| 1 |
#!/usr/bin/env perl |
| 2 |
|
| 3 |
# ==================================================================== |
| 4 |
# Written by Andy Polyakov <[email protected]> for the OpenSSL |
| 5 |
# project. The module is, however, dual licensed under OpenSSL and |
| 6 |
# CRYPTOGAMS licenses depending on where you obtain it. For further |
| 7 |
# details see http://www.openssl.org/~appro/cryptogams/. |
| 8 |
# ==================================================================== |
| 9 |
|
| 10 |
# March 2015 |
| 11 |
# |
| 12 |
# "Teaser" Montgomery multiplication module for ARMv8. Needs more |
| 13 |
# work. While it does improve RSA sign performance by 20-30% (less for |
| 14 |
# longer keys) on most processors, for some reason RSA2048 is not |
| 15 |
# faster and RSA4096 goes 15-20% slower on Cortex-A57. Multiplication |
| 16 |
# instruction issue rate is limited on processor in question, meaning |
| 17 |
# that dedicated squaring procedure is a must. Well, actually all |
| 18 |
# contemporary AArch64 processors seem to have limited multiplication |
| 19 |
# issue rate, i.e. they can't issue multiplication every cycle, which |
| 20 |
# explains moderate improvement coefficients in comparison to |
| 21 |
# compiler-generated code. Recall that compiler is instructed to use |
| 22 |
# umulh and therefore uses same amount of multiplication instructions |
| 23 |
# to do the job. Assembly's edge is to minimize number of "collateral" |
| 24 |
# instructions and of course instruction scheduling. |
| 25 |
# |
| 26 |
# April 2015 |
| 27 |
# |
| 28 |
# Squaring procedure that handles lengths divisible by 8 improves |
| 29 |
# RSA/DSA performance by 25-40-60% depending on processor and key |
| 30 |
# length. Overall improvement coefficients are always positive in |
| 31 |
# comparison to compiler-generated code. On Cortex-A57 improvement |
| 32 |
# is still modest on longest key lengths, while others exhibit e.g. |
| 33 |
# 50-70% improvement for RSA4096 sign. RSA2048 sign is ~25% faster |
| 34 |
# on Cortex-A57 and ~60-100% faster on others. |
| 35 |
|
| 36 |
$flavour = shift; |
| 37 |
$output = shift; |
| 38 |
|
| 39 |
$0 =~ m/(.*[\/\\])[^\/\\]+$/; $dir=$1; |
| 40 |
( $xlate="${dir}arm-xlate.pl" and -f $xlate ) or |
| 41 |
( $xlate="${dir}../../perlasm/arm-xlate.pl" and -f $xlate) or |
| 42 |
die "can't locate arm-xlate.pl"; |
| 43 |
|
| 44 |
open OUT,"| \"$^X\" $xlate $flavour $output"; |
| 45 |
*STDOUT=*OUT; |
| 46 |
|
| 47 |
($lo0,$hi0,$aj,$m0,$alo,$ahi, |
| 48 |
$lo1,$hi1,$nj,$m1,$nlo,$nhi, |
| 49 |
$ovf, $i,$j,$tp,$tj) = map("x$_",6..17,19..24); |
| 50 |
|
| 51 |
# int bn_mul_mont( |
| 52 |
$rp="x0"; # BN_ULONG *rp, |
| 53 |
$ap="x1"; # const BN_ULONG *ap, |
| 54 |
$bp="x2"; # const BN_ULONG *bp, |
| 55 |
$np="x3"; # const BN_ULONG *np, |
| 56 |
$n0="x4"; # const BN_ULONG *n0, |
| 57 |
$num="x5"; # int num); |
| 58 |
|
| 59 |
$code.=<<___; |
| 60 |
.text |
| 61 |
|
| 62 |
.globl bn_mul_mont |
| 63 |
.type bn_mul_mont,%function |
| 64 |
.align 5 |
| 65 |
bn_mul_mont: |
| 66 |
tst $num,#7 |
| 67 |
b.eq __bn_sqr8x_mont |
| 68 |
tst $num,#3 |
| 69 |
b.eq __bn_mul4x_mont |
| 70 |
.Lmul_mont: |
| 71 |
stp x29,x30,[sp,#-64]! |
| 72 |
add x29,sp,#0 |
| 73 |
stp x19,x20,[sp,#16] |
| 74 |
stp x21,x22,[sp,#32] |
| 75 |
stp x23,x24,[sp,#48] |
| 76 |
|
| 77 |
ldr $m0,[$bp],#8 // bp[0] |
| 78 |
sub $tp,sp,$num,lsl#3 |
| 79 |
ldp $hi0,$aj,[$ap],#16 // ap[0..1] |
| 80 |
lsl $num,$num,#3 |
| 81 |
ldr $n0,[$n0] // *n0 |
| 82 |
and $tp,$tp,#-16 // ABI says so |
| 83 |
ldp $hi1,$nj,[$np],#16 // np[0..1] |
| 84 |
|
| 85 |
mul $lo0,$hi0,$m0 // ap[0]*bp[0] |
| 86 |
sub $j,$num,#16 // j=num-2 |
| 87 |
umulh $hi0,$hi0,$m0 |
| 88 |
mul $alo,$aj,$m0 // ap[1]*bp[0] |
| 89 |
umulh $ahi,$aj,$m0 |
| 90 |
|
| 91 |
mul $m1,$lo0,$n0 // "tp[0]"*n0 |
| 92 |
mov sp,$tp // alloca |
| 93 |
|
| 94 |
// (*) mul $lo1,$hi1,$m1 // np[0]*m1 |
| 95 |
umulh $hi1,$hi1,$m1 |
| 96 |
mul $nlo,$nj,$m1 // np[1]*m1 |
| 97 |
// (*) adds $lo1,$lo1,$lo0 // discarded |
| 98 |
// (*) As for removal of first multiplication and addition |
| 99 |
// instructions. The outcome of first addition is |
| 100 |
// guaranteed to be zero, which leaves two computationally |
| 101 |
// significant outcomes: it either carries or not. Then |
| 102 |
// question is when does it carry? Is there alternative |
| 103 |
// way to deduce it? If you follow operations, you can |
| 104 |
// observe that condition for carry is quite simple: |
| 105 |
// $lo0 being non-zero. So that carry can be calculated |
| 106 |
// by adding -1 to $lo0. That's what next instruction does. |
| 107 |
subs xzr,$lo0,#1 // (*) |
| 108 |
umulh $nhi,$nj,$m1 |
| 109 |
adc $hi1,$hi1,xzr |
| 110 |
cbz $j,.L1st_skip |
| 111 |
|
| 112 |
.L1st: |
| 113 |
ldr $aj,[$ap],#8 |
| 114 |
adds $lo0,$alo,$hi0 |
| 115 |
sub $j,$j,#8 // j-- |
| 116 |
adc $hi0,$ahi,xzr |
| 117 |
|
| 118 |
ldr $nj,[$np],#8 |
| 119 |
adds $lo1,$nlo,$hi1 |
| 120 |
mul $alo,$aj,$m0 // ap[j]*bp[0] |
| 121 |
adc $hi1,$nhi,xzr |
| 122 |
umulh $ahi,$aj,$m0 |
| 123 |
|
| 124 |
adds $lo1,$lo1,$lo0 |
| 125 |
mul $nlo,$nj,$m1 // np[j]*m1 |
| 126 |
adc $hi1,$hi1,xzr |
| 127 |
umulh $nhi,$nj,$m1 |
| 128 |
str $lo1,[$tp],#8 // tp[j-1] |
| 129 |
cbnz $j,.L1st |
| 130 |
|
| 131 |
.L1st_skip: |
| 132 |
adds $lo0,$alo,$hi0 |
| 133 |
sub $ap,$ap,$num // rewind $ap |
| 134 |
adc $hi0,$ahi,xzr |
| 135 |
|
| 136 |
adds $lo1,$nlo,$hi1 |
| 137 |
sub $np,$np,$num // rewind $np |
| 138 |
adc $hi1,$nhi,xzr |
| 139 |
|
| 140 |
adds $lo1,$lo1,$lo0 |
| 141 |
sub $i,$num,#8 // i=num-1 |
| 142 |
adcs $hi1,$hi1,$hi0 |
| 143 |
|
| 144 |
adc $ovf,xzr,xzr // upmost overflow bit |
| 145 |
stp $lo1,$hi1,[$tp] |
| 146 |
|
| 147 |
.Louter: |
| 148 |
ldr $m0,[$bp],#8 // bp[i] |
| 149 |
ldp $hi0,$aj,[$ap],#16 |
| 150 |
ldr $tj,[sp] // tp[0] |
| 151 |
add $tp,sp,#8 |
| 152 |
|
| 153 |
mul $lo0,$hi0,$m0 // ap[0]*bp[i] |
| 154 |
sub $j,$num,#16 // j=num-2 |
| 155 |
umulh $hi0,$hi0,$m0 |
| 156 |
ldp $hi1,$nj,[$np],#16 |
| 157 |
mul $alo,$aj,$m0 // ap[1]*bp[i] |
| 158 |
adds $lo0,$lo0,$tj |
| 159 |
umulh $ahi,$aj,$m0 |
| 160 |
adc $hi0,$hi0,xzr |
| 161 |
|
