]> git.cryptolib.org Git - avr-crypto-lib.git/blobdiff - bigint/bigint.c
fixing bigint
[avr-crypto-lib.git] / bigint / bigint.c
index 6db0389d693490bdc818b4688e88b29370c6e4f7..d2d1fc76b75ff82c36e45c9606b15e65822cde04 100644 (file)
 
 #define DEBUG 0
 
-#if DEBUG
+#if DEBUG || 1
 #include "cli.h"
 #include "bigint_io.h"
+#include <stdio.h>
 #endif
 
 #ifndef MAX
 #define SET_NEG(a)   (a)->info |= BIGINT_NEG_MASK
 #define SET_POS(a)   (a)->info &= ~BIGINT_NEG_MASK
 #define XCHG(a,b)    do{(a)^=(b); (b)^=(a); (a)^=(b);}while(0)
-#define XCHG_PTR(a,b)    do{ a = (void*)(((bigint_ptr_int_t)(a)) ^ ((bigint_ptr_int_t)(b))); \
-                                b = (void*)(((bigint_ptr_int_t)(a)) ^ ((bigint_ptr_int_t)(b))); \
-                                a = (void*)(((bigint_ptr_int_t)(a)) ^ ((bigint_ptr_int_t)(b)));}while(0)
+#define XCHG_PTR(a,b)    do{ a = (void*)(((intptr_t)(a)) ^ ((intptr_t)(b))); \
+                                b = (void*)(((intptr_t)(a)) ^ ((intptr_t)(b))); \
+                                a = (void*)(((intptr_t)(a)) ^ ((intptr_t)(b)));}while(0)
 
 #define GET_SIGN(a) ((a)->info&BIGINT_NEG_MASK)
 
 /******************************************************************************/
 void bigint_adjust(bigint_t* a){
-       while(a->length_B!=0 && a->wordv[a->length_B-1]==0){
-               a->length_B--;
+       while(a->length_W!=0 && a->wordv[a->length_W-1]==0){
+               a->length_W--;
        }
-       if(a->length_B==0){
+       if(a->length_W==0){
                a->info=0;
                return;
        }
        bigint_word_t t;
        uint8_t i = BIGINT_WORD_SIZE-1;
-       t = a->wordv[a->length_B-1];
+       t = a->wordv[a->length_W-1];
        while((t&(1L<<(BIGINT_WORD_SIZE-1)))==0 && i){
                t<<=1;
                i--;
@@ -80,39 +81,39 @@ void bigint_adjust(bigint_t* a){
 
 /******************************************************************************/
 
-uint16_t bigint_length_b(bigint_t* a){
-       if(!a->length_B || a->length_B==0){
+uint16_t bigint_length_b(const bigint_t* a){
+       if(!a->length_W || a->length_W==0){
                return 0;
        }
-       return (a->length_B-1) * BIGINT_WORD_SIZE + GET_FBS(a);
+       return (a->length_W-1) * BIGINT_WORD_SIZE + GET_FBS(a);
 }
 
 /******************************************************************************/
 
-uint16_t bigint_length_B(bigint_t* a){
-       return a->length_B * sizeof(bigint_word_t);
+uint16_t bigint_length_B(const bigint_t* a){
+       return a->length_W * sizeof(bigint_word_t);
 }
 
 /******************************************************************************/
 
-uint32_t bigint_get_first_set_bit(bigint_t* a){
-       if(a->length_B==0){
+uint32_t bigint_get_first_set_bit(const bigint_t* a){
+       if(a->length_W==0){
                return (uint32_t)(-1);
        }
-       return (a->length_B-1)*sizeof(bigint_word_t)*8+GET_FBS(a);
+       return (a->length_W-1)*sizeof(bigint_word_t)*8+GET_FBS(a);
 }
 
 
 /******************************************************************************/
 
-uint32_t bigint_get_last_set_bit(bigint_t* a){
+uint32_t bigint_get_last_set_bit(const bigint_t* a){
        uint32_t r=0;
        uint8_t b=0;
        bigint_word_t x=1;
-       if(a->length_B==0){
+       if(a->length_W==0){
                return (uint32_t)(-1);
        }
-       while(a->wordv[r]==0 && r<a->length_B){
+       while(a->wordv[r]==0 && r<a->length_W){
                ++r;
        }
        if(a->wordv[r] == 0){
@@ -128,8 +129,10 @@ uint32_t bigint_get_last_set_bit(bigint_t* a){
 /******************************************************************************/
 
 void bigint_copy(bigint_t* dest, const bigint_t* src){
-       memcpy(dest->wordv, src->wordv, src->length_B*sizeof(bigint_word_t));
-       dest->length_B = src->length_B;
+    if(dest->wordv != src->wordv){
+           memcpy(dest->wordv, src->wordv, src->length_W * sizeof(bigint_word_t));
+    }
+    dest->length_W = src->length_W;
        dest->info = src->info;
 }
 
@@ -138,24 +141,26 @@ void bigint_copy(bigint_t* dest, const bigint_t* src){
 /* this should be implemented in assembly */
 void bigint_add_u(bigint_t* dest, const bigint_t* a, const bigint_t* b){
        uint16_t i;
-       bigint_wordplus_t t=0LL;
-       if(a->length_B < b->length_B){
+       bigint_wordplus_t t = 0LL;
+       if(a->length_W < b->length_W){
                XCHG_PTR(a,b);
        }
-       for(i=0; i<b->length_B; ++i){
+       for(i = 0; i < b->length_W; ++i){
 //             t = (bigint_wordplus_t)(a->wordv[i]) + (bigint_wordplus_t)(b->wordv[i]) + t;
                t += a->wordv[i];
                t += b->wordv[i];
                dest->wordv[i] = (bigint_word_t)t;
-               t>>=BIGINT_WORD_SIZE;
+               t >>= BIGINT_WORD_SIZE;
        }
-       for(; i<a->length_B; ++i){
+       for(; i<a->length_W; ++i){
                t += a->wordv[i];
                dest->wordv[i] = (bigint_word_t)t;
-               t>>=BIGINT_WORD_SIZE;
+               t >>= BIGINT_WORD_SIZE;
+       }
+       if(t){
+               dest->wordv[i++] = (bigint_word_t)t;
        }
-       dest->wordv[i++] = (bigint_word_t)t;
-       dest->length_B = i;
+       dest->length_W = i;
        bigint_adjust(dest);
 }
 
