Tuesday, 2020-02-04

T42<edp_17> Nope, the technologies=umts or gsm didn't work either.00:00
T42<edp_17> However when I used dataCallFormat=6 or 9 I didn't get the 'Network error' when I switched on the data.00:00
T42<edp_17> I have tried technology=gsm/utms, allowDataReq=on/off, dataCallFormat=6/9/11, lteNetworkMode=9/8/7/6/5 and networkSelectionManual0=true and some combinations, but no luck. When I turn on data its status (in Settings/Mobile network) jumping between connecting/enabled.00:27
T42<edp_17> I am trying to add my jolla account in Settings and I am getting the 'Unabe to login with this username and password' message while in the log: 'SailfishKeyProvider_storeKey(): error: unable to write scheme'00:44
T42<edp_17> Do you know what is it and how to fix?00:44
T42<edp_17> Oh, and at the beginning there is a 'SailfishKeyProvider_ini_read: unable to open file' message.00:45
T42<edp_17> @elros34 : I have found another module that was not included into the image. (cifs.ko) It manually needed to be transferred to the device. So it seems there is a problem somewhere in the line and the modules are not distributed properly into the sailfish rootfs.01:11
T42<edp_17> And this might be a similar reason why the mount points are still wrong in my fresh images however I have fixed them in the fixup_moutpoints script.01:13
r0kk3rzwhat steps do you go through to rebuild?01:14
T42<edp_17> I ran 'rpm/dhd/helpers/build_packages.sh'01:15
T42<edp_17> In Platform.01:15
T42<edp_17> Everything else were included (files from sparse) except the mount points and modules.01:16
T42<elros34> you know you can just unpack droid-hal*rpm and check content. Show diff of your fixup mountpoints01:18
T42<edp_17> I generated an image on Saturday (1 Feb) and now have unpacked it. It indeed has the wrong /dev/block/mmcblk0p9 in system.mount.01:24
T42<edp_17> In the fixup-mountpoints script I have -e 's bootdevice/by-name/FACTORYFS mmcblk0p9 ' \01:26
T42<edp_17> and I changed this on 25 Jan.01:26
T42<edp_17> the /efs, preload and /mnt/.lfs are also wrong. (Though I am not certain that I need this last one.)01:28
T42<elros34> From where id you get that path? You were supposed to add your device codename here: https://github.com/mer-hybris/hybris-boot/blob/master/fixup-mountpoints#L69001:29
T42<edp_17> I got them from android. I flashed the base I am using, booted into it then used the command from hadk 5.3. Configure Mountpoint Information01:30
T42<edp_17> Followed the hadk.01:30
T42<elros34> your fstab have  /dev/block/mmcblk https://github.com/edp17/android_device_samsung_galaxys2-common/blob/cm-12.1/rootdir/fstab.smdk4210.01:32
T42<edp_17> Yes, I haven't changed this. Should I?01:34
T42<elros34> remove changes you made in fixup and and add your device codename to mentioned place.01:36
T42<edp_17> Ok, I will try and create a new image, then extract droid-hal*.rpm to double check. Thanks.01:37
T42<elros34> wrong order, first build droid-hal, unpack rpm to make sure mount units are correct then create image01:39
T42<edp_17> I usually just run 'rpm/dhd/helpers/build_packages.sh', this does everything. I know I can give a parameter to build only droid-hal but I leave it running while I am sleeping and by the morning I will have the image what I can flash if the mount points are ok in droid-hal*rpm.01:42
T42<edp_17> Thanks for the help and good night. To be continued...01:42
r0kk3rzit doesnt rebuilt the hybris-hal, which you need to do01:44
T42<edp_17> Yes you are right. I forgot that should be done in the habuild. Oh my, I should sleep now because I mix up everything.01:45
T42<edp_17> Thanks for the heads up.01:45
T42<elros34> You probably don't need to make hybris-hal, /dev/block/mmcblk works in intramfs01:48
r0kk3rzisnt the whole boot image made by hybris-hal though?01:56
T42<edp_17> The make hybris-hal has just completed. Now I star the build_packages and got to sleep. 😄01:56
T42<adampigg> Mal:07:09
T42<adampigg> (Document) https://irc.thaodan.de/.imgstore/sGxlxou0r0.txt07:09
T42<adampigg> Mal: also, vol keys are nkt detected by csd07:23
T42<adampigg> So, unlikely a pa issue07:23
T42<adampigg> ^^ evdev trace07:28
