Saturday, 2021-04-17

riniguspiggz: working on build engine that can help with the ports as well. few more issues to resolve (faster to fix than explain) and I hope to be able to build port packages automatically on PC07:47
rinigusright now targeting something like to describe port packages (many packages are still missing). this is expected to be equivalent to the corresponding OBS repo07:49
rinigusfor maps & flatpak, that worked already. ports have bit more difficult case, but I think I figured out the solution as well. would just have to implement and see if it works07:50
T42<adampigg> Rinigus, you will have to tell me more07:51
rinigus@adampigg: I will. It is an attempt to have a build system that handles dependencies and allows to build interconnected packages as obs does. When it works for port as well,  I can explain in details here.07:57
T42<adampigg> Spiiroin: latest version any better?08:00
T42<b100dian> Hi, I don't understand this piece of advice from hadk:09:37
T42<b100dian> >  Then add  - droid-hal-hammerhead-detritus to droid-configs/patterns/jolla-hw-adaptation-hammerhead.yaml (substitute as appropriate for your device)09:37
T42<b100dian> First: I don't have jolla-hw-adaptation, I'd probably have to add them to patterns-sailfish-device-adaptation-tucana.inc09:37
T42<b100dian> Second: shouldn't there be a file named "droid-hal-tucana-detritus" ? how would one generate one?09:37
T42<b100dian> Just adding the reference produces this error:09:38
T42<b100dian> > Processing files: droid-hal-tucana-detritus-0.0.6-202104170918.aarch6409:38
T42<b100dian> error: File must begin with "/": {nil}09:38
T42<b100dian> when running "rpm/dhd/helpers/ --droid-hal"09:38
T42<b100dian> wow, just asking here makes me smarter:) Turnes out I had a {nil} in straggler_files (as opposed to %{nil})09:42
T42<elros34> I think detritus package is added automatically as a dependency if you have defined straggler_files09:48
mal@b100dian like elros34 said you only need to add the straggler_files things to the spec file as instructed, the dependency is added automatically10:55
T42<b100dian> Thanks mal and Elros, that's what happened probably - there was an error in my straggler_files list ^10:55
T42<b100dian> Ok, it does not boot.. I first thougth it was missing dtbo but I built that too. I get some 'device tree not found' errors, a mount("selinuxfs") error, and references to fts being input4 (was my guess good or I just sent it over to the wrong device?)
T42<b100dian> I did find some selinux-related props in my device tree, let me remove those..
T42M4rcosG4briel was added by: M4rcosG4briel11:31
mal@b100dian which android base?11:34
mal@b100dian based on your link you have hybris-17.1, did you remember to run "export TEMPORARY_DISABLE_PATH_RESTRICTIONS=true" before running make hybris-hal11:35
T42<b100dian> it's lineage-17.1, android 10 - I did not run that, thanks for pointing it out11:36
T42<b100dian> oooh you mean those were not warnings 🤦‍♂️11:38
T42<b100dian> (for reference, I saw "Disallowed PATH tool "cpio" used" etc. but I thought if it doesnt break it, it is just a warning..)11:39
mal@b100dian the warning about cpio causes kernel initrd to be empty :)11:40
T42<b100dian> I still see it for xmllint though.. rebuilding, my hopes are high again:D11:41
malxmllint is not causes issues, at least I haven't noticed any11:42
T42<b100dian> The "warnings" still appear. To be sure I added cpio to build/soong/ui/build/paths/config.go -> but this is not the issue at hand, even though the .img files grew a bit (so it would have been an issue in the future)11:56
T42<b100dian> so the output I got in /sys/fs/pstore is the same as above11:57
T42<elros34> did you disable skip_initramfs handling?12:14
T42<b100dian> not sure how to do that, so no. Let me duckduckgo it12:16
T42<b100dian> I do see in the log that "skip_initramfs rootwait ro init=/init androidboot.dtbo_idx=30" is appended to the cmdline but am not sure from where (not from my device tree)12:18
T42<elros34> from bootloader. In kernel/init/initramfs.c revert commit which adds support for skip_initramfs12:18
T42<b100dian> I was just going to write this.. ok, searching for the patch, thanks @elros3412:19
