**** BEGIN LOGGING AT Mon May 29 03:00:04 2017 May 29 05:53:52 morning all May 29 06:42:49 good morning May 29 08:24:15 hello, May 29 08:25:07 I'm using the krogoth branch of yocto and after adding systemd to my distro, udhcpc is not started automatically; I guess ifupdown used to do that but now /etc/network/interfaces does not exist anymore May 29 08:25:29 what is the right way to launch udhcpc with systemd ? May 29 08:43:04 top22 : you certainly need to do a service May 29 08:43:32 networkd is not provided by default, right ? May 29 08:43:55 I saw somebody do a bbappend for systemd to enable netword (https://lists.yoctoproject.org/pipermail/yocto/2015-July/025719.html) May 29 08:44:09 top22 : let me see for myself May 29 08:47:37 top22 : in my case I have a systemd-networkd.service on my board, maybe i can check for the recipe May 29 08:47:55 ChrysD: cool, thanks :) May 29 08:48:18 top22 : But certainly you will need a service for udhcpc May 29 08:49:21 top22 : somewhere i have that http://www.geexbox.org/forum/topic/networking-with-udhcpc/ May 29 08:49:35 top22 : I mean that PACKAGECONFIG ??= "sysusers lz4 kmod networkd resolved" May 29 08:51:07 ChrysD: if I have netword and udhcpc.service, networkd is for bringing the interface up and udhcpc.service is for getting an ip? May 29 08:51:56 top22 : I'm not familiar but I guess yes + the configuration of the interface May 29 08:52:57 top22 : you should see documentation about networking and systemD May 29 08:53:23 top22 : Because you will certainly need .socket also May 29 08:54:59 ChrysD: ok I'll check that out; but it's strange, I would have thought it would be really straightforward as with ifupdown :) May 29 08:55:20 top22 : Maybe is also straightforward, but in a different way as you are not used to May 29 08:56:05 top22 : I checked, and you have some files like "eth0.network" in /etc/systemd/network May 29 08:56:23 top22 : and it seems that /etc/systemd/network it's like the /etc/network/interfaces May 29 08:58:52 top22 : So i guess if you enable systemd-networkd, you will have everything needed, you will have to provide .network files to set up the interfaces and also a service for you uhdcpc May 29 09:00:38 Yes, I'll give it a shot, one more question; Shouldn't the networkd and resolved but added in the distro ? May 29 09:00:47 Or should I just do a bbappend of systemd in my layer ? May 29 09:02:50 top22 : you mean for adding the networkd? May 29 09:03:18 yes May 29 09:03:46 top22 : maybe a systemd_%.bbappend May 29 09:28:01 Is there anyone familiar with barebox here? May 29 09:31:08 top22 : have made a test, in /lib/systemd/network I putted a file and it configure my wlan0 May 29 09:34:47 top22 : I made a test May 29 09:35:01 top22 : i have put a file like wlan0.service in /lib/systemd/network May 29 09:35:05 top22 : and it worked in my case May 29 09:39:23 ChrysD: thanks for the info; I added networkd but the start fails with result 'exit-code' May 29 09:40:40 top22 : how did you add it? May 29 09:40:57 using a bbappend in my layer May 29 09:46:08 ChrysD: https://pastebin.com/7AbGFNkd May 29 09:49:08 top22 : and what was the error May 29 09:50:18 the systemd-tmpfiles-setup.service fails to start now because there is an unknown user 'systemd-network' May 29 09:50:40 so I should probably add that in the do_install_append May 29 10:01:04 ChrysD: My new bbappend with the creation of users: https://pastebin.com/fdDj7SRD May 29 10:01:57 top22 : i don't know if it is mandatory, but on lot of files i checked that there is "number " before the file. Like 10-eth0.network May 29 10:02:31 I think you're right May 29 10:02:36 top22 : so i don't know if it is just a name, or if there is a script that check for name file and the number specify a kind of priority or whatever May 29 10:03:06 top22 : I'm not the best that can help you... I'm new in yocto, new in systemd.. new in lot of things ahahah May 29 10:25:40 ChrysD: thanks a lot