**** BEGIN LOGGING AT Tue Oct 31 03:00:02 2017 Oct 31 08:54:41 hi, anyone experienced ssh_dispatch_run_fatal incorrect signature error in openssh recently ? Oct 31 08:58:37 nayfe: could it be related to the hard coded >2048bit RSA requirements? i remember getting signature errors on a 1024 bit keys, but not the exact same error Oct 31 08:59:00 >=2048* Oct 31 09:03:50 nrossi> logs are not helping :/ Oct 31 09:09:37 seems to be with rocko migration, maybe last 7.5p1 version Oct 31 10:30:26 during yocto build my ubuntu system logout without any reason, can anyone else ever saw this problem? Oct 31 10:32:47 usually because you're running out of RAM and the kernel is dumb and kills X instead of gcc Oct 31 10:33:40 rburton, i've 16Gb RAM 4 Gb was being used, Oct 31 10:34:05 read the dmesg or other system logs Oct 31 10:34:11 how to resume the process Oct 31 10:34:29 what process? Oct 31 10:34:38 build process? Oct 31 10:34:50 run the same bitbake command again Oct 31 10:37:14 rburton, so will it start from beginning? Oct 31 10:37:19 no Oct 31 10:38:38 rburton, looking at the dmesg Oct 31 10:38:40 [57872.961544] traps: xchat-gnome[2396] trap stack segment ip:7fbf50b9e858 sp:7ffc50b22830 error:0 Oct 31 10:38:41 [57872.961547] in libgobject-2.0.so.0.4800.2[7fbf50b6c000+52000] Oct 31 10:38:41 [63937.163511] perf: interrupt took too long (2505 > 2500), lowering kernel.perf_event_max_sample_rate to 79750 Oct 31 10:38:42 [64647.189461] perf: interrupt took too long (3132 > 3131), lowering kernel.perf_event_max_sample_rate to 63750 Oct 31 10:38:45 [69415.445224] perf: interrupt took too long (3919 > 3915), lowering kernel.perf_event_max_sample_rate to 51000 Oct 31 10:39:30 anything suspicious ? my CPU fans are still throttling Oct 31 10:39:54 that's just xchat crashing, presumably as the session died Oct 31 10:41:31 rburton, or should I expect the build can corrupted due to this interrupt? Oct 31 10:41:36 no Oct 31 10:41:38 it will be fine Oct 31 10:41:52 rburton, great ! Oct 31 10:45:24 rburton, bitbake command not found ?? Oct 31 10:45:50 sounds like you forgot to run . oe-init-build-env Oct 31 10:46:21 rburton, so I've to run this command again, I ran that before crash Oct 31 10:46:53 you always need to run that to setup the environment variables like $PATH Oct 31 10:47:07 RP1: master and master-next have diverged, what one should i rebase onto? Oct 31 10:47:35 master-sort-of-next Oct 31 10:48:22 rburton, command, right? Oct 31 10:48:32 actually this is my first yocto build Oct 31 10:48:48 . is more portable than source, but yes Oct 31 10:49:02 What are the rules to what packages are accepted and included in meta-oe? I mean, if I find something (zbackup to be specific), is contributions welcome? Oct 31 10:51:12 rburton, thanks, its resumed now Oct 31 10:52:04 sveinse: yes. in general if there's a better layer put it there, but meta-oe is the "everything else" layer Oct 31 10:52:59 rburton: repushed -next Oct 31 11:16:15 system is being crash during yocto build, how can I find the reason? Oct 31 11:50:10 i'm trying to use devtool deploy-target and it says "No files to deploy", recipe only do something like do_install(){ install -d -m 0777 ${D}${bindir} install -m 0755 mybinary ${D}${bindir}/mybinary } any reason ? Oct 31 11:53:51 hello, is there a way to disable update-alternatives postinst generation? Oct 31 13:25:15 ericb: don't use update-alternatives? the postinst generation is kind of the point. Oct 31 13:27:57 New news from stackoverflow: [yocto-Jethro]Neet to compile Oct 31 13:34:34 bluelightning: sorry was busy Oct 31 13:35:48 where can I enable or isntall the ps command to be used during bitbake? Oct 31 13:37:17 TurBoss[m]: a good question is why on earth does configure want ps as that is very odd. what package is this? add ps to HOSTTOOLS to get it in the sysroot. Oct 31 