I'm thinking of building a language for PIC microcontrollers. I want to be able to use non-fixed size arrays, like this:

  1. Declare the variable as int[]
  2. Wait for input from serial connection
  3. Make the variable input long

I think such a feature would be useful, but don't know how I should go about implementing this in a compiler which compiles to assembly. I want the arrays to be stored on consecutive register addresses, of course. Since I'm working with PICs, this has to be very memory efficient.

When someone writes int[], I don't think it's a good idea to reserve memory space for the variable yet, is it? Then the array would have a fixed maximum size. For example, when the array gets reserved memory address 5-100, and other variables get 4 and 101, the array has fixed borders and can't grow any bigger than 96 registers. Also, when I'd reserve memory from the start, let's say x bytes, and I in the end only need y bytes, I'm wasting x - y bytes. I don't want that.

That means the only option I see is initializing the array and reserving space in the microcontroller, on the fly. This will take up some memory and execution time, of course. I thought of a system like this:

  • Initialize an array int[x] = {int, int} which holds pointers to the start and end of arrays that are not initialized from the beginning - x would be the maximum amount of arrays (this is a concession, but is better than a maximum length for all arrays)
  • Store a variable c = 0 to indicate the number of arrays used
  • Store the borders of the initialized (reserved) memory in a variable somewhere
  • When an array gets a length:

    • Put pointers to the start (the current border) and end (the current border + the length) in the array from the first point at index c
    • Increment c

I think this would work (wouldn't it?), but there a few cons, mainly concerning memory: I need to store the array, c and the current memory borders as overhead.

Would there be a better way to implement non-fixed size arrays in a language for PIC microcontrollers? My requirements are:

  • Low memory overhead
  • The array length does not have to be changed on the fly
  • With the system I thought up, one cannot store values in arrays that haven't been initialized yet. If there would be a system that can store values in an array of undefined length, that would be an advantage
  • Faster systems (at runtime, compile time doesn't matter) are preferable
  • gcc.gnu.org/onlinedocs/gcc/Variable-Length.html - implementation is fairly trivial, arrays are allocated on stack.
    – SK-logic
    May 9, 2013 at 14:35
  • @SK-logic that's what I want! Unfortunately, it isn't supported by compilers for PICs AFAIK. It isn't that trivial to me, could you explain it (in an answer, perhaps)? A PIC is a very low-level thing, there's not just something like popping and pushing to or from a stack.
    – user76821
    May 9, 2013 at 14:37
  • 1
    Microchip provides a library of routines that provide a heap in Application Note 914.
    – Blrfl
    May 9, 2013 at 14:52
  • @Blrfl thanks, I stand corrected: it is possible in C, also for PICs. I edited my question; I would still like to know how this could look like from the compiler-side. The AN is very useful, but can't it be more efficient when built-in into the compiler?
    – user76821
    May 9, 2013 at 14:55
  • 1
    As @Blrfl noted, there's a heap available. Allocating non-fixed-length arrays on the stack is tricky, especially since it makes you even more vulnerable to stack buffer overrun errors than putting fixed-length arrays on the stack does. (Which, depending on the uses this device is being put to, can have security implications.) You ought to have a look at the way Delphi implements dynamic arrays as a language feature. They're always heap objects, managed by the compiler, and it works quite well. May 9, 2013 at 15:49

2 Answers 2


You will probably make it a lot easier on yourself and avoid a lot of user errors if you allow the declaration of variable-length arrays to be delayed until a point where the size is known. Then you don't have to invent a syntax for assigning the array size and you don't have to deal with differences in the order in which arrays are declared and in which they get their size.

If your compiler uses a stack-based allocation scheme for local variables (they are given addresses relative to the stack-frame of their function, like it is done for the 'big' platforms), the variable-length arrays can just be carved out of the stack frame as needed. Only if there are multiple variable-length arrays in a function do you need an additional pointer to indicate where the data of each starts.

If your compiler uses a fixed allocation scheme (all variables, including locals, are given a fixed (absolute) address by the compiler/linker), I would use an 'array stack' for carving the variable-length arrays from. As overhead, you would need a global pointer to indicate where the free space currently starts, a pointer for each variable-length array to indicate where its data is located and, in each function using variable-length arrays, a way to restore the global pointer to the value it had when entering the function.
The user would also need some way to indicate to the compiler/linker how much of a 'array stack' he thinks will be needed. A reasonable default could be to use whatever is left after allocating all fixed-size variables and variable-length array overhead.

I think this scheme uses the least amount of overhead, both space and time-wise. The scheme you came up with is essentially a malloc implementation.


Variable-length arrays in C99 are still fixed-size; you just don't know the size until runtime. They're not dynamic, so you can't declare int foo[] and start accessing elements. There's no way to know how much space to allocate for it and there's no magic that goes on behind the scenes that rearranges everything as the array grows.

In earlier versions of C, the compiler did a single move of the stack pointer at the start of scope because all of the variables were declared up front:

  int foo;
  int bar;
  // Stack pointer gets nudged 2 * sizeof(int)
  ... Code ...

What makes VLAs possible was a change in C99 that allowed declaration of variables anywhere in the code instead of just at the top of scope:

  int foo;  // Stack pointer gets nudged sizeof(int)
   ... Code ...
  int bar;  // Stack pointer gets nudged another sizeof(int)

Being able to do this means that to create a VLA, all that has to be done is evaluate the expression in the brackets (which can use previously-declared variables because the compiler knows where to find them), multiply the result by the size of the array type and nudge the stack pointer by that much:

  size_t size;  // Stack pointer gets nudged sizeof(size_t)
  size = get_number_of_ints_in_input();
   ... Code ...
  int data[size];  // Stack pointer gets nudged size * sizeof(int)
   ... Code ...

As long as you keep track of how much the stack pointer has been moved, you can pull it back where it should be anytime you go out of scope.

  • Thanks. That helped me solidify something I was working on. May 11, 2014 at 12:34

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.