**** BEGIN LOGGING AT Tue Jul 10 03:00:02 2018 Jul 10 05:08:02 New news from stackoverflow: YOCTO how to create a basic ubuntu 16.04 linux Jul 10 06:08:11 New news from stackoverflow: can't compile my old c-projects (with gcc) Jul 10 06:38:16 New news from stackoverflow: can't compile my old projects (with gcc) Jul 10 07:04:16 good morning Jul 10 07:26:15 morning! Jul 10 08:19:07 Hi every one, I'm currently working on AGL project and we have a problem about cmake (using it with a script and yocto SDK). I find a nice solution. Jul 10 08:19:10 https://patchwork.openembedded.org/patch/139551/ Jul 10 08:19:37 But unfortunatly, it 's not include in upstream Jul 10 08:20:02 and I don't find any comment on this fix. Jul 10 08:20:14 do you know an alternative solution? Jul 10 09:11:31 Hi guys and girls Jul 10 09:12:57 I am having a question regarding native recipes. So i have tis recipe named let's say recipe-native which includes the line "inherit native" and it works but if I remove that line and add " BBCLASSEXTEND="native" " then it suddenly cannot find some header files. Has anyone any idea? Jul 10 10:18:09 hello Jul 10 10:18:46 anyone that can help me with a quiestion on linux kernel build using the yocto exported SDK? Jul 10 10:19:10 dunno, ask and find out Jul 10 10:21:56 thanks, basically I've created my image SDK and installed on my dev machine Jul 10 10:22:34 With this SDK I'am able tu fully build a kernel tree that I have cloned in a specific directory, from start to end Jul 10 10:23:36 The problem is that if I try to run "make menuconfig" it complains for the lack of ncurses-devel which in fact appear to be missing from the SDK Jul 10 10:23:55 areola: that's a matter of your host OS Jul 10 10:24:03 So the question is, how do I control what goes in the SDK? Jul 10 10:26:08 areola: from where are you trying to run "make menuconfig" exactly? Jul 10 10:27:35 mckoan: From within the kernel tree that I have cloned from https://github.com/Xilinx/linux-xlnx Jul 10 10:28:27 So basically, I cloned the repository, installed the yocto SDK, run the sdk script and started a build Jul 10 10:28:39 areola: and where you cloned it? Jul 10 10:29:28 I've cloned it in a standalone separate directory Jul 10 10:31:14 areola: where? in your desktop PC? whick OS? Jul 10 10:31:25 With previos versions of yocto (I was on daisy) this worked beatifully because I used this standalone kernel dire to make all changes to the kernel, including configuration, and the generated my patches to be inserted in the yocto recipe Jul 10 10:32:01 mckoan: Yes the development PC is an ubuntu machine Jul 10 10:32:10 areola: I guess your problem is not YP, but your PC Jul 10 10:32:19 areola: To resolve this issue you have to first complete all update of you system then try this "sudo apt-get install libncurses5-dev" Jul 10 10:34:09 On my PC libncurses5-dev is already there....but the fact that seems to be that the SDK script uses the x86_64 sysroot under the SDK install dir for native stuff (and libncurses-dev is not therer) Jul 10 10:35:16 I was looking to a way to ensure that in the YP SDK install dir, libncurses-dev is installed Jul 10 10:35:49 Today it does not seem so and I don't know ho to control that Jul 10 10:35:52 areola: how did you generate the SDK? and how did you set it up? Jul 10 10:36:24 to generate : bitbake -c populate_sdk core-image-minimal Jul 10 10:37:09 to setup: ./poky-glibc-x86_64-core-image-minimal-cortexa9hf-neon-toolchain-2.4.3.sh Jul 10 10:37:45 areola: that's for install the SDK Jul 10 10:38:24 areola: you should run something like: source /opt/poky/2.4-cortexa9hf/environment-setup-cortexa9hf-neon-poky-linux-gnueabi Jul 10 10:38:33 to: prepare environment: source /opt/poky/2.4.3/environment-setup-cortexa9hf-neon-poky-linux-gnueabi Jul 10 10:38:43 yep, there Jul 10 10:39:09 After that I cd to the kernel standalone dir and build Jul 10 10:39:10 areola: poastebin your commands and the error please Jul 10 10:39:58 * mckoan BBL Jul 10 10:39:58 marcof@fax-ubu-1604:~/projects/Zynq/Kernel/linux-xlnx-mideal-new$ make menuconfig *** Unable to find the ncurses libraries