| deathmist1 | mal: https://github.com/mer-hybris/qt5-qpa-hwcomposer-plugin/pull/95 :) | 01:05 |
|---|---|---|
| T42 | <edp_17> @elros34 : Good morning. I've checked what you recommended. Have found this error: "droid-hal-init.service: Main process exited, code=exited, status=6/NOTCONFIGURED" Tried to find related communication in porters log and seen Vlad has had this issue but have not found the solution. | 06:03 |
| T42 | <edp_17> Some logs: | 06:03 |
| T42 | <edp_17> dmesg: https://paste.ubuntu.com/p/wbXSvhpm27/ | 06:03 |
| T42 | <edp_17> persistent journal: https://paste.ubuntu.com/p/Jxpv3XSm9B/ | 06:03 |
| T42 | <edp_17> findmnt: https://paste.ubuntu.com/p/36CG7TZVnk/ | 06:03 |
| T42 | <edp_17> strace /sbin/droid-hal-init: https://paste.ubuntu.com/p/Xnz7fg6vQt/ | 06:03 |
| T42 | <edp_17> df -h: https://paste.ubuntu.com/p/gXRZg8Gvf7/ | 06:03 |
| T42 | <edp_17> /usr/lib/systemd/system/system.mount: https://paste.ubuntu.com/p/jZmTnX4RC5/ | 06:03 |
| T42 | <edp_17> /usr/lib/systemd/system/system_root.mount: https://paste.ubuntu.com/p/cPScFgwns8/ | 06:03 |
| T42 | <edp_17> content of /system: https://paste.ubuntu.com/p/Jb2FYbMVJ2/ | 06:03 |
| T42 | <edp_17> mounts: https://paste.ubuntu.com/p/YPfQvMSRFK/ | 06:03 |
| T42 | <edp_17> content of /usr/libexec/droid-hybris/system/etc/init/ , /dev/mdss_rotator , /proc/cmdline and /etc/selinux/minimum/contexts/dbus_contexts: https://paste.ubuntu.com/p/c5scH5DzGg/ | 06:03 |
| T42 | <edp_17> It looks like /system now mounted correctly but droid-hal-init is crashing. | 06:04 |
| T42 | <edp_17> I don't have androidboot.selinux=permissive in kernel commandline (see /proc/cmdline in last url) | 06:08 |
| T42 | <edp_17> However, dmesg says: "SELinux: Starting in permissive mode" | 06:09 |
| T42 | <edp_17> and also says: "droid-hal-init: Could not read vendor SELinux version" | 06:09 |
| T42 | <edp_17> I don't have /vendor is it a problem? | 06:13 |
| T42 | <my90ds> Good day! Please tell me where you can see if there is a Sailfish OS port for a Xiaomi redmi note 5 smartphone and what state it is in now (what works and what does not work) I saw a port for redmi 5 plus on 4pda, there is about some redmi note 5 India is written, but there is nothing about how to install a sailfish OS on this smartphone, or is there the same firmware installation as on Certus, where are the firmware universal on redm | 06:16 |
| T42 | <elros34> @edp_17 hi, do you have /system/vendor? Maybe it's enough to create /vendor symlink | 06:20 |
| T42 | <edp_17> @elros34 : Yes, there is /system/vendor | 06:22 |
| T42 | <edp_17> https://paste.ubuntu.com/p/Jb2FYbMVJ2/ | 06:22 |
| T42 | <edp_17> Now I have /vendor but didn't change too much. | 06:33 |
| T42 | <edp_17> new dmesg: https://paste.ubuntu.com/p/H355S6fFzj/ | 06:33 |
| T42 | <edp_17> new journal: https://paste.ubuntu.com/p/qymysRGT9R/ | 06:33 |
| T42 | <elros34> looks better, now even apexd starts and some files form /vendor are parsed | 06:43 |
| T42 | <edp_17> Yeah and dmesg doesn't complain about not finding /vendor/etc/selinux/vendor_property_contexts. | 06:46 |
| T42 | <edp_17> However, droid-hal-init still crashing :( | 06:46 |
| T42 | <edp_17> Wait! Does this in journal "Started droid-hal-init" mean that droid-hal-init successfully completed/started? | 06:53 |
| T42 | <elros34> yeas but with flood of logs it's easy to miss it. You should use journalctl -b0 if you have persistent logging enabled | 06:54 |
| T42 | <elros34> @my90ds no idea if there is a port but you can search in this channel logs for device codename you are interested. Another way: search for droid-config-<device codename> in github/lab | 07:05 |
