**** BEGIN LOGGING AT Wed May 13 02:59:59 2015 May 13 03:13:10 chankit: that depends on where you want them to end up. you can submit them to a particular layer, or create your own May 13 03:16:57 kergoth: what if I just want it to be within say....recipes-devtools? May 13 03:18:20 that's a directory inside a layer, not a layer May 13 03:18:26 recipes-devtools where? May 13 03:18:34 oe-core/poky? May 13 03:21:47 kergoth: let me try to put it another way. I have now recipes for LLVM (both cross and host compiler) and beignet OpenCL May 13 03:22:05 and looking at http://layers.openembedded.org/layerindex/branch/master/layers/ makes me even more confused :-( May 13 03:22:14 how so? May 13 03:22:30 you haven't answered my question. recipes-devtools where? May 13 03:22:34 kergoth: now I don't even know where should I put my recipes May 13 03:23:18 that depends entirely on what the layers are. if you're updating versions of existing recipes, then obviously you already know where they should go — the same layer they're already in May 13 03:23:24 s/layers are/recipes are/ May 13 03:23:44 kergoth: I used to put my recipes under meta/recipes-devtools. Hence I said recipes-devtools May 13 03:23:53 okay, meta May 13 03:23:56 that's actually useful information May 13 03:24:03 recipes-devtools tells me nothing, that's a dir that exists in like 40 layers May 13 03:24:13 meta is the main layer inside oe-core May 13 03:24:16 kergoth: my bad. Wasn't too informed about meta thingy May 13 03:24:21 so submit your changes to the oe-core repository to the oe-core mailing list May 13 03:24:27 what directoryi s meta in? May 13 03:24:49 it's either poky or oe-core, which i already mentioned 2 sentences after you asked your question May 13 03:25:28 poky May 13 03:25:50 okay, the poky repository is an integration of bitbake, openembedded-core, and meta-yocto May 13 03:25:59 the meta directory in poky is part of openembedded-core May 13 03:26:53 your best bet is to make your changes in the poky git repository (or the openembedded-core one) and send them to the oe-core mailing list using git send-email, but how you send the patches to the mailing list is really up to you May 13 03:27:06 every layer generally has a readm eindicating how you contact the maintainers and how you submit changes May 13 03:27:56 the poky mailing list would have been okay, but they'll just redirect you to the oe-core list if you send the changes there May 13 03:28:05 https://lists.yoctoproject.org/ May 13 03:28:15 http://lists.openembedded.org/mailman/listinfo/openembedded-core specifically May 13 03:28:43 most likely the yocto docs woul dhave a section covering how to submit changes, i don't know offhand since i haven't read it recently May 13 03:29:20 ok....will definitely check it out. I read something about this as well in README in poky dir. May 13 03:29:59 thanks May 13 03:31:01 np May 13 08:02:09 morning all May 13 08:03:57 morn May 13 08:50:51 hi all, I have a question about a simple copy with bitbake. May 13 08:51:57 shoot May 13 08:52:44 I want to copy a folder with desctop pictures into a ipk, so opkg can install this. I have a local file location of this pictures, not a tar.gz. I did SRC_URI this with file:// but the locations get messed up in the working directory. May 13 08:53:39 Is there a tut anywhere how to get a whle directory into a ipk? May 13 08:55:25 sirnoname, can you show the recipe, sounds like you just need the copy commands May 13 08:56:10 please use pastebin ;) May 13 08:57:34 I need a ipk, a final copy to RFS is not my goal. The ipk is for auto update with opkg update/upgrade. I try pastebin this ... May 13 08:58:25 the procedure would be the same for both cases May 13 08:59:46 Will install do copy to ipk or RFS? May 13 09:00:19 hi @ all May 13 09:02:23 i am new to yocto and tried building a version for my Wandboard-Quad. It worked so far -> cool !. But now i like to install additional packages and found "smart" installed by default as package manager. when i update smart -db and then try to search for packages like openssh it dioesnt find anything !? is there an official repository i coul use ? May 13 09:03:33 sirnoname: our images are constructed from packages, so we always produce a package May 13 09:05:13 then I do something complete wrong :( http://pastebin.com/3h41YezJ May 13 09:06:23 sirnoname: do you receive any warnings or errors? May 13 09:06:57 yes, the lic files get messed because of directory structure behind