**** BEGIN LOGGING AT Wed Aug 06 02:59:59 2014 Aug 06 04:55:49 Hi yocto.. Aug 06 04:56:48 Hi yocto.. Aug 06 05:01:05 Hi.. Am compiling qt5 with support of meta-qt5 layer if i compile this i am getting error in qtwayland do_install error saying that ".h file" no such file or directory as mentioned in pastebin http://pastebin.com/pt5KTuhM please any one help me to rectify this error. Aug 06 05:13:32 please any one help us to resolve this problem.. Aug 06 05:17:42 looks like that glob isn't working as expected Aug 06 05:18:00 which branch are you building? Aug 06 05:20:41 embed: take a look in that directory and see if there any headers that match the glob Aug 06 05:24:08 how do I add xorg-lib to a custom layer? Aug 06 05:27:50 there are some X package groups, but they pull in quite a few packages Aug 06 05:29:06 you could try adding x11-base to your IMAGE_FEATURES Aug 06 05:30:11 not sure which package you need, so... Aug 06 05:31:05 some random libs for oracle java, don't even plan to use graphics, but don't know how to pull them out of the oracle java dependencies Aug 06 05:31:21 oh, oracle crap Aug 06 05:31:38 is there a layer for another java runtime? Aug 06 05:32:10 not sure, i haven't tried building anything with java Aug 06 05:32:30 probably not the best thing to run on a pi... Aug 06 05:33:02 IMAGE_FEATURES doesn't complain, but the RDEPENDS recipe says it can't build x11-base Aug 06 05:33:11 * mr_science hopes yocto kernel will finish cloning before dawn... Aug 06 05:33:39 x11-base wouldn't help for java Aug 06 05:33:56 unless that same lib is in another recipe Aug 06 05:34:25 mr_science, when the java rpm installs it somehow requires alsa-lib and some xlibraries Aug 06 05:35:13 okay, you'll need to look at the rdepends in the java recipe(s) and figure out which package they come from Aug 06 05:35:55 I'll try it again without the RDEPENDS, but I believe yocto somehow introspected those dependencies Aug 06 05:35:58 so back to the previous answer, but i'm not sure why it complains Aug 06 05:36:28 they're defined somewhere in a bb file Aug 06 05:37:12 trying.... so yocto doesn't do some ldconfig test upon rpm install? Aug 06 05:38:21 says it can't install oracle java: no package provides libXext.so Aug 06 05:38:29 it must be figuring it out automagically Aug 06 05:40:08 i don't have a build setup handy... lemme spin up a VM Aug 06 05:41:45 is there a flag to ignore rpm ld dependencies? Aug 06 05:42:35 mr_science:please look this wayland.bb file, it's mentioned to install header file in Qtcompositor but in qtwayland build dir there no Qtcompositor please check this pastebin http://pastebin.com/6BanUwVG Aug 06 05:42:43 not a good idea, even if it existed Aug 06 05:43:07 redengin: do a find in your workdir for that lib Aug 06 05:44:25 embed: i've built regular wayland/weston but not the qt stuff Aug 06 05:44:29 mr_science, running... Aug 06 05:45:52 embed: i have no idea what happened to qtcompositor Aug 06 05:46:17 if you can't get an answer here, i would post it to the mailing list Aug 06 05:46:26 mr_science : i have a doubt about qtwayland/5.3.0+gitAUTOINC+98dca3b54f-r0/build/ dir there is no qtcompositor Aug 06 05:46:30 ok.. Aug 06 05:47:01 are you building off master branch in all of your repos, or is it a release version? Aug 06 05:47:42 if master, then you could try switching to the last release branch Aug 06 05:47:53 or the other way round... Aug 06 05:48:18 * mr_science is building mostly master branches and custom layers Aug 06 05:50:19 mr_science, no libXext.so* found in tmp/work Aug 06 05:51:18 okay then, someone is missing a build dep Aug 06 05:52:02 you could build somethign like the sato image and see which package provides it Aug 06 05:52:31 that one is fairly small for an X image... Aug 06 05:54:41 i'm kinda stuck on a laptop in a hotel cloning a yocto kernel so i have no bandwidth... Aug 06 05:55:18 embed: the branch question was for