| 1 | #! /usr/bin/env perl
|
|---|
| 2 | # Copyright 2005-2021 The OpenSSL Project Authors. All Rights Reserved.
|
|---|
| 3 | #
|
|---|
| 4 | # Licensed under the Apache License 2.0 (the "License"). You may not use
|
|---|
| 5 | # this file except in compliance with the License. You can obtain a copy
|
|---|
| 6 | # in the file LICENSE in the source distribution or at
|
|---|
| 7 | # https://www.openssl.org/source/license.html
|
|---|
| 8 |
|
|---|
| 9 |
|
|---|
| 10 | # ====================================================================
|
|---|
| 11 | # Written by Andy Polyakov <[email protected]> for the OpenSSL
|
|---|
| 12 | # project. The module is, however, dual licensed under OpenSSL and
|
|---|
| 13 | # CRYPTOGAMS licenses depending on where you obtain it. For further
|
|---|
| 14 | # details see http://www.openssl.org/~appro/cryptogams/.
|
|---|
| 15 | # ====================================================================
|
|---|
| 16 |
|
|---|
| 17 | # December 2005
|
|---|
| 18 | #
|
|---|
| 19 | # Pure SPARCv9/8+ and IALU-only bn_mul_mont implementation. The reasons
|
|---|
| 20 | # for undertaken effort are multiple. First of all, UltraSPARC is not
|
|---|
| 21 | # the whole SPARCv9 universe and other VIS-free implementations deserve
|
|---|
| 22 | # optimized code as much. Secondly, newly introduced UltraSPARC T1,
|
|---|
| 23 | # a.k.a. Niagara, has shared FPU and concurrent FPU-intensive paths,
|
|---|
| 24 | # such as sparcv9a-mont, will simply sink it. Yes, T1 is equipped with
|
|---|
| 25 | # several integrated RSA/DSA accelerator circuits accessible through
|
|---|
| 26 | # kernel driver [only(*)], but having decent user-land software
|
|---|
| 27 | # implementation is important too. Finally, reasons like desire to
|
|---|
| 28 | # experiment with dedicated squaring procedure. Yes, this module
|
|---|
| 29 | # implements one, because it was easiest to draft it in SPARCv9
|
|---|
| 30 | # instructions...
|
|---|
| 31 |
|
|---|
| 32 | # (*) Engine accessing the driver in question is on my TODO list.
|
|---|
| 33 | # For reference, accelerator is estimated to give 6 to 10 times
|
|---|
| 34 | # improvement on single-threaded RSA sign. It should be noted
|
|---|
| 35 | # that 6-10x improvement coefficient does not actually mean
|
|---|
| 36 | # something extraordinary in terms of absolute [single-threaded]
|
|---|
| 37 | # performance, as SPARCv9 instruction set is by all means least
|
|---|
| 38 | # suitable for high performance crypto among other 64 bit
|
|---|
| 39 | # platforms. 6-10x factor simply places T1 in same performance
|
|---|
| 40 | # domain as say AMD64 and IA-64. Improvement of RSA verify don't
|
|---|
| 41 | # appear impressive at all, but it's the sign operation which is
|
|---|
| 42 | # far more critical/interesting.
|
|---|
| 43 |
|
|---|
| 44 | # You might notice that inner loops are modulo-scheduled:-) This has
|
|---|
| 45 | # essentially negligible impact on UltraSPARC performance, it's
|
|---|
| 46 | # Fujitsu SPARC64 V users who should notice and hopefully appreciate
|
|---|
| 47 | # the advantage... Currently this module surpasses sparcv9a-mont.pl
|
|---|
| 48 | # by ~20% on UltraSPARC-III and later cores, but recall that sparcv9a
|
|---|
| 49 | # module still have hidden potential [see TODO list there], which is
|
|---|
| 50 | # estimated to be larger than 20%...
|
|---|
| 51 |
|
|---|
| 52 | $output = pop and open STDOUT,">$output";
|
|---|
| 53 |
|
|---|
| 54 | # int bn_mul_mont(
|
|---|
| 55 | $rp="%i0"; # BN_ULONG *rp,
|
|---|
| 56 | $ap="%i1"; # const BN_ULONG *ap,
|
|---|
| 57 | $bp="%i2"; # const BN_ULONG *bp,
|
|---|
| 58 | $np="%i3"; # const BN_ULONG *np,
|
|---|
| 59 | $n0="%i4"; # const BN_ULONG *n0,
|
|---|
| 60 | $num="%i5"; # int num);
|
|---|
| 61 |
|
|---|
| 62 | $frame="STACK_FRAME";
|
|---|
| 63 | $bias="STACK_BIAS";
|
|---|
| 64 |
|
|---|
| 65 | $car0="%o0";
|
|---|
| 66 | $car1="%o1";
|
|---|
| 67 | $car2="%o2"; # 1 bit
|
|---|
| 68 | $acc0="%o3";
|
|---|
| 69 | $acc1="%o4";
|
|---|
| 70 | $mask="%g1"; # 32 bits, what a waste...
