| T42 | <smatkovi> is the content of this system_root.mount right https://paste.opensuse.org/pastes/23b4c73b5fdd | 14:15 |
|---|---|---|
| T42 | <smatkovi> i don't know how to mount it in / and not in /system, i've checked in my known places and it says / there and not /system | 14:51 |
| T42 | <smatkovi> can i somehow remount it in /, even if it is already mounted in /system? | 14:52 |
| T42 | <smatkovi> that is also strange: / # cd system/bin | 15:00 |
| T42 | <smatkovi> /bin/sh: cd: can't cd to system/bin: Too many levels of symbolic links | 15:00 |
| T42 | <smatkovi> maybe the system.mount is wrong? https://paste.opensuse.org/pastes/17bb5cfc3630 | 15:18 |
| T42 | <smatkovi> i could maybe package the system partition, but i don't know how | 16:08 |
| T42 | <elros34> I told you step by step what files to check, did you check all of them and last one in : droid-hal-device*rpm? | 16:11 |
| T42 | <smatkovi> I will say yes. But it wouldn't be the first time i thought i did something and then i didn't, so i will go through them again. So just to make sure i get everything, i should check the init.rc files for rootfs mounted in / instead of /system and then in the system_root.mount in the rpm if rootfs is mounted system_root? | 18:40 |
| T42 | <elros34> start with the latest: so system_root.mount in the rpm should NOT have this 'mount rootfs rootfs. | 18:43 |
| T42 | <elros34> If it has then go to init.rc files checking. | 18:43 |
| T42 | <smatkovi> now it's not even in the rpm | 19:01 |
| T42 | <smatkovi> maybe because i added %define makefstab_skip_entries / /system /vendor | 19:02 |
| T42 | <smatkovi> to droid-hal-device.inc? | 19:02 |
| T42 | <elros34> ... yeah | 19:02 |
| T42 | <elros34> btw you shouldn't edit droid-hal-device.inc | 19:02 |
| T42 | <smatkovi> that's what i understood from the hadk-hot | 19:03 |
| T42 | <elros34> you always add something to droid-hal-device.spec, *.inc file is in submodule | 19:04 |
| T42 | <smatkovi> here it says inc: - To prevent droid-hal creating incorrect systemd mount units add all dynamic partitions to makefstab_skip_entires similar to: https://github.com/mer-hybris/droid-hal-sony-lena/blob/master/rpm/droid-hal-common.inc#L20 | 19:06 |
| T42 | <smatkovi> maybe that should be updated then | 19:06 |
| T42 | <smatkovi> oh it says common | 19:07 |
| T42 | <elros34> it's no the best example but this is not droid-hal-device.inc | 19:07 |
| T42 | <smatkovi> as i didn't have common i used device | 19:07 |
| T42 | <smatkovi> so one could just create droid-hal-common.inc with that line? | 19:07 |
| T42 | <smatkovi> in that directory? | 19:07 |
| T42 | <elros34> you should not worry about it at all, remove what you added and move on with your simple issue to real porting | 19:08 |
| T42 | <smatkovi> where droid-hal-device.inc is | 19:08 |
| T42 | <smatkovi> ok thanks :) | 19:08 |
| T42 | <smatkovi> that's very interesting, the system.mount and system_root.mount files look completely different and more right now | 19:12 |
| T42 | <smatkovi> this is what the mount section for system_root.mount looks like: What=/dev/mmcblk0p25; Where=/system_root; Type=ext4; Options=ro,barrier=1 | 19:13 |
| T42 | <smatkovi> in droid-hal-device.rpm | 19:14 |
| T42 | <smatkovi> oh no What is wrong and should be /dev/block/bootdevice/by-name/system | 19:33 |
| T42 | <smatkovi> init.rc has still mount rootfs rootfs / remount bind ro nodev | 19:36 |
| T42 | <smatkovi> what should it have again? | 19:36 |
| T42 | <smatkovi> oh it should be outcommented | 19:37 |
| T42 | <smatkovi> i can't remember where i added that mmcblk0p25 | 19:40 |
| T42 | <smatkovi> omg i wrote it in two files in the tmp folder. maybe i should have deleted them anyway | 19:50 |
| T42 | <smatkovi> very strange it gets recreated and then states it gets this line from out/target/product/j5xnlte/system/vendor/etc/fstab.qcom but there clearly is something else | 19:53 |
| T42 | <elros34> system_root.mount finally looks correct | 19:57 |
| T42 | <smatkovi> i did make clean and have to wait an hour now for the kernel | 19:58 |
| T42 | <smatkovi> before it still had the error at system_root.mount but maybe a cleaner start will give better results tomorrow | 19:59 |
| T42 | <elros34> There is really no point to run make clean if youconly changed some files in droid repositories | 20:27 |
| T42 | <elros34> copy droid-hal-device*.rpm to device, install and test | 20:28 |
| T42 | <b100dian> piggz: abr: is https://forum.sailfishos.org/t/community-hardware-adaptations/14081 expected to have a line for each sfos release? Wouldn't a move instead of a copy suffice? We could specify initial available version as a separate column | 20:55 |
