**** BEGIN LOGGING AT Fri Feb 12 02:59:57 2021 Feb 12 06:13:20 Morning! Feb 12 06:13:46 Tofe: Somehow memnotify fails for me... https://paste.ubuntu.com/p/TPjDtwkrY9/ Feb 12 06:20:10 In general, do we still need memnotify might be another question Feb 12 06:20:42 There might be different/better approaches available in newer kernels already? Feb 12 06:20:56 It originates from the old 2.6/3.0 kernels Feb 12 07:08:23 Morning! Feb 12 07:11:25 Herrie: I can fix this error, it's an easy one; it's not even a really bad one, we could ignore it eventually. It's just the return type of vm_shrink should be changed to long unsigned int Feb 12 07:12:14 Tofe: Yeah I know it can be fixed, just wonder where we use memnotify.... Feb 12 07:12:17 as for the usefullness of this module, I quite don't know Feb 12 07:13:58 We probably used it somewhere together with fps showing in some dev options Feb 12 07:14:24 morphis: ^ You recall? Feb 12 07:14:45 Since we also need to patch every kernel to support it Feb 12 07:14:58 If we don't need it, we could just as well drop it :P Feb 12 07:22:15 I thought it was a kind of out-ot-memory thing Feb 12 07:23:05 It says "system-wide memory meter and notifier pseudo-device" Feb 12 07:26:52 So seems it doesn't do a whole lot except for monitoring and notifying (I wonder how/where it notifies though) Feb 12 07:28:24 I think it just tells the programs that keep the /dev/memnotify node open that the current value has changed/reached a limit Feb 12 07:28:55 but if noone in webos/luneos opens that node, it's quite useless anyway Feb 12 07:33:09 Yeah probably... Feb 12 07:33:17 TOfe: Audio & rotation are back on my tissot Feb 12 07:33:24 Vibration not working yet, wifi also not for me Feb 12 07:33:29 Will have a look into logs later Feb 12 07:54:01 Herrie: for wifi, I had to load the wlan module manually, not sure why Feb 12 07:54:21 OK Feb 12 07:54:34 Well will try that in a bit when I'm back Feb 12 07:55:18 I recall you fixed vibration previously? Feb 12 07:56:32 I fixed the crash, not sure I made it work properly Feb 12 07:56:42 https://github.com/Tofee/meta-smartphone-1/blob/testing/halium-9.0/meta-xiaomi/recipes-core/systemd/systemd-machine-units/wifi-module-load.service#L9 this should work... weird Feb 12 07:57:54 ah yes, vibrator worked when toying directly with /sys Feb 12 08:02:37 basically, our mainline nyx haptics module should work I guess Feb 12 08:04:16 https://github.com/Tofee/meta-webos-ports/blob/testing/halium-9.0/meta-luneos/recipes-webos/nyx-modules/nyx-modules/tissot.cmake#L35 this is probably the mistake Feb 12 08:33:33 Tofe: Yeah I recalled that worked partially at least Feb 12 08:33:38 I'll try soon as well Feb 12 09:41:39 Tofe: Hmmz vibration doesn't work still... I updated cmake, rebuilt the nyx-modules and nyx-modules-hybris and installed those, rebooted Feb 12 09:41:42 Let me check logs Feb 12 09:43:59 For WLAN I get: https://paste.ubuntu.com/p/4J6J9J4Sxz/ Feb 12 09:44:35 that's strange, I didn't get that Feb 12 09:44:42 let me retry this Feb 12 09:45:31 Unless I forgot something somehow Feb 12 09:45:35 While picking all your commits Feb 12 09:45:39 But I think I should be OK Feb 12 09:46:24 I have /userdata/luneos/lib/modules/4.9.188/kernel/drivers/staging/prima/wlan.ko and /lib/modules/4.9.188/kernel/drivers/staging/prima/wlan.ko Feb 12 09:46:56 ah, it works with insmod, but not modprobe Feb 12 09:47:41 (but I installed the module manually in the image, maybe I forgot a step) Feb 12 09:48:04 Herrie: I have the same .ko here Feb 12 09:49:50 Maybe we should check SFOS/UBPorts