I have a large linux elf program that doesn't work.  Well, ...
large parts of it work, but several different little errors/problems
happen (only on NetBSD).  I am wondering if the image is loaded right.

Should linux binary emulation mean that the text, data, and bss are
in the same place on both systems (they are)?  What about the 
ld.so interpreters are they loaded in the same place, are the libraries,
is the brk and "mmap" areas?  Should the interpreters get the same 
arguments?  So the basic question is should the memory images be
identical and are they?  I don't have a linux system (yet) do examine
this with.  I did look at the sources a bit.  It is clear that the
interpreter under linux gets a few extra vaules pushed onto its
stack: uid, euid, gid, and egid.  NetBSD loads the interpreter
at <start of data> + 0x10000000 (0x10000000==MAXDSIZ).  I dont
think linux does.