1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/security/nss/lib/freebl/mpi/mpi_i86pc.s Wed Dec 31 06:09:35 2014 +0100
1.3 @@ -0,0 +1,313 @@
1.4 +/
1.5 +/ This Source Code Form is subject to the terms of the Mozilla Public
1.6 +/ License, v. 2.0. If a copy of the MPL was not distributed with this
1.7 +/ file, You can obtain one at http://mozilla.org/MPL/2.0/.
1.8 +
1.9 +.text
1.10 +
1.11 + / ebp - 36: caller's esi
1.12 + / ebp - 32: caller's edi
1.13 + / ebp - 28:
1.14 + / ebp - 24:
1.15 + / ebp - 20:
1.16 + / ebp - 16:
1.17 + / ebp - 12:
1.18 + / ebp - 8:
1.19 + / ebp - 4:
1.20 + / ebp + 0: caller's ebp
1.21 + / ebp + 4: return address
1.22 + / ebp + 8: a argument
1.23 + / ebp + 12: a_len argument
1.24 + / ebp + 16: b argument
1.25 + / ebp + 20: c argument
1.26 + / registers:
1.27 + / eax:
1.28 + / ebx: carry
1.29 + / ecx: a_len
1.30 + / edx:
1.31 + / esi: a ptr
1.32 + / edi: c ptr
1.33 +.globl s_mpv_mul_d
1.34 +.type s_mpv_mul_d,@function
1.35 +s_mpv_mul_d:
1.36 + push %ebp
1.37 + mov %esp,%ebp
1.38 + sub $28,%esp
1.39 + push %edi
1.40 + push %esi
1.41 + push %ebx
1.42 + movl $0,%ebx / carry = 0
1.43 + mov 12(%ebp),%ecx / ecx = a_len
1.44 + mov 20(%ebp),%edi
1.45 + cmp $0,%ecx
1.46 + je L2 / jmp if a_len == 0
1.47 + mov 8(%ebp),%esi / esi = a
1.48 + cld
1.49 +L1:
1.50 + lodsl / eax = [ds:esi]; esi += 4
1.51 + mov 16(%ebp),%edx / edx = b
1.52 + mull %edx / edx:eax = Phi:Plo = a_i * b
1.53 +
1.54 + add %ebx,%eax / add carry (%ebx) to edx:eax
1.55 + adc $0,%edx
1.56 + mov %edx,%ebx / high half of product becomes next carry
1.57 +
1.58 + stosl / [es:edi] = ax; edi += 4;
1.59 + dec %ecx / --a_len
1.60 + jnz L1 / jmp if a_len != 0
1.61 +L2:
1.62 + mov %ebx,0(%edi) / *c = carry
1.63 + pop %ebx
1.64 + pop %esi
1.65 + pop %edi
1.66 + leave
1.67 + ret
1.68 + nop
1.69 +
1.70 + / ebp - 36: caller's esi
1.71 + / ebp - 32: caller's edi
1.72 + / ebp - 28:
1.73 + / ebp - 24:
1.74 + / ebp - 20:
1.75 + / ebp - 16:
1.76 + / ebp - 12:
1.77 + / ebp - 8:
1.78 + / ebp - 4:
1.79 + / ebp + 0: caller's ebp
1.80 + / ebp + 4: return address
1.81 + / ebp + 8: a argument
1.82 + / ebp + 12: a_len argument
1.83 + / ebp + 16: b argument
1.84 + / ebp + 20: c argument
1.85 + / registers:
1.86 + / eax:
1.87 + / ebx: carry
1.88 + / ecx: a_len
1.89 + / edx:
1.90 + / esi: a ptr
1.91 + / edi: c ptr
1.92 +.globl s_mpv_mul_d_add
1.93 +.type s_mpv_mul_d_add,@function
1.94 +s_mpv_mul_d_add:
1.95 + push %ebp