for WLAN Feb 12 09:50:01 Maybe we need some minor change in the .service Feb 12 09:50:06 I.e. start later or something as well Feb 12 09:51:50 could be. Also I saw that the detection of wait-for-android is not really working on tissot Feb 12 09:53:17 Tofe: Yeah it's a bit laggy Feb 12 09:53:22 I saw that previously as well Feb 12 09:54:15 I'll have to re-check is getting the property is the right way Feb 12 10:11:07 Or maybe it emits too late somehow Feb 12 10:11:15 I.e. held up by something on Android side Feb 12 10:13:54 like, because of time_daemon or something Feb 12 10:19:56 Yeah not sure Feb 12 10:20:04 But that's something I cna think of Feb 12 10:20:16 For vibrator, seems location might have changed..... Feb 12 10:24:32 As per discussion on https://irclogs.sailfishos.org/logs/%23sailfishos-porters/%23sailfishos-porters.2020-10-11.log.html Feb 12 10:25:22 I,e, if I do: sh-5.0# echo 500 > /sys/class/leds/vibrator/duration Feb 12 10:25:22 sh-5.0# echo 1 > /sys/class/leds/vibrator/activate Feb 12 10:25:24 It works Feb 12 10:25:46 So we'd need to update: https://github.com/webOS-ports/nyx-modules/blob/webOS-ports/webOS-OSE/src/haptics_timedoutput/haptics.c#L66 Feb 12 10:27:23 yes, we can add another match rule Feb 12 10:28:19 Well I think it needs a little more Feb 12 10:28:50 Because the actual vibrating is now a 2 step approach Feb 12 10:28:56 ah, right Feb 12 10:29:04 Write duration and activate Feb 12 10:29:55 when I was investigating this, I found a possible source code to copy from, with the timer callback and so on Feb 12 10:30:08 Probably mer Feb 12 10:30:54 This one? https://github.com/mer-hybris/ngfd-plugin-droid-vibrator/blob/3ff916bafef422a78ca51e68b8cd98c7638c6497/src/native.c Feb 12 10:33:29 no, something else Feb 12 10:33:47 I think it's somewhere in our logs, if the bot recorded it Feb 12 10:34:21 Tofe: Bot should be better now Feb 12 10:34:29 novaldex switched some stuff around Feb 12 10:34:35 There are still missing days, but a lot less Feb 12 10:35:04 ah but wait our current nyx implementation has all it needs, it seems. Feb 12 10:35:18 Does it? Feb 12 10:36:00 duration, timeout_callback, etc... it's what we what, isn't it Feb 12 10:36:06 Yes Feb 12 10:36:16 Just the device enumeration via udev doesn't work Feb 12 10:36:22 And the activation is different now Feb 12 10:36:50 These are the changes SFOS made to accomodate it at their side: https://github.com/mer-hybris/ngfd-plugin-droid-vibrator/commit/07c6d1aa9ce005b8bf35a68bc3bb4a730d6b50c1 Feb 12 10:37:30 https://github.com/webOS-ports/nyx-modules/blob/webOS-ports/webOS-OSE/src/haptics_timedoutput/haptics.c#L165 we'll just need to activate the vibrator here Feb 12 10:37:34 (and add a match rule) Feb 12 10:38:54 I'd prefer a udev rule, it's prettier Feb 12 10:38:58 and more robust Feb 12 10:39:53 I'll see what I can do Feb 12 10:44:56 so it would be subsystem "leds", then iterate to find a "vibrator" in the path I think Feb 12 10:46:37 Yes seems so Feb 12 10:46:43 Weird they put vibrator in leds but well Feb 12 10:46:57 yes, not pratical Feb 12 10:47:28 we could also add a udev rule to attach a property to the vibrator, but maybe we'll manage without it Feb 12 11:10:55 Herrie: https://github.com/Tofee/nyx-modules/commit/63aede26c941085e3c9551321b578346f012206e totally untested code Feb 12 11:12:20 https://github.com/Tofee/nyx-modules/commit/c2bb86fb53eb9625cbf971c6f13d4398532cffb3 a bit better, without a leak Feb 12 11:12:59 probably doesn't build at all, though ... Feb 12 12:12:16 Tofe: https://paste.ubuntu.com/p/YxS9D7Z7Bb/ Feb 12 