you Aug 06 05:55:35 adding x11 as the IMAGE_FEATURE appears to find the libraries Aug 06 05:55:49 ok.. Aug 06 05:55:53 doesn't look like I can just get xorg-lib Aug 06 05:56:09 i cloned from master.. Aug 06 05:56:27 x11 packagegroup will add some X packages but it's not horrendous Aug 06 05:56:36 i not cloned from branch.. Aug 06 05:56:40 sato image is still pretty small Aug 06 05:57:29 mr_science, I just wanted faster builds, the image size isn't too important Aug 06 05:57:36 embed: master isn't always in a sane state, but it normally works Aug 06 05:57:50 i cloned from this git https://github.com/meta-qt5/meta-qt5.git Aug 06 05:58:00 sometimes i have to fix stuff, depends on what you're trying to build Aug 06 05:58:26 then maybe an issue on that git repo? Aug 06 05:58:38 is the issue tracker enabled? Aug 06 05:59:29 redengin: still not too bad, since it should hopefully not pull in webkit... Aug 06 05:59:34 x11 is going to more than double my build times :( Aug 06 05:59:41 ouch Aug 06 06:00:18 anyone know how yocto figures out that a library is needed? (and perhaps a flag to ignore it?) Aug 06 06:00:30 then the answer might be to pull out that lib from the x11 package group and add it to your image manually Aug 06 06:01:16 if it's a build dep then it should be in a DEPENDS somwehere Aug 06 06:01:40 mr_science, it appears to show up from the root_fs rpm install only Aug 06 06:02:59 so you can find poky -name \*.bb\* -o -name \*.inc | xargs grep DEPENDS | grep Xext Aug 06 06:04:18 mr_science, I'm sure my recipe has no DEPENDS on those libs, its finding them at root_fs creation time Aug 06 06:04:37 i meant in the whole recipe tree Aug 06 06:05:04 that would be what you need to add to your install Aug 06 06:05:10 mr_science, if the oracle-java recipe has no depends how would it pick up others? Aug 06 06:05:45 no, a depends on libXext Aug 06 06:06:02 other stuff needs that, not just oracle Aug 06 06:06:27 it only complains when I add that recipe Aug 06 06:06:38 which recipe? Aug 06 06:06:48 my oracle-java.bb Aug 06 06:07:28 and that oracle-java.bb without depends still fails without finding xorg libs Aug 06 06:07:39 right, so oracle-java needs at least one more explicit depend on libXext or else you pull in x11 packagegroup Aug 06 06:08:20 if you add that depend to your java recipe then you don't need the whole packagegroup Aug 06 06:08:21 mr_science, ah, so you're saying write additional recipes just for libXext ? Aug 06 06:08:43 no, just add the depend to your oracle-java recipe Aug 06 06:09:22 or you can do a quick test and add it to your IMAGE_INSTALL in local.conf Aug 06 06:09:26 mr_science, but the only way I got the libXext dependency to resolve was by adding IMAGE_FEATURES += x11 which builds it all Aug 06 06:09:53 but that's a packagegroup that pulls in a bunch of x11 packages Aug 06 06:10:02 it's like a meta-recipe Aug 06 06:10:22 all you need is libXext and maybe whatever it needs Aug 06 06:11:16 so instead of adding the packagegroup to your image features, you add one depend to your oracle recipe Aug 06 06:12:20 ok, I'll try that again, seems like it didn't work before Aug 06 06:16:21 redengin: are you using http://git.yoctoproject.org/cgit/cgit.cgi/meta-oracle-java/ Aug 06 06:16:52 khem, I had to rewrite my own based on it, the current master was broken Aug 06 06:17:50 hmm Aug 06 06:18:10 libxext Aug 06 06:18:13 will clean drop all the cached files? Aug 06 06:18:32 add that to your DEPEND in the oracle recpe Aug 06 06:18:40 the last build using only the RDEPENDS libxext worked, but want to see it cleaned up Aug 06 06:19:22 not sure what the rdep name is, don't have a build handy Aug 06 06:19:50 RDEPENDS will invariably translate into DEPENDS Aug 06 06:20:00 khem, yes, I understand that Aug 06 06:20:28 but since its the rpm install that complains, I'm pretty sure DEPENDS won't work Aug 06 06:20:28 you can do -ccleanall Aug 06 06:20:49 khem, will