| T42 | <parmjotsinghrobot> Rip their 01.org blogs (re @SailfishFreenodeIRCBridgeBot: <piggz_>why the fsck...) | 00:20 |
|---|---|---|
| T42 | <parmjotsinghrobot> I remember the day it happened | 00:20 |
| T42 | <parmjotsinghrobot> I was reading some stuff on s2idle and soix when it happened | 00:20 |
| T42 | <parmjotsinghrobot> I reloaded and I got redirected to that crazy site | 00:21 |
| T42 | <parmjotsinghrobot> Never could find that post again (apart from wayback machine) | 00:21 |
| *** Danct12 is now known as Guest9335 | 03:11 | |
| piggz_ | mal: seen any audio routing headphone errors like this, happens when plugging in headphones. audio stop on device, but no audio in headphones https://paste.mozilla.org/r5CSjLKp | 10:17 |
| T42 | <b100dian> mal: I've installed the droidmedia build on the lineage 17.1 port with ACodec (instead of CCodec2). How do you test? Gallery or browser? If the latter, GECKO_CAMERA_DROID_NO_MEDIA_BUFFER=0 or GECKO_CAMERA_DROID_FORCE_MEDIA_BUFFER=1 ? | 19:45 |
| T42 | <b100dian> alright, reproduces with gallery. the one that doesn't work on my port. Seek or end of video. Cool :) | 20:01 |
| T42 | <b100dian> mal: fixed I guess | 21:32 |
| T42 | <b100dian> moved one clear that was only needed when not recycling buffers to only happen on the new codepath, not on releaseBufferResources | 21:33 |
| T42 | <b100dian> Unfortunately gst-droid still doesnt work on my 18.1 port :) I kinda hoped that would fix that too. Oh well | 21:34 |
| T42 | <Walid> for latest sailfish os, for device bullhead nexus 5x, i have problem with CONFIG_PID_NS, when disable it system start quickly but no gui, when i enabled it , got take long time to mount /system and vendor (1 h), and nothing work | 21:36 |
| mal | @b100dian I think you misunderstood what I said before, that still has the same bug based on the code, I didn't test it because it's obvious | 21:38 |
| T42 | <b100dian> mal: ok, about calling decStrong from inside buffer, I think it doesn't matter, I only use the buffer to get the queue and the queue calls decStrong on the buffer just as before | 21:39 |
| T42 | <b100dian> I agree about not understanding, but I did reproduce it and I think that .clear that I moved was also decStrong'ing the inner m_buffer, or that's my hypothesis | 21:40 |
| mal | I got segfault from decStrong when it was done like that | 21:40 |
| mal | hmm, ok | 21:40 |
| mal | maybe I need to try it now | 21:41 |
| mal | but backtrace of the crash before showed it happening when decStrong was called | 21:41 |
| T42 | <b100dian> Hmm I didnt gdb jolla-gallery indeed | 21:42 |
| T42 | <b100dian> Was a bit surprised that this was the repro | 21:42 |
| T42 | <b100dian> Please check and maybe share the android base where it happens, maybe I can repro on another device.. I only got a Xperia 10 where I didnt try to build .. | 21:43 |
| mal | I was using android 11 base, sony xperia 10 III | 21:44 |
| mal | I have also almost every other base available for testing (excluding android 10 base because the device has broken battery) | 21:46 |
| T42 | <b100dian> Xperia 10 iii has ACodec or CCodec2 ..? | 21:47 |
| T42 | <b100dian> Should be visible in logcat somehow | 21:47 |
| mal | it didn't show your logging output in logcat so it should be using the normal way | 21:48 |
| T42 | <b100dian> Ok | 21:48 |
| T42 | <elros34> Walid so nothing works with or without pid_ns? Are you sure this is the main problem? | 21:49 |
| mal | does your device in some cases end up in the else-branch of releaseBufferResources(? | 21:49 |
| T42 | <elros34> Walid what kernel version do you use? | 21:50 |
| mal | also which android base | 21:50 |
| mal | that last one was for Walid | 21:50 |
