**** BEGIN LOGGING AT Tue Feb 06 03:00:04 2018 Feb 06 03:00:28 SOURCE_MIRROR_FETCH was set.. Feb 06 03:01:28 manual has a warning. too bad the system doesn't Feb 06 03:21:45 * armpit_IST not enough sleep Feb 06 03:22:36 * armpit_IST wonders if this patch would be accepted Feb 06 03:22:39 - if not source_mirror_fetch: Feb 06 03:22:40 + if source_mirror_fetch: Feb 06 03:22:40 + bb.warn("You obviously did not read the manual\n SOURCE_MIRROR_FETCH is set and you might be shooting yourself in the foot\n - Armpit") Feb 06 03:22:40 + else: Feb 06 03:24:38 New news from stackoverflow: Yocto - Add Custom Driver Feb 06 07:25:24 New news from stackoverflow: Is there any way for building rt-kernel in yocto, other than using recipes-rt? Feb 06 10:10:32 Hi. I'm configuring Yocto images for 2 different systems: one will be a WiFi access point, the other will operate in station mode and connect to the AP. Feb 06 10:11:04 The Yocto build is based on the Variscite layer containing the machine configuration. This machine configuration includes both hostapd and wpa_supplicant. Feb 06 10:12:12 Should I create 2 new machine configurations based on the original configuration or can/should I change the MACHINE_EXTRA_RDEPENDS variable in my custom images? Feb 06 10:12:50 Of course I only want to install hostapd on the AP device and only wpa_supplicant on the STA device Feb 06 10:13:48 that sounds like you want two different images at least. Feb 06 10:15:53 LetoThe2nd: I have 2 different images. The problem is that my machine adds both hostapd and wpa_supplicant to the MACHINE_EXTRA_RDEPENDS: https://github.com/varigit/meta-variscite-fslc/blob/rocko/conf/machine/imx7-var-som.conf Feb 06 10:16:20 oh wait, it actualy happens at https://github.com/varigit/meta-variscite-fslc/blob/rocko/conf/machine/include/variscite.inc Feb 06 10:17:23 Can I change the MACHINE_EXTRA_RDEPENDS variable in my images and is this considered good practice for my use-case? Feb 06 10:18:22 Dear All, correct me if I am wrong, but it looks to me that the latest systemd recipe in openembedded-core does not allow disabling sysvinit compatibility: http://cgit.openembedded.org/openembedded-core/tree/meta/recipes-core/systemd/systemd_234.bb?h=master Feb 06 10:18:34 Is there any specific reason for that? Feb 06 10:19:34 seppe: nope, you can't. stuff thats set through conf files is available everywhere, whereas things in .bb files are recipe local. so if that is pulled in through some machine file, you basically have to either modify it, or derive and overwrite, in the form of a more specifc MACHINE Feb 06 10:20:55 i'd be fixing/wiping MACHINE_EXTRA_RDEPENDS in your distro conf and then adding the packages you want explicitly Feb 06 10:21:14 whilst moaning at your BSP provider that they should stop doing what they're doing Feb 06 10:21:23 LetoThe2nd: mmh, I see. Is there a thing like a machine confappend or do I need to copy the machine config and make my changes? Feb 06 10:21:25 that is the worst assignment for MACHINE_EXTRA_RDEPENDS i've ever seen Feb 06 10:22:14 seppe: nah, just create you machine conf and pull in the original at the top, then overwrite later. Feb 06 10:22:48 no need to create a new machine conf, just blast that assignment in your distro and set it to something sane Feb 06 10:22:59 seriously, moan at the BSP provider Feb 06 10:22:59 rburton: yeah, it's a typical case of "include ALL the features to show of our capabilities" iof making a nice configurable system. Feb 06 10:23:44 forced splashscreen, bluetooth, wifi stuff, all kernel modules, vnc server! Feb 06 10:23:45 idiots Feb 06 10:23:57 vnc server! vnc serve everything! Feb 06 10:25:02 