| osum4est_ | it looks like the missing .so files are in /vendor/lib64, but test_hwcomposer is looking for a 32-bit version. is this what the problem is? how do i get the 32bit versions/get hwcomposer to work with 64bit? | 01:44 |
|---|---|---|
| r0kk3rz | im not sure its possible to work with 64bit hwcomposer | 02:13 |
| osum4est_ | so it looks like i need to user surfaceflinger instead? is that the surfaceflinger-hack in the faq? | 02:28 |
| osum4est_ | running /system/bin/surfaceflinger just hangs. doesn't crash, but nothing on my screen changes. | 02:38 |
| r0kk3rz | i think thats what minisf is for | 03:09 |
| ghosalmartin | hmm ive setup a free amazon ec2 instance, lets see how porting on it works | 08:05 |
| ghosalmartin | although there is an 8gig storage limit, bleurgh | 08:39 |
| r0kk3rz | sounds like not enough | 08:40 |
| ghosalmartin | so tis $0.11 a gig me thinks | 08:43 |
| ghosalmartin | sfos work usually needs what, 120gigs? :P | 08:43 |
| ghosalmartin | 12 dollars a month seems reasonable, lets see how the build speed is though | 08:45 |
| r0kk3rz | check it can do sb2 before shelling out anything though | 08:46 |
| ghosalmartin | i dont really use sb2 tbh, i offload most of that work to magical obs :P but i will do | 08:49 |
| ghosalmartin | so far its just a ubuntu image | 08:49 |
| ghosalmartin | ill give it a test anyways just to know the limitations | 08:50 |
| T4 | <StillJustGerd> Just a quick check, is the hybris-15.1 tree still broken? | 13:15 |
| mal | no, it should work now | 13:16 |
| T4 | <StillJustGerd> Thanks, I'll set it up then. I was asking because the 14.1 lineage tree for my device is unmaintained, whereas the 15.1 tree is maintained | 13:19 |
| ghosalmartin | r0kk3rz, yo what was the gitlab project for the pipeline? | 14:37 |
| ghosalmartin | ahh found it | 14:39 |
| mal | ghosalmartin: is there a reason why you use so old libhybris in your bullhead OBS project? | 14:57 |
| T4 | <DSstill01> where are sailfish android tombstones located? | 16:10 |
| T4 | <elros34> mal: it might be this commit: https://github.com/libhybris/libhybris/commit/524ca76e0608028d0d64c2197bdbd93fb545b7cf. It cause hang and device reboot | 17:00 |
| T4 | <elros34> for test_hwcomposer | 17:00 |
| mal | @elros34 how does strace look like? | 17:12 |
| T4 | <elros34> mal: https://gist.github.com/elros34/a00fb33ade04077cb46d074481e9aa8b | 17:18 |
| T4 | <elros34> second log | 17:18 |
| T4 | <elros34> appplied few more gralloc related but no change: 8a25b0f 8c9fecd 2845ba3 f4aa7e0 | 17:27 |
| ghosalmartin | that amazon ec2 wasnt ideal, it ran out of memory :( | 17:58 |
| *** OhYash1 is now known as ohyash | 18:17 | |
| osum4est | mal, still haven't been able to get hwcomposer to work. everytime it runs it tells me there are libraries missing then segfaults. the libraries only exist on my phone as 64-bit, i don't have 32-bit ones to use. does this mean i need to use the surfaceflinger-hack in the faq? test_hwcomposer strace: https://pastebin.com/sV6fYi15 | 19:16 |
| mal | sometimes missing libs are not really an issue | 19:16 |
| mal | not sure though in this case | 19:16 |
| mal | looks like it fail when creating some color manager | 19:17 |
| osum4est | mal, heres the strace from minimer: http://pasted.co/8d2c28f8 looks like it gets stuck waiting for android.ha.... | 19:25 |
| mal | check logcat | 19:27 |
| mal | using /usr/libexec/droid-hybris/system/bin/logcat | 19:27 |
| mal | it should tell what it's waiting for | 19:28 |
| osum4est | its waiting for surfaceflinger as well as android.hardware.graphics.composer want me to pastebin it? | 19:30 |
