| Mister_Magister | oh great torch doesn't work on 4 of my devices out of 7 | 00:51 |
|---|---|---|
| Mister_Magister | because of the change to droidmedia-flashlight | 00:52 |
| T42 | <damnbruh2004> hello hello, reminding all of you that im trying to port onto fortuna3g with a hybris 18.1 base | 01:12 |
| T42 | <damnbruh2004> i am currently investigating a droid-hal-init failure | 01:12 |
| T42 | <damnbruh2004> it turns out the device is running system-as-root and i edited fstab to mount into / instead of /system | 01:12 |
| T42 | <damnbruh2004> droid-hal-init started giving signs of life, but i had to also create a symlink for /vendor -> /system/vendor | 01:12 |
| T42 | <damnbruh2004> and a symlink from /apex into /system/apex | 01:12 |
| T42 | <damnbruh2004> ~ # ls -l /system/bin/linker* | 01:12 |
| T42 | <damnbruh2004> lrwxr-xr-x 1 root shell 36 Jan 1 2009 */system/bin/linker* -> */apex/com.android.runtime/bin/linker* | 01:12 |
| T42 | <damnbruh2004> lrwxr-xr-x 1 root shell 36 Jan 1 2009 */system/bin/linker_asan* -> */apex/com.android.runtime/bin/linker* | 01:12 |
| T42 | <damnbruh2004> -rwxr-xr-x 1 root shell 694032 Jan 1 2009 */system/bin/linkerconfig* | 01:12 |
| T42 | <damnbruh2004> it then complained about boringssl self test failing and rebooting, so as irc logs suggested, if commented out reboot on failure defs in the rc files. and now it passes the boringssl checks but reboots with a message `bpfloader-failed` | 01:14 |
| Mister_Magister | system as root simply means you have /system in that partition no? so you mount it as system_root like very other sane person | 01:15 |
| Mister_Magister | dont' try to reinvent the wheel xd | 01:16 |
| T42 | <damnbruh2004> i did this to the fstab. before that there was nothing besides data mounting in early init | 01:18 |
| T42 | <damnbruh2004> https://github.com/fortuna-for-smoothie/android_device_samsung_msm8916-common/commit/505337a990947ded3fc79180defea64cef09cd39 | 01:18 |
| T42 | <damnbruh2004> this is so the hybris-hal scripts can create appropriate .mount files, as per hadk hot | 01:19 |
| Mister_Magister | mal: pages/about/components/MemInfoItem.qml adds up free space from two lists, PartitionModel.User and PartitionModel.System. miami has everything on one partition (/dev/sde26), and that gets classed as Mass (type 4), so it's in neither list. | 01:33 |
| Mister_Magister | so it is because it's not LVM setup | 01:33 |
| Mister_Magister | and on g2/op6 there's no dm-0 because it doesn't have the dmsetup so Storage comes up at 0GB | 01:38 |
| Mister_Magister | and 0GB available for same reason as miami | 01:38 |
| Mister_Magister | feels like jolla could give it bit more thought | 01:38 |
| Mister_Magister | it won't work anywhere but official ports and ports with LVM | 01:39 |
| Mister_Magister | fix for droidmedia-flashlight is supposed to be "update droidmedia dumbass" | 01:41 |
| Mister_Magister | my droidmedia on g2 is from 2024 xddd | 01:41 |
| Mister_Magister | ah droidmedia-flashlight won't work on 5.1 android | 02:08 |
| Mister_Magister | the interface is just not there | 02:08 |
| Mister_Magister | i wonder if i can fall back to the old one | 02:08 |
| Mister_Magister | now why would patterns-sailfish-device-configuration-titan contain it | 02:09 |
| Mister_Magister | Correct me if i'm wrong but droidmedia-flashlight is absolutely required and you cannot go back to sysfs way and on 5.2 flashlight is broken on at least 4 devices | 02:15 |
| Mister_Magister | let me repeat myself | 02:17 |
| Mister_Magister | [02:38] <Mister_Magister> feels like jolla could give it bit more thought | 02:17 |
| Mister_Magister | like at least provide old provider | 02:18 |
| Mister_Magister | wait does that mean on mainline you can't have flashlight? epic! | 02:18 |
| Mister_Magister | I won't repeat myself third time xd | 02:19 |
| Mister_Magister | I'm not upset i'm just disappointed | 02:24 |
| *** n2klb_ is now known as n2klb | 02:45 | |
