Attempt to create a virtual machine, but incus daemon throws following error messages to its logfile: Unable to run feature checks during QEMU initialization: Unable to locate a UEFI firmware Reproducible: Always Steps to Reproduce: Just create a virtual machine. Actual Results: Nothing happen except error messages. It seems that the upstream move some logic into the file "internal/server/instance/drivers/edk2/driver_edk2.go", the "src_prepare" function in ebuild should be updated to follow the upstream changes. As an ugly and quick workaround, I added a file (/etc/portage/env/app-container/incus) to patch the ebuild file, " function post_src_prepare() { sed -i \ -e "s:/usr/share/OVMF:/usr/share/edk2-ovmf:g" \ -e "s:OVMF_VARS.ms.fd:OVMF_VARS.fd:g" \ internal/server/instance/drivers/edk2/driver_edk2.go || die "Failed to fix hardcoded ovmf paths." } " But I am not sure incus-6.0.1-r1 works or not.
Awesome, thanks! I'll include the fix in 6.5 soon. Most likely LTS (6.0.1) isn't affected, but will have to see on the next version bump at least. Not sure if I've personally booted VMs in 6.0.1.
The bug has been closed via the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=33cbec09c4972302479826cecdfd26b094b2441b commit 33cbec09c4972302479826cecdfd26b094b2441b Author: Joonas Niilola <juippis@gentoo.org> AuthorDate: 2024-09-10 16:13:00 +0000 Commit: Joonas Niilola <juippis@gentoo.org> CommitDate: 2024-09-10 16:13:00 +0000 app-containers/incus: fix OVMF file paths in 6.5 Closes: https://bugs.gentoo.org/939428 Signed-off-by: Joonas Niilola <juippis@gentoo.org> app-containers/incus/incus-6.5-r1.ebuild | 215 +++++++++++++++++++++++++++++++ 1 file changed, 215 insertions(+)
Sometimes I use a patch w/ @GENTOO_BLAH_PATHS@ and sed that so I know when things have changed for this kind of situation.
(In reply to Sam James from comment #3) > Sometimes I use a patch w/ @GENTOO_BLAH_PATHS@ and sed that so I know when > things have changed for this kind of situation. I removed the OVMF file paths completely when bumping 6.5 because qa-sed notified be about them. Just figured upstream had made a better detection for them, since they also pushed support for some env variables determining the OVMF file paths prior. So really we wouldn't need to patch it if I updated init.d/service files, but I don't run 6.5 personally so it's a bit this and this to test everything.
Ah, I see - thanks!
