Re: [PATCH v2] physmem: allow cpu_memory_rw_debug to write to MMIO devices

[vringar]

On 20/12/2024 21:59, David Hildenbrand wrote:
> Good point, I suspect that will be problematic.

Looking at flatview_write_continue I see no early exit, so maybe it 
would still try to get through everything and work as we are hoping,
but that's why I'd like to write a test for it.
I'm just not sure if it should be a unit test or a QTest and
what examples I could look to copy.

> Maybe address_space_write() should be taught to be able "force write" to 
> ROM instead, so it can just handle everything as part of a single loop. 
> But that's a bit more churn ...

Tbh I'm not sure whether I understand the intricacies of this code well 
enough to write such a patch without significant guidance.

> Building another loop around address_space_write_rom+address_space_write 
> looks a bit suboptimal, but maybe it really is the low-hanging fruit here.

I don't understand what you mean by this
Do you mean keeping the current patch or going back to v1 or a different 
third approach?