High library cache waits kgllkdl1, kglhdgn2

BANNER                                                                               CON_ID

-------------------------------------------------------------------------------- ----------

Oracle Database 12c Enterprise Edition Release 12.2.0.1.0 - 64bit Production              0

PL/SQL Release 12.2.0.1.0 - Production                                                    0

CORE    12.2.0.1.0      Production                                                                0

TNS for Linux: Version 12.2.0.1.0 - Production                                            0

NLSRTL Version 12.2.0.1.0 - Production                                                    0

In our 3-node RAC cluster, we have seen recurring, sporadic spikes of Library Cache mutex X waits, during periods of high activity.  Our AWR reports consistently show that the code location of these waits is in "kglhdgn2 106" and "kgllkdl1 85".  I am unable to determine any information about what these code locations specify.

We have applied patch 26151631 to address the mutex X wait event bug with partitioned tables with deferred segment creation.  This patch did not address our issue.