Wednesday, 2023-07-05

T42<smatkovi> i think i just wanted to try to work with a minimally modified kernel, so i just got rid of the errors and didn't set it in kernel options earlier. this process takes so long that i forget what i wanted between sessions09:12
T42<smatkovi> okay i have a slight improvement: power_supply sm5703-fuelgauge: driver failed to report `charge_full' property: -2209:13
T42<smatkovi> should i also outcomment charge_full related parts?09:14
T42<smatkovi> or would this be dangerous in a way?09:14
T42<smatkovi> okay i am to square one i get the same error about energy_now as before, so i don't know how to patch the driver properly09:58
T42<smatkovi> i think i forgot to add some selinux dependencies, so i'm retrying with those now10:06
T42<smatkovi> i checked also for dependencies of dependencies and there is only CONFIG_SECURITY which i already had, but it still ends with a config on the device with selinx disabled. i set the dependencies all to =y, maybe i shouldn't do that10:44
T42<smatkovi> oh i had CONFIG_AUDIT=n earlier, that could be a problem10:46
T42<elros34> about power_supply: whether it's safe I think depends on how do you fix issue. There may be setter for particular property but without getter. In that case you do not comment out everything but return enodata or something like that. There are 2 ways to read particular property (check /sys/class/power_supply/*/property or uevent so you can't just uncomment everything11:45
T42<elros34> IIRC I also had 3 wrong properties. I do not know why this is not and issue on android but I suspect it's systemd udev which triggers it11:47
T42<elros34> Do you really get failed to report 'status' property? I do not not why you showing that line in kernel source11:54
T42<smatkovi> no i get failed to report energy_now property12:20
T42<smatkovi> oh i even found that property. i got an idea, thanks12:22
T42<smatkovi> yes it worked :) i got rid of that message12:51
T42<smatkovi> here is the output of output of "journalctl -b --no-pager"
T42<smatkovi> is selinux essential for a boot?12:55
T42<elros34> I think it maybe needed for android part IIRC since hybris-1712:56
T42<elros34> your first error is system_root.mount failure12:57
T42<smatkovi> I guess i have to check
T42<elros34> yeah or directly mount unit in /usr/lib/systemd/system/13:17
T42<smatkovi> If the paths are all correct and maybe different from each other13:17
T42<smatkovi> okay they seem to be correct when i check them in telnet13:24
T42<smatkovi> can this line in fstab be correct? /dev/root  /         None    defaults,noatime 0 013:25
T42<smatkovi> shouldn't it be something else than none?13:25
T42<smatkovi> and what do you mean by unit @elros34 ? i don't have such a partition13:26
T42<elros34> /etc/fstab? it's irrelevant. Show me system_root.mount13:26
T42<elros34> did you read hadk-hot? Sectiin about what to check first? systemd mount unit: /usr/lib/systemd/system/system_root.mount13:27
T42<smatkovi> no i didn't read that part13:28
T42<smatkovi> because i didn't assume that my device has dynamic partitions13:29
T42<smatkovi> how can i find out?13:29
T42<elros34> so read everything, few minuts but it will save you few hours blindly searching13:29
T42<smatkovi> oh and it is in the hybris-18 section13:32
T42<elros34> so this 'mount rootfs rootfs' is not hanlded properly. mount system partition manually now and check how it looks like: whether it's system-as-root or regular system13:34
T42<smatkovi> Example verbose script for mounting:  isn't available anymor13:43
T42<elros34> this part is probably not relevant for you device, mount system partitiona dn show the content instead13:44
T42<smatkovi> do you mean this?: / # mount rootfs rootfs / remount bind ro nodev13:54
