**** BEGIN LOGGING AT Sat Feb 13 03:02:54 2021 Feb 13 06:23:50 Morning! Feb 13 06:24:06 Tofe: Made 2 small comments on your PR to Halium devices Feb 13 06:24:47 I guess it's good to split it up into 2. 1 device specific (where you remove my unneeded remote for herrie gitlab (I guess I copied that from another manifest) and 1 for the setup script changes Feb 13 07:37:10 Tofe: I cleaned up the Halium directory ;) Feb 13 07:37:24 http://build.webos-ports.org/halium-luneos-9.0/ Feb 13 07:37:28 I guess anyway for Rosy and Mido we can do a 4.9 kernel build now too ;) Feb 13 08:10:30 Morning! Feb 13 08:10:50 oh, I forgot to tell you (but now it's too late), I also began to setup things for jenkins: https://github.com/Tofee/jenkins-jobs/commit/7bfe92061d5ac7d430f1c68ad5529fb589fa1694 Feb 13 08:11:57 Tofe: Well it wasn't a lot of work anyway ;) Feb 13 08:12:06 And we should end up with more or less the same Feb 13 08:12:20 I tried to factorize things a bit Feb 13 08:12:58 but I made it mainly with rosy in mind, at the time Feb 13 08:13:09 OK I will rework mine a bit Feb 13 08:13:22 I think for now we could still build 7.1 and 9.0 in parrallel for the Xiaomi targets Feb 13 08:13:47 Or anyway we have a Halium 7.1 build for it already, could skip it altogether Feb 13 08:14:47 yes, we can leave 7.1 in peace :) we nearly have same level of support with 9.0 right now - except bluetooth, and maybe wlan on tissot Feb 13 08:16:54 Herrie: I'll rename halium-luneos-9.0-20210211-1-tissot-4.9-new to halium-luneos-9.0-20210211 on build server, if that's the last one we have Feb 13 08:17:03 Yes it is Feb 13 08:17:12 ok; then I'll update the tissot recipe Feb 13 08:20:59 https://github.com/webOS-ports/jenkins-jobs/commit/51f84743169a4d8b5a83f3e94c6e3b75034c0545 Feb 13 08:21:06 That should incorporate your changes as well Feb 13 08:21:12 I'll squash it and PR I guess Feb 13 08:21:18 ok, great Feb 13 08:21:19 Already added the jobs on Jenkins Feb 13 08:26:44 now uploading my latest rosy halium build from january Feb 13 08:27:46 ...done (with 4G...) Feb 13 08:29:05 But I guess we should retry those with 4.9 kernel too? Feb 13 08:29:23 At some point Feb 13 08:29:59 on rosy ? was there any 4.9 kernel there ? Feb 13 08:33:34 ok, recipes updated, we should have a working build now for tissot and rosy Feb 13 08:33:46 Well found some: https://github.com/zexceed12300/android_kernel_xiaomi_rosy-4.9/commits/upstream-port Feb 13 08:37:05 right Feb 13 08:38:01 Not sure how good or usable it is... Feb 13 08:40:53 I think I'll postpone that till later; first I'd like to make sure we have a working build Feb 13 08:41:28 ah, damn, memnotify... Feb 13 08:48:55 Herrie: can you give me write access to memnotify-module repo ? Feb 13 08:48:58 Yes Feb 13 08:49:36 Tofe: Done Feb 13 08:59:16 Ok, tissot builds on my side, should be pretty much state-of-the-art Feb 13 09:02:36 BTW I'm not really sure what's up with the vibration bits... When I do them via command line they work OK, so I guess it's somewhere in LunaSysMgr or Nyx-Modules where it goes wrong. I'll dig into it Feb 13 09:02:38 Tofe: Nice Feb 13 09:44:05 Herrie: for booting twrp 3.5, we need AUDIT=y Feb 13 09:44:15 otherwise selinuxfs isn't getting mounted Feb 13 09:44:31 now, does it still boot luneos... let's try Feb 13 09:45:30 btw, getprop sys.init_boot_completed returns nothing , that's why it doesn't detect the end of boot Feb 13 09:46:06 ok, luneos still boots; I'll update the kernel defconfig Feb 13 09:48:20 Tofe: OK Feb 13 09:48:46 I also need to put this updated check_kernel_config script somewhere Feb 13 09:51:51 done. We now have a working recovery on tissot Feb 13 09:53:00 I'll reflash the whole luneos image and see what still needs work. Maybe wlan module loading... Feb 13 09:54:46 Yeah probably wlan and then stuff like BT which shouldn't be rocket science hopefully with bluebinder but who knows Feb 13 09:55:23 Not sure if other bits would need adjusting for that Feb 13 09:55:25 you already sketched a recipe ? Feb 13 09:55:39 No was late yesterday Feb 13 09:55:47 ok, I'll start it then Feb 13 09:56:08 It should be fairly straight forward since we have the other binder ones already Feb 13 09:56:13 Like