@@ -163,54 +168,51 @@ void bigint_add_u(bigint_t* dest, const bigint_t* a, const bigint_t* b){
 
 /* this should be implemented in assembly */
 void bigint_add_scale_u(bigint_t* dest, const bigint_t* a, uint16_t scale){
+       if(a->length_W == 0){
+               return;
+       }
+       if(scale == 0){
+               bigint_add_u(dest, dest, a);
+               return;
+       }
        bigint_t x;
 #if BIGINT_WORD_SIZE == 8
-       memset(dest->wordv + dest->length_B, 0, MAX(dest->length_B, a->length_B + scale) - dest->length_B);
+       memset(dest->wordv + dest->length_W, 0, MAX(dest->length_W, a->length_W + scale) - dest->length_W);
        x.wordv = dest->wordv + scale;
-       x.length_B = dest->length_B - scale;
-       if((int16_t)x.length_B < 0)
-               x.length_B = 0;
-       x.info = dest->info;
+       x.length_W = dest->length_W - scale;
+       if((int16_t)x.length_W < 0){
+               x.length_W = 0;
+               x.info = 0;
+       } else {
+               x.info = dest->info;
+       }
        bigint_add_u(&x, &x, a);
-       dest->length_B = x.length_B + scale;
+       dest->length_W = x.length_W + scale;
        dest->info = 0;
        bigint_adjust(dest);
 #else
-#error unimplemented!
-#endif
-
-
-/*     uint16_t i,j=0;
-       uint16_t scale_w;
-       bigint_word_t *dst;
-       bigint_wordplus_t t=0;
-       scale_w = (scale+sizeof(bigint_word_t)-1)/sizeof(bigint_word_t);
-       if(scale>dest->length_B*sizeof(bigint_word_t)){
-               memset(((uint8_t*)dest->wordv)+dest->length_B*sizeof(bigint_word_t), 0, scale-dest->length_B*sizeof(bigint_word_t));
-       }
-       // a->wordv = (const uint32_t*)(((uint8_t*)a->wordv)+(scale&3));
-       dst  = dest->wordv + (scale&(sizeof(bigint_word_t)-1));
-       for(i=scale/sizeof(bigint_word_t); i<a->length_B+scale_w; ++i,++j){
-               t += a->wordv[j];
-               if(dest->length_B>i){
-                       t += dst[i];
-               }
-               dst[i] = (bigint_word_t)t;
-               t>>=BIGINT_WORD_SIZE;
-       }
-       while(t){
-               if(dest->length_B>i){
-                       t += dst[i];
-               }
-               dst[i] = (bigint_word_t)t;
-               t>>=BIGINT_WORD_SIZE;
-               ++i;
-       }
-       if(dest->length_B < i){
-               dest->length_B = i;
+       bigint_t s;
+       uint16_t word_shift = scale / sizeof(bigint_word_t), byte_shift = scale % sizeof(bigint_word_t);
+       bigint_word_t bv[a->length_W + 1];
+       s.wordv = bv;
+       bv[0] = bv[a->length_W] = 0;
+       memcpy((uint8_t*)bv + byte_shift, a->wordv, a->length_W * sizeof(bigint_word_t));
+       s.length_W = a->length_W + 1;
+       bigint_adjust(&s);
+       memset(dest->wordv + dest->length_W, 0, (MAX(dest->length_W, s.length_W + word_shift) - dest->length_W) * sizeof(bigint_word_t));
+       x.wordv = dest->wordv + word_shift;
+       x.length_W = dest->length_W - word_shift;
+       if((int16_t)x.length_W < 0){
+               x.length_W = 0;
+               x.info = 0;
+       }else{
+               x.info = dest->info;
        }
+       bigint_add_u(&x, &x, &s);
+       dest->length_W = x.length_W + word_shift;
+       dest->info = 0;
        bigint_adjust(dest);
-       */
+#endif
 }
 
 /******************************************************************************/
@@ -220,32 +222,30 @@ void bigint_sub_u(bigint_t* dest, const bigint_t* a, const bigint_t* b){
        int8_t borrow=0;
        int8_t  r;
        bigint_wordplus_signed_t t=0LL;
-       uint16_t i, min, max;
-       min = MIN(a->length_B, b->length_B);
-       max = MAX(a->length_B, b->length_B);
-       r = bigint_cmp_u(a,b);
-       if(r==0){
-               bigint_set_zero(dest);
-               return;
-       }
-       if(b->length_B==0){
+       uint16_t i;
+       if(b->length_W == 0){
                bigint_copy(dest, a);
                SET_POS(dest);
                return;
        }
-       if(a->length_B==0){
+       if(a->length_W == 0){
                bigint_copy(dest, b);
                SET_NEG(dest);
                return;
        }
-       if(r<0){
+    r = bigint_cmp_u(a,b);
+    if(r == 0){
+        bigint_set_zero(dest);
+        return;
+    }
+       if(r < 0){
                bigint_sub_u(dest, b, a);
                SET_NEG(dest);
                return;
        }
-       for(i=0; i<max; ++i){
+       for(i = 0; i < a->length_W; ++i){
                t = a->wordv[i];
-               if(i<min){
+               if(i < b->length_W){
                        t -= b->wordv[i];
                }
                t -= borrow;
@@ -257,24 +257,24 @@ void bigint_sub_u(bigint_t* dest, const bigint_t* a, const bigint_t* b){
                }
        }
        SET_POS(dest);
-       dest->length_B = i;
+       dest->length_W = i;
        bigint_adjust(dest);
 }
 