| 162 |
mul $m1,$lo0,$n0 |
| 163 |
sub $i,$i,#8 // i-- |
| 164 |
|
| 165 |
// (*) mul $lo1,$hi1,$m1 // np[0]*m1 |
| 166 |
umulh $hi1,$hi1,$m1 |
| 167 |
mul $nlo,$nj,$m1 // np[1]*m1 |
| 168 |
// (*) adds $lo1,$lo1,$lo0 |
| 169 |
subs xzr,$lo0,#1 // (*) |
| 170 |
umulh $nhi,$nj,$m1 |
| 171 |
cbz $j,.Linner_skip |
| 172 |
|
| 173 |
.Linner: |
| 174 |
ldr $aj,[$ap],#8 |
| 175 |
adc $hi1,$hi1,xzr |
| 176 |
ldr $tj,[$tp],#8 // tp[j] |
| 177 |
adds $lo0,$alo,$hi0 |
| 178 |
sub $j,$j,#8 // j-- |
| 179 |
adc $hi0,$ahi,xzr |
| 180 |
|
| 181 |
adds $lo1,$nlo,$hi1 |
| 182 |
ldr $nj,[$np],#8 |
| 183 |
adc $hi1,$nhi,xzr |
| 184 |
|
| 185 |
mul $alo,$aj,$m0 // ap[j]*bp[i] |
| 186 |
adds $lo0,$lo0,$tj |
| 187 |
umulh $ahi,$aj,$m0 |
| 188 |
adc $hi0,$hi0,xzr |
| 189 |
|
| 190 |
mul $nlo,$nj,$m1 // np[j]*m1 |
| 191 |
adds $lo1,$lo1,$lo0 |
| 192 |
umulh $nhi,$nj,$m1 |
| 193 |
str $lo1,[$tp,#-16] // tp[j-1] |
| 194 |
cbnz $j,.Linner |
| 195 |
|
| 196 |
.Linner_skip: |
| 197 |
ldr $tj,[$tp],#8 // tp[j] |
| 198 |
adc $hi1,$hi1,xzr |
| 199 |
adds $lo0,$alo,$hi0 |
| 200 |
sub $ap,$ap,$num // rewind $ap |
| 201 |
adc $hi0,$ahi,xzr |
| 202 |
|
| 203 |
adds $lo1,$nlo,$hi1 |
| 204 |
sub $np,$np,$num // rewind $np |
| 205 |
adcs $hi1,$nhi,$ovf |
| 206 |
adc $ovf,xzr,xzr |
| 207 |
|
| 208 |
adds $lo0,$lo0,$tj |
| 209 |
adc $hi0,$hi0,xzr |
| 210 |
|
| 211 |
adds $lo1,$lo1,$lo0 |
| 212 |
adcs $hi1,$hi1,$hi0 |
| 213 |
adc $ovf,$ovf,xzr // upmost overflow bit |
| 214 |
stp $lo1,$hi1,[$tp,#-16] |
| 215 |
|
| 216 |
cbnz $i,.Louter |
| 217 |
|
| 218 |
// Final step. We see if result is larger than modulus, and |
| 219 |
// if it is, subtract the modulus. But comparison implies |
| 220 |
// subtraction. So we subtract modulus, see if it borrowed, |
| 221 |
// and conditionally copy original value. |
| 222 |
ldr $tj,[sp] // tp[0] |
| 223 |
add $tp,sp,#8 |
| 224 |
ldr $nj,[$np],#8 // np[0] |
| 225 |
subs $j,$num,#8 // j=num-1 and clear borrow |
| 226 |
mov $ap,$rp |
| 227 |
.Lsub: |
| 228 |
sbcs $aj,$tj,$nj // tp[j]-np[j] |
| 229 |
ldr $tj,[$tp],#8 |
| 230 |
sub $j,$j,#8 // j-- |
| 231 |
ldr $nj,[$np],#8 |
| 232 |
str $aj,[$ap],#8 // rp[j]=tp[j]-np[j] |
| 233 |
cbnz $j,.Lsub |
| 234 |
|
| 235 |
sbcs $aj,$tj,$nj |
| 236 |
sbcs $ovf,$ovf,xzr // did it borrow? |
| 237 |
str $aj,[$ap],#8 // rp[num-1] |
| 238 |
|
| 239 |
ldr $tj,[sp] // tp[0] |
| 240 |
add $tp,sp,#8 |
| 241 |
ldr $aj,[$rp],#8 // rp[0] |
| 242 |
sub $num,$num,#8 // num-- |
| 243 |
nop |
| 244 |
.Lcond_copy: |
| 245 |
sub $num,$num,#8 // num-- |
| 246 |
csel $nj,$tj,$aj,lo // did it borrow? |
| 247 |
ldr $tj,[$tp],#8 |
| 248 |
ldr $aj,[$rp],#8 |
| 249 |
str xzr,[$tp,#-16] // wipe tp |
| 250 |
str $nj,[$rp,#-16] |
| 251 |
cbnz $num,.Lcond_copy |
| 252 |
|
| 253 |
csel $nj,$tj,$aj,lo |
| 254 |
str xzr,[$tp,#-8] // wipe tp |
| 255 |
str $nj,[$rp,#-8] |
| 256 |
|
| 257 |
ldp x19,x20,[x29,#16] |
| 258 |
mov sp,x29 |
| 259 |
ldp x21,x22,[x29,#32] |
| 260 |
mov x0,#1 |
| 261 |
ldp x23,x24,[x29,#48] |
| 262 |
ldr x29,[sp],#64 |
| 263 |
ret |
| 264 |
.size bn_mul_mont,.-bn_mul_mont |
| 265 |
___ |
| 266 |
{ |
| 267 |
######################################################################## |
| 268 |
# Following is ARMv8 adaptation of sqrx8x_mont from x86_64-mont5 module. |
| 269 |
|
| 270 |
my ($a0,$a1,$a2,$a3,$a4,$a5,$a6,$a7)=map("x$_",(6..13)); |
| 271 |
my ($t0,$t1,$t2,$t3)=map("x$_",(14..17)); |
| 272 |
my ($acc0,$acc1,$acc2,$acc3,$acc4,$acc5,$acc6,$acc7)=map("x$_",(19..26)); |
| 273 |
my ($cnt,$carry,$topmost)=("x27","x28","x30"); |
| 274 |
my ($tp,$ap_end,$na0)=($bp,$np,$carry); |
| 275 |
|
| 276 |
$code.=<<___; |
| 277 |
.type __bn_sqr8x_mont,%function |
| 278 |
.align 5 |
| 279 |
__bn_sqr8x_mont: |
| 280 |
cmp $ap,$bp |
| 281 |
b.ne __bn_mul4x_mont |
| 282 |
.Lsqr8x_mont: |
| 283 |
stp x29,x30,[sp,#-128]! |
| 284 |
add x29,sp,#0 |
| 285 |
stp x19,x20,[sp,#16] |
| 286 |
stp x21,x22,[sp,#32] |
| 287 |
stp x23,x24,[sp,#48] |
| 288 |
stp x25,x26,[sp,#64] |
| 289 |
stp x27,x28,[sp,#80] |
| 290 |
stp $rp,$np,[sp,#96] // offload rp and np |
| 291 |
|
| 292 |
ldp $a0,$a1,[$ap,#8*0] |
| 293 |
ldp $a2,$a3,[$ap,#8*2] |
| 294 |
ldp $a4,$a5,[$ap,#8*4] |
| 295 |
ldp $a6,$a7,[$ap,#8*6] |
| 296 |
|
| 297 |
sub $tp,sp,$num,lsl#4 |
| 298 |
lsl $num,$num,#3 |
| 299 |
ldr $n0,[$n0] // *n0 |
| 300 |
mov sp,$tp // alloca |
| 301 |
sub $cnt,$num,#8*8 |
| 302 |
b .Lsqr8x_zero_start |
| 303 |
|
| 304 |
.Lsqr8x_zero: |
| 305 |
sub $cnt,$cnt,#8*8 |
| 306 |
stp xzr,xzr,[$tp,#8*0] |
| 307 |
stp xzr,xzr,[$tp,#8*2] |
| 308 |
stp xzr,xzr,[$tp,#8*4] |
| 309 |
stp xzr,xzr,[$tp,#8*6] |
| 310 |
.Lsqr8x_zero_start: |
| 311 |
stp xzr,xzr,[$tp,#8*8] |
| 312 |
stp xzr,xzr,[$tp,#8*10] |
| 313 |
stp xzr,xzr,[$tp,#8*12] |
| 314 |
stp xzr,xzr,[$tp,#8*14] |
| 315 |
add $tp,$tp,#8*16 |
| 316 |
cbnz $cnt,.Lsqr8x_zero |
| 317 |
|
| 318 |
add $ap_end,$ap,$num |
| 319 |
add $ap,$ap,#8*8 |
| 320 |
mov $acc0,xzr |
| 321 |
mov $acc1,xzr |
| 322 |
mov $acc2,xzr |
| 323 |
mov $acc3,xzr |
| 324 |
mov $acc4,xzr |
| 325 |
mov $acc5,xzr |
| 326 |
mov $acc6,xzr |
| 327 |
mov $acc7,xzr |
| 328 |
mov $tp,sp |
| 329 |
str $n0,[x29,#112] // offload n0 |
| 330 |
|
| 331 |
// Multiply everything but a[i]*a[i] |
| 332 |
.align 4 |
| 333 |
.Lsqr8x_outer_loop: |
| 334 |
// a[1]a[0] (i) |
| 335 |
// a[2]a[0] |
| 336 |
// a[3]a[0] |
| 337 |
// a[4]a[0] |
| 338 |
// a[5]a[0] |
| 339 |
// a[6]a[0] |
| 340 |
// a[7]a[0] |
| 341 |
// a[2]a[1] (ii) |
| 342 |
// a[3]a[1] |
| 343 |
// a[4]a[1] |
| 344 |
// a[5]a[1] |
| 345 |
// a[6]a[1] |
| 346 |
// a[7]a[1] |
| 347 |
// a[3]a[2] (iii) |
| 348 |
// a[4]a[2] |
| 349 |
// a[5]a[2] |
| 350 |
// a[6]a[2] |
| 351 |
// a[7]a[2] |
| 352 |
// a[4]a[3] (iv) |
| 353 |
// a[5]a[3] |
| 354 |
// a[6]a[3] |
| 355 |
// a[7]a[3] |
| 356 |
// a[5]a[4] (v) |
| 357 |
// a[6]a[4] |
| 358 |
// a[7]a[4] |
| 359 |
// a[6]a[5] (vi) |
| 360 |
// a[7]a[5] |
| 361 |
// a[7]a[6] (vii) |
| 362 |
|
| 363 |
mul $t0,$a1,$a0 // lo(a[1..7]*a[0]) (i) |
| 364 |
mul $t1,$a2,$a0 |
| 365 |
mul $t2,$a3,$a0 |
| 366 |
mul $t3,$a4,$a0 |
| 367 |
adds $acc1,$acc1,$t0 // t[1]+lo(a[1]*a[0]) |
| 368 |
mul $t0,$a5,$a0 |
| 369 |