libgbinder Feb 13 09:56:48 exactly Feb 13 09:57:22 speaking of which, those should probably be in meta-smartphone... Feb 13 09:57:28 but well, we'll see later Feb 13 10:05:08 damn, their makefile doesn't support cross-compile in the standard way... Feb 13 10:11:51 Tofe: Or maybe we should send it upstream to meta-oe/oe-core like Android tools? Feb 13 10:15:12 not sure, it's quite specific to halium-9.0 Feb 13 10:20:26 https://github.com/webOS-ports/jenkins-jobs/pull/14 LGTM, should I merge it now? Feb 13 10:20:45 I've merged that small lxc upgrade to gatesgarth branch (before it was only in master) Feb 13 10:23:16 JaMa: yes, let's merge that one Feb 13 10:53:38 Tofe: Let me start with a 4.9 Mido at my side I guess Feb 13 11:06:22 Tofe: https://github.com/webOS-ports/halium-helper-scripts/tree/master Feb 13 11:06:53 the check_config_file is useful for quickly fixing the defconfigs Feb 13 11:27:37 and we forgot to kill time_daemon Feb 13 11:29:24 Herrie: bluebooth already begins working Feb 13 11:30:00 Ah time_daemon yes we probably should kill it Feb 13 11:31:28 Tofe: You added bluebinder already? Feb 13 11:31:33 yes Feb 13 11:32:24 Ah I see the recipe... I don't understand why their Makefiles are always such a mess Feb 13 11:32:30 Must be laziness :P Feb 13 11:32:42 the recipe missed a couple of fixes, let me push it Feb 13 11:33:05 I guess we could send some improvement proposals upstream to Mer Feb 13 11:33:12 They're usually open to accept them Feb 13 11:34:31 ok, the recipe should now be what I have; it's not yet perfect (the service doesn't start for some rights error), but bluebinder itself works Feb 13 11:34:44 "bluebinder.service: Failed at step STDIN spawning /usr/bin/droid/bluebinder_wait.sh: Operation not permitted" Feb 13 11:35:13 oooh the path is incorrect... dammit Feb 13 11:35:57 Yeah that's Mer path I guess Feb 13 11:36:04 yes Feb 13 11:37:24 let me just compact the commits a bit Feb 13 11:38:42 done Feb 13 11:42:48 mmh the service still doesn't want to start /usr/sbin/bluebinder_wait.sh , but manually it succeeds Feb 13 11:43:47 Some permission issues somehow? Feb 13 11:45:27 maybe Feb 13 12:00:00 Shouldn't we remove the sandboxing bits? https://github.com/mer-hybris/bluebinder/blob/master/bluebinder.service#L13 Feb 13 12:20:28 I did just that some minutes ago, it helped a bit Feb 13 12:21:48 pushed Feb 13 15:20:17 Herrie: for wlan, it seems modprobe fails because it thinks the module is already in the kernel Feb 13 15:28:24 oh, my fault, I need to reflash luneos Feb 13 15:29:30 modprobe looks in /lib/modules/4.9.188 folder for informations about the builtin modules, so just installing the module won't update that table file Feb 13 15:36:01 Tofe: Hmmmz so Yocto needs to be enhanced somehow? Feb 13 15:38:36 well, the new module should ideally require also installing the new kernel ipk, but as the version is the same it's not complaining Feb 13 15:39:09 that's just because I modified brutally the kernel defconfig without pushing anything Feb 13 15:39:41 yep, now it works Feb 13 15:40:59 for bluetooth we should also disable the hciattach script Feb 13 15:47:21 Tofe: Ah weird it doesn't do that automatically when bumping srcrev of kernel Feb 13 15:47:26 Seems we're getting in pretty good shape :) Feb 13 15:47:45 Next thing that would be good to do is enable good old webos-telephonyd for dual sim Feb 13 15:49:47 ok, bluetooth VHCI activated in kernel (and pushed), so bluebinder should be nearly fine; I'll deactivate hciattach now Feb 13 15:54:49 bluebinder still doesn't work as-is with the service... Feb 13 15:55:01 also, let me disable that time_daemon stuff Feb 13 15:57:26 My mido build somehow fails with a weird include error Feb 13 16:10:18 ok, time_daemon should now be history Feb 13 16:10:29 Good :) Feb 13 16:11:49 yes, clean log :) Feb 13 16:20:01 Tofe: Maybe we need that timeout script? Feb 13 16:21:02 I,e,? https://github.com/SailfishOS-sagit/droid-config-sagit/blob/hybris-17.1/sparse/usr/bin/droid/droid-post-startup.sh Feb 13 16:21:39 Seems it simply restarts bluebinder and hci0 will appear afterwards Feb 13 16:23:04 it seems bluebinder is a bit sensible to the current state of things Feb 13 16:23:36 I still