**** BEGIN LOGGING AT Wed Aug 09 03:00:00 2017 Aug 09 12:28:46 Hello! I have a at91sam9263 based board which is not supported by meta-atmel Aug 09 12:29:22 now I have developed u-boot and kernel outside yocto to boot on my board Aug 09 12:29:40 now I want to integrate those two components into yocto Aug 09 12:29:50 shall I modify meta-atmel? Aug 09 12:30:09 wooo: well, probably you want them to integrate into your private layer, not into yocto :) Aug 09 12:30:18 yes... Aug 09 12:30:46 so I have a private layer.... and I add meta-atmel as dependency? Aug 09 12:30:51 wooo: and thats already the basic answer. if there is something in meta-atmel that you still want/need, then keep it. otherwise, get rig of it and just use it as a blueprint Aug 09 12:31:35 well this is true... there is basically nothing I want from meta-atmel Aug 09 12:31:57 then no need to keep it :) Aug 09 12:32:15 well... nice... i haven't look at it that way before :D Aug 09 12:32:20 just create your own meta-cool-bsp layer and pour everything you need into that. Aug 09 12:32:34 word! :D Aug 09 12:32:46 wooo: its just a bit of thinking and looking to reduce copypasta. Aug 09 12:34:43 Thanks LetoThe2nd for opening my eyes :D Aug 09 12:34:53 :-) Aug 09 12:35:36 wooo: once you feel that your layer is in good shape and you're willing to share, please don't forget to submit it to the layer index! Aug 09 12:35:47 I really don't want anything from meta-atmel... I just thought that because we have iMX6 based machine which uses meta-fslc I also needed meta-atmel for AT91 based machine Aug 09 12:37:34 OK LetoThe2nd ... I come from Buildroot background... so my yocto layer and recipes won't be perfect right away :D Aug 09 12:37:48 wooo: well, it always depends. the fsl stuff often brings specialized drivers etc, so it might make sense to keep them as a dependency. the at91 stuff is pretty mature and mainlined, though, so you'll probably be fine without meta-atmel altogether (our own at91 based systems do completely without meta-atmel, too) Aug 09 13:16:50 rburton: was there still a problem with "openssh: Fix key generation with systemd"? I see a revert in mut Aug 09 13:34:56 hum whats the current way to inspect the dependency graph of a recipe? something pulls in libsdl-mixer :-( Aug 09 13:39:03 in the do_preconfigure_prepend() function in my kernel recipe meta-mfr/recipes-kernel/linux/linux-mfr.bb, how do i reference the Aug 09 13:39:18 directory from which the function is running, i.e., meta-mfr/recipes-kernel/linux/linux-mfr.bb? Aug 09 13:40:56 LetoThe2nd: "-g -u taskexp" Aug 09 13:41:01 correction: i.e., meta-mfr/recipes-kernel/linux/ Aug 09 13:41:37 maxin: thx Aug 09 13:49:21 yates: THISDIR ? Aug 09 13:55:10 jku: thanks - that worked! after i asked i found it also referenced in http://www.yoctoproject.org/docs/1.6/kernel-dev/kernel-dev.html#recipe-space-metadata Aug 09 14:02:30 jku: something was breaking musl builds and that was on my hit list Aug 09 14:10:20 The more I try to use Jenkins with Yocto the less fun I seem to have :-( Aug 09 14:10:43 Try cursing at children. That's always fun. Aug 09 14:11:08 Im in the middle of an industrial estate and it would be a long walk to find some. Aug 09 14:12:25 I think I am begining to see why some many people hide all the build stuff inside a docker container away from the Jenkins environment stuff Aug 09 14:19:58 rburton: I see glibc missed out from latest master merge Aug 09 14:22:02 in current mut, don't worry :) Aug 09 14:22:11 something in there broke musl on the AB Aug 09 14:23:01 rburton: ok, I sent few patches just now Aug 09 14:23:09 was wondering if I should send this again Aug 09 14:23:20 i'll grab those before firing a fresh ab run Aug 09 14:24:16 ok Aug 09 14:39:45 i'm getting this error now in my build (attempting to use my own kernel recipe): Exception: FileNotFoundError: [Errno 2] No such file or directory: '/home/yocto-project-x5-r1/sources/poky/build_x5/tmp/sysroots/imx6ul-ebtron-x5/pkgdata/runtime-reverse/kernel-4.1.15-mx6ul+g65ef23a' Aug 09 14:40:27 however, this directory does have a kernel-4.1.15+g65ef23a Aug 09 14:41:15 i see mx6ul in the MACHINEOVERRIDES list for the machine configuration. is this problem related to that? Aug 09 14:42:44 there are no strings with mx6ul directly in the kernel recipe file