**** BEGIN LOGGING AT Thu Jun 12 02:59:59 2014 Jun 12 06:04:32 hi all Jun 12 06:35:08 hm, I have listed libsdl2 at IMAGE_INSTALL_appends += "libsdl2" but it will not get build Jun 12 06:35:17 I don't even see it in the dot files Jun 12 06:37:05 rink_: well, why should it? Jun 12 06:37:27 rink_: _append != _appends Jun 12 06:37:57 rink_: but you can figure these out by yourself. just do "bitbake -e your-image | less" and search for ^IMAGE_INSTALL Jun 12 06:39:40 ugh, typo - I do have '_append' in the .bb Jun 12 06:40:56 but I can see it in bitbake -e, which is why I don't understand it Jun 12 06:48:50 why isn't it being built - this used to work :-/ Jun 12 06:49:03 it drags in my RDEPENDS nicely Jun 12 07:19:18 good morning Jun 12 07:35:39 Good Morning! Jun 12 07:54:13 hi Denwid, all Jun 12 08:29:05 Why the heck is TOOLCHAIN_HOST_TASK called TOOLCHAIN_HOST_TASK and not TOOLCHAIN_HOST_PACKAGE_INSTALL or SDK_PACKAGE_INSTALL_HOST? :-( Jun 12 08:29:23 or something like that... it doesn't even list tasks Jun 12 08:38:16 morning all Jun 12 08:58:28 hi bluelightning, all Jun 12 08:58:30 JaMa: around? Jun 12 08:58:38 morning mckoan, RP Jun 12 08:58:43 RP: y Jun 12 08:59:06 JaMa: do you have the sstate for gcc and bintuils from that grub build available somewhere? Jun 12 08:59:52 yes, let me find them Jun 12 09:00:08 JaMa: readelf -e uhci.module on the broken build shows " [24] .note.GNU-stack PROGBITS 00000000 009180 000000 00 0 0 0" Jun 12 09:00:19 A good build shows 0 0 1 at the end Jun 12 09:00:39 I just don't know how in that specific build how the alignment ends up as 0 instead of 1 :/ Jun 12 09:03:22 bluelightning: hi! sorry for the duplicate yesterday. Jun 12 09:03:30 zecke: no worries Jun 12 09:04:37 RP: you need -cross state right? Jun 12 09:07:13 JaMa: yes Jun 12 09:08:48 RP: grub-sstate.tar.bz2 in the same location Jun 12 09:21:28 JaMa: thanks! Jun 12 09:24:03 bluelightning: do you know if a copy of pseudo was on the normal yocto mirror? Jun 12 09:32:02 zecke: yes, it's there also: http://downloads.yoctoproject.org/mirror/sources/pseudo-1.5.1.tar.bz2 Jun 12 10:49:10 JaMa: test case isolated to binutils... Jun 12 10:59:27 JaMa: its ld verses gold :/ Jun 12 11:57:52 Does this look wrong: DEPENDS_${PN}_class-nativesdk = "bison-native pkgconfig-native" ? It looks wrong to me... but do I have to make a -nativesdk for both of those dependencies now? Jun 12 12:04:56 i.e. should it be DEPENDS_${PN}_class-nativesdk = "bison-nativesdk pkgconfig-nativesdk" ? Jun 12 12:05:49 Denwid: yes, DEPENDS_${PN} is wrong, should just be DEPENDS Jun 12 12:06:01 oh, ok =) Jun 12 12:06:14 Denwid: I suspect they should be -native though not -nativesdk Jun 12 12:06:46 if it's expecting bison and pkgconfig to be used during the build, those would need to be the versions that run on the build host i.e. -native Jun 12 12:07:06 yes that makes sense Jun 12 12:56:54 hi Yocto, I'm trying to handle hotplugging of ethernet cables to my device, I see that there is no recipe for netplug or ifplugd (or did I miss them?). What should I use to manage this? Jun 12 13:03:24 PulkoMandy: netplug or ifplugd can be enabled in busybox .. Jun 12 13:16:16 maxin: oh, didn't think of looking there! thanks Jun 12 14:06:19 khem: ping? Jun 12 14:20:26 bluelightning: It's really easy to add auto completion in the SDK setup scripts: http://pastebin.com/WZVU7Jxd . I would really like this but i don't know if it can cause troubles with shells other than BASH. Toughts? Is it ok if I submit this to the mailinglist? Jun 12 14:22:25 Denwid: hmm... that does look like a good change, however I think those options may be bash-specific judging by the posix manpage for read: http://pubs.opengroup.org/onlinepubs/9699919799/utilities/read.html Jun 12 14:23:27 Yeah that's what I was afraid of... Jun 12 14:23:31 Too bad Jun 12 14:24:01 on the other hand, the script already starts with #!