Tuesday, 2018-12-04

mortalglitchI'm still stuck on boot I get to the telnet 2323, autologin@100000 fails as does a few dev services (dev-bfqio.mount, dev-cpuctl.mount, dev-stune.mount). If I start autologin@100000 using systemctl it will attempt to load the display and then just go to a black screen. Gathering logs now.00:14
malthose mount failures are not a problem00:18
mortalglitchsystemctl : https://pastebin.com/t9M8UBYJ JournalCtl : https://pastebin.com/k8QPzzP6 Dmesg: https://pastebin.com/4dS3vkWg00:20
mortalglitchI tried just in case, I can't seem to get rid of the mount failures at all00:20
mortalglitchI was getting a error droid-hal-init (Failed to initialize property area) so I removed the exit from the core/property_service.cpp and now it seems to go a bit further than it did00:23
malhmm, that error usually means some files are missing, do you have droid-hal detritus installed on the device, check with "zypper se droid-hal"00:27
malmortalglitch: just ignore those mount failures00:28
mortalglitchzypper se droid-hal gives error connection failed could not resolve host00:28
malI think there should be a way to ignore all failures and still get some output00:29
mortalglitchdetritus is in the list00:30
malit's installed?00:30
mortalglitchit shows under installed packages00:30
malmortalglitch: btw, you can only run droid-hal-init once per boot00:31
mortalglitchah, figured I was doing something bad earlier it would just instantly drop me to recovery00:31
malmortalglitch: not sure when you got the property area error, those logs don't show any errors from droid-hal00:32
mortalglitchit was prior to the rebuild, I could re-add the exit and test again to catch them00:33
maldid you do any other changes for the new build?00:34
mortalglitchnot this last round, just patched the properties cpp file00:34
mortalglitchfound an old chat log where sledge had suggested it and you had followed it up with a change to the droid-hal-startup.sh00:35
mortalglitchfor a similar issue00:35
malyes, that was a long time ago and the issue was fixed in another way00:36
mortalglitchok, will revert to see if I can get closer to the underlying issue ty once again00:36
malnot necessary00:39
malsomething is causing floating point errors from some processes, not sure what it could be00:40
malanyway it's late so we need to continue some other time00:45
mortalglitchwhen I run mic it does 100's of floating point exceptions didn't know if that normal00:45
malthat is normal00:46
mortalglitchalright mal thank you for all the help have a great night ^^00:46
*** ChanServ sets mode: +v T402:19
T4<adampigg> @eugenio_g7 , Mal: current status, boots nicely into language selection, size is nice, click next onto license agreement and hangs clicking accept, was gonna fix this morning but not got a USB c cable with me, will try and source one!!07:02
T4<adampigg> All those permission errors seem to have gone after removing the vibration plugin which seems odd07:03
T4<eugenio_g7> @adampigg hybris or mesa?07:22
T4<adampigg> Mesa07:22
T4<adampigg> I just left libhybris installed and over write the 3 conflicting files :)07:24
T4<adampigg> I wondered if mer /Jolla would accept a patch that built the eglfs plugin, which would stop you having to build qt07:24
T4<eugenio_g7> I guess it would be ok if we split the kms eglfs integration in its own package07:29
T4<eugenio_g7> But that would also mean updating mesa on mer (which would be great as it's pretty old, but we should be sure not to break the emulator)07:33
T4<adampigg> @eugenio_g7 did you have an issue accepting the license?07:37
T4<eugenio_g7> Not that I remember07:47
*** ChanServ sets mode: +v T408:34
T4<adampigg> @eugenio_g7 in the main UI now.... Touch seems bad. Did you have that08:57
