| T42 | <b100dian> mal: that zambezi reference does not fix the problem even after reflashing - still only two repos in my ssu lr... | 00:29 |
|---|---|---|
| T42 | <b100dian> Elros: I have to find time to build locally another device to compare the *.packages files, since I don't package that in github actions for tucana or sake. A quest for tomorrow | 00:29 |
| mal | quite sure something is wrong with the ssu configs, not sure what | 00:30 |
| mal | what does "ssu s" say? | 00:30 |
| mal | remove imei before pasting the output | 00:31 |
| mal | relevant part is the code name | 00:31 |
| T42 | <b100dian> https://paste.mozilla.org/Ued7KUSz/raw | 00:36 |
| Umeaboy | Hi! | 00:37 |
| Umeaboy | I realize that converting the PDF of the HADK into a program with a GUI is a daunting task, but if anyone is motivated and interested in helping out here's my start for it. I know the code sucks, but if the errors get corrected it will build into a nice GUI. :) | 00:39 |
| Umeaboy | https://github.com/sailinggalaxians/sailfishos-hadk-gui/tree/development | 00:39 |
| Umeaboy | Please only to commit to the development branch which is set to be standard. | 00:39 |
| Umeaboy | At least until it's stable enough to build. | 00:40 |
| T42 | <b100dian> mal: on another device I have "domain: sales" | "brand: Jolla" as opposed to nothing and "N/A" | 00:46 |
| mal | so ssu is completely messed up, just not sure exactly why, must be some mistake somewhere in the configs | 00:49 |
| mal | @b100dian what is the codename of your device? | 00:51 |
| T42 | <b100dian> pdx224 or xqcq54, depends on how you interpret it. | 00:53 |
| T42 | <b100dian> I got a diff between a sake package list and nagara: https://paste.mozilla.org/nLMTgss5/raw It seems even things like "less" appear with - (not present in nagara) | 00:58 |
| T42 | <elros34> MISSING patterns-sailfish-rnd ssu-vendor-data-jolla | 01:07 |
| T42 | <b100dian> indeed, it is what I am seeing as an effect, but of what, it is yet TBD | 01:16 |
| T42 | <b100dian> %define rpm_vendor qualcomm (in zambezi) vs %define rpm_vendor sony (in nagara). But it is not clear how it's used (MER_HA_VENDOR) | 01:23 |
| T42 | <b100dian> I will log out now, thanks for the help ! | 01:23 |
| T42 | <b100dian> another diff? Prefer: here only points to one device (1 V) https://build.sailfishos.org/project/prjconf/nemo:devel:hw:sony:nagara Kumano is a liitle more .. involved but not really device specific https://build.sailfishos.org/project/prjconf/nemo:devel:hw:sony:kumano | 01:28 |
| mal | @b100dian you are missing some things from patterns like patterns-sailfish-consumer-generic | 01:45 |
| mal | I checked the sources and that is pulling in ssu-vendor-data-jolla | 01:47 |
| mal | among a lot of other things | 01:48 |
| mal | @b100dian it seems you copied directly things from official repos, which don't need that because the official building system is different | 01:49 |
| mal | but community builds need patterns-sailfish-consumer-generic https://github.com/mer-hybris/droid-hal-configs/blob/master/patterns/templates/patterns-sailfish-device-configuration-%40DEVICE%40.inc#L28 | 01:49 |
| mal | while at it also add this https://github.com/mer-hybris/droid-hal-configs/blob/master/patterns/templates/patterns-sailfish-device-configuration-%40DEVICE%40.inc#L24C1-L24C59 | 01:50 |
| mal | and patterns-sailfish-device-tools is useful for early porting | 01:51 |
| rinigus | Morning! Nagara ssu config is based on official device and it wasn't tested on device. As @b100dian mentioned, we have nagara in repo name, not specific device. So common had to be changed. | 06:46 |
| rinigus | Thanks for tips regarding missing patterns. Will have to revise ssu as well | 06:47 |
| rinigus | @b100dian and mal: interesting that this linking error is across devices. mal, is this error coming together with r67 update? | 06:52 |
