| T42 | <A_T_R> @edp_17 did you find a solution for missing libdl_android.so rather than manual copy ? | 01:13 |
|---|---|---|
| T42 | <edp_17> @A_T_R , Yes. I didn't copy that. I've amended droid-hal-device.inc like: | 01:18 |
| T42 | <edp_17> %if 0%{?droid_target_aarch64:1} | 01:18 |
| T42 | <edp_17> rm -f $RPM_BUILD_ROOT%{_libexecdir}/droid-hybris/system/lib64/lib{dl,dl_android,c,m}.so | 01:18 |
| T42 | <edp_17> +# edp17 - dont copy libdl_android.so because that is elsewhere | 01:19 |
| T42 | <edp_17> cp -a \ | 01:19 |
| T42 | <edp_17> - $art_path/$apex_path/lib64/bionic/lib{dl,dl_android,c,m}.so \ | 01:19 |
| T42 | <edp_17> + $art_path/$apex_path/lib64/bionic/lib{dl,c,m}.so \ | 01:19 |
| T42 | <edp_17> $RPM_BUILD_ROOT%{_libexecdir}/droid-hybris/system/lib64 | 01:19 |
| T42 | <edp_17> +# edp17 copy libdl_android.so from elsewhere | 01:19 |
| T42 | <edp_17> + cp $ANDROID_ROOT/out/soong/.intermediates/bionic/libdl/libdl_android/android_arm64_armv8-a_core_shared/libdl_android.so $ANDROID_ROOT/installroot/usr/libexec/droid-hybris/system/lib64/ | 01:19 |
| T42 | <edp_17> + #cp $ANDROID_ROOT/out/soong/.intermediates/bionic/libdl/libdl_android/android_arm64_armv8-a_core_shared/libdl_android.so $RPM_BUILD_ROOT%{_libexecdir}/droid-hybris/system/lib64/ | 01:19 |
| T42 | <edp_17> %endif | 01:19 |
| T42 | <edp_17> Maybe this way is more readable: https://paste.ubuntu.com/p/8Zh8fPCQn9/ | 01:20 |
| entil | been crazy busy so haven't followed along, but is there any good news for fp6 and/or waydroid for any of the devices? | 05:01 |
| rinigus | in nagara/los21 port, we are unable to record video. while recording, there are complains in logcat regarding absence of package_native. (some other complains too). this is with camera1 api. absence of packages has been also observed with camera2 by @b100dian | 06:57 |
| rinigus | logs at https://github.com/sailfishos-sony-nagara/main/issues/34 | 06:58 |
| rinigus | any tips? | 06:59 |
| rinigus | b100dian: made some progress with the video. for that had to write fake_packagenative service and the recording started working. | 09:28 |
| rinigus | will cleanup a bit and then push sources | 09:28 |
| rinigus | except now we have the same issue as mal mentioned earlier - video looks to be recorded and frames are there, but playback is not fully functional. I think it was related to c2 codecs if I remember correctly | 09:29 |
| rinigus | hmm, when you pull to PC it works just fine. something with playback on device | 09:33 |
| T42 | <b100dian> Great idea with the fake package native ! (which is missing because it is native, but in java;) | 09:36 |
| T42 | <b100dian> I was wondering, about playback: No more Sailfish3 movie distributed to test? | 09:37 |
| T42 | <b100dian> Rinigus: is that a piece of hal you're pushing for all devices on obs, just like, say, hwcrypt? | 09:39 |
| rinigus | @b100dian: and when tested in SFOS browser video plays just fine! so, something is funky with playback | 09:52 |
| rinigus | in gallery | 09:52 |
| rinigus | re hal piece: have to think. I guess similar to fake_crypt and hwcrypt | 09:52 |
| rinigus | although, it could be great opportunity to learn gbinder and make it using that | 09:53 |
| rinigus | if it supports writing services | 09:53 |
| rinigus | any good service to take as an example? | 09:53 |
| rinigus | api at https://android.googlesource.com/platform/frameworks/native/+/master/libs/binder/aidl/android/content/pm/IPackageManagerNative.aidl | 09:55 |
| rinigus | and I suspect we don't even need to write most of it for sfos | 09:55 |
| rinigus | ... reading examples in libgbinder now | 09:58 |
| T42 | <elros34> @edp_17 what issue do you have with cursor? there is dconf key to enable visibility | 10:22 |
| T42 | <edp_17> @elros34 : the touchscreen is broken on my S4. I can control it with mouse and keyboard. However, as cursor is not displayed, don't know where it is positioned. | 14:48 |
| T42 | <edp_17> Tried to install your mousetracker but when installed as patch, device always reboots when patchmanager turns it on. | 14:49 |
| T42 | <edp_17> When I installed manually into qml file, I only got a white screen. As device is buggy, I couldn't ssh into it as wifi not started. | 14:51 |
| T42 | <elros34> did you use correct version for sfos 4.6? | 14:51 |
| T42 | <elros34> you need both patch file and library. 32 or 64 bit device? | 14:51 |
| T42 | <edp_17> The device is on sfos 4.4 and yes, I've downloaded correct version from your repo. | 14:52 |
| T42 | <elros34> ah 4.4 not 4.6 | 14:52 |
| T42 | <edp_17> Yes, I know needed the library and put it into the right place. It's arm32. | 14:53 |
| T42 | <elros34> AFAIK on 4.4 sailfish implementation is not enabled by default | 14:53 |
| T42 | <edp_17> Yeah, that's my pain. | 14:53 |
| T42 | <elros34> 'dconf read /desktop/sailfish/compositor/display_cursor' returns true: | 14:54 |
