Semaphore sleep problem

Today's test found a bug that would have errors when executing two commands with the module I was responsible for, such as "Command 1;command 2".

Analysis code found two questions,

First, command 1 in the creation of a background process, when the background process is created, the process continues to execute, ignoring the background for execution, so there may be a background for execution to complete, and command 2 to start execution of the situation, the order of execution is chaotic.

Second, command 1 sent a sigterm signal, and the module handler function has a sleep type function, when there is such a function, the process will sleep and give up the CPU, this time will continue to execute command 2, the order of the execution of the sequence of confusion.

Semaphore sleep problem