parent and child processes race each other

[jduck]

When synchronizing the parent and child processes using the pipe, the child might read the parent's status data before the parent is able to. I added a "sleep(1);" between the child's calls to "send_status_to_parent" and "wait_to_close" and that corrected it. The more correct solution would be to use a different mechanism for synchronization (perhaps a shared file mapping, with one entry per child denoting a state value).

[fi01]

I've updated to use non-blocking read to avoid race condition. Please check it.
Thanks.

[jduck]

Okay but now you are never closing the icmp sockets of the child (since wait_to_close never returns). Is that intended??

[jduck]

OH I see. The sockets get closed anyway when you kill the child process from the parent. In that case, all of the code after wait_for_close() is dead code and should be removed.