**** BEGIN LOGGING AT Wed Apr 06 02:59:58 2016 Apr 06 08:28:08 hi folks, I've one question similar to https://lists.yoctoproject.org/pipermail/yocto/2014-August/021368.html, which was not answered Apr 06 08:28:19 I've configured a "recovery" kernel for a board I'm working on (a kernel with an initramfs embedded into it and nothing else), so now I have to include the kernel (and its dtb) in my yocto project in order to build everything I need (uboot + standard_dtb + standard_kernel + rootfs + recovery_dtb + recovery_kernel) typing "bitbake myproject" Apr 06 08:28:59 is there an elegant way to do so? Apr 06 08:30:13 should I consider : "bitbake myproject" to build my standard env and then bitbake virtual/recovery_kernel? Apr 06 09:28:01 Hi all :) Concerning systemd recipe in oe-core: Looking through the recipe i noticed there was --disable-coredump in EXTRA_OECONF ( http://cgit.openembedded.org/cgit.cgi/openembedded-core/tree/meta/recipes-core/systemd/systemd_225.bb?h=jethro#n121 ). Does anyone know 1.) Why that is 2.) whether there's any chance i get it with that feature running on an ARM? Reason is i wanted to try systemd's coredumpctl and i suppose i need the coredump feature for that. Apr 06 09:28:45 It's no hard requirement for the project but it could be usefull and i'm basically still playing arround with that huge messy pile systemd offers ;) Apr 06 09:29:52 Anticom: can't you simply use EXTRA_OECONF_pn-systemd_remove that ? Apr 06 09:29:57 or is it not a list ? Apr 06 09:30:54 boucman_work: of course i could but the recipe author must have had a reason to disable it. Now i'm trying to figure out, whether it's simply to get systemd become a little slimmer and lighter or whether it's disabled bc it won't work Apr 06 09:31:50 is that to enable systemd itself to core-dump or is it to monitor core-dumps from systemd services ? Apr 06 09:32:57 boucman_work: if i knew i wouldn't ask :) Apr 06 09:33:13 :P Apr 06 09:33:37 I'd still consider myself a systemd noob. I'm just fiddling arround and exploring all the "new" things systemd and its companion tools/utilities offers Apr 06 09:33:44 ./configure --help in systmed's workdir might give you a clue, or a readme in there... Apr 06 09:34:10 boucman_work: good point. That might at least get me started. Cheers Apr 06 09:56:01 hi Apr 06 09:56:11 any one there Apr 06 09:56:51 shankar: hello, yes, just ask your question Apr 06 09:57:16 Im using intel edison board Apr 06 09:57:40 with yocto liux 1.6 version Apr 06 09:58:07 is it possible to configure wireless router in it Apr 06 10:02:40 any body there Apr 06 10:06:34 yocto is a tool for building linux systems, so it depends on what is in the edison software really Apr 06 10:06:57 looking for hostapd or connman is a good start Apr 06 10:18:49 thank you Apr 06 10:18:57 for u r support Apr 06 10:30:22 could to able to provide any additional information regarding wireless router configuration in intel edison Apr 06 10:31:19 personally, i don't have an edison, don't have the specific software that comes with the edison BSP out of the box, and have never configured an AP Apr 06 11:25:34 I'm having a hard time finding any docs on QEMU_TARGETS (from poky.conf) Apr 06 11:50:50 hi. so... i'm having a problem with wrong file being included again... i'm trying to bbappend barebox. the original recipe is in meta-phytec. here's my bbappend: https://bpaste.net/show/ea0d6a133c94 Apr 06 11:51:29 is 'IMAGE_FEATURES' modifiable and it's value is retained across recipes? Apr 06 11:52:13 i have one file in boardenv: env/boardenv/config-expansions Apr 06 11:52:55 but the original file from the barebox recipe is included no matter what i do. been banging my head on this for a few hours, could need some help