ORA-07445: Exception error: core dump [ldxsnf () + 625] [SIGSEGV

ORA-07445: Exception error: core dump [ldxsnf () + 625] [SIGSEGV

ORA-07445: Exception error: core dump [ldxsnf () + 625] [SIGSEGV

Error message in ALERT Log:

Mon Jan 20 15:03:22 2014
Incremental checkpoint up to RBA [0x442f. abd.0], current log tail at RBA [0x442f. 338a. 0]
Mon Jan 20 15:08:13 2014
Errors in file/Oracle/product/10.2.0/db/admin/PROD2_findbb/udump/prod2_ora_27268.trc:
ORA-07445: Rainbow rainbow square parent draft: draft article missing [ldxsnf () + 625] [SIGSEGV] [Address noo object] [0x2B28910C94F0] [] []
Mon Jan 20 15:08:15 2014
Trace dumping is refreshing Ming id = [cdmp_20140120150815]

To view the TRC file, there is a statement similar to the following: -- Some SQL statements are omitted:

Ksedmp: internal or fatal error
ORA-07445: Exception error: core dump [ldxsnf () + 625] [SIGSEGV] [Address not mapped to object] [0x2B28910C94F0] [] []
Current SQL statement for this session:
SELECT * FROM aa
Where
Trunc (sysdate)...
#######################################

Solution and Reference documents:

Database Version: Oracle Database 10g Enterprise Edition Release 10.2.0.3.0-64bi
Refer to MOS:
Bug 6641866-Dump from operations on DATE values (Document ID 6641866.8)
ORA-07445 [Ldxsnf] even with patch 6641866 applied (Document ID 1159983.1)

Description in Bug 6641866:
Description
A dump can occur (eg under ldxsnfcom <-ldxsnf <-ldxgpc <-evadrn)
From SQL which uses operations on DATEs such as TRUNC etc ..

Description in document 1159983.1:
The bug can be identified based on the following facts:

ORA-7445 is raised in ldxsnfcom ()
2. following functions at the top of the stack: ldxsnf ldxgpc evadrn
3. SQL trace (10046 event) on the session reveals "alter session set nls_language". More specifically, the incorrect value being set is similar
"Alter session set NLS_LANGUAGE = 'use'" directly or via "DBMS_SESSION.SET_NLS ('nls _ Language', 'us ')".

Bug 9746699 is fixed in 12.1 and patch set 11.2.0.2.
######################################## #
The and trunc (sysdate) Statement exists in conjunction with the SQL statement in the TRACE file during ORA-07455, which should be the reason.