T42 | <Zero %lastname%> (Photo, 734x572) https://irc.thaodan.de/.imgstore/Go0YIhnwI6.png 币圈电报微信群炸群、电报群指定群私聊、Twitter/Facebook点赞加粉等,需要的咨询 @he201501 | 03:11 |
---|---|---|
T42 | Ines Etter %lastname% was added by: Ines Etter %lastname% | 08:25 |
szopin | hello | 09:09 |
szopin | would anyone know what changes need to be made in gitlab CI from 3.2.1 to 3.4 to get a working image? (specifically for cosmocom-ci, as just changing @RELEASE@ results in a non-working image, would newer docker image be needed? it's using coderus' 3.1.0.12) | 09:11 |
szopin | I'm guessing 'user --name nemo --groups audio,input,video --password nemo' in .ks file needs changing to 'defaultuser'? | 09:17 |
T42 | <elros34> iirc this line was removed at some point. Compare ks with some more up to date device | 09:25 |
szopin | latest is enchilada but it only has 3.3 | 09:27 |
szopin | oh wait... that's created date | 09:28 |
szopin | nope... last update is vince-ci but also on 3.3 | 09:29 |
szopin | doesn't seem to be any ci setup for 3.4 build | 09:29 |
szopin | sadly the cosmo port doesn't update to either 3.3 or 3.4, ends up rebooting around 70% and no longer boots | 09:30 |
T42 | <elros34> yeah we don't have 3.4.0.x on obs so probbable many are still at 3.3.0.x. For sure you need to remove replace first su -c "/usr/bin/oneshot --mic" with /usr/bin/oneshot --mic | 09:31 |
T42 | <elros34> ah but did anybody update adaptation repos? | 09:31 |
kalube | what's the status with OBS? I got put off building 3.4 when they announced they were killing it | 09:40 |
szopin | elros34: thx, will try with that change | 09:48 |
szopin | or does the above mean trying to build 3.4 right away will not work and 3.3 is where it's at? | 09:57 |
szopin | oh one more thing, as all sensors seem dead on the 3.2.1 image no gyro/accel etc, all giving 0s in csd, is that something that can be troubleshot from within the port or does it need rebuilding the community-adaptation packages? | 10:06 |
szopin | (I was trying to follow hadk guide, but there is no android base for cosmo available, only kernel, so got stuck pretty early on on local manifests and repo setup, though most of it seems taken care of by NotKit) | 10:08 |
szopin | (even then, the guide says it should work out of the box :/) | 10:08 |
szopin89 | btw who is working on Volla Phone port? NotKit said it has very similar SoC, so could maybe reuse its android base to get the proper setup for things outside of the CI | 10:19 |
T42 | <edp_17> szopin: To make sensors working on 3.4 I have to downgrade sensorfw to 0.11.9. | 10:28 |
szopin | It was not working on 3.2.1... the initial NotKit release, about to try the 3.4 artifact with the su -c removed, will try to downgrade if it boots | 10:29 |
T42 | <elros34> I am afraid you need to upgrade adaptation repos to get it fully working with 3.4.0 otherwise some thinks might not work even if it boots like camera, videoplayback. Removing "su -c" is required to avoid some issues but will not help for not booting image. You would need to telnet to device and grab journalctl to see what fails | 10:32 |
szopin | can I grab journalctl from gemian? | 10:32 |
szopin | yeah must be... everything is in /.stowaways/sfos/ (except everything that's written to /data as it's also gemian rootfs) | 10:34 |
T42 | <elros34> I guess if you chroot to sailfish then it should be possible. But first you would need to enable persistent logging in sfos https://github.com/mer-hybris/hadk-faq/blob/master/README.rst#persistent-journalctl | 10:34 |
szopin | ok, will update that before the boot then | 10:35 |
szopin | alright, here goes nothing | 10:37 |
*** szopin89 is now known as szopin | 10:37 | |
szopin | at least keyboard in gemian still works :) | 10:39 |
szopin | hmm... var/log/journal is empty | 10:40 |
szopin | nope... it's there | 10:41 |
szopin | wrong dir | 10:41 |
szopin | 4 .journal files, but some binary format... | 10:43 |
T42 | <elros34> yes you need to use journalctl to read them | 10:43 |
szopin | ok... I can just chroot into /.stowaways, great stuff | 10:44 |
