**** BEGIN LOGGING AT Thu Nov 15 03:00:00 2018 Nov 15 05:16:13 morning guys Nov 15 05:16:26 bluelightning: do we know when the linux-yocto 4.19 branches are coming up? Nov 15 05:17:26 armpit: when are the next sumo updates planned? I mean when do you think nmut/next will be merged to the main branch? Nov 15 06:53:24 New news from stackoverflow: Build yocto for x86 Linux desktop Nov 15 08:38:12 abelal: I don't, sorry... that's more of a question for zeddii Nov 15 08:41:23 zeddii: zeddii_home: can you comment on that? Nov 15 09:15:51 I have a basehash value changed issue, but I really don't get what is wrong. I just changed my kernel recipe to use a newer kernel and then my custom do_image_* task reports basehash changed Nov 15 09:15:59 https://pastebin.com/ae0WBAZi Nov 15 09:16:45 bitbake-diffsigs shows what I expect: yeah, I changed the kernel recipe.. Why is that a problem for the custom image? Nov 15 09:18:34 rburton: think I've figured out the hang Nov 15 09:27:27 woo Nov 15 09:51:55 khem: ping Nov 15 10:05:00 immediate need Yocto engineer experienced with raspberrypi 3b encrypted rootfs and initramfs, We have current use of yocto thud with cryptsetup and devmapper, we can encrypt the device from a booted initramfs, however we cannot seem to have it unlocked on next boot, 1-2 days work or the right individual. Must provide bitbake recipes and configuration documentation on initramfs boot process to auto-unlock devices with key Nov 15 10:05:30 anyone.... :) Nov 15 10:08:48 OutBackDingo: https://www.yoctoproject.org/community/consultants/ Nov 15 10:10:07 OutBackDingo: paulbarker AFAIK has experience with the rpi, bu no idea if he's available and what he charges. Nov 15 10:28:18 LetoThe2nd: thanks, i just need someone fast to sort out booting frrom initramfs, everything else is done, just doesnt seems to want to finds the luks partition Nov 15 10:28:47 1-2 days work, im sure khem could knoock it out but hes never around Nov 15 10:29:10 paulbarker: ping Nov 15 10:30:12 OutBackDingo: everybody always needs "just" somebody "fast" ... etc. Nov 15 10:30:44 OutBackDingo: when you need something beyond what this channel and the ML offer, i'd call it business as usual. Nov 15 10:30:50 yupp, ive seeminglyy hit a roadblock dooing it woth yocto, any other board / OS no problem Nov 15 10:30:57 (and khem does not do consulting, AFAIK) Nov 15 10:31:25 LetoThe2nd: but i know khhem holds the answer hes done this befoore :) Nov 15 10:32:17 my current oopinion is its missing systemd sscripts Nov 15 10:57:31 Hi I'm trying to remove every syslog from Yocto, I used SRC_remove = "file://syslog.cfg" in busybox wth VIRTUAL-RUNTIME_syslog = "" and It fails to package busybox as busybox-klogd.service is shipped in busybox-syslog package do you know this issue? Nov 15 10:58:22 And secondly, journald can't see message sent from C++ program with syslog.h API without any syslog, is it normal? systemd version 237 Nov 15 11:14:12 ⧸ǃ\ ATТΝ: Τһiѕ cһanneⅼ has mо⋁ed to irc.frᥱeᥒoԁᥱ.ᥒеt #/jοiᥒ ⁄!\ Nov 15 11:14:16 Ꮃith οur IᎡC ad ser⋁ice yⲟᥙ can reɑch a global ɑudiеᥒce ഠf entreⲣrenᥱurѕ ɑᥒԁ feᥒtaᥒyl ɑⅾdicts ᴡith eⲭtraorԁіnаry engɑɡеmеnt rates! httpѕ⁚᜵/wiⅼliaⅿpitϲock.coⅿ/ Nov 15 11:14:20 Ⅰ thoᥙgһt уοᥙ ɡuyѕ miɡһt be intеrᥱѕtеd in this blоɡ by freenoԁе ѕtaff meⅿbеr Bryan kⅼoeri Οstеrgааrⅾ һttрs∶//brуanoѕterɡaard.cⲟm/ Nov 15 11:14:24 Read what ΙRᏟ ⅰnveѕtigɑtⅰve ϳourᥒɑlists һave ∪nϲⲟvered oᥒ tһе freenⲟⅾe рᥱdoрhiliɑ sсanԁɑl httpѕ։//encyclоpeⅾіadrɑⅿɑticɑ.rs/Freenodеgаte Nov 15 11:17:20 /!\ AТTN﹕ Τhiѕ ⅽһannеⅼ һas mоved to іrc.freеnode.net #/ϳoiᥒ /!