**** BEGIN LOGGING AT Fri Jun 15 03:00:11 2018 Jun 15 05:33:04 MOrning! Jun 15 05:33:15 elvispre: I would doubt that f2fs is the right way to go Jun 15 05:37:29 But maybe Tofe has something more useful to say ;) Jun 15 06:44:00 Morning Jun 15 06:45:12 elvispre: the initramfs script very probably uses an assumption for Ext2 Jun 15 06:50:49 We use this script https://github.com/Tofee/initramfs-tools-halium/blob/halium/scripts/halium which does most of the work, and in addition we have our ovverrides for LuneOS: https://github.com/Tofee/meta-webos-ports/blob/pyro/meta-luneos/recipes-core/initrdscripts/initramfs-boot-android/init.sh Jun 15 07:29:37 Tofe: Well I think we have ext4 on our targets ;) Jun 15 07:30:35 https://github.com/LineageOS/android_device_lge_hammerhead/blob/cm-12.1/fstab.hammerhead#L8 Jun 15 07:31:20 https://github.com/LineageOS/android_device_lge_mako/blob/cm-12.1/fstab.mako#L7 Jun 15 07:31:30 I think I saw Halium devices where they explicitly disabled f2fs Jun 15 07:37:06 elvispre: I think this works somehow better compared to previous, but you might want to try to revert back to ext4 and format/wipe cache instead from recovery Jun 15 07:37:19 For another OnePlus they did this in fstab to fix issues with formatting: https://github.com/Halium/android_device_oneplus_oneplus2/commit/5cdff457eb7241510aa003fb64f3b0f9b3bd7b59 Jun 15 08:07:35 Herrie|Laptop: do we really care about the cache partition, actually ? Isn't it only about the userdata one ? Jun 15 08:07:55 Tofe: Yeah I don't think we use cache ourselves for anything Jun 15 12:45:24 Herrie: Tofe: Yeah, I tried F2FS for cache because the error messages were F2FS ones. Reformatting cache did at least change the error message. Jun 15 12:45:43 I shall probably go back to ext4 cache and try a different approach. Jun 15 12:46:31 For info, I am picking my way through the bootstrap process so I can understand properly what is going on. Jun 15 12:47:31 You may be surprised to learn that I had never even looked inside our luneos-package zip files before this week. Jun 15 12:51:08 Herrie: Tofe: LuneOS does not use cache, but I was thinking that the android inside the container may want to. The big error I have is that the container fails to start. Jun 15 12:51:57 ah, ok ! I misunderstood the issue here Jun 15 15:08:05 Morning all, I was very sneaky & just did some patching of Jenkins, should all be back up & running now Jun 15 15:10:26 there are few new commits in webOS OSE, I've triggered new jobs, but it won't change (or even fix) anything important for us Jun 15 16:57:02 JaMa: Saw that Jun 15 16:57:08 Well they released memorymanager Jun 15 17:14:23 ping Herrie Jun 15 17:24:20 Herrie: https://github.com/webOS-ports/webos-keyboard/pull/85 will fix virtual keyboard Jun 15 17:42:13 Tofe: Fix vkb where? Jun 15 17:42:19 QT 5.11? Jun 15 17:47:13 elvispre|s: pong Jun 15 17:47:22 Hi Herrie Jun 15 17:47:49 I'm trying to build the Halium image and having some trouble Jun 15 17:48:13 Herrie: yes Jun 15 17:48:43 I accidentally typed "setup mako" when I meant "setup onyx" and it seems unrecoverable Jun 15 17:49:19 Can the repo repo be cleaned up? Jun 15 17:50:15 Tofe: Are you testing 5.11 on phone? Do you need LD_PRELOAD hack on it? Jun 15 17:51:21 nizovn: yes, and yes Jun 15 17:51:53 mmmh well actually I'm not sure I tested without it Jun 15 17:51:56 My current cleanup technique is "rm -rf" but that is rather expensive when there are many GB of downloads for each attempt Jun 15 17:51:59 Tofe: it doesn't work correctly it seems at least in emulator Jun 15 17:52:06 elvispre|s: delete halium/device/lge and halium/device/qualcom Jun 15 17:52:16 Same for vendor Jun 15 17:52:40 i see weird behavior: display gets updated only after next tap Jun 15 17:53:06 it's clearly visible in preferences page in browser when clicking on switches Jun 15 17:53:35 just not sure if it's only emu thing Jun 15 17:53:37 Herrie: Thanks. Actually, I have tweaked the build script you showed me so now I am trying that rather than typing commands by hand. Jun 15 18:25:49 elvispre: In general the build script & halium setup script should work OK-ish for most targets Jun 15 18:26:27 There are a few more tweaks to be added to setup script to automate the disabling of (Xiaomi/Samsung/OnePlus)Doze, CMActions/LineageActions and keyhandler Jun 15 18:26:36 But this is only for some targets Jun 15 18:33:38 nizovn: I don't have that issue on N5 Jun 15 18:34:01 though I have plenty of EGL errors like invalid context etc Jun 15 18:34:25 ok Jun 15 18:36:47 maybe it's already solved by Qt: https://github.com/qt/qtwebengine/commit/a66d4cd82972996d76edff52d17464c150dec6a6 Jun 15 