第三次 - ORA-07445: exception encountered: core dump [pfrinstr_BRNCNULL()+58]

saup007

alter 日志报警：

1. Thu Apr 09 16:32:45 2015
   
2. Exception [type: SIGSEGV, Address not mapped to object] [ADDR:0x35401040013] [PC:0x91636C0, pfrinstr_BRNCNULL()+58] [flags: 0x0, count: 1]
   
3. Errors in file /u01/app/oracle/diag/rdbms/sanguo/SANGUO/trace/SANGUO_ora_6859.trc  (incident=122753):
   
4. ORA-07445: exception encountered: core dump [pfrinstr_BRNCNULL()+58] [SIGSEGV] [ADDR:0x35401040013] [PC:0x91636C0] [Address not mapped to object] []
   
5. Incident details in: /u01/app/oracle/diag/rdbms/sanguo/SANGUO/incident/incdir_122753/SANGUO_ora_6859_i122753.trc
   
6. Use ADRCI or Support Workbench to package the incident.
   
7. See Note 411.1 at My Oracle Support for error and packaging details.
   
8. Thu Apr 09 16:33:09 2015
   
9. Dumping diagnostic data in directory=[cdmp_20150409163309], requested by (instance=1, osid=6859), summary=[incident=122753].
   
10. Thu Apr 09 16:33:10 2015
    
11. Sweep [inc][122753]: completed
    
12. Sweep [inc2][122753]: completed

复制代码

trace文件在附件中。

这是我之前遇到的：应该同一个错误，但根源可能不同。

https://community.oracle.com/thread/2460154?tstart=0
http://t.askmaclean.com/thread-3494-1-1.html

这次没有 create or replace package 这个操作。

Maclean Liu(刘相兵

odm finding:

Hdr: 10065549 11.2.0.1 PLSQL 11.2.0.1 PRODID-11 PORTID-46
Abstract: ORA-3113 AND ORA-7445 [PFRINSTR_BRNCNULL()+51]

*** 08/27/10 02:33 am ***
*** 08/27/10 02:33 am ***
  
  
  BUG TYPE CHOSEN
  ===============
  Code
  
*** 08/27/10 02:35 am ***
Customers application crashes with ORA-3113.
In the alert.log there is a ORA-7445: exception encountered: core dump
[pfrinstr_BRNCNULL()+51] [SIGSEGV] [ADDR:0x0] [PC:0x860E7C5] [Address not
mapped to object] []
Patch  9210925 for 11.2.0.1 is already applied.
Set "_disable_fast_validate"=TRUE does not help.
Problem isn't reproducible with a testcase.
After the crash customer recreates the failing package/procedure and it works
fine again for a while. This issue occurs with different procedures or
functions.


没有找到一个“confirm 为real bug"的bug note，就以往的案例来看：

1、考虑重建问题plSQL程序 即PKG_GATEWAY.unbind
2、尝试设置_disable_fast_validate=true

Maclean Liu(刘相兵

Dump continued from file: /u01/app/oracle/diag/rdbms/sanguo/SANGUO/trace/SANGUO_ora_6859.trc
ORA-07445: exception encountered: core dump [pfrinstr_BRNCNULL()+58] [SIGSEGV] [ADDR:0x35401040013] [PC:0x91636C0] [Address not mapped to object] []

11.2.0.2.0 + LINUX

flags: 0x0, count: 1]
Registers:
%rax: 0x00000000bf514bdc %rbx: 0x00002b7b92f03e78 %rcx: 0x00002b7b92efd950
%rdx: 0x00002b7b92f03ee8 %rdi: 0x00002b7b92f03e78 %rsi: 0x00000000bf514bd8
%rsp: 0x00007fff0be7bb50 %rbp: 0x00007fff0be7bb60  %r8: 0x0000035401040009
%r9: 0x0000000000000001 %r10: 0x00007fff0be7baf8 %r11: 0x00002b7b92d92c20
%r12: 0x00002b7b92f03ee8 %r13: 0x00002b7b92eef730 %r14: 0x00000000000085f0
%r15: 0x00002b7b92f03e78 %rip: 0x00000000091636c0 %efl: 0x0000000000010246
  pfrinstr_BRNCNULL()+41 (0x91636af) mov %rax,-0x10(%rbp)
  pfrinstr_BRNCNULL()+45 (0x91636b3) cmp $1,%r9d
  pfrinstr_BRNCNULL()+49 (0x91636b7) jnz 0x916382a
  pfrinstr_BRNCNULL()+55 (0x91636bd) mov (%rcx),%r8
> pfrinstr_BRNCNULL()+58 (0x91636c0) movzwl 0xa(%r8),%ecx
  pfrinstr_BRNCNULL()+63 (0x91636c5) test $0x6,%cl
  pfrinstr_BRNCNULL()+66 (0x91636c8) jnz 0x9163864
  pfrinstr_BRNCNULL()+72 (0x91636ce) mov (%r8),%rcx
  pfrinstr_BRNCNULL()+75 (0x91636d1) test %rcx,%rcx

*** 2015-04-09 16:32:46.037
dbkedDefDump(): Starting a non-incident diagnostic dump (flags=0x3, level=3, mask=0x0)
----- Current SQL Statement for this session (sql_id=frv4ksphxkx4k) -----
begin :1 := PKG_GATEWAY.unbind(:2, :3, :4);end;
----- PL/SQL Stack -----
----- PL/SQL Call Stack -----
  object      line  object
  handle    number  name
0x35cb643a8       114  package body ERATING_SWGLXH.PKG_OPTION
0x325e70ac0       594  package body ERATING_SWGLXH.PKG_GATEWAY
0x35b3c4f60         1  anonymous block


stack call
pfrrun=> pfrrun_no_tool=>pfrinstr_BRNCNULL