added tomsfastmath-0.12

changes.txt

@@ -1,3 +1,14 @@
+March 14th, 2007
+0.12 -- Christophe Devine contributed MIPS asm w00t
+     ++ quick release to get the MIPS code out there
+
+March 10th, 2007
+0.11 -- re-org'ed the source tree, it's cooler now
+     -- cleaned up mul/sqr so there is only one file per unit, should help building with older compilers and/or on slower boxes
+     -- [CRI] optimized fp_read_unsigned_bin
+     -- fixed -0 ... again ... I now have less hair on my head.
+     -- [CRI] renamed bn_reverse() -> fp_reverse()
+
 November 1st, 2006
 0.10 -- Unrolled Montgomery for 1..16 digits with TFM_SMALL_MONT_SET between 10% and 25% speedup depending on size.
      -- fixed fp_sqr_comba.c so it builds in ISO C mode [Andreas Lange]

demo/test.c

@@ -411,6 +411,93 @@ sqrtime:
       }
       printf("%5lu-bit: %9llu\n", t * DIGIT_BIT, t2);
   }
+invmodtime:
+  printf("Invmod:\n");
+  for (t = 2; t < FP_SIZE/2; t += 2) {
+     fp_zero(&a);
+     for (ix = 0; ix < t; ix++) {
+         a.dp[ix] = ix | 1;
+     }
+     a.used = t;
+     fp_zero(&b);
+     for (ix = 0; ix < t; ix++) {
+         b.dp[ix] = rand();
+     }
+     b.used = t;
+     fp_clamp(&b);
+     fp_zero(&c);
+     t2 = -1;
+     for (ix = 0; ix < 100; ++ix) {
+          t1 = TIMFUNC();
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          fp_invmod(&b, &a, &c);
+          t2 = (TIMFUNC() - t1)>>6;
+          if (t1<t2) { --ix; t2 = t1; }
+      }
+      printf("%5lu-bit: %9llu\n", t * DIGIT_BIT, t2);
+  }
 //#else
 monttime:
   printf("Montgomery:\n");
@@ -531,10 +618,10 @@ expttime:
      }
      printf("%5lu-bit: %9llu\n", t * DIGIT_BIT, t2);
   }
-  return;
+  return 0;
 #endif
 
-return;
+return 0;
 testing:
 
   fp_zero(&b); fp_zero(&c); fp_zero(&d); fp_zero(&e); fp_zero(&f); fp_zero(&a);

filter.pl

@@ -0,0 +1,30 @@
+#!/usr/bin/perl
+
+# we want to filter every between START_INS and END_INS out and then insert crap from another file (this is fun)
+
+$dst = shift;
+$ins = shift;
+
+open(SRC,"<$dst");
+open(INS,"<$ins");
+open(TMP,">tmp.delme");
+
+$l = 0;
+while (<SRC>) {
+   if ($_ =~ /START_INS/) {
+      print TMP $_;
+      $l = 1;
+      while (<INS>) {
+         print TMP $_;
+      }
+      close INS;
+   } elsif ($_ =~ /END_INS/) {
+      print TMP $_;
+      $l = 0;
+   } elsif ($l == 0) {
+      print TMP $_;
+   }
+}
+
+close TMP;
+close SRC;

fp_read_unsigned_bin.c

@@ -1,28 +0,0 @@
-/* TomsFastMath, a fast ISO C bignum library.
- * 
- * This project is meant to fill in where LibTomMath
- * falls short.  That is speed ;-)
- *
- * This project is public domain and free for all purposes.
- * 
- * Tom St Denis, tomstdenis@gmail.com
- */
-#include <tfm.h>
-
-void fp_read_unsigned_bin(fp_int *a, unsigned char *b, int c)
-{
-  /* zero the int */
-  fp_zero (a);
-
-  /* read the bytes in */
-  for (; c > 0; c--) {
-    fp_mul_2d (a, 8, a);
-    a->dp[0] |= *b++;
-    a->used += 1;
-  }
-  fp_clamp (a);
-}
-
-/* $Source$ */
-/* $Revision$ */
-/* $Date$ */

gen.pl

@@ -7,6 +7,7 @@ use strict;
 
 open( OUT, ">mpi.c" ) or die "Couldn't open mpi.c for writing: $!";
 foreach my $filename (glob "*fp_*.c") {
+   next if ($filename eq "fp_sqr_comba_generic.c");
    open( SRC, "<$filename" ) or die "Couldn't open $filename for reading: $!";
    print OUT "/* Start: $filename */\n";
    print OUT while <SRC>;

genlist.sh

@@ -0,0 +1,9 @@
+#!/bin/bash
+export a=`find . -type f | sort | grep "[.]/src" | grep "[.]c" | grep -v generators | sed -e 'sE\./EE' | sed -e 's/\.c/\.o/' | xargs`
+perl ./parsenames.pl OBJECTS "$a"
+export a=`find . -type f | grep [.]/src | grep [.]h | sed -e 'se\./ee' | xargs`
+perl ./parsenames.pl HEADERS "$a"
+
+# $Source: /cvs/libtom/tomsfastmath/genlist.sh,v $   
+# $Revision: 1.1 $   
+# $Date: 2006/12/31 21:31:40 $ 

