**** BEGIN LOGGING AT Fri Nov 25 03:00:00 2016 Nov 25 07:01:55 Why I am seeing The TMPDIR: /home/atomar/bin/Sysdk/build/tmp can't be located on nfs.? Nov 25 07:04:46 Ramose: because your tmp directory cant be located on a NFS mount. I suspect your home directory is located on a NFS server? Nov 25 07:10:31 nrossi: Thanks but How do I check it Nov 25 07:12:15 Ramose: you mean how to check if a specific directory is located on NFS? try the mount command Nov 25 07:12:42 yeah Nov 25 07:12:43 ok Nov 25 07:31:37 core-image-sysdk Nov 25 08:37:00 Good morning. Does Yocto have bug tracker for packages (i.e. packaging issues), like linux distros do? Nov 25 08:54:14 off-topic: I'm having issues with nano on minicom, it doesn't display properly. Any hint? Nov 25 08:54:38 rned Nov 25 08:59:35 aV_V, I have had problems using nano via serial too... just using vi helps Nov 25 09:00:35 eduardas_m: thats too old-school for me >.< Nov 25 09:01:40 it is not as bad as you might think... also available in basically any Linux environment Nov 25 09:35:36 Reading the docs on kernel incremental config I now know how to enable new features. But the documentation is not clear on how to disable a feature? Nov 25 09:35:47 I am referring to http://www.yoctoproject.org/docs/2.2/kernel-dev/kernel-dev.html#changing-the-configuration Nov 25 09:36:08 My guess is that I have to add the commenetd-out version of the parameter? Nov 25 09:36:26 eg # CONFIG_FOOBAR is not set Nov 25 09:36:38 or should I use CONFIG_FOOBAR=n? Nov 25 09:37:04 Either way an example in this specific part of the docs will probably help a lot for other users as well Nov 25 09:52:43 hello Nov 25 10:09:08 nross: Yeah,home directoy is mounted on NFS Nov 25 10:45:31 deva: the "is not set" variant Nov 25 10:45:40 AFAIK Nov 25 10:46:08 I agree we should improve the documentation if that's not made clear Nov 25 12:09:50 hia, is this channel active ? Nov 25 12:11:27 I have started to work with Yocto and I am coming from a buildroot mindset, I have just build my image with bitbake core-image-minimal but I would like to clean the build while keeping the generated toolchain and downloads. What's the right command to feed to bitbake ? Nov 25 12:20:13 pedr0_: bitbake has what's called the 'sstate' cache, which re-uses anything that didn't change. The typical cleaning is deleting the 'tmp' folder in the build directory Nov 25 12:20:29 The downloads will be kept, sstate will be kept, and the next build will only build what changed. Nov 25 12:23:00 right, I am also keen in keeping thte toolchain as I do not want to re-build the GCC, would that work in that respect ? Nov 25 12:23:34 but there is not bitbake command for it is it ? Something like "make clean" Nov 25 12:23:57 Getting these Failure expanding variable while bitbake, how it can be debugged Nov 25 12:26:24 Does Yocto have bug tracker for packages (i.e. packaging issues), like linux distros do? I can't find it Nov 25 12:26:44 OK instead of packages, in Yocto it's "recipes" Nov 25 12:26:47 SpaghettiCat: if its a recipe in oe-core (meta/) then bugzilla.yoctoproject.org Nov 25 12:27:02 terminology is that a recipe generates packages Nov 25 12:27:31 so source dpkg / rpm spec / bitbake recipe are all the same concept Nov 25 12:27:50 rburton: I see. Thanks for the info. I'm looking for bugs about connman, and I don't see a section for it in bugzilla, so I assume it's not in oe-core Nov 25 12:28:29 I have just removed the build/tmp directory and run the build again, it has just taken 3 minutes, it is hard to believe it recompiled everything but the toolchain. Are you sure ? Nov 25 12:31:20 it reused the sstate cache Nov 25 12:31:31 so it just restored the built packages, and then build an image