Tuesday, 2021-03-09

T42<edp_17> mal: Sorry, I wasn't near tg until now. Yes, symlinking that /etc/media_codecs_google_video_le.xml to /etc helped! Now the video recording works!00:17
T42<edp_17> Hi all! I am trying to fix video recording on the Nexus 5 (hammerhead, cm 12.1 base, sfos When trying to generate the jolla-camera-hw.txt, I am getting this error: https://paste.ubuntu.com/p/jbfJNZfPj6/01:15
T42<edp_17> When recording a video the camera hangs immediately (video recording doesn't start) and this is in the log: https://pastebin.ubuntu.com/p/X35Gp4JV4Y/01:36
T42<edp_17> The jolla-camera-hw.txt: https://paste.ubuntu.com/p/xYBR7kf2Bq/01:36
T42<edp_17> This error is flooding logcat: "E/qdhwcomposer( 1696): isValidDimension: layer handle is NULL"01:36
T42<edp_17> Do you have idea how to fix video recording?01:37
T42<edp_17> Another question. I've built a new image for the n7000 to include all changes so far. The initial setup exits right after selecting the language. In the log there is a strange error repeated "unable to open file '/etc/dconf/db/vendor' and seemingly cannot find the images for pixel ratio 1.0 (Although it is looking in the wrong place /u02:35
T42sr/share/themes/sailfish-default/meegotouch/z1.0/ instead of /usr/share/themes//meegotouch/z1.0/): https://paste.ubuntu.com/p/NvNMyMpZ3X/02:35
T42<edp_17> Anyone experienced similar?02:35
T42<elros34> @edp_17 'layer handle is NULL' is caused by qt5-qpa-hwcomposer-plugin, you can grep for that message in $ANDROID_ROOT/hardware and comment it out. If you use my external_display branch the this also will generate such an errors (a31 branch might be better)08:12
T42<elros34> @edp_17 do you have dconf installed and that /etc/dconf/db/vendor file exists?08:19
T42<edp_17> @elros34 : Yes, on the Nexus 5 I use your external_display branch as I wanted to try HDMI (although, so far I didn't have time for it).08:53
T42<edp_17> I'll comment that out, to stop that flood. Thanks.08:53
T42<elros34> and I will repeat again don't use that branch08:53
T42<edp_17> Okay, is the other branch has the hdmi stuff?08:54
T42<elros34> a3108:54
T42<edp_17> Okay, thanks.08:55
T42<edp_17> @elros34 [@edp_17 do you have dconf installed and that / …], Do you meant dconf installed in sdk or on device?08:55
T42<elros34> on device, check whether dconf update (as root) works08:56
T42<edp_17> I'll double check that. I expected to have it on device as this was a fresh build. I don't need to install them manually on other ports.08:58
* lbt happy to see rinigus' comments about OBS on sfos forum - makes a difference in internal discussions09:54
T42<adampigg> will need to see that09:55
riniguslbt: thanks! was thinking whether to raise here or at forum. will try to use forum more after your tip09:57
lbtyes - I don't think veskuh is on irc much anymore09:58
lbtessentially we're trying to understand what benefits the OBS brings to the community that can't be done in the SDK09:58
piggzlbt: ports for a start....09:59
piggzlbt: then complex app builds like puremaps and amazfish09:59
lbtmultiple releases10:00
piggzhosting port packages too, though i understand you may be worred about blobs10:00
riniguslbt: we tried to summarize it in that thread with piggz. but multiple releases we may have missed and actually are using OBS for it. although, it is not super trivial as SFOS tooling improves and it is tempting to drop support for some older SFOS releases10:01
piggzid like to see your maintenance buren reduced ... maybe that would happen if you ran the same version as you run internally?10:01
piggzre amazfish ... i have to build a bunch of supproting libs too, which isnt easy/possible in the sdk10:02
piggzplus those need different builds per sfos release10:03
lbtthe plan is/was to use the same infra processes for the community OBS - that's mainly there10:03
piggzand for managing ports .... i have mido, latte, pine, pro1, volla ..... each with devel: and testing: branches ..... cant imaging doing that without obs10:03
lbtYeah - I think the way in which OBS makes the porter's life easier is not fully appreciated10:06
riniguslbt: sounds like a good plan.10:06
lbttbh that's an area I've not been involved in for a long while10:06
* lbt is playing more with HA as home automation nowadays :D :D10:07
