**** BEGIN LOGGING AT Tue Nov 28 03:00:02 2017 Nov 28 04:04:58 hi friends. Im facing build errors with u-boot and linux-ti-staging in yocto morty branch. Nov 28 04:06:08 below is the build error with linux-ti-staging. Nov 28 04:06:40 CHK include/config/kernel.release Nov 28 04:06:40 GEN ./Makefile Nov 28 04:06:40 CHK include/generated/uapi/linux/version.h Nov 28 04:06:40 CHK include/generated/utsrelease.h Nov 28 04:06:40 Using /home/avi/Projects/BoardLabSetup/tmp/work-shared/beagleboard/kernel-source as source for kernel Nov 28 04:06:40 "4.4.91-gb9d9bc78b > /home/avi/Projects/BoardLabSetup/tmp/work/beagleboard-poky-linux-gnueabi/linux-ti-staging/4.4.91+gitAUTOINC+9d9bc78b62-r127a/build/.scmversion echo -gb9d9bc78b" exceeds 64 characters Nov 28 04:06:41 /home/avi/Projects/BoardLabSetup/tmp/work-shared/beagleboard/kernel-source/Makefile:1038: recipe for target 'include/generated/utsrelease.h' failed Nov 28 04:06:41 make[2]: *** [include/generated/utsrelease.h] Error 1 Nov 28 04:07:08 plz help. Nov 28 04:13:35 I was doing a core-image-minimal. Nov 28 07:29:56 good morning Nov 28 08:01:15 whowas darknite Nov 28 08:22:08 where are variables like `libdir` and `bindir` defined? and is there also a variable for `/usr/share`? Nov 28 08:28:13 brrm: datadir, they should be in bitbake.conf Nov 28 08:29:10 thank you nrossi Nov 28 09:03:10 I want to add a CMakeLists.txt to an 3rdparty library in order to build it with cmake. how can I add files to the build directory before building? Nov 28 09:04:23 brrm: prepare it as a patch to the original sources, and incorporate that into your recipe. Nov 28 09:05:44 brrm: then the usual mechanisms kick in properly Nov 28 09:13:47 New news from stackoverflow: madplay error in yocto Nov 28 09:15:31 LetoThe2nd: how do I have to create a patch for an non-existent file? I tried `diff --unified /dev/null CMakeLists.txt` Nov 28 09:17:51 it complains with 'can't find file to patch at input line 3' Nov 28 09:18:06 'No file to patch. Skipping patch.' Nov 28 09:27:18 brrm: i personally always do the patching in git, and then format-patch the commits that i want out. Nov 28 09:30:28 Hi all Nov 28 09:31:37 I have added a new layer (through build/conf/bblayers.conf and .gitmodules of my distro git repo) and I'd like to patch this layer Nov 28 09:38:19 LetoThe2nd: thank you, this works Nov 28 09:59:47 why can't we keep multiple recipes (*.bb files) in a single directory? Nov 28 10:03:28 Is it possible to inherit or bbappend a distro? Nov 28 10:04:22 xtron: we can. Nov 28 10:04:51 sveinse: nop, but you can override variables Nov 28 10:04:58 sveinse: you can create a new distro and require the original conf, then overwrite things Nov 28 10:06:27 I have added a new layer (through build/conf/bblayers.conf and .gitmodules of my distro git repo) and I'd like to patch this layer Nov 28 10:06:30 Anyone ? Nov 28 10:06:43 MajorGrub: "patch this layer"... please elaborate. Nov 28 10:06:49 ok Nov 28 10:07:05 you have to create a custom layer and use bbappend files to override recipes Nov 28 10:07:16 The layer is from a github repo (meta-swupdate) Nov 28 10:07:29 I'd like to modify a .bbappend inside this layer Nov 28 10:07:38 without having to modify the layer directly Nov 28 10:07:56 I don't want to create a branch for this layer and commit my changes Nov 28 10:08:19 MajorGrub: ah ok. well applying patches to the metadata is not a usecase we have in OE Nov 28 10:08:34 Hmmm Nov 28 10:08:36 MajorGrub: see above Nov 28 10:08:43 Duh Nov 28 10:08:54 MajorGrub: you can either do it through some setup script magic that you use, or create your own version of the layer. Nov 28 10:09:01 Ok