**** BEGIN LOGGING AT Fri Apr 05 02:59:57 2019 Apr 05 06:26:07 New news from stackoverflow: How to determine NTP communication status as a non root user Apr 05 06:37:59 psrcode: ptest for lttng-tools is much happier, thanks! :) Apr 05 06:38:11 psrcode: down to one failure now :) Apr 05 06:38:50 kanavin: https://autobuilder.yocto.io/pub/non-release/20190404-16/testresults/qemux86-64-ptest/testresults.json is the last set of results. My "-u" experiment seems to fix openssh. e2fsprogs still times out intermittently and still have the python3 problem :( Apr 05 06:40:15 * RP wonders if python3 is also abuffering problem Apr 05 07:03:18 good morning Apr 05 08:23:35 Heyho Apr 05 08:27:20 I'm a newcomer to the yocto project and I've encountered a problem when trying to include grpc which is located in the meta-openembedded/meta-network layer into my project: the master branch of meta-openembedded is incompatible with oe core but the rocko branch (which is compatible) does not contain the grpc recipes. Could someone please pinpoint me in the right direction? (Is there another way than merging the two branches of th Apr 05 08:31:17 BloodSurfer: copy the recipe in your custom layer and test it Apr 05 08:31:35 BloodSurfer: essentially 3 options: 1) upgrade all your layers to thud 2) include a layer that contains grpc in rocko (layers.openembedded.org says meta-iot-cloud does) 3) what mckoan says Apr 05 08:44:23 @mckoan and @jku: Thanks for the hints. Apr 05 09:25:29 RP: right, I wrote down my python findings from yesterday's in the bug https://bugzilla.yoctoproject.org/show_bug.cgi?id=13257 Apr 05 09:25:30 Bug 13257: normal, Medium+, 2.7 M4, ross.burton, IN PROGRESS DESIGN , [2.7 M3 rc1] python3 ptest facing timeout issue Apr 05 09:56:22 Hi, anybody here who managed to change the yocto config files for uboot from sd to qspi? Apr 05 10:01:18 I simply changed UBOOT_CONFIG ??= "sd" to UBOOT_CONFIG ??= "qspi1". But that lead to just compile errors. Apr 05 10:03:35 build-wayland/tmp/deploy/images/imx8mm/u-boot-spl.bin-imx8mm-qspi1 no such file or directory Apr 05 10:04:42 Is there a documentation page which explains how to switch the type of device for uboot? Apr 05 10:05:42 I googled but there's nothing found about how to adapt the yocto configuration for uboot Apr 05 10:09:08 Near the line 'UBOOT_CONFIG ??= "sd"' is another line: Apr 05 10:10:11 IMXBOOT_TARGETS = "${@bb.untils.contain{'UBOOT_CONFIG', 'fspi', 'flash_evk_flexspi', 'flash_evk', d)}" Apr 05 10:10:28 It does not contain 'qspi1'. Apr 05 10:10:44 This could be the reason for the compile error. Apr 05 10:11:34 But how to I add support for qspi? The PDFs of NXP say yocto supports uboot for qspi for that board Apr 05 10:11:54 s/to/do Apr 05 11:00:15 How do I use AUTOREV? When I set SRCREV to "HEAD", everything wokrs fine, but when I set it to "${AUTOREV}" then I get the following error: bb.data_smart.ExpansionError: Failure expanding variable SRCPV, expression was ${@bb.fetch2.get_srcrev(d)} which triggered exception FetchError: /usr/bin/env: 'svn': No such file or directory Apr 05 11:18:41 d_s_e: yeah that's a bug. workaround is to add svn to HOSTTOOLS in bitbake.conf. Apr 05 11:22:54 https://bugzilla.yoctoproject.org/show_bug.cgi?id=11442 fwiw Apr 05 11:22:55 Bug 11442: normal, Medium, 2.99, richard.purdie, NEW , svn AUTOREV recipe parse time explosion Apr 05 11:26:48 rburton: ok, thanks. will this be fixed in the next release? Apr 05 11:26:57 d_s_e: nope Apr 05 11:27:39 first thing is to decide how to fix it. i guess we could add svn to hosttools-not-fatal so it is used if present Apr 05 11:27:55 on the assumption that the breaking svn changes are old enough now that is a safe thing to do, and not actively reckless Apr 05 11:29:27 rburton: adding it to hosttools_nonfatal seem legit. what kind of breaking changes? Apr 05 11:31:27 d_s_e: a while back svn changed the repository format on disk. if your host has the old version then it won't be able to work with the checkouts. Apr 05 11:31:41 it was quite a while ago now, not sure if we support anything that ships that release