**** BEGIN LOGGING AT Tue Aug 06 02:59:58 2013 Aug 06 04:05:45 * liuyq is away: Away Aug 06 04:19:15 another one... Aug 06 05:17:50 * liuyq is back (gone 01:12:09) Aug 06 07:25:15 morning all Aug 06 07:25:34 hi eren Aug 06 07:28:05 O O Aug 06 07:28:05 * Aug 06 07:28:05 /¯\ Aug 06 07:34:31 shawn187: oh, what was that? there was an alightment error Aug 06 07:38:53 morning all Aug 06 07:42:03 morning silvio_ woglinde eren und #oe Aug 06 07:42:32 morning silviof Aug 06 08:06:45 hi silviof Aug 06 08:08:13 morning all Aug 06 10:28:03 hiii Aug 06 10:28:17 hi snkt Aug 06 10:36:26 Is it possible to build debian toolchain for ARM nokia900 machine? Aug 06 13:06:19 hi, is there a sure fire method to find which recipe out of 4000+b tasks depends on gtk? I'm running a headless app and would like to do some spring cleaning Aug 06 13:07:19 OWayne__: bitbake -g your-image-name Aug 06 13:07:20 OWayne__: will -g option work for you? Aug 06 13:07:27 :) Aug 06 13:07:39 OWayne__: that will produce some .dot graphs you can grep for "-> gtk" or similar Aug 06 13:19:55 bluelightning: seems to have worked, thanks, image is in need for a hair cut Aug 06 14:47:44 gm Aug 06 14:51:37 gm Aug 06 14:56:05 morning likewise_, Crofton|work Aug 06 14:57:04 hi crofton Aug 06 15:18:35 OT, but anyone else using an omap5 evm board? Aug 06 15:43:32 is somebody using kernel 3.10? here, dropbearkey never finishes which can be reproduced by 'dd if=/proc/timer_list bs=2048' Aug 06 15:44:14 ...argh... this does not finsih on my Fedora 19 kernel neither... Aug 06 16:15:16 * kergoth grumbles Aug 06 16:21:54 Hi there! Am I in the right channel to ask this question?: Aug 06 16:21:54 I wanted to add the following git repository to SRC_URI += ...: git://github.com/DFE/darmok.git;protocol=git;tag="darmok_v0.2"; Aug 06 16:23:08 But when building it with bitbake I get the following error: Aug 06 16:23:08 Log data follows: Aug 06 16:23:08 | DEBUG: Executing shell function do_kernel_checkout Aug 06 16:23:08 | ERROR. The branch 'meta' is required and was not Aug 06 16:23:08 | found. Ensure that the SRC_URI points to a valid linux-yocto Aug 06 16:23:08 | kernel repository Aug 06 16:23:09 | ERROR: Function failed: do_kernel_checkout (see ... Aug 06 16:23:35 any ideas what might have went wrong? Aug 06 16:23:42 I'm trying to debug it for hours now.. Aug 06 16:23:44 linux-yocto kernels have special handling for its kernel repository, not sure Aug 06 16:24:21 so it's rather a question for this channel?: #yocto Aug 06 16:24:37 correct.. if you are changing linux-yocto, you need to have the right 'meta' (meta data) layer to describe various configuration settings and such Aug 06 16:24:53 there is a guide on working with the BSP/kernel items, and it may be what you need.. Aug 06 16:26:25 chrysha: meaning you want multiple SRC_URI's in a kernel recipe? Aug 06 16:27:34 so bluelightning, back to yesterday's question... Aug 06 16:28:14 mr_science: sure - chrysha's situation is slightly different though Aug 06 16:28:29 if you have 2 git src_uri's in the same _git recipe, what about the source dir name defaulting to WORKDIR/git ? Aug 06 16:28:52 what happens to the second git repo? Aug 06 16:30:04 yes, one being a git repository. my project does not depend on the yocto kernel; therefore I don't have this meta information. so where do I get the meta information I need or how can I create it? Aug 06 16:30:44 chrysha: what kind of recipe are you making? Aug 06 16:30:56 chrysha: I'd suggest looking at the yocto project BSP guide and examining the linux-yocto repository Aug 06 16:31:49 I'm editing