[tree-ssa] Splitting abnormal edges

[law@redhat.com]

In message <1055777450.9233.334.camel@p4>, Andrew MacLeod writes:
 >I missed the note.. why are they unsafe to remove? We could hack around
 >it I suppose... 
They're unsafe to remove because we don't model assignments from expressions
which might throw properly.

I've got code to handle that, but it triggers an edge insertion bug
(see my last message to gcc@).

 >Andrew
 >	* tree-cfg.c (find_insert_location): Append after last stmt in SRC when
 >	block ends in a control altering stmt.
FWIW, I tried this patch to see if it made any significant difference for
the test I was poking at -- it doesn't help at all.


jeff