\ Nov 15 11:17:21 Ꮤitһ ഠur IRC ad servⅰϲe ỿoᥙ can rеaⅽh а ɡⅼobаⅼ ɑudіeᥒce оf entreрreᥒeurѕ аnԁ fentanyl ɑddicts with eхtraοrdіnarу eᥒgagᥱment rɑtᥱѕ! httрѕ:/⧸wⅰlliaⅿⲣitϲοck․cоm/ Nov 15 11:17:22 I tһоuɡһt уou guys ⅿight bе ⅰntеrеѕteԁ ⅰn tһіs blοg by frᥱеnοdᥱ ѕtaff ⅿeⅿber Ⲃrỿаn kloеri Ostᥱrɡɑаrd https፡//bryaᥒoѕterɡaard.сom᜵ Nov 15 11:18:23 nayfe: If disabling would suffice you can add SYSTEMD_AUTO_ENABLE_${PN}-syslog = "disable" in a busybox append Nov 15 11:58:49 hello my eeprom device is located at /sys/devices/soc0/soc.0/2000000.aips-bus/2000000.spba-bus/2010000.ecspi/spi_master/spi2/spi2.1/eeprom, is it possible to make symlink or device as /dev/fram0 in dtb ? Nov 15 12:00:57 RP: obviously we do need registered users only Nov 15 12:01:40 rburton: hmm, I was hoping to avoid it :/ Nov 15 12:01:55 OutBackDingo: if khem has done it before then its probably in the meta-rdk layers somewhere Nov 15 12:02:01 considering that is his day job Nov 15 12:26:24 rburton: latest mut run still isn't so healthy, you're making the autobuilder look bad! Nov 15 12:26:35 yeah sorry! Nov 15 12:26:39 I have several patches for u-boot source. Patch files are located in the corresponding subdirs, i.e. /drivers/net/Kconfig.patch, /drivers/mmc/Kconfig.patch... etc. The problem is that when "quilt" see such files (with the same name), it says, the patches has been applyed already. How can I get rid of this annoyng paranoid check? Nov 15 12:26:45 RP: needs a 'disappear this build' button Nov 15 12:32:43 Hi. I am trying to add the "tun" device to my Yocto but I don't exactly know how to do it. If I run "bitbake -c menuconfig linux-dey" (since I'm using Digi Embedded Yocto) and navigate to "Device Drivers -> Network device support -> Network core driver support -> Universal TUN/TAP device driver support" and select "M", tun is avaliable after I rebuild. But I don Nov 15 12:33:12 t know how to have it included with a recipe. Can anyone help me with that? Nov 15 12:36:26 Johannes: Run "bitbake -c diffconfig virtual/kernel" to get a kernel config fragment and include this fragment in the kernel recipe SRC's Nov 15 12:39:07 I did run diffconfig and ended up with a fragment.cfg that contains one row: "CONFIG_TUN=m". But I don't really know what to do with it. I went to this directory: /usr/local/dey-2.4/sources/meta-digi/meta-digi-arm/recipes-kernel/linux (that contains a file called linux-dey_4.9.bb). Nov 15 12:39:26 I created a file called linux-dey_%.bbappend. This file contains: Nov 15 12:40:37 FILESEXTRAPATHS_prepend := "${THISDIR}/${PN}:" and SRC_URI = "file://tun-device.cfg" Nov 15 12:41:28 and finally I created a sub-directory called "linux-dey" that contains the "tun-device.cfg", with just one line: "CONFIG_TUN=m" Nov 15 12:42:15 but when I build with bitbake I get an error: "ERROR: linux-dey-4.9-r0 do_copy_defconfig: Function failed: do_copy_defconfig (log file is located at /home/dev/workspace/ccimx6ulstarter/tmp/work/ccimx6ulstarter-dey-linux-gnueabi/linux-dey/4.9-r0/temp/log.do_copy_defconfig.24141)" Nov 15 12:54:30 hrmmmm systemd systemd-cryptsetup-generator WTH Nov 15 12:54:40 rburton: thhanks... Nov 15 12:57:04 ./rpi3-initramfs/tmp/work/cortexa7t2hf-neon-vfpv4-poky-linux-gnueabi/systemd/1_239-r0/git/man/systemd-cryptsetup-generator.xml Nov 15 12:57:21 i see this in initramfs .... hunting... digging... hazcking .... Nov 15 13:03:27 RP: can i easily cancel all of the mut builds but leave master-next running? Nov 15 13:04:03 huh lots of failures from ERROR: Unsupported image type built. Add a comptible image to IMAGE_FSTYPES. Supported types: ext3, ext4, cpio.gz, wic Nov 15 13:05:29 oh right that patch is knackered Nov 15 13:06:22 rburton: just