T42<b100dian> I'm actually surprised xiaomi published the full history for the kernel, git blame did find it:)12:30
T42<b100dian> already booting for 30 seconds woow:D12:31
nergzd723hey, all! I want to post SailfishOS to Sony Xperia Z5 ( sony-sumire ). Is there anything specific I should look at other than HADK?12:42
T42<b100dian> (reboots in 70 seconds now, exact reason unclear
T42<b100dian> shouldn't the vendor partition be mounted by then? I don't see any sda25 output (the vendor on tucana)12:51
T42<elros34> are you on 2323 port?13:03
T42<zinstack625> @b100dian is vendor partition fixed in fixup-mountpoints? How is it mentioned in fstab? Is systemd mount service present (/usr/lib/systemd/system/)?13:04
T42<elros34> there is no sign of preinit in dmesg, not to mention systemd and mount units13:06
T42<b100dian> the fixup-mountpoints is edited, let me see fstab. Havent yet tried telnet, trying next13:08
T42<zinstack625> Since systemd isn't even started, perhaps data isn't mounted?13:09
T42<b100dian> the /usr/lib/systemd folder is present in the rootfs, I believe I should check on device now13:11
T42<b100dian> I don't get the Mer debug telnet messages and no telnet doesn't work (hybris-boot.img)13:19
T42<b100dian> I have this line in but it doesn't seem to get into rootfs: `PRODUCT_COPY_FILES += \13:22
T42<b100dian>     $(LOCAL_PATH)/rootdir/etc/fstab.qcom:$(TARGET_COPY_OUT_RAMDISK)/fstab.qcom`13:22
T42<elros34> That and 70s boot suggest usb rndis issue13:22
T42<elros34> try to disable CONFIG_RNDIS_IPA and enable CONFIG_USB_CONFIGFS_RNDIS13:33
T42<b100dian> thanks, trying that next. While building - Any idea where the fstab should be generated from?13:51
T42<elros34> when you run make, fstab from your device/* should be installed somewhere in out/. Then it's used by droid-hal to generate systemd mount units13:54
T42<b100dian> Gothca. I have to backtrack why it worked in lineageos14:15
T42<elros34> but what is wrong with fstab? Sometimes it is not copied to out/ and you need to do it manually. You should skip it anyway because now you stuck in initramfs and tha tis irrelevant14:17
poetasterany one every experience the orientation sensor disappearing?14:21
T42<b100dian> ok, I got telnet, surprisingly on 23, that worked @elros3414:46
malpoetaster: what do you mean, randomly during use it stops working?14:46
poetasterno. clean install (volla phone or GS290) install stuff, fire up sdk, debug install app, shortly thereafter, no orientation sensor14:47
poetastersame behaviour two phone. volla port14:47
malanything interesting in logs?14:48
T42<b100dian> ```+ mount /dev//dev/block/sda32 /data14:49
T42<b100dian> mount: mounting /dev//dev/block/sda32 on /data failed: No such file or directory```14:49
T42<b100dian> oook, that was my fixup-mountmoints containing /dev/block's!14:52
malso you didn't look at the other lines for example how to do fixup14:53
poetasterlogs at:
T42<b100dian> Yeah, I was so happy with sublime multi-cursor editing that paste from ls -l :)14:54
poetastersorry mal, you were'nt talking to me.14:54
malpoetaster: I was asking logs from you14:55
poetasterin the repo I posted. big jctl
T42<b100dian> sorry I was so excited I got telenet:)14:56
poetastertelnet is exciting ;)14:56
poetaster<mal> piggz didn't see anything interesting in the logs.14:56
malpoetaster: yeah, not really much related to sensors in those logs14:57
T42<elros34> where in that 25minutes journal you lost sensors or started messwerk/csd to test sensor?14:57
malyeah, that would help also14:58
malalso getting verbose logging from sensorfw could help14:59
T42<elros34> you have flood of suspend failures that makes logs huge14:59
poetaster<T42> I always noticed to late. did systemcl status sensorfw as soon as I notice15:00
poetaster<mal> sensorfw doesn't seem to log at all?15:00
T42<b100dian> @nergzd723 as you can see, following the HADK and the then searching this IRC/telegram channel works great, and if not, asking here is really really helpful. I believe you can also look at another Xperia community port maybe similar to your Z515:01
poetaster<T42> I think piggz debug kernel is noisy by design.15:01
T42<elros34> poetaster: so I guess that journal might not even contains relevant logs.15:01
malpoetaster: is it even running?15:02