 /******************************************************************************/
 
 int8_t bigint_cmp_u(const bigint_t* a, const bigint_t* b){
-       if(a->length_B > b->length_B){
+       if(a->length_W > b->length_W){
                return 1;
        }
-       if(a->length_B < b->length_B){
+       if(a->length_W < b->length_W){
                return -1;
        }
-       if(a->length_B==0){
+       if(a->length_W==0){
                return 0;
        }
        uint16_t i;
-       i = a->length_B-1;
+       i = a->length_W-1;
        do{
                if(a->wordv[i] != b->wordv[i]){
                        if(a->wordv[i] > b->wordv[i]){
@@ -344,7 +344,7 @@ void bigint_sub_s(bigint_t* dest, const bigint_t* a, const bigint_t* b){
 
 int8_t bigint_cmp_s(const bigint_t* a, const bigint_t* b){
        uint8_t s;
-       if(a->length_B==0 && b->length_B==0){
+       if(a->length_W==0 && b->length_W==0){
                return 0;
        }
        s  = GET_SIGN(a)?2:0;
@@ -371,35 +371,38 @@ int8_t bigint_cmp_s(const bigint_t* a, const bigint_t* b){
 /******************************************************************************/
 
 void bigint_shiftleft(bigint_t* a, uint16_t shift){
-       uint16_t byteshift, word_alloc;
+       uint16_t byteshift, words_to_shift;
        int16_t i;
        uint8_t bitshift;
        bigint_word_t *p;
-       bigint_wordplus_t t=0;
-       if(shift==0){
+       bigint_wordplus_t t = 0;
+       if(shift == 0){
                return;
        }
-       byteshift = shift/8;
-       bitshift = shift&7;
-       for(i=0;i<=byteshift/sizeof(bigint_word_t); ++i){
-               a->wordv[a->length_B+i] = 0;
-       }
+       byteshift = shift / 8;
+       bitshift = shift & 7;
+
        if(byteshift){
-               memmove(((uint8_t*)a->wordv)+byteshift, a->wordv, a->length_B*sizeof(bigint_word_t));
+               memmove(((uint8_t*)a->wordv) + byteshift, a->wordv, a->length_W * sizeof(bigint_word_t));
                memset(a->wordv, 0, byteshift);
        }
-       p = (bigint_word_t*)(((uint8_t*)a->wordv)+byteshift);
-       word_alloc = a->length_B+(byteshift+sizeof(bigint_word_t)-1)/sizeof(bigint_word_t)+1;
-       a->wordv[word_alloc-1]=0;
-       if(bitshift!=0){
-               for(i=0; i<a->length_B; ++i){
-                       t |= ((bigint_wordplus_t)p[i])<<bitshift;
-                       p[i] = (bigint_word_t)t;
-                       t >>= BIGINT_WORD_SIZE;
-               }
-               p[i] = (bigint_word_t)t;
+       if(bitshift == 0){
+           a->length_W += (byteshift + sizeof(bigint_word_t) - 1) / sizeof(bigint_word_t);
+           bigint_adjust(a);
+           return;
        }
-       a->length_B = word_alloc;
+       p = a->wordv + byteshift / sizeof(bigint_word_t);
+       words_to_shift = a->length_W + (byteshift % sizeof(bigint_word_t)?1:0);
+    for(i=0; i < words_to_shift; ++i){
+        t |= ((bigint_wordplus_t)p[i]) << bitshift;
+        p[i] = (bigint_word_t)t;
+        t >>= BIGINT_WORD_SIZE;
+    }
+    if(t){
+        p[i] = (bigint_word_t)t;
+        a->length_W += 1;
+    }
+    a->length_W += (byteshift + sizeof(bigint_word_t) - 1) / sizeof(bigint_word_t);
        bigint_adjust(a);
 }
 
@@ -409,33 +412,34 @@ void bigint_shiftright(bigint_t* a, uint16_t shift){
        uint16_t byteshift;
        uint16_t i;
        uint8_t bitshift;
-       bigint_wordplus_t t=0;
-       byteshift = shift/8;
-       bitshift = shift&7;
-       if(byteshift >= a->length_B*sizeof(bigint_word_t)){ /* we would shift out more than we have */
+       bigint_wordplus_t t = 0;
+       byteshift = shift / 8;
+       bitshift = shift & 7;
+
+       if(byteshift >= a->length_W * sizeof(bigint_word_t)){ /* we would shift out more than we have */
                bigint_set_zero(a);
                return;
        }
-       if(byteshift == a->length_B*sizeof(bigint_word_t)-1 && bitshift>GET_FBS(a)){
+       if(byteshift == a->length_W * sizeof(bigint_word_t) - 1 && bitshift > GET_FBS(a)){
                bigint_set_zero(a);
                return;
        }
+
        if(byteshift){
-               memmove(a->wordv, (uint8_t*)a->wordv+byteshift, a->length_B-byteshift);
-               memset((uint8_t*)a->wordv+a->length_B-byteshift, 0,  byteshift);
+               memmove(a->wordv, (uint8_t*)a->wordv + byteshift, a->length_W * sizeof(bigint_word_t) - byteshift);
        }
-       byteshift /= sizeof(bigint_word_t);
-       if(bitshift!=0){
+
+    byteshift /= sizeof(bigint_word_t); /* byteshift is now wordshift */
+    a->length_W -= byteshift;
+       if(bitshift != 0 && a->length_W){
         /* shift to the right */
-               for(i=a->length_B-byteshift-1; i>0; --i){
-                       t |= ((bigint_wordplus_t)(a->wordv[i]))<<(BIGINT_WORD_SIZE-bitshift);
-                       a->wordv[i] = (bigint_word_t)(t>>BIGINT_WORD_SIZE);
+               i = a->length_W - 1;
+               do{
+                       t |= ((bigint_wordplus_t)(a->wordv[i])) << (BIGINT_WORD_SIZE - bitshift);
+                       a->wordv[i] = (bigint_word_t)(t >> BIGINT_WORD_SIZE);
                        t <<= BIGINT_WORD_SIZE;
-               }
-               t |= ((bigint_wordplus_t)(a->wordv[0]))<<(BIGINT_WORD_SIZE-bitshift);
-               a->wordv[0] = (bigint_word_t)(t>>BIGINT_WORD_SIZE);
+               }while(i--);
        }
-    a->length_B -= ((shift/8)+sizeof(bigint_word_t)-1)/sizeof(bigint_word_t);
        bigint_adjust(a);
 }
 
