Shulou(Shulou.com)06/01 Report--

This article is about how to manage RT-Thread interrupts. The editor thinks it is practical, so I share it with you. I hope you can get something after reading this article. Let's take a look at it with the editor.

The following is a summary of RT-Thread interrupt management, including a brief introduction to what interrupts are, what is the difference between bare metal interrupts and RT-Thread interrupts, how RT-Thread handles interrupts, what interrupt-related interfaces are provided by the RT-Thread kernel, and so on.

Summarize the learning process of RT-Thread interrupt management from the following aspects

Concept description related to interruption

What is an interruption? Interruption, as its name implies, is a work in progress, suddenly interrupted by other things, resulting in the original ongoing work can not continue normally, but need to deal with other things before they can come back to continue the original work.

How to understand interruption in a popular way? Imagine a scenario where you are happily writing code at home over the weekend, and suddenly your cell phone rings and you have to stop your work, record where the code is written, and then answer the call. "writing code" is ongoing work, and "phone ringing" is an interruption event.

This call is from my wife. She asked you to go to the vegetable market to buy some leeks and pork and make dumplings in the evening. Your wife dare not listen to her. So you think it is more important to buy things in the market, hang up the phone, go shopping, and then write the code that has not just been completed. "Market shopping" is to interrupt the service program, which is a typical interruption process.

With regard to the operation mode and privilege level of interrupts, Cortex-M processors can be divided into three states: privilege-level processing mode, privilege-level thread mode, and user-level thread mode. The relationship between these three states is shown in the following figure.

As can be seen from the figure above, service programs with interrupts or exceptions are always in privileged processing mode. On the other hand, the main thread (main thread) started when the kernel of the RT-Thread system is reset and powered on runs in privileged thread mode. Threads created by other users are running in user-level thread mode.

Why should the processor distinguish between the privilege level and the user level? Privilege, as the name implies, means that if the processor works at this level, it will have higher privileges and can access some special registers to prevent user-level code from accessing these special registers and destroying the data. Because of its particularity, the interrupt function works at the privileged level.

What's the difference between a bare metal interrupt and an operating system interrupt? When we deal with hardware interrupts in bare metal code, we usually only need to write interrupt handling functions, which is simple and direct.

However, with the operating system, everything has changed, and there are a lot of problems to consider. Because there are many threads running in the operation, after the interrupt comes, it is necessary to tell the operating system to save the information of the current running thread into the stack, then deal with the interrupt service program, and then go back to deal with the thread after dealing with the interrupt. At this time, thread switching scheduling may be involved, and thread switching itself requires PendSV interrupt participation.

Therefore, there is a big difference between handling interrupts in bare metal and handling interrupts in the operating system.

RT-Thread interrupt handling mechanism

Engineers who have known the Cortex-M series single-chip microcomputer generally know that there is an interrupt vector table in the assembly startup file startup_xxx.s of the chip, and all interrupts are processed through this interrupt vector table.

When an interrupt exception is triggered, the processor will determine which interrupt source it is, and then jump to a fixed location for processing. The address entry of each interrupt service program must be placed on a unified address, that is, it needs to be set to the interrupt vector offset register of NVIC.

In fact, once the hardware sends interrupts and exceptions, the interrupt entry is in this interrupt vector table, with or without the participation of the operating system. The difference is that in the bare metal environment, the interrupt service program is handled directly, while in the case of an operating system, it is necessary to retain the running condition of the thread first, then deal with the interrupt, and then restore the running environment of the thread.

The priority of hardware interrupt is the highest, and the priority of any thread is lower than that of hardware interrupt. Therefore, as long as a hardware interrupt event occurs, the system must deal with it accordingly.

RT-Thread generally has three stages when dealing with interrupts: the interrupt leader, the interrupt service program, and the interrupt subsequent program, as shown in the following figure.

The main job of the interrupt leader is that when an interrupt event occurs, the hardware of the processor automatically presses the current CPU-related register parameters into the interrupt stack. The program needs to call the rt_interrupt_enter () function to add 1 to the global variable rt_interrupt_nest, which is used to record the number of nesting layers of interrupts.

The main work of the user interrupting the service program is divided into two situations, one is not to switch threads, the other is to switch threads. If there is no thread switching, the interrupted thread will be returned after the interrupt service program and the interrupt subsequent program are completed.

If you want to switch threads, the rt_hw_context_switch_interrupt () function is called for context switching, which mainly sets the variable rt_interrupt_to_thread and then triggers the PendSV interrupt.

Note here: since the PendSV interrupt has the lowest priority and cannot be preempted by the interrupt, even if the interrupt is triggered, the PendSV interrupt is still in the waiting phase because it is still in the user interrupt handling function at this time. Only after exiting the subsequent program of the interrupt will the PendSV interrupt processing be performed and the thread context will be switched. Therefore, the context switch of the thread is not done within the user interrupt, but after the end of the interrupt.

