On Thu, Jun 16, 2022 at 12:14:01PM +0200, Paolo Bonzini wrote: > On 6/16/22 10:44, Daniel P. Berrangé wrote: > > On Thu, Jun 16, 2022 at 10:30:25AM +0200, Paolo Bonzini wrote: > > > When running from the build tree, the executable is able to find > > > the BIOS on its own; when running from the source tree, a firmware > > > blob should already be installed and there is no guarantee that > > > the one in the source tree works with the QEMU that is being used for > > > the installation. > > I think there is interaction with > > > > > > --firmwarepath=/usr/share/qemu-firmware:/usr/share/ipxe/qemu:/usr/share/seavgabios:/usr/share/seabios:/usr/share/sgabios > > > > as the firmware path listed there will be searched before the local > > build tree. > > > > I agree with your desire to remove the -bios option, but I'm not > > sure it is entirely safe unless logic in qemu_find_file is fixed > > to ignore the global search path when running from the build > > tree. > > Isn't this the same for any other invocation of QEMU, for example in qtest?
Yes, quite likely in fact, so a pre-existing widespread problem and thus not a reason to avoid this proposed change. > I admit I didn't think of the interaction with --firmwarepath, but "if it > hurts, don't do it" might apply here. That is, install compatible firmware > to the path _before_ trying to use a QEMU that specifies that path. I'm mostly thinking of downstream distro developers where they have an installed QEMU but are working on a build from a different QEMU version. We can address this separately though With regards, Daniel -- |: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
