I am starting a number of processes in the background of the scrip that I'm writing. I want to wait for some of them to finish before the end of each iteration of a for loop, but there is one process that I need to leave running and end separately. What I'm doing right now doesn't work as I seem to run out of memory, and more and more processes are sacrificed, including the script that I'm running. What I currently have is:


    loadAvg &
	pid=$!
	echo "LA pid $pid"
	for (( j=0; j < $units_to_fill; j=$((j+1)) )); do
		#Print all files in dir | shuffle file names | delete 1st 1024 files found
		if [ $from_zero -ne 1 ];then
			stat -c %n * | sort -rz | awk -v n=0 '/file_[0-9]*_[abcd].bin/ { if( ++n <= 1024) {print $1 ; }}' | xargs -r rm -f
		fi
		echo "Creating file $j"
		pids=$(writeFile "file_${j}.bin" $src_file $block_size $block_count "${wd}/${dd_log_file}") #Writes a file of (dd) block size 1MB, with 1024 blocks
		for job in `jobs -p`; do
			if [[ $job -ne $pid ]]; then
				echo "Waiting for $job"				
				wait $job
			fi
		done
	done
	kill $pid

The function `writeFile` spawns a number of child processes, and I can't guarantee the order that they'll exit in, as I was previously doing `wait $!` in place of the 

    for job in `jobs -p`

I can capture the pids of the child processes that are spawned in the function, and they are currently being returned in a format that's essentially `pid1 pid2 pid3 ...`, but running `wait $pids` doesn't work as some of the processes finish before that line is reached, so `wait` errors out.

Note: I'm running on an embedded system with busybox linux installed, and as such it's been quite cut down and a lot of standard functions are unavailable.