**** BEGIN LOGGING AT Thu Aug 15 03:00:00 2013 Aug 15 05:05:53 hello, i rebuild qt4e-demo-image (from begin), imx6qsabresd, it compiling, but when i run it - i got segmentation fault Aug 15 06:14:16 morning Aug 15 06:15:25 i just pull the latest changes from meta-openembedded Aug 15 06:15:40 now bitbake fails with this error: systemd.bbclass Aug 15 06:15:50 Parsing recipes..ERROR: ParseError at /home/jose/mcsdk/sources/meta-openembedded/meta-oe/recipes-graphics/lxdm/lxdm_git.bb:27: Could not inherit file classes/systemd.bbclass Aug 15 06:16:44 that file is actually missing in /meta-oe/classes Aug 15 06:18:34 melonipoika: do you have the current meta layer from either oe-core or poky? Aug 15 06:19:17 sgw_ i cloned it from here: systemd.bbclass Aug 15 06:19:19 ups Aug 15 06:19:21 git://git.openembedded.org/meta-openembedded Aug 15 06:19:51 melonipoika: yes, I get that, but you also need openembedded-core Aug 15 06:19:56 ah Aug 15 06:20:10 yes, i do have oe.core too Aug 15 06:20:13 I guess you have also cloned bitbake, Aug 15 06:20:25 yes Aug 15 06:20:52 melonipoika: when did you last pull oe-core? We have been merging the systemd changes from meta-oe to oe-core Aug 15 06:20:57 i am using TI's mcsdk, they use one scrip to clone the layers Aug 15 06:21:39 i am using this commit for oe-core Aug 15 06:21:42 0785489d558c34cacf5cb349949a15e27084efbf Aug 15 06:21:48 old stuff I guess, Aug 15 06:22:04 melonipoika: sorry I am not familiar with their tools, I know poky and oe-core master best. Aug 15 06:22:07 Looking at the ref Aug 15 06:22:19 let's me pull oe-core too, that might explain it Aug 15 06:22:57 yes, that is from 1.4, which is before we started to merge systemd into oe-core Aug 15 06:24:14 right Aug 15 06:27:19 ok, will see what happens if i pull oe-core too :-) Aug 15 06:39:00 OK, I updated oe-core and bitbake. Now i get a banch of QA errors in ti's meta-arago recipes :-) Aug 15 06:41:47 melonipoika: does the meta-arago have a dependency list that defines which layers and branches (or git revs?) maybe when you updated the meta-oe was the cause and you need to go back for meta-oe and oe-core Aug 15 06:42:55 sgw_ I am not sure, I will take a look. The QA errors I get are of the type "Variable RREPLACES is set as not being package specific, please fix this." Aug 15 06:43:58 melonipoika: it might be that you updated the meta-oe and (on my suggestion) updated oe-core, but really need to stay with the branches that meta-arago require Aug 15 06:44:12 I guess it will be just easier for me just to checrry pick the updated version of the package I need (opencv) on top of the old branches Aug 15 06:44:33 they might ahve their reasons to use old stuff :D Aug 15 06:44:58 yep :-) Aug 15 06:44:59 yeah, like the systemd integration in this case for one, there might be other issues Aug 15 06:45:43 but thanks for your help, I am completely new to yocto so this is a good learning for me :D Aug 15 06:46:49 melonipoika: some of the distro layers require older branches of the oe-core and meta-oe, while other stay more uptodate, it depends on your board (BSP) needs. Aug 15 06:47:11 i see Aug 15 08:50:37 hi all, i'm writing a new recipes to install external toolchain (based on external-sourcery-toolchain) Aug 15 08:50:53 but got problem when do_package_qa Aug 15 08:50:59 Error: Files/directories were installed but not shipped Aug 15 08:51:10 what does Files/directories were installed but not shipped mean? Aug 15 08:52:49 do_install() installed files, but they were not picked up by any FILES_* statements to get them into packages Aug 15 08:54:25 Ah, thanks rburton Aug 15 08:57:10 uvan: u-boot 2013.07 does not build for me. Aug 15 08:57:15 missing some config.h Aug 15 