Re: [Qemu-devel] [PATCH v6 4/4] hw/intc/arm_gicv3_kvm: Reset GICv3 cpu interface registers

[Vijay Kilari]

Hi Peter,

On Thu, Dec 1, 2016 at 3:40 PM, Vijay Kilari <address@hidden> wrote:
> On Wed, Nov 30, 2016 at 10:29 PM, Peter Maydell
> <address@hidden> wrote:
>> On 30 November 2016 at 16:23, Vijay Kilari <address@hidden> wrote:
>>> On Mon, Nov 28, 2016 at 10:05 PM, Peter Maydell
>>> <address@hidden> wrote:
>>>> Still I would prefer it if we did this with the same
>>>> mechanism for both TCG and KVM. A generic mechanism for
>>>> "let the CPU reset trigger reset of many other devices in the
>>>> system" isn't widely useful because real hardware doesn't
>>>> have that kind of action-at-a-distance behaviour.
>>>
>>> To make direct call from arm_cpu_reset() to reset CPUIF,
>>> I could not find a way to get GICv3CPUState from CPUARMState or
>>> ARMCPU struct.
>>
>> You don't want to directly call from arm_cpu_reset().
>> Coprocessor regs registered via cpregs can have
>> reset functions, which get called automatically.
>> This is what the TCG gicv3 code already does to reset
>> the CPU i/f, the relevant code just needs to be
>> arranged so it's used for KVM too.
>
> Yes, the reset functions of cpregs get CPUARMState as parameter
> and still we cannot fetch GICv3CPUState from it.

I propose to add new variable to CPUARMState to store
GICV3CPUState to able to access when cpregs reset is called.
Is it ok?

>
> The TCG code in arm_gicv3_cpuif.c is rely on el_hook to get
> GICv3CPUState.
>>
>>> Any idea how to get GICv3CPUState?
>>>
>>> In  hw/intc/arm_gicv3_cpuif.c implementation,
>>> el_hook function is registered to fetch GICv3CPUState
>>> from CPUARMState struct, but it is for TCG
>>
>> Yes, you don't need the el hook.
>
> Without this is there a way to get GICv3CPUState for KVM?
> I am not familiar with this code.
>
>>
>> thanks
>> -- PMM