**** BEGIN LOGGING AT Thu Nov 24 03:00:00 2016 Nov 24 05:57:48 Seems local qemux86 build was OK, so I guess it was an issue on the build server Nov 24 07:19:53 morning Nov 24 09:27:06 Morning! Nov 24 09:27:46 Herrie|Pre3: about the headers, what I wonder is how libhybris knows about the android version we are using in the build, if not with these headers Nov 24 09:30:05 Tofe: True Nov 24 09:30:11 Where's morphis when you need him :P Nov 24 09:30:17 morphis_: ^ Nov 24 09:30:34 here :-) Nov 24 09:30:42 Hey ! The invocation worked :) Nov 24 09:30:46 hah :-) Nov 24 09:30:48 it does Nov 24 09:31:00 sorry for being a bit absent but last weeks were quite crazy Nov 24 09:31:29 No problem, apart from these hybris things the rest is going quite smoothly :) Nov 24 09:32:18 great Nov 24 09:32:23 so what is the libhybris problem? Nov 24 09:32:28 So we're in the process of trying to upgrade to CM12.1 Nov 24 09:32:37 (On mako, so far) Nov 24 09:32:49 ok Nov 24 09:33:13 I've done a build, not-so-minimal but at least without java Nov 24 09:33:39 And based on top of the mer-hybris repos + some CM additional things Nov 24 09:34:05 (without some of they boot things that would be going in our way) Nov 24 09:34:49 And now I'm wondering about our libhybris recipe, and how it currently works to know what android version is being used and what headers are involved Nov 24 09:35:29 Looks like the phablets-headers recipe is involved, but not really sure Nov 24 09:35:33 morphis_: We have https://github.com/webOS-ports/phablet-headers/ Nov 24 09:35:45 Which seems to have some Android headers, just not sure where this originates from fully Nov 24 09:35:55 The Ubuntu source link seems to have changed/disappeared Nov 24 09:36:02 I would recommend https://code.launchpad.net/~libhybris-maintainers/android-headers/+git/android-headers/+ref/master Nov 24 09:36:06 So not sure how to make something similar for CM 12.1/13 Nov 24 09:36:11 that collects the headers for different Android versions Nov 24 09:36:29 and you shouldn't need to rebuild them as those are taken from AOSP and should be stable Nov 24 09:37:04 morphis_: wow, we missed that one, it's indeed really well adapted here Nov 24 09:37:08 and https://git.launchpad.net/android-headers/tree/debian/extract-headers.sh is the script to extract those headers from a Android tree Nov 24 09:37:41 looks like it currently has APIs 19 to 23 Nov 24 09:37:47 yes Nov 24 09:38:12 I'll have to check what CM12.1 actually is :p But that's a 10min job at most Nov 24 09:38:16 as you guys are not using that much of the android versions specific things in libhybris using 19 should be still ok Nov 24 09:39:01 Shouldn't we be as close as possible to the CM version ? What is the benefit of using an older API ? Nov 24 09:40:11 (CM12.1 is API level 22 it seems) Nov 24 09:44:05 Anyway, I'll try to migrate our headers to android-headers repo, that's quite promising Nov 24 09:44:43 I've also done a CM13 build, but it seems to be a bit too recent for libhybris :) Nov 24 09:45:13 (at least from what I can see on #sailfishos-porters) Nov 24 09:46:47 Tofe: it hardly depends on what you use and if the API has changed in that area Nov 24 09:46:58 Tofe: but more or less I am wondering why you want to upgrade to cm 12.1 at all Nov 24 09:47:09 any specific reason? Nov 24 09:49:31 For mako, not really; but that would open the way for hammerhead Nov 24 09:50:28 true, but I would more go the way of putting all love then into hammerhead Nov 24 09:50:49 so basically our approach in Ubuntu is that we try to abstract as much as possible of the Android API behind our own Nov 24 09:50:52 that makes sense too Nov 24 09:51:03 so that we have a way into the container which we control and don't have depend on a specific API version Nov 24 09:51:31 and for the linker you really don't need anything of the API Nov 24 09:53:55 yes, having an abstraction looks like a good idea for this Nov 24 15:59:54 Tofe: So If I understand it correctly we'll be replacing the phablet-headers with the ones above? Nov 24 17:49:18 Herrie: yup Nov 24 18:17:52 Tofe: OK :) Nov 24 18:32:01 BTW qemux86 on Jenkins was fine after re-run, so must have been some temporary glitch Nov 24 18:32:10 ok, good Nov 24 18:40:43 Herrie: flashing rootfs... Nov 24 18:41:15 Fingers crossed :D Nov 24 18:41:47 At least the headers are now at a decent api level Nov 24 18:44:04 bootloop :( Nov 24 18:44:22 RAMDISK: Couldn't find valid RAM disk image starting at 0. Nov 24 18:46:22 I'm a bit surprised Nov 24 18:46:43 Looks like an issue with initrd somehow Nov 24 19:14:03 Tofe: Not familiar with how all this works on lower level, but I recall seeing something in the recipes a while back Nov 24 19:14:09 Let me see if I can dig that up Nov 24 19:15:39 Tofe: You have these in defconfig: CONFIG_BLK_DEV_INITRD=y Nov 24 19:15:39 CONFIG_INITRAMFS_SOURCE="" Nov 24 19:15:39 CONFIG_RD_GZIP=y Nov 24 19:15:40 ? Nov 24 19:15:47 let me check Nov 24 19:15:50 THat's what we have currently in meta-smartphone it seems Nov 24 19:15:54 Not sure what they do Nov 24 19:16:26 yup, I have all this Nov 24 19:17:33 And CONFIG_BLK_DEV_RAM=y and CONFIG_BLK_DEV_RAM_COUNT=16 and CONFIG_BLK_DEV_RAM_SIZE=4096 ? Nov 24 19:18:09 yup Nov 24 19:18:25 OK well those were first guesses Nov 24 19:18:35 That was some good ideas yes Nov 24 19:18:43 My layers are copying to my machine here nw Nov 24 19:18:45 now Nov 24 19:18:59 I'll try to force rebuilding the initrd image Nov 24 19:19:05 Not sure any of this changed with CM 12.1 due to encryption and stuff? Nov 24 19:19:14 no idea at all Nov 24 19:59:49 Tofe: I assume you've seen https://github.com/shr-distribution/meta-smartphone/tree/krogoth/meta-lg/recipes-core/initrdscripts Nov 24 20:00:17 Not sure these partitions changed in newer CM? Nov 24 20:00:52 no they shouldn't Nov 24 20:00:59 and it shouldn't affect initrd anyway Nov 24 20:06:29 OK Nov 24 20:06:57 I didn't come across anything too useful otherwise tbh Nov 24 20:54:55 I'll look at that deeper tomorrow **** ENDING LOGGING AT Fri Nov 25 02:59:59 2016