**** BEGIN LOGGING AT Thu Jun 08 03:00:02 2017 Jun 08 04:02:08 hello all Jun 08 04:03:52 has anyone ran into a bitbake error about missing .h files that openssl needs? Jun 08 04:20:01 hello all? Jun 08 04:49:17 pastebin error Jun 08 04:49:22 so folks can see Jun 08 04:55:02 nemequ: ok, then I guess your host has issues you have packages compiled with old ( gcc 4.8 ) and new gcc 5.4+ compiler installed together, c++ std ABI has changes betwween these two versions Jun 08 04:55:43 I would suggest to check the libraries which are complaining about missing symbols as to who is providing them on host and if they are compiled using what compiler Jun 08 05:24:31 zeddii: the mips gcc7 fix needs to be applied to edgerouter branch of linux-yocto Jun 08 05:25:02 can you send a patch ? gcc7 patch is held up for just this change alone Jun 08 06:25:13 hi, Im having a issue with the populate_sdk. when i run the populate_sdk i get a check_data_file_clashes. : Package gdb-cross-canadian-arm wants *****: Package gdb-cross-canadian-arm wants Jun 08 06:26:11 it is erroring over all the files in /sysroots/x86_64-arago-linux/usr/share/gdb/system-gdbinit/ , /sysroots/x86_64-arago-linux/usr/share/gdb/syscalls/ and /sysroots/x86_64-arago-linux/usr/bin/arm-oe-linux/ Jun 08 06:35:19 morning all Jun 08 06:49:15 Good morning, i was wondering if anyone knows how to run a bootstrap file from a recipe? Jun 08 07:11:51 hi everyone Jun 08 07:12:45 I would like to know if there is a simple way to analyze a coredump with the sdk or the build environnement Jun 08 07:13:34 I don't know where to search any documentation on this Jun 08 07:15:04 no experience with this too, but i'd expect that using the sdk env source file should basically do the trick Jun 08 07:15:16 then you should get the required arch gdb Jun 08 07:16:21 thanks I will try the sdk Jun 08 07:50:25 Hi there. We are building a BSP for a custom board. We have two HW revisions (B-sample and C-sample) which are quite identical and I can break it down so that the only difference is in the device tree. What is the way to go to get two different images one for the A-sample and one for the C-sample? Jun 08 07:50:31 Thx Jun 08 07:51:28 Hi, how can I redo a build of a specific package ? The bitbake was interrupted in the moment an object file was generated, and now i have a message of truncated file. Jun 08 07:51:30 Thanks ! Jun 08 07:52:15 ChrysD_: I would do bitbake -c clean recipename in that situation Jun 08 07:52:33 bluelightning : Ok thanks Jun 08 07:52:37 Patrick___: depends a bit. you can either have the bootloader select different dtbs (u-boot can do that from FIT image, for example), or sometimes it is possible to strip out the changes into a dtb overlay that you can select in userspace Jun 08 07:56:10 LetoThe2nd I would rather go with two separate images than doing the decission in the bootloader or with a overlay, since this is only a temporary solution Jun 08 07:56:21 bluelightning : It worked thanks. Jun 08 07:56:39 bluelightning : Is it something linked to sstate ? Jun 08 07:58:02 Patrick___: ah so your actual question was how to do multi-machine builds in OE? ;-) Jun 08 07:59:10 Yes, sorry for the bad description Jun 08 07:59:50 Patrick___: no problem. :-) IIRC there are some possibillities, but i think it is not perfect... Jun 08 08:00:05 RP1: ping concerning multimachine-builds ^^^^^^ Jun 08 08:00:34 bluelightning : don't bother to answer, i'm looking the manual. thanks btw. Jun 08 08:00:57 Patrick___: i personally would just do two sepearate builds sharing the sstate-cache, but that my .02$ Jun 08 08:04:46 ChrysD_: it's OK - it's not in this case, just an unfortunate artifact of the file being written out and then