Saturday, 2018-06-30

rinigussomething is wrong in the current Jolla-latest-SailfishOS_Platform_SDK_Chroot-i486.tar.bz2. tar missing (easy to install though), ubu-chroot missing (which package provides that?). anything else? [was doing full refresh before starting to work on 2.2.x]. nothing seems relevant regarding it in the faq as well06:41
rinigus... or I managed to make a mistake(s) while unpacking platform chroot.06:41
rinigusturns out that android-tools-hadk was not installed in sfos platform06:57
r0kk3rzrinigus: yes theres a few things like that, which is a bit annoying08:58
rinigusr0kk3rz: ok, then at least its not some local install issue09:00
r0kk3rzyeah everyones been tripping up over that one09:02
elrosspiiroin: there is something wrong in mce. When I cover als sensor ( get "2" value from evdev_trace) then lock phone and unlock, brightness jumps for a while (line 231):
elrosso for some reason I get: modules/filter-brightness-als.c: fba_status_sensor_value_change_cb(): sensor: 40012:33
elrosI thought mce use /dev/input/event* for als sensor but looks like it use sensorfw12:36
elroslux data from evdev_trace and HybrisAlsAdaptor::processSample(const sensors_event_t& data) are correct12:44
branek_What should I do to get all .mount services in sailfish os image? I am missing everything from here:
branek_Even if I have: TARGET_RECOVERY_FSTAB := $(PLATFORM_PATH)/recovery.fstab in, .mount files are still missing from rootfs, so no /system, etc...14:02
r0kk3rzif you create your own mount services you should just be able to chuck them in your sparse in your config repo14:11
branek_Ok, my understanding was that should be generated during build time, from .fstab...14:23
r0kk3rzthey should be14:23
r0kk3rzbut clearly they arent14:23
r0kk3rzbut the end game is that you have .mount unit files in your rootfs, so how they get there doesnt really matter so much14:26
malso then you need to make sure the fstab file is found14:29
malbranek_: hybris-boot doesn't know anything about android side fstab variables, I mean that TARGET_RECOVERY_FSTAB is not used in hybris-boot14:29
malhybris-boot tries to search for the fstab files using some rules which might not always work14:30
branek_mal: I see now what you are speaking... Sorry for basic questions, but still getting all pieces...14:37
ghosalmartinhas anyone ever managed to get the kernel booting in qemu and telnet into it?16:05
r0kk3rzshould work in theory16:12
ghosalmartinbut theory is a treacherous thing and becomes see through when you put it on paper :P16:25
locusfhypothesis :p16:26
locusftheories require evidence16:26
ghosalmartinand now i have nothing :P16:27
ghosalmartinits all CONFIG_VTs fault and theres no obvious reasons why :(16:28
ghosalmartinis there anyway to check that a kernel is atleast booting even if it ends up panicing because it seems this kernel is just corrupt from the onset16:29
r0kk3rzyou really need uart access16:30
ghosalmartinso what your telling me is am fooked :P16:31
r0kk3rzbut an alternative is to get a kernel of the same version and diff the CONFIG_VT stuff to see if anything is mangled16:31
branek_mal: fstabs were correctly detected during hybris-hal build: /fstab is on device/oneplus/msm8998-common/init/vendor/etc/fstab.qcom device/oneplus/msm8998-common/recovery.fstab16:32
ghosalmartinr0kk3rz, well its weird since guhl had a working version, but he has one of those massive configs so can't tell whats required and whats not16:32
ghosalmartinr0kk3rz, also is it possible to diff a compiled kernel?16:32
r0kk3rzthat was an earlier version though wasnt it?16:32
r0kk3rzand im sure you can think of a way to dump just the enabled flags from a megaconfig16:33
ghosalmartinwell am sure a lovely grep would hook a brother up16:33
locusfif you really want a binary diff, gunzip the kernels and do a hexdump on them16:34
locusfpipe to diff16:34
locusfor redirect to files16:34
locusfit will help you zilch much16:34
r0kk3rzi guess you also want to pay attention to any set to n as well16:35
ghosalmartintheres like 5000 entries :P16:36
ghosalmartintime for a beet16:37
r0kk3rzpresumably you've tried using that config?16:39
ghosalmartinyeah theres errors all over the place16:39
ghosalmartinand while I can fix some, i think cpu_time_t was deprecated which makes it all a pita16:39
ghosalmartinwell may as well attempt it again16:42
branek_Ok, it looks like rpm/dhd/helpers/makefstab is the right thing to process .fstab and get .mount services...16:44
r0kk3rzghosalmartin: it probably wont help, since you've already identified a problem flag16:44
ghosalmartinr0kk3rz, thing is it doesnt even break like a normal flag though, it seems to just nuke the entire kernel, wont even call init16:46
ghosalmartinno console-ramoops16:46
ghosalmartini mean that must be crashing like super early16:46
r0kk3rzyeah thats not all that surprising16:48
r0kk3rzif you wanted to print the rest of the kernel init process to the VT, you'd want to start it really early16:49
ghosalmartinand am guessing vt is super important i guess itd explain the panic without it16:49
ghosalmartininteresting android-recommends.cfg is that config_vt is disabled16:52
ghosalmartinThe diff you're trying to view is too large. We only load the first 3000 changed files.17:02
ghosalmartinthat not even fair :P17:03
locusfdo you happen to have console=ttyFOOBAR in your kernel command line?17:14
locusfif you do, its quite useless as kernel cannot allocate a terminal to a tty17:15
ghosalmartinlocusf, so just tty?17:57
locusfno, remove it altogether18:26
locusfconsole and all18:27
ghosalmartinyeah no dice there ive had it removed for a while18:36