SRC_URI May 13 09:08:44 Steeltip: we don't officially provide package feeds since we aren't a distro May 13 09:09:00 Steeltip: there is however information on how to set up your own package feed and install from that: http://www.yoctoproject.org/docs/current/dev-manual/dev-manual.html#using-runtime-package-management May 13 09:09:08 ok, so i would have to build my own rpm's right ? May 13 09:09:40 Steeltip: unless you find a distro built by our system and use their images + packages, yes May 13 09:10:14 oh ... do you no such an "inofficial" / community distro / repo ? May 13 09:10:45 I don't know of one that has support for wandboard-quad, but that doesn't mean there isn't one May 13 09:10:55 our system builds packages automatically though, so it's not a difficult task for software we have a recipe for May 13 09:11:01 Steeltip: sounds like you actually want something like angstrom May 13 09:11:31 sirnoname: I'm not sure what that means I'm afraid, I was more after the actual message you receive May 13 09:12:08 have many many thanks for your informations and a nice day May 13 09:20:22 sirnoname: I think problem #1 is that your do_install closing } is not the first character on the line May 13 09:20:39 I wasn't aware but having just tested, that is significant May 13 09:25:08 other problems: (a) file:/// not file// in SRC_URI, (b) '*' as a source path in do_install isn't appropriate unless you set S to the subdirectory where the files are unpacked to (which will be ${WORKDIR}/home/myname/desktoppictures) and (c) if the destination location isn't one of the defaults you'll need to extend FILES_${PN} to point to that location so the files actually get packaged May 13 09:26:00 I will say that it is rather unorthodox to point a recipe to a file:// location outside of the metadata like this May 13 09:26:36 I changed the pastebin to http://pastebin.com/bK0CWzvW. It did successfully bitbake and do only a -dbg and -dev ipk in output. At /files there are two folders with pictures. May 13 09:27:15 yes, I would like do other and point to /files in the bb directory, but how? May 13 09:28:32 both ipk are empty May 13 09:28:51 that will be because either the files are not being installed, or they are not being packaged May 13 09:29:05 if it's the latter you will be getting a warning which you shouldn't ignore May 13 09:29:07 no recursive fuction? May 13 09:29:15 yes, there is May 13 09:29:30 looking at your updated file you haven't fixed the closing } May 13 09:29:34 as I wrote or different? May 13 09:29:46 hmm, I check May 13 09:39:40 now I get error message after moving the } to first character in line. But there is no error, only a warning that run.do_install quit with error code 1. May 13 09:43:18 sirnoname: could you please pastebin the entire output? May 13 09:44:33 Hello May 13 09:44:55 hi bryan May 13 09:45:34 i want to copy kernel dtb files in the /boot folder before preparing root file system May 13 09:45:54 can anybody tell me how can i achieve it May 13 09:46:30 i can see that the zImage is getting copied inside /boot folder automatically before creating tar.gz file May 13 09:47:35 I think that you need to set KERNEL_DEVICETREE May 13 09:48:58 i have one KERNEL_DEVICETREE_NAME which has dtb file name May 13 09:49:19 AFAICT that variable is not used by any of our core kernel classes May 13 09:49:29 only KERNEL_DEVICETREE May 13 09:50:10 maybe you have some custom logic in your BSP, I don't know - grep or git grep on the metadata you are using would confirm May 13 09:50:19 so you mean to say i need to set this variable with dtb file name May 13 09:50:27 yes May 13 09:50:54 and then i need to install it in ${D}/boot folder? May 13 09:51:20 the only thing you should need to do then is ensure that the kernel-devicetree package is included in your image May 13 09:52:40 actually i am able to get dtb file inside my DEPLOY_DIR_IMAGE folder.. May 13 09:53:06 The copy locations are invalid. I get messed directories. The SRC_URI do copy all * stuff with subdir to /home + complete SRC_URI path again. So the pictures are not in the /home directory where the workdir points to. Is this a problem if I do not use relative path in SRC_URI? May 13 09:54:27 but my requirement is that i want to copy this dtb file inside the tar ball root file system May 13 09:55:10 bryan: understood, that's what adding the kernel-devicetree package to your image will do May 13 09:56:17 sirnoname: that is a side-effect of that yes May 13 