**** BEGIN LOGGING AT Tue Sep 15 02:59:58 2015 Sep 15 06:11:12 hi all Sep 15 06:11:35 q: I want to add a file to my Yocto SDK (I make one with bitbake -c populate_sdk ) Sep 15 06:12:00 the idea is to add a 'toolchain.cmake' so my SDK will work with CMake by just including that file Sep 15 07:09:09 good morning Sep 15 08:14:14 hi all Sep 15 08:22:51 hey, could someone explain me how I can add an user that will be automatically logged-in in order to start a systemd user service as a normal user and not as root? Sep 15 08:25:56 you'll want one of extrausers or useradd Sep 15 08:25:58 https://www.yoctoproject.org/docs/1.8/ref-manual/ref-manual.html#ref-classes-extrausers Sep 15 08:26:02 https://www.yoctoproject.org/docs/1.8/ref-manual/ref-manual.html#ref-classes-useradd Sep 15 08:26:26 I'd guess useradd to create a user in the recipe that provides the service Sep 15 08:27:15 joshuagl: thank you! :) Sep 15 08:27:21 will it take care of auto-logging? Sep 15 08:37:00 kido: do you need that? I think you can use User= in the systemd service file. Sep 15 08:38:56 mhh indeed you are right, thanks! Sep 15 08:39:29 :) Sep 15 08:44:43 does anyone here know how to add things to the populate_sdk root ? Sep 15 08:44:53 I want to place a file there to instruct my CMake which compiler etc to use Sep 15 08:44:58 and I'd like it in the root Sep 15 08:45:38 * rink notes populate_sdk_base.bbclass seems to write things there, but I don't know ho to expand that Sep 15 08:47:44 morning all Sep 15 08:49:02 hi Sep 15 08:49:17 * rink points bluelightning upwards and hopes for help *puppy eyes* Sep 15 08:49:25 rink: :) Sep 15 08:49:50 rink: two ways - use TOOLCHAIN_TARGET_TASK_append to add packages, or create a function to generate the file and use POPULATE_SDK_POST_TARGET_COMMAND to call it Sep 15 08:50:23 hmm, my main confusion arises from that I don't really want to add a package per se Sep 15 08:50:46 just place file next to site-config-..., environment-setup-..., version-... Sep 15 08:51:03 right, I guess you probably want the latter in that case Sep 15 08:51:33 or maybe SDK_POSTPROCESS_COMMAND would be better Sep 15 08:53:46 hmm, that looks nifty *staring at buildhistory.bbbclass* Sep 15 08:54:19 and you'd use ${SDK_OUTPUT} within that function to get the path to where you want to create the file(s) Sep 15 08:55:03 I guess I can just do this in my image.bb file ? Sep 15 08:55:07 which inherits core-image Sep 15 08:59:02 yes Sep 15 09:55:25 bluelightning: hmm, I have 'SDK_POSTPROCESS_COMMAND += ' foo; ' Sep 15 09:55:27 and Sep 15 09:55:29 foo() { Sep 15 09:55:33 and I get Sep 15 09:55:43 function 'foo' doesn't exist Sep 15 09:55:54 both are in a .bb file Sep 15 10:09:24 rink: is the function properly closed? i.e. you don't have any stray }, or } indented instead of being at the beginning of the line? Sep 15 10:11:26 the implemention is Sep 15 10:11:28 foo() { Sep 15 10:11:30 echo 'hi there' Sep 15 10:11:31 exit 2 Sep 15 10:11:32 } Sep 15 10:13:38 if you run bitbake -e yourimage | less and search for "hi there" with '/' do you find the function? Sep 15 10:14:48 hmm Sep 15 10:15:32 * rink notes the variable name is 'delem_sdk_append_toolchain_cmake' Sep 15 10:15:38 and it showsup as 'delem_sdk' ? Sep 15 10:15:46 right, because of the _append Sep 15 10:16:01 # $delem_sdk Sep 15 10:16:01 # _append[toolchain_cmake] ast.py:161 [eval] Sep 15 10:16:01 # " echo ohai Sep 15 10:16:06 oh, that's nasty! Sep 15 10:16:20 what you're actually doing is saying "append to the function delem_sdk" if the override "toolchain_cmake" is set" Sep 15 10:16:43 another unfortunate consequence of _ being the override character Sep 15 10:17:00 well, thanks a lot man! Sep 15 10:18:15 * rink concludes _add_toolchain_cmake_ is a good function name