rburton: ok, thats a way too. i wasn't sure about the eval order of machine.conf / distro.conf Feb 06 10:25:22 worst case, _forcevariable works Feb 06 10:25:52 anyhow, I'll do my moaning after I get something working. So your suggestion is to create 2 seperate distros where I change the MACHINE_EXTRA_RDEPENDS? Feb 06 10:26:38 no Feb 06 10:26:57 in your distrio trim that value down to what you actually need Feb 06 10:27:02 (ideally, almost nothing) Feb 06 10:27:06 then in your images, pull in the bits you want Feb 06 10:27:20 ah, makes sense. Feb 06 10:27:23 that variable is for pacakges that *every* image for the machine will need, like a bootloader Feb 06 10:27:27 or a specific kernel module Feb 06 10:27:46 not 'bluetooth stack and wifi AP daemon and a splash screen and a vnc server' Feb 06 10:28:13 i smell a conspiracy here. Feb 06 10:30:01 "lets ship all our stuff with vnc enabled and lets create an embedded botnet!" Feb 06 10:30:17 ha Feb 06 10:30:23 worst botnet ever Feb 06 10:31:13 having a bad botnet is better than having no botnet. Feb 06 10:32:55 ¯\_(ツ)_/¯ Feb 06 10:40:31 meta-botnet Feb 06 10:41:22 Crofton|work: sshhhh Feb 06 11:04:01 anyhow, I think I only really need imx-kobs, pm-utils rng-tools and maybe u-boot-fw-utils. Does anyone with experience with the Freescale i.mx platform have feedback on that? Feb 06 12:23:09 I'm having a problem because base-files /etc/fstab defaults to half the ram size limit for /var/volatile in oe-core, so I want to reduce it so logs don't consume so much memory, should I just use a custom fstab or does reducing var volatile to say 64m or 128m make sense in oe-core? Feb 06 12:24:01 I mean does reducing the var/volatile size limit make sense in master oe-core? Feb 06 12:24:23 or should I just keep that change downstream Feb 06 13:05:13 rburton: What would you say are the best methods for trimming the MACHINE_EXTRA_RDEPENDS value down? I tried just overwriting it, but this means stuff added from other machine layers will also disappear. Feb 06 13:05:55 preferably I would just like to remove the stuff I don't want, but I haven't figured out how to do this yet Feb 06 13:24:45 is there a way to tell bitbake to *not* try to go and fetch version information? Feb 06 13:25:10 e.g. I just want to run a 'bitbake -c cleansstate x'. Feb 06 13:42:26 seppe: _remove can selectively remove bits Feb 06 13:42:56 T_UNIX: presumably because you're using autorev? easy way is "rm" in the sstate-cache Feb 06 13:43:50 @rburton I'm trying to integrate bitbake with gitlab-runner. Feb 06 13:45:19 the runner gets passed secret temporary credentials allowing it to access stuff. I simply want to cleanup recipes that bitbake presumebly messed up :-/ Feb 06 13:56:45 New news from stackoverflow: Bitbake: proper way to "patch" a function from a recipe? Feb 06 14:05:29 Hi All Feb 06 14:06:24 I am building an image based on "core-image-minimal", for the machine "qemux86", with IMAGE_FEATURES += "package-management" Feb 06 14:07:06 however, on the target machine i getting command not found when trying to call smart command (smart --help) Feb 06 14:07:24 distsys, what branch/release are you building ? Feb 06 14:07:44 smart has been replaced by dnf in the recent releases. Feb 06 14:08:00 branch rocko Feb 06 14:08:35 dnf then. Feb 06 14:08:45 ah really !! Feb 06 14:08:50 assuming you using rpm of course. Feb 06 14:09:05 yes yes ... trying dnf ... Feb 06 14:09:15 at the same time, rpm5 is now rpm4, but unless you are really into the internals of things, that's invisible. Feb 06 14:09:40 (y) Feb 06 14:11:54 by the way, dnf is it keeping the same global syntax as smart ? Feb 06 14:34:28 Hi, I