**** BEGIN LOGGING AT Tue Oct 14 02:59:59 2014 Oct 14 08:03:37 halstead: ping Oct 14 08:03:52 i've got some contacts for you WRT power adapters Oct 14 08:04:09 LetoThe2nd, pong! Oct 14 08:05:48 you can call +49 22188841520 (large music shop and rental) or +49 2103252300 (rental company) Oct 14 08:06:30 halstead: my old roadie friends thought those *might* be able to help you Oct 14 08:07:23 Thank you LetoThe2nd. We'll get in touch. Oct 14 08:07:37 no problem, good luck Oct 14 08:08:36 * halstead smiles. Oct 14 08:09:38 in case i mistyped the phone numbers, the company websites are http://www.musicstore.de/de_DE/EUR/ and http://aventem.net/ Oct 14 08:36:51 morning all Oct 14 08:38:37 Morning bluelightning ! Oct 14 08:38:51 morning halstead Oct 14 08:38:58 you're at ELCE I presume? Oct 14 08:39:28 bluelightning, Yes. Scrambling for DevDay again. At least my part of it is pretty solid. AFAIK. Oct 14 08:39:53 halstead: cool... good luck :) Oct 14 11:06:00 hi, I am slightly as to what installs this because the following command returns nothing, and I cannot find this file in my yocto tree either: opkg search /etc/rcS.d/S04modutils.sh -> Got a clue? Oct 14 11:06:16 it is using modprobe, but I am not really having depmod available on the target since it is 72 KB, and I would like to change that to insmod. Oct 14 11:06:23 but I need to find what exactly generates it! Oct 14 11:07:31 lpapp: isn't that a symlink to something like /etc/init.d/modutils.sh Oct 14 11:08:12 erbo: sure, it is, but still opkg search does not find that. Oct 14 11:08:19 so what installs this? Oct 14 11:08:38 opkg doesn't return anything for the link target either? Oct 14 11:08:49 nope :( Oct 14 11:09:13 ah! Oct 14 11:09:31 seems it is coming from modutils-initscripts based on a Yocto find command. Oct 14 11:09:33 I mean in the tree. Oct 14 11:09:40 opkg cannot identify the file, apparently. Oct 14 11:09:54 poky/meta/recipes-kernel/modutils-initscripts/modutils-initscripts.bb ? Oct 14 11:10:03 ah Oct 14 11:10:08 I was too late :) Oct 14 11:10:34 perhaps it is better to make modprobe optional? Oct 14 11:10:43 otherwise I need to install 72 KB with busybox, and our target is very limited. Oct 14 11:10:53 I would not waste it on such a trivial thing, personally. Oct 14 11:11:36 if you know exactly what modules you need at what time I guess insmod should be just fine Oct 14 11:11:37 [ -e /sbin/modprobe ] || LOAD_MODULE=insmod Oct 14 11:11:46 seems it is already have some kind of fallback. Oct 14 11:11:56 so I assume the best way is to find a way to get rid of modprobe on the target. Oct 14 11:12:10 if depmod is not coming with busybox, getting depmod on the board does not make so much sense after all. Oct 14 11:12:57 ah, ok, modprobe comes from busybox, so we either ship both or none. Oct 14 11:12:59 sounds fair. Oct 14 11:14:47 ah, heh, actually both fall back to /bin/kmod Oct 14 11:17:00 Is there a way of getting an autotools project to automatically source the environment-setup script from en OE sdk ? Oct 14 11:17:13 sorry for being slightly offtopic Oct 14 11:28:35 kroon: sure, why not, you just need to know the SDK install path, no? Oct 14 11:30:07 lpapp, but where would I include the env script ? Is there a autoconf macro for including it, so that for instance PATH in all generated Makefiles will be correctly set ? Oct 14 11:30:51 kroon: not sure, I prefer to source the environment script on demand. Oct 14 11:31:09 I mean when I build for the target, and usually do it manually for one shell session, not always. Oct 14 11:31:22 lpapp, thats what I kind of want to avoid, manually sourcing the env script Oct 14 11:31:31 since the developer would need to specify the SDK path, it is one-one step either way, isn't it? Oct 14 11:31:52 unless every dev has the same path, and you hard code it. Oct 14 11:34:09 lpapp, thats the case in our setup, every developer will have an identical environment, and we want to make it easy for them Oct 14 11:37:22 kroon: strange, but fair enough ^_^ Oct 14 12:04:01 erbo: grep -rn DEPMOD ../meta/recipes-core/busybox/busybox-1.20.2/defconfig -> I am not convinced that it is a good idea to disable that by default. Oct 14 12:04:17 given that modutils-initscripts sort of depends on it. Oct 14 12:05:12 basically here: if [ ! -f /lib/modules/`uname -r`/modules.dep ]; then [ "$VERBOSE" != no ] && echo "Calculating module dependencies ..." depmod -Ae Oct 14 12:05:15 fi Oct 14 12:29:36 Hello Oct 14 12:30:47 How do i fetch the source code and other files from recipe? Oct 14 12:34:22 like bitbake $RECIPE -c fetch ? Oct 14 12:36:09 LetoThe2nd: something like that, thanks Oct 14 12:48:48 hi Oct 14 12:48:51 I have added a new recipe file in meta/recipes-support/ folder Oct 14 12:49:01 When I am doing bitbake for the individual recipe I am getting below error: Oct 14 12:49:09 QA issue : Package was already stripped, this will prevent future debugging! Oct 14 12:51:25 CAn anyone please help me Oct 14 12:53:16 well just by reading the error message it sounds like the packages own build process is already stripping it, and therefore bitbake cannot control that Oct 14 12:53:32 (e.g. provide a stripped and a non-stripped version) Oct 14 12:53:53 so i personally would go and look at the build process used Oct 14 12:55:52 rjv: http://www.yoctoproject.org/docs/1.7/ref-manual/ref-manual.html#qa-issue-already-stripped Oct 14 12:56:14 (this section was just added to the in-development manual for the upcoming release) Oct 14 12:56:16 is there any way to change the error to warning Oct 14 12:56:36 I see the files are copied to the correct directory in tmp folder Oct 14 12:59:54 I have taken the atr file from below link Oct 14 12:59:55 http://www.citrix.com/downloads/citrix-receiver/linux/receiver-for-linux-131.html Oct 14 13:00:20 Sorry tar file Oct 14 13:02:51 my bb file contents are at http://pastebin.com/PvTj10M8 Oct 14 13:03:48 rjv: ideally, find out why the debug symbols are being stripped and disable it Oct 14 13:04:04 well look at the documentation that comes with the source on how to disable stripping Oct 14 13:04:15 there are hints in the link I posted above Oct 14 13:04:49 usually its some flag thats handed to the configuration or compilation stage, just as bluelightning's link points out, yes. Oct 14 13:05:30 alternatively you can set INSANE_SKIP to skip the error (as documented in the manual), but that's hacking around the problem Oct 14 13:05:55 well, insanely skipping can be entertaining at times, though. Oct 14 13:14:22 I have to made following cahnges in meta-yocto/conf/distro/poky.conf Oct 14 13:14:24 file Oct 14 13:15:07 moved options already-stripped and installed-vs-shipped from ERROR to WARN Oct 14 13:15:14 so now it is building Oct 14 13:15:17 thanks Oct 14 13:19:04 LetoThe2nd: heh Oct 14 13:29:56 Has the version numbering of RPMs changed to add a decimal point? e.g. "tzdata-2013d-r0" is now "tzdata-2013d-r0.0" Oct 14 13:30:16 smart upgrade is showing every single package needs upgrading Oct 14 13:30:21 the last number is from PR service Oct 14 13:30:55 presumably you just enabled that? Oct 14 13:31:43 I've taken over the project from someone else, I guess they didn't have it enabled Oct 14 15:29:03 Dear all, is there a way to pipe all stdout to console UI when running "bitbake -c testimage " ? Oct 14 15:30:54 I've noticed a few packages that are using src uris like ${VARIABLECONTAINERSERVER}/path/to/binary. Some packages like the yocto kernel beig able to easily specify a local server makes sense to me. Is there a risk of breaking something by doing this type of URL for like the kernel (master really hurts for updates)? Oct 14 15:44:33 mikhail_: -v ? Oct 14 15:45:22 blloyd: wouldn't PREMIRRORS be the best way to handle that kind of situation? Oct 14 15:52:45 blueligtning: my understanding of premirrors is it checks there for the download files (the ones places in downloads). For a yocto-kernel, this entails downloaing the entire git repo again when the hash is changed. Typically it's a small change from lasttime, but 600+ megs has to be downloaded. Oct 14 15:53:19 -v works during