**** BEGIN LOGGING AT Tue May 10 02:59:59 2016
May 10 03:59:15 <DougReeder> What copyright notice should we be putting in our source files?
May 10 04:34:45 <Herrie> DougReeder: Whatever you're comfortable with. We usually do Apache 2.0 or GPL v3
May 10 04:35:13 <DougReeder> But who are the apps we write copyright by?
May 10 04:38:09 <DougReeder> Obviously not a huge deal, but I’m having to create some new files for the Enyo 2.7 conversion.
May 10 04:59:04 <Herrie> DougReeder: The person who writes it
May 10 04:59:16 * DougReeder nods
May 10 04:59:26 <Herrie> https://github.com/webOS-ports/org.webosports.app.firstuse/blob/master/qml/FeedsPage.qml#l2
May 10 04:59:31 <Herrie> For example
May 10 04:59:39 <Herrie> We usually add the authors like that
May 10 06:32:44 <Tofe> Morning!
May 10 07:27:31 <Herrie|Pre3> Tofe: morning
May 10 08:14:48 <Herrie|Pre3> JaMa: ping
May 10 09:18:21 <Herrie|Pre3> JaMa: morning
May 10 09:21:40 <JaMa> morning
May 10 09:22:47 <Herrie|Pre3> JaMa: Tofe was able to run some traces on the issue we had with qtwebengine. Seems it's only occurring on ARM targets since qemux86 is fine. So we're guessing it's the cross compiler somehow.
May 10 09:23:26 <JaMa> do you have a backtrace?
May 10 09:23:45 <Herrie|Pre3> http://paste.ubuntu.com/16320677
May 10 09:24:11 <Herrie|Pre3> Tofe said: At frame #2 "other" is a reference to NULL
May 10 09:25:25 <Herrie|Pre3> May 09 16:49:36 <Tofe> and is called because of frame #12
May 10 09:25:46 <Herrie|Pre3> This works OK on qemux86 and was fine with QT 5.5 and Jethro.
May 10 09:26:04 <Herrie|Pre3> Started to act up on QT 5.6 it seems with Krogoth.
May 10 09:26:44 <Herrie|Pre3> Any app that uses a html5 db seems to cause the crash (for example Serene Notes in the Preware feed) but also html5test.com
May 10 09:26:53 <Herrie|Pre3> Both are fine on qemu
May 10 09:26:56 <JaMa> will try to check 5.7 if there are some changes in this area
May 10 09:29:30 <Herrie|Pre3> Since qemu seems fine and Tofe checked some bits we suspect it's the cross compiler somehow.
May 10 09:32:18 <Herrie|Pre3> I didn't really see anything obvious in 5.7 so far tbh
May 10 09:32:23 <Herrie|Pre3> Checked it a few times.
May 10 09:35:43 <Herrie|Pre3> Serene Notes uses IndexedDB
May 10 09:36:04 <Herrie|Pre3> Seems code for it in Chromium hasn't been touched in 2 years.
May 10 09:37:13 <Herrie|Pre3> Ah wait wrong branch :s
May 10 09:53:01 <Herrie|Pre3> Still that shouldn't explain why it works on 1 architecture but not on others. Not sure you have any other platforms you could test on?
May 10 09:53:14 <Herrie|Pre3> Simply loading html5test.com is enough to crash it.
May 10 12:22:08 <Herrie|Pre3> JaMa: You saw the email from Mark O'Donovan on Yocto ML just now? Is that maybe related to our issues with SCP?
May 10 12:27:12 <JaMa> just reading it now
May 10 12:27:34 <JaMa> but do I remember correctly that you had issues only with winscp not with normal scp or pscp from putty?
May 10 12:28:07 <JaMa> maybe winscp does something strange and that garbage makes it fail
May 10 12:28:58 <Herrie|Pre3> JaMa: I only tried WinSCP, Garfonso tried some others. WinSCP always worked before, Krogoth broke it somehow :P
May 10 12:30:49 <JaMa> do you have latest winscp?
May 10 12:31:26 <Herrie|Pre3> Let me update
May 10 12:31:43 <Herrie|Pre3> Fairly recent 5.7.4 but they got 5.7.7 and a 5.8 beta out
May 10 12:32:13 <JaMa> 5.8 supports some new openssh features
May 10 12:32:21 <JaMa> maybe worth trying the beta too
May 10 12:33:13 <Herrie|Pre3> I'll try 5.7.7 first, if that doesn't work 5.8 :)
May 10 12:34:15 <Garfonso> JaMa: pscp from putty works for me.