that keep all the downloaded packages? Aug 06 06:21:09 Yes, in oracle-java case you are repackaging not really building it Aug 06 06:21:24 redengin: no Aug 06 06:21:37 if you wan to keep them then do -ccleansstate Aug 06 06:22:15 I would encourage you to send any fixes you do upstream too. Aug 06 06:22:24 yeah, so they're really all rdeps anyway... Aug 06 06:22:58 khem, will do Aug 06 06:23:27 bitbake -ccleanstate doesn't work, nor bitbake -c cleanstate Aug 06 06:23:44 one more s in the second one Aug 06 06:24:02 ok, second s, still nothing to do Aug 06 06:24:26 bitbake -ccleanssate Aug 06 06:24:35 ah, got it Aug 06 06:24:49 what is it that you want to clean ? Aug 06 06:25:01 if its everything then delete your tmp/ Aug 06 06:25:13 and make sure sstate is hosted outside tmp Aug 06 06:25:19 I want to drop all the built X11 package and make sure just the libxext DEPENDS works Aug 06 06:27:42 perhaps its best to run with a new build dir Aug 06 10:05:38 hi guys, I am trying to add QT5 multimedia video support to my image. I compiled QT5 Multimedia + GStreamer1.0 and now I am trying to get the multimediawidgets/player example working. Aug 06 10:06:08 When I run it, I get following error: GStreamer-CRITICAL **: gst_object_ref_sink: assertion 'object != NULL' failed. No m_videoSink available. Aug 06 10:06:19 Does anyone know what that can be? Aug 06 10:09:59 ideas are also more than welcome :) Aug 06 10:47:40 is anyone there? Aug 06 10:49:38 yes Aug 06 10:59:25 this is kind of unrelated to yocto... but more general linux ... do you know if you can just put a boot.scr in the boot partition? and can I just put mmcargs in it without breaking any of the rest of the environment? Aug 06 10:59:41 Im just kind of an idiot when it comes to linux internals Aug 06 11:04:32 wherever you place boot.scr, you should set the path in boot-loader Aug 06 11:05:03 I really just want to change fb0.rotate to 1 ... Aug 06 11:05:26 although it seems like echoing it directly to the FS does not seem to rotate the screen :( Aug 06 11:05:45 if you are using u-boot, the environment for that is "script", so if you put it to /boot/boot.scr, you should set that one accordingly...e.g. setenv script /boot/boot.scr and saveenv Aug 06 11:05:49 I think i have to add it to the kernel boot args ... but am relatively clueless on how to do that Aug 06 11:06:51 are you using u-boot? Aug 06 11:07:09 yeah Aug 06 11:07:19 sortof ... I mean I built it and have a u-boot.img Aug 06 11:07:29 but all i really did was follow along the readme Aug 06 11:07:44 when it starts booting, you should stop booting when you see this "Hit any key to stop autoboot" Aug 06 11:08:14 and than you check what's the value of script environment: e.g. printenv script Aug 06 11:08:28 that's the path where your boot.scr should be Aug 06 11:08:56 script is the env var that stores the boot.scr Aug 06 11:09:02 location? Aug 06 11:09:47 that's right Aug 06 11:10:02 script is an environment variable Aug 06 11:10:08 ok Im going to try it Aug 06 11:10:13 ill let you know in a few Aug 06 11:29:35 theres no script var ...however there is a `bootenv=uEnv.txt` Aug 06 11:30:03 so can I edit that to set the bootargs? Aug 06 11:30:53 I think this is the variable I want to set Aug 06 11:30:55 mmcargs=setenv bootargs console=${console} ${optargs} root=${mmcroot} rootfstype Aug 06 11:31:40 here is the list of default u-boot env vars http://www.denx.de/wiki/view/DULG/UBootEnvVariables Aug 06 11:31:52 yes, you can set there whatever you want Aug 06 11:32:21 you change it with command setenv, e.g. setenv bootenv abc.txt Aug 06 11:32:40 after you change all you want to change, just type saveenv to save all variables Aug 06 11:33:12 if you type boot, it will boot with settings you changed but only once...next time it will boot with old settings if you didn't save variables Aug 06 11:52:19 03:06:25 AM) ivanstojanovic: hi guys, I am trying to add QT5 