The basics of embedded multitasking on a PIC: Part 1 - Reentrant multitasking -

The basics of embedded multitasking on a PIC: Part 1 – Reentrant multitasking

High-end concepts typically aren't applied to an 8-bit MCU. But don't rule them out just yet.

Applying multitasking to an embedded system based on a PIC microcontroller may look appealing. It permits full use of the PIC resources by serving more than one task at a time rather than staying idle waiting for an external event to occur and reacting accordingly from within an application. However, multitasking can turn applications into memory-hungry monsters that barely fit in your PIC. Task reentrancy is one approach that tackles expanded memory tasks, especially in small to medium PICs.

In this, the first part one of a three part series about reentrant multitasking and how you can realize it with two working examples of a Microchip's PICmicro MCU, I begin with an introduction to multitasking. In part two and three, available on, I'll get more into the meat of multitasking on the PIC.

Embedded multitasking is not new. It goes back some 30 years. The Intel 8080, Zilog Z80, and Motorola 68000 were popular microcontrollers 30 years ago. Interestingly, they were very similar in processing capabilities to many of today's microcontrollers. Those ancient devices had multiple general-purpose and index registers that could perform arithmetic and logic operations, and supported vectored interrupts and I/O ports. They had data buses with widths up to 16 bits, and address buses up to 24-bits wide. They also had large and powerful instruction sets that gave way to efficient programming and ran at moderate speeds of around 10 MHz. What's changed over the years is that today, that same processor is packaged with RAM and EPROM. Adding analog-to-digital modules and EEPROM gives us today's “microcontrollers.”

Thirty years ago, full-fledged real-time operating systems (RTOSs) existed and professional features like multitasking and multiprocessing were applied and implemented. This brings up an interesting point: where exactly does multitasking fit with this technology? With a little research, I found that products are available to handle multitasking with an RTOS. Developers must then question whether the RTOS is really required for their application, or near-real-time would suffice.

Back then, in the absence of multitasking, serial task execution was the only way to run embedded systems. Figure 1 shows two tasks running in a system. Task #1 executes until it concludes its operation, then passes control to Task #2. Task #2 runs and eventually terminates causing system operation to end.

By definition, multitasking allows an embedded system to share its resources between several programs running concurrently. The resources can be those of the processor itself (such as internal registers, memory, and arithmetic logic units) or the peripherals attached to the embedded system (such as display, printer, sensors, and actuators).

Figure 2 illustrates a simple form of multitasking. Here, Task #1 is the main task running in the system until some sort of interrupt or an event occurs at time t 1. This causes the system to put that task on hold, passing its control to another secondary Task #2, at time t 2, to serve the event and willingly return system control to the main task at time t 3. This kind of multitasking is based on cooperation between the running tasks, such that tasks are willing to give up system control at some point throughout their execution. Note that there's no mention of operating-system coordination between tasks; task delegation of system control is built in the task programs themselves. Nested subroutine calls and branching instructions are a means of implementing simple multitasking.

Multitasking in an operating-system environment is a different story. System resource sharing is realized under the operating system's control, rather than task delegation, by allocating those resources to a single task for a specific period of time before putting them temporarilyon hold and switching the resources to another task. The task allocated time and frequency depend on “multitasking policy.” Different policy schemes can be adopted, such as round-robin (equal task rights) and priority-based (higher priority tasks use resources more frequently than lower priority ones).

Cooperative vs. preemptive multitasking
The part of the operating system responsible for task switching is called the task scheduler. The scheduler enforces the multitasking policy whenever it regains system control from the user tasks. User tasks can be designed to either give up system control to the scheduler voluntarily (scenario #1) or forcibly (scenario #2).

Scenario #1, called cooperative multitasking , is where a breakpoint is implanted in the task program to transfer system control to the scheduler program. The scheduler can later resume task execution at that breakpoint. Figure 3 illustrates this scenario with three tasks. Task #1 returns control to the scheduler at time t 1. The scheduler decides that Task #2 gets the control, which is passed on to it at time t 2. At time t 3, Task #2 returns control to the scheduler and Task #3 gets control at t 4. Task #3returns control to the scheduler at t 5; the scheduler then decides that Task #1 resumes its execution. Task #1 regains control at t 6. This scenario repeats until all tasks conclude their operation or the system shuts down.

