**** BEGIN LOGGING AT Sun Feb 14 02:59:57 2021 Feb 14 07:11:01 Morning Feb 14 07:11:24 Tofe: Well I thought I had 2 working manifests, but they don't work anymore somehow... Feb 14 07:11:31 Even when starting from scratch Feb 14 07:11:51 I guess it's the common somehow, because I changed device and kernel, left common the same as for Tissot Feb 14 08:46:45 Morning! Feb 14 08:48:10 I'll retry with LOS msm8953 common, pretty sure a recent change we did to common somehow broke mido Feb 14 08:49:03 Because I was able to build Halium with 3.18/4.9 kernel before Feb 14 08:51:34 I wonder that kind of change could cause that Feb 14 08:52:28 you use the extended branch, right? Feb 14 08:53:05 ah, no, it's halium-9.0 for mido Feb 14 08:54:06 (and it matches my memory, where you alraedy disabled time_daemon in your repos) Feb 14 08:54:40 Unless I'm mistaken but I'm pretty sure I built 4.9 as well Feb 14 08:54:45 3.18 for sure Feb 14 08:54:58 I have mido booting Halium 9 Feb 14 08:56:37 yes, the mido status was quite good Feb 14 08:57:41 Herrie: so you've switch to halium-9.0 branch for common repo, right ? as it wasn't the case in you bpaste earlier Feb 14 08:58:27 At some point we should check the differences between those two, it's a bit silly to have slightly different branches for a common repo :) Feb 14 09:01:56 Yeah will get something to build and figure out what causes the issues Feb 14 09:02:34 The "los" 16 device tree with 4.9 kernel still fails with the headers it cannot find Feb 14 09:02:51 So my deduction is it can only be the common somehoow Feb 14 09:03:23 Next step is to use the LOS unmodified common to see what happens Feb 14 09:21:01 How do I get wifi to work? Feb 14 09:21:06 Not really clear to me Feb 14 09:21:13 Still doesn't work for me on TIssot Feb 14 09:37:38 Tofe: With it builds, however with it fails Feb 14 09:40:41 Seems it could only be those VNDK flags somehow: https://github.com/Herrie82/android_device_xiaomi_msm8953-common/compare/halium-9.0...Herrie82:halium-9.0-extended Feb 14 10:13:28 Herrie: for tissot, if you build the whole image and flash it, it should work now Feb 14 10:13:46 I think I pushed everything related to that Feb 14 10:14:26 yes the vndk flags sure change quite a bit what we build Feb 14 10:19:23 Tofe: Well I did that last night, but no luck Feb 14 10:19:31 ah :/ Feb 14 10:19:37 Still getting Sep 20 07:23:59 tissot modprobe[13689]: modprobe: ERROR: could not insert 'wlan': Exec format error Feb 14 10:19:59 it's weird, I never got that one Feb 14 10:20:32 Tofe: For bluebinder I added it just after bluez5, seems you added it after device... Feb 14 10:20:36 I guess your way is more correct Feb 14 10:20:46 Since we want to do it for specific devices only I guess Feb 14 10:21:23 well, in the future it'll be for all halium-9 devices I think Feb 14 10:21:31 Yeah probably Feb 14 10:21:34 Well let me try that... Feb 14 10:21:44 Maybe I should simply clean my tmp-glibc and rebuild Feb 14 10:24:03 you can offord that :p Feb 14 10:25:32 Tofe: Well deleting tmp-glibc should be OK in any case... It should quickly rebuild from sstate Feb 14 10:25:42 At least that's what JaMa told me sometime Feb 14 11:28:48 I did a cleanall for the linux-xiaomi-mido which should delete any artifacts also of the wlan module Feb 14 11:50:30 So it should rebuild it as one part Feb 14 13:11:01 Herrie: once https://github.com/Halium/hybris-patches/pull/10 is merged, our boot detection will work much better Feb 14 13:24:07 Ah yes that would work Feb 14 13:25:15 we can also apply this patch manually, and build our images, if we don't want to wait Feb 14 13:25:27 I'll do that for tissot Feb 14 13:29:19 Yeah that sounds good for now Feb 14 13:29:30 Though it should be merged soon I guess Feb 14 13:29:43 I noticed also shutdown seems stuck forever still Feb 14 13:29:52 We might want to have a look @ UBPOrts/SFOS for what they do Feb 14 13:29:53 