**** BEGIN LOGGING AT Tue Oct 03 03:00:01 2017 Oct 03 03:37:21 How can I auto start an application after the kernel loads? Oct 03 03:43:59 How can I reset bitbake to rebuild my image after changing a branch of poky? Oct 03 06:35:54 morning Oct 03 06:36:04 is this page up to date? https://wiki.yoctoproject.org/wiki/Releases Oct 03 06:36:33 I see that there is a tagged release yocto-2.3.2 but can't see it in the page above Oct 03 06:37:06 we are going to upgrade to pyro, just wondering if we should use 2.3.2 or 2.3.1 Oct 03 06:49:48 melonipoika: wikis are notoriously prone to getting out of date :) I suggest you follow yocto-announce list instead, and yes 2.3.2 is fine Oct 03 07:25:28 thank you :-) Oct 03 08:04:57 hello, how do I checkout source via git in a bitbake recipe when git submodules are in use (i.e. I need a git clone --recursive to be done automatically)? Oct 03 08:34:43 i all, I am starting with poky recently, working through the quick start tutorial. During my first bitbake, I am struggling with very low download/fetch speeds (0: binutils-cross-i586-2.28-r0 do_fetch (pid 3105) 8% |####### | 13.5K/s). My normal connection is 48 MBit/s (checked). All I could find was tutorials/hints on setting up my own local mirror. Oct 03 08:34:53 When I download files from upstream sources mirror http://downloads.yoctoproject.org/mirror/sources/ directly, it gives the same slow speeds... Oct 03 09:16:06 Hi there Oct 03 09:16:14 I'm trying to bump yocto version to pyro Oct 03 09:16:29 what influence can have that sysroot is splitted to packages not per machine Oct 03 09:16:45 I was using yocto generated sysroot to cross compile my app using cmake Oct 03 09:17:49 http://www.yoctoproject.org/docs/2.3/ref-manual/ref-manual.html#moving-to-the-yocto-project-2.3-release Oct 03 09:19:52 hnje: ok fine Oct 03 09:20:03 open-nandra I am starting the upgrade too. I run into the first issue: Execution of event handler 'staging_taskhandler' failed Oct 03 09:20:16 ${@oe.utils.build_depends_string(d.getVar('PACKAGE_WRITE_DEPS'), 'do_populate_sysroot')} which triggered exception AttributeError: 'NoneType' object has no attribute 'split' Oct 03 09:20:41 too bad pastebin is blocked in my company's network :-( Oct 03 09:20:48 but it means also that I need to somehow change cmakelist.txt to be able to work with sysroot per package (when I want to recompile by hand only app and not whole yocto) Oct 03 09:21:26 i haven't investigated this further yet, but just a heads up if you see something similar Oct 03 09:21:27 or it will be done automagically? Oct 03 09:21:42 melonipoika: thx Oct 03 09:22:11 I am on this rev for poky: 6a1f33cc40bfac33cf030fe41e1a8efd1e5fad6f Oct 03 09:22:16 2.3.2 Oct 03 09:23:12 melonipoika: er - I just checked out that revision, PACKAGE_WRITE_DEPS isn't even referred to there... ? Oct 03 09:24:14 open-nandra: you shouldn't need to change your cmakelists.txt - just ensure you specify all of your dependencies and it should build Oct 03 09:24:29 (by "specify all of your dependencies" I mean in DEPENDS within the recipe) Oct 03 09:24:32 bluelightning: ok thanks Oct 03 09:24:38 yes sure Oct 03 09:25:50 thanks for info Oct 03 10:17:19 ups, sorry bluelightning... i was on the wrong path Oct 03 10:17:40 this is the commit I have checked out for my pyro build: ce26a57 (HEAD, tag: refs/tags/yocto-2.3.2, tag: refs/tags/pyro-17.0.2) Revert "expat: Don't use getrandom() in the -native case" Oct 03 10:28:00 melonipoika: can you see which recipe this is triggering in? Oct 03 10:29:26 melonipoika: this is weird... the error you mentioned pretty much means that PACKAGE_WRITE_DEPS is not set, and yet I can't see how that is possible - it gets defaulted to "" Oct 03 10:30:45 specifically it gets defaulted in package.bbclass, and the code referred to in the error is in the package_deb/package_ipk/package_rpm classes, which all inherit package, so the default should definitely be set Oct 03 10:37:46 thanks bluelightning. I will check if any of the recipes in our layers are changing this variable Oct 03 10:38:17 the first recipe that returns this error is python3-pycrypto_2.6.1.bb