1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/security/nss/lib/freebl/mpi/mpi_mips.s Wed Dec 31 06:09:35 2014 +0100
1.3 @@ -0,0 +1,472 @@
1.4 +/* This Source Code Form is subject to the terms of the Mozilla Public
1.5 + * License, v. 2.0. If a copy of the MPL was not distributed with this
1.6 + * file, You can obtain one at http://mozilla.org/MPL/2.0/. */
1.7 +#include <regdef.h>
1.8 + .set noreorder
1.9 + .set noat
1.10 +
1.11 + .section .text, 1, 0x00000006, 4, 4
1.12 +.text:
1.13 + .section .text
1.14 +
1.15 + .ent s_mpv_mul_d_add
1.16 + .globl s_mpv_mul_d_add
1.17 +
1.18 +s_mpv_mul_d_add:
1.19 + #/* c += a * b */
1.20 + #void s_mpv_mul_d_add(const mp_digit *a, mp_size a_len, mp_digit b,
1.21 + # mp_digit *c)
1.22 + #{
1.23 + # mp_digit a0, a1; regs a4, a5
1.24 + # mp_digit c0, c1; regs a6, a7
1.25 + # mp_digit cy = 0; reg t2
1.26 + # mp_word w0, w1; regs t0, t1
1.27 + #
1.28 + # if (a_len) {
1.29 + beq a1,zero,.L.1
1.30 + move t2,zero # cy = 0
1.31 + dsll32 a2,a2,0 # "b" is sometimes negative (?!?!)
1.32 + dsrl32 a2,a2,0 # This clears the upper 32 bits.
1.33 + # a0 = a[0];
1.34 + lwu a4,0(a0)
1.35 + # w0 = ((mp_word)b * a0);
1.36 + dmultu a2,a4
1.37 + # if (--a_len) {
1.38 + addiu a1,a1,-1
1.39 + beq a1,zero,.L.2
1.40 + # while (a_len >= 2) {
1.41 + sltiu t3,a1,2
1.42 + bne t3,zero,.L.3
1.43 + # a1 = a[1];
1.44 + lwu a5,4(a0)
1.45 +.L.4:
1.46 + # a_len -= 2;
1.47 + addiu a1,a1,-2
1.48 + # c0 = c[0];
1.49 + lwu a6,0(a3)
1.50 + # w0 += cy;
1.51 + mflo t0
1.52 + daddu t0,t0,t2
1.53 + # w0 += c0;
1.54 + daddu t0,t0,a6
1.55 + # w1 = (mp_word)b * a1;
1.56 + dmultu a2,a5 #
1.57 + # cy = CARRYOUT(w0);
1.58 + dsrl32 t2,t0,0
1.59 + # c[0] = ACCUM(w0);
1.60 + sw t0,0(a3)
1.61 + # a0 = a[2];
1.62 + lwu a4,8(a0)
1.63 + # a += 2;
1.64 + addiu a0,a0,8
1.65 + # c1 = c[1];
1.66 + lwu a7,4(a3)
1.67 + # w1 += cy;
1.68 + mflo t1
1.69 + daddu t1,t1,t2
1.70 + # w1 += c1;
1.71 + daddu t1,t1,a7
1.72 + # w0 = (mp_word)b * a0;
1.73 + dmultu a2,a4 #
1.74 + # cy = CARRYOUT(w1);
1.75 + dsrl32 t2,t1,0
1.76 + # c[1] = ACCUM(w1);
