arch/c6x/lib/csum_64plus.S

   1 ; SPDX-License-Identifier: GPL-2.0-only
   2 ;
   3 ;  linux/arch/c6x/lib/csum_64plus.s
   4 ;
   5 ;  Port on Texas Instruments TMS320C6x architecture
   6 ;
   7 ;  Copyright (C) 2006, 2009, 2010, 2011 Texas Instruments Incorporated
   8 ;  Author: Aurelien Jacquiot (aurelien.jacquiot@jaluna.com)
   9 ;
  10 #include <linux/linkage.h>
  11
  12 ;
  13 ;unsigned int csum_partial_copy_nocheck(const char *src, char * dst,
  14 ;                                       int len, int sum)
  15 ;
  16 ; A4:   src
  17 ; B4:   dst
  18 ; A6:   len
  19 ; B6:   sum
  20 ; return csum in A4
  21 ;
  22
  23         .text
  24 ENTRY(csum_partial_copy_nocheck)
  25         MVC     .S2     ILC,B30
  26
  27         ZERO    .D1     A9              ; csum (a side)
  28 ||      ZERO    .D2     B9              ; csum (b side)
  29 ||      SHRU    .S2X    A6,2,B5         ; len / 4
  30
  31         ;; Check alignment and size
  32         AND     .S1     3,A4,A1
  33 ||      AND     .S2     3,B4,B0
  34         OR      .L2X    B0,A1,B0        ; non aligned condition
  35 ||      MVC     .S2     B5,ILC
  36 ||      MVK     .D2     1,B2
  37 ||      MV      .D1X    B5,A1           ; words condition
  38   [!A1] B       .S1     L8
  39    [B0] BNOP    .S1     L6,5
  40
  41         SPLOOP          1
  42
  43         ;; Main loop for aligned words
  44         LDW     .D1T1   *A4++,A7
  45         NOP     4
  46         MV      .S2X    A7,B7
  47 ||      EXTU    .S1     A7,0,16,A16
  48         STW     .D2T2   B7,*B4++
  49 ||      MPYU    .M2     B7,B2,B8
  50 ||      ADD     .L1     A16,A9,A9
  51         NOP
  52         SPKERNEL        8,0
  53 ||      ADD     .L2     B8,B9,B9
  54
  55         ZERO    .D1     A1
  56 ||      ADD     .L1X    A9,B9,A9        ;  add csum from a and b sides
  57
  58 L6:
  59   [!A1] BNOP    .S1     L8,5
  60
  61         ;; Main loop for non-aligned words
  62         SPLOOP          2
  63  ||     MVK     .L1     1,A2
  64
  65         LDNW    .D1T1   *A4++,A7
  66         NOP             3
  67
  68         NOP
  69         MV      .S2X    A7,B7
  70  ||     EXTU    .S1     A7,0,16,A16
  71  ||     MPYU    .M1     A7,A2,A8
  72
  73         ADD     .L1     A16,A9,A9
  74         SPKERNEL        6,0
  75  ||     STNW    .D2T2   B7,*B4++
  76  ||     ADD     .L1     A8,A9,A9
  77
  78 L8:     AND     .S2X    2,A6,B5
  79         CMPGT   .L2     B5,0,B0
  80   [!B0] BNOP    .S1     L82,4
  81
  82         ;; Manage half-word
  83         ZERO    .L1     A7
  84 ||      ZERO    .D1     A8
  85
  86 #ifdef CONFIG_CPU_BIG_ENDIAN
  87
  88         LDBU    .D1T1   *A4++,A7
  89         LDBU    .D1T1   *A4++,A8
  90         NOP             3
  91         SHL     .S1     A7,8,A0
  92         ADD     .S1     A8,A9,A9
  93         STB     .D2T1   A7,*B4++
  94 ||      ADD     .S1     A0,A9,A9
  95         STB     .D2T1   A8,*B4++
  96
  97 #else
  98
  99         LDBU    .D1T1   *A4++,A7
 100         LDBU    .D1T1   *A4++,A8
 101         NOP             3
 102         ADD     .S1     A7,A9,A9
