When management tools (e.g. libvirt) query QEMU capabilities,
they start QEMU with a minimalistic configuration and issue
various commands on monitor. One of the command issued is/might
be "query-sev-capabilities" to learn values like cbitpos or
reduced-phys-bits. But as of
v9.0.0-1145-g16dcf200dc the monitor
command returns an error instead.
This creates a chicken-egg problem because in order to query
those aforementioned values QEMU needs to be started with a
'sev-guest' object. But to start QEMU with the values must be
known.
I think it's safe to assume that the default path ("/dev/sev")
provides the same data as user provided one. So fall back to it.
Fixes: 16dcf200dc951c1cde3e5b442457db5f690b8cf0
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Link: https://lore.kernel.org/r/157f93712c23818be193ce785f648f0060b33dee.1719218926.git.mprivozn@redhat.com
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
}
sev_common = SEV_COMMON(MACHINE(qdev_get_machine())->cgs);
- if (!sev_common) {
- error_setg(errp, "SEV is not configured");
- return NULL;
+ if (sev_common) {
+ sev_device = object_property_get_str(OBJECT(sev_common), "sev-device",
+ &error_abort);
+ } else {
+ sev_device = g_strdup(DEFAULT_SEV_DEVICE);
}
- sev_device = object_property_get_str(OBJECT(sev_common), "sev-device",
- &error_abort);
fd = open(sev_device, O_RDWR);
if (fd < 0) {
error_setg_errno(errp, errno, "SEV: Failed to open %s",