adcs $acc2,$acc2,$t1 |
| 370 |
mul $t1,$a6,$a0 |
| 371 |
adcs $acc3,$acc3,$t2 |
| 372 |
mul $t2,$a7,$a0 |
| 373 |
adcs $acc4,$acc4,$t3 |
| 374 |
umulh $t3,$a1,$a0 // hi(a[1..7]*a[0]) |
| 375 |
adcs $acc5,$acc5,$t0 |
| 376 |
umulh $t0,$a2,$a0 |
| 377 |
adcs $acc6,$acc6,$t1 |
| 378 |
umulh $t1,$a3,$a0 |
| 379 |
adcs $acc7,$acc7,$t2 |
| 380 |
umulh $t2,$a4,$a0 |
| 381 |
stp $acc0,$acc1,[$tp],#8*2 // t[0..1] |
| 382 |
adc $acc0,xzr,xzr // t[8] |
| 383 |
adds $acc2,$acc2,$t3 // t[2]+lo(a[1]*a[0]) |
| 384 |
umulh $t3,$a5,$a0 |
| 385 |
adcs $acc3,$acc3,$t0 |
| 386 |
umulh $t0,$a6,$a0 |
| 387 |
adcs $acc4,$acc4,$t1 |
| 388 |
umulh $t1,$a7,$a0 |
| 389 |
adcs $acc5,$acc5,$t2 |
| 390 |
mul $t2,$a2,$a1 // lo(a[2..7]*a[1]) (ii) |
| 391 |
adcs $acc6,$acc6,$t3 |
| 392 |
mul $t3,$a3,$a1 |
| 393 |
adcs $acc7,$acc7,$t0 |
| 394 |
mul $t0,$a4,$a1 |
| 395 |
adc $acc0,$acc0,$t1 |
| 396 |
|
| 397 |
mul $t1,$a5,$a1 |
| 398 |
adds $acc3,$acc3,$t2 |
| 399 |
mul $t2,$a6,$a1 |
| 400 |
adcs $acc4,$acc4,$t3 |
| 401 |
mul $t3,$a7,$a1 |
| 402 |
adcs $acc5,$acc5,$t0 |
| 403 |
umulh $t0,$a2,$a1 // hi(a[2..7]*a[1]) |
| 404 |
adcs $acc6,$acc6,$t1 |
| 405 |
umulh $t1,$a3,$a1 |
| 406 |
adcs $acc7,$acc7,$t2 |
| 407 |
umulh $t2,$a4,$a1 |
| 408 |
adcs $acc0,$acc0,$t3 |
| 409 |
umulh $t3,$a5,$a1 |
| 410 |
stp $acc2,$acc3,[$tp],#8*2 // t[2..3] |
| 411 |
adc $acc1,xzr,xzr // t[9] |
| 412 |
adds $acc4,$acc4,$t0 |
| 413 |
umulh $t0,$a6,$a1 |
| 414 |
adcs $acc5,$acc5,$t1 |
| 415 |
umulh $t1,$a7,$a1 |
| 416 |
adcs $acc6,$acc6,$t2 |
| 417 |
mul $t2,$a3,$a2 // lo(a[3..7]*a[2]) (iii) |
| 418 |
adcs $acc7,$acc7,$t3 |
| 419 |
mul $t3,$a4,$a2 |
| 420 |
adcs $acc0,$acc0,$t0 |
| 421 |
mul $t0,$a5,$a2 |
| 422 |
adc $acc1,$acc1,$t1 |
| 423 |
|
| 424 |
mul $t1,$a6,$a2 |
| 425 |
adds $acc5,$acc5,$t2 |
| 426 |
mul $t2,$a7,$a2 |
| 427 |
adcs $acc6,$acc6,$t3 |
| 428 |
umulh $t3,$a3,$a2 // hi(a[3..7]*a[2]) |
| 429 |
adcs $acc7,$acc7,$t0 |
| 430 |
umulh $t0,$a4,$a2 |
| 431 |
adcs $acc0,$acc0,$t1 |
| 432 |
umulh $t1,$a5,$a2 |
| 433 |
adcs $acc1,$acc1,$t2 |
| 434 |
umulh $t2,$a6,$a2 |
| 435 |
stp $acc4,$acc5,[$tp],#8*2 // t[4..5] |
| 436 |
adc $acc2,xzr,xzr // t[10] |
| 437 |
adds $acc6,$acc6,$t3 |
| 438 |
umulh $t3,$a7,$a2 |
| 439 |
adcs $acc7,$acc7,$t0 |
| 440 |
mul $t0,$a4,$a3 // lo(a[4..7]*a[3]) (iv) |
| 441 |
adcs $acc0,$acc0,$t1 |
| 442 |
mul $t1,$a5,$a3 |
| 443 |
adcs $acc1,$acc1,$t2 |
| 444 |
mul $t2,$a6,$a3 |
| 445 |
adc $acc2,$acc2,$t3 |
| 446 |
|
| 447 |
mul $t3,$a7,$a3 |
| 448 |
adds $acc7,$acc7,$t0 |
| 449 |
umulh $t0,$a4,$a3 // hi(a[4..7]*a[3]) |
| 450 |
adcs $acc0,$acc0,$t1 |
| 451 |
umulh $t1,$a5,$a3 |
| 452 |
adcs $acc1,$acc1,$t2 |
| 453 |
umulh $t2,$a6,$a3 |
| 454 |
adcs $acc2,$acc2,$t3 |
| 455 |
umulh $t3,$a7,$a3 |
| 456 |
stp $acc6,$acc7,[$tp],#8*2 // t[6..7] |
| 457 |
adc $acc3,xzr,xzr // t[11] |
| 458 |
adds $acc0,$acc0,$t0 |
| 459 |
mul $t0,$a5,$a4 // lo(a[5..7]*a[4]) (v) |
| 460 |
adcs $acc1,$acc1,$t1 |
| 461 |
mul $t1,$a6,$a4 |
| 462 |
adcs $acc2,$acc2,$t2 |
| 463 |
mul $t2,$a7,$a4 |
| 464 |
adc $acc3,$acc3,$t3 |
| 465 |
|
| 466 |
umulh $t3,$a5,$a4 // hi(a[5..7]*a[4]) |
| 467 |
adds $acc1,$acc1,$t0 |
| 468 |
umulh $t0,$a6,$a4 |
| 469 |
adcs $acc2,$acc2,$t1 |
| 470 |
umulh $t1,$a7,$a4 |
| 471 |
adcs $acc3,$acc3,$t2 |
| 472 |
mul $t2,$a6,$a5 // lo(a[6..7]*a[5]) (vi) |
| 473 |
adc $acc4,xzr,xzr // t[12] |
| 474 |
adds $acc2,$acc2,$t3 |
| 475 |
mul $t3,$a7,$a5 |
| 476 |
adcs $acc3,$acc3,$t0 |
| 477 |
umulh $t0,$a6,$a5 // hi(a[6..7]*a[5]) |
| 478 |
adc $acc4,$acc4,$t1 |
| 479 |
|
| 480 |
umulh $t1,$a7,$a5 |
| 481 |
adds $acc3,$acc3,$t2 |
| 482 |
mul $t2,$a7,$a6 // lo(a[7]*a[6]) (vii) |
| 483 |
adcs $acc4,$acc4,$t3 |
| 484 |
umulh $t3,$a7,$a6 // hi(a[7]*a[6]) |
| 485 |
adc $acc5,xzr,xzr // t[13] |
| 486 |
adds $acc4,$acc4,$t0 |
| 487 |
sub $cnt,$ap_end,$ap // done yet? |
| 488 |
adc $acc5,$acc5,$t1 |
| 489 |
|
| 490 |
adds $acc5,$acc5,$t2 |
| 491 |
sub $t0,$ap_end,$num // rewinded ap |
| 492 |
adc $acc6,xzr,xzr // t[14] |
| 493 |
add $acc6,$acc6,$t3 |
| 494 |
|
| 495 |
cbz $cnt,.Lsqr8x_outer_break |
| 496 |
|
| 497 |
mov $n0,$a0 |
| 498 |
ldp $a0,$a1,[$tp,#8*0] |
| 499 |
ldp $a2,$a3,[$tp,#8*2] |
| 500 |
ldp $a4,$a5,[$tp,#8*4] |
| 501 |
ldp $a6,$a7,[$tp,#8*6] |
| 502 |
adds $acc0,$acc0,$a0 |
| 503 |
adcs $acc1,$acc1,$a1 |
| 504 |
ldp $a0,$a1,[$ap,#8*0] |
| 505 |
adcs $acc2,$acc2,$a2 |
| 506 |
adcs $acc3,$acc3,$a3 |
| 507 |
ldp $a2,$a3,[$ap,#8*2] |
| 508 |
adcs $acc4,$acc4,$a4 |
| 509 |
adcs $acc5,$acc5,$a5 |
| 510 |
ldp $a4,$a5,[$ap,#8*4] |
| 511 |
adcs $acc6,$acc6,$a6 |
| 512 |
mov $rp,$ap |
| 513 |
adcs $acc7,xzr,$a7 |
| 514 |
ldp $a6,$a7,[$ap,#8*6] |
| 515 |
add $ap,$ap,#8*8 |
| 516 |
//adc $carry,xzr,xzr // moved below |
| 517 |
mov $cnt,#-8*8 |
| 518 |
|
| 519 |
// a[8]a[0] |
| 520 |
// a[9]a[0] |
| 521 |
// a[a]a[0] |
| 522 |
// a[b]a[0] |
| 523 |
// a[c]a[0] |
| 524 |
// a[d]a[0] |
| 525 |
// a[e]a[0] |
| 526 |
// a[f]a[0] |
| 527 |
// a[8]a[1] |
| 528 |
// a[f]a[1]........................ |
| 529 |
// a[8]a[2] |
| 530 |
// a[f]a[2]........................ |
| 531 |
// a[8]a[3] |
| 532 |
// a[f]a[3]........................ |
| 533 |
// a[8]a[4] |
| 534 |
// a[f]a[4]........................ |
| 535 |
// a[8]a[5] |
| 536 |
// a[f]a[5]........................ |
| 537 |
// a[8]a[6] |
| 538 |
// a[f]a[6]........................ |
| 539 |
// a[8]a[7] |
| 540 |
// a[f]a[7]........................ |
| 541 |
.Lsqr8x_mul: |
| 542 |
mul $t0,$a0,$n0 |
| 543 |
adc $carry,xzr,xzr // carry bit, modulo-scheduled |
| 544 |
mul $t1,$a1,$n0 |
| 545 |
add $cnt,$cnt,#8 |
| 546 |
mul $t2,$a2,$n0 |
| 547 |
mul $t3,$a3,$n0 |
| 548 |
adds $acc0,$acc0,$t0 |
| 549 |
mul $t0,$a4,$n0 |
| 550 |
adcs $acc1,$acc1,$t1 |
| 551 |
mul $t1,$a5,$n0 |
| 552 |
adcs $acc2,$acc2,$t2 |
| 553 |
mul $t2,$a6,$n0 |
| 554 |
adcs $acc3,$acc3,$t3 |
| 555 |
mul $t3,$a7,$n0 |
| 556 |
adcs $acc4,$acc4,$t0 |
| 557 |
umulh $t0,$a0,$n0 |
| 558 |
adcs $acc5,$acc5,$t1 |
| 559 |
umulh $t1,$a1,$n0 |
| 560 |
adcs $acc6,$acc6,$t2 |
| 561 |
umulh $t2,$a2,$n0 |
| 562 |
adcs $acc7,$acc7,$t3 |
| 563 |
umulh $t3,$a3,$n0 |
| 564 |
adc $carry,$carry,xzr |
| 565 |
str $acc0,[$tp],#8 |