debugging this! Apr 06 11:56:34 can we add to IMAGE_FEATURES in one recipe and access the values in another recipe? Apr 06 11:57:52 can I use a kernel append such as linux-$ or linux.bbappend to attach to multiple kernel recipes in the form linux-nameofboardhere ? Apr 06 11:58:28 I've got a patch for the kernel, which I want to be able to append to multiple bsp kernels Apr 06 11:59:27 where the names are linux-raspberrypi linux-intel etc Apr 06 12:03:02 CTtpollard: don't think so (though don't take my word for it). but you can reuse the same patch on multiple kernels. Apr 06 12:03:46 ionte_: yeh, just wanted to save maintenance having one append instead of multiple that's all :/ Apr 06 12:08:22 CTtpollard: Dunno whether % only replaces version string or is a generic wildcard. Have you tried linux-%.bbappend ? Apr 06 12:09:10 I've got a build running on my localhost currently, hence i can't look it up myself. You should be able to find it in the bb manual or yocto mega manual Apr 06 12:09:33 CTtpollard: well, restructure your recipes so that the images all share a common name, and then selectively append inside the bbappands by using the trailing selectors of the variables you need to modify Apr 06 12:10:12 LetoThe2nd: the kernel recipes are not under my control Apr 06 12:10:21 CTtpollard: bad luck then. Apr 06 12:11:46 it seems quite common across bsp's to name the kernel recipes linux-$target, so I thought they would be a capture for all Apr 06 12:12:44 common does not necessarily imply a good and maintainable practise Apr 06 12:13:03 how do i easiest see what paths are used when searching for a file? Apr 06 12:13:49 if the file is missing i get a debug message with all paths that were searched, but not so if it was found Apr 06 12:33:30 (write here again maybe someone can answer) hi folks, I've one question similar to https://lists.yoctoproject.org/pipermail/yocto/2014-August/021368.html, which was not answered Apr 06 12:33:38 I've configured a "recovery" kernel for a board I'm working on (a kernel with an initramfs embedded into it and nothing else), so now I have to include the kernel (and its dtb) in my yocto project in order to build everything I need (uboot + standard_dtb + standard_kernel + rootfs + recovery_dtb + recovery_kernel) typing "bitbake myproject" Apr 06 12:34:02 is there an elegant way to do this? Apr 06 12:47:57 ionte_: IMAGE_FEATURE are features of images, so they only apply to image recipes. Apr 06 12:48:55 ionte_: sorry, that was kryozen's question and I missed him Apr 06 13:10:41 zero_note: I haven't done that but maybe this works: http://permalink.gmane.org/gmane.linux.embedded.yocto.meta-arago/3981 Apr 06 13:11:02 or at least is agood place to start: [PATCH 1/5] recovery-kernel: recovery kernel build for keystone Apr 06 13:13:09 vmeson: thank you, its really a good place to start Apr 06 13:14:59 zero_note: don't thank me, thanks be to google. :) Apr 06 13:17:42 vmeson: google is a great tool, but I have to thank who makes the right query ;-) Apr 06 14:07:15 kergoth: around? Apr 06 14:07:54 Hi, I'm running into some issues while trying to install the eclipse adt plugin, it fails due to a dependency to tm.terminal. I came accross this mail https://lists.yoctoproject.org/pipermail/yocto/2016-January/027951.html which didn't get any answer. I built the plugin with this patch but it feels like a dirty hack. Should I report this issue? Apr 06 14:09:00 mathieu3: if it doesn't have a bug in bugzilla it might be an idea, we're struggling to find people to work on the eclipse plugin atm :( Apr 06 14:10:58 kergoth: turns out that messages during parsing don't trigger logging prefix change. I'm wondering about http://git.yoctoproject.org/cgit.cgi/poky-contrib/commit/?h=rpurdie/wip&id=b14b3bcc14f7209cc2477f916f1360738cab6895 