**** BEGIN LOGGING AT Mon Apr 19 02:59:56 2021 Apr 19 08:22:16 Morning! Apr 19 08:22:24 pinephone build launched locally Apr 19 09:17:27 Morning! Apr 19 09:17:35 Tofe: Seems that the modified kernel also builds inside Halium ;) Apr 19 09:17:51 Preparing the meta-webos-ports and meta-smartphone changes now Apr 19 09:51:01 Tofe: BTW I noticed that ofono-conf for Hammerhead wasn't switched to ril binder, is this on purpose or you forgot? Apr 19 09:52:26 Herrie: I forgot Apr 19 09:52:46 Tofe: OK I'll add it then Apr 19 09:52:50 thanks Apr 19 10:03:20 I guess I should also add this right? https://github.com/Tofee/android_device_lge_hammerhead/commit/5a7086365fcc14722938ff69ea1b803fd4efa7bf Apr 19 10:05:04 yes, I think persist and firmware are also needed in mako Apr 19 10:05:23 and if they're not, it's not really an issue :) Apr 19 10:06:19 OK will add those commits from Hammerhead as well then, need to add the SYSTEM_AS_ROOT as well anyway Apr 19 10:06:22 Didn't do that yet it seems Apr 19 10:23:33 OK Halium side builds, added it to the build server Apr 19 10:24:34 :) Apr 19 10:45:32 Tofe: OK building image Apr 19 10:54:11 memnotify isn't happy, disabled it for now Apr 19 10:55:23 Same for kernel-module-wlan Apr 19 10:57:07 you reuse the 5.1 defconfig ? Apr 19 10:57:18 Yes Apr 19 10:57:21 ok good Apr 19 10:57:38 Ah wait maybe not :P Apr 19 10:59:33 Seems I somehow didn't commit it or to wrong branch Apr 19 10:59:34 Retry Apr 19 11:01:25 Hello there Apr 19 11:01:30 Jonek: Hi! Apr 19 11:01:54 Do you have an idea if eSIM handling implementation is google proprietary or open in aosp? Apr 19 11:03:10 On my side, I don't know Apr 19 11:16:08 Jonek: Also no idea to be honest Apr 19 11:17:33 On way to know would be to find a lineageos supported device with esim Apr 19 11:17:38 One* Apr 19 11:17:48 Tofe: Kernel is failing on https://github.com/Herrie82/android_kernel_lge_mako/blob/halium-9.0/kernel/pid_namespace.c#L279 Apr 19 11:18:07 "/home/herrie/LuneOS/gatesgarth/webos-ports/tmp-glibc/work-shared/mako/kernel-source/kernel/pid_namespace.c:279:21: error: 'struct pid_namespace' has no member named 'user_ns'" Apr 19 11:19:02 Should I revert to the easier pidns_install which we use for Hammerhead? Apr 19 11:20:32 https://github.com/shr-distribution/linux/blob/3a27d83814c6f419f0b21ee5fb17434d24bd9abe/kernel/pid_namespace.c#L269-L277 ? Apr 19 11:21:53 Seems LOS supports it: https://review.lineageos.org/c/LineageOS/android_packages_apps_SetupWizard/+/305725 Apr 19 11:22:18 Though only very recently? Apr 19 11:22:38 https://review.lineageos.org/c/LineageOS/android_device_google_muskie/+/234542/2 Apr 19 11:25:35 priv-app/EuiccGoogle/esim0.img Apr 19 11:25:35 -priv-app/EuiccGoogle/EuiccGoogle.apk;PRESIGNED Apr 19 11:25:57 that means that there is closed-source pre-packed google app, right (the 2nd row)? Apr 19 11:26:20 how about .img file in android sources? what does it mean? is it vm or what Apr 19 11:30:15 Jonek: I'm not sure, you'd need to inspect that .img to see what's inside Apr 19 11:30:22 I guess some small file system/os Apr 19 11:35:38 Tofe: OK compiles now after I fixed another defconfig issue + pidns_install from Hammerhead Apr 19 11:43:21 Tofe: Uploading image ;) Apr 19 12:10:21 Herrie: great ! I am still building qtwebengine -- of course Apr 19 12:11:27 Jonek: the .img looks like a firmware, I'd say. Uploaded to the eSIM chip, like a dedicated little OS Apr 19 12:11:58 thats what I imagine, but its still a