Friday, 2019-08-23

*** ahjolinna_ is now known as ahjolinna04:46
spiiroin@DylanVanAssche about that mce & volume keys. mce itself does not emit or even use volume key events (except a bit, for display state management). but mce does policy logic that informs (over dbus) whether compositor/applications are allowed to use / should ignore volkey events they receive as input events04:50
spiiroinin olden times the policy decisions were expressed by disabling hw interrupts etc, later on by having mce grab input device nodes, and currently by doing that dbus signaling04:52
spiiroinsome of the vocabulary in code / comments reflects this legacy (talk about grabbing and interrupts etc)04:53
spiiroinbut shortly: if lockscreen is active, everybody should be ignoring volume keys - except if there is ongoing "music playback" activity04:53
spiiroinfor more details, see
T42<birdzhang> mal, i found this udev rules cause my device's speaker no sound, after delete this line, all is working06:32
T42<birdzhang> ^^^
T42<birdzhang> some dmesg logs:
Mister_Magister but that supposed to be improvement…06:33
T42<birdzhang> to make speaker has sound, i need put tas2557_uCDSP.bin to /etc/firmware/ , it's ugly but working06:35
Mister_Magisteri had to copy whole /system/etc/firmware to /etc so that ain't new thing06:35
mal@birdzhang where is the firmware?08:57
T42<birdzhang> on hybris 15.1, i need put bin file to hybris-boot/initramfs/etc/firmware/, so that it can be loaded09:00
T42<birdzhang> i modified /usr/bin/droid/, add /etc/firmware to FIRMWARE_FOLDERS, no luck09:01
mal@birdzhang but where are the files originally on the device?09:12
T42<birdzhang> this is no /sys$DEVPATH/loading on my device, so always failed09:13
T42<DylanVanAssche> @spiiroin since our lockscreen is still broken, it's normal that lipstick is ignoring the volume keys :P09:32
T42<DylanVanAssche> So the volume keys are actually working but they act broken due to the lockscreen troubles09:32
mal@birdzhang you still didn't answer my question09:37
T42<birdzhang> mal, it's on /system/etc/firmware/09:39
malso symlinking to /etc/firmware doesn't work anymore after the config change?09:39
T42<birdzhang> yup09:43
malwhy would that break it09:52
T42<birdzhang> firmware tas2557_uCDSP.bin: firmware_loading_store: unexpected value (0)09:55
mal@birdzhang when did you update the config submodule? can you show journal log also11:23
T42<birdzhang> mal, about 7 days ago. i will show you when i get home11:34
T42<birdzhang> mal, journalctl log
T42<birdzhang> wlan module also broken14:50
electro575NFS is needed for sailfishOS ? in the kernel defconfig15:04
electro575network file system is just for remote control a linux system, no ?15:05
UmeaboyHi! I made a booboo with the repo command:
UmeaboyHow can I solve it?15:09
mal@birdzhang does your device have this already
mal@birdzhang maybe you are missing this CONFIG_FW_LOADER_USER_HELPER_FALLBACK=y from kernel config?16:06
Umeaboymal: I'm following the wiki page for Xperia X so that's why Iäm asking.16:06
malUmeaboy: maybe your build device doesn't have ssh key generated?16:08
Umeaboymal: You mean inside the HABUILD SDK?16:09
UmeaboyBecause I just generated a ssh key on my build host and I still get the same error.16:10
malit uses the things from your normal home16:10
UmeaboyDeleting the repo config did the trick16:14
Umeaboymal: Building with Dalvik support is not permitted or where are the instructions to do so?16:16
malnot allowed16:18
electro5751all devices need watchdog to launch the kernel ?17:11
albertus1vknecht: did you have to figure out how to solve the conflicting files in the MIC step for your own porting to a new device, or did you just assemble it as ideas how to solve my problem? (I'd like to know if the problem is unique to me or rather based in the current software versions for "everybody")18:42
malalbertus1: there is really something wrong in your setup as nobody has had same issues18:50
malalbertus1: also there are many packages which have similar things that two or more packages provide the same folder and why are those not causing issues only if you have built one of those then there is a problem18:52
deathmisthow are people properly dealing with ofono-ril-binder-plugin needing more up-to-date versions of libgrilio and ofono? should I just downgrade this specific package?18:59
albertus1so do you think there it might be worth to start from scratch again?18:59
albertus1is there any sort of mechanism like reproducible builds or just checksums so I could compare files?19:03
maldeathmist: usually people just build older ofono-ril-binder-plugin19:18
maldeathmist: just like I have mentioned many times in channel logs19:18
T42<elros34> albertus1: you still didn't run mic with --verbose --debug? maybe it will reveal something20:00
malalbertus1: can you check in platfrom sdk which grep is installed, using "sudo zypper se grep"20:07
deathmistmal: alright seems binder version 1.0.7 works a-ok, sorry for annoying. I was just doing things a bit differently before and wanted to make sure20:09
albertus1@elros34: I did20:13
malhmm, that is normal20:13
albertus1running again (with --debug and --verbose) so I can paste the output20:16
albertus1after vknecht's suggested adjustment of the spec file, I got rid of all conflicts but one20:17
albertus1because all but one involved geoclue-provider(s)20:18
albertus1this is what it was before:
malalbertus1: the thing is that manyy other packages have similar conflicts and don't seem to be a problem20:18