| abr | The droidmedia flashlight implementation just provides a d-bus service implementing org.sailfish.flashlight.provider. You can write your own alternative doing that however you need to. It's more flexible than the old way, which was awful and needs to stay dead. | 05:43 |
| T42 | <elros34> @bruhforreal don't you have flattened apexes (don't make /apex symlink) error when droid-hal-ini-starts? | 09:30 |
| T42 | <damnbruh2004> Initially i have nothing in /apex and folders like com.android.* in /system/apex | 09:38 |
| T42 | <elros34> so what is output from droid-hal-init without symlink? | 09:51 |
| T42 | <damnbruh2004> First it complianed about missing selinux policies files in vendor. I crsted the /vendor -> /system_root/vendor symlink. Then it complained about boringssl test failing. I tried running the boringssl_self_test32 file manually and it complained about missing linker. I checked and turnns out /system/bin/linker (or smth like that) is linked to /apex/com.android.something/linker. /apex is empty so linked /apex -> /system/apex. Now boringssl | 09:57 |
| T42 | <damnbruh2004> And now i get further in init and it says somehting about the bpf loader failing | 09:58 |
| T42 | <code_bht> /apex | 10:00 |
| T42 | <damnbruh2004> Got em | 10:02 |
| T42 | <elros34> just show whole log if you can't find it | 10:02 |
| T42 | <elros34> but I bet its common non updatable apexes issue | 10:03 |
| T42 | <damnbruh2004> I'll send later today. Do I capture logs with / without the symlinks I already made? | 10:05 |
| T42 | <elros34> I think without /apex symlink it should be good but maybe I am wrong. I guess apex-bootstrap starts with error | 10:06 |
| Mister_Magister | abr: there's no other way to turn on flashlight than the old way though | 16:38 |
| Mister_Magister | and I don't have the source to write the alternative because it's closed sourcer | 16:38 |
| abr | there usually is | 16:38 |
| Mister_Magister | just sysfs, as it was doing before, thats the only way | 16:38 |
| abr | some sort of sysfs or debugfs maybe? | 16:38 |
| Mister_Magister | ye | 16:39 |
| Mister_Magister | what it was doing in 5.1 | 16:39 |
| abr | the old way was *not* sysfs | 16:39 |
| abr | it started a camcorder session with a small offscreen output surface, and turned on the light | 16:39 |
| Mister_Magister | imho jolla should write sysfs backend alternative | 16:39 |
| abr | which is why the phone got hot | 16:39 |
| Mister_Magister | I see | 16:39 |
| Mister_Magister | I thought it was sysfs | 16:39 |
| Mister_Magister | anyway we need sysfs | 16:40 |
| abr | the sysfs paths are different for each device, and only needed for old ones. everything from i think android 6 can use the proper way | 16:40 |
| abr | so you can do it yourself for your device. it's very simple | 16:40 |
| Mister_Magister | so the actual code doesn't matter i just need to expose same dbus? | 16:40 |
| abr | yup. and the package needs the same Provides as the droidmedia was | 16:40 |
| Mister_Magister | got it | 16:41 |
| abr | and it'll get pulled in | 16:41 |
| Mister_Magister | still think jolla should do that not me xd | 16:41 |
| abr | it's device specific | 16:41 |
| abr | see it was quite well thought out really | 16:41 |
| abr | unlike your mean comments about it | 16:42 |
| mal | abr: the commit message mentioned that it should fall back to old way but couldn't find it, also android 5.1 doesn't seem to contain the new api | 16:48 |
| abr | mal: maybe we decided to take the old way out later because it was so old. | 16:49 |
| abr | can't remember, it was a while ago | 16:49 |
| abr | but i thought we built it against 5.1? | 16:49 |
| mal | hmm, need to recheck it but I grepped the sources | 16:49 |
| abr | ah well, a sysfs would be better for the really old stuff anyway. that old method was a really terrible idea | 16:51 |
| mal | yeah, if there is possibility for sysfs then that should be used | 16:52 |
| abr | Mister_Magister: is it the g2 still? | 16:52 |
| mal | the fact that we still keep as old base as 5.1 still working, and old kernels, should be seen as a good thing | 16:53 |