The main work of the interrupt subsequent program is to tell the system kernel to leave the interrupt state, subtract the global variable rt_interrupt_nest by 1 by calling the rt_interrupt_leave () function, and then recover the CPU-related register parameters from the interrupt stack.

When restoring the CPU register parameters, it should be noted that if thread switching is involved in the user interrupt, then you need to restore to the new thread CPU register parameters instead of to the CPU register parameters of the thread interrupted by the interrupt.

When dealing with interrupts, the RT-Thread operating system usually adopts the methods of "upper part (Top Half)" and "bottom part (Bottom Half)". The reason is that the operating system itself will not make any assumptions and restrictions on the processing time of the interrupt service program, but in order to ensure the real-time performance of the system, users need to ensure that the interrupt service program is completed in as short a time as possible.

How to understand the interrupt handling methods of "upper half (Top Half)" and "bottom half (Bottom Half)"? Take buying vegetables as an example. Your wife calls you to buy food in the vegetable market (interruption event), but you consider that if you don't write the code for a long time, it will lead to a broken chain of thoughts. In order to avoid this situation (to avoid long-time processing and service interruption), you can place an order online to purchase (short-term interruption processing). When the fresh supermarket receives the order information (semaphore, mail, message queue), it will arrange for the delivery boy to deliver the goods to your door. The time-consuming work of buying vegetables is done by other people (other threads).

The interrupt handling methods of "upper part (Top Half)" and "bottom part (Bottom Half)" are mainly used in some situations where time-consuming interrupt transactions are needed, such as data reception and processing. Usually, the time to receive the data is relatively short, as long as the received data is saved, but the process of processing the data may be time-consuming, so it needs to be processed separately, and the upper part is to receive the data. the bottom half is time-consuming data processing.

API function Interface related to RT-Thread interrupt

In order to isolate the interrupt anomalies from the underlying hardware, the RT-Thread kernel encapsulates interrupts and exceptions into a set of abstract interfaces, as shown in the following figure.

Application examples related to RT-Thread interrupts

The application examples related to RT-Thread interrupts are mainly to verify the API interface functions related to interrupts, such as the use of global interrupt switches, and to verify the interrupt handling methods of "upper part (Top Half)" and "bottom part (Bottom Half)" through the key interrupt example.

The main purpose of the global interrupt switch example is to verify that the global variable is protected by the switch global interrupt when the same variable is accessed by multiple threads.

The key interrupt example is mainly to verify the interrupt handling of "upper part (Top Half)" and "bottom part (Bottom Half)". The interrupt event is triggered by the key press, and an email is sent in the interrupt service function to inform the thread to deal with it accordingly.

In the irq_ test. H header file, recompile the project source code by opening the corresponding macro definition switch, and download it to the development board to verify the experimental phenomenon, as shown in the following figure.

Matters needing attention when RT-Thread interrupts the application

Interrupt is a kind of exception, which must be dealt with when an interrupt exception occurs in the system. If the interrupt is not handled in time or improperly handled in the RT-Thread real-time operating system, it will cause system error or logic confusion, and serious will lead to destructive paralysis of the system.

When handling RT-Thread interrupt exceptions, there are the following considerations:

1. The interrupt service program works in privilege-level processing mode, the priority is higher than any thread, and no thread can preempt the interrupt service program.

two。 In the operating system, interrupt nesting can be supported, high-priority interrupts can preempt low-priority interrupts, and thread rescheduling is restarted after all interrupts have been processed.

3. In Cortex-M architecture, when an interrupt occurs, the CPU register stacks are automatically completed by the hardware, and the leading program of the interrupt usually only records the number of nested layers of the interrupt.

4.RT-Thread uses independent memory space as interrupt stack instead of thread stack. With the increase of threads, the effect of reducing memory footprint is more obvious.

5. It is suggested that the "upper part (Top Half)" and "bottom part (Bottom Half)" should be used to handle interrupt exceptions, and the processing time of interrupt service programs should be as short as possible.

6. The use of global interrupt switch is the simplest way to prohibit multi-thread access to the critical area, which can be used in any situation, but it should be noted that this method has a great impact on the real-time performance of the system, and improper use will destroy the real-time performance of the system. The time to use the global interrupt lock should be as short as possible.

7. The global interrupt switch supports the nested use of multi-level interrupts, and each call to the rt_hw_interrupt_enable () function can restore the system to the state it was before the off interrupt (which may be off interrupt or on interrupt).

8. The interrupt service program runs in privileged processing mode, in which the relevant functions that suspend the current thread operation cannot be used, because there are no threads in the running environment of the interrupt service program.

The above is how to carry out RT-Thread interrupt management, the editor believes that there are some knowledge points that we may see or use in our daily work. I hope you can learn more from this article. For more details, please follow the industry information channel.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.