I just asked for a renaming of the patch file Feb 14 13:30:04 Yup saw that Feb 14 13:33:45 SFOS is doing a lot of Mer specific patch triggering; I prefer the PR approach (which I believe is how ubports handles it) Feb 14 13:34:06 ok, new tissot halium image built & uploaded Feb 14 13:34:39 Nice :) Feb 14 13:34:50 For Mido really seems to be the VNDK flags somehow Feb 14 13:34:50 hurray for 4g Feb 14 13:35:02 I'll try to see what it does exactly Feb 14 13:35:20 Maybe some camera flags in BoardConfig.mk or device.mk somehow Feb 14 13:35:27 I'll compare them a bit later Feb 14 13:35:53 Will be out with kids in about an hour for some time skating, so probably tonight or tomorrow Feb 14 13:36:16 don't forget I also added some VNDK related commits to tissot's device.mk Feb 14 13:36:16 and BoardConfig too iirc Feb 14 13:36:24 so that we get a vendor image, but keep strong links between vendor and system Feb 14 13:36:58 Ah maybe I need those simply Feb 14 13:37:01 Ok :) In the meantime I'll try to see what goes wrong with bluebinder Feb 14 13:37:55 Locally I bumped libgbinder and libglibutils Feb 14 13:38:14 good idea Feb 14 13:38:14 Let's see how my bluebinder works if at all Feb 14 13:41:25 ok, boot works a bit better with the patch Feb 14 13:42:08 As in quicker? Feb 14 13:42:31 yes, doesn't wait until systemd's timeout (90s) is reached Feb 14 14:04:20 for bluebinder, looks like it was simply the PostStart script that was always failing and was problematic for systemd. Also, the WantedBy was incorrect. Feb 14 14:07:39 I already was thinking about graphical.target Feb 14 14:07:45 That didn't look familiar with me Feb 14 14:08:32 WAnted to mention it but forgot ;) Feb 14 14:08:37 Good it seems sorted now Feb 14 14:09:19 yes :) Feb 14 14:29:33 Tofe: Oohh it boots a lot quicker! Feb 14 14:30:41 I still don't have wifi or BT though somehow Feb 14 14:30:58 Anyway need to run now will pull some logs tonight probably Feb 14 14:31:20 How to test BT? Regular BT app and turn it on there or....? Feb 14 14:37:53 turn it on, yes :) Feb 14 14:40:42 Doesn't seem to do anything here somehow Feb 14 14:40:59 Well that's for later to check then Feb 14 15:40:39 Herrie: it could be you'll need to disable the service and reenable it again, if you only installed the ipk Feb 14 15:41:48 I reflashed whole image with all changes you pushed Feb 14 15:57:49 fingers crossed Feb 14 16:03:22 Tofe: Well wifi and bt not working is after flashing whole image with all changes..... Feb 14 16:03:38 Will pull some logs when back home tonight or tomorrow morning Feb 14 16:03:49 strange Feb 14 16:04:02 speaking of which, my wifi suddently stopped working Feb 14 16:04:28 I suspend a radio partition got corrupted, somehow Feb 14 16:04:31 suspect* Feb 14 16:17:33 Tofe: I might have something similar will see Feb 14 16:20:04 I'll just reflash the radio/modem partitions Feb 14 16:20:17 just in case. But it's quite unlikely, though Feb 14 16:27:35 damn, now I have neither wlan nor bluetooth ! Feb 14 16:31:59 mmmh no firmware loading... Feb 14 16:36:42 Tofe: Well it's good it's not just me then :P Feb 14 16:36:49 At least some consistency Feb 14 16:36:53 hehe yes Feb 14 16:37:10 Random not working is harder to debug Feb 14 16:37:28 Weird it worked and stopped working though Feb 14 16:37:38 yes, I'm quite surprised here Feb 14 16:37:51 Could it be we need these binder names in defconfig? Feb 14 16:38:13 I didn't read in full detail but saw them in some UBPorts kernels Feb 14 16:39:25 but I didn't touch that in https://github.com/Tofee/tissot/commits/tissot/4.9/halium-9.0/arch/arm64/configs/tissot_defconfig Feb 14 16:42:54 No but I mean it could be needed to keep things stable and running over time. I'm not sure Feb 14 17:06:02 I don't think wifi is doing anything with the binder Feb 14 17:06:12 Probably not indeed Feb 14 17:06:26 or, could it be that the "boot_completed" is just happening