Re: [Qemu-devel] [PATCH for-2.12] hw/arm/allwinner-a10: Do not use nd_table in instance_init function

[Peter Maydell]

On 4 April 2018 at 18:20, Thomas Huth  wrote:
> The instance_init function of a device can be called at any time, even
> if the device is not going to be used (i.e. not going to be realized).
> So a instance_init function must not do things that could cause QEMU
> to exit, like calling qemu_check_nic_model(_table[0], ...) for example.
> But this is what the instance_init function of the allwinner-a10 device
> is currently doing - and this causes QEMU to quit unexpectedly when
> you run the 'device-list-properties' QMP command for example:
>
> $ echo "{'execute':'qmp_capabilities'}"\
>"{'execute':'device-list-properties',"\
>" 'arguments':{'typename':'allwinner-a10'}}" \
>| arm-softmmu/qemu-system-arm -M mps2-an505,accel=qtest -qmp stdio
> {"QMP": {"version": {"qemu": {"micro": 91, "minor": 11, "major": 2},
>  "package": "build-all"}, "capabilities": []}}
> {"return": {}}
> Unsupported NIC model: lan9118
>
> ... and QEMU quits after printing the last line (which should not happen
> just because of running 'device-list-properties' here).
>
> And with the cubieboard, this even causes QEMU to abort():
>
> $ echo "{'execute':'qmp_capabilities'}"\
>"{'execute':'device-list-properties',"\
>" 'arguments':{'typename':'allwinner-a10'}}" \
>| arm-softmmu/qemu-system-arm -M cubieboard,accel=qtest -qmp stdio
> {"QMP": {"version": {"qemu": {"micro": 91, "minor": 11, "major": 2},
>  "package": "build-all"}, "capabilities": []}}
> {"return": {}}
> Unexpected error in error_set_from_qdev_prop_error() at 
> hw/core/qdev-properties.c:1095:
> Property 'allwinner-emac.netdev' can't take value 'hub0port0', it's in use
> Aborted (core dumped)
>
> To fix the problem we've got to move the offending code to the realize
> function instead.
>
> Signed-off-by: Thomas Huth 
> ---
>  I know, an even cleaner fix would likely be to remove serial_hds and
>  nd_table from the device completely and wire it up from the board code
>  instead - but that's a major rework compared to this simple fix here,
>  which we likely should avoid at this point in time of the hard freeze
>  period.
>



Applied to target-arm.next, thanks.

-- PMM

[Qemu-devel] [PATCH for-2.12] hw/arm/allwinner-a10: Do not use nd_table in instance_init function

[Thomas Huth]

The instance_init function of a device can be called at any time, even
if the device is not going to be used (i.e. not going to be realized).
So a instance_init function must not do things that could cause QEMU
to exit, like calling qemu_check_nic_model(_table[0], ...) for example.
But this is what the instance_init function of the allwinner-a10 device
is currently doing - and this causes QEMU to quit unexpectedly when
you run the 'device-list-properties' QMP command for example:

$ echo "{'execute':'qmp_capabilities'}"\
   "{'execute':'device-list-properties',"\
   " 'arguments':{'typename':'allwinner-a10'}}" \
   | arm-softmmu/qemu-system-arm -M mps2-an505,accel=qtest -qmp stdio
{"QMP": {"version": {"qemu": {"micro": 91, "minor": 11, "major": 2},
 "package": "build-all"}, "capabilities": []}}
{"return": {}}
Unsupported NIC model: lan9118

... and QEMU quits after printing the last line (which should not happen
just because of running 'device-list-properties' here).

And with the cubieboard, this even causes QEMU to abort():

$ echo "{'execute':'qmp_capabilities'}"\
   "{'execute':'device-list-properties',"\
   " 'arguments':{'typename':'allwinner-a10'}}" \
   | arm-softmmu/qemu-system-arm -M cubieboard,accel=qtest -qmp stdio
{"QMP": {"version": {"qemu": {"micro": 91, "minor": 11, "major": 2},
 "package": "build-all"}, "capabilities": []}}
{"return": {}}
Unexpected error in error_set_from_qdev_prop_error() at 
hw/core/qdev-properties.c:1095:
Property 'allwinner-emac.netdev' can't take value 'hub0port0', it's in use
Aborted (core dumped)

To fix the problem we've got to move the offending code to the realize
function instead.

Signed-off-by: Thomas Huth 
---
 I know, an even cleaner fix would likely be to remove serial_hds and
 nd_table from the device completely and wire it up from the board code
 instead - but that's a major rework compared to this simple fix here,
 which we likely should avoid at this point in time of the hard freeze
 period.

 hw/arm/allwinner-a10.c | 12 ++--
 1 file changed, 6 insertions(+), 6 deletions(-)

diff --git a/hw/arm/allwinner-a10.c b/hw/arm/allwinner-a10.c
index 43a3f01..5dbbacb 100644
--- a/hw/arm/allwinner-a10.c
+++ b/hw/arm/allwinner-a10.c
@@ -38,11 +38,6 @@ static void aw_a10_init(Object *obj)
 
 object_initialize(>emac, sizeof(s->emac), TYPE_AW_EMAC);
 qdev_set_parent_bus(DEVICE(>emac), sysbus_get_default());
-/* FIXME use qdev NIC properties instead of nd_table[] */
-if (nd_table[0].used) {
-qemu_check_nic_model(_table[0], TYPE_AW_EMAC);
-qdev_set_nic_properties(DEVICE(>emac), _table[0]);
-}
 
 object_initialize(>sata, sizeof(s->sata), TYPE_ALLWINNER_AHCI);
 qdev_set_parent_bus(DEVICE(>sata), sysbus_get_default());
@@ -91,6 +86,11 @@ static void aw_a10_realize(DeviceState *dev, Error **errp)
 sysbus_connect_irq(sysbusdev, 4, s->irq[67]);
 sysbus_connect_irq(sysbusdev, 5, s->irq[68]);
 
+/* FIXME use qdev NIC properties instead of nd_table[] */
+if (nd_table[0].used) {
+qemu_check_nic_model(_table[0], TYPE_AW_EMAC);
+qdev_set_nic_properties(DEVICE(>emac), _table[0]);
+}
 object_property_set_bool(OBJECT(>emac), true, "realized", );
 if (err != NULL) {
 error_propagate(errp, err);
@@ -118,7 +118,7 @@ static void aw_a10_class_init(ObjectClass *oc, void *data)
 DeviceClass *dc = DEVICE_CLASS(oc);
 
 dc->realize = aw_a10_realize;
-/* Reason: Uses serial_hds in realize and nd_table in instance_init */
+/* Reason: Uses serial_hds and nd_table in realize function */
 dc->user_creatable = false;
 }
 
-- 
1.8.3.1