Saturday, 2019-08-17

ghosalmartinyeah pulseaudio is always crashing for what feels like the same reason as ngfd and qmlthread10:58
malghosalmartin: focus on fixing the android side first11:08
ghosalmartinmal: for what though, the android side seems fine11:08
ghosalmartinmal: for rild? since that seems to be actually working now and i can make calls, but messages fail11:09
malghosalmartin: did you have that qti/qmi issue at some point?11:17
ghosalmartinmal: yeah but think that was a miss diagnosis11:17
ghosalmartinor it can explain why texts fail11:20
malhow does logcat -b radio look when you try to send sms?11:21
ghosalmartinmal: am about to give it another test11:22
ghosalmartini serious need a spare sim to do this with11:23
wdehooghi, I am trying to get working on my phone (oneplusone/bacon/12.1). I have a UI but it is stuttering11:24
ghosalmartin4g also works11:24
wdehoogand lots of errors like:  W/Adreno-EGL( 2198): <qeglDrvAPI_eglSwapBuffers:3709>: EGL_BAD_SURFACE11:25
ghosalmartinmal: texts just worked...this is awkward11:25
wdehoogjournal: , dmesg: , logcat:
ghosalmartinwdehoog, is surfaceflinger accidently running?11:26
wdehoogghosalmartin: no idea. how can I check?11:27
ghosalmartinps -ax | grep surfaceflinger11:27
wdehoognot running11:27
wdehoogon previous 3.0 versions I needed to use old libhybris (5.25) that does not work anymore11:28
ghosalmartinwdehoog, does the latest not work? also do you have the performance stuff in the this file11:29
wdehoogonly gives a UI and allows t finish the startup wizard11:29
wdehoog0.0.5.32 made the startup-wizard crash11:29
ghosalmartinwdehoog, what android base11:29
ghosalmartinahh olddd :P11:29
wdehoogmy droid-hal-device.conf  already has those11:30
ghosalmartinyeah am outta ideas11:30
wdehoogthanks anyway11:30
ghosalmartinmal: it seems rild or the whatever takes a long time to update the UI element11:32
ghosalmartinalso the order to make bluetooth do stuff is11:35
ghosalmartinstart android service, modprobe the bluetooth-power module, then start bluebinder, then when thats running run rfkill to unblock11:36
ghosalmartinbut even then that still doesnt make it work :P11:36
ghosalmartinive gotta go, bye bye11:36
Mister_Magisterany ideas how to fix dying ssh and not working usb networking?12:55
malMister_Magister: is the device using configfs based usb?12:57
Mister_Magisterno idea12:57
Mister_Magistermal: how do i check that?12:58
malkernel config for example13:17
malor .rc files13:17
malMister_Magister: do you have /config/usb_gadget/g1/configs directory on the device?13:35
Mister_Magistersec i'm sorting out things13:35
Mister_Magisteri do13:36
malMister_Magister: which usb-moded packages do you have installed?13:46
malMister_Magister: you should only have the base usb-moded package, none of the config packages13:46
Mister_Magisterusb-moded-connection-sharing-android-config usb-moded-defaults-android13:47
Mister_Magisterremove those?13:47
malyes, remove those of possible, if not then update patterns and also add
malthat goes to spec of course in your case13:48
malthe patterns might prevent manual removal, not sure13:48
malso update config packages just in case13:48
malMister_Magister: which folder do you have inside that /config/usb_gadget/g1/configs ?13:49
malis it b.1?13:49
maland maybe functions13:49
Mister_MagisterUDC  bDeviceClass  bDeviceProtocol  bDeviceSubClass  bMaxPacketSize0  bcdDevice  bcdUSB  configs  functions  idProduct  idVendor  os_desc  strings13:49
mallooks ok probably13:49
malhmm, but no b.1 or c.1?13:50
malwait, which folder is that, is that /config/usb_gadget/g1 ?13:50
malso does the configs subfolder have b.1 or c.1?13:51
malMister_Magister: just for reference how usb was added to xa2
malprobably do the same13:52
Mister_Magistermal: did exact same thing :P
Mister_Magisterlbt: could you please patternize fxtec:t5?13:58
Mister_Magistermal: still no networking14:15
Mister_Magisterdhcp can't get the ip14:16
Mister_Magisterand even if i set ip phone is not in the network14:16
malMister_Magister: but does the device appear in usb at least?14:18
Mister_Magisterye sure14:19
