handling overflow in the fixnum*fixnum case of sexp_mul, which is never reached directly from the vm

This commit is contained in:
Alex Shinn 2011-11-28 13:12:46 +00:00
parent e7dbddadef
commit d9b57ab6b1

View file

@ -1113,6 +1113,7 @@ sexp sexp_sub (sexp ctx, sexp a, sexp b) {
}
sexp sexp_mul (sexp ctx, sexp a, sexp b) {
sexp_lsint_t prod;
int at=sexp_number_type(a), bt=sexp_number_type(b), t;
sexp r=SEXP_VOID;
sexp_gc_var1(tmp);
@ -1127,7 +1128,11 @@ sexp sexp_mul (sexp ctx, sexp a, sexp b) {
r = sexp_type_exception(ctx, NULL, SEXP_NUMBER, a);
break;
case SEXP_NUM_FIX_FIX:
r = sexp_fx_mul(a, b);
prod = (sexp_lsint_t)sexp_unbox_fixnum(a) * sexp_unbox_fixnum(b);
if ((prod < SEXP_MIN_FIXNUM) || (prod > SEXP_MAX_FIXNUM))
r = sexp_mul(ctx, tmp=sexp_fixnum_to_bignum(ctx, a), b);
else
r = sexp_make_fixnum(prod);
break;
case SEXP_NUM_FIX_FLO:
r = sexp_make_flonum(ctx, sexp_fixnum_to_double(a)*sexp_flonum_value(b));