1.77 + sw t1,4(a3)
1.78 + # c += 2;
1.79 + addiu a3,a3,8
1.80 + sltiu t3,a1,2
1.81 + beq t3,zero,.L.4
1.82 + # a1 = a[1];
1.83 + lwu a5,4(a0)
1.84 + # }
1.85 +.L.3:
1.86 + # c0 = c[0];
1.87 + lwu a6,0(a3)
1.88 + # w0 += cy;
1.89 + # if (a_len) {
1.90 + mflo t0
1.91 + beq a1,zero,.L.5
1.92 + daddu t0,t0,t2
1.93 + # w1 = (mp_word)b * a1;
1.94 + dmultu a2,a5
1.95 + # w0 += c0;
1.96 + daddu t0,t0,a6 #
1.97 + # cy = CARRYOUT(w0);
1.98 + dsrl32 t2,t0,0
1.99 + # c[0] = ACCUM(w0);
1.100 + sw t0,0(a3)
1.101 + # c1 = c[1];
1.102 + lwu a7,4(a3)
1.103 + # w1 += cy;
1.104 + mflo t1
1.105 + daddu t1,t1,t2
1.106 + # w1 += c1;
1.107 + daddu t1,t1,a7
1.108 + # c[1] = ACCUM(w1);
1.109 + sw t1,4(a3)
1.110 + # cy = CARRYOUT(w1);
1.111 + dsrl32 t2,t1,0
1.112 + # c += 1;
1.113 + b .L.6
1.114 + addiu a3,a3,4
1.115 + # } else {
1.116 +.L.5:
1.117 + # w0 += c0;
1.118 + daddu t0,t0,a6
1.119 + # c[0] = ACCUM(w0);
1.120 + sw t0,0(a3)
1.121 + # cy = CARRYOUT(w0);
1.122 + b .L.6
1.123 + dsrl32 t2,t0,0
1.124 + # }
1.125 + # } else {
1.126 +.L.2:
1.127 + # c0 = c[0];
1.128 + lwu a6,0(a3)
1.129 + # w0 += c0;
1.130 + mflo t0
1.131 + daddu t0,t0,a6
1.132 + # c[0] = ACCUM(w0);
1.133 + sw t0,0(a3)
1.134 + # cy = CARRYOUT(w0);
1.135 + dsrl32 t2,t0,0
1.136 + # }
1.137 +.L.6:
1.138 + # c[1] = cy;
1.139 + jr ra
1.140 + sw t2,4(a3)
1.141 + # }
1.142 +.L.1:
1.143 + jr ra
1.144 + nop
1.145 + #}
1.146 + #
1.147 + .end s_mpv_mul_d_add
1.148 +
1.149 + .ent s_mpv_mul_d_add_prop
1.150 + .globl s_mpv_mul_d_add_prop
1.151 +
1.152 +s_mpv_mul_d_add_prop:
1.153 + #/* c += a * b */
1.154 + #void s_mpv_mul_d_add_prop(const mp_digit *a, mp_size a_len, mp_digit b,
1.155 + # mp_digit *c)
1.156 + #{
1.157 + # mp_digit a0, a1; regs a4, a5
1.158 + # mp_digit c0, c1; regs a6, a7
1.159 + # mp_digit cy = 0; reg t2
1.160 + # mp_word w0, w1; regs t0, t1
1.161 + #
1.162 + # if (a_len) {
1.163 + beq a1,zero,.M.1
1.164 + move t2,zero # cy = 0
1.165 + dsll32 a2,a2,0 # "b" is sometimes negative (?!?!)
1.166 + dsrl32 a2,a2,0 # This clears the upper 32 bits.