| 566 |
adds $acc0,$acc1,$t0 |
| 567 |
umulh $t0,$a4,$n0 |
| 568 |
adcs $acc1,$acc2,$t1 |
| 569 |
umulh $t1,$a5,$n0 |
| 570 |
adcs $acc2,$acc3,$t2 |
| 571 |
umulh $t2,$a6,$n0 |
| 572 |
adcs $acc3,$acc4,$t3 |
| 573 |
umulh $t3,$a7,$n0 |
| 574 |
ldr $n0,[$rp,$cnt] |
| 575 |
adcs $acc4,$acc5,$t0 |
| 576 |
adcs $acc5,$acc6,$t1 |
| 577 |
adcs $acc6,$acc7,$t2 |
| 578 |
adcs $acc7,$carry,$t3 |
| 579 |
//adc $carry,xzr,xzr // moved above |
| 580 |
cbnz $cnt,.Lsqr8x_mul |
| 581 |
// note that carry flag is guaranteed |
| 582 |
// to be zero at this point |
| 583 |
cmp $ap,$ap_end // done yet? |
| 584 |
b.eq .Lsqr8x_break |
| 585 |
|
| 586 |
ldp $a0,$a1,[$tp,#8*0] |
| 587 |
ldp $a2,$a3,[$tp,#8*2] |
| 588 |
ldp $a4,$a5,[$tp,#8*4] |
| 589 |
ldp $a6,$a7,[$tp,#8*6] |
| 590 |
adds $acc0,$acc0,$a0 |
| 591 |
ldr $n0,[$rp,#-8*8] |
| 592 |
adcs $acc1,$acc1,$a1 |
| 593 |
ldp $a0,$a1,[$ap,#8*0] |
| 594 |
adcs $acc2,$acc2,$a2 |
| 595 |
adcs $acc3,$acc3,$a3 |
| 596 |
ldp $a2,$a3,[$ap,#8*2] |
| 597 |
adcs $acc4,$acc4,$a4 |
| 598 |
adcs $acc5,$acc5,$a5 |
| 599 |
ldp $a4,$a5,[$ap,#8*4] |
| 600 |
adcs $acc6,$acc6,$a6 |
| 601 |
mov $cnt,#-8*8 |
| 602 |
adcs $acc7,$acc7,$a7 |
| 603 |
ldp $a6,$a7,[$ap,#8*6] |
| 604 |
add $ap,$ap,#8*8 |
| 605 |
//adc $carry,xzr,xzr // moved above |
| 606 |
b .Lsqr8x_mul |
| 607 |
|
| 608 |
.align 4 |
| 609 |
.Lsqr8x_break: |
| 610 |
ldp $a0,$a1,[$rp,#8*0] |
| 611 |
add $ap,$rp,#8*8 |
| 612 |
ldp $a2,$a3,[$rp,#8*2] |
| 613 |
sub $t0,$ap_end,$ap // is it last iteration? |
| 614 |
ldp $a4,$a5,[$rp,#8*4] |
| 615 |
sub $t1,$tp,$t0 |
| 616 |
ldp $a6,$a7,[$rp,#8*6] |
| 617 |
cbz $t0,.Lsqr8x_outer_loop |
| 618 |
|
| 619 |
stp $acc0,$acc1,[$tp,#8*0] |
| 620 |
ldp $acc0,$acc1,[$t1,#8*0] |
| 621 |
stp $acc2,$acc3,[$tp,#8*2] |
| 622 |
ldp $acc2,$acc3,[$t1,#8*2] |
| 623 |
stp $acc4,$acc5,[$tp,#8*4] |
| 624 |
ldp $acc4,$acc5,[$t1,#8*4] |
| 625 |
stp $acc6,$acc7,[$tp,#8*6] |
| 626 |
mov $tp,$t1 |
| 627 |
ldp $acc6,$acc7,[$t1,#8*6] |
| 628 |
b .Lsqr8x_outer_loop |
| 629 |
|
| 630 |
.align 4 |
| 631 |
.Lsqr8x_outer_break: |
| 632 |
// Now multiply above result by 2 and add a[n-1]*a[n-1]|...|a[0]*a[0] |
| 633 |
ldp $a1,$a3,[$t0,#8*0] // recall that $t0 is &a[0] |
| 634 |
ldp $t1,$t2,[sp,#8*1] |
| 635 |
ldp $a5,$a7,[$t0,#8*2] |
| 636 |
add $ap,$t0,#8*4 |
| 637 |
ldp $t3,$t0,[sp,#8*3] |
| 638 |
|
| 639 |
stp $acc0,$acc1,[$tp,#8*0] |
| 640 |
mul $acc0,$a1,$a1 |
| 641 |
stp $acc2,$acc3,[$tp,#8*2] |
| 642 |
umulh $a1,$a1,$a1 |
| 643 |
stp $acc4,$acc5,[$tp,#8*4] |
| 644 |
mul $a2,$a3,$a3 |
| 645 |
stp $acc6,$acc7,[$tp,#8*6] |
| 646 |
mov $tp,sp |
| 647 |
umulh $a3,$a3,$a3 |
| 648 |
adds $acc1,$a1,$t1,lsl#1 |
| 649 |
extr $t1,$t2,$t1,#63 |
| 650 |
sub $cnt,$num,#8*4 |
| 651 |
|
| 652 |
.Lsqr4x_shift_n_add: |
| 653 |
adcs $acc2,$a2,$t1 |
| 654 |
extr $t2,$t3,$t2,#63 |
| 655 |
sub $cnt,$cnt,#8*4 |
| 656 |
adcs $acc3,$a3,$t2 |
| 657 |
ldp $t1,$t2,[$tp,#8*5] |
| 658 |
mul $a4,$a5,$a5 |
| 659 |
ldp $a1,$a3,[$ap],#8*2 |
| 660 |
umulh $a5,$a5,$a5 |
| 661 |
mul $a6,$a7,$a7 |
| 662 |
umulh $a7,$a7,$a7 |
| 663 |
extr $t3,$t0,$t3,#63 |
| 664 |
stp $acc0,$acc1,[$tp,#8*0] |
| 665 |
adcs $acc4,$a4,$t3 |
| 666 |
extr $t0,$t1,$t0,#63 |
| 667 |
stp $acc2,$acc3,[$tp,#8*2] |
| 668 |
adcs $acc5,$a5,$t0 |
| 669 |
ldp $t3,$t0,[$tp,#8*7] |
| 670 |
extr $t1,$t2,$t1,#63 |
| 671 |
adcs $acc6,$a6,$t1 |
| 672 |
extr $t2,$t3,$t2,#63 |
| 673 |
adcs $acc7,$a7,$t2 |
| 674 |
ldp $t1,$t2,[$tp,#8*9] |
| 675 |
mul $a0,$a1,$a1 |
| 676 |
ldp $a5,$a7,[$ap],#8*2 |
| 677 |
umulh $a1,$a1,$a1 |
| 678 |
mul $a2,$a3,$a3 |
| 679 |
umulh $a3,$a3,$a3 |
| 680 |
stp $acc4,$acc5,[$tp,#8*4] |
| 681 |
extr $t3,$t0,$t3,#63 |
| 682 |
stp $acc6,$acc7,[$tp,#8*6] |
| 683 |
add $tp,$tp,#8*8 |
| 684 |
adcs $acc0,$a0,$t3 |
| 685 |
extr $t0,$t1,$t0,#63 |
| 686 |
adcs $acc1,$a1,$t0 |
| 687 |
ldp $t3,$t0,[$tp,#8*3] |
| 688 |
extr $t1,$t2,$t1,#63 |
| 689 |
cbnz $cnt,.Lsqr4x_shift_n_add |
| 690 |
___ |
| 691 |
my ($np,$np_end)=($ap,$ap_end); |
| 692 |
$code.=<<___; |
| 693 |
ldp $np,$n0,[x29,#104] // pull np and n0 |
| 694 |
|
| 695 |
adcs $acc2,$a2,$t1 |
| 696 |
extr $t2,$t3,$t2,#63 |
| 697 |
adcs $acc3,$a3,$t2 |
| 698 |
ldp $t1,$t2,[$tp,#8*5] |
| 699 |
mul $a4,$a5,$a5 |
| 700 |
umulh $a5,$a5,$a5 |
| 701 |
stp $acc0,$acc1,[$tp,#8*0] |
| 702 |
mul $a6,$a7,$a7 |
| 703 |
umulh $a7,$a7,$a7 |
| 704 |
stp $acc2,$acc3,[$tp,#8*2] |
| 705 |
extr $t3,$t0,$t3,#63 |
| 706 |
adcs $acc4,$a4,$t3 |
| 707 |
extr $t0,$t1,$t0,#63 |
| 708 |
ldp $acc0,$acc1,[sp,#8*0] |
| 709 |
adcs $acc5,$a5,$t0 |
| 710 |
extr $t1,$t2,$t1,#63 |
| 711 |
ldp $a0,$a1,[$np,#8*0] |
| 712 |
adcs $acc6,$a6,$t1 |
| 713 |
extr $t2,xzr,$t2,#63 |
| 714 |
ldp $a2,$a3,[$np,#8*2] |
| 715 |
adc $acc7,$a7,$t2 |
| 716 |
ldp $a4,$a5,[$np,#8*4] |
| 717 |
|
| 718 |
// Reduce by 512 bits per iteration |
| 719 |
mul $na0,$n0,$acc0 // t[0]*n0 |
| 720 |
ldp $a6,$a7,[$np,#8*6] |
| 721 |
add $np_end,$np,$num |
| 722 |
ldp $acc2,$acc3,[sp,#8*2] |
| 723 |
stp $acc4,$acc5,[$tp,#8*4] |
| 724 |
ldp $acc4,$acc5,[sp,#8*4] |
| 725 |
stp $acc6,$acc7,[$tp,#8*6] |
| 726 |
ldp $acc6,$acc7,[sp,#8*6] |
| 727 |
add $np,$np,#8*8 |
| 728 |
mov $topmost,xzr // initial top-most carry |
| 729 |
mov $tp,sp |
| 730 |
mov $cnt,#8 |
| 731 |
|
| 732 |
.Lsqr8x_reduction: |
| 733 |
// (*) mul $t0,$a0,$na0 // lo(n[0-7])*lo(t[0]*n0) |
| 734 |
mul $t1,$a1,$na0 |
| 735 |
sub $cnt,$cnt,#1 |
| 736 |
mul $t2,$a2,$na0 |
| 737 |
str $na0,[$tp],#8 // put aside t[0]*n0 for tail processing |
| 738 |
mul $t3,$a3,$na0 |
| 739 |
// (*) adds xzr,$acc0,$t0 |
| 740 |
subs xzr,$acc0,#1 // (*) |
| 741 |
mul $t0,$a4,$na0 |
| 742 |
adcs $acc0,$acc1,$t1 |
| 743 |
mul $t1,$a5,$na0 |
| 744 |
adcs $acc1,$acc2,$t2 |
| 745 |
mul $t2,$a6,$na0 |
| 746 |
adcs $acc2,$acc3,$t3 |
| 747 |
mul $t3,$a7,$na0 |
| 748 |
adcs $acc3,$acc4,$t0 |
| 749 |
umulh $t0,$a0,$na0 // hi(n[0-7])*lo(t[0]*n0) |
| 750 |
adcs $acc4,$acc5,$t1 |
| 751 |
umulh $t1,$a1,$na0 |
| 752 |
adcs $acc5,$acc6,$t2 |
| 753 |
umulh $t2,$a2,$na0 |
| 754 |
adcs $acc6,$acc7,$t3 |
| 755 |
umulh $t3,$a3,$na0 |
| 756 |
adc $acc7,xzr,xzr |
| 757 |
adds $acc0,$acc0,$t0 |
| 758 |
umulh $t0,$a4,$na0 |
| 759 |