|
|---|
| 71 | $tmp0="%g4";
|
|---|
| 72 | $tmp1="%g5";
|
|---|
| 73 |
|
|---|
| 74 | $i="%l0";
|
|---|
| 75 | $j="%l1";
|
|---|
| 76 | $mul0="%l2";
|
|---|
| 77 | $mul1="%l3";
|
|---|
| 78 | $tp="%l4";
|
|---|
| 79 | $apj="%l5";
|
|---|
| 80 | $npj="%l6";
|
|---|
| 81 | $tpj="%l7";
|
|---|
| 82 |
|
|---|
| 83 | $fname="bn_mul_mont_int";
|
|---|
| 84 |
|
|---|
| 85 | $code=<<___;
|
|---|
| 86 | #ifndef __ASSEMBLER__
|
|---|
| 87 | # define __ASSEMBLER__ 1
|
|---|
| 88 | #endif
|
|---|
| 89 | #include "crypto/sparc_arch.h"
|
|---|
| 90 |
|
|---|
| 91 | .section ".text",#alloc,#execinstr
|
|---|
| 92 |
|
|---|
| 93 | .global $fname
|
|---|
| 94 | .align 32
|
|---|
| 95 | $fname:
|
|---|
| 96 | cmp %o5,4 ! 128 bits minimum
|
|---|
| 97 | bge,pt %icc,.Lenter
|
|---|
| 98 | sethi %hi(0xffffffff),$mask
|
|---|
| 99 | retl
|
|---|
| 100 | clr %o0
|
|---|
| 101 | .align 32
|
|---|
| 102 | .Lenter:
|
|---|
| 103 | save %sp,-$frame,%sp
|
|---|
| 104 | sll $num,2,$num ! num*=4
|
|---|
| 105 | or $mask,%lo(0xffffffff),$mask
|
|---|
| 106 | ld [$n0],$n0
|
|---|
| 107 | cmp $ap,$bp
|
|---|
| 108 | and $num,$mask,$num
|
|---|
| 109 | ld [$bp],$mul0 ! bp[0]
|
|---|
| 110 | nop
|
|---|
| 111 |
|
|---|
| 112 | add %sp,$bias,%o7 ! real top of stack
|
|---|
| 113 | ld [$ap],$car0 ! ap[0] ! redundant in squaring context
|
|---|
| 114 | sub %o7,$num,%o7
|
|---|
| 115 | ld [$ap+4],$apj ! ap[1]
|
|---|
| 116 | and %o7,-1024,%o7
|
|---|
| 117 | ld [$np],$car1 ! np[0]
|
|---|
| 118 | sub %o7,$bias,%sp ! alloca
|
|---|
| 119 | ld [$np+4],$npj ! np[1]
|
|---|
| 120 | be,pt SIZE_T_CC,.Lbn_sqr_mont
|
|---|
| 121 | mov 12,$j
|
|---|
| 122 |
|
|---|
| 123 | mulx $car0,$mul0,$car0 ! ap[0]*bp[0]
|
|---|
| 124 | mulx $apj,$mul0,$tmp0 !prologue! ap[1]*bp[0]
|
|---|
| 125 | and $car0,$mask,$acc0
|
|---|
| 126 | add %sp,$bias+$frame,$tp
|
|---|
| 127 | ld [$ap+8],$apj !prologue!
|
|---|
| 128 |
|
|---|
| 129 | mulx $n0,$acc0,$mul1 ! "t[0]"*n0
|
|---|
| 130 | and $mul1,$mask,$mul1
|
|---|
| 131 |
|
|---|
| 132 | mulx $car1,$mul1,$car1 ! np[0]*"t[0]"*n0
|
|---|
| 133 | mulx $npj,$mul1,$acc1 !prologue! np[1]*"t[0]"*n0
|
|---|
| 134 | srlx $car0,32,$car0
|
|---|
| 135 | add $acc0,$car1,$car1
|
|---|
| 136 | ld [$np+8],$npj !prologue!
|
|---|
| 137 | srlx $car1,32,$car1
|
|---|
| 138 | mov $tmp0,$acc0 !prologue!
|
|---|
| 139 |
|
|---|
| 140 | .L1st:
|
|---|
| 141 | mulx $apj,$mul0,$tmp0
|
|---|
| 142 | mulx $npj,$mul1,$tmp1
|
|---|
| 143 | add $acc0,$car0,$car0
|
|---|
| 144 | ld [$ap+$j],$apj ! ap[j]
|
|---|
| 145 | and $car0,$mask,$acc0
|
|---|
| 146 | add $acc1,$car1,$car1
|
|---|
| 147 | ld [$np+$j],$npj ! np[j]
|
|---|
| 148 | srlx $car0,32,$car0
|
|---|
| 149 | add $acc0,$car1,$car1
|
|---|
| 150 | add $j,4,$j ! j++
|
|---|
| 151 | mov $tmp0,$acc0
|
|---|
| 152 | st $car1,[$tp]
|
|---|
| 153 | cmp $j,$num
|
|---|
| 154 | mov $tmp1,$acc1
|
|---|
| 155 | srlx $car1,32,$car1
|
|---|
| 156 | bl %icc,.L1st
|
|---|
| 157 | add $tp,4,$tp ! tp++
|
|---|
| 158 | !.L1st
|
|---|
| 159 |
|
|---|
| 160 | mulx $apj,$mul0,$tmp0 !epilogue!
