T42 | <Hound77> @adampigg [<reply to media>], Nokia ? | 05:02 |
---|---|---|
T42 | <adampigg> Gm | 05:51 |
T42 | <adampigg> Mal, if youre busy, i could implement activity manager in droidmedia....its only 6 methods | 06:04 |
masha11 | When starting lipstick i get the message: "ServiceManager: service 'display.qservice' died".How to fix it? lipstick+logcat: https://pastebin.com/fe8eRDh2 , strace lipstick: https://paste.ubuntu.com/p/ctvy6bDJH3/ | 07:28 |
masha11 | After running test_hwcomposer, the device (griffin) reboots and after running minimer, doing nothing. This is normal?est_hwcomposer+logcat: https://pastebin.com/fLpu4riD , strace test_hwcomposer: https://paste.ubuntu.com/p/sV5zwBXTZJ/ | 07:33 |
masha11 | dmesg: https://pastebin.com/dA6DGNvEjournalctl: https://pastebin.com/5X9v56NDsystemctl: https://pastebin.com/rNbwrfTz | 07:33 |
mal | masha11: do you have all needed symlinks so firmware files are found, including the ones mentioned in faq | 07:39 |
mal | based on logcat most things are broken in your android side | 07:40 |
masha11 | mal, I think yes. How to check it? | 07:41 |
mal | there is no exact rule what is needed, it depends on the device and android base | 07:46 |
mal | first check the faq for 14.1 porting notes | 07:47 |
masha11 | mal, my symlinks for services and firmware files: https://pastebin.com/cgFPAHmC | 07:49 |
mal | masha11: have you checked the android-config.h header in droid-hal rpm, what does it contain? | 08:08 |
mal | does it have QCOM_BSP in it | 08:08 |
masha11 | mal, %define android_config \ #define MALI_QUIRKS 1\. I'll add QCOM_BSP now. | 08:19 |
mal | masha11: is that device using mali gpu? | 08:22 |
mal | masha11: based on lineage info page that is a qcom device | 08:23 |
mal | masha11: https://github.com/LineageOS/android_device_motorola_griffin/blob/lineage-16.0/BoardConfig.mk#L38 | 08:23 |
mal | masha11: how did you think that needs mali quirks? | 08:23 |
mal | masha11: you don't add it directly to that file, please answer my questions first | 08:24 |
masha11 | mal, sorry, I mixed it up. My core does not use Mali. | 08:26 |
mal | masha11: so read faq lines 888-> | 08:26 |
masha11 | mal, ok, thanks! | 08:26 |
mal | masha11: remove any mention of mali quirks from that spec if you added something there | 08:27 |
masha11 | mal, ok, deleted. | 08:27 |
mal | masha11: before running build_packages.sh remove some sources | 08:28 |
mal | masha11: to be sure it rebuilds everything | 08:28 |
mal | from hybris/mw folder delete the folders qt5-qpa-hwcomposer-plugin and libhybris | 08:29 |
masha11 | ok | 08:30 |
mal | at least those, it doesn't hurt to remove the whole hybris/mw folder either | 08:30 |
masha11 | yes | 08:30 |
mal | sometimes it doesn't rebuild things properly if only some define has changed | 08:31 |
T42 | <adampigg> Mal, u listen to cppcast? Current episode on cmake is relatively interesting | 09:05 |
mal | I have never even heard of such a thing | 09:13 |
mal | @adampigg if you want to implement that then do :) | 09:22 |
T42 | <adampigg> Mal, ill give it a go..... | 09:23 |
T42 | <adampigg> Cppcast is a podcast for c++ | 09:23 |
liux | mal, May I apply permission for uploading sfos roms to http://images.devaamo.fi/sfa ? | 14:23 |
mal | liux: it seems the person giving permission is not here anymore, the nick is tbr, seems to be on #nemomobile at least | 14:32 |
liux | mal, ok. just ping him there. thank you | 14:38 |
mal | liux: I think you understood that you should not have vendor blobs in the images you share there, you can check that in the droid-hal rpms, the ones build from open source are ok but not ones that might come from vendor repo, usually those do not end up in image but there have been some cases that blobs are copied and in such case build needs to be fixed | 15:24 |
mal | for example I had to edit some scripts to make sure the blobs do not end up to image in fairphone 2 build | 15:25 |
liux | mal, ok. i'll check droid-hal rpms | 15:29 |
liux | mal, Just go through all rpms in droid-hal-aries but seems no vendor blobs there. could you help show your scripts? | 15:40 |
mal | liux: fairphone 2 has a special vendorsetup.sh in vendor/ repo | 15:46 |
mal | so I just hacked it to only copy needed headers and stuff | 15:47 |
mal | liux: I checked your droid-hal rpm, looks fine | 15:48 |
liux | mal, sounds good. thank you. just a little confuse. I have no qcom vendor blobs in my hadk source but there are vendor blobs built in cm rom (i've built cm-11.0 myself). do you think it's all right for sfos? | 15:53 |
*** ahjolinna_ is now known as ahjolinna | 15:56 | |
mal | liux: some devices need vendor repo to be present when building, some don't | 16:01 |
kimmoli | you need to find another way to distribute that cm-11.0 base (if it needs to be installed prior sfos) | 16:23 |
mal | liux: ^ | 16:31 |