/bin/bash Jun 12 14:24:05 so actually it should be fine Jun 12 14:25:31 Ok I'll submit it to the mailinglist then and look out for other feedback, it would be really cool :) Jun 12 15:16:38 blloyd, I figured out the problem I was having yesterday Jun 12 15:16:50 I needed to explicitly set KMACHINE_beaglebo ?= “beaglebone" Jun 12 15:16:58 (where beaglebo was my custom machine definition) Jun 12 15:17:08 because normally it does KMACHINE_$machine = “$machine” Jun 12 15:17:40 Hi is there any documentation about adding (usb) wifi functionality to the image :) Jun 12 15:17:57 I've tried using the linux-firmware- Jun 12 15:18:34 but that's not enough (as it's only the firmware and I in my test the wifi usb module didn't trigger any loading of module, nor could I insmod one) Jun 12 15:18:46 or modprobe Jun 12 16:08:41 that's nice to know mkeeter Jun 12 16:09:23 My understanding is that if dev-pkgs is included in an image, all sources should be included on the image. Is this correct? Jun 12 16:10:05 blloyd: no. Jun 12 16:10:12 dev-pkgs means every -dev package Jun 12 16:10:16 so headers and .so symlinks Jun 12 16:14:22 that makes sense I have core-image-sato slightly customized so I end up with IMAGE_FEATURES="debug-tweaks package-management dev-pkgs x11-sato splash x11-base ssh-server-dropbear hwcodecs tools-sdk tools-profile dbg-pkgs tools-debug". Jun 12 16:14:30 But I'm not seeing any headers at all. Jun 12 16:15:40 use bitbake -e to see what value IMAGE_FEATURES actually gets set to? Jun 12 16:16:00 that's where that line came from. Final setting for IMAGE_FEATURES Jun 12 16:16:12 First thing I did when it didn't work as expected. Jun 12 16:31:56 find a bunch of headers, so it mostly working. But it appears yocto-kernel-dev doesn't get brought in. Jun 12 16:35:36 right, so dev-pkgs actually means "for every package installed, install the corresponding -dev package" Jun 12 16:36:58 so what subtle difference am I missing? This image boots, linux-yocto is the kernel used. So why wouldn't it be considered installed? Jun 12 16:37:16 because the kernel package is called linux-yocto Jun 12 16:37:54 do you just want kernel headers, or really yocto-kernel-dev Jun 12 16:37:57 oops, typo. linux-yocto-dev is what I am looking for. Again the package with kernel headers is missing Jun 12 16:38:14 kernel headers should work Jun 12 16:38:45 I need to be able to build an out of source device driver and run gdb (and hopefully gdb-server) on the device. Jun 12 16:41:14 blloyd: you *should* have a kernel and kernel-dev package then. check your installed package list? Jun 12 16:42:14 blloyd: fair warning, changes to IMAGE_FEATURES alone don't cause the image do_rootfs task to re-run right now, you'll want to clean and rebuild the image or force it with -f/-C Jun 12 16:42:26 https://github.com/MentorEmbedded/meta-mentor/pull/219 Jun 12 16:42:44 i think you're already past that, but for the future Jun 12 17:16:49 Jefro: halstead: are there details about the voting portal that is being turned on for YP? Jun 12 17:17:56 darknighte, I know how it works but I don't know what's been written up about it. Jun 12 17:19:40 darknighte, It's basically a private web poll on