**** BEGIN LOGGING AT Wed Apr 24 02:59:57 2019 Apr 24 06:42:56 good morning Apr 24 06:45:00 New news from stackoverflow: Yocto: RRECOMMENDS_${PN} = "kernel-module-g-ffs" didn't install g_ffs.ko into image Apr 24 08:31:05 building linux kernel, ${workdir}/git/.../dtb/xx.dtb is not available in ${workdir}/build/.../dtb/ where kernel_devicetree = xx.dtb got error "install: cannot stat 'xx.dtb' no such file" Apr 24 08:31:50 how to get this file in ${workdir}/build ? Apr 24 10:45:41 New news from stackoverflow: BitBake recipe for custom setup.py Apr 24 11:51:08 rburton, not sure which patchbomb is that? Apr 24 11:51:42 kanavin: not sure what i was asking now Apr 24 11:52:02 rburton, the mesa change Apr 24 11:52:28 ah yes, your series with mako in seemed to change mesa Apr 24 11:58:04 rburton, tried reverting the mako change, buildhistory-diff stays silent Apr 24 11:58:15 ok weird randomness here Apr 24 11:58:19 cheers Apr 24 12:25:31 morning Apr 24 13:04:24 Hrm... what would be the bitbake workflow to fetch all sources to build a sdk target matching an image? Apr 24 13:16:10 New news from stackoverflow: Populate_ext_error in yocto Apr 24 13:17:21 bitbake -c populate_sdk image-name Apr 24 13:17:26 florian, ^^^ Apr 24 13:19:03 Crofton: sure... but this actually builds the sdk - I would like to fetch whatever is required and yes it differs from what its required to build the image. Apr 24 13:20:01 Like bitbake -c fetchall -c populate_sdk $image Apr 24 13:20:24 but from what I know multiple -c are not supported Apr 24 13:20:42 ah yes Apr 24 13:20:52 I thought your question was too easy :) Apr 24 13:21:08 don't suppose anyone would like to expand/categorise some of https://wiki.yoctoproject.org/wiki/FutureMigrationGuide ? Apr 24 13:21:17 rburton: RP: ^ Apr 24 13:21:27 have run out of time for today, will pick it back up tomorrow Apr 24 13:21:34 but help would be appreciated Apr 24 13:47:25 a recipe which DEPENDS on my new wxwidgets 3.0.4 recipe is failing: /bin/sh: 1: wx-config: not found Apr 24 13:47:47 this is in the do_compile Apr 24 13:48:18 and of course I am indeed using wx-config in the make file to generate the compile options Apr 24 13:49:28 i do not understand how the wx-config, which is apparently supposed to be generated by the wxwidgets recipe, gets included into the build environment for my dependent recipe Apr 24 13:49:59 here is the wxwidgets recipe: https://paste.fedoraproject.org/paste/BqlST8sMRAVOMoQaYb89tA Apr 24 13:50:37 i'm thinking in the binconfig_sysroot_preprocess() function, right? Apr 24 13:51:55 yates: have a look in sysroot-components/, find the wxwidgets folder and check that wx-config is in usr/bin Apr 24 13:52:07 (need to add a helper to do that0 Apr 24 13:55:36 rburton: is sysroot-components somewhere under the sources/poky/build/tmp folder? Apr 24 13:55:41 it's not in mine.. Apr 24 13:55:45 remember this is morty Apr 24 13:55:54 i have a sysroots Apr 24 13:56:00 is that what you mean? Apr 24 13:56:01 oh right just dig around the target sysroot then Apr 24 13:56:05 (and upgrade) Apr 24 13:56:53 also have you submitted your new wxwidgets recipe to meta-oe yet Apr 24 13:57:05 you wouldn't want this one! Apr 24 13:57:08 it don't work.. Apr 24 13:57:15 yet Apr 24 13:58:26 wx-config is nowhere under my tmp/sysroots Apr 24 13:59:01 how do i start the submission process ? Apr 24 13:59:50 is there any interaction between what's in the sysroots and the FILES_xyz of the wx recipe? Apr 24 13:59:51 you send the patch Apr 24 14:00:11 send it via email? Apr 24 14:00:24 no FILES is the packages, the sysroot is a subset of whatever is installed Apr 24 14:00:37 have a look at the original wxwidgets recipe and see what it does that you're