+ * Revision 1.16 2002/01/13 13:40:31 mdw
+ * Avoid trashing arguments before we've used them.
+ *
+ * Revision 1.15 2001/06/16 13:00:20 mdw
+ * Use the generic exponentiation functions.
+ *
+ * Revision 1.14 2001/02/22 09:04:26 mdw
+ * Cosmetic fix.
+ *
+ * Revision 1.13 2001/02/03 12:00:29 mdw
+ * Now @mp_drop@ checks its argument is non-NULL before attempting to free
+ * it. Note that the macro version @MP_DROP@ doesn't do this.
+ *
+ * Revision 1.12 2000/10/08 15:48:35 mdw
+ * Rename Karatsuba constants now that we have @gfx_kmul@ too.
+ *
+ * Revision 1.11 2000/10/08 12:04:27 mdw
+ * (mpmont_reduce, mpmont_mul): Cope with negative numbers.
+ *
+ * Revision 1.10 2000/07/29 17:05:43 mdw
+ * (mpmont_expr): Use sliding window exponentiation, with a drop-through
+ * for small exponents to use a simple left-to-right bitwise routine. This
+ * can reduce modexp times by up to a quarter.
+ *
+ * Revision 1.9 2000/06/17 11:45:09 mdw
+ * Major memory management overhaul. Added arena support. Use the secure
+ * arena for secret integers. Replace and improve the MP management macros
+ * (e.g., replace MP_MODIFY by MP_DEST).
+ *
+ * Revision 1.8 1999/12/22 15:55:00 mdw
+ * Adjust Karatsuba parameters.
+ *
+ * Revision 1.7 1999/12/11 01:51:14 mdw
+ * Use a Karatsuba-based reduction for large moduli.
+ *
+ * Revision 1.6 1999/12/10 23:18:39 mdw
+ * Change interface for suggested destinations.
+ *
+ * Revision 1.5 1999/11/22 13:58:40 mdw
+ * Add an option to disable Montgomery reduction, so that performance
+ * comparisons can be done.
+ *
+ * Revision 1.4 1999/11/21 12:27:06 mdw
+ * Remove a division from the Montgomery setup by calculating
+ * %$R^2 \bmod m$% first and then %$R \bmod m$% by Montgomery reduction of
+ * %$R^2$%.
+ *
+ * Revision 1.3 1999/11/21 11:35:10 mdw
+ * Performance improvement: use @mp_sqr@ and @mpmont_reduce@ instead of
+ * @mpmont_mul@ for squaring in exponentiation.
+ *
+ * Revision 1.2 1999/11/19 13:17:26 mdw
+ * Add extra interface to exponentiation which returns a Montgomerized
+ * result.
+ *