T4<eugenio_g7> Had it happen sometimes just after the UI start-up, but then it works fine09:17
T4<eugenio_g7> Does the lock screen work? Mine shows only a spinning circle, but I can swipe from the edges and get to the home screen (which works fine)09:18
T4<adampigg> Yes09:20
T4<adampigg> I can swipe sideways ok09:20
T4<adampigg> But can't swipe up for menu09:20
T4<adampigg> Or click icons09:20
T4<adampigg> If I swipe right from the bottom it works09:21
T4<adampigg> But can't click anything09:21
T4<adampigg> There is something to do with touch protocols but I can't find it09:22
T4<adampigg> Mal?09:22
r0kk3rz@eugenio_g7: in the past broken lockscreen is an ofono issue, but i thought that was fixed09:31
T4<adampigg> R0KK3RZ, what's the touch protocol a b stuff?09:37
T4<adampigg> I had spinning icon, but then unmasked ofono09:37
r0kk3rzdoes it has a modem?09:43
T4<eugenio_g7> r0kk3rz oh that might explain it09:43
r0kk3rzofono needs to be running, but otherwise it should be ok i think09:44
r0kk3rz@adampigg: i have no idea tbh, might have to do some digging09:45
T4<eugenio_g7> No, it doesn't have a modem, my ofono is masked (and I thought I disabled the modem from one of the droid-hal spec files)09:45
T4<adampigg> No modem09:45
r0kk3rzyeah thats fine, just configure rilplugin to have no modems09:46
T4<eugenio_g7> Great, will try this evening :) thanks for the pointer09:46
T4<eugenio_g7> Also Sailfish-browser seems broken, pages seem to load but nothing displays09:47
r0kk3rzadreno quirks libhybris option ;)09:47
T4<adampigg> It looks so nice, but my touch is so buggy!!09:47
T4<eugenio_g7> On mesa?09:47
r0kk3rzoh, its not qcom09:47
T4<adampigg> Not usinglibhybris ;)09:47
r0kk3rzoh, cool09:48
r0kk3rzthat could get interesting though09:48
r0kk3rzwith media, cameras, sensors09:49
T4<adampigg> Well, camera doesn't work in lineage anyway!09:52
r0kk3rzyeah that wont help09:52
krnlyngpiggz_, is cache mounted?13:01
T4<adampigg> Krnyng let me check13:02
T4<adampigg> Thoughts on touchscreen weirdness?13:02
T4<adampigg> Btw, actdead charging animation working perfectly... That's usually the last thing to fix!13:04
T4<adampigg> Krnlyng, no, and there is no mount unit for it13:08
T4<adampigg> Fs_mgr tries to mount it tho13:09
krnlyngpiggz_, what's inside it?13:15
krnlyngmaybe you need to add a mount for it13:15
T4<adampigg> krnlyng: https://bpaste.net/show/4341a2051e0113:25
krnlyngpiggz_, not sure if it helps but maybe you can figure out where that hwc.reg comes from13:26
T4<adampigg> Will do13:34
T4<adampigg> Is this for hwcomposer support?13:34
*** ChanServ sets mode: +v T414:00
T4<adampigg> Krnlyng,.the reason the android mounts fail is that the symlinls in /dev are all off by onex but im probably better doing it in systemd anyway..n16:19
T4<adampigg> Got 4 partitions mounted16:19
T4<adampigg> One contains the wifi mac details16:19
krnlyngpiggz_, if you can please fix the udev rules16:20
T4<adampigg> No joy with toichscreen yet tho...its like i can touch things,  but they stay sellected, and cant be deselected16:20
T4<adampigg> I probably need evdev_trace installed16:22
T4<adampigg> Krnlyng, does the rule go in droid-config-dev/sparse...?16:25
krnlyngpiggz_, there are already rules there, you probably have to adapt those16:28
T4<adampigg> Ok16:30
T4<adampigg> Krnlyng, so, should i just leave it to android  init/fs_mgr to do the mounts?16:32
T4<adampigg> @eugenio_g7 it's very odd....it's like all my swipe gestures are handled perfectly, just clicking stuff doesnt work. It almost seems like an app issue. Not a hw/driver issue16:40
piggz_krnlyng: mal: what does this look like? https://bpaste.net/show/75b91728de1618:40
krnlyngpiggz_, touch input18:40
piggz_krnlyng: yes, but it doesnt work correctly....its probably hard to explain without a video!18:46
