded image, the end of the file looks like: | 22:44 |
| uvos | i installed ltris just yesterday and it worked fine | 22:44 |
| freemangordon | I had to restart hildon-desktop in order modest icon to appear | 22:44 |
| uvos | hmm | 22:44 |
| uvos | strange | 22:44 |
| freemangordon | mhm | 22:44 |
| uvos | i gues maybe its broken outside the debian menu? | 22:45 |
| freemangordon | yeah, modest is outside of debian | 22:45 |
| freemangordon | so maybe it is broken like that | 22:45 |
| sicelo | i'm a bit lost though - what's missing in the bq27200 driver? | 22:45 |
| freemangordon | ok, will play a bit with it when I have some spare time | 22:45 |
| _inky | sorry for interrupting you, but the last two lines look like: | 22:46 |
| _inky | /-,<L`````$E%3D2N0F"" | 22:46 |
| _inky | ` | 22:46 |
| uvos | let me install modest | 22:46 |
| _inky | only one symbol of backtick on the last line. | 22:46 |
| freemangordon | wait few more minutes | 22:46 |
| _inky | should i remove it? | 22:46 |
| freemangordon | uvos: armhf/arm64 are still building | 22:46 |
| sicelo | freemangordon: one thing that's really broken is desktop widgets. possibly something that must be fixed as part of Gio port | 22:46 |
| uvos | ok ill install something else then | 22:46 |
| freemangordon | yeah, must have broken it when I did the port. or, was it me? I don't really remember | 22:47 |
| uvos | freemangordon: i just installed dorian | 22:47 |
| uvos | and it worked fine | 22:47 |
| uvos | looks like its just broken on your machine | 22:48 |
| freemangordon | in the VM? | 22:48 |
| uvos | no on d4 | 22:48 |
| uvos | true might be 64 bit issue or something | 22:48 |
| freemangordon | mhm, will look at it | 22:48 |
| _inky | ok i need a repo already | 23:36 |
| _inky | it builds. | 23:36 |
| _inky | that's a gemini client that uses sdl. | 23:36 |
| _inky | i used it on droid for a couple of weeks, but now i can commit a package. | 23:37 |
| _inky | i also built an emulator for 8bit computer: oric atmos. | 23:37 |
| _inky | use it on droid. woud like to package too. | 23:37 |
| _inky | but wait. i am not sure how to deal with submodules. | 23:47 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!