Mister_Magisterrndis is there14:19
Mister_Magisterand nm tries to connects but dhcp doesn't work14:19
Mister_Magisterit was even before removing android14:20
malMister_Magister: what configfs configs do you have enabled in kernel?14:21
malMister_Magister: oh, I thought you didn't have working usb at all14:23
Mister_Magisteri have CONFIG_USB_CONFIGFS_RNDIS14:23
Mister_Magisterstill add CONFIG_USB_CONFIGFS_QCRNDIS?14:23
Mister_Magistermal: nah i do have usb14:24
Mister_Magisterand it's working fine in ramdisk14:24
Mister_Magisterbut not in booted sfos14:24
malMister_Magister: remove RNDIS and add QCRNDIS, assuming you have qcom device14:24
maland see what happens14:24
Mister_Magisterwell okay14:24
Mister_Magisterpro1 has so many things to fix14:26
Mister_Magistermal: side quest while i'm building kernel. wifi not always works i need to reboot couple of times to get it working. any delay hack or something?14:27
malMister_Magister: could it be some timing issue in firmware loading issue?14:29
Mister_Magisterit could dunno14:29
maldo you use the new firmware loading method?14:29
Mister_Magisterfirmware loading?14:30
mali.e. do you have latest config submodule and have you removed any firmware symlinks from sparse?14:30
Mister_Magistermal: i'm at this commit in submodule
Mister_Magisterwhy would i remove firmware links?14:32
Mister_Magisteraaaaaaaaaaaaand? i don't follow ;-;14:36
malwhich means firmware symlinks are obsolete and can/should be removed14:37
malis now it finds the firmware for all usual android locations automatically14:37
mal*it now14:37
Mister_Magisterbut if they are not removed they will break things?14:38
Mister_Magisterno wait14:38
Mister_Magisteri don't have firmware symlinks14:39
Mister_Magistermal: so some delay needed?14:42
malnot sure what the issue is14:43
maldmesg would be useful14:43
Mister_Magisteri will give you dmesg if that helps14:43
malfrom a failure14:43
Mister_Magisterye we thought bout same xd14:43
Mister_Magistercan't find anything reasonable in that14:59
Mister_Magisteralso sensors stop working :P15:03
Mister_Magisteri need to start fixing failing android services probably15:04
Mister_Magistermal: lol now nm connects but ssh doesn't work (telnet does) but after i disconnect and reconnect phone it stops working (telnet only works if i boot with usb)15:07
Mister_Magisterkernel broke when building with changed parameters15:10
malbroke how?15:12
malMister_Magister: maybe you need to configure usb-moded, for example function_rndis value which is by default rndis_bam.rndis and could be different for you15:14
Mister_Magistermal: but how do i configure?15:15
malMister_Magister: read the source
malso you need see which ones you need to modify15:16
Mister_Magisteralso mal i have g1 and g215:17
Mister_Magistermal: but how do i know what is correct option?15:17
Mister_Magisteri'm confuse15:17
malnot sure, I would look at kernel sources15:18
malyou need to see first which kernel configs you should use15:19
Mister_Magistermal: here are kernel sources if you want
malmaybe the android .rc files of your device tell which configfs values you need to use in usb-moded config15:20
Mister_Magisterokay CONFIG_USB_CONFIGFS_RMNET_BAM breaks things15:20
Mister_Magisterit makes kernel do BAAM15:21
Mister_Magisterthere is tons of rc files ;p15:22
malbreaks at build time or runtime?15:22
malhow exactly does it break15:23
Mister_Magisterdrivers/usb/gadget/function/f_rmnet.c:492: undefined reference to `ipa_data_resume'15:24
Mister_Magisterand some others15:24
malyou might need CONFIG_USB_F_QDSS15:37
Mister_Magisterwell with QCRNDIS kernel is not booting15:37
malwhy didn't you just grep the kernel sources and see what is needed15:38
Mister_Magisterwhat do you mean15:38
malthe missing reference to `ipa_data_resume', grepping that then checking the Makefile of that folder for the file it was found easily told which config flag is missing15:39
Mister_Magisterbut i have CONFIG_USB_CONFIGFS_F_QDSS=y15:39
Mister_Magisterit should enable everything required i think15:40
Mister_Magisterwhy do we modify defconfig directly anyway lol15:40