Nov 28 10:09:04 MajorGrub: you can look here https://github.com/sbabic/meta-swupdate-boards how to customize SWupdate Nov 28 10:09:25 nayfe: thx. Nov 28 10:09:38 pas de soucis Nov 28 10:13:58 New news from stackoverflow: Yocto build error for python and qt5 Nov 28 10:19:29 a `${@bb.fetch2.get_srcrev(d)}', varname='SRCPV'` job fails with `exit code 127, output: /usr/bin/env: 'svn': No such file or directory`. The provided $PATH does not contain the build machine's '/usr/bin'. Is that correct? Nov 28 10:19:59 i.e. to prevent contamination? Nov 28 10:20:04 yes. svn should have been buit by subversion-native and put in the native sysroot Nov 28 10:22:49 do I thus need to depend on subversion-native within the recipe that uses 'svn://' in SRC_URI? Nov 28 10:24:16 hm.. might be because I switched MACHINE Nov 28 10:29:37 just having a svn: url in SRC_URI will cause it to be added as a do_fetch dependency Nov 28 10:29:41 so it should be there already Nov 28 10:33:54 hm... looking at the used PATH, it's filled with cortexa9hf-neon-kls-linux-gnueabi subpaths Nov 28 10:34:28 I guess that should be something among the lines of 'x86_64' right? Nov 28 10:43:05 nayfe and LetoThe2nd, one say no, one say yes. Perhaps a use case is warranted: We have a vendor which provides us with the BSP for custom machines. They also provide and maintain a complete Linux system, and thus have their own distro. However, I need to build on top of this distro, so I would also like to amend changes and additions to the distro. What is the best approach for this? Nov 28 10:43:36 I could manually sync and merge their distro with mine, but that is, well manual work. Nov 28 10:44:03 New news from stackoverflow: libmad error in yocto krogoth 2.1.2 Nov 28 10:44:04 assuming they provide a git repo, worse case is fork it Nov 28 11:44:13 New news from stackoverflow: Is there a way to check the exact list of packages that will be installed in the image in Yocto? Nov 28 13:40:15 at install phase, which is better? to copy the files needed at ${D} or make a hard links in order to save space? Nov 28 13:57:10 Is the do_patch executed by default by bitbake even if there is no "do_patch ()" defined in the .bb of the recipe being cooked ? Nov 28 14:04:06 Tamis: if you want to save space, have a look at rm_work.bbclass Nov 28 14:04:58 kanavin: thanks I'll do Nov 28 14:14:42 MajorGrub: yes as the default classes define that Nov 28 15:48:52 LetoThe2nd: I fixed issue with nm we discuss yesterday (by disabel mac address randomization) now it works fine Nov 28 15:50:49 open-nandra: ah, interesting. where was that feature set? Nov 28 15:51:25 LetoThe2nd: by default in nm (you can disable it by: [device] Nov 28 15:51:25 wifi.scan-rand-mac-address=no) Nov 28 15:51:40 LetoThe2nd: seem this confuse my kernel driver Nov 28 15:51:49 now it works perfectly ;) Nov 28 15:52:01 open-nandra: ok, so its an nm configuration thing, not a compile time issue? Nov 28 15:52:25 LetoThe2nd: seem it's enabled by default and can be disabled by this option not with compile time Nov 28 15:53:03 ok, thanks for reporting back then. are you using a super specific wifi peripheral, or how comes? Nov 28 15:53:23 esp8089 Nov 28 15:53:46 LetoThe2nd: there is out of tree driver for that, there was an attempt to get into kernel but ... Nov 28 15:53:59 * LetoThe2nd can already sense the story. Nov 28 15:54:27 LetoThe2nd: yes you know but customer is waiting for solution .... :D Nov 28 15:58:48 im having difficulty building a uimage kernel, is there a package that provides "mkimage" from u-boot? i cannot seem to find it, apparently "u-boot-mkimage" and "u-boot-mkimage-native" dont