1.167 + # a0 = a[0];
1.168 + lwu a4,0(a0)
1.169 + # w0 = ((mp_word)b * a0);
1.170 + dmultu a2,a4
1.171 + # if (--a_len) {
1.172 + addiu a1,a1,-1
1.173 + beq a1,zero,.M.2
1.174 + # while (a_len >= 2) {
1.175 + sltiu t3,a1,2
1.176 + bne t3,zero,.M.3
1.177 + # a1 = a[1];
1.178 + lwu a5,4(a0)
1.179 +.M.4:
1.180 + # a_len -= 2;
1.181 + addiu a1,a1,-2
1.182 + # c0 = c[0];
1.183 + lwu a6,0(a3)
1.184 + # w0 += cy;
1.185 + mflo t0
1.186 + daddu t0,t0,t2
1.187 + # w0 += c0;
1.188 + daddu t0,t0,a6
1.189 + # w1 = (mp_word)b * a1;
1.190 + dmultu a2,a5 #
1.191 + # cy = CARRYOUT(w0);
1.192 + dsrl32 t2,t0,0
1.193 + # c[0] = ACCUM(w0);
1.194 + sw t0,0(a3)
1.195 + # a0 = a[2];
1.196 + lwu a4,8(a0)
1.197 + # a += 2;
1.198 + addiu a0,a0,8
1.199 + # c1 = c[1];
1.200 + lwu a7,4(a3)
1.201 + # w1 += cy;
1.202 + mflo t1
1.203 + daddu t1,t1,t2
1.204 + # w1 += c1;
1.205 + daddu t1,t1,a7
1.206 + # w0 = (mp_word)b * a0;
1.207 + dmultu a2,a4 #
1.208 + # cy = CARRYOUT(w1);
1.209 + dsrl32 t2,t1,0
1.210 + # c[1] = ACCUM(w1);
1.211 + sw t1,4(a3)
1.212 + # c += 2;
1.213 + addiu a3,a3,8
1.214 + sltiu t3,a1,2
1.215 + beq t3,zero,.M.4
1.216 + # a1 = a[1];
1.217 + lwu a5,4(a0)
1.218 + # }
1.219 +.M.3:
1.220 + # c0 = c[0];
1.221 + lwu a6,0(a3)
1.222 + # w0 += cy;
1.223 + # if (a_len) {
1.224 + mflo t0
1.225 + beq a1,zero,.M.5
1.226 + daddu t0,t0,t2
1.227 + # w1 = (mp_word)b * a1;
1.228 + dmultu a2,a5
1.229 + # w0 += c0;
1.230 + daddu t0,t0,a6 #
1.231 + # cy = CARRYOUT(w0);
1.232 + dsrl32 t2,t0,0
1.233 + # c[0] = ACCUM(w0);
1.234 + sw t0,0(a3)
1.235 + # c1 = c[1];
1.236 + lwu a7,4(a3)
1.237 + # w1 += cy;
1.238 + mflo t1
1.239 + daddu t1,t1,t2
1.240 + # w1 += c1;
1.241 + daddu t1,t1,a7
1.242 + # c[1] = ACCUM(w1);
1.243 + sw t1,4(a3)
1.244 + # cy = CARRYOUT(w1);
1.245 + dsrl32 t2,t1,0
1.246 + # c += 1;
1.247 + b .M.6
1.248 + addiu a3,a3,8
1.249 + # } else {
1.250 +.M.5:
1.251 + # w0 += c0;
1.252 + daddu t0,t0,a6
1.253 + # c[0] = ACCUM(w0);
1.254 + sw t0,0(a3)
1.255 + # cy = CARRYOUT(w0);
1.256 + dsrl32 t2,t0,0
1.257 + b .M.6
1.258 + addiu a3,a3,4
1.259 + # }
1.260 + # } else {
1.261 +.M.2:
1.262 + # c0 = c[0];
1.263 + lwu a6,0(a3)
1.264 + # w0 += c0;
1.265 + mflo t0
1.266 + daddu t0,t0,a6
1.267 + # c[0] = ACCUM(w0);
1.268 + sw t0,0(a3)
1.269 + # cy = CARRYOUT(w0);
1.270 + dsrl32 t2,t0,0
1.271 + addiu a3,a3,4
1.272 + # }
1.273 +.M.6:
1.274 +
1.275 + # while (cy) {
1.276 + beq t2,zero,.M.1
1.277 + nop
1.278 +.M.7:
1.279 + # mp_word w = (mp_word)*c + cy;
1.280 + lwu a6,0(a3)
1.281 + daddu t2,t2,a6
1.282 + # *c++ = ACCUM(w);
1.283 + sw t2,0(a3)
1.284 + # cy = CARRYOUT(w);
1.285 + dsrl32 t2,t2,0
1.286 + bne t2,zero,.M.7
1.287 + addiu a3,a3,4
1.288 +
1.289 + # }
1.290 +.M.1:
1.291 + jr ra
1.292 + nop
1.293 + #}
1.294 + #
1.295 + .end s_mpv_mul_d_add_prop
1.296 +
1.297 + .ent s_mpv_mul_d
1.298 + .globl s_mpv_mul_d
1.299 +
1.300 +s_mpv_mul_d:
1.301 + #/* c = a * b */
1.302 + #void s_mpv_mul_d(const mp_digit *a, mp_size a_len, mp_digit b,
1.303 + # mp_digit *c)
1.304 + #{
1.305 + # mp_digit a0, a1; regs a4, a5
1.306 + # mp_digit cy = 0; reg t2
1.307 + # mp_word w0, w1; regs t0, t1
1.308 + #
1.309 + # if (a_len) {
1.310 + beq a1,zero,.N.1
1.311 + move t2,zero # cy = 0
1.312 + dsll32 a2,a2,0 # "b" is sometimes negative (?!?!)
