[RegCNET] Memory allocation fix in postproc.f

[Raymond Arritt]

Dear RegCNET friends,

There is a small bug in postproc.f that may create a memory allocation 
problem in some cases.  In subroutine CALCMSE the working arrays tv, 
p, and z are variable-size arrays but are only declared locally in the 
subroutine. As a result memory is not allocated in a consistent way 
and postproc may fail with a "segmentation fault" error.

One way to fix this is to add the following line to postproc.f 
somewhere in the middle of the type declarations near the beginning of 
the program:

       real tv(nx,ny,nz), p(nx,ny,nz), z(nx,ny,nz)

I put this right before the line that begins with "real vmisdat (etc)" 
but anywhere before the first executable statement is OK.

Then, add the variables tv, p and z to the arguments of CALCMSE in 
each call (there are two calls).  In each case replace the 
continuation line
      &       , zs,sigh,pt,nti,nqvi,npsi,nmsei)
with
      &       , zs,sigh,pt,nti,nqvi,npsi,nmsei,tv,p,z)

Finally, add the variables tv, p, and z to the parameters of CALCMSE 
where the subroutine is defined.  To do this replace
       SUBROUTINE CALCMSE(f2d,f3d,nx,ny,nz,nfld2d,nfld3d
      &       , zs,sigh,pt,nta,nqva,npsa,nmse)
with
       SUBROUTINE CALCMSE(f2d,f3d,nx,ny,nz,nfld2d,nfld3d
      &       , zs,sigh,pt,nta,nqva,npsa,nmse,tv,p,z)

I hope this is helpful.  You may or may not have had crashes with 
postproc because of this.  It seems most likely to make postproc fail 
when the domain size is large.

Regards
Ray Arritt




-- 
  Raymond W. Arritt                           tel:  +1-424-242-0329
  Professor, Department of Agronomy                 +1-515-294-9870
  Iowa State University                       Skype:       rwarritt
  3010 Agronomy Hall, Ames, Iowa 50011 USA    fax:  +1-515-294-2619
   ---------------------------------------------------------------
   "You miss 100% of the shots you don't take." -- Wayne Gretzky