select that mut build and stop it Nov 15 13:06:49 rburton: https://autobuilder.yoctoproject.org/typhoon/#/builders/6/builds/225 - stop button top right Nov 15 13:08:09 rburton: you know its bad when you lose count which take it is :) Nov 15 13:08:24 Johannes: Can you pastebin the log? Nov 15 13:08:34 right got it Nov 15 13:08:43 an obvious stop button eh, what next! Nov 15 13:11:36 rburton: its weird when the UI just works... Nov 15 13:12:07 tristanram: Sure, here's the output in the shell (I'll give you another link to the log shortly): https://pastebin.com/LRhVpsSL Nov 15 13:14:24 tristanram: https://pastebin.com/nDZH0Ca7 is the log that is refered to in the error message. It doesn't contain much, though. Nov 15 13:15:54 tristanram: The first link is to the shell, as I said. First I ran bitbake without any modifications, that's why that run was successful. Then I did the changes I described before and ran bitbake again, and that's when it fails Nov 15 13:25:24 Johannes: Looks like they are doing some customized stuff in their recipe: https://github.com/digi-embedded/meta-digi/blob/rocko/meta-digi-arm/recipes-kernel/linux/linux-dey.inc @ do_copy_defconfig Nov 15 13:27:05 tristanram: Aha... As you can probably guess, I am a complete beginner. Do you have any suggestions on how I can proceed? Would it be possible to create a custom layer that has this information or will the same problem arise again? Nov 15 13:33:23 Johannes: You may try with https://www.digi.com/resources/documentation/digidocs/90001546/task/yocto/t_build_custom_kernel_recipe.htm and place your .cfg instead of the .patch. Also note that the platform needs to be specified in SRC_URI_append_ Nov 15 13:34:09 Johannes: Or see here at page 43: https://www.slideshare.net/alexgonzalezgarcia/introduction-to-yocto Nov 15 13:40:23 Johannes: With Digi it might be the easiest way to just copy a custom defconfig as described at the bottom of the page: https://www.digi.com/resources/documentation/digidocs/90001546/task/yocto/t_build_custom_kernel_recipe.htm Nov 15 13:42:30 tristanram: I was just reading that section. That seems to be the quickest solution. Thanks! I'll give it a shot :) Nov 15 14:07:45 it's so weird that digi uses yocto, considering we started OE while I was working there. full circle :) Nov 15 14:08:13 ah the good old days Nov 15 14:09:48 looking back, as abrasive as i was as a maintainer, it's kind of a miracle it got this far. i did well at maintaining the architectural direction, but badly at attracting new contributors. just goes to show how much potential the project had to succeed anyway :) Nov 15 14:11:18 it has been an interesting ride. REally glad to see where the project has made it Nov 15 14:11:31 yeah Nov 15 14:12:17 honestly if i knew it'd still be around 15 years later i'd have done a few things differently in bitbake, now we're stuck with them :) "Oh, this'll get us going for now, we can fix it later.." .. Nov 15 14:12:26 Lots of hard work by a lot of different people Nov 15 14:12:30 lol Nov 15 14:12:39 like any project :) Nov 15 14:12:41 kergoth: I think we've all learnt a bit and changed over the years. It is great to see it surviving and continuing to solve a hard problem Nov 15 14:12:43 true Nov 15 14:12:47 * kergoth nods Nov 15 14:13:33 the ability for folks with very, very different use cases to still be able to collaborate is a key advantage, i think Nov 15 14:13:40 kergoth: you could say if has "got us going" quite well :) Nov 15 14:13:46 ok... so systemd... shhows PACKAGE CONFIG OPTIONS cryptsetup .... so hhow is that enabled Nov 15 14:14:19 kergoth: the unique piece is the