[Bug tree-optimization/30911] VRP fails to eliminate range checks in Ada code

[baldrick at free dot fr]

------- Comment #64 from baldrick at free dot fr  2008-03-30 16:02 -------
Subject: Re:  VRP fails to eliminate range checks in Ada code

> > But there are no checks on the array access.  It looks like the f-e
> > doesn't generate them in the first place (as opposed to fold or gigi
> > making a mistake).
> 
> This is as documented in the GNAT manual, section 3.2.4 Validity Checking.
> You need to pass -gnatVs to have them.

Consider the following test case:

procedure Overflow (X : Positive) return Integer is
   A : array (Positive) of Integer;
   pragma Import (Ada, A);
begin
   A (X) := X;
end;

(for which no checks are generated).  In the user guide

 GNAT GPL User's Guide 
 The GNAT Ada Compiler
 GNAT GPL Edition, Version 2007
 Document revision level 1.422
 Date: 2007/03/30 03:28:29

in section 3.2.4 it says:

 In GNAT, the result of such an evaluation in normal default mode is to
 either use the value unmodified, or to raise Constraint_Error in those
 cases where use of the unmodified value would cause erroneous execution.
 The cases where unmodified values might lead to erroneous execution are
 case statements (where a wild jump might result from an invalid value),
 and subscripts on the left hand side (where memory corruption could occur
 as a result of an invalid value).

This testcase is a clear example of erroneous execution, being of the type
explicitly mentioned.  So according to section 3.2.4 Constraint_Error should
be raised "in normal default mode".  Yet this is not the case.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30911