| osum4est | currently i have user@10000 masked. upon boot it says "Waiting for SurfaceFlinger". here's the logcat after booting and then running hw_composer https://pastebin.com/FzqRSHDC | 19:38 |
| osum4est | with GL_PLATFORM set of course | 19:38 |
| wdehoog | mal: elros34: last working libhybris for me is a205c27711f9493d756ddbdc7134251337393c0c | 19:40 |
| wdehoog | 35976c201741cfc21c114615b8524db5367e034a fails with lipstick[1654]: [F] unknown:0 - QPA-HWC: hw_get_module(HWC_HARDWARE_MODULE_ID, (const hw_module_t **)(&hwc_module)) in create returned -2 | 19:41 |
| osum4est | well the faq says to do surfaceflinger hack if logcat says its waiting for surfaceflinger so i'm going to try that | 19:45 |
| mal | osum4est: wait | 19:50 |
| osum4est | mal, alright. waiting! | 19:51 |
| mal | osum4est: build the droidmedia stuff mentioned in chapter 13.2, those provided minisf service, not sure if that could help | 19:51 |
| mal | maybe do the whole 13.2 while you are at it, you will need it later | 19:52 |
| osum4est | will do, thanks! | 19:53 |
| osum4est | and yea i also did "ps aux" and no minisf programs were running. /system/bin/service check SurfaceFlinger also did not find anything. just saw those on an old irc log if it helps at all | 19:54 |
| mal | we disable normal surfaceflinger because it's not needed but we have a mini version of it for some things | 19:55 |
| osum4est | got it. doing 13.2 now | 19:56 |
| piggz | seriously, how do i respond when someone claims my workaround for 3.0.3(installing the old libicu) causes esxessive battery drain | 20:10 |
| kimmoli | powertop | 20:11 |
| piggz | no toher chainges, just the existance of a unicode library causes power drain | 20:11 |
| T4 | <elros34> wdehoog: yeah 35976c2 breaks even getprop but few later commits fix it | 20:34 |
| mal | it's completely possible that individual commits in libhybris are broken and only the complete set works | 20:37 |
| wdehoog | mal: elros34: thanks. now I wonder why libhybris 0.0.5.31 gives me a UI. going to retry .30 | 20:40 |
| mal | wdehoog: maybe because in .31 the debugging support in enabled only if user specifically activates it | 20:42 |
| osum4est | mal, with user@100000 masked, the logs look the same. with it not masked it looks like surfaceflinger? keeps restarting. logcat: https://pastebin.com/W0n5gL7P hwcomposer strace: https://pastebin.com/7Xaq70FQ looks like colormanager is still having problems though. is this due to libsdm-color.so not being found? (found in logcat) | 20:59 |
| osum4est | "ps aux | grep mini" still shows nothing though. not sure if thats a valid test of minisf though | 21:00 |
| mal | osum4est: hmm, now that logs gives maybe some hint, "E qdgralloc: Init: Failed to open ion device - Permission denied" | 21:17 |
| osum4est | yea, i saw that. not sure why file it is referencing though | 21:18 |
| mal | osum4est: in your $ANDROID_ROOT run this: find out/ -name ueventd* | 21:18 |
| mal | osum4est: I have a feeling make hybris-hal didn't build some things that are needed, it happens occasionally | 21:18 |
| osum4est | mal, https://pastebin.com/3Tfdbcsd | 21:19 |
| osum4est | output of the find command ^ | 21:19 |
| mal | osum4est: now check in HABUILD_SDK output of "make modules" and look for ueventd | 21:20 |
| osum4est | mal, ueventd is not in there. closest thing is just "uevents" | 21:21 |
| mal | hmm, odd, pastebin the whole output | 21:22 |
| mal | also pastebin content of out/target/product/marlin/root/ueventd.rc | 21:22 |
| osum4est | mal, https://pastebin.com/9aEkb4eQ | 21:22 |
