**** BEGIN LOGGING AT Tue Aug 28 03:00:03 2018 Aug 28 05:28:11 khem: sorry my computer died again and couldnt see if you replied to the compiler-rt-native stuff Aug 28 05:28:23 khem: is this a known issue? Aug 28 05:30:31 Hello guys n girls!! Aug 28 05:30:58 Does anyone know how to get an initramfs.img out of the yocto core-image-minimal ?? Aug 28 05:31:00 aehs29: I did not reply yes, Aug 28 05:31:39 compiler-rt is disabled due to a bug that was reported by some end user Aug 28 05:34:12 if we enable compiler-rt for native then it will also mean that cross-clang will link with compiler-rt too Aug 28 05:34:22 eventually I want to be there Aug 28 05:34:33 test it out and let me know if it all pans ok Aug 28 05:45:26 Hi, With some helpful advice from this IRC channel (thanks @kergoth ), I used local variables and OVERRIDES to create a single recipe which could build from multiple sources of the same software package (a specific tagged release, the HEAD of our develop branch, or local source). The override worked perfectly except that bitbake also rebuilt all the packages in the image, including many of the native Aug 28 05:45:32 packages. So not so good as a time saver... Aug 28 05:45:35 Next I created 3 recipes: foo-release, and foo-git-head, and foo-local. I could build each one individually from a clean tmp/, but otherwise bitbake complains that the target files are already in tmp/sysroots and refuses to build. (Even if I delete tmp/sysroots and the manifest file, knowledge of what used to be in tmp/sysroots persists.) Aug 28 05:45:40 Next I created 3 versions of the same recipe: foo_0.3.1.bb , foo_0.3.1+git.bb and foo_0.3.1+local. I can build each fine if I do a bitbake -c cleansstate foo between them, but if I build more than one I get the following error: Aug 28 05:45:44 ERROR: Multiple versions of foo are due to be built (/home/ubuntu/fsl-release-bsp/sources/meta-foo/recipes-foo/foo/foo_0.3.1.bb /home/ubuntu/fsl-release-bsp/sources/meta-foo/recipes-foo/foo/foo_0.3.1+local.bb). Only one version of a given PN should be built in any given build. You likely need to set PREFERRED_VERSION_foo to select the correct version or don't depend on multiple versions. Aug 28 05:45:50 Did I miss something implementing one of these three approaches? Is there another way to accomplish this (ex: maybe find a practical use case for BBVERSIONS)?. Aug 28 05:54:16 you can use only one recipe at a time for this since practically you are building same s/w Aug 28 05:54:43 but it will be good to understand your usecase.. why do you need these different versions Aug 28 06:04:39 khem, the use case is that I want to switch frequently from one source of the software to another. For example I want to try a feature on my local branch. If I find a bug, I want to quickly rebuild the package with the release or HEAD version and see if it has the same behavior. Aug 28 06:05:06 Previously we had different versions of SRC_URI and a few other variables commented out and would edit this to switch sources. Predictably this is error prone and requires an extra step before checking in legitimate changes to the recipe. Aug 28 06:06:58 We only ever use one of these versions at a time, but being able to build one and then the other is something we do all the time--sometimes to incorporate in an image, but sometimes I just update the target rpm on my board using smart install. Aug 28 06:09:14 khem: yeah I tried enabling it but compiler-rt-native doesnt build, I think cmake was complaining about not finding llvmConfig.toolchain or something Aug 28 06:09:29 khem: I can give it another shot once I get my computer back Aug 28 06:09:52 it should not build it it, infact that a bug Aug 28 06:10:17 rohanmahy: look at devtool I think you can get this done effectively with it Aug 28 06:10:59 which version are you on ? Aug 28 06:13:17 khem: we are on morty. it looks like devtool just creates the recipes. I have all the variants of the recipes. I just need to be able to switch quickly among them. Did I miss something in devtool that does that? Aug 28 06:17:13 rohanmahy: you could be in better shape with yocto versions newer than morty becase it uses recipe specific syroot