**** BEGIN LOGGING AT Thu Apr 14 02:59:58 2016 Apr 14 03:16:18 respawn1: the Yocto Project provides tools to build your own custom Linux distribution, yes Apr 14 03:16:36 respawn1: if you want to try it out I would suggest following our quickstart guide, it's fairly straightforward Apr 14 03:16:46 respawn1: http://www.yoctoproject.org/docs/current/yocto-project-qs/yocto-project-qs.html Apr 14 03:17:50 thank you Apr 14 03:20:53 is it hard to maintain such a system? Apr 14 03:22:26 there's a learning curve, certainly... but it'll be significantly easier to maintain than Linux from Scratch though Apr 14 07:28:42 hi, how can I enable barebox features in a recipe? Apr 14 07:30:10 I have a patch that applies and changes the arch/arm/configs/am335x_defconfig file. But that does not seem to work. Apr 14 07:58:38 hi, I'm updating my tree with latest jethro branch updates and diffing the manifest files shows that libssp0 got removed from my images. Which commit in jethro is causing this? I don't have any hardening features enabled. Apr 14 08:00:34 hi all Apr 14 08:00:49 we are building qt5 with qemuarm Apr 14 08:01:18 and we have enabled egl with qtbaase and trying to run opengl examples but they are not running Apr 14 08:01:54 we are trying to set depth using fbset fbset -depth 24 Apr 14 08:01:54 ioctl FBIOPUT_VSCREENINFO: Invalid argument Apr 14 08:01:59 we get this error Apr 14 08:02:06 Can anybody help Apr 14 08:02:13 we are on the master branch Apr 14 09:32:01 what is the name of that command, that lets me automagically bbappend a reciepe? Apr 14 09:39:36 raykinsella78: I'm not exactly sure what you mean there Apr 14 09:51:47 raykinsella78: bitbake atutomatically apply bbappend to the recipe Apr 14 10:45:43 raykinsella78: recipetool newappend ? Apr 14 10:46:21 Noor: that sounds like the framebuffer driver simply doesn't support the video mode you're trying to set Apr 14 10:57:39 RP we were running qt5 opengl examples that come with qt5 Apr 14 10:58:30 Noor: Try different hardware, I suspect qemuarm only supports 16bit displays Apr 14 10:59:44 RP: yeah on other boards those are working Apr 14 11:00:01 RP: how can we confirm that qemuarm only support 16bit Apr 14 11:12:28 is there any nvidia supported bsp layers, for the jetson tk1 for example? Apr 14 11:15:13 CTtpollard: https://github.com/cubicool/meta-jetson-tk1 and its many forks Apr 14 11:15:20 none of which are supported by nvidia, obviously Apr 14 11:15:54 i'd trust kraj's fork to be usabl Apr 14 11:16:24 cheers rburton Apr 14 11:17:17 nice Apr 14 11:17:25 (yay google :) Apr 14 11:17:47 I saw the cubicool one, just wanted to ask the 'official' question Apr 14 11:18:23 afaik, there isn't an offiicial one Apr 14 11:18:27 harass nvidia for that Apr 14 11:18:37 if enough people ask, they may look at providing one Apr 14 11:23:03 RP: go it, thanks Apr 14 12:30:54 hello Apr 14 12:32:26 kergoth: are you here? Apr 14 12:33:42 Hi, I'm trying to install the yocto plugin into Eclipse (with CDT) Mars.2, a pretty fresh install. Apparently the plugin cannot be installed from the update site, and I don't understand why. Apr 14 12:34:04 davis: he will be available in about 3 hours Apr 14 12:34:11 Any suggestions for what I may be doing wrong or what I can do to debug it? Apr 14 12:34:20 Using this update site: http://downloads.yoctoproject.org/releases/eclipse-plugin/2.0.1/mars/ Apr 14 12:36:39 Noor: many thanks Apr 14 13:11:12 I used kergoth's suggestion to build an appends setup for my kernel config. Apr 14 13:11:44 I've made a gist of the result. However, it does not work. Apr 14 13:11:51 https://gist.github.com/netskink/7700c036eec034281bab009a7f68bc33 Apr 14 13:11:59 any suggestions is much appreciated Apr 14 13:46:12 davis: what are you trying to achieve and what does not work? It is unclear from your question Apr 14 13:47:15 marex-cloud____: i'm trying to get a kernel to build with a modified kernel config Apr 14 13:48:37 i started off doing $bitbake -c menuconfig virtual/kernel, saving my