| *** Sellerie0 is now known as Sellerie | 05:13 | |
| rinigus | Thaodan thanks, I'll look it up how it worked. Will first split the packages into common and port specific ones | 08:43 |
|---|---|---|
| *** IHAVENONICK is now known as Zucca | 08:45 | |
| rinigus | I presume that anything which does not depend on droid hal/libhybris for building can be called common. And if it is from mer repositories, not adjusted for specific port | 08:45 |
| rinigus | Please correct if I'm wrong | 08:46 |
| T42 | J_Jul_T was added by: J_Jul_T | 10:20 |
| rinigus | update re tbuilder & ports: common packages are covered at https://github.com/sailfishos-sony-tama/tbuilder-common | 15:22 |
| rinigus | hopefully, we can setup hosting soonish and then it can be used for providing common packages for ports. | 15:22 |
| T42 | <b100dian> folks, I've moved my usb ssd to another computer and cannot resume from the exact place I was with my attempt to port. For one, I deleted ~/hadk/installroot and it doesnt seem to be created by build_packages —droid-hal. Was it created by this or by some other step? | 15:30 |
| T42 | <b100dian> Second question would be: I don't seem to get systemd to mount the units that it generated. How can I debug this? I dont' see a specific systemd failure in dmesg or journalctl | 15:32 |
| T42 | <elros34> inspect mount units manually: cat /usr/lib/systemd/system/*.mount. Do they contain path to block device or by-name path? | 15:39 |
| T42 | <b100dian> some are, such as metadata: | 15:43 |
| T42 | <b100dian> ``` # find /usr/lib/systemd/system/ -type f -iname \*.mount -exec grep tucana -q {} \; -print -exec grep What {} \; | 15:43 |
| T42 | <b100dian> /usr/lib/systemd/system/vendor.mount | 15:43 |
| T42 | <b100dian> What=/dev/sda24 | 15:43 |
| T42 | <b100dian> /usr/lib/systemd/system/vendor-dsp.mount | 15:43 |
| T42 | <b100dian> What=/dev/sde47 | 15:43 |
| T42 | <b100dian> /usr/lib/systemd/system/sys-fs-pstore.mount | 15:43 |
| T42 | <b100dian> What=pstore | 15:43 |
| T42 | <b100dian> /usr/lib/systemd/system/metadata.mount | 15:43 |
| T42 | <b100dian> What=/dev/block/by-name/metadata | 15:43 |
| T42 | <b100dian> /usr/lib/systemd/system/vendor-bt_firmware.mount | 15:43 |
| T42 | <b100dian> What=/dev/sde26 | 15:43 |
| T42 | <elros34> I guess metadata is not important but if you have "waiting for complete device initialization" in logs you can try to remove symlink from /etc/systemd/system/local-fs.target.wants/metadata.mount and reboot. If that doesn't help then show your full journal/dmesg | 15:49 |
| T42 | <b100dian> (the filter on tucana is for this line: "# From ./device/xiaomi/tucana/rootdir/etc/fstab.qcom :" | 15:49 |
| T42 | <elros34> IIRC mount units are also created from init*rc files so grep tuncana can hide something | 15:50 |
| T42 | <b100dian> that's correct, here's the full list https://pastebin.ubuntu.com/p/xNhzzGMQYk/. I don'thave a "waiting or complete (..)" in my logs | 15:52 |
| T42 | <b100dian> hmm maybe they are just not mounted because `# ls /usr/lib/systemd/system/local-fs.target.wants/ | 15:54 |
| T42 | <b100dian> systemd-remount-fs.service tmp.mount` | 15:54 |
| T42 | <elros34> symlinks are in /etc/systemd/system/local-fs.target.wants/ | 15:54 |
| T42 | <elros34> at least for droid mounts | 15:55 |
| T42 | <b100dian> Oh, I see.. I don't have a local-fs.target.wants in /etc/systemd/system/ | 15:58 |
| T42 | <elros34> you don't? What rpm -q --scripts droid-hal-$DEVICE returns? DO you have /usr/lib/droid/all-units.txt? | 16:01 |
| T42 | <b100dian> on the device? | 16:02 |
| T42 | <elros34> best would be on device but droid-local repo should be fine too | 16:02 |
| T42 | <b100dian> output on device https://pastebin.ubuntu.com/p/YgwtgJcbrw/ | 16:04 |
| T42 | <b100dian> (includes the all-units.txt contents) | 16:06 |
| T42 | <elros34> so it should works, try reinstalling droid-hal-device package to see if symlinks are created | 16:06 |