makefile

@@ -1,9 +1,9 @@
 #makefile for TomsFastMath
 #
 #
-VERSION=0.10
+VERSION=0.12
 
-CFLAGS += -Wall -W -Wshadow -I./ 
+CFLAGS += -Wall -W -Wshadow -Isrc/headers
 
 ifndef MAKE
    MAKE=make
@@ -22,35 +22,34 @@ CFLAGS += -fomit-frame-pointer
 
 endif
 
-OBJECTS = \
+#START_INS
-fp_set.o \
+OBJECTS=src/addsub/fp_add.o src/addsub/fp_add_d.o src/addsub/fp_addmod.o src/addsub/fp_cmp.o \
-\
+src/addsub/fp_cmp_d.o src/addsub/fp_cmp_mag.o src/addsub/fp_sub.o src/addsub/fp_sub_d.o \
-fp_rshd.o fp_lshd.o fp_div_2d.o fp_mod_2d.o fp_mul_2d.o fp_2expt.o \
+src/addsub/fp_submod.o src/addsub/s_fp_add.o src/addsub/s_fp_sub.o src/bin/fp_radix_size.o \
-fp_mul_2.o fp_div_2.o  \
+src/bin/fp_read_radix.o src/bin/fp_read_signed_bin.o src/bin/fp_read_unsigned_bin.o \
-\
+src/bin/fp_reverse.o src/bin/fp_s_rmap.o src/bin/fp_signed_bin_size.o src/bin/fp_to_signed_bin.o \
-fp_cnt_lsb.o \
+src/bin/fp_to_unsigned_bin.o src/bin/fp_toradix.o src/bin/fp_unsigned_bin_size.o src/bit/fp_cnt_lsb.o \
-\
+src/bit/fp_count_bits.o src/bit/fp_div_2.o src/bit/fp_div_2d.o src/bit/fp_lshd.o src/bit/fp_mod_2d.o \
-fp_add.o fp_sub.o fp_mul.o fp_sqr.o fp_div.o fp_mod.o \
+src/bit/fp_rshd.o src/divide/fp_div.o src/divide/fp_div_d.o src/divide/fp_mod.o src/divide/fp_mod_d.o \
-s_fp_add.o s_fp_sub.o \
+src/exptmod/fp_2expt.o src/exptmod/fp_exptmod.o src/misc/fp_ident.o src/misc/fp_set.o \
-\
+src/mont/fp_montgomery_calc_normalization.o src/mont/fp_montgomery_reduce.o \
-fp_cmp_d.o fp_add_d.o fp_sub_d.o fp_mul_d.o fp_div_d.o fp_mod_d.o \
+src/mont/fp_montgomery_setup.o src/mul/fp_mul.o src/mul/fp_mul_2.o src/mul/fp_mul_2d.o \
-fp_addmod.o fp_submod.o fp_mulmod.o fp_sqrmod.o fp_invmod.o \
+src/mul/fp_mul_comba.o src/mul/fp_mul_comba_12.o src/mul/fp_mul_comba_17.o src/mul/fp_mul_comba_20.o \
-fp_gcd.o fp_lcm.o fp_prime_miller_rabin.o fp_isprime.o \
+src/mul/fp_mul_comba_24.o src/mul/fp_mul_comba_28.o src/mul/fp_mul_comba_3.o src/mul/fp_mul_comba_32.o \
-fp_prime_random_ex.o fp_mul_comba.o fp_sqr_comba.o \
+src/mul/fp_mul_comba_4.o src/mul/fp_mul_comba_48.o src/mul/fp_mul_comba_6.o src/mul/fp_mul_comba_64.o \
-\
+src/mul/fp_mul_comba_7.o src/mul/fp_mul_comba_8.o src/mul/fp_mul_comba_9.o \
-fp_montgomery_setup.o fp_montgomery_calc_normalization.o fp_montgomery_reduce.o \
+src/mul/fp_mul_comba_small_set.o src/mul/fp_mul_d.o src/mul/fp_mulmod.o src/numtheory/fp_gcd.o \
-\
+src/numtheory/fp_invmod.o src/numtheory/fp_isprime.o src/numtheory/fp_lcm.o \
-fp_exptmod.o \
+src/numtheory/fp_prime_miller_rabin.o src/numtheory/fp_prime_random_ex.o src/sqr/fp_sqr.o \
-\
+src/sqr/fp_sqr_comba.o src/sqr/fp_sqr_comba_12.o src/sqr/fp_sqr_comba_17.o src/sqr/fp_sqr_comba_20.o \
-fp_cmp.o fp_cmp_mag.o \
+src/sqr/fp_sqr_comba_24.o src/sqr/fp_sqr_comba_28.o src/sqr/fp_sqr_comba_3.o src/sqr/fp_sqr_comba_32.o \
-\
+src/sqr/fp_sqr_comba_4.o src/sqr/fp_sqr_comba_48.o src/sqr/fp_sqr_comba_6.o src/sqr/fp_sqr_comba_64.o \
-fp_unsigned_bin_size.o fp_read_unsigned_bin.o fp_to_unsigned_bin.o \
+src/sqr/fp_sqr_comba_7.o src/sqr/fp_sqr_comba_8.o src/sqr/fp_sqr_comba_9.o \
-fp_signed_bin_size.o fp_read_signed_bin.o fp_to_signed_bin.o \
+src/sqr/fp_sqr_comba_generic.o src/sqr/fp_sqr_comba_small_set.o src/sqr/fp_sqrmod.o 
-fp_read_radix.o fp_toradix.o fp_radix_size.o fp_count_bits.o fp_reverse.o fp_s_rmap.o \
-\
-fp_ident.o 
 
-HEADERS=tfm.h
+HEADERS=src/headers/tfm.h 
+
+#END_INS
 
 ifndef LIBPATH
    LIBPATH=/usr/lib
@@ -97,6 +96,14 @@ test: $(LIBNAME) demo/test.o mtest/mtest
 timing: $(LIBNAME) demo/test.o
 	$(CC) $(CFLAGS) demo/test.o $(LIBNAME) $(PROF) -o test
 
+profiled:
+	CFLAGS="${CFLAGS} -fprofile-generate" MAKE=${MAKE} ${MAKE} timing
+	./test
+	rm -f `find . -type f | grep "[.]o" | xargs`
+	rm -f `find . -type f | grep "[.]a" | xargs`
+	rm -f test
+	CFLAGS="${CFLAGS} -fprofile-use" MAKE=${MAKE} ${MAKE} timing
+	
 stest: $(LIBNAME) demo/stest.o 
 	$(CC) $(CFLAGS) demo/stest.o $(LIBNAME) -o stest
 
@@ -148,5 +155,5 @@ zipup: no_oops docs clean
 	mv -f tfm* ~ ; rm -rf tomsfastmath-$(VERSION)
 
 # $Source: /cvs/libtom/tomsfastmath/makefile,v $ 
-# $Revision: 1.29 $ 
+# $Revision: 1.38 $ 
-# $Date: 2006/10/22 13:27:10 $ 
+# $Date: 2007/03/13 01:23:03 $ 