make assuming it's OK Jun 08 08:09:35 LetoThe2nd, Patrick___: Using multiconfig and two MACHINE values would allow you to build two different machines in one build Jun 08 08:10:25 bluelightning : yeah i see that do_clean just remove outputs of the workdir. So presumly, it should change a kind of " bitbake build history" to let know that this task need to be done i guess. Jun 08 08:11:14 thx, i will give multiconfig a try Jun 08 08:15:11 ChrysD_: well, I'm not sure there's a perfect mitigation for this kind of problem Jun 08 08:15:17 RP1: i remembered we discussed it :-) but you know my brains getting older and older ;-) Jun 08 08:16:02 LetoThe2nd: applies to us all :) Jun 08 08:16:06 ChrysD_: we could say just clear out ${B} if do_compile is interrupted, but then you would be forced to sit through all of the compile on the next build even if there was no corruption Jun 08 08:20:05 bluelightning : yeah and for me it's logical to redo the steps Jun 08 08:20:13 bluelightning : to avoid further problems Jun 08 08:23:30 bluelightning : BTW, i don't really understand when you say : well, I'm not sure there's a perfect mitigation for this kind of problem Jun 08 08:24:10 well some people like do_compile being able to be interrupted and then have it pick up where it left off rather than restarting from scratch Jun 08 08:24:28 I guess you could have bitbake do something special on force interruption Jun 08 08:24:39 bluelightning: we need a command line switch: --slack=true or false Jun 08 08:24:55 bluelightning: if true, clean out {B} for more time to slacking off Jun 08 08:25:11 --chair-swordfights=true Jun 08 08:25:13 ;) Jun 08 08:25:34 exactly. Jun 08 08:25:53 bluelightning : it's like if we aren't speaking about the same thing. I feel like still a big newbie in this ecosystem. Jun 08 08:27:10 ChrysD_: at the moment, a force interruption of a task isn't treated specially, it's the same as if it failed on its own Jun 08 08:27:27 oO( interruptions in the force ) Jun 08 08:27:57 and interrupting do_compile doesn't always result in corrupted files being written out, I'd expect that's actually quite a rare occurrence Jun 08 08:28:03 annoying when it does happen though I agree Jun 08 08:28:22 bluelightning : my feeling is that i don't know what I say that make you telling me that. Do you understand where is my misunderstanding ? As if we started to speak about something and then I don't see what are the link between what i said and what you say. So presumly, i don't understand at all what 'im doing ahah. Jun 08 08:29:38 ChrysD_: it's a problem I am familiar with so perhaps I am unloading a lot of detail in my responses, apologies for that Jun 08 08:30:09 bluelightning : I guess the responsability came also from me. That's why i said i will look into manual. I just was wondering what the -c clean is doing. Jun 08 08:30:59 well -c clean deletes the work directory and all stamps for the recipe's tasks (stamps being the files we write out to tell bitbake that tasks have already successfully completed) Jun 08 08:31:11 that should hopefully be mentioned in the manual somewhere Jun 08 08:31:20 bluelightning : yeah its written Jun 08 08:31:25 ok good Jun 08 08:31:41 bluelightning : that's why I said : [10:10] bluelightning : yeah i see that do_clean just remove outputs of the workdir. So presumly, it should change a kind of " bitbake build history" to let know that this task need to be done i guess. Jun 08 08:32:04 bluelightning : So i didn't understand after what you are seing linked to the fact of my curiosity to "how" -c clean works ahahah Jun 08 08:32:50 what I launched into was an explanation of why we don't have a fix for this problem, rather than what -c clean does