not Apr 24 14:00:39 ok, well then i'm stumped Apr 24 14:00:46 meta-oe has README for submission Apr 24 14:00:47 this used to work... Apr 24 14:00:54 ok, i'll check the README Apr 24 14:02:26 rburton: does the "gtk2" in "BINCONFIG_GLOB = "arm-fslc-linux-gnueabi-gtk2-unicode-3.0" refer to gtk library 2.0? Apr 24 14:02:36 presumably Apr 24 14:02:38 that may be the problem - this is gtk3 Apr 24 14:03:25 tias? Apr 24 14:04:18 this gtk version naming is the death of me.. Apr 24 14:04:27 just glob that bit Apr 24 14:04:37 all of that is overly-precise Apr 24 14:04:43 what if you're not doing a arm-fsl build Apr 24 14:06:05 for now i'll just s/gtk2/gtk3 Apr 24 14:06:43 get specific to work first, then generalize.. Apr 24 14:08:20 plus our hardware guy needs this build quickly, i didn't expect to have to deal with this. Apr 24 14:08:35 as i said, i THOUGHT it was working, but it's been a few weeks since i messed with it. Apr 24 14:08:45 yates you saw the existing wx recipe? I think in meta-ros? Apr 24 14:09:07 Crofton: presumably at one point, yes, i think so. it's 2.9.5? Apr 24 14:10:13 not sure, just something I saw ageas ago Apr 24 14:17:19 Hmm, I'm trying to parse recipes with tinfoil to examine variable values and it's complaining that I don't have all the hosttools installed Apr 24 14:17:42 Do I really need the host tools just for parsing? Apr 24 14:19:31 Crofton: what is ros? meta-ros? Apr 24 14:21:01 https://github.com/bmwcarit/meta-ros Apr 24 14:21:06 is this the right place to look at the existing meta-oe recipes? https://layers.openembedded.org/layerindex/branch/master/layer/meta-oe/ Apr 24 14:26:27 that will point you to the git repo Apr 24 14:32:19 ok, thank you Apr 24 14:32:47 i'm still batting 0. change gtk2 to gtk3, rebuilt, and i still get no wx-config Apr 24 14:33:25 and there is no wx-config anywhere in tmp Apr 24 14:33:59 is it possible i'm battling cache state? Apr 24 14:46:29 after building just the wxwidgets recipe, there is no wx-config anywhere under tmp Apr 24 14:47:04 there is not *wx*config* anywhere under tmp Apr 24 14:47:11 s/not/no/ Apr 24 14:47:38 the wxwidgets recipe builds without error though Apr 24 15:07:04 Hi, I'm getting an error each time I build my image. I have never had this error for 2 years. "opkg_conf_load: Could not lock /home/xxx/xxx/poky/build-xxx/tmp/work/beaglebone-poky-linux-gnueabi/image-xxx/1.0-r0/rootfs//var/lib/opkg/lock: Resource temporarily unavailable.". I've tried dropping thread BB_NUMBER_THREADS ?= "1" but I've got the same problem. I've also remove the "tmp" folder, but again I had the same error. Apr 24 15:20:24 I maybe found the problem, I'm calling "opkg status ..." in the "pkg_postinst_${PN}". Apr 24 15:24:05 rburton: in a recipe under morty, where is the ${B} path, typically? is it under tmp? Apr 24 15:24:43 yates: as per the documentation, its the same as ${S} unless you change it yourself or a class does (autotools.bbclass does) Apr 24 15:24:48 bitbake -e will show this for you Apr 24 15:25:35 ok Apr 24 15:25:55 ok, and what the heck is ${@get_binconfig_mangle(d)} ??? Apr 24 15:26:30 $ git grep get_binconfig_mangle Apr 24 15:26:30 meta/classes/binconfig.bbclass:def get_binconfig_mangle(d): Apr 24 15:26:31 that Apr 24 15:27:48 ok Apr 24 16:33:14 is there a way to *just* update yocto's recipes without modifying things in bitbake and scripts Apr 24 16:33:15 ? Apr 24 16:46:43 Striking7: use bitbake + oe-core instead of poky Apr 24 16:46:53 poky is bitbake+oe-core glued together into a single repo Apr 24 16:47:06 thanks rburton, will do Apr 24 16:47:14 of course, recipe updates may depend on changes to bitbake, if youre moving between branches Apr 24 16:47:20 Yet again rburton saves me