On 11/20/14 14:05, Eduardo Habkost wrote:
On Thu, Nov 20, 2014 at 01:21:18PM -0500, Don Slutz wrote:
[...]
@@ -242,9 +243,16 @@ static void pc_q35_init(MachineState *machine)
pc_register_ferr_irq(gsi[13]);
+ assert(pc_machine->vmport != ON_OFF_AUTO_MAX);
+ if (pc_machine->vmport == ON_OFF_AUTO_AUTO) {
+ no_vmport = xen_enabled();
+ } else {
+ no_vmport = (pc_machine->vmport != ON_OFF_AUTO_ON);
+ }
+
/* init basic PC hardware */
pc_basic_device_init(isa_bus, gsi, &rtc_state, &floppy,
- !pc_machine->vmport, 0xff0104);
+ no_vmport, 0xff0104);
Sorry for suggesting yet another change, but: what about changing
pc_machine->vmport here instead of using a no_vmport variable, so the
actual vmport configuration may be queried by anybody later using the
QOM property? It would even make the code shorter.
I mean:
if (pc_machine->vmport == ON_OFF_AUTO_AUTO) {
pc_machine->vmport = xen_enabled() ? ON_OFF_AUTO_OFF : ON_OFF_AUTO_ON);
}
pc_basic_device_init(isa_bus, gsi, &rtc_state, &floppy,
(pc_machine->vmport != ON_OFF_AUTO_ON), 0xff0104);
Look better to me also. So will adjust the patch.
-Don Slutz
