**** BEGIN LOGGING AT Sat Jan 02 02:59:56 2021 Jan 02 10:27:29 For the ueventd issue, it's not clear yet why we have this issue with Halium-9 and not the 7.1 or 5.1 versions... The conflict should also occur there for the netlink socket... Jan 02 10:28:10 (also I confirm that systemd creates a UEVENT netlink with pid 1, that's the one that may conflict with ueventd in lxc) Jan 02 10:49:33 Tofe: I do recall we had some ueventd patch in 5.1/7.1 that would sort some buffer issues or something Jan 02 10:50:10 Halium 10 is from NotKit for a single OnePlus device I think so probably done clean and not based upon Mer that much Jan 02 10:50:47 I can have a look later for ueventd, now heading out for some shopping Jan 02 10:50:56 Or let me just quickly check now Jan 02 10:51:49 Herrie: yes, ueventd is also patched to fix a "uevent events peak", the buffer being too small Jan 02 10:52:29 I'll wait for NotKit to show up, then I'll ask about his status Jan 02 10:53:29 Tofe: Yeah that one I meant Jan 02 10:54:15 I understood it's very much WIP and not even alpha at this stage, at least that was my understanding earlier this week Jan 02 10:55:29 With NotKit it usually helps to tag him ;) Jan 02 10:55:58 Or PM Jan 02 10:56:29 Ok, we'll see :) Jan 02 13:02:36 Tofe: Seems NotKit woke up Jan 02 13:03:03 Couldn't it be some systemd feature we use that others don't? Just thinking out loud here Jan 02 14:41:17 Herrie: I don't know, I can't really explain why Halium-7.1, which has exactly the same ueventd code, would work well with the same systemd Jan 02 15:04:08 Yeah that's true Jan 02 15:16:41 Tofe: I came across this one: https://github.com/Halium/android_system_core/commit/cb721e441d046600964e9639742e387b14887501 Jan 02 15:16:51 This is what we have in 7.1 Jan 02 15:16:55 Not sure that's in 9.0 as well? Jan 02 15:17:39 And this: https://github.com/Halium/android_system_core/commit/f35441ad149623455b3e59e64e5b2ba942d5fd97 Jan 02 15:18:46 And this one: https://github.com/Halium/android_system_core/commit/49f8d2b51904214c5f8a9879c6e4cc2729e0b454 Jan 02 15:18:58 Just going through some patches and pasting what I think might be related Jan 02 16:18:18 Herrie: I thin kthe /socket thing is also in the patches Jan 02 16:19:11 which is also a bit messy, because it needs to modify the rootfs directory, i.e. the system.img mount Jan 02 17:41:32 Tofe: Seems the scripts in Mobian for lxc might be of help, seems they do quite some things Jan 02 17:41:54 The ones in lxc-android NotKit pointed to Jan 02 17:44:22 yes, notably they pre-mount a lot of android things Jan 02 17:45:12 currently I'm trying to build the GSI image myself, but "lunch" doesn't find lineageos_halium_arm64-userdebug Jan 02 17:47:08 ah, there's no vendorsetup.sh Jan 02 17:49:11 ok, let's go Jan 02 17:53:25 Herrie: we could even fork mobian's lxc-android to webos-ports Jan 02 17:54:11 (but I can't do it with my account) Jan 02 17:55:41 Tofe: Give me a minute Jan 02 17:59:21 Tofe: Done Jan 02 18:05:52 thx Jan 02 22:11:26 Herrie: do I understand correctly that if take my GSI image, and the vendorimage from my rosy build, it should work ? Jan 02 22:12:09 Tofe: Not sure but could be yes Jan 02 22:12:32 I never really understood that part completely Jan 02 22:12:43 I'm just not sure if additional modifications have to be made to better separate the vendor image from the rest Jan 02 22:12:58 Well worth the try? Jan 02 22:13:02 for sure yes Jan 02 22:13:26 I have a GSI system.img now, and still have my previous vendor.img of course Jan 02 22:14:12 oh, it's not system-as-root Jan 02 22:21:32 sorry my fault, it is system-as-root Jan 02 22:26:56 The whole idea of GSI (in theory) is that only the vendor differs Jan 02 22:27:15 Though NotKit said some hacks were needed with Whyred Jan 02 22:28:16 There's this for Whyred: https://github.com/Linux-On-Sdm6Series/Linux_manifest/releases Jan 02 22:31:08 This account also has a device tree for Whyred with a lot of commits https://github.com/Linux-On-Sdm6Series/device_halium_whyred/commits/halium-9.0 Jan 02 22:38:42 yes, but as long as these files go to vendor.img, that's fine.... not sure if they do, though Jan 02 22:39:35 hey, my rosy lxc almost starts! Jan 02 22:40:46 ah, "init: critical process 'healthd' exited 4 times in 4 minutes" Jan 02 22:40:52 let's just disable that one Jan 02 22:40:59 Well I think you can kill that one Jan 02 22:41:45 yes, the only issue is that it's marked as critical Jan 02 22:42:00 so when it crashes it's fatal Jan 02 23:13:35 damn, can't find what starts healthd... otherwise it's not that bad, overall **** ENDING LOGGING AT Sun Jan 03 02:59:56 2021