**** BEGIN LOGGING AT Wed Feb 14 03:00:02 2018
Feb 14 06:18:26 <Herrie> Tofe: Got test_hwcomposer to work!
Feb 14 07:58:08 <Tofe> Herrie: oh great ! What was the missing point ?
Feb 14 08:11:00 <Herrie|Laptop> Tofe: I got the bootloop back due to firmware not being fixed in new image yet, but it was the android_property_service PR I pointed at yesterday
Feb 14 08:44:00 <Herrie|Laptop> I still don't have luna-next working though... Any clues?
Feb 14 08:56:47 <Herrie|Laptop> Let me first push firmware back
Feb 14 08:56:52 <Herrie|Laptop> TO get rid of bootloop
Feb 14 08:59:12 <Herrie|Laptop> Tofe: I still needed to manually do export EGL_PLATFORM=hwcomposer in order to get test_hwcomposer to work
Feb 14 08:59:21 <bshah> that's correct
Feb 14 08:59:31 <bshah> that's how it is supposed to work
Feb 14 08:59:43 <bshah> Herrie|Laptop: re luna-next, do you have logs?
Feb 14 09:00:55 <Tofe> Herrie|Laptop: for luna-next, keep in mind you'll need to rebuild the qpa recipe with the correct QTI_* cflags
Feb 14 09:03:13 <Herrie|Laptop> Tofe: Ah yes
Feb 14 09:03:36 <Herrie|Laptop> bshah: Still have bootloop due to wcnss after placing files in /etc/firmware
Feb 14 09:03:42 <Herrie|Laptop> So retracting my steps from last week
Feb 14 09:04:00 <bshah> hm
Feb 14 09:04:15 <bshah> I find it curious that it still bootloops after that kernel patch
Feb 14 09:04:18 <bshah> :/
Feb 14 09:04:25 <Herrie|Laptop> bshah: It was my firmware bits
Feb 14 09:04:34 <Herrie|Laptop> I was playing around last week for that
Feb 14 09:04:38 <Herrie|Laptop> And did some things
Feb 14 09:04:43 <Herrie|Laptop> Could be I forgot a step now
Feb 14 09:05:11 <bshah> Herrie|Laptop: persist partiton?
Feb 14 09:08:00 <Herrie|Laptop> It would help if I would put stuff in lib/firmware instead of etc/firmware I guess
Feb 14 09:10:46 <Herrie|Laptop> Seems it doesn't bootloop now at least :P
Feb 14 09:11:56 <Herrie|Laptop> bshah: test_hwcomposer gives me a white rectangle artifact after a while on top right. Is this "normal" ?
Feb 14 09:12:09 <bshah> probably not
Feb 14 09:12:24 <bshah> Herrie|Laptop: --enable-adreno-quirks in libhybris please
Feb 14 09:12:47 <Herrie|Laptop> bshah: OK let me try
Feb 14 09:17:48 <Herrie|Laptop> Tofe: You mean I need this one? https://github.com/webOS-ports/meta-webos-ports/blob/pyro/meta-luneos/recipes-qt/qt5/qt5-qpa-hwcomposer-plugin/0002-Add-QCOM_BSP-define-switch.patch
Feb 14 09:17:58 <bshah> yeah
Feb 14 09:18:09 <bshah> with QTI_BSP as well included
Feb 14 09:21:47 <Herrie|Laptop> Would be good when GitHub would show search results in forks :P
Feb 14 09:23:15 <Herrie|Laptop> bshah: The --enable-adreno-quirks didn't do the trick here
Feb 14 09:23:22 <bshah> :-(
Feb 14 09:24:09 <bshah> if you strace hwcomposer test... do you see any messagees about it running out of file descripters eg?
Feb 14 09:24:28 <bshah> (when flickering or rectangle artifact is shown that is..)