adcs $acc1,$acc1,$t1 |
| 760 |
umulh $t1,$a5,$na0 |
| 761 |
adcs $acc2,$acc2,$t2 |
| 762 |
umulh $t2,$a6,$na0 |
| 763 |
adcs $acc3,$acc3,$t3 |
| 764 |
umulh $t3,$a7,$na0 |
| 765 |
mul $na0,$n0,$acc0 // next t[0]*n0 |
| 766 |
adcs $acc4,$acc4,$t0 |
| 767 |
adcs $acc5,$acc5,$t1 |
| 768 |
adcs $acc6,$acc6,$t2 |
| 769 |
adc $acc7,$acc7,$t3 |
| 770 |
cbnz $cnt,.Lsqr8x_reduction |
| 771 |
|
| 772 |
ldp $t0,$t1,[$tp,#8*0] |
| 773 |
ldp $t2,$t3,[$tp,#8*2] |
| 774 |
mov $rp,$tp |
| 775 |
sub $cnt,$np_end,$np // done yet? |
| 776 |
adds $acc0,$acc0,$t0 |
| 777 |
adcs $acc1,$acc1,$t1 |
| 778 |
ldp $t0,$t1,[$tp,#8*4] |
| 779 |
adcs $acc2,$acc2,$t2 |
| 780 |
adcs $acc3,$acc3,$t3 |
| 781 |
ldp $t2,$t3,[$tp,#8*6] |
| 782 |
adcs $acc4,$acc4,$t0 |
| 783 |
adcs $acc5,$acc5,$t1 |
| 784 |
adcs $acc6,$acc6,$t2 |
| 785 |
adcs $acc7,$acc7,$t3 |
| 786 |
//adc $carry,xzr,xzr // moved below |
| 787 |
cbz $cnt,.Lsqr8x8_post_condition |
| 788 |
|
| 789 |
ldr $n0,[$tp,#-8*8] |
| 790 |
ldp $a0,$a1,[$np,#8*0] |
| 791 |
ldp $a2,$a3,[$np,#8*2] |
| 792 |
ldp $a4,$a5,[$np,#8*4] |
| 793 |
mov $cnt,#-8*8 |
| 794 |
ldp $a6,$a7,[$np,#8*6] |
| 795 |
add $np,$np,#8*8 |
| 796 |
|
| 797 |
.Lsqr8x_tail: |
| 798 |
mul $t0,$a0,$n0 |
| 799 |
adc $carry,xzr,xzr // carry bit, modulo-scheduled |
| 800 |
mul $t1,$a1,$n0 |
| 801 |
add $cnt,$cnt,#8 |
| 802 |
mul $t2,$a2,$n0 |
| 803 |
mul $t3,$a3,$n0 |
| 804 |
adds $acc0,$acc0,$t0 |
| 805 |
mul $t0,$a4,$n0 |
| 806 |
adcs $acc1,$acc1,$t1 |
| 807 |
mul $t1,$a5,$n0 |
| 808 |
adcs $acc2,$acc2,$t2 |
| 809 |
mul $t2,$a6,$n0 |
| 810 |
adcs $acc3,$acc3,$t3 |
| 811 |
mul $t3,$a7,$n0 |
| 812 |
adcs $acc4,$acc4,$t0 |
| 813 |
umulh $t0,$a0,$n0 |
| 814 |
adcs $acc5,$acc5,$t1 |
| 815 |
umulh $t1,$a1,$n0 |
| 816 |
adcs $acc6,$acc6,$t2 |
| 817 |
umulh $t2,$a2,$n0 |
| 818 |
adcs $acc7,$acc7,$t3 |
| 819 |
umulh $t3,$a3,$n0 |
| 820 |
adc $carry,$carry,xzr |
| 821 |
str $acc0,[$tp],#8 |
| 822 |
adds $acc0,$acc1,$t0 |
| 823 |
umulh $t0,$a4,$n0 |
| 824 |
adcs $acc1,$acc2,$t1 |
| 825 |
umulh $t1,$a5,$n0 |
| 826 |
adcs $acc2,$acc3,$t2 |
| 827 |
umulh $t2,$a6,$n0 |
| 828 |
adcs $acc3,$acc4,$t3 |
| 829 |
umulh $t3,$a7,$n0 |
| 830 |
ldr $n0,[$rp,$cnt] |
| 831 |
adcs $acc4,$acc5,$t0 |
| 832 |
adcs $acc5,$acc6,$t1 |
| 833 |
adcs $acc6,$acc7,$t2 |
| 834 |
adcs $acc7,$carry,$t3 |
| 835 |
//adc $carry,xzr,xzr // moved above |
| 836 |
cbnz $cnt,.Lsqr8x_tail |
| 837 |
// note that carry flag is guaranteed |
| 838 |
// to be zero at this point |
| 839 |
ldp $a0,$a1,[$tp,#8*0] |
| 840 |
sub $cnt,$np_end,$np // done yet? |
| 841 |
sub $t2,$np_end,$num // rewinded np |
| 842 |
ldp $a2,$a3,[$tp,#8*2] |
| 843 |
ldp $a4,$a5,[$tp,#8*4] |
| 844 |
ldp $a6,$a7,[$tp,#8*6] |
| 845 |
cbz $cnt,.Lsqr8x_tail_break |
| 846 |
|
| 847 |
ldr $n0,[$rp,#-8*8] |
| 848 |
adds $acc0,$acc0,$a0 |
| 849 |
adcs $acc1,$acc1,$a1 |
| 850 |
ldp $a0,$a1,[$np,#8*0] |
| 851 |
adcs $acc2,$acc2,$a2 |
| 852 |
adcs $acc3,$acc3,$a3 |
| 853 |
ldp $a2,$a3,[$np,#8*2] |
| 854 |
adcs $acc4,$acc4,$a4 |
| 855 |
adcs $acc5,$acc5,$a5 |
| 856 |
ldp $a4,$a5,[$np,#8*4] |
| 857 |
adcs $acc6,$acc6,$a6 |
| 858 |
mov $cnt,#-8*8 |
| 859 |
adcs $acc7,$acc7,$a7 |
| 860 |
ldp $a6,$a7,[$np,#8*6] |
| 861 |
add $np,$np,#8*8 |
| 862 |
//adc $carry,xzr,xzr // moved above |
| 863 |
b .Lsqr8x_tail |
| 864 |
|
| 865 |
.align 4 |
| 866 |
.Lsqr8x_tail_break: |
| 867 |
ldr $n0,[x29,#112] // pull n0 |
| 868 |
add $cnt,$tp,#8*8 // end of current t[num] window |
| 869 |
|
| 870 |
subs xzr,$topmost,#1 // "move" top-most carry to carry bit |
| 871 |
adcs $t0,$acc0,$a0 |
| 872 |
adcs $t1,$acc1,$a1 |
| 873 |
ldp $acc0,$acc1,[$rp,#8*0] |
| 874 |
adcs $acc2,$acc2,$a2 |
| 875 |
ldp $a0,$a1,[$t2,#8*0] // recall that $t2 is &n[0] |
| 876 |
adcs $acc3,$acc3,$a3 |
| 877 |
ldp $a2,$a3,[$t2,#8*2] |
| 878 |
adcs $acc4,$acc4,$a4 |
| 879 |
adcs $acc5,$acc5,$a5 |
| 880 |
ldp $a4,$a5,[$t2,#8*4] |
| 881 |
adcs $acc6,$acc6,$a6 |
| 882 |
adcs $acc7,$acc7,$a7 |
| 883 |
ldp $a6,$a7,[$t2,#8*6] |
| 884 |
add $np,$t2,#8*8 |
| 885 |
adc $topmost,xzr,xzr // top-most carry |
| 886 |
mul $na0,$n0,$acc0 |
| 887 |
stp $t0,$t1,[$tp,#8*0] |
| 888 |
stp $acc2,$acc3,[$tp,#8*2] |
| 889 |
ldp $acc2,$acc3,[$rp,#8*2] |
| 890 |
stp $acc4,$acc5,[$tp,#8*4] |
| 891 |
ldp $acc4,$acc5,[$rp,#8*4] |
| 892 |
cmp $cnt,x29 // did we hit the bottom? |
| 893 |
stp $acc6,$acc7,[$tp,#8*6] |
| 894 |
mov $tp,$rp // slide the window |
| 895 |
ldp $acc6,$acc7,[$rp,#8*6] |
| 896 |
mov $cnt,#8 |
| 897 |
b.ne .Lsqr8x_reduction |
| 898 |
|
| 899 |
// Final step. We see if result is larger than modulus, and |
| 900 |
// if it is, subtract the modulus. But comparison implies |
| 901 |
// subtraction. So we subtract modulus, see if it borrowed, |
| 902 |
// and conditionally copy original value. |
| 903 |
ldr $rp,[x29,#96] // pull rp |
| 904 |
add $tp,$tp,#8*8 |
| 905 |
subs $t0,$acc0,$a0 |
| 906 |
sbcs $t1,$acc1,$a1 |
| 907 |
sub $cnt,$num,#8*8 |
| 908 |
mov $ap_end,$rp // $rp copy |
| 909 |
|
| 910 |
.Lsqr8x_sub: |
| 911 |
sbcs $t2,$acc2,$a2 |
| 912 |
ldp $a0,$a1,[$np,#8*0] |
| 913 |
sbcs $t3,$acc3,$a3 |
| 914 |
stp $t0,$t1,[$rp,#8*0] |
| 915 |
sbcs $t0,$acc4,$a4 |
| 916 |
ldp $a2,$a3,[$np,#8*2] |
| 917 |
sbcs $t1,$acc5,$a5 |
| 918 |
stp $t2,$t3,[$rp,#8*2] |
| 919 |
sbcs $t2,$acc6,$a6 |
| 920 |
ldp $a4,$a5,[$np,#8*4] |
| 921 |
sbcs $t3,$acc7,$a7 |
| 922 |
ldp $a6,$a7,[$np,#8*6] |
| 923 |
add $np,$np,#8*8 |
| 924 |
ldp $acc0,$acc1,[$tp,#8*0] |
| 925 |
sub $cnt,$cnt,#8*8 |
| 926 |
ldp $acc2,$acc3,[$tp,#8*2] |
| 927 |
ldp $acc4,$acc5,[$tp,#8*4] |
| 928 |
ldp $acc6,$acc7,[$tp,#8*6] |
| 929 |
add $tp,$tp,#8*8 |
| 930 |
stp $t0,$t1,[$rp,#8*4] |
| 931 |
sbcs $t0,$acc0,$a0 |
| 932 |
stp $t2,$t3,[$rp,#8*6] |
| 933 |
add $rp,$rp,#8*8 |
| 934 |
sbcs $t1,$acc1,$a1 |
| 935 |
cbnz $cnt,.Lsqr8x_sub |
| 936 |
|
| 937 |
sbcs $t2,$acc2,$a2 |
| 938 |
mov $tp,sp |
| 939 |
add $ap,sp,$num |
| 940 |
ldp $a0,$a1,[$ap_end,#8*0] |
| 941 |
sbcs $t3,$acc3,$a3 |
| 942 |
stp $t0,$t1,[$rp,#8*0] |
| 943 |
sbcs $t0,$acc4,$a4 |
| 944 |
ldp $a2,$a3,[$ap_end,#8*2] |
| 945 |
sbcs $t1,$acc5,$a5 |
| 946 |
stp $t2,$t3,[$rp,#8*2] |
| 947 |
sbcs $t2,$acc6,$a6 |
| 948 |
ldp $acc0,$acc1,[$ap,#8*0] |