May 10 12:34:40 <Herrie|Pre3> 5.7.7 no go
May 10 12:35:31 <Herrie|Pre3> 5.8.2 also not. Stuck on "starting the session"
May 10 12:35:45 <Herrie|Pre3> Authenticated OK though :s
May 10 12:37:26 <JaMa> normal scp works for me too
May 10 12:37:59 <JaMa> martin@jama ~ $ scp -P 5522 hw.txt root@localhost:
May 10 12:37:59 <JaMa> The authenticity of host '[localhost]:5522 ([127.0.0.1]:5522)' can't be established.
May 10 12:38:02 <JaMa> ECDSA key fingerprint is SHA256:OOeWKpj4CLpP6qei0I6QGhMU0rs/5wkBF32V2UAAr1g.
May 10 12:38:05 <JaMa> Are you sure you want to continue connecting (yes/no)? yes
May 10 12:38:08 <JaMa> Warning: Permanently added '[localhost]:5522' (ECDSA) to the list of known hosts.
May 10 12:38:11 <JaMa> hw.txt                                                                                                                      100%   72     0.1KB/s   00:00
May 10 12:41:49 <JaMa> and I've tried sound in VirtualBox and I see snd card in alsamixer as pulseaudio, but playing wav with aplay doesn't show any error, but I don't hear it on host
May 10 12:43:22 <JaMa> FWIW I also see some segfaults in dmesg
May 10 12:43:22 <Herrie|Pre3> JaMa: Same here :P
May 10 12:43:23 <JaMa> [    3.967752] mediaindexer[503]: segfault at 24 ip b7092f18 sp bfb51750 error 4 in libglib-2.0.so.0.4600.2[b7061000+123000]
May 10 12:43:26 <JaMa> [    5.187442] LunaAppManager[529]: segfault at 11 ip b67f894e sp bfb54e10 error 4 in libc-2.23.so[b6788000+172000]
May 10 12:43:35 <Herrie|Pre3> JaMa: This is first boot
May 10 12:43:41 <Herrie|Pre3> These should disappear after a reboot
May 10 12:43:57 <Herrie|Pre3> It's because some db8 kinds aren't available yet
May 10 12:44:16 <Herrie|Pre3> Not very graceful though they segfault :P
May 10 12:44:27 <JaMa> Default Sample Specification: s16le 2ch 44100Hz
May 10 12:44:27 <JaMa> Default Channel Map: front-left,front-right
May 10 12:44:27 <JaMa> Default Sink: auto_null
May 10 12:44:27 <JaMa> Default Source: auto_null.monitor
May 10 12:48:15 <JaMa> will try to add kernel-module-snd-intel8x0 which fixed it for me somewhere else
May 10 12:48:24 <Herrie|Pre3> auto_null doesn't sound very good to me, but I'm not familiar with it.
May 10 12:50:37 <JaMa> LunaAppManager segfaults even after reboot
May 10 12:52:57 <Herrie|Pre3> JaMa: Hmmmz... It shouldn't do that. What build you're using?
May 10 12:54:06 <JaMa> adding that module fixes aplay -L to show additional outputs
May 10 12:54:08 <JaMa> and now it works
May 10 12:54:09 <JaMa> root@qemux86:~# aplay -D front  40_Horkyze\ Slize\ -\ Banda\ tupych\ hlav\ \(Bonus\).wav
May 10 12:54:12 <JaMa> Playing WAVE '40_Horkyze Slize - Banda tupych hlav (Bonus).wav' : Signed 16 bit Little Endian, Rate 44100 Hz, Stereo
May 10 12:55:06 <JaMa> pulseaudio needs some configuration to find it as well, but will modify our config to at least install and load kernel-module-snd-intel8x0 by default
May 10 13:00:10 <Herrie|Pre3> :)
May 10 13:02:02 <JaMa> Herrie|Pre3: if I change /etc/default/pulseaudio.conf to use default.pa instead of webos-system.pa and restart pulseaudio.service then it works
May 10 13:02:48 <JaMa> at least with paplay, not through the html5 Testr app
May 10 13:04:20 <JaMa> and after restart it works from Testr as well
May 10 13:04:27 <JaMa> and there is some sound when booting
May 10 13:06:16 <Herrie|Pre3> JaMa: Nice :D
May 10 13:06:43 <Herrie|Pre3> Can you push changes for that?