| T42 | <Herrie1982> These are the variants we have covered in LuneOS now that go from very old (ie 3.4 kernel to current 7.x mainline) and everything in between on various Android versions: https://github.com/webOS-ports/nyx-modules/blob/webOS-ports/webOS-OSE/src/led_torch/led_torch.c | 16:54 |
| T42 | <Herrie1982> Small and lean enough to piggy back useful bits from | 16:54 |
| abr | that's a nice collection of all of them | 16:55 |
| mal | at some point 3.4 kernel might become too problematic | 16:55 |
| Mister_Magister | abr: yes | 16:56 |
| abr | I think you want /sys/kernel/debug/ledflash | 16:56 |
| Mister_Magister | well the torch is not the issue i know the sysfs path | 16:57 |
| mal | and those old devices don't support volte so those are becoming a bit useless already | 16:57 |
| T42 | <Herrie1982> You can get voicecalls via WhatsApp, Telegram, Signal and Teams working :P (re @SailfishFreenodeIRCBridgeBot: <mal>and those old d...) | 16:59 |
| T42 | <Herrie1982> We have those on LuneOS now, videocalls too btw :) | 16:59 |
| mal | true if waydroid or something works | 16:59 |
| T42 | <Herrie1982> Needs more testing, but so far pretty good. We're using the various libpurple plugins at our side, heavily patched, patches already PR-ed | 17:00 |
| mal | ok | 17:00 |
| mal | I really should find time to review and test some libhybris PRs | 17:01 |
| T42 | <Herrie1982> Will send some PRs for the various SFOS and Halium bits too | 17:01 |
| abr | libpurple's doing video calling now? that's very cool | 17:01 |
| T42 | <Herrie1982> gstdroid etc needed some work too :) | 17:01 |
| abr | that's not using telepathy too is it? | 17:01 |
| mal | @Herrie1982 what kind of work? codec stuff? | 17:01 |
| T42 | <Herrie1982> libpurple not, we have own backend in LuneOS, but all up for grabs | 17:02 |
| T42 | <Herrie1982> We just use the libpurple plugins for each of the networks with a rtc-engine in the back for the video bits, using gstdroid etc | 17:02 |
| mal | nice | 17:03 |
| T42 | <Herrie1982> https://github.com/webOS-ports/org.webosports.messaging is how the Messaging part looks like which does IM, pics, videos, voice messages, reactions, polls etc :) | 17:03 |
| T42 | <Herrie1982> Phone app takes care of the audio and video calls :) | 17:03 |
| T42 | <Herrie1982> mal: For gstdroid it are a lot.... | 17:13 |
| T42 | <Herrie1982> We currently carry 24 patches | 17:13 |
| mal | wow | 17:13 |
| mal | make a PR or PRs and we'll check | 17:13 |
| T42 | <Herrie1982> Will do in next couple of days (re @SailfishFreenodeIRCBridgeBot: <mal>make a PR or PR...) | 17:16 |
| T42 | <Herrie1982> Some are probably because of qt6, cleanup, codecs all kinds of reasons | 17:16 |
| T42 | <Herrie1982> For our rtc engine: https://github.com/webOS-ports/luneos-rtc-engine | 17:16 |
| mal | don't make PRs for each commit, that would be terrible to merge, at most a couple, maybe just one is ok, depends on the changes | 17:18 |
| T42 | <Herrie1982> Yeah will group them and squash where appropriate :) | 17:19 |
| T42 | <Herrie1982> Been testing this on everything from HP TouchPad, Nexus 4, Nexus 5 all the way to new devices with Android 15/16, qcom, mtk, exynos etc | 17:20 |
| mal | I should test my android 17 base | 17:22 |
| mal | I have it build already, includeing droidmedia | 17:22 |
| T42 | <Herrie1982> Ah droidmedia also have a few patches but thats only a handful | 17:23 |
| T42 | <Herrie1982> We have everything running on a 16.0 GSI now, even the 3.4 kernels with various vendors dating back to non-treble all converted to treble | 17:23 |
| mal | I have a16 fixes for droidmedia also, to make camera work on newer a16 revisions | 17:23 |
| T42 | <Herrie1982> So we only need to build armv7 and aarch64 for Android and mainline + vbox | 17:24 |
| mal | ok | 17:25 |
| mal | you have patched GSI kernel or what? | 17:25 |
| T42 | <Herrie1982> patched in all vndk versions mostly and some light shimming here and there | 17:33 |
| T42 | <Herrie1982> Few minor kernel patches but nothing major that we didnt have yet | 17:33 |