|
|---|
| 161 | mulx $npj,$mul1,$tmp1
|
|---|
| 162 | add $acc0,$car0,$car0
|
|---|
| 163 | and $car0,$mask,$acc0
|
|---|
| 164 | add $acc1,$car1,$car1
|
|---|
| 165 | srlx $car0,32,$car0
|
|---|
| 166 | add $acc0,$car1,$car1
|
|---|
| 167 | st $car1,[$tp]
|
|---|
| 168 | srlx $car1,32,$car1
|
|---|
| 169 |
|
|---|
| 170 | add $tmp0,$car0,$car0
|
|---|
| 171 | and $car0,$mask,$acc0
|
|---|
| 172 | add $tmp1,$car1,$car1
|
|---|
| 173 | srlx $car0,32,$car0
|
|---|
| 174 | add $acc0,$car1,$car1
|
|---|
| 175 | st $car1,[$tp+4]
|
|---|
| 176 | srlx $car1,32,$car1
|
|---|
| 177 |
|
|---|
| 178 | add $car0,$car1,$car1
|
|---|
| 179 | st $car1,[$tp+8]
|
|---|
| 180 | srlx $car1,32,$car2
|
|---|
| 181 | |
|---|
| 182 |
|
|---|
| 183 | mov 4,$i ! i++
|
|---|
| 184 | ld [$bp+4],$mul0 ! bp[1]
|
|---|
| 185 | .Louter:
|
|---|
| 186 | add %sp,$bias+$frame,$tp
|
|---|
| 187 | ld [$ap],$car0 ! ap[0]
|
|---|
| 188 | ld [$ap+4],$apj ! ap[1]
|
|---|
| 189 | ld [$np],$car1 ! np[0]
|
|---|
| 190 | ld [$np+4],$npj ! np[1]
|
|---|
| 191 | ld [$tp],$tmp1 ! tp[0]
|
|---|
| 192 | ld [$tp+4],$tpj ! tp[1]
|
|---|
| 193 | mov 12,$j
|
|---|
| 194 |
|
|---|
| 195 | mulx $car0,$mul0,$car0
|
|---|
| 196 | mulx $apj,$mul0,$tmp0 !prologue!
|
|---|
| 197 | add $tmp1,$car0,$car0
|
|---|
| 198 | ld [$ap+8],$apj !prologue!
|
|---|
| 199 | and $car0,$mask,$acc0
|
|---|
| 200 |
|
|---|
| 201 | mulx $n0,$acc0,$mul1
|
|---|
| 202 | and $mul1,$mask,$mul1
|
|---|
| 203 |
|
|---|
| 204 | mulx $car1,$mul1,$car1
|
|---|
| 205 | mulx $npj,$mul1,$acc1 !prologue!
|
|---|
| 206 | srlx $car0,32,$car0
|
|---|
| 207 | add $acc0,$car1,$car1
|
|---|
| 208 | ld [$np+8],$npj !prologue!
|
|---|
| 209 | srlx $car1,32,$car1
|
|---|
| 210 | mov $tmp0,$acc0 !prologue!
|
|---|
| 211 |
|
|---|
| 212 | .Linner:
|
|---|
| 213 | mulx $apj,$mul0,$tmp0
|
|---|
| 214 | mulx $npj,$mul1,$tmp1
|
|---|
| 215 | add $tpj,$car0,$car0
|
|---|
| 216 | ld [$ap+$j],$apj ! ap[j]
|
|---|
| 217 | add $acc0,$car0,$car0
|
|---|
| 218 | add $acc1,$car1,$car1
|
|---|
| 219 | ld [$np+$j],$npj ! np[j]
|
|---|
| 220 | and $car0,$mask,$acc0
|
|---|
| 221 | ld [$tp+8],$tpj ! tp[j]
|
|---|
| 222 | srlx $car0,32,$car0
|
|---|
| 223 | add $acc0,$car1,$car1
|
|---|
| 224 | add $j,4,$j ! j++
|
|---|
| 225 | mov $tmp0,$acc0
|
|---|
| 226 | st $car1,[$tp] ! tp[j-1]
|
|---|
| 227 | srlx $car1,32,$car1
|
|---|
| 228 | mov $tmp1,$acc1
|
|---|
| 229 | cmp $j,$num
|
|---|
| 230 | bl %icc,.Linner
|
|---|
| 231 | add $tp,4,$tp ! tp++
|
|---|
| 232 | !.Linner
|
|---|
| 233 |
|
|---|
| 234 | mulx $apj,$mul0,$tmp0 !epilogue!
