On 9/6/2012 6:42 AM, Rajagopal Venkat wrote:
(305177) (4380506988525) (4380506988525) (4380507324219) (4380507019042) (0) <<<<<<< (4380507354736) (4380596923827) (4380596893310) (4380507415771) (4380606964111) (4380606903076)
cpu usage from total_cpu_time() - (425033.990889%)
As mentioned in my earlier comments, next patch will be submitted to handle these events(for which start time is not recorded) instead of ignoring them.
now THIS explains a lot, and makes the fix (the one in the other mail you just sent) make a lot more sense than the things from earlier ;-)
thanks for chasing this down!