normal build tasks, but with testimage, it does not add much useful info Oct 14 15:54:50 where a bare local repo can be effectively kept up to date efficiently and only the changes get pulled across the network. Then a .tar.gz can be made of that image no big deal. Oct 14 15:56:42 mikhail_: I think that's just the way it has been implemented Oct 14 15:57:33 bluelightning: hm, is there a workaround that does not involve piping log files to stdout manually? Oct 14 15:58:06 mikhail_: not that I am aware of at the moment, sorry Oct 14 15:58:28 looks like the meta-kernel-dev layer implements just what I asked about. Oct 14 15:59:20 blloyd: I'm concerned that if the current mirroring system is not working we should try to fix that... meta-kernel-dev is for development, it's not really intended as a workaround for this kind of issue Oct 14 16:03:46 why I am asking here. It seems inefficient and one of the sites I'm working from has a very poor inet connection. The downloading the full git repo to get the next revision is a painful way to get a minor change. Once the revision is retrieved, later recompiles use it fine. But that first get is awful. Oct 14 16:04:47 the kernel is just the one I've noticed as it is very big and I've been working directly with it lately (trying to move a machine from 3.4 to 3.14) Oct 14 16:06:03 the git mirror tarball doesn't contain the hash Oct 14 16:06:04 hmm, I'm not sure I'd class that as a minor change though Oct 14 16:06:23 we keep each linux-yocto kernel version in separate repositories currently, that probably doesn't help matters Oct 14 16:06:24 it will download whatever it can find, then if necessary git fetch to get what's missing if the SRCREV isn't included in that repo Oct 14 16:06:28 the dev also brings over ${AUTOREV} which I don't want. Oct 14 16:06:36 at least, thats how it typically works, maybe linux-yocto is special Oct 14 16:08:56 so if it is working properly it should extract the download, git update if needed, then update the download again? Oct 14 16:17:26 bluelightning: thanks for clarifying, we really need this functionality, would this question be appropriate for the subscribers of the yocto mailing list? Oct 14 16:22:58 mikhail_: I would suggest filing this as an enhancement request in bugzilla Oct 14 16:23:40 blloyd: if that's not the way it currently works, then it seems to me that it ought to Oct 14 16:45:25 thanks blue. I will try to do some checking into what it is really doing and file bug/enhancement request as appropriate. If it isn't pulling the whole thing, I will have to see what the huge bandwidith is caused by. Oct 14 16:47:16 bluelightnin: thanks, Created https://bugzilla.yoctoproject.org/show_bug.cgi?id=6841 Oct 14 16:47:17 Bug 6841: enhancement, Undecided, ---, richard.purdie, NEW , '-v' switch and testimage Oct 14 16:47:41 blloyd: thanks Oct 14 16:49:29 mikhail_: thanks, I've adjusted the categorisation slightly (it'll almost certainly be a change to the testimage implementation rather than bitbake) Oct 14 16:50:25 bluelightnin: thanks, :) I was not sure - that sounds good! Oct 14 19:13:49 Is David here? Oct 14 19:51:58 Hi all, I have a general question about yocto/bitbake and how an image is prepared Oct 14 19:52:43 I have a cmake package that builds a library and some executables. My recipe results in all the files being built and staged in TMPDIR/image Oct 14 19:53:34 Library in TMPDIR/image/usr/lib, binaries in TMPDIR/image/usr/bin, and a file in TMPDIR/image/etc/udev/rules.d/89-device.rules Oct 14 19:54:18 However, when the final image is put together only the library is present. Do I need to do anything special to make sure those other files are included? Oct 14 19:54:56 Are binaries omitted by default or anything like that? Oct 14 20:27:27 return_to_sender, it should work right away, unless your CMAKE rules miss something. As a workaround you can append a do_install() {...} to your recipe to install the missing files **** BEGIN LOGGING AT Wed Oct 15 01:30:49 2014 **** ENDING LOGGING AT Wed Oct 15 03:00:00 2014