**** BEGIN LOGGING AT Thu Apr 27 03:00:00 2017 Apr 27 08:27:59 hey Apr 27 08:32:59 i'm still having issues with a deploy task that I defined in a package Apr 27 08:33:22 i'm building a distro that's supposed to run on rpi but also other target platforms Apr 27 08:34:14 for that i created a package that deploys the device tree files into the deploy dir as a task Apr 27 08:35:07 but now when i'm not building for rpi the .bb file fails to expand because apparently some code in there is invalid for other targets Apr 27 08:35:31 sounds like you manually recreated the kernel-devicetree package Apr 27 08:35:33 the weird thing is, the package is not even a dependency for anything on non rpi. so i'm not quite understanding what's going on Apr 27 08:38:17 https://github.com/agherzan/meta-raspberrypi/blob/master/recipes-bsp/bootfiles/bcm2835-bootfiles.bb#L38 Apr 27 08:38:22 this might be what i'm after Apr 27 08:42:08 or not :D Apr 27 08:46:46 aah but https://github.com/agherzan/meta-raspberrypi/blob/master/recipes-bsp/bootfiles/bcm2835-bootfiles.bb#L12 Apr 27 09:00:06 khem: ping Apr 27 11:14:27 is there a way to *set* the IMAGE_FSTYPES based on the machine type? i can do IMAGE_FSTYPES_append_raspberrypi e.g. but i would like to remove everything else from the IMAGE_FSTYPES Apr 27 11:14:57 IMAGE_FSTYPES_raspberrypi Apr 27 11:19:47 rburton, hi there Apr 27 11:20:07 rburton, I am under the impression there is a silent bug Apr 27 11:20:32 it appears building MTD images from scratch (jffs2, ubi) Apr 27 11:21:28 the first build size is 10% bigger than the second build from sstate Apr 27 11:22:06 only for these IMAGE_FSTYPES: .tar.gz is unaffected Apr 27 11:22:35 I am wondering about the many changes to the compressors, ets. Apr 27 11:22:55 ant_work: that's interesting Apr 27 11:24:42 afais the tarball is done with the contents of ${IMGDEPLOYDIR} Apr 27 11:25:02 while for jffs2 etc we process ${IMAGE_ROOTFS} Apr 27 11:28:36 rburton, I'll retest in a few hours Apr 27 11:29:10 the point was, musl images were *bigger* than glibc images Apr 27 12:05:39 i deleted tmp/deploy/images/raspberryi3 - which apparently was a bad idea Apr 27 12:05:53 now i'm getting errors about missing Image-* files Apr 27 12:06:04 is there any way to repopulate that dir? Apr 27 12:06:32 bitbake myimage -C rootfs Apr 27 12:06:37 change myimage as appropriate Apr 27 12:06:51 that will tell it to re-run the rootfs stage and rebuild the images Apr 27 12:08:30 rburton: ah, needded to clean and rebuild linux-raspberrypi Apr 27 13:55:06 when i'm generating a qcow2 image, the grub inside will try to boot from /dev/hda.. when booting qcow2 with kvm via virtio device driver, it will be available as /dev/vda .. is there any way to automatically adjust the value in grub config? Apr 27 14:11:28 ah, SYSLINUX_ROOT Apr 27 20:43:49 domme: pong Apr 27 20:44:14 IMAGE_FSTYPE_rpi = "xyz" Apr 27 20:46:15 rburton: I looked at the gcc7 failure in AB Apr 27 20:46:25 ah yes, basically i wanted to ask about those suffixes.. why does _raspberrypi (or apparently _rpi too) work for all raspberrypi flavours when the machine type is actually raspberrypi3 e.g.? is that string magic or defined.somewhere? Apr 27 20:46:31 khem ^ Apr 27 20:47:30 rburton: I see two problems. the lsb image failures are due to kernel 4.1 needing to be patched for gcc7 and other 1 failure is seen in mips64 where it seems the test harness never sees the qemu come to a login prompt Apr 27 20:48:17 rburton: I built qemumps64 for sysvinit and I could log in no issues. systemd it came to login prompt but logins did not work. Are we testng systemd with poky ? Apr 27 20:48:39 domme: some overrides are SOC_FAMILY specific Apr 27 20:49:36 rburton: can you rerun the mips64 once again ? Apr 27 20:50:52 khem: will grep for that tomorrow Apr 27 20:50:56 Thanks Apr 27 20:51:36 I find the bitbake syntax often easy to read but hard to know when to use what exactly Apr 27 20:51:52 and to google for how things work Apr 27 21:02:04 domme: there is a bitbake manual which can come handy Apr 27 21:02:14 its simple BNF syntax for most part Apr 27 21:02:18 except overrides Apr 27 21:02:27 which is quite innovative Apr 27 21:03:11 Hehe, yeah.. it's still overwhelming when you're just starting Apr 27 21:38:56 khem, hi Apr 27 21:39:45 trying to reproduce those oversized images...it did already happen in the past Apr 27 21:42:01 but now I think there is smthg wrong in the linux-libc-headers that blocks the build of klibc with musl Apr 27 21:42:03 https://pastebin.com/nfXcwhBf Apr 27 21:46:05 in6.h Apr 27 21:53:42 khem: yes we're testing with systemd but not on all arches, and sure i'll refire mips64 now Apr 27 21:55:49 khem: https://autobuilder.yocto.io/builders/nightly-mips64/builds/246 **** ENDING LOGGING AT Fri Apr 28 03:00:01 2017