08:57:22 when trying to build the fw-utils. Aug 15 08:57:42 same with the sourcery toolchain versions 2009q1 and 2013.07 Aug 15 08:57:45 same with the sourcery toolchain versions 2009q1 and 2013.05* Aug 15 08:58:38 morning all Aug 15 08:59:48 lpapp: i used our own u-boot with support arm64 Aug 15 09:00:16 uvan: can you try my latest patch on the mailing list? Aug 15 09:01:07 lpapp: could you give me the link. i seach but not found Aug 15 09:03:00 uvan: http://patches.openembedded.org/patch/55653/ Aug 15 09:06:40 lpapp: ok. let me try Aug 15 09:07:44 uvan: bitbake u-boot-fw-utils just in case. Aug 15 09:19:43 rburton: Time to improve that message I think, this is becomming a FAQ Aug 15 09:21:14 RP: yes Aug 15 09:29:44 hello, i rebuild qt4e-demo-image (from begin), imx6qsabresd, it compiling, but when i run it - i got segmentation fault Aug 15 09:43:03 lpapp; i got the same error with both 2013.07 and 2013.01 Aug 15 09:45:16 uvan: perhaps we should drop fw-utils for now then. Aug 15 09:45:20 I do not have time to fix that build error. Aug 15 09:46:28 it's ok lpapp, actually i'm not using that tool Aug 15 10:22:04 lpapp: are you going to fix qtserialport or should I drop it from meta-qt5? Aug 15 10:22:37 JaMa: ? Aug 15 10:23:29 can anybody say, is it my mistake with qt4e-demo-image, or it broken in master? Aug 15 10:23:29 lpapp: I've replied to your patch on oe-devel ML Aug 15 10:23:43 JaMa: I am not subscribed. Aug 15 10:24:11 's the issue Aug 15 10:24:53 uvan: k, thanks for verifying. Any clue at any rate how to fix this? Aug 15 10:24:57 lpapp: then use archives to find it Aug 15 10:25:03 ivan: should I bring it up in #u-boot? Aug 15 10:25:15 JaMa: do not have time atm. Aug 15 10:25:25 ok, reverted Aug 15 10:25:48 why ? Aug 15 10:27:06 it wasn't working Aug 15 10:27:25 JaMa: then you are doing something wrong. :) Aug 15 10:27:33 JaMa: I do not see any reply to be honest on the mailing list. Aug 15 10:27:49 although probably it is worth having more patient than "you do not have right now, so I revert it" Aug 15 10:27:56 yes accepting patches without properly testing them because I was too busy Aug 15 10:27:56 patience* Aug 15 10:28:30 lpapp: http://lists.openembedded.org/pipermail/openembedded-devel/2013-August/091671.html Aug 15 10:29:00 lpapp: i tried it with u-boot separate and also got the same thing. Aug 15 10:29:49 uvan: now that is strange. Aug 15 10:29:53 bitbake u-boot works for me. Aug 15 10:29:59 JaMa: pff Aug 15 10:30:19 JaMa: wrong push I guess. Aug 15 10:30:27 u-boot work, but u-boot-fw-utits not work Aug 15 10:30:34 cannot write "not tested" when I think it was tested. Aug 15 10:30:39 uvan: yeah Aug 15 10:30:45 uvan: let us ask in #u-boot tonight. Aug 15 10:31:46 lpapp: wrong commit http://patchwork.openembedded.org/patch/55035/ Aug 15 10:32:39 lpapp: i have to leave now. see you tomorrow Aug 15 10:32:46 uvan: see you, thanks again. Aug 15 10:33:20 JaMa: well, someone will add back anyway. Aug 15 10:33:58 yes hopefully next version will be tested before sending it Aug 15 10:34:56 JaMa: as I said it was tested. ;) Aug 15 10:36:08 I wonder how, as it clearly doesn't have right LIC_FILES_CHKSUM Aug 15 10:36:17 this discussion is waste of time, bye Aug 15 10:36:34 wrong push != untested Aug 15 10:37:12 are you saying that I cannot push test from e-mail? Aug 15 10:37:26 s/test/patch/ Aug 15 10:37:47 no, I am saying I made a wrong push. Aug 15 10:37:54 I pushed a broken version. Aug 15 10:38:00 even though I fixed that localy. Aug 15 10:38:02 locally* Aug 15 10:38:07 and I should have pushed the fixed. Aug 15 10:38:33 push == sent email, right? Aug 15 10:38:43 yes Aug 15 10:39:14 anyway, this is pointless, reverting it is not a big harm. Aug 