uvos | tmlind: my mz617 hangs at [ 14.474121] calling emif_driver_init+0x0/0x1000 [emif] @ 491 | 00:08 |
---|---|---|
uvos | we had this before | 00:08 |
uvos | what was the solution again | 00:08 |
uvos | (the hang is right after udev probes the modules | 00:09 |
uvos | ) | 00:09 |
uvos | perhaps unsuprisingly removing emif from dts worked | 00:29 |
uvos | i gues its missconfiguring the dram | 00:29 |
uvos | tmlind: http://uvos.xyz/maserati/iGuesThisIsTCBridgeEeprom.txt | 00:50 |
uvos | your values | 00:50 |
sixwheeledbeast | so i got 4.5hours this time until low battery it said 7% remaining within minutes of that switched off. will do another charge cycle | 01:11 |
Wizzup | yeah we need to work on that :) | 01:13 |
Wizzup | it's not easy, it needs a lot of measurement and careful checking to see what prevents ret/off mode | 01:13 |
Wizzup | also possibly wifi power saving is not on? | 01:14 |
* Wizzup zzz | 01:14 | |
tmlind | uvos: great thanks a lot :) that fixed the lcd, i uploaded the tc358765 bridge restore script to: http://muru.com/linux/d4/mz617-lcd-restore.sh | 05:55 |
tmlind | uvos: no luck getting the bridge driver to talk with the bridge over i2c so far, will debug more when i get a chance | 05:56 |
freemangordon | uvos: https://github.com/libsdl-org/SDL/commit/c12f46b100d22a0e06a64c5b6d1baa3f446d34e6 | 08:05 |
sicelo | sixwheeledbeast: you can see average consumption in /sys/class/power_supply/bq27200-0/*_avg | 08:16 |
freemangordon | we may need https://github.com/libsdl-org/SDL/commit/cf7eef37b045bb3f841e26879fdc6d865c8aaf9a as well | 08:24 |
freemangordon | uvos: with first patch applied it no longer segfaults, but "INFO: Failure to create window (Could not create EGL window surface)" instead | 08:30 |
freemangordon | it seems we need newer libsdl for this to work | 08:43 |
sixwheeledbeast | ah wifi power saving I'll check | 10:03 |
sixwheeledbeast | no it's on maximum which was the default on fremantle. the advance connection dialog seems a bit ropy to use with the touch screen. I have just disabled the connection search to see if that makes a difference, I wouldn't expect it to if I already have a connection but maybe it does. | 10:07 |
Wizzup | sixwheeledbeast: the dialog probably isn't correct | 10:10 |
Wizzup | iw get wlan0 power_save | 10:10 |
Wizzup | is I htink the way to get | 10:10 |
Wizzup | freemangordon: we could pull in a newer one if you and uvos want | 10:10 |
Wizzup | tmlind: great to hear it's restored | 10:14 |
freemangordon | Wizzup: I think it will be needed, but lets wait for uvos | 10:16 |
freemangordon | I as not able to backport those two patches | 10:16 |
sixwheeledbeast | iw isn't there does it need installing? | 10:25 |
Wizzup | yeah | 10:27 |
sixwheeledbeast | it's on - iw dev wlan0 get power_save | 10:43 |
Wizzup | ok | 10:43 |
sicelo | (yes, power_save is the driver default) | 11:15 |
tmlind | Wizzup, freemangordon, uvos: can you guys also disabling dma for the serial port that causes issues on shutdown, i guess that's the console uart3? | 12:59 |
tmlind | &uart { | 12:59 |
tmlind | /delete-property/ dma-names; | 12:59 |
tmlind | }; | 12:59 |
tmlind | i mean &uart3 | 12:59 |
tmlind | and i mean can you guys also try disabling dma :) | 13:00 |
uvos | sure by all means update sdl if it helps with the bug | 13:46 |
freemangordon | I am trying to build the one from sid, no success so far | 13:49 |
freemangordon | uvos: could you try what tmlind asked for (uart dma)? | 13:49 |