May 10 13:07:03 <Herrie|Pre3> Will then kick of testing build with latest meta-oe Krogoth updates
May 10 13:10:38 <JaMa> I wonder if webos-system.pa exists only because of this modification:
May 10 13:10:40 <JaMa> ### Automatically load the Pulseaudio Droid
May 10 13:10:40 <JaMa> .ifexists module-droid-card.so
May 10 13:10:40 <JaMa> load-module module-droid-card
May 10 13:10:40 <JaMa> .endif
May 10 13:10:52 <JaMa> the meta-webos-ports commit doesn't say much :/
May 10 13:11:09 <JaMa> will update webos-system.pa with latest system.pa and thiese 4 lines and see what happens
May 10 13:13:39 <JaMa> pushed
May 10 13:15:17 <JaMa> Herrie|Pre3: please stop updating "testing" branch like this
May 10 13:15:37 <JaMa> use update.sh script to include the commit log and update "krogoth" branch and then just rebase testing branch on top of it
May 10 13:15:40 <JaMa> fixing it now
May 10 13:21:27 <Herrie|Pre3> JaMa: Which update.sh?
May 10 13:33:11 <Herrie|Pre3> JaMa: For the webos-system.pa changes: You dropped quite a few load-modules, are these in the standard system.pa already?
May 10 13:33:24 <Herrie|Pre3> Quite sure morphis put them there for a reason?
May 10 13:35:00 <morphis> JaMa, Herrie|Pre3: we need the droid modules in the .pa file and I am not sure what else we need, do you have a diff handy?
May 10 13:35:32 <Herrie|Pre3> morphis: see JaMa's commit in meta-webos-ports/krogoth branch
May 10 13:36:37 <JaMa> Herrie|Pre3: https://github.com/webOS-ports/webos-ports-setup/blob/master/update.sh
May 10 13:37:07 <Herrie|Pre3> JaMa, I run this locally and then push updated krogoth branch?
May 10 13:37:22 <JaMa> https://github.com/webOS-ports/meta-webos-ports/commit/a9c7130ee405060d8bd66b91bb2cd143be27a663
May 10 13:37:33 <Herrie|Pre3> JaMa: Thnx for QA fixes by the way :)
May 10 13:37:41 <Herrie|Pre3> Nice to have the build be
May 10 13:37:47 <Herrie|Pre3> a bit cleaner with less QA
May 10 13:38:12 <morphis> JaMa, Herrie|Pre3: only important one I see lost are the bluez ones
May 10 13:38:43 <JaMa> Herrie|Pre3: yes, you might need to modify update.sh to right location of your checkouts (cd ~/$DIR)
May 10 13:40:23 <JaMa> Herrie|Pre3: or add parameter for it if you want and you might want to run "git remote update" in each layer, before first use (I don't need it, because I always update the layers first for other projects and then just run update.sh here to use the latest I already have and tested somewhere else)
May 10 13:42:04 <JaMa> comparing ./src/daemon/system.pa.in in upstream pulseaudio shows only one small additon of module-coreaudio-detect since pulse 5.0
May 10 13:42:12 <Herrie|Pre3> JaMa: OK will look a bit, still learning the ropes and new things every day :)
May 10 13:42:30 <JaMa> and module-dbus-protocol since 4.0
May 10 13:42:41 <JaMa> not sure what was the base for our webos-system.pa
May 10 13:49:34 <Herrie|Pre3> JaMa: I'm sure it's somewhere in the back of morphis' mind :P
May 10 14:01:40 <Herrie|Pre3> JaMa: Any clues for the qtwebengine crash?
May 10 14:01:56 <JaMa> nope :/
May 10 14:01:59 <Herrie|Pre3> Really strange it works on qemu but crashes on Maguro/Mako/T=
May 10 14:02:01 <Herrie|Pre3> TP
May 10 14:02:12 <JaMa> but I wouldn't bet on cross compiler yet
May 10 14:02:51 <JaMa> there is a lot of arm specific code in qtwebengine, so it might fail just because the values passed down the backtrace are different
May 10 14:04:21 <Herrie|Pre3> If you have any specific debugging suggestions, I'm sure Tofe is happy to try :)
May 10 14:07:41 <JaMa> it would be interesting to add breakpoint on storage/browser/quota/client_usage_tracker.cc:351 and compare the values passed in qemux86 and maguro/mako/tp
May 10 14:08:31 <JaMa> if it's NULL only for the later, then the issue is somewhere above
May 10 14:09:57 <JaMa> but there is also check in void ClientUsageTracker::AccumulateOriginUsage which would prevent passing NULL origin down
May 10 14:11:22 <JaMa> only changes I've seen in this area for 5.7 is switch from int64 to int64_t
May 10 14:11:59 <JaMa> and new method ClientUsageTracker::GetCachedOriginsUsage
May 10 14:13:33 <Tofe> JaMa: on mako, origin isn't NULL in ClientUsageTracker
May 10 14:13:47 <Tofe> It's only NULL in GURL's copy constructor
May 10 14:15:29 <Tofe> I'm a bit sad that my backport didn't work, because we're really really close to https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69841 use case and symptoms
May 10 14:16:01 <Tofe> That being said, the patch doesn't *fix* the issue itself, it just makes it disappear for their use case
May 10 14:25:06 <JaMa> Tofe: maybe we should try to rewrite "int64* usage = &cached_usage_by_host_[host][origin];" in 2 steps like ClientUsageTracker::GetCachedHostUsage is doing to see where exactly it fails?