Feb 14 09:25:47 <Tofe> Herrie|Laptop: This patch was made for tenderloin, and only activates QCOM_BSP; you'll need QTI_BSP and so on, exactly like in the libhybris recipe
Feb 14 09:26:03 <Herrie|Laptop> bshah: Bad file descriptor I see a number of times
Feb 14 09:26:15 <Herrie|Laptop> https://bpaste.net/
Feb 14 09:26:25 <Herrie|Laptop> Oops https://bpaste.net/show/4d3d699d9f20
Feb 14 09:26:25 <bshah> wrong paste ^
Feb 14 09:29:16 <bshah> hm possibly that is not --enable-adreno-quirks thingi
Feb 14 09:29:54 <bshah> Herrie|Laptop: see: https://github.com/libhybris/libhybris/issues/352
Feb 14 09:30:44 <Herrie|Laptop> bshah: That makes sense
Feb 14 09:31:21 <Herrie|Laptop> I guess I can try that
Feb 14 09:32:44 <Herrie|Laptop> bshah: Might be good to introduce another flag for libhybris for this?
Feb 14 09:32:55 <Herrie|Laptop> Seeing it can take forever to sort upstream and trickle down?
Feb 14 09:33:06 <bshah> maybe dunno.. well first let's verify if it is same issue
Feb 14 09:33:25 <Herrie|Laptop> Yeah
Feb 14 09:33:27 <Herrie|Laptop> Doing that now
Feb 14 09:49:23 <Herrie|Laptop> bshah: With that change the spiral doesn't spin and lock up
Feb 14 09:49:26 <Herrie|Laptop> So does strace
Feb 14 09:49:59 <Herrie|Laptop> Then again, the code snippet that krnlyng pasted differed quite a bit from my libhyrbis code
Feb 14 09:50:04 <Herrie|Laptop> So could be I patched it wrongly
Feb 14 09:50:37 <Herrie|Laptop> I.e. I didn't have _hybris_hook_my_printf and also no #ifdef WANT_INITIALIZE_BIONIC in my hook.c from upstream
Feb 14 09:51:06 <Herrie|Laptop> Did have static struct _hook hooks_common in there, so simply pasted the added lines there, but could be I need more work there
Feb 14 09:51:23 <Herrie|Laptop> I.e. this is what I did: https://github.com/Herrie82/libhybris/commit/1d6b16dc194078c49b50c3721ce80d5c26d2ff52
Feb 14 10:36:03 <Herrie|Laptop> Tofe: I'm inclined to drop the BSP patch for tenderloin in favor of https://github.com/Herrie82/qt5-qpa-hwcomposer-plugin/commit/5c048c23223abb9c10b7b1480dcd6e8b74b9c560
Feb 14 10:36:45 <Herrie|Laptop> And then using https://github.com/webOS-ports/meta-webos-ports/blob/pyro/meta-luneos/recipes-luneos/webos-keyboard/webos-keyboard.bb#L32
Feb 14 10:36:52 <Herrie|Laptop> with _append_devicename
Feb 14 10:36:56 <Herrie|Laptop> How does that sound?
Feb 14 10:37:01 <Herrie|Laptop> Looks like a more generic approach to me?
Feb 14 10:47:04 <Herrie|Laptop> bshah/Tofe: This is what I get when I grep around: https://bpaste.net/show/9fc2c3be28c2
Feb 14 10:47:28 <Herrie|Laptop> So that would mean I should do QCOM_BSP and QCOM_HARDWARE? And not QTI_BSP?
Feb 14 10:47:52 <Herrie|Laptop> If I understood correctly QTI_BSP replaced QCOM_BSP that existed previously?
Feb 14 10:48:22 <Herrie|Laptop> However there are targets that still use QCOM_BSP?