View the full-size image

Scenario #2, called preemptive multitasking , is where task execution is interrupted at any point in time by an internal or external system event. System control is then handed over to the scheduler to decide which task is to resume execution. Figure 4 illustrates this scenario with two tasks. Here, Task #1 runs until an interrupt occurs at time t 1, suspending the task. Control is passed on to the scheduler program at t 2, at which time the program decides to pass the control to Task #2 based on some policy and activates that task at t 3. Task #2 runs until t 4, when another interrupt occurs causing it to suspend and the control is passed on the scheduler again at t 5. The scheduler decides to resume Task #1 execution at t 6, which takes back control until another interrupt occurs. Frequent interrupts to running tasks enable the scheduler to manage system control switching between tasks.

View the full-size image

Multitasking reentrant programs
The tasks employed for a multitasking embedded system can be replicate control applications or a combination of control, data crunching, and user interfacing. But we know that not all embedded systems have an abundance of memory that fits multiple task programs. In addition, many PICs have limited interrupts, work registers, and stack capabilities that keep them form competing against full-fledged processors. This may explain why it's hard to find operating systems that support multitasking.

One way around the PIC's limitations allows better manipulation of its resources. In computer jargon, it's called reentrant programming . Reentrant programs realize a simple form of multitasking while minimizing memory overhead. When you consider running a number [x] of tasks of similar functionality, each of which requiring [y] memory bytes to run, you may think at first glance that you would need [x * y] bytes of memory. However, with reentrancy, you would actually require only a fraction [?] of memory over and above that of a single task, giving a total of [(1 + ?) * y] bytes for the [x] tasks.

Figure 5 illustrates the reentrant program concept as a special case of multitasking, which was demonstrated in Figure 4. Here, all tasks are similar in functionality and are represented by a single reentrant program. The scheduler switches the execution of this program among the distinct tasks upon interrupt events.

View the full-size image

A reentrant program is a generic data-driven task. If you manage to separate the task data from the code and make the code register-dependent, you can end-up with a generic task. Having done this, another question arises: how can user tasks invoke their reentrant program simultaneously? The answer is twofold:

• First, you must initialize the registers of the generic task with each task's data.

• Second, you need some mechanism, such as interrupts, to switch the generic task between user tasks.

A user task gaining control of the generic task loads its processor state (its registers' settings) and resumes its previously interrupted operation. But that only happens after it the user task saves its predecessor state. The task in control of the processor may set a pointer to its descendant task to take over control when the interrupt occurs.

System limitations
My first encounter with microcontrollers was the Microchip's PIC18F452 MCU. It was a challenge to implement my version of multitasking on such a device. I was faced with limitations that required a workaround that costs memory and execution time. I didn't encounter such limitations while working with those ancient microprocessor devices.

On one hand, most of the PIC data transfer and arithmetic instructions involve the W register. To facilitate indirect addressing on a PIC, the address must first be loaded into the FSR register. This makes a stack implementation inefficient, comprising many instructions for one indirect memory access. The programmer has to work around this by passing parameters in fixed memory locations.

The PIC's small hardware return stack allows only minor nesting of calls, so the program structure is almost flat, relying on macros. Also, the PIC lacks support for automatically saving and restoring its state on interrupts. This makes interrupt processing extremely delicate. It's up to the programmer to save the processor state on interrupt, but this entails a strict instruction sequence.

Figure 6 shows how the memory in a PIC-based embedded system is allocated for a multitasking environment. Two cases are shown for the program memory: the memory map on the left shows a multiple of similar user tasks with replicated code.

View the full-size image

The map on the right uses a reentrant code. Common program modules in both cases are the interrupt handler, the task initialization, and the task scheduler. Those modules are almost identical in both cases. The data memory holding the tasks' work areas is also similar but with minor differences. The software stack in the data memory solves the drawbacks of the PIC's hardware stack.

Gamal Ali Labib is an IT consultant in Cairo, Egypt. He specializes in IT security and IT turn-key projects management. He is also interested in parallel processing and VLSI. Dr. Labib has a B.Sc. and M.Sc. in computer engineering and electronics from Ain Shams University, Egypt, and a PhD in computer science from University of London, U.K. You can reach Dr. Labib at .

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.