or the *** required header files. *** 'make menuconfig' requires the ncurses libraries. *** *** Install ncurses (ncurses-devel) and try again. *** scripts/kconfig/Makefile:201: recipe for target 'scripts/kconfig/dochecklxdialog' failed make[1]: *** [scripts/kconfig/dochecklxdialog] Error 1 Makefile:546: rec Jul 10 10:40:11 areola: pastebin Jul 10 10:40:24 how do I do that? Jul 10 10:40:39 areola: google->pastebin Jul 10 10:41:28 * mckoan (food) Jul 10 10:43:09 https://pastebin.com/qLvj2xaS Jul 10 10:48:49 mckoan: This is confirms that, apart from the problem with ncurses and menuconfig, I am able to run a full build https://pastebin.com/k1DSRaFE Jul 10 10:49:08 enjoy your meal :) Jul 10 11:05:09 hi there, I am currently looking for a proper and clean solution for my Yocto git repository and would like to ask how you handle this. Jul 10 11:06:34 I have a cloned morty repository with a lot of changes and would now upgrade to a newer version of yocto. Jul 10 11:07:37 how do you set up your git repository? Jul 10 11:09:04 New news from stackoverflow: Yocto sumo, meta-virtualization with vmdk Jul 10 11:19:35 Hey guys. Does anyone know the diff between "inherit" a class and "BBCLASSEXTEND" ? Jul 10 12:00:28 illarios: bbclassextend only works for native/nativesdk and effectively makes a new recipe, so you write foo.bb with bbclassextend and you get a foo-native and nativesdk-foo for free Jul 10 12:00:40 whereas if you want to write a recipe which is only native you can write foo-native.bb and inherit native in it Jul 10 12:09:16 New news from stackoverflow: Build Yocto image on a RPi B+ Jul 10 13:08:22 how do i get at the code in meta-swupdate? my thought was to do a devtool modify -x , where is something provided in the meta-swupdate layer, but there is nothing PROVIDEd in that layer. Jul 10 13:25:46 yates: what code in particular? Jul 10 13:26:26 the swupdate tool itself is 'swupdate' Jul 10 13:26:36 https://github.com/sbabic/meta-swupdate/blob/master/recipes-support/swupdate/ Jul 10 13:26:59 which if you read the recipe is https://github.com/sbabic/swupdate Jul 10 13:30:52 rburton: well, i just tried "devtool modify -x swupdate src/swupdate" and that worked! Jul 10 13:31:19 PROVIDES is only in a recipe if the default isnt enough, the default provide is PN Jul 10 13:37:10 rburton: thanks for that Jul 10 13:37:34 that resolves that. Jul 10 13:39:26 how do i determine what "components" the meta-swupdate layer provides? for example could be all of the follwoing: a) an initrd kernel, b) a static library foo.a for doing , c) ... Jul 10 13:44:52 areola: which settings you have used in local.conf to generate the SDK? Jul 10 13:45:01 is it "swupdate" and "swupdate-image"? Jul 10 13:45:08 areola: could you try: bitbake meta-toolchain ? Jul 10 13:52:34 mckoan: there is something in particular I should look into local.conf? Jul 10 13:53:45 yates: by looking at what files are in the layer? Jul 10 13:55:19 mckoan: took a quick look now and there's nothing apparently related to the SDK generation in there... Jul 10 13:56:43 mckoan: fact is that if I run menuconfig in the kernel devshell it does work, but in that case all the references point to a fully fledged yocto build env Jul 10 14:03:51 areola: are you able to run menuconfig for a vanilla (x86) kernel on your host? Jul 10 14:05:49 mckoan: uhm, it takes some time to try that as, I have no vanilla kernel source installed... Jul 10 14:07:33 work in progress... Jul 10 14:08:20 areola: and with the kernel you are using does it work at least soes it open the screen Jul 10 14:09:38 New news from stackoverflow: i.Mx6Quad add simple audio card to ALSA || how to install Kernel patch locally in yocto? Jul 10 14:31:15 mckoan: Yes, it works when doing a vanilla build on my dev pc (the very same with the issue occurring when building with the yocto exported SDK) Jul 10 14:32:04 areola: please try bitbake meta-toolchain and test using it Jul 10 14:33:12 areola: can you pastebin your environment-setup... script? Jul 10 14:39:49 mckoan: Here is the env script https://pastebin.com/KiZtSrW1 Jul 10 15:09:49 New news from