1.96 + mov %esp,%ebp
1.97 + sub $28,%esp
1.98 + push %edi
1.99 + push %esi
1.100 + push %ebx
1.101 + movl $0,%ebx / carry = 0
1.102 + mov 12(%ebp),%ecx / ecx = a_len
1.103 + mov 20(%ebp),%edi
1.104 + cmp $0,%ecx
1.105 + je L4 / jmp if a_len == 0
1.106 + mov 8(%ebp),%esi / esi = a
1.107 + cld
1.108 +L3:
1.109 + lodsl / eax = [ds:esi]; esi += 4
1.110 + mov 16(%ebp),%edx / edx = b
1.111 + mull %edx / edx:eax = Phi:Plo = a_i * b
1.112 +
1.113 + add %ebx,%eax / add carry (%ebx) to edx:eax
1.114 + adc $0,%edx
1.115 + mov 0(%edi),%ebx / add in current word from *c
1.116 + add %ebx,%eax
1.117 + adc $0,%edx
1.118 + mov %edx,%ebx / high half of product becomes next carry
1.119 +
1.120 + stosl / [es:edi] = ax; edi += 4;
1.121 + dec %ecx / --a_len
1.122 + jnz L3 / jmp if a_len != 0
1.123 +L4:
1.124 + mov %ebx,0(%edi) / *c = carry
1.125 + pop %ebx
1.126 + pop %esi
1.127 + pop %edi
1.128 + leave
1.129 + ret
1.130 + nop
1.131 +
1.132 + / ebp - 36: caller's esi
1.133 + / ebp - 32: caller's edi
1.134 + / ebp - 28:
1.135 + / ebp - 24:
1.136 + / ebp - 20:
1.137 + / ebp - 16:
1.138 + / ebp - 12:
1.139 + / ebp - 8:
1.140 + / ebp - 4:
1.141 + / ebp + 0: caller's ebp
1.142 + / ebp + 4: return address
1.143 + / ebp + 8: a argument
1.144 + / ebp + 12: a_len argument
1.145 + / ebp + 16: b argument
1.146 + / ebp + 20: c argument
1.147 + / registers:
1.148 + / eax:
1.149 + / ebx: carry
1.150 + / ecx: a_len
1.151 + / edx:
1.152 + / esi: a ptr
1.153 + / edi: c ptr
1.154 +.globl s_mpv_mul_d_add_prop
1.155 +.type s_mpv_mul_d_add_prop,@function
1.156 +s_mpv_mul_d_add_prop:
1.157 + push %ebp
1.158 + mov %esp,%ebp
1.159 + sub $28,%esp
1.160 + push %edi
1.161 + push %esi
1.162 + push %ebx
1.163 + movl $0,%ebx / carry = 0
1.164 + mov 12(%ebp),%ecx / ecx = a_len
1.165 + mov 20(%ebp),%edi
1.166 + cmp $0,%ecx
1.167 + je L6 / jmp if a_len == 0
1.168 + cld
1.169 + mov 8(%ebp),%esi / esi = a
1.170 +L5:
1.171 + lodsl / eax = [ds:esi]; esi += 4
1.172 + mov 16(%ebp),%edx / edx = b
1.173 + mull %edx / edx:eax = Phi:Plo = a_i * b
1.174 +
1.175 + add %ebx,%eax / add carry (%ebx) to edx:eax
1.176 + adc $0,%edx
1.177 + mov 0(%edi),%ebx / add in current word from *c
1.178 + add %ebx,%eax
1.179 + adc $0,%edx
1.180 + mov %edx,%ebx / high half of product becomes next carry
1.181 +
1.182 + stosl / [es:edi] = ax; edi += 4;
1.183 + dec %ecx / --a_len
1.184 + jnz L5 / jmp if a_len != 0
1.185 +L6:
1.186 + cmp $0,%ebx / is carry zero?