do that correctly Nov 28 16:09:52 : KERNEL_IMAGETYPE = "uImage" is not enough ? Nov 28 16:10:14 nayfe: no, ive done that already Nov 28 16:11:18 where is ${D} defined ? Nov 28 16:11:24 uglyoldbob: ok then you have DEPENDS = "u-boot-mkimage-native" Nov 28 16:11:27 ? Nov 28 16:11:50 ${D} is set internally by bitbake as deploy folder Nov 28 16:11:53 nayfe: yes Nov 28 16:12:54 what is your error when exec. mkimage ? Nov 28 16:13:04 maybe pastebin your recipe Nov 28 16:22:50 nayfe: hmm i think it was something dumb i did, i did a cleansstate on all u-boot related recipes and rebuilt the image, now it works Nov 28 16:24:39 uglyoldbob> glad to hear Nov 28 16:45:19 I'm building two images (recovery and normal), i wish to add VOLATILE_BINDS_append conditionally, is there any way to do this ? duplicate volatile-binds recipe ? Nov 28 16:45:36 (same distro/machine) Nov 28 16:48:43 how do i load the vmlinux symbols when debugging a uimage kernel on qemu? (load vmlinux apparently writes things to the qemu system causing bad things) Nov 28 17:10:35 it could be interesting to do something similaor to tmpfile.d for volatile-binds ... Nov 28 17:13:28 On i.MX6 board, when i use ramdisk (recovery system), when i reboot, it waits 30sec watchdog, and when i run standard system, it works fine, any idea ? Nov 28 17:16:15 otavio: hi, any help on this ? Nov 28 17:27:24 nayfe: I didn't follow what you meant Nov 28 17:29:12 I've got a question from someone trying to build an external kernel module from an sdk Nov 28 17:29:41 otavio: i have 2 images, a recovery image that generates a FIT image, and loaded in ram, and a standard image. When i load FIT image in u-boot, and runs it, when i type "reboot", it hangs 30 sec that WDOG resets. When i run standard image, when i type "reboot", it reboot asap without hanging Nov 28 17:29:44 what do we need to inject into the sdk to build a kernel module that wants to see the kernel headers? Nov 28 17:30:10 otavio: i use an external WDOG + wdog1 Nov 28 17:45:18 New news from stackoverflow: How do I generate an ordered lists of the executed tasks when bitbaking a package? Nov 28 17:50:31 nayfe: aren't your initramfs overriding the fdt address space? Nov 28 17:50:43 Crofton|work: kernelsrc iir Nov 28 17:50:47 Crofton|work: kernelsrc iirc Nov 28 17:52:58 TOOLCHAIN_TARGET_TASK_append = " kernel-devsrc" Nov 28 17:53:00 maybe? Nov 28 17:53:08 otavio: i use following uboot script: setenv bootargs '.......'; run findfdt; bootm ${ramdisk_addr_r}#conf@${fdtfile} Nov 28 17:54:23 Crofton|work: I think so Nov 28 17:55:25 nayfe: does not help much; you must check this Nov 28 17:55:33 nayfe: which board? Nov 28 18:00:47 otavio: custom imx6dl, i use standard fdt_high=0xffffffff fdt_addr_r=0x18000000 fdt_addr=0x18000000 u-boot variables, and in FIT its file, i don't set any fdt adress. We have some issues with Stefan Agner concerning fdt relocation, as i need to add a dummy DTB entry between my ramdisk and first DTB entry Nov 28 18:02:06 otavio: https://lists.denx.de/pipermail/u-boot/2016-August/263689.html Nov 28 18:09:00 hmm, I find a kernel-dev package Nov 28 18:45:28 New news from stackoverflow: Yocto using OpenCV Nov 28 20:15:07 i'm confused about ${PN}-dev packages, can someone point me to docs about the behind the scene actions done for recipes included in DEPENDS Nov 28 20:18:08 sum1; that https://www.yoctoproject.org/docs/latest/mega-manual/mega-manual.html#splitting-an-application-into-multiple-packages ? Nov 28 20:18:16 you're mixing up two different things. DEPENDS is build time, ${PN}-dev is runtime, target packages Nov 28 20:21:26 are yocto -dev packages not the same as