Feb 14 10:49:59 <bshah> I think some code does yeah
Feb 14 10:52:22 <Herrie|Laptop> Well I guess I could define all 3
Feb 14 10:52:29 <Herrie|Laptop> Not sure that makes a difference
Feb 14 10:55:16 <Herrie|Laptop> Doesn't seem to make a difference if I use QCOM_BSP or QTI_BSP for test_hwcomposer
Feb 14 10:56:12 <Herrie|Laptop> Checking now for qt5-qpa-hwcomposer-plugin
Feb 14 10:56:23 <Herrie|Laptop> bshah: FYI strace on luna-next: https://bpaste.net/show/577b6f75b74e
Feb 14 11:05:38 <Herrie|Laptop> Hmmz no luck with qt5-qpa-hwcomposer it seems
Feb 14 11:06:45 <Herrie|Laptop> New strace: https://bpaste.net/show/ab7f7ecbfdab
Feb 14 11:38:21 <Herrie|Laptop> OK some more clue as to why the UI doesn't come up: ERROR: Caught a segmentation fault while loading plugin file: /usr/lib/gstreamer-1.0/libgstlibav.so
Feb 14 11:56:17 <Tofe> Herrie|Laptop: pulseaudio-droid-module also needs this kind of define
Feb 14 11:56:32 <Herrie|Laptop> Tofe: I thought I had that
Feb 14 11:56:33 <Herrie|Laptop> Let me check
Feb 14 11:57:02 <Tofe> Herrie|Laptop: as for your proposal, well, why not
Feb 14 11:58:23 <Herrie|Laptop> Tofe: Well we shouldn't be the only ones needing this I guess?
Feb 14 11:58:30 <Herrie|Laptop> SFOS would also need it at some point
Feb 14 12:24:58 <Herrie|Laptop> Tofe: Well I do, but PA still not happy
Feb 14 12:25:44 <Herrie|Laptop> https://bpaste.net/show/eef965faa61e
Feb 14 12:32:27 <Tofe> well, not sure then...
Feb 14 12:35:55 <Herrie|Laptop> Tofe: Well PA is a bit of a black box at time
Feb 14 12:35:57 <Herrie|Laptop> +s
Feb 14 12:36:08 <Herrie|Laptop> I guess that can cause the gstreamer to segfault :P
Feb 14 12:36:38 <Herrie|Laptop> For mako we had some alsa-lib configuration: https://github.com/webOS-ports/meta-webos-ports/blob/pyro/meta-luneos/recipes-multimedia/alsa/alsa-lib/0002-Add-UCM-configuration-for-Nexus-4-Mako.patch
Feb 14 12:36:42 <Herrie|Laptop> Not sure I might need something similar
Feb 14 12:37:01 <Herrie|Laptop> I'll browse a bit around on halium mer-hybris etc to see
Feb 14 12:37:30 <Tofe> Herrie|Laptop: I remember we also patched our PA webos config to disable alsa completly
Feb 14 12:38:09 <Herrie|Laptop> Tofe: Ah could be
Feb 14 12:50:30 <Herrie|Laptop> Tofe: We did that for Mako yes
Feb 14 12:50:35 <Herrie|Laptop> I guess I could try the same here?
Feb 14 12:51:04 <Herrie|Laptop> And for Tenderloin but not for Hammerhead it seems :S
Feb 14 12:53:52 <Herrie|Laptop> It doesn't hurt to try I guess
Feb 14 12:59:19 <Herrie|Laptop> Log seems better now
Feb 14 12:59:26 <Herrie|Laptop> paplay allows me to play .wav
Feb 14 12:59:34 <Herrie|Laptop> So sound seems to work through PA then?
Feb 14 13:03:27 <Tofe> yep
Feb 14 13:03:28 <Tofe> good !
Feb 14 13:03:47 <Herrie|Laptop> Still same crash in luna-next though :S
Feb 14 13:03:53 <Tofe> Herrie|Laptop: yes we could do the same patch for all libhybris targets I think
Feb 14 13:06:18 <Herrie|Laptop> Tofe: OK so we'd need to test it for Hammerhead...
Feb 14 13:06:35 <Herrie|Laptop> The environment.conf seems fine. Same values as the mer guys: https://github.com/piggz/droid-config-mido/blob/master/sparse/var/lib/environment/compositor/droid-hal-device.conf
Feb 14 13:06:57 <Herrie|Laptop> https://github.com/webOS-ports/meta-webos-ports/blob/pyro/meta-luneui/recipes-luneui/luna-next/luna-next-conf/mido/environment.conf
Feb 14 13:07:02 <Herrie|Laptop> So it seems to be gstreamer somehow
Feb 14 13:13:46 <Tofe> Do we have anything related to libhybris in gstreamer ?