12:42:50 fixed Feb 12 14:48:24 Another error: /usr/src/debug/nyx-modules/7.1.0-1+gitAUTOINC+522f65cadf-r0/git/src/haptics_timedoutput/haptics.c:116: error: undefined reference to 'g_strcmp' Feb 12 15:03:35 ah, damn, I forgot to check if that one existed :D Feb 12 15:04:47 pushed fix Feb 12 15:04:57 but I'll start my own build soon too Feb 12 17:29:44 Herrie: I don't understand you halium-devices:halium-9.0 branch history Feb 12 17:29:56 looks like it doesn't match upstream's history ? Feb 12 17:30:01 your* Feb 12 17:33:23 Herrie: If I PR mine, here's what I have: https://github.com/Halium/halium-devices/compare/halium-9.0...Tofee:halium-9.0?expand=1 Feb 12 17:37:00 ok, PR'ed Feb 12 17:55:09 Tofe: Ah well I cloned 7.1 one at some point Feb 12 17:55:18 When 9.0 didn't exist yet Feb 12 17:55:32 I need to force push my local updated one Feb 12 19:51:13 Tofe: Seems nyx-modules needs some tweak(s) still: Sep 21 00:32:47 tissot LunaSysMgr[3127]: [] [pmlog] NYXHAPTICS_NODEVICE_ERR {} Did not find any devices matching the vibrator subsystem Feb 12 20:05:58 Herrie: can't say I'm surprised ! Feb 12 20:26:21 Tofe: Well you said it was untested, so yes, me neither ;) Feb 12 20:26:39 I've built it locally, let me debug that quickly Feb 12 20:27:48 Tofe: For BT found this in SFOS: https://github.com/SailfishOS-sagit/droid-config-sagit/blob/hybris-16.0/sparse/usr/bin/droid/droid-post-startup.sh Feb 12 20:27:53 Could be we have the same issue Feb 12 20:28:27 well we don't have any bluebinder, maybe that's the issue p Feb 12 20:28:28 :p Feb 12 20:34:23 Could very well be... "bluebinder is a simple proxy for using android binder based bluetooth through vhci." Feb 12 20:36:40 P.s. Seems this is what they use for Xiaomi Sagit on SFOS: https://github.com/SailfishOS-sagit/android_kernel_xiaomi_msm8998/blob/a04247efda37e85667003e872a1e3d9ccdd26b20/arch/arm64/configs/sagit_defconfig#L794 Feb 12 20:36:50 I guess we should try the same and bluebinder to be added Feb 12 20:36:58 Seems lots of things are via binder going forward Feb 12 20:37:09 Luckily we should be able to piggyback from Mer :) Feb 12 20:42:02 yes, and binder is easy enough to use Feb 12 20:42:38 so, it seems my udev listing of "leds" is fine, but the sub-systems have no name, so "udev_list_entry_get_name" is returning NULL or something like that Feb 12 20:42:50 I think I'll just simplify all this Feb 12 20:55:12 https://github.com/webOS-ports/nyx-modules/pull/18 seems to work at least partially Feb 12 20:55:40 the "300" 0.6sec vibration worked, but the others didn't Feb 12 20:57:29 Herrie: ah, yes, I see "CONFIG_BT_HCIVHCI=y" Feb 12 21:01:58 Tofe: Well it's a start I guess Feb 12 21:02:07 When I manually tried values it worked it seems Feb 12 21:02:12 Can try a bit more tomorrow Feb 12 21:06:00 Herrie: we'll need a clearer status for builds.webos-ports.org/halium-luneos-9.0 content: what's up-to-date ? what needs a last upload ? Feb 12 21:06:34 I don't think we're pointing the latest things for tissot in the recipe, and also not for rosy Feb 12 21:09:07 and for rosy I think I'll need to rebuild halium and upload the result, just to be sure Feb 12 21:11:07 anyway, that's for tomorrow Feb 12 21:19:42 Tofe: Yeah. already cleaned what I knew could be cleaned Feb 12 21:19:52 Can be done a bit more Feb 12 21:20:02 I guess I could stat to setup things on Jenkins Feb 12 21:20:09 s/stat/start Feb 12 21:46:36 I guess this would be a good start: https://github.com/webOS-ports/jenkins-jobs/commit/97f07ac2c4db612ec74bcbb40786ed299607f576 **** ENDING LOGGING AT Sat Feb 13 03:02:54 2021