@@ -443,7 +447,7 @@ void bigint_shiftright(bigint_t* a, uint16_t shift){
 
 void bigint_xor(bigint_t* dest, const bigint_t* a){
        uint16_t i;
-       for(i=0; i<a->length_B; ++i){
+       for(i=0; i<a->length_W; ++i){
                dest->wordv[i] ^= a->wordv[i];
        }
        bigint_adjust(dest);
@@ -452,7 +456,7 @@ void bigint_xor(bigint_t* dest, const bigint_t* a){
 /******************************************************************************/
 
 void bigint_set_zero(bigint_t* a){
-       a->length_B=0;
+       a->length_W=0;
 }
 
 /******************************************************************************/
@@ -460,96 +464,102 @@ void bigint_set_zero(bigint_t* a){
 /* using the Karatsuba-Algorithm */
 /* x*y = (xh*yh)*b**2n + ((xh+xl)*(yh+yl) - xh*yh - xl*yl)*b**n + yh*yl */
 void bigint_mul_u(bigint_t* dest, const bigint_t* a, const bigint_t* b){
-       if(a->length_B==0 || b->length_B==0){
+       if(a->length_W == 0 || b->length_W == 0){
                bigint_set_zero(dest);
                return;
        }
-       if(dest==a || dest==b){
+       if(dest == a || dest == b){
                bigint_t d;
-               bigint_word_t d_b[a->length_B+b->length_B];
+               bigint_word_t d_b[a->length_W + b->length_W];
                d.wordv = d_b;
                bigint_mul_u(&d, a, b);
                bigint_copy(dest, &d);
                return;
        }
-       if(a->length_B==1 || b->length_B==1){
-               if(a->length_B!=1){
+       if(a->length_W == 1 || b->length_W == 1){
+               if(a->length_W != 1){
                        XCHG_PTR(a,b);
                }
-               bigint_wordplus_t t=0;
+               bigint_wordplus_t t = 0;
                uint16_t i;
                bigint_word_t x = a->wordv[0];
-               for(i=0; i < b->length_B; ++i){
-                       t += ((bigint_wordplus_t)b->wordv[i])*((bigint_wordplus_t)x);
+               for(i=0; i < b->length_W; ++i){
+                       t += ((bigint_wordplus_t)b->wordv[i]) * ((bigint_wordplus_t)x);
                        dest->wordv[i] = (bigint_word_t)t;
-                       t>>=BIGINT_WORD_SIZE;
+                       t >>= BIGINT_WORD_SIZE;
+               }
+               dest->length_W = i;
+               if(t){
+                   dest->wordv[i] = (bigint_word_t)t;
+                   dest->length_W += 1;
                }
-               dest->wordv[i] = (bigint_word_t)t;
-               dest->length_B = i+1;
                dest->info = 0;
                bigint_adjust(dest);
                return;
        }
-       if(a->length_B * sizeof(bigint_word_t) <= 4 && b->length_B * sizeof(bigint_word_t) <= 4){
+       if(a->length_W * sizeof(bigint_word_t) <= 4 && b->length_W * sizeof(bigint_word_t) <= 4){
                uint32_t p=0, q=0;
                uint64_t r;
-               memcpy(&p, a->wordv, a->length_B*sizeof(bigint_word_t));
-               memcpy(&q, b->wordv, b->length_B*sizeof(bigint_word_t));
+               memcpy(&p, a->wordv, a->length_W*sizeof(bigint_word_t));
+               memcpy(&q, b->wordv, b->length_W*sizeof(bigint_word_t));
                r = (uint64_t)p * (uint64_t)q;
-               memcpy(dest->wordv, &r, (dest->length_B = a->length_B + b->length_B)*sizeof(bigint_word_t));
+               memcpy(dest->wordv, &r, (dest->length_W = a->length_W + b->length_W)*sizeof(bigint_word_t));
                bigint_adjust(dest);
                return;
        }
-       bigint_set_zero(dest);
        /* split a in xh & xl; split b in yh & yl */
-       const uint16_t n = (MAX(a->length_B, b->length_B)+1)/2;
+       const uint16_t n = (MAX(a->length_W, b->length_W)+1)/2;
        bigint_t xl, xh, yl, yh;
        xl.wordv = a->wordv;
        yl.wordv = b->wordv;
-       if(a->length_B<=n){
+       if(a->length_W<=n){
                bigint_set_zero(&xh);
-               xl.length_B = a->length_B;
+               xl.length_W = a->length_W;
                xl.info = a->info;
        }else{
-               xl.length_B=n;
+               xl.length_W=n;
                xl.info = 0;
                bigint_adjust(&xl);
                xh.wordv = &(a->wordv[n]);
-               xh.length_B = a->length_B-n;
+               xh.length_W = a->length_W-n;
                xh.info = a->info;
        }
-       if(b->length_B<=n){
+       if(b->length_W<=n){
                bigint_set_zero(&yh);
-               yl.length_B = b->length_B;
+               yl.length_W = b->length_W;
                yl.info = b->info;
        }else{
-               yl.length_B=n;
+               yl.length_W=n;
                yl.info = 0;
                bigint_adjust(&yl);
                yh.wordv = &(b->wordv[n]);
-               yh.length_B = b->length_B-n;
+               yh.length_W = b->length_W-n;
                yh.info = b->info;
        }
        /* now we have split up a and b */
        /* remember we want to do:
-        * x*y = (xh*yh)*b**2n + ((xh+xl)*(yh+yl) - xh*yh - xl*yl)*b**n + yh*yl
-        *          5          9     2   4   3    7   5   6   1         8   1
+        * x*y = (xh * b**n + xl) * (yh * b**n + yl)
+        *     = (xh * yh) * b**2n + xh * b**n * yl + yh * b**n * xl + xl * yl
+        *     = (xh * yh) * b**2n + (xh * yl + yh * xl) * b**n + xl *yl
+        *     // xh * yl + yh * xl = (xh + yh) * (xl + yl) - xh * yh - xl * yl
+        * x*y = (xh * yh) * b**2n + ((xh+xl)*(yh+yl) - xh*yh - xl*yl)*b**n + xl*yl
+        *          5              9     2   4   3    7   5   6   1         8   1
         */
-       bigint_word_t  tmp_b[2*n+2], m_b[2*(n+1)];
+       bigint_word_t  tmp_b[2 * n + 2], m_b[2 * (n + 1)];
        bigint_t tmp, tmp2, m;
        tmp.wordv = tmp_b;
-       tmp2.wordv = &(tmp_b[n+1]);
+       tmp2.wordv = &(tmp_b[n + 1]);
        m.wordv = m_b;
 
