Thunks for Closures

[darko20]

So the last piece of the puzzle for my C based closures is generating a thunk that inserts a pointer to the data for the closure in the argument list. I'm thinking I'll compile the thunk function once I have the address of the data (represented by 0x12345678 here):

typedef struct cl {
  int a, b, c;
} *cl;

int c(const cl l, int d, int e) {
  l->a = d;
  return l->a + l->b + l->c + e;
}

int thunk(int d, int e) {
  return c((const cl)0x12345678, d, e);
}

I know that MIR will generate a thunk for all functions and I would like to avoid doubling up the thunks. My question is, will thunk be optimized away somehow or is there a better approach, such as some kind of inlining trickery? The position of the data link l in c is unimportant of course.