Bug in CA client VAX architecture conversion routines

[hill@luke.atdiv.lanl.gov (Jeff Hill)]

Hello,

This message is directed to persons running EPICS on the VAX. 


Chris Mayer (cjm@mail.ast.cam.ac.uk) has discovered a bug in the 
channel access client VAX architecture dependent data conversion 
routines. The problem will show up if you are using one of types
dbr_short, dbr_enum, or dbr_long.


This is perhaps cryptic and confusing but ok:

    for(i=0; i<num; i++){
	*d++ = *s++;
    }

However this wasnt:

    for(i=0; i<num; i++){
        *d++ = ntohs( *s++ );
    }

The problem is that ntohs() is a macro and there is an 
unexpected side effect where the ++ operator is executed twice
so we get one byte correctly and the other byte is garbage:

#  define ntohs(SHORT)\
   ((unsigned short)(((short)(SHORT))<<8 | ((unsigned short)(SHORT))>>8 ))

I will apply this fix to 3 instances in convert.c for EPICS 3.12.

Jeff

______________________________________________________________________
Jeffrey O. Hill			Internet	johill@lanl.gov
LANL MS H820			Voice		505 665 1831
Los Alamos, NM USA 87545	FAX		505 665 5107