_svenwhen i'm connected to the gemini via serial console i can see the boot process and after finish that i get a prompt like 130|aeon6797_6m_n: User is "shell". If i try journal -f i get the error "error: object '' from LD_PRELOAD cannot be preloaded (cannot open shared object file): ignored". Any idea how i can see logs? I can't access /run/log/journal/<id>, devel-su doesn't work18:38
Nokiusanything to keep in mind when moving from to ?19:06
r0kk3rzthe csd ini file, but thats it really19:18
maland updating submodules19:23
r0kk3rzthat goes without saying19:26
Nokiussomehow I get confused wehn looing into other repos history19:29
Nokiusmal: why did you set  sparse/etc/pulse/xpolicy.conf.d/fmradio.conf to fmradio.conf.disabled ?19:38
Nokiusyour devices has a fmradio19:38
malNokius: you miss the point of that19:42
Nokiusmal: yes I do, so if I disable it it means it enabled like vise versa?19:43
malNokius: I created a symlink that doesn't have .disabled as the file ending so the parser finds the file, if there is only the file with .disabled ending then parser skips that and fm radio is disbaled19:43
malNokius: it only parses files ending with .conf so the default situation with only fmradio.conf.disabled means fm radio not enabled19:45
Nokiusokay thanks19:45
r0kk3rzmal: for what purpose?19:49
malr0kk3rz: ?19:52
piggzmal: comments welcome
* Nokius it's hard to keep track of port19:53
Nokiusmal: why did you remove sparse/lib/udev/rules.d/999-droid-vidc.rules ? this is not in dcd?20:01
malNokius: you should learn to check submodule commit history :)20:06
NokiusI do right now :D20:07
malif I remove something it's quite common it's because of some change in submodule20:07
malpiggz: we'll need to discuss the naming of the dbus service with people from jolla20:08
r0kk3rzdoes sensorfw do dbus?20:10
Nokiusmal: I belive you do so, I was just couriuse thats why I asked ;)20:24
malNokius: did you really need to led configuration? I only have it because my device has some issues with led20:31
malNokius: it's not good to just randomly copy things20:31
malsame for the vibrator configuration, did you have very weak keypad feedback in the past?20:33
Nokiusokay thanks20:38
malNokius: you didn't answer my question20:44
malNokius: I need custom vibrator config because feedback from virtual keyboard was not noticeable20:45
Nokiusmal: no we faced no issue with the vibrator20:45
malNokius: the led fix was because the blinking didn't work correctly because of a kernel bug20:45
r0kk3rzhmm i might try that on scorpion20:45
malNokius: so you can feel the vibration when you type with virtual keyboard?20:46
malr0kk3rz: which one?20:46
r0kk3rzvibra config20:46
Nokiusmal no I dont20:46
malr0kk3rz: ok, I used the values from xperia x repo as the base and then adapted until I had the same feedback as on xperia x20:47
Nokiusmh obs gives a No space left on device20:48
T4<abhishek_0> i successfully created 64bit sources ... soon i'll start porting with 64bit sources as base20:48
malNokius: usually there should be a small feedback from virtual keyboard20:48
r0kk3rzyeah pretty sure i broke obs by building new python :)20:49
Nokiusmal: when I type I feel nothing20:49
malNokius: maybe test with different values in the feedback config to see if something works, maybe start from this
Nokiusmal: okay20:50
Nokiusmal: thanks20:50
Nokiusmal: increasing Press ( I assume it's the one when typing) to an higher value did help I still feel nothing20:58
Nokiusah looks like the deivce is vibrating after type for a while21:02
piggzmal: good topic for meeting ... would u like to bring it up? i often cant make them21:04
piggzyou can link to the repos21:04
r0kk3rzdoes it really need its own service?21:05
TAMplus39Can anyone tell me how and where to symlink mixpaths.xml?22:23
malTAMplus39: what is the problem, it should be quite cleat that you need to create some symlinks in /etc which point to file in /system/etc22:46
mal*files in22:46
TAMplus39It always says permission denied22:54
TAMplus39And im directing it to where the device tree has the file22:55
malwhat exact command are you running?22:56
malwhat do you mean?22:56
TAMplus39That I dont know what system/etc folder you refer to22:57
malyou don't seem to understand how symlinks work?22:57
TAMplus39I only used time ago on OS X really dont know them on linux22:58
malso either on device (in /etc folder) or in config repo sparse (in $ANDROID_ROOT/hybris/droid-configs/sparse/etc) you run ln -s /system/etc/some_filename22:59
malnormally it's ln -s target linkname but since the link name we want is the same as the original filename and we run it in the correct folder we can skip link name from the command23:01
TAMplus39Oh, that makes it easier23:02
TAMplus39I was running it as ln -s /system/etc/mixer_paths.xml /etc/mixer_paths.xml23:02
TAMplus39But running ln -s /system/etc/mixer_paths.xml23:02
TAMplus39Did the trick23:02
malare you sure?23:03
malthe command you gave works also23:03
malTAMplus39: or did you mean you ran that in your build enviroment and not on the device?23:04
TAMplus39Build environment23:04
malthen of course you cannot use the absolute path for the link name23:06
TAMplus39mal: Oh and also I have another question, Building wlan as module is the only way to get wifi working?23:08
T4<adampigg> r0kk3rz, I dont know, thats why im opening up the debate ... seems like a usefull function that allows integration with sports type apps easily23:09
TAMplus39Because enabling kernel modules derps kernel building23:10
malTAMplus39: depends on the device, not all device need it as module but many do23:12
TAMplus39Then I guess I have to get more knowledge on building android kernels to get modules enabled without error :/23:20
malwe really can't help if you don't show the errors23:25
malbut anyway it's very late now so not today anymore23:25
TAMplus39I posted it on here a few days ago:c23:27

Generated by 2.17.1 by Marius Gedminas - find it at!