Michal Prívozník wrote:
On 1/28/26 18:11, Roman Bogorodskiy wrote:
Do not hardcode "/usr/local/share" directory in the configuration file for the firmware and uboot paths.
Use meson's configure_file()/configuration_data() to substitute it with the "datadir" value.
...
This is a bit tricky. Because of the change made to bhyve.conf.in the expected output for check-augeas-libvirtd_bhyve test changes:
libvirt.git/_build/src/test_libvirtd_bhyve.aug:8.2-11.35:exception thrown in test libvirt.git/_build/src/test_libvirtd_bhyve.aug:8.7-.34:exception: Iterated lens matched less than it should Lens: /home/zippy/tmp/libvirt.git/src/bhyve/libvirtd_bhyve.aug:36.13-.43: Last match: /home/zippy/tmp/libvirt.git/src/bhyve/libvirtd_bhyve.aug:32.17-.31: Error encountered at 2:0 (42 characters into string) < "/usr/share/uefi-firmware"\n|=|uboot_path = "/usr/share/u-b>
Tree generated so far: /firmware_dir = "/usr/share/uefi-firmware"
We'd need to make the test output also depend on @DATADIR@ expansion and only after that it can be fed to scripts/augeas-gentest.py. Alternatively, we can just live with the .conf not reflecting the defaults loaded by the driver. BTW: I'm testing these on FreeBSD 15, hence difference in our outputs.
Interesting. For some reason augeas was missing on my system. Regardless of DATADIR expansion the aug file needs to be updated in 1/2. I'll send out a v3 of the 1/2 patch for that. I think I'll stash 2/2 for now, we can live with that indeed. Esp. I can do path substitution on the port/package level anyway. By the way, I think the difference we are seeing are caused probably not by FreeBSD 15 vs 16 difference, but because I configure paths in meson in the same way as the port does that, and you are probably using the default paths.
Michal