view tomsfastmath/src/mul/fp_mul_comba.c @ 645:8622ee48fab5 dropbear-tfm

- Work around broken asm constraint behaviour on 32bit x86 OS X
author Matt Johnston <matt@ucc.asn.au>
date Wed, 30 Nov 2011 22:27:26 +0800
parents a362b62d38b2
children
line wrap: on
line source
/* 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, [email protected]
 */

/* About this file...

*/

#include <tfm.h>

#if defined(TFM_PRESCOTT) && defined(TFM_SSE2)
   #undef TFM_SSE2
   #define TFM_X86
#endif

/* these are the combas.  Worship them. */
#if defined(TFM_X86)
/* Generic x86 optimized code */

/* anything you need at the start */
#define COMBA_START

/* clear the chaining variables */
#define COMBA_CLEAR \
   c0 = c1 = c2 = 0;

/* forward the carry to the next digit */
#define COMBA_FORWARD \
   do { c0 = c1; c1 = c2; c2 = 0; } while (0);

/* store the first sum */
#define COMBA_STORE(x) \
   x = c0;

/* store the second sum [carry] */
#define COMBA_STORE2(x) \
   x = c1;

/* anything you need at the end */
#define COMBA_FINI

/* this should multiply i and j  */
#define MULADD(i, j)                                      \
	do { fp_digit ti = (i); \
		fp_digit tj = (j); \
asm(                                                      \
     "movl  %6,%%eax     \n\t"                            \
     "mull  %7           \n\t"                            \
     "addl  %%eax,%0     \n\t"                            \
     "adcl  %%edx,%1     \n\t"                            \
     "adcl  $0,%2        \n\t"                            \
     :"=r"(c0), "=r"(c1), "=r"(c2): "0"(c0), "1"(c1), "2"(c2), "m"(ti), "m"(tj)  :"%eax","%edx","%cc"); } while (0);

#elif defined(TFM_X86_64)
/* x86-64 optimized */

/* anything you need at the start */
#define COMBA_START

/* clear the chaining variables */
#define COMBA_CLEAR \
   c0 = c1 = c2 = 0;

/* forward the carry to the next digit */
#define COMBA_FORWARD \
   do { c0 = c1; c1 = c2; c2 = 0; } while (0);

/* store the first sum */
#define COMBA_STORE(x) \
   x = c0;

/* store the second sum [carry] */
#define COMBA_STORE2(x) \
   x = c1;

/* anything you need at the end */
#define COMBA_FINI

/* this should multiply i and j  */
#define MULADD(i, j)                                      \
asm  (                                                    \
     "movq  %6,%%rax     \n\t"                            \
     "mulq  %7           \n\t"                            \
     "addq  %%rax,%0     \n\t"                            \
     "adcq  %%rdx,%1     \n\t"                            \
     "adcq  $0,%2        \n\t"                            \
     :"=r"(c0), "=r"(c1), "=r"(c2): "0"(c0), "1"(c1), "2"(c2), "g"(i), "g"(j)  :"%rax","%rdx","%cc");

#elif defined(TFM_SSE2)
/* use SSE2 optimizations */

/* anything you need at the start */
#define COMBA_START

/* clear the chaining variables */
#define COMBA_CLEAR \
   c0 = c1 = c2 = 0;

/* forward the carry to the next digit */
#define COMBA_FORWARD \
   do { c0 = c1; c1 = c2; c2 = 0; } while (0);

/* store the first sum */
#define COMBA_STORE(x) \
   x = c0;

/* store the second sum [carry] */
#define COMBA_STORE2(x) \
   x = c1;

/* anything you need at the end */
#define COMBA_FINI \
   asm("emms");