May 10 14:25:50 <JaMa> maybe it wouldn't trigger this whole issue and we would be fine :)
May 10 14:27:00 <Tofe> JaMa: I've done that already; the problem occurs on the second map index: "host" is already in cached_usage_by_host_, but it's just an empty map, meaning that "origin" isn't in that map and a new item need to be inserted, thus the GURL constructor. I've separated the two steps, and it crashes on the second one, as guessed.
May 10 14:27:58 <Tofe> I will try to separate things even more, to avoid using [] so directly to insert an item. But I fear it won't help much.
May 10 14:28:49 <JaMa> alternatively can we temporary disable caching for this?
May 10 14:29:34 <Tofe> Eventually. i'll have to find out what option triggers it.
May 10 14:42:34 <JaMa> finally got fix for random failure from phonesim
May 10 14:42:58 <Herrie|Pre3> JaMa: Random failure while building?
May 10 14:45:32 <JaMa> yes
May 10 14:45:45 <JaMa> somehow I was able to trigger it quite often in local builds
May 10 15:19:03 <Herrie|Pre3> JaMa: OK :)
May 10 16:56:58 * DougReederG3 checks in
May 10 17:26:58 <Tofe> Herrie: 495 at html5test !
May 10 17:31:54 <DougReederG3> That's better than FF on my Android
May 10 17:36:03 <GodGinrai> morning EricBlade
May 10 17:41:28 <Tofe> SereneNotes now seems to work
May 10 17:41:40 <Tofe> It crashed at startup? Or adding a new note?
May 10 17:45:51 <Tofe> Herrie, JaMa: here is what I could do about it: https://github.com/webOS-ports/qtwebengine-chromium/pull/10
May 10 17:46:10 <Tofe> It's no more than a weak workaround, but well, it works
May 10 17:47:26 <JaMa> Tofe: looks good to me, lets watch that gcc ticket or retry with 5.3.1 when it's in oe-core (if it gets upgraded before gcc 6.1 is added)
May 10 17:47:50 <JaMa> thanks for spending the time on this (I know it couldn't be quick because of how long webengine builds)
May 10 17:48:08 <Herrie> Tofe: It would crash right on load
May 10 17:48:12 <Herrie> html5test.com also OK?
May 10 17:48:17 <Tofe> Herrie: both ok
May 10 17:48:23 <Herrie> Tofe: I see now!
May 10 17:49:01 <Herrie> Tofe: Great work
May 10 17:49:02 <Tofe> JaMa: oh, always a pleasure :p And with my new machine, updating my webengine build is now a matter of minutes
May 10 17:49:06 <Herrie> I'm OK with this workaround
May 10 17:49:15 <Herrie> Tofe: Minutes to compile webengine?
May 10 17:49:33 <Tofe> Herrie: to update an already built package
May 10 17:50:19 <Herrie> Tofe: Ah :P
May 10 17:50:20 <Tofe> I've got enough RAM for the whole linking to be done in one step, and my SSD handles the huge file copy just fine
May 10 17:50:26 <Herrie> :D
May 10 17:50:40 <Herrie> How much RAM you have? I think my build machine will be limited to 8
May 10 17:50:53 <Tofe> 16GB, can be later expanded up to 64
May 10 17:51:23 <Herrie> Tofe: How long would building luna-webappmgr incl. qtwebengine take for you?
May 10 17:51:36 <Herrie> For qemux86 that is :P
May 10 17:51:59 <Tofe> Herrie: I'd have to build the rest of the image :) Which would mean about 2h I guess
May 10 17:52:10 <Tofe> or maybe a little bit more, iirc
May 10 17:52:15 <Herrie> Ah ok
May 10 17:52:19 <Herrie> I'll try at my end
May 10 17:52:41 <Herrie> Got a hunch for the HTML5 File API problem DougReeder reported, but need to test it
May 10 17:52:58 <Tofe> you need to rebuild qtwebengine for that?