| rinigus | @b100dian: started to look into ssu. very handy diff with sake | 07:54 |
| rinigus | @b100dian: %define rpm_vendor sony is used to create url for OBS repo. as we have :sony: in the repo name, we have to use sony | 08:01 |
| rinigus | @b100dian: prjconf should be OK, same scheme worked for tama devices | 08:23 |
| rinigus | @elros34: I wonder whether patterns-sailfish-rnd ssu-vendor-data-jolla are pulled by something else. will try with added patterns-sailfish-consumer-generic | 08:27 |
| rinigus | @b100dian: I have adjusted patterns based on the feedback from last night and sake config. also, we had old pulseaudio (same as I used on tama); replaced with common on OBS | 08:51 |
| rinigus | looking into ssu | 09:07 |
| rinigus | @b100dian: ssu should be fixed now. I have dropped our custom board mappings and let scripts to generate them | 09:46 |
| rinigus | current ssu config: https://paste.mozilla.org/tGSB4RTG/raw | 09:46 |
| rinigus | new config and pulseaudio are at obs waiting for your mic image generation :) | 09:47 |
| rinigus | only concern is Device UID in ssu s. namely it is not IMEI but some uuid, I think. but that could be related to the absence of proper ril on our aosp | 09:50 |
| T42 | <b100dian> Thanks rinigus, will try them soon:) so we should have gdb and strings and less and all the things now \o/ | 09:52 |
| rinigus | @b100dian: gdb, strings, less - all there :) | 09:52 |
| T42 | <b100dian> Thanks mal for the answers above, I was logged out by then, seems they helped a lot | 09:52 |
| T42 | <b100dian> rinigus: do you think we can use %{rpm_device} here ? https://github.com/sailfishos-sony-nagara/community-adaptation/blob/updated/adaptation-community.ini.in#L7 Or @HABUILD_DEVICE@ | 09:59 |
| rinigus | @b100dian: no, I don't think it is defined | 10:09 |
| rinigus | why do you want to use it? | 10:09 |
| T42 | <b100dian> Just to not hardcode stuff. But my intuition also was that it wouldn't be defined (or FAMILY etc) | 10:10 |
| rinigus | I think right now it should be all OK. | 10:11 |
| rinigus | @b100dian: looks like by accident it booted to actdead mode and that seem to work | 10:25 |
| T42 | <b100dian> Screen is on:D? With charging indicator? | 10:26 |
| rinigus | yes, charging indicator is shown | 10:30 |
| rinigus | @b100dian and mal : libmedia.so missing symbol is in libstagefright_foundation.so . from just looking on ldd output and checking that required libstagefright_foundation.so are all in /system/lib64, it is not clear what prevents loading it | 12:25 |
| rinigus | corresponding paste: https://paste.mozilla.org/JSpdovp7/raw | 12:26 |
| rinigus | oops, it has slightly different name. probably that's the issue | 12:30 |
| rinigus | sorry, name is the same _ZN7android12MetaDataBase13writeToParcelERNS_6ParcelE | 12:31 |
| T42 | <b100dian> So the symbol is actually there? Interestring - does ldd actually know where to look in the case of android libs? | 12:49 |
| rinigus | @b100dian: with minimer I am getting a different error, will paste in a sec | 12:52 |
| rinigus | paste at https://paste.mozilla.org/6Bf74jv8/raw | 12:57 |
| rinigus | I wonder, it does remind me of https://github.com/sonyxperiadev/bug_tracker/issues/851 | 12:57 |
| T42 | <b100dian> Hmn so you do not get the symbol error with minimer, interesting, I had that yesterday | 13:08 |
| rinigus | so, if you read that aosp issue - it was sufficient to copy libs into corresponding folder and keyboard opened. linker is rather picky regarding location of the libs and it does not want to follow symlinks (due to selinux?) | 13:09 |
| rinigus | I copied libandroidicu and friends to /overlays and now the error is different: https://paste.mozilla.org/kKPbCBrp/raw | 13:10 |
| rinigus | sorry for longer strace | 13:11 |
