Paolo Bonzini <pbonz...@redhat.com> writes: > On 5/3/23 11:12, Alex Bennée wrote: >> From: Fabiano Rosas <faro...@suse.de> >> >> The 'virt' machine uses virtio-net-pci as a fallback when no other >> network driver has been selected via command line. Select VIRTIO_NET >> and VIRTIO_PCI from CONFIG_ARM_VIRT to avoid errors when PCI_DEVICES=n >> (due to e.g. --without-default-devices): >> >> $ ./qemu-system-aarch64 -M virt -accel tcg -cpu max >> qemu-system-aarch64: Unsupported NIC model: virtio-net-pci > > With respect to patches 5-17, very few devices need to be present when > configuring --without-default-devices, and thus need to be "select"ed by > Kconfig. You should select a device only if you cannot even start the > machine without --nodefaults. >
There are some devices that are not explicitly under the scope of -nodefaults, i.e. they are not part of the "default" logic at vl.c, but still some code deep within QEMU uses them as fallback in some situations. > Anything else should be added by hand to configs/ if you use > --nodefaults. In particular, failures of "make check" when configured > --without-default-devices are *test* bugs, not configuration bugs. > Yes, that makes sense, just keep in mind that this have lead to us not testing the --without-default-devices build and people just assuming some devices will always be present. So there's genuine scenarios of us providing a CONFIG that can never be turned off because everything breaks.