| T42 | <edp_17> @elros34 : Thanks. I'll use journalctl -b0. Do you want me to upload a new journal? | 07:13 |
| T42 | <edp_17> Oh, and how can I add symlinking /system/vendor to /vendor into the build? | 07:16 |
| T42 | <edp_17> Just create a symlink in sparse? Will it work? | 07:17 |
| T42 | <elros34> it will. now check logcat | 07:22 |
| T42 | <edp_17> Thanks. Logcat: https://paste.ubuntu.com/p/HCm6xR5jKK/ | 07:34 |
| T42 | <edp_17> Is this the issue "SELinux : Unknown class hwservice_manager"? | 07:38 |
| T42 | <edp_17> Another logcat (I let it run until the device rebooted): https://paste.ubuntu.com/p/z4hj7HfB46/ | 07:39 |
| T42 | <elros34> no idea about that selinux msg, have you tried getenforce command to confirm selinux is in permissive mode? For sure this one is relevant and common "libandroidicu.so" not found | 07:42 |
| T42 | <edp_17> /system/bin/getenforce : Permissive | 07:43 |
| Thaodan | The unknown class issue is irrelevant its a common message that there are unknown classes but its not doing anything. | 07:44 |
| T42 | <edp_17> That missing "libandroidicu.so" rings a bell. Let me check my notes. | 07:44 |
| Thaodan | Please use the channel grep about the icu.so missing there where solutions for that. | 07:44 |
| Thaodan | this is what we do for Xperia10II/Xperia10III and I do on Xperia 1/1II | 07:46 |
| Thaodan | https://github.com/mer-hybris/droid-config-sony-lena/blob/master/sparse/system/lib64/libandroidicu.so | 07:46 |
| T42 | <edp_17> Thanks Thaodan. | 07:47 |
| T42 | <edp_17> For another device I had to create a /odm/lib folder and add a symlink to /apex/com.android.runtime/lib/libandroidicu.so | 07:49 |
| T42 | <edp_17> Is this not proper solution? | 07:49 |
| T42 | <elros34> it is, you can't use /system because you didn't package it | 07:50 |
| T42 | <edp_17> Oh, I see. Thanks. I've done it on device and will replicate in sparse too. | 07:50 |
| T42 | <edp_17> New logs: logcat: https://paste.ubuntu.com/p/C8Rw4kY8tC/ | 07:54 |
| T42 | <edp_17> journal: https://paste.ubuntu.com/p/YWrKgzKbGn/ | 07:54 |
| T42 | <edp_17> Sorry, this is the new logcat (the other one has the previous logs too where libandroidicu.so wasn't found): https://paste.ubuntu.com/p/bjnRb5GDhf/ | 07:57 |
| Thaodan | there's still an issue with apex that is marked as updateable I think, you could ask mal | 07:58 |
| T42 | <edp_17> Now there is another one not found "CANNOT LINK EXECUTABLE "/apex/com.android.media.swcodec/bin/mediaswcodec": library "libcutils.so"" | 08:00 |
| T42 | <edp_17> I'll go thorugh these until I symlined all of them. Will be back after. | 08:00 |
| T42 | <elros34> libcutil.so seems to be related to namespace so adding symlik will probably not help | 08:09 |
| T42 | <edp_17> Yep, it didn't. :) | 08:11 |
| T42 | <edp_17> There is one thing I noticed. Since I added those symlinks into /odm/lib, the device consloe became very sluggish. | 08:12 |
| T42 | <edp_17> So, what should I do now? Ask mal as Thaodan suggested? | 08:12 |
| T42 | <elros34> didn't you fight already with ldconfig.*txt with similar namespace issue? | 08:16 |
| T42 | <edp_17> No, I haven't. I've checked my notes and there is nothing about ldconfig.*txt in there. | 08:25 |
| Thaodan | Is it A11? There's no ld.config.txt there anymore | 08:26 |
| T42 | <edp_17> Thaodan : No, it is Android10 (Los17.1) | 08:50 |
| Thaodan | then it its /systeme/etc | 08:50 |
| Thaodan | *then it its /system/etc | 08:50 |
| T42 | <edp_17> Thanks Thaodan! | 08:58 |