Aug 09 14:43:48 here is my kernel recipe file: https://da.gd/mU3S -> https://paste.fedoraproject.org/paste/jC4QiQ47ft7gHohX-mIqEQ/ Aug 09 14:54:09 How do I make qt5base-native available to another recipe? I'm trying to add qt support to gstreamer1.0-plugins-bad and I need it to find all of qt5 libraries and binaries. Aug 09 15:02:50 i have a package that built fine under dizzy with autotools-brokensep. now trying to build it under krogoth, configure complains about not being able to finde install.sh and friends. did we have any change in that area? Aug 09 15:26:03 WARNING: /mounthere/fsl-community-bsp/sources/meta-fsl-arm-extra/recipes-kernel/linux/linux-wandboard_3.14.28.bb: Unable to get checksum for linux-wandboard SRC_URI entry Makefile.patch: file could not be found Aug 09 15:26:16 Does that mean it can't find Makefile.patch or the checksum file? Aug 09 15:26:58 there's no such thing as a checksum file for us Aug 09 15:27:02 it can't find Makefile.patch Aug 09 15:27:15 Thanks. Aug 09 15:27:18 it wants to generate a checksum for the file, but there's no file to be checksummed Aug 09 15:27:19 np Aug 09 15:27:22 Gotcha. Aug 09 15:30:23 how does yocto determine the kernel directory name, e.g., the "kernel-4.1.15-mx6ul+g65ef23a" in Exception: FileNotFoundError: [Errno 2] No such file or directory: '/home/yocto-project-x5-r1/sources/poky/build_x5/tmp/sysroots/imx6ul-ebtron-x5/pkgdata/runtime-reverse/kernel-4.1.15-mx6ul+g65ef23a' Aug 09 15:30:39 ? Aug 09 15:32:46 here is the entire error: https://paste.fedoraproject.org/paste/OgKJ--wfOokTAEAkvDyUIg Aug 09 15:33:53 it's just PN and PV, the same that's used everywhere else, including WORKDIR Aug 09 15:33:57 that error makes no sense Aug 09 15:34:15 how so? Aug 09 15:35:06 the runtime-reverse directory does have a kernel-4.1.15+g65ef23a, i've discovered. Aug 09 15:48:26 I'm doing something simple wrong. I have a custom layer with a .bbappend file and it should apply a patch Makefile.patch. However, the bitbake doesn't appear to look in that custom layer directory for Makefile.patch and therefor doesn't find it. Aug 09 15:48:32 Any questions or comments appreciated. Aug 09 15:49:30 your bbappend needs to set FILESEXTRAPATHS so bitbake knows where to find your patch Aug 09 15:49:36 FILESEXTRAPATHS_prepend := "${THISDIR}/${PN}:" typically Aug 09 15:50:11 Ahhhh. Ok. The guide I'm working off is a few years old, it didn't mention that. I'll try that out. Thanks., Aug 09 15:52:53 rburton: Could I specify the full path in the SRC_URI_append? Aug 09 15:53:20 no, set FILESEXTRAPATH as above, and put the patch in a subdirectory like usual Aug 09 15:53:28 kk Aug 09 15:57:01 Hmmm, it seems to want the patch to be in a subdirectory. Here's one of the search paths: /mounthere/fsl-community-bsp/sources/meta-bsp-custom/recipes-kernel/linux/linux-wandboard-3.14.28/mx6dl Aug 09 15:57:34 Makefile.patch is in the parent directory, linux-wandboard-3.14.28 Aug 09 15:57:53 I can move it there easily enough. Aug 09 16:01:34 nvm, it's searching that directory as well and not finding it. wtf. Aug 09 16:10:44 what is the "d" variable used here? does that variable name have some special significance? https://da.gd/seQ4 -> https://paste.fedoraproject.org/paste/E7gEDXwyc9nQm2rl5v-NZQ/ Aug 09 16:15:02 what/where is oe.rootfs in the line "from oe.rootfs import image_list_installed_packages" ? Aug 09 16:17:06 yates: is it the data of packagedata? Guessing though. Aug 09 16:20:54 perhap Aug 09 16:20:56 s Aug 09 16:37:44 so where is image_list_installed_packages built? Aug 09 16:38:15 thta final path/file name is taken from there Aug 09 16:38:46 pkg_info = os.path.join(d.getVar('PKGDATA_DIR', True), 'runtime-reverse', pkg) Aug 09 16:39:04 zomg my patch worked. I'm basically a wizard, Harry. /dance Aug 09 16:39:27 easy boy. Aug 09 16:39:53 \o/ Aug 09 16:40:02 * majuk turns yates into a newt with his Yocto voodoo Aug 09 16:58:22 So here's a more meta question: I have a bunch of lines to add to the .dtb file. Should I patch them into the existing .dtb or make a new .dtb file and reference that somehow? Aug 09 16:58:49 I guess that's more of a style choice? Aug 09 17:03:06 s/dtb/dts Aug 09 17:11:00 is ok to share the same SSTATE_DIR directory between different builds (different target machines and/or different values at