Jun 08 08:33:31 the "bitbake build history" presumably is the stamps, but in this situation it's not the presence of stamps causing the problem - the failing task isn't stamped as having been done Jun 08 08:34:19 bluelightning : ahhh ok Jun 08 08:34:33 bluelightning : I wasn't seing that there is a problem ahah Jun 08 08:34:54 bluelightning : For me the problem was me by interruping the process and not because of a bitbake problem. Jun 08 08:35:08 ultimately if there's an underlying fault here it's that the combination of the compiler and OS write out a junk file instead of atomically writing either a complete file or no file at all Jun 08 08:37:19 blueligthning : that's why there is a mismatch between what you say and what i say. I wasn't pointed that there is a problem. Just to know how to redo a recipe. And then curious of how -c clean works so that bitbake restart from there ahah. But thanks for the patience. Jun 08 08:38:23 ChrysD_: ok, no worries :) Jun 08 08:40:02 bluelightning : The only things i would remember, is that there are some interruptions in the force. Leto ;) Jun 08 08:40:41 :D Jun 08 09:37:53 hi, Im having a issue with the populate_sdk. when i run the populate_sdk I get: check_data_file_clashes. : Package gdb-cross-canadian-arm wants *****: Package gdb-cross-canadian-arm Jun 08 09:40:50 MarcWe: since when? which layers/machine/image? something we can look at? Jun 08 09:45:51 I'm on a oder release krogoth Jun 08 09:46:24 LetoThe2nd: db-cross-canadian-arm wants to install file *****sysroots/x86_64-arago-linux/usr/share/gdb/system-gdbinit/elinos.py sysroots/x86_64-arago-linux/usr/share/gdb/system-gdbinit/elinos.py Jun 08 09:48:03 it has multiple issues with all files in: sysroots/x86_64-arago-linux/usr/bin/arm-oe-linux-gnueabi/ ,,,,,,,, /sysroots/x86_64-arago-linux/usr/share/gdb/syscalls/ ,, and,, /sysroots/x86_64-arago-linux/usr/share/gdb/system-gdbinit/ Jun 08 09:48:21 MarcWe: sure that you are not mixing up layer revisions or such? or "incompatible" layers? Jun 08 09:49:39 everything is on krogoth Jun 08 09:50:06 MarcWe: which layers, then? Jun 08 09:50:44 Hi Jun 08 09:51:05 LetoThe2nd: but i have some layers of my own Jun 08 09:51:13 I'm trying out the testing framework of yocto and i'm getting a strange error: https://da.gd/vSePw Jun 08 09:51:31 MarcWe: well my personal attempt would be trying to deactivate layers one after the other to see what clashes. Jun 08 09:51:39 it's strange cause if the test issues a `reboot` command on the target, the ssh connection is expected to break Jun 08 09:51:46 MarcWe: but that might also be a bad idea. Jun 08 09:51:55 just the only thing that comes to my mind) Jun 08 09:52:43 LetoThe2nd: oke, p.s. if found that if i edit oe-core/meta/recipes-core/packagegroups/packagegroup-cross-canadian.bb and comment out the binutils it wil build Jun 08 09:52:53 but then i have sdk issues Jun 08 09:53:34 MarcWe: no surprise. probably something in your non-core layers just also tries to insert the binutils, and clashes. the point is to find out, who and why. Jun 08 09:54:23 a ok so i need to finde the other layer that incluedes binutils to reslolve the clashe Jun 08 09:54:51 MarcWe: basically, yes. IMHO Jun 08 09:55:35 tnx now i know that I'm on the ride track Jun 08 10:52:07 I'm trying to setup a SCons build with my SDK. Programs build fine, but they have the wrong interpreter set. Jun 08 10:52:31 Any idea what environment variable I may have misconfigured? Jun 08 11:14:00 Dear all, we are looking for gstreamer expert in my company and i would like to know if someone knows a mailinng list in yocto community for this kind of request ... Also if you are interested, please let know in pm Jun 08 11:17:17 dabrain34 : you will have maybe a better answer in #gstreamer Jun 08 11:18:59 ChrysD_>We already also asked but its good also to enlarge the circle :) Jun 08 11:19:12 ChrysD_>We already asked but its good also to enlarge the circle :) Jun 08 11:35:34 dabrain34: https://www.collabora.com/about-us/open-source/open-source-projects/gstreamer.html Jun 08 11:39:49 kanavin: thank you. Jun 08 11:41:55 dabrain34: you will have to pay them though :-) Jun 08 11:42:25 you mean gstreamer experts ? ;) Jun 08 11:42:30 dabrain34: yes Jun 08 11:42:42 dabrain34: that's the idea :) Jun 08 11:43:04 kanavin: are you interested ? Jun 08 11:43:37 dabrain34: I am not a gstreamer expert, collabora folks are Jun 08 11:43:45 dabrain34: but they will not help you for free :) Jun 08 11:45:17 we dont want help, we want to hire ! :) Jun 08 11:45:29 I know collabora ;) Jun 08 11:46:36 also centricular are gst consultants Jun 08 11:46:42 dabrain34: I am afraid this is more difficult than you think, you're better off giving collabora a consulting contract Jun 08 11:49:13 kanavin: thank you for your advise :) Jun 08 11:57:23 Bah. My programs end up with an interpreter that is not in my SDK at all. Jun 08 11:57:38 ed2: hi again Jun 08 11:58:19 diego_r: hi Jun 08 12:00:43 ed2: in the heat of the discussion yesterday I forgot to mention the other problem I was hitting: sometimes, if I build an image multiple times, I get duplicated (or n-plicated) entries in the fstab for the additional partition (not "/" or "/boot") Jun 08 12:03:09 diego_r: that sounds like a bug to me. Jun 08 12:03:36 diego_r: you can either dig into it yourself or create a bug for me in bugzilla. Jun 08 12:08:27 ed2: considering my current tasklist, I'd go for the bugzilla for now. Hope you don't feel offended :-) Jun 08 12:11:22 diego_r: i don't :) I'm a maintainer of wic code, so I appreciate people finding issues and reporting them. Jun 08 12:19:44 Oh I just checked that meta-qt5 has no pyro version :s Jun 08 12:21:27 dabrain34: dabrain34: Jun 08 12:21:46 you want to hire? I'm available :) Jun 08 12:24:42 I have an query on, using docker for yocto builds Jun 08 12:28:22 ChrysD_: master is probably good still Jun 08 12:28:49 rburton : How it works if I miss pyro layer with master? Jun 08 12:28:55 rburton : if I mix* Jun 08 12:29:18 Ravi: https://www.youtube.com/watch?v=JXHLAWveh7Y Jun 08 12:29:27 khem, I see your message from yesterday. I can do that. Jun 08 12:29:29 For some reason qmake builds from my sdk choose the correct dynamic loader, but I have to set it in the linker options for my other builds. Jun 08 12:29:44 ChrysD_: master and pryo have only just branched, and unless there's been stuff in meta-qt5 since the branch that breaks under pyro you'll be fine Jun 08 12:31:18 rburton : so basically, nothing bad about mixing layer branches expect the fact that it could have problems? Jun 08 12:32:25 you need to be aware that changes may have been made. mixing master after API changes have been made wouldn't work for example Jun 08 12:32:38 right now i doubt there is anything in meta-qt5 that will break pyro oe Jun 08 12:32:56 better yet ask a meta-qt5 maintainer such as jama Jun 08 12:33:27 rburton : i thought there is a kind of mecanism that oblige you to have all the meta from the same branch... So basically I can have a morty project for example, and add a pyro layer, and if it doesn't break is good. Jun 08 12:33:40 rburton : Nice to know thanks. Jun 08 12:34:51 rburton : So the branch name give you just a hint that it was tested on a specific poky version? Jun 08 12:36:45 its not a hint Jun 08 12:36:57 its a statement of what the intended version is Jun 08 12:37:12 ChrysD_: You're very likely to see issue with any