Feb 14 13:21:36 <Herrie|Laptop> Let me check
Feb 14 13:21:42 <Herrie|Laptop> Tofe: I think we could drop: https://github.com/webOS-ports/meta-webos-ports/blob/herrie/qt5qpa/meta-luneos/recipes-qt/qt5/qt5-qpa-hwcomposer-plugin/0002-Fix-4.4.2-hwcomposer-build.patch ?
Feb 14 13:22:37 <Herrie|Laptop> Seems to be for 4.4 specifically?
Feb 14 13:31:24 <Herrie|Laptop> Tofe: If I read https://github.com/mer-hybris/pulseaudio-modules-droid/blob/d53995823d335a06a25b98d9a7ecd25f06afbcea/src/common/droid-util-audio.h correctly, we'd only need to define QCOM_BSP? QCOM_DIRECTTRACK doesn't seem to do anything. Or we could define QCOM_HARDWARE directly I guess
Feb 14 13:31:42 <Herrie|Laptop> QTI_BSP doesn't seem to do anything for pulseaudio-modules-droid
Feb 14 13:39:00 <Herrie|Laptop> Tofe: Seems we don't do anything gstreamer ourselves, but qtmultimedia used by camera and luna-next depends on it
Feb 14 13:43:43 <Tofe> Herrie|Laptop: I think the safest way to handle the android headers is that whenever they are used, we define the same things as in Android for that target.
Feb 14 13:44:17 <Herrie|Laptop> Tofe: OK so everywhere where we have virtual/android ?
Feb 14 13:44:25 <Tofe> android-headers
Feb 14 13:44:55 <Herrie|Laptop> Yeah
Feb 14 13:45:57 <Tofe> doid-util-audio.h includes hardware/audio.h, which maybe includes other things, and looking into each one to check if there is an #ifdef or not is tedious; it's far simpler to just have the same build environment as during the Halium build
Feb 14 13:46:05 <Herrie|Laptop> libhybris, pulseaudio-modules-droid, qt5-qpa-hwcomposer-plugin, qtscenegraph-adaptation, qtsensors-sensorfw-plugin, sensorfw and nyx-modules-hybris based on grep
Feb 14 13:46:22 <Tofe> ok, yes, looks correct
Feb 14 13:47:08 <Tofe> you'll probably already find tenderloin patches for these
Feb 14 13:47:14 <Herrie|Laptop> Tofe: ehm nope
Feb 14 13:47:36 <Tofe> oh?
Feb 14 13:47:54 <Tofe> no QCOM_BSP and QCOM_DIRECTTRACK ?
Feb 14 13:53:13 <Herrie|Laptop> Tofe for Tenderloin we have the QCOM_BSP for qt5-qpa-hwcomposer-plugin and QCOM_DIRECTTRACK  and QCOM_HARDWARE for pulse-audio-modules-droid
Feb 14 13:53:23 <Herrie|Laptop> We don't have anything in the other recipes
Feb 14 13:53:44 <Herrie|Laptop> Then again the 5.1 build for Tenderloin isn't a stable as the 4.4 used to be
Feb 14 13:53:49 <Herrie|Laptop> That could be explained by this :P
Feb 14 13:54:11 <Tofe> Herrie|Laptop: maybe, or when I began working I got a bit lazy...
Feb 14 13:54:17 <Tofe> s/I/it/
Feb 14 13:54:54 <Herrie|Laptop> Tofe: Also possible
Feb 14 13:55:07 <Herrie|Laptop> But you suggest to add these flags to all recipes?
Feb 14 14:10:20 <Tofe> Herrie|Laptop: do you think it's a bit too much?