piggzlbt: HA runs my heating ;)10:07
lbtI do have Silica running on my Raspi 4 with a 7" display...10:07
lbtyeah - me too - I have a really nice python/MQTT/Qt5/Silica setup10:08
piggzneat ... there is a sfos HA abb btw....10:08
piggzthe same box runs my dns with the adguard plugin10:08
lbtMQTT <> signal/slot library too10:08
lbtall asyncio aswell10:08
piggzi have a tasmota temp/humitiy sensor with the flashed firmware, connected to a relay to control the boiler10:09
piggzwith all the network presence detection stuff thrown in to confuse things10:09
lbtoh - and it's all running django too (as an ORM layer) so web ui will be easy10:09
lbtah - I built all my own stuff with pi-zeros10:10
piggzi dont have the time for that!10:10
piggzanyway, i guess you can be our OBS advocate!10:10
lbthehe ... why do you think I'm not around much anymore10:10
lbtwell, I do advocate some - but everyone knows I'm pro-OBS and I need others to make their voices heard10:11
lbtthat's the really important thing - openrepos took a lot of the community away from OBS and it's lost relevance as the community has moved more to being a single-app developer type place10:12
lbtwhat OBS used to do was make it easier to pull in a SFOS core package and 'fix' it when your app needed something in a lib fixing. That barrier is higher now I think10:13
lbtso OBS as a gateway into the OS itself10:14
T42<edp_17> @elros34 [on device, check whether dconf update (as root …], This is what I got: https://paste.ubuntu.com/p/ypPgzWzmTF/10:16
T42<edp_17> dconf cannot be found in any folder10:17
riniguslbt: interesting angle regarding single-app development. I wonder whether we have too few developers going for larger apps requiring more libs? although, looking at some matrix apps, they require some support libs as well.10:17
T42<edp_17> @edp_17 [dconf cannot be found in any folder], Sorry, it is there in /usr/bin/dconf10:17
T42<elros34> error is quite clear, you have wrong value in jolla-camera-hw.txt which cause dconf update fail10:18
rinigusbut indeed, larger projects - maybe can even look into qt updates - do require OBS.10:18
rinigusI wonder, how much we are heard with piggz or are we also now just in a box of OBS advocates :)10:18
T42<edp_17> @elros34 [error is quite clear, you have wrong value in …], yeah, I just read that too. But that is what the droid-camres -w wrote into it. What should I use instead of those values?10:19
T42<elros34> I guess you need to amend it based on GST_DEBUG output or remove that at all. I don't have that key but I generated jolla-camera-hw.text long time ago10:21
lbtrinigus: you're both extremely important voices actually.10:23
T42<edp_17> Oh, I wasn't aware of that I could remove those. I'll experiment with them. Thanks.10:25
T42<elros34> I don't know whether you can but runing 'GST_DEBUG=6 mk-cam-conf 0 /dev/null 2>&1 | grep params_parse | sed -e 's/.*param\s//' | sort -u | grep -i frame' should be more proper10:26
riniguslbt: thank you. we will do our best to be active10:26
T42<elros34> lbt: do you know whether obs use createrepo_c for creating repodata/*xml.gz or something else?10:28
T42<edp_17> @elros34 : That was the issue! Once I fixed the jolla-camera-hw.txt and did a dconf update, the initial setup didn't stop. 😃 Thanks!10:28
lbt99% sure it uses createrepo (no _c)10:29
T42<elros34> lbt: repodata/*-primary.xml.gz on obs is different than the one generated on platform sdk by creatrepo_c. primary.xml.gz is missing recommended packages. Maybe that createrepo is the culprit10:31
piggzlbt: thanks!10:33
piggzon that note, given how important I am, someone could look at my geck-dev MRs :D10:33
lbtthat's a good spot elros3410:38
lbtthe OBS is quite old so I'm pushing to do an upgrade atm - that may also fix aarch6410:38
riniguslbt: would be great!10:41
T42<elros34> I have tried to confirm that by testing createrepo from older platform sdk (2.1.4) but it's not installed by default and jolla repos  for that release are down so so could be wrong guess. Great that you want to upgrade whole thing10:41
piggzlbt: that will be great10:42
T42<hacker12455> mal: the device now seems to hang on bootlogo10:53
T42<hacker12455> did not change the kernel config10:53
T42<hacker12455> only the cmdline10:53
T42<hacker12455> and added the files10:53