Apr 06 14:11:06 of there is one already : https://bugzilla.yoctoproject.org/show_bug.cgi?id=9105 Apr 06 14:11:09 Bug 9105: normal, Medium+, 2.1 M4, timothy.t.orling, IN PROGRESS IMPLEMENTATION , [MARS] Cannot install eclipse-poky plugin Apr 06 14:29:31 there is a way to copy files into the kernel source dir before the kernel will be built? Apr 06 14:33:17 zero_note: making and applying it as a patch might be the simplest way Apr 06 14:38:46 boucman_work: ok, but in my case the file is binary: following this as example http://permalink.gmane.org/gmane.linux.embedded.yocto.meta-arago/3981, I'm trying to write a recipe for my recovery kernel, and the way I obtain initramfs wants the busybox binary (statically linked) under initramfs/ in kernel source dir Apr 06 14:39:54 so what I thought: compile busybox -> copy the binary into the kernel source dir -> compile kernel Apr 06 14:41:20 but maybe I can obtain this by modify kernel's do_compile? Apr 06 14:49:49 zero_note: I don't think do_compile is the right place, but I never had to deal with the internals of yocto's kernel building logic Apr 06 14:51:19 boucman_work: thanks anyway Apr 06 15:12:48 hello channel Apr 06 15:13:44 I am doing a multilib (lib and lib64) build of poky and I'm having issues with CMake modules, I'm not sure if they should be placed in /usr/lib/cmake or /usr/lib64/cmake Apr 06 15:14:22 if someone knows off the top of their heads what's the correct path it would be greatly appreciated :) Apr 06 15:20:30 RP: i was thinking about something like that too, not sure what the best approach is for it. there's also messages coming from event handlers in recipe context that it'd be nice to prefix too :\ Apr 06 15:20:33 afk, coffee Apr 06 15:40:55 fredcadete_ Apr 06 15:41:09 You mean the find XXX.cmake files? Apr 06 15:41:29 betchou: yes, those are the ones Apr 06 15:41:51 It should be in /usr/share. Apr 06 15:41:57 haha Apr 06 15:42:03 And it should not depend on 32 or 64 bits. Apr 06 15:42:25 and if I put them there they will be put in the sysroot by populate_sysroot? Apr 06 15:42:47 Your FindXXX.cmake files or whatever have to handle 32 or 64 bits by themself. Apr 06 15:42:53 I tried putting them in /usr/lib regardless and that was the problem I was having Apr 06 15:43:08 Those are not librairies. Apr 06 15:46:30 I'll give it a try Apr 06 15:46:42 come on :D Apr 06 15:46:48 meta-qt5 seems to be putting things in /usr/lib by the way Apr 06 15:47:02 I'm checking. Apr 06 15:47:21 thanks a lot for the tips betchou Apr 06 15:47:28 meta-qt5 fido here Apr 06 15:47:53 sorry I mean jethro Apr 06 15:49:05 /usr/share/cmake-2.8/Modules Apr 06 15:49:19 Fuuuuuuuuuuuuuuulllll of FindXXX.cmake guys. Apr 06 15:49:52 from cmake itself I think Apr 06 15:51:24 Do as you wish then Apr 06 15:52:34 I'll try /usr/share, sounds like a good idea Apr 06 15:52:38 thanks Apr 06 15:52:39 There are Find files and Config files. Apr 06 15:52:57 Check what you are generating first. Apr 06 15:53:25 but I can't put them in /usr/share/cmake-2.8 just because in this build it's cmake 2.8 :) Apr 06 15:53:47 For sure Apr 06 15:53:51 though I think /usr/share/cmake will be picked up too Apr 06 15:55:41 that's solved! betchou, you get good karma Apr 06 15:55:45 thanks a lot Apr 06 15:57:31 From OpenCV: Apr 06 15:57:44 /usr/share/OpenCV/OpenCVConfig.cmake Apr 06 15:58:19 fredcadete_: You might have package config files generated also. Apr 06 15:58:28 *.pc filrd Apr 06 15:58:31 *files Apr 06 15:58:36 I do Apr 06 16:00:10 but they'll be arch-specific, they'll mention lib64 libraries Apr 06 16:00:26 by the same criteria the cmake files would also go to /usr/lib64 Apr 06 16:00:38 and this is when I give up and