Re: [RFC v1 00/38] arm cleanup experiment for kvm-only build

[Claudio Fontana]

On 3/3/21 7:39 PM, Richard Henderson wrote:
> On 3/3/21 10:20 AM, Claudio Fontana wrote:
>> On 3/3/21 7:17 PM, Claudio Fontana wrote:
>>> On 3/3/21 7:09 PM, Peter Maydell wrote:
>>>> On Wed, 3 Mar 2021 at 17:57, Claudio Fontana <cfontana@suse.de> wrote:
>>>>> One thing I noticed is that tests try to run qemu-system-aarch64 with 
>>>>> accel "qtest" and machine "virt",
>>>>> and the thing tries to create a cortex-a15 cpu model for some unknown 
>>>>> reason.
>>>>
>>>> That is expected. The default CPU type for 'virt' is cortex-a15; if you 
>>>> want
>>>> something else then you need to specify the -cpu option.
>>>>
>>>> -- PMM
>>>>
>>>
>>> I see, I'll experiment a bit thanks.
>>>
>>> I assume changing the default to "max" is out of the question,
>>> and we should instead feed the -cpu option from the tests?
>>>
>>
>> And since we are on topic, should the qemu-system-aarch64 still contain the 
>> cortex-a15 cpu model for some reason?
> 
> The goal is for qemu-system-arm and qemu-system-aarch64 to be as compatible 
> as 
> possible.  That's why the default is the same for both.
> 
> 
> r~
> 

Ah too bad, then I need to rework some code,
and we need to keep lots of stuff that otherwise could have gone away.

It is a bit weird that qemu-system-aarch64 runs with a cortex-a15 model tbh, as 
cortex-a15 is not capable of aarch64.

Thanks,

Claudio