|
|---|
| 235 | mulx $npj,$mul1,$tmp1
|
|---|
| 236 | add $tpj,$car0,$car0
|
|---|
| 237 | add $acc0,$car0,$car0
|
|---|
| 238 | ld [$tp+8],$tpj ! tp[j]
|
|---|
| 239 | and $car0,$mask,$acc0
|
|---|
| 240 | add $acc1,$car1,$car1
|
|---|
| 241 | srlx $car0,32,$car0
|
|---|
| 242 | add $acc0,$car1,$car1
|
|---|
| 243 | st $car1,[$tp] ! tp[j-1]
|
|---|
| 244 | srlx $car1,32,$car1
|
|---|
| 245 |
|
|---|
| 246 | add $tpj,$car0,$car0
|
|---|
| 247 | add $tmp0,$car0,$car0
|
|---|
| 248 | and $car0,$mask,$acc0
|
|---|
| 249 | add $tmp1,$car1,$car1
|
|---|
| 250 | add $acc0,$car1,$car1
|
|---|
| 251 | st $car1,[$tp+4] ! tp[j-1]
|
|---|
| 252 | srlx $car0,32,$car0
|
|---|
| 253 | add $i,4,$i ! i++
|
|---|
| 254 | srlx $car1,32,$car1
|
|---|
| 255 |
|
|---|
| 256 | add $car0,$car1,$car1
|
|---|
| 257 | cmp $i,$num
|
|---|
| 258 | add $car2,$car1,$car1
|
|---|
| 259 | st $car1,[$tp+8]
|
|---|
| 260 |
|
|---|
| 261 | srlx $car1,32,$car2
|
|---|
| 262 | bl,a %icc,.Louter
|
|---|
| 263 | ld [$bp+$i],$mul0 ! bp[i]
|
|---|
| 264 | !.Louter
|
|---|
| 265 |
|
|---|
| 266 | add $tp,12,$tp
|
|---|
| 267 | |
|---|
| 268 |
|
|---|
| 269 | .Ltail:
|
|---|
| 270 | add $np,$num,$np
|
|---|
| 271 | add $rp,$num,$rp
|
|---|
| 272 | sub %g0,$num,%o7 ! k=-num
|
|---|
| 273 | ba .Lsub
|
|---|
| 274 | subcc %g0,%g0,%g0 ! clear %icc.c
|
|---|
| 275 | .align 16
|
|---|
| 276 | .Lsub:
|
|---|
| 277 | ld [$tp+%o7],%o0
|
|---|
| 278 | ld [$np+%o7],%o1
|
|---|
| 279 | subccc %o0,%o1,%o1 ! tp[j]-np[j]
|
|---|
| 280 | add $rp,%o7,$i
|
|---|
| 281 | add %o7,4,%o7
|
|---|
| 282 | brnz %o7,.Lsub
|
|---|
| 283 | st %o1,[$i]
|
|---|
| 284 | subccc $car2,0,$car2 ! handle upmost overflow bit
|
|---|
| 285 | sub %g0,$num,%o7
|
|---|
| 286 |
|
|---|
| 287 | .Lcopy:
|
|---|
| 288 | ld [$tp+%o7],%o1 ! conditional copy
|
|---|
| 289 | ld [$rp+%o7],%o0
|
|---|
| 290 | st %g0,[$tp+%o7] ! zap tp
|
|---|
| 291 | movcs %icc,%o1,%o0
|
|---|
| 292 | st %o0,[$rp+%o7]
|
|---|
| 293 | add %o7,4,%o7
|
|---|
| 294 | brnz %o7,.Lcopy
|
|---|
| 295 | nop
|
|---|
| 296 | mov 1,%i0
|
|---|
| 297 | ret
|
|---|
| 298 | restore
|
|---|
| 299 | ___
|
|---|
| 300 | |
|---|
| 301 |
|
|---|
| 302 | ########
|
|---|
| 303 | ######## .Lbn_sqr_mont gives up to 20% *overall* improvement over
|
|---|
| 304 | ######## code without following dedicated squaring procedure.
|
|---|
| 305 | ########
|
|---|
| 306 | $sbit="%o5";
|
|---|
| 307 |
|
|---|
| 308 | $code.=<<___;
|
|---|
| 309 | .align 32
|
|---|
| 310 | .Lbn_sqr_mont:
|
|---|
| 311 | mulx $mul0,$mul0,$car0 ! ap[0]*ap[0]
|
|---|
| 312 | mulx $apj,$mul0,$tmp0 !prologue!
|
|---|
| 313 | and $car0,$mask,$acc0
|
|---|
| 314 | add %sp,$bias+$frame,$tp
|
|---|
| 315 | ld [$ap+8],$apj !prologue!