szopin100 | probably related to not working sensors: Nov 05 10:38:06 localhost kernel: (6)[1:swapper/0]i2c i2c-1: of_i2c: modalias failure on /i2c@11011000/gsensor@6aNov 05 10:38:06 localhost kernel: (6)[1:swapper/0]i2c i2c-1: of_i2c: modalias failure on /i2c@11011000/gyro@69 | 11:01 |
szopin100 | I think this is where it fails: ov 05 10:38:08 localhost (5)[1072:droid-hal-init]droid-hal-init: /init.rc: 1: Could not import file 'vendor.mediatek.hardware.power@2.0-init.rc': No such file or directoryNov 05 10:38:08 localhost (5)[1072:droid-hal-init]droid-hal-init: Parsing file /vendor/etc/init/vendor.mediatek.hardware.pq@2.2-service.rc...Nov 05 | 11:17 |
szopin100 | 10:38:08 localhost (5)[1072:droid-hal-init]droid-hal-init: Parsing file /vendor/etc/init/vndservicemanager.rc...Nov 05 10:38:08 localhost systemd[1]: dsme.service: Unit entered failed state. | 11:17 |
mal | show full log | 11:22 |
szopin100 | where best to upload it? | 11:22 |
szopin100 | pastebin? | 11:23 |
mal | yes | 11:29 |
szopin100 | too big for pastebin, but this works https://privatebin.net/?7bbb00f111c4a0d2#8cWsqcbbTn9P86RvRAfdAEBRLw2CkjoXMhcZj3JBX7BQ | 11:31 |
mal | so are sensors the only issue or something else also? | 11:40 |
szopin100 | sensors and BT, but with the 3.4 it's just not booting | 11:40 |
mal | also get output of journalctl -b --no-pager | 11:41 |
szopin100 | updating from terminal also results in unbootable state whether to 3.3 or 3.4, and same with artifact from gitlab | 11:41 |
szopin100 | one sec | 11:41 |
szopin100 | Failed to get boot id: No such file or directory(I'm chrooting into /.stowaways/sfos as no recovery/telnet option on cosmo) | 11:43 |
T42 | <elros34> just for future use journalctl -b-1 to get logs only from previous boot not all | 11:43 |
mal | yeah | 11:43 |
mal | good point | 11:43 |
szopin100 | this is -b-1: https://privatebin.net/?863447e0ee4864ab#DT9JMXVLNih8wYdG5fFLwTcc2TNy6CepyAvBM86j71FH | 11:45 |
szopin100 | and --no-pager from last boot https://privatebin.net/?da53ad067c5d9ffd#DFRbLWx8YGhjod794dEzkVfVrhiG22dUHcUqgfPgdDxN | 11:46 |
szopin100 | seems identical | 11:48 |
szopin100 | I guess > file does the nopager | 11:48 |
T42 | <elros34> systemd-* and dsme fails with: "Failed at step NETWORK spawning". THis si probably the first issue to solve | 11:49 |
rinigus | kalube: I saw by accident last night 3.4... target at OBS. cannot comment on whether it is ready or not. maybe someone from Jolla can provide bg | 11:55 |
kalube | rinigus: cheers, hopefully I can get proper A/B support in sailfish and make some headway :D | 11:56 |
T42 | <elros34> @szopin100: what "zcat /proc/config.gz | grep CONFIG_NET_NS" shows? | 12:02 |
mal | which device and kernel is that? | 12:03 |
mal | ah, 4.4 kernel | 12:03 |
szopin100 | no such file or directory | 12:04 |
szopin100 | on gemian it's set to y | 12:05 |
T42 | <elros34> you need to telnet to device get it from there or read it from defconfig from kernel sources if they are up to date | 12:06 |
szopin | can't telnet to it as it has no recovery mode | 12:06 |
szopin | or at least unaware how to access it | 12:06 |
szopin | tried volume up/down on booting into sfos but got nowhere | 12:07 |
szopin | could try restoring the working 3.2.1 should populate /proc then | 12:07 |
T42 | <elros34> you don't need recovery or any buttons for telnet, when it's booting you should see in dmesg that usb interface is setup, then "telnet 192.168.2.15 2323" | 12:07 |
mal | what kind of build is that? | 12:08 |
szopin | no idea, I need to boot into gemian to setup directory /.stowaways/sailfishos/ with all the rootfs of sfos, then boot from a different partition into it | 12:09 |
szopin | maybe there is some way to trigger recovery menu, haven't found it yet | 12:09 |
szopin | yeah... would need to dd the hybris-recovery img | 12:10 |
T42 | <elros34> like I said you don't need recovery /init-debug script in sfos rootfs setup telnet for you when you boot regular sailfish. | 12:10 |
szopin | I don't think there is one in the artifacts from CI | 12:10 |
szopin | it reboots in seconds when booting 3.4 | 12:11 |
szopin | can try to get it from the wroking 3.2.1 | 12:11 |
T42 | <elros34> ok so for now maybe try as a workaround to remove PrivateNetwork=true from all services which fails with "Failed at step NETWORK spawning" to see if that helps | 12:12 |
