**** BEGIN LOGGING AT Thu Jun 28 03:00:02 2018 Jun 28 07:40:19 Hi, is there a version of yocto for imx platform with a browser+hw accelerated video playback. thanks Jun 28 07:41:33 suvirb: every YP version support all IMX features, I would suggest 'rocko' Jun 28 07:43:43 will rocko is a huge upgrade? i tried image from boundry device with yocto pyro and honestly, video was terrible in performance on chromium and less terrible but unacceptable quality on firefox. kernel 4.1.15 Jun 28 08:45:10 Hi everyone, i work on ubuntu 16.04, my collegue works with debian scratch. She's having an issue with meta-java openjre-8, autoconf fails to execute javac Test.java and it works ok for me. Any idea? Jun 28 08:48:04 maxin? :) Jun 28 08:48:06 New news from stackoverflow: Yocto sumo, meta-virtualization with vmdk Jun 28 08:48:49 nayfe: hi.. Jun 28 08:50:19 nayfe: you mean executing javac on target ? Jun 28 08:52:34 nayfe: have you checked the differences in java related configurations in local.conf Jun 28 08:56:19 maxin: I mean when it fails to buildsjavac-native Jun 28 08:56:58 maxin: i'm gonna check libstdc++6-$compilerversion-dev is installed brb Jun 28 08:57:05 nayfe: ok.. Jun 28 08:58:23 it is .. Jun 28 09:11:05 maxim: config.log : https://pastebin.com/8gDswSvk // log.do_configure https://pastebin.com/UcaDGvxK Jun 28 09:11:22 oops maxin Jun 28 09:16:43 nayfe: checking.. Jun 28 09:20:31 nayfe: fails with Illegal instruction .. Jun 28 09:25:45 nayfe: (checking) other than the OS differences, do anyone of you installed JDK in the build machine? Jun 28 09:33:40 maxin: when both have openjdk-8-jdk **** BEGIN LOGGING AT Thu Jun 28 11:35:20 2018 Jun 28 12:42:24 maxin: i reproduce the problem with crops/poky debian-9-base Jun 28 12:42:47 sumo branch Jun 28 12:50:17 nayfe: ok Jun 28 12:50:53 i just added libx11-dev to Dockerfile as it is requested for meta-java Jun 28 12:51:33 by default there is no java installed (no java/javac in path)) Jun 28 12:52:13 Hi folks! Jun 28 12:52:18 I wonder what are x86_64_nativesdk used for ? is it the rpm's used to build the SDK (cross tool chain) ? Jun 28 12:53:03 I am talking about build/tmp-glibc/deploy/rpm/x86_64_nativesdk Jun 28 12:53:12 you can generate an SDK for another arch than your host Jun 28 12:53:19 nayfe: java in host is not a requirement for building with meta-java Jun 28 12:53:48 maxin: yep but i try to found where it fails :) Jun 28 12:54:27 nayfe: so, is it SDK to build cross compiler? Jun 28 12:54:34 nayfe: me too . :) Jun 28 12:54:45 \o/ Jun 28 12:56:16 astrunin[m]: https://www.yoctoproject.org/docs/2.5/mega-manual/mega-manual.html#ref-classes-nativesdk Jun 28 12:56:19 Or is it crosscompiler and Yocto uses build tools from host? Jun 28 13:00:06 nayfe: So, this RPM's related with populate_sdk command. It's all about creating your own SDK. Am I right? Jun 28 13:00:33 indead Jun 28 13:00:41 indeed Jun 28 13:02:07 maxin: maybe just a problem of free mem Jun 28 13:04:06 nop :/ Jun 28 13:04:42 /tmp/work/x86_64-linux/inetlib-native/1.1.1-r0/recipe-sysroot-native/usr/bin/javac: line 21: 692 Illegal instruction (core dumped) ${RUNTIME} -Xmx1024m -cp ${ECJ_JAR} org.eclipse.jdt.internal.compiler.batch.Main ${1+"$@"} Jun 28 13:32:29 maxin: Is it normal javavm-native provides java 1.5 ? is it enough to execute ECJ compiler ? Jun 28 13:35:32 nayfe: depends on PREFERRED_PROVIDER_virtual/java-initial-native settings Jun 28 13:35:39 i set cacao-initial-native jamvm-native ecj-bootstrap-native as preferred providers Jun 28 13:35:53 as described in README Jun 28 13:35:56 nayfe: ok Jun 28 13:37:22 nayfe: that combination works in Fedora Jun 28 13:37:58 gcc on stretch is 6.3 and 5.4 on my ubuntu Jun 28 14:07:19 I don't understand the examples shown here: https://www.yoctoproject.org/docs/latest/bitbake-user-manual/bitbake-user-manual.html#variable-interaction-worked-examples Jun 28 14:07:42 There are two examples using A_foo_append and they don't have the same behaviours Jun 28 14:08:14 should OVERRIDES = "foo" be removed