**** BEGIN LOGGING AT Mon Apr 25 02:59:56 2022 Apr 25 07:36:29 Morning! Apr 25 07:38:23 Morning! Apr 25 07:45:19 Tofe: But I guess working USB doesn't help much anyway ;) Apr 25 07:45:27 But at least good it's not rebooting anymore Apr 25 11:26:09 Tofe: I reverted the cardshell commit like you suggested on qemu now and I'm getting some clues in logs by the looks of it Apr 25 11:27:03 It might be com.webos.app.notification ? Apr 25 11:27:50 ah, I always had issues with that one Apr 25 11:28:21 See last lines in log: https://pastebin.ubuntu.com/p/FjqyTg2Rms/ Apr 25 11:28:32 I can confirm app menu & rotation work for me now Apr 25 11:30:03 I don't get VKB though still Apr 25 11:30:52 And HW KBD also doesn't work it seems in qemu Apr 25 11:31:17 Some logs: https://pastebin.ubuntu.com/p/3bzDMHtJv9/ Apr 25 11:32:35 Seems some permission errors for org.maliit somehow Apr 25 11:32:58 Apr 25 12:23:44 qemux86-64 ls-hubd[288]: [] [pmlog] ls-hubd LSHUB_NO_NAME_PERMS {"APP_ID":"org.maliit","ROLE_ID":"/usr/sbin/MaliitServer","EXE":"/usr/sbin/MaliitServer"} Executable: "/usr/sbin/MaliitServer" (cmdline: /usr/sbin/MaliitServer -instance 0) does not have permission to register name: "org.maliit" Apr 25 11:33:04 Apr 25 12:23:44 qemux86-64 MaliitServer[816]: [] [pmlog] LS_REQ_NAME {"FILE":"transport.c","LINE":1931} Invalid permissions for org.maliit Apr 25 11:33:04 Apr 25 12:23:44 qemux86-64 sh[816]: WARNING: Failed to register service handle: Invalid permissions for org.maliit Apr 25 11:33:38 If you want I can share the image Apr 25 11:33:53 But it should be the same as yesterday's one with the QML change just Apr 25 11:37:18 Seems we also need to call something in SAM to get some DB kinds registered: Apr 25 12:23:40 qemux86-64 mojodb-luna[590]: [] [pmlog] DB8 MOJ_SERVICE_WARNING {"sender":"com.webos.applicationManager","method":"find","payload":{"query":{"from":"com.webos.applicationManager.launchpoints:2","orderBy":"_rev"}},"error":"kind not registered: 'com.webos.applicationManager.launchpoints:2'","reqErr":-3970} Apr 25 11:37:38 Which are defined in a conf file here: https://github.com/webosose/sam/blob/af986e600cba012d9ad67a50222174725fb242c3/files/conf/sam-conf.json.in Apr 25 11:37:57 And some methods in here: https://github.com/webosose/sam/blob/af986e600cba012d9ad67a50222174725fb242c3/src/conf/SAMConf.cpp Apr 25 11:40:03 Hmmz the file is there, but I see this in logs: Apr 25 12:23:38 qemux86-64 sam[687]: [W][Settings][loadBlockedList][/etc/palm/sam-conf.json] Failed to parse blocked-file sam-conf Apr 25 11:41:30 oops was away for some minutes, let me catch up :p Apr 25 11:43:05 so for MaliitServer, you now have the manifest and so on ? Apr 25 11:51:32 so your SAM error is here: https://github.com/webosose/sam/blob/master/src/conf/SAMConf.cpp#L104= Apr 25 11:52:11 doesn't seem fatal or problematic, I'd say Apr 25 12:35:53 Tofe: I have: https://bpa.st/WRKQ which seems 1 file too many in services.d somehow Apr 25 12:36:49 Tofe: No SAM error shouldn't be fatal, but we're missing the DB kind, which might not be fatal for now Apr 25 12:36:51 right, I'd say "./usr/share/luna-service2/services.d/com.webos.service.ime.service" shouldn't be there. Can you show me the manifest content? Apr 25 12:37:10 We might want to use that kind going forward instead of the legacy one Apr 25 12:37:13 But we'd have to see Apr 25 12:37:31 https://bpa.st/OMYQ Apr 25 12:38:04 ok, so I was wrong, it's "./usr/share/luna-service2/services.d/com.webos.service.ime.role.json" that is superfluous Apr 25 12:38:33 Yes that one Apr 25 12:38:37 I just was about to paste that Apr 25 12:39:03 however, it shouldn't trigger any error I think Apr 25 12:43:01 No I guess so too Apr 25 12:44:52 I might simply need to add org.maliit to com.webos.service.ime.role.json ? Apr 25 12:44:58 Under allowedNames Apr 25 12:45:43 And seems that we need to fix some appinfo.json for our QML Settings app, I'll fix those Apr 25 12:45:54 They fail validation, hence don't show up in the launcher Apr 25 13:13:02 but, why did it work before ? it's what confuses me Apr 25 13:14:59 Tofe: Well SAM seems to do different validation of appinfo.json Apr 25 13:15:18 Seems its expecting vendorurl instead of vendor_url Apr 25 13:15:51 And it might not like our "useLuneOSStyle" or "visible": "true" Apr 25 13:15:57 Will test a few things Apr 25 13:16:18 no I meant for maliit Apr 25 13:16:34 for settings, it's probably visible: "true" instead of visible: true Apr 25 13:17:07 Tofe: Well Maliit worked for you, not for me ;) Apr 25 13:17:18 I posted here a link to a json validator with the appinfo schema, not sure how to find it again Apr 25 13:19:42 Yeah probably it's the visible one Apr 25 13:19:47 Seeing it's expecting a boolean Apr 25 13:19:57 https://github.com/webOS-ports/sam/blob/webOS-ports/webOS-OSE/files/schema/ApplicationDescription.schema#L96 Apr 25 13:20:34 Yup that fixes it, but UI doesn't look good Apr 25 13:21:48 Let me PR this fixes at least Apr 25 13:28:46 https://github.com/webOS-ports/org.webosports.app.settings/pull/110 Apr 25 13:34:00 Tofe: For validation, this one I guess? https://www.jsonschemavalidator.net/ Apr 25 13:38:56 ah yes, correct Apr 25 13:39:08 but in the URI I gave, the schema on the left was already filled in Apr 25 13:40:23 Tofe: Well it cached it at my end ;) Apr 25 13:41:01 Anyway seems it's sorted now for these apps Apr 25 13:41:10 There are other issues with qml-runner but those seem to be in other apps too Apr 25 13:45:02 ok good Apr 25 13:45:39 merged Apr 25 14:00:18 I do seem to get window decoration (-X etc) for the Settings QML apps somehow Apr 25 14:03:15 ah ! it's because it doesn't use WebOSWindow Apr 25 14:03:43 it's quite easy to fix: "import Eos.Window 0.1" then replace "LuneOSWindow" with "WebOSWindow" Apr 25 14:04:22 I don't think we use any weird QML property from LuneOSWindow, so it should work as-is Apr 25 14:05:41 Tofe: Ah OK I'll try that Apr 25 14:08:10 I have a "ApplicationWindow" in GenericCategoryWindow.qml, no LuneOSWindow Apr 25 14:10:12 mmh put WebOSWindow instead anyway Apr 25 14:10:19 weird that we used that Apr 25 14:14:00 Downloading my new image and will test it Apr 25 14:23:25 Well: file:///usr/palm/applications/org.webosports.app.settings.wifi/qml/GenericCategoryWindow.qml:140 Cannot assign to non-existent property "background" Apr 25 14:23:31 So now it doesn't launch at all ;) Apr 25 14:27:28 When I delete that one I get another one for categoryheader Apr 25 14:27:38 file:///usr/palm/applications/org.webosports.app.settings.wifi/qml/GenericCategoryWindow.qml:131 Cannot assign to non-existent property "header" Apr 25 14:44:55 ah, it's a property of ApplicationWindow which doesn't exist in WebOSWindow... I'll do the rest of the migration then Apr 