| T42 | <elros34> not so proper but easiest way would be removing gst-droid and then git clone --recurse https://github.com/sailfishos/gst-droid -b 0.20201104.0 | 00:02 |
|---|---|---|
| mal | I think running git submodule update --recursive in the gst-droid folder might also work | 00:03 |
| mal | or maybe git submodule update --init | 00:03 |
| T42 | <ankaos> mal: my "rpm / dhd / helpers / build_packages.sh" files are installed with version 3.4.0.24. | 00:05 |
| T42 | <elros34> mal looks like that is not enough, common for some reason can't be correctly updated | 00:08 |
| mal | @elros34 ok, submodules are sometimes annoying, at least I got rid of that in gst-droid by moving to meson build, it's just that that seems to have some issue with older sfos versions | 00:10 |
| T42 | <b100dian> hmm dbus does not seem to fail until `systemd[1]: Started User Manager for UID 100000.` or `systemd[4006]: Listening on D-Bus Session Message Bus Socket` https://pastebin.ubuntu.com/p/59FcwV67Cb/ and then every 90 secs there's `systemd[4006]: Failed to get initial list of names: Connection timed out`. Food for thought for tomorrow:) | 00:11 |
| T42 | <elros34> @b100dian do you have some logs from droid-hal-init in dmesg? | 00:38 |
| T42 | <b100dian> sure, https://pastebin.ubuntu.com/p/7g9t3Mgz6N/ | 00:49 |
| T42 | <b100dian> I'm going to reboot now, se you all tomorrow:) | 01:04 |
| T42 | <b100dian> And thanks again for your time | 01:04 |
| T42 | <elros34> sure, see you | 01:07 |
| Umeaboy | mal: You mind helping me solving my hybris-boot problem I mentioned last time? I don't know why it sees it as unset when it is in fact set. | 01:18 |
| Umeaboy | https://pastebin.ubuntu.com/p/Yd8ZJVgCCT/ | 01:19 |
| T42 | <A_T_R> https://paste.ubuntu.com/p/s6BVYSCjZv/ | 05:49 |
| T42 | <A_T_R> sample error - ``` clang-9: error: linker command failed with exit code 1 (use -v to see invocation)``` | 05:49 |
| T42 | <elros34> @A_T_R did you export needed env variable before building? | 09:08 |
| T42 | <b100dian> Morning - dmesg with kernel cmdline `printk.devkmsg=on` https://pastebin.ubuntu.com/p/YwjRCSVDHH/ is richful of droid-hal-init messages now. | 09:33 |
| T42 | <b100dian> (I had this message before: "droid-hal-init: 1457 output lines suppressed due to ratelimiting" before) | 09:33 |
| T42 | <b100dian> one thing I know now is that droid-hal-init is the one trying to modeprobe the vendor/lib/modules. Maybe this can be redirected to /lib/modules/$kernel-version? | 09:34 |
| rinigus | I enabled hybris debug + trace options and still don't see any indications of loading libhybris linker in flatpak env. as described yesterday over here. | 09:59 |
| rinigus | if there are any ideas why libhybris is not kicking in while strace shows loading libhardware and libwayland-egl from hybris, please tell. | 10:00 |
| Thaodan | b100dian: You can have any subdir in /lib/modules/$kernel-version something like /lib/modules/$kernel-version/extramodules/$vendor could work. | 10:06 |
| T42 | <elros34> *.rc files use modprobe -a -d /vendor/lib/modules so I think that will not work but you can just put all these modules in /etc/modules-load.d/*.conf | 10:12 |
| Thaodan | If it is packaged it should go to /usr/lib/modules-load.conf.d./] | 10:29 |
| Thaodan | If it is packaged it should go to /usr/lib/modules-load.conf.d | 10:29 |
| Thaodan | If it is packaged it should go to /usr/lib/modules-load.d | 10:29 |
| * Thaodan is still not fully awake | 10:29 | |
| T42 | <b100dian> Thanks, I'll try that next when at the keyboard. The only modules I have are for sound (a dozen) and wifi, do you think they are related to droid-hal-init not finishing startup? | 10:34 |
| T42 | <elros34> no | 10:38 |
| T42 | <A_T_R> @elros34 | 10:48 |
| T42 | <A_T_R> I have exported that ( export TEMPORARY_DISABLE_PATH_RESTRICTIONS=true ) , | 10:48 |