makefile.shared

@@ -1,11 +1,11 @@
 #makefile for TomsFastMath
 #
 #
-VERSION=0:10
+VERSION=0:12
 
 CC=libtool --mode=compile --tag=CC gcc
 
-CFLAGS += -Wall -W -Wshadow -I./ 
+CFLAGS += -Wall -W -Wshadow -Isrc/headers
 
 ifndef IGNORE_SPEED
 
@@ -20,36 +20,35 @@ CFLAGS += -fomit-frame-pointer
 
 endif
 
+#START_INS
+OBJECTS=src/addsub/fp_add.o src/addsub/fp_add_d.o src/addsub/fp_addmod.o src/addsub/fp_cmp.o \
+src/addsub/fp_cmp_d.o src/addsub/fp_cmp_mag.o src/addsub/fp_sub.o src/addsub/fp_sub_d.o \
+src/addsub/fp_submod.o src/addsub/s_fp_add.o src/addsub/s_fp_sub.o src/bin/fp_radix_size.o \
+src/bin/fp_read_radix.o src/bin/fp_read_signed_bin.o src/bin/fp_read_unsigned_bin.o \
+src/bin/fp_reverse.o src/bin/fp_s_rmap.o src/bin/fp_signed_bin_size.o src/bin/fp_to_signed_bin.o \
+src/bin/fp_to_unsigned_bin.o src/bin/fp_toradix.o src/bin/fp_unsigned_bin_size.o src/bit/fp_cnt_lsb.o \
+src/bit/fp_count_bits.o src/bit/fp_div_2.o src/bit/fp_div_2d.o src/bit/fp_lshd.o src/bit/fp_mod_2d.o \
+src/bit/fp_rshd.o src/divide/fp_div.o src/divide/fp_div_d.o src/divide/fp_mod.o src/divide/fp_mod_d.o \
+src/exptmod/fp_2expt.o src/exptmod/fp_exptmod.o src/misc/fp_ident.o src/misc/fp_set.o \
+src/mont/fp_montgomery_calc_normalization.o src/mont/fp_montgomery_reduce.o \
+src/mont/fp_montgomery_setup.o src/mul/fp_mul.o src/mul/fp_mul_2.o src/mul/fp_mul_2d.o \
+src/mul/fp_mul_comba.o src/mul/fp_mul_comba_12.o src/mul/fp_mul_comba_17.o src/mul/fp_mul_comba_20.o \
+src/mul/fp_mul_comba_24.o src/mul/fp_mul_comba_28.o src/mul/fp_mul_comba_3.o src/mul/fp_mul_comba_32.o \
+src/mul/fp_mul_comba_4.o src/mul/fp_mul_comba_48.o src/mul/fp_mul_comba_6.o src/mul/fp_mul_comba_64.o \
+src/mul/fp_mul_comba_7.o src/mul/fp_mul_comba_8.o src/mul/fp_mul_comba_9.o \
+src/mul/fp_mul_comba_small_set.o src/mul/fp_mul_d.o src/mul/fp_mulmod.o src/numtheory/fp_gcd.o \
+src/numtheory/fp_invmod.o src/numtheory/fp_isprime.o src/numtheory/fp_lcm.o \
+src/numtheory/fp_prime_miller_rabin.o src/numtheory/fp_prime_random_ex.o src/sqr/fp_sqr.o \
+src/sqr/fp_sqr_comba.o src/sqr/fp_sqr_comba_12.o src/sqr/fp_sqr_comba_17.o src/sqr/fp_sqr_comba_20.o \
+src/sqr/fp_sqr_comba_24.o src/sqr/fp_sqr_comba_28.o src/sqr/fp_sqr_comba_3.o src/sqr/fp_sqr_comba_32.o \
+src/sqr/fp_sqr_comba_4.o src/sqr/fp_sqr_comba_48.o src/sqr/fp_sqr_comba_6.o src/sqr/fp_sqr_comba_64.o \
+src/sqr/fp_sqr_comba_7.o src/sqr/fp_sqr_comba_8.o src/sqr/fp_sqr_comba_9.o \
+src/sqr/fp_sqr_comba_generic.o src/sqr/fp_sqr_comba_small_set.o src/sqr/fp_sqrmod.o 
 
-OBJECTS = \
+HEADERS=src/headers/tfm.h 
-fp_set.o \
+
-\
+#END_INS
-fp_rshd.o fp_lshd.o fp_div_2d.o fp_mod_2d.o fp_mul_2d.o fp_2expt.o \
-fp_mul_2.o fp_div_2.o  \
-\
-fp_cnt_lsb.o \
-\
-fp_add.o fp_sub.o fp_mul.o fp_sqr.o fp_div.o fp_mod.o \
-s_fp_add.o s_fp_sub.o \
-\
-fp_cmp_d.o fp_add_d.o fp_sub_d.o fp_mul_d.o fp_div_d.o fp_mod_d.o \
-fp_addmod.o fp_submod.o fp_mulmod.o fp_sqrmod.o fp_invmod.o \
-fp_gcd.o fp_lcm.o fp_prime_miller_rabin.o fp_isprime.o \
-fp_prime_random_ex.o fp_mul_comba.o fp_sqr_comba.o \
-\
-fp_montgomery_setup.o fp_montgomery_calc_normalization.o fp_montgomery_reduce.o \
-\
-fp_exptmod.o \
-\
-fp_cmp.o fp_cmp_mag.o \
-\
-fp_unsigned_bin_size.o fp_read_unsigned_bin.o fp_to_unsigned_bin.o \
-fp_signed_bin_size.o fp_read_signed_bin.o fp_to_signed_bin.o \
-fp_read_radix.o fp_toradix.o fp_radix_size.o fp_count_bits.o fp_reverse.o fp_s_rmap.o \
-\
-fp_ident.o 
 
-HEADERS=tfm.h
 
 ifndef LIBPATH
    LIBPATH=/usr/lib
@@ -105,6 +104,6 @@ stest: $(LIBNAME) demo/stest.o
 	$(CC) $(CFLAGS) demo/stest.o $(LIBNAME_S) -o stest
 
 # $Source: /cvs/libtom/tomsfastmath/makefile.shared,v $ 
-# $Revision: 1.14 $ 
+# $Revision: 1.19 $ 
-# $Date: 2006/10/22 13:27:10 $ 
+# $Date: 2007/03/13 01:23:03 $ 
 

mtest/mtest.c

@@ -39,7 +39,7 @@ mulmod
 #include <time.h>
 #include <tommath.h>
 #define CRYPT
-#include "../tfm.h"
+#include "../src/headers/tfm.h"
 
 FILE *rng;
 

parsenames.pl

@@ -0,0 +1,26 @@
+#!/usr/bin/perl
+#
+# Splits the list of files and outputs for makefile type files 
+# wrapped at 80 chars 
+# 
+# Tom St Denis
+@a = split(" ", $ARGV[1]);
+$b = "$ARGV[0]=";
+$len = length($b);
+print $b;
+foreach my $obj (@a) {
+   $len = $len + length($obj);
+   $obj =~ s/\*/\$/;
+   if ($len > 100) {
+      printf "\\\n";
+      $len = length($obj);
+   }
+   print "$obj ";
+}
+#if ($ARGV[0] eq "HEADERS") { print "testprof/tomcrypt_test.h"; }
+
+print "\n\n";
+
+# $Source: /cvs/libtom/tomsfastmath/parsenames.pl,v $   
+# $Revision: 1.1 $   
+# $Date: 2006/12/31 21:31:40 $ 

src/bin/fp_read_unsigned_bin.c

@@ -0,0 +1,66 @@
+/* TomsFastMath, a fast ISO C bignum library.
+ * 
+ * This project is meant to fill in where LibTomMath
+ * falls short.  That is speed ;-)
+ *
+ * This project is public domain and free for all purposes.
+ * 
+ * Tom St Denis, tomstdenis@gmail.com
+ */
+#include <tfm.h>
+
+void fp_read_unsigned_bin(fp_int *a, unsigned char *b, int c)
+{
+  /* zero the int */
+  fp_zero (a);
+
+  /* If we know the endianness of this architecture, and we're using
+     32-bit fp_digits, we can optimize this */
+#if (defined(ENDIAN_LITTLE) || defined(ENDIAN_BIG)) && !defined(FP_64BIT)
+  /* But not for both simultaneously */
+#if defined(ENDIAN_LITTLE) && defined(ENDIAN_BIG)
+#error Both ENDIAN_LITTLE and ENDIAN_BIG defined.
+#endif
+  {
+     unsigned char *pd = (unsigned char *)a->dp;
+
+     if ((unsigned)c > (FP_SIZE * sizeof(fp_digit))) {
+        int excess = c - (FP_SIZE * sizeof(fp_digit));
+        c -= excess;
+        b += excess;
+     }
+     a->used = (c + sizeof(fp_digit) - 1)/sizeof(fp_digit);
+     /* read the bytes in */
+#ifdef ENDIAN_BIG
+     {
+       /* Use Duff's device to unroll the loop. */
+       int idx = (c - 1) & ~3;
+       switch (c % 4) {
+       case 0:	do { pd[idx+0] = *b++;
+       case 3:	     pd[idx+1] = *b++;
+       case 2:	     pd[idx+2] = *b++;
+       case 1:	     pd[idx+3] = *b++;
+                     idx -= 4;
+	 	        } while ((c -= 4) > 0);
+       }
+     }
+#else
+     for (c -= 1; c >= 0; c -= 1) {
+       pd[c] = *b++;
+     }
+#endif
+  }
+#else
+  /* read the bytes in */
+  for (; c > 0; c--) {
+     fp_mul_2d (a, 8, a);
+     a->dp[0] |= *b++;
+     a->used += 1;
+  }
+#endif
+  fp_clamp (a);
+}
+
+/* $Source$ */
+/* $Revision$ */
+/* $Date$ */

