While working on a system with multiple BSPs, but a single distro configuration, I struggled with an issue of PR numbers increasing quickly and different behavior from boot to boot.
The systemd one, it looks like there are two alternative solutions. I choose the one that was similar to a connman fix. Add a machine specific configuration recipe. The alternative would have been to adjust the test cases to pass the argument for the longer timeout via the kernel command line. I was worried that this could end up being a requirement for more then just test cases. Note in this patch, I also moved the 'machine-id' file to the config stuff, with the note it appeared to me this made the most sense as someone may end up defining some specific semantics to that file in some configurations. The remaining changes made sense in my configuration, but I'm not entirely sure they are the right solution. The other alternative I toyed with was to declare the package to be machine specific -only- for QEMU BSPs. But I'm not sure this really would have solved my PR number/hash problem. Mark Hatle (4): systemd: Remove items that made this machine (qemu) specific mesa: Remove machine specific append weston: Remove machine specific append gstreamer: Remove machine specific append meta/recipes-core/systemd/systemd-conf.bb | 51 +++++++++++++++++++ ...ange-the-default-device-timeout-to-2.patch | 35 ------------- meta/recipes-core/systemd/systemd_239.bb | 28 ++++------ meta/recipes-graphics/mesa/mesa.inc | 4 +- meta/recipes-graphics/wayland/weston_5.0.0.bb | 4 +- .../gstreamer/gstreamer1.0_1.14.2.bb | 2 - 6 files changed, 64 insertions(+), 60 deletions(-) create mode 100644 meta/recipes-core/systemd/systemd-conf.bb delete mode 100644 meta/recipes-core/systemd/systemd/0001-core-device.c-Change-the-default-device-timeout-to-2.patch -- 2.18.0 -- _______________________________________________ Openembedded-core mailing list [email protected] http://lists.openembedded.org/mailman/listinfo/openembedded-core