/* this should multiply i and j  */
#define MULADD(i, j)                                     \
asm(                                                     \
    "movd  %6,%%mm0     \n\t"                            \
    "movd  %7,%%mm1     \n\t"                            \
    "pmuludq %%mm1,%%mm0\n\t"                            \
    "movd  %%mm0,%%eax  \n\t"                            \
    "psrlq $32,%%mm0    \n\t"                            \
    "addl  %%eax,%0     \n\t"                            \
    "movd  %%mm0,%%eax  \n\t"                            \
    "adcl  %%eax,%1     \n\t"                            \
    "adcl  $0,%2        \n\t"                            \
    :"=r"(c0), "=r"(c1), "=r"(c2): "0"(c0), "1"(c1), "2"(c2), "m"(i), "m"(j)  :"%eax","%cc");

#elif defined(TFM_ARM)
/* ARM code */

#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(                                                          \
"  UMULL  r0,r1,%6,%7           \n\t"                         \
"  ADDS   %0,%0,r0              \n\t"                         \
"  ADCS   %1,%1,r1              \n\t"                         \
"  ADC    %2,%2,#0              \n\t"                         \
:"=r"(c0), "=r"(c1), "=r"(c2) : "0"(c0), "1"(c1), "2"(c2), "r"(i), "r"(j) : "r0", "r1", "%cc");

#elif defined(TFM_PPC32)
/* For 32-bit PPC */

#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 
   
/* untested: will mulhwu change the flags?  Docs say no */
#define MULADD(i, j)              \
asm(                              \
   " mullw  16,%6,%7       \n\t" \
   " addc   %0,%0,16       \n\t" \
   " mulhwu 16,%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");

#elif defined(TFM_PPC64)
/* For 64-bit PPC */

#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 
   
/* untested: will mulhdu change the flags?  Docs say no */
#define MULADD(i, j)              \
asm(                              \
   " mulld  r16,%6,%7       \n\t" \
   " addc   %0,%0,16       \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):"r16");

#elif defined(TFM_AVR32)

/* ISO C code */

#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(                             \
   " mulu.d r2,%6,%7        \n\t"\
   " add    %0,r2           \n\t"\
   " adc    %1,%1,r3        \n\t"\
   " 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 */

#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)                                                              \
   do { fp_word t;                                                                \
   t = (fp_word)c0 + ((fp_word)i) * ((fp_word)j); c0 = t;                         \
   t = (fp_word)c1 + (t >> DIGIT_BIT);            c1 = t; c2 += t >> DIGIT_BIT;   \
   } while (0);

#endif

#ifndef TFM_DEFINES

/* generic PxQ multiplier */
void fp_mul_comba(fp_int *A, fp_int *B, fp_int *C)
{
   int       ix, iy, iz, tx, ty, pa;
   fp_digit  c0, c1, c2, *tmpx, *tmpy;
   fp_int    tmp, *dst;

   COMBA_START;
   COMBA_CLEAR;
   
   /* get size of output and trim */
   pa = A->used + B->used;
   if (pa >= FP_SIZE) {
      pa = FP_SIZE-1;
   }

   if (A == C || B == C) {
      fp_zero(&tmp);
      dst = &tmp;
   } else {
      fp_zero(C);
      dst = C;
   }

   for (ix = 0; ix < pa; ix++) {
      /* get offsets into the two bignums */
      ty = MIN(ix, B->used-1);
      tx = ix - ty;

      /* setup temp aliases */
      tmpx = A->dp + tx;
      tmpy = B->dp + ty;

      /* this is the number of times the loop will iterrate, essentially its 
         while (tx++ < a->used && ty-- >= 0) { ... }
       */
      iy = MIN(A->used-tx, ty+1);

      /* execute loop */
      COMBA_FORWARD;
      for (iz = 0; iz < iy; ++iz) {
          MULADD(*tmpx++, *tmpy--);
      }

      /* store term */
      COMBA_STORE(dst->dp[ix]);
  }
  COMBA_FINI;

  dst->used = pa;
  dst->sign = A->sign ^ B->sign;
  fp_clamp(dst);
  fp_copy(dst, C);
}

#endif

/* $Source$ */
/* $Revision$ */
/* $Date$ */