| osum4est | ueventd.rc: https://pastebin.com/kYtSXCtF | 21:23 |
| mal | so just like I guessed, another ueventd.rc is missing | 21:25 |
| mal | osum4est: do you have some vendor folder in out/target/product/marlin/root/ or out/target/product/marlin/ | 21:27 |
| osum4est | i have vendor folders in both places. out/target/product/marlin/root/vendor is empty. out/target/product/marlin/vendor has a build.prop, default.prop and a lib folder | 21:29 |
| osum4est | in the lib folders there is a modules folder, and in there is just a modules.dep | 21:29 |
| osum4est | and modules.dep is empty | 21:30 |
| mal | ok, as a temporary ugly hack copy file device/google/marlin/ueventd.common.rc to out/target/product/marlin/vendor/ with name ueventd.rc | 21:30 |
| mal | then run build_packages.sh -d and rebuild image | 21:31 |
| mal | I probably need to make a test build with that device repo at some point and figure out how we need to modify build scripts to fix the issue, this is now happening on many devices and should be fixed | 21:34 |
| wdehoog | mal: 0.30 also gave me a UI. then I removed the .jolla-xxx files and then the troubles came back. startup wizard was broken and phone started to reboot | 21:40 |
| mal | wdehoog: what files? | 21:41 |
| wdehoog | with 0.31 startupwizard is works better but still not ok. UI is there but not correct | 21:41 |
| mal | hmm | 21:41 |
| wdehoog | the ones from the startup-wizard | 21:41 |
| wdehoog | tomorrow I'll try to get sme logs | 21:41 |
| osum4est | mal, it worked. i have ui. this is amazing, thanks!!! i need to fix the touch event in my spec file though, touch isn't working - but i think i know how to do that. as far as the ueventd.rc goes, can't i just add it to that sparse folder? or symlink it? | 21:49 |
| mal | osum4est: no, sparse won't help with that | 21:52 |
| mal | osum4est: it's used on droid-hal rpm build, sparse is in droid-config repos | 21:52 |
| mal | different repos | 21:52 |
| mal | osum4est: touch input is fixed in one config file in sparse, you can find it there | 21:53 |
| osum4est | awesome, i got input working and set my pixel_ratio. i have another build problem though, i have to manually add /vendor and /system to my /etc/fstab and make the system symlink after flashing. how can i integrate this into the build process? | 23:06 |
| r0kk3rz | you might be able to add those to sparse | 23:23 |
| osum4est | would i need to duplicate the current fstab and put it there? because currently there isn't one in sparse | 23:49 |
| T4 | <DSstill01> GUYS I MADE ALIENDALVIK 1.0.82 FOR HYBRIS-13.0 BASE TO WORK ON GALAXY A5. IM THE FIRST IN THE WORLD WITH ALIENDALVIK ON AN UNSUPPORTED DEVICE WITH 6.0 BASE, IM A GOD WOOO. | 23:49 |
| T4 | <DSstill01> It had been a difficult work, but finally i did it, i could install and run an android app, esketittt | 23:50 |
| r0kk3rz | osum4est: yeah you would | 23:50 |
| T4 | <DSstill01> And btw it's impossible to make aliendalvik 8.1 work to every device since it's a device-specific system.img that runs with lxc | 23:52 |
| T4 | <DSstill01> But 1.0.82 works after i copied some binder-based stuff from aliendalvik 0.46 (arm for hybris 11.0 only) to make 32bit binder works (because aliendalvik 1.0.82 system requires 64bit) | 23:54 |
| T4 | <DSstill01> [Edit] But 1.0.82 works after i copied some binder-based stuff from aliendalvik 0.46 (arm for hybris 11.0 only) to make 32bit binder works (because aliendalvik 1.0.82 system requires 64bit binder) | 23:55 |
| T4 | <DSstill01> So, after this i can call myself a genius | 23:55 |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!