the lower 128 bytes of the 8032 uC into blocks, 8 bytes allocated to each task to be precise (trying to speed up the switching operation by avoiding the use of external memory)
And slowing down all routines by forcing them to use external memory.
Also 8 bytes can not hold 8 registers, psw, acc, b, dph, dpl

The only problem I am facing
I think "only" is wrong, a RTOS is a daunting task.

some kind of priority implementation for each task whereby each task is given a value and is dispatched on its priority
you will need priority inversion protection or you will hang the system.

Erik