malmaybe you should use the RNDIS and not QCRNDIS15:40
maland adapt usb-moded config15:40
Mister_Magisterdunno how to adapt it though :C15:41
malI told you already15:41
Mister_Magisterby checking kernel15:41
malcheck .rc files for usb initialization15:41
Mister_Magisterdunno what to search for15:41
malit should have a file that handles setting all different usb modes15:42
Mister_Magistermal: i have CONFIG_USB_CONFIGFS_F_QDSS=y and it still gives missing ipa_data_resume16:01
malwell check the kconfig if there is some dependency missing16:05
Mister_Magisteror conflict16:05
Mister_Magisteri hate configfs16:05
Mister_Magistergreat my kernel is not booting to gui16:21
Mister_Magisterokay pro1 goes to sleep untill i get working kernel16:28
Mister_Magisterso mal how is your weekend? i won't bother you with debugging gui on another phone?16:29
Mister_Magisterwait no i updated dcd submodule on 5z not on pro116:46
Mister_Magisteralso mal doesn't look like that udev rule is anywhere near working
Mister_Magistermal: also is not starting17:02
Mister_Magisterit says success but file isn't mounted17:05
T42<DylanVanAssche> In the HADK I found some explanation about hooking up the vibration motor to the `ngfd` daemon. On the PinePhone, the vibration motor is exposed as an input device with FF_RUMBLE and FF_PERIODIC (as mentioned in the HADK). However, the HADK describes the other approach with `ffmemless`. Is there any information available about17:07
T42this approach (with an event input device)?17:07
malMister_Magister: most devices don't have, it's optional17:19
Mister_Magisterstill it's working17:19
Mister_Magisteri need it17:20
malMister_Magister: it's normal to have those DEVLINKS warnings17:20
Mister_Magisteralso mal17:20
Mister_Magistermagic thing happened17:20
malMister_Magister: what is needed in that?17:20
Mister_Magisteri got gui17:20
Mister_Magisterwait lemme explain17:20
Mister_Magisterwhat just happened: i copied keymaster service from /vendor to /usr/libexec in order to add ld reload and then maigaclly gui started working! why? i don't have idea i just coped it and bum it's working17:21
mal@DylanVanAssche this might be good example also needs
malMister_Magister: ok, that was probably preventing android side from starting further and it was just waiting for that service17:23
Mister_Magisterbut that service was segfaulting17:23
T42<DylanVanAssche> @mal Thanks! I will look at those in depth, they are a nice starting point17:23
Mister_Magisterand i added nothing just copied it17:23
malMister_Magister: yes, some services need selinux_stubs preload17:23
Mister_Magisteri did not add preload17:24
Mister_Magisterit started when i was adding it in vim17:24
Mister_Magisterit's without preload17:24
malthat makes no sense17:24
Mister_Magisteri know!17:24
malunless you had keymaster disabled before?17:24
Mister_Magisteri don't think so17:24
malMister_Magister: was that 15.1 or 16.0?17:25
Mister_Magisteri just copied file from vendor and when i was editing it gui booted17:25
Mister_Magisterit's the 5z device17:25
Mister_Magisteri need to mount the file to system17:25
malMister_Magister: ah, ok17:25
malyes, that can be needed then17:25
Mister_Magisterand it's segfaulting again after reboot17:26
Mister_Magistersomething other caused it to work17:26
mal@DylanVanAssche those can probably be copied directly (well renaming might be nice), might need some small adjustments later based on vibrator strength17:27
mal@DylanVanAssche note the filenames in that one file in case you rename the config files17:28
Mister_Magistermal: how can i make droid-hal-early-init working?17:28
malMister_Magister: what is wrong with it?17:28
Mister_Magisterwell it's not working17:28
Mister_Magisternot mounting file17:28
T42<DylanVanAssche> @mal Alright :) I took a look at the src code of the ngfd. If I'm understanding it right; the daemon detects automatically the vibrator by searching for the right event input device.17:29
malMister_Magister: you show no information, messages in dmesg or journal related to that? the actual .sh file you use etc17:29
