linux/meta-virtualization.git/recipes-core, branch scarthgap

libvirtd: don't allow named.service listening on libvirt network interface

2024-03-15T17:17:20+00:00

libvirtd has its own network interface named virbr0, and it using dnsmasq to setup the DNS. the named.service also listen interface and try to bind the port 59 on virtbr0, that cause dnsmasq report following error: dnsmasq: failed to create listening socket for 192.168.122.1: Address already in use Signed-off-by: Xiangyu Chen Signed-off-by: Bruce Ashfield

runv: set COMPATIBLE_HOST

2023-11-06T16:21:12+00:00

runv rdepends on qemu, so update COMPATIBLE_HOST for runv to accord with qemu. Signed-off-by: Kai Kang Signed-off-by: Bruce Ashfield

packagegroup-container: require ipv6 for podman

2023-11-06T16:21:12+00:00

The recipe *podman* requires the distro feature *ipv6*. Using a distro without it causes the build of *packagegroup-container* fails, even if *packagegroup-podman* is not used: ERROR: Nothing RPROVIDES 'podman' (but /build/../work/layers-3rdparty/meta-virtualization/recipes-core/packagegroups/packagegroup-container.bb RDEPENDS on or otherwise requires it) podman was skipped: missing required distro feature 'ipv6' (not in DISTRO_FEATURES) NOTE: Runtime target 'podman' is unbuildable, removing... Missing or unbuildable dependency chain was: ['podman'] NOTE: Runtime target 'packagegroup-docker' is unbuildable, removing... Missing or unbuildable dependency chain was: ['packagegroup-docker', 'podman'] Signed-off-by: Jörg Sommer Signed-off-by: Bruce Ashfield

packagegroup-kubernetes: set COMPATIBLE_HOST

2023-09-15T17:30:14+00:00

packagegroup-kubernetes requires kubernetes which is only compatible with x86_64, arm and aarch64, so set COMPATIBLE_HOST for packagegroup-kubernetes to align with it. Signed-off-by: Kai Kang Signed-off-by: Bruce Ashfield

packagegroup-container: not compatible with mips and riscv

2023-09-15T17:30:14+00:00

The recipe docker-moby which is required by packagegroup-container is not compatible with mips. And it inherits goarch.bbclass, so it is not compatible with riscv32 too. Update COMPATIBLE_HOST accordingly for packagegroup-container. Signed-off-by: Kai Kang Signed-off-by: Bruce Ashfield

oci-image-tools: don't build for riscv64

2023-09-04T03:05:49+00:00

Currently oci-image-tools has a do_compile error for riscv64. The problem could be reproduced by: MACHINE=qemuriscv64 bitbake oci-image-tools So explicitly set COMPATIBLE_HOST here to avoid it building for riscv64. When someone interested in using this recipe for riscv64 fixes the compile issue, this setting could be removed. Also don't build packagegroup-container/packagegroup-kubernetes since they depends on oci-image-tools Signed-off-by: Changqing Li Signed-off-by: Bruce Ashfield

busybox-initrd: auto pv from busybox

2023-05-17T15:44:51+00:00

Not hardcoding the version of busybox from the core layer in this layer is a small improvement for maintenance. But the main motivation is to support the following layer combination without getting a parse error from bitbake: - poky LTS, branch = kirkstone - meta-lts-mixins, branch = kirkstone/go - meta-lts-mixins, branch = kirkstone/rust - meta-virtualization, branch = master Signed-off-by: Adrian Freihofer Signed-off-by: Bruce Ashfield

busybox-initrd: update include

2023-02-01T03:50:41+00:00

OEcore has an update to busybox, and since we include the core recipe directly, we have to update our PV to match. Signed-off-by: Bruce Ashfield

treewide: bulk update patches with status field

2023-01-27T15:32:06+00:00

While the insane.bbclass upstream-status check hasn't been made default, users of meta-virtualization may have it enabled in their distros .. so the effect is the same. We must have this tracking tag in out patches. This is a bulk update to add the tag and silence the QA message. As packages get updated, the normal/routine process of checking the patches will continue, and the status fields may (or may not) get more useful. Signed-off-by: Bruce Ashfield

xen/sysvinit: add hvc0 console only if not there already

2022-07-07T14:55:15+00:00

The base inittab recipe is creating inittab entries for consoles listed in SERIAL_CONSOLES. For qemu, this contains "115200,hvc0" so an entry is created in inittab for it. Prevent to create a second entry if hvc0 is present in SERIAL_CONSOLES. On qemuarm, this solves issues with the console when starting on top of Xen as dom0. Signed-off-by: Bertrand Marquis Signed-off-by: Bruce Ashfield