| T42 | <edp_17> @elros34 : I've found some stuff in my notes related to ld.config* (Previously I was looking for ldconfig*) | 08:59 |
| T42 | <edp_17> Well, it seems on the other device I couldn't manage to solve the libcutils.so related issue with ld.config.txt. :( I'll try on this one. | 10:02 |
| Thaodan | Maybe try the same solution as done for icu. | 10:07 |
| T42 | <edp_17> Thaodan : I've tried (created a symlink in /odm/lib) but didn't work. Or you were referring to something else? | 10:10 |
| T42 | <edp_17> This is the content of /system/etc/ld.config.txt : https://paste.ubuntu.com/p/DvtK7Dn2gC/ | 10:11 |
| T42 | <edp_17> There is no /usr/libexec/droid-hybris/system/${LIB} in namespace.media.permitted.paths | 10:11 |
| Thaodan | Yes that's what I meant. | 10:11 |
| T42 | <edp_17> and libcutils.so is not listed in namespace.media.link.default.shared_libs | 10:11 |
| Thaodan | edp_17: try this patch https://github.com/mer-hybris/droid-system-sony-template/blob/master/patches-system/ld.config.29.patch | 10:12 |
| Thaodan | Packaging droid-system seems to be so much easier | 10:12 |
| T42 | <edp_17> I don't have /system/etc/ld.config.29.txt | 10:15 |
| Thaodan | Ah sorry ld.config.28.txt then | 10:16 |
| Thaodan | try the other patch in the very same repo. | 10:16 |
| T42 | <edp_17> What about if I add namespace.media.permitted.paths += /usr/libexec/droid-hybris/system/${LIB} and namespace.media.link.default.shared_libs += libcutils.so into ld.config.txt? (on device) | 10:16 |
| Thaodan | https://github.com/mer-hybris/droid-system-sony-template/blob/master/patches-system/ld.config.28.patch | 10:17 |
| Thaodan | use this first | 10:17 |
| T42 | <edp_17> No, this didn't help. Logcat still complains about libcutils.so : https://paste.ubuntu.com/p/NH5XSxW2Cq/ (re @edp_17: What about if I add ...) | 10:21 |
| T42 | <edp_17> Thaodan : How I supposed to know whether I need ld.config.28 or 29? (On device I only have ld.config.txt but there is no either *.28.txt nor *.29.txt) | 10:22 |
| T42 | <edp_17> I do have /usr/libexec/droid-hybris/system/etc/ld.config.29.txt but don't have /system/etc/ld.config.29.txt | 10:31 |
| T42 | <edp_17> Thanodan : Okay, this is what I've done. I created an empty /system/etc/ld.config.29.txt file. After reboot the droid-hal-early-init.sh bid mounted the ld.config.29.txt from /usr/libexec/droid-hybris/system/etc/ to /system/etc/ | 10:41 |
| T42 | <edp_17> However, the same error about libcutils.so is still in logcat. :( | 10:42 |
| T42 | <edp_17> Thaodan ^^ (Sorry, earlier mistyped your name) | 10:43 |
| T42 | <edp_17> Now, this is the content of /system/etc/ld.config.29.txt : https://paste.ubuntu.com/p/QYMqXywtC4/ | 10:45 |
| T42 | <edp_17> mal : Do you have idea how to fix this libcutils.so issue? | 10:46 |
| mal | @edp_17 if you don't have 28 or 29 then do similar for ld.config.txt | 12:18 |
| T42 | <edp_17> mal : Hi. Now I have ld.config.29.txt in /system/etc/ because I've created a blank file which them "overwritten" by bind mount. Or how do you meant "do similar for ld.config.txt"? | 12:50 |
| T42 | <edp_17> mal : Did you mean adding these into ld.config.so? : | 12:50 |
| T42 | <edp_17> namespace.media.permitted.paths += /usr/libexec/droid-hybris/system/${LIB} | 12:50 |
| T42 | <edp_17> and | 12:50 |
| T42 | <edp_17> namespace.media.link.default.shared_libs += libcutils.so | 12:50 |
| mal | @edp_17 do not create random files like that, if you only have ld.config.txt originally there then you override that | 12:55 |
| T42 | <edp_17> mal : Oh, so override with what? With /usr/libexec/droid-hybris/system/etc/ld.config.29.txt ? | 13:01 |
