v$wait_event: wait_time < 0

I know this question was asked before but I don't think the answer was clear. I have the following returned from a query of v$session_wait

SID EVENT WAIT_TIME SECONDS_IN_WAIT STATE

--- --------------  ---------  --------------- -----------------
19 latch free              -1              361 WAITED KNOWN TIME

The seconds_in_wait is constantly increasing. According to the 9i docs, the wait_time of -1 and state indicate that the wait event was very short and is now over. But if I look at the server process for sid 19, it's eating up cpu like there's no tomorrow (12.4% on an 8 cpu server). What exactly is the server process doing? Normally you can tell by the wait events but apparently not in this case.

This process sat on the "latch free" event for 20 minutes with the seconds_in_wait increasing. If I run it again it may show up as sitting on "scattered read" for 20 minutes before finishing. A third time it may sit on "sql*net message from client" all the while the server process is running away with cpu. What is going on

DB Version is 9.2.0.5. O/S is solaris.

TIA Received on Thu Jun 14 2007 - 09:06:33 CDT