time :) Apr 24 16:47:56 Yeah, I figured as much. I *know* how bad it sounds, but I modified things in scripts (mostly devtool) and fetchers in bitbake/lib/bb/fetch2 Apr 24 16:49:32 surely dealing with that is what branches are for Apr 24 16:49:39 just rebase Apr 24 16:51:52 good call Apr 24 17:42:24 denix: did you have a chance to look at the 2.5.3 release? Apr 24 18:08:45 RP: looks good to me Apr 24 18:34:20 Hi, how may I declare an environment variable in the script "pkg_preinst" that will be accessible to the "pkg_postinst" ? I tried export MY_VAR=1 in pkg_preinst and then echo $MY_VAR in pkg_postinst, but it show nothing. Apr 24 18:40:17 they're completely different shell scripts run at completely different times. Apr 24 18:48:32 kergoth: When the rootfs is produced I understand, but when we install the package manually, it's not the same shell between pre_inst/post_inst? Apr 24 18:48:49 why would it be? they're still run at two different times Apr 24 18:48:52 kergoth: If I use a file in /tmp to keep the information, this should work ? Apr 24 18:48:55 one before files are extracted, one after Apr 24 18:49:30 * kergoth that could work, yeah, but you'd have to somehow make usre it gets cleaned up, even if preinst ran but postinst didn't due to a failure part way through the install Apr 24 18:50:49 kergoth: why would it be? they're still run at two different times -> I thought that when running "opkg install my_package" all the trigger functions were executed with the same shell session (the one of the user who executed the command). Apr 24 18:51:20 no, you run a pcakage manager, which is already a different process from the shell you're running it Apr 24 18:51:28 and then that process spawns a new shell instance to run the script Apr 24 18:51:35 so you're already at *least* two processes away from your shell Apr 24 18:51:47 a child process can't change the environment of its parent Apr 24 18:52:03 kergoth: Interesting, I didn't know. Apr 24 18:53:04 https://en.wikipedia.org/wiki/Environment_variable Apr 24 18:54:01 kergoth: In pkg_postinst, I would like to test for the presence of another package. But as long as I execute "opkg status" or "opkg list", this produce a kind of deadlock because the opkg lock file. So the yocto build image failed. So I though that getting this information on pkg_preinst, put it into /tmp/xxx and getting information back on pkg_postinst would do the trick. Do you have a better idea ? Apr 24 18:54:33 i wouldn't use the package manager to check that, i'd check for the actual binaries or files installed by the other package Apr 24 18:54:38 but *shrug*, your call Apr 24 18:57:33 kergoth: I'm agree with you. By the way, thank you for the lesson :) Apr 24 18:57:43 no problem Apr 24 18:58:20 if you ever want to learn more about processes, environment vars, etc, the book advanced programming in the unix environment is a good, though verbose, resource. good reference material Apr 24 19:01:56 kergoth: This is a huge book, I bought it in epub version 2 years ago, but with the arrival of my 2 children and I could not finish it. But thank you for confirming that it was a good buy. Apr 24 19:05:01 its basically the bible, indeed. Apr 24 19:11:36 any suggestions on the best yocto/openembedded book? Apr 24 19:42:11 erakis: i don't know many that Apr 24 19:42:19 ve read it cover to cover, but at least you can look stuff up as reference Apr 24 19:42:54 i think i read about 50% + a few specific sections beyond that Apr 24 20:41:43 how can I check whether UPSTREAM_CHECK_URI for a recipe works locally? Apr 25 02:18:26 New news from stackoverflow: Yocto: Nothing provides python-re-native **** ENDING LOGGING AT Thu Apr 25 03:00:02 2019