**** BEGIN LOGGING AT Thu Oct 23 03:00:00 2014 Oct 23 06:07:47 anyone active in the channel? Oct 23 06:08:04 nope, we're all in standby Oct 23 06:13:21 I am trying to bootup yocto kernel and ext3 files using -kernel and -hda option. Oct 23 06:13:51 I want to increase the size of the file system by creating .img file using the qemu-img command and passing the image with the -hdb option Oct 23 06:14:02 but that resulting in kernel panic Oct 23 06:14:09 any idea how to solve this? Oct 23 06:16:58 not me, specifically. it sounds a bit more like qemu-specific issue, i've got little experience there. Oct 23 06:30:45 it is possible that OE servers might be unavailable from 30minutes from now (0700UTC till 1300UTC) due to emergency Network maintenance (bad fibre) Oct 23 06:47:00 Only tangentially related, but I bet some embedded systems people occasionally use USB<->serial gizmos: http://www.eevblog.com/forum/reviews/ftdi-driver-kills-fake-ftdi-ft232/ Oct 23 07:34:23 bufferunderflow, if you won't mind having oe create a larger image for you, you could look into IMAGE_ROOTFS_SIZE or IMAGE_ROOTFS_EXTRA_SPACE Oct 23 07:35:57 AndersD: i thought about suggesting that, but that actually is another effect. e.g. a bigger rootfs versus a second filesystem Oct 23 07:56:21 LetoThe2nd, ah, I missed the -hdb option in his question. Yes, you're right, that's another question. Oct 23 08:03:36 AndersD: of course its possible that it was a xy-questionm then your answer would probably be fitting best :) Oct 23 08:05:58 is pkg-config config supposed to work with an sdk? For me the bin exists but my sysroot has no pkgconfig/*.pc files. Is there an option to populate the sdk sysroot with the pkg-config files? Oct 23 08:27:39 irontia, I've got plenty of *.pc files in my sdk generated using populate_sdk Oct 23 08:27:59 so, yes, pkgconfig should work Oct 23 08:33:59 thanks AndersD, so i need to figure out where i broke it Oct 23 08:36:09 Yep, I guess so. I just checked, and I've not only got the pc-files in my own sdk, but also in a standard sdk generated from poky usibn bitbake core-image-minimal -c populate_sdk Oct 23 08:45:15 morning all Oct 23 09:40:56 morning everyone. Is it possible to run a board specific non-qemu image in qemu? And how do I specify the hardware in that case? Oct 23 10:36:53 Hi ! bitbake -c devshell does not start-up . I see the window close instantly again. (1.6.1) - there is no error msg. tmux is installed Oct 23 10:37:53 dl9pf: anything in log.do_devshell for the recipe? Oct 23 10:38:59 http://paste.debian.net/128283/ Oct 23 10:39:52 nothing useful Oct 23 10:40:08 hmm yes, it's not being very helpful is it Oct 23 10:44:44 dl9pf: it will be trying to run: tmux split-window "{command}" Oct 23 10:47:39 dl9pf: is there a temp/run.do_terminal* file for the recipe? Oct 23 10:48:15 hmm - yes Oct 23 10:48:17 sec Oct 23 10:52:33 http://paste.debian.net/128284/ Oct 23 10:55:47 pseudo not in path/library path it seems (pseudo-native was built actually) Oct 23 11:00:41 ok, that definitely shouldn't happen Oct 23 11:01:07 let me try a fresh build in a new folder. Oct 23 11:19:00 i wonder wether i can distribute the sdk in a different way. I'd prefer to build a debian package instead of self-extracting shell script. Oct 23 11:45:05 When I have a sysroot, is there a good way to find our where a certain file came from? Oct 23 11:45:40 The sysroot was created by bitbake and I still have all logs etc. Oct 23 11:46:16 I can use find and grep, but that's slow. Oct 23 11:49:49 Hello! I have an issue with building for the raspberrypi. I'm using the daisy branch and the meta-raspberrypi from yoctoproject.org. The issue is that I can't get hardfloat to work Oct 23 11:50:34 if I set TUNE_FEATURES = "armv6 arm1176jzfs callconvention-hard" then TARGET_FPU is set to "soft" Oct 23 11:50:41 no matter what I set in local.conf Oct 23 12:04:17 grubasy: I don't thin the rasperrypi is able to do hard float Oct 23 12:06:57 abelloni, AFAIK it can and there is a meta that enables it, but its older than daisy :( Oct 23 12:07:01 see http://www.raspberrypi.org/forums/viewtopic.php?f=56&t=59282 Oct 23 12:07:39 yeah, basically, you want armhf, not gnueabihf Oct 23 12:07:54 I'm not sure this is supported anywhere :) Oct 23 12:10:21 abelloni: I see Oct 23 12:11:48 I'll try this https://github.com/VCTLabs/meta-raspberrypi/wiki/Raspberry-Pi-Image-Setup (they mention: Cross-toolchain and SDK: poky-eglibc-x86_64-meta-toolchain-gmae-arm1176jzfshf-vfp-toolchain-gmae-1.5.sh (armv6 / vfp, raspberrypi compatible) ) and see what happens :) Oct 23 12:23:28 uhm Oct 23 12:23:56 http://www.gentoogeek.org/rpihf-ipk/all/Packages Maintainer tag is set to poky@yoctoproject.org so perhaps it is