Mister_Magistertest_hwc2 works17:29
Mister_Magistermal: everything copied from sony17:30
Mister_Magisterstarted manually works, systemctl shows success, no file mounted17:30
mal@DylanVanAssche yes, it should autodetect it as long you have the correct packages installed which are installed by that define in droid-hal-version spec I linked17:30
T42<DylanVanAssche> @mal that makes sense :) I think that line is missing in the droid-config package, need to check :p17:31
mal@DylanVanAssche for reference
Mister_Magistermal: here is the sh file
Mister_Magisterit works when started manually17:32
mal@DylanVanAssche no need to add any packages in patterns for that, only that define to droid-hal-version spec17:32
mal@DylanVanAssche do note that it's the droid-hal-version spec that has the define, not config spec17:33
T42<DylanVanAssche> @mal Oh yeah small difference in naming :p I'm checking now the spec file for the PinePhone specific package17:34
mal@DylanVanAssche if course if you have some different way in use in that pinephone then adapt as needed17:35
maljust make sure correct package is somehow installed17:35
T42<DylanVanAssche> @mal this package isn't normally a device specific device? Like the config one I mean?17:36
Mister_Magistermal: there is nothing aboupt it in dmesg or journal17:36
T42<DylanVanAssche> @mal hmmmm need to build that package, it's not yet in our repos17:42
T42<adampigg> Dylan, need to move everything from my repo to the new real repo18:02
T42<DylanVanAssche> Yeah we have some device repo now18:02
Mister_Magisteroh i hate android18:12
Mister_Magisterand debugging android services ;-;18:12
Mister_Magisteri think i might be missing yet another udev rules ;-;18:16
mal@adampigg shall i move those packages if you are busy?18:48
mal@adampigg hmm, you seem to have removed the packages from your home project18:53
T42<adampigg> Mal, hom3 sub project dontbeevil19:02
mal@adampigg only two packages there19:07
malis that correct?19:07
Mister_Magistertest_hwc2 working but no gui i'm so cloose ;-; but debugging blobs sugz19:14
T42<adampigg> Mal, i deleted everything provided by native-common, so yes19:17
mal@adampigg so kernel is not in that repo?19:21
mal@adampigg it only has config repo and mesa19:22
T42<Zosenko> (Video, 39s)
T42<adampigg> Mal,.no, we arnt building a kernel package yet, doing it manually and copying to sdcard19:22
mal@adampigg is that mesa obsolete also, it should come from native-common19:24
T42<adampigg> Mal, afk :)19:24
mal@adampigg looks like it's the bisect branch19:26
T42<adampigg> Mal, not quite...was trying different revisions19:27
T42<adampigg> Mal, yeah, not resolved the issues yet19:29
mal@adampigg hmm, should I add community-adaptation package to native common or include common in native-common19:29
T42<adampigg> (Photo, 2560x1440)
mal@adampigg do you still use droid-hal-version?19:35
T42<adampigg> We should, just not yet19:36
mal@adampigg just noticed that the hybrisless support to droid-hal-version submodule is not merged19:42
T42<adampigg> Mal,.any reviews / approvals?19:43
mal@adampigg I seem to have approved it months ago :D19:43
mal@DylanVanAssche ffmemless package is available in jolla repo so no need to build it, the issue is that you don't have hybrisless enabled in droid-hal-version20:07
mal@DylanVanAssche but hybrisless build is not yet merged to the submodule so it needs a custom submodule or I need to merge the changes, need to do some testing first20:08
T42<adampigg> Mal, i guess i have it forked20:14
mal@adampigg probably your own repo as submodule20:19
mal@adampigg it's good to remind me of those PRs, I sometimes forget those20:20
T42<adampigg> Mal,.i forget too!20:20
T42<DylanVanAssche> Okay so I need the PR submodule version without hybris, get it to build, add those files you linked and normally the vibrator should start working :)20:25
maland then you need to "%define native_build 1" in spec after changing the submodule20:27
T42<DylanVanAssche> I will try it out tomorrow :) hopefully I get that working already since sensors and other stuff are a dead end atm. But that's for later, vibrator seems to be 'easy', babysteps :p20:28