1.313 + dsrl32 a2,a2,0 # This clears the upper 32 bits.
1.314 + # a0 = a[0];
1.315 + lwu a4,0(a0)
1.316 + # w0 = ((mp_word)b * a0);
1.317 + dmultu a2,a4
1.318 + # if (--a_len) {
1.319 + addiu a1,a1,-1
1.320 + beq a1,zero,.N.2
1.321 + # while (a_len >= 2) {
1.322 + sltiu t3,a1,2
1.323 + bne t3,zero,.N.3
1.324 + # a1 = a[1];
1.325 + lwu a5,4(a0)
1.326 +.N.4:
1.327 + # a_len -= 2;
1.328 + addiu a1,a1,-2
1.329 + # w0 += cy;
1.330 + mflo t0
1.331 + daddu t0,t0,t2
1.332 + # cy = CARRYOUT(w0);
1.333 + dsrl32 t2,t0,0
1.334 + # w1 = (mp_word)b * a1;
1.335 + dmultu a2,a5
1.336 + # c[0] = ACCUM(w0);
1.337 + sw t0,0(a3)
1.338 + # a0 = a[2];
1.339 + lwu a4,8(a0)
1.340 + # a += 2;
1.341 + addiu a0,a0,8
1.342 + # w1 += cy;
1.343 + mflo t1
1.344 + daddu t1,t1,t2
1.345 + # cy = CARRYOUT(w1);
1.346 + dsrl32 t2,t1,0
1.347 + # w0 = (mp_word)b * a0;
1.348 + dmultu a2,a4
1.349 + # c[1] = ACCUM(w1);
1.350 + sw t1,4(a3)
1.351 + # c += 2;
1.352 + addiu a3,a3,8
1.353 + sltiu t3,a1,2
1.354 + beq t3,zero,.N.4
1.355 + # a1 = a[1];
1.356 + lwu a5,4(a0)
1.357 + # }
1.358 +.N.3:
1.359 + # w0 += cy;
1.360 + # if (a_len) {
1.361 + mflo t0
1.362 + beq a1,zero,.N.5
1.363 + daddu t0,t0,t2
1.364 + # w1 = (mp_word)b * a1;
1.365 + dmultu a2,a5 #
1.366 + # cy = CARRYOUT(w0);
1.367 + dsrl32 t2,t0,0
1.368 + # c[0] = ACCUM(w0);
1.369 + sw t0,0(a3)
1.370 + # w1 += cy;
1.371 + mflo t1
1.372 + daddu t1,t1,t2
1.373 + # c[1] = ACCUM(w1);
1.374 + sw t1,4(a3)
1.375 + # cy = CARRYOUT(w1);
1.376 + dsrl32 t2,t1,0
1.377 + # c += 1;
1.378 + b .N.6
1.379 + addiu a3,a3,4
1.380 + # } else {
1.381 +.N.5:
1.382 + # c[0] = ACCUM(w0);
1.383 + sw t0,0(a3)
1.384 + # cy = CARRYOUT(w0);
1.385 + b .N.6
1.386 + dsrl32 t2,t0,0
1.387 + # }
1.388 + # } else {
1.389 +.N.2:
1.390 + mflo t0
1.391 + # c[0] = ACCUM(w0);
1.392 + sw t0,0(a3)
1.393 + # cy = CARRYOUT(w0);
1.394 + dsrl32 t2,t0,0
1.395 + # }
1.396 +.N.6:
1.397 + # c[1] = cy;
1.398 + jr ra