T42<elros34> Enable persistent logging and restart device or  IIRC systemd-journal15:02
poetaster<mal> is what running?15:02
malpoetaster: sensorfw15:02
poetasterof course.15:02
malpoetaster: does messwerk app show sensor data?15:02
poetaster<mal> sec.15:03
poetaster<mal> I lie!    Active: inactive (dead)15:04
poetasterbut this is new15:04
malpoetaster: which android base version does that device use?15:05
poetaster<mal> volla's release is a 9 I believe.15:05
malno need to add those <> to the nick when highlighting in irc15:07
poetasterI'm seeing Failed to restart sensorfw.service: Unit sensorfw.service failed to load: No such file or directory.15:07
poetasterbut this is not the volla but the gs290 (same hardware)15:08
T42<elros34> it sensorfwd.service not sensorfw.service15:08
malhmm, check "pkcon search sensorfw"15:09
poetastert42:it's there.15:10
poetastermal: it's also not generating the errors I was accustomed to seeing (about failing to register the orienation sensor)15:11
malwhat does test_sensors return? it's in libhybris-tests package15:11
poetastersec. I'm installing messwerk...15:12
poetastereverything but gps and proximity seems dead.15:14
poetasterMy app does some library stuff but all of it Jolla conform.15:15
T42<elros34> but did you start sensorfwd first?15:15
poetasterT42: restatarted15:16
T42<elros34> what about test_sensors mal was asking?15:16
poetasterT42: where to I find that package?15:17
malpoetaster: T42 is a telegram tunnel bot, to highlight the person talking via that use the nick after it inside those <>15:17
T42<elros34> @nickname is enough. just run test_sensors, most likely you have it installed15:17
poetasterOk, I have enough trouble keeping my xmpp bot and mattermost bridge from driving me mad.15:17
malpoetaster: yeah, like this @elros3415:18
poetaster@mal that's what we do at work :)15:18
poetaster@mal test_sensors not found15:18
malpoetaster: but that @ is only then talking to those using telegram, for irc @ is not needed15:19
malthis those thing of using different systems is complicated15:19
malpoetaster: like I said, it's in libhybris-tests package, you might need to install that first15:19
poetastermal: where? I mean, I've got the source so I guess I could compile it.15:23
malpoetaster: I think this should work assuming repos are correctly setup: pkcon install libhybris-tests15:23
poetasterI tried that first.15:23
maldid you build the image yourself?15:24
poetasterSorry, got it. No the image is one supplied by Volla (it's piggz though)15:24
poetasterI got15:24
poetastermal: `hwmod != NULL' failed.15:25
malhmm, ok15:26
malso then you probably need to try to run sensorfwd with verbose logging to see what it sees15:27
poetastermal: so modify the unit file ?15:27
malyes or stop the service and run the command manually15:28
poetastermal: that is a better idea ;:)15:29
malchange the log level15:31
poetasterto error or info15:31
poetaster--log-level=error has the same output.15:33
malno to show more, can't remember if next ones are verbose or test or what, sensorfwd --help should tell15:33
poetastermal: test/debug15:36
nergzd723<T42 "<b100dian> @nergzd723 as you can"> thanks for tips!15:36
mal@adampigg ping15:38
poetaster@mal, he's in the loop15:40
* nergzd723 < >15:40
T42<elros34> shouldn't sensorfw try to load binder instead hybris adaptors? According to they switched to binder plugin15:41
malpkcon search sensorfwd should tell what has been installed15:44
malactually pkcon search sensorfw15:44
poetaster<fetching beer>15:45
poetastermal: hybris-libsensorfw-qt5-binder-0.12.2-1.4.1.jolla.armv7hl (installed)Sensor Framework Qt515:48
T42<elros34> I wonder whether original image used non binder plugin and then it gets updated to binder when you installed some apps (know bug in pkcon)15:51
poetaster@elros, that sounds plausible. it 'seems' to happen after I've done some debugging with one of the apps I'm developing.15:52
poetasterI do some post install stuf ... install -D -t %{buildroot}/%{_datadir}/%{name}/lib/ \15:54
poetasterbut looking at 'this version' that doesn't seem to be it.15:54
T42<elros34> you can unpack original sfos image to see which plugin was initially installed15:55
T42Michael %lastname% was added by: Michael %lastname%16:04