albertus1hmm, that's very strange then20:18
malalbertus1: I checked on my device and there were many cases with similar conflicts and those exact ones you have and all worked fine20:19
malso the issue is something else than the conflicts20:19
albertus1this is with --verbose and --debug
albertus1and all those conflicts do not cause any problems? if you say there exist many of them and it still works fine?20:21
malalbertus1: yes, no issues from those, I mean nothing reports those as conflicts20:22
malalbertus1: but manually checking package contents shows those ones you have and other similar ones but those are not considered conflicts20:23
albertus1hm, very strange20:23
albertus1this is about the `rpm` that i have:
deathmist well this is new O_o didn't find anything in logs either. used droidmedia source (w/ kimmoli's AudioPolicy patch)
malalbertus1: check with "sudo zypper if rpm"20:30
malalbertus1: i just did a clean reinstall of platform sdk and doing some testing20:30
malalbertus1: I'm also now setting up clean build target to see what happens20:55
malalbertus1: mic build using OBS built packages worked fine so now trying locally built packages20:58
deathmistfound the culprit, the issue was indeed new. anyone with a Snapdragon 835 (MSM8998) device building droidmedia getting AudioMix errors w/ CAF sources: revert in hardware/qcom/audio-caf/msm8998/21:24
maldeathmist: what did the errors cause?21:27
deathmistmal: during droidmedia building w/ AudioPolicy patch21:28
malok, interesting21:30
malalbertus1: I really can't understand your issue, I did a clean rebuild of packages using clean sdk and target and then built image using mic (for different device though) and I had no conflicts during mic build21:32
albertus1hm, that's odd. thanks for your effort21:43
albertus1maybe something wrong in the guide or maybe I just got some important thing(s) wrong while going through it. maybe I should start from scratch...21:44
malalbertus1: is there anything special in your build machine, like which filesystem does the disk used for sfos build environment have and so on?21:49
albertus1build environment should be ext4, just $ANDROID_ROOT is on ntfs due to its need for space21:54
albertus1is there anything to be seen from the verbose/debug output of mic I posted above?
albertus1there's a stacktrace looking like python stuff21:56
albertus1it doesn't even sound like the error is due to the conflicts (which are only warnings)21:59
deathmistalbertus1: I really try hard to avoid anything NTFS when it comes to Linuxy stuff. imo this could be the cause of all your issues. first off it doesn't have any concept of the Linux permission system and assumes 777 on everything22:00
malyes, it could be that during installation it somehow thinks the folder is not really a folder because of the non-linux filesystem22:02
albertus1true, but what's the way to go if I don't have enough space on my ext4 partition? should I create an "iso file" on the NTFS partition and build an ext4 file system "inside"?22:02
albertus1I've never seen any linux command fail to distinguish folders and non-folders on ntfs...22:03
albertus1I've even mounted it with options "suid,exec" (because usually I mount that with nosuid and noexec)22:05
malalbertus1: well that is the only thing that seems to be different in your setup, I probably should have seen the same issues in my test if it was something generic22:09
albertus1hm, yes, and the device/family, which caused me to do some manual builds as hinted here22:10
malalbertus1: maybe you could try to move only some stuff out of the ntfs disk?22:10
albertus1but how do I know which?22:10
albertus1I don't really understand all those things in $ANDROID_ROOT22:11
malalbertus1: I mean keep only android sources in ntfs22:12
albertus1are those the largest part?22:12
malother parts are only 1-2 gb max maybe22:13
albertus1hmm, "prebuilts" directory has 18 GB22:14
albertus1"out" hat 15.7 GB22:14
malthat is android stuff22:14
malso the point is that all the platform sdk stuff would happen in ext4 disk22:15
albertus1well, platform sdk itself (the chroot) is on ext422:15
malit might need some hacking but shouldn't be too difficult22:15
malalbertus1: yes but some stuff happens in android_root22:16
albertus1those hybris/ things22:16
malyes, hybris and rpm folders, and droid-local-repo, and you need to symlink or bind mount out/ from ntfs to the other android root, maybe also some other folder, like device and build22:18
albertus1hm hm, but should I then just copy the existing stuff over, or would I rather need to start anew from some earlier point?22:20
T42<adampigg> Mal: meta package would be a a repo containing only an.rpm/ folder and .spec, with focus on the Requires: lines?22:21
albertus1one more thing is that in ubu-chroot, $ANDROID_HOME needed to be /parentroot/parentroot/mnt/..... while in platform sdk it was only /parentroot/mnt/....22:22
albertus1(I thought using /parentroot more convenient than bind-mounting all the way)22:22
malalbertus1: you can have two copies of some things if it's easier, like hybris folder22:33
malalbertus1: it might need some thinking to figure out the best way, it's quite late already so maybe not today22:34
mal@adampigg that is one way, other would be to just have a new package in config repo with those22:35
albertus1maybe i'll try the entire build process on a remote virtual server with more space....22:42

Generated by 2.17.1 by Marius Gedminas - find it at!