[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] [PATCH] Avoid cpu_physical_memory_rw() with a constant is_write argument



On Tue, 18 Feb 2020 at 20:07, Philippe Mathieu-Daudé <philmd@xxxxxxxxxx> wrote:
> I don't understand well cpu_physical_memory*(). Aren't these obsolete?
> They confuse me when using multi-core CPUs.

They sort of are, but there is no simple mechanical replacement
for them -- you need to look at the individual use to see what
address space it should really be using. For instance the cases
in hw/dma/ probably would require the device to be updated to
the new pattern where it takes a MemoryRegion defining what
it should be doing DMA to, and then it can create an AddressSpace
and use that with address_space_*. But that's a bunch of work
on older devices which mostly people don't care very much about.

In theory we could do a textual replacement of cpu_physical_memory*
with address_space_rw(&address_space_memory,...)
but that's usually not the right address space, so I'm not
sure that churn is worthwhile.

thanks
-- PMM

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/mailman/listinfo/xen-devel

 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.