09:56:43 Howto use file:// relative to my bb file? May 13 09:57:23 lets say I have a files folder in same directory: file://files/*? May 13 09:58:22 or this : SRC_URI = "file:/./files/*" ? May 13 09:59:08 sirnoname: no, the "files" part is already in the path May 13 09:59:48 sirnoname: so just files://* May 13 10:02:38 @bluelightning : what is the value which needs to be set for KERNEL_DEVICETREE variable May 13 10:03:01 will it be only *.dtb file or ${S} path? May 13 10:06:10 yes i have some custom logic in bsp May 13 10:06:34 ncgs: you could try PACKAGECONFIG_append_pn-qtbase = " sql-mysql" - it did not help( May 13 10:07:05 bryan: full path starting with ${S} May 13 10:07:12 ok May 13 10:07:15 thanks May 13 10:08:03 ok, deems both file://files/* and file:// do same result. Now I got the working dir and some additional dirs inside with lic files /patches etc. The install ${WORKDIR} ${D} do not work, is this wrong? May 13 10:08:41 sirnoname: I've lost track now of what you have in the recipe, can you please pastebin the current version? May 13 10:10:39 no problem: http://pastebin.com/6Q9LrV1d May 13 10:10:44 @bluelightning : i have set the value with my dtb file but still it does not get into the image :( May 13 10:10:54 path May 13 10:11:18 i mean in tar ball May 13 10:11:47 bryan: did you also add kernel-devicetree to IMAGE_INSTALL in your image recipe? May 13 10:13:03 sirnoname: simply installing the entire ${S} to ${D} will not work - there are other files there, you are going to need to ensure they end up in a subdirectory May 13 10:13:17 i am using core-image-sato, you mean in core-image-sato.bb file? May 13 10:13:20 sirnoname: perhaps put them into a subdirectory within "files" and point to those May 13 10:13:40 bryan: you should really be creating your own custom image recipe at this point May 13 10:14:02 bryan: however there are other options just for testing, see http://www.yoctoproject.org/docs/current/dev-manual/dev-manual.html#usingpoky-extend-customimage May 13 10:14:59 sirnoname: I don't think install is recursive either, you'll probably need to use cp -R May 13 10:15:08 if you have paths within that May 13 10:15:34 ok I trie May 13 10:15:36 try May 13 10:21:29 ok, now I got the problem, that some files double exists and I want to insure the desktop picture ipk is installed as last package in my open embedded image receipt. May 13 10:26:16 sirnoname: if you mean other packages already install those images - that's not possible, the package manager expects that only one package provides a particular file May 13 10:26:39 there's also no mechanism for ordering package installation other than dependencies which would not help here May 13 10:27:33 At least I can do at my image bb a after installation script which overwrites them, as I do now ;) May 13 10:28:11 you could use a postinstall to do it, but that would potentially be overridden on any upgrade May 13 10:28:42 the proper way to replace files in other packages is to replace them at the source using a bbappend for the recipe that packages them May 13 10:33:18 ok, thanks and have a nice day May 13 11:12:42 good morning May 13 11:13:13 is there something written down what can be done when I'm in a devshell? May 13 11:13:31 I basically would like to run the different steps from bitbake manually May 13 11:15:39 ncgs: I don't know why that would not work... I would suggest examining the final value of PACKAGECONFIG by looking at bitbake -e qtbase | less May 13 11:16:34 jaeckel: you can run the run.do_* files in ../temp/ to re-run a task May 13 11:20:20 is it somewhere written down in which order they are executed? May 13 11:22:50 log.task_order I suppose :) May 13 11:29:32 jaeckel: right yes May 13 11:30:17 mh, how do I run these scripts that are nor bash neither python? May 13 11:30:25 like run.do_package May 13 11:37:07 jaeckel: hmm, I suspect you cannot... May 13 11:37:21 we do have devpyshell as well in 1.7+ May 13 11:38:26 so from 1.8 on they can be run? May 13 11:38:39 but I'm not sure how you would use that to run tasks easily either, bb.build.exec_func('do_taskname', d) doesn't seem to work May 13 11:38:59 mh okay May 13 11:38:59 no, that does not affect what goes into the runfiles for python tasks May 13 11:39:12 what are you trying to do? May 13 11:40:11 I'm trying to get find out why my rpm spec file contains stuff it shouldn't May 13 11:40:21 what sort of stuff? May 13 11:40:46 old path references of the recipe May 13 11:41:32 