Prior to Oracle Database 10g, all latch waits show up as the latch free wait event . In Oracle Database 10g, common latches are broken out and have independent wait event names and statistics but documentation about this independent wait event statistics are forgotten.
Yesterday I saw long waits of “latch: KCL gc element parent latch ” event on 10046 trace output of one of our sqls. I searched the documentation, metalink and google, all I have found is 15 bug pages on metalink, no page on documentation, 6 pages on google and all of this pages are not talking about defintion of event. They only show that this event has a place on the tkproof outputs. But my sql is suffering from this event what should I do now to lower this long waits???.
I hope documentation about latch sub events will be more definitive on upcoming 11G.
Ps: Any comment about definition of this wait event will be welcome.