Big changes for performance. mp_digits are now always unsigned ints.

mp_words are used only on machines that support long long arithmetic.
s_mp_mod_d() was deleted.  It was not being used and was not part of the
public API. The code that computes squares in s_mp_sqr was broken out
into a separate new function s_mpv_sqr_add_prop(), which is a target for
assembly language optimization.  New function s_mpv_div_2dx1d(), also a
target for assembly optimization.  These changes made X86 benchmark time
go from 22.5 seconds to 8.3 seconds on my reference test system.

security/nss/lib/freebl/mpi/mpi-priv.h

@@ -38,7 +38,7 @@
  * the GPL.  If you do not delete the provisions above, a recipient
  * may use your version of this file under either the MPL or the GPL.
  *
- *  $Id: mpi-priv.h,v 1.9 2000-08-22 01:57:33 nelsonb%netscape.com Exp $
+ *  $Id: mpi-priv.h,v 1.10 2000-08-31 02:51:23 nelsonb%netscape.com Exp $
  */
 #ifndef _MPI_PRIV_H_
 #define _MPI_PRIV_H_ 1
@@ -191,14 +191,13 @@ void     s_mp_div_2d(mp_int *mp, mp_digit d);  /* divide by 2^d in place  */
 void     s_mp_mod_2d(mp_int *mp, mp_digit d);  /* modulo 2^d in place     */
 void     s_mp_div_2(mp_int *mp);               /* divide by 2 in place    */
 mp_err   s_mp_mul_2(mp_int *mp);               /* multiply by 2 in place  */
-mp_digit s_mp_norm(mp_int *a, mp_int *b);      /* normalize for division  */
+mp_err   s_mp_norm(mp_int *a, mp_int *b, mp_digit *pd); 
+                                               /* normalize for division  */
 mp_err   s_mp_add_d(mp_int *mp, mp_digit d);   /* unsigned digit addition */
 mp_err   s_mp_sub_d(mp_int *mp, mp_digit d);   /* unsigned digit subtract */
 mp_err   s_mp_mul_d(mp_int *mp, mp_digit d);   /* unsigned digit multiply */
 mp_err   s_mp_div_d(mp_int *mp, mp_digit d, mp_digit *r);
 		                               /* unsigned digit divide   */
-mp_err   s_mp_mod_d(mp_int *mp, mp_digit d, mp_digit *r);
-                                               /* unsigned digit rem      */
 mp_err   s_mp_reduce(mp_int *x, const mp_int *m, const mp_int *mu);
                                                /* Barrett reduction       */
 mp_err   s_mp_add(mp_int *a, const mp_int *b); /* magnitude addition      */
@@ -232,6 +231,9 @@ void     s_mpv_mul_d_add(const mp_digit *a, mp_size a_len, mp_digit b,
 			 mp_digit *c);
 void     s_mpv_mul_d_add_prop(const mp_digit *a, mp_size a_len, mp_digit b, 
 			      mp_digit *c);
+void     s_mpv_sqr_add_prop(const mp_digit *a, mp_size a_len, mp_digit *sqrs);
+mp_err   s_mpv_div_2dx1d(mp_digit Nhi, mp_digit Nlo, mp_digit divisor, 
+		         mp_digit *quot, mp_digit *rem);
 
 /* c += a * b * (MP_RADIX ** offset);  */
 #define s_mp_mul_d_add_offset(a, b, c, off) \

security/nss/lib/freebl/mpi/mpi.h

@@ -36,7 +36,7 @@
  * may use your version of this file under either the MPL or the
  * GPL.
  *
- *  $Id: mpi.h,v 1.8 2000-08-22 01:57:33 nelsonb%netscape.com Exp $
+ *  $Id: mpi.h,v 1.9 2000-08-31 02:51:23 nelsonb%netscape.com Exp $
  */
 
 #ifndef _H_MPI_
@@ -72,7 +72,7 @@ typedef unsigned int      mp_sign;
 typedef unsigned int      mp_size;
 typedef int               mp_err;
 
-#ifndef MP_USE_32
+#ifndef MP_NO_MP_WORD
 #if defined(ULONG_LONG_MAX)			/* GCC, HPUX */
 #define MP_ULONG_LONG_MAX ULONG_LONG_MAX
 #elif defined(ULLONG_MAX)			/* Solaris */
@@ -81,10 +81,10 @@ typedef int               mp_err;
 #define MP_ULONG_LONG_MAX ULONGLONG_MAX
 #endif
 
-#if defined(MP_ULONG_LONG_MAX) && MP_ULONG_LONG_MAX > UINT_MAX
 typedef unsigned int      mp_digit;
 #define MP_DIGIT_MAX      UINT_MAX
 
+#if defined(MP_ULONG_LONG_MAX) && MP_ULONG_LONG_MAX > UINT_MAX
 #if MP_ULONG_LONG_MAX == ULONG_MAX
 typedef unsigned long     mp_word;
 typedef          long     mp_sword;
@@ -94,29 +94,29 @@ typedef unsigned long long mp_word;
 typedef          long long mp_sword;
 #define MP_WORD_MAX       MP_ULONG_LONG_MAX
 #endif
+#else /* MP_ULONG_LONG_MAX not defined */
+#define MP_NO_MP_WORD 1
 #endif
-#endif /* !USE_32 */
+#endif /* !MP_NO_MP_WORD */
 
-#if !defined(MP_DIGIT_MAX)
-#if ULONG_MAX == UINT_MAX
-typedef unsigned short    mp_digit;
+#if !defined(MP_WORD_MAX) && defined(MP_DEFINE_SMALL_WORD)
 typedef unsigned int      mp_word;
 typedef          int      mp_sword;
-#define MP_DIGIT_MAX      USHRT_MAX
 #define MP_WORD_MAX       UINT_MAX
-#else
-typedef unsigned int      mp_digit;
-typedef unsigned long     mp_word;
-typedef          long     mp_sword;
-#define MP_DIGIT_MAX      UINT_MAX
-#define MP_WORD_MAX       ULONG_MAX
-#endif
 #endif
 
 #define MP_DIGIT_BIT      (CHAR_BIT*sizeof(mp_digit))
 #define MP_WORD_BIT       (CHAR_BIT*sizeof(mp_word))
 #define MP_RADIX          (1+(mp_word)MP_DIGIT_MAX)
 
+#define MP_HALF_DIGIT_BIT (MP_DIGIT_BIT/2)
+#define MP_HALF_DIGIT_MAX USHRT_MAX
+#define MP_HALF_RADIX     (1+(mp_digit)MP_HALF_DIGIT_MAX)
+/* MP_HALF_RADIX really ought to be called MP_SQRT_RADIX, but it's named 
+** MP_HALF_RADIX because it's the radix for MP_HALF_DIGITs, and it's 
+** consistent with the other _HALF_ names.
+*/
+
 #if MP_DIGIT_MAX == USHRT_MAX
 #define MP_DIGIT_FMT      "%04X"     /* printf() format for 1 digit */
 #else