can't have get a reliable start of the service Feb 13 16:46:24 Tofe: It could be that this getprop call doesn't reply timely or accurately yet? https://github.com/mer-hybris/bluebinder/blob/master/bluebinder_wait.sh Feb 13 16:46:45 Should be able to see something in journal/logcat for that? Feb 13 16:52:30 Tofe: Maybe we should define some binder devices as per https://brauner.github.io/2019/01/09/android-binderfs.html and https://github.com/SailfishOS-sagit/android_kernel_xiaomi_msm8998/blob/a04247efda37e85667003e872a1e3d9ccdd26b20/arch/arm64/configs/sagit_defconfig#L710 Feb 13 16:52:31 ? Feb 13 17:04:38 Tofe: I noticed libgbinder seems quite active, might be worth a bump: https://github.com/mer-hybris/libgbinder/commits/master Feb 13 17:07:38 Herrie: getprop works fine I tried it in command line Feb 13 17:08:18 OK just pasting some things I came across that might be of help.... Feb 13 17:08:27 Need to build something here myself as well I guess Feb 13 17:08:36 Mido doesn't seem to like 4.9 with 16.0 somehow Feb 13 17:08:44 So I guess I'll focus on Tissot here as well Feb 13 17:30:58 btw, I've already re-uploaded tissot's halium image on build server, with time_daemon deactivated Feb 13 17:31:30 for mido, can you paste the error ? maybe I can help here ? Feb 13 17:39:28 Tofe: Yeah I saw Feb 13 17:39:48 Tofe: It's complaining about some missing includes in QCamera2 Feb 13 17:40:01 I'll paste error in a bit Feb 13 17:42:06 Https://bpa.st/QSSQ Feb 13 17:43:04 This is with https://bpa.st/FB4A Feb 13 18:02:16 Tofe: Camera works on my Tissot :P Feb 13 18:02:22 Front and back ;) Feb 13 18:08:11 Tofe: Should we add bluebinder to the packagegroup somewhere? Feb 13 18:25:17 yes, like line 154 Feb 13 18:26:15 I already have it locally, let me push it Feb 13 18:27:53 done Feb 13 18:29:12 Herrie: oh, but that the halium build... isn't that kernel already associated with a working 16.0 android rom ? Feb 13 18:29:54 Tofe: no seems 16.0 only with 3.18 on Mido Feb 13 18:29:54 Camera works on Tissot ?! oh my... I'll just drop the work I was doing to migrate to gstreamer and libdroidmedia :p Feb 13 18:32:15 At least the ones I've seen. I've seen 16.0-4.9 kernels though but no device tree I think Feb 13 18:36:00 Well I guess gstreamer and libdroidmedia is better for long run Feb 13 18:36:19 Also OSE has camera bits and services with gsteamer I think Feb 13 18:37:56 true; but so far it's a near-zero success :p Feb 13 18:38:03 ok let me see that kernel... Feb 13 18:39:13 ah but it's the device tree that fails to build, not the kernel Feb 13 18:39:17 I misunderstood Feb 13 18:41:25 though camera/CameraParameters.h should be in the device tree, too Feb 13 18:44:06 Herrie: this is what it's looking for: https://android.googlesource.com/platform/frameworks/av/+/refs/tags/android-cts-9.0_r15/camera/include/camera/CameraParameters.h Feb 13 18:44:12 in frameworks/av Feb 13 18:44:52 I have it locally, so you should too... Feb 13 18:48:19 Tofe: Well I have the files, need to check where exactly in a minute, but QCamera2 doesn't seem to find them Feb 13 18:50:27 but I don't see how changing the kernel plays a role here Feb 13 18:50:53 Me neither... And I know I had it building before somehow Feb 13 18:51:08 It could be some interaction between device tree and common maybe Feb 13 18:52:22 but you only changed kernel, right? Feb 13 19:10:06 Yes Feb 13 19:10:23 Well we did changes to common between last time I built mido and now Feb 13 19:10:25 For Tissot Feb 13 19:11:33 maybe you should retry with usual kernel, just to be sure it doesn't come from there Feb 13 19:12:08 3.18 you mean? Feb 13 19:41:11 I tried so many kernels before, don't remember which was the correct one... Feb 13 19:41:15 WIll just try them one by one :P Feb 13 19:51:58 Not sure, seems none build Feb 13 19:52:09 I guess I can just start from LOS from scratch Feb 13 19:56:33 Seems I can start with https://github.com/nE0sIghT/android_device_xiaomi_mido and https://github.com/nE0sIghT/android_kernel_xiaomi_msm8953 Feb 13 20:48:07 Herrie: oh, you didn't have the working manifest under the hand ? Feb 13 20:48:23 I meant Herrie82/kernel_xiaomi_mido **** ENDING LOGGING AT Sun Feb 14 02:59:57 2021