T42<smatkovi> mount: bad usage13:54
T42<smatkovi> Try 'mount --help' for more information.13:54
T42<elros34> no more like 'mount /dev/(system partition block device from fixup-mountpoints) /mnt; ls -al /mnt13:55
T42<smatkovi> i think it gets mounted properly14:02
T42<elros34> so it's system-as-root14:02
T42<elros34> find init.rc in device repository and that mount rootfs part14:03
T42<smatkovi> skip_initramfs handling so kernel can use hybris-boot initramfs. In kernel init/initramfs.c history find 2 following commits and revert them: "call free_initrd() when skipping init" and "Add skip_initramfs command line option". Alternatively, you can force in code do_skip_initramfs = 0.14:03
T42<elros34> unrelated, your problem is: ' mount rootfs rootfs /" I think systemd do not know what 'rootfs' is14:04
T42<smatkovi> okay thanks14:05
T42<elros34> or maybe you should add "/dev/block/bootdevice/by-name/system         /          ext4    ro,barrier=1  " to android's fstab, make sure it's updated also in out/ and then build droid-hal again and check if system_root.mount changed14:12
T42<smatkovi> this? /home/sebastian/hadk/hybris/hybris-boot/initramfs/etc/fstab14:16
T42<smatkovi> there are so many^^14:16
T42<TheVancedGamer> it's supposed to be in your device tree14:17
T42<TheVancedGamer> so device/samsung/yourdevicehere14:17
T42<TheVancedGamer> it could be either in your common tree or device specific tree14:17
T42<smatkovi> /home/sebastian/hadk/device/samsung/msm8916-common/recovery/recovery.fstab14:17
T42<smatkovi> /home/sebastian/hadk/device/samsung/msm8916-common/recovery/twrp.fstab14:17
T42<smatkovi> /home/sebastian/hadk/device/samsung/msm8916-common/rootdir/etc/fstab.qcom14:17
T42<smatkovi> i have only these 314:17
T42<TheVancedGamer> fstab.qcom is what you're after14:17
T42<smatkovi> okay thanks14:17
T42<TheVancedGamer> add the line that Elros suggested there and rebuild droid-hal14:18
T42<smatkovi> it actually starts with this exact line14:19
T42<TheVancedGamer> oh :P14:19
T42<Mister_Magister> 8916 lol14:19
T42<smatkovi> so this could be the appropriate init.rc then /home/sebastian/hadk/device/samsung/msm8916-common/rootdir/etc/init.qcom.rc14:21
T42<smatkovi> there i have this line     mount_all /system/vendor/etc/fstab.qcom14:21
T42<k1gen> msm8916 has great mainline support, is it possible to use mainline kernel with sailfish for him?14:24
T42<smatkovi> what's so funny about 8916? is there already a port for that or what?14:24
T42<elros34> hm why this wasn't problem on other devices.. As dirty hack you can comment out that 'mount rootfs rootfs / remount bind ro node' from init.rc in out/ and in system/core or whether place its copied from and then build droid-hal14:24
T42<elros34> more proper way: makefstab_skip_entries14:25
T42<smatkovi> what happens if i comment that out, shouldn't it  be mounted?14:26
T42<smatkovi> i'll continue later with that14:26
T42<elros34> in that case droid helper script should use correct entry from fstab14:28
T42<TheVancedGamer> mal: have you had any issues with non-existent call audio in hybris-20?15:13
T42<TheVancedGamer> I checked audiosystem-passthrough-qti and i'm getting some weird "failed to call set_parameters()"15:16
T42<smatkovi> so i could just remove this line in init.rc on the device and see what happens15:38
T42<smatkovi> but actually this line is in a comment which says that this line is removed during droid-hal-device build15:40
T42<smatkovi> oh you're right, they're in the outpaths15:44
T42<k1gen> I had this in 18, issue was hidl_compat-related (re @TheVancedGamer: mal: have you had an...)16:11
malnot sure what is causing that hidl compat issue16:15
T42<smatkovi> i think it's properly mounted now. now lipstick ui failes to start. could this be because of selinux not being enabled?