as well Sep 15 10:26:50 hmmm Sep 15 10:27:41 using ${SDK_OUTPUT} places the file in sdk/image/ ... Sep 15 10:28:36 * rink tries ${SDK_OUTPUT}/${SDKPATH} Sep 15 10:32:34 rink: ah yes, sorry, that is almost certainly the path you want Sep 15 11:29:23 <[Sno]> otavio: around? wrt. openjdk-8 ... Sep 15 11:40:12 <[Sno]> otavio: when I understood your mail correctly, you mean: logically prepare your patchsets and mail oe-dev (meta-java) mailing list - didn't I? Sep 15 11:43:52 hi, is it possible that opkg search is buggy or I am doing something wrong? My package installs some files into /usr/lib which I can see in the output of opkg files mypkg, but then opkg search /usr/lib/libfoo.... does not return mypkg. Sep 15 11:47:58 oh, I am doing it wrong.. it needs the full path Sep 15 11:48:19 i.e. opkg search /usr/lib/libfoo...., it was not enough to use opkg search ./libfoo.... in /usr/lib. Sep 15 11:48:29 would be nice to have feature though :) Sep 15 11:49:01 [Sno]: yes. And also repport the segfault issue there. Sep 15 11:49:34 [Sno]: keeping one clean branch on top of master allows for more people to use this as base work and ultimately help to reproduce and fix the issue Sep 15 11:49:46 [Sno]: maybe khem could help in the segfault problem Sep 15 11:50:00 <[Sno]> otavio: I agree Sep 15 11:50:28 <[Sno]> I figured out some internal exception's which occur with zero (without shark) either Sep 15 11:50:51 <[Sno]> will fix them to have a zero running and rebase to master and make some clean, logically sane commits ;) Sep 15 11:51:41 <[Sno]> otavio: unfortunately it requires meta-oe updates for llvm ^^ Sep 15 11:51:45 [Sno]: thanks. Also if you want to cleanup the other recipes, be my guest ;-) Sep 15 11:51:50 [Sno]: not a problem Sep 15 11:51:51 <[Sno]> I rebase that patch against master either Sep 15 12:00:35 [Sno]: thx a lot Sep 15 14:19:02 hi! I have an extremely newbish question to ask. I need to build a version of linux for a freescale ARM processor on a WB-EDM-iMX6 board. Sep 15 14:20:19 I downloaded the build appliance and am using Hob to target qemuarm. Can I take the .ext4 files that are generated and put them on an sdcard to get linux on that platform? Sep 15 14:25:38 I could not get Hob to run with either iso or hddimg selected and manually editing the local.conf file to include IMAGE_FSTYPES ?= "sdcard" did not work as Hob overwrote those lines Sep 15 14:25:55 how do I use the .ext4 files? Sep 15 14:31:58 first of all - don't use hob ;) Sep 15 14:32:28 unfortunately Hob has a bit of a blind spot where it comes to custom image types Sep 15 14:32:28 and no, a file for qemu almost certainly won't boot on a random board. Sep 15 14:33:24 I have tried googling but have not been able to find a decent tutorial on how to do this Sep 15 14:33:28 it might drop into a rudimentary shell however, if there is a preexisting bootloader and kernel in use that just tries to jump into the image. if the kernel is sufficiently self-contained. Sep 15 14:33:57 Mike_Newb: well what does the vendor say, and offer as bsp/documentation? Sep 15 14:35:00 I inherited the board and it has an sdcard with yocto linux on there, but I wanted to do a clean install (on a different sd card) Sep 15 14:35:47 no additional documentation, or leftovers how the sdcard was created? Sep 15 14:36:10 because i guess you're in for a rough ride, then. Sep 15 14:37:08 on the other hand, isn't it just a standard wandboard? Sep 15 14:38:23 yes Sep 15 14:38:26 FWIW, otavio has a nice book out which even uses the wandboard as an example Sep 15 14:39:24 what is the title of the book? Sep 15 14:39:31 is Otavio the author? Sep 15 14:40:52 I'm guessing Embedded Linux Development with Yocto Project by Otavio Salvador is the book Sep 15 14:41:57 I will