**** BEGIN LOGGING AT Thu Nov 28 02:59:58 2019 Nov 28 06:12:43 Morning Nov 28 07:52:28 Morning! Nov 28 07:53:46 JaMa: damn, I didn't know yocto already did this ! I have my own scripts that do that :p Nov 28 08:06:24 Herrie|2: interesting. My first guess would be that GetWLSurface() returns NULL in https://github.com/webosose/chromium68/blob/master/src/ozone/wayland/shell/wl_shell_surface.cc#L32 Nov 28 08:07:54 Note that it could possibly be the fault of my quick compositor implementation Nov 28 08:08:23 at this point, it might be worth it to investigate using SAM with our cardshell Nov 28 08:09:15 Ah I forgot, they use qtwayland 5.6... damn... Nov 28 08:11:49 Tofe: Yeah they do it seems Nov 28 08:14:42 and it's luna-surfacemanager I'm talking about, not SAM, I always mix them up Nov 28 08:16:57 Yeah luna-surfacemanager is their compositor Nov 28 08:17:02 SAM should be luna-sysmgr replacement mainly Nov 28 08:39:22 Tofe: Should I simply put some logging around that to see what's in there? Nov 28 08:41:39 Herrie: first you can insert a "DCHECK(GetWLSurface());" before the call, to be sure Nov 28 08:42:09 you can also activate the client wayland trace (export WAYLAND_DEBUG=client) Nov 28 08:42:41 in /etc/luna-next/environment.conf, you can also activate WAYLAND_DEBUG=server -- but that's quite verbose Nov 28 08:53:01 Tofe: The WL ones from journalctl: https://paste.ubuntu.com/p/DZWf2Gd9xZ/ Nov 28 09:02:49 it looks ok so far (some wl_surface are created by the compositor) Nov 28 09:03:10 but it could also be that these wl_surface are created too late somehow Nov 28 09:11:52 Tofe: I'll add the DCHECK shortly. AGL has some commits that might be of help? https://github.com/webosose/chromium68/commits/%4039.agl.guppy Nov 28 09:16:56 most of these are already in main branch, aren't they? Nov 28 09:25:03 Tofe: Not really it seems :S https://github.com/webosose/chromium68/compare/@39.agl.guppy Nov 28 09:36:10 Though from an architecture perspective we're a lot closer to OSE v.s. AGL. Since they only use WAM & Webruntime from OSE, nothing else. So we should need a lot less of their custom AGL specific stuff Nov 28 09:54:17 Herrie: correct; also they use ivi_shell, which is neither what OSE uses nor what we use Nov 28 09:55:01 "ivi" is roughly a "one fullscreen app at a time" kind of shell Nov 28 09:57:14 Yes ivi is for cars Nov 28 09:57:19 So that makes sense Nov 28 10:01:55 Tofe: Where DCHECK should output it's results? Nov 28 10:02:19 ewww... I guess it should crash if it's NULL Nov 28 10:06:47 Well by BT: https://paste.ubuntu.com/p/jG2426h4cN/ Nov 28 10:07:08 And journalctl: https://paste.ubuntu.com/p/shWYnrNmyY/ Nov 28 10:13:32 that just means GetWLSurface() does return a valid pointer, and the issue is elsewhere Nov 28 10:22:14 at this piont, I'm not sure what would be the cause... Nov 28 11:29:19 Tofe: Adding some logging around the various lines to see if it gives any clues Nov 28 11:40:54 Tofe: Issues seems actually similar to https://bugzilla.redhat.com/show_bug.cgi?id=1723017 Nov 28 11:41:02 I might just want to update my build to Qt 5.13;) Nov 28 11:41:07 Since it's still 5.12.x based Nov 28 11:43:03 It's worth a try Nov 28 12:05:13 OK it's building Nov 28 13:07:38 