Aug 28 06:18:30 khem: that is useful. I have been trying to convince some managers to switch to rocko where we also have board support. Aug 28 06:18:47 so I would suggest that you make SRC_URI overridsabele, eg SRC_URI += "${NOW_URI}" Aug 28 06:18:57 and then override NOW_URI from local.conf Aug 28 06:19:17 NOW_URI_pn- = "git://,,,," Aug 28 06:19:29 overriding from local.conf triggers almost a full build. I had that working on friday. ;-) Aug 28 06:21:32 Is anyone planning to add package-specific overrides anytime soon? (Just like OVERRIDES, but scope is limited to the current package). Aug 28 06:26:13 khem: thanks for the info. I will try out the multiple recipes approach in rocko and see if it works better there. Aug 28 06:29:19 rohanmahy: the rebuild will depend on how deep the dependencies on this recipe are Aug 28 06:29:49 rohanmahy: if you were to write three recipes for say glibc then good luck with rebuilds they will happen Aug 28 06:30:07 but if the recipe is more or less a leaf package in dep tree then it is better Aug 28 06:31:13 khem: the recipe is a leaf which depends on 3 other leaves, but still it rebuilt the compiler and all sorts of other things that did not change. Aug 28 06:31:38 all the dependencies where our own code in our layer. Aug 28 06:35:07 Hello. Aug 28 06:38:39 This may not be the right place to ask but I'm trying to find a Automotive hardware platform for Yocto. I've been doing some heavy research into the renasas H2/H3 and I think it's does everything I want but it's way to much just to get started on the development side. Any suggestions on alternative hardware platforms that 4 camera support parking assist and infotainment cockpits with dual screens? Aug 28 06:46:40 Hi All, Aug 28 06:46:40 Which ubuntu version is best for yocto development ? Aug 28 06:56:56 webreformer_: it depends which release of yocto project you are building Aug 28 06:57:24 generally look at supported distros page Aug 28 06:59:03 you can check that via looking at SANITY_TESTED_DISTROS in meta-poky/conf/distro/poky.conf Aug 28 06:59:42 assuming latest release then always check https://yoctoproject.org/docs/current/ref-manual/ref-manual.html#detailed-supported-distros Aug 28 07:01:14 rohanmahy: if it is rebuilding compiler etc. you have a different problem. I would not expect it to rebuild Aug 28 07:05:07 Phreaknes: #automotive channel might have more suggestions but I think besides renesas H2/H3 see https://wiki.automotivelinux.org/agl-distro/source-code Aug 28 08:06:38 good morning **** BEGIN LOGGING AT Tue Aug 28 09:19:08 2018 Aug 28 09:46:07 New news from stackoverflow: How to access the build target name in bitbake Aug 28 10:11:09 Hi, anyone facing oom-killer problems with MariaDB using Rocko? Aug 28 10:12:21 my system is unusable after 1 hour runtime because oom-killer is killing MariaDB Aug 28 10:13:02 versions prior to rocko (5.5.57-MariaDB) were ok Aug 28 10:23:34 mckoan: i think that you need to verify the maria db settings related to mem usage Aug 28 10:24:02 MarcWe: yes, I suspected that and I'm verifying, thx Aug 28 11:45:58 Writing a recipe for webrtc, I need to write my own do_fetch. One of the steps in fetching, using Google's tools, uses the xz binary to unzip something after downloading. Even though I depend on xz-native, the xz binary is only available some time after do_fetch. Aug 28 11:46:39 Should I do the part which requires xz in do_unpack even though it downloads stuff, or is there a way to make xz available to do_fetch? Aug 28 12:06:47 mort: DEPENDS is actually short-hand for adding a dependency on xz-native for do_configure Aug 28 12:07:08 mort: you'll need to do do_fetch[depends] += xz-native:do_populate_sysroot Aug 28 12:07:42 or if you write a proper fetcher for bitbake, see base.bbclass for the logic that looks at the fetch url and extends the dependencies automatically Aug 28 12:07:45 rburton: I see, thanks - though why does it work in do_unpack then? Is that just a race condition? Aug 28 12:08:23 yeah probably Aug 28 12:09:50 rburton: another thing, how does bitbake know if it should re-run the fetch step or not? When I just have a custom do_fetch with no SRC_URI, it seems to