| 949 |
sbcs $t3,$acc7,$a7 |
| 950 |
ldp $acc2,$acc3,[$ap,#8*2] |
| 951 |
sbcs xzr,$topmost,xzr // did it borrow? |
| 952 |
ldr x30,[x29,#8] // pull return address |
| 953 |
stp $t0,$t1,[$rp,#8*4] |
| 954 |
stp $t2,$t3,[$rp,#8*6] |
| 955 |
|
| 956 |
sub $cnt,$num,#8*4 |
| 957 |
.Lsqr4x_cond_copy: |
| 958 |
sub $cnt,$cnt,#8*4 |
| 959 |
csel $t0,$acc0,$a0,lo |
| 960 |
stp xzr,xzr,[$tp,#8*0] |
| 961 |
csel $t1,$acc1,$a1,lo |
| 962 |
ldp $a0,$a1,[$ap_end,#8*4] |
| 963 |
ldp $acc0,$acc1,[$ap,#8*4] |
| 964 |
csel $t2,$acc2,$a2,lo |
| 965 |
stp xzr,xzr,[$tp,#8*2] |
| 966 |
add $tp,$tp,#8*4 |
| 967 |
csel $t3,$acc3,$a3,lo |
| 968 |
ldp $a2,$a3,[$ap_end,#8*6] |
| 969 |
ldp $acc2,$acc3,[$ap,#8*6] |
| 970 |
add $ap,$ap,#8*4 |
| 971 |
stp $t0,$t1,[$ap_end,#8*0] |
| 972 |
stp $t2,$t3,[$ap_end,#8*2] |
| 973 |
add $ap_end,$ap_end,#8*4 |
| 974 |
stp xzr,xzr,[$ap,#8*0] |
| 975 |
stp xzr,xzr,[$ap,#8*2] |
| 976 |
cbnz $cnt,.Lsqr4x_cond_copy |
| 977 |
|
| 978 |
csel $t0,$acc0,$a0,lo |
| 979 |
stp xzr,xzr,[$tp,#8*0] |
| 980 |
csel $t1,$acc1,$a1,lo |
| 981 |
stp xzr,xzr,[$tp,#8*2] |
| 982 |
csel $t2,$acc2,$a2,lo |
| 983 |
csel $t3,$acc3,$a3,lo |
| 984 |
stp $t0,$t1,[$ap_end,#8*0] |
| 985 |
stp $t2,$t3,[$ap_end,#8*2] |
| 986 |
|
| 987 |
b .Lsqr8x_done |
| 988 |
|
| 989 |
.align 4 |
| 990 |
.Lsqr8x8_post_condition: |
| 991 |
adc $carry,xzr,xzr |
| 992 |
ldr x30,[x29,#8] // pull return address |
| 993 |
// $acc0-7,$carry hold result, $a0-7 hold modulus |
| 994 |
subs $a0,$acc0,$a0 |
| 995 |
ldr $ap,[x29,#96] // pull rp |
| 996 |
sbcs $a1,$acc1,$a1 |
| 997 |
stp xzr,xzr,[sp,#8*0] |
| 998 |
sbcs $a2,$acc2,$a2 |
| 999 |
stp xzr,xzr,[sp,#8*2] |
| 1000 |
sbcs $a3,$acc3,$a3 |
| 1001 |
stp xzr,xzr,[sp,#8*4] |
| 1002 |
sbcs $a4,$acc4,$a4 |
| 1003 |
stp xzr,xzr,[sp,#8*6] |
| 1004 |
sbcs $a5,$acc5,$a5 |
| 1005 |
stp xzr,xzr,[sp,#8*8] |
| 1006 |
sbcs $a6,$acc6,$a6 |
| 1007 |
stp xzr,xzr,[sp,#8*10] |
| 1008 |
sbcs $a7,$acc7,$a7 |
| 1009 |
stp xzr,xzr,[sp,#8*12] |
| 1010 |
sbcs $carry,$carry,xzr // did it borrow? |
| 1011 |
stp xzr,xzr,[sp,#8*14] |
| 1012 |
|
| 1013 |
// $a0-7 hold result-modulus |
| 1014 |
csel $a0,$acc0,$a0,lo |
| 1015 |
csel $a1,$acc1,$a1,lo |
| 1016 |
csel $a2,$acc2,$a2,lo |
| 1017 |
csel $a3,$acc3,$a3,lo |
| 1018 |
stp $a0,$a1,[$ap,#8*0] |
| 1019 |
csel $a4,$acc4,$a4,lo |
| 1020 |
csel $a5,$acc5,$a5,lo |
| 1021 |
stp $a2,$a3,[$ap,#8*2] |
| 1022 |
csel $a6,$acc6,$a6,lo |
| 1023 |
csel $a7,$acc7,$a7,lo |
| 1024 |
stp $a4,$a5,[$ap,#8*4] |
| 1025 |
stp $a6,$a7,[$ap,#8*6] |
| 1026 |
|
| 1027 |
.Lsqr8x_done: |
| 1028 |
ldp x19,x20,[x29,#16] |
| 1029 |
mov sp,x29 |
| 1030 |
ldp x21,x22,[x29,#32] |
| 1031 |
mov x0,#1 |
| 1032 |
ldp x23,x24,[x29,#48] |
| 1033 |
ldp x25,x26,[x29,#64] |
| 1034 |
ldp x27,x28,[x29,#80] |
| 1035 |
ldr x29,[sp],#128 |
| 1036 |
ret |
| 1037 |
.size __bn_sqr8x_mont,.-__bn_sqr8x_mont |
| 1038 |
___ |
| 1039 |
} |
| 1040 |
|
| 1041 |
{ |
| 1042 |
######################################################################## |
| 1043 |
# Even though this might look as ARMv8 adaptation of mulx4x_mont from |
| 1044 |
# x86_64-mont5 module, it's different in sense that it performs |
| 1045 |
# reduction 256 bits at a time. |
| 1046 |
|
| 1047 |
my ($a0,$a1,$a2,$a3, |
| 1048 |
$t0,$t1,$t2,$t3, |
| 1049 |
$m0,$m1,$m2,$m3, |
| 1050 |
$acc0,$acc1,$acc2,$acc3,$acc4, |
| 1051 |
$bi,$mi,$tp,$ap_end,$cnt) = map("x$_",(6..17,19..28)); |
| 1052 |
my $bp_end=$rp; |
| 1053 |
my ($carry,$topmost) = ($rp,"x30"); |
| 1054 |
|
| 1055 |
$code.=<<___; |
| 1056 |
.type __bn_mul4x_mont,%function |
| 1057 |
.align 5 |
| 1058 |
__bn_mul4x_mont: |
| 1059 |
stp x29,x30,[sp,#-128]! |
| 1060 |
add x29,sp,#0 |
| 1061 |
stp x19,x20,[sp,#16] |
| 1062 |
stp x21,x22,[sp,#32] |
| 1063 |
stp x23,x24,[sp,#48] |
| 1064 |
stp x25,x26,[sp,#64] |
| 1065 |
stp x27,x28,[sp,#80] |
| 1066 |
|
| 1067 |
sub $tp,sp,$num,lsl#3 |
| 1068 |
lsl $num,$num,#3 |
| 1069 |
ldr $n0,[$n0] // *n0 |
| 1070 |
sub sp,$tp,#8*4 // alloca |
| 1071 |
|
| 1072 |
add $t0,$bp,$num |
| 1073 |
add $ap_end,$ap,$num |
| 1074 |
stp $rp,$t0,[x29,#96] // offload rp and &b[num] |
| 1075 |
|
| 1076 |
ldr $bi,[$bp,#8*0] // b[0] |
| 1077 |
ldp $a0,$a1,[$ap,#8*0] // a[0..3] |
| 1078 |
ldp $a2,$a3,[$ap,#8*2] |
| 1079 |
add $ap,$ap,#8*4 |
| 1080 |
mov $acc0,xzr |
| 1081 |
mov $acc1,xzr |
| 1082 |
mov $acc2,xzr |
| 1083 |
mov $acc3,xzr |
| 1084 |
ldp $m0,$m1,[$np,#8*0] // n[0..3] |
| 1085 |
ldp $m2,$m3,[$np,#8*2] |
| 1086 |
adds $np,$np,#8*4 // clear carry bit |
| 1087 |
mov $carry,xzr |
| 1088 |
mov $cnt,#0 |
| 1089 |
mov $tp,sp |
| 1090 |
|
| 1091 |
.Loop_mul4x_1st_reduction: |
| 1092 |
mul $t0,$a0,$bi // lo(a[0..3]*b[0]) |
| 1093 |
adc $carry,$carry,xzr // modulo-scheduled |
| 1094 |
mul $t1,$a1,$bi |
| 1095 |
add $cnt,$cnt,#8 |
| 1096 |
mul $t2,$a2,$bi |
| 1097 |
and $cnt,$cnt,#31 |
| 1098 |
mul $t3,$a3,$bi |
| 1099 |
adds $acc0,$acc0,$t0 |
| 1100 |
umulh $t0,$a0,$bi // hi(a[0..3]*b[0]) |
| 1101 |
adcs $acc1,$acc1,$t1 |
| 1102 |
mul $mi,$acc0,$n0 // t[0]*n0 |
| 1103 |
adcs $acc2,$acc2,$t2 |
| 1104 |
umulh $t1,$a1,$bi |
| 1105 |
adcs $acc3,$acc3,$t3 |
| 1106 |
umulh $t2,$a2,$bi |
| 1107 |
adc $acc4,xzr,xzr |
| 1108 |
umulh $t3,$a3,$bi |
| 1109 |
ldr $bi,[$bp,$cnt] // next b[i] (or b[0]) |
| 1110 |
adds $acc1,$acc1,$t0 |
| 1111 |
// (*) mul $t0,$m0,$mi // lo(n[0..3]*t[0]*n0) |
| 1112 |
str $mi,[$tp],#8 // put aside t[0]*n0 for tail processing |
| 1113 |
adcs $acc2,$acc2,$t1 |
| 1114 |
mul $t1,$m1,$mi |
| 1115 |
adcs $acc3,$acc3,$t2 |
| 1116 |
mul $t2,$m2,$mi |
| 1117 |
adc $acc4,$acc4,$t3 // can't overflow |
| 1118 |
mul $t3,$m3,$mi |
| 1119 |
// (*) adds xzr,$acc0,$t0 |
| 1120 |
subs xzr,$acc0,#1 // (*) |
| 1121 |
umulh $t0,$m0,$mi // hi(n[0..3]*t[0]*n0) |
| 1122 |
adcs $acc0,$acc1,$t1 |
| 1123 |
umulh $t1,$m1,$mi |
| 1124 |
adcs $acc1,$acc2,$t2 |
| 1125 |
umulh $t2,$m2,$mi |
| 1126 |
adcs $acc2,$acc3,$t3 |
| 1127 |
umulh $t3,$m3,$mi |
| 1128 |
adcs $acc3,$acc4,$carry |
| 1129 |
adc $carry,xzr,xzr |
| 1130 |
adds $acc0,$acc0,$t0 |
| 1131 |
sub $t0,$ap_end,$ap |
| 1132 |
adcs $acc1,$acc1,$t1 |
| 1133 |
adcs $acc2,$acc2,$t2 |
| 1134 |