order a copy. Thank you for the pointer. Sep 15 14:45:22 there are stock images at wandbord.org, but their download links have been broken for a while now and they don't reply to requests to fix this Sep 15 14:49:19 those should be reproductible with the howto howto of the wandboard wiki Sep 15 14:49:36 IIRC it has pretty much a exact steps howto Sep 15 14:50:25 Mike_Newb: yeah, see http://wiki.wandboard.org/index.php/Getting_started_with_Yocto_on_Wandboard Sep 15 14:59:00 that looks good. thank you Sep 15 15:40:18 there is a way to force bitbake to parse again all the recipes? Sep 15 15:51:46 aj_c: before I answer that, what problem are you trying to solve? Sep 15 15:54:42 I rewrite a recipe that have a RPROVIDES to a package, i changed to point to another package but after run bitbake it still target to the old one Sep 15 15:57:26 aj_c: ok, reparsing isn't going to fix that Sep 15 15:58:04 can you pastebin the RPROVIDES part of the recipe? Sep 15 15:58:15 sure Sep 15 16:00:01 RP: autobuilder looks fairly idle, mind if I queue a build? Sep 15 16:06:22 joshuagl: yes, please hold off Sep 15 16:06:33 joshuagl: trying to sort the M3 build atm Sep 15 16:06:40 RP: ack, will hold Sep 15 16:12:09 joshuagl: looking a bit more, go for it, I think the M3 build is good Sep 15 16:13:21 RP: oh, great. Thanks! - build queued Sep 15 16:19:14 Mike_Newb: :) Sep 15 16:31:45 remind me, what's the script/tool for listing overlayed recipes? Sep 15 16:32:57 aha, bitbake-layers show-overlayed Sep 15 16:34:53 bluelightning: I'm editing packagegroup-core-boot.bb trying to get some flexibility to commands utils http://pastebin.com/dL4YgNvZ# Sep 15 16:36:03 aj_c: ok so you were actually changing RDEPENDS_${PN} ? Sep 15 16:37:00 yes instead of hardcoded busybox it points to ${VIRTUAL-RUNTIME_anybox} Sep 15 16:42:38 my mistake it seems my variable at local.conf was poiting to busybox and that is why it was trying to pull it... Sep 15 16:43:01 aj_c: right... I did just try something similar and it worked Sep 15 16:43:35 aj_c: bitbake -e packagegroup-core-boot | less is a good way to check what's really being set Sep 15 19:20:39 halstead: qemuarm runs on the VM but keyboard mapping is whackyroot Sep 15 19:21:31 challinan, Interesting. The keyboard mapping via vnc? Sep 15 19:22:06 heh, good question - vmc seems to be working for LXTerminal, etc Sep 15 19:22:47 but when I launch qemu, typing 'b' yields s, 'c' yields d, and so on. no pattern that is obvious Sep 15 19:24:59 must be something w/ vnc. when I launch qemuarm from an ssh session, it works ok Sep 15 19:25:06 probably can ignore this for now Sep 15 19:29:36 hmm, building for minnowboard using 1.8/Fido yields this fetcher error: Sep 15 19:29:43 ERROR: Fetcher failure: Unable to find revision c33d39561807e1073ca412f1c771f43e4da75994 in branch meta even from upstream Sep 15 19:33:37 challinan, I don't have a quick solution but I'll look into it. Sep 15 19:34:04 i'm more interested in getting past this fetcher failure ;) Sep 15 19:34:47 challinan, That hash is missing from which repo? Sep 15 19:35:16 duh Sep 15 19:35:28 ERROR: Function failed: Fetcher failure for URL: 'git://git.yoctoproject.org/linux-yocto-3.14.git;bareclone=1;branch=standard/base,meta;name=machine,meta'. Unable to fetch URL from any source Sep 15 19:59:24 wonder how far along rburton got with bug 4634 Sep 15 19:59:26 Bug https://bugzilla.yoctoproject.org/show_bug.cgi?id=4634 normal, Low, 1.9, ross.burton, NEW , BitBake default 'dirs' in exec_func (${B}) causes build failures Sep 15 20:13:40 * kergoth ponders Sep 15 20:37:18 have we branched 2.0 yet? I'm wondering if I should hold back upstream submissions which are features that might not be appropriate for 2.0? Sep 15 20:56:52 * kergoth wonders why