25 14:47:43 It should be low hanging fruit in general Apr 25 14:57:14 yes, it's pretty easy QML stuff, and if you're tempted you can give it a try Apr 25 14:57:26 otherwise I'll handle it tonight Apr 25 15:00:50 I was trying to tackle a few other small things here and there now Apr 25 15:03:10 Seems Tenderloin doesn't boot indeed as per report Apr 25 15:03:13 Trying to get some logs now Apr 25 15:08:31 https://bpa.st/TWSQ Apr 25 15:10:47 "/scripts/local-bottom/K99-move-boot.sh: line 6: echo: write error: Invalid argument" not sure we had that one before Apr 25 15:12:05 Well that log doesn't look too shocking, this one shows more issues... Apr 25 15:12:14 "Checking was requested for "rootfs", but it is not a device" I wonder what systemd thinks is "rootfs" actually Apr 25 15:12:26 https://pastebin.ubuntu.com/p/rBV8fQkR8M/ Apr 25 15:17:11 That one does seem to ring a bell Apr 25 15:17:16 And I thought it was fixed :S Apr 25 15:18:34 yes, I patched the kernel to avoid this NULL pointer Apr 25 15:18:49 Let me check again Apr 25 15:18:53 Maybe using old ref somehow for kernel Apr 25 15:19:16 https://github.com/shr-distribution/meta-smartphone/commit/ca48ba54c06e12511e88730a4910017cf667e822 Apr 25 15:20:08 Seems to be OK in my recipe... I will bump to https://github.com/shr-distribution/linux/commit/9ef22c826b8b72f7caa4358f32e6b2356679439f just to be sure Apr 25 15:22:44 you updated the uImage, right ? Apr 25 15:24:53 Yeah well I should have ;) Apr 25 15:24:59 I'll do again just to be sure Apr 25 15:29:33 OK now flashed a new uImage Apr 25 15:29:42 Time to be patient, Tenderloin is slow to boot anyway Apr 25 15:30:21 yeah, sorry, it's partially my fault I never pushed the required change Apr 25 15:33:42 What do you mean? Apr 25 15:34:33 Well I have a more proper log now, but nothing on screen ;) Apr 25 15:34:50 https://pastebin.ubuntu.com/p/fs4XtxNxD8/ Apr 25 15:34:53 Let me check a bit Apr 25 15:35:12 I remember you said you needed to fix some udev for display somehow? Apr 25 15:36:14 well surface-manager.sh is mainly working well with DRM devices Apr 25 15:36:41 that was also a reason why we switched our display driver on VirtualBox Apr 25 15:37:11 Tofe: Well on PP it worked Apr 25 15:37:16 Does it make sense to test Mido? Apr 25 15:37:20 PP(P) are Mesa based Apr 25 15:37:25 Just as a comparison to see what it does? Apr 25 15:37:27 and probably use DRM Apr 25 15:37:38 yes, it makes sense Apr 25 15:37:43 OK let me build that Apr 25 15:37:51 Tissot is still with my son, so cannot test that Apr 25 15:38:07 no pb, I'll also test on hammerhead at some point Apr 25 15:38:28 but we'll probably need to adapt a bit the shell script that does the detection Apr 25 15:38:36 OK Apr 25 16:10:58 looking forward to trying it out when its ready! Apr 25 16:13:13 me too ! Apr 25 16:31:39 codepoet: Well qemux86-64 is starting to look good also for Enyo 1 and 2 apps Apr 25 16:31:45 We'll give you a shout when it's ready ;) Apr 25 16:31:53 Still issues to iron out Apr 25 18:05:02 back home, time to flash my PP Apr 25 18:08:16 Herrie: here is what we should revert, but only for touchpad: https://github.com/openembedded/openembedded-core/commit/e9715d4234eb7b45dee8b323799014646f0a1b07 Apr 25 18:08:48 it also means the recipe will become MARCHINE_ARCH, so we'll have to reclare it ABI safe, like JaMa said some time ago Apr 25 18:09:01 That