stopping instrumentation in callgrind

Question

I'm spawning multiple processes and starting the instrumentation in each of them. When I try to stop the instrumentation just before the process exits, the instrumentation program seems to hang in the shell as if the process has already finished and it doesn't have a process to stop the instrumentation for. Here is the code:

from os import system,fork,getpid
from glob import glob
from sys import exit

for filename in glob("py/*.py"):
  f=fork()
  if f==0:
    system("callgrind_control --instr=on "+str(getpid()))
    execfile(filename,{})
    system("callgrind_control --instr=off "+str(getpid()))
    exit()

How can I solve the hanging problem? Do I really need to stop the instrumentation?

Answer 1

I solved the callgrind_control hanging problem by using call instead of system, with the parameter shell=True

from os import system,fork,getpid
from glob import glob
from subprocess import call
from multiprocessing import Process

def caller(filename):
  pid=getpid()
  call(["callgrind_control","--instr=on",str(pid)],shell=True)
  execfile(filename,{})
  call(["callgrind_control","--instr=off",str(pid)],shell=True)

for filename in glob("py/*.py"):
  p=Process(target=caller,args=(filename,))
  p.start()
  p.join()