non-trivial layer that doesn't match versions (with the exception of using master with a recent release branch) Jun 08 12:37:14 *but* pyro was released just weeks ago, and meta-qt5 doesn't move fast, so they'll still work Jun 08 12:37:40 we're adding markup now so layers can say what versions they work with and you'll get nice errors Jun 08 12:38:07 So the markup of layers is a kind of pictures of the state of the project Jun 08 12:38:44 and the maintener of other layer when they put the name of the branch as the poky, is to say that it made with the aim of working in this version. Jun 08 12:39:40 but doesn't mean that couldn't work with other, "as our own risk" Jun 08 12:40:22 yeah. changes in newer versions which break older layers include stuff like new classes being added to oe-core, or behaviour changes (such as how FILES_dbg works) Jun 08 12:40:44 Ok Jun 08 12:41:07 so basically, maybe the morty version of meta-qt5 can work with my "pyro" project. Jun 08 12:41:26 or the master as it's supposly the work made for pyro Jun 08 12:41:40 i'd go with master and expect a pyro branch to appear shortly, most likely with the same content as master. Jun 08 12:41:57 ok nice to know thanks rburton / jku. Jun 08 12:52:37 khem: I just checked and I see the gcc-7 ilog2 change on all the relevant kernels, all branches. 4.1/4.4/4.9/4.10 .. where is this being seen ? I bet it is just a SRCREV issue. I'll check that now. Jun 08 12:54:54 hi all , i need to set environment variable during kernel module build process, i have already tried to add export FOO="BAR" in do_configure() function but if launch devshell var are not exported... any clues ? Jun 08 12:56:30 export FOO="bar" in the recipe itself Jun 08 12:56:47 if you put it in do_configure then it won't happen in the devshell becase that doesnt' run do_configure Jun 08 12:56:57 (and the export will only be present for the duration of do_configure) Jun 08 13:01:20 @rburton thx for info :) but doesn't seem to work if we want to set for example -> export KERNELARCH=${ARCH} Jun 08 13:04:31 if i put export FOO="bar" in a recipe and then do a devshell, echo $FOO says "bar" Jun 08 13:04:34 so that works Jun 08 13:05:13 and if you do the same think with Yocto env var ? but i agree FOO="bar" works like a charm ;) Jun 08 13:13:45 Where do is usually defined the IMAGE_CMD for IMAGE_FSTYPES? Jun 08 13:15:06 ChrysD_, classes/image_types.bbclass Jun 08 13:15:42 ant_work : ok thanks. Jun 08 13:15:42 then you can change it in machine.conf Jun 08 13:18:23 ant_work : i was using .sdcard and it sounds that the new meta doesn't support .sdcard, i have to check if there is a new .wic ! Thanks. Jun 08 13:19:10 yw Jun 08 13:20:16 diego_r: Thanks, wil check yr link Jun 08 13:21:45 Ravi: you're welcome Jun 08 13:55:29 so anyone has any clue about my issue with oe's testimage unable to reboot the target? Jun 08 13:55:51 the exact command works fine if i launch manually: Jun 08 13:56:00 $ ssh -l root -o UserKnownHostsFile=/dev/null -o StrictHostKeyChecking=no -o LogLevel=ERROR 172.31.173.111 'export PATH=/usr/sbin:/sbin:/usr/bin:/bin; { sleep 1; reboot; } > /dev/null &' Jun 08 13:56:00 zeenix@build1:~/pelux/build$ echo $? Jun 08 13:56:00 0 Jun 08 14:06:40 also, why would you shutdown before rebooting the target? http://git.yoctoproject.org/cgit.cgi/poky/tree/meta/lib/oeqa/controllers/masterimage.py#n106 Jun 08 15:44:18 MarcWe: looks like you may be getting a clash with linaro binutils... Jun 08 17:23:28 zeddii_home: see http://errors.yoctoproject.org/Errors/Latest/?filter=beaea15cf2429291b0f29ba3d1a12729f1d13ba3&type=commit Jun 08 17:24:11 see linux-yocto failure its using these srcrevs linux-yocto-4.1.38+gitAUTOINC+c55c877244_2ce56d130d Jun 08 