T42<adampigg> (Document) https://irc.thaodan.de/.imgstore/S5l5kaZK3E.txt07:28
T42<Bdj %lastname%> Hi everybody12:15
T42<Bdj %lastname%> Is there something like GSI  that we could use it on project treble enabled device12:15
T42<Bdj %lastname%> Sailfish OS GSI12:15
T42<Bdj %lastname%> @Bdj %lastname% [Hi everybody12:15
T42<Bdj %lastname%> Is there something like GSI  that …], ?12:15
malnot really sure how that GSI works and if it could be used somehow13:09
T42<Kalubeee> Bdj a GSI replaces the /system_root partition, this would limit your space to install software, technically it would probably be possible to install sailfish os to  system but there aren't really any benefits13:17
T42<edp_17> What is that GSI?13:17
T42<Kalubeee> As it has to meddle with a lot of device specific firmware which GSIs don't touch it wouldn't make sense13:18
T42<Kalubeee> @edp_17 Generic System Image, basically the idea is an Android ROM can be built just once and be installed on all devices13:19
T42<edp_17> I see, thanks.13:20
T42<Bdj %lastname%> @Kalubeee that would be great to do this for sailfish13:26
T42<Bdj %lastname%> And create hardware abstraction13:26
T42<Bdj %lastname%> Then a generic sfos image can be installed on lots of devices13:26
T42<Bdj %lastname%> An updating system  also is more easier13:26
T42<Bdj %lastname%> For example we can create a repository for main os13:26
T42<Bdj %lastname%> An a repo for every device specific libs13:26
T42<Kalubeee> @bdj you'd need still to create a hardware abstraction for every device, SFOS isn't quite modular enough to have that level of abstraction and it wouldn't be flashable as a GSI13:28
T42<Kalubeee> Another key feature of GSIs being that you use the same kernel your phone shipped with - selinux enabled and often with SFOS incompatibilities13:28
T42<NotKit> kernel limitation is the biggest problem in terms of making pure GSI13:47
T42<NotKit> configs could be probably dealt with some extra scripts13:47
T42<Lukapanio> @NotKit [kernel limitation is the biggest problem in te …], But mb we can port just by recompiling kernel13:49
T42<NotKit> maybe, but someone should work on this then13:50
T42<Kalubeee> @Lukapanio that would technically make it not a GSI, you would essentially have all of the issues compiling SFOS as a GSI would cause (2gb of space on most devices - event if you move /home to /data) and you would need sparse files per device at a minimum. And none of the benefits (flashing a ROM with a few fastboot commands without15:03
T42having to recompile)15:03
*** Sage is now known as Guest7458615:18
*** Guest74586 is now known as Sage_15:18
*** Sage_ is now known as Sage__15:18
*** Sage__ is now known as Sage_15:18
*** Oksana_ is now known as Oksana15:54
T42<infixremix> @Kalubeee [Bdj a GSI replaces the /system_root partition, …], Could be very useful, sailfish os could provide updates easily, no cache partition, available for all device16:53
T42ankaos was added by: ankaos17:12
ThaodanI think the hybris part could be made a gsi  but not the parts above I don't think that gsi's are made to work with anything that is not Android.17:47
ThaodanSailfishOS also doesn't fit into the Android partition layout and hacks around it. So there's no real single image.17:49
T42<Akatsu %lastname%> is there binder implementation of audio in packages? im not sure if i should build pulseaudio-modules-droid18:06
T42<Akatsu %lastname%> Hybris 15.118:07
T42<Deathmist1337> Akatsu there is `pulseaudio-modules-droid-hidl` but that's for call audio on >=15.1, not audio in general. without `pulseaudio-modules-droid` you will guaranteed get absolutely 0 audio out of your device18:22
T42<Akatsu %lastname%> ok, i thought there are some "binder" implementation of this, seems like its same like in Halium18:23
T42<Akatsu %lastname%> 6minutes passed on building rpm/ngfd-plugin-native-vibrator.spec should i cancel it and just comment this package in build script or wait, wait and wait?18:39
Thaodanyou could look which processes below the build scripts are still active and get hint of what's happening18:57
ThaodanAlso please set a username in telegram so that mentions from and to irc work18:58