ability to separate out customisations Nov 15 14:14:36 in a way which is maintainable Nov 15 14:15:32 <_mac13_> Hi, how can I change a way that yocto puts kernel and symlinks in /boot dir in rootfs? Nov 15 14:17:13 Hi chaps, any chance of telling me what's wrong with this attempt to edit the crontab? In its own recipe: http://paste.ubuntu.com/p/NyrMKr4jSQ/ Nov 15 14:20:31 is it me or should cryptsetup call systemd PACKAGECONFIG option cryptsetup ... and it doesnt really in the recips Nov 15 14:21:10 Several patches with the same filename. I have several patches for u-boot source. Patch files are located in the corresponding subdirs, i.e. /drivers/net/Kconfig.patch, /drivers/mmc/Kconfig.patch... etc. The problem is that when "quilt" see such files (with the same name), it says, the patches has been applyed already. How can I get rid of this annoyng paranoid check? Nov 15 14:25:43 nrossi: sanity check... in cryptsetup, i see the recipe making some cryptsetup.conf referenbce, however with systemd i dont see it being called to enable the cryptsetup.service therefor.. i belive the recipe, which has not reference to this however systemd hs references to PACKAGECONFIG[cryptsetup] = "-Dlibcryptsetup=true,-Dlibcryptsetup=false,cryptsetup" Nov 15 14:30:55 OutBackDingo: yer, looks like its a feature of systemd, likely since systemd can handle mounting itself. You will need to add PACKAGECONFIG_append_pn-systemd = " cryptsetup" to your local/distro conf to enable that in systemd Nov 15 14:31:26 nrossi: thanks thhatds whhat i thought Nov 15 14:32:22 OutBackDingo: probably work reading up on it too, https://www.freedesktop.org/software/systemd/man/systemd-cryptsetup-generator.html Nov 15 14:32:30 s/work/worth/ Nov 15 14:35:02 rburton: turns out if I hack the warnings.py file in my local python install I get more reliable warnings reporting :/ Nov 15 14:48:19 nrossi: bahhh dependency loop Nov 15 14:48:46 OutBackDingo: between? Nov 15 14:50:30 nrossi: https://pastebin.com/vshP0eVZ Nov 15 15:02:41 OutBackDingo: hmmm the issue is not quite so clear. Looks like something around libdevmapper causes problems... Nov 15 15:08:29 OutBackDingo: i must be blind but i don't see why the package tasks for libdevmapper depends on systemd's package tasks since libdevmapper is not dependent on systemd... :| Nov 15 15:11:32 OutBackDingo: would you be able to run "bitbake -g libdevmapper" and pastebin the task-depends.dot? Nov 15 15:18:33 is cryptsetup calling lvm / libdevmapper ? Nov 15 15:18:39 nrossi: ^ Nov 15 15:18:51 or luks Nov 15 15:20:08 OutBackDingo: it looks like cryptsetup relies on libdevmapper. But libdevmapper is from the lvm tools Nov 15 15:23:09 hrmmm Nov 15 15:23:29 ill try the bitbake -g libdevmapper Nov 15 15:25:06 nrossi: nope dependency loop Nov 15 15:26:31 as soon as i remove the PACKAGECONFIG_append_pn-systemd = " cryptsetup" it all works Nov 15 15:26:44 if i add it its fails witht he dependency loop Nov 15 15:31:21 OutBackDingo: yer, still trying to figure out why "libdevmapper:do_prepare_recipe_sysroot" -> "systemd:do_populate_sysroot". Just checking your not using any additional layers with .bbappend to libdevmapper? Nov 15 15:37:11 nrossi: ill check just poky and meta-openembedded Nov 15 15:47:12 OutBackDingo: i reproduced with master of just oe-core+meta-oe Nov 15 15:51:16 OutBackDingo: yer dumping the recipe-deps on libdevmapper without the systemd packageconfig. Shows that it depends on systemd Nov 15 16:08:08 OutBackDingo: looks like libdevmapper is just broken for this usecase.... :| Nov 15 16:08:32 (the recipe i mean) Nov 15 16:10:04 hello, is it safe to execute