piggz_im just gonna grab a trace from my mido to see how that looks18:46
piggz_krnlyng: see this from mido https://bpaste.net/show/3acf396ad14d18:48
piggz_notice how the gap between BTN_TOUCH 1 and 0  is a lot less on mido18:49
piggz_could that cause the weird behaviour where swipes seem to work ok, but touches dont?18:49
krnlyngpiggz_, hard to tell what the problem is18:51
piggz_krnlyng: pls watch this... https://www.piggz.co.uk/nextcloud/index.php/s/oSA9MaBAiiyk2kt19:02
piggz_@eugenio_g7 ^^ is this how your touch was?19:02
malpiggz_: so it somehow missed the touch releases maybe?19:05
piggz_mal: seems like that19:08
krnlyngpiggz_, that's really odd, could be that pixel_ratio is one thing, but the delayed touches are weird19:11
krnlyngis the system busy or smth?19:11
krnlyngi.e. does it lag?19:11
krnlyngmaybe some setting in the touch driver? no idea19:12
piggz_krnlyng: no, there doesnt seemt o be lag when dragging19:14
eugeniopiggz_, when I had touch troubles only drag worked19:20
piggz_eugenio: so, similar ish?19:21
eugeniostupid question... have you tried unplugging the USB cable?19:21
eugenioperhaps, I managed only to open the application launcher, but taps didn't seem to register19:22
eugeniobut it went away in seconds19:22
piggz_yes, had cables in/out19:22
malI have had random touch issues on fp2 when some usb cable or charger is connected19:22
mali.e. touch fails completely when using some charger19:23
piggz_ive been disconnected with no change19:23
piggz_the releases are in the evdev_trace tho19:23
piggz_lookin at the trace looks fine19:23
eugeniopiggz_, these are my silica settings19:28
piggz_eugenio: where are those set?19:28
eugeniobut you can modify them on the fly (not from the nemo user as those dconf values are locked)19:30
eugenioif I'm not mistaken there is a section about it on the hadk-faq19:30
piggz_eugenio: 1.5 didint help :(20:07
piggz_eugenio: can you do dmesg | grep ft5x4620:10
r0kk3rzpiggz_: cat the input event and look at the touch output20:24
piggz_r0kk3rz: ??20:25
*** ChanServ sets mode: +v T420:25
piggz_r0kk3rz: evdev_trace looks ok ... catting it gives garbage doesnt it20:25
r0kk3rzbut its a bit like the matrix, if you look at the garbage long enough20:26
r0kk3rzyou can see a woman in a red dress20:26
eugeniopiggz_, https://pastebin.com/WXZhX4mH20:29
piggz_eugenio: looks the same20:30
eugenioI'm still on 2.2.1, but I don't think that should matter20:30
piggz_it does feel like a qt ish issue20:31
piggz_evdev looks all fine20:31
piggz_i wonder if there is any config for lipstick20:31
piggz_eugenio: r0kk3rz: krnlyng: mal: fixed!20:36
krnlyngwhat was it?20:37
piggz_i feel it should be added to the faq!20:37
piggz_i found this https://forum.qt.io/topic/63167/evdev-vs-evdevtouch-touchscreen-driver-linux-embedded20:37
piggz_which says, for this TS, evdev works better than evdevtouch20:37
krnlyngpiggz_, do that :)20:37
piggz_so, i changed the lipstick env line20:37
eugenioI do have evdevtouch though :/20:38
piggz_well, you _had_ weird issued20:38
piggz_but, atleast this works20:38
piggz_i now have a working-ish tablet!20:39
mortalglitchis the lipstick-hack(aka surfaceflinger-hack) from the faq still valid in some cases?20:40
piggz_eugenio: might go back to ratio 1.0 ... get more icons on a row :)20:40
piggz_mortalglitch: that has only been valid for 2 or 3 devices iirc20:40
piggz_not very relevant at all20:41
mortalglitchI saw previously it had been used in some htc builds and I'm having no luck with the display on mine20:41
eugeniolol evdev doesn't work at all here20:41
piggz_eh, how odd20:42
r0kk3rzmortalglitch: does yours need hwc2?20:42
mortalglitchnot certain, how would I verify the requirements for it?20:42
piggz_eugenio: this is my full config https://bpaste.net/show/8fe3a6bfb1dd20:43