changes and then $bitbake yocto-nsdk-ip-image-sfu Apr 14 13:48:55 the image build will build the kernel, but when I program my device with the resultant image Apr 14 13:49:42 boot it and then $cp /proc/config.gz /tmp and then $gunzip /tmp/.../config.gz I do not see my mods to the kernel config Apr 14 13:50:42 marex-cloud____: Marex did you see those messages? Apr 14 13:51:28 I wonder if kergoth joined or that is just an autojoin after a netsplit Apr 14 13:52:29 You can use zgrep FOO /proc/config.gz , it's quicker Apr 14 13:52:52 hmm.. i did not know that. thanks Apr 14 13:53:34 You provided whole new kernel defconfig to the build system, right? Apr 14 13:54:01 fwiw, this target does not have zgrep, but I recorded this in my notes for future reference. thanks. Apr 14 13:54:09 marex-cloud____: yes, kind of Apr 14 13:54:25 https://gist.github.com/netskink/7700c036eec034281bab009a7f68bc33 Apr 14 13:54:30 zcat |grep works too Apr 14 13:54:40 at the bottom it shows the recipe Apr 14 13:55:05 it shows a config, not a .config or defconfig Apr 14 13:55:12 maybe that is the problem. Apr 14 13:56:12 Ha ... What is in that bbappend file? Apr 14 13:56:39 im looking Apr 14 13:57:41 pfft, i noticed all my edits where I had foo/goo and rewrote them as foo_goo since you can't have / in the file names were tossed Apr 14 13:58:01 i'll have post the bbapend file again. one sec Apr 14 13:59:16 actually it is there. it just reordered it Apr 14 13:59:24 its about 2/3 through the gist Apr 14 13:59:59 meta-JFD-KERNEL/recipes-kernel/linux/linux-stream800_4.1.bbapend Apr 14 14:00:09 it specifies the file as config Apr 14 14:01:07 the last file which is a tree of the meta-JFD-KERNEL fs shows a complete config file in linux-stream800 subdir Apr 14 14:05:47 Hi everyone, I use dart 6 ul som module whis is variscite kit, and i use yocto fido 1.8, and i installed apache, tomcat7, after, i deployed my webapps, but device is locked when my webapp is started, so i research a little and i saw mmcqd and swap process on top, the output of free -h command is swap byte is zero, what is swap process and mmcqd, Apr 14 14:06:21 mmcqd and swap process is took a large cpu percent Apr 14 14:07:53 is mmcqd somethign to do with flash file systems? Apr 14 14:09:19 i have this custom device tree that i fiddle with. each time i change it the whole kernel is fetched and rebuilt. anyway to avoid that, and just build the device tree? Apr 14 14:10:19 ahh, is a kernel thread for managing io queues on the sd Apr 14 14:10:29 (apart from doing it manually using dtc of course) Apr 14 14:17:53 marex-cloud____: you stil around? Apr 14 14:19:15 Hi everyone, I use dart 6 ul som module whis is variscite kit, and i use yocto fido 1.8, and i installed apache, tomcat7, after, i deployed my webapps, but device is locked when my webapp is started, so i research a little and i saw mmcqd and swap process on top, the output of free -h command is swap byte is zero, what is swap process and mmcqd, [17:06] mmcqd and swap process is took a large cpu percent Apr 14 14:26:08 since i have done a .bbapend is there anyway to see if it was run and appended to the results in log? Apr 14 14:29:19 so i was thinking maybe the name of by bbapend file was incorrect, but it appears to match Apr 14 14:29:29 davis: I usually check with bitbake -e and check in the output if the variables touched by the bbappend are correctly modified Apr 14 14:29:35 virtual/kernel resolves to linux-stream800 Apr 14 14:31:15 boucman_work: let me see if that looks correct. in this case im guessing its my SRC_URI since that is the only variable in my appends file Apr 14 14:31:38 probably Apr 14 14:31:53 also look at the comment above SRC_URI it tells you all the places the variable was changed Apr 14 14:31:56 pretty handy Apr 14 14:32:40 im looking Apr 14 14:33:01 i found iit in the -e output, but I dont see my new layer in this area yet Apr 14 14:34:13 perhaps its not beeing applied, I see file://defconfig but I dont see file://config which is