topic Re: arguments passing. in Software Archive

arguments passing.

Intel_C_Intel — Tue, 08 May 2001 03:15:34 GMT

Why my previous post has no subject? I hope this one works.

------------

Hi, I got this run error at line 2: "call mysub(...)".
line 1 print gives " 37617776 u1: 33423424"
debug at line 30 gives:
d=1 LOC(x0) = 33423424, LOC(x1)=2147348480

It looks like something wrong at passing arguments
Does anyone know what may cause this?
thanks a bunch!

--------------------------------------------------

1: print *, LOC(u0), ' u1=', LOC(u1)

2: call mysub(1,u0,u1, n)

........................

30: subroutine mysub(d, x0,x1,p)

Re: arguments passing.

Steven_L_Intel1 — Tue, 08 May 2001 03:40:32 GMT

I don't quite understand the question, nor do I see a "run error"

I can surmise that at line 30, x1 has been copied onto the stack, but I'd need to see the whole program to understand why.

Steve

Re: arguments passing.

Intel_C_Intel — Tue, 08 May 2001 04:33:01 GMT

Hi Steve

The runtime error is "forrtl: severe (157): Program Exception - access violation"
Here is the code:

............

 type t_bigarrays 
            	double complex	u0(ntotal/p1) 
              	double complex	u1(ntotal/p1) 
 end type t_bigarrays 
 type (t_bigarrays), pointer :: pbigarrays 
 ALLOCATE(pbigarrays) 
 
 call fft(1, pbigarrays%u1, pbigarrays%u0, pnumber%np) 
 
!--------------------------------------------------------------------- 
fft(d, x0, x1, p) 
!--------------------------------------------------------------------- 
 
      implicit none 
      include 'global.h' 
      integer p 
      integer d 
      double complex x0(ntotal/p), x1(ntotal/p)

I can see LOC(u0), LOC(u1) push into stack correctly. But after jump to fft, I can not figure out how those values are popped.
Before passing, u0: 37617776 u1: 33423424
After jump to fft, LOC(x0) = 33423424, LOC(x1)=2147348480

Thanks,

Re: arguments passing.

Steven_L_Intel1 — Tue, 08 May 2001 05:17:58 GMT

Not enough information - please send a ZIP file of your project and a description of the problem to us at vf-support@compaq.com and we'll take a look.

Steve

Re: arguments passing.

Intel_C_Intel — Tue, 08 May 2001 05:29:59 GMT

I'd like to send the whole file, but the first part of code needs special hardware. It won't be able to run at your machine.

thanks anyway.

Re: arguments passing.

Steven_L_Intel1 — Tue, 08 May 2001 05:57:51 GMT

Well, in that case I suggest you start with my article on Access Violations at http://www.compaq.com/fortran/visual/vfn06/index.html#accvio

Please send the project anyway - or at least the source files needed to compile the caller and callee, and the project settings used. I may spot something that doesn't need it to be run.

Steve