| T42 | <smatkovi> unfortunately i still get this in journalctl: Jun 23 03:00:00 localhost systemd[1]: system_root.mount: Mount process exited, c | 21:33 |
| T42 | <smatkovi> ode=exited status=32 | 21:33 |
| T42 | <smatkovi> maybe this is also relevant: Jun 23 03:00:00 localhost systemd[1]: Failed to mount Droid mount for /system_ro | 21:35 |
| T42 | <smatkovi> ot. | 21:35 |
| T42 | <smatkovi> al/17a855b54e4047b2b3c800939bd7fd04) is 1.5M, max 1.0M, 0B free. | 21:35 |
| T42 | <smatkovi> Jun 23 03:00:00 localhost mount[574]: mount: /system_root: mount point not mount | 21:35 |
| T42 | <smatkovi> ed or bad option. | 21:35 |
| mal | show the system_root.mount file | 21:43 |
| T42 | <smatkovi> https://paste.opensuse.org/pastes/27fc60875478 | 21:50 |
| T42 | <smatkovi> you're right, it looks wrong again | 21:50 |
| mal | yeah, that What= seems wrong | 21:51 |
| mal | how did you get that file? did you create it or was it generated automatically | 21:51 |
| T42 | <smatkovi> it was generated | 21:51 |
| T42 | <smatkovi> i mean the last one | 21:51 |
| T42 | <smatkovi> the one that looked more right, i guess that one i created | 21:52 |
| mal | so create a working one | 21:52 |
| T42 | <smatkovi> this is the line in fstab /dev/block/bootdevice/by-name/system / ext4 ro,barrier=1 | 21:52 |
| T42 | <smatkovi> what should be in the working one | 21:53 |
| T42 | <smatkovi> What=/dev/block/bootdevice/by-name/system ? | 21:53 |
| mal | the actual device node instead | 21:53 |
| T42 | <smatkovi> you mean /dev/mmcblk0p25? | 21:53 |
| mal | something like that, if that is what your device has for the system partition | 21:54 |
| T42 | <smatkovi> yes that is for the system partition | 21:54 |
| T42 | <smatkovi> is that the only thing i should change in the system_root.mount? | 21:55 |
| T42 | <smatkovi> it still can't mount correctly :/ | 21:57 |
| mal | hmm, that is wrong otherwise also | 21:58 |
| mal | the options part | 21:58 |
| mal | that part and also Type from some other service like vendor.mount | 21:58 |
| T42 | <smatkovi> Options=remount,bind,ro,nodev | 21:59 |
| T42 | <smatkovi> Type=rootfs | 21:59 |
| T42 | <smatkovi> i have no vendor.mount, i have efs.mount for example | 22:00 |
| T42 | <elros34> obviously if your droid-hal-device*rpm in local repo have correct mount unit and sfos on device not then either you are creating image using different rpms or installing wrong image | 22:01 |
| T42 | <smatkovi> should options be Options=ro,barrier=1 ? | 22:02 |
| T42 | <elros34> why do you ask, you have correct file in local repo, you even pasted it, same probably for other mount units | 22:03 |
| T42 | <elros34> zypper info droid-hal-<device> should tell you rpm version which you can compare to good one in local repo | 22:04 |
| T42 | <smatkovi> it wants to refresh repositories :/ | 22:07 |
| T42 | <elros34> so 'rpm -q <package name>' | 22:09 |
| T42 | <smatkovi> it really installs another one | 22:09 |
| T42 | <elros34> you should have list of packages in same directory as sfos image | 22:11 |
| T42 | <smatkovi> maybe it has to do with the fact that i extract and repack the tar.gz and hybris-boot.img | 22:14 |
| T42 | <smatkovi> maybe it is in update-binary? | 22:15 |
| T42 | <elros34> what? | 22:16 |
| T42 | <smatkovi> the list of packages | 22:16 |
| T42 | <smatkovi> but it's very strange | 22:16 |
| T42 | <elros34> no, list of packages used in image are in same directory as sfos image, you can't miss it | 22:17 |
| T42 | <smatkovi> in the .packages file the most recent correct package version is listed | 22:20 |
| T42 | <elros34> so obviously you recreated or installed wrong one | 22:21 |
| T42 | <smatkovi> i see now that the .tar.gz in the .zip was modified two days ago | 22:23 |
| T42 | <smatkovi> which should be in fact today | 22:23 |
| T42 | <smatkovi> i found the correct packaged tarball | 22:27 |
| T42 | <smatkovi> but it had sfe instead of sailfishos in its name | 22:27 |
| T42 | <smatkovi> and it doesn't install because of error 7 :/ | 22:33 |
| T42 | <smatkovi> i'll try it as .tar and not .tar.gz | 22:36 |
| T42 | <elros34> this is common error and very old issue still not fixed. There many ways to overcome it but nothing simple | 22:38 |
| T42 | <smatkovi> i messed with the kikstarter and probably something else, but i didn't remember where | 22:41 |
| T42 | <smatkovi> it worked when i extracted the .bz2 and just used the .tar, then i edited a file to get the .gz and now it creates a .gz and a .bz2 and doesn't work properly | 22:43 |