T42 | <elros34> I have check https://build.merproject.org/package/binary/nemo:devel:hw:planet:cosmocom/droid-hal-cosmocom?arch=armv8el&filename=droid-hal-cosmocom-0.0.6-202002102318.armv7hl.rpm&repository=sailfish_latest_armv7hl and it has /init-debug like it should | 12:14 |
szopin | # it or set to false? | 12:15 |
T42 | <elros34> probably it doesn't matter | 12:15 |
szopin | k | 12:15 |
szopin | works! | 12:18 |
szopin | wifi connects but no connectivity, but build shows 3.4.0.24 | 12:22 |
szopin | maybe will change them one at a time | 12:22 |
szopin | hmmm orientation is also locked, components app doesn't seem to change it | 12:23 |
mal | is sensorfw running? | 12:27 |
szopin | hmm two of the PrivateNetwork settings resetted themselves back to true, leaving dsme.service with it commented for now | 12:36 |
szopin | I think I broke the Page.qml when trying to change it landscape inverted :/ | 12:37 |
szopin | that was weird, display was just a line, then it seemed like a bootloop as got chinese 'Goodbye(or welcome?)' black screen and only vibra when pressing power key... but it's all fine lol | 12:39 |
szopin | weird, the Page.qml seems fine, maybe it's the second boot issue | 12:42 |
szopin | couldn't save developer password in settings with all the PrivateNetwork commented out | 12:43 |
mal | first thing to fix is that dsme | 12:43 |
szopin | yeah, with it commented out and Orentation.All back in getting the tutorial again | 12:43 |
szopin | this time developer password got saved and sensorfwd is running | 12:46 |
szopin | fixing landscape from within worked, network still not working | 12:51 |
szopin | (connects to wifi but nothing opens in either browser or setting up account) | 12:51 |
szopin | should I try to edit the PrivateNetwork back in and restart dsme? | 12:54 |
szopin | that causes a restart | 13:02 |
spiiroin | szopin: does the dsme process start at all? or does it fail already when systemd is settings up things? | 13:05 |
szopin | hard to tell, tried systemctl after uncommenting the PrivateNetwork bit, it told me to daemon-reload(?)... 10-15 seconds later it showed lockscreen and then blackscreen/reboot | 13:06 |
szopin | can try to have top open in another window and see | 13:07 |
szopin | systemctl restart dsme* | 13:07 |
spiiroin | szopin: until things stablize, you might want comment out "StartLimitAction=reboot" from dsme.service | 13:11 |
szopin | Ok, with that commented it just goes into blackscreen and need to power off | 13:15 |
spiiroin | top is probably not that useful if the process dies immediately/is not started at all.. | 13:17 |
szopin | seen only dsme-server process running | 13:17 |
spiiroin | there ought to be two: dsme-wdd and dsme-server. former is watchdog kicking parent process, and latter the outwards visible actual service | 13:19 |
spiiroin | about the first thing that should happen on dsme startup: parent writes that "DSME 0.81.4 starting up" to stderr. journald should get it, but that was not in that paste | 13:20 |
szopin | I can access now the journal from within 3.4, one sec | 13:21 |
szopin | Starting DSME/Failed to start/Stopped/Failed to start/Stopped/Starting DSME/Failed/Stopped/Starting/Failed in this order (grep on DSME) | 13:24 |
szopin | If you want full logs will be easier to get from gemian as fingerterm is sideways and select/copy will be a pain if not impossible | 13:25 |
szopin | Let me pull the full log | 13:26 |
spiiroin | szopin: I think all of those are progress reports logged from systemd.. so nothing from dsme process itself? | 13:26 |
szopin | in journal no, there was no DMSE 0.8... | 13:26 |
spiiroin | but but but... there were other units that were failing with that PrivateNetwork? was there some unit that has it and works? i.e. might this be some sort of systemd vs kernel version issue or stte? | 13:27 |
szopin | there is 5 or six .service files with PrivateNetwork, 2 or 3 of them reset themselves to uncommented after booting into sfos, commenting dsme.service is enough to boot in without reboot, let me check that log again | 13:29 |