|
|---|
| 316 |
|
|---|
| 317 | mulx $n0,$acc0,$mul1 ! "t[0]"*n0
|
|---|
| 318 | srlx $car0,32,$car0
|
|---|
| 319 | and $mul1,$mask,$mul1
|
|---|
| 320 |
|
|---|
| 321 | mulx $car1,$mul1,$car1 ! np[0]*"t[0]"*n0
|
|---|
| 322 | mulx $npj,$mul1,$acc1 !prologue!
|
|---|
| 323 | and $car0,1,$sbit
|
|---|
| 324 | ld [$np+8],$npj !prologue!
|
|---|
| 325 | srlx $car0,1,$car0
|
|---|
| 326 | add $acc0,$car1,$car1
|
|---|
| 327 | srlx $car1,32,$car1
|
|---|
| 328 | mov $tmp0,$acc0 !prologue!
|
|---|
| 329 |
|
|---|
| 330 | .Lsqr_1st:
|
|---|
| 331 | mulx $apj,$mul0,$tmp0
|
|---|
| 332 | mulx $npj,$mul1,$tmp1
|
|---|
| 333 | add $acc0,$car0,$car0 ! ap[j]*a0+c0
|
|---|
| 334 | add $acc1,$car1,$car1
|
|---|
| 335 | ld [$ap+$j],$apj ! ap[j]
|
|---|
| 336 | and $car0,$mask,$acc0
|
|---|
| 337 | ld [$np+$j],$npj ! np[j]
|
|---|
| 338 | srlx $car0,32,$car0
|
|---|
| 339 | add $acc0,$acc0,$acc0
|
|---|
| 340 | or $sbit,$acc0,$acc0
|
|---|
| 341 | mov $tmp1,$acc1
|
|---|
| 342 | srlx $acc0,32,$sbit
|
|---|
| 343 | add $j,4,$j ! j++
|
|---|
| 344 | and $acc0,$mask,$acc0
|
|---|
| 345 | cmp $j,$num
|
|---|
| 346 | add $acc0,$car1,$car1
|
|---|
| 347 | st $car1,[$tp]
|
|---|
| 348 | mov $tmp0,$acc0
|
|---|
| 349 | srlx $car1,32,$car1
|
|---|
| 350 | bl %icc,.Lsqr_1st
|
|---|
| 351 | add $tp,4,$tp ! tp++
|
|---|
| 352 | !.Lsqr_1st
|
|---|
| 353 |
|
|---|
| 354 | mulx $apj,$mul0,$tmp0 ! epilogue
|
|---|
| 355 | mulx $npj,$mul1,$tmp1
|
|---|
| 356 | add $acc0,$car0,$car0 ! ap[j]*a0+c0
|
|---|
| 357 | add $acc1,$car1,$car1
|
|---|
| 358 | and $car0,$mask,$acc0
|
|---|
| 359 | srlx $car0,32,$car0
|
|---|
| 360 | add $acc0,$acc0,$acc0
|
|---|
| 361 | or $sbit,$acc0,$acc0
|
|---|
| 362 | srlx $acc0,32,$sbit
|
|---|
| 363 | and $acc0,$mask,$acc0
|
|---|
| 364 | add $acc0,$car1,$car1
|
|---|
| 365 | st $car1,[$tp]
|
|---|
| 366 | srlx $car1,32,$car1
|
|---|
| 367 |
|
|---|
| 368 | add $tmp0,$car0,$car0 ! ap[j]*a0+c0
|
|---|
| 369 | add $tmp1,$car1,$car1
|
|---|
| 370 | and $car0,$mask,$acc0
|
|---|
| 371 | srlx $car0,32,$car0
|
|---|
| 372 | add $acc0,$acc0,$acc0
|
|---|
| 373 | or $sbit,$acc0,$acc0
|
|---|
| 374 | srlx $acc0,32,$sbit
|
|---|
| 375 | and $acc0,$mask,$acc0
|
|---|
| 376 | add $acc0,$car1,$car1
|
|---|
| 377 | st $car1,[$tp+4]
|
|---|
| 378 | srlx $car1,32,$car1
|
|---|
| 379 |
|
|---|
| 380 | add $car0,$car0,$car0
|
|---|
| 381 | or $sbit,$car0,$car0
|
|---|
| 382 | add $car0,$car1,$car1
|
|---|
| 383 | st $car1,[$tp+8]
|
|---|
| 384 | srlx $car1,32,$car2
|
|---|
| 385 | |
|---|
| 386 |
|
|---|
| 387 | ld [%sp+$bias+$frame],$tmp0 ! tp[0]
|
|---|
| 388 | ld [%sp+$bias+$frame+4],$tmp1 ! tp[1]
|
|---|
| 389 | ld [%sp+$bias+$frame+8],$tpj ! tp[2]
|
|---|
| 390 | ld [$ap+4],$mul0 ! ap[1]
|
|---|
| 391 | ld [$ap+8],$apj ! ap[2]
|
|---|