| T42 | <A_T_R> https://t.me/sailfishos_porters/115835 | 10:48 |
| T42 | <A_T_R> solved this issue (by | 10:48 |
| T42 | <A_T_R> comment-out the clang parameter from Boardconfig.mk to force use GCC -) | 10:48 |
| T42 | <A_T_R> but I got other compile errors due to an outdated device tree that I used | 10:50 |
| T42 | <elros34> not everybody use telegram so please in future use irclogs.sailfishos.org urls. You have many issues when building android part so I wonder whether you use same sources which where used to build lineage image. Is lineage-17 really better than lineage-16 on your device? | 11:16 |
| T42 | <A_T_R> i do not build android roms | 11:18 |
| T42 | <elros34> but you tried some | 11:19 |
| T42 | <A_T_R> Is lineage-17 really better than lineage-16 on your device? - idont know actually | 11:19 |
| T42 | <elros34> so why do you choose 17 as base for sailfish then? | 11:20 |
| T42 | <A_T_R> i compiled halium9 + hybris-16.0 | 11:20 |
| T42 | <A_T_R> just curious to try with lineage-17.1 🙄 | 11:21 |
| T42 | <elros34> ah ok:) | 11:22 |
| Mister_Magister | oh dang it now LED isn't working | 11:49 |
| rinigus | looks like eglGetPlatformDisplay is not implemented in hybris and that's why it fails to work in qt5.15/flatpak. | 11:57 |
| rinigus | mal ^ | 11:58 |
| mal | rinigus: hmm, that is only supported in egl 1.5, does it really need that new egl? | 12:04 |
| T42 | <elros34> I don't have access to notes now so can't be sure but I think when I was trying qt5.15 in chroot I had to rebuild qtwayland with libhybris installed or/and disable something to get rid of eglgetplatformdisplay requirement. Take this with grain of salt. | 12:05 |
| mal | rinigus: https://github.com/libhybris/libhybris/pull/430 see my comment there, I had the implementation, maybe we should add it back | 12:06 |
| mal | rinigus: try adding the changes to egl.c from here https://github.com/libhybris/libhybris/commit/2d40517ca87786a1b979d83c6fa1123220d468f1 | 12:08 |
| mal | rinigus: if that works I can make a PR of those old changes | 12:08 |
| rinigus | mal: don't know whether it is needed, but that what is called by /usr/lib/plugins/wayland-graphics-integration-client/libqt-plugin-wayland-egl.so | 12:09 |
| rinigus | mal: thanks, will test with those changes | 12:09 |
| rinigus | @elros34: (taking it with the grain of salt) thanks, sounds similar then | 12:10 |
| mal | rinigus: that seems to be conditional in qtwayland https://github.com/qt/qtwayland/blob/d10d7bab434b6917cd20ea57355570981974c54d/src/hardwareintegration/client/wayland-egl/qwaylandeglclientbufferintegration.cpp#L79 | 12:11 |
| rinigus | mal: interesting - just read your comment. | 12:11 |
| rinigus | mal: which means that they compiled flatpak platform with it | 12:12 |
| mal | rinigus: wondering how older android bases would work with that | 12:12 |
| mal | rinigus: or maybe not | 12:13 |
| rinigus | these days only aarch64 is supported for flatpaks (from arm side) | 12:13 |
| mal | rinigus: you can see there are two ways it can reach eglGetPlatformDisplay in the code | 12:13 |
| mal | rinigus: check the egl extensions you have listed to see which one it reaches | 12:14 |
| rinigus | mal: will check the strace... | 12:15 |
| mal | rinigus: wondering if we should make a fallback to those devices which don't support that | 12:16 |
| rinigus | mal: let's get it working first at least somewhere - one step in a time :) | 12:16 |
| mal | ok | 12:16 |
| mal | I'll be away for some time, need to go cycling | 12:17 |
| rinigus | me too - have to move a bit. but I think the error is actually quite clear as reported by ldd: plugins/wayland-graphics-integration-client/libqt-plugin-wayland-egl.so is missing that symbol | 12:17 |
| rinigus | will continue later today, enjoy being out! | 12:18 |
| Thaodan | mal rinigus: there is a glvnd pr pending that should be merged before doing stuff like that IMHO | 12:18 |