stackoverflow: yocto dlib error: undefined reference to symbol 'cblas_dgemm' Jul 10 17:10:08 New news from stackoverflow: Removing busybox completely from a Yocto generated image Jul 10 18:43:46 swupdate has two recipes in its recipes-support/swupdate folder: swupdate_2017.11.bb and swupdate_2018.03.bb. bitbake is using 2018.03.bb - how is this determined? Jul 10 18:50:52 yates: A lot of things factor into that Jul 10 18:51:50 For two recipes in the same layer, it will choose the higher version, unless there is a PREFERRED_VERSION or one specifies a DEFAULT_PREFERENCE Jul 10 18:52:37 https://www.yoctoproject.org/docs/latest/mega-manual/mega-manual.html#var-PREFERRED_VERSION https://www.yoctoproject.org/docs/latest/mega-manual/mega-manual.html#var-DEFAULT_PREFERENCE Jul 10 18:52:55 ok, thanks man. Jul 10 18:54:18 JPEW: version == 2017.11 ? Jul 10 18:54:22 (e.g.) Jul 10 18:54:43 Yes Jul 10 18:55:12 IIRC bitbake uses the debian version parsing rules.... I might be wrong about that though Jul 10 19:00:59 does PREFERRED_VERSION belong in a .conf file or a .bb file? Jul 10 19:01:16 yates: putting it in a .bb won't do anything Jul 10 19:01:20 .conf file (machine, distro, or local) Jul 10 19:01:39 if you want to control default behavior when no PREFERRED_VERSION is set, that's what DEFAULT_PREFERENCE is for Jul 10 19:02:03 PREFERRED_VERSION controls recipe version selection globally. one recipe can't control what version of some other recipe is going to be used Jul 10 19:02:08 i see Jul 10 19:04:26 DEFAULT_PREFERENCE would go in the layer.conf for the layer, then? Jul 10 19:04:52 no Jul 10 19:05:35 DEFAULT_PREFERENCE goes in a recipe to control whether that recipe is preferred by default in the absence of user preferences. mainly used for recipes we want available but not built by default, i.e. development versions Jul 10 19:07:11 what is the meaning of the value DEFAULT_PREFERENCE is assigned? Jul 10 19:07:31 an integer specifying the preference priority? Jul 10 19:07:39 with negative maning "don't do this" Jul 10 19:07:44 meaning Jul 10 19:07:58 essentially yes. common values are -1 or 1 Jul 10 19:09:07 more often than not you don't need DEFAULT_PREFERENCE at all. also worth noting that layer priority has a bigger effect on default recipe selection than DEFAULT_PREFERENCE does Jul 10 19:09:36 an old version of a recipe in a layer with a higher BBFILE_PRIORITY will be used in preference to a newer version of a recipe in a layer with a lower BBFILE_PRIORITY in the absence of PREFERRED_VERSION Jul 10 19:10:04 say that five times fast.. Jul 10 19:10:06 cause it's not complicated enough ;) Jul 10 19:10:12 yah. Jul 10 19:10:57 almost all of the complexity and learning curve comes back to the oe/yocto priority of high flexibility. it covers a lot of use cases, a lot of machines, distros, layers, and works for tons of companies with differing priorities, which makes it difficult to keep things simple Jul 10 19:11:09 just the nature of the beast, really Jul 10 19:11:58 where another tool might have one and only one way to accomplish any particular thing, in this one there are usually multiple options depending on your requirements and priorities Jul 10 19:12:32 there are a lot of advantages to that, but some disadvantages as well, as is usual with such things Jul 10 19:16:01 would conf/local.conf of my build environment be the right place to put a PREFERRED_VERSION? Jul 10 19:24:34 as with most configuration, that depends on whether the chnage is temporary or not. local.conf isn't a great place to maintain settings over the long term, but it's great for temporary changes or those local to this particular build Jul 10 19:24:40 i.e. yes, for now Jul 10 19:28:19 yes, local to a particular build is why i want this Jul 10 19:29:00 kergoth: and it's "PREFERRED_VERSION", not "PREFFERRED_VERSION"! Jul 10 19:30:00 that's true, yes, it works better without the typo :) Jul 10 19:37:46 i know documentation is tedius, but it "Would Be Nice[TM]" if some of this was put into the variables portion of the manual(s). e.g., where the variable should go, wwhat other variables affect