| 392 | ld [$np],$car1 ! np[0]
|
|---|
| 393 | ld [$np+4],$npj ! np[1]
|
|---|
| 394 | mulx $n0,$tmp0,$mul1
|
|---|
| 395 |
|
|---|
| 396 | mulx $mul0,$mul0,$car0
|
|---|
| 397 | and $mul1,$mask,$mul1
|
|---|
| 398 |
|
|---|
| 399 | mulx $car1,$mul1,$car1
|
|---|
| 400 | mulx $npj,$mul1,$acc1
|
|---|
| 401 | add $tmp0,$car1,$car1
|
|---|
| 402 | and $car0,$mask,$acc0
|
|---|
| 403 | ld [$np+8],$npj ! np[2]
|
|---|
| 404 | srlx $car1,32,$car1
|
|---|
| 405 | add $tmp1,$car1,$car1
|
|---|
| 406 | srlx $car0,32,$car0
|
|---|
| 407 | add $acc0,$car1,$car1
|
|---|
| 408 | and $car0,1,$sbit
|
|---|
| 409 | add $acc1,$car1,$car1
|
|---|
| 410 | srlx $car0,1,$car0
|
|---|
| 411 | mov 12,$j
|
|---|
| 412 | st $car1,[%sp+$bias+$frame] ! tp[0]=
|
|---|
| 413 | srlx $car1,32,$car1
|
|---|
| 414 | add %sp,$bias+$frame+4,$tp
|
|---|
| 415 |
|
|---|
| 416 | .Lsqr_2nd:
|
|---|
| 417 | mulx $apj,$mul0,$acc0
|
|---|
| 418 | mulx $npj,$mul1,$acc1
|
|---|
| 419 | add $acc0,$car0,$car0
|
|---|
| 420 | add $tpj,$sbit,$sbit
|
|---|
| 421 | ld [$ap+$j],$apj ! ap[j]
|
|---|
| 422 | and $car0,$mask,$acc0
|
|---|
| 423 | ld [$np+$j],$npj ! np[j]
|
|---|
| 424 | srlx $car0,32,$car0
|
|---|
| 425 | add $acc1,$car1,$car1
|
|---|
| 426 | ld [$tp+8],$tpj ! tp[j]
|
|---|
| 427 | add $acc0,$acc0,$acc0
|
|---|
| 428 | add $j,4,$j ! j++
|
|---|
| 429 | add $sbit,$acc0,$acc0
|
|---|
| 430 | srlx $acc0,32,$sbit
|
|---|
| 431 | and $acc0,$mask,$acc0
|
|---|
| 432 | cmp $j,$num
|
|---|
| 433 | add $acc0,$car1,$car1
|
|---|
| 434 | st $car1,[$tp] ! tp[j-1]
|
|---|
| 435 | srlx $car1,32,$car1
|
|---|
| 436 | bl %icc,.Lsqr_2nd
|
|---|
| 437 | add $tp,4,$tp ! tp++
|
|---|
| 438 | !.Lsqr_2nd
|
|---|
| 439 |
|
|---|
| 440 | mulx $apj,$mul0,$acc0
|
|---|
| 441 | mulx $npj,$mul1,$acc1
|
|---|
| 442 | add $acc0,$car0,$car0
|
|---|
| 443 | add $tpj,$sbit,$sbit
|
|---|
| 444 | and $car0,$mask,$acc0
|
|---|
| 445 | srlx $car0,32,$car0
|
|---|
| 446 | add $acc1,$car1,$car1
|
|---|
| 447 | add $acc0,$acc0,$acc0
|
|---|
| 448 | add $sbit,$acc0,$acc0
|
|---|
| 449 | srlx $acc0,32,$sbit
|
|---|
| 450 | and $acc0,$mask,$acc0
|
|---|
| 451 | add $acc0,$car1,$car1
|
|---|
| 452 | st $car1,[$tp] ! tp[j-1]
|
|---|
| 453 | srlx $car1,32,$car1
|
|---|
| 454 |
|
|---|
| 455 | add $car0,$car0,$car0
|
|---|
| 456 | add $sbit,$car0,$car0
|
|---|
| 457 | add $car0,$car1,$car1
|
|---|
| 458 | add $car2,$car1,$car1
|
|---|
| 459 | st $car1,[$tp+4]
|
|---|
| 460 | srlx $car1,32,$car2
|
|---|
| 461 | |
|---|
| 462 |
|
|---|
| 463 | ld [%sp+$bias+$frame],$tmp1 ! tp[0]
|
|---|
| 464 | ld [%sp+$bias+$frame+4],$tpj ! tp[1]
|
|---|
| 465 | ld [$ap+8],$mul0 ! ap[2]
|
|---|
| 466 | ld [$np],$car1 ! np[0]
|
|---|
| 467 | ld [$np+4],$npj ! np[1]
|
|---|
| 468 | mulx $n0,$tmp1,$mul1
|
|---|
| 469 | and $mul1,$mask,$mul1
