**** BEGIN LOGGING AT Fri Nov 03 03:00:02 2017 Nov 03 03:43:00 If my os did an NTP sync what log would that be in/ Nov 03 03:43:00 ? Nov 03 07:51:45 hi guys Nov 03 07:52:08 how can i force virtual/kernel to produce vmlinux again Nov 03 07:52:32 ? Nov 03 09:25:21 Costin1: clean it and then rerun it? Nov 03 09:25:45 Costin1: or do you mean to recompile? Nov 03 09:38:17 New news from stackoverflow: error while compiling my own recipe in yocto (pyro) Nov 03 10:52:00 I've got a bbappend in my distro layer which adds a task. Is it possible to make addtask conditional on the value of DISTRO? Trying to make my layer YP compatible Nov 03 10:52:30 I've looked at the mega-manual and the bitbake manual. Can't see anything about combining overrides and addtask Nov 03 10:52:33 DISTRO override? Nov 03 10:53:02 oh, addtask Nov 03 10:56:45 joshuagl: Yea, I've figured everything else out, just not how to make addtask distro-specific Nov 03 10:57:46 empty task with a distro override for your distro? Nov 03 10:57:59 on a call atm but I'm sure there's an example somewhere Nov 03 10:58:23 you can add tasks from python Nov 03 10:59:23 joshuagl: has your kvm patch been through a run on the ab? Nov 03 10:59:37 rburton: nope Nov 03 11:00:18 lets see what happens! :) Nov 03 11:00:23 \o/ Nov 03 11:01:04 rburton: I can see bb.build.addtask() used in image.bbclass, I'll throw together an anonymous python function to check DISTRO and call that as needed Nov 03 11:01:12 was hoping for a simpler solution but that should work Nov 03 11:02:58 joshuagl: there's not a way to run bitbake-selftest without triggering nightly on the ab right? Nov 03 11:08:31 New news from stackoverflow: Add lib to SDK generated from meta-toolchain-qt5 with Yocto Nov 03 11:19:24 rburton: not atm Nov 03 11:43:10 Now I'm a bit lost. `do_deploy_oryx() {...}` is resulting in a function named do_deploy_oryx instead of do_deploy, even though "oryx" is in OVERRIDES Nov 03 11:44:22 Well, the function name doesn't change because of overrides. What gets executed should change, though, i.e. do_deploy_oryx should be run instead of the normal do_deploy Nov 03 11:45:12 But isn't do_deploy empty by default anyway and your recipe should just implement it? Nov 03 11:48:11 neverpanic: I'm adding a do_deploy task in a bbappend in my distro layer, trying to make it only take effect if DISTRO is actually set to my distro Nov 03 11:52:20 Ok, it works if there's an empty do_deploy function defined, but not if do_deploy is undefined Nov 03 11:54:09 That's expected, I think. Functions in Yocto are really just variables with a flag set. d.getVar('FOO') probably also doesn't give you FOO_override if FOO isn't set at all. Nov 03 11:55:25 Hmmmm. I'm trying to make my distro layer YP-compatible which means the changes in my bbappend should only take effect if DISTRO is set to the distro defined in the layer Nov 03 11:55:33 I can't just go forcing do_deploy to some value Nov 03 11:56:04 trying to do my good deed and be YP-compatible after the discussions at the dev meeting in Prague Nov 03 11:58:08 rburton: RP and I just had a chat and decided that we should drop f25 from the clusters (and the supported distro list). Please ignore my runqemu patch. Nov 03 11:58:11 paulbarker: no good deed goes unpunished, you should'be known Nov 03 11:59:21 paulbarker: I know there have been discussions about adding syntactic sugar to improve the ease of compat, I'm still distracted but hoping to circle back on this herte Nov 03 11:59:37 LetoThe2nd: Ah the root cause is my intent not the code haha Nov 03 11:59:43 paulbarker: I think the problem is more about adding the task conditionally rather than the function name Nov 03 11:59:44 paulbarker: pohly1 made the refkit layers all yocto compatible, might be some inspiration in there? Nov 03 12:00:04 paulbarker: there is a bb.build.addtask() function you could call conditionally from anon python iirc Nov 03 12:00:22 I've got the addtask conditional by using anonymous python Nov 03 