malcan you telnet in?10:54
mal@edp_17 that camera hang log has "Waiting for service media.audio_policy" so that tells what you need to do11:33
T42<edp_17> mal: I guess need to add MINIMEDIA_AUDIOPOLICYSERVICE_ENABLE := 1 and AUDIOPOLICYSERVICE_ENABLE := 1.11:37
mal@elros34 thanks for the hint about ssu domain in sdk, that seems to be a bug in there and now we know why it was wrong11:38
andreysysHi, everyone11:38
andreysysI found some strange for me into gst-droid, so it is about fps-range for image and video modes. Image and video modes have different settings for the frame rate like fps-range. In the image mode, the viewfinder shows frames much more slowly then in the video mode. I see that on many devices. I set condition for the image mode according to the video mode. After that the frame rate became much more faster. My question for the experts, Why that11:38
andreysysconditions are different? And, would be better to combine it into the one function or do the image setting the same of the video setting? Here I post link to the code https://github.com/sailfishos/gst-droid/pull/17/files#diff-08720511b477eca8bd977129cafcc4eb7ba233bc7ff5b852f234d1e5576e06e9R540    https://github.com/sailfishos/gst-droid/pull/17/files#diff-08720511b477eca8bd977129cafcc4eb7ba233bc7ff5b852f234d1e5576e06e9R48911:38
T42<edp_17> mal: Okay. I though need the other one too  (AUDIOPOLICYSERVICE_ENABLE := 1).11:39
mal@edp_17 code has this: ifeq ($(MINIMEDIA_AUDIOPOLICYSERVICE_ENABLE),1)11:40
T42<edp_17> What does the AUDIOPOLICYSERVICE_ENABLE := 1 control?11:40
T42<edp_17> Oh, and I read somewhere to set FORCE_HAL:=1. What's that?11:41
malthat is some rare thing, usually not needed11:41
T42<edp_17> mal: Okay, thanks. I'll remove the AUDIOPOLICYSERVICE_ENABLE.11:42
T42<edp_17> mal: Another question, how should I decide whether I need audiosystem-passthrough-dummy-af?11:43
malit depends again what you see in logcat and if camera and video codecs work11:44
T42<edp_17> mal: I see, so we don't know for sure in advance.11:45
malandreysys: I need to check that part of the code to see why those are different11:46
T42<hacker12455> mal: no usb detected11:53
T42<hacker12455> so can't telnet11:53
andreysysmal:  Thank You for the answer. I have put here two links on those conditions, to quick search.11:59
*** vup2 is now known as vup13:06
techgeeksterHi! In the process after the repo sync, /mer-kernel-check/mer_verify_kernel_config cannot be found. Is there something wrong with the repo sync process?14:03
techgeeksterI mean from my side. I am sure that the process works fine. Maybe I did something wrong14:03
T42<elros34> if repo would finish without issues then you would have in place when repos where cloned  $ANDROID_ROOT/hybris/mer-kernel-check: https://github.com/mer-hybris/android/blob/hybris-17.1/default.xml#L2414:17
T42<edp_17> Could you please advise me what setup file is used or testing the camera in csd? (Both pictures/video do work on both front/main cameras, but in csd both only show a black screen.)14:30
T42<jgibbon> @lbt Wait, Silica on a Raspberry Pi 4 with (the official, I presume) touch screen? I guess that's a bit more recent than the old "SailPi" experiment that's lost in the nether regions of the internet… Is there someone I can bribe to get my hands on an sd image or build script? ;)14:38
techgeekster<T42> I think it did not complete the syn correctly. It exceded the maximum of 120GBs of my VM so yes....14:51
lbtjgibbon tbh I assumed it wouldn't be shareable ... I'll take a look14:51
lbtit's a bit ugly to build. I hacked the src a bit and then used rpmbuild in debian to make rpms which I converted to debs using alien14:53
T42<adampigg> (Document) https://irc.thaodan.de/.imgstore/DA0xKNXXKF.mp414:53
lbtI'm running it fullscreen egl at the minute and having problems with wayland... specifically this bug: https://bugreports.qt.io/browse/QTBUG-8461414:54
piggztbt: on that version of qt, or good old 5.6?14:56
lbtand yeah ui/sailfish-silica is closed :(14:59
lbtI don't think I'm allowed to share a blob but I have asked15:02
T42<jgibbon> @lbt thanks, much appreciated – I don't expect a positive answer, as well.15:22
T42<jgibbon> OTOH I can sign an NDA if needed :D15:22
piggzi guess the interesting part is saifish running on qt 5.15 at all!15:27