Oct 03 10:38:45 but i can see from the error that some our our classes are being execute... let's see what we are doing here :-) Oct 03 10:41:05 no, we are not modifying PACKAGE_WRITE_DEPS Oct 03 10:42:54 thanks for your help, without being able to share logs I am afraid it will be very difficult for you to help more... Oct 03 11:32:41 bluelightning, I think I found the problem... we are importing package.bbclass in one of our layers insted of using the one from poky... Oct 03 11:33:13 Hi, I have a question about yocto linux on raspberry pi. I am using systemd as an init manager, which works fine for the most part. But when I run systemd-analyze directly after i logged in, it says that it is still booting. After 1:30 minutes this is done. I have checked the boot plot from systemd-analyze plot, and i see that /dev contains a lot of devices that aren't on the py. What is causing this? Oct 03 11:35:41 melonipoika: that would do it. now you get to figure out what was changed in your layer and how to port it to the new package.bbclass... Oct 03 11:46:47 Hi. Is it correctly assumed that custom IMAGE_FEATURES cannot be added (seems to be a validation of existing IMAGE_FEATURES)? I can of course use DISTRO_FEATURE or MACHINE_FEATURES instead, but IMAGE_FEATURES naming convention feels more correct in my situation. Oct 03 11:50:18 melonipoika: ah that would explain it Oct 03 12:01:23 yep, that was it... building now :-D thank you! Oct 03 12:02:22 so we copied that class to our layer since we don't want to modify upstreaming layers directly. Now I need to see if the changes we made to package.bbclass need to be upstreamed Oct 03 12:04:53 so all changes we made to package.bbclass are so that we can produce a single file with all the open source licenses for the open source packages that we ship with our devices. I guess this could be useful upstream too Oct 03 12:05:23 why does that needs to be done in package.bbclass? Oct 03 12:08:19 i don't know, let me see the git history Oct 03 12:09:15 not sure *how* that can work in do_package, as images contain different sets of licenses Oct 03 12:09:36 as a rootfs hook, sure Oct 03 12:09:37 Regarding my question about IMAGE_FEATGURES. If I remember it correct, MACHINE_FEAURES cannot be set from an image recipe context. So what I need is probably IMAGE_FEATURES after all. Oct 03 12:10:09 So back to the original question, can I make my own customized IMAGE_FEATURES without edit the validation part in image.bbclass? Oct 03 12:11:10 rburton we tried using the manifest but there were some fields that need to be read directly from the recipe Oct 03 12:11:49 i am trying to understand why the developer that made these changes had to use this class... Oct 03 12:34:30 rburton we use package.bbclass to extract some info about the packages from the recipe, then we added another class to generate the package list Oct 03 12:34:41 and we use a rootfs hook: ROOTFS_POSTPROCESS_COMMAND_append = "create_packagelist" Oct 03 12:34:55 you can almost definitely do that without needing to patch package.bbclass Oct 03 12:35:19 I will add a story to our backlog to try upstreaming our changes, probably there are better ways to do it :-) Oct 03 12:54:52 denix: thx for adding the patch to meta-ti Oct 03 12:55:06 denix: /but/ we need to branch for pyro Oct 03 12:55:35 denix: 511b167f6683b76ea8f3853e3b8ea0199b0fa25d is good for pyro Oct 03 12:57:12 rburton: a prove about lttng issues: https://ci.ossystems.com.br/view/All%20failing%20jobs/job/fsl-community-bsp-master_x11/lastBuild/console Oct 03 12:58:15 Hi all ! I'm doing a meta that can be share for two differend board. So sometimes i could have recipes that could apply only for one board and not for the other one. Such as for exemple u-boot-variscite that is for only variscite board... What is the best way ? Oct 03 12:58:21 Thanks by advance. Oct 03 12:58:53 What is the best way, to apply recipes depending machine*? Oct 03 13:00:42 ChrysD: check how we did for meta-freescale Oct 03 13:00:49 ChrysD: it has a bunch of them Oct 03 13:01:08 ChrysD: meta-freescale-3rdparty offers variscite