should solve the long boot time Apr 25 18:12:26 Herrie: same status as PPP, same bug Apr 25 18:17:35 Tofe: OK well that's good then at least, consistency ;) Apr 25 18:18:00 yes Apr 25 18:18:38 touchscreen doesn't seem to work, for me Apr 25 18:19:01 Tofe: Well touch work for me in gneeral Apr 25 18:19:14 Just the issue with FirstUse mode Apr 25 18:20:32 Let's see where my mido image is at Apr 25 18:20:38 It was busy with webruntime Apr 25 18:20:44 ah, yes, it actually works Apr 25 18:20:50 I just can't interact with the app Apr 25 18:21:08 Tofe: Yeah I had the same Apr 25 18:21:21 Mido is ready, I put it to charge, so I should be able to flash soon Apr 25 18:23:46 web apps seem fine, execpt incorrect DPI Apr 25 18:24:30 Does your VKB work on PP? Apr 25 18:24:54 nope Apr 25 18:27:18 Yeah in general the apps side looks pretty OK Apr 25 18:27:33 Biggest issue is some of the QML apps for now and the notifications that aren't working yet Apr 25 18:27:49 Dashboards work somehow, but regular notifications in Testr don't Apr 25 18:30:59 Seems the charging cable wasn't good, finding another one Apr 25 18:32:23 I managed to start maliit with just adding the manifest file Apr 25 18:32:35 Weird, because I have the manifest in mine Apr 25 18:32:57 Herrie: I'm not using the very latest honister here Apr 25 18:33:05 missing your last commit for maliit for instance Apr 25 18:33:36 Tofe: Ah Apr 25 18:33:38 OK Apr 25 18:34:03 still, it looks like LSM isn't aware there is a maliit in there Apr 25 18:38:10 Maybe a reboot needed? Apr 25 18:39:49 let's try Apr 25 18:42:25 nope Apr 25 18:42:27 "[pmlog] surface-manager LSM {} WaylandInputMethod* WaylandTextModelFactory::findInputMethod(int), Trying to findInputMethod 0" Apr 25 18:43:12 oh, "surface-manager LSM {} void WebOSSurfaceItem::setWindowProperty(const QString&, const QVariant&), WebOSSurfaceItem(0x2e16250, name="surfaceItem__WEBOS_WINDOW_TYPE_KEYBOARD", parent=0x0, geometry=0,0 -1x-1) m_shellSurface not available, use surface() instead" Apr 25 18:45:03 Well that's a clue at least Apr 25 19:29:22 This will help the accounts app to add IM accounts at least: https://github.com/webOS-ports/messaging-accounts/pull/13 Apr 25 19:29:29 C+DAV is another issue by the looks of it ;) Apr 25 19:41:31 Tofe: FWIW on Mido also no /sys/class/drm Apr 25 19:43:48 mido log for reference: https://pastebin.ubuntu.com/p/qqm252NWhp/ Apr 25 20:15:27 Seems we might need some of the CONFIG_DRM_MSM defconfig flags Apr 25 20:21:26 I foudn some other Xiaomi device with 4.9 kernel with soem of these options enabled, let me try them on Mido and see what happens Apr 25 20:27:21 not sure enabling DRM_MSM would be a good idea Apr 25 20:27:32 if could very well break the whole display Apr 25 20:27:59 Tofe: OK... well let's see Apr 25 20:29:02 Worth a try at least Apr 25 20:29:10 It seems enabled in other kernels, so it might work Apr 25 20:29:23 but we don't really need DRM I think; it's just surface-manager.sh that uses it to auto-configure LSM Apr 25 20:29:52 we can probably hard-code the config and skip the DRM part Apr 25 20:30:08 Tofe: Yeah probably Apr 25 20:31:50 the first thing that is probed is configd; it LSM can retrieve the display config from there, then it won't even try to look for DRM infos Apr 25 20:31:57 if* Apr 25 20:32:30 luna-send -n 1 -a com.webos.surfacemanager luna://com.webos.service.config/getConfigs '{"configNames":["com.webos.surfacemanager.displayConfig"]}' <-- this is the call Apr 25 20:34:26 Yeah I saw that one Apr 25 20:34:33 Not sure why we don't get anything back there Apr 25 20:35:57 I have ./etc/configd/layers/base/com.webos.surfacemanager.json Apr 25 20:36:01 With a correct value Apr 25 20:36:11 maybe it's not read, for some reason ? Apr 25 20:39:18 There are some funny configd warnings in logs Apr 25 20:39:40 Nov 19 17:19:41 mido configd[1736]: 1736 (configd) PBNJSON src/pbnjson_c/jobject.c:646 :: Attempt to retrieve size from something not an object 0x7faed94428 Apr 25 20:43:55 Herrie: actually I think our configd-data is incomplete or so Apr 25 20:45:12 but I don't really know how these layers and stuff work Apr 25 20:45:47 for instance, we want "com.webos.surfacemanager.displayConfig" here. The json is called "com.webos.surfacemanager.json", fine. But what about the displayConfig sub-selector ? Apr 25 20:47:02 Tofe: Well we have to go by what LG provided but it can be incomplete of course Apr 25 20:47:08 I would not be surprised Apr 25 20:49:50 what we provided was "com.webos.surfacemanager.compositorGeometry" actually Apr 25 20:50:42 so we'll want to rename that json key to displayConfig, or duplicate it, something like that Apr 25 20:51:28 # luna-send -n 1 -a com.webos.surfacemanager luna://com.webos.service.config/getConfigs '{"configNames":["com.webos.surfacemanager.compositorGeometry"]}' Apr 25 20:51:28 {"subscribed":false,"configs":{"com.webos.surfacemanager.compositorGeometry":"720x1440+0+0r0s1"},"returnValue":true} Apr 25 20:52:36 it would be nice to try that simple change on mido or tenderloin Apr 25 20:52:54 It could be the values in compositorGeometry contain all the required values Apr 25 20:53:12 Not sure what those values after the resolution mean to be honest Apr 25 20:53:45 ah, it's a bit like the X11 geometry stuff Apr 25 20:54:38 width x height + posX + posY, then some homemade stuff, probably the rotation and screen Apr 25 20:54:46 no, scale Apr 25 20:55:03 ah OK Apr 25 20:55:24 Not "screen"? I thought it was used for that anyway Apr 25 20:55:30 scale makes sense Apr 25 20:56:26 I'm not sure :) Apr 25 21:03:18 anyway I'll stop here for tonight Apr 25 21:04:46 Same here, another day tomorrow Apr 25 21:04:53 Thnx for pointers Apr 25 22:32:48 Tofe: Well still had another look Apr 25 22:32:57 Seems that surface-manager doesn't like stuff Apr 25 22:33:21 However if I start to toy around with EGL_PLATFORM=hwcomposer and QT_QPA_PLATFORM=hwcomposer stuff starts to work Apr 25 22:33:38 I.e. i get stuff on screen Apr 25 22:33:40 And rotation works Apr 25 22:33:53 Not sure if that's what we need in the end, but it gives me UI on mido Apr 25 22:34:09 Not the correct resolution because I manually kick it off now with gdb Apr 25 23:02:44 Seems that if I start with hwcomposer just it works Apr 25 23:03:59 Basically setting https://github.com/webosose/luna-surfacemanager/blob/master/base/startup/product.env.in#L31 to hwcomposer instead of eglfs and removing L45-145 Apr 25 23:04:07 Touch doesn't work though then Apr 25 23:04:12 But UI is fine on Mido Apr 25 23:04:33 Well fine, displays OK, similar to PP(P), but FirstUse also funny size there Apr 25 23:15:48 Seems to do the trick on Tenderloin as well Apr 25 23:15:53 Anyway now time for bed **** ENDING LOGGING AT Tue Apr 26 02:59:56 2022