bit suprising to have it inside linaege Apr 19 12:12:30 I thought eSIM is a bit more closed and its not that easy to just upload a firmware into it Apr 19 12:13:09 +Tofe how long does it take for you to build qtwebengine and whats your buildmachine? Apr 19 12:14:04 Jonek: it's 2-3 hours, with my i5-6600 CPU with SSD and 16GB ram Apr 19 12:14:38 k, I have same timing on my laptop Apr 19 12:15:00 qt builds longer than the rest of the system.. Apr 19 12:15:19 yes, far longer. qtbase take its time too Apr 19 12:20:02 Tofe: http://build.webos-ports.org/halium-luneos-9.0/luneos-dev-package-mako-0-20210419.zip Apr 19 12:20:12 Herrie: thanks ! will test tonight :) Apr 19 12:22:49 Jonek: On my (now already dated) builder it's a lot quicker, but yes it takes quite some HD, memory and CPU ;) Apr 19 12:34:43 Herrie: not sure what I'll upgrade to next time; The i5 choice was based on a TDP compromise, but AMD has improved a lot also in this area Apr 19 12:36:47 well if I could spend some extra money I would probably got for x5950 :) Apr 19 12:37:27 I think you cant get anything comparable without going to super-expensive servers Apr 19 12:47:53 Jonek: I got a 2990WX 2,5 years ago... Apr 19 12:48:00 128GB RAM, 2* 1TB M.2 SSD ;) Apr 19 12:48:17 And 10TB "regular disks" in raid mirror for some storage Apr 19 12:49:45 nice Apr 19 12:49:57 and here I come, with my lenovo x1 carbon laptop... :D Apr 19 12:50:08 Jonek: I have the same for work ;) Apr 19 12:50:22 And an old one too ;) Apr 19 14:06:34 Tofe: Now Halium build was failing on kernel, but seems I just need to disable apparmor like you did for Hammerhead ;) Apr 19 16:45:50 Herrie: pinephone works just fine with gatesgarth; no bluetooth, but maybe it was the same before Apr 19 16:46:08 Tofe: Ah nice! Apr 19 16:46:19 now flashing mako Apr 19 16:46:21 But this is without any updates right? Or you updated kernel already? Apr 19 16:46:34 I didn't change anything Apr 19 16:46:39 Tofe: Mako kernel build for Halium fails, not really sure yet what it is Apr 19 16:46:45 I thought apparmor, but seems something else Apr 19 16:47:13 can you paste the output ? maybe I recognize it Apr 19 16:49:01 https://paste.ubuntu.com/p/8G7f6BFT3f/ Apr 19 16:57:17 Herrie: https://bpa.st/RRPA Apr 19 16:58:15 Herrie: it's crypto, probably something we can disable Apr 19 16:58:33 ah no, it's arch/arm/kernel/sys_oabi-compat.c, ok Apr 19 16:59:44 can you disable CONFIG_OABI_COMPAT ? Apr 19 17:05:05 Tofe: someone submitted to me some results from R9 5950x, but it wasn't very impressive for the price, but it had only 32G ram, so I guess it was in swap quite a lot as well, so maybe it could do better Apr 19 17:06:56 and they submitted only the build with 32 BB_NUMBER_THREADS, which harmed the time significantly as well with this ram Apr 19 17:09:13 JaMa: that's way too many threads for 32GB Apr 19 17:10:12 Herrie: for the mako boot issue, it could be a missing piece in the kernel, which is needed by systemd when mounting the /dev,/sys and /proc filesystems Apr 19 17:17:17 Tofe: Well it doesn't look that bad to start with ;) Apr 19 17:21:03 Herrie: might be related to fd_is_mount_point in systemd... we had a patch for this kind of thing in our old kernels, right? Apr 19 17:21:42 Ehm we had some vfs patch I think Apr 19 17:21:59 right Apr 19 17:22:56 https://github.com/shr-distribution/linux/commit/1ad88b041787d8ce8407a021271ef1031e95cba6 this one Apr 19 17:24:24 