| T42 | <b100dian> hmm any pointers how to do that? on the device I have no connection and in the host Platform SDK prompt I get "No provider of 'droid-hal-tucana' found." (with zypper in) | 16:15 |
| T42 | <b100dian> should I just run all the 7.2.1 commands on host | 16:16 |
| T42 | <elros34> copy droid-hal-device*rpm from local repo to device and install it or just execute that postinst script | 16:16 |
| T42 | <elros34> you didn't change anything so you don't need to run 7.2.1 at all now | 16:18 |
| T42 | <b100dian> I've pushed `hadk/droid-local-repo/tucana/droid-hal-tucana/droid-hal-tucana-0.0.6-202105011144.aarch64.rpm` through twrp to /data/.stowaywais/sailfishos/ and `rpm -i /droid-hal-tucana..` and still they dont' appear | 16:26 |
| T42 | <elros34> I usually use rpm -U but that should be irrelevant. Does 'systemctl --no-reload preset vendor.mount' creates symlink? | 16:29 |
| T42 | <b100dian> that command takes an unusually long time.. | 16:33 |
| T42 | <b100dian> but yes, it creates the etc link! | 16:34 |
| T42 | <elros34> didn't you had anything in logs during that? | 16:34 |
| T42 | <b100dian> lots of "binder.. transaction failed" and then this https://pastebin.ubuntu.com/p/rPzTym8YpG/ | 16:39 |
| T42 | <b100dian> > Failed to get initial list of names: Connection timed out | 16:40 |
| T42 | <b100dian> dbus seems to not be running | 16:40 |
| T42 | <b100dian> or it is.. systemctl status dbus.service says it is | 16:41 |
| T42 | <elros34> I meant during that symlinks creation but I think most important would be some errors during droid-hal-device.rpm installation. | 16:42 |
| T42 | <elros34> maybe rpm -vv -U droid-hal-device.rpm will show something | 16:43 |
| T42 | <b100dian> added a —force since normally it shows 'already installed' https://pastebin.ubuntu.com/p/DXWTs2SbjS/ | 16:47 |
| T42 | <b100dian> I think + '[' 2 -eq 1 ] is not good at about line ~1234 | 16:48 |
| T42 | <elros34> ah sorry I miss it, it's because upgrade | 16:50 |
| T42 | <b100dian> @elros34 , this may be something I have changed without knowing on my machine since I ported the ssd from another computer and the mounts used to work.. So the droid-hal-$DEVICE package should also make the local-fs.wants appear in etc in the PlatformSDK, isn't it? I believe I should start from here. Remove and re-add the package? | 16:58 |
| T42 | <b100dian> (sorry for the noise created from this, I am traveling and added the ssd to the laptop..) | 16:59 |
| T42 | <elros34> not in platform but in sdk target if you built droid-hal. Force removal droid-hal-device without uninstalling everything else then installing it might be good idea | 17:02 |
| T42 | <b100dian> ok, this is something I need to look into very attentively since I do remember that I had to reinstall the tooling and the target because that was also referenced by ~/.scratchbox2 and that I did not port from another machine.. | 17:09 |
| T42 | <elros34> so build_packages.sh -d then sb2 -t $VENDOR-$DEVICE-$PORT_ARCH -m sdk-install -R ls -al /etc/systemd/system/local-fs.target.wants/ | 17:12 |
| T42 | <b100dian> hmm `$ sb2 -t $VENDOR-$DEVICE-$PORT_ARCH -m sdk-install -R ls /etc/systemd/system | 17:21 |
| T42 | <b100dian> basic.target.wants getty.target.wants multi-user.target.wants nemo-devicelock.service` | 17:21 |
| T42 | <b100dian> How can I uninstall it, if I set after -R `zypper rm droid-hal-tucana` it says No provider of 'droid-hal-tucana' found. | 17:22 |
| T42 | <elros34> did you run build_packages.sh -d first? | 17:24 |
| T42 | <b100dian> yes - and ` zypper se droid-hal-tucana` shows the pacakge (and some more, -devel etc) | 17:25 |
| T42 | <elros34> so why zypper rm fails if it's installed? Didn't you mix up arm32/64 targets? | 17:26 |
| T42 | <elros34> assuming it is installed, zypper se -s droid-hal will tell you | 17:27 |
| T42 | <b100dian> it's aarch64 indeed | 17:28 |