src/bin/fp_reverse.c

@@ -10,7 +10,7 @@
 #include <tfm.h>
 
 /* reverse an array, used for radix code */
-void bn_reverse (unsigned char *s, int len)
+void fp_reverse (unsigned char *s, int len)
 {
   int     ix, iy;
   unsigned char t;

src/bin/fp_to_unsigned_bin.c

@@ -21,7 +21,7 @@ void fp_to_unsigned_bin(fp_int *a, unsigned char *b)
       b[x++] = (unsigned char) (t.dp[0] & 255);
       fp_div_2d (&t, 8, &t, NULL);
   }
-  bn_reverse (b, x);
+  fp_reverse (b, x);
 }
 
 /* $Source$ */

src/bin/fp_toradix.c

@@ -47,7 +47,7 @@ int fp_toradix(fp_int *a, char *str, int radix)
   /* reverse the digits of the string.  In this case _s points
    * to the first digit [exluding the sign] of the number]
    */
-  bn_reverse ((unsigned char *)_s, digs);
+  fp_reverse ((unsigned char *)_s, digs);
 
   /* append a NULL so the string is properly terminated */
   *str = '\0';

src/divide/fp_div_d.c

@@ -13,7 +13,12 @@ static int s_is_power_of_two(fp_digit b, int *p)
 {
    int x;
 
-   for (x = 1; x < DIGIT_BIT; x++) {
+   /* fast return if no power of two */
+   if ((b==0) || (b & (b-1))) {
+      return 0;
+   }
+
+   for (x = 0; x < DIGIT_BIT; x++) {
       if (b == (((fp_digit)1)<<x)) {
          *p = x;
          return 1;

src/headers/tfm.h

@@ -315,13 +315,13 @@ void fp_set(fp_int *a, fp_digit b);
 #define fp_copy(a, b)      (void)(((a) != (b)) && memcpy((b), (a), sizeof(fp_int)))
 #define fp_init_copy(a, b) fp_copy(b, a)
 
-/* negate and absolute */
-#define fp_neg(a, b)  { fp_copy(a, b); (b)->sign ^= 1; }
-#define fp_abs(a, b)  { fp_copy(a, b); (b)->sign  = 0; }
-
 /* clamp digits */
 #define fp_clamp(a)   { while ((a)->used && (a)->dp[(a)->used-1] == 0) --((a)->used); (a)->sign = (a)->used ? (a)->sign : FP_ZPOS; }
 
+/* negate and absolute */
+#define fp_neg(a, b)  { fp_copy(a, b); (b)->sign ^= 1; fp_clamp(b); }
+#define fp_abs(a, b)  { fp_copy(a, b); (b)->sign  = 0; }
+
 /* right shift x digits */
 void fp_rshd(fp_int *a, int x);
 
@@ -457,7 +457,7 @@ int fp_toradix_n(fp_int * a, char *str, int radix, int maxlen);
 /* VARIOUS LOW LEVEL STUFFS */
 void s_fp_add(fp_int *a, fp_int *b, fp_int *c);
 void s_fp_sub(fp_int *a, fp_int *b, fp_int *c);
-void bn_reverse(unsigned char *s, int len);
+void fp_reverse(unsigned char *s, int len);
 
 void fp_mul_comba(fp_int *A, fp_int *B, fp_int *C);
 

src/mont/fp_montgomery_reduce.c

@@ -311,6 +311,7 @@ asm(                               \
     " MOVCC %0,#0            \n\t" \
 :"=r"(cy),"=m"(_c[0]):"0"(cy),"1"(_c[0]):"r0","%cc");
 
+/******************************************************************/
 #elif defined(TFM_PPC32)
 
 /* PPC32 */
@@ -341,6 +342,7 @@ asm(                                 \
    " addze    %0,%0         \n\t"    \
 :"=r"(cy),"=m"(_c[0]):"0"(cy),"1"(_c[0]):"16","%cc");
 
+/******************************************************************/
 #elif defined(TFM_PPC64)
 
 /* PPC64 */
@@ -352,27 +354,26 @@ asm(                                 \
 
 #define INNERMUL                     \
 asm(                                 \
-   " mulld    16,%3,%4       \n\t"   \
+   " mulld    r16,%3,%4       \n\t"   \
-   " mulhdu   17,%3,%4       \n\t"   \
+   " mulhdu   r17,%3,%4       \n\t"   \
-   " addc     16,16,%0       \n\t"   \
+   " addc     r16,16,%0       \n\t"   \
-   " addze    17,17          \n\t"   \
+   " addze    r17,r17          \n\t"   \
-   " ldx      18,0,%1        \n\t"   \
+   " ldx      r18,0,%1        \n\t"   \
-   " addc     16,16,18       \n\t"   \
+   " addc     r16,r16,r18       \n\t"   \
-   " addze    %0,17          \n\t"   \
+   " addze    %0,r17          \n\t"   \
-   " sdx      16,0,%1        \n\t"   \
+   " sdx      r16,0,%1        \n\t"   \
-:"=r"(cy),"=m"(_c[0]):"0"(cy),"r"(mu),"r"(tmpm[0]),"1"(_c[0]):"16", "17", "18","%cc"); ++tmpm;
+:"=r"(cy),"=m"(_c[0]):"0"(cy),"r"(mu),"r"(tmpm[0]),"1"(_c[0]):"r16", "r17", "r18","%cc"); ++tmpm;
 
 #define PROPCARRY                    \
 asm(                                 \
-   " ldx      16,0,%1       \n\t"    \
+   " ldx      r16,0,%1       \n\t"    \
-   " addc     16,16,%0      \n\t"    \
+   " addc     r16,r16,%0      \n\t"    \
-   " sdx      16,0,%1       \n\t"    \
+   " sdx      r16,0,%1       \n\t"    \
    " xor      %0,%0,%0      \n\t"    \
    " addze    %0,%0         \n\t"    \
-:"=r"(cy),"=m"(_c[0]):"0"(cy),"1"(_c[0]):"16","%cc");
+:"=r"(cy),"=m"(_c[0]):"0"(cy),"1"(_c[0]):"r16","%cc");
 
 /******************************************************************/
-
 #elif defined(TFM_AVR32)
 
 /* AVR32 */
@@ -402,6 +403,40 @@ asm(                                 \
    " acr      %0            \n\t"    \
 :"=r"(cy),"=r"(&_c[0]):"0"(cy),"1"(&_c[0]):"r2","%cc");
 