Tofe: Your attempts were with 5.13 already right? Nov 28 14:03:48 yes Nov 28 14:04:13 Ah Ok that might explain my issue with 5.12.x then ;) Nov 28 14:04:37 Based on above bugzilla it seems they had issues with 5.12.4 and not with 5.13 ;) Nov 28 14:04:47 Just my QtBase was compiling for over 1 hr, so I restarted it now Nov 28 14:04:51 Shouldn't be that long ;) Nov 28 14:04:56 right; but their issue might be a different one too, we'll see Nov 28 14:05:16 Well the error was exactly the same ;) Nov 28 14:53:03 Hmmz qtwayland patches are failing, but I'm not seeing why :S Nov 28 14:55:44 all your meta* are up to date for zeus ? Nov 28 14:55:56 looks like a mistmatch between meta-qt5 and meta-wop Nov 28 15:02:30 Tofe: Yeah should be :S Nov 28 15:19:40 Seems my meta are up to date Nov 28 15:19:44 Not sure what's up here Nov 28 15:22:58 Tofe: https://paste.ubuntu.com/p/XHfgV8xT7q/ Nov 28 15:23:24 615aa20 for qtwayland seems to suggest: https://github.com/qt/qtwayland/commit/615aa208d131ab99e967725504fcb16fdda4ea83 Nov 28 15:25:52 https://github.com/webOS-ports/meta-webos-ports/blob/zeus/meta-luneui/recipes-qt/qt5/qtwayland/0004-QWaylandXdgSurface-handle-ExtendedSurface-window-pro.patch#L38 and https://github.com/qt/qtwayland/blob/5.13.2/src/plugins/shellintegration/xdg-shell/qwaylandxdgshell.cpp#L264 Nov 28 16:04:33 I'll check what my layers are at home in a bit Nov 28 17:17:24 I'll start with fresh clone just in case Nov 28 18:35:11 Herrie: still interested with my values ?... Nov 28 18:43:36 Tofe: sure Nov 28 18:43:43 For reference never hurts Nov 28 18:45:19 /media/LuneOS/testing/webos-ports/meta-qt5 432ad2a (HEAD -> zeus, origin/zeus-next, origin/zeus, origin/master-next, origin/master, origin/jansa/zeus, origin/jansa/master, origin/HEAD) qttools: don't install tst_qtattributionsscanner Nov 28 18:45:39 /media/LuneOS/testing/webos-ports/meta-webos-ports 925189840 (HEAD -> zeus, tofe/zeus) luneos-dev-image: remove gtk+3-demo Nov 28 18:47:53 Tofe: it's now building qtwebengine Nov 28 18:48:04 So it must have been some artifacts Nov 28 18:48:31 ok, well, good Nov 28 18:50:06 So should have a new image soon hopefully Nov 28 18:50:26 I pushed all changes you had to various places so they're in the image Nov 28 18:51:29 ah, thanks, I was too lazy :p Nov 28 18:53:50 I should make the Chromium one a patch instead of forking, but that's easy enough Nov 28 18:54:15 Once stuff is working anyway would need to clean up things Nov 28 18:57:11 And hopefully the Chromium 72 bits get fixed Nov 28 19:22:25 Tofe: It does seem to be a timing issue somehow as well Nov 28 19:22:37 When I run the .sh multiple times I get differnet results it seems Nov 28 19:22:51 Installing the .dbg again to see if I can bt Nov 28 19:22:59 mmh ok Nov 28 19:23:45 See https://paste.ubuntu.com/p/bYVVH9d7TY/ for what I mean Nov 28 19:27:34 well it's multi-threaded and multi-process, so why not, after all Nov 28 19:29:28 Yeah Nov 28 19:29:40 Let me try to build without neva-media like lokesh suggested Nov 28 19:29:42 JaMa: FYI, here are my scripts helpers for installing debugfs : https://paste.ubuntu.com/p/qPYyQkF9zG/ Nov 28 19:31:45 Seems I forgot that one ;) **** ENDING LOGGING AT Fri Nov 29 02:59:58 2019