15 10:39:29 it will be readded pretty quickly. Aug 15 10:39:29 then my comment was correct :), someone will send tested version, not some random version from his local branch Aug 15 10:39:51 both versions are tested Aug 15 10:39:53 broken and fixed Aug 15 10:40:05 problem is the push, not that those changes are not tested. Aug 15 10:40:39 in which case, obviously the author cannot write I have not tested it. Aug 15 10:40:42 when he thinks he did. Aug 15 10:40:56 anyway, I am gone. Aug 15 10:41:41 oh, one thing: it would be nice to have scenegraph packaged. Aug 15 10:41:44 JaMa: ^ Aug 15 10:41:49 from playground and git. Aug 15 10:41:58 even though it will be dumped soon hopefully. Aug 15 11:10:49 How to add dependency from class to recipe? Aug 15 11:11:01 I have this... do_package[depends] += xaf-native:do_populate_sysroot Aug 15 11:11:16 but it gives compilation error Aug 15 11:11:23 what is correct syntax Aug 15 11:11:25 ? Aug 15 14:12:13 khem: HOORAY QEMU PPC FIX Aug 15 14:24:41 Hum.. I'm trying to get git requests for git:// to go through our company's proxy server. It starts up, then immediately hangs without transferring much of any data. Aug 15 14:25:04 (I'm following https://wiki.yoctoproject.org/wiki/Working_Behind_a_Network_Proxy#Git_Setup_.28with_nc.29) Aug 15 14:26:01 the 'nc' command is running (nc -x proxy-server:8080 -X 5 git.yoctoprojecto.org 9418) Aug 15 14:26:48 does git.yoctoproject.org support http? Aug 15 14:35:00 is there any way to execute yocto build on 2 different machines connected together over ethernet? Aug 15 14:35:39 Hi. I'm getting this error when building pseudo-native on a x86_64 system (using dylan): http://paste.debian.net/25507/ . Do you know what may be causing it? Aug 15 14:37:22 try this: http://www.cyberciti.biz/faq/x86_64-linux-error-gnustub-32h-missing-error-and-solution/ Aug 15 14:39:09 Krz: before collecting that error message I had installed libc6-dev-i386 (Debian). Should it have done the trick? Aug 15 14:43:20 strange then; what do you have under /usr/include/gnu ? Aug 15 14:44:23 Garibaldi|work: http is supported yes Aug 15 14:47:24 Krz: I don't have it. I have /usr/include/gnu-versions.h though. Aug 15 14:48:26 you must be missing something; i would try reinstalling libc-dev + libc-dev-i386 Aug 15 14:48:50 whatever they are called under your linux distro Aug 15 14:51:19 Krz: I probably am. Thanks for the tips. Aug 15 14:55:36 I guess it helps if I install it on the chroot I'm working. Aug 15 14:55:40 * mario-goulart facepalms Aug 15 15:01:12 Now I have /usr/include/gnu ( http://paste.debian.net/25515/ ) and this error: http://paste.debian.net/25514/ Aug 15 15:03:59 http://stackoverflow.com/questions/14795608/asm-errno-h-no-such-file-or-directory Aug 15 15:04:12 looks like linux headers missing Aug 15 15:05:07 Yeah, I tried some, but it seems they are the wrong ones. Aug 15 15:07:21 do: find / -name asm.h Aug 15 15:27:43 hey guys, quick question: where is the do_compile() task located for linux-yocto.bb? the obvious place I thought would be the recipe itself but unless I'm missing something I haven't been able to see it in there. Aug 15 15:29:43 simar: it'll be generated from the classes inherited by the recipe Aug 15 15:30:19 simar: most likely kernel_do_compile() in meta/classes/kernel.bbclass Aug 15 15:40:04 ah got it. thanks bluelightning! Aug 15 15:43:31 if I change SDKMACHINE do I have to wipe out whole tmp/ dir? Aug 15 15:46:41 So, I'm writing a recipe that needs to pull source from a git repo. I just updated the SRCREV to a new hash. It's definitely downloading the new version. But it doesn't complain about the now incorrect md5 checksums. Are checksums simply ignored when fetching from