T42<Akatsu %lastname%> `!! Please set the desired RELEASE variable in ~/.hadk.env to build an image for ` How does this variable looks like inside of it? export RELEASE= ? this variable is not part of SFOSHADK guide but later its mentioned during kickstart file creation19:02
T42<edp_17> Akatsu : Yes, this is how I use.19:04
T42<Akatsu %lastname%> so if i got this thing should i add this export into hadk logout from it and run build_packages once again or i can just go straight to packaging ignoring this error?19:05
T42<edp_17> If you added the RELEASE in hadk.env you can go build-packages. By the way that is not an error more like instruction how to go forward.19:08
T42<Akatsu %lastname%> can anyone give me guide on systemd/system/local-fs.target.wants/*  ? i need to create vendor and system mount20:40
T42<edp_17> You can find an example here: https://github.com/edp17/droid-config-hammerhead/tree/master/sparse/lib/systemd/system20:49
T42<elros34> They are usually generated during droid-hal build. If that is not the case for you check $ANDROID_ROOT/droid-hal-device.log whether fstab and uevent* are found. You shoudn't need to create them manually20:49
T42<Akatsu %lastname%> nice but i have /vendor inside od /system/ so how should i mount it?20:51
T42<Akatsu %lastname%> i have /system mounted just fine with /dev/mmcblk0p2720:52
T42<elros34> this is generated from your fstab and rc files20:52
T42<elros34> If you don't have vendor partition then check out/*/root whether you have there vendor symlink and add it to straggler files20:53
T42<Akatsu %lastname%> there is symlink20:53
T42<Akatsu %lastname%> from what i see i have issue with hal build log because it searches for ueventd.*.rc and my root has ueventd.rc * probably should be qcom20:55
T42<elros34> If you have ueventd.rc then it should be fine20:59
T42<adampigg> mal: around?21:04
T42<edp_17> @elros34 : You were  right with the fixup-mountpoints! I built a new image and it is booted up into the initial setup.21:04
T42<edp_17> However the app hangs (because of the invoker thing) and cannot completed. 😔21:05
T42<elros34> try to remove files which silica boost rpm  provides, maybe it will be enough21:06
T42<edp_17> Thanks. Last time when I tried to remove the other package you suggested it wanted to remove the half system.21:07
T42<edp_17> So, I kept that.21:07
mal@adampigg yes21:08
T42<elros34> @edp_17 it's probably in patterns or dependecies of some core app21:09
T42<adampigg> mal: pulse/vol keys ... pulse is playing system sounds, but i think the media volume is muted as no music is heard.....21:12
T42<adampigg> volume keys are not seen by CSD21:12
malshow output of evdev_trace -i21:14
T42<edp_17> @elros34 : Thanks. It seems there was problem with the modules too. I manually copied them (wifi and cifs) over to $ANDROID_ROOT//out/target/product/i9100/system/lib/modules/ but the ones in the final image were not compatible. I needed to transfer to the device again.21:14
T42<adampigg> mal: https://bpaste.net/GMEQ21:15
mal@adampigg ok, that looks normal21:17
piggzi know :)21:18
T42<elros34> @edp_17 weird. From where did you copy them? In your sparse you have wrong script with sudo insmod /system/lib/modules21:21
T42<edp_17> Finnaly! With my last image I can login to the jolla store. 😄21:21
mal@adampigg just to be sure, is hw config file correct?21:22
T42<adampigg> mal: maybe not21:22
T42<edp_17> I copied them from here mer/android/droid/out/target/product/i9100/obj/KERNEL_OBJ/  (drivers/net/wireless/bcmdhd  and fs/cifs) to mer/android/droid/out/target/product/i9100/system/lib/modules/21:23
T42<edp_17> @elros34 : what is wrong with that script?21:25
T42<elros34> everything:)21:25
T42<edp_17> If the dhd.ko module is correct, the script loads it correctly.21:25
T42<edp_17> 😄21:25
mal@adampigg the file seems reasonable, except maybe Receiver=0, I assume you have a normal phone speaker for calls?21:25
T42<adampigg> yes21:26
T42<elros34> @edp_17 sudo is not needed, use modprobe instead insmod, use your compiled module form /lib/modules/ not form cyanogenmod21:27
