| genericv3 | today i messed with it on my own. Everything looks okay except display.qservice. it was working well on our previous attempts. It's probably caused because of vndservicemanager, which depends on /dev/vndbinder. But /dev/vndbinder is empty so it's waiting for it forever. There is a section about this on hadk-faq, that was worked in previous package but not helping in this one. i'll check it tomorrow | 01:03 |
|---|---|---|
| genericv2 | mal: about /dev/vndbinder issue: https://paste.opensuse.org/pastes/492cfbc182bd https://paste.opensuse.org/pastes/a29fe48733ad | 18:54 |
| mal | hmm, ndksupport: Could not load /vendor/lib64/hw/hwcomposer.msm8998.so from default namespace: dlopen failed: cannot locate symbol "__get_tls_hooks" referenced by "/odm/lib64/libEGL.so".... | 18:58 |
| mal | that is bad | 18:58 |
| mal | @elros34 ^ | 18:58 |
| genericv2 | Now i copied those /odm files from /system instead of /usr/libexec, they seem okay now. | 19:04 |
| mal | copied? what? | 19:05 |
| mal | what have you done to odm? | 19:06 |
| genericv2 | it don't exist in my phone normally | 19:06 |
| genericv2 | i created it | 19:06 |
| mal | what have you added there? | 19:06 |
| genericv2 | and copied some files while trying by myself. | 19:06 |
| genericv2 | https://paste.opensuse.org/pastes/37dc672053d5 these, while trying to make hwcomposer work (and it worked in the previous package) | 19:09 |
| genericv2 | if i did something wrong i can remove them | 19:09 |
| mal | that is not how you should do it | 19:11 |
| genericv2 | how i must do it | 19:12 |
| mal | usually only the libandroidicu.so symlink is needed there, and in some cases couple more symlinks | 19:13 |
| genericv2 | when i don't do that, test_hwcomposer for an example throws a lot of cannot find library errors. And i could make it work like this sometime. | 19:14 |
| genericv2 | But i'll try anyways | 19:14 |
| genericv2 | With only libandroidicu.so symlink on /odm: https://paste.opensuse.org/pastes/76a9adf89c40 https://paste.opensuse.org/pastes/80ccf159ffea | 19:25 |
| genericv2 | oops | 19:25 |
| genericv2 | https://paste.opensuse.org/pastes/80ccf159ffea https://paste.opensuse.org/pastes/76a9adf89c40 | 19:25 |
| genericv2 | ah it was okay actually, i saw like there is no space for a moment | 19:25 |
| mal | are those new ones really correct? "cannot locate symbol "__get_tls_hooks" referenced by "/odm/lib64/libEGL.so"...." | 19:28 |
| genericv2 | i really forget to pull new files. pff | 19:28 |
| mal | ? | 19:28 |
| genericv2 | logcat: https://paste.opensuse.org/pastes/5d0e478680ed | 19:30 |
| genericv2 | journal: https://paste.opensuse.org/pastes/624eaa9d918d | 19:30 |
| genericv2 | okay these are correct | 19:31 |
| mal | so hwcomposer is up now based on logcat, but lipstick is not starting, nothing about that in log | 19:31 |
| mal | does test_hwcomposer work? | 19:32 |
| mal | EGL_PLATFORM=hwcomposer test_hwcomposer | 19:32 |
| genericv2 | nope, it does not work. (libbinder.so not found) | 19:33 |
| genericv2 | i copied these files to /odm for this yesterday. | 19:34 |
| mal | so does binder-list show anything for the vndbinder? | 19:36 |
| genericv2 | / # binder-list | grep vnd | 19:36 |
| genericv2 | / # binder-list -d /dev/vndbinder | 19:36 |
| genericv2 | nothing related to vnd | 19:36 |
| genericv2 | hwbinder and binder are okay | 19:37 |
| mal | has there ever been anything in vdnbinder? | 19:37 |
| mal | *vndbinder | 19:38 |
| genericv2 | no in the newest package, i don't remember previous ones | 19:38 |
| mal | so does test_hwcomposer crash or what? does logcat show anything when you run that | 19:39 |
| genericv2 | library "libbinder.so" not found | 19:41 |
| genericv2 | test_hwcomposer: test_common.cpp:370: HWComposer* create_hwcomposer_window(): Assertion `err == 0' failed. | 19:41 |
| genericv2 | Aborted (core dumped) | 19:41 |
| genericv2 | and logcat gets some new output, i'll try to catch thrm | 19:42 |
| genericv2 | also, right now i can make test_hwcomposer work again with my previous /odm | 19:43 |
| mal | so from where do you copy the files? | 19:45 |
| mal | strace output would be nice from non-working attempt | 19:46 |
| genericv2 | Most of them from /apex, and some of them from /usr/libexec/droid-hybris. And these libEGL & LibGLES* dudes from /system. | 19:52 |
| genericv2 | https://paste.opensuse.org/pastes/fa29f4aba060 | 19:52 |
| mal | hmm, odd that it doesn't attempt /system at all | 19:53 |
| mal | are you really sure you need all of those files? | 19:55 |
| mal | and do you need to copy the files or does symlinking work? | 19:56 |
| genericv2 | test_hwcomposer doesn't work if they don't exist. Didin't tried symlinking but i can try if you want. | 19:56 |
| genericv2 | i saw some selinux denials and other selinux related errors. And that "not accessiable by namespace default" etc. Can these related to these errors? | 20:06 |
| genericv2 | selinux config and policy are okay right now, checked it twice | 20:06 |
| mal | linker config is probably related to this, just not sure why it's causing issues on your device | 20:07 |
| genericv3 | yeah, it's strange. /linkerconfig/ld.config thingy exists. | 20:09 |
| T42_ | <b100dian> hmm test_common line 370, I remember that I commented out parts of that because it would not get to read the version so that it would use hwcomposer2. However, in your case it used to work so that may not be required 🤔 | 20:27 |
| genericv3 | quiting now, if anyone gets an idea please write, i'll check irclogs tomorrow. | 20:32 |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!