CREATE OR REPLACE PACKAGE PKG_P
AS
	G_SAL NUMBER(7,2):=3000;
END;
/
-----------------------------------

----------------------------------
CREATE OR REPLACE PROCEDURE P(v_deptno NUMBER,v_dname VARCHAR2)
AS
CURSOR c_emp(c_deptno NUMBER,c_dname VARCHAR2)
IS
	SELECT E.Ename,D.Dname,E.Sal 
	FROM Emp E,Dept D
	WHERE E.Deptno=D.Deptno 
	AND
	E.Deptno=c_deptno 
	AND
	D.Dname=c_dname
	AND
	E.Sal=PKG_P.G_SAL ;
BEGIN
FOR i IN c_emp(v_deptno,v_dname)
LOOP
	DBMS_OUTPUT.PUT_LINE(i.Ename||' '||i.Dname||' '||i.Sal);
END LOOP;
END;

CREATE OR REPLACE PROCEDURE P(v_deptno NUMBER,v_dname VARCHAR2)
AS
CURSOR c_emp(c_deptno NUMBER,c_dname VARCHAR2)
IS
	SELECT E.Ename,D.Dname,E.Sal 
	FROM Emp E,Dept D
	WHERE E.deptno=D.deptno 
	AND 
	E.Deptno=c_deptno 
	AND
	D.Dname=c_dname;
BEGIN
FOR i IN c_emp(v_deptno,v_dname)
LOOP
	IF i.sal=PKG_P.G_SAL THEN
		DBMS_OUTPUT.PUT_LINE(i.Ename||' '||i.Dname||' '||i.Sal);
	END IF;
END LOOP;
END;

CREATE OR REPLACE PROCEDURE P(v_deptno NUMBER,v_dname VARCHAR2)
AS
CURSOR c_emp(c_deptno NUMBER,c_dname VARCHAR2)
IS
	SELECT E.Ename,D.Dname,E.Sal 
	FROM Emp E,Dept D
	WHERE E.deptno=D.deptno 
	AND 
	E.Deptno=c_deptno 
	AND
	D.Dname=c_dname;
TYPE t is RECORD
(
	v_ename VARCHAR2(30),
	v_dname VARCHAR2(30),
	v_sal NUMBER(7,2)
);
TYPE t1 IS TABLE OF t;
t2 t1;

BEGIN
OPEN c_emp(v_deptno,v_dname);
FETCH c_emp BULK COLLECT INTO t2;
FOR i in t2.FIRST..t2.LAST
LOOP
	IF t2(i).V_sal=PKG_P.G_SAL THEN
		DBMS_OUTPUT.PUT_LINE(t2(i).v_ename||' '||t2(i).V_dname||' '||t2(i).V_sal);
	END IF;
END LOOP;
END;
---------------

1. bulk fetch is always better for 10 to 1000 rows, over just one row at a time.  Indeed, there have been automatic PL/SQL optimizations added to recent versions of the PL/SQL compiler that will translate a FOR LOOP into its BULK VERSION automatically.

2. which brings us to style of coding.  Generally I prefer FOR LOOP to independent cursors.  I think they read better, and as is noted in #1, there is the automatic optimization in its favor now too.

3. it seems strange to me that you need both the DEPTNO and DNAME values at the same time.  I am going to assume this is just a fluke of your test case.

4. as Michel points out, putting it all in SQL gives the optimizer the most options so you want the value of the package variable in the SQL code, not tested as an IF statement.  However...

5. calling PL/SQL from SQL is expensive in many ways.  It would be vastly better for you to have passed the value of the package variable into the routine as another parameter.

create or replace procedure p (dept_no_p in number, dname_p in varchar2, sal_p in number)
as
begin
   for r1 in (
                SELECT E.Ename,D.Dname,E.Sal 
                FROM Emp E,Dept D
                WHERE E.Deptno=D.Deptno 
                AND E.Deptno=deptno_p
                AND D.Dname=dname_p
                AND E.Sal=sal_p
             ) loop
      DBMS_OUTPUT.PUT_LINE(i.Ename||' '||i.Dname||' '||i.Sal);
   end loop;
end;
/