T42<edp_17> @elros34 : I know that is not nice, but it does the trick. However I am willing to do it properly if you let me know.21:27
T42<adampigg> mal: changing that didnt effect the vol keys21:28
T42<edp_17> @elros34 : Oh, I didn't realize there were all of the modules. First I'll manually try to load dhd from there.21:29
mal@adampigg I know, just happened to notice that so thought I should mention it21:30
T42<edp_17> @elros34 : Yep, the dhd loaded, so I will update the script to use the ones from there. One issue less. 😄21:31
T42<adampigg> mal: so, what route do you think the volume keys take?21:31
T42<elros34> use it same way as in faq example so just: modprobe dhd (parameters)21:31
T42<edp_17> I have tried 'modprobe dhd firmware_path=/system/etc/wifi/bcmdhd_sta.bin nvram_path=/system/etc/wifi/nvram_net.txt' but it didn't load the module. What's wrong using the insmod?21:34
T42<Akatsu %lastname%> Can anyone look into this log?21:34
T42<Akatsu %lastname%> https://pastebin.com/MntsqJqE21:34
T42<Akatsu %lastname%> For now i can say i need to bundle libminisf.21:34
T42<Akatsu %lastname%> But i do need to solve this error with "Failed to initialize property area" inside second stage init too.21:34
T42<elros34> @edp_17 insmod requires full path and does't handle module dependecies21:37
T42<adampigg> mal:  i found _something_ ... i dont know what it means21:37
T42<adampigg> on mido, when i press vol keys, nothing of interest is logged in journal21:38
T42<adampigg> in pine, i get this21:38
T42<adampigg> Feb 04 21:28:06 PinePhone mce[2246]: tklock.c: tklock_notif_slot_set(): notification 'mce_volume_key' added21:38
T42<adampigg> Feb 04 21:28:08 PinePhone mce[2246]: tklock.c: tklock_notif_slot_free(): notification 'mce_volume_key' removed21:38
mal@adampigg csd app monitors Qt::Key_VolumeDown and Up21:38
T42<edp_17> @elros34 : Thanks. Then I need to find out how to call modprobe properly. 😄21:38
mal@adampigg https://git.sailfishos.org/mer-core/mce/blob/master/tklock.c#L1538 not sure what that does21:40
T42<elros34> Akatsu please set nickname in telegram settings. Did you add all needed files to stragglers?21:40
T42<adampigg> mal: maybe false alert, i added -v to mce startup on pine21:41
T42<adampigg> mal: yeah, theyve gone now21:42
T42<adampigg> :/21:42
mal@adampigg I see those on x10 also21:42
T42<adampigg> ok, so, im very confiused where the events go!21:44
mal@adampigg check the VerificationKey.qml in /usr to see what it does, there is some resource it asks21:46
malmaybe something related to that is the issue?21:46
T42<edp_17> @elros34 : Interesting. If I call the 'invoker --type=qt5 jolla-settings' from command line, it works and doesn't hang. If I update the .desktop and use the same command, it hangs.21:48
T42<edp_17> Plus, some apps still hangs if I remove the silica from the --type=silica-qt5 in their .desktop, but some others do work.21:48
T42<elros34> Have tried to get rid of that booster by removing files or masking service: ln -s /dev/null /etc/systemd/user/booster-silica-qt5.service21:53
T42<edp_17> This is what I am trying currently. I have not found the package yet. I am doing this masking. What is it for and why can it help?21:55
T42<edp_17> Nope, I ran this command, then opened jolla-settings. It is the same, hangs after the first few touch.21:56
T42<elros34> reboot...21:56
T42<elros34> or stop service21:56
T42<adampigg> mal: maybe spot something in a full log? https://bpaste.net/Q3CQ21:57
mal@adampigg nothing obvious22:02
T42<adampigg> yeah, i cant see anything either22:02
mal@adampigg did you check the file I mentioned and add some debug printing the to see what the variables are22:07
T42<edp_17> @elros34 : after the reboot the jolla-settings doesn't hang and works. How can I make this permanent in for the new images?22:09
T42<elros34> add symlink to sparse or better strace/gdb and fix issue22:09
T42<edp_17> strace/gdb?22:10
T42<adampigg> mal: volumeKeysResource.acquired = false22:11
T42<elros34> @edp_17 most basic stuff for debugging applications22:12
T42<edp_17> Oh, I see you meant use the strace or gdb to find out the root cause and fix.22:14
T42<adampigg> mal: so, does that mean i dont get the scale button resource?22:15