17:24:44 http://errors.yoctoproject.org/Errors/Details/143869/ Jun 08 17:27:29 rburton: I was looking at gcc7/core-image-sato testimage failure and it seems infra issue to me Jun 08 17:27:31 https://autobuilder.yocto.io/builders/nightly-deb/builds/274/steps/Running%20Sanity%20Tests/logs/stdio Jun 08 17:29:25 ok. I’ll have a look. Jun 08 17:49:36 zeddii_home: while you are here http://sprunge.us/XIgi Jun 08 17:49:47 this is what I am seeing with latest master Jun 08 17:55:59 khem. yep. saw that as well, will do a fix for it. Jun 08 18:00:45 zeddii_home: khem: I pinged aehs29 about that. I believe he's working on it Jun 08 18:00:58 cool Jun 08 18:03:06 ok Jun 08 18:18:17 hello, quick question: I have a package (libdce) where on my host, it generates a .pc package file in sysroots/MACHINE_NAME/usr/lib/pkgconfig, however when I try to find that on my target filesystem, it is not there. Jun 08 18:18:37 I'd like for the .pc file to be available on my target filesystem Jun 08 18:23:00 twall: how does your recipe look like for libdce ? Jun 08 18:24:35 @khem: extremely basic, just a SRC_URI, simple dependencies and IPC_HEADERS. https://pastebin.com/UmWBc7ir Jun 08 18:27:09 khem: extremely basic, just a SRC_URI, simple dependencies and IPC_HEADERS. https://pastebin.com/UmWBc7ir Jun 08 18:29:13 twall: ok Jun 08 18:29:46 can you check the content of your packages-split directory in tmp/ dir for this recipe ? Jun 08 18:30:04 especially libdce-dev Jun 08 18:32:19 usually they are packaged automatically if the makefile of the component installs them Jun 08 18:32:45 if not then you might have to write a do_install_bbappend to copy the .pc files from ${B} into ${D} Jun 08 18:38:03 there doesn't appear to be a packages-split directory in tmp for this recipe. Correct me if I'm wrong but I'm looking in build/{toolchain}/work/{armv7-blah-blah compiler}/libdce/version/temp Jun 08 18:39:17 last dirname is not temp but packages-split/libdce-dev Jun 08 18:40:50 the libdce-dev folder does not exist. There's a libdce-dev file in libdce/version/pkgdata/runtime that looks like it specifies the location of where a .pc file would go though Jun 08 18:42:17 hmmm so there is no -dev folder at all ? Jun 08 18:46:54 Hi guys. Is there any way to have custom sanity tests? For example I have a variable that I want to deprecate and have a warning everything a user still has it in the config.conf Jun 08 18:49:18 I want these tests to be ran at every build so similar to check_sanity_everybuild Jun 08 18:56:44 khem: doesn't appear to be Jun 08 19:00:20 there are plenty of libdce-dev files though, they're all in the pkgdata folder in the libdce folder Jun 08 19:04:39 alephan: you can easily write your own class to do that Jun 08 19:05:27 ERRATA: I want to deprecate and have a warning every time a user still has it in the config.conf (auto correct is...) Jun 08 19:07:14 I'm not sure how it hooks into the build system. Do you have any links that I can investigate? Don't see anything obvious. @rburton Jun 08 19:07:28 rburton: ^ Jun 08 19:09:10 alephan: simply python() { if d.getVar("DEPRECATED"): bb.warn("Don't use deprecated, use NEWNAME")} in a class that you add to INHERIT might well work Jun 08 19:09:49 put that in a class, add classname to INHERIT, bitbake executes the class for every recipe, which then checks if the variable exists and calls bb.warn if it does Jun 08 19:09:51 simples Jun 08 19:09:54 rburton: that was my first try. The problem is that it throws the error many times. Jun 08 19:10:00 bb.error :) Jun 08 19:10:04 (fatal) Jun 08 19:10:08 Still. Even with fatal. Jun 08 19:10:10 yeah classe execute more than once Jun 08 19:10:23 It doesn't stop at first as I would have wanted. Jun 