https://github.com/Herrie82/android_kernel_lge_mako/blob/halium-9.0/fs/stat.c#L61 mmmh let's use the raw version Apr 19 17:26:12 Tofe: Yeah I wasn't sure there Apr 19 17:26:18 I saw the light so let it for now Apr 19 17:26:21 You think raw is better? Apr 19 17:26:41 yes Apr 19 17:27:17 or, at least, it'll work better; iirc the "light" version has been abandonned later on Apr 19 17:27:57 Should I drop the &fput_needed? Apr 19 17:28:01 So it becomes struct file *f = fget_raw(fd); ? Apr 19 17:28:50 https://git.ustc.edu.cn/chendotjs/linux/-/commit/d228dd20d4c968a688b29cf59d2450c2bdcac018 see that commit for instance Apr 19 17:28:55 Or I'll just copy the whole function? vfs_fstat Apr 19 17:29:41 for the moment, let's just use fget_raw_light instead of fget_light Apr 19 17:29:57 if that doesn't work, we'll copy the whole method with fget_raw Apr 19 17:30:37 PS: I just need the fastboot image to test this, don't bother building the rest Apr 19 17:31:33 Tofe: OK Apr 19 17:38:02 Tofe: http://build.webos-ports.org/halium-luneos-9.0/zImage--3.4.113+gitr0+05e937c217-r0-mako-0-0.fastboot Apr 19 17:40:49 better, I have adb Apr 19 17:41:20 lxc seems to fail though Apr 19 17:41:26 but it's already quite a nice start! Apr 19 17:41:55 Yeah not bad for now Apr 19 17:42:02 It could be I need to tweak some more defconfig Apr 19 17:42:06 Or it's something Android side Apr 19 17:43:07 ah, proc_ns_follow_link, NULL pointer, reminds me something I did for hammerhead... Apr 19 17:44:17 Yeah Apr 19 17:45:43 I think it's fixed by the commits I did https://github.com/Tofee/android_kernel_lge_hammerhead/commits/hammerhead/3.4/halium-9.0 Mars 8th Apr 19 17:45:46 This one? https://github.com/shr-distribution/linux/commit/b70a94373212e69f928721c7b8efb2d612f548a8 Apr 19 17:47:33 I was more thinking about the ones on PID namespace Apr 19 17:48:00 Ah OK Apr 19 17:48:04 Yeah seems that might be the one Apr 19 17:49:45 Tofe: I'm out for most of the evening, I sent you an invite for my Mako kernel repo. Feel free to cherry pick some things there Apr 19 17:50:16 ok thanks ! I'll see if I have the time to tinker with it Apr 19 17:52:43 Otherwise I'll pick it up later Apr 19 18:10:27 Herrie: can you just paste or push your latest meta-smartphone? Apr 19 18:10:46 (for the kernel) Apr 19 18:10:58 Let me do that shortly Apr 19 18:11:03 I have it locally Apr 19 18:25:21 Tofe: https://github.com/Herrie82/meta-smartphone/commit/a5bd959d76ed25cf564afd79282bbe094e9fc75b Apr 19 19:37:22 Herrie: thanks, building Apr 19 20:41:56 Herrie: I've pushed some commits to the kernel, mainly taken from hammerhead Apr 19 20:42:05 now the android container almost starts well Apr 19 20:42:44 we still need libnetutils in the halium build, and then it might just be ok Apr 19 20:43:53 c.f. https://github.com/Tofee/android_device_lge_hammerhead/commit/8f667b95c07cd1a49c7ab6342b6455da73d4af8d#diff-247ee86229a709a6e2eedfc1d3c4a557825aee073e20b0112ab76f4ca8e4bc4eR129 and maybe https://github.com/Tofee/android_device_lge_hammerhead/commit/8f667b95c07cd1a49c7ab6342b6455da73d4af8d#diff-247ee86229a709a6e2eedfc1d3c4a557825aee073e20b0112ab76f4ca8e4bc4eR129 Apr 19 20:44:17 that's all for tonight for me :) I think mako will be quick to converge, hopefully **** BEGIN LOGGING AT Mon Apr 19 21:08:15 2021 Apr 19 21:20:52 Tofe: great! Apr 19 21:21:07 I wasn't sure about libnetutils so didn't include it yet Apr 19 21:22:14 Will do a fresh Android build tomorrow and push it **** ENDING LOGGING AT Tue Apr 20 02:59:56 2021