13:37:41 Is linuxcnc Oct 31 13:37:51 i'm building a recipe for it Oct 31 13:38:20 rburton: this goes on the recipe or the local conf? Oct 31 13:39:48 ah see, i said check *why* Oct 31 13:40:11 linuxcnc is officially really stupid btw Oct 31 13:40:22 https://github.com/LinuxCNC/linuxcnc/blob/master/src/configure.ac#L820 Oct 31 13:40:28 it looks for a load of binaries Oct 31 13:40:48 you'll need to figure out if each one of those is used at *build time* or embedded in a package for *runtime* Oct 31 13:41:05 ah Oct 31 13:41:07 oh Oct 31 13:41:25 I see Oct 31 13:41:36 if runtime then you'll need to tell it where the binaries will end up, by passing ac_cv_path_PS="${bindir}/ps" etc Oct 31 13:42:07 as where the binaries are on your build host may not (with hosttools, definitely will not match) where they are on the target Oct 31 13:42:23 if its failing here i foresee more cross-compilation problems, so good luck Oct 31 13:42:37 ty Oct 31 14:11:57 <_taw_> in my siginfo: 'AUTOREV value is ${@bb.fetch2.get_autorev(d)}' Oct 31 14:12:51 <_taw_> my _git.bb recipe is never rebuilt although the top SRCREV changed, probably since the siginfo stores AUTOREV as the script not the actual SRCREV value, how can I fix? Oct 31 14:14:03 _taw_: did you set SRCREV to ${AUTOREV}? Oct 31 14:15:33 <_taw_> https://pastebin.com/84FMh2GX -- I followed the mega manual Oct 31 14:16:39 <_taw_> I'm using yocto 2.0.2 this might be fixed with the latest version, if so I would propose to upgrade to the team Oct 31 14:19:53 <_taw_> I want to use default_preference=-1 on devel feed that will follow master branch, but somehow the fetch task never triggers again. although master of the project changed. http://www.yoctoproject.org/docs/2.2/mega-manual/mega-manual.html#var-DEFAULT_PREFERENCE Oct 31 14:25:41 How is wic able to populate the file system without being root? I understand how the mkfs is done and how the partitions is assembled, but I cannot seem to find how wic is able to mount the fs without using a loopback device and hence need root. Is this something pseudo emulates? Oct 31 14:33:16 sveinse: from a quick look, it appears to copy the partition image (eg the .ext4) into the partition image and then expand it if required Oct 31 14:33:29 engine.py, write() Oct 31 14:34:04 the mkfs is actually pre-wic Oct 31 14:42:29 rburton: Yes, It creates the filesystem in a unprivileged file using mkfs, and when everything is done it copies this file into the final partitioned image. But the middle part: how it mounts this file and populates the rootfs is the mystery I'm curious about Oct 31 14:42:46 why would it need to mount it? Oct 31 14:43:05 creates a partition image on disk, copies bits from ext4 image into the right bit of the partition image Oct 31 14:43:31 oh you mean the ext4 generation? Oct 31 14:43:32 rburton: how else do you populate the extfs image with data? Oct 31 14:43:41 rburton: yes Oct 31 14:43:43 well that's nothing to do with wic :) Oct 31 14:45:05 sveinse: man mkfs.ext4, look up what -d does Oct 31 14:50:51 i'm trying to create wakaama client recipe (LWM2M), without success https://pastebin.com/4YzB2uH1 if someone can help? DTLS option tries to build tinydtls submodule, and it fails with "configure: error: cannot run C compiled programs." Oct 31 14:52:25 rburton: -d is neither documented or a recoginized option to mkfs.ext4 on my system. Aha, right, yocto is relying on newer mkfs.