12:00:53 paulbarker: but it only works if do_deploy is set even if its empty? :/ Nov 03 12:01:02 that sounds odd :( Nov 03 12:01:03 Also got "inherit deploy" conditional using base_conditional Nov 03 12:01:05 but it's now not finding the do_deploy function when my distro is selected Nov 03 12:01:25 paulbarker: conditional inherits are very error prone :( Nov 03 12:01:28 I may just be crazy trying to add do_deploy in a bbappend Nov 03 12:02:11 paulbarker: The trouble is the inherit is effectively an immediate expansion operation and the system may or may not be ready for that Nov 03 12:02:53 I think that's part of the issue. There's also some confusion as do_deploy is an sstate task Nov 03 12:10:52 Heyho, I try to install packages via opkg on my yocto. However, I always get the error: "nothing provides libc". I have installed glibc but that didnt help. Any ideas? Nov 03 12:11:16 The error happens since pyro Nov 03 12:31:44 FabKna: the package is probably called libc6 ? Nov 03 12:31:55 RP: the OE-Core seems frozen for a while; when 2.5 patches will start to be merged? I want to work on few more updates but want to avoid duplicating work Nov 03 12:32:26 otavio: the autobuilder hasn't been giving stable results so we're kind of stalled. I also have some "distractions" at the moment :( Nov 03 12:32:53 :-( Nov 03 12:32:53 * LetoThe2nd does his very best to distract RP Nov 03 12:33:10 RP: and there is a 'next' branch somewhere ? Nov 03 12:33:21 so I could look and see if I reproduce something? Nov 03 12:33:41 RP: Thanks, will try that. Nov 03 12:34:09 otavio: ross/mut is what rburton has been working on Nov 03 12:34:36 otavio: https://autobuilder.yocto.io/tgrid?length=20 - note no green builds for ages :( Nov 03 12:34:46 it is not being in master-next anymore? Nov 03 12:34:56 otavio: it hasn't made it to -next Nov 03 12:35:13 otavio: I've been letting ross try and get a patchset that works, no point in us both doing it Nov 03 12:35:36 Ok; gotcha Nov 03 12:43:02 RP: I cant find a package called libc6. Do you know how to install it? Nov 03 12:44:22 FabKna: I find it hard to believe you don't have libc installed already, unless you're using musl? Nov 03 12:47:30 RP: shouldn't that fill in? Nov 03 12:50:16 LetoThe2nd: fill in? Nov 03 12:51:15 RP: shouldn't musl technically provide libc for a given distribution? or am i misguided here? Nov 03 12:51:46 LetoThe2nd: it is not the same Nov 03 12:51:52 LetoThe2nd: right, yes, but it would not necessarily be called libc6 Nov 03 12:51:56 FabKna, it looks like your opkg doesn't find its installed packages registry Nov 03 12:52:08 (or yes, libc is renamed) Nov 03 12:52:11 LetoThe2nd: I think it requires rebuild Nov 03 12:52:27 otavio: of course it does Nov 03 12:52:50 i'm just speaking about the package management dependencies Nov 03 12:52:52 LetoThe2nd: it can provide a virtual/libc or something Nov 03 12:53:22 luneff: Do you know how to fix this? Nov 03 12:53:28 LetoThe2nd: and runtime dependencies, after rebuld, will be catch with shlib Nov 03 12:53:43 rburton, sgw: What happened to the ppc random number gen issue? Nov 03 12:54:00 LetoThe2nd: so PROVIDES, yes; RPROVIDES, no Nov 03 12:54:25 LetoThe2nd: it is not a drop-in Nov 03 12:54:55 otavio: it sounds like mixing distro builds then. i mean, given a coherent build, there should be no package depending on a form of libc that is not provided by musl. Nov 03 12:55:03 at least thats my understanding Nov 03 12:55:42 LetoThe2nd: yes and if there is, it is a bug ... or a binary package which was not rebuild Nov 03 12:56:17 otavio: right. thats what i thnk too. Nov 03 13:04:31 otavio: what updates you're thinking of? Nov 03 13:18:30 kanavin: I want to upgrade cmake again and will go over the packages we use most to check for new releases Nov 03 13:18:41 kanavin: we have few patches waiting already Nov 03 13:20:47 otavio: do you know about 'bitbake -c checkpkg world'? I just ran it to see what your recipes in need of update are. Nov 03 13:21:01 otavio: linux-firmware, cmake, go-dep, libdrm Nov 03 13:25:40 rburton: coming back to your earlier comment, what's the value of running bitbake-selftest on the AB outside of the nightly? i.e. should we add it somewhere else too? Nov 03 13:35:53 kanavin: I didn't know about it Nov 03 13:36:37 otavio: that's how http://recipes.yoctoproject.org/rrs/recipes/2.4/M4/ works, but don't trust the website too much Nov 03 13:36:56 otavio: the website has been broken or out of date more often than it's been working, so run the check locally Nov 03 15:09:15 New news from stackoverflow: Can't generate SDK for Windows with Yocto Nov 03 15:19:03 * fray answers the stackoverlfow question.. Nov 03 15:19:11 wrong version of meta-mingw (branch) for the oe-core they are using Nov 03 15:29:10 kanavin: where does checkpkg outputs? Nov 03 15:29:30 otavio: tmp/log/checkpkg.csv Nov 03 15:30:11 * RP puts "overhaul logging" onto the todo list :/ Nov 03 15:30:58 RP: that file in particular should be in json format Nov 03 15:31:04 would be nice if checkpkg would output an INFO ours v1.0 theirs v1.1 or similar Nov 03 15:31:13 it's currently in csv, which is rather awful to read Nov 03 15:31:16 moto-timo: it does Nov 03 15:31:26 to the console, not in the log Nov 03 15:31:41 kanavin: that is one of I suspect many issues Nov 03 15:32:00 if i run chkpkg on one recipe, I shouldn't have to go find the log file just to see the result Nov 03 15:32:17 my 2 cents Nov 03 15:32:41 moto-timo: no disagreement, just a sad lack of time issue :/ Nov 03 15:32:54 if someone wanted to file a bug or bugs... Nov 03 15:33:04 might at least remind us it could be improved Nov 03 15:33:15 RP: agreed. time is a big problem. Nov 03 15:33:30 * moto-timo needs to file more bugs Nov 03 15:33:57 * moto-timo thinks time problem is going to get worse before better Nov 03 15:34:23 it never gets better Nov 03 15:35:31 reminds me of a well-known saying in cycling: it never gets easier, you just ride faster Nov 03 15:35:38 lol Nov 03 15:36:01 that moment when your legs stop hurting as much so you ride faster Nov 03 15:36:07 ah, it's from Greg LeMond Nov 03 15:41:09 I want to use the SPI3 on my board, but problem is my board is custom made and was not configured for SPI3. The reference board use SPI1. How do I get SPI3 to work? Nov 03 16:13:25 Greetings! I am getting the following when running bitbake....I haven't seen these before, and haven't been able to find a workaround so far. Can anyone point me in the right direction? Nov 03 16:15:59 Hm....errors not posting. Nov 03 16:16:32 "/usr/bin/ld: Cannot find -lz" Nov 03 16:17:03 Also "/usr/bin/ld: Cannot find -lcrypto" and "Cannot find -lssl" Nov 03 16:21:02 JimH: what did you change? Upgraded to a new version? Nov 03 16:21:33 JimH: at a guess it sounds like a missing zlib-native and openssl-native dependency Nov 03 16:33:52 Do we have any RNG knowledgeable people here? Nov 03 16:33:58 I'm wondering about http://git.yoctoproject.org/cgit.cgi/poky-contrib/commit/?h=rpurdie/wip9&id=2bc28ebdbe972e820153441694259c0f8fced5cf Nov 03 16:39:17 RP, I enabled utilities like 'vim', 'lsof', and 'screen'. That was about it. Nov 03 16:41:29 RP: https://wiki.archlinux.org/index.php/Rng-tools seems to agree, linking to https://lwn.net/Articles/525459/ which has more details (section "Administrator recommendations") Nov 03 16:43:04 RP: I agree with this fwiw, why on god's green earth would you do that - the kernel already uses "good" entropy to salt the rng Nov 03 16:43:29 looks like we agree, ship it! Nov 03 16:43:55 although I believe rng-tools has the ability to use CPU entropy instructions if they exist too Nov 03 16:44:06 not sure if that needs you to /not/ specify a hwrng device Nov 03 16:44:24 also: if you have a PC from like,the past 10 years, modprobe tpm_rng Nov 03 16:44:28 and you have a /dev/hwrng that works Nov 03 16:45:30 why the hell doesn't that get modprobed magically Nov 03 16:45:36 * rburton adds to initscript Nov 03 16:50:53 ramcq: in this case we