| Thaodan | mal: good idea, should do the same, should explore more of Finland | 12:19 |
| mal | Thaodan: ah, that is not draft anymore | 12:19 |
| mal | maybe I should finally make a PR to libhybris for vulkan support | 12:21 |
| mal | rinigus: Thaodan: btw, that glvnd PR implements eglGetPlatformDisplay | 12:22 |
| Thaodan | Rinigus: test pls^^ | 12:23 |
| rinigus | mal and Thaodan : I will get in touch later tonight when I get the chance to sit down and test. thanks a lot for pointers! | 12:29 |
| Thaodan | rinigus: always happy help, looking forward to your results | 12:31 |
| T42 | <elros34> rinigus: just to confirm: when building qtwayland against libhybris I got Checking for EGL 1.5 with Wayland Platform... no" so eglGetPlatformDisplay wasn't needed anymore | 13:08 |
| rinigus | @elros34: thanks! as building qt for flatpak is out of my hands (at the moment), I will have to get eglGetPlatformDisplay working. let's see if it will work | 13:49 |
| Umeaboy | Hi! | 15:20 |
| Umeaboy | Thaodan: Are you perhaps around? | 15:22 |
| Umeaboy | I can't seem to successfully build the kernel without warnings even thou I did change in the defconfig first. | 15:22 |
| T42 | Yoursensai was added by: Yoursensai | 15:39 |
| T42 | <Yoursensai> Is sailfish port available for whyred | 15:43 |
| Thaodan | I am | 15:46 |
| Thaodan | please post logs Umeaboy | 15:46 |
| Thaodan | Some warnings are fine they are intentionally ignored | 15:47 |
| mal | Umeaboy: just an idea, export the variable just before running make | 15:48 |
| mal | not sure what the env setup does before that | 15:48 |
| Thaodan | e.g. see here: https://github.com/sonyxperiadev/kernel/blob/aosp/LA.UM.7.1.r1/Makefile#L649 | 15:49 |
| Thaodan | mal: You the one about using external tools? | 15:50 |
| mal | Thaodan: ? | 15:53 |
| Thaodan | TEMPORARY_DISABLE_PATH_RESTRICTIONS | 15:55 |
| Thaodan | https://android.googlesource.com/platform/build/+/master/Changes.md#path_tools | 15:55 |
| mal | what about that? | 16:05 |
| Thaodan | I thought he got the warning because a tool that was from path was used in our projects e.g. cpio in hybris-boot | 16:34 |
| T42 | <A_T_R> https://pastebin.ubuntu.com/p/mvg6NRwvwC/ | 16:39 |
| mal | Thaodan: yes, and exporting that should fix it but didn't seem to do it afaik | 16:59 |
| rinigus | mal and Thaodan : started with libglvnd-pr. now it loads hybris (so, required symbols are exported and pass the first check). however, after some search, it states undefined symbol: eglCreatePlatformWindowSurface | 17:08 |
| rinigus | mal's changes I haven't tried yet | 17:08 |
| rinigus | for this test, I compiled only SFOS bits, without recompiling droid-hal | 17:10 |
| rinigus | now, I am not 100% sure it is correct way of doing it | 17:10 |
| mal | rinigus: try adding my changes on top of the glvnd | 17:11 |
| mal | at least relevant parts of it | 17:11 |
| mal | meaning do not add another one of eglGetPlatformDisplay | 17:11 |
| rinigus | OK, I can try to add them on top. its basically HYBRIS_IMPLEMENT_FUNCTION ... lines, right? | 17:12 |
| rinigus | mal: ^ | 17:12 |
| rinigus | but these (glvnd and yours) changes don't require droid-hal rebuild, right? | 17:13 |
| mal | rinigus: of course not, libhybris is built using droid-hal not the other way :) | 17:13 |
| mal | well except the android side libhybris things like the compat stuff but those should not be affected | 17:14 |
| rinigus | that's what I thought - I had a feeling that there are bits in libhybris compiled on android side and packaged there... into droid-hal- packages, I presume | 17:15 |
| mal | yes, hwc2 compat layer but we haven't changed it | 17:17 |
| rinigus | mal: no go - we end up defining the same symbols twice: https://dpaste.org/MWNP | 17:19 |
| mal | rinigus: 20:11 < mal> meaning do not add another one of eglGetPlatformDisplay | 17:20 |