08 19:10:28 save a "I warned" value into the data store and check that first Jun 08 19:10:39 And if it is a warning... it throws thousands. Jun 08 19:11:04 Yeah... that's a cool workaround. Jun 08 19:11:19 Would work. But Can't I do it with an new event handler? Jun 08 19:11:31 Sunt searching for a `proper` way - if there is such thing. Jun 08 19:11:37 Just* Jun 08 19:13:38 there's a configparsed handler you can hook but that still fires more than once Jun 08 19:13:41 just not as many times Jun 08 19:14:09 hm try BuildStarted event Jun 08 19:14:09 Defining a new one is useless because nobody would fire it. Jun 08 19:15:28 BuildStarted looks good. Jun 08 19:15:34 Thanks rburton . Jun 08 19:26:47 rburton: Here is a beer for you! Works like a charm. Jun 08 19:50:23 Hello I have a problem with quilt-native Jun 08 19:50:34 I get the errer that checking the directory failed Jun 08 19:50:52 | Error during checking directory: -d dir/build/tmp/sysroots/x86_64-linux/etc/ Jun 08 19:51:02 anyone has an idea how i can solve that Jun 08 19:51:03 pastebin the full log please Jun 08 19:53:54 https://pastebin.com/R9iQMDDi Jun 08 19:55:29 i added the output or do u want more? and thank you for your help Jun 08 19:58:19 @rburton Jun 08 19:59:39 do you have a weird setup? virtual machine, unusual file system, etc? Jun 08 19:59:53 nope Jun 08 20:00:04 i have a similuar machine and there it is working Jun 08 20:00:38 try mkdir -p'ing that directory yourself and see what happens Jun 08 20:00:49 sounds like something really weird has happened to the permissions or something Jun 08 20:02:45 the funny part is that the directory exists Jun 08 20:03:08 one thing would be to delete your entire tmp/ and see if that helps Jun 08 20:03:19 did it Jun 08 20:03:28 i also downloaded my complete repo Jun 08 20:03:37 pastebin the stat of that directory? Jun 08 20:04:25 ls -n? Jun 08 20:04:32 oh ok wait Jun 08 20:04:33 stat Jun 08 20:05:29 https://pastebin.com/vc5i4fzp Jun 08 20:06:13 well thats annoyingly normal isn't it Jun 08 20:06:29 it is Jun 08 20:08:04 i have no idea Jun 08 20:15:16 gnu install doesn't have the phrase "checking directory" in Jun 08 20:15:57 neither does quilt Jun 08 20:16:29 google has never seen that error message Jun 08 20:16:59 the way its formed Jun 08 20:17:00 | Error during checking directory: -d /home/LinuxUser/git_ws/yocto/build/tmp/work/x86_64-linux/quilt-native/0.64-r0/image/home/LinuxUser/git_ws/yocto/build/tmp/sysroots/x86_64-linux/etc/ Jun 08 20:17:13 makes me think that *something* thinks the directory is called "-d /home/LinuxUser...." Jun 08 20:17:28 do you have a custom binary called install? Jun 08 20:21:30 thanks man Jun 08 20:21:32 ... Jun 08 20:21:40 custom install ... Jun 08 20:21:45 thanks a lot Jun 08 20:26:33 rburton: zeddii_home has sent the gcc7 fixed for kernel 4.1 Jun 08 20:26:44 rburton: with this I think we should be good Jun 08 20:44:11 gettext sucks 110 seconds for do_configure, and lone task running on 42 core machine Jun 08 20:44:32 seems criminal Jun 08 20:45:04 you cant choke such a monster for 2 minutes doing just 1 package Jun 08 21:29:11 khem: it runs five configures Jun 08 21:29:23 khem: if you use taskset to pin configure to a specific cpu it goes faster for me Jun 08 21:31:17 a lot faster, tried to get some kernel guys interested in this as its clearly (to me) pathalogically bad scheduling Jun 08 21:33:05 zeddii_home: ^ if you know anyone who understands the scheduler Jun 08 21:33:46 http://git.yoctoproject.org/cgit/cgit.cgi/poky-contrib/commit/?h=ross/staging&id=eaacbda2533b88f4220d30c9b5394dd3e0025dc2 makes gettext configure go from two minutes to 40 seconds on my hardware Jun 08 21:40:08 Hunk: sorted? Jun 08 23:07:04 Might