talk with a guy at fosdem , but I lost (or havent yet found) his card Feb 06 14:34:53 I asked about to procedure to co maintain a layer (meta-oic) Feb 06 14:35:13 I am sure this is documented somewhere I will RTFM Feb 06 14:36:24 as in, have commit access to? Feb 06 14:36:29 yes Feb 06 14:36:50 my patches were merged by maintainer Feb 06 14:37:02 but he seems too busy to keep doing it Feb 06 14:37:09 contact halstead (michael@yoctoproject.org) to request push access there, with the maintainer on CC to verify your request Feb 06 14:37:17 ok thx Feb 06 14:37:49 presuming that you've discussed this with the maintainer that is Feb 06 14:37:55 if you haven't then hey, its git, fork it Feb 06 14:46:11 yea I have fork but I dont want pple to use it Feb 06 14:46:37 I contacted halstead by email, let's wait and see I am not hurry Feb 06 14:46:42 thx for the tips anyway Feb 06 14:59:47 Hi, folks! I don't know why, but i got that my linux kernel builds before all the toolchains... And i got mipsisa64r6el-wrs-linux-gcc: command not found Feb 06 15:00:18 How can i force dependencies build before kernel build? Feb 06 15:01:17 i tried to remove all the files from build directory except downloads and conf directories Feb 06 15:01:27 and re-run image building Feb 06 15:11:58 i have no more questions. It was my own fault. Feb 06 15:12:13 i ruined DEPENDS Feb 06 15:12:23 have nice day, guys! Feb 06 15:58:17 YPTM armin is on Feb 06 15:58:36 * moto-timo joining soon Feb 06 15:58:39 YPTM: Stephano joined. Feb 06 15:58:51 YPTM: Alex joined Feb 06 15:58:53 YPTM: Maxin joined Feb 06 15:59:05 YPTM: Amanda joined Feb 06 15:59:24 YPTM: Stephen Joined Feb 06 15:59:43 YPTM: Richard joined Feb 06 16:00:06 YPTM: Pascal joined Feb 06 16:00:52 YPTM ross joining Feb 06 16:01:38 YPTM: Joshua Watt here Feb 06 16:01:47 YPTM: Martin, Oliver, Marcus (of Phoenix Contact, Germany) here, participating for the first time Feb 06 16:01:49 stephano: you're not stephen, right? Feb 06 16:02:29 YPTM wmills is here Feb 06 16:02:54 access code? it says the conference has ended with 88748961 Feb 06 16:03:06 YPTM: Denys joined Feb 06 16:03:11 moto-timo: worked for me with 88748961 Feb 06 16:03:26 * moto-timo shakes fist at skype Feb 06 16:04:13 negative Feb 06 16:05:01 moto-timo: here in .fi we're both using regular phones and dialling regular numbers here, and it worked fine Feb 06 16:05:12 moto-timo: https://www.yoctoproject.org/tools-resources/community/monthly-technical-call Feb 06 16:05:30 moto-timo, I had to call dial into: 1-800-262-0778/404-397-1527 Feb 06 16:09:05 status report is here http://lists.openembedded.org/pipermail/openembedded-core/2018-February/147215.html Feb 06 16:09:18 YPTM: https://wiki.yoctoproject.org/wiki/Weekly_Status Feb 06 16:11:01 spectre thread from last month is http://lists.openembedded.org/pipermail/openembedded-core/2018-January/146441.html Feb 06 16:11:09 * moto-timo in finally Feb 06 16:11:09 (read the entire thread) Feb 06 16:11:20 user error I guess. skype interface bad Feb 06 16:16:29 Hi All Feb 06 16:18:07 Is there a way to build a package individually ? I mean not by re-invoking bitbake image_name. Actually, i tried bitbake package_name, but this seems to just check the package metadata files (.bb) Feb 06 16:18:30 distsys: nope, you do rebuild recipes not packages. Feb 06 16:18:36 distsys: bitake Feb 06 16:18:42 images are just recipes, bitbake builds recipes Feb 06 16:18:51 okay Feb 06 16:18:58 you can't build a *package* as a package is the result of chopping up a recipe's output Feb 06 16:19:45 (some distros use package to refer to source and output forms, we use recipe for source and package for