Created attachment 2444 [details] log.do_rootfs git: master/df0672b038e6942c97d266af9ca175a06f2b0423 Steps: Add the following lines in conf/local.conf: PREFERRED_VERSION_rpm = "4.11.2" PREFERRED_VERSION_rpm-native = "4.11.2" RPMROOTFSDEPENDS_remove = "rpmresolve-native:do_populate_sysroot" bitbake core-image-minimal for qemuarm Error log: ERROR: Unable to install packages. Command '/buildarea/poky/build-qemuarm/tmp/sysroots/x86_64-linux/usr/bin/smart --quiet --data-dir=/buildarea/poky/build-qemuarm/tmp/work/qemuarm-poky-linux-gnueabi/core-image-minimal/1.0-r0/rootfs/var/lib/smart install -y run-postinsts@all packagegroup-core-boot@qemuarm' returned 1: error: package update-alternatives-opkg-0.1.8+git0+53274f0875-r0.armv5e is intended for a different operating system error: package netbase-1:5.3-r0.armv5e is intended for a different operating system error: package libc6-2.21-r0.armv5e is intended for a different operating system error: package base-files-3.0.14-r89.qemuarm is intended for a different operating system error: package busybox-1.23.1-r0.armv5e is intended for a different operating system error: package libz1-1.2.8-r0.armv5e is intended for a different operating system error: package libuuid1-2.25.2-r1.armv5e is intended for a different operating system error: package initscripts-functions-1.0-r155.armv5e is intended for a different operating system error: package busybox-hwclock-1.23.1-r0.armv5e is intended for a different operating system error: package base-passwd-3.5.29-r0.armv5e is intended for a different operating system error: package init-ifupdown-1.0-r7.qemuarm is intended for a different operating system error: package sysvinit-pidof-2.88dsf-r14.armv5e is intended for a different operating system error: package sysvinit-inittab-2.88dsf-r10.qemuarm is intended for a different operating system error: package modutils-initscripts-1.0-r7.armv5e is intended for a different operating system error: package libkmod2-19+git0+fd56638aed-r0.armv5e is intended for a different operating system error: package libblkid1-2.25.2-r1.armv5e is intended for a different operating system error: package initscripts-1.0-r155.armv5e is intended for a different operating system error: package sysvinit-2.88dsf-r14.armv5e is intended for a different operating system error: package udev-182-r9.armv5e is intended for a different operating system error: package packagegroup-core-boot-1.0-r17.qemuarm is intended for a different operating system ERROR: Function failed: do_rootfs ERROR: Logfile of failure stored in: /buildarea/poky/build-qemuarm/tmp/work/qemuarm-poky-linux-gnueabi/core-image-minimal/1.0-r0/temp/log.do_rootfs.11584 ERROR: Task 7 (/buildarea/poky/meta/recipes-core/images/core-image-minimal.bb, do_rootfs) failed with exit code '1' NOTE: Tasks Summary: Attempted 1964 tasks of which 9 didn't need to be rerun and 1 failed. More details, please see the attachment
I don't know who this should be assigned to. The RPM 4 alternative was implemented by others and needs to be improved to handle better cross-platform support. The issue is that either rpm4 is starting and determining the target == host, or that it simply doesn't know what "armv5e" is. Traditionally rpm4 has a hard coded list of known architectures, and when you deviate from that it gets confused and can't do various validations. RPM5 is controlled by the /etc/rpm/platform file. This file will will contain the entries: <line1>host primary architecture(similar to uname) <lineN>all compatible canonical archs(regex) I'm going to assign this to Saul Wold, as he was the one who introduced the original version of the code. Hopefully he or someone he knows will be able to handle the defect.
I am closing this bug since the original use case for rpm4 was to work with other layers and distro's using other package management tools not with smart, I am not sure this is an actual regression as I just tested this with 1.8+M1 and got the same failure.
(In reply to comment #2) > I am closing this bug since the original use case for rpm4 was to work with > other layers and distro's using other package management tools not with > smart, > I am not sure this is an actual regression as I just tested this with 1.8+M1 > and got the same failure. Verified as per above comment.