go home Apr 06 16:02:04 surely the "if its arch specific then put it in $libdir" rule is good enough here Apr 06 16:02:06 Package config DO precise the exact version of a library. Apr 06 16:02:37 Which is why the path might include 32 or 64. Apr 06 16:03:00 but so do the cmake files being installed in this specific case Apr 06 16:03:14 :'( Apr 06 16:03:17 Then you have shitty CMake files :P Apr 06 16:03:19 woe is me Apr 06 16:03:49 Find CMake => global Apr 06 16:03:54 believe me, if it were up to me they would not be there Apr 06 16:03:58 Config CMake => specific Apr 06 16:04:05 Package Config => specific Apr 06 16:04:07 the package is using this wonderful cmake feature: https://cmake.org/cmake/help/v3.5/command/export.html?highlight=export Apr 06 16:04:15 Hi. I've build core-image-base with standard MACHINE ?= beaglebone. I'm trying to use runqemu like this: runqemu qemuarm zImage-beaglebone.bin core-image-base-beaglebone.jffs2 without success. Should be it ok or I should try another way? Apr 06 16:04:30 versatile123: beaglebone != qemuarm Apr 06 16:04:53 versatile123: if you want to run stuff in qemu, it's much easier to just use MACHINE=qemuarm Apr 06 16:04:59 kergoth: do you think that patch would be an improvement? Apr 06 16:06:55 Small question for me now: Apr 06 16:06:59 rburton: so how can I run beaglebone image? Using simple qemu-arm or runqemu script, doesn't matter.. Apr 06 16:08:35 I want to build a BSP on top of raspberrypi. I'm creating a new machine. But for now I get: bcm2835-bootfiles was skipped: incompatible with machine toto Apr 06 16:08:44 versatile123: run a beaglebone image on a beaglebone Apr 06 16:08:53 not in COMPATIBLE_MACHINE Apr 06 16:09:05 What is the proper way? Apr 06 16:09:33 versatile123: you could probably work out the right incantations to make qemu run a beagle image, but it won't be optimal. a beagle is not a qemuarm. Apr 06 16:09:58 (build a qemuarm image and you get a kernel that knows how to do stuff fast when its inside qemu) Apr 06 16:10:40 rburton: Wait a minute.. Beaglebone is arm, quemu is doing arm easily I guess.. So what's the problem? Apr 06 16:10:44 I'm really confused now Apr 06 16:11:13 versatile123: well, for a start ARM is an instruction set and nothing more. qemu doesn't emulate a beaglebone. Apr 06 16:11:22 http://pastebin.com/UPrQgtsj Apr 06 16:11:30 toto => lightpixels Apr 06 16:13:41 rburton: I don't want qemu to emulate beaglebone. I want it to emulate ARM cortex a8. Why qemu can't emulate this processor? Apr 06 16:14:30 well i presume you'd like input and output Apr 06 16:14:42 the a8 is just a processor Apr 06 16:14:48 so qemu has to emulate an entire board Apr 06 16:14:58 rburton: No, I don't need this at all, never said I need this. Apr 06 16:15:31 all i can suggest is to try invoking qemu-arm directly and see if you can make it boot Apr 06 16:15:45 (or switch to MACHINE=qemuarm for stuff you want to run in a qemu) Apr 06 16:15:46 rburton: I only want to modify standard beagleboard device-tree, to try write simple device driver using .compatible pointer Apr 06 16:16:27 http://qemu.weilnetz.de/qemu-doc.html#ARM-System-emulator <— this is the hardware that qemu emulates Apr 06 16:16:28 and I notice that standard yocto beaglebone machine is the only one with specified devicetree file, so it is the easiest way for me Apr 06 16:17:53 How much easier question if someone knows: How to handle COMPATIBLE_MACHINE when building a BSP on top of another? Apr 06 16:19:37 betchou: you can probably add the parent machine to MACHINEOVERRIDES Apr 06 16:21:02 so toto.conf which includes raspberrypi.conf can set MACHINEOVERRIDES .