It is possible to transform the pure Rosetta form of GNU nested function similar to the pure C, such that it doesn't need any stack trampoline. I wonder if that would be closer in performance to the pure C form.

(I can't be bothered to run his benchmarks)

    #include <stdio.h>
    typedef struct env_ E;
    typedef struct fat_ptr_ Fp;
    typedef int fn(E*);
    struct fat_ptr_ {
      fn *f;
      E  *e;
    };
    #define INT(body) ({ int lambda(E*){ return body; }; (Fp){lambda,0}; })
    struct env_ {
      int k;
      Fp xl; Fp x2; Fp x3; Fp x4;
    };
    #define FpMk(fn,e) {fn, e}
    #define FpCall(fn) (fn.f(fn.e))
    int main(){
      int a(E env, Fp x5){
        int b(E *ep){
          return a( (E){--(ep->k), FpMk(b, ep), ep->xl, ep->x2, ep->x3}, ep->x4 );
        }
        return env.k<=0 ? FpCall(env.x4) + FpCall(x5) : b(&env);
      }
      printf(" %d\n", a( (E){10, INT(1), INT(-1), INT(-1), INT(1)}, INT(0)) );
    }