piggz_i wonder if the physical size options broke it....20:43
piggz_ill try without20:43
r0kk3rzmortalglitch: what base are you using?20:43
eugenioyup mine is the same withouth the size envvars20:43
eugenioanyway, back working on sensors :)20:44
mortalglitchand Lineage20:44
piggz_eugenio: any progreess?20:45
eugenionope, but I least I got the kernel modules to load (switched to the lineage 13 kernel)20:46
eugenioreading more about IIO to try obtaining events from them manually without sensorfwd20:47
mortalglitchhaha, just tried to make libsurfaceflinger and i'm getting ninja error: /SHARE_LIBRARIES/libvulkan_intermediates/export_includes missing or no known rule to make it20:48
r0kk3rzmortalglitch: i think it was 15.1 that really needed the sf hack20:52
r0kk3rzmortalglitch: whats your issue?20:52
piggz_eugenio: no, the physical env didnt make any difference ... evdev for me!20:53
mortalglitchgetting odd behavior during boot. autologin@100000 fails and occasionally user@100000 fails  I can then systemctl start each of those services and it will switch to showing an all black display20:54
eugeniowonder how it will behave on other Mi Pads20:54
r0kk3rzmortalglitch: tried test_hwcomposer?20:55
piggz_r0kk3rz: on your tablet build, is there a .sepc setting to not install the voicecall stuff?20:56
r0kk3rzpiggz_: its in patterns20:57
r0kk3rzpiggz_: https://github.com/mer-hybris/droid-hal-configs/blob/master/patterns/templates/jolla-configuration-%40DEVICE%40.yaml#L920:58
r0kk3rzmurder it20:58
piggz_r0kk3rz: yeah, found it ta20:58
malmortalglitch: you issue is different level than actual graphics20:58
mortalglitchah, more around the user issue?20:59
r0kk3rzblack screen sounds like hwcomposer is probably working20:59
malpiggz_: his device fails even before it gets to graphics level, some floating point errors from some processes20:59
malr0kk3rz: ^21:00
r0kk3rzmortalglitch: what device?21:01
mortalglitchhtc 10 (pme)21:01
mortalglitchwill grab some newer logs21:03
r0kk3rzyeah, dmesg and journal21:03
mortalglitchdmesg: https://pastebin.com/Ltmqw5xX21:05
mortalglitchjournalctl : https://pastebin.com/LByEPXHq21:06
r0kk3rzmortalglitch: kill time_daemon and ipacm21:12
r0kk3rzfrom the android init rcs21:13
mortalglitchtrying to locate the file (sorry fairly new to this type of building)21:18
piggz_r0kk3rz: can you give broadcom BT tips21:28
mortalglitchr0kk3rz thank you for that21:28
r0kk3rzpiggz_: maybe21:29
piggz_r0kk3rz: so, the device is /dev/ttyHSU021:30
r0kk3rzsounds usual, got yo firmwares?21:30
piggz_i have firmware linked to /etc/firmware/bt/....21:30
r0kk3rzgood good, what name?21:30
piggz_sh-3.2# ls /etc/firmware/bt/21:31
piggz_i cant find a file with a mac address atm21:32
piggz_i have /data/misc/bluedroid and /data/misc/bluetooth but noth are empty21:32
piggz_hangon, it might be in /config21:33
r0kk3rzok cool21:33
r0kk3rzprod it with hci_attach21:33
eugeniopiggz_, could you please pastebin me the output of find /sys/bus/iio/devices/iio\:device*/ ?21:35
r0kk3rzpiggz_: https://github.com/r0kk3rz/droid-config-scorpion_windy/commit/063779b2ff7a899dee8b0fadeb69e9de6b2b77fe21:35
piggz_r0kk3rz: i think im supposed to have a /config/bt/bd_addr.conf but it isnt there21:37
piggz_eugenio: https://bpaste.net/show/6c894e3ccf3f21:38
piggz_r0kk3rz: when i run hciattach i get a time out21:41
r0kk3rziirc you need to disable rfkill first21:41
r0kk3rzwhat args did you give hciattach?21:42
r0kk3rzpiggz_: https://github.com/r0kk3rz/droid-config-scorpion_windy/blob/master/sparse/usr/bin/droid/droid-bt-init.sh21:43
piggz_r0kk3rz: yeah, thats what i did21:44
piggz_but i need that package built21:45
