**** BEGIN LOGGING AT Sun Feb 12 03:00:02 2017 Feb 12 06:48:47 Hi! Feb 12 07:23:14 Does LuneOS have any known issues with ARMv8's 64-bit architecture? Feb 12 08:22:41 morning Feb 12 08:23:03 linux_unix-10: We haven't done a port yet for ARMv8's 64 bit architecture to be honest Feb 12 08:23:08 What issues are you having? Feb 12 08:23:23 In general we should have similar setup like SailfishOS and Ubuntu Touch Feb 12 08:34:06 Since we use the same basis Feb 12 09:33:19 Morning Feb 12 09:33:31 Herrie: do you know when http://build.webos-ports.org/luneos-testing/ipk/hammerhead/ gets synchronized? Feb 12 09:37:43 Tofe: Hmmz that should be part of rsync normally I'd say Feb 12 09:38:12 Seems it doesn't happen now for mako & Hammerhead Feb 12 09:38:17 So must be some issue in our 12.1 bits Feb 12 09:38:24 Let me have a look Feb 12 09:38:28 well, for instance sensorfw is still at git 6e5... whereas you bumped it two weeks ago Feb 12 09:38:56 Tofe: Yeah Feb 12 09:39:10 Well it seems that for both Mako & Hammerhead it didn't sync after 25th of Jan Feb 12 09:39:15 While for others it did :S Feb 12 09:43:59 what does scripts/staging_sync.sh look like? Feb 12 09:45:04 Tofe: Those are in here: https://github.com/webOS-ports/webos-ports-setup/tree/krogoth/scripts Feb 12 09:45:23 Last line in that file is commented out Feb 12 09:47:58 yes, but previous line should be recursive Feb 12 09:50:01 in http://jenkins.nas-admin.org/view/luneos-testing/job/luneos-testing_workspace-rsync/lastBuild/consoleText it's like the mako and hammerhead directories are just not there Feb 12 09:50:16 (look at MD5 sums for example) Feb 12 09:50:55 Tofe: Issue might be: http://jenkins.nas-admin.org/job/luneos-testing_update-manifest/307/console Feb 12 09:51:43 I removed hammerhead from supported machines there now Feb 12 09:51:46 Let me kick off a new build Feb 12 09:52:38 OK running Feb 12 09:53:18 Tofe: Problem is I don't have access to the build server directly Feb 12 09:53:24 Would need novaldex|away to check that Feb 12 09:54:29 http://build.webos-ports.org/luneos-testing/device-images.json seems to lack hammerhead entirely Feb 12 09:59:07 Tofe: That should be triggered by http://jenkins.nas-admin.org/job/luneos-testing_update-manifest/configure which runs https://github.com/webOS-ports/jenkins-jobs/blob/master/jenkins-job.sh and then in turn runs: https://github.com/webOS-ports/webos-ports-setup/blob/master/scripts/update-manifest.py Feb 12 09:59:15 At least if I understood correctly Feb 12 10:00:44 yes, I'm trying to figure out why the manifest is wrong Feb 12 10:03:19 Tofe: What I can think of is that the dirs aren't available with proper permissions on the build server in "Syncing deploy dir from /home/jenkins/workspace/luneos-testing/webos-ports/tmp-glibc/deploy" Feb 12 10:03:54 mmh Feb 12 10:05:14 But we'd need someone with proper access to check that... Feb 12 10:11:08 ok the issue you pointed to seems to be a consequence of the fact that the MD5 sum were not generated; and the issue seems to happens since, well, quite some time (at least June 2016), looking at the number of the last retrieved md5 sum for mako Feb 12 10:15:12 It seems the MD5 isn't generated at all in general :S Feb 12 10:15:16 It's all a bit weird Feb 12 10:15:30 I don't really touch the Jenkins bits that much usually Feb 12 10:15:41 It was all setup by JaMa mainly in the past Feb 12 10:16:03 well, they are generated for maguro, grouper, qemu... But not for mako and hammerhead, strangely enough Feb 12 10:17:03 but... then we have two separate rsync jobs ? because something is sync'ing the testing images for mako and hammerhead, after all Feb 12 10:23:37 Tofe: Well we're running the same twice Feb 12 10:23:46 In order to make sure we don't run out of space if I remember correctly Feb 12 10:23:55 ah ok :) Feb 12 10:24:39 Tofe: See http://jenkins.nas-admin.org/job/luneos-testing_build/configure Feb 12 10:25:02 Does builds, rsync, cleanup, builds, rsync Feb 12 10:25:28 ok :) Feb 12 10:26:27 Herrie: well, no wonder the md5 are