linux-yocto_3.4.bbappend Aug 06 16:32:21 chrysha: if you just want to build a kernel without the linux-yocto style meta branch you can do that by taking a copy of the linux-yocto-custom recipe under meta-skeleton and using that for your kernel Aug 06 16:33:14 I want to build the yocto kernel, but also to include code from an external git repository Aug 06 16:34:02 might be better off with a second recipe, if it's an out of tree kernel module or similar Aug 06 16:34:06 * kergoth shrugs Aug 06 16:34:43 yup Aug 06 16:37:50 our code is based on the yocto kernel, but we want to make some additions to it. the original code needs to be patched so that the device drivers can function properly (the code in the git repo adds new device drivers) Aug 06 16:38:23 chrysha: ok, in that case I think you do need to have a corresponding meta branch Aug 06 16:38:43 chrysha: can the new drivers be built as modules? Aug 06 16:39:00 chrysha: then you can make a separate recipe for them Aug 06 16:39:01 it can be both, compiled into the kernel and as kernel modules Aug 06 16:39:25 both should be possible, based on the kernel configuration Aug 06 16:39:28 I don't think we support dual GIT repo's as SRC_URI... not 100% sure. Aug 06 16:39:53 but didn't it work 2 years ago? Aug 06 16:41:11 likewise_: we definitely do Aug 06 16:43:02 so how does (busybox) start-stop-daemon end up putting the wrong PID in the pidfile? Aug 06 16:43:37 mr_science: hmm... maybe it's the pid of the shell that starts the process? Aug 06 16:43:42 of course when i stop it, i get warning: killing process 0: No such process Aug 06 16:44:02 it's not started with a shell... Aug 06 16:44:20 oh, you mean the init /bin/sh ? Aug 06 16:44:38 bluelightning: I'm not well informed anymore. Thanks for correcting me. Is there an example recipe demonstrating dual GIT checkout? Aug 06 16:46:09 likewise_: linux-yocto does this; I don't know of any other examples Aug 06 16:46:37 mr_science: hmm but if it's showing 0 then that can't be right Aug 06 16:49:40 that's when the pid is wrong Aug 06 16:49:59 if i echo the correct one, then stop works Aug 06 16:57:04 so the pid is always one less than the actual pid... Aug 06 17:04:21 bluelightning: the pid problem is the same whether i --exec /bin/foo or --exec /bin/bash -- -c "foo" Aug 06 17:04:39 mr_science: hmm, I'm not sure what is going on Aug 06 17:04:46 mr_science: definitely sounds like a bug Aug 06 17:04:51 sometimes i even get two PIDs... Aug 06 17:05:00 er... ?! Aug 06 17:05:17 this is with an in-house bin... Aug 06 17:05:49 i think the two pid thing was with my homespun init script that didn't use start-stop-daemon Aug 06 17:06:31 the old init scripts were all working fine until they threw a couple of new binaries in the mix... Aug 06 17:07:39 hmm, looks like an old busybox bug... Aug 06 17:07:58 i didn't think think our classic stuff was *that* old... Aug 06 17:09:49 ah I forgot you were mostly working with OE-Classic Aug 06 17:11:21 i switch back and forth almost daily... Aug 06 17:11:36 classic at work, poky master at home... Aug 06 17:12:27 mr_science: that's a pretty big gap to cover :) Aug 06 17:12:44 yup Aug 06 17:13:07 still a surprising amount of overlap, but many thinks have changed... Aug 06 17:13:21 mr_science: yes the devils are in the details Aug 06 17:13:27 but a F lot of them :) Aug 06 17:14:21 yeah, don't look too closely at my meta-rpi layer Aug 06 17:14:50 probably a lot of "don't do it *that* way now..." in there... Aug 06 17:15:31 * mr_science puts down the inti stuff to install a fresh build **** ENDING LOGGING AT Wed Aug 07 02:59:58 2013