KALUBE | mal: I fixed wifi on j3xnlte, there was a binary in /system/bin that needed to be executed, adding the .rc to sparse fixed it. I see there isn't much documentation (or any?) on Spreadtrum devices, would it be helpful to add a section to the FAQ? | 16:54 |
KALUBE | I will get other stuff working first I think... | 16:54 |
liux | kimmoli, some developers distribute cm roms on servers provided by android community. do you think it's right way? | 16:54 |
mal | KALUBE: nice | 16:54 |
mal | liux: that is the proper way | 16:54 |
mal | liux: for the cm rom you build | 16:55 |
mal | KALUBE: you can add things to faq, it's good to have all there so others can benefit from it | 16:55 |
liux | mal, glad to hear that. | 16:56 |
mal | liux: why do you need custom cm build? | 16:56 |
mal | liux: or are the original builds gone already? | 16:56 |
mal | KALUBE: not sure if you need to add a .rc file, maybe you can just add call to the binary to the same wifi systemd service you use to modprobe it, assuming you modprobe it like that | 16:57 |
mal | but either way probably works | 16:58 |
liux | mal, gone already, and i also need patch in frameworks to make camera work just as abranson said. https://github.com/liuxdev/android_frameworks_av/commit/e14ced03854f72aad2e3b3d17ddd0f2875862d7a | 16:59 |
liux | mal, it can be patched in hadk but so many .so need to reconfigure during build and need a script to replace original .so. before first boot | 17:01 |
mal | liux: that can be handled in sfos side also like I have done it for fp2 | 17:01 |
mal | no need to replace .so files, it loads those from /usr/libexec if done in a certain way | 17:02 |
mal | liux: https://github.com/mer-hybris/android_system_core/commit/3456e019173b426927baf606d7b817d6ffdb7fdd | 17:03 |
mal | that patch can be used to make sure the patched frameworks/av is used | 17:05 |
mal | without the need to use patched android system.img | 17:05 |
liux | liux, yes, but another question is original builds gone already. so I built it myself.... | 17:05 |
KALUBE | mal: I figured it was easier as it can be added to sparse files or ideally included in the script that creates all the .rc symlinks from the FAQ. It has to be called at least 10 seconds before insmod-ing the wifi module otherwise it will fail... Spreadtrum drivers are rough | 17:07 |
liux | mal, thank you for your patch it make easier for my porting work next time | 17:07 |
mal | KALUBE: ok, that could all be done in systemd service as those can contain multiple commands in one service, but rc file works also, I personally prefer systemd service but it's a matter of opinion | 17:12 |
KALUBE | mal: I suppose it's good to avoid using android methods where possible. And you're right it is simpler to implement. I'll check that it works that way and probably use that for future builds | 17:16 |
piggz | mal: ping | 20:54 |
mal | piggz: pong | 20:56 |
piggz | mal: my activity manager isnt working, looking for suggestions..... | 20:57 |
piggz | https://paste.gnome.org/pp3mseoc0 | 20:57 |
piggz | it doesnt print the "Starting..." message in logcat | 20:58 |
piggz | it definitelt compiles that line as i added a #warning inside the #if | 20:59 |
mal | piggz: hmm, maybe try ALOGW? | 20:59 |
mal | piggz: also have you added any debug logging to the fake service to see if any of those get called? | 21:00 |
piggz | ok | 21:00 |
mal | ALOGW is better since at least then it should always be printed | 21:02 |
piggz | mal: nope, nothing in logcat....i killed minimediaservice to allow it to be restarted | 21:13 |
piggz | mal: binary is def new... | 21:15 |
piggz | sh-3.2# strings /system/bin/minimediaservice | grep activity | 21:15 |
piggz | Starting fake activity manager | 21:15 |
piggz | Getting fake activity manager name | 21:15 |
mal | piggz: I think that should be before cameraservice, not after | 21:25 |
mal | piggz: if it's cameraservice that hangs? | 21:25 |
piggz | ah maybe | 21:26 |
mal | also before mediaplayerservice just in case | 21:26 |
piggz | i have,,, | 21:27 |
piggz | it also wasnt printing the message later on in the code, so i did think it must be stuck somewhere, or logging was horribly broken! | 21:27 |
mal | piggz: also explains the issue you had with sensormanager? | 21:28 |
piggz | yes | 21:28 |
piggz | mal: ha, it works too! | 21:32 |
piggz | mal: in the logcat, it only calls registerUidObserver | 21:34 |
piggz | i will tidy up the patch and PR it tomorrow evening | 21:34 |
piggz | but, feel free to try it out | 21:34 |
mal | piggz: sometimes the mistake is quite small and difficult to see | 21:38 |
vknecht | need more eyeballs :-) | 21:39 |
mal | piggz: the onTransact is not needed at all then? | 21:40 |
mal | I mean does it get called often? | 21:40 |
T42 | <adampigg> mal, only twice in the log | 21:45 |
T42 | <adampigg> gnight | 21:45 |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!