gone then ;) Feb 12 10:28:46 I added an update_manifest after the first rsync too Feb 12 10:29:36 Herrie: could we restart the build job? so that the new flow is taken into account Feb 12 10:32:00 Let me check Feb 12 10:33:02 OK running Feb 12 12:32:10 Herrie: I was asking because I'm thinking of getting someone (or try myself) to port LuneOS to the Nexus 6P, which uses a 64-bit ARMv8 processor. Feb 12 12:34:25 Herrie: I also asked because Ubuntu Touch and Sailfish OS both have issues with ARMv8 (as far as I know), but Plasma Mobile doesn't. Feb 12 14:04:26 linux_unix-10: I'm a bit surprised the Plasma Mobile guys didn't have issues, as they use the same base Feb 12 14:09:46 Tofe: For the Nexus 5X port, they apparently ran 32-bit code and gradually transitioned the code to 64-bit. Feb 12 14:10:20 Tofe: At least, that's what I heard. Feb 12 14:15:40 that's also what the others did; anyway the support for aarch64 in libhybris is pretty new, and the Nexus 5X now seems to begin to behave fairlywell, and I guess it should be quite the same status for Plasma Mobile Feb 12 14:17:35 Tofe: Plasma Mobile's libhybris has support for Android 6.x, which helps a lot. Feb 12 14:19:44 linux_unix-10: it isn't the main libhybris project? (which also has Android 6 support, though still a bit experimental) Feb 12 14:21:30 Tofe: As far as I know, Plasma Mobile's libhybris was patched with better support. I think the patches were or are in the process of being merged into the main code. Feb 12 14:21:50 ah, ok Feb 12 14:21:56 maybe they're merged now :) Feb 12 14:23:20 I already tried Nexus 5 with a CM13 build; it almost worked (sound & sensors were ok), but I had issues with the hwcomposer, so I came back to CM12.1 for the moment Feb 12 14:24:39 But maybe these issues are device-specific; if you want to try a port for Nexus 6P, you're very welcome :) Feb 12 14:25:37 I wrote a little guide https://gist.github.com/Tofee/f3a65d5ea391da20817d3746f0031357 when I did the Nexus 5 port -- I'll merge it with our official guide when I'll have some time Feb 12 15:50:22 Tofe: Fix in Jenkins job seems to have worked :P Feb 12 15:50:28 http://build.webos-ports.org/luneos-testing/ipk/hammerhead/sensorfw_0.8.5%2Bgitr1%2B9efd97d298-r0.1_hammerhead.ipk Feb 12 15:53:56 I just wonder if the ignore(FAILURE) { build("luneos-testing_update-manifest", SUPPORTED_MACHINES: "mako grouper maguro hammerhead") } shouldn't be adjusted for the machine names it just built ? Feb 12 16:42:52 Herrie: good question Feb 12 17:24:48 Herrie: actually, the weird behavior I had on my build was pretty accurate: it's now in the jenkins build. It's because of the libhybris bump, from 085cd7... to a37a36... Feb 12 17:28:02 Tofe: Hmmz OK Feb 12 17:28:51 I'm now looking for the culprit commit Feb 12 17:35:43 Herrie: it's b766b8943db9a265d171417a20840f50a4c7c15d, a huge merge commit from morphis Feb 12 17:36:13 ah no this one isn't so big Feb 12 17:42:43 Well the morphis one adds Android 6/CM13 support Feb 12 17:42:51 We could do without I guess for now Feb 12 17:43:06 Since you anyway still have the issue with hwcomposer on CM13 Feb 12 17:52:45 I... don't understand the commit :p Feb 12 17:53:05 it's not a big commit, but the hooks modification is what you did a while ago Feb 12 17:54:16 Tofe: Yeah and we had that running locally already Feb 12 17:54:19 As a patch before Feb 12 17:57:56 ok, now I understand a little bit. Well, we need to solve that, anyway. Feb 12 17:58:17 I'm just glad to know what happened :) Feb 12 18:16:41 I'll ask morphis for some help tomottow Feb 12 18:16:44 tomorrow* Feb 12 18:17:03 with a commit that big, it's just not possible to see what could have gone wrong Feb 12 18:36:30 Tofe: Yeah Feb 12 18:36:42 It was a quite big one Feb 12 18:39:50 One year of dev on Ubuntu side ;) Feb 12 18:40:41 One good news is that with this newer libhybris, the sensors seem to work without my dirty patch Feb 12 18:40:57 but the UI crashes, so the compromise isn't that good! **** ENDING LOGGING AT Mon Feb 13 03:00:01 2017