| T42 | <elros34> ? | 14:54 |
| rinigus | @b100dian: video recording works now. I have added fake-package-native to config and its already at obs. all you have to do is zypper ref && zypper up. and then either reboot or start the service | 15:58 |
| rinigus | fake-package-native is implemented using libgbinder. in our case, it was queued regarding android.content.pm.IPackageManagerNative/hasSystemFeature | 16:00 |
| rinigus | if someone has issues due to missing package_native, this service should help. for nagara it was needed for LOS21 | 16:00 |
| rinigus | seems that IPackageManagerNative API is not stable. so, its possible that other android versions may need different code->method map. | 16:01 |
| rinigus | sources at https://github.com/sailfishos-sony-nagara/fake-package-native | 16:01 |
| mal | hmm,.there must be something.wrong in adaptatiom or.somewhere else if it needs that | 17:18 |
| mal | rinigus: when does it start to complain that? | 17:19 |
| mal | I suspect some fake service might be needed in droidmedia | 17:20 |
| mal | some service which depends on some java side stuff | 17:21 |
| T42 | <b100dian> mal: it was implemented in java indeed | 18:01 |
| T42 | <b100dian> Rinigus: thanks! I will test as soon as I get to wifi | 18:02 |
| rinigus | mal: it starts complaining when you try to start recording video (camera or advancedcamera) | 18:12 |
| rinigus | before you hit record button, UI shows camera output nicely. | 18:13 |
| rinigus | as soon as you hit it, it freezes. | 18:13 |
| rinigus | logs are at https://github.com/sailfishos-sony-nagara/main/issues/34 | 18:13 |
| rinigus | mal: in the logs, we have Control message: Could not find 'aidl/package_native' ... | 18:14 |
| rinigus | after implementing this service, I have recording working just fine | 18:14 |
| rinigus | what camera drivers do, they call android.content.pm.IPackageManagerNative/hasSystemFeature and have arg android.hardware.touchscreen | 18:15 |
| rinigus | and few other things with that hasSystemFeature: android.hardware.type.automotive, android.hardware.type.television, android.hardware.type.watch | 18:16 |
| rinigus | it gets a reply and then proceeds with recording | 18:16 |
| rinigus | mal: droidmedia could be a good place as well. I used it as an option to write in gbinder | 18:17 |
| rinigus | so, can be compiled at obs directly | 18:17 |
| mal | rinigus: there have been cases that the error is about package native bt the real service needing it something else | 18:22 |
| mal | rinigus: is that wth camera2? | 18:23 |
| rinigus | mal: no, I had with camera1 or whatever it was before | 18:23 |
| rinigus | b100dian had it in the log below with camera2 | 18:24 |
| mal | I know one issue in camera2 which is similar | 18:24 |
| rinigus | he had "CameraService: No packages for calling UID 100000" | 18:24 |
| mal | can you grab full logcat if you haven't yet | 18:24 |
| rinigus | try to use that service, it may help | 18:25 |
| mal | I know how to solve the camera2 issue | 18:25 |
| mal | it was possible by disabling one feature with a property value | 18:26 |
| rinigus | I don't have camera2. with camera1? without fake service? | 18:26 |
| rinigus | note that it works great right now with that service :) | 18:26 |
| T42 | <b100dian> mal: that prop was automotibe related:)? | 19:30 |
| T42 | <b100dian> *Automotive | 19:30 |
| mal | nope | 19:37 |
| T42 | <b100dian> the playback problems of the recorded video are very similar to sake and presumably fp5 | 19:51 |
| T42 | <b100dian> Starts playing, stutters a bit the freezes | 19:51 |
| T42 | <b100dian> *then | 19:57 |
| T42 | <edp_17> @elros34 : The 'dconf read /desktop/sailfish/compositor/display_cursor' returns blank | 20:16 |
| T42 | <elros34> you must set it to true in sfos 4.4 for cursor | 20:16 |
| T42 | <edp_17> Thanks, that did the trick. I have cursor! Great. :) | 20:18 |
| mal | I need to try to figure out that playback issue again, I know it is caused by not-released buffers | 20:53 |
| T42 | <edp_17> mal, any idea of that backports issue? | 21:00 |
| T42 | <b100dian> mal: regarding the unreleased buffers: nagara does not need changes for browser playback. It works withe Media Buffers enabled. | 21:32 |
| mal | ok | 21:32 |
| mal | need to have a better look at the backports makefiles | 21:32 |
| T42 | <b100dian> So the gallery and browser playback causes may be different. I will try to see gst logs for gallery though | 21:33 |
| T42 | <edp_17> mal: thanks. If you need any input from me about backports, please let me know. (I outputted all variables that .mk file uses and they look sane.) | 22:22 |
| T42 | <elros34> so run make twice with these variables and you will have backport module ready to use | 22:42 |
| T42 | <edp_17> Tried already. :) | 23:10 |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!