Sunday, 2020-10-11

T42<maffeen> What should I do?06:05
T42<edp_17> @Parallellhistory : If you are interested in harbour-containers, you can find some info here:
*** monich_ is now known as monich10:33
sambiakHello, is porting sailfish OS to a device that only supports android 10 is impossible ?13:07
malat the moment android 10 support is not yet ready13:07
T42<Anton %lastname%> Is it possible to port native vibrator to 4.9 kernel? I'm not sure how to debug ngfd13:12
T42<Anton %lastname%> Or, perhaps, droid one's fine?13:12
malvibrator should work fine on 4.9 kernel13:17
malit has worked on several devices already13:17
maldroid-vibrator really shouldn't be used13:17
T42<Anton %lastname%> qt5-feedback-haptics isn't used across the default system, am I right?13:20
T42<Anton %lastname%> There's no ffmemless across /dev/input/event*, as evdev_trace reports, and it says something about every one of events, perhaps my kernel config is incorrect?13:28
malmost devices don't support ffmemless13:28
malso do you have native vibrator packages installed?13:28
T42<Anton %lastname%> At the moment, no. Working on it.13:29
malI mean ngfd-plugin-native-vibrator and qt5-feedback-haptics-native-vibrator13:30
T42<Anton %lastname%> I have them both13:33
malso anything in logs?13:46
malI assume you are using hybris-16.0? when did you update hybris-patches the last time?13:46
T42<Anton %lastname%> This wednesday, I believe, I checked out the whole hadk folder13:48
T42<Anton %lastname%> logcat screams that:
malnative vibrator has nothing to do with android side13:50
T42<Anton %lastname%> dmesg has references to vendor.vibrator-1.0 constantly dying13:50
malit uses directly the sysfs interface13:50
maljournalctl is more relevant13:50
malor dmesg13:51
T42<Anton %lastname%> there's "ngfd[4742]: libprofile: session bus connection requested while blocked" right before hwcomposer13:52
maltry what happens if you do "systemctl restart ngfd"14:00
T42<Anton %lastname%> That's fun, there's no such service14:00
T42<Anton %lastname%> Systemctl status, however, reports starting it as a part of user@100000.service14:01
T42<elros34> use systemctl-user14:02
malyes, forgot it was a user service14:02
T42<Anton %lastname%> restarts and runs fine14:03
T42<Anton %lastname%> How should I test it though?14:03
T42<Anton %lastname%> Perhaps it is working??14:03
malthere is the components app which has feedback testing possibility14:04
T42<Anton %lastname%> That means there's no vibration then. At least my sanity is still in place14:06
malcan you manually control vibrator in sysfs14:07
maleither in /sys/class/timed_output/vibrator/enable or under /sys/class/leds/vibrator/14:08
malin latter case you first need to write a value to duration and then 1 to activate14:10
malyou should also check the permissions of those14:10
T42<Anton %lastname%> There are none of those14:10
T42<Anton %lastname%> My kernel is not configured properly then?14:11
malwhat device is that?14:12
T42<Anton %lastname%> motorola moto g7, river14:13
malI assume vibrator worked in android?14:14
T42<Anton %lastname%> Yep, one of the best phones for lineage os14:15
maldo you see any firmware loading issues in dmesg or journalctl?14:15
T42<Anton %lastname%> Other than logcat spamming, because I turned android support for vibrator, no14:16
malthat seems to have aw869x vibrator based on device repo, check that it has that driver enabled in kernel14:16
T42<Anton %lastname%> It just so happens, that I turned it off right before getting here14:19
T42<Anton %lastname%> Building it now14:19
malCONFIG_AW869X_HAPTIC=y that is in lineage defconfig at least14:20
T42<Anton %lastname%> Exactly14:20
T42<Anton %lastname%> I've had a bad read from hadk and hadk-faq14:20
malhmm, where did you see anything like that in those?14:21
T42<Anton %lastname%> All this ffmemless and driver adapting stuff just somehow flipped a switch somewhere14:21
malffmemless is very rare now14:22
T42<elros34> it has 16kB memory so no memless bridge:)14:22
T42<Anton %lastname%> It's the only way, according to hadk and hadk-faq14:23
malbut pattern templates have the native vibrator which makes it the default14:26
malthe ffmemless is mentioned as an optional thing if supported14:26
malfaq even says " you need a kernel haptics driver that supports a memless interface "14:27
T42<Anton %lastname%> I read it as "If you don't want to use droid-vibrator, you better prepare for writing memless haptics driver yourself", especially considering point 114:28
malhopefully it will work now14:30
maldid the dummy_netd help with mobile data?14:30
T42<Anton %lastname%> dummy_netd made everything perfect14:33
maldo you have any broken features left?14:33
T42<Anton %lastname%> Just vibrator14:33
T42<Anton %lastname%> Booted with the new kernel, there is /sys/class/leds/vibrator now, and it works, ngfd, still not so much14:34
T42<Anton %lastname%> Okay, I figured out how to make vibration work, it involved setting permissions to /sys/class/led/vibrator/ contents. I'll add that to udev sparse file tomorrow, as well as play with some configs. After that, the system looks like it will be ready for release19:14
T42<elros34> are you sure you have this patch applied?
malI did ask about that earlier19:20
T42<Anton %lastname%> the init.rc file in out/target/product/*/init.rc has it19:21
maldoes the init.rc on device have it?19:21
xmngot to say sailfish looks to be running very nicely on the pinephone. So thanks to whoever has been working on that.19:23
T42<Anton %lastname%> init.rc on device has it too19:24

Generated by 2.17.1 by Marius Gedminas - find it at!