T42<adampigg> What about wifi?20:28
malwhat is the problem with sensors?20:28
T42<DylanVanAssche> @adampiggz my chip isn't doing anything, while other devkits are doing fine with the dame setup.20:31
T42<DylanVanAssche> @mal I copied the config files over from the native port of Adam but Messwerk always shows 0's of every sensor data.20:32
T42<adampigg> Mal, iio config i think? Dylan, what are the names of the devices exposed by iio?20:32
T42<DylanVanAssche> *same20:32
T42<DylanVanAssche> device0, device1, device2 (maybe with iio in front of it, not sure, don't have the devkit running atm)20:33
T42<adampigg> Dylan, cat decice?/name20:34
T42<DylanVanAssche> I will boot the devkit real quick :p20:35
T42<DylanVanAssche> ```[root@Sailfish ~]# cat /sys/bus/iio/devices/iio\:device1/name20:39
T42<DylanVanAssche> stk331020:39
T42<DylanVanAssche> [root@Sailfish ~]# cat /sys/bus/iio/devices/iio\:device2/name20:39
T42<DylanVanAssche> mpu605020:39
T42<DylanVanAssche> [root@Sailfish ~]# cat /sys/bus/iio/devices/iio\:device3/name20:39
T42<DylanVanAssche> lis3mdl```20:39
T42<DylanVanAssche> `iio\:device0` is the battery charger. It doesn't have the file `name`20:39
T42<DylanVanAssche> You can read the data through the sysfs kernel, but connecting them to sensorfw seems to be an issue20:39
T42<adampigg> Doesnt the latte sensorfw config use names like als to determin the device to use? None of those names match the names used in latte, so look at that20:43
T42<DylanVanAssche> `input_match = accel_3d` must be `input_match = mpu6050` for example?20:45
T42<adampigg> I think....mal?20:46
T42<DylanVanAssche> Just tried it, no difference20:47
malI would guess it's input_match, do those devices have other parameters in that /sys/bus/iio/devices/iio\:device1/20:48
T42<DylanVanAssche> These are all the things that are available for the sensors20:51
T42<adampigg> Device 1 looks like als/prox?20:58
T42<DylanVanAssche> yes, proximity and luminance sensor20:58
T42<DylanVanAssche> Only the `in_illuminance_raw` returns 0 (prox_raw is fine), all the rest of the sensors (raw data) are returning something.20:59
mal@adampigg maybe you can show what latte returns for those  /sys/bus/iio/devices/iio* paths21:05
T42<DylanVanAssche> If something is coming up later, I will check the logs tomorrow21:06
malwe need to see if the issue is sensorfw code or something else21:07
JustGerdI'm unable to follow chapter 6 in the hardware adaptation guide right now, i for some reason can't install any targets. I'm getting the following error: Toolchain 'patterns-sailfish-sb2-armv7hl' doesn't exist or is already installed - can not install.21:12
JustGerdam i doing something wrong?21:12
T42<adampigg> @mal [@adampigg maybe you can show what latte return …], Not atm, drinking mojito ;) will have the latte tomorrow21:12
mal@adampigg no hurry :)21:14
malJustGerd: do you use correct naming mentioned in the warning box?21:16
JustGerdyep. I named the tooling SailfishOS-, and tried to name the target $VENDOR-$DEVICE-$PORT_ARCH which evaluates to motorola-montana-arm6421:17
JustGerdokay i just figured it out: for some reason it tried to use arguments on grep that busybox grep didn't support, installing regular grep fixed it.22:09
malJustGerd: hmm, which command was using those22:13
JustGerdmal: sdk-manage (and sdk-assistant through sdk-manage)22:13
malJustGerd: you were using latest sdk? with clean installation?22:14
malI'll have to do some testing, that is a bug if it really happens on clean installations22:15
JustGerdmal: probably i just upgraded my platformsdk incorrectly or something. it should've been clean, but i might've made a mistake somewhere when installing (like a month ago)22:15
malok, I'll test with clean sdk installation tomorrow and see what happens22:16
JustGerdi was using the latest platform sdk chroot when i installed, which was on july 24th22:16
malI think with that information I can test both upgrading and clean installation and see if there is any difference22:17

Generated by 2.17.1 by Marius Gedminas - find it at!