diff --git a/bn_fast_mp_montgomery_reduce.c b/bn_fast_mp_montgomery_reduce.c
index 54d9b0a..43a4d37 100644
--- a/bn_fast_mp_montgomery_reduce.c
+++ b/bn_fast_mp_montgomery_reduce.c
@@ -28,6 +28,10 @@ int fast_mp_montgomery_reduce(mp_int *x, const mp_int *n, mp_digit rho)
    int     ix, res, olduse;
    mp_word W[MP_WARRAY];
 
+   if (x->used > MP_WARRAY) {
+      return MP_VAL;
+   }
+
    /* get old used count */
    olduse = x->used;
 
diff --git a/bn_fast_s_mp_mul_digs.c b/bn_fast_s_mp_mul_digs.c
index 558d151..875798e 100644
--- a/bn_fast_s_mp_mul_digs.c
+++ b/bn_fast_s_mp_mul_digs.c
@@ -87,7 +87,7 @@ int fast_s_mp_mul_digs(const mp_int *a, const mp_int *b, mp_int *c, int digs)
    {
       mp_digit *tmpc;
       tmpc = c->dp;
-      for (ix = 0; ix < (pa + 1); ix++) {
+      for (ix = 0; ix < pa; ix++) {
          /* now extract the previous digit [below the carry] */
          *tmpc++ = W[ix];
       }
diff --git a/bn_mp_lshd.c b/bn_mp_lshd.c
index 888989a..b49b545 100644
--- a/bn_mp_lshd.c
+++ b/bn_mp_lshd.c
@@ -24,6 +24,10 @@ int mp_lshd(mp_int *a, int b)
    if (b <= 0) {
       return MP_OKAY;
    }
+   /* no need to shift 0 around */
+   if (mp_iszero(a) == MP_YES) {
+      return MP_OKAY;
+   }
 
    /* grow to fit the new digits */
    if (a->alloc < (a->used + b)) {
diff --git a/bn_mp_montgomery_reduce.c b/bn_mp_montgomery_reduce.c
index a38173e..a9c7752 100644
--- a/bn_mp_montgomery_reduce.c
+++ b/bn_mp_montgomery_reduce.c
@@ -29,6 +29,7 @@ int mp_montgomery_reduce(mp_int *x, const mp_int *n, mp_digit rho)
     */
    digs = (n->used * 2) + 1;
    if ((digs < MP_WARRAY) &&
+       (x->used <= MP_WARRAY) &&
        (n->used <
         (1 << ((CHAR_BIT * sizeof(mp_word)) - (2 * DIGIT_BIT))))) {
       return fast_mp_montgomery_reduce(x, n, rho);