Feb 14 14:10:40 <Herrie|Laptop> Tofe: You ask me :P ?
Feb 14 14:10:46 <Herrie|Laptop> I've got no clue whatsoever
Feb 14 14:14:14 <Tofe> I'd say, yes, we should include it in these recipes. Though I'd like to find a better way, like putting it directly in android-headers somehow
Feb 14 14:16:54 <Herrie|Laptop> To me it seems to make sense.. Whenever I add these flags things seem to start magically working :P
Feb 14 14:17:43 <Herrie|Laptop> bshah: Are you aware of QCOM_DIRECTTRACK? From what I can tell it seems to be mainly 8960 chips (like our good old Tenderloin) and not something for recent targets
Feb 14 14:17:54 <Herrie|Laptop> QCOM_BSP and QTI_BSP seem to be interchangable?
Feb 14 14:26:28 <Tofe> Herrie|Laptop: QTI_BSP doesn't exist for the tenderloin build
Feb 14 14:26:54 <Herrie|Laptop> Tofe: OK. QCOM_BSP does?
Feb 14 14:26:59 <Tofe> yup
Feb 14 14:27:19 <Herrie|Laptop> And if you grep the DIRECTTRACK?
Feb 14 14:30:28 <Tofe> I didn't grep, I say that from memories :p DIRECTTRACK does exist too
Feb 14 14:41:58 <Herrie|Laptop> Tofe: Seems you might be right: https://github.com/mer-hybris/audioflingerglue/blob/master/services/audioflinger_5_1_0.h#L75
Feb 14 14:42:05 <Herrie|Laptop> Seems to be there for 5.1, not for 7.1 anymore
Feb 14 14:48:11 <Herrie|Laptop> QCOM_DIRECTTRACK seems audio only though
Feb 14 15:46:58 <Herrie|Laptop> bshah: You have any idea how I can get test_sensors to find my sensors?
Feb 14 15:47:18 <bshah> see if sensorservice is running?
Feb 14 15:47:25 <Herrie|Laptop> https://bpaste.net/show/3e0d4bf15f4f
Feb 14 15:49:46 <Herrie|Laptop> bshah: In logcat or the sensorfw one?
Feb 14 15:50:06 <bshah> in android side
Feb 14 15:51:01 <Herrie|Laptop> slim_daemon: [NDK] bindNDKSensors: Sensor server is unavailable.
Feb 14 15:51:15 <bshah> no I mean
Feb 14 15:51:17 <Herrie|Laptop> 02-12 08:54:00.307     0  3053 W qti_sensors_hal: sensors_get_sensors_list: handle 17 is not registered!
Feb 14 15:51:21 <bshah> is sensorservice process is running?
Feb 14 15:52:14 <Herrie|Laptop> How I figure that out?
Feb 14 15:52:37 <bshah> ps? :P
Feb 14 15:56:04 <Herrie|Laptop> I guess I need some params there because it only returns very little info...
Feb 14 15:56:14 <Herrie|Laptop> https://bpaste.net/show/b1eb0c57c358
Feb 14 15:56:30 <bshah> ps -ef
Feb 14 15:56:39 <bshah> ps -ef | grep sensorservice
Feb 14 15:59:03 <Herrie|Laptop> root     10861  7732  0 09:11 pts/2    00:00:00 grep sensorservice
Feb 14 15:59:14 <bshah> okay
Feb 14 16:00:50 <bshah> Herrie|Laptop: please add followig in system/core/rootdir/init.rc ; https://ptpb.pw/QQG-
Feb 14 16:03:11 <Herrie|Laptop> Does it matter where?
Feb 14 16:03:36 <bshah> not really no
Feb 14 16:04:05 <bshah> I just have it at end
Feb 14 16:04:49 <Herrie|Laptop> OK
Feb 14 16:04:56 <Herrie|Laptop> Any reason why this is not upstream yet?
Feb 14 16:04:59 <Herrie|Laptop> Or it's device specific?
Feb 14 16:09:16 <Herrie|Laptop> Seems it's in the 5.1 branch but not in 7.1 one yet?
