Monday, 2021-03-01

*** Bloop is now known as Guest839505:05
rinigusmal: morning! ended up factoring audiopolicy into a separate service as a result, camera works as needed including shutter sounds.06:14
rinigusall required changes are in, if someone has the same issues06:14
rinigusmal ^06:14
T42<hacker12455> should selinux be disabled on 17.1 too?08:03
malno 17.1 assumes selinux is enabled08:46
T42<hacker12455> keep audit off or on09:19
T42<Verevka86> <mal>
T42<Verevka86> building hybris16.0 broken10:11
mal@Verevka86 when did you update submodule?10:13
malupdate it now, just merged a fix I did yesterday10:14
T42<Verevka86> @mal [update it now, just merged a fix I did yesterd …], Thanks, fix10:27
piggzmal: i kernel config check on the PP kernel doesnt show anything that might break sailfail, afaict10:29
malpiggz: time for some strace?10:53
piggzyeah, maybe11:02
T42<zinstack625> pketo: could you please give me store access?11:09
piggzmal: ps aux shows jolla-mediaplayer with -prestart param ... but thats not in the .desktop file11:09
piggzmal: i dont think its started from the .desktop in usr/share/applications, it seems to be done via dbus????11:13
piggzmal: yes, adding -v there helped, in dbus-1/services/com.jolla.mediaplayer.service11:15
piggzmal: .... i wonder if it is all the dbus launched apps which fail11:17
piggzcontacts loads, uses jail, but doesn use dbus11:18
piggzcalendar uses dbus, doesnt open11:18
piggzgalery works, and doesnt use dbus afaict11:19
piggznotes doesnt work, and uses dbus11:20
T42<swanux> @mal: I kinda stuck with 16.0 base. So what do you recommend: try to go with 15.1, stick with the current situation or wait for stable 17.1?13:11
malwhat is the problem with 16.0?13:11
T42<swanux> I can get telnet (with some forced connect, and only if I hält the boot process before droid hal / systemd init), but I have no idea where to continue (tried some config modifications, manual continue, journalctl, etc, but no luck)13:13
T42<swanux> *halt13:13
maldo you have the .rc files as mentioned in faq?13:14
T42<swanux> In the init folder? Yes, I pushed them after flashing upon recommendation.13:16
malhow did you disable selinux?13:21
malcheck what you have in /etc/selinux/13:22
T42<swanux> I disabled it on boot time if that's what you meant. Config says also that it's disabled (and dmesg)13:30
T42<hacker12455> hm13:37
T42<hacker12455> does 17.1 base boot to UI?13:37
mal@swanux ok, can you provide logs? have you tried starting droid-hal-init manually to see what it prints13:38
mal@hacker12455 it can boot to UI, I have a device which boots to UI with hybris-17.113:38
T42<hacker12455> good13:38
T42<hacker12455> was just checking cause I upped motorola-cedric to 17.1 base13:38
swanuxmal: Yeah, sure. Which logs would you like (journals, live)? Tried droid-hal-init from telnet, no output (looks like it ran properly though as I didn't loose connecton)13:42
maljournal is useful13:45
swanuxmal: Well it's interesting. Now I can't even get logging to work. (tried both persistent and automatic but nothing in var/log/journal). Is the output of journalctl --no-tail -f from telnet good instead?13:58
malI usually use journalctl -b --no-pager14:01
swanuxNo journal files were found (as for some reason it doesn't work now despite my /etc/systemd/journald.conf storage setting)14:05
malwait, how did you stop the boot?14:07
malyou should instead mask droid-hal-init service and not stop the boot to before systemd14:08
swanuxI did touch /data/.stowaways/sailfishos/init_enter_debug214:12
swanuxSo should I mask now instead and delete that file?14:13
pketo@zinstack625 moto g7 added14:13
swanuxOk, ran systemctl mask droid-hal-init and now I don't need to force connection after reboot (great!)14:16
swanuxmal: Here's the log
T42<Verevka86> @hacker12455 [does 17.1 base boot to UI?], Yes14:29
T42<zinstack625> pketo: awesome, thanks14:46
rinigusmal: not much is working on aosp10/aarch64 with sim card for me. mainly I can send/receive sms'es, but the rest is quite broken15:57
rinigussuch as cell data does not connect; calls are muted; even ringing stops after a single ring tone.15:58
rinigussome I have seen mentioned for seine (mute calls), but the rest is not there15:58
rinigusmal: is that normal for aosp10/aarch64 or are there some extras found for tama?15:59
rinigusissues at
malrinigus: I seem to have same issue on my fp3 but works on seine16:06
malabout call audio16:06
malI haven't tested that otherwise much16:06
malrinigus: for me mobile data works16:09
malrinigus: do you have the ohm fixes?16:09