| T42 | <elros34> I remember somebody had similar issue that symlink in odm vs copied library have worked differently | 13:16 |
| T42 | <elros34> did you check logcat for new errors? | 13:16 |
| T42 | <elros34> it can't find any hwcomposer library | 13:24 |
| T42 | <b100dian> maybe the overlay should only have empty files and we mount-bind the originals over them. If I remember correctly, there's a difference in detecting the namespace when you have a link vs a mount bind, maybe overlays are more like links | 13:31 |
| rinigus | @b100dian and @elros34: forgot to check logcat. will do. as for overlay and mount-bind - maybe. right now we can experiment with just a copy | 13:53 |
| rinigus | maybe we need also to look at linker. but I'll start with logcat and one more error/issue that was there | 13:54 |
| rinigus | @elros34: logcat a bit before minimer and after available at https://paste.mozilla.org/jyva8GY3/raw | 14:24 |
| rinigus | I'll check where hwcomposer lib could be | 14:25 |
| rinigus | One more observation: in AOSP, we have a message in logcat "linkerconfig generated /linkerconfig with mounted APEX modules info". in SFOS it is absent | 14:29 |
| rinigus | in SFOS we have: linker : Warning: failed to find generated linker configuration from "/linkerconfig/ld.config.txt" | 14:30 |
| T42 | <b100dian> @elros34 this sounds familiar with some linkerconfig patch you had, right? | 14:46 |
| rinigus | @elros32: here is the list of composer-related files that I found on /system and /vendor. in addition there is a list from binder: https://paste.mozilla.org/U1TajEMz/raw | 14:46 |
| T42 | <b100dian> The patch I mentioned I used in hybris-18.1 as here https://github.com/sailfishos-on-sake/hybris-patches/blob/hybris-18.1/system/core/0051-linkerconfig-switch-no-updatable-apex.patch but maybe it is not applicable to 21 | 14:48 |
| T42 | <elros34> IIRC it's not | 14:48 |
| T42 | <b100dian> Ok, we might need to re-read the android init | 14:51 |
| T42 | <b100dian> rinigus, about that message in init https://android.googlesource.com/platform/system/core/+/refs/tags/android-14.0.0_r67/init/builtins.cpp#1242 | 15:37 |
| T42 | <b100dian> It seems that we also have in logcat | 15:37 |
| T42 | <b100dian> droid-hal-init: Command 'perform_apex_config' action=post-fs-data (/usr/libexec/droid-hybris/system/etc/init/hw/init.rc:1029) took 81ms and failed: Failed to find current mount namespace | 15:37 |
| rinigus | @b100dian: not sure what can we do about it | 15:46 |
| T42 | <b100dian> the last message is init reading its own /proc/self/ns/mnt and that fails | 15:47 |
| T42 | <b100dian> yeah, no low hanging fix found quickly | 15:48 |
| T42 | <elros34> unless something changed in hybris patch there will be no linker namespaces for android part. No idea how this work in hybris21 but do you have some directories in /linkerconfig? Btw did you mask user@100000 before trying minimer? | 15:54 |
| T42 | <elros34> have you tried whether surfaceflinger shows something? | 15:55 |
| T42 | <b100dian> yes, /linkerconfig seems to be fairly 'well' https://pastebin.ubuntu.com/p/Qx5qRcWXrf/ | 15:57 |
| rinigus | @elros34: I haven't masked anything. how would I mask user@100000 | 15:58 |
| T42 | <elros34> just systemctl mask user@100000; reboot. Just precaution if lipstick starts | 15:59 |
| rinigus | will try | 16:00 |
| rinigus | my /linkerconfig looks as the one from @b100dian. maybe its fine... | 16:01 |
| rinigus | as for what changed, from aosp sony bugs replies sounded like _r67 had some changes in linker related aspects. what exactly, I don't know | 16:03 |
| rinigus | example bug: https://github.com/sonyxperiadev/bug_tracker/issues/851 | 16:03 |
| rinigus | if I mask user@100000, I have SFOS logo which doesn't disappear. | 16:07 |
| rinigus | and minimer is not crashing, but https://paste.mozilla.org/Sx7Vk6FG/raw | 16:08 |
| rinigus | I guess I need to get rid of logo :( | 16:08 |