Nov 25 12:46:40 What Toaster calls a "Project", is that a distro? Nov 25 12:47:02 never used it but i think distro is just a varialbe you set in a distro Nov 25 12:47:05 erm Nov 25 12:47:12 i mean a distro is just the variable you set in a project Nov 25 12:47:34 Ok Nov 25 12:47:52 I was just firing up toaster for the first time.. I think I'll leave it alone for now. Nov 25 12:49:30 Just seemed something to mess around with on a Friday afternoon ;) Nov 25 12:51:17 I think I'd rather spend time learning more about the Distro layer. :p Nov 25 12:53:08 Hi, could anybody hint me in the right direction on how to generate a checksum like sha1sum for my build output via COMPRESS_CMD_sha1sum e.g. in poky/meta/classes/image_types.bbclass. Can't figure it out somehow... :/ Nov 25 12:56:07 build output = my-image.rootfs.tar.bz2 Nov 25 12:59:05 As the bz2 suffix is also a compression type (COMPRESS_CMD_bz2) I tried "tar.bz2.sha1sum" as IMAGE_FSTYPES, but that didn't work either. Nov 25 13:12:36 How do I invalid the sstate cache in order to build everyhing but keeping the toolchain and the downloads ? Nov 25 13:14:02 I just want to be sure my image comes from the latest sources - but I do not want the toolchain to be recompiled and downloads to be fetched again Nov 25 13:14:54 rebuilding the toolchain doesn't take very long so the easy way is to just mv the sstate Nov 25 13:15:50 you could delete everything in the sstate folder that isn't in a native directory (named after your distro, or 'universal', so rm sstate/?? -rf Nov 25 13:16:02 that should delete everything that is for the target Nov 25 13:16:03 bibake -c cleansstate , shouldn't that do the trick? Nov 25 13:16:12 if you list every recipe that you need to clean, yes Nov 25 13:16:45 personally i'd rebuild the toolchain too, it's only going to add a few minutes if you want to be sure you rebuild from scratch Nov 25 13:17:01 same for me. Nov 25 13:28:43 would that result in a re-downloads of all the tarballs ? Nov 25 13:29:28 can I just build the toolchain so I can have an idea about how long it takes on my machine ? Nov 25 13:32:34 it's a bit of a pain to rebuild something I do not actually need to, as the toolchain, it looks the quickest way though. Nov 25 13:34:07 your downloaded tarballs are in a different directory, so a delete of tmp won't delete them Nov 25 13:35:04 or a delete of sstate. Nov 25 13:35:50 tarballs are kept in whatever you set DL_DIR to, so if thats under tmp/ then move it Nov 25 13:36:52 gcc-cross-[arch] binutils-cross-[arch] is most of the toolchain i guess Nov 25 13:37:06 what's the bb command I can issue to get the cross gcc being compiled ? Nov 25 13:37:13 is there any ? Nov 25 13:37:16 ^ Nov 25 13:37:43 I have tried gcc-crosssdk-x86_64 but I am not sure that's the right one Nov 25 13:37:47 no, its not Nov 25 13:38:14 what's the right one ? Nov 25 13:38:23 "gcc-cross-[arch] binutils-cross-[arch] is most of the toolchain i guess" Nov 25 13:39:38 thanks, let me hava a go Nov 25 13:41:14 is the arch the host's one or the target's one ? Nov 25 13:42:02 nevermind, sorry Nov 25 13:52:03 sorry for keep hammering newbye's question, but where is the DL_DIR set ? is it in the local.cong ? Nov 25 13:52:09 local.conf file? Nov 25 13:52:25 for me it is. Nov 25 13:52:54 oks, sorry it was commented out and it points to the {TOPDIR}/download folder Nov 25 13:54:14 you sure you are looking at your local.conf and not your local.conf.sample? Nov 25 13:54:56 is there a way to list all the recipes containing in a layer ? and to show all the layers ? :-) Nov 25 13:59:23 pedr0_: you might find the bitbake-layers tool helpful Nov 25 13:59:46 i.e. bitbake-layers show-layers Nov 25 14:00:32 and bitbake-layers show-recipes