        bigint_mul_u(dest, &xl, &yl);  /* 1: dest <= xl*yl     */
        bigint_add_u(&tmp2, &xh, &xl); /* 2: tmp2 <= xh+xl     */
        bigint_add_u(&tmp, &yh, &yl);  /* 3: tmp  <= yh+yl     */
        bigint_mul_u(&m, &tmp2, &tmp); /* 4: m    <= tmp2*tmp  */
-       bigint_mul_u(&tmp, &xh, &yh);  /* 5: h    <= xh*yh     */
+       bigint_mul_u(&tmp, &xh, &yh);  /* 5: tmp  <= xh*yh     */
        bigint_sub_u(&m, &m, dest);    /* 6: m    <= m-dest    */
-    bigint_sub_u(&m, &m, &tmp);    /* 7: m    <= m-h       */
-       bigint_add_scale_u(dest, &m, n*sizeof(bigint_word_t)); /* 8: dest <= dest+m**n*/
-       bigint_add_scale_u(dest, &tmp, 2*n*sizeof(bigint_word_t)); /* 9: dest <= dest+tmp**(2*n) */
+    bigint_sub_u(&m, &m, &tmp);    /* 7: m    <= m-tmp     */
+       bigint_add_scale_u(dest, &m, n * sizeof(bigint_word_t));       /* 8: dest <= dest+m**n*/
+       bigint_add_scale_u(dest, &tmp, 2 * n * sizeof(bigint_word_t)); /* 9: dest <= dest+tmp**(2*n) */
 }
 
 /******************************************************************************/
@@ -582,66 +592,103 @@ void bigint_mul_s(bigint_t* dest, const bigint_t* a, const bigint_t* b){
 
 /******************************************************************************/
 
+
+#if DEBUG_SQUARE
+unsigned square_depth = 0;
+#endif
+
 /* square */
 /* (xh*b^n+xl)^2 = xh^2*b^2n + 2*xh*xl*b^n + xl^2 */
 void bigint_square(bigint_t* dest, const bigint_t* a){
-       if(a->length_B*sizeof(bigint_word_t)<=4){
-               uint64_t r=0;
-               memcpy(&r, a->wordv, a->length_B*sizeof(bigint_word_t));
-               r = r*r;
-               memcpy(dest->wordv, &r, 2*a->length_B*sizeof(bigint_word_t));
+       if(a->length_W * sizeof(bigint_word_t) <= 4){
+               uint64_t r = 0;
+               memcpy(&r, a->wordv, a->length_W * sizeof(bigint_word_t));
+               r = r * r;
+               memcpy(dest->wordv, &r, 2 * a->length_W * sizeof(bigint_word_t));
                SET_POS(dest);
-               dest->length_B=2*a->length_B;
+               dest->length_W = 2 * a->length_W;
                bigint_adjust(dest);
                return;
        }
-       if(dest==a){
+       if(dest->wordv == a->wordv){
                bigint_t d;
-               bigint_word_t d_b[a->length_B*2];
+               bigint_word_t d_b[a->length_W*2];
                d.wordv = d_b;
                bigint_square(&d, a);
                bigint_copy(dest, &d);
                return;
        }
+
+#if DEBUG_SQUARE
+       square_depth += 1;
+#endif
+
        uint16_t n;
-       n=(a->length_B+1)/2;
+       n=(a->length_W+1)/2;
        bigint_t xh, xl, tmp; /* x-high, x-low, temp */
        bigint_word_t buffer[2*n+1];
        xl.wordv = a->wordv;
-       xl.length_B = n;
+       xl.length_W = n;
        xl.info = 0;
        xh.wordv = &(a->wordv[n]);
-       xh.length_B = a->length_B-n;
-       xh.info = 0;
+       xh.length_W = a->length_W-n;
+       xh.info = a->info;
        bigint_adjust(&xl);
-       bigint_adjust(&xh);
        tmp.wordv = buffer;
 /* (xh * b**n + xl)**2 = xh**2 * b**2n + 2 * xh * xl * b**n + xl**2 */
-
-//     cli_putstr("\r\nDBG (a): xl: "); bigint_print_hex(&xl);
-//     cli_putstr("\r\nDBG (b): xh: "); bigint_print_hex(&xh);
+#if DEBUG_SQUARE
+       if(square_depth == 1){
+        cli_putstr("\r\nDBG (a): xl: "); bigint_print_hex(&xl);
+        cli_putstr("\r\nDBG (b): xh: "); bigint_print_hex(&xh);
+       }
+#endif
        bigint_square(dest, &xl);
-//     cli_putstr("\r\nDBG (1): xl**2: "); bigint_print_hex(dest);
-       bigint_square(&tmp, &xh);
-//     cli_putstr("\r\nDBG (2): xh**2: "); bigint_print_hex(&tmp);
-       bigint_add_scale_u(dest, &tmp, 2*n*sizeof(bigint_word_t));
-//     cli_putstr("\r\nDBG (3): xl**2 + xh**2*n**2: "); bigint_print_hex(dest);
+#if DEBUG_SQUARE
+       if(square_depth == 1){
+           cli_putstr("\r\nDBG (1): xl**2: "); bigint_print_hex(dest);
+       }
+#endif
+    bigint_square(&tmp, &xh);
+#if DEBUG_SQUARE
+    if(square_depth == 1){
+        cli_putstr("\r\nDBG (2): xh**2: "); bigint_print_hex(&tmp);
+    }
+#endif
+       bigint_add_scale_u(dest, &tmp, 2 * n * sizeof(bigint_word_t));
+#if DEBUG_SQUARE
+       if(square_depth == 1){
+           cli_putstr("\r\nDBG (3): xl**2 + xh**2*n**2: "); bigint_print_hex(dest);
+       }
+#endif
        bigint_mul_u(&tmp, &xl, &xh);
-//     cli_putstr("\r\nDBG (4): xl*xh: "); bigint_print_hex(&tmp);
+#if DEBUG_SQUARE
+       if(square_depth == 1){
+           cli_putstr("\r\nDBG (4): xl*xh: "); bigint_print_hex(&tmp);
+       }
+#endif
        bigint_shiftleft(&tmp, 1);
-//     cli_putstr("\r\nDBG (5): xl*xh*2: "); bigint_print_hex(&tmp);
-       bigint_add_scale_u(dest, &tmp, n*sizeof(bigint_word_t));
-//     cli_putstr("\r\nDBG (6): x**2: "); bigint_print_hex(dest);
-//     cli_putstr("\r\n");
+#if DEBUG_SQUARE
+       if(square_depth == 1){
+           cli_putstr("\r\nDBG (5): xl*xh*2: "); bigint_print_hex(&tmp);
+       }
+#endif
+       bigint_add_scale_u(dest, &tmp, n * sizeof(bigint_word_t));
+#if DEBUG_SQUARE
+       if(square_depth == 1){
+           cli_putstr("\r\nDBG (6): x**2: "); bigint_print_hex(dest);
+           cli_putstr("\r\n");
+       }
+       square_depth -= 1;
+#endif
 }
 