| T42 | <elros34> you mean yamuisplash logo? | 16:09 |
| rinigus | @elros34: yes. so, stopped the service. still the messages are the same | 16:11 |
| rinigus | will paste logcat | 16:12 |
| rinigus | https://paste.mozilla.org/K2L1KF7s/raw | 16:15 |
| T42 | <elros34> pressing power button doesn't help? Maybe you can compare this to surfaceflinger test(if it even works) | 16:18 |
| rinigus | tried, but it didn't. would try again | 16:23 |
| T42 | <elros34> did you use correct command with bootanim service according to hadk-hot? | 16:26 |
| T42 | <b100dian> https://sailfishos.wiki/link/20#bkmrk-%E6%88%91%E6%9C%89-telnet-%E4%BD%86%E6%B2%A1%E6%9C%89-gui%EF%BC%9A | 16:30 |
| T42 | <b100dian> nice chapter name ;) | 16:30 |
| T42 | <b100dian> so, we it might be the SailfishOS logo that's already on display, don't remember how that is called, I'll disable it next | 16:31 |
| T42 | <b100dian> yamuisplash | 16:32 |
| rinigus | so, as soon as I stop yamuisplash, I am getting @@@@ panel power off @@@@ and it doesn't come on anymore | 16:35 |
| rinigus | running /system/bin/surfaceflinger did not produce anything on the screen | 16:36 |
| T42 | <b100dian> no, I mean, I've masked it as user@100000 and rebooted | 16:36 |
| T42 | <b100dian> I have some progress to report: a static image of test_hwcomposer (instead of animating) | 16:36 |
| rinigus | ooooo | 16:37 |
| T42 | <b100dian> but with a changed test_hwcomposer https://github.com/b100dian/libhybris/tree/fix/test_hwcomposer | 16:37 |
| T42 | <b100dian> which initializes HWC2 first. | 16:37 |
| T42 | <elros34> rinigus but did your bootanimation also? | 16:37 |
| T42 | <b100dian> no, surfaceflinger and bootanim didn't start for me either (I mean, didn't display anything) | 16:38 |
| T42 | <b100dian> this build of test_hwcomposer is actually.. binary copied from sake :D where I debugged this more. I will temporarily include that branch in our hybris to see if changes anything | 16:38 |
| rinigus | @elros34: I'll have to test android bits as well. I presume that bootanim has to run | 16:38 |
| T42 | <b100dian> but the fact that it is not animating is not encouraging | 16:38 |
| T42 | <b100dian> logcat when running that test_hwcomposer https://paste.mozilla.org/edCmS1GK/raw | 16:40 |
| T42 | <b100dian> rinigus: that hybris fork with modified HWC2 first test_hwcomposer would be this from sake, would you be ok to add it to our OBS .. temporarily ? https://build.sailfishos.org/package/view_file/nemo:devel:hw:asus:sake/libhybris/_service?expand=1 | 16:42 |
| rinigus | sure | 16:42 |
| T42 | <b100dian> or since the change is in the test executable, maybe it would make more sense to build it locally, just to check 🤔. | 16:46 |
| T42 | <b100dian> I still have all the errors when running minimer, the symbol not found, the namespace one and a new(?) tls one https://paste.mozilla.org/CUuzvc7C/raw | 16:47 |
| T42 | <b100dian> rinigus: if you have fewer errors on minimer it might be worth masking yamuisplash and user@100000 and reboot and test minimer again? | 16:47 |
| rinigus | @elros34: android /system/bin/surfaceflinger and in parallel after that ANDROID_ROOT="/system" /system/bin/bootanimation shows android boot animation | 16:48 |
| rinigus | @b100dian: will test minimer again with splash masked | 16:50 |
| rinigus | @elros34 and @b100dian: minimer worked | 16:51 |
| mal | how did you solve the linking issue? | 16:51 |
| T42 | <b100dian> Yay! Is it interactive? | 16:51 |
| T42 | <b100dian> I mean animating | 16:51 |
| rinigus | but that is in the session that was without splashscreen and after I stopped surfaceflinger | 16:52 |
| T42 | <b100dian> asking because I found more lines in logcat on my test_hwcomposer, and one is "kgsl-3d0: possible deadlock. Context 4 might be blocked for itself" - I would not hope to investigate such things :D | 16:52 |