www.yoctoproject.org. Jun 12 17:35:28 can somebody explain Jun 12 17:35:38 why modprobe doesn't work on yocto :) Jun 12 17:35:41 generated image Jun 12 17:35:58 or why modules-load.d doesn't load my modules :/ Jun 12 17:39:32 are the .ko files for the modules actually present on your image? Jun 12 17:40:12 yep they are Jun 12 17:40:21 If I insmod them, they actually work ;) Jun 12 17:40:32 it's for in-tree kernel modules Jun 12 17:40:44 worth a shot :P Jun 12 17:41:07 so linux-mainline (using beaglebone (black)) compiles the rt2800usb module Jun 12 17:41:24 but I need to chain load a lot of modules to have this one work (obviously) Jun 12 17:41:32 and what's worth a shot? Jun 12 17:41:48 the question? Jun 12 17:41:51 my stupid answer Jun 12 17:42:58 maxtothemax: Don't worry about it ;-) Jun 12 17:43:48 modules.dep issue maybe? Jun 12 17:44:42 maxtothemax: I think you're right Jun 12 17:45:26 maxtothemax: the modules I need aren't in there, now I could add them manually. But I don't think that's the purpose Jun 12 17:54:07 halstead: thx. I was curious how it compared to the heliosvoting site we are considering for OE. Jun 12 17:54:12 so the question now is whether Yocto should be spitting out valid and complete modules.dep files or if this is as designed, and to be honest I don't know the answer Jun 12 17:54:39 iirc depmod-cross is run at image creation time to create modules.dep in the rootfs Jun 12 17:55:45 I'm still not that into yocto :D Jun 12 17:58:28 dvhart, maybe you can weight in on this Jun 12 17:59:05 fenrig says rt2800usb isn't showing up in modules.dep on his yocto images Jun 12 17:59:19 and he can insmod it but not modprobe it Jun 12 17:59:39 i can insmod it with a pathname (but you guys know this) Jun 12 17:59:51 i'm editing the modules.dep file Jun 12 17:59:55 and after Jun 12 17:59:59 executing "depmod" Jun 12 18:00:07 I can modprobe the modules :/ Jun 12 18:00:25 well, this is strange: I manually add linux-yocto-dev to IMAGE_INSTALL and get NOTHING RPROVIDES linux-yocto-dev, and it complains that linux-yocto-dev incompatible with machine. Jun 12 18:00:41 so yocto is somehow not catching that particular module when building the image Jun 12 18:00:44 okay :D Jun 12 18:00:48 after doing a depmod Jun 12 18:00:50 manually Jun 12 18:00:55 I have a proper working modprobe Jun 12 18:01:04 (the editting wasn't necessary) Jun 12 18:01:14 I think we can classify this as soms sort of bug? Jun 12 18:01:21 I believe that's what depmod does Jun 12 18:01:38 :D haven't got much experience with it Jun 12 18:02:03 * maxtothemax searches the bugzilla for anything related Jun 12 18:02:16 thx man ;) Jun 12 18:02:43 I can include ${PREFERRED_PROVIDER_virtual/kernel}, but ${PREFERRED_PROVIDER_virtual/kernel}-dev causes an incompatible machine? I'm confused Jun 12 18:04:29 blloyd, linux-yocto-dev is a different recipe Jun 12 18:04:35 so it has a different COMPATIBLE_MACHINE value Jun 12 18:05:12 fenrig, what yocto release are you using? Jun 12 18:09:18 fenrig, can't find any related bugs sorry Jun 12 18:09:48 euhm the newest yocto release Jun 12 18:09:53 but i'm using the angstrom variant :) Jun 12 18:10:00 the plot thickens Jun 12 18:10:07 I think it was 1.6 Jun 12 18:10:09 I don't know anything about angstrom :) Jun 12 18:10:12 yeah 1.6 is newest Jun 12 18:12:00 :) Jun 12 18:12:19 though I'm making my own image, based on the core-boot-minimal Jun 12 18:12:28 but I'm using angstrom distro Jun 12 18:12:56 maxtothemax: for now it's fixed. It's just for demonstrating something to my teachers :p