 103         SHL     .S1     A8,8,A0
 104
 105         STB     .D2T1   A7,*B4++
 106 ||      ADD     .S1     A0,A9,A9
 107         STB     .D2T1   A8,*B4++
 108
 109 #endif
 110
 111         ;; Manage eventually the last byte
 112 L82:    AND     .S2X    1,A6,B0
 113   [!B0] BNOP    .S1     L9,5
 114
 115 ||      ZERO    .L1     A7
 116
 117 L83:    LDBU    .D1T1   *A4++,A7
 118         NOP             4
 119
 120         MV      .L2X    A7,B7
 121
 122 #ifdef CONFIG_CPU_BIG_ENDIAN
 123
 124         STB     .D2T2   B7,*B4++
 125 ||      SHL     .S1     A7,8,A7
 126         ADD     .S1     A7,A9,A9
 127
 128 #else
 129
 130         STB     .D2T2   B7,*B4++
 131 ||      ADD     .S1     A7,A9,A9
 132
 133 #endif
 134
 135         ;; Fold the csum
 136 L9:     SHRU    .S2X    A9,16,B0
 137   [!B0] BNOP    .S1     L10,5
 138
 139 L91:    SHRU    .S2X    A9,16,B4
 140 ||      EXTU    .S1     A9,16,16,A3
 141         ADD     .D1X    A3,B4,A9
 142
 143         SHRU    .S1     A9,16,A0
 144    [A0] BNOP    .S1     L91,5
 145
 146 L10:    MV      .D1     A9,A4
 147
 148         BNOP    .S2     B3,4
 149         MVC     .S2     B30,ILC
 150 ENDPROC(csum_partial_copy_nocheck)
 151
 152 ;
 153 ;unsigned short
 154 ;ip_fast_csum(unsigned char *iph, unsigned int ihl)
 155 ;{
 156 ;       unsigned int checksum = 0;
 157 ;       unsigned short *tosum = (unsigned short *) iph;
 158 ;       int len;
 159 ;
 160 ;       len = ihl*4;
 161 ;
 162 ;       if (len <= 0)
 163 ;               return 0;
 164 ;
 165 ;       while(len) {
 166 ;               len -= 2;
 167 ;               checksum += *tosum++;
 168 ;       }
 169 ;       if (len & 1)
 170 ;               checksum += *(unsigned char*) tosum;
 171 ;
 172 ;       while(checksum >> 16)
 173 ;               checksum = (checksum & 0xffff) + (checksum >> 16);
 174 ;
 175 ;       return ~checksum;
 176 ;}
 177 ;
 178 ; A4:   iph
 179 ; B4:   ihl
 180 ; return checksum in A4
 181 ;
 182         .text
 183
 184 ENTRY(ip_fast_csum)
 185         ZERO    .D1     A5
 186  ||     MVC     .S2     ILC,B30
 187         SHL     .S2     B4,2,B0
 188         CMPGT   .L2     B0,0,B1
 189   [!B1] BNOP    .S1     L15,4
 190   [!B1] ZERO    .D1     A3
 191
 192   [!B0] B       .S1     L12
 193         SHRU    .S2     B0,1,B0
 194         MVC     .S2     B0,ILC
 195         NOP     3
 196
 197         SPLOOP  1
 198         LDHU    .D1T1   *A4++,A3
 199         NOP     3
 200         NOP
 201         SPKERNEL        5,0
 202  ||     ADD     .L1     A3,A5,A5
 203
 204 L12:    SHRU    .S1     A5,16,A0
 205   [!A0] BNOP    .S1     L14,5
 206
 207 L13:    SHRU    .S2X    A5,16,B4
 208         EXTU    .S1     A5,16,16,A3
 209         ADD     .D1X    A3,B4,A5
 210         SHRU    .S1     A5,16,A0
 211   [A0]  BNOP    .S1     L13,5
 212
 213 L14:    NOT     .D1     A5,A3
 214         EXTU    .S1     A3,16,16,A3
 215
 216 L15:    BNOP    .S2     B3,3
 217         MVC     .S2     B30,ILC
 218         MV      .D1     A3,A4
 219 ENDPROC(ip_fast_csum)
 220
 221 ;
 222 ;unsigned short
 223 ;do_csum(unsigned char *buff, unsigned int len)
 224 ;{