| T42 | <b100dian> I don't know why it fails at rm though, am continuing looking | 17:30 |
| T42 | <b100dian> this is search output https://pastebin.ubuntu.com/p/vXnPBrqts9/ | 17:33 |
| T42 | <elros34> so obviously install it first. I was pretty sure that build_packages.sh will install it automatically. Are you sure script didn't fail? | 17:36 |
| T42 | <elros34> but maybe I mix up something | 17:36 |
| T42 | <elros34> also echo $VENDOR-$DEVICE-$PORT_ARCH prints xiaomi-tucana-aarch64 for you? | 17:39 |
| T42 | <b100dian> yes - to the last question | 17:42 |
| T42 | <b100dian> I touched the spec file and run build_packages -d again, * Building successful, adding packages to repo | 17:42 |
| T42 | <b100dian> I'm not sure if this is really installing, using the previous sb2 invocation with -R zypper in droid-hal-tucana prompts me to upgrade to a newer version (different timestamp) | 17:44 |
| T42 | <b100dian> aaand yes, after upgrade `$ sb2 -t xiaomi-tucana-aarch64 -m sdk-install -R ls /etc/systemd/system | 17:44 |
| T42 | <b100dian> basic.target.wants getty.target.wants local-fs.target.wants multi-user.target.wants nemo-devicelock.service` | 17:44 |
| T42 | <b100dian> so maybe only the —mic command installs everything..? | 17:44 |
| T42 | <b100dian> the build_packages —configs does install the configs package though | 17:46 |
| T42 | <elros34> so to be sure droid-hal-tuncana is installed in target and still symlinks are missing? | 17:46 |
| T42 | <b100dian> nope, this zypper in droid-hal-tucana triggered an update that added the symlinks, thanks! | 17:47 |
| T42 | <b100dian> here is the output from build_packages.sh -d https://pastebin.ubuntu.com/p/S64ct7cnqs/ | 17:47 |
| T42 | <b100dian> So I had manually updated it. But I still don't know what was wrong _before_ this | 17:48 |
| T42 | <elros34> that still doesn't explain why postinstall script wasn't executed when you built image | 17:48 |
| T42 | <elros34> you can unpack sfos tarball and check for symlinks if they are missing run mic again and search for errors in output | 17:51 |
| T42 | <b100dian> Yes, I'll manually check the tar before installing now. As for explanation - there was at least 1 time in the past when I had `%define makefstab_skip_entries /vendor`in the spec file, and probably that version of the package was wrong but even after removing the line it it was not updated..? | 17:56 |
| T42 | <elros34> but you are missing all symlinks not only vendor so for sure not only that is/was wrong | 18:01 |
| T42 | <b100dian> Sure, and I was rebuilding with that set to #define (so commented out) for some time.. I don't have the full explanation. But thanks again for navigating me through how this works, hopefully I won't end up in this situation without understanding! | 18:08 |
| T42 | <b100dian> I have a build issue in build_packages —mw now... `/srv/mer/toolings/SailfishOS-4.0.1//opt/cross/bin/aarch64-meego-linux-gnu-ld: build/release/main.o:(.data.rel.ro+0x8): undefined reference to `_nfc_plugin_dbus_neard'` | 18:29 |
| T42 | <elros34> not sure which package fails but you may try to run git clean -xdf in hybris/mw/"that package" | 18:49 |
| T42 | <b100dian> nfcd. Thanks again for the advice, spot on! | 19:23 |
| T42 | <b100dian> humph.. the final tar does not have the /etc/systemd/system/local-fs.target.wants, even though it exists in sb2 target $VENDOR-$DEVICE.. | 19:56 |
| T42 | <elros34> what is in target is not realy relevant for mic. Have you checked for errors from mic? | 20:11 |
| T42 | LORD_ELSHAD was added by: LORD_ELSHAD | 20:49 |
| T42 | <b100dian> This is the output from the mic command https://pastebin.ubuntu.com/p/7TJcMKjqxG/ | 21:05 |
| T42 | <elros34> @b100dian you probably have partition mounted with nodev or nosuid which cause issue | 21:25 |
| T42 | iiiiiiilian was added by: iiiiiiilian | 22:59 |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!