**** BEGIN LOGGING AT Mon Dec 07 02:59:57 2020 Dec 07 05:39:28 exit Dec 07 06:30:36 Morning. I'm having trouble building python3-setuptools-native on dunfell. I get "Illegal instruction (core dumped)" which I've debugged down to line 69 in random.py (of cpython) (RECIP_BPF=2**-BPF). Anyone seen anything similar? Using Ubuntu 20.04 as host. Dec 07 07:02:54 good morning Dec 07 07:41:24 yo dudX Dec 07 08:06:42 RP: want a nice goodmorningly surprise? Dec 07 08:07:28 * LetoThe2nd could offer a pseudo.log.... namely, a 98GiB large one? Dec 07 08:18:36 it seems to be some kind of infinite depth recursion with esdk+buildhistory+a specific directory layout. will investigate. Dec 07 08:22:47 hello all, I'm very new to yocto. after following the steps on this pages https://variwiki.com/index.php?title=Yocto_Build_Release&release=RELEASE_SUMO_V1.1_DART-6UL Dec 07 08:22:57 I'm not able to compile uboot :( Dec 07 08:23:08 bitbake u-boot-variscite -f -c compile Dec 07 08:23:36 Sorry if this isn't the correct forum, but really need the help Dec 07 08:24:43 can you advise where is the best place for asking help ? Dec 07 08:26:03 simon: If you're following the setup instructions from Variscite then you probably need to ask them for support Dec 07 08:26:23 The sumo release of Yocto is now EOL, if you can upgrade to something newer I'd recommend doing so Dec 07 08:26:52 The only thing I can think to check off the top of my head is your host distro version. If you're using something new like Ubuntu 20.04, the sumo release of Yocto might not support that Dec 07 08:27:34 Thanks paulbarker, I'm asking Variscite as we chat :) Dec 07 08:27:44 cheers :) Dec 07 08:36:58 * LetoThe2nd ponders slapping variscite for the -f flag anyways. Dec 07 08:53:31 LetoThe2nd: Good morning. I think I can manage without, thanks ;-) Dec 07 08:54:03 hehe Dec 07 09:05:58 seems qemu-5.1.0-r0 do_package_write_deb has an issue with Subprocess output:dpkg-deb: error: package name has characters that aren't lowercase alphanums or '-+.' Dec 07 09:07:50 OutBackDingo: seems odd nobody else is hitting that, what configuration do you have which is different? Dec 07 09:09:59 OutBackDingo: Could you try with the rpm or ipk package manager? See if it's reproducible with either of those Dec 07 09:15:23 paulbarker: i added meta-browser meta-clang and meta-rust, added chromium to the build Dec 07 09:16:08 building with package_rpm now to test Dec 07 09:31:14 paulbarker: it seems that a directory recursion is created if on is using kas inside the layer to build an esdk Dec 07 09:32:35 LetoThe2nd: Could you file a bug in https://github.com/siemens/kas/issues? Dec 07 09:33:22 paulbarker: i could, but its not really a kas issue in itself i think. its more like, being triggered by a practise kas encourages. Dec 07 10:02:38 RP: Looks like https://git.openembedded.org/openembedded-core/commit/?id=cc30537bb4d0815c4d69f0b1a1a8a200085902d5 breaks building inside a layer directory Dec 07 10:03:46 When using kas this is especially common Dec 07 10:05:17 paulbarker: hmm, I wasn't particularly happy about doing this either :/ Dec 07 10:05:21 Saur: ^^^ Dec 07 10:05:22 paulbarker: do you know if that went into dunfell? Dec 07 10:05:34 LetoThe2nd: it shouldn't have done yet Dec 07 10:05:40 (anyways it is certainly not of relevance to my particular problem) Dec 07 10:06:17 i'd guess the situation of building the esdk inside the layer has been broken ever since, but really, just guessing! Dec 07 10:07:53 paulbarker: may have to revert for now? Dec 07 10:09:44 RP: I don't understand the issue it's trying to fix Dec 07 10:10:11 Let me look back at the emails and see what I think Dec 07 10:10:25 I don't like suggesting reverts for things if I don't understand why they were added Dec 07 10:29:21 RP: I've replied to Peter's patch, copying in you and the list Dec 07 10:33:38 For now I'll do a local revert of that patch Dec 07 10:48:59 paulbarker: seem package_rpm works fine.... so whats the breakage in package_deb Dec 07 10:50:10 package_deb is always a bit broken as nobody uses it Dec 07 10:50:12 OutBackDingo: So others here may disagree with me but in my experience package_deb is not as well