output) Feb 06 16:19:55 what i want actually is to have, say my rpm package, added to the list of rpm packages without rebuilding the image entirely ? Feb 06 16:21:22 so shall i suppose this not possible ? Feb 06 16:21:31 what do you mean by "list of rpm packages"? if you mean the package feed used to construct the image, there's a recipe to update it after you rebuild the recipe. if not, i have no idea what you're referring to Feb 06 16:22:08 yes i am referring to the package feed Feb 06 16:24:10 you mean, if i recall *bitbake image_recipe* again, after setting up my new recipe, it will just rebuild the recently added one, somehing like that ? Feb 06 16:26:46 distsys: do you want your newly added package to be included in the image at the time when the image gets built, or do you want it to be available via package feed and installed later using package management on target hardware? those are your two options Feb 06 16:27:10 say i am adding a new recipe, hello.bb, and then i am going to rebuild the image recipe *bitbake core-image-minimal*, will this rebuild only the new recipe and generate the rpm package for it ? what if i just do*bitbake hello* ? will this generate the rpm package for hello recipe and add it to the package feed ? Feb 06 16:28:03 i want to the first then second :) Feb 06 16:28:22 distsys: if hello is not included in core-image-minimal, then core-image-minimal will not be rebuilt. Otherwise, both hello and the image (with the hello included) will be rebuilt. Feb 06 16:28:33 distsys: if you only issue 'bitbake hello' then nothing happens to the image Feb 06 16:28:53 Ok Feb 06 16:29:02 now what about the second option :) Feb 06 16:29:24 what about it? what is the question? Feb 06 16:29:33 distsys: yes it will build the recipe, make packages and put them in the feed Feb 06 16:29:47 if you want to use the repo as an actual feed on the target, bitbake package-index to force it to rebuild the indexes Feb 06 16:29:56 (pretty sure this is in the docs) Feb 06 16:29:58 If i want it to be available only on the package feed, and install it via ackage mgmt Feb 06 16:30:38 distsys: please do read the manual :) Feb 06 16:30:43 oO( 'ackage feed' is kinda wonderful as a typo) Feb 06 16:30:48 there's a whole section there about package feeds Feb 06 16:31:23 https://bugzilla.yoctoproject.org/show_bug.cgi?id=12273 Feb 06 16:31:24 Bug 12273: normal, Medium+, 2.5 M2, ross.burton, NEW , non-deterministic search for package dependencies Feb 06 16:31:28 RP: looking at AB logs now for eclipse Feb 06 16:31:34 rburton: yes it will build the recipe, make packages and put them in the feed ? which command you mean ? is it bitbake recipe_name ? Feb 06 16:31:52 distsys: yes. Feb 06 16:31:54 RP: last time I looked the Old AB was down, probably being rebooted Feb 06 16:32:15 https://bugzilla.yoctoproject.org/show_bug.cgi?id=12516 Feb 06 16:32:16 Bug 12516: normal, Medium, 2.5, juro.bystricky, NEW , e2fsprogs: POT-​Creation-​Date removed non-deterministically Feb 06 16:32:21 rburton: ok very good thanks :) Feb 06 16:33:13 RP: I got bogged down testing poky-lsb and mulsc yesterday, but my RFC/pull request for the new kernels is out. Feb 06 16:33:22 zeddii: its on the AB already :) Feb 06 16:33:42 can you tell from the sea of red? Feb 06 16:33:52 rburton. I have a meta-poky pull request I'm typing up now. Feb 06 16:34:07 moto-timo: https://autobuilder.yoctoproject.org/main/builders/eclipse-plugin-neon/builds/382/steps/Building%20Eclipse%20Plugin/logs/stdio Feb 06 16:34:08 yeah turns out i didn't test it locally ;) Feb 06 16:34:31 * rburton is sorry for the ab spam :) Feb 06 16:34:40 RP: yes, that is the one I am