18:48:46 nizovn: I tested, and I need LD_PRELOAD Jun 15 18:55:26 It turns out that the halium onyx root filesystem's fstab has *both* an F2FS and an ext4 entry for /cache. Jun 15 18:55:54 Presumably, that's why I get an error with it formatted either way. Jun 15 18:56:24 Moving on. At least I learned something :-) Jun 15 19:10:52 Hehe Jun 15 19:10:58 Any logs you can post we can have a look too Jun 15 19:11:03 Some things might ring a bell ;) Jun 15 19:21:00 Herrie: If you're interested, https://pastebin.com/vs5aG5nR is from just now Jun 15 19:21:52 Really, I think I just have to follow the boot process through and understand what is supposed to happen and so identify what does not happen. Jun 15 19:36:08 Well I see various error messages that could give a clue Jun 15 19:37:35 mount: mounting /dev/disk/by-partlabel/boot on /rfs/android//boot failed: No such device Jun 15 19:37:45 mount: mounting /dev/disk/by-partlabel/recovery on /rfs/android//recovery failed: No such device Jun 15 19:37:53 Double // seems suspicious Jun 15 19:38:05 And /halium-boot.sh: line 550: [: /rfs/usr/lib/lxc-android-config/70-onyx: binary operator expected Jun 15 19:38:11 May 16 19:18:57 onyx kernel[842]: [ 6.876251] systemd[1]: Failed to insert module 'autofs4': No such file or directory Jun 15 19:39:18 Seems you might some of the PID namespace patches in kernel: May 16 19:18:55 onyx systemd[566]: systemd-timesyncd.service: Failed at step NAMESPACE spawning /lib/systemd/systemd-timesyncd: Bad file descriptor Jun 15 19:39:45 I.e. https://github.com/Halium/android_kernel_lge_hammerhead/commit/5754614eb43dea44a99e54898e3b83d4d96d8b83 and https://github.com/Halium/android_kernel_lge_hammerhead/commit/bd221854de33b75db7a7fa01cb34274b62a7cbf8 Jun 15 19:40:37 Ah wait that's in my kernel Jun 15 19:41:56 I guess you might need to bump srcrev a bit: https://github.com/shr-distribution/meta-smartphone/blob/pyro/meta-oneplus/recipes-kernel/linux/linux-oneplus-onyx_git.bb#L22 Jun 15 19:42:21 Should become: 355e7065c91b9a2d1e52d85b3c2263ed785111a9 Jun 15 20:06:40 Herrie: it's a bit strange that this line fails this way https://github.com/Tofee/initramfs-tools-halium/blob/halium/scripts/halium#L550 Jun 15 20:09:50 Tofe: Well there could be a syntax error here: https://github.com/shr-distribution/meta-smartphone/blob/pyro/meta-oneplus/recipes-core/udev/udev-extraconf/70-onyx.rules Jun 15 20:19:59 elvispre: Seems that there's an issue mounting your persist partition which contains firmware and stuff Jun 15 20:21:19 Herrie: I think the log says it was mounted OK (line 989) Jun 15 20:23:51 elvispre: Yeah :S Jun 15 20:27:03 Herrie: Do you build your halium image using the same machine as your luneos things? Jun 15 20:27:10 elvispre: Well luna-service2 seems to fail to start Jun 15 20:27:19 ls-hubd_public.service Jun 15 20:27:59 Herrie: That sounds bad :-) Jun 15 20:30:10 elvispre: Well it seems that the mmcblk0p28 might be the issue Jun 15 20:30:14 That's system partition it seems Jun 15 20:30:17 As per http://www.merproject.org/logs/%23sailfishos-porters/%23sailfishos-porters.2016-06-01.log.html Jun 15 20:30:24 humrobenhome: try mounting that partition as rw once and then again as ro and then reboot Jun 15 20:31:10 See from 19:27 onwards in that log Jun 15 20:31:20 That might give some hints on the issues and how to troubleshout Jun 15 20:31:22 shoot Jun 15 20:33:05 Hmmz seems system should be 27 Jun 15 20:33:08 data = 28 Jun 15 20:33:39 At least according to https://jira.lineageos.org/browse/BUGBASH-1203?focusedCommentId=20519&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-20519 Jun 15 20:33:47 But could be different ROM Jun 15 20:37:41 Thanks for the suggestions. I'll try your most up to date kernel first... Jun 15 20:38:16 Unfortunately it's trial & error with some of these things Jun 15 20:38:26 The OnePlus situation with switching OS didn't help I'm sure Jun 15 20:38:55 Oh I've built my own linux desktop several times. I know all about trial and error. Jun 15 20:39:43 But in general logs don't look too far off tbh Jun 15 20:39:52 Could be a few small fixes and it would actually give UI Jun 15 20:40:00 Then stuff like wifi, bt needs fixing Jun 15 20:40:04 And some firmware stuff Jun 15 20:40:11 But nothing too shocking by the looks of it Jun 15 20:40:35 I think we're using /data not /system. LuneOS is a chroot of /data/luneos. Jun 15 20:41:11 Yes Jun 15 20:41:28 But not sure about the Android bits where these are mounted etc Jun 15 20:41:37 LuneOS is in data for sure Jun 15 20:41:54 That's what I was saying about trying to understand exactly how the bootstrap works. **** ENDING LOGGING AT Sat Jun 16 03:00:04 2018