BUG: Repeated EQUIVALENCE's Cause GP Fault During Compile

• Microsoft FORTRAN for OS/2, version 5.1

SYMPTOMS

Code compiled with Microsoft FORTRAN 5.1 that contains repeated EQUIVALENCE statements using the same variables may cause the compiler to generate a Trap D protection violation during pass two of the compile.

The test case requires that the same variables be used in three different EQUIVALENCE statements effectively equivalencing three different arrays indirectly.

RESOLUTION

Simplify EQUIVALENCE statements to directly equivalence arrays instead if indirectly through equivalencing identical elements to the same separate variable.

STATUS

Microsoft has confirmed this to be a problem in FORTRAN version 5.1. We are researching this problem and will post new information here in the Microsoft Knowledge Base as it becomes available.

MORE INFORMATION

The following code can be used to demonstrate the problem.

Sample Code #1

The following code reproduces the problem:

      CHARACTER*1 ASTDAT(70),HEDR01(70),HEDR07(70)

      EQUIVALENCE (HEDR01(1),NUMRES), (HEDR01(3),CUSIS),
     1        (HEDR01(11),CRADTS)

      EQUIVALENCE (HEDR07(1),NUMRES), (HEDR07(3),CUSIS),
     1        (HEDR07(11),PFDESS)

      EQUIVALENCE (ASTDAT(1),NUMRES), (ASTDAT(3),CUSIS),
     1        (ASTDAT(11),TICKES)

      stop
      end

Sample Code #2

The following code illustrates the solution:

      CHARACTER*1 ASTDAT(70),HEDR01(70),HEDR07(70)

      EQUIVALENCE (ASTDAT(1),HEDR01(1),HEDR07(1),NUMRES)
      EQUIVALENCE (HEDR01(3),CUSIS),
     1            (HEDR01(11),CRADTS,PFDESS,TICKES)
      stop
      end