uvos | yes i plan to | 13:52 |
Wizzup | we passed the last phase of the eu evaluation | 13:53 |
Wizzup | grant thing | 13:53 |
freemangordon | cool | 13:54 |
freemangordon | Wizzup: do you know what that salsa debian repo url is? | 14:03 |
Wizzup | do you have a link? | 14:06 |
Wizzup | or do you need a link? | 14:06 |
freemangordon | I need a link | 14:32 |
Wizzup | https://salsa.debian.org/sdl-team/libsdl2 ? | 14:35 |
Wizzup | freemangordon: ^^ | 14:35 |
freemangordon | thanks | 14:40 |
freemangordon | Wizzup: how do we fork that in our repo? pull/push? | 14:41 |
Wizzup | we already have a fork of it | 14:42 |
freemangordon | yes, but it is old | 14:42 |
Wizzup | we just need to pull in new source and tag and see if we need to update debian rules | 14:42 |
Wizzup | do you want me to do it? | 14:42 |
freemangordon | yes, we do, to enable gles1 :) | 14:42 |
freemangordon | yes, please | 14:42 |
freemangordon | umm, wait | 14:42 |
freemangordon | lemme first see if it fixes the issue | 14:42 |
freemangordon | Wizzup: ^^^ | 14:42 |
Wizzup | ok | 14:43 |
freemangordon | tests are being build :) | 14:44 |
freemangordon | Wizzup: btw, are d4 batteries still available? | 14:47 |
Wizzup | yeah, I think so | 14:47 |
Wizzup | in fact dreamer has some other ones we bought from aliexpress as well | 14:48 |
Wizzup | I haven't tried them yet | 14:48 |
Wizzup | is yours flat? | 14:48 |
freemangordon | !800 mAh are being reported | 14:49 |
freemangordon | *~800 | 14:49 |
freemangordon | IIUC it should be 3000 | 14:49 |
freemangordon | POWER_SUPPLY_CHARGE_FULL=816965 | 14:50 |
Wizzup | def not 3000 I think | 14:51 |
Wizzup | mine reports like 1049mAh | 14:51 |
freemangordon | well, unfortunately it didn't help: | 14:51 |
freemangordon | INFO: Failure to create window (Can't window GBM/EGL surfaces on window creation.) | 14:51 |
freemangordon | POWER_SUPPLY_CHARGE_FULL_DESIGN=3000000 | 14:52 |
Wizzup | yeah whatever that is, it's probably not mAh | 14:52 |
freemangordon | it should be | 14:53 |
Wizzup | you can read the battery, it will not say 3000mAh | 14:54 |
freemangordon | you mean the sticker on it? | 14:54 |
* Wizzup hcecks | 14:55 | |
Wizzup | 1735mAh i what the stickers say | 14:55 |
Wizzup | https://www.ebay.com/itm/293964657026 | 14:56 |
dreamer | Wizzup: yes! how should I get them to you? :# | 14:57 |
dreamer | although so far not super impressed. maybe I need to "train" the charging circuit and do a full discharge cycle first | 14:57 |
freemangordon | Wizzup: I don't have any capacity info I see | 14:57 |
dreamer | (so far have been recharching it at ~10-15% again. but maybe that's actually still 30-40%) | 14:58 |
freemangordon | but yeah, google saus 1735 | 14:58 |
freemangordon | mtg, ttyl | 14:58 |
Wizzup | dreamer: I will be in .nl on 3-4 march | 14:58 |
Wizzup | dreamer: right we also cap how much we charge the battery | 14:59 |
dreamer | I'm in amst on the 3rd | 14:59 |
Wizzup | that might work then :) | 15:05 |
uvos | freemangordon: its 1740 mah nominal | 15:56 |
uvos | for d4 | 15:56 |
freemangordon | ok | 15:57 |
freemangordon | stll, 800 sounds like I need a new battery :) | 15:57 |
uvos | the "new" batteries are from 2014 at the latest | 15:57 |
uvos | and they have like 1200mah | 15:57 |
uvos | if you want a real new battery you have to wire in some other cell | 15:58 |
uvos | freemangordon: the reported capacity is wrong since i added auto detection because we blacklist the one wire module thats required to read the battery eeprom | 15:58 |
