Issues with limbo DIEs in dwarf2out.c

Richard Kenner kenner@vlsi1.ultra.nyu.edu
Sun Jan 20 19:42:00 GMT 2002


I have the following 16-line Ada program that will cause an ICE on
x86 with -g in limbo_die processing.

 procedure Test_1 is

 type SIMPLE_BOOLEAN_ARRAY_TYPE is array (NATURAL range <>) of BOOLEAN;

 type SIMPLE_BOOLEAN_ARRAY_ACCESS_TYPE is access SIMPLE_BOOLEAN_ARRAY_TYPE;

 Boolean_Array_Access : SIMPLE_BOOLEAN_ARRAY_ACCESS_TYPE;

 Boolean_Array_Access_2 : SIMPLE_BOOLEAN_ARRAY_ACCESS_TYPE;

   begin
         Boolean_Array_Access.all :=
            (Boolean_Array_Access.all OR
             Boolean_Array_Access_2.all);

 end Test_1;

The problem is the new_die call in add_bound.  The tree there is an
expression, in this case MAX_EXPR, so there's no way to get a 
context for it.

I'm confused as to how this is supposed to work.  Can somebody help?



More information about the Gcc mailing list