from the first example? Jun 28 14:12:45 When you put OVERRIDES = "foo", it says A is replaced by A_foo but overriding is done after normal assignment Jun 28 14:12:49 dkc Jun 28 14:13:53 okay Jun 28 14:15:13 in a recipe you can have MYVAR = "value0" MYVAR_poky = "value1" so if you are using poky, MYVAR equals value1, value0 otherwise. But if you add OVERRIDES = "poky" it will automatically select _poky value1 Jun 28 14:15:18 sort of Jun 28 14:24:34 got it, I think I just misread the first example. But with your explanation, I understand the behaviour of overrides Jun 28 14:25:51 regarding my issue from back then: To build a uncomressed uImage, set KERNEL_IMAGETYPE to uImage.bin. This however will fail, as oe will no longer set u-boot-mkimage as dependency then (IMHO a bug, will likely submit a patch sooner or later) and fail the compilation. I am now working around by setting KERNEL_ALT_IMAGETYPE to uImage and simply not using this image in the end. Jun 28 14:26:01 Just in case anyone has the same problem too Jun 28 14:38:28 Hey guys, one really small stupid question i think. If you build an image via bitbake core-image-minimal f.e., there are multiple arch dependend folders in the tmp/deploy/ipk folder, how is this defined? Or in another words, there was someday a rpi jethro based image, which got updated to morty later. The supported architecture during the jethro time was cortexa7hf-vfp-vfpv4-neon, this got changed to cortexa7hf-neon-vfpv4 since Jun 28 14:38:28 krogoth. This results in an issue with opkg which does not recognize that new packages are availabe. One solution would be to add cortexa7hf-neon-vfpv4 into the arch.conf but this is not possible because of missing access to the devices. Is there another solution that opkg on images build for cortexa7hf-vfp-vfpv4-neon also recognizes images build for cortexa7hf-neon-vfpv4, e.g. how can say that the new packages are also valid Jun 28 14:38:29 for cortexa7hf-vfp-vfpv4-neon? Jun 28 14:56:06 maxin: with jdb, i see it crashes with "Set uncaught java.lang.Throwable" in inner class Logger of https://git.eclipse.org/c/jdt/eclipse.jdt.core.git/tree/org.eclipse.jdt.core/batch/org/eclipse/jdt/internal/compiler/batch/Main.java. We use a 2011 version for libecj-bootstrap-native, maybe it has been fixed on more recent versions Jun 28 15:05:19 rburton_, is there any issues with the bind and dhcp updates? Jun 28 15:05:40 armpit: yeah mailed you yesterday Jun 28 15:06:04 oh, let me look Jun 28 15:06:22 took the day off and didn't open the computer Jun 28 15:08:41 RP: I'll take https://bugzilla.yoctoproject.org/show_bug.cgi?id=12808 . Want to look it over before I volunteered :) Jun 28 15:08:42 Bug 12808: normal, Medium+, 2.6 M2, richard.purdie, NEW , bitbake-layers create-layer Jun 28 15:21:18 JPEW: cool, thanks! Jun 28 15:22:01 is es possible to create unpartitioned images using wic? Jun 28 15:23:15 armpit: I just tried replicating the eclipse issue and couldn't. See https://autobuilder.yocto.io/builders/nightly/builds/1195 Jun 28 15:24:10 armpit: what is odd is that in your failing builds, e.g. https://autobuilder.yocto.io/builders/nightly/builds/1176 commit_resolvedhead_eclipse-poky-neon is "" Jun 28 15:24:42 yet the command there, https://autobuilder.yocto.io/builders/nightly/builds/1176/steps/ResolveLayerHead_4/logs/stdio works and gives the right revision Jun 28 15:27:47 i'm trying to get opkg tools installed on my ubuntu, the guides i have founf are all stale relative to the opkg source hosted on yocto project org, is there a guide? Jun 28 15:28:39 does anybody here have experience with mLinux? Multitech's build using Yocto? Jun 28 15:28:45 zarzar, you may also use dpkg to query ipks if this helps Jun 28 15:29:08 fsdun: i need to build packages Jun 28 15:29:19 oh. ok. Jun 28 15:29:34 why not use oe for that ... ;) Jun 28 15:30:03 RP, the url was changed yesterday or this morning by MH Jun 28 15:30:30 fsdun: working with a yocto built system that has opkg Jun 28 15:30:35 armpit: hmm, changed