| T42 | <Herrie1982> Mostly to keep systemd 257 happy | 17:33 |
| Mister_Magister | [17:53] <mal> the fact that we still keep as old base as 5.1 still working, and old kernels, should be seen as a good thing | 18:55 |
| Mister_Magister | it absolutely is! | 18:55 |
| Mister_Magister | G2 is political statement not just a port at this point | 18:55 |
| Mister_Magister | device from 2014 can STILL be used today and that should be awesome | 18:56 |
| Mister_Magister | [17:57] <mal> and those old devices don't support volte so those are becoming a bit useless already | 18:56 |
| Mister_Magister | that is also sadly true, even op6 is borderline unusable | 18:56 |
| Mister_Magister | g2 is still useful as THE cheapest way to get your hands on sfos | 18:57 |
| Mister_Magister | 10$ is not much | 18:57 |
| Mister_Magister | [17:59] <mal> true if waydroid or something works | 18:58 |
| Mister_Magister | my telegram app does voicecalls without android | 18:58 |
| Mister_Magister | [18:01] <mal> I really should find time to review and test some libhybris PRs | 18:59 |
| Mister_Magister | like the sensorfw one from one certain dumbass called magister | 18:59 |
| Mister_Magister | meanwhile my other 2014 device currently runs kernel 7.3 | 19:01 |
| mal | I still don't quite understand why I was still supposed to review the sensorfw PR, I'll try to find some time | 19:11 |
| mal | I have some important PRs to review, connman, some gps stuff etc | 19:11 |
| Mister_Magister | mal: it was reviewed but i guess spiiroin asked for green light | 19:13 |
| Mister_Magister | just gotta type lgtm and merge xd | 19:13 |
| T42 | <Herrie1982> mal: Submitted what we carried and didnt submit yet. Feel free to pick or reject what you want, happy to rework based on feedback. Just want to reduce the patches we carry for most components | 19:55 |
| T42 | <Herrie1982> Also submitted fm radio to droidian for the pa-droid | 19:56 |
| T42 | <Herrie1982> Might be of interest in case you have any devices with fm radio | 19:56 |
| mal | we have functioning fm radio, what did you have to fix? | 19:58 |
| T42 | <Herrie1982> Ah ok didnt know that, we didnt have anything in LuneOS | 19:59 |
| Mister_Magister | oh i should probably test if the radio still works on g2 | 19:59 |
| mal | Mister_Magister: some old qcom devices have issues with fm radio because of sandboxing, those that need to write to sysfs | 20:00 |
| Mister_Magister | lovely | 20:00 |
| mal | there were two ways devices handled that, some did it in kernel and some used the sysfs way | 20:01 |
| mal | I think it's possible to switch that in kernel with some work | 20:01 |
| mal | @Herrie1982 thanks for the PRs, need to check those, some need some thorough review, those gst-droid ones change so much | 20:03 |
| T42 | <Herrie1982> I know take your time, some might not be very noticable at your end in your usecases, some were really what we saw in some edge cases like the voicecalls etc (re @SailfishFreenodeIRCBridgeBot: <mal>@Herrie1982 tha...) | 20:13 |
| T42 | <Herrie1982> Some others are proper and good hygiene and fixes, tried to split it that way | 20:13 |
| Mister_Magister | mal: you don't happen to know how can i, on mainline, preserve video aspect ratio withen doing software decoding | 20:21 |
| T42 | <adampigg> obs down again | 20:25 |
| T42 | <adampigg> had a new halium rootf to upload too | 20:26 |
| Mister_Magister | so it's because of you | 20:26 |
| Mister_Magister | mal: abr: in /sys/class/leds/ i have led:flash_0 and led:flash_torch so one of those should work | 21:02 |
| Mister_Magister | but they don't and /sys/kernel/debug/ledflash actually works so abr wins again i guess | 21:02 |
| Mister_Magister | apparently they are not supposed to be working lol so i'll use debug thingy | 21:07 |
| T42 | <adampigg> mister_magister: i think abr has some new hook for handling torch | 21:09 |
| Mister_Magister | nope | 21:09 |
| Mister_Magister | you're not caught up with our convo | 21:09 |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!