spiiroin | reset to uncommented? as in, your manual edits disappear on reboot? | 13:35 |
* spiiroin wonders if fs sync happens ... | 13:35 | |
szopin | yes, for bluebinder and that obexdservice I think, the # before PrivateNetwork was gone | 13:36 |
szopin | even though originally I commented them all out | 13:36 |
T42 | <elros34> @szopin so if it doesn't reboot now then check that /proc/config.gz | 13:43 |
szopin | no such setting | 13:45 |
T42 | <elros34> no CONFIG_NET_NS=y? then this needs to be enabled in defconfig | 13:47 |
szopin | hmm... that's not something I can do on device? | 13:47 |
T42 | <elros34> no, you need to build kernel | 13:48 |
szopin | is this needed on 3.4? can check if the 3.2.1 has it, but don't think the CI rebuilds the kernel | 13:49 |
T42 | <elros34> not sure as I have it enabled for a long time | 13:50 |
szopin | would this be the android kernel or the cosmo-linux kernel? is it built on obs with the community-adaptation part? | 13:50 |
T42 | <elros34> ci only create image, it doesn't build anything | 13:51 |
szopin | will have to reach out to NotKit to see if he has that setup for sfos still to just add one option | 13:51 |
T42 | <elros34> usually we build kernel in pc then package it and upload rpm to obs | 13:52 |
szopin | btw what options are needed for flatpaks? if notkit agrees to rebuild it would be good to include those to for angelfish | 14:04 |
szopin | only PR_SET_NO_NEW_PRIVS/ | 14:06 |
szopin | ok kernel is in droid-hal and there is also droid-hal-cosmocom-img-recovery-0.0.6-202002102318.armv7hl.rpm I guess this would be needed for recovery mode? | 14:11 |
szopin | I need to get that whole setup myself, but NotKit supposedly used some private planet mediatek BSP, the HAL guide supports only lineageOS now, but halfway through goes to CyanogenMod as it's on nexus | 14:14 |
szopin | and there is neither lineage nor Cm for cosmo (maybe pure AOSP Pie will work?) | 14:14 |
T42 | <Simon %lastname%> Does anybody know where I can find xulrunner-qt5-dev? I'm trying to compile qtmozembed but it doesn't have this dependency | 14:25 |
szopin | piggz: are you using public android base for yggdrasil or some private mediatek bsp like cosmo? Volla is using the same Helio p23 so ws wondering if maybe the same base would work to setup hadk properly? | 14:27 |
szopin | wouldn't have to bother NotKit all the time as it seems BT also needs another defconfig change :/ | 14:28 |
T42 | <NotKit> I have public tree based on yggdrasil. Need some time to clean up and upload | 14:29 |
szopin | Oh wow you're here too? | 14:29 |
szopin | oops | 14:29 |
szopin | I guess T42 is telegram bridge? | 14:30 |
T42 | <NotKit> yes | 14:30 |
Thaodan | T42 because 42 is the answer to everything | 14:32 |
szopin | from the HADK daq CONFIG_BT_HCIVHCI=y is supposedly needed for BT on 8.1, though cosmo is 9 so not sure | 14:32 |
szopin | :) | 14:32 |
szopin | faq* | 14:32 |
Thaodan | https://www.dictionary.com/e/slang/42/ | 14:32 |
szopin | I guess IRC is old fashioned and telegram has all the answers now | 14:33 |
szopin | wait for discourse bridge | 14:33 |
Thaodan | discord? For discord there is bitlbee discord. I just needed a name for the telegram bridge and thught 42 would fit. | 14:34 |
szopin | 6cord too | 14:35 |
szopin | I think the only client on sfos | 14:35 |
szopin | Thaodan: there should be one XMPP like bridge that covers telegram, matrix, discord, qq and AOL and whatsapp | 14:37 |
szopin | instead of one protocol covers all, one bridge to rule them all | 14:38 |
Thaodan | There is you can use xmpp transports for that | 14:38 |
szopin | kek | 14:38 |
Thaodan | bitlbee is like that just for IRC | 14:38 |
szopin | cosmo guys are fighting with matrix bridge alone for the last 72hrs and it's not looking good | 14:38 |
szopin | (that's telegram <-> matrix, two newish and popular and supported frameworks, add all that legacy support to that) | 14:40 |
samsep10l | hello people. i am not sure if i can get answer here. but do anyone know what to do with this? | 14:46 |
samsep10l | halium/out/soong/host/linux-x86/bin/d8', needed by 'halium/out/soong/.intermediates/frameworks/base/packages/EasterEgg/EasterEgg/android_common/dex/EasterEgg.jar' | 14:47 |