in what way? Jun 28 15:30:39 so its possible its fixed Jun 28 15:31:08 I've updated git://git.yoctoproject.org/eclipse-poky to be current with git://git.yoctoproject.org/eclipse-yocto. Does this fix the problem? Jun 28 15:31:11 fsdun: what is oe, is it supported on ubuntu also? Jun 28 15:31:17 armpit: oh :/ Jun 28 15:31:27 zarzar, SDKs provided scripts to create ipks in the past. don't know if this is still the case Jun 28 15:31:28 armpit: those should be one and the same repo Jun 28 15:31:41 its noted in the bug. I put all the email exchanges Jun 28 15:32:09 halstead: can we make eclipse-poky a symlink to eclipse-yocto? Jun 28 15:37:06 armpit: is that bug still open? I think we need to ask michael to symlink them Jun 28 15:37:29 RP, it's a mirror with different hooks so clones/pulls/etc work but pushes need to update to eclipse-yocto. Jun 28 15:37:58 halstead: why can't we just symlink it? Jun 28 15:38:20 RP, yes its still opened Jun 28 15:39:40 RP, I suppose we can. The idea was to phase out the old name. Jun 28 15:40:11 halstead: We can just list and use the new one. I really don;t like the idea these can get out of sync though Jun 28 15:40:16 apparently swupdate has a dependency on libconfig, and my oe layer is using libconfig v1.5, and v1.5 moved from the hyperrealm website to git. Jun 28 15:41:00 RP, Okay. I'll change the setup. Jun 28 15:41:30 halstead: thanks. I just don't want to end up chasing this down again in the future when some new change gets pushed Jun 28 15:42:05 i changed the SRC_URI accordingly, set the right SRCREV, but i'm getting problems like "LIC_FILES_CHKSUM points to an invalid file" and, after fixing that, now just a failure in which make is run with no arguments (missing Makefile file?). are there other tweaks to the layer to go from a web url to git? Jun 28 15:43:28 here's the libconfig_1.5.bb file i'm currently using: https://paste.fedoraproject.org/paste/fPuK9K6-cz3hs40ToN6FkA Jun 28 15:43:38 yates: sounds like you didn't set S properly Jun 28 15:43:56 "S"? Jun 28 15:44:13 what is "S"? Jun 28 15:44:19 yes, the S variable, defined in half the recipes in oe-core, is the path to the extracted source tree Jun 28 15:44:32 when using a git uri, it'll be unpacked to ${WORKDIR}/git, so S has to be set to that Jun 28 15:44:56 where is that S variable defined? Jun 28 15:45:03 i don't understand the question Jun 28 15:45:13 * RP is missing his old build resources, this is making me so unproductive :( Jun 28 15:45:25 by definition the path to the source tree is recipe specific Jun 28 15:46:06 oh, so you mean i now need to set S in libconfig_1.5.bb? Jun 28 15:46:43 i have a look, thanks kergoth Jun 28 15:49:47 if it's using a git uri now, yes Jun 28 15:52:32 kergoth: i notied that mozjs_17.0.0.bb in the oe layer is setting S to S = "${WORKDIR}/${BPN}${PV}/js/src" Jun 28 15:52:55 oh wait, that's not a git uri. Jun 28 15:53:31 so just ${WORKDIR}/git Jun 28 15:53:48 yates: that setting is only valid for mozjs, libconfig will need a different setting Jun 28 15:53:54 and yes, ${WORKDIR}/git will likely work Jun 28 15:56:44 RP, Change complete. I'm trying to remember why Tim and I didn't do this from the beginning. Jun 28 15:57:21 halstead: thanks! Jun 28 15:57:25 I think he wanted it to work for everything already out there but intended to find and correct the name everywhere going forward. Jun 28 15:57:48 I'd like to write a test case for 'bitbake-layers add-layer', would the best place be oe-selftest? If so, how do I figure out what the test ID should be? Jun 28 15:58:11 So we made a mirror. But this is far more simple. Jun 28 15:58:47 kergoth: yep, that fixed the "new git download" problem(s), now i'm on to the next issue. thank you. Jun 28 15:59:02 JPEW: just don't have an ID and yes, it could be a good place Jun 28 16:03:38 JPEW, good question.. we should define the test case ID groupings Jun 28 16:12:49 halstead: are there issues or large backlog in mailman currently on yp.org? Jun 28 16:19:56 denix, Possibly. We need