Re: [PATCH v3] memory: Directly dispatch alias accesses on origin memory region

[Mark Cave-Ayland]

On 19/04/2021 22:11, Philippe Mathieu-Daudé wrote:

> > My patch might not be the proper fix, but we need to figure out how
> > to avoid others to hit the same problem, as it is very hard to debug.
> >
> > At least an assertion and a comment.
>
> Something like:
>
> -- >8 --
> diff --git a/softmmu/memory.c b/softmmu/memory.c
> index d4493ef9e43..e031ac6e074 100644
> --- a/softmmu/memory.c
> +++ b/softmmu/memory.c
> @@ -1442,6 +1442,7 @@ MemTxResult memory_region_dispatch_read(MemoryRegion *mr,
>      unsigned size = memop_size(op);
>      MemTxResult r;
>
> +    assert(!(mr->alias && !mr>alias_offset)); /* Use AddressSpace API
> instead */
>      if (!memory_region_access_valid(mr, addr, size, false, attrs)) {
>          *pval = unassigned_mem_read(mr, addr, size);
>          return MEMTX_DECODE_ERROR;

AFAICT the dispatch functions don't handle rom/ram or aliases so you might need to go 
a bit stronger. The check is further complicated by the fact that you can have 
rom/ram devices which do define the underlying mr->ops.

I'm wondering for memory_region_dispatch_read() if this would work?

   assert(!mr->alias && !memory_access_is_direct(mr, false));


ATB,

Mark.