a git repo? Aug 15 15:47:52 mulhern:cksum is ignored; you want to use commit SHA Aug 15 15:49:29 Krz: no, you don't need to do that Aug 15 15:49:47 Krz: (I mean, you should just be able to change it and run the SDK build again) Aug 15 15:50:10 bluelighting: nice, can I build two SDKs in one bitbake command? Aug 15 15:50:29 damn, always type your nick wrong :/ Aug 15 15:50:56 Krz: not for two different SDKMACHINEs no - but you can build multiple SDKs with different contents for the same SDKMACHINE at once Aug 15 15:50:57 Krz: no nick completion with ESC? Aug 15 15:51:05 Krz: use tab completion ;) Aug 15 15:51:17 sorry tab, bluelightning got me Aug 15 15:51:21 works, didn't expect webclient to be so clever :) Aug 15 15:51:45 ah nice... I've never used the web client, good to know it has basic functionality :) Aug 15 15:51:53 Krz: Deleting those SRC_URI lines, then. But what do you mean by commit SHA. The git commit SHA that I set SRCREV to? Aug 15 15:52:31 mulhern: yes, either SHA for SRCREV, or "${AUTOREV}" to pick latest&greatest commit from choosen branch Aug 15 15:53:32 mulhern: Krz: AUTOREV is not recommeded nor allowed in oe-core since it causes network activity, oe-core should build self-contained once all the files are downloaded Aug 15 15:58:30 khem: Woot on the PPC issue Aug 15 17:00:55 I have my own layer with bbappend inside and .patch file. Now bbappend applies to original poky recipe and fails during fetching my patch... patch is my layer in: recipes-core/expat/expat/ Aug 15 17:01:18 Yocto does not even try to look there, what can be wrong? Yocto searches only poky/ paths Aug 15 17:01:29 you need to set FILESEXTRAPATHS appropriately Aug 15 17:01:35 grep around, it's used in many bbappends Aug 15 17:02:05 e.g. https://git.linaro.org/gitweb?p=openembedded/meta-linaro.git;a=blob;f=meta-linaro/recipes-core/busybox/busybox_1.21.1.bbappend;h=0fd4571e7b09161c3fd8a2ccd42cd9cfea9babb6;hb=refs/heads/master Aug 15 17:12:23 Krz_: something like this at the top of your bbappend will make it look in a subdir called $PN Aug 15 17:12:27 FILESEXTRAPATHS_prepend := "${THISDIR}/${PN}:" Aug 15 17:12:40 ndec: btw, are you in charge of meta-linaro now after hrw? Aug 15 17:13:23 mr_science: now you wonder if that line should be a default for every bbappend... :) Aug 15 17:15:09 denix: things that make you go "hmm..." Aug 15 17:33:27 mr_science: thanks, works Aug 15 17:34:13 mr_science: well, partially, patch is downloaded and "applied" into my configure script Aug 15 17:34:45 mr_science: the only thing is my configure script is not changed :) Aug 15 17:35:11 don't patch the configure script if you're using autoconf. the autotools bbclass re-runs autoreconf, which regenerates configure from configure.{in,ac} Aug 15 17:36:37 mr_science: now I have that song stuck in my head, thanks :) Aug 15 17:37:14 kergoth: damn, so how do I achieve the same? I need small mod in configure script Aug 15 17:37:41 Krz_: you're not modifying the generated configure script by any chance? for an autotooled software package that won't work Aug 15 17:38:21 you'd need to modify configure.ac or configure.in, assuming you're not already Aug 15 17:38:34 s/modify/patch/ Aug 15 17:39:04 well, I took expat from Yocto downloads directory, untarred, modified existing configure and added patch to the layer Aug 15 17:39:23 patch the configure.ac or configure.in Aug 15 17:39:30 ok, will do Aug 15 17:39:31 right, what he said ;) Aug 15 17:40:05 just repeating what one another said with different wording :) Aug 15 17:40:20 * kergoth tries a meta-ivi image with qemuarm Aug 15 17:41:05 sgw_: I just sent a patch to fix default tune for qemuppc it depends on kernel patches from Bruce that he send