 225 ;       int odd, count;
 226 ;       unsigned int result = 0;
 227 ;
 228 ;       if (len <= 0)
 229 ;               goto out;
 230 ;       odd = 1 & (unsigned long) buff;
 231 ;       if (odd) {
 232 ;#ifdef __LITTLE_ENDIAN
 233 ;               result += (*buff << 8);
 234 ;#else
 235 ;               result = *buff;
 236 ;#endif
 237 ;               len--;
 238 ;               buff++;
 239 ;       }
 240 ;       count = len >> 1;               /* nr of 16-bit words.. */
 241 ;       if (count) {
 242 ;               if (2 & (unsigned long) buff) {
 243 ;                       result += *(unsigned short *) buff;
 244 ;                       count--;
 245 ;                       len -= 2;
 246 ;                       buff += 2;
 247 ;               }
 248 ;               count >>= 1;            /* nr of 32-bit words.. */
 249 ;               if (count) {
 250 ;                       unsigned int carry = 0;
 251 ;                       do {
 252 ;                               unsigned int w = *(unsigned int *) buff;
 253 ;                               count--;
 254 ;                               buff += 4;
 255 ;                               result += carry;
 256 ;                               result += w;
 257 ;                               carry = (w > result);
 258 ;                       } while (count);
 259 ;                       result += carry;
 260 ;                       result = (result & 0xffff) + (result >> 16);
 261 ;               }
 262 ;               if (len & 2) {
 263 ;                       result += *(unsigned short *) buff;
 264 ;                       buff += 2;
 265 ;               }
 266 ;       }
 267 ;       if (len & 1)
 268 ;#ifdef __LITTLE_ENDIAN
 269 ;               result += *buff;
 270 ;#else
 271 ;               result += (*buff << 8);
 272 ;#endif
 273 ;       result = (result & 0xffff) + (result >> 16);
 274 ;       /* add up carry.. */
 275 ;       result = (result & 0xffff) + (result >> 16);
 276 ;       if (odd)
 277 ;               result = ((result >> 8) & 0xff) | ((result & 0xff) << 8);
 278 ;out:
 279 ;       return result;
 280 ;}
 281 ;
 282 ; A4:   buff
 283 ; B4:   len
 284 ; return checksum in A4
 285 ;
 286
 287 ENTRY(do_csum)
 288            CMPGT   .L2     B4,0,B0
 289    [!B0]   BNOP    .S1     L26,3
 290            EXTU    .S1     A4,31,31,A0
 291
 292            MV      .L1     A0,A3
 293 ||         MV      .S1X    B3,A5
 294 ||         MV      .L2     B4,B3
 295 ||         ZERO    .D1     A1
 296
 297 #ifdef CONFIG_CPU_BIG_ENDIAN
 298    [A0]    SUB     .L2     B3,1,B3
 299 || [A0]    LDBU    .D1T1   *A4++,A1
 300 #else
 301    [!A0]   BNOP    .S1     L21,5
 302 || [A0]    LDBU    .D1T1   *A4++,A0
 303            SUB     .L2     B3,1,B3
 304 ||         SHL     .S1     A0,8,A1
 305 L21:
 306 #endif
 307            SHR     .S2     B3,1,B0
 308    [!B0]   BNOP    .S1     L24,3
 309            MVK     .L1     2,A0
 310            AND     .L1     A4,A0,A0
 311
 312    [!A0]   BNOP    .S1     L22,5
 313 || [A0]    LDHU    .D1T1   *A4++,A0
 314            SUB     .L2     B0,1,B0