Jun 12 18:12:56 so Jun 12 18:16:12 Oh yeah, do you guys know how to work with connman ? Jun 12 18:16:37 on the arch wiki I found the connmanctl command, but it isn't working on my embedded board Jun 12 18:45:44 fenrig: try #connman Jun 12 18:46:14 rburton: thx Jun 12 18:47:11 connmanctl should work, and if it doesn't then that needs to be fixed if its a problem in how we're building it Jun 12 18:47:17 don't try and run it as a non-root user, you need to be root Jun 12 18:47:19 (or xuser) Jun 12 18:52:04 i'm using the serial connection to my board Jun 12 18:52:10 so I'm root by default Jun 12 19:52:31 I see gawk got uprev'd a couple days ago. Anyone else seeing build failures as it errantly tries to remove files from the host builder in /usr ? Jun 12 19:59:52 ok, root cause found. The removal (rm -f) "passes" if the build host doesn't have those files, e.g. ubuntu 14.10. But my host has those gawk files and hence it fails -EPERM. Jun 12 20:00:06 ouch Jun 12 20:00:16 gotta love host contamination bugs Jun 12 20:00:32 make[4]: Entering directory '/mnt/home/paul/poky/build/tmp/work/corei7-64-poky- Jun 12 20:00:32 linux/gawk/4.1.1-r0/build/extension' Jun 12 20:00:32 for i in filefuncs.la fnmatch.la fork.la inplace.la ordchr.la readdir.la readfi Jun 12 20:00:32 le.la revoutput.la revtwoway.la rwarray.la testext.la time.la ; do \ Jun 12 20:00:32 rm -f /usr/lib/gawk/$i ; \ Jun 12 20:00:35 done Jun 12 20:00:37 rm: cannot remove '/usr/lib/gawk/filefuncs.la': Permission denied Jun 12 20:00:39 rm: cannot remove '/usr/lib/gawk/fnmatch.la': Permission denied Jun 12 20:00:41 ... etc Jun 12 20:01:24 * paulg will raise a bug now and perhaps sniff around on it more tomorrow AM. Jun 12 20:09:11 https://bugzilla.yoctoproject.org/show_bug.cgi?id=6440 in case anyone was following along. Jun 12 20:09:14 Bug 6440: normal, Undecided, ---, saul.wold, NEW , gawk fails to install on some hosts who install files similar to yocto paths Jun 12 20:09:31 * paulg pats yocti on the head. Jun 12 20:10:30 what are yocto paths? Jun 12 20:11:36 i'm assuming he means when the target paths match up with the layout of the host. if both use /usr/lib/gawk Jun 12 20:13:53 Yeah, I getting amused by the number of different ways people use the word "Yocto" Jun 12 20:15:46 yeah, the paths quoted above: '/usr/lib/gawk/filefuncs.la' and such. Jun 12 20:16:52 I meant "yocto paths" as simply the default paths yocto placed the above binaries at. Nothing more, nothing less. Jun 12 20:18:21 checking ubuntu, they don't install any .la in their gawk deb it seems. Jun 12 20:21:41 * paulg moves the files aside as a workaround Jun 12 20:24:41 Hi Jun 12 20:24:57 any thoughts on how I should obtain a connmanctl command on my yocto build :) Jun 12 20:25:08 I did a find | grep connman Jun 12 20:25:17 and didn't find any connmanctl :/ Jun 12 20:51:23 for connmanctl to be in the image you'll have to use "connman-client" in the IMAGE_INSTALL (or somewhere else depending on how you're used to doing it) Jun 13 01:22:38 Question: I did a "bitbake sudo" to create a new rpm for my minimal image (which has rpm). When I try to install it with rpm, I get a list of failed dependencies. How do I determine the recipe I need to back to fulfill each dependency? It's not straightforward. Jun 13 01:23:29 For example: error: Failed dependencies: libc.so.6(GLIBC_2.15) is needed by sudo-1.8.9p5-r0.armv6_vfp Jun 13 01:26:46 Or: How do I determine which recipe provides a particular file? Is that information somewhere? **** ENDING LOGGING AT Fri Jun 13 02:59:59 2014