freemangordon | uvos: with latest upstream libsdl, I get "INFO: Failure to create window (Can't window GBM/EGL surfaces on window creation.)" | 15:58 |
uvos | when its blacklisted cpacp-battery uses the maximum that can be in a mapphone | 15:58 |
uvos | (ie the thablets) | 15:58 |
uvos | so that charging works | 15:58 |
freemangordon | uvos: ok, it is not that important (battery) | 15:58 |
freemangordon | see libsdl | 15:58 |
uvos | we blacklist the one wire module | 15:58 |
uvos | because it blocks idle | 15:58 |
uvos | ok well, thats still a bit unfortionate | 15:59 |
freemangordon | it is KMSDRM_CreateSurfaces that fails, will debug that | 16:01 |
freemangordon | uvos: are you sure your test case is correct? | 16:01 |
freemangordon | like, everything is inited correctly? | 16:02 |
uvos | i mean it works in gles mode, its also taken from the init of neverball. | 16:02 |
uvos | and it works everywhere else | 16:02 |
uvos | (ie x11/ wayland) | 16:02 |
freemangordon | see https://pastebin.com/wUx8xLb9 | 16:02 |
freemangordon | just to confirm | 16:03 |
freemangordon | hmm, what is SDL_WINDOW_OPENGL ? | 16:03 |
uvos | accellerated window | 16:03 |
uvos | just means opengl context is required | 16:04 |
uvos | dosent specity if thats desktop gl | 16:04 |
uvos | or gles | 16:04 |
freemangordon | I think we need SDL_WINDOW_FULLSCREEN as well | 16:04 |
uvos | no we dont | 16:04 |
uvos | the backend just makes it fullscreen by default | 16:04 |
freemangordon | because you can't have non-fullscreen kms window | 16:04 |
freemangordon | ok | 16:04 |
uvos | dosent matter | 16:04 |
freemangordon | so, is the code ^^^ ok? | 16:05 |
uvos | yeah sure, i mena i would remove SDL_GL_SetAttribute to make it more minimal, but its fine | 16:06 |
freemangordon | I will | 16:06 |
uvos | SDL_GL_SetAttribute stuff is just so that we get a specifc context we want | 16:06 |
freemangordon | mhm | 16:06 |
uvos | but obv we dont care in this test | 16:06 |
freemangordon | mhm | 16:06 |
freemangordon | umm... | 16:09 |
uvos | hmm? | 16:09 |
freemangordon | sec | 16:09 |
freemangordon | I was thinking about 960x540 | 16:10 |
freemangordon | it should be 540x960 | 16:11 |
freemangordon | uvos: this simply cannot work as you want it to https://salsa.debian.org/sdl-team/libsdl2/-/blob/master/src/video/SDL_video.c#L3485 | 16:33 |
freemangordon | uvos: seems with kms backend you should try GL first and then if it fails try GLES | 16:37 |
freemangordon | this is crazy https://salsa.debian.org/sdl-team/libsdl2/-/blob/master/src/video/kmsdrm/SDL_kmsdrmopengles.c#L44 | 16:39 |
freemangordon | uvos: I don;t know what hacks are there for WL/X11, but by looking in the code of kms backend, there is no way it can auto-detect OGL/GLES | 16:41 |
freemangordon | Wizzup: please pull latest libsdl | 16:44 |
freemangordon | I will push the needed patches for it later on (have to do some shopping first) | 16:45 |
Wizzup | freemangordon: pull, from debian/ | 16:49 |
Wizzup | ? | 16:49 |
freemangordon | pull from salsa, yes | 16:56 |
freemangordon | (15,42,11) Wizzup: we just need to pull in new source and tag and see if we need to update debian rules | 16:57 |
freemangordon | (15,42,15) Wizzup: do you want me to do it? | 16:57 |
freemangordon | no I want | 16:57 |
freemangordon | *now | 16:57 |
Wizzup | ok | 16:59 |
freemangordon | uvos: what do you think: instead of preventing d4 from boot to desktop on < 20% charge, isn't it better do disable one of the cores? | 17:02 |