say ubuntu? where -dev includes libraries&headers Nov 28 20:21:39 In Y recipe, if you need X in runtime, you add RDEPENDS_${PN} += "X-dev" ? Nov 28 20:22:00 If you need it at compile time you add DEPENDS = "X" Nov 28 20:24:00 if a package is required and both compile time and runtime, i would add X to both DEPENDS & RDEPENDS, right? Nov 28 20:24:29 indeed but syntax is not identical :) Nov 28 20:24:55 then, it's up to the Y recipe creator to split correctly packages Nov 28 20:25:15 hmm? Nov 28 20:27:01 if you need something non-standard you need to add files manually, like FILES_${PN}-dev = "${libdir}/toto.la" Nov 28 20:29:54 sum1: DEPENDS lists *recipes*, RDEPENDS *packages*. Nov 28 20:34:54 aren't packages just built recipes? Nov 28 20:36:16 in recipes, you tell where generated files are deployed and in which package Nov 28 20:36:39 so you can create toto recipe, and create toto-a toto-b toto-c packages Nov 28 20:41:48 ahh kk, i think that clears up my understanding somewhat. Nov 28 20:42:08 apologies for the noob questions, thanks for being patient and answering them :) Nov 28 20:42:42 sum1: np ive been there, and still learning a lot ;p Nov 28 20:46:33 Hey folks, I have a quick question about yocto behavor: Github tags with underscores have been mucking up my builds. Nov 28 20:47:09 Is the proper behavior here that the github tag should not have underscores, or should it be handled by legitimize_package_name? Nov 28 20:48:21 austen: what ill effects have you seen? Nov 28 20:48:55 Specifically on rocko with meta-atmel, I'm seeing the kernel modules fail opkg-build Nov 28 20:49:49 The tag linux4sam_5.7 occurs in the package name, which is failing the name check. Nov 28 20:50:28 otherwise use sha version Nov 28 20:50:38 austen: the question would be what is the exact mechanism of the tag getting into the package name? is it set in SRCREV? Nov 28 20:52:08 Hm. Let me check. Nov 28 21:21:04 Er, how do I determine where a specific package is coming from? It looks like the packages are dynamically added and I can't find anything in the meta-layer Nov 28 21:30:35 So, I think I've confirmed that the eSDK seems to be ignoring my BBMASK as set in bblayers.conf Nov 28 21:31:18 if you want to exclude layers from the sdk then you can use SDK_LAYERS_EXCLUDE Nov 28 21:32:01 nathani_: BBMASK probably isn't whitelisted through, it really ought to be Nov 28 21:32:11 it builds sdk fine as well as 'minimal' eSDK Nov 28 21:32:40 but I had to physically delete the BBMASK layers to get my full eSDK to build Nov 28 21:32:48 nathani_: SDK_LOCAL_CONF_WHITELIST += "BBMASK" ? Nov 28 21:33:47 let me try those methods. Thanks rburton, bluelightning Nov 28 21:33:58 austen: enable buildhistory and then build the image, you'll get a graph of the dependencies produced Nov 28 21:34:32 austen: https://wiki.yoctoproject.org/wiki/Technical_FAQ#How_do_I_find_out_why_something_is_in_my_image.3F Nov 28 21:34:47 I'll give it a shot. Thanks! Nov 28 21:45:39 bluelightning, doesn't look like whitelist works Nov 28 21:47:25 nathani_: it should, I'm not sure why it wouldn't :( Nov 28 21:50:00 again, not for the full eSDK, the minimal still works Nov 28 21:55:01 the SDK_LAYERS_EXCLUDE doesn't seem to work either. Although, I would think this is used for layers and can't be applied to specific recipes Nov 28 21:59:05 bluelightning, should I file a bug report somewhere? Nov 28 21:59:48 nathani_: yes please at bugzilla.yoctoproject.org - reproduction instructions would be helpful in the report Nov 28 23:46:21 New news from stackoverflow: where are the sources downloaded for a yocto **** ENDING LOGGING AT Wed Nov 29 03:00:02 2017