not be the best place to ask this but does anyone know the status of meta-qt5 for pyro? I see an wiki that references a pyro branch but the github isn't showing that branch as existing Jun 08 23:15:35 dscully: i'd expect it to be a copy of master when it appears, and be appearing soon Jun 08 23:15:51 asking on the list would be good to prod the maintainer Jun 08 23:15:55 anyone idling? Jun 08 23:15:59 even if it is the same as master is sends a good message Jun 08 23:16:05 j0nny5: ask, don't ask to ask Jun 08 23:16:12 right Jun 08 23:17:25 having trouble with a bitbake core-image-sato and it is erroring around openssl and has warnings of missing .h files - i haven't found someone with a similar enough problem yet, so, no major guidance, but i've tired a few things Jun 08 23:17:40 pastebin the logs please Jun 08 23:17:41 rburton: thanks, will do Jun 08 23:19:17 i did a bitbake -c cleansstate openssl and retried, no luck Jun 08 23:19:26 ah, k Jun 08 23:27:35 hello, I'm running into an issue with a build. I'm trying to install a package, but I keep getting the 'installed but not shipped' error. I've tried the fix from here: https://lists.yoctoproject.org/pipermail/yocto/2016-January/028127.html to no avail Jun 08 23:28:19 @rburton would it be bad if i tried a "bitbake openssl" next as an attempt to build just that and sent those logs? Jun 08 23:28:46 I'm trying to install the libdce package, and currently I have a libdce.so file in my output filetree, and there is a libdce.pc file that gets generated on the host, but is not shipped to the target Jun 08 23:42:59 where do i go to set a path for the bitbake process to use? Jun 08 23:45:12 it is missing .h files because it is looking in the wrong directories Jun 08 23:58:59 rburton: that change looks good to me regardless what kernel schedulers think about Jun 09 00:00:51 rburton: how do i add or fix a path for several .h files Jun 09 00:05:34 j0nny5: you can use -I/path/to/header in CFLAGS Jun 09 00:06:00 or you can change the source files to #include "path/to/header/foo.h" Jun 09 00:06:02 and so on Jun 09 00:06:23 what is the issue you are seeing btw. Jun 09 00:08:40 doing a build of core-image-sato and openssl fails Jun 09 00:09:53 khem: the files are located in /usr/include/linux/ not /usr/include on ubuntu server 16.04 Jun 09 00:10:12 stddef.h and two others Jun 09 00:41:08 stddef.h is compiler provided header and it knows how to incude it Jun 09 00:41:25 are you using --sysroot option or not ? Jun 09 00:43:52 he's gone Jun 09 00:44:02 and so am i, i think Jun 09 00:44:47 rburton: good night Jun 09 00:45:02 good things come to those who wait :) Jun 09 00:45:04 too bad Jun 09 01:06:42 when I give up on my raspi running yocto, i usualy leave it running, no wifi, no chrony, nothing. Then I get back the next day, lo and behold, i have wifi Jun 09 01:06:55 henc ei also have time of day at stratum 3 Jun 09 01:08:11 i dont understand why letting is sit and stew will gain it wifi after some hours Jun 09 01:30:32 berndhs: do you have rfkill installed Jun 09 01:37:26 i have it turned off Jun 09 01:37:48 so wifi is unblocked Jun 09 01:38:18 but you are correct, by default it is bloked Jun 09 01:39:41 see if n/w service is retrying relentlessly Jun 09 01:39:46 I usually do this Jun 09 01:39:53 sudo rfkill unblock wifi Jun 09 01:39:57 sudo ip link set wlan0 up Jun 09 01:40:06 wpa_cli reconfigure Jun 09 01:40:13 wpa_supplicant -iwlan0 -c /etc/wpa_supplicant.conf Jun 09 01:43:49 the reconfigure complaines like so: Jun 09 01:43:50 Failed to connect to non-global ctrl_ifname: (nil) error: No such file or directory Jun 09 01:45:03 i need some kind of key managemtn don't i **** ENDING LOGGING AT Fri Jun 09 03:00:03 2017