freemangordon | then it will still boot and be usable, albeit a bit slower | 17:03 |
freemangordon | also, we may limit the working core frequency | 17:03 |
freemangordon | decreasing the power usage even more | 17:03 |
uvos | we dont do that | 17:04 |
uvos | also its 3.4v | 17:04 |
uvos | thats like 5% charge | 17:04 |
uvos | depends on current ofc | 17:05 |
uvos | also i dont think that the core is really the main user here | 17:05 |
uvos | we have sgx and the display etc too | 17:05 |
freemangordon | that was just an idea | 17:06 |
uvos | tmlind: disabeling dma for uart3 dosent help any | 17:06 |
freemangordon | uvos: did you see my notes ^^^ re sdl? | 17:06 |
tmlind | uvos: ok thanks for testing | 17:07 |
uvos | freemangordon: yeah looks like its pretty broken, somehow it works on x11 tho | 17:08 |
freemangordon | well, it is not that bad, you either insist on GLES, or call SDL_CreateWindow() and if it fails do SDL_GL_SetAttribute(SDL_GL_CONTEXT_PROFILE_MASK, SDL_GL_CONTEXT_PROFILE_ES); and call SDL_CreateWindow() again | 17:10 |
uvos | right | 17:10 |
uvos | except it segfaulted before | 17:10 |
uvos | but yeah | 17:10 |
freemangordon | now it will not | 17:10 |
freemangordon | we are going to pull the latest | 17:11 |
uvos | ok | 17:11 |
freemangordon | hopefully later today it will be in the repo | 17:11 |
freemangordon | ttyl, shopping | 17:11 |
uvos | freemangordon: anyhow eglGetPlatformDisplayEXT is still broken | 17:12 |
uvos | freemangordon: as in it gets you a platformDisplay for gbm with swrast if you eglGetDisplayDriverName | 17:13 |
uvos | with afaik no way to get pvr | 17:14 |
uvos | (other than swapping cards around | 17:14 |
uvos | ) | 17:14 |
uvos | but it looks like mesa is just broken | 17:14 |
uvos | becuase its the same problem on amdgpu | 17:14 |
freemangordon | yeah, it is mesa, nothing I can do there | 17:15 |
uvos | if you have time later, please fix the build of the ddk-um libs, so that we can have gles1 again | 17:18 |
uvos | (ie the re'd part) | 17:18 |
uvos | or just retract the re untill it works and reissue a build of the binary package, thats fine too | 17:19 |
freemangordon | I'll fix the build | 17:21 |
freemangordon | Wizzup: may I force-push ddk-um tag with build fix? | 19:59 |
freemangordon | hmm, I guess no, -devel branch is affected too | 20:00 |
freemangordon | scratch that | 20:00 |
freemangordon | umm, some suse guy acked my omapdrm patch, still nothing from Tomi. I hope he is doing well. | 20:29 |
Wizzup | freemangordon: yeah force push tag is ok | 20:44 |
freemangordon | too late, I made another changelog entry :) | 20:44 |
freemangordon | uvos: ddk-um build is fixed, new package is in the repo | 20:45 |
freemangordon | Wizzup: please lmk when you are ready with libsdl | 20:45 |
Wizzup | ok | 20:47 |
Wizzup | freemangordon: what version do you want, 2.0.20? | 20:49 |
freemangordon | yes | 20:49 |
freemangordon | tmlind: could you please do a PR to https://github.com/maemo-leste/sgx-ddk-um for the render node patch? | 20:55 |
uvos | Wizzup: freemangordon: i think there is something very wrong with hildons paring of .desktop files | 23:30 |
uvos | Wizzup: freemangordon: http://uvos.xyz/maserati/sh.desktop | 23:31 |
uvos | this dosent work | 23:31 |
uvos | it just runs sh | 23:31 |
uvos | so the options are discarded in fact it dosent seam to matter, any applicaion gets striped of its options | 23:32 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!