samsep10l | i looked at halium page, but it didn't give me solution | 14:48 |
szopin | not sure if halium is used in any sfos distro, ubports uses it afaik | 14:50 |
samsep10l | i tried before, failed. might be a bug. | 14:52 |
szopin | I mean there are people here who port all kinds of stuff, so not a bad place to ask, just that sfos doesn't use halium | 14:54 |
szopin | (then again ubports documentation is supposedly non-existant, so might as well get better answers elsewhere) | 14:55 |
szopin | try /join #halium 364 people there, here out of a 100 only a percentage will have experience | 15:02 |
szopin | that 364 is weird though as they don't have a bridge, but instead NickTelegram, so could be doubles | 15:02 |
szopin | yeah the bridging solutions seem to be non-standardised | 15:03 |
rinigus | kalube: what do you expect from A/B support? for me it mainly looks as an option to expand LVM into those "unused" volumes | 16:26 |
rinigus | jolla devs: any update on OBS 3.4. target? | 16:26 |
kalube | rinigus currently you have to hard code partitions, meaning in order to dual boot with android you are forced to install sailfish on a particular slot which is kinda a pain. ubports and postmarketos can both be installed to either slot making it way easier to dual boot | 16:29 |
pketo | rinigus: sorry, no other news about 3.4 target than it was tried but there was weird issues, and then other stuff came up and nobody has had time yet to look at it again | 16:37 |
rinigus | pketo: thank you very much for update! and there is nothing to be sorry about - it is all understandable. | 16:38 |
rinigus | kalube: I see. haven't considered dual boot myself. | 16:39 |
T42 | <elros34> kalube so do not hardcode partitions. Amend mount units before systemd start. I use such a hack in init script to support 2 similar devices with different system partition number | 16:42 |
kalube | @elros seems to be the right approach | 16:50 |
szopin | pketo: does it mean trying to build 3.4 images is doomed to fail? | 17:01 |
szopin | as in: there no ported device that starts as 3.4 and you can only get there through OTA | 17:04 |
szopin | don't want to waste NotKit's time if that's so | 17:04 |
mal | building 3.4.0 is possible locally, just OBS is the problem | 18:16 |
T42 | <Simon %lastname%> Where can I find "nemo-test-tools"? | 20:34 |
mal | where do you need those? not even sure what those are | 20:34 |
T42 | <Simon %lastname%> it's a dependency of qtmozembed-tests | 21:24 |
T42 | <Simon %lastname%> I thought maybe one of those tests helps narrowing down why the browser stays white on pinephone | 21:25 |
mal | probably now | 21:27 |
mal | *probably not | 21:28 |
mal | we know roughly where the issue is happening but not why | 21:28 |
T42 | <Simon %lastname%> is that documented anywhere accessible? | 21:31 |
T42 | <Simon %lastname%> or how can I help debugging this? It's the last big missing piece as far as I can see | 21:33 |
mal | just some discussion I had with piggz about the issue some time ago | 21:37 |
Thaodan | You can use the channel logs to search for that | 21:38 |
mal | I think it was in a private chat | 21:39 |
mal | maybe I should try to find the links from my own irc logs | 21:40 |
T42 | <Simon %lastname%> this might not be the most appropriate place but I put an issue here some while ago: https://github.com/sailfish-on-dontbeevil/issues-dontbeevil/issues/15 | 21:42 |
Thaodan | Please set a user name and attach some logs to the issue | 21:46 |
T42 | <Simon %lastname%> How do I set a user name? | 21:50 |
Thaodan | Just search for telegram username | 22:06 |
T42 | <simonschmeisser> thanks, done, I also added some minimal logs to the issue but I can add more if you guide me | 22:14 |
Thaodan | I'm not sure what the problem is but you can look in to the documenation of the package for more debug output | 22:51 |
mal | the issue is not very simple | 23:01 |
Thaodan | hm ok | 23:23 |
Thaodan | Is it related to not having gmp-droid? | 23:24 |
mal | no | 23:33 |
mal | gmp-droid is optional but nice to have | 23:33 |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!