adcs $acc3,$acc3,$t3 |
| 1135 |
//adc $carry,$carry,xzr |
| 1136 |
cbnz $cnt,.Loop_mul4x_1st_reduction |
| 1137 |
|
| 1138 |
cbz $t0,.Lmul4x4_post_condition |
| 1139 |
|
| 1140 |
ldp $a0,$a1,[$ap,#8*0] // a[4..7] |
| 1141 |
ldp $a2,$a3,[$ap,#8*2] |
| 1142 |
add $ap,$ap,#8*4 |
| 1143 |
ldr $mi,[sp] // a[0]*n0 |
| 1144 |
ldp $m0,$m1,[$np,#8*0] // n[4..7] |
| 1145 |
ldp $m2,$m3,[$np,#8*2] |
| 1146 |
add $np,$np,#8*4 |
| 1147 |
|
| 1148 |
.Loop_mul4x_1st_tail: |
| 1149 |
mul $t0,$a0,$bi // lo(a[4..7]*b[i]) |
| 1150 |
adc $carry,$carry,xzr // modulo-scheduled |
| 1151 |
mul $t1,$a1,$bi |
| 1152 |
add $cnt,$cnt,#8 |
| 1153 |
mul $t2,$a2,$bi |
| 1154 |
and $cnt,$cnt,#31 |
| 1155 |
mul $t3,$a3,$bi |
| 1156 |
adds $acc0,$acc0,$t0 |
| 1157 |
umulh $t0,$a0,$bi // hi(a[4..7]*b[i]) |
| 1158 |
adcs $acc1,$acc1,$t1 |
| 1159 |
umulh $t1,$a1,$bi |
| 1160 |
adcs $acc2,$acc2,$t2 |
| 1161 |
umulh $t2,$a2,$bi |
| 1162 |
adcs $acc3,$acc3,$t3 |
| 1163 |
umulh $t3,$a3,$bi |
| 1164 |
adc $acc4,xzr,xzr |
| 1165 |
ldr $bi,[$bp,$cnt] // next b[i] (or b[0]) |
| 1166 |
adds $acc1,$acc1,$t0 |
| 1167 |
mul $t0,$m0,$mi // lo(n[4..7]*a[0]*n0) |
| 1168 |
adcs $acc2,$acc2,$t1 |
| 1169 |
mul $t1,$m1,$mi |
| 1170 |
adcs $acc3,$acc3,$t2 |
| 1171 |
mul $t2,$m2,$mi |
| 1172 |
adc $acc4,$acc4,$t3 // can't overflow |
| 1173 |
mul $t3,$m3,$mi |
| 1174 |
adds $acc0,$acc0,$t0 |
| 1175 |
umulh $t0,$m0,$mi // hi(n[4..7]*a[0]*n0) |
| 1176 |
adcs $acc1,$acc1,$t1 |
| 1177 |
umulh $t1,$m1,$mi |
| 1178 |
adcs $acc2,$acc2,$t2 |
| 1179 |
umulh $t2,$m2,$mi |
| 1180 |
adcs $acc3,$acc3,$t3 |
| 1181 |
adcs $acc4,$acc4,$carry |
| 1182 |
umulh $t3,$m3,$mi |
| 1183 |
adc $carry,xzr,xzr |
| 1184 |
ldr $mi,[sp,$cnt] // next t[0]*n0 |
| 1185 |
str $acc0,[$tp],#8 // result!!! |
| 1186 |
adds $acc0,$acc1,$t0 |
| 1187 |
sub $t0,$ap_end,$ap // done yet? |
| 1188 |
adcs $acc1,$acc2,$t1 |
| 1189 |
adcs $acc2,$acc3,$t2 |
| 1190 |
adcs $acc3,$acc4,$t3 |
| 1191 |
//adc $carry,$carry,xzr |
| 1192 |
cbnz $cnt,.Loop_mul4x_1st_tail |
| 1193 |
|
| 1194 |
sub $t1,$ap_end,$num // rewinded $ap |
| 1195 |
cbz $t0,.Lmul4x_proceed |
| 1196 |
|
| 1197 |
ldp $a0,$a1,[$ap,#8*0] |
| 1198 |
ldp $a2,$a3,[$ap,#8*2] |
| 1199 |
add $ap,$ap,#8*4 |
| 1200 |
ldp $m0,$m1,[$np,#8*0] |
| 1201 |
ldp $m2,$m3,[$np,#8*2] |
| 1202 |
add $np,$np,#8*4 |
| 1203 |
b .Loop_mul4x_1st_tail |
| 1204 |
|
| 1205 |
.align 5 |
| 1206 |
.Lmul4x_proceed: |
| 1207 |
ldr $bi,[$bp,#8*4]! // *++b |
| 1208 |
adc $topmost,$carry,xzr |
| 1209 |
ldp $a0,$a1,[$t1,#8*0] // a[0..3] |
| 1210 |
sub $np,$np,$num // rewind np |
| 1211 |
ldp $a2,$a3,[$t1,#8*2] |
| 1212 |
add $ap,$t1,#8*4 |
| 1213 |
|
| 1214 |
stp $acc0,$acc1,[$tp,#8*0] // result!!! |
| 1215 |
ldp $acc0,$acc1,[sp,#8*4] // t[0..3] |
| 1216 |
stp $acc2,$acc3,[$tp,#8*2] // result!!! |
| 1217 |
ldp $acc2,$acc3,[sp,#8*6] |
| 1218 |
|
| 1219 |
ldp $m0,$m1,[$np,#8*0] // n[0..3] |
| 1220 |
mov $tp,sp |
| 1221 |
ldp $m2,$m3,[$np,#8*2] |
| 1222 |
adds $np,$np,#8*4 // clear carry bit |
| 1223 |
mov $carry,xzr |
| 1224 |
|
| 1225 |
.align 4 |
| 1226 |
.Loop_mul4x_reduction: |
| 1227 |
mul $t0,$a0,$bi // lo(a[0..3]*b[4]) |
| 1228 |
adc $carry,$carry,xzr // modulo-scheduled |
| 1229 |
mul $t1,$a1,$bi |
| 1230 |
add $cnt,$cnt,#8 |
| 1231 |
mul $t2,$a2,$bi |
| 1232 |
and $cnt,$cnt,#31 |
| 1233 |
mul $t3,$a3,$bi |
| 1234 |
adds $acc0,$acc0,$t0 |
| 1235 |
umulh $t0,$a0,$bi // hi(a[0..3]*b[4]) |
| 1236 |
adcs $acc1,$acc1,$t1 |
| 1237 |
mul $mi,$acc0,$n0 // t[0]*n0 |
| 1238 |
adcs $acc2,$acc2,$t2 |
| 1239 |
umulh $t1,$a1,$bi |
| 1240 |
adcs $acc3,$acc3,$t3 |
| 1241 |
umulh $t2,$a2,$bi |
| 1242 |
adc $acc4,xzr,xzr |
| 1243 |
umulh $t3,$a3,$bi |
| 1244 |
ldr $bi,[$bp,$cnt] // next b[i] |
| 1245 |
adds $acc1,$acc1,$t0 |
| 1246 |
// (*) mul $t0,$m0,$mi |
| 1247 |
str $mi,[$tp],#8 // put aside t[0]*n0 for tail processing |
| 1248 |
adcs $acc2,$acc2,$t1 |
| 1249 |
mul $t1,$m1,$mi // lo(n[0..3]*t[0]*n0 |
| 1250 |
adcs $acc3,$acc3,$t2 |
| 1251 |
mul $t2,$m2,$mi |
| 1252 |
adc $acc4,$acc4,$t3 // can't overflow |
| 1253 |
mul $t3,$m3,$mi |
| 1254 |
// (*) adds xzr,$acc0,$t0 |
| 1255 |
subs xzr,$acc0,#1 // (*) |
| 1256 |
umulh $t0,$m0,$mi // hi(n[0..3]*t[0]*n0 |
| 1257 |
adcs $acc0,$acc1,$t1 |
| 1258 |
umulh $t1,$m1,$mi |
| 1259 |
adcs $acc1,$acc2,$t2 |
| 1260 |
umulh $t2,$m2,$mi |
| 1261 |
adcs $acc2,$acc3,$t3 |
| 1262 |
umulh $t3,$m3,$mi |
| 1263 |
adcs $acc3,$acc4,$carry |
| 1264 |
adc $carry,xzr,xzr |
| 1265 |
adds $acc0,$acc0,$t0 |
| 1266 |
adcs $acc1,$acc1,$t1 |
| 1267 |
adcs $acc2,$acc2,$t2 |
| 1268 |
adcs $acc3,$acc3,$t3 |
| 1269 |
//adc $carry,$carry,xzr |
| 1270 |
cbnz $cnt,.Loop_mul4x_reduction |
| 1271 |
|
| 1272 |
adc $carry,$carry,xzr |
| 1273 |
ldp $t0,$t1,[$tp,#8*4] // t[4..7] |
| 1274 |
ldp $t2,$t3,[$tp,#8*6] |
| 1275 |
ldp $a0,$a1,[$ap,#8*0] // a[4..7] |
| 1276 |
ldp $a2,$a3,[$ap,#8*2] |
| 1277 |
add $ap,$ap,#8*4 |
| 1278 |
adds $acc0,$acc0,$t0 |
| 1279 |
adcs $acc1,$acc1,$t1 |
| 1280 |
adcs $acc2,$acc2,$t2 |
| 1281 |
adcs $acc3,$acc3,$t3 |
| 1282 |
//adc $carry,$carry,xzr |
| 1283 |
|
| 1284 |
ldr $mi,[sp] // t[0]*n0 |
| 1285 |
ldp $m0,$m1,[$np,#8*0] // n[4..7] |
| 1286 |
ldp $m2,$m3,[$np,#8*2] |
| 1287 |
add $np,$np,#8*4 |
| 1288 |
|
| 1289 |
.align 4 |
| 1290 |
.Loop_mul4x_tail: |
| 1291 |
mul $t0,$a0,$bi // lo(a[4..7]*b[4]) |
| 1292 |
adc $carry,$carry,xzr // modulo-scheduled |
| 1293 |
mul $t1,$a1,$bi |
| 1294 |
add $cnt,$cnt,#8 |
| 1295 |
mul $t2,$a2,$bi |
| 1296 |
and $cnt,$cnt,#31 |
| 1297 |
mul $t3,$a3,$bi |
| 1298 |
adds $acc0,$acc0,$t0 |
| 1299 |
umulh $t0,$a0,$bi // hi(a[4..7]*b[4]) |
| 1300 |
adcs $acc1,$acc1,$t1 |
| 1301 |
umulh $t1,$a1,$bi |
| 1302 |
adcs $acc2,$acc2,$t2 |
| 1303 |
umulh $t2,$a2,$bi |
| 1304 |
adcs $acc3,$acc3,$t3 |
| 1305 |
umulh $t3,$a3,$bi |
| 1306 |
adc $acc4,xzr,xzr |
| 1307 |
ldr $bi,[$bp,$cnt] // next b[i] |
| 1308 |
adds $acc1,$acc1,$t0 |
| 1309 |
mul $t0,$m0,$mi // lo(n[4..7]*t[0]*n0) |
| 1310 |
adcs $acc2,$acc2,$t1 |
| 1311 |
mul $t1,$m1,$mi |
| 1312 |
adcs $acc3,$acc3,$t2 |
| 1313 |
mul $t2,$m2,$mi |
| 1314 |
adc $acc4,$acc4,$t3 // can't overflow |
| 1315 |
mul $t3,$m3,$mi |
| 1316 |
adds $acc0,$acc0,$t0 |
| 1317 |
