[Bug-gforth] [bug #52085] Invalid memory address error in valid code. Equivalent code has no error.

[Anton Ertl]

Update of bug #52085 (project gforth):

                  Status:                    None => Invalid                
             Assigned to:                    None => anton                  
             Open/Closed:                    Open => Closed                 

    _______________________________________________________

Follow-up Comment #1:

I could reproduce the problem.  This is a bug in rin.fth (also in rin2.fth and
rin3.fth, but there the bug does not lead to a crash).  The problem is that
PAD points to a little behind HERE, and code is compiled at HERE, moving HERE
and PAD, but not the former PAD stored in IO.  What happens in rin.fth is that
the READ-FILE overwrites the ";S" compiled by the ";" in RIN.  You can check
by doing a SIMPLE-SEE RIN before and after the call to RIN.

Fix: Use the address returned by PAD directly.  It is only valid until HERE
changes (e.g., by compilation, ALLOT, or ",").

    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?52085>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.gnu.org/