from my appends Apr 14 14:39:01 hello. I had some questions yesterday about trying out Yocto even though I am an average linux user, not a programmer and not a developer. I was referred to the quick start guide. I forgot to ask though, can this be used to install a linux system on a laptop? I feel as if it is geared towards less powerful devices. And what would be the advantage of installing one's 'own' system over something like say gentoo? I am truly Apr 14 14:39:01 blown away by what the Yocto Project has accomplished, putting this power into our hands, and if there is even a slight advantage to me, an average user, over a precompiled distro, i will go for it and try it out Apr 14 14:39:13 if it's not there, your bbappend is probably not applied. is your layer correct ? do you use it for something else ? do the names match ? Apr 14 14:40:00 respawn1: I don't think yocto is the right tool for that. It can do it, but it's not really meant as its primary use-case Apr 14 14:40:17 boucman_work: so I modified the filename so instead of config, i called it defconfig. I modifed the recipe so that it used the new name as well. Apr 14 14:40:38 boucman_work: i restarted the build of the image and it detected the kernel needed to rebuild Apr 14 14:41:01 when its done, I'll test it and see what the result is. Many thanks pal, i appreciate your expertise Apr 14 14:41:40 respawn1: its not for laptops. If you want to up your linux skills for a laptop, I suggest gentoo or arch linux. Apr 14 14:41:50 respawn1: but to answer your original question : yes yocto can create images that you can install on a PC, it's just... not the tool I would use. you'll learn a lot about how linux works, though :P Apr 14 14:42:00 fwiw, yocto came from open embedded which came from gentoo Apr 14 14:42:48 archlinux is somewhat different but I prefer a arch distro to gentoo. It is very flexible and well documented. Apr 14 14:43:18 davis, boucman_work thank you Apr 14 14:43:34 you are welcome. good luck! Apr 14 14:43:44 I already use arch, but as per your suggestion I will try out gentoo Apr 14 14:45:18 tripzero: Hi, you're the right person for mraa/upm recipes update, right? Apr 14 14:45:22 respawn1: i was looking for the link relating yocto/openembedded poky to gentoo's portage, but its not here Apr 14 14:45:35 https://en.wikipedia.org/wiki/Yocto_Project Apr 14 14:47:44 I think there is some mention of the relation in the BitBake wiki Apr 14 14:48:23 yes, http://www.aosabook.org/en/yocto.html Apr 14 14:49:24 thank you for the information, it has helped me tremendously Apr 14 14:49:43 heh, your question helped me Apr 14 14:50:20 that aosabook looks like I should read it again now that i have some working knowledge Apr 14 14:50:31 it kind of put me to sleep when I first found it Apr 14 14:51:11 The kernel defconfig is always called defconfig iirc Apr 14 14:52:04 sounds reasonable Apr 14 14:52:20 You cannot relate yocto to portage Apr 14 14:52:21 im programming the board now Apr 14 14:52:29 will see in a bit Apr 14 14:52:31 You also cannot relate poky to portage Apr 14 14:53:27 yeah, i'm just echoing this line from that url, BitBake is a Gentoo-Portage-inspired build tool, used by both the Yocto Project and OpenEmbedded communities to utilize metadata to create Linux images from source. Apr 14 14:53:55 i only know what i read and not the histoory. i was assuming they knew what they were talking about Apr 14 14:54:41 damn, zcat /proc/config.gz | grep USB_ETH Apr 14 14:54:49 CONFIG_USB_ETH is not set Apr 14 14:55:24 uname -a also does not show my custom version append. Apr 14 14:56:40 so if I do bitbake -e virtual/kernel and grep for JFD Apr 14 14:56:49 I see my new layer at top Apr 14 14:57:59 i see in some lines which correspond to my layer.conf Apr 14 14:58:42 bitbake -c virtual/kernel Apr 14 15:00:23 marex-cloud____: i did not do that. After I made the mods to recipe, I did the build of the image and I thought i saw it rebuild the kernel. I was doing -e to see if it appended