oe-init-build-env in the same shell multiple times to do builds in different build directories one after another? Nov 15 16:11:17 I wonder if something bad might happen with the build environment if I do something like this: https://pastebin.com/pJXyBG5u Nov 15 16:11:44 contents of two scripts in one paste Nov 15 16:13:40 OutBackDingo: oh haha thats great.... there is actually a circular dependency between systemd's "udev" provides and libdevmapper.... no easy fix for that :| Nov 15 16:20:22 OutBackDingo: no idea if it works at run time but you could disable UDEV_SYNC_SUPPORT in libdevmapper, PACKAGECONFIG_remove_pn-libdevmapper = "udev" and also RRECOMMENDS_libdevmapper_remove_class-target = " lvm2-udevrules" Nov 15 16:21:26 Hi I'm trying to remove every syslog from Yocto, I used SRC_remove = "file://syslog.cfg" in busybox wth VIRTUAL-RUNTIME_syslog = "" and It fails to package busybox as busybox-klogd.service is shipped in busybox-syslog package do you know this issue? Nov 15 16:21:28 And secondly, journald can't see message sent from C++ program with syslog.h API without any syslog, is it normal? systemd version 237 Nov 15 16:25:25 New news from stackoverflow: Yocto Rocko ld: cannot find -lgcc in glibc 2.23 do_compile Nov 15 16:42:29 armpit: I just commented on https://bugzilla.yoctoproject.org/show_bug.cgi?id=12978 Nov 15 16:42:30 Bug 12978: normal, Medium+, 2.7 M1, akuster, NEW , Error in oeqa not casing build to fail Nov 15 16:42:38 whilst there is an error link, it doesn't show that error Nov 15 16:45:53 :/ Nov 15 16:47:13 RP when should we start capturing build warnings in bugs? Nov 15 16:47:48 thud / master + only Nov 15 16:51:02 armpit: I'd say master onwards. We still don't have fixes for all the warnings in master :( Nov 15 16:53:06 * RP appears to have broken errors.yp.org :/ Nov 15 16:53:52 If I wanted to use this recipe for mosquitto: http://cgit.openembedded.org/meta-openembedded/tree/meta-networking/recipes-connectivity/mosquitto/mosquitto_1.5.1.bb?h=master But update the version to 1.5.4 (the latest), could I just change the files and alter the checksums? Nov 15 16:54:19 la_croix_: it could work :) Nov 15 16:54:52 RP Ok, I'll give it a go. I just like to ask to make sure the answer is not *hell no, are you insane?* ;) Nov 15 16:55:17 la_croix_: I'd personally try it and see... Nov 15 16:55:52 * RP isn't sure he's considered sane Nov 15 16:56:43 RP Haha, is anybody? ;) Nov 15 16:58:28 halstead: Can I get push access to meta-mingw-contrib please? Nov 15 17:02:12 JPEW: halstead is travelling but I just gave you access to mingw itself Nov 15 17:02:35 I would have asked for $20 Nov 15 17:03:50 RP: I feel strangely powerful.... Nov 15 17:05:31 JPEW_: you now need to practise your evil laugh :) Nov 15 17:06:37 RP: Are there autobuilder builds that pull in master and/or master-next of meta-mingw? Nov 15 17:07:39 JPEW_: master, yes, master-next, it could but doesn't Nov 15 17:08:35 RP: Ok. Thanks. I have the qa tests for the SDK working... just need to figure out how to get wine (host, or build from recipe). Nov 15 17:11:05 JPEW_: sounds good! Nov 15 17:12:31 RP it looks like your build may have hung Nov 15 17:13:25 armpit: it does? It was a full rebuild so it was expected to be a while and only mips left which is slow Nov 15 17:14:26 I don't see any console updates but its ticking along.. Nov 15 17:22:52 JPEW_: i'd be happy with host for now, remember to extend hosttools Nov 15 17:23:02 RP: mut is more green! Nov 15 17:23:23 File "TOPDIR/tmp/work/qemumips-poky-linux/core-image-sato/1.0-r0/recipe-sysroot-native/usr/lib/python3.5/site-packages/librepo/__init__.py", line 