 /******************************************************************************/
 void bigint_sub_u_bitscale(bigint_t* a, const bigint_t* b, uint16_t bitscale){
        bigint_t tmp, x;
-       bigint_word_t tmp_b[b->length_B + 1];
+       bigint_word_t tmp_b[b->length_W + 1];
        const uint16_t word_shift = bitscale / BIGINT_WORD_SIZE;
 
-       if(a->length_B < b->length_B + word_shift){
+       if(a->length_W < b->length_W + word_shift){
 #if DEBUG
                cli_putstr("\r\nDBG: *bang*\r\n");
 #endif
@@ -654,7 +701,7 @@ void bigint_sub_u_bitscale(bigint_t* a, const bigint_t* b, uint16_t bitscale){
 
        x.info = a->info;
        x.wordv = &(a->wordv[word_shift]);
-       x.length_B = a->length_B - word_shift;
+       x.length_W = a->length_W - word_shift;
 
        bigint_sub_u(&x, &x, &tmp);
        bigint_adjust(a);
@@ -669,40 +716,40 @@ void bigint_reduce(bigint_t* a, const bigint_t* r){
 #if DEBUG
        cli_putstr("\r\nDBG: (a) = "); bigint_print_hex(a);
 #endif
-       if(r->length_B==0 || a->length_B==0){
+       if(r->length_W==0 || a->length_W==0){
                return;
        }
-       if((r->length_B*sizeof(bigint_word_t)<=4) && (a->length_B*sizeof(bigint_word_t)<=4)){
+       if((r->length_W*sizeof(bigint_word_t)<=4) && (a->length_W*sizeof(bigint_word_t)<=4)){
                uint32_t p=0, q=0;
-               memcpy(&p, a->wordv, a->length_B*sizeof(bigint_word_t));
-               memcpy(&q, r->wordv, r->length_B*sizeof(bigint_word_t));
+               memcpy(&p, a->wordv, a->length_W*sizeof(bigint_word_t));
+               memcpy(&q, r->wordv, r->length_W*sizeof(bigint_word_t));
                p %= q;
-               memcpy(a->wordv, &p, a->length_B*sizeof(bigint_word_t));
+               memcpy(a->wordv, &p, a->length_W*sizeof(bigint_word_t));
                bigint_adjust(a);
 //             cli_putstr("\r\nDBG: (0) = "); bigint_print_hex(a);
                return;
        }
        uint16_t shift;
-       while(a->length_B > r->length_B){
-               shift = (a->length_B - r->length_B) * 8 * sizeof(bigint_word_t) + GET_FBS(a) - rfbs - 1;
+       while(a->length_W > r->length_W){
+               shift = (a->length_W - r->length_W) * 8 * sizeof(bigint_word_t) + GET_FBS(a) - rfbs - 1;
                /*
-               if((a->wordv[a->length_B-1] & ((1LL<<GET_FBS(a)) - 1)) > r->wordv[r->length_B-1]){
+               if((a->wordv[a->length_W-1] & ((1LL<<GET_FBS(a)) - 1)) > r->wordv[r->length_W-1]){
                        // cli_putc('~');
                        cli_putstr("\r\n ~ [a] = ");
-                       cli_hexdump_rev(&a->wordv[a->length_B-1], 4);
+                       cli_hexdump_rev(&a->wordv[a->length_W-1], 4);
                        cli_putstr("  [r] = ");
-                       cli_hexdump_rev(&r->wordv[r->length_B-1], 4);
+                       cli_hexdump_rev(&r->wordv[r->length_W-1], 4);
                        shift += 1;
                }
                */
 //             cli_putstr("\r\nDBG: (p) shift = "); cli_hexdump_rev(&shift, 2);
-//             cli_putstr(" a_len = "); cli_hexdump_rev(&a->length_B, 2);
-//             cli_putstr(" r_len = "); cli_hexdump_rev(&r->length_B, 2);
+//             cli_putstr(" a_len = "); cli_hexdump_rev(&a->length_W, 2);
+//             cli_putstr(" r_len = "); cli_hexdump_rev(&r->length_W, 2);
 //             uart_flush(0);
                bigint_sub_u_bitscale(a, r, shift);
 //             cli_putstr("\r\nDBG: (1) = "); bigint_print_hex(a);
        }
-       while((GET_FBS(a) > rfbs) && (a->length_B == r->length_B)){
+       while((GET_FBS(a) > rfbs) && (a->length_W == r->length_W)){
                shift = GET_FBS(a)-rfbs-1;
 //             cli_putstr("\r\nDBG: (q) shift = "); cli_hexdump_rev(&shift, 2);
                bigint_sub_u_bitscale(a, r, shift);
@@ -722,12 +769,12 @@ void bigint_reduce(bigint_t* a, const bigint_t* r){
 /* calculate dest = a**exp % r */
 /* using square&multiply */
 void bigint_expmod_u(bigint_t* dest, const bigint_t* a, const bigint_t* exp, const bigint_t* r){
-       if(a->length_B==0 || r->length_B==0){
+       if(a->length_W==0 || r->length_W==0){
                return;
        }
 
