Extended sql_trace output reporting incorrect elapsed time

In some 9.2.0.5 on AIX (single instance, dedicated server) trace files, I am seeing incorrect elapsed times for database calls (not wait events). Diagnosis becomes difficult when tkprof reports the query took 35 years, 218 days, 14 hours, 59 minutes and 31 seconds (I figure a new index will get that puppy running in sub-decade time).

In looking at the raw trace file, the e= value is the same as the tim= value for a given execution call. This of course causes the elapsed time values to be corrupted.

I know I have seen this somewhere, but I can't find it on metalink, google did not show up anything. I am obviously not using the right search criteria. Any pointers on a specific bug or note is greatly appreciated.

Regards,
Daniel Fink

--
http://www.freelists.org/webpage/oracle-l