Wednesday, 2019-12-18

T42<edp_17> When do you know we can use 3.2.1.20 version to build?00:00
r0kk3rznot yet00:05
T42<edp_17> I have flashed the new image and got a boot loop.00:06
mautzoneedp_17:Then flash the old image and copy the files over ssh. Don't forget o set the password for nemo.00:11
T42<edp_17> I can do that but the 'qt5-qtdeclarative-qmlscene' package will be missing. Isn't a problem?00:12
mautzoneedp_17: Wait a sec00:14
T42<edp_17> Thanks.00:15
T42<edp_17> Ok, I have pushed the minimer3.tar.gz and all three unzipped files to the device.00:17
mautzoneedp_17: http://www.mediafire.com/file/1f1emxj933td64j/qt5-qtdeclarative-qmlscene-5.6.3%2Bgit10-1.6.2.jolla.armv7hl.rpm00:19
T42<edp_17> Thank you!00:20
T42<edp_17> Ok, qt5-... is installed on device.00:25
T42<elros34> few sugestions: use strace with -f option, grab full logcat which include early boot messages and keep it running when testing hwcomposer/minimer/surfaceflinger. Also unlikly it will work because your device is not that old but you can try EGL_PLATFORM=fbdev00:25
r0kk3rzelros, the previous port for this device did manage to use hwcomposer00:28
r0kk3rzso i wouldnt bother with fbdev00:28
T42<edp_17> True, I have installed DRiim's image and worked. But old. :-(00:29
T42<elros34> right so maybe its not worth to try00:29
T42<edp_17> I ran this: 'EGL_PLATFORM=hwcomposer /usr/lib/qt5/bin/qmlscene -platform hwcomposer main.qml'00:29
T42<edp_17> and got this: 'library "libgui.so" wasn't loaded and RTLD_NOLOAD prevented it00:29
T42<edp_17> Segmentation fault'00:29
T42<edp_17> I ran as root.00:30
T42<edp_17> And the same message if I run it as nemo.00:31
T42<edp_17> Isn't this the problem: 'start-user-session@USER.service not-found inactive dead'00:34
T42<edp_17> Or this is just a symptom?00:35
r0kk3rzunrelated00:35
T42<edp_17> ok, thanks.00:35
T42<elros34> and as yesterday you should mask user@100000.service00:36
T42<edp_17> Yeah, at this very moment I scrolled up for the command and wanted to ask. 👍00:37
T42<edp_17> Done00:37
T42<edp_17> I ran 'EGL_PLATFORM=hwcomposer /system/xbin/strace /usr/lib/qt5/bin/qmlscene -platform hwcomposer main.qml' and got: https://pastebin.com/eVXpkB3j00:52
T42<edp_17> It looks like the problem with the gralloc files.01:00
T42<edp_17> It is looking for the gralloc.exynos5.so file in /usr/libexec/droid-hybris/system/lib/hw/. There is no hw folder in /usr/libexec/droid-hybris/system/lib/.01:04
T42<edp_17> Should I create this folder and put a symlink to the gralloc.exynos5.so?01:04
T42<edp_17> Plus into /vendor/lib/hw/ and /odm/lib/hw/?01:05
T42<edp_17> It didn't help.01:08
T42<edp_17> The gralloc.samsungexynos8890.so is missing from the device. Maybe this is the issue?01:08
T42<edp_17> I just ran an 'strace systemctl --user restart lipstick' and got: https://pastebin.com/iyVE76qC01:34
T42<edp_17> Maybe this is red herring but I don't have '/sys/fs/selinux' and '/sys/fs/kdbus/0-user/bus'01:34
T42<edp_17> The only folders I have in /sys/fs: cgroup    ecryptfs  ext4    fuse  pstore01:34
r0kk3rzit will often look for a blob in several directories02:05
r0kk3rzbut you see it does eventually find one02:05
mautzoneIf driims version worked correctly, maybe you could try to downgrade some of the packages?02:09
mautzoneqt5-qpa-hwcomposer-plugin to be exactly.02:12
mautzoneI would try revision bb95d09b893761c25409363e15f7048739c436ba02:17
T42Richardson %lastname% was added by: Richardson %lastname%02:22
T42恭然幄 %lastname% was added by: 恭然幄 %lastname%02:56
T42<谭渊轸 %lastname%> (Photo, 591x408) https://irc.thaodan.de/.imgstore/qj1kGb9BeZ.png 非小号Mytoken推特脸书增关注,菠菜棋牌推广霸屏强推APP内置广告,微信电报增粉私聊、推广,交易所上币投票,交易所平台账号代实名,如有需要请联系:➡️    @yx0000103:01
T42程蓉棠 %lastname% was added by: 程蓉棠 %lastname%03:38
T42<edp_17> @mautzone: Thank you. I was not aware that I could downgrade only some packages. How can I do that?07:57
T42<edp_17> How can I go back to that revision? Thanks.07:57
Mister_Magisterninja: error: 'libhwc2_compat_layer', needed by 'hybris-hal', missing and no known rule to make it08:22
Mister_Magisterummm what. hybris1508:22
Mister_Magisterah wait08:26
Mister_Magisteri got it08:26
Mister_Magisterwtf i cloned libhybris to external and it didn't help08:31