want to run fetch every time Aug 28 12:11:28 oh btw pretty sure there's a depot fetcher out there somewhere Aug 28 12:15:50 it complains about "unparsed line: 'do_fetch[depends] += xz-native:do_populate_sysroot'" Aug 28 12:16:29 you'll need quotes around teh xz-native:... bit Aug 28 12:16:51 was just about to comment that it works with quotes, but I wasn't sure if that means exactly the same Aug 28 12:20:22 rburton: is there an easy way to tell the system that it's downloaded though? I might check out the depot fetcher, but if it's just a tiny change I'd rather just use my current do_fetch until I take the time to find the fetcher Aug 28 12:30:12 mort: use the same variables as the standard fetcher and the variable depends should do the right thing Aug 28 12:35:54 I can't find any variable, but https://github.com/openembedded/bitbake/blob/master/lib/bb/fetch2/git.py#L349 - it looks like the git fetcher just exits early if it detects that the repo is already downloaded. Is the mechanism for not re-fetching simply that the fetcher just exits if the local files already exist? Aug 28 12:39:42 Does anyone have any experience with initramfs out of Yoctoand EFI boot ? Aug 28 12:42:47 if there are any glib-savvy developers with a few hours to kill, fixing https://bugzilla.yoctoproject.org/show_bug.cgi?id=12904 would be much appreciated Aug 28 12:42:48 Bug 12904: normal, Undecided, ---, ross.burton, NEW , Port panel from apm/acpi to sysfs Aug 28 12:46:15 hello, I am trying to use the Yocto-generated SDK on a CMake project, however Boost is not found: https://pastebin.com/m45nrJ0n Aug 28 12:46:37 the Boost headers exist in the target sysroot, though Aug 28 12:46:45 rburton: cc for now, I'm trying to catch up with all the other stuff Aug 28 12:46:56 such as, oh, recipe updates :D Aug 28 12:47:24 the same CMake project builds fine via a Yocto recipe though Aug 28 13:46:53 New news from stackoverflow: how to add a third party library as a package in Yocto build Aug 28 14:47:06 New news from stackoverflow: Yocto - development build vs production build Aug 28 15:17:50 https://www.linuxfoundation.org/press-release/2018/08/yocto-project-welcomes-new-members-advances-open-source-embedded-systems-through-momentum/ Aug 28 15:18:40 awesome Aug 28 15:20:36 uh, cool Aug 28 16:10:26 https://autobuilder.yoctoproject.org/new/#/builders/29/builds/48 Aug 28 16:19:33 The server failed the authenticity check (autobuilder.yoctoproject.org). Aug 28 16:47:30 New news from stackoverflow: Yocto Build fails to at `go get` command within Bitbake Recipe for InfluxDB Aug 28 17:47:41 New news from stackoverflow: How to deploy a Find*.cmake file for an Autotools library in the correct place for Yocto? Aug 28 17:48:35 RP: thats great Aug 28 18:17:46 New news from stackoverflow: if conditional statement in bbappend file Aug 28 18:29:19 RP: https://gist.github.com/kraj/17fe3a56d95fff9a72772fe869a67965 Aug 28 18:29:35 seeing these errors on risc-v alone, ideas ? Aug 28 21:02:21 khem: no qemu support making on target postinstalls inevitable Aug 28 21:02:24 kanavin_: ^^^ Aug 28 22:10:13 kanavin_: https://autobuilder.yocto.io/builders/nightly-x86-64/builds/1282 :( Aug 28 22:23:52 hmm... I'm getting this at the moment: https://pastebin.com/Yz1hSarV Aug 28 23:34:13 bluelightning: it seems acl needs to compile with -fPIC -DPIC options Aug 28 23:34:33 khem: has nobody else seen this though? Aug 28 23:35:07 in any case, what changed? Aug 28 23:42:01 bluelightning: we are using security flags on poky by default now Aug 28 23:42:29 so its possible it will expose these kind of errors but I have not see this in my world builds Aug 28 23:43:57 I build for qemux86-64 and your build is using genericx86-64 Aug 28 23:44:13 dont think that should matter Aug 29 01:19:04 New news from stackoverflow: Yocto ubuntu bitbake error Aug 29 01:40:01 bluelightning: did it help ? Aug 29 01:40:36 khem: did what help? Aug 29 01:46:37 your acl error .. did yor forget ? Aug 29 01:55:07 khem: I didn't try anything further, moved onto something else Aug 29 01:56:01 ok Aug 29 01:56:20 I was interested to narrow it down since I am not seeing it here **** ENDING LOGGING AT Wed Aug 29 03:00:03 2018