swanuxmal I may have found something, I just don't know how to use it. Here are the logs after manually running droid-hal-init (as I masked it): (this was in a separate telnet terminal - in the terminal where I ran the command it produced this: nohup: appending output to nohup.out, and hanging there.) Also, I can't do16:19
swanuxanything through telnet after this command, but I still have an active usb network connection.16:19
malswanux: which device is that?16:25
swanuxdipper. Xiaomi Mi 8.16:25
maldid you try 17.1 before? did you remove any changes to config repo before doing 16.0?16:28
swanuxI use separate droid config repos for 16.0. I tried 17.1 before, that resulted in bootloop without telnet. you said something above about SELinux in 17.1 though (?), because I did 17.1 the same way as 16.0 (last time on Saturday I guess)16:33
rinigusmal: not sure about ohm fixes. (would have to pull logcat and journal still)16:43
rinigusmal: where do we get ohm fixes?16:44
rinigusmal: looks like I have ohm
rinigusI can see that ohm is updated to 1.3.0. any plugins I should look into as well?16:47
T42<elros34> @swanux about that cgroup errors you have you can try adding  systemd.legacy_systemd_cgroup_controller=yes to kernel boot arguments16:49
swanux@elros34 Thanks for the advice. Dumb question, but how can I add boot arguments?16:52
T42<elros34> BOARD_KERNEL_CMDLINE in BoardConfig*.mk in your device repo16:55
swanuxThanks, (found it as under devices/xiaomi/sdm845-common) I'm building now.16:59
swanuxmal: Thanks to @elros34 I've got a lot more promising log now :
malrinigus: I'll let you know about those a bit later today17:19
T42<elros34> @swanux yeah not really:) android part fails same way17:23
T42<elros34> you could try strace  logcat to see why it fails to start17:24
swanux@elros34 Yeah, I see. I meant it's now shorter because it has one less error in it thanks to you. You know, small step for a senior a giant step for beginner.17:30
swanuxRegarding logcat, I can't really use it...17:30
swanuxlogcat read failure17:30
rinigusmal: great! I have to look into FPD before anyway17:33
T42<elros34> @swanux again do you start it after droid-hal-init appears in logs? Also strace should still shows something more than 1 line17:34
swanuxSo, if I understand correctly, I should run (strace) droid-hal-init and then try with logcat?17:36
T42<elros34> I was talking about stracing logcat but you can strace droid-hal-init also. Be aware you can start droid-hal-init only once per boot17:37
swanuxI see. Here's the strace of logcat before I run droid-hal-init (as I can't run anything after it since it breaks telnet) :
swanuxAnd here's the strace for droid-hal-init:
swanuxAnd here's journal log for droid-hal-init (the same as before, just to have everything at one place):
T42<elros34> please: cat /usr/libexec/droid-hybris/system/etc/init/vndservicemanager.rc17:58
swanux@elros34 You, mean the output? ( It's from the faq, so it's the same18:02
T42<elros34> I want to be sure it's same because you refuse to upload your up to date repos so "cat /path/to/file"  on device18:03
T42<elros34> obviously that paste is wrong, if you have same on device then that is the reason why droid-hal-init fails18:04
T42<elros34> If you want to download some file directly from github then you must first use RAW button18:10
malyeah, easy to do it wrong otherwise18:10
swanuxThank you, I guess I'd have never thought about this stupid mistake... I'm gonna replace them now.18:24
T42<elros34> I wonder why parsing error is visible in strace but not in your journal.18:28
T42<elros34> Could be that you have some messages from droid-hal-init in dmesg?18:30
malmessages should show in journal also18:33
swanuxNow I'm recreating journal (after running droid-hal-init), I'll post it, but this line looked familiar to me18:37
swanuxdroid-hal-init: couldn't write 7958 to /dev/cpuset/system-background/tasks: No space left on device18:37
swanuxI'll post the whole log in a minute. Also dmesg.18:37
swanuxHere's the journal (it's huge this time) :
swanuxAnd the whole dmesg -w output:
malswanux: did you fix all of the files you copied from github?18:46
swanuxmal: Yes, I copied them from raw github and pushed after checking. Here's the raw output of disabled_services from cat:
T42<elros34> your journal is huge because you have persistent logging enabled. use -b0 switch to get journal only from current boot18:59
swanuxThanks for the tip! Here it is:
T42<elros34> droid-hal-init is still not very verbose. If you logcat still fails try again stracing droid-hal-init. At least that way it prints something19:09
swanux@elros34 logcat still the same, here's the new strace:
T42<elros34> it still can't read rc files..19:17
swanuxAre the permissions correct?