May 10 17:53:44 <Herrie> No, but not sure I did an initial qemux86 webengine build here :P
May 10 17:54:23 <Herrie> Anyway it's building in my vbox
May 10 17:54:29 <Herrie> Shouldn't be too long
May 10 17:54:54 <Herrie> Tofe: I assume no issues are to be expected with your PR for qemux86?
May 10 17:55:19 <Tofe> No, it's completely generic
May 10 17:55:24 <Herrie> :D
May 10 17:55:45 <Tofe> and it does quite the same thing as the previously, in addition
May 10 17:55:55 <Tofe> previous code*
May 10 17:57:04 <Tofe> The only detail is that my gcc is a patched one, with the little fix I backported
May 10 17:57:12 <Tofe> but which was useless in our case
May 10 17:57:25 <Tofe> so if we see it still crashes, we'll add that fix too
May 10 17:58:51 <Herrie> LOL
May 10 17:58:52 <Herrie> OK
May 10 17:58:57 <Herrie> Well I sneaked it in just now
May 10 17:59:04 <Herrie> It'll start building Maguro soon ;)
May 10 17:59:07 <Herrie> So I can test :P
May 10 17:59:12 <Tofe> ok great :)
May 10 17:59:39 <Herrie> At least I think it picks up changes for a new device that's in the same build run :P
May 10 18:00:06 <Herrie> First batch does Mako, Tenderloin & RPi2, then it syncs and does Maguro, qemux86 & Grouper
May 10 18:01:04 <nizovn> Tofe: strange that there was no crash on pri2, but nice it now works!
May 10 18:01:28 <Tofe> nizovn: it's the same architecture?
May 10 18:01:52 <nizovn> afaik, it's armv7
May 10 18:01:55 <Herrie> nizovn: You got a RPi2 ?
May 10 18:02:07 <Tofe> nizovn: with Neon ?
May 10 18:02:17 <Herrie> nizovn: It's slightly different I think
May 10 18:02:23 <Herrie> It's armv7ve or something
May 10 18:02:24 <nizovn> Herrie: no, i remember Andolamin was testing it
May 10 18:02:57 <Herrie> nizovn: Those were older builds ;)
May 10 18:02:59 <Tofe> nizovn: I think we really did hit a corner use case of gcc, no luck here
May 10 18:03:02 <Herrie> New builds fail :S
May 10 18:03:31 <Herrie> Gstreamer it seems: http://jenkins.nas-admin.org/job/luneos-testing_raspberrypi2/lastBuild/consoleFull
May 10 18:04:54 <Herrie> Andolamin is checking it already
May 10 18:05:26 <Tofe> "libgstgl-1.0.so: error: undefined reference to 'vc_dispmanx_update_start'" ok, maybe GL thingy
May 10 18:06:09 <Tofe> We updated gstreamer recently?
May 10 18:06:22 <Tofe> Ah, maybe the switch to krogoth?
May 10 18:06:27 <Herrie> Yeah could be
May 10 18:12:16 <Tofe> Ok well, I earned a beer.
May 10 19:17:14 <Herrie> Tofe: You do, but don't you prefer wine :P ?
May 10 19:17:23 <Herrie> Or even some champagne for the occasion :p
May 10 19:19:20 <JaMa> Herrie: we should lock meta-raspberrypi revision in conf/layers.txt so that new issues don't leak from meta-raspberrypi upstream
May 10 19:19:32 <JaMa> Herrie: especially when we're tracking master branch from krogoth builds
May 10 19:19:43 <JaMa> Herrie: I've locked it in stable branch already
May 10 19:46:45 <Herrie> JaMa: OK that's a good idea
May 10 20:23:14 <Herrie> elvispre: Didn't have much time today, only was able to conclude it should be oFono, not connMan
May 10 20:23:19 <Herrie> Hopefully some more time tomorrow
May 10 20:31:36 <elvispre> Herrie: No problem.
May 10 20:39:09 <Tofe> Herrie: I generally prefer wine, but tonight I didn't want to drink a whole bottle :p
May 10 20:41:39 <scoutcamper> Tofe, forget beer and wine. Shots!
May 10 20:41:47 <scoutcamper> (says the college student)
May 10 20:43:25 <Tofe> scoutcamper: well, I should have, because I've watched eurovision song contest, and I was definitely too sober for that
**** ENDING LOGGING AT Wed May 11 02:59:58 2016