Re: [qemu-s390x] hw/s390x/ipl: Dubious use of qdev_reset_all_fn

[Philippe Mathieu-Daudé]

On 5/24/19 8:04 PM, David Hildenbrand wrote:
> On 24.05.19 19:54, Philippe Mathieu-Daudé wrote:
>> Hi Christian,
>>
>> I'm having hard time to understand why the S390_IPL object calls
>> qemu_register_reset(qdev_reset_all_fn) in its realize() method, while
>> being QOM'ified (it has a reset method).
>>
>> It doesn't seem to have a qdev children added explicitly to it.
>> I see it is used as a singleton, what else am I missing?
>>
>> Thanks,
>>
>> Phil.
>>
> 
> Looks like I added it back then (~4 years ago) when converting it into a
> TYPE_DEVICE.
> 
> I could imagine that - back then - this was needed because only
> TYPE_SYS_BUS_DEVICE would recursively get reset.

Thanks for the quick response :)

> Did you try removing it, to see if anything breaks?

>From build POV it is OK, but I have now idea of the effects with KVM.

I don't know how to test on s390x systems, but luckily Patchew/s390x is
back up so I'll try my luck with a RFC patch, but I'm not sure the
default tests cover this specific device uses.

Regards,

Phil.