* host tools Oct 31 14:52:55 now it makes sense, thanks Oct 31 15:53:55 well i surrender, i'll remove dtls support for now :x Oct 31 16:20:41 What is the linaro toolchain compared to the stock toolchain in yocto? Why does some platforms rely on it, such as the Intel FPGA SoC systems? Oct 31 17:11:23 does pyro have any restrictions to minimal gcc version? Oct 31 17:58:44 New news from stackoverflow: List packages that were updated with the new poky version? Oct 31 19:54:07 sveinse: yes, we build our own e2fstools to ensure that we get -d (which originally was a patch we wrote, iirc) Oct 31 20:02:57 rburton: brilliant, thank you Oct 31 20:05:02 I want to make my own distro (to avoid over-complex local.conf) and require poky.conf, how can I come around constructs like these: DISTRO_EXTRA_RDEPENDS += " ${POKY_DEFAULT_EXTRA_RDEPENDS}". First I want my own MYDISTRO_DEFAULT_EXTRA_RDEPENDS, secondly it is +='ed in poky.conf, so I cannot really come around that can I? Oct 31 20:06:30 sveinse: don't require poky.conf, just copy/paste/edit as required Oct 31 20:07:05 also pay attention to the number of lines in poky.conf which are just assigning the defaults anyway Oct 31 20:09:15 rburton: ok, the require statements at the bottom of poky.conf, what significanse have they? Oct 31 20:09:41 rburton: I hope we've reduced those to a minimum by now Oct 31 20:10:25 sveinse: world build excludes, disabling static libraries, turning on uninative Oct 31 20:11:09 what is uninative? Oct 31 20:11:33 sveinse: https://wiki.yoctoproject.org/wiki/What%27s_New_In_2.1 Oct 31 20:13:40 so generally speaking disabling static libs and turning on uninative is a good thing in a custom distro then? Oct 31 20:14:35 your choice! Oct 31 20:14:47 do you want static libs? poky decided no, as it saves build time. Oct 31 20:15:08 uninative increases sstate reuse which is good when you've an autobuilder cluster with eight different distros Oct 31 20:15:18 (as uninative makes them all the same) Oct 31 20:15:23 again, good for poky. Oct 31 20:15:32 rburton: yes, seems smart Oct 31 20:15:42 if you don't then its just overhead you don't need Oct 31 20:16:03 if you ever want to link statically then no-static-libs is obviously a bad idea Oct 31 20:18:33 rburton: I'm curious to that, does no-static-libs enforces that no package can ever link statically? Or is it just that it stops generating .a and .la for -dev pacakges? Oct 31 20:18:44 .la are not static libs Oct 31 20:18:56 it does its best to stop packages building static libraries Oct 31 20:18:59 ah, yes, sorry, its late Oct 31 20:19:19 nothing is forced and you can change it for specific recipes Oct 31 20:19:25 right Oct 31 20:19:25 just have a look at the inc, it's not magic :) Oct 31 20:26:51 The DISTRO_FEATURE multiarch, this is for the target system and not for the host? So it wouldn't make too much sense on an arm system? Oct 31 20:29:12 New news from stackoverflow: Installing PHP Composer on Yocto/Arm Beaglebone Oct 31 23:08:05 I find something really interesting: I'm creating a distro which is among other is setting some PREFERRED_PROVIDER_xxx for a machine layer. The machine layer use ?= to set some default, and I'd like to the same from the distro, so I'd also use ?=. But that clashes, and the BSP default takes precendence. How do resolve these? Oct 31 23:09:21 What I need is some stronger assignment than ?=, like ??= is weaker. Oct 31 23:09:56 (granted that these weak, weaker, strong, stronger, tend to make thing extremely complicated) Nov 01 02:29:57 New news from stackoverflow: How to change the config of u-boot in Yocto Nov 01 02:46:14 I'm trying to create a source mirror of my packages however I have one particular package that contains a git submodule. My issue is that everytime yocto tries to fetch the package (which contains the submodule), it attempts to access the git repo over the network even though it has already been previously fetched. ( It's already located in my downloads folder) My goal is to be able to do a complete build entire Nov 01 02:46:14 offline. Any ideas on what I can do? Nov 01 02:58:36 Is it normal for toaster to be stuck at "Parsing 0% complete" for several minutes when building recipes? **** ENDING LOGGING AT Wed Nov 01 03:00:01 2017