| T42 | <edp_17> (Of course keeping its original ld.config.txt name.) | 13:02 |
| mal | you copy the ld.config.txt from system and then edit it with the needed paths | 13:09 |
| T42 | <edp_17> mal : I've mounted /system for rw and added the paths into it but didn't work. | 13:10 |
| mal | are you sure you edited it correctly? | 13:10 |
| T42 | <edp_17> Give me a minute, I'll get the original and the modified one. | 13:11 |
| T42 | <edp_17> Original: https://paste.ubuntu.com/p/K2GY76VChF/ | 13:12 |
| T42 | <edp_17> Modified: https://paste.ubuntu.com/p/QNN4Vk7N4k/ | 13:13 |
| T42 | <edp_17> The difference is these two lines: | 13:14 |
| T42 | <edp_17> namespace.media.permitted.paths += /usr/libexec/droid-hybris/system/${LIB} | 13:14 |
| T42 | <edp_17> namespace.media.link.default.shared_libs += libcutils.so | 13:14 |
| mal | you added the path to wrong namespace, it should be in default namespace not media | 13:14 |
| mal | also quite likely the shared_libs change is not needed if the paths are correct | 13:15 |
| T42 | <edp_17> mal : Okay, I'll try again. Thanks. | 13:18 |
| T42 | <edp_17> mal : There is no [default] namespace in there. Should I add it? | 13:25 |
| T42 | <edp_17> There is [legacy] and [postinstall] nothing else. | 13:25 |
| T42 | <edp_17> Oh, I see now. So not "namespace.media.permitted.paths += *" but "namespace.default.permitted.paths += *" | 13:26 |
| T42 | <edp_17> Got it. | 13:26 |
| T42 | <edp_17> Is what I wrote correct? (Currently there is no "namespace.default.permitted.paths" in the ld.config.txt) | 13:28 |
| mal | so you add it | 13:30 |
| mal | just without the + | 13:30 |
| T42 | <edp_17> mal : This is how it looks like now : https://paste.ubuntu.com/p/GGwJ2nYzdK/ | 13:31 |
| mal | seems reasonable | 13:32 |
| T42 | <edp_17> mal : and the error is still there: https://paste.ubuntu.com/p/HhbGXXzTpT/ | 13:34 |
| T42 | <edp_17> :( | 13:34 |
| mal | hmm, it says permitted_paths="" | 13:44 |
| T42 | <edp_17> Should I add "namespace.default.permitted_paths = /usr/libexec/droid-hybris/system/${LIB}" ? | 14:02 |
| T42 | <edp_17> This is the second device I am getting exactly the same error that results in reboot after a few minutes. Both are on Andriod10 base (Los17.1) | 14:03 |
| mal | maybe you could try to start the failing process manually with strace to see what it loads | 14:04 |
| T42 | <edp_17> What's the failing process? Here is an older logcat: https://paste.ubuntu.com/p/PYJyNPWnyY/ | 14:06 |
| T42 | <edp_17> Or is that shown in journal? | 14:06 |
| T42 | <edp_17> here is the journal relevant for that logcat: https://paste.ubuntu.com/p/WvwbVQbTCN/ | 14:07 |
| deathmist1 | mal: would you mind testing https://github.com/mer-hybris/qt5-qpa-hwcomposer-plugin/pull/95? | 14:27 |
| deathmist1 | I'd do that myself but I'm not in the mood to setup the env manually right now and am still salty about sfbootstrap breaking quite a bit the next SFOS release right after I made it public while nothing about the build process changed for years previously 🥲 | 14:36 |
| deathmist1 | oh also: anything new about OBS still? is it staying up for porters to use or has no decision been made yet? | 14:39 |
| deathmist1 | also I'm not sure if I've missed something or is the HADK PDF .rst source not uploaded anywhere (as a git repo)? I know previously in this channel a diff was uploaded between 4.3.0.12 and 4.3.0.15, but I'd like to see those diffs properly and not have them e.g. on a pastebin that disappears after a year | 14:43 |