1.187 + jz L8
1.188 + mov 0(%edi),%eax / add in current word from *c
1.189 + add %ebx,%eax
1.190 + stosl / [es:edi] = ax; edi += 4;
1.191 + jnc L8
1.192 +L7:
1.193 + mov 0(%edi),%eax / add in current word from *c
1.194 + adc $0,%eax
1.195 + stosl / [es:edi] = ax; edi += 4;
1.196 + jc L7
1.197 +L8:
1.198 + pop %ebx
1.199 + pop %esi
1.200 + pop %edi
1.201 + leave
1.202 + ret
1.203 + nop
1.204 +
1.205 + / ebp - 20: caller's esi
1.206 + / ebp - 16: caller's edi
1.207 + / ebp - 12:
1.208 + / ebp - 8: carry
1.209 + / ebp - 4: a_len local
1.210 + / ebp + 0: caller's ebp
1.211 + / ebp + 4: return address
1.212 + / ebp + 8: pa argument
1.213 + / ebp + 12: a_len argument
1.214 + / ebp + 16: ps argument
1.215 + / ebp + 20:
1.216 + / registers:
1.217 + / eax:
1.218 + / ebx: carry
1.219 + / ecx: a_len
1.220 + / edx:
1.221 + / esi: a ptr
1.222 + / edi: c ptr
1.223 +
1.224 +.globl s_mpv_sqr_add_prop
1.225 +.type s_mpv_sqr_add_prop,@function
1.226 +s_mpv_sqr_add_prop:
1.227 + push %ebp
1.228 + mov %esp,%ebp
1.229 + sub $12,%esp
1.230 + push %edi
1.231 + push %esi
1.232 + push %ebx
1.233 + movl $0,%ebx / carry = 0
1.234 + mov 12(%ebp),%ecx / a_len
1.235 + mov 16(%ebp),%edi / edi = ps
1.236 + cmp $0,%ecx
1.237 + je L11 / jump if a_len == 0
1.238 + cld
1.239 + mov 8(%ebp),%esi / esi = pa
1.240 +L10:
1.241 + lodsl / %eax = [ds:si]; si += 4;
1.242 + mull %eax
1.243 +
1.244 + add %ebx,%eax / add "carry"
1.245 + adc $0,%edx
1.246 + mov 0(%edi),%ebx
1.247 + add %ebx,%eax / add low word from result
1.248 + mov 4(%edi),%ebx
1.249 + stosl / [es:di] = %eax; di += 4;
1.250 + adc %ebx,%edx / add high word from result
1.251 + movl $0,%ebx
1.252 + mov %edx,%eax
1.253 + adc $0,%ebx
1.254 + stosl / [es:di] = %eax; di += 4;
1.255 + dec %ecx / --a_len
1.256 + jnz L10 / jmp if a_len != 0
1.257 +L11:
1.258 + cmp $0,%ebx / is carry zero?
1.259 + jz L14
1.260 + mov 0(%edi),%eax / add in current word from *c
1.261 + add %ebx,%eax
1.262 + stosl / [es:edi] = ax; edi += 4;
1.263 + jnc L14
1.264 +L12:
1.265 + mov 0(%edi),%eax / add in current word from *c
1.266 + adc $0,%eax
1.267 + stosl / [es:edi] = ax; edi += 4;
1.268 + jc L12
1.269 +L14:
1.270 + pop %ebx
1.271 + pop %esi
1.272 + pop %edi
1.273 + leave
1.274 + ret
1.275 + nop
1.276 +
1.277 + /
1.278 + / Divide 64-bit (Nhi,Nlo) by 32-bit divisor, which must be normalized
1.279 + / so its high bit is 1. This code is from NSPR.
1.280 + /
1.281 + / mp_err s_mpv_div_2dx1d(mp_digit Nhi, mp_digit Nlo, mp_digit divisor,
1.282 + / mp_digit *qp, mp_digit *rp)
1.283 +
1.284 + / esp + 0: Caller's ebx
1.285 + / esp + 4: return address
1.286 + / esp + 8: Nhi argument
1.287 + / esp + 12: Nlo argument
1.288 + / esp + 16: divisor argument
1.289 + / esp + 20: qp argument
1.290 + / esp + 24: rp argument
1.291 + / registers:
1.292 + / eax:
1.293 + / ebx: carry
1.294 + / ecx: a_len
1.295 + / edx:
1.296 + / esi: a ptr
1.297 + / edi: c ptr
1.298 + /
1.299 +
1.300 +.globl s_mpv_div_2dx1d
1.301 +.type s_mpv_div_2dx1d,@function
1.302 +s_mpv_div_2dx1d:
1.303 + push %ebx
1.304 + mov 8(%esp),%edx
1.305 + mov 12(%esp),%eax
1.306 + mov 16(%esp),%ebx
1.307 + div %ebx
1.308 + mov 20(%esp),%ebx
1.309 + mov %eax,0(%ebx)
1.310 + mov 24(%esp),%ebx
1.311 + mov %edx,0(%ebx)
1.312 + xor %eax,%eax / return zero
1.313 + pop %ebx
1.314 + ret
1.315 + nop
1.316 +
