Re: [PULL 15/28] hw/intc: arm_gicv3_hvf: save/restore Apple GIC state

[Mohamed Mediouni <mohamed@unpredictable.fr>]

> On 13. Jun 2026, at 13:45, Philippe Mathieu-Daudé <philmd@oss.qualcomm.com> wrote:
> 
> Hi Mohamed,
> 
> On 1/5/26 12:14, Peter Maydell wrote:
>> From: Mohamed Mediouni <mohamed@unpredictable.fr>
>> On HVF, some of the GIC state is in an opaque Apple-provided structure.
>> Save/restore that state to be able to save/restore VMs that use the hardware GIC.
>> Signed-off-by: Mohamed Mediouni <mohamed@unpredictable.fr>
>> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
>> Reviewed-by: Manos Pitsidianakis <manos.pitsidianakis@linaro.org>
>> Message-id: 20260429190532.26538-3-mohamed@unpredictable.fr
>> Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
>> ---
>>  hw/intc/arm_gicv3_common.c         |  1 +
>>  hw/intc/arm_gicv3_hvf.c            | 94 ++++++++++++++++++++++++++++--
>>  hw/intc/arm_gicv3_hvf_stub.c       | 25 ++++++++
>>  hw/intc/meson.build                |  1 +
>>  include/hw/intc/arm_gicv3_common.h |  3 +
>>  5 files changed, 120 insertions(+), 4 deletions(-)
>>  create mode 100644 hw/intc/arm_gicv3_hvf_stub.c
> 
> 
>> diff --git a/hw/intc/arm_gicv3_hvf.c b/hw/intc/arm_gicv3_hvf.c
>> index 22f19d274d..ae881092ea 100644
>> --- a/hw/intc/arm_gicv3_hvf.c
>> +++ b/hw/intc/arm_gicv3_hvf.c
>> @@ -13,6 +13,7 @@
>>  #include "qemu/error-report.h"
>>  #include "qemu/module.h"
>>  #include "system/runstate.h"
>> +#include "migration/vmstate.h"
>>  #include "system/hvf.h"
>>  #include "system/hvf_int.h"
>>  #include "hvf_arm.h"
>> @@ -37,8 +38,13 @@ struct HVFARMGICv3Class {
>>    typedef struct HVFARMGICv3Class HVFARMGICv3Class;
>>  -/* This is reusing the GICv3State typedef from ARM_GICV3_ITS_COMMON */
>> -DECLARE_OBJ_CHECKERS(GICv3State, HVFARMGICv3Class,
>> +typedef struct HVFGICv3State {
>> +    GICv3State parent_obj;
>> +    uint32_t size;
>> +    void *state;
>> +} HVFGICv3State;
>> +
>> +DECLARE_OBJ_CHECKERS(HVFGICv3State, HVFARMGICv3Class,
>>                       HVF_GICV3, TYPE_HVF_GICV3);
>>    /*
>> @@ -668,7 +674,7 @@ static const ARMCPRegInfo gicv3_cpuif_reginfo[] = {
>>  static void hvf_gicv3_realize(DeviceState *dev, Error **errp)
>>  {
>>      ERRP_GUARD();
>> -    GICv3State *s = HVF_GICV3(dev);
>> +    GICv3State *s = (GICv3State *)HVF_GICV3(dev);
>>      HVFARMGICv3Class *kgc = HVF_GICV3_GET_CLASS(s);
>>      int i;
>>  @@ -715,6 +721,86 @@ static void hvf_gicv3_realize(DeviceState *dev, Error **errp)
>>      }
>>  }
>>  +/*
>> + * HVF doesn't have a way to save the RDIST pending tables
>> + * to guest memory, only to an opaque data structure.
>> + */
>> +static bool gicv3_is_hvf(void *opaque)
>> +{
>> +    return hvf_enabled() && hvf_irqchip_in_kernel();
>> +}
> Unfortunately this commit (48396ad6ce9) breaks bisection:
> 
> ../../hw/intc/arm_gicv3_hvf.c:730:29: error: call to undeclared function 'hvf_irqchip_in_kernel'; ISO C99 and later do not support implicit function declarations [-Wimplicit-function-declaration]
>  730 |     return hvf_enabled() && hvf_irqchip_in_kernel();
>      |                             ^
> 1 error generated.

Hi,

My mistake. Will try to see in which rev of the patchset I intro’d this one…
and will see how to avoid it going forward...

On a side note, two bugs affecting HVF right now:

- looks like on M1 there’s an issue affecting detecting 36-bit PAs

think that bug was intro’d by me but will have to look more into it

- and qemu master is broken rn with HVF (and there’s an unmerged patch for
that)

> 
> The method is declared / defined in the following commit (2a3c965516f).
> 
> Please be careful to avoid that in future to avoid manual operations.
> 
> Regards,
> 
> Phil.