Feb 14 16:56:12 <Herrie|Laptop> bshah: That partially worked.... I.e. It seems it's trying to start it but gets Feb 12 08:53:52 mido init: cannot find '/system/bin/sensorservice.disabled' (No such file or directory), disabling 'sensorservice'
Feb 14 16:56:23 <Herrie|Laptop> I can manually start it and it seems to run then
Feb 14 16:56:40 <bshah> /system/bin/sensorservice.disabled .. o.o
Feb 14 16:56:51 <Herrie|Laptop> But test_sensors still the same output
Feb 14 16:56:55 <Herrie|Laptop> Any suggestions?
Feb 14 16:56:55 <bshah> that looks wrong patch in initrc?
Feb 14 16:57:43 <Herrie|Laptop> I'm grepping around a bit now ;)
Feb 14 16:58:00 <Herrie|Laptop> In my halium-luneos-7.1 folder to see where that comes from
Feb 14 17:05:15 <Herrie|Laptop> grep doesn't find any sensorservice.disabled in my halium dir :P
Feb 14 17:07:37 <Herrie|Laptop> OK found the culprit
Feb 14 17:08:48 <Herrie|Laptop> Seems to come from https://github.com/shr-distribution/meta-smartphone/blob/pyro/meta-android/recipes-core/android-system/android-system/20-remove-services#L6
Feb 14 17:10:07 <Herrie|Laptop> I guess I need to have a look at this with Tofe
Feb 14 17:10:12 <Herrie|Laptop> Seems we need some cleanup of bits
Feb 14 17:42:03 <Tofe> Herrie: ah, too bad I didn't look at your discussion earlier, I could have pointed to that lxc patch right away
Feb 14 17:42:42 <Tofe> For the last two, we might want to remove the patching
Feb 14 17:43:03 <Tofe> now that we use a Halium system/core, it's not so much ubuntu oriented
Feb 14 18:01:38 <Herrie|Pre3> Tofe: ok will try that. We need the sensorservice from Android in general right?
Feb 14 18:02:03 <Tofe> well, it seems so... it worked well until now, but maybe we were lucky
Feb 14 18:03:08 <Herrie|Pre3> Tofe: Do sensors work? Not 100% sure
Feb 14 18:03:20 <Herrie|Pre3> Recall we had some issues
Feb 14 18:03:50 <Tofe> rotation is a sensor
Feb 14 18:10:13 <Herrie|Pre3> Tofe is armhf = aarch64*n
Feb 14 18:11:09 <Herrie|Pre3> pmos has issues with luna-next on armhf
Feb 14 18:15:15 <nizovn> i know nothing about libhybris, but probably we don't need sensorservice in android, as sensorfw/libhybris uses libhardware.so directly
Feb 14 18:16:49 <Tofe> Herrie: I didn't understand the question... was is a question ?
Feb 14 18:17:37 <Tofe> nizovn: it could be the android side does some initialization, though
Feb 14 18:19:16 <Tofe> Herrie: our builds are all armhf, since the hardware supports it (i.e. there is a FPU). Aarch64 is no exception here.
Feb 14 18:41:58 <Herrie> Tofe: ok
Feb 14 18:52:24 <Herrie> I guess should run test_sensors on other targets
Feb 14 19:04:03 <Herrie> Tofe: Was just wondering if their armhf issue was related to my gstreamer issue somehow
Feb 14 19:04:06 <Herrie> But seems it's not
Feb 14 19:07:05 <Herrie> Tofe: Can you see if test_sensors works for you on N4/N5?
Feb 14 19:07:27 <Herrie> My N5 doesn't work with adb it seems
Feb 14 19:11:03 <Tofe> ah hum I won't be much available tonight :) big diner to prepare and so on
Feb 14 19:12:37 <Herrie> Tofe: No rush
Feb 14 19:13:05 <Herrie> Lazy sushi here, some champagne and other things :)
**** ENDING LOGGING AT Thu Feb 15 03:00:01 2018