        bigint_t res, base;
-       bigint_word_t t, base_b[MAX(a->length_B,r->length_B)], res_b[r->length_B*2];
+       bigint_word_t t, base_b[MAX(a->length_W,r->length_W)], res_b[r->length_W*2];
        uint16_t i;
        uint8_t j;
 //     uint16_t *xaddr = &i;
@@ -740,16 +787,16 @@ void bigint_expmod_u(bigint_t* dest, const bigint_t* a, const bigint_t* exp, con
 //     cli_putstr("\r\npost-copy");
        bigint_reduce(&base, r);
        res.wordv[0]=1;
-       res.length_B=1;
+       res.length_W=1;
        res.info = 0;
        bigint_adjust(&res);
-       if(exp->length_B == 0){
+       if(exp->length_W == 0){
                bigint_copy(dest, &res);
                return;
        }
        uint8_t flag = 0;
-       t=exp->wordv[exp->length_B - 1];
-       for(i=exp->length_B; i > 0; --i){
+       t=exp->wordv[exp->length_W - 1];
+       for(i=exp->length_W; i > 0; --i){
                t = exp->wordv[i - 1];
                for(j=BIGINT_WORD_SIZE; j > 0; --j){
                        if(!flag){
@@ -760,12 +807,12 @@ void bigint_expmod_u(bigint_t* dest, const bigint_t* a, const bigint_t* exp, con
                        if(flag){
                                bigint_square(&res, &res);
                                bigint_reduce(&res, r);
-                               if(t & (1<<(BIGINT_WORD_SIZE-1))){
+                               if(t & (1 << (BIGINT_WORD_SIZE - 1))){
                                        bigint_mul_u(&res, &res, &base);
                                        bigint_reduce(&res, r);
                                }
                        }
-                       t<<=1;
+                       t <<= 1;
                }
        }
 
@@ -775,24 +822,43 @@ void bigint_expmod_u(bigint_t* dest, const bigint_t* a, const bigint_t* exp, con
 }
 
 /******************************************************************************/
-
+#if 1
 #define cli_putstr(a)
+#define cli_putstr_P(a)
 #define bigint_print_hex(a)
 #define cli_hexdump_rev(a,b)
 #define uart_flush(a)
-
+#define printf_P(...)
+#endif
 /* gcd <-- gcd(x,y) a*x+b*y=gcd */
 void bigint_gcdext(bigint_t* gcd, bigint_t* a, bigint_t* b, const bigint_t* x, const bigint_t* y){
-        bigint_t g, x_, y_, u, v, a_, b_, c_, d_;
-        uint16_t i=0;
-        if(x->length_B==0 || y->length_B==0){
+        uint16_t i = 0;
+        printf_P(PSTR("\nDBG: gcdext( "));
+        bigint_print_hex(x);
+     printf_P(PSTR(", "));
+     bigint_print_hex(y);
+     printf_P(PSTR(")\n"));
+        if(x->length_W == 0 || y->length_W == 0){
+            printf_P(PSTR("\nDBG: got zero in gcd <%s %s %d>\n"), __FILE__, __func__, __LINE__);
+            if(gcd){
+                bigint_set_zero(gcd);
+            }
+            if(a){
+                bigint_set_zero(a);
+            }
+         if(b){
+             bigint_set_zero(b);
+         }
                 return;
         }
-        if(x->length_B==1 && x->wordv[0]==1){
-                gcd->length_B = 1;
-                gcd->wordv[0] = 1;
+        if(x->length_W == 1 && x->wordv[0] == 1){
+            if(gcd){
+             gcd->length_W = 1;
+             gcd->wordv[0] = 1;
+             gcd->info = 0;
+            }
                 if(a){
-                        a->length_B = 1;
+                        a->length_W = 1;
                         a->wordv[0] = 1;
                         SET_POS(a);
                         bigint_adjust(a);
@@ -802,11 +868,14 @@ void bigint_gcdext(bigint_t* gcd, bigint_t* a, bigint_t* b, const bigint_t* x, c
                 }
                 return;
         }
-        if(y->length_B==1 && y->wordv[0]==1){
-                gcd->length_B = 1;
-                gcd->wordv[0] = 1;
+        if(y->length_W == 1 && y->wordv[0] == 1){
+                if(gcd){
+             gcd->length_W = 1;
+             gcd->wordv[0] = 1;
+             gcd->info = 0;
+                }
                 if(b){
-                        b->length_B = 1;
+                        b->length_W = 1;
                         b->wordv[0] = 1;
                         SET_POS(b);
                         bigint_adjust(b);
@@ -817,27 +886,28 @@ void bigint_gcdext(bigint_t* gcd, bigint_t* a, bigint_t* b, const bigint_t* x, c
                 return;
         }
 