the variable. i'll do a -c virtual/kernel and then then bitbake imagename and see if it shows up. Apr 14 15:01:05 bitbake -c virtual/kernel fails, do you mean bitbake virtual/kernel? Apr 14 15:01:06 -c clean, sorry Apr 14 15:01:25 ahh that makes sense Apr 14 15:01:50 fwiw, maybe i should mention this Apr 14 15:02:24 just be ware if you do manual changes to the kernel source and do -c clean, it will overwrite them Apr 14 15:02:32 when I do these commands, even before I made my mod, the last line of the laers would always show ":" Apr 14 15:02:42 now it shows meta-JFD-KERNEL = ":" Apr 14 15:03:14 anyway bitbake virtual/kernel is now definitely building. Apr 14 15:03:47 0: linux-stream800-4.1-sure008 do_compile (pid 6436) Apr 14 15:04:07 i alwos wonder about that bit where it appends sue008 Apr 14 15:04:13 You can check in work if the kernel was built with the correct kernel config of course Apr 14 15:04:35 its like this kernel has already got an appends setup Apr 14 15:04:48 You do know git grep, right? Apr 14 15:05:09 but if my layer is last i think mine is applied as well Apr 14 15:05:21 marex-cloud____: kind of, I noted it in my notes. Apr 14 15:05:23 one sec Apr 14 15:06:38 marex-cloud____: actuallly no. I only have entries like this bitbake -e virtual/kernel | grep '^FILE' Apr 14 15:08:11 i do a lot of find . -type f | xargs grep FOO and find . -type d | grep FOO_DIR Apr 14 15:22:51 marex-cloud____: I booted the image after the clean of kernel and rebuild of image Apr 14 15:23:10 and yet again, my USB_ETH mod to /proc/config.gz does not show up. Apr 14 15:25:38 Which version of eclipse should be supported by the adt plugin? I have tried with Mars.2 and that doesn't seem to work. Apr 14 15:27:54 fabo: yes. Apr 14 15:28:05 obiwan, where are you? Apr 14 15:28:36 fabo: I can talk with others about mraa/upm recipe updates Apr 14 15:34:08 fabo, I see your patches. They will likely need to be rebased and can you submit the patch as a github PR (github.com/tripzero/meta-intel-iot-middleware). Apr 14 15:39:15 tripzero: sure. I'll do that in a few minutes. Thanks Apr 14 15:54:46 tripzero: mraa update submitted as a pull request. UPM update isn't needed since your repository had already latest version. Apr 14 16:05:16 zeddii_home: around ? Apr 14 16:05:46 zeddii_home: docker that you have in meta-virt can I create docker images for host using it Apr 14 16:06:07 zeddii_home: or is it only for adding docker infra to yocto based images Apr 14 16:06:59 README tells me its the latter Apr 14 16:07:19 but I am looking for first option Apr 14 16:16:24 when configuring the kernel using menuconfig the using diffstat, soes the fragment.cfg file that is generated just show the difference from your last changes or from the generic kernel configuration? Apr 14 16:20:25 good question Apr 14 16:20:41 Any folks around that know EFI well? Looking for input on bug 9384 Apr 14 16:20:42 Bug https://bugzilla.yoctoproject.org/show_bug.cgi?id=9384 enhancement, Medium, 2.2, eduard.bartosh, ACCEPTED , wic should write startup.nsh for bootimg-efi Apr 14 16:22:15 kergoth: you can just boot kernel directly Apr 14 16:25:26 your approach is sane in the sense that u will get a boot menu Apr 14 16:27:35 * kergoth knows squat about efi and just copied the bits from live in bootimg which are also the same as those in mkefidisk.sh :) Apr 14 16:29:31 kernel itself is a EFI app so thats not needed if you dont want to go through hoops Apr 14 16:53:43 kergoth: yay! i've been looking for you. And regarding UEFI, i did some work on uefi and used to read the tianocore website all the time. I enjoyed the work but not enough to miss it. Apr 14 16:55:44 regarding that bug. i've got a regular minowboard. I used it as a uefi reference board. do you need someone to test the patch? Apr 14 17:43:13 anyone tried building qt5 for multiple machines lately? Apr 14 17:59:39 What's the way to interactively adjust the configuration of a yocto kernel. I have found that bitbake -c menuconfig goes