Conference turris::decc_bugs

1269.0. "Compiler loop with 7.1 VMS AXP 5.5-002 C" by COMICS::EDWARDSN (Dulce et decorum est pro PDP program) Fri Mar 07 1997 10:59

As the title suggests, compiling the following puts the compiler into a loop.
It works on VMS 7.0 DEC C 5.5-002 on VAX, but fails on AXP 7.1, 5.5-002

Any suggestions?


Here it is:

/*** COMPILES OK IF STRING.H IS NOT INCLUDED ***/
#include   <string.h>

#include   <types.h>


struct BI_rules_message_struct
{
   unsigned char           message_type;
   unsigned char           header_len;
   unsigned char           rule_data [1];                  /*** MAKES NO
DIFFERENCE IF THIS ARRAY IS LARGER THAN 1 ***/
};
typedef struct BI_rules_message_struct BI_rules_message_rec;


void bi_client_create_rules_list
(
   unsigned char *rules_message
)
{
   BI_rules_message_rec *bi_rules_message;
   unsigned char *message_ptr;
   unsigned char type_terminator_flag;
   size_t size;

   /* Cast message pointer to a rules message structure */

   bi_rules_message = (BI_rules_message_rec *) rules_message;

   /* Set message pointer to the 1st byte of the variable length part of the
message */

   message_ptr = bi_rules_message->rule_data;

   /* Extract rule type terminator flag from the message */

   size = sizeof type_terminator_flag;

/*** OR IF MEMCPY IS NOT CALLED ***/
   memcpy ( & type_terminator_flag, message_ptr, size );

/*** OR IF THIS WHILE LOOP IS COMMENTED OUT ***/
   while ( type_terminator_flag != 0 )
   {
   }

   return;
}

    
    I have confirmed this is a problem in V5.5.  The compiler was hanging
    in the back end.  Our V5.6 compiler, which will be available for
    field test very soon, does not hang.  I will send mail to the
    back-end people to see if they can confirm a fix.
    
    It seems that you have several workarounds.  Is that enough to keep
    the customer happy?
    
    						Ed
    
    

    
    To follow-up on my .1.  I understand that two fixes were made to
    the back-end module in question that involved potential hanging. 
    Looks like the problem has certainly been corrected in V5.6.
    
    Again, we are sorry for the difficulty this problem has caused the
    customer.
    
    					Ed