T42<k1gen> mal: damn, it's been three months16:17
T42<k1gen> I met NotKit in another chat some time ago, he didn't debug it also iirc16:18
mal@k1gen well, I haven't looked into it for quite a while16:18
T42<k1gen> no worries :)16:18
T42<TheVancedGamer> mal: oddly in this same port call audio works on Halium 1116:22
T42<elros34> @smatkovi log is too short to tell anything. Are you sure it's mounted properly now, have you checked: 'findmnt' output?16:23
T42<k1gen> if I would want to return to sailfish and stoll don't have audiocalls, I'll just have to ask for help with setting up 100% legally acquired aliendalvik, so I have at least reliable audio/video calls in another app like google meet16:23
T42<k1gen> *still16:23
T42<elros34> AFAIK there is no such a thing as aliendalvik on new ports16:24
T42<smatkovi> /                            /dev/mmcblk0p28[/.stowaways/sailfishos]16:25
T42<smatkovi>                                              ext4    rw,noatime,data=ordered16:25
T42<smatkovi> |-/data                      /dev/mmcblk0p28 ext4    rw,noatime,data=ordered16:25
T42<smatkovi> |-/system                    /dev/mmcblk0p25 ext4    ro,noatime16:25
T42<smatkovi> i think16:25
T42<k1gen> @elros34: well, I don't want to fire up a waydroid vm every time I want to call someone - or have it running in the background 24/7 to be able to receive calls16:25
T42<smatkovi> that means / and system are mounted16:26
T42<k1gen> @elros34: also, can you explain why?16:26
T42<elros34> there is nothing to explain, it's jolla decision/lack of man power to provide it16:28
T42<k1gen> but there is still a way to set up ad on community port, right?16:29
T42<elros34> that is the point, no there is not16:29
T42<smatkovi> here is the journalctl
T42<k1gen> I wasn't asking about it back in April because I remember you or someone else here saying we don't discuss ad here16:30
T42<smatkovi> the complete16:30
T42<elros34> we don't discus about cracking jolla's software on jolla's channel and that is all16:30
T42<elros34> @smatkovi did you read that log? it's completely useless16:31
T42<smatkovi> i thought the errors about lipstick and pulseaudio were somehow useful16:31
T42<smatkovi> but thanks for pointing that out so there is somewhere else to look16:32
T42<smatkovi> for the problem16:32
T42<elros34> there is nothing in this log except kernel spam to look. use grep -v "sm5703" next time before you post it. Did you check system_root.mount is now correct?16:32
T42<elros34> I see kernel driver is full pr_info. I am not sure this is correct, maybe it should be pr_debug for example16:37
T42<smatkovi> i tried to compile the kernel with debug options and to be as verbose as possible16:39
T42<smatkovi> system_root.mount is identical to the one i pasted earlier16:40
T42<smatkovi> i think those are the kernel options16:41
T42<smatkovi> but maybe they don't have to do anything with the driver16:47
T42<elros34> I see no point unless you debug some kernel issues. (re @smatkovi: i tried to compile t...)16:48
T42<elros34> If system_root.mount is same then why do you say system is mounted correctly now? DDid you remove line we were talking about? out/target/product/j5xnlte/root/init.rc16:50
T42<smatkovi> yes i removed it16:51
T42<smatkovi> i think it was already commented out in in system_root.mount16:51
T42<elros34> and also in system/core something?16:52
T42<elros34> chek also droid-hal-device.log16:53
T42<elros34> ah I get it now, you forgot about hybris-patched don't you?16:54
T42<smatkovi> you mean this: hybris-patches/ --mb ?16:56
T42<smatkovi> this one i did16:57
T42<elros34> I do not think so. Unless you forgot about make hybris-hal after16:57
T42<smatkovi> a can do that again. i ran without arguments, doesn't this make hybris-hal also, becaus it should make everything?17:00
T42<smatkovi> maybe this one is interesting too: [defaultuser@j52016 log]$ /usr/libexec/droid-hybris/system/bin/logcat17:01
T42<smatkovi> -bash: /usr/libexec/droid-hybris/system/bin/logcat: Too many levels of symbolic links17:01
T42<elros34> no it doesn't, android part is build in android environment17:01
T42<smatkovi> oh i see, i misunderstood you. i always run make -j$(nproc --all) hybris-hal droidmedia17:02
T42<smatkovi> after setting a kernel option17:03
T42<elros34> that doesn't make sense, you must forgot about something, make hybris-hal again and show out/target/product/j5xnlte/root/init.rc then build droid-hal17:03
T42<smatkovi> error: rootdir/init.rc: does not match index17:05
T42<smatkovi> Patch failed at 0003 (hybris) Mer can specify mis-alignment handling - this is the wrong place to set it17:05
T42<smatkovi> oh you were right17:05
T42<smatkovi> maybe when i did repo -sync once i should have patched it again17:06
T42<smatkovi> i don't get what repo -sync overwrites. when i want to restore a file version from the sources it doesn't overwrite it, sometimes also not when i delete it, but other files get overwritten17:18
T42<smatkovi> i know the proper way would be git diff17:18
T42<smatkovi> but i'm too lazy to make a patch and then reapply it17:19
T42<smatkovi> now i get this in journalctl
T42<smatkovi> maybe this could be the relevant line: mount: /system/etc/ld.config.29.txt: mount(2) system call failed: Too many levels of symbolic links.17:46
T42<elros34> I had impression you didn't bother to read hadk-hot? What is the issue we are trying to fix? mount points so check them for real17:48
T42<smatkovi> it says something about dynamic partitions and to add %define makefstab_skip_entries / /product /system /system_ext /vendor to droid-hal-common.inc17:58
T42<smatkovi> only that i don't find the appropriate file, don't know where to look for it and how it can be named and i don't even know if the partitions are actually dynamic17:59
T42<smatkovi> but maybe from android 10 there are dynamic partitions for sure?17:59
T42<elros34> its' hybris-18 not related to hybris-17. how about this or this
T42<smatkovi> /home/sebastian/hadk/rpm/dhd/droid-hal-device.inc18:01
T42<smatkovi> this maybe?18:01
T42<elros34> ok lets get it simple: you were supposed to check system_root.mount, now show it to me18:02
T42<smatkovi> okay
T42<elros34> so you see still see ' mount rootfs rootfs / remount bind ro nodev' in last line. Now show me out/target/product/j5xnlte/root/init.rc18:09
T42<smatkovi> is it okay to be downloadable?
