Re: [PATCH 09/12] hw/s390x/css-bridge: Realize sysbus device before accessing it

[Thomas Huth <thuth@redhat.com>]

On 18/10/2023 16.11, Philippe Mathieu-Daudé wrote:
> sysbus_mmio_map() should not be called on unrealized device.

Can you elaborate? I don't see a sysbus_mmio_map() in this code here...?

  Thomas


> Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
> ---
>   hw/s390x/css-bridge.c | 7 +++----
>   1 file changed, 3 insertions(+), 4 deletions(-)
> 
> diff --git a/hw/s390x/css-bridge.c b/hw/s390x/css-bridge.c
> index 4017081d49..15d26efc95 100644
> --- a/hw/s390x/css-bridge.c
> +++ b/hw/s390x/css-bridge.c
> @@ -95,7 +95,6 @@ static const TypeInfo virtual_css_bus_info = {
>   
>   VirtualCssBus *virtual_css_bus_init(void)
>   {
> -    VirtualCssBus *cbus;
>       BusState *bus;
>       DeviceState *dev;
>   
> @@ -103,19 +102,19 @@ VirtualCssBus *virtual_css_bus_init(void)
>       dev = qdev_new(TYPE_VIRTUAL_CSS_BRIDGE);
>       object_property_add_child(qdev_get_machine(), TYPE_VIRTUAL_CSS_BRIDGE,
>                                 OBJECT(dev));
> -    sysbus_realize_and_unref(SYS_BUS_DEVICE(dev), &error_fatal);
>   
>       /* Create bus on bridge device */
>       bus = qbus_new(TYPE_VIRTUAL_CSS_BUS, dev, "virtual-css");
> -    cbus = VIRTUAL_CSS_BUS(bus);
>   
>       /* Enable hotplugging */
>       qbus_set_hotplug_handler(bus, OBJECT(dev));
>   
> +    sysbus_realize_and_unref(SYS_BUS_DEVICE(dev), &error_fatal);
> +
>       css_register_io_adapters(CSS_IO_ADAPTER_VIRTIO, true, false,
>                                0, &error_abort);
>   
> -    return cbus;
> +    return VIRTUAL_CSS_BUS(bus);
>    }
>   
>   /***************** Virtual-css Bus Bridge Device ********************/