to move to groups.io soon. Mailman and/or our self hosted mailserver often has trouble these days. Jun 28 17:33:56 somehow a recipe I created rebuilds every time, what do i have to include to get it cached ? Jun 28 17:54:49 desert: save the sstate signature between 2 builds and use bitbake-diffsigs to see what's changing between them Jun 28 18:56:31 greetings Jun 28 18:56:39 Having a little trouble with a build Jun 28 18:57:04 building an altera userland, rdma-core Jun 28 18:57:26 and getting an error: Could NOT find PythonInterp (missing: PYTHON_EXECUTABLE) Jun 28 18:57:32 any thoughts? Jun 28 18:59:08 this happens during a do_config step Jun 28 19:22:23 New news from stackoverflow: Best practice for building firmware using Yocto Jun 28 20:09:14 can devtool be used to test mtd-utils? Jun 28 20:10:21 after "devtool modify -x mtd-utils src/mtd-utils", should i then run "devtool build mtd-utils" Jun 28 20:10:22 ? Jun 28 20:15:44 i am getting an error, pointing to this log file: https://paste.fedoraproject.org/paste/evD7SZuXlFnW-h8MxHW5aA Jun 28 20:16:28 why would this happen? Jun 28 20:16:52 i.e., why would there be "No rule to make target 'install" Jun 28 20:21:35 bad make file? Jun 28 20:32:50 ya' think? :) Jun 28 20:33:19 i guess the sub-surface question is, if this layer ever worked, why is it not working now? Jun 28 20:34:31 Ah, I don't think I've been following the conversation long enough to know what layer you are talking about. Jun 28 20:35:04 mtd Jun 28 20:35:21 https://paste.fedoraproject.org/paste/Eq8h5gKjIT5OFfGFkvalSw Jun 28 20:36:21 i'm not sure - is it even valid to try to devtool-build that? Jun 28 20:38:10 I don't know. I'm not very familiar with devtool Jun 28 20:38:25 I assume you're using devtool to make custom changes? Jun 28 20:38:50 or apply a patch or some such? Jun 28 20:38:51 when i try to bitbake the swupdate-image, i get a "install: cannot stat '/Storage/swupdate-test/sources/poky/build-hw-test-image/tmp/work/armv7at2hf-neon-fslc-linux-gnueabi/mtd-utils/1.5.2-r0/git//include/libubi.h': No such file or directory" Jun 28 20:39:04 so i was going to try to track it down with devtool Jun 28 20:39:20 /git//include/libubi.h Jun 28 20:39:24 Extra slash Jun 28 20:39:42 I don't think thats probably the issue Jun 28 20:39:46 kay Jun 28 20:40:26 Its a little disconcerting that the swupdate-image recipe is poking into the mtd-utils build directory Jun 28 20:40:35 AFAIK, thats not supposed to be happening Jun 28 20:50:32 I see EXTRA_IMAGE_FEATURES and include dbg-pkgs and dev-pkgs but what about bin-pkgs ? I need a -bin pkg in my image Jun 28 20:50:53 yates: You might try dumping the environment of swupdate-image (bitbake -e swupdate-image > env.txt) to see if you can see what it is doing. Jun 28 20:53:26 It seems a little unlikely that mtd-utils is necessarily to blame for this. I'm not sure why it doesn't build with devtool, but if it builds fine normally, I would look at swupate-image instead Jun 28 20:54:19 JPEW: ok, i'll try that -e option - never heard of it. Jun 28 20:54:24 thank you. Jun 28 20:54:47 Very useful... but very verbose. Make sure you redirect to a file :) Jun 28 20:55:43 i thought bitbake was very verbose WITHOUT -e! Jun 28 20:55:56 especially when an error happens.. Jun 28 21:22:49 New news from stackoverflow: create symbolic link in bitbake recipe Jun 28 22:02:57 JPEW: in case its not clear, I'm not trying to be picky about that tests patch, just trying to share some of the things I've learnt with unittest... Jun 28 22:03:09 JPEW: I'm actually really pleased to see new tests! Jun 28 22:03:31 RP: No, thats fine. Always good to get feedback :) Jun 28 22:04:04 JPEW: originally we never used msg either and it was a total nightmare to debug Jun 28 22:04:57 RP: Ya I could see that. It looks like its just deriving from python unittest so you can use all the facilities it has? Jun 28 22:05:28 JPEW: it is python unitttest! Jun 28 22:05:39 JPEW: so yes, and we're 3.x **** ENDING LOGGING AT Fri Jun 29 03:00:00 2018