= ":raspberrypi" Apr 06 16:23:11 rburton: Just trying now! Thanks a lot! I've seen the line COMPATIBLE_MACHINE = "raspberrypi" in at least ten files. I would have had to write 10 .bbappend?! Apr 06 16:26:34 rburton: It does not seem to work. :/ Apr 06 16:26:40 betchou: no, set machineoverrides in your machine conf like i said, and when compatible machine is evaulated it will look at toto and raspberripi Apr 06 16:26:56 check that MACHINEOVERRIDES is being set correctly, with bitbake -e Apr 06 16:30:46 rburton: Seems to work now. I did not commit your modif. Apr 06 16:31:02 rburton: yep!! Thanks :) Apr 06 16:34:05 rburton: This is the proper way to do it by the way? Apr 06 17:10:25 is anyone aware of existing recipes for building zfs-on-linux? Apr 06 18:14:18 or rather does anyone have experience building an out-of-tree kernel module thats built using autotools? Apr 06 18:14:41 it seems like I probably can't follow examples for recipes inheriting `module`, nor those inheriting `autotools` exactly Apr 06 18:42:37 Hello. It seems that newer Yocto/raspberry-pi-bsp/rpi-basic-image releases no longer produce a finished (partitioned etc.) image as an result but "just" the rootFS. Or am I missing something? Apr 06 18:43:05 It's just what I would expect to come out, taking the name "rpi-basic-image" into account. Apr 06 21:57:49 Is there a standard for handling of user supplied config files? Separate recipe? merge-files? bbappend? Keeping it in the same recipe with an append results in checksums changing when the machine changes. I know we've used separate recipe sin some cases, i.e. alsa-state (though admittedly that's not config), but what about the general case? Apr 06 21:58:39 standard? hahahahaha Apr 06 21:58:56 was that helpful? Apr 06 22:01:35 rburton1: hehe Apr 06 22:02:04 "convention" would perhaps have been a better choice Apr 06 22:07:14 kergoth - Is there a way to have all the *-dev-*.rpms not installed in the final image easily?? Apr 06 22:07:37 not installed? Apr 06 22:07:42 don't install them in the first place Apr 06 22:07:51 sounds like you've got dev-pkgs in IMAGE_FEATURES Apr 06 22:08:17 I will look to see if dev-pkgs is tucked away in some *.bb file. Apr 06 22:08:44 its an image feature, so your distro or local conf, or the image recipe itself Apr 06 22:12:32 something in your confict is set to ADD the dev-pkgs.. (or you have a custom image that is set that way) Apr 06 22:13:10 One of the *.bb files has "require recipes-core/images/core-image-minimal-dev.bb". It appears I need a line: PACKAGE_EXCLUDE = "dev-pkgs" Apr 06 22:13:36 dev-pkgs isn't a package, it's an image feature Apr 06 22:14:46 clearly pulling in the core-image-minimal-dev config isn't the right thing to do Apr 06 22:14:58 core-image-* are examples, copy instead of including Apr 06 22:15:22 i wonder if RDEPENDS supports "foo | bar" to install foo or bar Apr 06 22:15:40 thats dpkg syntax, no idea what rpm has (fray?) Apr 06 22:15:43 Hmmm, ok. I will steer away from using PACKAGE_EXCLUDE, and look into better approach. Apr 06 22:16:08 PACKAGE_EXCLUDE is a method that was added to say "never install this list of packages, trigger an error if they are required" Apr 06 22:16:38 fray - Makes sense. Thx Apr 06 22:17:30 session over.. ogtta go.. Apr 06 22:18:48 Duh, sorry guys. Just require core-image-minimal. Thanks for kick starting me in the right direction. It helped. Apr 06 22:19:16 don't require an existing image, just copy the contents into your image recipe and hack it from there Apr 06 22:19:32 or you'll be really annoyed when someone changes whats in core-image-minimal Apr 06 22:19:51 rburton1 - Is that the standard practice by most everyone! Apr 06 22:20:02 yes Apr 06 22:20:19 core-image-* are examples, and also mostly trivial Apr 06 22:20:39 Ok. That helps. **** ENDING LOGGING AT Thu Apr 07 02:59:58 2016