swanuxI have no other idea, I checked them again, and their content is fine. I literally ctrl+a, ctrl+c, ctrl+v-d it from raw github then adb push-ed them.19:21
T42<elros34> if they are readable the should be fine. Strace doesn't show any ermission issue on rc files.? I see you start whole script. Could you try just strace -f /sbin/droid-hal-init instead.  Do you have any additional content in dmesg when you run strace that way?19:25
malswanux: try changing the permissions to 44419:26
mal@elros34 it seems it can read the files with 444 permissions based on log19:27
swanuxmal: I changed the permissions. Now running droid-hal-init with strace as @elros34 suggested.19:29
T42<elros34> mal: right 0644 are parsed but 0666 fails19:31
swanuxWoah, now everything is different. I'll share the new data, just need a minute19:32
swanuxThis is the first part of strace -f (probably not so relevant as it continued, just in case):
swanuxThen hundreds of lines of this:19:35
swanux[pid  6003] newfstatat(AT_FDCWD, "/dev/.coldboot_done", 0x7fdc6e40c8, 0) = -1 ENOENT (No such file or directory)19:35
swanux[pid  6003] nanosleep({tv_sec=0, tv_nsec=10000000}, 0x7fdc6e4088) = 019:35
swanuxEnd the end is:
swanuxHere's the short journal :
swanuxAnd the short as well dmesg:
swanuxAnd mal @elros34, huge difference: Now my telnet interface didn't die.19:38
T42<elros34> you need to execute commands from before stracing, you can see in logs it waits for /dev/.coldboot_done19:40
piggzrinigus: congrats19:49
mal@elros34 I think unmasking droid-hal-init might make sense already19:50
swanux@elros34 mal: I guess unmasking really makes sense, as it's been running for a couple of minutes now and still not dead19:51
swanuxI'll wait for the end of this, and if no problems I'll unmask it19:51
swanuxOk, I guess I unmasked it and rebooted. Now it's not dead, but I can't see much difference (I mean visually, on the device's screen)19:57
swanuxdroid-hal seems to have inited successfully. Thank you for your help @elros34 and @mal20:02
malany indication why UI fails?20:03
T42<swanux> Well, I don't know what would be an indication.. (the journal you mean? Or dmesg?)20:20
maljournal of course, also logcat might tell something20:27
T42<swanux> OK, thanks. I'll continue tomorrow I guess.20:32
T42Sasank %lastname% was added by: Sasank %lastname%20:40
T42<hacker12455> any idea why usb would not open during hybris-boot on 17.120:45
T42<hacker12455> only thing we get is this20:46
maldid you set the environmental variable before building hybris-hal?20:46
T42<hacker12455> which one?20:47
T42<hacker12455> nope20:47
T42<hacker12455> what does that change?20:48
malit makes kernel work20:48
malotherwise the initrd doens't contain what it should contain and boot fails very early20:48
T42<hacker12455> sure gonna try it20:48
malyou only need to run make hybris-boot and then flash the hybris-boot.img manually20:49
T42<hacker12455> well that image does not create an interface too20:57
T42<hacker12455> we have nothing on usb20:57
T42<hacker12455> device stays on the oem logo21:00
T42<hacker12455> it worked back when we were running a nougat base21:01
maldo you see anything in jost dmesg? does it even show up in there21:12
T42<hacker12455> nope21:16
malno sign of the device in dmesg? does the device remain on or reboot?21:29
T42<hacker12455> remains on21:41
T42<hacker12455> no sign of the device21:41
T42<elros34> @hacker12455 by 'no sign' you mean there is no hybris-boot information in host dmesg or nothing at all from usb. Your paste suggest that you are not booting hybris-boot.22:36

Generated by 2.17.1 by Marius Gedminas - find it at!