support Oct 03 13:02:04 ChrysD: COMPATIBLE_MACHINE etc Oct 03 13:19:15 Hey everyone. Does anyone know, does Yocto have the Qualcomm Diag support installed? Oct 03 13:19:17 hello guys. Could somebody help me with syslog and systemd in krogoth? I send a message to the remote host using logger, I see the message in the tcpdump, but I don't see it in the logs :-( Oct 03 13:20:34 btw syslogd is from the busybox package Oct 03 13:30:01 otavio : thanks Oct 03 13:30:04 rburton : ok thanks Oct 03 14:58:15 YPTM: - dial into: 1-800-262-0778/404-397-1527 - enter the bridge number: 88748961# Oct 03 14:58:26 YPTM: sjolley joined Oct 03 14:58:37 YPTM armin is on Oct 03 14:58:44 YPTM: stephano joined Oct 03 15:00:37 YPTM: Saul joined Oct 03 15:00:56 * sgw waves morning all Oct 03 15:00:57 YPTM: Mark here.. Oct 03 15:02:03 YPTM: Michael on the call. Oct 03 15:02:07 YPTM: Joshua joined Oct 03 15:02:31 YPTM: brian joined Oct 03 15:02:46 YPTM ross trying to join but fatfingering Oct 03 15:03:07 in! Oct 03 15:05:35 YPTM: Joshua Watt here Oct 03 15:07:17 YPTM: Randy joined the YP Tech call Oct 03 15:07:33 ypttm, Bill Mills here Oct 03 15:10:07 sjolley for next YP -- my intention is to resume to bitbake/setup/layerindex working Oct 03 15:14:21 fyi: https://www.openembedded.org/wiki/OEDEM_2017 Oct 03 15:14:27 MArch 12-14 2018 Oct 03 15:15:00 thanks Oct 03 15:15:16 np :), ptr to oedam prague? I keep finding the old ones Oct 03 15:15:42 nvm Oct 03 15:15:49 silly search :/ Oct 03 15:21:23 What does "COMPATIBLE_MACHINE="(label1|label2)" means ? Does label1 is something that contain in the MACHINE variable or is the exact name ? Oct 03 15:22:53 ChrysD: COMPATIBLE_MACHINE contains a regular expression. Oct 03 15:23:36 Saur : ok perfect thanks. Oct 03 15:26:06 Is there any reason to use d.expand("${FOO}") instead of d.getVar("FOO") ? My guess is that the result is the same, but that d.getVar() may be slightly faster. Oct 03 15:36:06 Saur: the only reason to use expand is if it's more complex and you don'tw ant to assemble the string piecemeal, i.e. you have a bunch of vars involved Oct 03 16:08:21 kergoth: As I assumed then. Oct 03 16:31:36 khem: seen http://undeadly.org/cgi?action=article;sid=20170930133438 ? Oct 03 16:31:43 rather interesting Oct 03 16:53:22 Hey everyone. Oct 03 16:53:53 Im running yocto on my dragonboard 410c. How can I reset it back to factory? Oct 03 17:00:45 Are here some Rust user? I've a question about the meta-rust layer? Oct 03 17:01:27 tcpdump: delete your tmp within your build. Oct 03 17:01:37 dir Oct 03 17:06:30 kergoth: we force -fvisibility=hidden in OE Oct 03 18:04:46 zzeroo: are you using rust? Oct 03 18:27:17 Is it possible to have base-files override a config file provided by a different package in a straightforward way? Oct 03 18:42:58 Are there any native apps in Yocto that show cpu volatages, power usage, temps, etc? Oct 03 18:51:22 otavio: fyi lttng-modules has a tagged 2.9.4 release now, which has the fixes for supporting 4.13 Oct 03 18:52:07 khem: ah, right, forgot about that Oct 03 18:52:24 ntl: great! rburton would you guys accept this upgrade as it is a bugfix release? Oct 03 18:55:13 tcpdump: I believe lcdproc (in meta-oe) will show you that sort of information Oct 03 18:55:34 possibly not the kind of UI you are after though Oct 03 18:57:17 Ill check it out bluelightning thanks. Oct 03 19:04:45 otavio: yes, if you make the 'bugfix release' clear in the commit message, and follow the policy of master first, releases second when sending patches Oct 03 19:05:10 I think there is a rocko branch now already Oct 03 20:27:14 otavio: Yes Oct 03 20:35:49 zzeroo: how has been the experience? we are using go a lot Oct 03 20:36:02 but my interest for rust is growing lately Oct 03 22:14:57 otavio: that patch from mattsm(?) for go and musl on x86-64 did not work, I guess he is working another fix for bug #12136 Oct 03 22:14:58 Bug https://bugzilla.yoctoproject.org/show_bug.cgi?id=12136 critical, High, 2.4 M4, matt, ACCEPTED , go-dep and go-helloworld fail to link with MUSL on x86-64 machines typesj **** ENDING LOGGING AT Wed Oct 04 03:00:02 2017