| rinigus | :) | 17:20 |
| rinigus | mal: that seemed to make all linked and it just crashes without any major error | 18:19 |
| rinigus | would have to look into it, maybe gdb can help | 18:19 |
| mal | rinigus: have you checked if it really return anything useful functions for those | 18:24 |
| rinigus | mal: no. how to check it? | 18:24 |
| mal | not sure, try gdb first to see how it fails | 18:25 |
| mal | but maybe libhybris debug printing could tell something | 18:26 |
| rinigus | mal: crash looks to be happening with my maliit plugin active. after uninstalling it, I get application hanging and not showing anything | 18:42 |
| rinigus | although, without hybris acceleration, apps work fine with maliit | 18:42 |
| rinigus | so, rather weird, but hard to help from outside - let's see if I can dig up something | 18:43 |
| Mister_Magister | hmm led doesn't work in CSD but works when i charge phone | 20:53 |
| Mister_Magister | interesting | 20:53 |
| Umeaboy | mal: I'm back now. | 21:14 |
| Umeaboy | I'm a little lost....... export sunrpc.......... | 22:05 |
| Umeaboy | ? | 22:05 |
| mal | ? | 22:05 |
| Umeaboy | I have to export something to build successfully right? | 22:09 |
| Umeaboy | You did read the entire log right? | 22:11 |
| Umeaboy | https://pastebin.ubuntu.com/p/Yd8ZJVgCCT/ | 22:12 |
| mal | yes, I asked you to try to run "export TEMPORARY_DISABLE_PATH_RESTRICTIONS=true" just before make hybris-hal, in case envsetup.sh sets some default things | 22:14 |
| Umeaboy | mal: Does it make any difference if I run it before or after because I did run it. | 22:19 |
| mal | Umeaboy: I don't know, that is why I asked you to try | 22:20 |
| mal | I haven't checked what envsetup.sh does | 22:20 |
| Umeaboy | mal: The envsetup.sh script imports several commands that enable you to work with the Android source code, including the commands used in this exercise. | 22:22 |
| Umeaboy | https://source.android.com/setup/build/building | 22:22 |
| mal | yes, but does it set some default value for env vars etc | 22:22 |
| mal | can't you just try? | 22:22 |
| Umeaboy | Sure. | 22:31 |
| Umeaboy | Do I run make clean first? | 22:31 |
| mal | no need | 22:31 |
| mal | hopefully | 22:31 |
| Umeaboy | OK. | 22:32 |
| Umeaboy | I need to rebuild hybris-hal? hybris-boot is the failing part. | 22:34 |
| mal | probably hybris-boot should be enough | 22:35 |
| Umeaboy | OK. | 22:35 |
| Umeaboy | mal: Here you can see that I did as you told me and I still got the same error: https://pastebin.com/Kui0FfGN | 23:07 |
| Thaodan | The warnings are harmless in this case | 23:13 |
| Thaodan | the kernel is self contained and does not use anything else then cc/the compiler, make and the shell. It ignores most of the env. | 23:13 |
| Thaodan | Is that really android 10? AOSP can't build in tree kernel anymore with 10. | 23:14 |
| Thaodan | Also don't use pastebin | 23:15 |
| Umeaboy | Thaodan: Lineage 18.1 is based of AOSP 10 right? | 23:15 |
| Thaodan | No idea I don't use Lineage | 23:16 |
| Umeaboy | 17.1 | 23:16 |
| Umeaboy | Sorry. | 23:16 |
| Umeaboy | My bad. | 23:16 |
| Thaodan | I any case I guess that code is broken in the kernel of the vendor. I would compare the kernel tree with one where it is working. E.g. Sony AOSP kernel where have that enabled. | 23:17 |
| Thaodan | I would recomment looking at white commit state that code is and get the fixes missing apply them and submit the patches to the kernel maintainer for that tree. | 23:18 |
| T42 | <elros34> the error is quite clear if you read the logs, sunrpc driver have error but you don't need it anyway so just do not enable it. defconfig is also wrong but I have already told you few days ago what to change | 23:18 |
| Thaodan | Yeah nfs is nice to have if you use it. | 23:19 |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!