+/******************************************************************/
+#elif defined(TFM_MIPS)
+
+/* MIPS */
+#define MONT_START 
+#define MONT_FINI
+#define LOOP_END
+#define LOOP_START \
+   mu = c[x] * mp
+
+#define INNERMUL                     \
+asm(                                 \
+   " multu    %3,%4          \n\t"   \
+   " mflo     $12            \n\t"   \
+   " mfhi     $13            \n\t"   \
+   " addu     $12,$12,%0     \n\t"   \
+   " sltu     $10,$12,%0     \n\t"   \
+   " addu     $13,$13,$10    \n\t"   \
+   " lw       $10,%1         \n\t"   \
+   " addu     $12,$12,$10    \n\t"   \
+   " sltu     $10,$12,$10    \n\t"   \
+   " addu     %0,$13,$10     \n\t"   \
+   " sw       $12,%1         \n\t"   \
+:"=r"(cy),"=m"(_c[0]):"0"(cy),"r"(mu),"r"(tmpm[0]),"1"(_c[0]):"$10","$12","$13"); ++tmpm;
+
+#define PROPCARRY                    \
+asm(                                 \
+   " lw       $10,%1        \n\t"    \
+   " addu     $10,$10,%0    \n\t"    \
+   " sw       $10,%1        \n\t"    \
+   " sltu     %0,$10,%0     \n\t"    \
+:"=r"(cy),"=m"(_c[0]):"0"(cy),"1"(_c[0]):"$10");
+
+/******************************************************************/
 #else
 
 /* ISO C code */
@@ -428,7 +463,7 @@ asm(                                 \
 #define LO  0
 
 #ifdef TFM_SMALL_MONT_SET
-#include "fp_mont_small.c"
+#include "fp_mont_small.i"
 #endif
 
 /* computes x/R == x (mod N) via Montgomery Reduction */

src/mul/fp_mul_comba.c

@@ -205,15 +205,15 @@ asm(                              \
 
 #define COMBA_FINI 
    
-/* untested: will mulhwu change the flags?  Docs say no */
+/* untested: will mulhdu change the flags?  Docs say no */
 #define MULADD(i, j)              \
 asm(                              \
-   " mulld  16,%6,%7       \n\t" \
+   " mulld  r16,%6,%7       \n\t" \
    " addc   %0,%0,16       \n\t" \
-   " mulhdu 16,%6,%7       \n\t" \
+   " mulhdu r16,%6,%7       \n\t" \
    " adde   %1,%1,16       \n\t" \
    " addze  %2,%2          \n\t" \
-:"=r"(c0), "=r"(c1), "=r"(c2):"0"(c0), "1"(c1), "2"(c2), "r"(i), "r"(j):"16");
+:"=r"(c0), "=r"(c1), "=r"(c2):"0"(c0), "1"(c1), "2"(c2), "r"(i), "r"(j):"r16");
 
 #elif defined(TFM_AVR32)
 
@@ -243,6 +243,39 @@ asm(                             \
    " acr    %2              \n\t"\
 :"=r"(c0), "=r"(c1), "=r"(c2):"0"(c0), "1"(c1), "2"(c2), "r"(i), "r"(j):"r2","r3");
 
+#elif defined(TFM_MIPS)
+
+#define COMBA_START
+
+#define COMBA_CLEAR \
+   c0 = c1 = c2 = 0;
+
+#define COMBA_FORWARD \
+   do { c0 = c1; c1 = c2; c2 = 0; } while (0);
+
+#define COMBA_STORE(x) \
+   x = c0;
+
+#define COMBA_STORE2(x) \
+   x = c1;
+
+#define COMBA_FINI 
+   
+#define MULADD(i, j)              \
+asm(                              \
+   " multu  %6,%7          \n\t"  \
+   " mflo   $12            \n\t"  \
+   " mfhi   $13            \n\t"  \
+   " addu    %0,%0,$12     \n\t"  \
+   " sltu   $12,%0,$12     \n\t"  \
+   " addu    %1,%1,$13     \n\t"  \
+   " sltu   $13,%1,$13     \n\t"  \
+   " addu    %1,%1,$12     \n\t"  \
+   " sltu   $12,%1,$12     \n\t"  \
+   " addu    %2,%2,$13     \n\t"  \
+   " addu    %2,%2,$12     \n\t"  \
+:"=r"(c0), "=r"(c1), "=r"(c2):"0"(c0), "1"(c1), "2"(c2), "r"(i), "r"(j):"$12","$13");
+
 #else
 /* ISO C code */
 
@@ -270,6 +303,7 @@ asm(                             \
 
 #endif
 
+#ifndef TFM_DEFINES
 
 /* generic PxQ multiplier */
 void fp_mul_comba(fp_int *A, fp_int *B, fp_int *C)
@@ -326,21 +360,7 @@ void fp_mul_comba(fp_int *A, fp_int *B, fp_int *C)
   fp_copy(dst, C);
 }
 
-#include "fp_mul_comba_small_set.i"
+#endif
-#include "fp_mul_comba_3.i"
-#include "fp_mul_comba_4.i"
-#include "fp_mul_comba_6.i"
-#include "fp_mul_comba_7.i"
-#include "fp_mul_comba_8.i"
-#include "fp_mul_comba_9.i"
-#include "fp_mul_comba_12.i"
-#include "fp_mul_comba_17.i"
-#include "fp_mul_comba_20.i"
-#include "fp_mul_comba_24.i"
-#include "fp_mul_comba_28.i"
-#include "fp_mul_comba_32.i"
-#include "fp_mul_comba_48.i"
-#include "fp_mul_comba_64.i"
 
 /* $Source$ */
 /* $Revision$ */

src/mul/fp_mul_comba_12.c

@@ -1,3 +1,6 @@
+#define TFM_DEFINES
+#include "fp_mul_comba.c"
+
 #ifdef TFM_MUL12
 void fp_mul_comba12(fp_int *A, fp_int *B, fp_int *C)
 {

src/mul/fp_mul_comba_17.c

@@ -1,3 +1,6 @@
+#define TFM_DEFINES
+#include "fp_mul_comba.c"
+
 #ifdef TFM_MUL17
 void fp_mul_comba17(fp_int *A, fp_int *B, fp_int *C)
 {

src/mul/fp_mul_comba_20.c

@@ -1,3 +1,6 @@
+#define TFM_DEFINES
+#include "fp_mul_comba.c"
+
 #ifdef TFM_MUL20
 void fp_mul_comba20(fp_int *A, fp_int *B, fp_int *C)
 {

src/mul/fp_mul_comba_24.c

@@ -1,3 +1,6 @@
+#define TFM_DEFINES
+#include "fp_mul_comba.c"
+
 #ifdef TFM_MUL24
 void fp_mul_comba24(fp_int *A, fp_int *B, fp_int *C)
 {

src/mul/fp_mul_comba_28.c

@@ -1,3 +1,6 @@
+#define TFM_DEFINES
+#include "fp_mul_comba.c"
+
 #ifdef TFM_MUL28
 void fp_mul_comba28(fp_int *A, fp_int *B, fp_int *C)
 {

src/mul/fp_mul_comba_3.c

@@ -1,3 +1,6 @@
+#define TFM_DEFINES
+#include "fp_mul_comba.c"
+
 #ifdef TFM_MUL3
 void fp_mul_comba3(fp_int *A, fp_int *B, fp_int *C)
 {