too soon now Feb 14 17:09:24 I see "subsys-pil-tz a21b000.qcom,pronto: Subsystem error monitoring/handling services are up" Feb 14 17:10:06 That's also an option of course, being too soon Feb 14 17:10:31 But then the patch maybe isn't 100% correct Feb 14 17:10:42 nope, it's not that; I've disabled the wlan systemd service, and we still have the issue Feb 14 17:11:04 Could it be I messed up the tissot halium image somehow Feb 14 17:11:26 How? You just added the patch right? Feb 14 17:11:42 yes Feb 14 17:12:33 I didn't change that many things latetly: wlan module, a couple of halium patches, bluebinder Feb 14 17:22:59 Tofe: Could be we need a few defconfig tweaks for fw_helper_path etc? Feb 14 17:23:09 Not sure I included those already Feb 14 17:23:24 we have it in cmdline Feb 14 17:23:41 and it worked yesterday... Feb 14 17:24:07 let me rollback a bit the halium image I have Feb 14 17:24:36 I mainly took Mer's kernel check script and changed those values Feb 14 17:25:00 Could be we need some more for lxc or other bits that ubports have Feb 14 17:35:08 But that wouldn't explain things working before Feb 14 17:35:22 Maybe it was time_daemon :P Feb 14 17:57:59 I'm reflashing luneos with yesterday's halium image Feb 14 17:58:55 ... and I have wlan back Feb 14 17:59:36 and bluebinder works Feb 14 17:59:45 so I messed up halium's image somehow Feb 14 18:04:25 it can't seriously be time_daemon... can it Feb 14 18:09:31 in the meantime, let me remove the commit that points to the newest halium image Feb 14 18:11:27 ok it's not that, it seems. Let's see if I remove the "init_bootcompleted" patch Feb 14 18:21:21 ok, it's that patch Feb 14 18:41:46 Herrie: I pushed a new halium image; it doesn't include the patch for init_boot_completed, we'll have to find a better way Feb 14 18:42:25 also forced-pushed meta-smartphone:testing/halium-9.0 to update the checksums for that image Feb 14 18:42:27 OK so that patch is buggy Feb 14 18:42:39 it triggers too early I guess Feb 14 18:42:43 Might be good to mention in the PR our findings Feb 14 18:45:08 done Feb 14 18:45:42 I'm now trying to put in place the camera stuff via gstreamer, in case it would just work Feb 14 18:46:07 basically I'm just adding the omx service to device.mk Feb 14 18:53:59 Just got back home,... Rebuilding with the force-pushed Halium image Feb 14 19:23:59 Herrie: I think the right thing to do would be to use the Mer patch that start the init_droid_done.sh script; only we also will set the property there Feb 14 19:26:57 basically, we change slightly https://github.com/Halium/hybris-patches/blob/halium-9.0/system/core/0008-hybris-Notify-Mer-s-systemd-that-we-re-done.patch so that it also does setprop sys.init_boot_completed 1 Feb 14 19:27:04 and then, everyone is happy Feb 14 19:29:08 That would probably work nicely yes Feb 14 19:29:25 testing that right now Feb 14 19:30:08 nope :( Feb 14 19:33:09 still too soon for our wlan module Feb 14 19:54:28 Might be worth to poke around some of the UBPorts Halium 9 targets to see how they deal with it Feb 14 20:00:39 I think it's not only the wlan module loading, it's coupled to other things Feb 14 20:00:54 I masked the service that loads the module, and it didn't avoid the issue Feb 14 20:01:17 so there's another service that is activated too soon, and triggers this kind of problem Feb 14 20:04:16 NotKit or Erfan Abdi might have some thoughts Feb 14 20:04:29 I doubt this would be specific to us? Feb 14 20:06:17 https://github.com/Halium/android_device_halium_halium_arm64/commit/cc536b527c2585ec040505fca661cd5bcffe4bef ? Feb 14 20:06:35 https://github.com/Halium/android_device_halium_halium_arm64/commit/475eb70561989c80060680b6577a75c8e182998d Feb 14 20:07:42 https://github.com/Halium/android_device_halium_halium_arm64/commit/1e2b7c8d083c1eb2ba54834956c010f556b48a21 Feb 14 20:09:16 Might be some pointers **** ENDING LOGGING AT Mon Feb 15 02:59:58 2021