multimedia video support to my image. I compiled QT5 Multimedia + GStreamer1.0 and now I am trying to get the multimediawidgets/player example working. Aug 06 11:52:19 (03:06:55 AM) ivanstojanovic: When I run it, I get following error: GStreamer-CRITICAL **: gst_object_ref_sink: assertion 'object != NULL' failed. No m_videoSink available. Aug 06 11:52:19 (03:07:06 AM) ivanstojanovic: Does anyone know what that can be? Aug 06 12:14:08 Hi there, I have a question about a strange boot log message I get when my Kernel loads for a Yocto meta-layer/board support package I’m working on: ‘Failed to execute /init (error -13)’. There is no kernel panic, and the image seems to function properly. I’ve a copy of the full log here: https://www.dropbox.com/s/gp40ewxc90y48r9/messages.txt . Anyone encountered something like this before? or any suggestions on where to start? Aug 06 15:32:44 hi, why am I getting this? :( Error: 'meta-yocto/conf' must be a directory containing local.conf & bblayers.conf Aug 06 15:32:53 I am not using anything from meta-yocto. Why should I have it? Aug 06 15:33:18 can that be hacked out of ./scripts/oe-setup-builddir ? Aug 06 15:34:03 lpapp: looking at the code, that's where it thinks TEMPLATECONF points to Aug 06 15:34:20 so check if that is set to what you think it should be set to Aug 06 15:34:32 bluelightning: I am using TEMPLATECONF=meta-foo/conf . oe-init-build-env Aug 06 15:35:05 well, apparently it's not seeing that because the error message uses $TEMPLATECONF Aug 06 15:35:39 bluelightning: right, so TEMPLATECONF is not handled properly in Yocto. Aug 06 15:35:48 because it does work to create the configs based on my sample. Aug 06 15:35:57 and that is also what I should pass on. Aug 06 15:36:09 somehow it does not end up right in oe-setup-builddir from the look of it :( Aug 06 15:36:49 naturally, this is not a common case that someone removes meta-yocto, so that may explain why this issue has not yet been discovered. Aug 06 15:37:57 bluelightning: correct, it is not passed to the script from the main... Aug 06 15:38:02 I think it should, right? Aug 06 15:38:16 . $OEROOT/scripts/oe-buildenv-internal && \ Aug 06 15:38:16 $OEROOT/scripts/oe-setup-builddir && \ Aug 06 15:38:16 [ -n "$BUILDDIR" ] && cd $BUILDDIR Aug 06 15:38:16 dawd, #define EACCES 13 /* Permission denied */ Aug 06 15:38:37 (or maybe not, I do not know, sorry) Aug 06 15:38:55 your /init (init script in your initrd) isn't chmod 755 by the sounds of it. Aug 06 15:41:12 lpapp: does $BUILDDIR/conf/templateconf.cfg exist in your setup? Aug 06 15:41:41 bluelightning: conf is empty due to the failure. Aug 06 15:42:14 I think if that file exists and is empty that will lead to this exact problem... Aug 06 15:42:57 bluelightning: nothing exists in builddir other than an empty conf dir Aug 06 15:43:08 basically the steps to reproduce: Aug 06 15:43:12 1) rm -rf build/* Aug 06 15:43:45 2) TEMPLATECONF=meta-foo/conf . oe-init-build-env Aug 06 15:44:13 bluelightning: sorry, my fault, I must apologize... Aug 06 15:44:30 never mind, I was running oe-init... directly rather than setup.sh (my one-liner script) Aug 06 15:45:05 it is interesting to see "bash" on core-image-minimal? Aug 06 15:45:17 (that does not look right, does it ... ) Aug 06 15:45:43 it should not really be in there no Aug 06 15:46:27 yeah, what is more: opkg search /bin/bash or opkg search bash returns empty. Aug 06 15:47:01 lrwxrwxrwx 1 root root 12 Jul 29 05:58 /bin/bash -> /bin/busybox Aug 06 15:47:04 ah, right, that is fine then. Aug 06 15:47:07 busybox will use "ash". Aug 06 15:48:09 this is probably done for compatibility with old, but compatible scripts having the bash shebang in a left-over state. Aug 06 15:48:12 I'm surprised to see a /bin/bash link as well.. does busybox provide bash these days, or did someone just add the link in hopes it willw work? Aug 06 15:48:30 AFAIK busybox only provides ash as a functional