need to ensure virtual machines don't hang waiting for entropy... Nov 03 17:06:02 RP: rburton: F25 worker has been disabled Nov 03 17:06:08 armpit: ^^ Nov 03 17:06:50 joshuagl, k. thanks Nov 03 17:07:22 issues left to resolve, but that's a big one worked around Nov 03 17:07:32 * joshuagl goes for dinner Nov 03 17:31:04 RP: i'll pick that commit and re-fire a mut without f25 Nov 03 17:31:21 rburton: sounds good, thanks Nov 03 18:34:00 Hi Nov 03 18:36:06 I am a bit confused. In buildroot the opkg packge references http://git.yoctoproject.org/cgit/cgit.cgi/opkg/about/. Where to get support for opkg apart from the mailing list Nov 03 18:51:17 ski7777: the mailing list, or here i guess Nov 03 18:51:41 I was searching for here or another # Nov 03 18:51:45 Thank you Nov 03 18:56:15 Using github's raw function it is possible to proide files/tree structure just via a repository. Does opkg need the index of the webserver? Nov 03 18:57:04 If not it should be possible to provide an opkg repo on github? Nov 03 20:01:24 ... now I've done it. Cloned rocko, followed directions and built a kernel. It booted, too. Then I picked up a copy of Streif's book on the Yocto Project and everything I thought I knew became scrambled. Nov 03 20:03:08 Is there, somewhere on the net, (and I hate to write this) a "teaspoon-by-teaspoon" tutorial on the Yocto build process? Nov 03 20:06:45 For example: The target is an Intel 3825, one of their Atom devices. After cloning meta-intel into the poky directory, I go into build/conf/bblayers.conf and add $HOME/Yocto/poky/meta-intel to the list. That's it? That seems too simple. Nov 03 20:07:10 and set MACHINE="foo" Nov 03 20:07:17 where foo is machine you build for Nov 03 20:08:07 I set MACHINE="intel-corei7-64". Nov 03 20:08:16 That's in local.conf . Nov 03 20:08:23 Can somebody help me with opkg? Nov 03 20:09:05 so ideally you create an image and it works oon that hardware Nov 03 20:09:46 I would hope it would work on the hardware. It's still chugging along building linux-intel-4.9.56+stuff. Nov 03 20:10:05 But... isn't there something specific to the 3825 that I must set to get feature on that device? Nov 03 20:10:11 ..features.. Nov 03 20:12:31 need to ask the meta-intel guys Nov 03 20:12:46 generally the intel stuff tends to be pretty generic Nov 03 20:17:11 I started the current build using the -k switch in bitbake .. and there are a mess of warnings that it couldn't retrieve packages. Nov 03 20:18:20 Eventually bitbake gets to the end... and I've just been rerunning it. Is that safe? (Safe in the sense that another package has a dependency on the package that wasn't loaded .. and miscompiled because of it). Nov 03 20:18:35 Can somebody help me with opkg? Nov 03 20:44:30 Crofton|work: Thank you. You've settled down some of my confusion. Will probably be beating on this over the weekend (think I have an older Pentium-class computer I can resurrect for the cause. Nov 03 20:44:51 Uh. Will Yocto compile for a plain Pentium 4 (single thread) computer? Nov 03 20:47:00 Guess I'll find out over the weekend. A pleasant evening to all. Nov 03 21:11:31 updating to pyro, and now my image recipe is getting skipped because "username blah does not have a static ID defined. skipping it" Nov 03 21:14:34 seems to be related to a recent change in openembedded-core/meta/classes/useradd-staticids.bbclass Nov 03 21:18:36 EXTRA_USERS_PARAMS = "useradd -u 1000 -d /home/blah -r -U -s /bin/sh -P '' blah;" Nov 03 21:21:33 i'm guess i can't use -u -U at the same time? Nov 03 21:53:19 nathani_: sounds like you're driving staticids wrong Nov 03 22:04:50 Hmm I'm trying to readd a drive in my array after manually wiping its first and last 20Mbs, but I then got an error that the drive didnt have a proper GPT table. Ok. Then I ran zpool replace tank sdf sdf and then it sais that sdf is busy and then I get CHKSUM error in the pool. What am I doing wrong? Nov 03 22:05:47 Wrong channel, please ignore **** ENDING LOGGING AT Sat Nov 04 03:00:01 2017