src/mul/fp_mul_comba_32.c

@@ -1,3 +1,6 @@
+#define TFM_DEFINES
+#include "fp_mul_comba.c"
+
 #ifdef TFM_MUL32
 void fp_mul_comba32(fp_int *A, fp_int *B, fp_int *C)
 {

src/mul/fp_mul_comba_4.c

@@ -1,3 +1,6 @@
+#define TFM_DEFINES
+#include "fp_mul_comba.c"
+
 #ifdef TFM_MUL4
 void fp_mul_comba4(fp_int *A, fp_int *B, fp_int *C)
 {

src/mul/fp_mul_comba_48.c

@@ -1,3 +1,6 @@
+#define TFM_DEFINES
+#include "fp_mul_comba.c"
+
 #ifdef TFM_MUL48
 void fp_mul_comba48(fp_int *A, fp_int *B, fp_int *C)
 {

src/mul/fp_mul_comba_6.c

@@ -1,3 +1,6 @@
+#define TFM_DEFINES
+#include "fp_mul_comba.c"
+
 #ifdef TFM_MUL6
 void fp_mul_comba6(fp_int *A, fp_int *B, fp_int *C)
 {

src/mul/fp_mul_comba_64.c

@@ -1,3 +1,6 @@
+#define TFM_DEFINES
+#include "fp_mul_comba.c"
+
 #ifdef TFM_MUL64
 void fp_mul_comba64(fp_int *A, fp_int *B, fp_int *C)
 {

src/mul/fp_mul_comba_7.c

@@ -1,3 +1,6 @@
+#define TFM_DEFINES
+#include "fp_mul_comba.c"
+
 #ifdef TFM_MUL7
 void fp_mul_comba7(fp_int *A, fp_int *B, fp_int *C)
 {

src/mul/fp_mul_comba_8.c

@@ -1,3 +1,6 @@
+#define TFM_DEFINES
+#include "fp_mul_comba.c"
+
 #ifdef TFM_MUL8
 void fp_mul_comba8(fp_int *A, fp_int *B, fp_int *C)
 {

src/mul/fp_mul_comba_9.c

@@ -1,3 +1,6 @@
+#define TFM_DEFINES
+#include "fp_mul_comba.c"
+
 #ifdef TFM_MUL9
 void fp_mul_comba9(fp_int *A, fp_int *B, fp_int *C)
 {

src/mul/fp_mul_comba_small_set.c

@@ -1,3 +1,6 @@
+#define TFM_DEFINES
+#include "fp_mul_comba.c"
+
 #if defined(TFM_SMALL_SET)
 void fp_mul_comba_small(fp_int *A, fp_int *B, fp_int *C)
 {

src/numtheory/fp_isprime.c

@@ -63,6 +63,7 @@ int fp_isprime(fp_int *a)
    }
 
    /* now do 8 miller rabins */
+   fp_init(&b);
    for (r = 0; r < 8; r++) {
        fp_set(&b, primes[r]);
        fp_prime_miller_rabin(a, &b, &res);

src/sqr/fp_sqr_comba.c

@@ -14,7 +14,6 @@
    #define TFM_X86
 #endif
 
-
 #if defined(TFM_X86)
 
 /* x86-32 optimized */
@@ -390,25 +389,25 @@ asm(                              \
 /* multiplies point i and j, updates carry "c1" and digit c2 */
 #define SQRADD(i, j)             \
 asm(                             \
-   " mulld  16,%6,%6       \n\t" \
+   " mulld  r16,%6,%6       \n\t" \
-   " addc   %0,%0,16       \n\t" \
+   " addc   %0,%0,r16       \n\t" \
-   " mulhdu 16,%6,%6       \n\t" \
+   " mulhdu r16,%6,%6       \n\t" \
-   " adde   %1,%1,16       \n\t" \
+   " adde   %1,%1,r16       \n\t" \
    " addze  %2,%2          \n\t" \
-:"=r"(c0), "=r"(c1), "=r"(c2):"0"(c0), "1"(c1), "2"(c2), "r"(i):"16","%cc");
+:"=r"(c0), "=r"(c1), "=r"(c2):"0"(c0), "1"(c1), "2"(c2), "r"(i):"r16","%cc");
 
 /* for squaring some of the terms are doubled... */
 #define SQRADD2(i, j)            \
 asm(                             \
-   " mulld  16,%6,%7       \n\t" \
+   " mulld  r16,%6,%7       \n\t" \
-   " mulhdu 17,%6,%7       \n\t" \
+   " mulhdu r17,%6,%7       \n\t" \
-   " addc   %0,%0,16       \n\t" \
+   " addc   %0,%0,r16       \n\t" \
-   " adde   %1,%1,17       \n\t" \
+   " adde   %1,%1,r17       \n\t" \
    " addze  %2,%2          \n\t" \
-   " addc   %0,%0,16       \n\t" \
+   " addc   %0,%0,r16       \n\t" \
-   " adde   %1,%1,17       \n\t" \
+   " adde   %1,%1,r17       \n\t" \
    " addze  %2,%2          \n\t" \
-:"=r"(c0), "=r"(c1), "=r"(c2):"0"(c0), "1"(c1), "2"(c2), "r"(i), "r"(j):"16", "17","%cc");
+:"=r"(c0), "=r"(c1), "=r"(c2):"0"(c0), "1"(c1), "2"(c2), "r"(i), "r"(j):"r16", "r17","%cc");
 
 #define SQRADDSC(i, j)            \
 asm(                              \
@@ -419,12 +418,12 @@ asm(                              \
 
 #define SQRADDAC(i, j)           \
 asm(                             \
-   " mulld  16,%6,%7       \n\t" \
+   " mulld  r16,%6,%7       \n\t" \
-   " addc   %0,%0,16       \n\t" \
+   " addc   %0,%0,r16       \n\t" \
-   " mulhdu 16,%6,%7       \n\t" \
+   " mulhdu r16,%6,%7       \n\t" \
-   " adde   %1,%1,16       \n\t" \
+   " adde   %1,%1,r16       \n\t" \
    " addze  %2,%2          \n\t" \
-:"=r"(sc0), "=r"(sc1), "=r"(sc2):"0"(sc0), "1"(sc1), "2"(sc2), "r"(i), "r"(j):"16", "%cc");
+:"=r"(sc0), "=r"(sc1), "=r"(sc2):"0"(sc0), "1"(sc1), "2"(sc2), "r"(i), "r"(j):"r16", "%cc");
 
 #define SQRADDDB                  \
 asm(                              \
@@ -504,6 +503,113 @@ asm(                              \
    " adc    %2,%2,%5        \n\t" \
 :"=r"(c0), "=r"(c1), "=r"(c2) : "r"(sc0), "r"(sc1), "r"(sc2), "0"(c0), "1"(c1), "2"(c2) : "%cc");
 
