EH merge

[Benjamin Kosnik]

Hey Mark.

I'll let Richard give the definitive answer on this question. 

I still have two additional FAILS in libstdc++-v3 testresults. Here is
Jason's analysis:

> The static problems seem to be due to getting wrong offsets from the action
> table; the regions are represented properly, but wind up wrapping the wrong
> instructions.  I suspect a problem with either as or ld.

The plot thicks: we're getting two FDEs for the same base address in the
executable.  One has the right info, the other not; presumably it would be
right for some other function.  This definitely seems like a linker issue.



He'll be back next week and will probably be able to give more info then.

-benjamin