today so club them together Aug 15 17:43:30 kergoth: yes actually I initially missed that you answered the question already, sorry :) Aug 15 17:44:57 :) Aug 15 17:48:59 hey guys, I've been working on building 3.10 from linux-yocto-dev for the BeagleBoard. I'm currently dealing with an breakage in the build that is caused by this upstream patch: http://goo.gl/zdAnaC Aug 15 17:50:08 hmmm... don't see a lot that could go wrong with that patch. Aug 15 17:50:44 so your UIMAGE_LOADADDR is 0 words? Aug 15 17:50:46 I'm not sure why it breaks but it complains of having not UBOOT_LOADADDRESS at the time of making the image. Aug 15 17:50:53 yeah jkridner Aug 15 17:51:22 I reverted that very patch and built it. I get a random load address of 0xffffff2 Aug 15 17:52:04 I looked into the kernel.bbclass which is inherited by linux-yocto-dev.bb but it hasn't been touched since. Aug 15 17:52:16 started using tmpfs for my Yocto builds, really feeling the difference Aug 15 17:52:57 I'm not sure if moving all layers there will make it way faster (as they are read only) Aug 15 17:53:18 I thought the do_uboot_mkimage() task would be responsible for making the uimage. But doesn't seem to be the case so far. Aug 15 17:58:43 bluelightning: kergoth: configure.in patch works like a charm, thanks Aug 15 18:00:24 great! Aug 15 18:00:25 bbl Aug 15 18:00:54 simar, the uImage is just passed to the kernel to build, can be overridden, see KEEPUIMAGE Aug 15 18:01:08 but we do want the kernel to be smart enough to build it itself, for simple cases. Aug 15 18:12:21 how do I trim down eglibc? I would like to remove WCHAR support and some others Aug 15 18:22:48 * mr_science needs to trim down his beard... Aug 15 18:25:42 * rburton needs to trim down his gut Aug 15 18:25:48 on that note, run time! Aug 15 18:43:14 rburton: don't hit any run-time errors Aug 15 18:44:11 * kergoth groans Aug 15 18:44:46 * mranostay ducks the rotten fruit flying from kergoth Aug 15 18:55:42 * mr_science reminds kergoth to gpg-sign his fruit first Aug 15 19:05:50 :-) Aug 15 19:32:00 zeddii: I see. I changed KEEPUIMAGE to something other than 'yes' so it would not run the do_uboot_mkimage() but it still fails. Aug 15 19:32:32 If this is something that will fix itself I can get around with it, I was just curious why it was breaking. Aug 15 19:34:48 I guess manually running mkimage on a vmlinux is the only way to get a uImage? Aug 15 19:45:33 urk no. Aug 15 19:45:42 everything we run builds uImage Aug 15 19:46:02 if your image type is uimage, then that is passed to the kernel build, which takes care of it. Aug 15 19:46:43 what you are looking at is around to allow boards that have specific requirements, like an entry point, to be easily set and passed to the mkimage phase. Aug 15 20:07:26 anyone started to see do_rootfs (ipk) failures lately with: ERROR: ld.so: object 'libpseudo.so' from LD_PRELOAD cannot be preloaded: ignored.? Aug 15 21:41:02 can I use my PV as "1.0+AUTOINC" ? Aug 15 21:41:16 will it do the auto incrementation if build multeple times? Aug 15 22:54:27 hi guys, can anyone on the channel help me with a eclipse/yocto plugin issue? Aug 15 22:55:32 the helloworld won't configure "C compiler cannot create executable" Aug 15 22:55:58 I think my sysroot location is wrong Aug 15 23:02:44 brm: FWIW the config.log file usually gives better info about the exact problem Aug 15 23:03:07 could not see any error message in the .log file Aug 15 23:05:23 I've been seeing a strange RPM dependency problem which I'd chalked up to some local patches of mine, but I've finally downloaded danny and it's doing the same thing Aug 15 23:05:46 actually I lie .. I get the following error: configure:3166: arm-poky-linux-gnueabi-gcc -march=armv7-a -mthumb-interwork -mfloat-abi=softfp -mfpu=neon --sysroot=/home/bmentink/beagleboard_black/build/tmp/sysroots/beaglebone -qversion >&5 arm-poky-linux-gnueabi-gcc: error: unrecognized command line option '-qversion' arm-poky-linux-gnueabi-gcc: fatal error: no input files Aug 15 23:06:41 for example, "rpm -qp --requires perl-dev-*" says that package depends on "perl-lib-dev" Aug 15 23:06:56 RP: So regarding the sysroot location, it is currently pointing at "/home/bmentink/beagleboard_black/build/tmp/sysroots" Aug 15 23:07:15 but no RPM actually provides that Aug 15 23:08:18 I can't help but notice that the debian package rename hook is changing perl-lib to libperl5, and if I look closely I see there's a RPROVIDES_perl-lib = "perl-lib" in the recipe Aug 15 23:09:17 is this a known problem? Aug 15 23:09:59 RP: My build produces a .xz root file image that I put on a SD card i see no way to use the tools to produce the rootfs somewhere cause the tool expects .tar.bz2 Aug 15 23:13:45 evanp: there are some dependencies generated for things which don't exist :/ Aug 15 23:14:50 brm: the above -qversion test sounds like one of the standard configure tests, I'd look later in the log Aug 15 23:18:10 RP: Would you mind having a look at http://pastebin.com/DNfb8Vfr Aug 15 23:18:38 RP: it seems to happen for lots of stuff that splits itself into several packages. libsegfault and libstdc++ produce similar problems, for example.... Is there an intended solution for it? I mean, I could add RPROVIDES everywhere I encounter it--is that how you'd characterize this? The recipes which do it are missing RPROVIDES? Seems more like a band-aid than a real solution to me. Aug 15 23:21:09 RP: Around line 1291 in the pastbin is where the errors are Aug 15 23:21:18 that is, the real problem feels like a broken algorithm for adding -dev RDEPENDS. however, finding a "correct" algorithm looks...nontrivial.... Aug 15 23:23:39 evanp: They are recommendations, not hard dependencies Aug 15 23:23:59 evanp: as you say, the algorithm is flawed but we don't know how to do anything better Aug 15 23:24:18 evanp: If you want to find the code, I think we called it dep_chains or something like that Aug 15 23:25:30 brm: the real problem is lines 1297-1308 - the linker can't find libraries it needs Aug 15 23:25:48 RP: are you sure they're recommendations? people have been telling me these RPMs refuse to install without --nodeps, at least on 1.2.... Aug 15 23:26:24 brm: --sysroot=/home/bmentink/beagleboard_black/build/tmp/sysroots sounds wrong Aug 15 23:26:39 brm: it should be pointing at some subdirectory of that shouldn't it? Aug 15 23:27:02 maybe an arm-xxx directory under there? Aug 15 23:29:01 * RP -> Zzzz Aug 15 23:30:52 gn Aug 15 23:51:59 where does bitbake put a kernel image by default? Aug 16 00:36:33 RP: Sorry got disconnected ... yes I agree sysroot sounds wrong Aug 16 00:37:47 cjosephson: build/tmp/deploy/images, where "build" is your build dir and possibly really is "build" Aug 16 00:40:40 RP: So the manual says sysroot must point to my root filesystem .. but all I have in the image directory is a brm-image-beaglebone.tar.xz file, so how do I get the tools to break it out somewhere? Aug 16 00:47:53 RP: Got it !!! As you say it should have been ... /sysroots/beaglebone .. conbfures fine now and built binary, Yay .. now all I have to do is get it to talk to my target board. Aug 16 02:07:19 Anyone know how I use runqemu-extract-sdk with a rootfs.tar.xz file instead of what it expects .... tar.bz2 ? Aug 16 02:16:26 i am asking cause when I run qemu from eclipse it is complaining about where I told it to find the kernel **** ENDING LOGGING AT Fri Aug 16 02:59:58 2013