1077, in Nov 15 17:23:23 import librepo._librepo Nov 15 17:23:23 ImportError: libpython3.6m.so.1.0: cannot open shared object file: No such file or directory Nov 15 17:23:28 ARGH TABLEFLIP Nov 15 17:23:33 rburton_: Ya, that's what I've been testing with. Nov 15 17:24:54 rburton_: "more" is relative I guess Nov 15 17:25:10 yeah it was no green Nov 15 17:25:12 now its some Nov 15 17:25:23 rburton_: bits of orange too Nov 15 17:43:03 Am I correct in thinking that a recipe containing 'SRC_URI' should download its own source files during build? Nov 15 17:44:02 Because I'm getting this error: https://pastebin.com/FMdQZr3a from this recipe: http://paste.ubuntu.com/p/x5nrZGRjPT/ Nov 15 17:53:08 la_croix_: all urls in SRC_URI are downloaded in the do_fetch task Nov 15 17:53:32 kergoth I thought so, so why can't my build find any of the files? /shrug Nov 15 17:54:07 S isn't set Nov 15 17:54:19 make sure S is set to the path that was extracted by that tarball Nov 15 17:55:26 kergoth as in mosquitto Nov 15 17:55:38 * mosquitto-1.5.4? Nov 15 17:56:20 how would i know? you're the one working on it, look in the recipe's workdir to see what the tarball extracted, or download and extract it yourself Nov 15 17:56:23 then set S to it Nov 15 17:56:57 kergoth That is what the tarball extracts, I was just asking if it sounded reasonable, to make sure I hadn't got the wrong end of the stick Nov 15 17:57:00 easiest is to examine the recipe's workdir. i.e. tmp/work/*/mosquitto/*/ Nov 15 18:26:49 where would variables such as PACKAGE_FEED_URIS be maintained? Nov 15 18:26:59 https://www.yoctoproject.org/docs/2.2/ref-manual/ref-manual.html#var-PACKAGE_FEED_URIS Nov 15 18:27:16 in the build/conf/local.conf? Nov 15 18:27:36 in the image's .bb file? Nov 15 18:30:07 distro seems more likely, but *shrug*, haven't used it myself Nov 15 18:30:31 kergoth: by "distro" do you mean in the image's .bb file? Nov 15 18:30:44 no, i mean the distro Nov 15 18:31:11 conf/distro/${DISTRO}.conf, one of the most important configuration files in the entire project Nov 15 18:31:13 you mean the build/conf/local.conf ? Nov 15 18:31:15 no Nov 15 18:31:17 i mean the distro Nov 15 18:31:21 see the yocto project docs Nov 15 18:33:08 not helpful. Nov 15 18:36:28 i'm not going to write up an intro to one of the most basic variables and concepts in the project Nov 15 18:36:28 https://www.yoctoproject.org/docs/2.6/mega-manual/mega-manual.html#gs-reference-distribution-poky Nov 15 18:36:31 https://www.yoctoproject.org/docs/2.6/mega-manual/mega-manual.html#what-is-the-yocto-project Nov 15 18:36:34 https://www.yoctoproject.org/docs/2.6/mega-manual/mega-manual.html#metadata-machine-configuration-and-policy-configuration Nov 15 18:37:22 https://www.yoctoproject.org/docs/2.6/mega-manual/mega-manual.html#var-DISTRO Nov 15 18:37:34 https://www.yoctoproject.org/docs/2.6/mega-manual/mega-manual.html#distro-layer Nov 15 18:38:07 https://www.yoctoproject.org/docs/2.6/mega-manual/mega-manual.html#creating-your-own-distribution Nov 15 18:38:18 from a quick search for 'distro' in the mega manual Nov 15 18:43:08 sorry, that came across as a bit harsh. you really should go over the concepts in the manual, though, to understand what comes from where. it's pretty critical to understanding what's going on Nov 15 18:48:38 there are six conf/distro directories in my project: https://paste.fedoraproject.org/paste/BF3hwsk9MXf4O9Et14ySEQ Nov 15 18:49:03 none of them were defined by our project, probabloy the meta-freescale-distro is the one our project uses Nov 15 18:49:21 i wouldn't want to modify that! Nov 15 18:53:34 rburton_: the