hmm, well, that should only be dependent on what gets installed May 13 11:41:55 right May 13 11:42:13 can I somehow extract the spec file from an rpm? May 13 11:42:35 well, you can see the spec file that we use to produce the rpm, you should find it in the workdir for the recipe May 13 11:43:05 pretty sure you can extract it as well, but it ought to be exactly the same as the specfile in the workdir May 13 11:43:25 mh, no there isn't one in the workdir May 13 11:44:34 well, unless you have rm_work enabled, there really ought to be one there if do_package_write_rpm has executed (which it would as part of a normal build of the recipe with RPM packaging enabled) May 13 11:44:43 I just checked and I can see one here for ncurses for example May 13 11:44:58 in which state is it created? May 13 11:45:10 which state? sorry I don't understand the question May 13 11:45:17 which step, sorry May 13 11:45:24 do_package_write_rpm May 13 11:46:56 now it's there May 13 11:47:01 perfect May 13 11:47:02 thanks May 13 12:40:04 interesting, after the update to daisy, I am getting this for ssh and scp commands: packet_write_wait: Connection to 192.168.0.1: Broken pipe lost connection May 13 12:40:35 this is happening after typing the password for the specified user. May 13 12:50:19 is this familiar to anyone? May 13 12:56:56 Try a different ServerAliveInterval ? May 13 12:57:28 or TCP keepalive? May 13 13:02:31 Hm, what's the right way to over-write /etc/default/ssh? I can't see an example but is set in image.bbclass. Is it fair game just to create a new copy in a bbappend somewhere or is that not playing nicely? May 13 13:04:06 pev: that should be fine May 13 13:06:36 bluelightning: Thanks, will give that a crack! May 13 13:20:59 pev: on the server side? Because there is not even a connection, so what to keep alive? :) May 13 13:21:06 this is when trying to connect. May 13 13:32:04 I am encountering an issue I don't know how to troubleshoot. I'm triggering a script when a USB drive is inserted via udev. The script performs an opkg install, but I get the error " opkg_configure: mypackage.postinst returned 255." However I don't get that error when I manually call opkg install. Does anyone have any tips on why the postinst fails when run via udev? May 13 13:45:32 FileNotFound: I would have to assume it's something about the environment that opkg is being called from - e.g. maybe PATH isn't the same, or something like that May 13 13:45:59 bluelightning: how do you see what the PATH is in udev? May 13 13:46:10 it doesn't seem straight forward :-/ May 13 13:46:35 FileNotFound: the way I'd do it is just print out the whole environment from that udev script, then you can go over it at your leisure May 13 13:46:56 "env" would do that May 13 13:50:00 is it safe to read a file in /var/run when booting, so basically in an init service given that the service itself writes the file in question, too? May 13 13:50:13 or can there by any race condition by not yet sync'd content of some sort? May 13 13:50:19 bluelightning: hmm, I'm not seeing PATH when I print out env May 13 13:50:38 that seems odd right? May 13 13:53:03 FileNotFound: I don't know, I'm not sure what is normal for the udev script environment May 13 13:53:43 lpapp: I doubt it, if any caching is involved the cache would be read from if it had been written to May 13 13:54:25 lpapp: only issue I can think of is that /var/run might be mounted at some point during boot, so before then it would just be a directory in the rootfs and after it would probably be tmpfs May 13 13:55:26 FileNotFound: it is possible that it is normal though; it's not as if you are running under any kind of user context there (other than naturally being root) May 13 14:00:22 bluelightning: you doubt race condition, right? May 13 14:01:07 lpapp: it might possibly be a race if mounting is involved and there's no dependency on that mounting service... May 13 14:06:51 bluelightning: it used to be /var/volatile, tmpfs. Is that still the case? May 13 14:21:21 lpapp: I don't know off the top of my head May 13 14:22:41 Anyone know why pkg_postinst and pkg_postrm are in PACKAGEVARS, but not pkg_preinst and pkg_prerm? Shouldn't the latter two affect the sstate just like the former two? May 13 14:24:02 Saur: they should yes, probably just an oversight May 13 14:24:19 bluelightning: Ok, thought it was something like that. May 13 14:24:44 I'll add a ticket, have a couple of more vars I think belong in