Recursive Karatsuba Pseudocode

This is pseudocode for a space-optimized recursive implementation of the Karatsuba multiplication algorithm.

MUL(X,Y,N):
;X points to the first multiplicand of N digits
;Y points to the second multiplicand of N digits
;N is an integer power of 2
;Assumes yucky big endian
    X.loc = scrap-6+4N  ;\
    Y.loc = X.loc+N     ; |The location of the local X,Y, and Z vars
    Z.loc = Y.loc+N     ; |M is the location of the results of sub multiplies
    M.loc = X.loc-N     ;/
    IF N==1:            ;\
        X*Y->Z[0:2]     ; |Base Case
        RETURN          ;/
    COPY(X,inputX,N)    ;Copy the input X to the local X variable
    COPY(Y,inputY,N)    ;Copy the input Y to the local Y variable
    MUL(X[N/2:N],Y[N/2:N],N/2)  ;Multiply the bottom half of the digits of X and Y
    COPY(Z[N:2N],M,N)           ;Copy the result of the previous multiply to the bottom half of Z
    MUL(X[0:N/2],Y[0:N/2],N/2)  ;Multiply the top half of the digits of X and Y
    COPY(Z[0:N],M,N)            ;Copy the result of the previous multiply to the top half of Z
    sign=0
    SUB(X[0:N/2],X[N/2:N],N/2)  ;Subtract the bottom half of X from the top half, returning flags
    IF underflow
        NEG(X[0:N/2],N/2)
        sign=1
    SUB(Y[0:N/2],Y[N/2:N],N/2)  ;Subtract the bottom half of Y from the top half, returning flags
    IF underflow
        NEG(Y[0:N/2],N/2)
        sign=~sign
    MUL(X[0:N/2],Y[0:N/2],N/2)
    SUB(M[0:N],Z[0:N],N)
    SUB(M[0:N],Z[N:2N],N)
    IF sign==1
        ADDir(Z[0:3N/2],M[0:N],N) ;ADD N digits, then continue as long as there is overflow
        RETURN
    ELSE
        SUBir(Z[0:3N/2],M[0:N]) ;SUB N digits, then continue as long as there is underflow
        RETURN