for your help; I got the network working May 29 10:26:06 however, I'm under the impression that there is a dhcp client integrated in systemd as I dont need udhcpc to run to get an ip May 29 10:52:37 top22 : Have you checked into some files? May 29 10:52:43 top22 : like /lib/systemd/network May 29 10:52:50 top22 : or /usr/lib/systemd/network May 29 10:53:56 top22 : https://wiki.archlinux.org/index.php/Dhcpd how to configure it on systemd May 29 10:54:04 top22 : maybe you have it directly May 29 11:02:41 Hi I am trying to comiple a OpenSSL engine with Yocto May 29 11:02:46 could anyone help me with that? May 29 11:03:19 I am not sure if I should do that in the OpenSSL recipe, or in a new one. The Engine code seems to use header files from the openssl source code (internal ones, not only the ones exposed in the SDK) May 29 11:17:42 what is IMAGE_FEATURES[validitems] for? Specifiying the features that I can set with IMAGE_FEATURES? May 29 11:18:05 phako[m]: yes May 29 11:18:56 ttllkk: if it uses internal headers from openssl you'd probably have to modify the recipe to install them somewhere the other recipe can find them May 29 11:19:55 RP1: thx May 29 11:20:18 phako[m]: its there so the code can check the options specified are valid May 29 11:28:38 RP1 which recipe, the openssl one? May 29 11:28:58 you mean modifying that recipe so that the interal headers used also go to the SDK? May 29 11:41:48 ttllkk: yes May 29 11:42:09 ttllkk: needing headers which aren't installed is usually a bad sign though :/ May 29 12:31:03 Why do several recipes in OE-core reference license files as file://${COREBASE}/meta/... ? May 29 12:31:27 Is that the right fallback when nothing in the component's source code specifies the license? May 29 12:31:53 To me it seems that such LIC_FILES_CHKSUM entries defeat the purpose of the variable. May 29 12:32:11 Well, at least the "checksum" part. May 29 12:32:36 It still ensures that there is a valid license text, in case that someone wants to gather those. May 29 13:01:42 pohly: I think that is used when there is no license text, e.g. a packagegroup? May 29 13:03:04 RP1: yes, it seems to be some kind of fallback. I just wonder whether using this approach should be discourage and instead the authors of the code should add proper license declarations. May 29 13:03:47 I've been "guilty" (?) of using the COREBASE license for initramfs-framework modules. May 29 13:04:03 pohly: it certainly does get scrutiny, I don't really see the point of copying a MIT text into lots of directories though May 29 13:04:46 RP: but a licensing comment in the source code would be suitable, right? May 29 13:05:20 pohly: if there are sources, yes May 29 13:05:21 Now that I think of it, my initramfs-framework modules don't have source code - the small scripts are generated by the recipe. May 29 13:05:30 pohly: that would indeed be preferable May 29 13:07:25 It is a holiday guys! May 29 13:08:33 hi Crofton|work :) May 29 13:10:20 RP1 I know, unfortunately this openssl engine is not written by me =( May 29 13:10:23 Thanks for your help! May 29 13:13:56 florian, Hi! May 29 13:14:02 I hope things are well May 29 13:14:19 Holiday here, so having a barbecue, going to be a long day :) May 29 13:18:04 Ah very good... I could need some day off as well. May 29 13:20:53 Is there any way to tell a recipe in what sysroot the usr/include folder is located? Currently a file called "x86_64-poky-linux-g++" looks in the "x86_64-linux" sysroot for header files. But it has to look in the "genericx86-64" sysroot May 29 13:31:13 What does it imply to configure network interface on Barebox directly? May 29 13:31:34 Does it create files like .network files automatically once in the rootfs? May 29 13:51:14 Crofton|work: I could quite fancy a bbq but the weather here isn't right May 29 13:51:41 rj_: are you passing the --sysroot= option to the compiler? May 29 13:53:50 RP: I don't, at least not that i know of. However