Pro*C and Oracle 8i, fetch coredumps

From: Christian Bantzer <bantzer_at_webpre.com>
Date: Mon, 15 Nov 1999 08:59:23 -0800
Message-ID: <G6XX3.159$B5.47206@WReNphoon3>

Hi,

We are in the process of migrating from Oracle 7.3 to 8i. We are running into some problems with Pro*C programs. I have not much of a clue when it come to Pro*C, neither does our DBA.

The problem manifests itself by coredumping, when a query result is fetched into a struc, only when the query returns more than a certain number of rows. The struc is large enough to hold all rows, which lead me to suspect some buffer size issue between Oracle and the C program. But again I don't really have a clue on Pro*C.

If anyone has experienced similar problems, I would appreciate hearing about them. Also, on the off chance that somebody would be willing to look at the code I include it below.
Compiled with cc: WorkShop Compilers 5.0 98/12/15 C 5.0 on a Sun Solaris.

It might be a compiler issue too, I am just using the sample makefile provided by Pro*C.

Any hints where to look next would be great.

Thanks

   Christian

Pro*C code follows (posted via remarque, I hope they don't put in line breaks, just ignore if they do)

#include <stdio.h>
#include <string.h>
#include <stdlib.h>
#include <ctype.h>
#include <sqlca.h>

#define MAXREC 1200
#define UIDLEN 20

EXEC SQL BEGIN DECLARE SECTION;
 struct emp_stru
  { varchar ename[MAXREC][10];

     varchar job1[MAXREC][9];
 };
  varchar sqlstmt1[1000];
EXEC SQL END DECLARE SECTION; struct emp_stru emprec;

char logon_db_II();
void sql_error(char *);

/* MAIN PROGRAM */
main(int argc, char *argv[])
{

    logon_db_II(); /*logon to Oracle*/

    /* Hard code select statement. The program will core dump, if the number of rows is not limited

        to a number smaller than 136. The select statement is not supposed to make sense, it just

        retrieves two string constants for all matching rows. The point is to show that the core

        dumping depends on the number of rows returned.
        Change the rownum limit above 136 to see this program core dump!

    */
    sprintf(sqlstmt1.arr, "select 'asdf', 'ertt' from all_catalog where rownum < 136");

    sqlstmt1.len = strlen(sqlstmt1.arr);     sqlstmt1.arr[sqlstmt1.len] = '\0';

    /* just showing select statement on the screen */     printf("\nTrying to execute the following select statement:\n %s\n\n",sqlstmt1.arr);

    /* Print SQL error, should one occur */     EXEC SQL WHENEVER SQLERROR DO sql_error("ar");

    /* Do the retrieval, do some debug printouts*/     EXEC SQL PREPARE s1 FROM :sqlstmt1;     EXEC SQL DECLARE c1 CURSOR FOR s1;
    EXEC SQL OPEN c1;
    printf("Cursor opened \n");
    EXEC SQL FETCH c1 INTO :emprec;
    printf("Cursor contents fetched into struc \n");     EXEC SQL CLOSE c1;
    printf ("%-10s\n",emprec.ename[0].arr);     printf("Cursor closed \n");

    EXEC SQL COMMIT WORK RELEASE;     exit(0);
}

void sql_error(char *tname) /* Issue the Oracle error message */ {
  char ora_msg[120];
  int tname_len, buf_len, msg_len;

  tname_len = strlen(tname);
  buf_len = sizeof(ora_msg);
  printf("\n\n***** Area Report terminating due to Oracle error on table %.*s :", tname_len,tname);
  sqlglm(ora_msg, &buf_len, &msg_len );   printf("\n********* %.*s\n", msg_len, ora_msg );   EXEC SQL WHENEVER SQLERROR CONTINUE;
  EXEC SQL ROLLBACK WORK RELEASE;
  exit(1);
}

char logon_db_II()

    /* Logon to Oracle */
{

    EXEC SQL BEGIN DECLARE SECTION;

        varchar     uid[UIDLEN];
	varchar     pwd[UIDLEN];

    EXEC SQL END DECLARE SECTION;     strcpy(uid.arr, "scott");
    strcpy(pwd.arr, "tiger");
    uid.len = strlen(uid.arr);
    pwd.len = strlen(pwd.arr);

    EXEC SQL CONNECT :uid IDENTIFIED BY :pwd;     if (sqlca.sqlcode != 0)

        {
	    if (sqlca.sqlcode == -1034)
		{
		    printf("Oracle is not available\n"); exit(1);
		}
            else
		{
		    printf("Error connecting to Oracle: %s\n", sqlca.sqlerrm.sqlerrmc);
		    exit(1);
		}
        }

    return(printf("\nConnect to ORACLE as user %s\n", uid.arr)); } /* End of logon_db_II */

   -**** Posted from RemarQ, http://www.remarq.com/?c ****-  Search and Read Usenet Discussions in your Browser Received on Mon Nov 15 1999 - 10:59:23 CST