[PATCH] Eliminate duplicate dwarf2 data

[Andi Kleen]

Daniel Berlin <dberlin@redhat.com> writes:
> 
> First, we split the DIES across compilation units, one compilation
> unit for each include file (or some better heuristic), with the DIES
> for things in that include file as children of that compilation unit.
> 
> Then, we do a  checksum of the DIE (It's md5 right now), and
> output the DIE in a linkonce section with the checksum in the name
> (like .gnu.linkonce.wi._DW_192e1d5adfa).

So it'll break when you have a hash collision ?  (even MD5 can have 
collisions...)

-Andi