splunking Feb 06 16:34:53 * zeddii built those v4.14 and v4.15 kernels everywhere. so they should be minimally sane :D Feb 06 16:34:53 moto-timo: http://git.yoctoproject.org/cgit.cgi/yocto-autobuilder-helper/tree/scripts/publish-artefacts#n161 is the script that gets run Feb 06 16:35:09 moto-timo: I thought I'd translated the current code but its clearly not working :/ Feb 06 16:35:32 zeddii: great, we'll pull them in and see what happens :) Feb 06 16:35:40 RP: my suspicion is the docs are not building, but that is only a hunch so I need to keep reading Feb 06 16:35:53 zeddii, any news on 4.12 and 4.10 updates ? Feb 06 16:36:15 armpit_IST. There's a 4.12 series in test now. so it is 85% done. Feb 06 16:36:52 we did 4.1 and working to upstream Feb 06 16:37:03 meltdown Feb 06 16:37:13 rburton. I did the dev on one machine, and builds on another. so I hope all that red isn't from bad REVs. or something really went bad in my SRCREV update scripts. Feb 06 16:37:22 armpit_IST. cool, is that via CoreyM ? Feb 06 16:37:29 Yes Feb 06 16:37:53 I can pull the 4.1 stuff into my linux-yocto tree then. I haven't looked for 4.1 in the past week or so, since I was in 4.14+ hell. Feb 06 16:38:57 zeddii, you can preview via https://github.com/MontaVista-OpenSourceTechnology/linux-nonlts-secfix/tree/linux-4.1.y-secfix Feb 06 16:39:14 awesome. pulled it up. will have a look. Feb 06 16:39:54 we did 2.6.32 too : ) Feb 06 16:40:12 ouch! :D Feb 06 16:40:16 livin' the dream!! Feb 06 16:44:02 zeddii: ERROR: No recipes available for: Feb 06 16:44:02 /home/pokybuild/yocto-autobuilder/yocto-worker/nightly-mips/build/meta-yocto-bsp/recipes-kernel/linux/linux-yocto_4.10.bbappend Feb 06 16:44:03 over and over and over Feb 06 16:44:14 ah... the bbappend. Feb 06 16:44:23 I have dangling bbappends allowed. Feb 06 16:44:34 I'll delete them and send it. Feb 06 16:44:45 cheers Feb 06 16:45:01 I *knew* I'd miss something in that tangle. Feb 06 16:47:57 patch sent. grabbing lunch, will be back in 15 mins for the next thing ... Feb 06 16:52:01 zeddii: ERROR: No recipes available for: Feb 06 16:52:01 /home/ross/Yocto/meta-intel/recipes-kernel/linux/linux-yocto-rt_4.10.bbappend Feb 06 16:52:05 oh sorry meta-intel Feb 06 16:52:06 ignore me Feb 06 16:53:34 RP: hunch was wrong, docs built ok. I'll keep digging in a bit when I am in the office Feb 06 16:54:12 moto-timo: I was hoping it might be obvious to someone who actually knows what that step is doing! Feb 06 16:54:33 moto-timo: are the commands at the end looking for the right files and they're not getting generated? Feb 06 16:54:38 moto-timo: (and thanks!) Feb 06 16:56:55 RP: all the kernel bits in ross/zedd and getting a spin on the ab now that isn't exploding yet Feb 06 17:56:31 RP: I know better than anybody what is happening, so I will figure it out Feb 06 18:17:36 is order of entries in PACKAGES the only determining factor to how files are picked up during package-split? is ${PN}-dbg any special and doesn't follow the order of PACKAGES? Feb 06 18:22:30 denix: -dbg is special Feb 06 18:22:54 denix: ended up special casing it as setting FILES_${PN}-dbg in all cases was a pain Feb 06 18:28:11 RP: how can I override it so I can package some files separately - i.e. I have subpackage and I want subpackage-dbg to work, but it doesn't Feb 06 18:28:31 denix: policy is no separate -dbg packages Feb 06 18:28:53 denix: I don't remember if the code has any overrides Feb 06 18:29:07 denix: its in package.bbclass Feb 06 18:30:34 RP: thanks, I'll look there Feb 06 18:36:23 Is there any guidance as to when a pull request should be used instead of a stack of patches? Feb 06 19:28:02 zeddii: zeddii_home: does -lsb not need 4.9 ltsi? Feb 06 19:28:24 lsb has no kernel requirements.. Feb 06 19:28:34 lsb only has application/libraries requirements Feb 06 19:28:51 you can be LSB compliant w/ BSD if you really wanted to be, as long as you can run Linux apps and meet the required APIs Feb 06 19:28:59 fray: misunderstanding on my part then Feb 06 19:29:34 (lsb is pretty much on life support anyway.. it's still reasonable 'sanity test' in many cases.. but everything has standardized for the point it's not a big deal) Feb 06 20:24:13 rburton: we should apply glibc 2.27 before binutils update Feb 06 20:24:26 rburton: that way we can ignore the 2.26 build failure Feb 06 20:27:44 khem: ok Feb 06 20:28:03 New news from stackoverflow: Python is Missing gzip.py in Yocto/Bitbake Build || how to add an existing recipe to a custom BSP layer in Yocto Feb 06 21:17:24 moto-timo, Had to upgrade our Nexus3 for the recent security issue. Looks good to me. Can you take a peak? Feb 06 21:17:35 moto-timo, Maybe we can make a post-update checklist. Feb 06 21:22:02 halstead: will do thank you. Good idea on checklist. Feb 06 21:22:25 Thank you! Feb 06 22:28:28 New news from stackoverflow: Bitbake: "Nothing PROVIDES" error Feb 06 22:58:36 New news from stackoverflow: yocto/bitbake recipe for adding empty directory to rootfs embedded linux Feb 06 23:12:19 hello, What is the reason for calling getty instead of start_getty in recipes-core/sysvinit/sysvinit-inittab_2.88dsf.bb for tty* devices? Feb 06 23:52:03 armpit_IST: We're seeing grub-efi and ruby-native failures on morty still :( Feb 06 23:56:22 RP, is that on stable/morty-next ? Feb 06 23:57:36 armpit_IST: and master morty Feb 06 23:57:58 armpit_IST: newer hosts, older release problem Feb 06 23:58:28 I have the gnu-efi gcc 7 fix in morty-next Feb 06 23:59:03 ruby: https://autobuilder.yoctoproject.org/main/builders/nightly-musl/builds/618/steps/RunConfig/logs/stdio, grub-efi-native: https://autobuilder.yoctoproject.org/main/builders/nightly-musl-x86-64/builds/73/steps/RunConfig/logs/stdio Feb 06 23:59:06 armpit_IST: ah, great! Feb 06 23:59:33 armpit_IST: Is anything ready for merging? Feb 06 23:59:38 the ruby is a package update Feb 07 00:00:01 rocko-next was clean last weekend Feb 07 00:00:21 pryo and morty have build failures Feb 07 00:01:41 armpit_IST: you mean an upgrade to fix it? Feb 07 00:02:12 no, it was upgraded to 2.4 for webkitgtk update to fix spectra Feb 07 00:02:35 I don't know if we got a clean build with that update Feb 07 00:03:17 armpit_IST: we have and haven't, depends which host gets it to build. I think it broke before too :( Feb 07 00:03:18 * armpit_IST brain is scattered this week. in India and not sleeping much Feb 07 00:03:40 RP, ah, ok Feb 07 00:03:50 armpit_IST: np, sorry to ask tricky questions Feb 07 00:04:20 armpit_IST: I know there is another ruby upgrade in master with claims of fixing some other issues :/ Feb 07 00:04:37 I'll look Feb 07 00:04:57 I have not seen ruby fail on pyro or has it ? Feb 07 00:05:23 armpit_IST: It may have, I'm lost off though :/ Feb 07 00:06:09 have we seen ccache fail on other releases? Feb 07 00:06:19 armpit_IST: I have the added complexity that I'm testing the new AB code :/ Feb 07 00:06:32 armpit_IST: I don't remember ccache for a while Feb 07 00:07:11 it failed on pyro.. maybe morty and the fix seems to be in a version newer than master Feb 07 00:08:17 armpit_IST: "great" :/ Feb 07 00:21:08 ccache version 3.3 + included c99 support so may be why pyro and up are fine Feb 07 00:28:56 New news from stackoverflow: Bluetooth over serial (SPP) configuration and setup **** ENDING LOGGING AT Wed Feb 07 03:00:02 2018