T42<elros34> ok now unpack droid hal rpm droid droid-local-repo/ and check how system_root.mount looks like18:14
T42<smatkovi> what do you mean by unpack?18:20
T42<smatkovi> do you mean the rpms?18:21
T42<elros34> yes droid-hal-device*rpm  via some archive manager you normally use18:21
T42<smatkovi> droid-hal-j5xnlte-0.0.6-202307051715.armv7hl.rpm18:21
T42<smatkovi> :S the file is not there18:23
T42<smatkovi> it should be there right: droid-hal-j5xnlte-0.0.6-202307051715.armv7hl.rpm18:24
T42<smatkovi> /./usr/lib/systemd/system18:24
T42<smatkovi> there is only system.mount18:25
T42<smatkovi> and it says [Mount]18:25
T42<smatkovi> What=/dev/mmcblk0p2518:25
T42<smatkovi> Where=/system18:25
T42<smatkovi> this should be longer right?18:26
T42<elros34> no but it's wrong anyway18:26
T42<elros34> show your fstab from out/18:26
T42<elros34> do you create and install sfos image manually somehow?18:27
T42<smatkovi> i extract the zip and then put hybris-boot and the .gz file to a directory where i extracted the zip already and modified the and updater-script to handle the .gz file and then i pack this directory to a zip18:30
T42<smatkovi> i didn't find the places where these two scripts are generated18:31
T42<smatkovi> i found some places but they ended up with the old scripts18:31
T42<elros34> how do you create image? with
T42<smatkovi> yes18:31
T42<smatkovi> without arguments18:32
T42<smatkovi> do you mean this fstab? /home/sebastian/hadk/out/target/product/j5xnlte/system/vendor/etc/fstab.qcom18:32
T42<elros34> yes18:33
T42<smatkovi> i think there is the first line the line you wanted me to put in there18:35
T42<elros34> no that is not the line I told  you to put there18:36
T42<elros34> you were supposed to put "/dev/block/bootdevice/by-name/system         /          ext4    ro,barrier=1  "18:37
T42<elros34> in this case replace existing18:37
T42<smatkovi> so change the mountpoint from /system to /18:38
T42<elros34> yes18:39
T42<smatkovi> thanks18:39
T42<elros34> of course also in device repo but you should know that already18:40
T42<smatkovi> i don't get why some rpms have a mismatching checksum and get redownloaded sometimes when building the image18:59
T42<smatkovi> now i get this in journalctl
T42<smatkovi> maybe this is the first relevant line Device /dev/sailfish/home does not exist or access denied19:14
T42<smatkovi> so maybe i need to fix the other mountpoints too?19:14
T42<elros34> dev/sailfish/home is annoying but irrelevant. You really need to start getting proper logs, either via fixing kernel driver or simply  adding grep -v to filter out this spam and removing rate limit in /etc/systemd/journald.conf, enabling persisten logging.19:18
T42<smatkovi> maybe this is the relevant line Mar 14 18:08:17 j52016 mount[628]: mount: /system_root: mount point not mounted or bad option.19:45

Generated by 2.17.1 by Marius Gedminas - find it at!