| T42 | <b100dian> Mal: no, I dont think so. I included it for.. backward compatibility? I recently excluded it with ifdef android 5 | 21:51 |
| mal | @b100dian I'm building that new version now, if I still get the crash I can tell exactly what you should do to the code | 21:52 |
| T42 | <Walid> I have used pie kernel its work with Ubuntu touch xenial . I used hybrid -16.9 | 21:53 |
| T42 | <Walid> I have used pie kernel its work with Ubuntu touch xenial . I used hybrid -16.0 (edited) | 21:53 |
| T42 | <b100dian> mal: thank you | 21:53 |
| mal | @b100dian so in frameAvailable you always have unslotted on your device | 21:53 |
| mal | @b100dian can you add debug print next to the decStrong to be sure | 21:53 |
| T42 | <b100dian> Ah, I responded to the wrong else | 21:54 |
| T42 | <b100dian> I thought of if m_queue | 21:54 |
| T42 | <b100dian> I understand your question now. | 21:54 |
| T42 | <b100dian> Yes, initially buffers are not found in unslotted | 21:54 |
| T42 | <Walid> https://github.com/bullhead-ubuntu-touch | 21:55 |
| T42 | <elros34> Walid what is version of this 'pie kernel'? | 21:55 |
| T42 | <Walid> Her IS my work | 21:55 |
| T42 | <Walid> 3.10 | 21:55 |
| mal | @b100dian ok, that makes my idea a bit more problematic then, need to think | 21:56 |
| T42 | <Walid> https://github.com/bullhead-ubuntu-touch/kernel_lge_bullhead/tree/hybris-16.0 | 21:56 |
| mal | @b100dian hmm, it doesn't crash anymore, still odd why the backtrace suggested otherwise, maybe it somehow hid the fact that the crash was from that clear | 22:00 |
| T42 | <b100dian> Maybe clear makes another decStrong.. | 22:01 |
| mal | @b100dian I still suggest you reorder the _DroidMediaBufferQueue::buffersReleased so that the locker and buffers_released are last like they used to be | 22:01 |
| T42 | <b100dian> I will again testvthat my answer to the' else' question was 100% correct tomorrow | 22:02 |
| T42 | <b100dian> Ok, I will try that too | 22:02 |
| T42 | <b100dian> It lead me to crashes sometimes, but maybe because of the clear..? | 22:02 |
| mal | maybe | 22:03 |
| mal | @b100dian ah, maybe you need to have the locker first but the buffers_released should be last | 22:05 |
| mal | that frameAvailable uses the m_lock also | 22:06 |
| T42 | <elros34> Walid what is your current branch in droid-config? What was the error without pid_ns? | 22:10 |
| T42 | <Walid> i will rebuild and check log, but the main error is lipstick can't start, droid-hal-init status is active and running (re @elros34: Walid what is your c...) | 22:12 |
| T42 | <elros34> did you add files needed files according to hadk-faq hybris-16 section? did you check logcat? | 22:13 |
| T42 | <elros34> still don't know which is your current branch in droid config | 22:13 |
| T42 | <Walid> yes i do it, i will check again log and put it in pastbin (re @elros34: did you add files ne...) | 22:37 |
| T42 | <Walid> master branch (re @elros34: still don't know whi...) | 22:44 |
| T42 | <elros34> then you are missing all hybris-16 related changes | 22:50 |
| T42 | <Walid> what change must do ? have any related docs? (re @elros34: then you are missing...) | 22:53 |
| T42 | <elros34> these from hadk-faq, 'master' branch is empty unless you use 'main' branch. If you don't know then maybe you also confuse it in sfos image buil | 22:55 |
| T42 | <elros34> d | 22:56 |
| T42 | <Walid> you mean this docs https://etherpad.wikimedia.org/p/hadk-hot ? i just see it now, (re @elros34: these from hadk-faq,...) | 23:05 |
| T42 | <elros34> no hadk-faq changes you have in 'main' but you sad that you use 'master' branch. if this is really the case then you are missing all this changes in device | 23:08 |
| T42 | <elros34> so lipstick will not work | 23:08 |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!