piggzany licenses being broke there?15:28
T42<jgibbon> It sounded like debian, though...15:37
riniguspiggz: no licenses broken until distributed. but that is very promising info16:12
T42<calebccff> Damn I'd be /very/ interested in getting that UI running on the OnePlus 6. Lomiri runs really well on Mainline there16:34
T42<calebccff> My last attempt at building sailfishOS for mainline didn't go so well, almost definitely Qt related, Lomiri only works because of the newer mesa and Qt supported by Manjaro ARM16:36
lbtjgibbon: I'm afraid the answer was "no" :/   Sorry ... but it was worth asking about. To be honest it makes little sense to base against it on RPi until/unless silica gets opened.16:36
lbtoh and I was a touch off - my silica Qt is 5.11 - the 5.15 one is what I'm using to try and debug wayland on the pi16:38
T42BNONA_20 was added by: BNONA_2016:57
T42<elros34> @Mister_Magister: probably fix for not working test_hwcomposer. Please leave a feedback if you will test it. https://github.com/elros34/libhybris/commit/3b024539a7b1d1eb9b7b1d36079f8920a14d1cb918:02
Mister_Magister@elros34 what you talking about?18:02
T42<elros34> didn't yuo have issue that test_hwcomposer show nothing after you tested qcom_bsp fix?18:03
rinigusmal, piggz: that /dev/cg2_bpf patch fixed it and I don't need systemd.legacy_systemd_cgroup_controller boot option anymore.18:22
rinigusSummary with the fix for IRC archive18:22
rinigusIf you get Failed to create /user.slice/user-100000.slice/user@100000.service/init.scope control group: Permission denied and failing to get to GUI18:22
rinigusand you have newer kernel (according to piggz, 4.5 or above) then make sure to apply https://github.com/mer-hybris/hybris-patches/commit/54008fe9fc67c13eff6d3d48f78bb268af7883be18:22
techgeeksterAnother question. During the make session (make -j1 hybris-hal droidmedia) this error occurred "FAILED: ninja: 'out/target/product/raphael/kernel', needed by 'out/target/product/raphael/boot.img', missing and no known rule to make it18:26
techgeekster18:01:32 ninja failed with: exit status 118:26
techgeekster". What should I do in this occasion18:26
piggzrinigus: does that do effectively the same as using legacy CG though?  you just swapped the cg2 for a cg1 mount?18:27
malrinigus: good that you got it fixed18:33
riniguspiggz: I don't think so, see comment in the beginning of the patch. mal: thanks18:46
T42<elros34> mal: looks like this commit broke test_hwcomposer for me: https://github.com/libhybris/libhybris/commit/bd08484162d4c5ab9906cfa573105635c9eb8884. Adding <android-config.h> to test_common.h seems to fix the issue18:46
mal@elros34 argh, so many places where that is needed19:01
mal@elros34 so which header needs to have it19:01
malI mean what part breaks it19:01
malagain the gralloc issue19:01
T42<elros34> mal: without that include test_hwcomposer didn't show anything and would probably also crash but I have some null check in hwcomposer. Not sure which excatly header needed it.19:05
mal@elros34 did you add it as the first include in test_common.h?19:06
T42<elros34> ah yes19:08
mal@elros34 do you want to make a PR or should I?19:13
mal@elros34 I wonder if the <android-config.h> include should really be in hwcomposer_window.h which includes gralloc.h19:14
T42<elros34> I can test that and report later if you think that is better19:14
malI need to check the code a little19:15
T42<elros34> sure no rush, I am happy that I can use latest libhybris now19:16
mal@elros34 even better idea, remove the gralloc.h include from hwcomposer_window.h19:19
malthat header doesn't seem to contain any gralloc code anyway19:20
malalso the fb.h include might not be needed there19:22
malalthough still it might make sense to have android-config.h include in test_common.h also19:27
piggzmal: hows the pinephone hacking?21:45
T42<F1a5H> Are yall using irc on sailfish os?21:57
T42<jgibbon> lbt no worries, thanks for asking – I couldn't imagine someone in charge possibly being cool with that, these things can easily become a licensing nightmare. having silica run on an sbc is definitely interesting.22:28
T42<F1a5H> (Photo, 720x1520) https://irc.thaodan.de/.imgstore/bCCaUC0R49.png22:34
T42<F1a5H> Bitchx on sailfishos22:34

Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!