Re: [Qemu-devel] [RFC PATCH v2 3/6] memory: add memory_region_to_address()

[Scott Wood <scottwood@freescale.com>]

On 04/16/2013 03:25:50 AM, Peter Maydell wrote:
> On 16 April 2013 00:19, Scott Wood <scottwood@freescale.com> wrote:
> > This is useful for when a user of the memory region API needs to
> > communicate the absolute bus address to something outside QEMU
> > (in particular, KVM).
> >
> > Signed-off-by: Scott Wood <scottwood@freescale.com>
> > ---
> > TODO: Use add/del memory listeners later in the patchset, which  
> would
> > eliminate the need for this patch.
> 
> Yes, please do.

I will.  Alex wanted a respun patchset to work with the latest kernel  
API, and I didn't have a chance to address this yet.  It's still RFC.

> > +/* memory_region_to_address: Find the full address of the start of  
> the
> > + *      given #MemoryRegion, ignoring aliases.  There is no  
> guarantee
> > + *      that the #MemoryRegion is actually visible at this  
> address, if
> > + *      there are overlapping regions.
> > + *
> > + * @mr: #MemoryRegion being queried
> > + * @asp: if non-NULL, returns the #AddressSpace @mr is mapped in,  
> if any
> > + */
> > +hwaddr memory_region_to_address(MemoryRegion *mr, AddressSpace  
> **asp);
> 
> A MemoryRegion can appear in more than one AddressSpace (or none at  
> all),
> so I don't think this is a very clearly defined API to put in the
> memory API itself. (It's ok to make that kind of assumption as a user
> of the memory APIs for particular cases, eg in how a memory listener
> callback function behaves. But we shouldn't be baking those  
> assumptions
> into new API functions.)

I'm not sure how it's different from memory_region_to_address_space in  
that regard, other than its ability to gracefully handle the "or none  
at all" case.

How does it work to be in more than one AddressSpace, if there's a  
parent chain?  More than one "as" with the same "as->root"?  Or are you  
referring to aliases?

-Scott