1.399 + sw t2,4(a3)
1.400 + # }
1.401 +.N.1:
1.402 + jr ra
1.403 + nop
1.404 + #}
1.405 + #
1.406 + .end s_mpv_mul_d
1.407 +
1.408 +
1.409 + .ent s_mpv_sqr_add_prop
1.410 + .globl s_mpv_sqr_add_prop
1.411 + #void s_mpv_sqr_add_prop(const mp_digit *a, mp_size a_len, mp_digit *sqrs);
1.412 + # registers
1.413 + # a0 *a
1.414 + # a1 a_len
1.415 + # a2 *sqr
1.416 + # a3 digit from *a, a_i
1.417 + # a4 square of digit from a
1.418 + # a5,a6 next 2 digits in sqr
1.419 + # a7,t0 carry
1.420 +s_mpv_sqr_add_prop:
1.421 + move a7,zero
1.422 + move t0,zero
1.423 + lwu a3,0(a0)
1.424 + addiu a1,a1,-1 # --a_len
1.425 + dmultu a3,a3
1.426 + beq a1,zero,.P.3 # jump if we've already done the only sqr
1.427 + addiu a0,a0,4 # ++a
1.428 +.P.2:
1.429 + lwu a5,0(a2)
1.430 + lwu a6,4(a2)
1.431 + addiu a2,a2,8 # sqrs += 2;
1.432 + dsll32 a6,a6,0
1.433 + daddu a5,a5,a6
1.434 + lwu a3,0(a0)
1.435 + addiu a0,a0,4 # ++a
1.436 + mflo a4
1.437 + daddu a6,a5,a4
1.438 + sltu a7,a6,a5 # a7 = a6 < a5 detect overflow
1.439 + dmultu a3,a3
1.440 + daddu a4,a6,t0
1.441 + sltu t0,a4,a6
1.442 + add t0,t0,a7
1.443 + sw a4,-8(a2)
1.444 + addiu a1,a1,-1 # --a_len
1.445 + dsrl32 a4,a4,0
1.446 + bne a1,zero,.P.2 # loop if a_len > 0
1.447 + sw a4,-4(a2)
1.448 +.P.3:
1.449 + lwu a5,0(a2)
1.450 + lwu a6,4(a2)
1.451 + addiu a2,a2,8 # sqrs += 2;
1.452 + dsll32 a6,a6,0
1.453 + daddu a5,a5,a6
1.454 + mflo a4
1.455 + daddu a6,a5,a4
1.456 + sltu a7,a6,a5 # a7 = a6 < a5 detect overflow
1.457 + daddu a4,a6,t0
1.458 + sltu t0,a4,a6
1.459 + add t0,t0,a7
1.460 + sw a4,-8(a2)
1.461 + beq t0,zero,.P.9 # jump if no carry
1.462 + dsrl32 a4,a4,0
1.463 +.P.8:
1.464 + sw a4,-4(a2)
1.465 + /* propagate final carry */
1.466 + lwu a5,0(a2)
1.467 + daddu a6,a5,t0
1.468 + sltu t0,a6,a5
1.469 + bne t0,zero,.P.8 # loop if carry persists
1.470 + addiu a2,a2,4 # sqrs++
1.471 +.P.9:
1.472 + jr ra
1.473 + sw a4,-4(a2)
1.474 +
1.475 + .end s_mpv_sqr_add_prop