we haven't just renamed fetch2 to fetch yet Sep 15 20:58:40 could use a review of the patch i just sent to the bitbake list, it's another one that touches the fetcher, so careful review is important Sep 15 21:06:20 kergoth: *cough* tests *cough* Sep 15 21:25:55 yeah, ran the selftests, no problems there Sep 15 21:25:57 hrm Sep 15 21:31:53 * kergoth starts work on a selftest to test this case Sep 15 21:34:24 khem: are you around? Sep 15 21:44:41 where can i find a simple open source dual/quad core sbc? Sep 15 21:53:01 malkauns: there are a number of options available afaik. wandboard seems like one viable option for that Sep 15 22:11:02 kergoth: I've just realised we have a problem where values in OVERRIDES (like TARGET_ARCH) are in turn influenced by values in OVERRIDES Sep 15 22:11:27 that hurts my head Sep 15 22:11:49 kergoth: It hurts mine too. Do we just keep iterating the expansion until it stabilises ? Sep 15 22:12:21 kergoth: We've some bugs in multilib in particular related to this :( Sep 15 22:13:26 I'm tempted to say we just don't allow it, but there are probably valid use cases for applying overrides to vars in overrides. Eg. say I want to forcevariable TARGET_ARCH or something.. in which case I think we'd have to keep recursing until it stops changing, with some form of recursion check not unlike how we do MIRRORS Sep 15 22:13:27 hmm Sep 15 22:14:37 kergoth: the circular reference I've found is DEFAULTTUNE_virtclass-multilib-libXXX which changes TARGET_ARCH (through a whole set of vars) Sep 15 22:16:59 kergoth: probably put a limit of say 5 recursions and then if we hit that print an error asking for their configuration so we can laugh at it as being crazy? :) Sep 15 22:17:39 Yeah, exactly. the limit would be arbitrary, but as we've seen with mirrors, we don't want to risk infinite recursion Sep 15 22:17:45 hmm Sep 15 22:18:42 kergoth: There is also a secondary problem with the ??= operator not updating the dependency code properly meaning TARGET_ARCH isn't being tracked properly anyway :/ Sep 15 22:19:00 kergoth: I think I'll sleep on it and hope that helps ;-) Sep 15 22:34:00 can you continue expansion -removing- the current expansion from futures Sep 15 22:34:10 that would limit the number of expansions to the number of entries? Sep 15 22:34:42 (that is more or less what I sent for a patch to the mirror processing code BTW... seems to do a good job stopping infinite recursion or adding arbitrary limits) Sep 15 22:34:46 good idea, same approach you just took to fix mirrors Sep 15 22:34:49 heh :) Sep 15 22:35:16 and I think it's "obvious" to a user that once the iem has been processed, it's not expected to be processed again Sep 15 22:35:25 (at least I'd hope) Sep 15 22:35:26 aside: does https://gist.github.com/kergoth/aa5acae96424ac57236f seem reasonable? I'm not entirely happy with it, but unless we can figure out how to get env vars set in the devshell process but not be exported to its children, i can't think of a better way to do it Sep 15 22:37:30 only fear.. would the environment topdir or workdir conflict with anything running in the devshell, orwise it sounds good Sep 15 22:56:13 screw it, i'm just sending patches to teh list and if it's for stuff that doesnt' belong in 2.0, i'll leave it to the folks merging them to decide what to do with it if we haven't merged it yet, rather than holding off on emailing Sep 16 01:15:10 Is it OK to share the sstate cache directory as well as the downloads with different yocto builds. For example, if I am using the TI yocto and the freescale fsl-community etc? I am guessing it will be but just want to make sure there are no side effects of build A populating incorrect cache for build B.. **** ENDING LOGGING AT Wed Sep 16 02:59:59 2015