**** BEGIN LOGGING AT Wed Dec 16 02:59:59 2015 Dec 16 08:58:58 otavio: in case you didn't see it, fsl-ppc is failing on the AB: https://autobuilder.yoctoproject.org/main/builders/nightly-fsl-ppc/builds/556/steps/BuildImages/logs/stdio Dec 16 10:07:36 Hi all, does anyone know how to ignore a bbappend file without delete it? Dec 16 10:08:15 bluelightning: Hi, I am getting an error which you might be able to help with. tzdata receipe under meta/recipes-exteneded is trying to install stuff under ${datadir}/zoneinfo/ but that is already installed by the external-linaro-toolchain recipe. Dec 16 10:08:27 bluelightning: how can I fix this properly Dec 16 10:12:56 clement_: look for BBMASK Dec 16 10:15:05 CTtpollard: yes i tried it, but it seems only working in the local.conf file Dec 16 10:16:48 clement_: is the a problem? you can give a path as the value to mask a specific layers append Dec 16 10:18:40 clement_: thanks a lot, it is working actually Dec 16 10:18:49 CTtpollard: thanks a lot, it is working actually Dec 16 10:19:02 clement_: np Dec 16 10:26:39 anyone have any ideas what instance is overwriting /etc/hosts on boot in yocto/genivi? Dec 16 10:28:58 jani, maybe in base_files.bb? Dec 16 10:39:26 aragua: thats a runtime configuration? i'll check that when i get my hw back - college seems to have taken it. Dec 16 10:39:30 thanks. Dec 16 14:17:27 would it be possible to move the initramfs manifest generated during the build of core-image-minimal into a generated OS's rootfs? Dec 16 15:18:27 who can update https://www.yoctoproject.org/tools-resources/community/mailing-lists ? Dec 16 15:18:41 it doesn't mention yocto-security, secutity, openembedded-architecture MLs Dec 16 15:19:06 typo in secutity is intentional as the ML is private :) Dec 16 15:20:26 JaMa: I had some luck recently filing bugs in b.yp.o for the web site Dec 16 15:21:06 JaMa: https://bugzilla.yoctoproject.org/enter_bug.cgi?product=Website Dec 16 15:22:53 https://bugzilla.yoctoproject.org/show_bug.cgi?id=8819 Dec 16 15:22:54 Bug 8819: normal, Undecided, ---, jeffrey.osier-mixon, NEW , https://www.yoctoproject.org/tools-resources/community/mailing-lists should mention new MLs Dec 16 15:22:55 ok Dec 16 15:29:35 is it possible in wic to specify a partition without a mountpoint? Dec 16 15:29:54 alternatively what happens if you specify the same mountpoint for two different partitions? Dec 16 15:44:16 <_4urele_> hello everyone! Dec 16 15:45:35 <_4urele_> does anyone knows how to add ssh public key to an image (in order to connect without password) Dec 16 17:03:54 * pedroalvarez subscribes to _4urele_ question Dec 16 17:17:51 denix: so I deleted my build folder to do a clean compilation for the BB X15, but at the end I still have the compilation errors for omapdrm-pvr Dec 16 17:18:10 denix: should I stay on the daisy branch for every meta? Dec 16 17:19:08 Anarky: why Daisy specifically? are you locked to 3.14 kernel for any reason? Dec 16 17:22:37 denix: no, but it is the version indicated in the "setting up build environment", so I was thinking it is the latest stable version Dec 16 17:23:13 Anarky: fido with 4.1 is the latest official release Dec 16 17:24:15 denix: ok, I will recompile with this version then Dec 16 18:00:16 _4urele_: for your ssh keys question there is this yocto recipe you could incorporate into your own layer: http://git.yoctoproject.org/cgit/cgit.cgi/meta-web-kiosk/tree/recipes-common/ssh-keys/ssh-keys_0.1.bb?h=master Dec 16 18:22:48 so I'm still trying to figure out how to programmatically build different machine types without modifying my local.conf file. it seems the first thing bitable does is unset the MACHINE variable and pulls it in from the =? in local.conf Dec 16 18:23:09 it's set with ?