| T42 | <edp_17> mal : Tried to understand how this linker-namespace works but I am not sure. (Found this site https://source.android.com/devices/architecture/vndk/linker-namespace) | 15:51 |
| T42 | <edp_17> On the other device I got the same issue, at that time tried to generate the corresponding configuration with PRODUCT_TREBLE_LINKER_NAMESPACES and BOARD_VNDK_VERSION in BoardConfig.mk, but building hybris-hal failed so I gave up. | 15:52 |
| T42 | <edp_17> I've also tried to add "namespace.default.permitted.paths += /usr/libexec/droid-hybris/system/${LIB}" into "/system/apex/com.android.runtime.release/etc/ld.config.txt" but didn't help. The same error in logcat. | 15:53 |
| T42 | <edp_17> If you are interested content of "/system/apex/com.android.runtime.release/etc/ld.config.txt" : https://paste.ubuntu.com/p/HPQxC8tyD4/ | 15:54 |
| mal | the problematic process is /apex/com.android.media.swcodec/bin/mediaswcodec | 16:09 |
| T42 | <edp_17> mal : thanks. So just simply strace it or should I use some switches? | 16:15 |
| Thaodan | -ff helps in case the process creates children | 16:21 |
| T42 | <edp_17> Thanks Thaodan. | 16:31 |
| T42 | <edp_17> How can I restart persistent journal? (If it's getting too big/long and would like to do a fresh start.) | 16:33 |
| mal | how do you check it? "journalctl -b --no-pager" only shows parts from current boot | 16:33 |
| T42 | <nephros> `journalctl --vacuum-size=10M ` should truncate the files themselves. | 16:55 |
| T42 | <nephros> You can also `journalctl --vacuum-time=1day` or so. | 16:55 |
| mal | deathmist1: I need to check that PR | 17:19 |
| T42 | <elros34> deathmist1 never see hadk pdf source, only diff for some releases | 18:28 |
| T42 | <elros34> @edp_17 maybe you could remove "/usr/libexec/droid-hybris/system/lib/libcutils.so" to see whether mediaswcodec will start and anything else change | 18:50 |
| T42 | <elros34> no idea if that so contains any hybris changes. If not then mediaswcodec should load it from /system or something | 18:52 |
| T42 | <edp_17> mal , @nephros , @elros34 : Thanks. :) | 19:04 |
| deathmist1 | @elros34 I'd like for that to change :/ what's the point of having it not be accessible to the public? | 19:08 |
| deathmist1 | you can already read it, just makes checking for necessary changes to build steps tedious and time consuming | 19:09 |
| Thaodan | Did someone try to sign hal packages and preinstall the pgp used for it into a port? | 20:33 |
| T42 | <edp_17> @elros34 : I've renamed that libcutils.so and the sympton has remained. However, the error disappeared from logcat. | 20:56 |
| T42 | <edp_17> logcat: https://paste.ubuntu.com/p/gn7zd23Rrw/ | 20:56 |
| T42 | <edp_17> journal -b0: https://paste.ubuntu.com/p/ppBkDwxxqy/ | 20:57 |
| T42 | <edp_17> In "/system/apex/com.android.runtime.release/etc/ld.config.txt" and "/system/etc/ld.config.txt" I have left the added "namespace.default.permitted.paths += /usr/libexec/droid-hybris/system/${LIB}" entry. | 20:58 |
| T42 | <edp_17> mal : strace mediaswcodec : https://paste.ubuntu.com/p/JCGqdy7wxQ/ | 21:01 |
| T42 | <edp_17> Now, it is not crashing because I renamed "/usr/libexec/droid-hybris/system/lib/libcutils.so" | 21:01 |
| T42 | <elros34> so could be irrelevant for now. I wonder about: "ServiceManager: add_service('gpu',1) uid=1072 - PERMISSION DENIED" and other similar. | 22:01 |
| T42 | <Helena> 𝐈 𝐟𝐢𝐧𝐚𝐥𝐥𝐲 𝐟𝐨𝐮𝐧𝐝 𝐚 𝐫𝐞𝐚𝐥 𝐟𝐨𝐫𝐞𝐱 𝐭𝐫𝐚𝐝𝐢𝐧𝐠 𝐡𝐞𝐫𝐞 𝐨𝐧 𝐭𝐞𝐥𝐞𝐠𝐫𝐚𝐦 𝐚𝐟𝐭𝐞𝐫 𝐛𝐞𝐢𝐧𝐠 𝐬𝐜𝐚𝐦𝐦𝐞𝐝 𝐚 𝐥𝐨𝐭 𝐨𝐟 𝐭𝐢𝐦𝐞𝐬. 𝐈’𝐯𝐞 𝐟𝐢𝐧𝐚𝐥𝐥𝐲 𝐟𝐨𝐮𝐧𝐝 𝐬𝐨𝐦𝐞𝐨𝐧𝐞 𝐰𝐡𝐨 𝐢𝐬 𝐥𝐞𝐠� | 23:27 |
| T42 | <Helena> 👇👇👇👇👇👇👇👇 | 23:28 |
| T42 | <Helena> @F3v_HSlGYiI0MGFk | 23:28 |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!