|
|---|
| 470 | mov 8,$i
|
|---|
| 471 |
|
|---|
| 472 | mulx $mul0,$mul0,$car0
|
|---|
| 473 | mulx $car1,$mul1,$car1
|
|---|
| 474 | and $car0,$mask,$acc0
|
|---|
| 475 | add $tmp1,$car1,$car1
|
|---|
| 476 | srlx $car0,32,$car0
|
|---|
| 477 | add %sp,$bias+$frame,$tp
|
|---|
| 478 | srlx $car1,32,$car1
|
|---|
| 479 | and $car0,1,$sbit
|
|---|
| 480 | srlx $car0,1,$car0
|
|---|
| 481 | mov 4,$j
|
|---|
| 482 |
|
|---|
| 483 | .Lsqr_outer:
|
|---|
| 484 | .Lsqr_inner1:
|
|---|
| 485 | mulx $npj,$mul1,$acc1
|
|---|
| 486 | add $tpj,$car1,$car1
|
|---|
| 487 | add $j,4,$j
|
|---|
| 488 | ld [$tp+8],$tpj
|
|---|
| 489 | cmp $j,$i
|
|---|
| 490 | add $acc1,$car1,$car1
|
|---|
| 491 | ld [$np+$j],$npj
|
|---|
| 492 | st $car1,[$tp]
|
|---|
| 493 | srlx $car1,32,$car1
|
|---|
| 494 | bl %icc,.Lsqr_inner1
|
|---|
| 495 | add $tp,4,$tp
|
|---|
| 496 | !.Lsqr_inner1
|
|---|
| 497 |
|
|---|
| 498 | add $j,4,$j
|
|---|
| 499 | ld [$ap+$j],$apj ! ap[j]
|
|---|
| 500 | mulx $npj,$mul1,$acc1
|
|---|
| 501 | add $tpj,$car1,$car1
|
|---|
| 502 | ld [$np+$j],$npj ! np[j]
|
|---|
| 503 | srlx $car1,32,$tmp0
|
|---|
| 504 | and $car1,$mask,$car1
|
|---|
| 505 | add $tmp0,$sbit,$sbit
|
|---|
| 506 | add $acc0,$car1,$car1
|
|---|
| 507 | ld [$tp+8],$tpj ! tp[j]
|
|---|
| 508 | add $acc1,$car1,$car1
|
|---|
| 509 | st $car1,[$tp]
|
|---|
| 510 | srlx $car1,32,$car1
|
|---|
| 511 |
|
|---|
| 512 | add $j,4,$j
|
|---|
| 513 | cmp $j,$num
|
|---|
| 514 | be,pn %icc,.Lsqr_no_inner2
|
|---|
| 515 | add $tp,4,$tp
|
|---|
| 516 |
|
|---|
| 517 | .Lsqr_inner2:
|
|---|
| 518 | mulx $apj,$mul0,$acc0
|
|---|
| 519 | mulx $npj,$mul1,$acc1
|
|---|
| 520 | add $tpj,$sbit,$sbit
|
|---|
| 521 | add $acc0,$car0,$car0
|
|---|
| 522 | ld [$ap+$j],$apj ! ap[j]
|
|---|
| 523 | and $car0,$mask,$acc0
|
|---|
| 524 | ld [$np+$j],$npj ! np[j]
|
|---|
| 525 | srlx $car0,32,$car0
|
|---|
| 526 | add $acc0,$acc0,$acc0
|
|---|
| 527 | ld [$tp+8],$tpj ! tp[j]
|
|---|
| 528 | add $sbit,$acc0,$acc0
|
|---|
| 529 | add $j,4,$j ! j++
|
|---|
| 530 | srlx $acc0,32,$sbit
|
|---|
| 531 | and $acc0,$mask,$acc0
|
|---|
| 532 | cmp $j,$num
|
|---|
| 533 | add $acc0,$car1,$car1
|
|---|
| 534 | add $acc1,$car1,$car1
|
|---|
| 535 | st $car1,[$tp] ! tp[j-1]
|
|---|
| 536 | srlx $car1,32,$car1
|
|---|
| 537 | bl %icc,.Lsqr_inner2
|
|---|
| 538 | add $tp,4,$tp ! tp++
|
|---|
| 539 |
|
|---|
| 540 | .Lsqr_no_inner2:
|
|---|
| 541 | mulx $apj,$mul0,$acc0
|
|---|
| 542 | mulx $npj,$mul1,$acc1
|
|---|
| 543 | add $tpj,$sbit,$sbit
|
|---|
| 544 | add $acc0,$car0,$car0
|
|---|
| 545 | and $car0,$mask,$acc0
|
|---|
| 546 | srlx $car0,32,$car0
|
|---|
| 547 | add $acc0,$acc0,$acc0
|
|---|
| 548 | add $sbit,$acc0,$acc0
|
|---|
| 549 | srlx $acc0,32,$sbit
|
|---|
| 550 | and $acc0,$mask,$acc0