| mal | @elros34 I was thinking, would it be possible to include busybox in the install zip and use that to extract the rootfs? | 22:47 |
| T42 | <Mister_Magister> everything is possible mal :) | 22:47 |
| T42 | <Mister_Magister> especially that | 22:47 |
| T42 | <Mister_Magister> like adding busybox to hybris-boot target and then copying binaries/libs should be piece of cake | 22:48 |
| T42 | <elros34> mal: yes, I did that but why not using tar.gz which AFAIK works correctly | 22:48 |
| mal | we have the prebuilt busybox in external/ already | 22:48 |
| mal | @elros34 yeah, both are possible, just need to choose one | 22:49 |
| T42 | <Mister_Magister> switching to gz would be better imo | 22:49 |
| T42 | <Mister_Magister> no overcomplicating just to keep format | 22:49 |
| mal | is there any significant difference in size? | 22:49 |
| T42 | <Mister_Magister> I don't think anyone worries about size honestly | 22:50 |
| T42 | <Mister_Magister> otherwise it would be lz4 | 22:50 |
| T42 | <smatkovi> of course it was the build_packages.sh | 22:50 |
| T42 | <smatkovi> it's about 10-30Mb difference in size | 22:51 |
| T42 | <smatkovi> so neglectible i would say | 22:51 |
| mal | yeah, that is quite small difference | 22:53 |
| T42 | <smatkovi> maybe 40MB so less than 10% of the whole size of the .zip file | 22:55 |
| T42 | <smatkovi> now i have error 6 :/ | 22:59 |
| T42 | <smatkovi> can i somehow only reset the building of sailfish os packages? | 23:05 |
| mal | what do you mean? | 23:17 |
| T42 | <smatkovi> when i build sailfish os the .zip file ends up with a .tar.gz and a tar.bz2 (which is the correct one) but when i bunzip2 that and also change the updater-unpack.sh and remove the -z flag from the tar command and change the mentioned filenames accordingl also in updater-script i still get an error and two days ago this was working | 23:20 |
| mal | if you remove some package from droid-local-repo and rebuild it then it should at least be latest | 23:22 |
| T42 | <smatkovi> yes i have that, but i want to be able to install it again | 23:23 |
| mal | so why doesn't the new image contain the new changes? | 23:24 |
| mal | or are you missing some change? | 23:24 |
| T42 | <smatkovi> now it works :) | 23:26 |
| T42 | <smatkovi> i just cloned dhd again | 23:26 |
| T42 | <smatkovi> i think now i got rid of the mount error, but still it doesn't boot into gui. this is the output of journalctl -b —no-pager: https://paste.opensuse.org/pastes/20dd74ee01d2 | 23:40 |
| T42 | <smatkovi> and this of dmesg after boot: https://paste.opensuse.org/pastes/41453b990898 | 23:41 |
| Mister_Magister | lmao the video problem in sake, that is "Dropping frame due to QoS", is funny because it's interleaved underflow and overflow | 23:43 |
| Mister_Magister | so it seems like frame is not in time but then it arrives and there are two frames, but then next frame is also late | 23:43 |
| Mister_Magister | so it just keeps dropping them | 23:44 |
| Mister_Magister | funnily what i don't think cause issue is audioless video stream, like camera viewfinder or microtube for that matter, files cause issue but streams don't | 23:44 |
| Mister_Magister | maybe perhaps due to buffer? | 23:45 |
| Mister_Magister | like its getting couple qos messages but only at the beginning | 23:45 |
| Mister_Magister | for local files you wouldn't have large buffer | 23:45 |
| T42 | <smatkovi> interesting | 23:46 |
| Mister_Magister | ikr | 23:46 |
| Mister_Magister | it doesn't entirely explain anything but its clues | 23:47 |
| Mister_Magister | decoder frame list getting long: 52 frames,possible internal leaking? | 23:47 |
| Mister_Magister | and then when it fills up, it crashes the playback entirely? | 23:47 |
| Mister_Magister | neither gst-droid nor droidmedia has message "decoder frame list getting long" wtf | 23:51 |
| Mister_Magister | oh its in gstreamer | 23:52 |
| T42 | <elros34> @smatkovi don't tell me you still didn't apply all hybris patches? | 23:55 |
| Mister_Magister | so jolla bothered to add av1 to droidmedia but not to gstdroid lol | 23:56 |
| T42 | <smatkovi> maybe the patches get overwritten? | 23:56 |
| T42 | <smatkovi> oh yes the make clean maybe | 23:56 |
| T42 | <elros34> no make clean doesn't overwrite anything | 23:57 |
| T42 | <elros34> I am just asking because logs looks suspicious | 23:57 |
| T42 | <smatkovi> fatal: previous rebase directory .git/rebase-apply still exists but mbox given. | 23:58 |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!