PACKAGEVARS... May 13 14:24:59 bluelightning: http://cgit.openembedded.org/openembedded-core/commit/?id=0e326280a15b0f2c4ef2ef4ec441f63f55b75873 I think /var/volatetile does not apply anymore May 13 14:28:06 volatile* May 13 15:04:13 bluelightning: I think I found some strange thing going on in our layer again ^_^ May 13 15:04:30 bluelightning: remember that very long I mentioned that we have to override fstab to add partitions, like /home May 13 15:04:34 ago* May 13 15:05:08 now the change above that I showed added /run and since we were overriding fstab with .bbappend, I forgot to sync it up May 13 15:05:26 but this gets me to the question of: is there a nicer and more future proof way to handle file extensions, not .bb extensions? May 13 15:06:04 I think the rootfs command variable could work out; is that it? Or is there something recommended for this? I would assume that others might also want custom partitions, you see. May 13 15:09:48 morning May 13 15:10:10 good morning, kergoth May 13 15:14:48 hrrm. seems this commit breaks things: May 13 15:14:50 commit dce88e4a2c0b28022ef7fa63896aad35cc011cb8 May 13 15:14:50 Author: Chunrong Guo May 13 15:14:50 Date: Fri May 8 10:49:20 2015 +0800 May 13 15:14:50 polkit: do not hardcode the libdir May 13 15:14:50 May 13 15:15:18 ERROR: Command Error: exit status: 1 Output: May 13 15:15:18 Applying patch polkit-use-instead-of-hardcoded-lib.patch May 13 15:15:19 patching file data/Makefile.am May 13 15:15:19 patching file src/polkitagent/Makefile.am May 13 15:15:19 Hunk #1 FAILED at 78. May 13 15:16:08 both zeddii and I see it independently... May 13 15:19:00 lpapp: if you do it that way it will be wiped out on upgrade, so it really needs to come from the recipe May 13 15:19:25 lpapp: (well, if you don't have target packaging as you probably don't, that may not be an issue) May 13 15:20:01 there isn't a generic mechanism for extending individual files, nor a specific one for /etc/fstab May 13 15:20:04 paulg: strange that it didn't fail in my world builds May 13 15:20:59 in process of testing a simple revert.... May 13 15:22:06 oh wait... May 13 15:22:17 looks like local bbappend fallout. Ignore me. May 13 15:22:44 NOTE: recipe polkit-0.112-r0: task do_populate_sysroot_setscene: Started May 13 15:22:53 good May 13 15:23:54 * zeddii swims around the bowl May 13 15:24:04 watch out for the log May 13 15:27:49 paulg. that's an o'henry bar May 13 15:28:09 bluelightning: so what would you recommend? What would you do if you were me? May 13 15:29:44 what is the best way or at least most practical with Yocto to have custom partitions? May 13 15:44:54 lpapp: for now, just update your fstab copy May 13 15:46:02 ok, do you see this changing in the future though? May 13 15:46:37 I don't know of any plans to May 13 15:47:00 doesn't mean it won't, though May 13 15:47:02 and I cannot think of any simple solution either May 13 15:47:12 right, it's a tricky problem May 13 17:09:15 hello all May 13 17:10:41 i need a little help with using a freescale layer, first time openembedded and yocto user. May 13 17:12:18 hello belen2 May 13 17:13:59 hello jc_need_layer_he May 13 17:14:32 i am looking for some really 'simple' help to get started with a freescale layer May 13 17:15:22 i am very new to openembedded and yocto May 13 17:15:53 jc_need_layer_he: most likely I will not be able to help. But ask away: if someone can help, they'll answer May 13 17:18:56 if i want to compile for a board supported in the meta-fsl-arm layer, do i change the distro = "poky" in the local.conf? May 13 17:19:55 or just the MACHINE ??= "quemuarm" May 13 17:22:06 jc_need_layer_he: you should only need to change the MACHINE value May 13 17:22:22 I have used the Yocto Project Quick Start to setup my development pc (debian 8, gnome), the example built (core-image-sato) May 13 17:22:58 jc_need_layer_he: here's an old but likely to be valid quick tutorial: http://www.ossystems.com.br/blog/2013/04/15/yocto-with-boundary-devices-nitrogen6x-5-steps-only.html May 13 17:24:00 Just use fido instead of daisy as branch. May 13 17:24:01 when i change the machine to MACHINE ??= "imx23evk", do i still bitbake core-image-minimal ? May 13 17:30:28 checking out the ossystem help now, thank you. May 13 18:06:05 hmm, it seems openssh is broken in Yocto for me May 13 18:06:23 Changing UsePAM yes to UsePAM no in the sshd_config made my ssh/scp problems vanihs May 13 18:06:26 vanish*, even **** ENDING LOGGING AT Thu May 14 02:59:58 2015