|
|---|
| 551 | add $acc0,$car1,$car1
|
|---|
| 552 | add $acc1,$car1,$car1
|
|---|
| 553 | st $car1,[$tp] ! tp[j-1]
|
|---|
| 554 | srlx $car1,32,$car1
|
|---|
| 555 |
|
|---|
| 556 | add $car0,$car0,$car0
|
|---|
| 557 | add $sbit,$car0,$car0
|
|---|
| 558 | add $car0,$car1,$car1
|
|---|
| 559 | add $car2,$car1,$car1
|
|---|
| 560 | st $car1,[$tp+4]
|
|---|
| 561 | srlx $car1,32,$car2
|
|---|
| 562 | |
|---|
| 563 |
|
|---|
| 564 | add $i,4,$i ! i++
|
|---|
| 565 | ld [%sp+$bias+$frame],$tmp1 ! tp[0]
|
|---|
| 566 | ld [%sp+$bias+$frame+4],$tpj ! tp[1]
|
|---|
| 567 | ld [$ap+$i],$mul0 ! ap[j]
|
|---|
| 568 | ld [$np],$car1 ! np[0]
|
|---|
| 569 | ld [$np+4],$npj ! np[1]
|
|---|
| 570 | mulx $n0,$tmp1,$mul1
|
|---|
| 571 | and $mul1,$mask,$mul1
|
|---|
| 572 | add $i,4,$tmp0
|
|---|
| 573 |
|
|---|
| 574 | mulx $mul0,$mul0,$car0
|
|---|
| 575 | mulx $car1,$mul1,$car1
|
|---|
| 576 | and $car0,$mask,$acc0
|
|---|
| 577 | add $tmp1,$car1,$car1
|
|---|
| 578 | srlx $car0,32,$car0
|
|---|
| 579 | add %sp,$bias+$frame,$tp
|
|---|
| 580 | srlx $car1,32,$car1
|
|---|
| 581 | and $car0,1,$sbit
|
|---|
| 582 | srlx $car0,1,$car0
|
|---|
| 583 |
|
|---|
| 584 | cmp $tmp0,$num ! i<num-1
|
|---|
| 585 | bl %icc,.Lsqr_outer
|
|---|
| 586 | mov 4,$j
|
|---|
| 587 | |
|---|
| 588 |
|
|---|
| 589 | .Lsqr_last:
|
|---|
| 590 | mulx $npj,$mul1,$acc1
|
|---|
| 591 | add $tpj,$car1,$car1
|
|---|
| 592 | add $j,4,$j
|
|---|
| 593 | ld [$tp+8],$tpj
|
|---|
| 594 | cmp $j,$i
|
|---|
| 595 | add $acc1,$car1,$car1
|
|---|
| 596 | ld [$np+$j],$npj
|
|---|
| 597 | st $car1,[$tp]
|
|---|
| 598 | srlx $car1,32,$car1
|
|---|
| 599 | bl %icc,.Lsqr_last
|
|---|
| 600 | add $tp,4,$tp
|
|---|
| 601 | !.Lsqr_last
|
|---|
| 602 |
|
|---|
| 603 | mulx $npj,$mul1,$acc1
|
|---|
| 604 | add $tpj,$acc0,$acc0
|
|---|
| 605 | srlx $acc0,32,$tmp0
|
|---|
| 606 | and $acc0,$mask,$acc0
|
|---|
| 607 | add $tmp0,$sbit,$sbit
|
|---|
| 608 | add $acc0,$car1,$car1
|
|---|
| 609 | add $acc1,$car1,$car1
|
|---|
| 610 | st $car1,[$tp]
|
|---|
| 611 | srlx $car1,32,$car1
|
|---|
| 612 |
|
|---|
| 613 | add $car0,$car0,$car0 ! recover $car0
|
|---|
| 614 | add $sbit,$car0,$car0
|
|---|
| 615 | add $car0,$car1,$car1
|
|---|
| 616 | add $car2,$car1,$car1
|
|---|
| 617 | st $car1,[$tp+4]
|
|---|
| 618 | srlx $car1,32,$car2
|
|---|
| 619 |
|
|---|
| 620 | ba .Ltail
|
|---|
| 621 | add $tp,8,$tp
|
|---|
| 622 | .type $fname,#function
|
|---|
| 623 | .size $fname,(.-$fname)
|
|---|
| 624 | .asciz "Montgomery Multiplication for SPARCv9, CRYPTOGAMS by <appro\@openssl.org>"
|
|---|
| 625 | .align 32
|
|---|
| 626 | ___
|
|---|
| 627 | $code =~ s/\`([^\`]*)\`/eval($1)/gem;
|
|---|
| 628 | print $code;
|
|---|
| 629 | close STDOUT or die "error closing STDOUT: $!";
|
|---|