Nov 25 14:01:26 cheers Nov 25 14:02:13 to get recipes specific to a layer, you'd need to filter the output a bit I think Nov 25 14:03:02 btw compiling the cross gcc is taking its bit of time, it has finished now after about 20 minutes on my machine Nov 25 14:03:23 now a lot of time, but some time. Nov 25 14:18:25 is there a way to show all recipes associated with a certain layer ? Nov 25 14:19:46 pedr0: you could find/grep inside a layer directory for *.bb Nov 25 14:21:29 pedr0_: It's not really a common usecase, you would usually keep sstate and let bitbake figure out whether something needs to be re-compiled for you. If the sources changed, bitbake will recompile automatically. Nov 25 14:22:13 right, I am keeping pushing this into the way I used to work but I should give up Nov 25 14:22:56 bitbake-layers show-recipes gives you a list of recipes and their layers Nov 25 14:22:56 I was a bit bothered buy the fact that after removing the sstate-cache and tmp directories the toolchain is recompiled, but nevermind Nov 25 14:23:27 yeah :bitbake-layers show-recipes | grep -B 1 'meta-whatever' Nov 25 14:24:17 basically = rm -rf tmp sstate-cache = build from scratch - download_packages Nov 25 14:24:34 pedr0_: it is possible to use an external toolchain if you don't want build it all the time Nov 25 14:25:25 yap, but I won't go for it as I do not want to overcomplicate the problem for now Nov 25 14:25:36 that said, bitbake is pretty good at detecting when rebuilds are needed---I don't find the need to clean build that often Nov 25 14:26:01 how do you manage to get a build out of a CI server ? Nov 25 14:26:10 do you build from scratch all the time ? Nov 25 14:26:25 yeah I do not think I will clean that often in my dev machine Nov 25 14:27:00 Usually, you keep the sstate cache but delete tmp/ Nov 25 14:27:15 Without sstate, CI builds are going to be slow. Nov 25 14:27:53 I used to run a clean build (i.e. no sstate or tmp but keep the DL_DIR) on a CI server but it could afford to take hours in that case Nov 25 14:30:04 from time to time i have to do a clean build if i like it or not... Nov 25 14:32:48 hello, I am using TFTP and NFS boot for testing purposes of my Yocto build Nov 25 14:32:50 what is the point in deleting the tmp folder ? Nov 25 14:33:01 however, when I try lsmod I get an error Nov 25 14:33:04 Error: could not get list of modules: No such file or directory Nov 25 14:33:18 even though I can check that the .ko files exist Nov 25 14:33:34 on the mounted NFS sysroot Nov 25 14:33:55 any ideas why that can happen? Nov 25 14:35:33 pedr0_: when you move files from one package to another and remove the old package from your build, you'll get a file conflict in the shared sysroot if you don't delete tmp Nov 25 14:35:52 "I was a bit bothered buy the fact that after removing the sstate-cache and tmp directories the toolchain is recompiled, but nevermind" <— you deleted everything, so yes the toolchain was rebuilt. Nov 25 14:43:36 yes sorry, bad writing - I was looking for a way where everything BUT the toolchain would get rebuilt and rm the two folders did not make ti Nov 25 14:43:41 make it Nov 25 15:24:40 I get "mount: only root can mount proc on /proc" during NFS boot. Am I missing a kernel config or something? Nov 25 15:25:35 eduardas_m: are you seeing this error when you do something like 'depmod' on the system? Nov 25 15:26:23 I got this during boot Nov 25 15:26:29 eduardas_m: my guess is you extracted the rootfs as a user, not root, so the ownership is wrong, which breaks setuid binaries Nov 25 15:26:42 yes, I did Nov 25 15:26:58 you might be right Nov 25 15:27:27 btw, kergoth, you are the tslib maintainer, no? Nov 25 15:27:58 was, yes, recently it ownership