Instrumentation issue when the program has P2P + Coll
Example:
P0 P1 P2
Send(2) Bcast Bcast
Bcast recv(0)
P0 P1 P2
Isend(2) Bcast Bcast
Bcast recv(0)
wait
P0 P1 P2
Send(2) Bcast Bcast
Bcast
P0 P1 P2
recv(2) Bcast Bcast
Bcast send(0)
We should improve the instrumentation to handle these cases