eugeniopiggz_, thanks. Could you please insmod heci.ko, heci-ish.ko, hid-heci-ish.ko, sens-col-core.ko, hid-sensor-hub.ko, hid-sensor-iio-common.ko, hid-sensor-trigger.ko, hid-sensor-accel-3d.ko and hid-sensor-als.ko (in that order) and retry that find command?21:45
maleugenio: did you alread try adding the config I mentioned yesterday?21:46
eugenioyup, nothing changed. But I don't think that sensorfwd is the problem21:47
eugenioI think is in the kernel / my kernel config21:47
mortalglitchr0kk3rz: I disabled the ipacm and time_daemon and rebooted, is there any particular changes I should be on the look out for in the logs?21:48
eugeniolooking on the net, IIO accelerometer devices should expose at least in_accel_{x,y,z}_raw in the iio:device sysfs path21:49
eugeniothey aren't in my case (even though I do have them in sensor_elements as they should)21:49
r0kk3rzmortalglitch: not really, they were just spamming21:49
maleugenio: well based on your log yesterday the sensors were found by sensorfwd21:49
malIioAdaptor::setup() Accelerometer found21:50
maleugenio: without the proper hwsettings.ini the system just won't use the sensors21:50
malat least in the past it didn't21:50
malbut that log from yesterday looked fine21:51
eugenioyes the devices are there, but they miss the objects on which the actual values are reported (in_accel_x_raw, in_accel_y_raw, etc), at least looking at other sysfs tree for other accelerometers exposed on IIO21:52
maleugenio: how did you test those? using messwerk or csd app?21:52
malshow how the sysfs looks for those21:52
eugenioI had also enabled them manually from sensorfw, so perhaps that's why sensorfwd detected them before creating the hw-settings.ini21:53
eugeniohttps://www.kernel.org/doc/Documentation/hid/intel-ish-hid.txt <-- bottom, that should how it should look21:53
piggz_eugenio: https://bpaste.net/show/ef162edddd6121:53
eugeniomal, this how they look in my device: https://pastebin.com/r80Zibfm21:54
eugeniook exactly the same, thanks piggz_21:54
eugeniouh sorry I opened my very same pastebin link :/21:54
eugeniohmm they should be more21:55
eugeniohmmm I do have this kernel oops after loading heci-hish.ko https://pastebin.com/657iJMde22:01
malmortalglitch: just making sure, did you do any modifications to droid-hal-startup.sh?22:10
mortalglitchdouble checking22:11
mortalglitchI had tried previously removing the "exec nohup" from the /sbin/droid-hal-init. But right now it is the default set up22:15
piggz_r0kk3rz: eugenio: i now can enable the BT device...but scanning doesnt find anything22:31
piggz_r0kk3rz: https://bpaste.net/show/d57dfa9ab48822:33
mortalglitchsemi good sign after systemctl starting autologin@100000 and the screen blacks out the buttons react with vibrate feedback and the buttons register in dmsg ^^22:37
malmortalglitch: what did you change?22:39
piggz_r0kk3rz: found a device, but then bt turned off, and now cant find anything...but progress is progress!22:40
mortalglitchnothing this time just rebooted, waited about 2 minutes and then ran "systemctl start autologin@100000"22:40
piggz_oh, it found  a phone22:41
piggz_nice, double-tap-wake works too22:42
piggz_r0kk3rz: is the droid-bt-init.sh script needed with the other config?22:42
r0kk3rzpiggz_: no, i dont use the script anymore22:50
piggz_r0kk3rz: so the service goes too i guess22:50
r0kk3rzthe config is for use with this https://build.merproject.org/package/show/nemo:devel:hw:sony:scorpion/bluetooth-rfkill-event-hciattach22:50
r0kk3rzwhich is a much better solution22:51
piggz_r0kk3rz: yes, i have that now22:52
r0kk3rzbut you should be able to init it manually with hciattach22:54
r0kk3rzgot your kernel config somewhere?22:55
piggz_r0kk3rz: https://github.com/piggz/android_kernel_xiaomi_latte/blob/hybris-13.0-latte/arch/x86/configs/lineageos_latte_defconfig23:07

Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!