> It seems like you also need to describe RO segments, at least those which
> you cannot intuit from the exec file, though you may not have access to all
> the necessary info (or you could just dump the segment into the core file).
> 

Additionally, a RO segment may have been writable at some point in the past
(the run-time link-editor is one example of an agent that is liable to do
manipulations like this). I suppose you could try and find out if this is
the case by looking at shadow object chains for object that are ultimately
backed by the vnode pager.

Anyway, an object format that natively support multiple segments would be
infinitely better suited to unify dumps and executables.

-pk

------------------------------------------------------------------------------