I have a fortran code calling a C function. In Fortran, the 6th
parameter in the calling sequence is declared as,

INTEGER     ISIZE20
PARAMETER   (ISIZE20 = 20)

In C it was declared as int *sizePtr.

The problem was that when *sizePtr is assigned to, a SEGV was generated.
I compared &ISIZE20 in Fortran and sizePtr in C. sizePtr is not pointing
to the address of ISIZE20.

It seems that this is caused by ISIZE20 is a constant PARAMETER.
But how does C function get to know this?

Thanks,

Think of a parameter as being a compile-time construct.  Since you
mention C, I'd draw analogy with a #define in the C preprocessor.
If you did a

  #define ISIZE20 20

then what would you say about the address of ISIZE20?  The question
is meaningless - it doesn't have an address.

Depending on how it is used, the compiler *MIGHT* find it convenient
to define some address that contains the value 20 and use that in
some contexts.  But this is certainly not guaranteed to be a unique
address.  There might be multiple addresses where the value is
stored (for example, one for each of several places where the
parameter is used), or there might be none.

    As Richard Maine pointed out, named constants do not have addresses in
some Fortran implementations.  In some other implementations, named constants
DO have addresses but they are located in write-protected (AKA read-only)
areas of memory.  If you attempt to modify them, you will get weird and FATAL
run-time errors.  I once did what you did in DEC Fortran under VMS back in
1985.  I got a {*filter*} program crash along with an error message something like,
"Attempt to modify read-only program section".  It took me a while to figure
out exactly what mistake I had made.

--
----------
Sincerely,

Elmbrook Computer Services Voice Phone: (414) 783-5869
17130 W. Burleigh Place  Fax Phone:   (414) 783-5928
Brookfield, WI   53005  Disclaimer:  These opinions are mine alone.
USA    They do NOT represent any organization.