+#elif defined(TFM_MIPS)
+
+/* MIPS */
+
+#define COMBA_START
+
+#define CLEAR_CARRY \
+   c0 = c1 = c2 = 0;
+
+#define COMBA_STORE(x) \
+   x = c0;
+
+#define COMBA_STORE2(x) \
+   x = c1;
+
+#define CARRY_FORWARD \
+   do { c0 = c1; c1 = c2; c2 = 0; } while (0);
+
+#define COMBA_FINI
+
+/* multiplies point i and j, updates carry "c1" and digit c2 */
+#define SQRADD(i, j)              \
+asm(                              \
+   " multu  %6,%6          \n\t"  \
+   " mflo   $12            \n\t"  \
+   " mfhi   $13            \n\t"  \
+   " addu    %0,%0,$12     \n\t"  \
+   " sltu   $12,%0,$12     \n\t"  \
+   " addu    %1,%1,$13     \n\t"  \
+   " sltu   $13,%1,$13     \n\t"  \
+   " addu    %1,%1,$12     \n\t"  \
+   " sltu   $12,%1,$12     \n\t"  \
+   " addu    %2,%2,$13     \n\t"  \
+   " addu    %2,%2,$12     \n\t"  \
+:"=r"(c0), "=r"(c1), "=r"(c2):"0"(c0), "1"(c1), "2"(c2), "r"(i):"$12","$13");
+
+/* for squaring some of the terms are doubled... */
+#define SQRADD2(i, j)            \
+asm(                             \
+   " multu  %6,%7          \n\t" \
+   " mflo   $12            \n\t" \
+   " mfhi   $13            \n\t" \
+                                 \
+   " addu    %0,%0,$12     \n\t" \
+   " sltu   $14,%0,$12     \n\t" \
+   " addu    %1,%1,$13     \n\t" \
+   " sltu   $15,%1,$13     \n\t" \
+   " addu    %1,%1,$14     \n\t" \
+   " sltu   $14,%1,$14     \n\t" \
+   " addu    %2,%2,$15     \n\t" \
+   " addu    %2,%2,$14     \n\t" \
+                                 \
+   " addu    %0,%0,$12     \n\t" \
+   " sltu   $14,%0,$12     \n\t" \
+   " addu    %1,%1,$13     \n\t" \
+   " sltu   $15,%1,$13     \n\t" \
+   " addu    %1,%1,$14     \n\t" \
+   " sltu   $14,%1,$14     \n\t" \
+   " addu    %2,%2,$15     \n\t" \
+   " addu    %2,%2,$14     \n\t" \
+:"=r"(c0), "=r"(c1), "=r"(c2):"0"(c0), "1"(c1), "2"(c2), "r"(i), "r"(j):"$12", "$13", "$14", "$15");
+
+#define SQRADDSC(i, j)            \
+asm(                              \
+   " multu  %6,%7          \n\t"  \
+   " mflo   %0             \n\t"  \
+   " mfhi   %1             \n\t"  \
+   " xor    %2,%2,%2       \n\t"  \
+:"=r"(sc0), "=r"(sc1), "=r"(sc2):"0"(sc0), "1"(sc1), "2"(sc2), "r"(i),"r"(j) : "%cc");
+
+#define SQRADDAC(i, j)           \
+asm(                             \
+   " multu  %6,%7          \n\t" \
+   " mflo   $12            \n\t" \
+   " mfhi   $13            \n\t" \
+   " addu    %0,%0,$12     \n\t" \
+   " sltu   $12,%0,$12     \n\t" \
+   " addu    %1,%1,$13     \n\t" \
+   " sltu   $13,%1,$13     \n\t" \
+   " addu    %1,%1,$12     \n\t" \
+   " sltu   $12,%1,$12     \n\t" \
+   " addu    %2,%2,$13     \n\t" \
+   " addu    %2,%2,$12     \n\t" \
+:"=r"(sc0), "=r"(sc1), "=r"(sc2):"0"(sc0), "1"(sc1), "2"(sc2), "r"(i), "r"(j):"$12", "$13", "$14");
+
+#define SQRADDDB                  \
+asm(                              \
+   " addu    %0,%0,%3       \n\t" \
+   " sltu   $10,%0,%3       \n\t" \
+   " addu    %1,%1,$10      \n\t" \
+   " sltu   $10,%1,$10      \n\t" \
+   " addu    %1,%1,%4       \n\t" \
+   " sltu   $11,%1,%4       \n\t" \
+   " addu    %2,%2,$10      \n\t" \
+   " addu    %2,%2,$11      \n\t" \
+   " addu    %2,%2,%5       \n\t" \
+                                  \
+   " addu    %0,%0,%3       \n\t" \
+   " sltu   $10,%0,%3       \n\t" \
+   " addu    %1,%1,$10      \n\t" \
+   " sltu   $10,%1,$10      \n\t" \
+   " addu    %1,%1,%4       \n\t" \
+   " sltu   $11,%1,%4       \n\t" \
+   " addu    %2,%2,$10      \n\t" \
+   " addu    %2,%2,$11      \n\t" \
+   " addu    %2,%2,%5       \n\t" \
+:"=r"(c0), "=r"(c1), "=r"(c2) : "r"(sc0), "r"(sc1), "r"(sc2), "0"(c0), "1"(c1), "2"(c2) : "$10", "$11");
 
 #else
 
@@ -566,19 +672,6 @@ asm(                              \
 
 #endif
 
-#include "fp_sqr_comba_generic.c"
+/* $Source$ */
-#include "fp_sqr_comba_small_set.i"
+/* $Revision$ */
-#include "fp_sqr_comba_3.i"
+/* $Date$ */
-#include "fp_sqr_comba_4.i"
-#include "fp_sqr_comba_6.i"
-#include "fp_sqr_comba_7.i"
-#include "fp_sqr_comba_8.i"
-#include "fp_sqr_comba_9.i"
-#include "fp_sqr_comba_12.i"
-#include "fp_sqr_comba_17.i"
-#include "fp_sqr_comba_20.i"
-#include "fp_sqr_comba_24.i"
-#include "fp_sqr_comba_28.i"
-#include "fp_sqr_comba_32.i"
-#include "fp_sqr_comba_48.i"
-#include "fp_sqr_comba_64.i"

src/sqr/fp_sqr_comba_12.c

@@ -1,7 +1,13 @@
+#define TFM_DEFINES
+#include "fp_sqr_comba.c"
+
 #ifdef TFM_SQR12
 void fp_sqr_comba12(fp_int *A, fp_int *B)
 {
    fp_digit *a, b[24], c0, c1, c2, sc0, sc1, sc2;
+#ifdef TFM_ISO
+   fp_word tt;
+#endif
 
    a = A->dp;
    COMBA_START; 
@@ -133,3 +139,6 @@ void fp_sqr_comba12(fp_int *A, fp_int *B)
 #endif
 
 
+/* $Source$ */
+/* $Revision$ */
+/* $Date$ */

src/sqr/fp_sqr_comba_17.c

@@ -1,7 +1,13 @@
+#define TFM_DEFINES
+#include "fp_sqr_comba.c"
+
 #ifdef TFM_SQR17
 void fp_sqr_comba17(fp_int *A, fp_int *B)
 {
    fp_digit *a, b[34], c0, c1, c2, sc0, sc1, sc2;
+#ifdef TFM_ISO
+   fp_word tt;
+#endif
 
    a = A->dp;
    COMBA_START; 
@@ -183,3 +189,6 @@ void fp_sqr_comba17(fp_int *A, fp_int *B)
 #endif
 
