Re: [poky] [PATCH 2/4] conf/fragments: add fragments for qemux86-64 and qemuarm64
Alexander Kanavin <[email protected]> Wed, 4 Jun 2025 11:05:37 +0200
| Newsgroups | org.yoctoproject.lists.poky |
|---|---|
| Message-ID | <CANNYZj-jTjD281=McMDzQtABizNqZ5csbbE9aiJZLRA2U6BX5g@mail.gmail.com> |
On Wed, 4 Jun 2025 at 10:48, Richard Purdie <[email protected]> wrote: > > I think bitbake-config-build should be extended to have > > 'list-machines/set-machine/list-distros/set-distro', and then > > bitbake-setup would call out to that, like it already does for > > fragments. There's already 'show-machines' in bitbake-layers, which > > should be renamed and moved, so that bitbake-layers is less > > over-stuffed with sub-commands. > > Maybe. I do like the idea of being able to express a configuration with > just a list of fragments so a magic mapping of machine/XXX and > distro/XXX does still have some appeal. We could code something: > > BBCONFIG_FRAGMENT_VARIABLE_MAPS = "machine/:MACHINE and distro/:DISTRO" > > to make the code generic? My idea was to do away with fragments for machine/distro, and have 'bitbake-config-build set-machine XXX' write 'MACHINE = 'XXX' into conf/auto.conf. And then bitbake-setup config files would have a particular syntax for that, e.g. replace the current: "qemux86-64-poky": { "description": "Poky reference distro build for qemux86-64", "oe-template": "poky-blank", "oe-fragments": ["core/machines/qemux86-64","yocto/distros/poky","core/yocto/sstate-mirror-cdn","core/qemu/enable-native-sdl"], "targets": ["bitbake core-image-minimal core-image-full-cmdline core-image-sato-sdk core-image-weston"] }, with "qemux86-64-poky": { "description": "Poky reference distro build for qemux86-64", "oe-template": "poky-blank", "oe-distro": "poky", "oe-machine": "qemux86-64", "oe-fragments": ["core/yocto/sstate-mirror-cdn","core/qemu/enable-native-sdl"], "targets": ["bitbake core-image-minimal core-image-full-cmdline core-image-sato-sdk core-image-weston"] }, I don't have a strong opinion either way, perhaps it's best to 'upload discussion data in brain and process', while fixing further do_unpack rework fails :) If you can respond with further points I'd appreciate that! (so I better see the forest) Alex