nowhere Apr 14 17:59:56 what does "goes nowhere" mean? Apr 14 18:06:14 kergoth: not where I expected. However, it seems the documentation for the Yocto 1.8.x I am using is no longer online, so I am looking at the wrong docs here. Apr 14 18:06:50 -c menuconfig should open menuconfig. changes made there will end up in the results of that build, but of course wiping tmp will lose the change unless you persist it into a layer Apr 14 18:07:32 kergoth: that did not work for me. It did not end up in the build. This used to be the case, but with Yocto-config-fragments, things seem to work differently. Apr 14 18:08:37 I now ran the kernel_configme, menuconfig, diffconfig steps for Yocto 1.8.1, and I got a .cfg fragment dumped in my work dir. However, I am still figuring out how to get it picked up in an incremental build step. (i.e. compile, deploy) Apr 14 18:08:44 it's supposed to taint the tasks to force them to re-run after menuconfig changes the config, but it's possible you're on an old enough version that it didn't do so. if that's the case, you can bitbake -C configure virtual/kernel after running menuconfig to get it to rerun the later tasks Apr 14 18:09:39 you'd need to create a layer, put a bbappend for the kernel in that layer, put the fragment next to it, add the fragment to SRC_URI in the bbappend, and add to FILESEXTRAPATHS so it can be found Apr 14 18:09:52 recipetool's appendsrcfile is the easiest way to do this, but not sure it existed in 1.8 Apr 14 18:10:27 kergoth: but will that not rebuild the complete kernel as the recipe changed? Apr 14 18:10:50 kergoth: I am in development mode, fine-tuning the kernel in test-cycles against the target. Apr 14 18:11:31 I don't understand the question. if you append it and the bbappend modifies SRC_URI, yes, it'll rebuild, as the sources were changed Apr 14 18:11:57 if not, and you just want to test a quick change without persisting it, then you can use the -C command i gave before with menuconfig Apr 14 18:12:49 kergoth: I am used to fixing bugs, adding code, and modifying the kernel config in the work dir, then compile,deploy, and reboot my target to pick up the new kernel. I would like this cycle to be short. Apr 14 18:13:23 yes, like i said, it's trivial to force bitbake to re-run the build tasks after you make such a change by using -C Apr 14 18:13:37 once you get it nailed down, then you could do the diffconfig and save a fragment to a layer to persist those changes for later Apr 14 18:14:13 kergoth: ok, so after menuconfig I can -c config, -c compile, -c deploy and my changes should be affective? Apr 14 18:14:44 if you really want to do it manually, yes, but that's what -C does, it tells bitbake to taint the specified task which will result in re-running all the tasks that come after it Apr 14 18:18:06 kergoth: -C on menuconfig does not work for 1.8.x: ERROR: Task do_menuconfig is marked nostamp, cannot invalidate this task Apr 14 18:19:31 no, that's not what i said Apr 14 18:20:00 -c menuconfig, then -C configure Apr 14 18:20:18 tainting menuconfig won't do anything, you need to taint a task that's actually run normally so tasks after it get re-run Apr 14 18:20:23 there are no tasks after menuconfig Apr 14 18:23:10 kergoth: ok. My confusion is I thought "configure" configures the kernel purely from the metadata, not from anything in the work directories. So tainting/running configure after menuconfig didn't make sense to me. (It still doesn't, until I understand the underlying mechanisms.) Apr 14 18:24:01 afaik configure writes .config from defconfig if it's not there already and then runs make oldconfig from that, but it varies slightly from one kernel to the next Apr 14 18:24:14 if configure overwrites your changes for this kernel, you could presumably just taint compile instead Apr 14 18:25:07 kergoth: yes, that's with non-Yocto'ized-kernels. That's actually what I am used to. The Yocto config fragment systems seems pretty convoluted though. Apr 14 18:25:41 its true, the