when compiling it does show --sysroot= being set to the genericx86_64 directory. But when checking the libtool program, it has the wrong sysroot May 29 13:55:16 RP: I'm currently rebuilding the programm with "--with-libtool-sysroot=Path/To/Sysroot" passed to the configure script May 29 13:57:10 RP: Which results in the same error sadly... May 29 13:58:35 rj_: did you reautoconf it? May 29 13:58:51 RP: how would i do that? May 29 13:59:02 rj_: run reautoconf? May 29 13:59:59 RP: well, i deleted the full tmp folder before, so it is pretty much a clean build May 29 14:01:06 RP: The thing is, everytime the recipe can't find stdlib.h even though it is there May 29 14:01:40 rj_: if its looking in the wrong sysroot, that isn't surprising May 29 14:02:10 rj_: My question was about whether your do_configure runs autoreconf or libtoolize (assuming its an autoconf based recipe) May 29 14:02:26 If it doesn't run those, it may not be finding the correct libtool version for example May 29 14:03:14 RP: i don't have a do_configure in my recipe May 29 14:03:29 RP: i inherit autotools-brokensep May 29 14:03:43 rj_: ok, so it should be being reautoconf'd already May 29 14:03:53 (the autotools class handles it) May 29 14:03:58 Ah alright ^^ May 29 15:43:57 this isn't right, i see bitbake rpi-test-image fail because of May 29 15:44:20 NOTE: make -j 4 install DESTDIR=/home/bernd/dawork/yocto/poky/rpi-build/tmp/work/cortexa7hf-neon-vfpv4-poky-linux-gnueabi/libtalloc/2.1.9-r0/image May 29 15:46:44 for those that are interested and awake https://da.gd/9HJE May 29 15:47:32 that is the script giving, eventually, the "no rule to make target install" May 29 16:15:27 NOTE: make -j 4 install DESTDIR=/home/bernd/dawork/yocto/poky/rpi-build/tmp/work/cortexa7hf-neon-vfpv4-poky-linux-gnueabi/libtalloc/2.1.9-r0/image May 29 16:15:35 for those that are interested and awake https://da.gd/9HJE May 29 16:42:06 Anyone available to help a beginner? May 29 16:42:49 I've built an image (core_image_base) for a Beaglebone Black, and put it onto an SD card with dd May 29 16:43:31 but U-Boot sits at "Trying to boot from MMC1" May 29 16:44:25 It is using the SD card because the date U-Boot prints is todays May 29 16:58:22 can you intercept u-boot before it tries to auto boot and run some mmc commands manually? May 29 17:00:41 no i cant see anything, monitor is not working May 29 17:00:48 wrong resolution probably May 29 17:02:36 sorry wrong window May 29 17:04:18 how are you cureently set up? 3.3V USB to serial adapter, or ... ? May 29 17:17:16 I've got an FTDI cable on the J1 header of the Beaglebone, and I'm using minicom to receive the text May 29 17:18:00 all that I receive from the board is: May 29 17:18:28 U-Boot SPL 2017.01 (May 29 2017 - 14:20:46) May 29 17:18:39 Trying to boot from MMC1 May 29 17:19:05 to get that I plug it in with button S2 pressed May 29 17:19:15 ok, so it isn't that it fails to boot a kernel ; it fais to load u-boot itself. May 29 17:19:23 ah right May 29 17:19:49 what do you think would cause that? May 29 17:20:17 it could be a regression in u-boot that hasn't been noticed yet ; it could be a slightly crappy SD card.... May 29 17:20:46 is the text I see from the first stage of U-Boot, which is then failing to load the second stage? May 29 17:21:15 I'll see if I can hunt down an alternative SD card to try May 29 17:21:17 I'm not 100% sure on the boot sequence of the BB. May 29 17:21:36 OK no worries May 29 17:21:42 thanks for your help so far :) May 29 17:21:44 but chances are it tries to load a small chunk to then run and continue loading more. May 29 17:24:04 here is what I was thinking of running mmc commands manually, if you ever get u-boot to bootstrap itself... May 29 17:24:09 https://pastebin.com/3TWXYVNC May 29 17:26:11 hi, i wrote a own layer with recipes-bsp/u-boot/u-boot-mydistro.bb set PREFERRED_PROVIDER_virtual/bootloader ="u-boot-mydistro" in build/conf/local.conf and run bitbake -c devshell u-boot-mydistro and in the new shell i run make mydistro_defconfig and then make but then i get an error about bad value -march=armv5 the arm version for the build seemed to be armv7-a May 29 17:28:38 whenn i run bitbake u-boot-mydistro it runs without errors but in the tmp/work dir is no binary May 29 17:29:23 Thanks paulg May 29 17:29:42 I was expecting some kind of console for U-Boot, as that's what I've seen before May 29 17:30:19 barry_scott, no prob ; next step might be to get a known valdated image (with u-boot) for that board and put taht on your card. May 29 17:30:47 rule out the card vs. u-boot itself being borked. May 29 17:32:19 Yeah that's a good idea May 29 17:32:35 I think I have a spare SD card I can try too, but I'll try a different image first May 29 17:44:19 how can I restrict packages locale? I don't want to spend the time up front to support everything in 129 different languages May 29 18:01:39 berndhs, I believe you can put something like ... May 29 18:01:43 IMAGE_LINGUAS="" May 29 18:02:00 in one of your conf files. May 29 18:02:09 should at least give you something to google for. May 29 19:13:13 So I downloaded the Debian Jessie IoT image from here: https://beagleboard.org/latest-images May 29 19:13:34 copied it to the SD card with dd May 29 19:13:43 and it behaves exactly the same as my Yocto image May 29 19:13:58 so I guess either my SD card or Beaglebone is faulty May 29 19:24:41 barry_scott: double check the memory address you are using to flash kernel into memory May 29 19:29:27 lsandov, he isn't even getting that far ; u-boot itself isn't managing to post. May 29 19:30:49 barry_scott, on the bright side, it is probably easier to buy another cheap/small u-SD card than it is to try and bisect u-boot. :) May 29 19:31:22 paulg: I see. Agree, bisecting uboot is not the best approach. May 29 20:17:20 Thanks for the advice, I'll try a new SD card and see what happens then :) May 29 22:34:07 i'm seeing a recurring problem in building libtalloc May 29 22:34:40 comes down to calling make, and no make install label defined May 29 22:34:46 that seems really wrong May 29 22:44:22 berndhs: are you using the libtalloc recipe in meta-networking ? May 29 22:45:08 probably, i'm not sure wehre it comes from. Should I not use it? May 29 22:45:24 no, I guess I was wondering if you'd written your own perhaps May 29 22:45:38 no this is all stick stuff right now May 29 22:45:47 ok May 29 22:45:54 just want to make a minimum thig for raspi May 29 22:46:18 then add things as needed, to keep things minimum size and complexity May 29 22:46:45 hmm, well it looks like make install is expected to work there May 29 22:47:21 oh i'm sure of that too, i just dont know where it went May 29 22:48:14 makefile got lost perhaps, but how would i know? May 29 22:50:31 where is the Makefile for that in any case? or is it built on the fly ? May 29 22:56:30 i think it's built on the fly May 29 22:57:28 it is, you'll find it in the build directory for the recipe May 29 22:57:37 bitbake -e libtalloc | grep ^B= May 29 22:58:04 (according to the recipe, it's using waf to build so I guess waf is producing the makefile) May 29 23:01:51 I went into the tmp directory, with a find i blew away everything called libtalloc* May 29 23:02:10 to make it rebuild that library May 29 23:21:02 berndhs: it's much safer to do bitbake -c cleansstate libtalloc May 29 23:25:27 ok will do that. That rebuilds only that target from scratch? May 29 23:48:24 i need to find out how to supress wayland and romanian ant turkish support. I don't need any of that, and its dozens of packages May 30 02:35:17 what do you do when a package or recipe and the website is down or gone? May 30 02:36:12 i cant file a site called www.lartmaker.nl May 30 02:36:32 hope the stuff has been archived elsewhere May 30 02:36:44 and archive it yourself May 30 02:37:44 its at another github **** ENDING LOGGING AT Tue May 30 03:00:04 2017