 315 ||         SUB     .S2     B3,2,B3
 316 ||         ADD     .L1     A0,A1,A1
 317 L22:
 318            SHR     .S2     B0,1,B0
 319 ||         ZERO    .L1     A0
 320
 321    [!B0]   BNOP    .S1     L23,5
 322 || [B0]    MVC     .S2     B0,ILC
 323
 324            SPLOOP  3
 325            SPMASK  L1
 326 ||         MV      .L1     A1,A2
 327 ||         LDW     .D1T1   *A4++,A1
 328
 329            NOP     4
 330            ADD     .L1     A0,A1,A0
 331            ADD     .L1     A2,A0,A2
 332
 333            SPKERNEL 1,2
 334 ||         CMPGTU  .L1     A1,A2,A0
 335
 336            ADD     .L1     A0,A2,A6
 337            EXTU    .S1     A6,16,16,A7
 338            SHRU    .S2X    A6,16,B0
 339            NOP             1
 340            ADD     .L1X    A7,B0,A1
 341 L23:
 342            MVK     .L2     2,B0
 343            AND     .L2     B3,B0,B0
 344    [B0]    LDHU    .D1T1   *A4++,A0
 345            NOP     4
 346    [B0]    ADD     .L1     A0,A1,A1
 347 L24:
 348            EXTU    .S2     B3,31,31,B0
 349 #ifdef CONFIG_CPU_BIG_ENDIAN
 350    [!B0]   BNOP    .S1     L25,4
 351 || [B0]    LDBU    .D1T1   *A4,A0
 352            SHL     .S1     A0,8,A0
 353            ADD     .L1     A0,A1,A1
 354 L25:
 355 #else
 356    [B0]    LDBU    .D1T1   *A4,A0
 357            NOP     4
 358    [B0]    ADD     .L1     A0,A1,A1
 359 #endif
 360            EXTU    .S1     A1,16,16,A0
 361            SHRU    .S2X    A1,16,B0
 362            NOP     1
 363            ADD     .L1X    A0,B0,A0
 364            SHRU    .S1     A0,16,A1
 365            ADD     .L1     A0,A1,A0
 366            EXTU    .S1     A0,16,16,A1
 367            EXTU    .S1     A1,16,24,A2
 368
 369            EXTU    .S1     A1,24,16,A0
 370 ||         MV      .L2X    A3,B0
 371
 372    [B0]    OR      .L1     A0,A2,A1
 373 L26:
 374            NOP     1
 375            BNOP    .S2X    A5,4
 376            MV      .L1     A1,A4
 377 ENDPROC(do_csum)
 378
 379 ;__wsum csum_partial(const void *buff, int len, __wsum wsum)
 380 ;{
 381 ;       unsigned int sum = (__force unsigned int)wsum;
 382 ;       unsigned int result = do_csum(buff, len);
 383 ;
 384 ;       /* add in old sum, and carry.. */
 385 ;       result += sum;
 386 ;       if (sum > result)
 387 ;               result += 1;
 388 ;       return (__force __wsum)result;
 389 ;}
 390 ;
 391 ENTRY(csum_partial)
 392            MV      .L1X    B3,A9
 393 ||         CALLP   .S2     do_csum,B3
 394 ||         MV      .S1     A6,A8
 395            BNOP    .S2X    A9,2
 396            ADD     .L1     A8,A4,A1
 397            CMPGTU  .L1     A8,A1,A0
 398            ADD     .L1     A1,A0,A4
 399 ENDPROC(csum_partial)
 400
 401 ;unsigned short
 402 ;ip_compute_csum(unsigned char *buff, unsigned int len)
 403 ;
 404 ; A4:   buff
 405 ; B4:   len
 406 ; return checksum in A4
 407
 408 ENTRY(ip_compute_csum)
 409            MV      .L1X    B3,A9
 410 ||         CALLP   .S2     do_csum,B3
 411            BNOP    .S2X    A9,3
 412            NOT     .S1     A4,A4
 413            CLR     .S1     A4,16,31,A4
 414 ENDPROC(ip_compute_csum)