posix shell.. but I haven't kept up with it recently Aug 06 15:48:42 it is no harm, and it can be positive for some scripts. Aug 06 15:49:13 there is harm if bash isn't actually supported by busybox.. any /bin/bash script will expect to be able to use various bash-isms.. and if they don't work the system behavior could be incorrect.. Aug 06 15:49:41 if the bash-isms are supported, then I have no concern with that.. (like I said, I haven't kept up with busybox work for about the last year or so) Aug 06 15:49:49 no, scripts will not only work if they use bashism; if they do not (like many), then it is a no issue. Aug 06 15:50:04 and if there was no bash symlink, no scripts would work anyway, so it is a reasonable improvement IMHO. Aug 06 15:50:08 in OE, the scripts that don't use bash-isms are supposed to be set to /bin/sh Aug 06 15:50:33 we've gone so far as to remove bash-isms from various scripts in the past, specificaly to ensure they work with the /bin/sh link to busybox Aug 06 15:50:51 well, there is difference between the ideal and real worlds. Aug 06 15:51:19 I am actually working now with a script that has bash shebang, but it does not use bashism. Aug 06 15:51:46 yes, but I need to provide my direct users with software that works.. if THEY want to deviate, thats fine.. I'll warn them -- but it's up to them.. but within YP and OE, I don't want us to be creating a link that will give a false impression for users.. Aug 06 15:51:47 the script would be breaking without the symlink. Aug 06 15:52:12 it is not false impression Aug 06 15:52:12 I'd suggest you change the script then to reference /bin/sh, so it's clear to people in the future it's not intended to be bash specific.. Aug 06 15:52:22 because the intention behind it is not what you think it should be. Aug 06 15:52:44 if I provide /bin/bash to my customers, they will expect it to run bash syntax shell scripts.. Aug 06 15:53:07 if you implement /bin/bash as something else for your product that is fine.. it's a one-off for a device.. but for a general system it's not a good idea Aug 06 15:53:12 the customers will not care as long as it works. Aug 06 15:53:18 mine do Aug 06 15:53:23 fray_: at least the core busybox recipe does not register itself as an alternative for /bin/bash Aug 06 15:53:24 if you do not create the symlink, the customers will not be happy since it will not work. Aug 06 15:53:50 but thats the point, it won't work with the symlink if you use bash-isms.. and if you don't selecting /bin/bash doesn't make sense Aug 06 15:54:29 paulg, Thanks! Yeah the init has does not have execute permission... I'll get that fixed. Aug 06 15:54:52 fray_: like I said, there is a huge difference between real and ideal worlds. Aug 06 15:55:10 one has to be pragmatic here, and not perfectionist. Perfection is the good of enemy which can make customers unhappy. Aug 06 15:55:14 lpapp: I'm pretty sure fray_ knows the difference Aug 06 15:55:32 enemy of good* Aug 06 15:55:34 bluelightning: I am not. Aug 06 15:56:41 in addition, breaking scripts now would be overly unacceptable due to "perfectionism". Aug 06 16:17:08 * mranostay scrolls up Aug 06 23:59:47 Hi Aug 07 00:00:03 I have a patchset adding the fitImage support into Yocto, which ML do I send that one to ? Aug 07 00:02:34 oe-core i think... Aug 07 00:02:52 is it part of the poky repo? Aug 07 00:18:48 mr_science: yes, it's patching meta/ Aug 07 00:19:04 sorry, I am still a bit confused by what goes where Aug 07 00:19:38 if it goes in meta-oe, then it's a de Aug 07 00:19:43 differnt list Aug 07 00:20:14 check yocto.org to be sure Aug 07 00:20:44 it sounds like it should go to openembedded-devel@lists.openembedded.org Aug 07 00:25:45 okay, time to bail back to the hotel and find food Aug 07 00:38:54 Marex__: the meta directory in poky is part of oe-core, yes Aug 07 00:56:50 kergoth: all right then, thanks! **** ENDING LOGGING AT Thu Aug 07 02:59:58 2014