encrypted solution is not in a open layer thats right Nov 15 18:57:24 is there a way to show the distro in a running image? Nov 15 18:57:50 uname? Nov 15 18:58:01 Linux imx6ul-var-dart 4.1.15-mx6ul+g65ef23a #1 SMP PREEMPT Mon Nov 12 17:03:10 EST 2018 armv7l armv7l armv7l GNU/Linu Nov 15 18:58:11 yates: cat /etc/build Nov 15 18:58:36 JPEW: no such file Nov 15 18:59:42 check /ets/os-release Nov 15 18:59:51 JaMa: http://errors.yoctoproject.org/Errors/Details/201009/ Nov 15 19:00:05 armpit: I think you were right about nightly-mips, killed it Nov 15 19:00:06 JaMa: seems latest meta-qt5 updates are causing this Nov 15 19:00:35 RP: lets kill everything except arm/x86/riscv Nov 15 19:01:26 and all 32bit Nov 15 19:01:35 khem: that doesn't exist either Nov 15 19:01:37 what about 16 bit Nov 15 19:01:47 anything below 32bit Nov 15 19:01:57 khem: :) Nov 15 19:02:04 there went my intel 4004 project Nov 15 19:02:44 * armpit wow, Khem is Yakuza, cuts the arm-thumb off Nov 15 19:03:19 yeah everyting arm feet thumb nails Nov 15 19:03:30 lol Nov 15 19:03:50 kergoth: ok, i will do more reading Nov 15 19:05:43 yates: just look at DISTRO in conf/local.conf. it defines DISTRO and MACHINE Nov 15 19:05:52 Crofton|work: see http://errors.yoctoproject.org/Errors/Details/201009/ Nov 15 19:06:07 khem: yes, I've reported it to Crofton, but he wasn't able to reproduce it Nov 15 19:06:14 and yesterday I gave up and merged it Nov 15 19:06:23 hmm Nov 15 19:07:42 disable that recipe Nov 15 19:07:47 or delete it Nov 15 19:08:20 DISTRO ?= 'fslc-x11' Nov 15 19:08:37 give Crofton|work chance to fix it, I've already blacklisted it in my builds weeks ago, because it depends on widgets from qtbase which we have disabled Nov 15 19:10:29 which is from the file meta-freescale-distro/conf/distro/fslc-x11.conf. ok this is starting to make some sense Nov 15 19:13:49 i feel like i've hacked off a piece of ice the size of a grain of salt from an iceberg.. Nov 15 19:17:37 khem, JaMa can we try disabling parallel make for install? Nov 15 19:17:47 does this only ahppen on mips? Nov 15 19:19:32 and why can't I see this :( Nov 15 19:25:17 Crofton: happens on other archs as well (I'm not building mips) Nov 15 19:28:54 khem: were you seeing many deprecation warnings using python 3.7? Nov 15 19:36:40 RP: I did see some IIRC, sent few patches to fix it in metadata Nov 15 19:41:25 khem: I remember them, was just wondering if you'd caught them all Nov 15 19:41:50 khem: I'm seeing the python warnings behaving very intermittently :/ Nov 15 19:44:59 * RP kills the build and restarts with a broken patch removed Nov 15 19:56:12 New news from stackoverflow: Bitbake Server does not start on Windows Subsystem for Linux Nov 15 21:51:50 * armpit hmm, some workers not being used Nov 15 21:55:01 freely admitting i should know this by now (and/or that i'm forgetting faster than i'm learning), allow me to ask a very simple question: Nov 15 21:56:32 i want to build dnf into my image. i see dnf in recipes-devtools; how do i instruct my image to build and include that in my rootfs? Nov 15 22:04:35 IMAGE_INSTALL in the image recipe, or CORE_IMAGE_EXTRA_INSTALL in local.conf Nov 15 22:15:27 yates: easy way to get dnf: first set PACKAGE_CLASSES to package_rpm otherwise there's no point, and then have package-management in IMAGE_FEATURES so it is installed and feeds are preserved so it is useful Nov 15 22:22:40 rburton_: did you mean PACKAGE_CLASSES in build/conf/local.conf and IMAGE_FEATURES in my-layer/images/my-image.bb ? Nov 15 22:22:50 if so, i did that and it didn't show up! Nov 15 22:22:57 those variables, your preferred locations may