mal@adampigg interesting, https://git.sailfishos.org/mer-core/nemo-qml-plugin-policy22:21
malcheck what that does22:21
T42<edp_17> Is the content in jolla store kernel or device related?22:22
T42<edp_17> I am talking about sfos apps not the AD.22:22
T42<adampigg> mal: ill head down that path tomorrow!22:23
mal@adampigg still wondering if this is related the user/group issues you had22:26
T42<adampigg> as am i22:26
T42<adampigg> but22:26
T42<adampigg> ive already created and added to groups media, media_rw, camera22:26
T42<adampigg> now, groups look very similar to mido22:26
mal@adampigg did you make sure those are created early enough during image build?22:51
T42<edp_17> @elros34 : Do you have an idea how I can use hwcomposer instead of fbdev? (When I use the screen it gets very hot and the battery drain is also huge. However when screen is off, battery life is quite good.)22:57
T42<elros34> I guess you can't. Take it with a grain of salt but I think your slow ui can be related to how vsync is handled in v10 backend in qt5-qpa-hwcomposer-plugin. All other devices use v11 or v20 and they schedule repaint on every vsync but that is not the case for v10. I might be totaly wrong because of my little understanding how it works23:02
T42<edp_17> Thanks for explaining. Yeah, the other annoying thing is related to the screen refresh that I already described. In some apps on certain screens I need to touch the screen to force a refresh and to see the updated screen. I already got used to it but if somebody would like to try this out, this doesn't give a good impression. 😄23:06
T42<edp_17> Oh, and the charging messages are also completely mixed up. When the device is off the charger sometimes I get a message "too low power to charge". (Lol, of course as the phone is not plugged in.)23:08
T42<edp_17> When the phone is on charger and I pull it off, the icon still shows charging.23:09
T42<elros34> Check what udevadm monitor prints when you dissconnect usb. Maybe there is no POWER_SUPPLY_PRESENT/ONLINE events23:16
T42<edp_17> It seems there is no POWER_SUPPLY_PRESENT/ONLINE event: https://pastebin.com/tX58Z2Wp23:20
T42<elros34> udevadm monitor --help then add --udev23:24
T42<elros34> sorry  I mean --property23:26
T42<edp_17> Thanks. This way there is online 0 and : https://pastebin.com/435HB4pd23:29
T42<edp_17> I have a feeling that what I described happens because of the screen refresh issue.23:31
T42<edp_17> @elros34 : Can you help me out with the bluetooth? I have followed hadk-faq 10.5 but when I tried to 'Build rfkill middleware and add to patterns' I got this: https://pastebin.com/UqN6y63d23:42
r0kk3rzyou cant use that with bluez423:42
T42<edp_17> Oh.23:42
T42<edp_17> Hi r0kk3rz. Thanks.23:43
T42<elros34> there is second spec file there without bluez5 dependecies23:43
T42<edp_17> How can I make this work then. Similar way how mautsala did for the hammerhead?23:43
T42<elros34> I don't know what mautsala did but you should be able to build it with -s rpm/bluetooth-rfkill-event.spec option23:46
r0kk3rzyou can try bringing it up manually before worrying about stuff like that23:47
r0kk3rzusing hciattach23:47
T42<edp_17> For hammerhead we created two services dev-ttyHS99.device and hciattach.service in /lib/systemd/system. (Example: https://github.com/edp17/droid-config-hammerhead/tree/master/sparse/lib/systemd/system) and the bt started working.23:48
T42<edp_17> @elros34 : I'll try this now.23:48
T42<edp_17> @r0kk3rz: Yeah, this is what I meant how matusala (and later I) did it for the hammerhead.23:49
r0kk3rzthat rfkill-event middleware is the more 'standard' way to do that23:50
r0kk3rzbut it more or less performs the same action23:51
T42<edp_17> I have tried the command: 'rpm/dhd/helpers/build_packages.sh --mw=https://github.com/mer-hybris/bluetooth-rfkill-event -s rpm/bluetooth-rfkill-event-hciattach.spec' but got this https://pastebin.com/Pn6hD6Pg23:54
r0kk3rzwrong spec23:55
r0kk3rzuse the other one23:55
T42<edp_17> My bad, I see the difference now.23:56
T42<edp_17> Guess I should go to bed now. 😄23:56

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