supported after all? :) Oct 23 12:46:55 bluelightning: seems like the line 152 is the culprit - multi-line (http://paste.debian.net/128284/) Oct 23 12:47:56 yes - that is it . Oct 23 12:48:01 dl9pf: aha... that would break things Oct 23 12:48:54 unset mc did the trick . Oct 23 12:51:22 hmm, looks like we will need to fix the script creation to handle this Oct 23 12:51:55 how did that value get in there? from googling it seems related to the recent shellshock fixes at least Oct 23 12:52:12 well, as "mc" is not part of the target environment anyway Oct 23 12:55:34 the wrapper is part of midnight commander. it is defined in the host env as mc () (from /etc/profile.d/mc.sh Oct 23 14:01:17 dl9pf: I'm still not quite sure how to reproduce this Oct 23 14:02:12 dl9pf: I've looked at what that script is doing (at least in the version I can find) and at least with my version of bash I don't get that BASH_FUNC variable set Oct 23 14:02:28 dl9pf: which distro are you running this on? Oct 23 14:35:13 suse 13.1 and debian 7.6 Oct 23 14:35:36 as soon as they have mc.sh in profile.d both have that defined. Oct 23 14:35:52 mc () Oct 23 14:35:52 { Oct 23 14:35:52 . /usr/share/mc/mc-wrapper.sh Oct 23 14:35:52 } Oct 23 15:10:15 Hey, I need to create a wrapper for a cross compiler that supplies certain flags, one of the build tools used doesn't have the ability to pass -L flags to the compiler (and executes it via a fixed name it looks up in the path) Oct 23 17:31:08 Are there any examples of/best practices for doing this in bitbake/yocto/oe? Oct 23 19:18:37 Hmm, bitbake-layers show-appends could use a bit of love. e.g. argument to hide skipped stuff, or to show only appends in a given layer. /me adds to todo Oct 23 19:23:57 any yocdroid folks around? Oct 23 19:32:37 Should PREFERED_PROVIDER go in machine.conf or distro.conf? Does it matter? Oct 23 19:32:50 (Same query with PREFERED_VERSION) Oct 23 19:33:36 most often thats a distro decision, with the exception of kernel and u-boot and the like, though those are usually handled via COMPATIBLE_MACHINE and DEFAULT_PREFERENCE Oct 23 19:35:13 kergoth: but both should work? Oct 23 19:35:37 yep, I'm isolating the machine related things in machine presently Oct 23 19:35:48 as long as the var is set in the config metadata, bitabke doesn't care what file happens to set it Oct 23 19:35:56 ordering matters if its set in multiple places, obviously Oct 23 20:06:58 oh http://git.openembedded.org/openembedded-core/commit/?h=daisy&id=5163eb2c3c492d3640aa0628c29f6b8d446a8261 Oct 23 20:07:04 how was it tested ? Oct 23 20:08:24 it adds couple of patches to src uri but I dont see them in commit Oct 23 20:10:35 I dont see it on mailing list either Oct 23 20:13:20 [OE-core] [URGENT] Daisy is broken Oct 23 20:16:19 well someone's git cherry-pick was messed I can see that :) Oct 23 20:16:41 conflict in cherry-pick are PITA Oct 23 20:30:08 The reason I'm asking is that `PREFERRED_PROVIDER_kernel-devicetree-overlays = "linux-ti"` doesn't seem to want to work Oct 23 20:34:24 nvm, bblayer interference issue with includes Oct 23 20:36:19 Is there a preference between meta-beagleboard and meta-ti? Oct 23 20:57:21 nvm again, using the beaglebone, not the beagleboard so I've removed meta-beagleboard in favor of meta-ti Oct 23 21:08:53 Q: looking for hints on how to debug terminal popping up then dying immediately with "bitbake virtual/kernel -c menuconfig" after trying several different OE_TERMINAL setups with no success, log simply states "Python function do_menuconfig finished". Kernel builds okay, everything else about the setup is fine, just can't figure out why menuconfig is borked (worked a couple of weeks ago). Oct 23 21:10:15 fishey2: http://article.gmane.org/gmane.linux.embedded.yocto.general/21564 Oct 23 21:12:24 denix: thanks Oct 23 21:49:39 hi Oct 23 21:49:46 need some yocto help Oct 23 21:50:02 Found my menuconfig problem; was due to shellshock fixes to bash. poky/meta/classes/terminal.bbclass was not properly passing thru BASH_FUNC_xx to run script. Oct 23 21:50:15 hello Oct 23 21:50:19 anyone there Oct 23 23:37:29 hello there Oct 23 23:37:42 has anyone tried running SPEC INT2006 under yocto-linux? Oct 23 23:38:31 I got spectools to compile file but once I start running the benchmaks... perlbench fails badly Oct 24 00:47:59 Hi. I have working Yocto installation with my project that I didn't touch for about a month. Now I tried to continue work on that project and got sudden "AttributeError: 'module' object has no attribute 'pythonCacheLine'" Oct 24 00:48:39 I did a quick google search and found same error message in the logs of this room, but unfortunately it was unanswered :( **** ENDING LOGGING AT Fri Oct 24 03:00:00 2014