**** BEGIN LOGGING AT Mon Feb 11 03:00:11 2019 Feb 11 16:49:43 Hi, I'm testing SDK created after rocko->thud upgrade. I'm generating it with 'bitbake - populate_sdk'. There are only minor changes in our layer, but for some reason a few -dev packages are missing in the target part of SDK. Feb 11 16:50:30 Namely opencv-dev, systemd-dev and poco-dev. Other (like qt dev stuff) are there. Is there anything changed in SDK configuration? Feb 11 17:36:16 tnovotny: make sure these packages are in your image against which the SDK is being generated Feb 11 17:36:26 usually this is common reason Feb 11 17:36:40 dependencies get refined over time Feb 11 17:38:59 khem: I've checked this already. Especially systemd is there for sure. It is true that opencv wasn't there (only libopencv-xxx), but even adding opencv into IMAGE_INSTALL didn't help. Feb 11 17:40:03 is there any change in complementary packages between rocko->thud? We have layer since ~dizzy, so we have some copy&paste error most probably Feb 11 17:44:23 I also saw target-sdk-provides-dummy which is new to me. Maybe I forgot to update something? Feb 11 18:07:15 are you constructing your image using some reference image from core ? if so then it should not be anything special needed I suppose Feb 11 18:10:03 khem: we have own image with quite a lot of stuff and modification. But this is a good idea - start with some image from core, add stuff and find the root cause Feb 11 19:25:27 re Feb 11 19:52:53 * pahartik still wonders why "git.openembedded.org" refuses TCP6 connections at port "9418" Feb 11 19:56:12 * pahartik referred to this problem on 2019-02-04 Feb 11 19:59:28 halstead: any ideas ^ Feb 11 21:18:51 I'm having problems with the u-boot recipe in sumo. The patch that added native python is breaking the build on ubuntu. Is anyone seeing anything similar? Feb 11 21:21:43 how does it break? Feb 11 21:22:53 Crofton: recipe for u-boot compailed about "unable to execute 'x86_64-linux-gnu-gcc': No such file or directory" Feb 11 21:24:17 https://pastebin.com/3h1Jxx8a Feb 11 21:24:44 If I remove the lines that add the native python, it's happy again Feb 11 21:25:18 even weirder, its on the last 3 (at least) versions of Ubuntu. Yet, works without a problem on Fedora Feb 11 21:42:45 pahartik, No good reason. I've added the IPv6 flag to enable. It should work now. (cc: bluelightning) Feb 11 21:43:04 Fedora is better :) Feb 11 21:43:13 pahartik, Was this reported anywhere else where I can add an update? Feb 11 21:45:05 Crofton: ha! unfortunately, me telling the people demanding I fix this to not use ubuntu is a show stopper Feb 11 21:46:29 heh Feb 11 21:47:03 I wonder why it is using the X86..lgcc? Feb 11 21:47:24 maybe unser CCACHE before uild and rename ccache binary? Feb 11 21:47:27 I'm crosscompling for arm on x86 Feb 11 21:48:15 I see in the u-boot makefile for libfdt, it unsets CC and CROSSCOMPILE. That is when it blows up Feb 11 21:48:29 I can remove this, and then I got down a different rabbit hole Feb 11 21:48:48 jonmason: This sounds really familiar.... Feb 11 21:48:53 yeah Feb 11 21:48:54 I think that recipe may be wrong. I think you need to build tools separately from the rest of u-boot. Feb 11 21:49:30 the tools need to be built as -native IIRC Feb 11 21:49:34 the u-boot recipe looks the same in thud Feb 11 21:50:05 yeah but I suspect you're building a tool that isn't normally built. I think I ran into that Feb 11 21:50:19 one sec... Feb 11 21:50:23 thanks Feb 11 21:52:05 jonmason: 3b0b16300b351878790729d6270cd113bca73eff is what I was thinking of Feb 11 21:52:32 yeah, reverting that "solves" my problem Feb 11 21:53:06 ah that again Feb 11 21:53:35 Oh, I think there might have been a bug in u-boot.... Feb 11 21:53:37 I know reverting that is not a real solution Feb 11 21:54:07 I fixed it bu upgrading u-boot, but let me see if I can remember the fix Feb 11 21:54:16 I could try the latest (instead of v2017.7) Feb 11 21:54:28 jonmason: IIRC that fixed it for me Feb 11 21:54:53 I *did* narrow down the exact problem a while ago Feb 11 21:55:17 jonmason: actually yeah I think they fixed that upstream at some point.. I might even have that commit where they did it Feb 11 21:55:23 I does appear to be your patch, so I assume you beat it to death back in july ;-) Feb 11 21:56:30 If there is a stable version of sumo/thud, it might be good to update the recipe to the working version (or are we way too late at this point?) Feb 11 21:59:29 yeah so they changed the way that stuff worked in u-boot so I was getting the same error you're getting and I needed this to fix it: https://paste.debian.net/1067103/ Feb 11 22:00:05 IIRC it worked fine if I built the version of u-boot matching the oe-core recipe. Feb 11 22:00:54 georgem: thanks! I'll give this a try Feb 11 22:02:01 the made quite a few changes in a short period of time to that Makefile so hopefully the patch applies to exactly what you're trying to build Feb 11 22:02:04 they* Feb 11 22:32:08 looks like libfdt might've been depricated atound may of 2017 Feb 11 22:32:32 " Now that this module has been accepted upstream we should stop using the local U-Boot one. In preparation for this, rename it to indicate it is for legacy use." Feb 11 22:32:48 commit 555ba4889c8ffc5bac933976db9fd915c796f72d in u-boot Feb 11 22:33:09 anyway, I'll keep digging on this but I appreciate all the help and insight! **** BEGIN LOGGING AT Mon Feb 11 23:40:41 2019 Feb 12 01:40:37 halstead: I can confirm that "git.openembedded.org" started accepting TCP6 connections at port "9418" Feb 12 01:41:20 Thank you pahartik. Feb 12 01:41:22 halstead: I have only asked about it on this IRC channel Feb 12 01:42:01 pahartik, Thanks for asking. I would not have known otherwise. Feb 12 01:42:13 halstead: Thank you for fixing Feb 12 01:42:24 Certainly :) Feb 12 01:50:13 guess we will charge more for this free service now ; ) **** BEGIN LOGGING AT Tue Feb 12 02:24:26 2019 **** ENDING LOGGING AT Tue Feb 12 02:59:59 2019