was handed off to someone who actually has touchscreen hardware to test with :) Nov 25 15:28:03 the github repo is the same, though Nov 25 15:28:32 I am actually trying to get my Qt app to work with resistive touchscreen Nov 25 15:28:51 ts_test works wonderfully, but on Qt app there is y axis inversion Nov 25 15:29:00 any ideas? Nov 25 15:29:36 I know its something with environment variable configuration, but can not get a feel of what it actually is Nov 25 15:32:05 Hi, is there anybody here, who can educate me about environment (or make?) variable SDKTARGETSYSROOT? For example, should it be set in `bitbake -c devshell...` ? Should it be set automatically, or do I need to set it somehow? Nov 25 15:33:19 that variable is set in the environment setup script for the sdk, which is used to do builds without bitbake. it has nothing to do with bitbake builds. Nov 25 15:33:24 Because I have a kernel module recipe, which uses it, but fails to compile, because '"__LINUX_ARM_ARCH__" is not defined', and at this point it seems to me, the reason is, that the variable is blank. Nov 25 15:34:09 that's not appropriate. a kernel module recipe should inherit module and use the kernel source tree path Nov 25 15:35:13 yeah. so what is this "sdk" thing, compared to building with bitbake? (keywords for google or relevant link would be great) Nov 25 15:38:28 (yocto/openembedded newbie here). Nov 25 15:44:20 eduardas_m: I've used something like https://fedoraproject.org/wiki/Input_device_configuration#Example:_Axis_inversion Nov 25 15:45:06 eduardas_m: to invert for X11 but I suspect there is a Qt-specific config Nov 25 15:55:29 hyde: an sdk is used for development outside of bitbake. i.e. to build with make and do active development rather than mucking about with recipes Nov 25 15:55:41 useful to hand off to app developers who don't want to deal with systems stuff Nov 25 15:58:39 ash_charles, yeah, that is unapplicable in my case since I use Qt with linuxfb Nov 25 16:04:50 eduardas_m: ah okay---'fraid I don't know a Qt conf off hand Nov 25 16:05:33 Hello everybody! Nov 25 16:05:37 I'm trying to generate an extensible SDK for my image, it complains it can't find the module oe.recipeutils. Nov 25 16:05:44 e.g., I get that error if I run bitbake core-image-minimal -c populate_sdk_ext Nov 25 16:05:48 Has anyone run into that kinda of problem before? Nov 25 16:06:51 That happens when it tries running the devtool command Nov 25 16:43:41 is there a fix for the getVar mess lurking on some -next branch somewhere? Nov 25 16:45:13 what mess specifically Nov 25 16:45:25 Exception: AttributeError: module 'bb.data' has no attribute 'getVar' Nov 25 16:45:38 getting reams of those in the last 2d or so. Nov 25 16:45:40 on master Nov 25 16:46:09 I think zeddii mentioned it was some flag day type change that was still making its way through... Nov 25 16:47:07 for what recipes? oe-core only had one user left which is fixed already, and there's a patch for the two recipes in meta-oe o the list Nov 25 16:49:38 let me check... Nov 25 16:50:24 meta-networking/recipes-support/nis/ypbind-mt_1.38.bb Nov 25 16:50:48 meta-networking/recipes-support/nis/yp-tools_3.3.bb Nov 25 16:51:00 they're fixed in the patch Nov 25 16:51:13 search oe-devel for patches from RP, it's the latest one Nov 25 16:51:28 ok, guess it has yet to be merged by joe or similar. Nov 25 16:55:24 * paulg looks at joeythesaint Nov 25 19:54:32 paulg: git-am https://patchwork.openembedded.org/patch/134230/ Nov 25 19:54:39 gm all Nov 25 19:58:10 khem, thanks. Nov 25 23:58:02 * vmeson has a reverse epiphany.... Nov 25 23:59:36 anyway see the email list for ross' epiphany and libwnck4 reverts to morty. **** ENDING LOGGING AT Sat Nov 26 03:00:00 2016