anymore Apr 05 11:31:57 then again, suse and centos are known to be ancient Apr 05 11:49:53 rburton: you mean the breaking changes in the working copy format? that were some time ago, yes. Apr 05 11:50:28 that was with svn 1.8 I guess. I'm on Debian here, and even that one has svn 1.9 Apr 05 11:54:45 hello, does anyone know why some files installed in ${IMAGE_ROOTFS} end up with a different md5 than those in the .ipk files ? Apr 05 12:04:25 rrerolle: do you have prelink enabled? Apr 05 12:04:57 rrerolle: or there are some other rootfs postprocess tasks which might modify them Apr 05 12:05:12 yes prelink is enabled Apr 05 12:05:14 like mklibs Apr 05 12:05:53 yes image-prelink and image-mklibs are enabled on this image Apr 05 12:08:42 in fact what I'm trying to do is generating an md5sum file of the whole rootfs, I'm doing it in a ROOTFS_POSTPROCESS_COMMAND, but the md5s don't match Apr 05 12:11:23 JaMa: don't mklibs and prelink also run as a ROOTFS_POSTPROCESS_COMMAND ? Apr 05 12:22:04 rrerolle: no, they run as IMAGE_PREPROCESS_COMMAND Apr 05 12:35:32 * armpit heavy metal... Apr 05 12:36:00 armpit: what? where? Apr 05 12:36:47 here in India.. they seem not to be phased by it.. Apr 05 12:37:49 It almost sounds like Queensyce but hasher Apr 05 12:37:57 armpit: here, have some punjabi metal https://www.youtube.com/watch?v=i4FqGPRQWFM Apr 05 12:38:31 JaMa: so I need to find a way to run as IMAGE_PREPROCESS_COMMAND, but *after* mklibs and prelink Apr 05 12:38:57 rrerolle: may i suggest IMAGE_POSTPROCESS_COMMAND Apr 05 12:39:07 then they pack some pop Apr 05 12:39:26 not sure why i'm surprised punjabi metal exists Apr 05 12:39:30 that's quite a combination of sounds Apr 05 12:39:48 rburton: rule666: if it exists, there is a metal version of it. Apr 05 12:40:03 ahah Apr 05 12:40:17 happy birthday song ? Apr 05 12:42:03 the funny thing about the heavy metal is the elderly in the pub are not bothered by it Apr 05 12:43:12 armpit: here you go https://www.youtube.com/watch?v=ENuWdhXgeGI Apr 05 12:43:41 does anyone here use ISO images from bitbake? Apr 05 12:44:07 which standard? 9001 or something else Apr 05 12:44:09 armpit kergoth fray: know any customers that actually use the iso images? Apr 05 12:44:11 yeah Apr 05 12:44:29 specifically the iso image type, mkisofs Apr 05 12:45:24 very tempted to rip out cdrecord from core and replace the iso stuff with xorriso Apr 05 12:45:41 got the recipes but the class needs a port. wondering if anyone actually uses it though Apr 05 12:45:50 rburton, it was big back in the day.. I have not heard anything lately Apr 05 12:46:08 right Apr 05 12:46:33 secret plan is to shunt iso generation into meta-filesystems with the rest of the deps :) Apr 05 12:46:35 Hi guys, I'm trying to build this: https://github.com/fluent/fluent-bit/blob/master/fluent-bit_git.bb but bitbake (from rocko) throws this at me: fluent-bit-1.1-git-r0 do_package: SYSTEMD_SERVICE_fluent-bit value td-agent-bit.service does not exist . Any ideas? Apr 05 12:46:37 I would go for ti Apr 05 12:47:21 rburton, put it in meta-gplv2 Apr 05 12:47:48 armpit: xorriso is nicely licensed Apr 05 12:47:55 and has a cdrecord-compat mode Apr 05 12:48:17 rburton, I meant hide it in meta-gplv2 Apr 05 12:48:33 lol Apr 05 12:48:38 nobody will find it there! Apr 05 12:48:50 exactly Apr 05 12:51:39 * armpit great getting fumigated.. it killing the beer flavor Apr 05 13:14:18 RP: could you get me a link to that failure? Apr 05 13:25:15 Hi everyone ! Apr 05 13:26:42 RP: nevermind found it Apr 05 13:32:09 sveinse: the service file doesn't exist Apr 05 13:32:58 rburton: ok. Upstream inconsistencies then. Thanks. Apr 05 13:59:07 Has anyone experimented with using yocto/wic to generate images for cloud execution, AWS, Azure or similar? Apr 05 14:01:45 sveinse: not a lot to experiment to be honest Apr 05 14:02:09 should be pretty trivial to do the aws integration pieces too Apr 05 14:02:21 there's an image-container class to help Apr 05 14:02:32 I come to you because I've a problem with zlib in ruby ! Someone can help me with the integration of a ruby's gem (zlib here) ? Apr 05 14:02:56 rburton: is there any for azure too? Apr 05 14:05:10 rburton, the python3 test timeout issue seems bisectable Apr 05 14:05:22 it worked fine when python 3.7.2 was just introduced Apr 05 14:05:41 so I am trying to get down to a single commit now Apr 05 14:25:10 anyone using cmake bbclass from allarch recipes? seems to be broken for a while Variable OECMAKE_C_COMPILER value changed from 'arm-webos-linux-gnueabi-gcc' to 'aarch64-webos-linux-gcc' Apr 05 14:25:23 the same with OECMAKE_CXX_COMPILER Apr 05 14:39:08 +1 for xorriso Apr 05 14:47:57 hello there. i just started using yocto and its a brilliant project. but can anyone tell me how to change sysv to systemd ? Apr 05 14:49:00 lastaid: You can find the init manager documentation here: Apr 05 14:49:01 https://www.yoctoproject.org/docs/latest/mega-manual/mega-manual.html#selecting-an-initialization-manager Apr 05 14:52:09 thank you so much! Apr 05 14:52:29 can i also add single package as distro features? Apr 05 14:57:28 lastaid: When you want to add a package to an image you usually add the package to IMAGE_INSTALL Apr 05 14:58:21 Something like IMAGE_INSTALL_append = " pkg" (notice the extra space just after the quote) Apr 05 14:59:30 i am currently configuring my local.conf, i guess normally i would modify or create a bitbake recipe? Apr 05 15:00:50 lastaid, normally, you would create your own image recipe, and place exactly what you want in it. It's easier to start with an existing image, and modify that. Apr 05 15:01:14 yeah. but i have to say i am impressed Apr 05 15:01:21 yocto is solid stuff! Apr 05 15:01:47 yes it is. It is also frequently misunderstood, and considered to be 'complicated to use'. Apr 05 15:02:16 i though so as well. i build a kernel once, but not an entire distribution. Apr 05 15:02:53 yeah, it's a distribution toolkit aimed at embedded linux use cases. Apr 05 15:05:53 it does beat fiddling about with the official raspbian image Apr 05 15:27:49 New news from stackoverflow: Yocto recipe fails to install npm package Apr 05 16:17:32 * RP wakes up Apr 05 16:22:00 lol which continent are you in this week? Apr 05 16:24:16 fray: on irc, not physically :) Apr 05 16:24:58 lol Apr 05 16:48:16 How does IMAGE_TYPES and IMAGE_FSTYPES relate to CONVERSIONTYPES? Is is how a suffix is handled? Can it be combined arbitrarely? Is it the IMAGE_TYPES in image_types.bbclass which ultimately define all the allowable combinations? Apr 05 16:49:16 Specifically, I'm figuring out how to generate a vmdk image file. Not sure if I need to go via a IMAGE_FSTYPE wic or a "container" (whatever that is) Apr 05 16:50:19 sveinse: wic.vmdk in IMAGE_FSTYPES Apr 05 16:50:53 with recent Yocto releases Apr 05 16:51:41 JaMa: right, so my first assuption is correct? IMAGE_FSTYPE="type.conversion" where conversion is from the available CONVERSIONTYPES Apr 05 16:52:02 before rocko it was just vmdk in IMAGE_FSTYPES Apr 05 16:52:15 yeah, I'm at rocko Apr 05 16:58:24 Anyways, "wic.wmdk" was the way to go. Perfect, thanks JaMa Apr 05 16:59:03 Fired straight up in VirtualBox. Cloud VM next Apr 05 17:54:56 RP: I ran ptest manually inside the image with the util-linux patches that I sent and the "core-image-sato-sdk-ptest: Fix free space issues causing test failures" patch and it worked \o/ Apr 05 17:55:05 I'll try with testimage now Apr 05 18:34:29 Hey all, this is a simple question but it's eluding me Apr 05 18:35:07 I set up yocto on my machine so I could use bitbake to check for all upstream versions of all packages Apr 05 18:35:13 Works great on my computer Apr 05 18:36:06 I'm ready to push it to our production server, but when I copied the directory over and sourced oe-init-build-env it still has all the absolute paths from my local machine Apr 05 18:36:27 So any time I run devtool it's still looking for the path from my dev machine Apr 05 18:36:37 How do I re-initialize that path? Apr 05 18:39:54 nvm, found it: build/conf/bblayers.conf Apr 05 18:52:35 Hmm. Seems that wasn't the only place it was keeping that path... Apr 05 18:58:39 Error, TMPDIR has changed