=, which means its set there only if it's not already set Dec 16 18:23:17 by default it's allowed to come in through the environment, in which case it'd already be set Dec 16 18:23:20 just MACHINE=foo bitbake bar Dec 16 18:24:20 that doesn't seem tow work for me, I always see unset MACHINE as an early entry in my build.log Dec 16 18:25:35 denix: sorry to bother you again, but which version of the linaro toolchain should I use? with the 2013.03 and 2013.10 I can't build zlib ("ld: cannot find ../../lib/libc.so.6") Dec 16 18:26:13 deviosity: yes, and? MACHINE isn't exported into tasks Dec 16 18:26:21 it's still used just fine by bitbake itself Dec 16 18:26:27 run bitbake -e | grep MACHINE= Dec 16 18:28:45 all I get from that is my UBOOT_MACHINE= Dec 16 18:29:53 it's always listed in bitbake -e, just commented out. i'd suggest looking again Dec 16 18:29:58 same as with DISTRO Dec 16 18:36:03 I wonder if I am on a bad version of bitbake: bitbake version 1.26.0 Dec 16 18:36:30 so the closest thing I have is unset MACHINE in the bitbake -e Dec 16 18:38:12 And then $MACHINE_OVERRIDES which is my arch type and MACHINE?= definition Dec 16 18:40:32 regardless, setting MACHINE in the environment works just fine, it's only unset in exported shell scripts. iirc -e used the same code to emit variables as the shell scripts do, hence the unset. but it should also emit the commented out original value. ie.. '# MACHINE=foo' Dec 16 18:43:32 I'm not sure it does. I've never been able to get the value of MACHINE out of bitbake -e. Dec 16 18:44:11 MACHINE=foo bitbake -e | grep MACHINE= Dec 16 18:44:11 UBOOT_MACHINE="" Dec 16 18:44:20 mario-goulart: sounds like it's just a bug we should address Dec 16 18:44:31 need to be able to inspect such values, so we don't want the unexport flag to affect -e Dec 16 18:44:55 deviosity: again, that's likely just a bug, but it doesn't change the fact that it's still set in the bitbake metadata Dec 16 18:45:45 I can confirm that setting MACHINE in the environment works fine (unless you forcebly set it in the config). Dec 16 18:45:59 forcibly* Dec 16 18:45:59 I still don't understand why it won't use the defined MACHINE variable, only MACHINE ?= Dec 16 18:46:10 ? Dec 16 18:46:23 again, ?= *means* it will only get that value if the variable isn't already set to something else Dec 16 18:46:29 setting it in the environment sets it to something else. Dec 16 18:46:32 so that ?= will do nothing Dec 16 18:48:08 would it pull this in from somewhere other than local.con? Dec 16 18:48:15 local.conf even Dec 16 18:54:05 variables set in the process/shell environment will flow into the metadata, if the variable names are either in the default whitelist or are in BB_ENV_EXTRAWHITE. MACHINE is in BB_ENV_EXTRAWHITE as set by scripts/oe-buildenv-internal Dec 16 18:54:55 MACHINE won't be set elsewhere, it has no default value Dec 16 18:55:18 so if I set MACHINE ?= do a deliberate invalid machine and run MACHINE=imx6dlsabresd bitbake core-image-base, all I get is Please set a valid MACHINE in your local.conf or environment Dec 16 18:56:18 sounds like you have a problem in your setup. try MACHINE=imx6dlsabresd BB_ENV_EXTRAWHITE="$BB_ENV_EXTRAWHITE MACHINE" bitbake core-image-base Dec 16 18:57:34 okay that gets me somehwhere Dec 16 18:58:26 ah, okay, so either you're using custom setup scripts, or the version of oe-buildenv-internal in your oe-core didn't include MACHINE in BB_ENV_EXTRAWHITE by default Dec 16 18:58:30 that'd explain it Dec 16 18:58:58 another option, assuming you're scripting this, would be to pass it in via auto.conf rather than the environment Dec 16 18:59:10 i.e. echo 'MACHINE=imx6dlsabresd' >conf/auto.conf Dec 16 18:59:17 I'm just using oe-init-build-env Dec 16 18:59:38 I see that whitelisting in oe-buildenv-internal Dec 16 18:59:48 oe-init-build-env sources oe-buildenv-internal under the hood Dec 16 19:55:02 * reanguiano works on getting tcpdump from oe onto a dizzy build Dec 16 19:56:33 tcpdump is in meta-networking.. shouldnt' be any real work to do, just add meta-networking to bblayers, and add it to your image Dec 16 19:58:04 kergoth: cool Dec 16 19:58:50 kergoth: so roughly: clone meta-openembedded, branch to dizzy, include path to meta-openembedded/meta-networking in bblayers, add tcpdump package to the appropriate PACKAGE/IMAGE_INSTALL statement in build/conf/local.conf? Dec 16 19:59:35 yep Dec 16 20:00:04 CORE_IMAGE_EXTRA_INSTALL += "tcpdump" is probably easiest, though IMAGE_INSTALL_append can work too (but that can also apply to initramfs images) Dec 16 20:01:12 :nod: Dec 16 20:02:19 isn't it better to just make a .bbappend to your image recipe and just append to IMAGE_INSTALL ? Dec 16 20:02:36 what are the pros/cons of IMAGE_INSTALL_APPEND versus IMAGE_INSTALL Dec 16 20:02:44 