linux-yocto build process is definitely complex and I don't know it well, I'm sure others in here could be more helpful with those specifically Apr 14 18:25:59 still the menuconfig to test changes process should work just fine with it Apr 14 18:26:26 kergoth: thanks for helping out Apr 14 18:26:34 no worries, good luck :) Apr 14 18:50:25 Hmm, wonder why buildhistory-diff doesn't diff the pre/post inst/rm scripts rather than showing hte before / after values Apr 14 18:50:35 i htink a diff would be more readable than a hundred lines of before and after Apr 14 19:37:29 I'm getting 'Computing transaction...error: Can't install...no package provides libavformat.so.53'. Where is the version number for this .so determined, because i have a version 54 in my sysroot? Apr 14 20:10:53 Has anyone had issues with Ubuntu 14 trusty's version of matchbox-keyboard showing up very small? Apr 14 20:11:08 I'm running a 64-bit OS. Apr 14 20:11:16 It's so tiny. Apr 14 20:15:29 Look at how small it appears: http://i.imgur.com/wKL7nK3.png Apr 14 20:41:22 I'm thinking we should rework the color handling a bit. Rather than only enabling color on ttys, instead explicitly filter out color codes on fds that aren't ttys Apr 14 20:41:35 then we could add color to messages in more places than just the prefix added in knotty Apr 14 20:42:00 i.e. the new recipe+task and filename prefixing could be colored for easier visual parsing, rather than only coloring NOTE/ERROR/etc Apr 14 20:45:38 My issue appears to be a bug, I've reported it here: https://bugs.launchpad.net/ubuntu/+source/matchbox-keyboard/+bug/1570588 Apr 14 20:49:37 anybody have an idea why I would be getting 'computing transaction...error'...'no package provides libavformat.so.53' from running smart during a do_rootfs? Apr 14 20:50:29 I'm trying to make an oracle jdk 8 recipe Apr 14 20:54:30 that sounds like something is either depending on libav or ffmpeg's library and it's not in the build Apr 14 20:54:51 or that version of it, specifically Apr 14 20:55:34 (unfortunately the library API versions have nothing to do with actual release versions of either, unless you have a very good memory of which API version got released where) Apr 14 20:56:10 but yeah, see if you have FFmpeg or Libav in your build, and if yes which API versions of libraries it provides Apr 14 20:56:22 not sure why it would be a dependency of JDK, though :P Apr 14 20:56:49 since I'm pretty sure nothing in it should depend on libav* Apr 14 20:58:05 hi there Apr 14 20:58:10 (thus I recommend seeing yuor dependency chain in general) Apr 14 20:58:27 weird thing - I'm using some old Yocto branch and when I try to build mpfr it fails to apply a patch Apr 14 20:59:29 poky/meta/recipes-support/mpfr/mpfr-3.1.1/long-long-thumb.patch Apr 14 20:59:32 that's where patch sits Apr 14 21:00:09 but Yocto says patch doesn't exist there and it searches directories like poky/meta/recipes-support/mpfr/mpfr-3.1.1/x86 etc. Apr 14 21:00:27 any idea why the patch is not getting found? Apr 14 21:01:26 JEEB: thanks, i'll take another look. I'm using a bunch of .inc files from meta-oracle-java which I think's my problem Apr 14 21:02:58 and I have a libavformat.so.53 in my sysroot's /usr/lib, which is frustrating Apr 14 21:03:00 it actually searches directory where that patch is stored also! Apr 14 21:03:06 and still no luck Apr 14 21:03:24 err .54 rather Apr 14 21:09:36 karobar: happen to have any prebuilt binaries or packages that were not rebuilt against libavformat.so.54? Apr 14 21:27:20 okay, this was amusing. i accidentally added commercial to LICENSE_FLAGS instead of LICENSE_FLAGS_WHITELIST in local.conf, which marked every recipe as commercial, making no recipes anywhere buildable Apr 14 21:27:22 oops.. Apr 14 21:46:30 Note to self: don't do that. Apr 14 22:08:43 possibly should add a sanity check for that.. that's a var that only really has meaning in recipe context Apr 14 22:31:26 kergoth: thats interesting caveat **** ENDING LOGGING AT Fri Apr 15 02:59:58 2016