umulh $t0,$m0,$mi // hi(n[4..7]*t[0]*n0) |
| 1318 |
adcs $acc1,$acc1,$t1 |
| 1319 |
umulh $t1,$m1,$mi |
| 1320 |
adcs $acc2,$acc2,$t2 |
| 1321 |
umulh $t2,$m2,$mi |
| 1322 |
adcs $acc3,$acc3,$t3 |
| 1323 |
umulh $t3,$m3,$mi |
| 1324 |
adcs $acc4,$acc4,$carry |
| 1325 |
ldr $mi,[sp,$cnt] // next a[0]*n0 |
| 1326 |
adc $carry,xzr,xzr |
| 1327 |
str $acc0,[$tp],#8 // result!!! |
| 1328 |
adds $acc0,$acc1,$t0 |
| 1329 |
sub $t0,$ap_end,$ap // done yet? |
| 1330 |
adcs $acc1,$acc2,$t1 |
| 1331 |
adcs $acc2,$acc3,$t2 |
| 1332 |
adcs $acc3,$acc4,$t3 |
| 1333 |
//adc $carry,$carry,xzr |
| 1334 |
cbnz $cnt,.Loop_mul4x_tail |
| 1335 |
|
| 1336 |
sub $t1,$np,$num // rewinded np? |
| 1337 |
adc $carry,$carry,xzr |
| 1338 |
cbz $t0,.Loop_mul4x_break |
| 1339 |
|
| 1340 |
ldp $t0,$t1,[$tp,#8*4] |
| 1341 |
ldp $t2,$t3,[$tp,#8*6] |
| 1342 |
ldp $a0,$a1,[$ap,#8*0] |
| 1343 |
ldp $a2,$a3,[$ap,#8*2] |
| 1344 |
add $ap,$ap,#8*4 |
| 1345 |
adds $acc0,$acc0,$t0 |
| 1346 |
adcs $acc1,$acc1,$t1 |
| 1347 |
adcs $acc2,$acc2,$t2 |
| 1348 |
adcs $acc3,$acc3,$t3 |
| 1349 |
//adc $carry,$carry,xzr |
| 1350 |
ldp $m0,$m1,[$np,#8*0] |
| 1351 |
ldp $m2,$m3,[$np,#8*2] |
| 1352 |
add $np,$np,#8*4 |
| 1353 |
b .Loop_mul4x_tail |
| 1354 |
|
| 1355 |
.align 4 |
| 1356 |
.Loop_mul4x_break: |
| 1357 |
ldp $t2,$t3,[x29,#96] // pull rp and &b[num] |
| 1358 |
adds $acc0,$acc0,$topmost |
| 1359 |
add $bp,$bp,#8*4 // bp++ |
| 1360 |
adcs $acc1,$acc1,xzr |
| 1361 |
sub $ap,$ap,$num // rewind ap |
| 1362 |
adcs $acc2,$acc2,xzr |
| 1363 |
stp $acc0,$acc1,[$tp,#8*0] // result!!! |
| 1364 |
adcs $acc3,$acc3,xzr |
| 1365 |
ldp $acc0,$acc1,[sp,#8*4] // t[0..3] |
| 1366 |
adc $topmost,$carry,xzr |
| 1367 |
stp $acc2,$acc3,[$tp,#8*2] // result!!! |
| 1368 |
cmp $bp,$t3 // done yet? |
| 1369 |
ldp $acc2,$acc3,[sp,#8*6] |
| 1370 |
ldp $m0,$m1,[$t1,#8*0] // n[0..3] |
| 1371 |
ldp $m2,$m3,[$t1,#8*2] |
| 1372 |
add $np,$t1,#8*4 |
| 1373 |
b.eq .Lmul4x_post |
| 1374 |
|
| 1375 |
ldr $bi,[$bp] |
| 1376 |
ldp $a0,$a1,[$ap,#8*0] // a[0..3] |
| 1377 |
ldp $a2,$a3,[$ap,#8*2] |
| 1378 |
adds $ap,$ap,#8*4 // clear carry bit |
| 1379 |
mov $carry,xzr |
| 1380 |
mov $tp,sp |
| 1381 |
b .Loop_mul4x_reduction |
| 1382 |
|
| 1383 |
.align 4 |
| 1384 |
.Lmul4x_post: |
| 1385 |
// Final step. We see if result is larger than modulus, and |
| 1386 |
// if it is, subtract the modulus. But comparison implies |
| 1387 |
// subtraction. So we subtract modulus, see if it borrowed, |
| 1388 |
// and conditionally copy original value. |
| 1389 |
mov $rp,$t2 |
| 1390 |
mov $ap_end,$t2 // $rp copy |
| 1391 |
subs $t0,$acc0,$m0 |
| 1392 |
add $tp,sp,#8*8 |
| 1393 |
sbcs $t1,$acc1,$m1 |
| 1394 |
sub $cnt,$num,#8*4 |
| 1395 |
|
| 1396 |
.Lmul4x_sub: |
| 1397 |
sbcs $t2,$acc2,$m2 |
| 1398 |
ldp $m0,$m1,[$np,#8*0] |
| 1399 |
sub $cnt,$cnt,#8*4 |
| 1400 |
ldp $acc0,$acc1,[$tp,#8*0] |
| 1401 |
sbcs $t3,$acc3,$m3 |
| 1402 |
ldp $m2,$m3,[$np,#8*2] |
| 1403 |
add $np,$np,#8*4 |
| 1404 |
ldp $acc2,$acc3,[$tp,#8*2] |
| 1405 |
add $tp,$tp,#8*4 |
| 1406 |
stp $t0,$t1,[$rp,#8*0] |
| 1407 |
sbcs $t0,$acc0,$m0 |
| 1408 |
stp $t2,$t3,[$rp,#8*2] |
| 1409 |
add $rp,$rp,#8*4 |
| 1410 |
sbcs $t1,$acc1,$m1 |
| 1411 |
cbnz $cnt,.Lmul4x_sub |
| 1412 |
|
| 1413 |
sbcs $t2,$acc2,$m2 |
| 1414 |
mov $tp,sp |
| 1415 |
add $ap,sp,#8*4 |
| 1416 |
ldp $a0,$a1,[$ap_end,#8*0] |
| 1417 |
sbcs $t3,$acc3,$m3 |
| 1418 |
stp $t0,$t1,[$rp,#8*0] |
| 1419 |
ldp $a2,$a3,[$ap_end,#8*2] |
| 1420 |
stp $t2,$t3,[$rp,#8*2] |
| 1421 |
ldp $acc0,$acc1,[$ap,#8*0] |
| 1422 |
ldp $acc2,$acc3,[$ap,#8*2] |
| 1423 |
sbcs xzr,$topmost,xzr // did it borrow? |
| 1424 |
ldr x30,[x29,#8] // pull return address |
| 1425 |
|
| 1426 |
sub $cnt,$num,#8*4 |
| 1427 |
.Lmul4x_cond_copy: |
| 1428 |
sub $cnt,$cnt,#8*4 |
| 1429 |
csel $t0,$acc0,$a0,lo |
| 1430 |
stp xzr,xzr,[$tp,#8*0] |
| 1431 |
csel $t1,$acc1,$a1,lo |
| 1432 |
ldp $a0,$a1,[$ap_end,#8*4] |
| 1433 |
ldp $acc0,$acc1,[$ap,#8*4] |
| 1434 |
csel $t2,$acc2,$a2,lo |
| 1435 |
stp xzr,xzr,[$tp,#8*2] |
| 1436 |
add $tp,$tp,#8*4 |
| 1437 |
csel $t3,$acc3,$a3,lo |
| 1438 |
ldp $a2,$a3,[$ap_end,#8*6] |
| 1439 |
ldp $acc2,$acc3,[$ap,#8*6] |
| 1440 |
add $ap,$ap,#8*4 |
| 1441 |
stp $t0,$t1,[$ap_end,#8*0] |
| 1442 |
stp $t2,$t3,[$ap_end,#8*2] |
| 1443 |
add $ap_end,$ap_end,#8*4 |
| 1444 |
cbnz $cnt,.Lmul4x_cond_copy |
| 1445 |
|
| 1446 |
csel $t0,$acc0,$a0,lo |
| 1447 |
stp xzr,xzr,[$tp,#8*0] |
| 1448 |
csel $t1,$acc1,$a1,lo |
| 1449 |
stp xzr,xzr,[$tp,#8*2] |
| 1450 |
csel $t2,$acc2,$a2,lo |
| 1451 |
stp xzr,xzr,[$tp,#8*3] |
| 1452 |
csel $t3,$acc3,$a3,lo |
| 1453 |
stp xzr,xzr,[$tp,#8*4] |
| 1454 |
stp $t0,$t1,[$ap_end,#8*0] |
| 1455 |
stp $t2,$t3,[$ap_end,#8*2] |
| 1456 |
|
| 1457 |
b .Lmul4x_done |
| 1458 |
|
| 1459 |
.align 4 |
| 1460 |
.Lmul4x4_post_condition: |
| 1461 |
adc $carry,$carry,xzr |
| 1462 |
ldr $ap,[x29,#96] // pull rp |
| 1463 |
// $acc0-3,$carry hold result, $m0-7 hold modulus |
| 1464 |
subs $a0,$acc0,$m0 |
| 1465 |
ldr x30,[x29,#8] // pull return address |
| 1466 |
sbcs $a1,$acc1,$m1 |
| 1467 |
stp xzr,xzr,[sp,#8*0] |
| 1468 |
sbcs $a2,$acc2,$m2 |
| 1469 |
stp xzr,xzr,[sp,#8*2] |
| 1470 |
sbcs $a3,$acc3,$m3 |
| 1471 |
stp xzr,xzr,[sp,#8*4] |
| 1472 |
sbcs xzr,$carry,xzr // did it borrow? |
| 1473 |
stp xzr,xzr,[sp,#8*6] |
| 1474 |
|
| 1475 |
// $a0-3 hold result-modulus |
| 1476 |
csel $a0,$acc0,$a0,lo |
| 1477 |
csel $a1,$acc1,$a1,lo |
| 1478 |
csel $a2,$acc2,$a2,lo |
| 1479 |
csel $a3,$acc3,$a3,lo |
| 1480 |
stp $a0,$a1,[$ap,#8*0] |
| 1481 |
stp $a2,$a3,[$ap,#8*2] |
| 1482 |
|
| 1483 |
.Lmul4x_done: |
| 1484 |
ldp x19,x20,[x29,#16] |
| 1485 |
mov sp,x29 |
| 1486 |
ldp x21,x22,[x29,#32] |
| 1487 |
mov x0,#1 |
| 1488 |
ldp x23,x24,[x29,#48] |
| 1489 |
ldp x25,x26,[x29,#64] |
| 1490 |
ldp x27,x28,[x29,#80] |
| 1491 |
ldr x29,[sp],#128 |
| 1492 |
ret |
| 1493 |
.size __bn_mul4x_mont,.-__bn_mul4x_mont |
| 1494 |
___ |
| 1495 |
} |
| 1496 |
$code.=<<___; |
| 1497 |
.asciz "Montgomery Multiplication for ARMv8, CRYPTOGAMS by <appro\@openssl.org>" |
| 1498 |
.align 4 |
| 1499 |
___ |
| 1500 |
|
| 1501 |
print $code; |
| 1502 |
|
| 1503 |
close STDOUT; |