**** BEGIN LOGGING AT Sat Mar 09 03:00:34 2019 Mar 09 06:04:47 Morning! Mar 09 06:04:57 Tofe: Interesting work on the mainline kernel bits :) Mar 09 06:05:33 Maybe silly question, but did you update the defconfig for the Halium values? We'd still need most of those since we're using systemd, lxc etc Mar 09 07:17:18 Herrie: I've taken the defconfig of postmarketos's plasma mobile for mainline hammerhead Mar 09 07:17:27 and no, there's no lxc anymore :p Mar 09 07:35:27 Tofe: AH OK. Yeah PMOS should work normally Mar 09 07:35:53 We might need LXC later again for Anbox ;) Mar 09 07:37:54 Yes, right Mar 09 07:38:01 well that'll be another story... Mar 09 07:38:23 True Mar 09 07:39:02 Curious to see how the mainline will work. Tenderloin mainline seems to be an option too, just like Mako Mar 09 07:39:16 Could kill 3.4 altogether :P Mar 09 07:40:10 Well elvispre's Onyx is also on 3.4 Mar 09 08:11:30 Right now my main issue is that I can't see how far the initrd went Mar 09 08:12:00 I have a 3.4-based recovery (i.e. no pstore) and a 5.0 kernel (i.e. no last_kmsg).... yay Mar 09 10:46:28 Tofe: That's not very helpful Mar 09 10:51:16 I'm sure the PMOS guys found some ways to debug? Mar 09 10:51:29 Or maybe they have some custom recovery Mar 09 10:52:11 Kernel version shouldn't matter for pstore vs last_kmsg, this was Android version related, not kernel specific I think? Mar 09 10:55:29 android began to use pstore after 3 Mar 09 10:55:31 3.4 Mar 09 10:55:48 last_kmsg is android only Mar 09 11:02:56 Ah ok Mar 09 11:24:03 bshah: ping Mar 09 11:35:11 Tofe: nothing in /var/log such as boot.log? Mar 09 11:36:42 nope, I don't even know if initrd is started Mar 09 15:27:20 That's not very helpful indeed Mar 09 15:27:31 There must be some ways to debug I'd say Mar 09 15:27:51 You're for sure not the first one Mar 09 15:30:30 I may have found something: maybe there needs to be a padding between the zImage and the dtb Mar 09 15:30:54 currently I just do a dumb "cat zImage dtb > zImage-dtb" Mar 09 15:44:10 Tofe: You should be able to compare with our working fastboot image Mar 09 15:44:22 And get an idea of the padding Mar 09 15:45:16 I'm comparing with postmarketos, and they use a padding of 2048 Mar 09 15:45:38 Their mkbootimg tool also seems more up to date, and has a dtb option Mar 09 15:49:54 We could use theirs in our Yocto Mar 09 15:50:22 it's what I'm doing now Mar 09 15:51:01 but it seem to produce a non-bootable kernel... Mar 09 15:58:54 or should I upgrade my N5 to Android 6 ? does it bring an updated bootloader ? Mar 09 16:00:41 nope, didn't help Mar 09 16:01:23 that's strange. Mar 09 16:02:09 Maybe another recovery version somehow Mar 09 16:02:12 , Mar 09 16:02:26 only fastboot is involved here Mar 09 16:02:27 PMOS guys usually document things properly Mar 09 16:03:33 Can you boot their image? Mar 09 16:03:40 That would be a good test Mar 09 16:03:59 And see if you can use some tools to compare properties of both images Mar 09 16:04:08 That would be my approach Mar 09 16:04:18 I don't have theirs, but I could boot the one from flto Mar 09 16:05:01 I compared the two images: https://bpaste.net/show/e157b1a5b3b1 Mar 09 16:05:39 I'll answer your next question: yes, I've tried flto's cmdline :) Mar 09 16:06:40 but here when I say "it doesn't boot", it's pretty severe: it blocks fastboot completely, before even starting the kernel Mar 09 16:07:16 with my non-padding and our good ol' abootimg, it start the kernel, at least Mar 09 16:48:48 Weird Mar 09 16:49:04 I noticed quite some size difference in ramdisk Mar 09 16:49:27 yes, he included all the firmwares in the world :) Mar 09 16:52:34 LOL Mar 09 17:24:43 Tofe: do you remember details behind systemd patch: Mar 09 17:24:44 commit e6c535ba5c270f9e594df53b672eeeda083fbd6c Mar 09 17:24:44 Author: Christophe Chapuis Mar 09 17:24:44 Date: Sat Apr 30 18:53:20 2016 +0200 Mar 09 17:24:44 fd_fdinfo_mnt_id: disable fdinfo stat Mar 09 17:24:46 ? Mar 09 17:25:44 the meta-webos-ports commit which added it is actually from Herrie|Phone|2 with "Since it causes boot loop issues on some targets (Mako)" Mar 09 17:42:04 JaMa: it's because fd_info caused problems on 3.4 kernels Mar 09 17:42:36 It might be that, with our brand new fstat fix, this patch isn't needed anymore. But I can't be sure. Mar 09 18:40:48 Herrie: if I'm correct about https://github.com/Halium/android_kernel_lge_hammerhead/blob/halium-5.1/arch/arm/boot/Makefile#L67 , then it seems that the kernel itself does "cat" instead of padding Mar 09 18:50:41 JaMa: It's easy enough to test without this patch Mar 09 18:50:48 Boot/not boot is easy to establish ;) Mar 09 18:50:59 Pretty sure it was for systemd at some point Mar 09 18:51:05 yes Mar 09 19:02:11 Tofe: Indeed looks like it Mar 09 19:02:16 Re cat for kernel Mar 09 19:39:15 oh, damn, I just understood... They don't use a real mainline kernel: https://gitlab.com/postmarketOS/linux-postmarketos/commits/postmarketos-linux-qcom Mar 09 21:21:31 with their kernel, at least I get until initrd it seems **** ENDING LOGGING AT Sun Mar 10 02:59:56 2019