location. You need to either move it back to /home/timesys/poky/build/tmp or rebuild Apr 05 18:58:59 I like the "or rebuild" option, but "bitbake clean" fails with the above error Apr 05 18:59:20 ls Apr 05 19:24:17 armpit2, rburton_ RP, ready to build on the AB again. Apr 05 19:27:58 RP, rburton_: Jason looked at my gobject-introspection build failure mystery and cracked the case Apr 05 19:29:34 I was hitting a file handle limit in my $HOME which was on a netapp with a low quota Apr 05 19:30:20 the issue only started popping up now as the latest GO uprev turned on its cache (which is mandatory now) Apr 05 19:30:53 so between GO using $HOME/cache and g-ir-scanner I was hitting the quota Apr 05 19:32:07 I already had plans to update GO to not use $HOME, since at the moment I can't say if the cache is organized enough to avoid host contamination Apr 05 19:32:57 so I was going to move the cache to the recipe's go-tmp, or create a go-cache or something similar Apr 05 19:36:55 anyway, just thought I would post an update, I hate to leave things dangling and this should wrap it up Apr 05 19:57:05 marka: ah neat. yeah moving the cache seems like a wise idea. Apr 05 19:57:30 Striking7: just delete tmp Apr 05 19:57:44 Thank you rburton_ Apr 05 19:57:52 maybe we should make that message clearer Apr 05 19:58:09 rburton_: "get your hands off my $HOME" was uttered in the office at least once Apr 05 19:59:08 Yeah, a clearer message would be helpful Apr 05 20:00:10 Striking7: patch on the list Apr 05 20:00:15 basically "delete and rebuild" Apr 05 20:00:31 tmp is entirely generated so it can be deleted at any point and restored from sstate-cache Apr 05 20:09:50 Good to know Apr 05 20:10:09 Thanks again rburton_ . You've saved me a lot of time Apr 05 20:28:46 New news from stackoverflow: Yocto PR Service do not increment ${PR} Apr 05 20:48:17 marka: aha, sounds like a good move to move it somewhere in TMPDIR Apr 05 20:49:09 yup, hopefully you'll see some patches shortly Apr 05 20:49:12 justanotherboy: great! I'm going to look at the patches tomorrow, have a migraine that isn't helping concentration tonight Apr 05 20:49:50 Thanks RP Apr 05 21:18:27 RP: there ? we need https://patchwork.openembedded.org/patch/160125/ for warrior+master Apr 05 21:19:10 its fine to pull it after release for warrior but will make clang SDKs work better on target Apr 05 21:19:41 building chromium takes 22 hrs on rpi :) Apr 05 21:20:06 cross compilation is still relevant Apr 05 21:21:38 RP: (lttng-tools) weird, i cannot reproduce the failure locally with poky master-next and the same image (sato-sdk-ptest), let me know if if sometime it passes, might be a problem on our end. Thanks khem and RP for all the feedback and help. Apr 05 21:22:22 psrcode: which targets did you try Apr 05 21:22:49 qemux86_64 i think, was it on something else? Apr 05 21:23:12 you might want to try other arches too Apr 05 21:23:24 maybe qemuarm64 Apr 05 21:24:29 khem: based on the testresult RP linked it was on a qemux86-64 Apr 05 21:26:01 is there mechanism in place to receive email on failure for particular recipe for the autobuilder? Apr 05 21:35:44 psrcode: I see, Apr 05 21:36:16 psrcode: I do not know how to get notifications automatically, but halstead or RP might be able to help Apr 05 21:36:25 might be a matter of disk space, this test does produce more trace data than the rest of the test suite Apr 05 21:38:42 I might be interested in getting those notification for the lttng-* project recipe and babeltrace also. Given the wide array of configuration OE build against we would be foolish not to leverage it. Apr 05 21:38:47 would be quite nice Apr 05 21:56:48 sure enough, I dont know if builders have disk issues, but its a good point to check with RP Apr 05 22:06:10 psrcode, Disk space should not be an issue as far as I know. I can't speak to ease of adding notifications right now. Apr 05 22:07:13 halstead: okai will look elsewhere then, as for the notification I can do something on my own, will need to check how to scrape results. thanks **** ENDING LOGGING AT Sat Apr 06 02:59:57 2019