像这样的？
测试用例：

SQL> create table mac
  2    (state number,
  3     rpd varchar2(2),
  4     mac_address varchar2(20) constraint uk_mac unique,
  5     total_count number
  6    );

Table created.

SQL> insert into mac
  2    select 26, 'aa', 'aa:bb:12:cc:ab:aa', 1 from dual union
  3    select 26, 'bb', 'aa:bb:12:cc:ab:ab', 1 from dual union
  4    select 26, 'cc', 'aa:bb:12:cc:ab:ac', 1 from dual;

3 rows created.

一个过程：我把整行选成一个局部变量。你可能会得到 NO_DATA_FOUND ，我正在处理它（什么也不做）。如果有匹配，我就减去 TOTAL_COUNT 将同一mac地址行的列值乘以1，然后插入整个新行。
然而，这个模型似乎支持重复的mac地址。这合法吗？在这种情况下你想怎么办？例如，如果您再次输入相同的mac地址？ SELECT 会回来的 TOO_MANY_ROWS 应该处理的。一个选项是只更新一行（哪一行？）；还是。。。？

SQL> create or replace procedure p_mac
  2    (par_state in number, par_rpd in varchar2, par_mac_address in varchar2)
  3  is
  4    l_row mac%rowtype;
  5  begin
  6    select *
  7      into l_row
  8      from mac
  9      where mac_address = par_mac_address;
 10
 11    update mac set
 12      total_count = total_count - 1
 13      where mac_address = par_mac_address;
 14
 15    insert into mac (state, rpd, mac_address, total_count)
 16    values
 17    (par_state, par_rpd, par_mac_address, 1);
 18  exception
 19    when no_data_found then null;
 20  end;
 21  /

Procedure created.

测试：

SQL> select * from mac;

     STATE RP MAC_ADDRESS          TOTAL_COUNT
---------- -- -------------------- -----------
        26 aa aa:bb:12:cc:ab:aa              1
        26 bb aa:bb:12:cc:ab:ab              1
        26 cc aa:bb:12:cc:ab:ac              1

SQL> exec p_mac(26, 'ab', 'aa:bb:12:cc:ab:ac');

PL/SQL procedure successfully completed.

SQL> select * From mac;

     STATE RP MAC_ADDRESS          TOTAL_COUNT
---------- -- -------------------- -----------
        26 aa aa:bb:12:cc:ab:aa              1
        26 bb aa:bb:12:cc:ab:ab              1
        26 cc aa:bb:12:cc:ab:ac              0
        26 ab aa:bb:12:cc:ab:ac              1

SQL>