whats the relation between ISR and callback function?

Question

I am a little confused about ISR and callback function. I just know the ISR have no input params and return value. But i use TI MCU SDK register a timer params containing callback function with input params. I want to know how does this timer ISR run the callback function? What is the actual flow CPU runs?

Answer 1

As answered the hardware for a specific chip or architecture has a very specific way that interrupts are handled. A fixed address where the first instruction for a specific handler is or a fixed address where the address to the handler lives.

Then you add to this the mcu company wants to make your experience easier and tries to do much of the work for you. So they may have a generic handler for each thing and a library system where you tell their library where your code is, their software then has some, likely ram based scheme, if your handler is registered runtime, that is used by the real handler to branch/jump/call yours.

In this case the actual ISR would be TI code.

If you, in your code, have indicated you want to "handle" the interrupt then via library calls or some other scheme you indicate the interrupt you want to handle and the function you want to be called (and sure, perhaps even parameters can be part of this design scheme). This call to indicate your callback function likely also enables that interrupt into the cpu. Depending on the architecture there may be one big handler and only one interrupt or hundreds of individual interrupts and possibly a handler for each, or somewhere in the middle. So it may enable an individual interrupt or it may just enable an if-then-else path in some code the mcu vendor wrote.

interrupt comes into the cpu

cpu finds a good stopping point (between instructions usually)

cpu specific solution for finding the interrupt service routine code, either hardcoded address or hardcoded address in a vector table, or a way to register an address in the logic.

No generally you do not have parameters passed to the real isr. The isr will need to know what caused the interrupt and this is very much processor architecture specific. It is generally not "passed" to the handler like a compiled function call, instead control and status registers are often used.

In this scheme ti writes the isr, but then in some part of their software design they then call your function as if it were a normal compiled function. Possibly even with parameters depending on their design scheme.

Your function eventually finishes and returns

The ti isr finishes and indicates to the processor to continue with the code that was interrupted.

Answer 2

An interrupt (service routine) is called by hardware. A callback function is called by software, which could be a driver, OS or library.

Some libs or drivers might be designed so that you pass along a function pointer to a custom function. Then the driver implements the ISR so that the ISR calls your callback. This provides an abstraction layer at the expensive of some extra function call overhead.

I don't know how your TI driver is implemented, but it is common to have a timer driver register a list of callbacks with corresponding timeouts, then inside the ISR run a loop and check if it is time to execute a certain callback. With such designs it's important to keep the callback functions small.