Thursday, 2022-09-29

budrzH, I'm trying to build Sailfish OS for Google Pixel 3a XL using the hybris-18.1 branch and I am stuck on section 5.4 of the HADK guide. particularly the make part13:18
budrzHere is my local manifest:
budrzI deleted the "paths" folder because I was getting an error involving GoogleParts:
budrzI applied the patch for hybris:
budrzHere is my make log file:
budrzthere seems to be an  error with BinderService.h:40:72: error: allocating an object of abstract class type 'android::FakeAudioPolicyService'         return sm->addService(String16(SERVICE::getServiceName()), new SERVICE(), allowIsolated,13:26
budrzbut the build continued so I'm not sure if that's really the probelm13:26
budrzI do see some warnings with depmod not being able to open modules.order and modules.builtin but those are just warnings13:29
maljust a moment, I will integrate the fix for that fakeaudiopolicyservice issue13:37
malbudrz: run "repo sync external/droidmedia" and then try again13:55
budrzmal: thanks I will try the fix14:07
T42<hasajacy> im trying to install this on nexus 514:46
T42<hasajacy> and its showing nexus logo for about 2 minutes already14:46
T42<hasajacy> sorry wrong link14:49
budrzmal: the build got a little closer to completing but there's now a ValueError:
budrzIt says "DTB image must not be empty."14:54
budrzSorry, wrong one. Here is the correct onr:
mallooks like there is an issue with kernel build, it uses new header format which assumed dtb to be provided as a parameter, need to check which value to give the parameter15:07
malbudrz: I have a possible fix, wait a moment15:14
malbudrz: go to hybris/hybris-boot folder and apply this patch and see if it helps15:19
budrzmal: thanks I'll apply the patch15:25
ThaodanXperia 10 IV and Xperia 1 IV hav similar thing from what I know.16:01
Thaodandevices older than that just append the dtbo16:01
Thaodanto the kernel image16:01
malThaodan: some device use separate dtbo.img which that doesn't support either, I have possible fix for both16:08
budrzmal: thanks for the help. the build completed successfully16:18
piggzmal: Thaodan: why is there no debuginfo package for gst-droid in adaptation-community?18:26
piggzgmp-droid-debuginfo is there, but no gst18:26
malpiggz: a bug in release scripts that handle those18:33
piggzmal: do you happen to have such package for 0.20211101.0-1.3.1.jolla ?18:33
malpiggz: you can easily build one if you need it18:34
maljust build on community obs from correct tag18:34
piggzwondering if i can without droidmedia18:35
piggzprobably can18:35
malpiggz: droidmedia-devel is in here
*** Daaanct12 is now known as Danct1219:21
mal@elros34 I tested the apex patch, some strange issues, I seem to have some race condition during boot which causes vndservicemenager to load incorrect libs which breaks things20:38
malprobably the linkerconfig bind mounting is not done yet when that starts20:38
T42<elros34> it's done on post-fs-data:
mal@elros34 ah, that is removed by build20:45
T42<elros34> are you sure?20:46
T42<elros34> patch brings bind mount back and it's called in that line20:46
malI mean it's commented out from init.rc20:47
malor am I missing something20:49
T42<elros34> I do not remember enter_default_mount_ns being commented out20:50
malah, it was bootstrap that was commented out20:50
T42<elros34> yes 'mount none /linkerconfig/bootstrap /linkerconfig' so this may affect vndservicemanager20:51
mal@elros34 you don't see any issues related to that?20:52
malI have to kill vndservicemanager once to get UI up20:53
T42<elros34> I have many many issues on my device so not sure, sometimes gui doesn't start so I have to restart lipstick but it's random20:53
malI do also have issues with droid-hal-init timing out20:55
malsomething is making it slow20:55
malbut could be related to vnd20:56
T42<elros34> btw: on my device /linkerconfig/bootstrap/ld.config.txt is same as the one in /linkerconfig/default/. The only difference is  /linkerconfig/bootstrap/*/ld.config.tx files but no idea how this should be handled21:00
malI need to figure out why the boot is so slow21:02
malit's very slow, it takes a lot of time to even get telnet21:02
T42<elros34> maybe some wait_for in init*.rc?21:03
mal@elros34 it took about 1 minute 38 sec for usb interface to get an ip21:05
T42<elros34> did you commented out all include init*usb*.rc?21:07
T42<elros34> your device specific init*rc can still have that21:08
malcould be but that doesn't seem to cause the slowness, everything seems slow, I see some not so nice things in journal21:10
malsome internal storage timeout for example21:10
T42<elros34> sounds bad especially if it happens before droid-hal-init-starts21:12
malyes, and it's /vendor partition timing out21:13
malit does mount it later though21:13
malTimed out waiting for device dev-mmcblk0p32.device and one other also21:14
T42<elros34> maybe some flood in udev: udevadm monitor?21:17
malalso audit seems it active causing a lot of flood21:17
malI now disabled some useless services and disabling audit from kernel commandline, let's see if that helps21:18
maludevadmin monitor seems quiet now21:21
malonly some rare power supply messages21:21
T42<elros34> check if you have 'output lines suppressed due to ratelimiting' this can hide some issues21:22
malI already disabled ratelimiting earlier21:24
malkernel is suppressing a lot of audit messages21:25
mal@elros34 some improvement in boot speed but vndservicemanager still has issues, trying to debug more21:58

Generated by 2.17.1 by Marius Gedminas - find it at!