| rinigus | mal: "solved" was to copy libraries into overlayed /system/lib64 | 16:53 |
| T42 | <b100dian> but we can do that on boot :P | 16:53 |
| rinigus | @b100dian: interactive? not sure - tapping does nothing | 16:53 |
| T42 | <b100dian> No, I mean does it move anything on screen | 16:53 |
| rinigus | yes, it does :) | 16:53 |
| mal | this must be some sony issue, not r67 issue since fp5 boots fine with r67 | 16:54 |
| T42 | <b100dian> rinigus: perfect. ship it :) | 16:54 |
| rinigus | OK, will try now to boot again and see if it works from start | 16:54 |
| mal | rinigus: which files did you copy? | 16:55 |
| rinigus | mal: all here https://github.com/sailfishos-sony-nagara/droid-config-sony-nagara/tree/main/sparse/overlays/system/lib64 were with symlinks replaced by their real libs | 16:56 |
| rinigus | later, on boot, this is overlaid with /system/lib64 | 16:57 |
| rinigus | @b100dian: nothing to ship. if you do this overlay it should work for you too | 16:57 |
| rinigus | from boot, minimer works | 16:59 |
| rinigus | let's see if user@100000 can come up as well | 17:00 |
| T42 | <adampigg> glad progress is being made, then you can get back to qt6 with me :) | 17:00 |
| rinigus | and we have GUI | 17:02 |
| rinigus | @adampigg: not too soon, but looking forward! | 17:02 |
| T42 | <b100dian> rinigus: great! So only actual libs and disable yamuisplash? | 17:03 |
| rinigus | @b100dian: exactly! | 17:04 |
| T42 | <b100dian> Might mean we need to restart hwc service before lipstick, if you want to keep the splash, but I thought that was already happening | 17:07 |
| rinigus | @b100dian: SFOS is reporting that I have 19% battery. will have to figure out whether I can trust it or not | 17:37 |
| T42 | <b100dian> rinigus: how are you reading that? | 17:37 |
| T42 | <b100dian> (using the GUI:)? | 17:37 |
| rinigus | GUI :) | 17:37 |
| T42 | <b100dian> cat /sys/class/power_supply/battery/capacity maybe? | 17:39 |
| rinigus | mal: we also have a script for mounting the overlays: https://github.com/sailfishos-sony-nagara/droid-config-sony-nagara/blob/main/sparse/usr/sbin/mount-overlays . so, it should be easy to use | 17:42 |
| rinigus | @b100dian: maybe | 17:43 |
| rinigus | @b100dian: didn't take a chance and flashed stock back. that one reported the same 15% as were reported by SFOS | 17:58 |
| rinigus | charging... | 17:58 |
| T42 | <b100dian> good. Tomorrow you can try our image on stock ;) My problem with AOSP is getting hot, not the charge. | 17:59 |
| rinigus | @b100dian: not sure hwc restart after splash will help. looks like panel goes off and then we cannot start GUI anymore. need to get into GUI then can switch on/off as we please | 17:59 |
| rinigus | AOSP is getting hot which is similar issue with the charge disappearing faster than it should. maybe need faster charges to keep it up | 18:00 |
| rinigus | chargers | 18:01 |
| rinigus | but maybe we should move to stock. I will look around a bit and see how to get kernel for it | 18:02 |
| T42 | <b100dian> Ah, I always forget that theres kernel mismatch.. | 18:03 |
| rinigus | there is, unfortunately. | 18:05 |
| rinigus | I wonder if has anyone around here tried to compile kernel for sony stock? maybe we already have experts in-house | 18:23 |
| T42 | <edp_17> I also would need a change in xa2 kernel, so, am too interested where from I can get the source. | 20:24 |
| T42 | <elros34> what do you mean? xa2 kernel sources like for all sailfish x sony devices are in mer-hybris | 20:32 |
| mal | @edp_17 what kind of change? | 21:19 |
| T42 | <edp_17> Mal: there is a missing flag. (I can't remember which one.) | 21:30 |
| T42 | <edp_17> @elros34 , Thanks, I forgot about that it is there. I'll try to build one. | 21:31 |
| mal | @edp_17 what is the flag needed for? | 22:00 |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!