local.conf) ? Aug 09 17:13:07 clopez: thats the actual point Aug 09 17:29:27 cool :) Aug 09 18:22:30 can someone please give me a clue where the python variable image_list_installed_packages, as used in exec_python_func(), is generated? Aug 09 18:22:47 each entry is apparently a path Aug 09 18:23:32 from sources/poky/meta/classes/license.bbclass Aug 09 18:28:44 or am i going off into the weeds? Aug 09 19:32:37 ok apparently LOCALVERSION from the kernel recipe is used to generate. once i got that right, it worked. Aug 09 19:33:57 so i got past that error but now, again in do_rootfs()/license_create_manifest(), i'm getting this Exception: FileNotFoundError: [Errno 2] No such file or directory: '/home/yocto-project-x5-r1/sources/poky/build_x5/tmp/sysroots/imx6ul-ebtron-x5/pkgdata/runtime-reversException: FileNotFoundError: [Errno 2] No such file or directory: '/home/yocto-project-x5-r1/sources/poky/build_x5/tmp/sysroots/imx6ul-ebtron-x5/pkgdat Aug 09 19:33:58 a/runtime-reverse/kernel-module-bridgee/kernel-module-bridge Aug 09 19:35:28 rather, /home/yocto-project-x5-r1/sources/poky/build_x5/tmp/sysroots/imx6ul-ebtron-x5/pkgdata/runtime-reverse/kernel-module-bridge Aug 09 19:37:16 there is not kernel-module-bridge directory/file there, but there is in the original working directory files for the recipe from which this was taken, and i have not a clue why they are divergent Aug 09 20:39:53 So I have multiple files to patch in the process of building my kernel. bitbake doesn't appreciate multiple SRC_URI_append directives in my .bbappend file. Can someone assist me with the proper syntax/method here? https://pastebin.com/hEBt75fG Aug 09 20:40:30 With line 3 uncommented, it fails to find Makefile.patch Aug 09 20:41:19 majuk: _append doesn't add whitespace for you Aug 09 20:41:28 so put _append = " file://..." Aug 09 20:41:41 space space Aug 09 20:41:43 Ooooooooooo Aug 09 20:43:55 Yup, I see now, ERROR: linux-wandboard-3.14.28-r0 do_fetch: Function failed: Fetcher failure for URL: 'file://Makefile.patchfile://wandboard_dts.patch'. Unable to fetch URL from any source. Aug 09 20:44:58 Coooool. Thanks yall. Maybe one day I'll have enough of an idea how these directives are used. Aug 09 20:45:44 how is an image under meta-foo/recipes-foo/images/myimage.bb linked with a specific kernel under /meta-bar/recipes-kernel/linux/linux-bar_4.1.bb (e.g.)? Aug 09 20:46:15 yates: kernel is decided by machine usually Aug 09 20:46:32 yates: usually virtual/kernel is required and this is specified in the machine config Aug 09 20:46:48 PREFERRED_PROVIDER_virtual/kernel setting usually decides which recipe provides it for given machine Aug 09 20:48:38 gotcha. Aug 09 21:54:49 Does anyone know how a toolchain and RFS wordsize could get out of sync? I'm doing meta-bitbake and the toolchain and RFS have different wordsizes which is manifesting itself in strange compiler errors Aug 09 21:55:05 bitbake meta-toolchain* Aug 09 21:58:30 ythl: what error do you see Aug 09 22:01:29 @khem: https://stackoverflow.com/questions/45424272/gnu-gcc-bug-when-using-both-sysroot-and-c11 Aug 09 22:02:23 my sysroot wordsize.h is 4, but everywhere else (toolchain, etc.) it's 8 Aug 09 22:04:26 ythl: does it happen with OE toolchain ? Aug 09 22:04:58 it seems to me that its related to how the toolchain is configured/compiled Aug 09 22:06:44 I'm just baffled because premade toolchains work, but as soon as I want to use a sysroot (so I can link to libraries in my RFS), it breaks because the wordsize.h in my sysroot is incorrect Aug 09 22:07:04 Premade toolchains work Aug 09 22:07:14 Unless I add my specific sysroot Aug 09 22:07:19 Then it doesn't work Aug 09 22:08:12 So it seems to me that my sysroot is being generated incorrectly Aug 09 22:08:46 And actually the generated toolchain works fine, so long as I don't use my sysroot Aug 09 22:12:06 However, I know my RFS is good because I am able to load it to my SBC and boot... Aug 10 02:56:34 Today my supervisor was reviewing the project and sez, "I hear a lot about something called 'Yocto Linux'. Why not use that in the next iteration?" Aug 10 02:58:41 I have been looking into using it.. but not with any urgency. Now? I find the material on the Yocto site a little dated and somewhat out-of-order. **** ENDING LOGGING AT Thu Aug 10 03:00:01 2017