-        while(x->wordv[i]==0 && y->wordv[i]==0){
+        while(x->wordv[i] == 0 && y->wordv[i] == 0){
                 ++i;
         }
-        bigint_word_t g_b[i+2], x_b[x->length_B-i], y_b[y->length_B-i];
-        bigint_word_t u_b[x->length_B-i], v_b[y->length_B-i];
-        bigint_word_t a_b[y->length_B+2], c_b[y->length_B+2];
-        bigint_word_t b_b[x->length_B+2], d_b[x->length_B+2];
+        bigint_word_t g_b[i + 2], x_b[x->length_W - i], y_b[y->length_W - i];
+        bigint_word_t u_b[x->length_W - i], v_b[y->length_W - i];
+        bigint_word_t a_b[y->length_W + 2], c_b[y->length_W + 2];
+        bigint_word_t b_b[x->length_W + 2], d_b[x->length_W + 2];
+     bigint_t g, x_, y_, u, v, a_, b_, c_, d_;
 
         g.wordv = g_b;
         x_.wordv = x_b;
         y_.wordv = y_b;
-        memset(g_b, 0, i*sizeof(bigint_word_t));
-        g_b[i]=1;
-        g.length_B = i+1;
-        g.info=0;
+        memset(g_b, 0, i * sizeof(bigint_word_t));
+        g_b[i] = 1;
+        g.length_W = i + 1;
+        g.info = 0;
         x_.info = y_.info = 0;
-        x_.length_B = x->length_B-i;
-        y_.length_B = y->length_B-i;
-        memcpy(x_.wordv, x->wordv+i, x_.length_B*sizeof(bigint_word_t));
-        memcpy(y_.wordv, y->wordv+i, y_.length_B*sizeof(bigint_word_t));
-        for(i=0; (x_.wordv[0]&(1<<i))==0 && (y_.wordv[0]&(1<<i))==0; ++i){
+        x_.length_W = x->length_W - i;
+        y_.length_W = y->length_W - i;
+        memcpy(x_.wordv, x->wordv + i, x_.length_W * sizeof(bigint_word_t));
+        memcpy(y_.wordv, y->wordv + i, y_.length_W * sizeof(bigint_word_t));
+        for(i = 0; (x_.wordv[0] & (1 << i)) == 0 && (y_.wordv[0] & (1 << i)) == 0; ++i){
         }
 
         bigint_adjust(&x_);
@@ -859,46 +929,69 @@ void bigint_gcdext(bigint_t* gcd, bigint_t* a, bigint_t* b, const bigint_t* x, c
         bigint_copy(&u, &x_);
         bigint_copy(&v, &y_);
         a_.wordv[0] = 1;
-        a_.length_B = 1;
+        a_.length_W = 1;
         a_.info = 0;
         d_.wordv[0] = 1;
-        d_.length_B = 1;
+        d_.length_W = 1;
         d_.info = 0;
         bigint_set_zero(&b_);
         bigint_set_zero(&c_);
+     printf_P(PSTR("\nloop: x_ = "));
+     bigint_print_hex(&x_);
+     printf_P(PSTR("; y_ = "));
+     bigint_print_hex(&y_);
         do{
-                cli_putstr("\r\nDBG (gcdext) 0");
-                while((u.wordv[0]&1)==0){
-                        cli_putstr("\r\nDBG (gcdext) 0.1");
+                printf_P(PSTR("\nDBG (gcdext) 0"));
+                while((u.wordv[0] & 1) == 0){
+                        printf_P(PSTR("\nDBG (gcdext) 0.1"));
                         bigint_shiftright(&u, 1);
-                        if((a_.wordv[0]&1) || (b_.wordv[0]&1)){
+                        if((a_.wordv[0] & 1) || (b_.wordv[0] & 1)){
                                 bigint_add_s(&a_, &a_, &y_);
                                 bigint_sub_s(&b_, &b_, &x_);
                         }
                         bigint_shiftright(&a_, 1);
                         bigint_shiftright(&b_, 1);
+                        printf_P(PSTR(" a_ = "));
+                        bigint_print_hex(&a_);
+                        printf_P(PSTR("; b_ = "));
+                        bigint_print_hex(&b_);
                 }
-                while((v.wordv[0]&1)==0){
-                        cli_putstr("\r\nDBG (gcdext) 0.2");
+                while((v.wordv[0] & 1) == 0){
+                        printf_P(PSTR("\nDBG (gcdext) 0.2"));
                         bigint_shiftright(&v, 1);
-                        if((c_.wordv[0]&1) || (d_.wordv[0]&1)){
+                        if((c_.wordv[0] & 1) || (d_.wordv[0] & 1)){
                                 bigint_add_s(&c_, &c_, &y_);
                                 bigint_sub_s(&d_, &d_, &x_);
                         }
+             printf_P(PSTR(" c* = "));
+             bigint_print_hex(&c_);
                         bigint_shiftright(&c_, 1);
                         bigint_shiftright(&d_, 1);
-
+             printf_P(PSTR(" c_ = "));
+             bigint_print_hex(&c_);
+             printf_P(PSTR("; d_ = "));
+             bigint_print_hex(&d_);
                 }
-                if(bigint_cmp_u(&u, &v)>=0){
+                if(bigint_cmp_u(&u, &v) >= 0){
+             printf_P(PSTR("\nDBG (gcdext) 0.3"));
                        bigint_sub_u(&u, &u, &v);
                        bigint_sub_s(&a_, &a_, &c_);
                        bigint_sub_s(&b_, &b_, &d_);
+            printf_P(PSTR(" a_ = "));
+            bigint_print_hex(&a_);
+            printf_P(PSTR("; b_ = "));
+            bigint_print_hex(&b_);
                 }else{
+             printf_P(PSTR("\nDBG (gcdext) 0.4"));
                        bigint_sub_u(&v, &v, &u);
                        bigint_sub_s(&c_, &c_, &a_);
                        bigint_sub_s(&d_, &d_, &b_);
+            printf_P(PSTR(" c_ = "));
+            bigint_print_hex(&c_);
+            printf_P(PSTR("; d_ = "));
+            bigint_print_hex(&d_);
                 }
-        }while(u.length_B);
+        }while(u.length_W);
         if(gcd){
                 bigint_mul_s(gcd, &v, &g);
         }
@@ -924,7 +1017,7 @@ void bigint_inverse(bigint_t* dest, const bigint_t* a, const bigint_t* m){
 void bigint_changeendianess(bigint_t* a){
        uint8_t t, *p, *q;
        p = (uint8_t*)(a->wordv);
-       q = ((uint8_t*)p)+a->length_B*sizeof(bigint_word_t)-1;
+       q = p + a->length_W * sizeof(bigint_word_t) - 1;
        while(p<q){
                t = *p;
                *p = *q;