tested as rpm or ipk. I recommend switching over to one of those Dec 07 10:51:58 If you want to continue using package_deb (which, really, I don't recommend) then you'll need to modify the qemu recipe so the package names comply with the debian rules (uppercase letters and certain punctuation characters are banned) Dec 07 10:51:59 OutBackDingo: do you know the package name its breaking on? Dec 07 10:52:07 Patches welcome Dec 07 10:52:33 RP: I think it's this email: https://lists.yoctoproject.org/g/yocto/message/51667 Dec 07 10:52:35 So qemu Dec 07 10:52:36 paulbarker: we really should put a sanity test in core so all packages "obey" the rules Dec 07 10:53:29 paulbarker: package name, not recipe name... Dec 07 10:55:07 ah, it is there, qemu-x86_64 Dec 07 10:55:33 so its the "_" in that name Dec 07 10:56:07 which layer bbappends qemu to split per arch? or do we not test building target qemu? Dec 07 10:57:13 RP: Controversial opinion: I'd rather get rid of package_deb then we don't have to worry about those rules. But I understand some folks do use it so that's probably not viable Dec 07 10:57:32 paulbarker: not an option and it does basically work Dec 07 10:57:58 paulbarker: why would you recommend package_rpm over package_deb ? Id prefer to stay on the debianish side of the river.... Dec 07 10:58:20 sicne the whole premise of overc appears to be based off minimal debian Dec 07 10:58:32 OutBackDingo: of the three backends, its the least tested/used Dec 07 10:58:48 OutBackDingo: Because it works and it's well tested :) But sarcasm aside we should fix this one Dec 07 10:58:52 welp its about to get a ton of testing :) Dec 07 10:59:24 also a lot of fixing? Dec 07 10:59:28 yes, something as simple as _ in a name should be resolved Dec 07 10:59:40 LetoThe2nd: obviously ill fix whatever im capoable of Dec 07 11:00:20 im still wrapping my head around the whole overc "distro" but getting more in depth Dec 07 11:06:43 OutBackDingo: unless you're using the package manager on the target, the choice of package manager is literally just a choice of what format do you want intermediate files to be in. Dec 07 11:09:42 rburton: hey, don't blame meta-ros for xilinx stuff :) Dec 07 11:11:16 rburton: i do use the package manager on the target Dec 07 11:15:29 OutBackDingo: i suggest you work out what broke deb and how the automatic QA can detect that in the future then :) Dec 07 11:16:12 JaMa: the fortran bug? i blame everyone for fortran Dec 07 11:16:36 JaMa: in my defence i said it could be any number of things Dec 07 11:42:22 rburton: yes, it could be and with the information he shared in meta-ros and bugzilla it's difficult to debug, I'm just saying that I'm building meta-ros with fortran enabled couple times per day and haven't seen his issue on x86,x86_64,arm,aarch64 Dec 07 11:42:54 and I've also built it with xilinx BSP a while ago, but only the BSP layer not their poky based distro Dec 07 11:48:24 it does seem odd. i'm still 50/50 on user error: the tweak to turn on fortran is minimal and hard to do wrong Dec 07 13:04:29 rburton: agreed, and meta-ros does just what local.conf.sample.extended does, because it needs fortran for some of the recipes (including lapack from meta-oe) Dec 07 13:35:49 rburton: one of his logs shows that gcc-cross is from locked sstate, I think that's the reason why fortran never got really enabled in his build Dec 07 13:36:26 JaMa: yes good catch! Dec 07 16:16:46 Argh, I love how my ISP will block connected for 10 seconds if I open too m any too quickly.. RIP do_fetch.. Dec 07 16:17:04 (luckily there is do_fetch[number_threads] = "..." so I can limit it Dec 07 17:08:53 sakoman: what rburton's fix to find more CVE does with the raffle? does it mean that Ross now needs to grow something in his garden as well? :) Dec 07 17:09:29 it looks like he grows some dirty bikes mostly :) Dec 07 17:09:55 JaMa: it just means there are more targets to enter fixes for :-) Dec 07 23:11:45 JaMa: before posting the patch to oe-core i fixed the CPEs of 8 that don't appear in the report just days before the raffle was announced. if i'd known I'd have waited ;) Dec 07 23:12:33 :) **** ENDING LOGGING AT Tue Dec 08 02:59:57 2020