Mister_Magisterwhat is this even08:31
Mister_Magisteroh submodule is missing08:33
jusaPSA: pull request 258 to droid-hal-device will break compatibility with 3.2.1 release. there exists now branch "upgrade-3.2.1" in the repository which is from the 3.2.1 release level. adaptations should use this branch until the next sailfishos release. if you have some patches you need in your adaptation please make sure they get to both master and upgrade-3.2.1 branch. reasoning for09:23
jusathe change is easier and better handling of hw adaptation groups for users, and supporting both post 3.2.1 and earlier would just cause kludges bound for failure09:23
jusahttps://github.com/mer-hybris/droid-hal-device/pull/25809:24
jusaI will add this information to hadk-faq as well09:25
Mister_Magisterjusa: so master is now for post 3.2.1 versions?09:28
jusaMister_Magister: yes09:28
Mister_Magisterokai09:30
deathmistjusa: mal: I feel like dhd upgrade-3.2.1 branch should be at 482ea1852ebabf578799a0a1fad00f6e1284eae2 to make droidmedia & audioflingerglue builds work locally in online mode as well because it was broken before :p10:17
jusadeathmist left already, but as a comment to ^ that one, those patches need to be cherry-picked to upgrade-3.2.1 branch. IMO better to have 3.2.1 as the split point and just take patches as needed10:54
jusaunless the history is 1:1 until then, haven't checked10:55
jusaI mean, if all the commits are really necessary :) I'll stop talking now10:56
Mister_Magisterargh why new devices have to be so annoying11:35
Mister_Magistermoto z is not booting my kernel for no reason11:35
T42<adampigg> hmm, my double negative circuit just activated!11:47
Mister_Magisterthat's not double negative11:51
vknechtanything obvious to you in these straces for test_hwc2 https://pastebin.com/2UZ8u5Jn and test_hwcomposer https://pastebin.com/9EnAxzXD ? (suzu, aosp9 base)12:11
vknechttest_hwc2 just seem to loop, waiting for these:12:12
vknechtdroid-hal-init: Received control message 'interface_start' for 'android.hardware.graphics.composer@2.1::IComposer/default' from pid: 2693 (/system/bin/hwservicemanager)12:12
vknechtdroid-hal-init: Could not find service hosting interface android.hardware.graphics.composer@2.1::IComposer/default12:12
vknechtand this is logcat messages related to test_hwcomposer: https://pastebin.com/8M5XxUjt12:12
malvknecht: the real error is somewhere in logcat probably "Waited one second for android.ha"12:19
malvknecht: no need to use test_hwcomposer, test_hwc2 is the one to use on new bases12:20
vknechthttps://pastebin.com/HyX0zGuS : it's waiting for android.hardware.graphics.composer@2.1::IComposer/default12:23
vknechtthe list of filenames containing composer: https://pastebin.com/bxnpTBKi12:30
vknechtam I missing something, or should some be overridden in sparse ?12:30
Mister_MagisterFFS LENOVO12:33
vknechtthat the android.hardware.graphics.composer@2.1-service.rc : https://pastebin.com/Jibp98Sk12:37
Mister_Magisterzhxt: yo ping, help me with griffin please it's not even trying to boot my kernel ;-;12:38
zhxtMister_Magister: give me seconds to find the logs, I hadn't play with it for a looooooong time  @.@12:47
Mister_Magisterzhxt: kay12:51
Mister_Magisterzhxt: it's weird because it's not bootlooping or anything it freezes before booting kernel even12:51
Mister_Magisterzhxt: i built lineage15 for my port even and it's booting fine12:54
r0kk3rzno wai13:00
Mister_Magisterr0kk3rz: this isn't funny13:03
zhxtMister_Magister: finally found my logs, https://gist.github.com/zhxt/0dae076b73ec711d3ba7610957e5cd0713:05
zhxtIIRC, I use lineage 14.1 at that time, not 1513:06
Mister_Magisterye13:06
Mister_Magisteri'm doing it on 15 cause 14 wasn't there or something13:06
Mister_Magisterhmm i forgot cgroups but it still should boot13:06
zhxtfreezes before the kernel?  bootloader is broken?13:13
Mister_Magisternah it's not broken13:14
zhxthmmm,13:16
T42<elros34> maybe broken CONFIG_VT?13:21
Mister_Magisterwhat do you mean broken?13:21
Mister_Magisterconfing_vt is y13:22
T42<elros34> commit in kernel which crash kernel very early so you can't grab last_kmsg when CONFIG_VT is enabled13:22
Mister_Magisterkernel doesn't crash13:23