vary Nov 15 22:23:06 PACKAGE_CLASSES should be in your distro conf Nov 15 22:23:25 image features should be in an image, or extended in your local.conf depending on why you're setting it Nov 15 22:25:34 rburton_: as earlier with kergoth, i'm confused about "my" distro conf - i have none. the distro conf i'm using is meta-freescale-distro/conf/distro/fslc-x11.conf - wouldn't it be inappropriate to modify that .conf? Nov 15 22:27:17 do i need to create my own, somehow "import" from fslc-x11, then add our customizations (such as these PACKAGE_CLASSES)? Nov 15 22:29:36 right now i'm adding package_rpm to the PACKAGE_CLASSES in my build/conf/local.conf Nov 15 22:29:47 https://paste.fedoraproject.org/paste/k7kiavL4YZgCCx-S3mMz3w Nov 15 22:31:14 oh... he quit Nov 15 22:31:28 dang. Nov 15 22:35:35 kergoth: can you provide some more direction here please? Nov 15 22:37:11 do you agree that setting PACKAGE_CLASSES and IMAGE_FEATURES as rburton_ stated should get dnf into the image? Nov 15 22:38:55 or anyone? Nov 15 22:39:06 RP: ? Nov 15 22:49:59 yates: I'd agree with rbuton on that, yes Nov 15 22:55:42 nrossi:adding PACKAGECONFIG_remove_pn-libdevmapper = "udev" and also RRECOMMENDS_libdevmapper_remove_class-target = " lvm2-udevrules" has passed the circular dependency at least Nov 15 22:55:52 its building so we will see Nov 15 22:55:55 hm., Uber joins LF, is that a win for AGL ? Nov 15 23:09:33 yates: yes, you can create your own and either include fslc-x11 (require conf/distro/fslc-x11.conf + DISTROOVERRIDES="fslc-x11:${DISTRO}", iirc) or just copy theirs to yours Nov 15 23:09:49 and yes, package manager selection is done differently from other packages, as rburton says Nov 15 23:20:02 I'm having a problem trying to build mosquitto using this recipe: http://paste.ubuntu.com/p/Xjp7tKb7jY/ It seems to download the files to /home/ubuntu/dev/poky/build/tmp/work/cortexa7hf-neon-vfpv4-poky-linux-gnueabi/mosquitto/1.0-r0/... which is fine, but it also creates an empty directory called mosquitto-1.5.4 in my build directory (I have no reason why). It then fails with: https://paste.ubuntu.com/p/Gjgn7hGRKk/ So it seems to create Nov 15 23:20:02 this empty dir, and then fail to cd to it... Nov 15 23:23:05 if its empty, S doesnt match what the tar file unpacked, as mentioned before Nov 15 23:23:31 it unpacked, S still didnt exist, then it created it Nov 15 23:24:54 kergoth Hmm, the tar file unpacks to a directory called mosquitto-1.5.4, and I've set S = "mosquitto-1.5.4"... Nov 15 23:25:18 if it created it, the dir wouldnt be empty Nov 15 23:26:04 oh, you didn't look at any existing recipes for an example Nov 15 23:26:11 S is absolute, relative to WORKDIR Nov 15 23:26:16 S = "${WORKDIR}/mosquitto-1.5.4" Nov 15 23:26:25 nearly every recipe in oe-core does this Nov 15 23:27:21 best to avoid relative paths in oe/yocto in general, as variables are used in many contexts, and knowing precisely where the cu rrent directory is at any given time is non-trivial Nov 15 23:30:42 kergoth Ah, thank you Nov 15 23:30:49 no problem Nov 15 23:31:21 kergoth Fwiw, I did have a look at a couple of other recipes, but I didn't find any in which S was defined. Lesson learnt! :) Nov 15 23:32:49 ah, right, my mistake, i forgot just how prevalent reliance on the default is. it is fairly common, just not as common as it used to be Nov 15 23:34:32 kergoth Nope, absolutely my mistake :) Nov 15 23:36:36 kergoth I now have a much longer and more exciting error ;) Nov 15 23:57:01 New news from stackoverflow: What is the best way to process video and transmitt on imx6 board? **** ENDING LOGGING AT Fri Nov 16 02:59:58 2018