T42<elros34> poetaster oh you might have too new plugin version installed16:07
poetasterelros: that sounds possible too.16:10
T42<elros34> I can see that on obs there is no versioned testing repo and on devel sensorfw plugin was updated16:11
poetasterelros: how to loop mount an android lvm image ;) (I'm reading adb's man)16:13
T42<elros34> ah it's lvm so I do not know but for sure on pc it will be easier then using adb. Do you use image from sailfishos-porters-ci and if yes which one?16:14
poetasterelros: just looking at  /usr/lib/sensord-qt5/ I can see it's probably that ... almost all from okt. 2020 and then some from march 13.16:18
T42<elros34> chroot and use rpm -q info hybris-libsensorfw-qt5-binder. If that returns 0.11.9 then updated version might cause your issues16:21
T42<b100dian> got into the 2323 telnet next.. (w/o dhcp). The init.log is and last 200 kernel logs are
T42<b100dian> I notice you guys are talking about  binder: is this the one used for nfc?16:23
poetasterVersion     : 0.12.216:24
poetaster1.4.1.jolla that's probably it. hmmm.16:25
T42<elros34> on unpacked image?16:25
poetasterno that's on the phone.16:25
poetasterI can't loopback mount the image. :)16:25
T42<b100dian> try sdat2img input.img -o output.img16:26
T42<elros34> yeah I know you have 12.2 on phone you already showed that long time ago. So Can you show me url for image you are using?16:26
malwhich sfos version is that image?16:27
T42<elros34> 3.4 that is why  I asked16:27
poetasterelros: sorry. it's from the ubports installer ... looking for it.16:28
malso another reason to not use devel repos in images provided to users16:29
T42<elros34> yeah, there is no testing repo set up for that device16:29
poetasterI'm not using the devel updates. but that's not what you mean?16:30
T42<elros34> you are using devel repo and due to bug in pkcon (or backend), every time you install package, everything else is upgraded16:31
poetasterof course.16:31
poetasterit's odd that it 'appears to happen' when I install my own apps via sdk. or is that what you mean?16:32
poetasterI'm probably just overthinking it.16:33
poetasterelros: I believe
T42<elros34> every time you packagekit installs something so probably your own app via sdk also triggers it that is why it's so annoying16:40
poetasterelros: so I need to make those read only.16:44
poetasteror add a depends :)16:45
T42<elros34> I would disable devel repos after first boot16:45
T42<b100dian> It seems I don't have a /system folder in the rootfs. shouldn't it be there instead of mounted?16:50
T42<elros34> @poetaster ssu lr to show repos, ssu dr reponame dto disable repository16:56
poetaster@elros, thanks. if only I could mount the image.16:58
poetaster@elros: these are sparse android images ?17:00
poetaster@elros, that's what file says.17:00
T42<elros34> I am not sure what do you mean now. If you want to confirm version installed on image then find build you are using here: Then open it and you will find build log which also contains list of installed packages with versions17:03
poetaster@elros, I thought I'd just loop mount the image file and that turned out to be difficult. I was hoping to find the libraries too.17:05
poetaster@elros, as you suggested it's 0.11.9 in the image.  and the img is a sparse image.17:09
poetasterelros34: thanks for your help.17:34
poetasterelros: fixed.17:51
T42<elros34> be reflashing?:)17:54
poetasterelros: adaptation-community adaptation-community-common18:08
poetasterelros: off, copied the libs back and all's well.18:08
poetasterthe loopback mount wasn't the issue. without lvm you're toast, no? :)18:09
poetasterthe app (imageworks) is in QA...18:09
piggzmal: poetaster: hi ... ive been painting the house .... i do not enjoy painting a rough render18:31
piggzwhats the summary .. have u fixed it yet?18:31
piggzah, becuase i havnt got round to a testing: image yet....18:44
piggzwell, port is still wip18:44
malpiggz: yeah, the problem was just the devel repo being too new19:20
malpiggz: even if the port if wip it's probably better to use testing repos19:21
piggzmal: yeah, youre right ... i just havnt got to it ... will try this weekend19:36

Generated by 2.17.1 by Marius Gedminas - find it at!