it, etc. Jul 10 19:38:01 what Jul 10 19:38:26 wait, the suggester gets the job?!? :) Jul 10 19:39:42 i must admit, though PREFERRED_VERSION is pretty complete, except for where it should go. Jul 10 19:45:14 not everyone is good as putting themselves into a beginner mindset, unfortunately. the yocto project does have technical writers on staff, however. i'd suggest opening a bug in bugzilla to propose the documentation improvement Jul 10 19:59:01 stephano ... babic? Jul 10 20:01:01 where does yocto/devtool maintain the git repo used for the source when doing a devtool modify -x ? this is somewhat of a mystery to me. it isn't the original github repo, is it? Jul 10 20:06:07 is it just a temporary clone? Jul 10 20:07:43 stephano: why do you configure the kernel as part of your meta-swupdate? my vendor-supplied kernel does not have the config Jul 10 20:07:56 (variscite imx6ul) Jul 10 20:08:13 yates: the sources end up whereever you told devtool to put them. default is under workspace/sources/ Jul 10 20:08:21 the exact contents of the git repo depend on SRC_URI in the recipe Jul 10 20:10:05 kergoth: ok, so the source under workspace/sources/swupdate (e.g.) is just a temporary git clone? the deltas are used to generate the .bbappend files and then the repo is blown away? Jul 10 20:10:50 yes. Jul 10 20:12:04 the repo exists whle you're using devtool to develop. devtool finish/reset/etc will result in the build no longer using it Jul 10 20:12:48 even then it doesn't remove it, just stops building from it, to avoid losing your changes Jul 10 20:15:23 kergoth: yes, it is usually I who blow away the source/swupdate copy. but the next time you do a devtool modify -x, it rebuilds it from the bbappens, i guess Jul 10 20:15:58 that is, it will clone the repo, then apply the bbappends, so you are back at your current working version Jul 10 20:18:00 in other words, the repo is never updated with your changes. rather, your changes are the composite of the original (probably hash-tagged) repo plus your bbappends Jul 10 20:18:30 you may say "of course", but i am still beginner thinking mode... Jul 10 20:19:22 even i say "of course" to this. the repo would become corrupt if everyone and his brother were pushing changes up. Jul 10 20:20:44 devtool sync will sync up / update a prevoiusly extracted source tree for an existing recipe. no idea how it does it, or the impact on the existing repo Jul 10 21:41:09 Why can't Yocto build directory TMPDIR be in NFS? Jul 10 21:41:20 That seems like an arbitrary limitation Jul 10 21:43:58 it's not arbitrary at all, it has to due with limitations of the filesystem Jul 10 21:44:05 among other things, b itbake needs to be able to use file locking reliably Jul 10 21:44:53 bitbake can't build on non-case-sensitive filesystems either, also the opposite of arbitrary Jul 10 21:45:12 though that one has more to do with what we're buiding than bitbake itself Jul 10 21:46:24 Looks like I have to open a ticket with IT, then... Jul 10 21:46:28 Thanks kergoth Jul 10 22:00:59 note that the metadata can live in nfs just fine, as can DL_DIR and SSTATE_DIR, just not TMPDIR Jul 10 22:01:00 np Jul 10 22:01:09 afaik anyway Jul 10 22:08:25 Yeah, I kind of figured after reading a bunch of mailing threads. I wonder if NFS v4 file locking is sufficient for Yocto? Would probably be a pretty big feature to bitbake though. Jul 10 22:09:20 file i/o performance could be a concern too. a number of folks put TMPDIR in a tmpfs/ramfs, after all Jul 10 22:09:42 course that's relatively minor in comparison to the locking issue, and obviously isn't fatal, just worth noting Jul 10 22:11:04 New news from stackoverflow: Yocto system stuck, Waiting for root device PARTUUID Jul 10 22:21:45 ythl: we have huge problems with performance of nfs and tools like git fwiw Jul 10 22:22:07 khem: huh, https://github.com/arsv/minibase seems interesting Jul 10 22:32:26 anybody with experience on the Rockchip 64 ? Jul 10 22:33:12 i have issues flashing, i got yocto to build a base image, but i cant find the right mechanism to flash my images **** ENDING LOGGING AT Wed Jul 11 03:00:02 2018