 
+/* $Source$ */
+/* $Revision$ */
+/* $Date$ */

src/sqr/fp_sqr_comba_20.c

@@ -1,7 +1,13 @@
+#define TFM_DEFINES
+#include "fp_sqr_comba.c"
+
 #ifdef TFM_SQR20
 void fp_sqr_comba20(fp_int *A, fp_int *B)
 {
    fp_digit *a, b[40], c0, c1, c2, sc0, sc1, sc2;
+#ifdef TFM_ISO
+   fp_word tt;
+#endif
 
    a = A->dp;
    COMBA_START; 
@@ -213,3 +219,6 @@ void fp_sqr_comba20(fp_int *A, fp_int *B)
 #endif
 
 
+/* $Source$ */
+/* $Revision$ */
+/* $Date$ */

src/sqr/fp_sqr_comba_24.c

@@ -1,7 +1,13 @@
+#define TFM_DEFINES
+#include "fp_sqr_comba.c"
+
 #ifdef TFM_SQR24
 void fp_sqr_comba24(fp_int *A, fp_int *B)
 {
    fp_digit *a, b[48], c0, c1, c2, sc0, sc1, sc2;
+#ifdef TFM_ISO
+   fp_word tt;
+#endif
 
    a = A->dp;
    COMBA_START; 
@@ -253,3 +259,6 @@ void fp_sqr_comba24(fp_int *A, fp_int *B)
 #endif
 
 
+/* $Source$ */
+/* $Revision$ */
+/* $Date$ */

src/sqr/fp_sqr_comba_28.c

@@ -1,7 +1,13 @@
+#define TFM_DEFINES
+#include "fp_sqr_comba.c"
+
 #ifdef TFM_SQR28
 void fp_sqr_comba28(fp_int *A, fp_int *B)
 {
    fp_digit *a, b[56], c0, c1, c2, sc0, sc1, sc2;
+#ifdef TFM_ISO
+   fp_word tt;
+#endif
 
    a = A->dp;
    COMBA_START; 
@@ -293,3 +299,6 @@ void fp_sqr_comba28(fp_int *A, fp_int *B)
 #endif
 
 
+/* $Source$ */
+/* $Revision$ */
+/* $Date$ */

src/sqr/fp_sqr_comba_3.c

@@ -1,7 +1,13 @@
+#define TFM_DEFINES
+#include "fp_sqr_comba.c"
+
 #ifdef TFM_SQR3
 void fp_sqr_comba3(fp_int *A, fp_int *B)
 {
    fp_digit *a, b[6], c0, c1, c2, sc0, sc1, sc2;
+#ifdef TFM_ISO
+   fp_word tt;
+#endif
 
    a = A->dp;
    COMBA_START; 
@@ -43,3 +49,6 @@ void fp_sqr_comba3(fp_int *A, fp_int *B)
 #endif
 
 
+/* $Source$ */
+/* $Revision$ */
+/* $Date$ */

src/sqr/fp_sqr_comba_32.c

@@ -1,7 +1,13 @@
+#define TFM_DEFINES
+#include "fp_sqr_comba.c"
+
 #ifdef TFM_SQR32
 void fp_sqr_comba32(fp_int *A, fp_int *B)
 {
    fp_digit *a, b[64], c0, c1, c2, sc0, sc1, sc2;
+#ifdef TFM_ISO
+   fp_word tt;
+#endif
 
    a = A->dp;
    COMBA_START; 
@@ -333,3 +339,6 @@ void fp_sqr_comba32(fp_int *A, fp_int *B)
 #endif
 
 
+/* $Source$ */
+/* $Revision$ */
+/* $Date$ */

src/sqr/fp_sqr_comba_4.c

@@ -1,7 +1,13 @@
+#define TFM_DEFINES
+#include "fp_sqr_comba.c"
+
 #ifdef TFM_SQR4
 void fp_sqr_comba4(fp_int *A, fp_int *B)
 {
    fp_digit *a, b[8], c0, c1, c2, sc0, sc1, sc2;
+#ifdef TFM_ISO
+   fp_word tt;
+#endif
 
    a = A->dp;
    COMBA_START; 
@@ -53,3 +59,6 @@ void fp_sqr_comba4(fp_int *A, fp_int *B)
 #endif
 
 
+/* $Source$ */
+/* $Revision$ */
+/* $Date$ */

src/sqr/fp_sqr_comba_48.c

@@ -1,7 +1,13 @@
+#define TFM_DEFINES
+#include "fp_sqr_comba.c"
+
 #ifdef TFM_SQR48
 void fp_sqr_comba48(fp_int *A, fp_int *B)
 {
    fp_digit *a, b[96], c0, c1, c2, sc0, sc1, sc2;
+#ifdef TFM_ISO
+   fp_word tt;
+#endif
 
    a = A->dp;
    COMBA_START; 
@@ -493,3 +499,6 @@ void fp_sqr_comba48(fp_int *A, fp_int *B)
 #endif
 
 
+/* $Source$ */
+/* $Revision$ */
+/* $Date$ */

src/sqr/fp_sqr_comba_6.c

@@ -1,7 +1,13 @@
+#define TFM_DEFINES
+#include "fp_sqr_comba.c"
+
 #ifdef TFM_SQR6
 void fp_sqr_comba6(fp_int *A, fp_int *B)
 {
    fp_digit *a, b[12], c0, c1, c2, sc0, sc1, sc2;
+#ifdef TFM_ISO
+   fp_word tt;
+#endif
 
    a = A->dp;
    COMBA_START; 
@@ -73,3 +79,6 @@ void fp_sqr_comba6(fp_int *A, fp_int *B)
 #endif
 
 
+/* $Source$ */
+/* $Revision$ */
+/* $Date$ */

src/sqr/fp_sqr_comba_64.c

@@ -1,7 +1,13 @@
+#define TFM_DEFINES
+#include "fp_sqr_comba.c"
+
 #ifdef TFM_SQR64
 void fp_sqr_comba64(fp_int *A, fp_int *B)
 {
    fp_digit *a, b[128], c0, c1, c2, sc0, sc1, sc2;
+#ifdef TFM_ISO
+   fp_word tt;
+#endif
 
    a = A->dp;
    COMBA_START; 
@@ -653,3 +659,6 @@ void fp_sqr_comba64(fp_int *A, fp_int *B)
 #endif
 
 
+/* $Source$ */
+/* $Revision$ */
+/* $Date$ */

src/sqr/fp_sqr_comba_7.c

@@ -1,7 +1,13 @@
+#define TFM_DEFINES
+#include "fp_sqr_comba.c"
+
 #ifdef TFM_SQR7
 void fp_sqr_comba7(fp_int *A, fp_int *B)
 {
    fp_digit *a, b[14], c0, c1, c2, sc0, sc1, sc2;
+#ifdef TFM_ISO
+   fp_word tt;
+#endif
 
    a = A->dp;
    COMBA_START; 
@@ -83,3 +89,6 @@ void fp_sqr_comba7(fp_int *A, fp_int *B)
 #endif
 
 
+/* $Source$ */
+/* $Revision$ */
+/* $Date$ */