Mister_Magisterit's not even booting13:23
T42<elros34> do you have uart and early debug enabled in kernel so you know it's not booting or just guessing?13:24
Mister_Magisterhow can i have it. Obviously i don't have uart13:25
Mister_Magisterwhen anything is booting there is vibration and also i would get any thing on usb / at least reboot/bootlooop13:25
T42<elros34> some devices have it via jack port, anyway you can just disable config_vt and yo will know13:25
T42<elros34> this is the faulty commit which cause issue on other device: https://github.com/LineageOS/android_kernel_google_marlin/commit/1bb88fcb4961166195e2cb1973901ecf931983a113:27
Mister_Magisterwait a minute it's 15.113:29
Mister_Magisterit might be because of lacking android and spl version13:29
Mister_Magisterbut device didn't came with 15.1 so that's unlikely13:30
zhxtsince lineage 15 is booting fine, I guess it maybe something wrong with kernel configs , but not sure which one is.13:34
Mister_Magisterzhxt: sfos kernel is different than lineage one not just by configs13:35
Mister_Magisteri will try comparing hex13:35
Mister_Magisterzhxt: you were using rendering on surface flinger? hwc wasn't working?13:42
zhxtnope, both work :P13:49
Mister_Magisteryep13:49
Mister_Magisterandroid and spl version are in lineage bootimg13:49
Mister_Magisterand since bootloaders are hecking slowflakes it might cause the problem13:50
Mister_Magisterit's not spl either14:01
KabouikI thought the fingerprint part of SFOS was closed source and couldn't be ported to other devices, but in the libhybris table (https://wiki.merproject.org/wiki/Adaptations/libhybris) I see a few phones with working fingerprint15:04
KabouikDid I misunderstand?15:04
Mister_Magisterit's proably error on website15:07
deathmistKabouik: there was a new column added to the table at some point, the very old ports that haven't been updated in a while have incorrect tables15:10
KabouikSad. :< Thanks for your answers15:17
*** lynxis_ is now known as lynxis20:01
vknechtMal, found a piece of the puzzle, init.loire.rc disables hwcomposer service in early-boot and waits for a surfaceflinger "running" property to start it:20:02
vknechthttps://github.com/sonyxperiadev/device-sony-loire/commit/75811b2c2d85cc8147e1b4d6b29d22bfb5ac2c8420:02
vknechtstarting the service manually from telnet gives these logcat message, then crash/reboot: https://pastebin.com/Wr7Uqjb420:03
vknechttried just commenting the hack but not working/sufficient20:06
vknecht(minisf is running)20:07
Mister_Magisterzhxt[m]: ramdisk was empty :P20:12
Mister_Magisterfoken CONFIG_VT21:04
Mister_Magister@elros34 u were motherfucking right partially21:04
Mister_Magister@elros34 so if i revert this https://github.com/LineageOS/android_kernel_google_marlin/commit/1bb88fcb4961166195e2cb1973901ecf931983a1 it gonna work?21:06
T42<elros34> yes, it should be ok21:06
Mister_Magisterokay thanks21:06
Mister_Magister@elros34 for the info, second part was empty ramdisk :P21:07
Mister_Magister@elros34 well gotta search more error: 'struct tty_port' has no member named 'worker_thread'21:15
Mister_Magisterwaaait patch didn't load whole21:17
Mister_Magisterjesus so annoying21:17
T42<elros34> worker_thread is part of that patch so you are mixing something21:18
Mister_Magisterjust patch didn't download whole21:19
Mister_Magisterah patch isn't fully compatybile even21:22
T42<elros34> what patch are you talking about? use git revert "that commit"21:25
Mister_Magisterno it doesn't work like that21:25
Mister_Magisterargh why does it have to be so complicated21:30
* vknecht tries to picture jesus annoying MM while patching kernel... « thou shalt not use but files, use git am »21:31
Mister_Magisterwhy even21:31
Mister_Magisterit's not even about single commit21:32
Mister_Magisterfinally21:36
Mister_Magisterafter reverting 3 commits21:36
Mister_Magisterwhere is my gui baby21:37
Mister_Magistermy goodness dmesg is just magic21:37
Mister_Magistertried selecting one milisecond of dmesg and failed https://paste.opensuse.org/3816098721:38
vknechtOT: citop, top-like command for CI... https://github.com/nbedos/citop21:52

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