Re: memory_object_lock_request and memory_object_data_return fnord

[Roland McGrath]

> I think that that optimization could be easily integrated into
> pager_memcpy.

There was once a theory that memcpy would use vm_copy for large
page-aligned copies.  There is in fact code for this in libc, but I think
it doesn't get used because it's only the generic code and there are
optimized x86 versions that don't do any page-alignment checks.  Certainly
if memcpy gets inlined it doesn't happen.

I think we should certainly use vm_copy for whole-page copies in
pager_memcpy because of the badly suboptimal behavior you've described.