depends what you mean by better. for long term maintenance yes Dec 16 20:02:46 the best is to have your own completely separate image recipe and edit that Dec 16 20:02:49 local.conf is good for a quick test/hack/whatever Dec 16 20:02:51 thats about it Dec 16 20:03:09 but generally it's better to append in your own layer right? Dec 16 20:03:13 for long term Dec 16 20:03:15 not for image recipes Dec 16 20:03:17 ah Dec 16 20:03:34 so i should have my own big image recipe from scratch? Dec 16 20:03:54 you can copy an existing one and modify it as desired Dec 16 20:04:01 image recipes are so trivial, and you usually want full control over what goes into them, so you might as well do it that way Dec 16 20:04:19 that's true, Dec 16 20:04:37 i usually just appended core-image-minimal whatever i needed Dec 16 20:04:44 i didn't need much though Dec 16 20:04:57 just things like GNU screen, mdadm, btrfs progs Dec 16 20:05:14 that's why it seemed overkill to have my own image Dec 16 20:05:32 core-image-minimal contains almost nothing as is Dec 16 20:05:38 copying it is equally trivial, and eaiser to maintain :) Dec 16 20:05:49 fair enough Dec 16 20:05:59 right... and to me, if you're adding stuff like that it's no longer "minimal" so it might as well have a different name :) Dec 16 20:06:32 indeed Dec 16 20:06:56 is it ok to just change the commit ID in src rev in bbappend if i want things like latest sources for btrfs progs? Dec 16 20:07:04 or do i need my own recipe for that too Dec 16 20:08:22 you could, but in my experience you'll forget to update it after upstream bumps it, and now you're using an older version than upstream :) Dec 16 20:08:29 * kergoth has been bitten by that multiple times Dec 16 20:08:51 is there a way to make it just checkout on the latest sources? Dec 16 20:09:23 a recipe you mean? Dec 16 20:09:42 if so, just set SRCREV = "${AUTOREV}" within the recipe Dec 16 20:09:47 yeah, just give a git url to SRC_URI and no SRC_REV, just always get the latest in do_fetch Dec 16 20:09:52 ah Dec 16 20:09:58 thanks Dec 16 20:11:11 SRCREV rather than SRC_REV btw Dec 16 20:11:19 (don't you just love consistency...) Dec 16 20:11:49 oh there's 2? Dec 16 20:11:53 heck Dec 16 20:12:44 another thing, i have a recipe for a custom kernel (non yocto), and it boots, but i don't have any loadable modules, IF i had a lot of modules, is there an easy way to just include them all in the image without having to append all the modules one by one? Dec 16 20:13:05 btw, I've seen a SRCREV that contained a git commit hash followed by a single space today... Dec 16 20:13:18 and it failed to parse with gitpkgv Dec 16 20:13:53 is that something that would be considered a bug? Dec 16 20:14:13 that someone wrote a space or that it can't parse out a space? Dec 16 20:14:25 That is can't ignore the space. Dec 16 20:14:27 maybe it should only read 40 chars Dec 16 20:14:32 s/is/it/ Dec 16 20:14:57 sound like a feature request to me. Dec 16 20:15:11 one that would encourage bad practice though Dec 16 20:15:24 Personally, I don't even like changing that. If people don't pay attention to whitespace, let it fail. Dec 16 20:15:34 But maybe the message should be improved. Dec 16 20:17:55 that's like saying always put rvalues on the left side of a conditional, has ANYONE here ever really mistakenly written = instead of == in an if statement? Dec 16 20:18:59 you let them write one space after the hash, next thing you know they're be writting carriage returns, AND WHERE DOES IT STOP?? Dec 16 20:19:09 they'll* Dec 16 20:19:44 obsrwr: yep, I have ;) Dec 16 20:20:00 * obsrwr argument lies smashed to pieces Dec 16 20:20:34 luckily these days I think you get a compiler warning for such things, at least from gcc Dec 16 20:20:46 I think clang actually started this warning Dec 16 20:21:00 If you want the assignment, you need to use double braces now Dec 16 20:22:02 in C as well as C++? Dec 16 20:22:23 I write very little C++, so I can't tell without trying. Dec 16 21:44:11 kergoth: Thanks again. It did turn out be easy to get tcpdump built and onto this image. Dec 16 21:45:32 * reanguiano did have to include a handful of dependencies in the bblayer but those were in the meta-openembedded dir. **** ENDING LOGGING AT Thu Dec 17 02:59:58 2015