old-autovect-branch/gcc/testsuite/gfortran.dg/g77/970125-0.f

   1 c { dg-do compile }
   2 c
   3 c Following line added on transfer to gfortran testsuite
   4 c { dg-excess-errors "" }
   5 c
   6 C JCB comments:
   7 C g77 doesn't accept the added line "integer(kind=7) ..." --
   8 C it crashes!
   9 C
  10 C It's questionable that g77 DTRT with regarding to passing
  11 C %LOC() as an argument (thus by reference) and the new global
  12 C analysis.  I need to look into that further; my feeling is that
  13 C passing %LOC() as an argument should be treated like passing an
  14 C INTEGER(KIND=7) by reference, and no more specially than that
  15 C (and that INTEGER(KIND=7) should be permitted as equivalent to
  16 C INTEGER(KIND=1), INTEGER(KIND=2), or whatever, depending on the
  17 C system's pointer size).
  18 C
  19 C The back end *still* has a bug here, which should be fixed,
  20 C because, currently, what g77 is passing to it is, IMO, correct.
  21
  22 C No options:
  23 C ../../egcs/gcc/f/info.c:259: failed assertion `ffeinfo_types_[basictype][kindtype] != NULL'
  24 C -fno-globals -O:
  25 C ../../egcs/gcc/expr.c:7291: Internal compiler error in function expand_expr
  26
  27 c     Frontend bug fixed by JCB 1998-06-01 com.c &c changes.
  28
  29         integer i4
  30         integer(kind=8) i8
  31         integer(kind=8) max4
  32         data max4/2147483647/
  33         i4 = %loc(i4)
  34         i8 = %loc(i8)
  35         print *, max4
  36         print *, i4, %loc(i4)
  37         print *, i8, %loc(i8)
  38         call foo(i4, %loc(i4), i8, %loc(i8))
  39         end
  40         subroutine foo(i4, i4a, i8, i8a)
  41         integer(kind=7) i4a, i8a
  42         integer(kind=8) i8
  43         print *, i4, i4a
  44         print *, i8, i8a
  45         end