[erlang-questions] Understanding a record generated by fprof:analyse/*

[Rich Neswold]

Hello,

I'm running the fprof profiler on our newly developed system and found
a record in the profiler output that I don't understand. The record in
question is:

{[{{gen_server,loop,6},                        1630,   84.765,   26.450}],
 { {gen_server,decode_msg,8},                  1630,   84.765,   26.450},
 [{{gen_server,handle_msg,5},                  1630,    0.018,   25.549}]}.

The second column of numbers is the accumulated time used by the
function and the last column is the time taken in the function itself
(not functions it has called.) In 99% of the records, the accumulated
time is greater than the function's time, which makes sense. In the
case of genserver:handle_msg/5, however, the accumulated time is 18 uS
(!) and the function took 25 mS. How can this be?

Is anyone familiar enough with fprof to know what this is telling me?

Thank you!

--
Rich