arch/sparc64/lib/U1memcpy.S

   1 /* U1memcpy.S: UltraSPARC-I/II/IIi/IIe optimized memcpy.
   2  *
   3  * Copyright (C) 1997, 2004 David S. Miller (davem@redhat.com)
   4  * Copyright (C) 1996, 1997, 1998, 1999 Jakub Jelinek (jj@ultra.linux.cz)
   5  */
   6
   7 #ifdef __KERNEL__
   8 #include <asm/visasm.h>
   9 #include <asm/asi.h>
  10 #define GLOBAL_SPARE    g7
  11 #else
  12 #define GLOBAL_SPARE    g5
  13 #define ASI_BLK_P 0xf0
  14 #define FPRS_FEF  0x04
  15 #ifdef MEMCPY_DEBUG
  16 #define VISEntry rd %fprs, %o5; wr %g0, FPRS_FEF, %fprs; \
  17                  clr %g1; clr %g2; clr %g3; subcc %g0, %g0, %g0;
  18 #define VISExit and %o5, FPRS_FEF, %o5; wr %o5, 0x0, %fprs
  19 #else
  20 #define VISEntry rd %fprs, %o5; wr %g0, FPRS_FEF, %fprs
  21 #define VISExit and %o5, FPRS_FEF, %o5; wr %o5, 0x0, %fprs
  22 #endif
  23 #endif
  24
  25 #ifndef EX_LD
  26 #define EX_LD(x)        x
  27 #endif
  28
  29 #ifndef EX_ST
  30 #define EX_ST(x)        x
  31 #endif
  32
  33 #ifndef EX_RETVAL
  34 #define EX_RETVAL(x)    x
  35 #endif
  36
  37 #ifndef LOAD
  38 #define LOAD(type,addr,dest)    type [addr], dest
  39 #endif
  40
  41 #ifndef LOAD_BLK
  42 #define LOAD_BLK(addr,dest)     ldda [addr] ASI_BLK_P, dest
  43 #endif
  44
  45 #ifndef STORE
  46 #define STORE(type,src,addr)    type src, [addr]
  47 #endif
  48
  49 #ifndef STORE_BLK
  50 #define STORE_BLK(src,addr)     stda src, [addr] ASI_BLK_P
  51 #endif
  52
  53 #ifndef FUNC_NAME
  54 #define FUNC_NAME       memcpy
  55 #endif
  56
  57 #ifndef PREAMBLE
  58 #define PREAMBLE
  59 #endif
  60
  61 #ifndef XCC
  62 #define XCC xcc
  63 #endif
  64
  65 #define FREG_FROB(f1, f2, f3, f4, f5, f6, f7, f8, f9)           \
  66         faligndata              %f1, %f2, %f48;                 \
  67         faligndata              %f2, %f3, %f50;                 \
  68         faligndata              %f3, %f4, %f52;                 \
  69         faligndata              %f4, %f5, %f54;                 \
  70         faligndata              %f5, %f6, %f56;                 \
  71         faligndata              %f6, %f7, %f58;                 \
  72         faligndata              %f7, %f8, %f60;                 \
  73         faligndata              %f8, %f9, %f62;
  74
  75 #define MAIN_LOOP_CHUNK(src, dest, fdest, fsrc, len, jmptgt)    \
  76         EX_LD(LOAD_BLK(%src, %fdest));                          \
  77         EX_ST(STORE_BLK(%fsrc, %dest));                         \
  78         add                     %src, 0x40, %src;               \
  79         subcc                   %len, 0x40, %len;               \
  80         be,pn                   %xcc, jmptgt;                   \
  81          add                    %dest, 0x40, %dest;             \
  82
  83 #define LOOP_CHUNK1(src, dest, len, branch_dest)                \
  84         MAIN_LOOP_CHUNK(src, dest, f0,  f48, len, branch_dest)
  85 #define LOOP_CHUNK2(src, dest, len, branch_dest)                \
  86         MAIN_LOOP_CHUNK(src, dest, f16, f48, len, branch_dest)
  87 #define LOOP_CHUNK3(src, dest, len, branch_dest)                \
  88         MAIN_LOOP_CHUNK(src, dest, f32, f48, len, branch_dest)
  89
  90 #define STORE_SYNC(dest, fsrc)                          \
  91         EX_ST(STORE_BLK(%fsrc, %dest));                 \
  92         add                     %dest, 0x40, %dest;
  93
  94 #define STORE_JUMP(dest, fsrc, target)                  \
  95         EX_ST(STORE_BLK(%fsrc, %dest));                 \
  96         add                     %dest, 0x40, %dest;     \
  97         ba,pt                   %xcc, target;
  98
  99 #define FINISH_VISCHUNK(dest, f0, f1, left)     \
 100         subcc                   %left, 8, %left;\
 101         bl,pn                   %xcc, 95f;      \
 102          faligndata             %f0, %f1, %f48; \
 103         EX_ST(STORE(std, %f48, %dest));         \
 104         add                     %dest, 8, %dest;
 105
 106 #define UNEVEN_VISCHUNK_LAST(dest, f0, f1, left)        \
 107         subcc                   %left, 8, %left;        \
 108         bl,pn                   %xcc, 95f;              \
 109          fsrc1                  %f0, %f1;
 110
 111 #define UNEVEN_VISCHUNK(dest, f0, f1, left)             \
 112         UNEVEN_VISCHUNK_LAST(dest, f0, f1, left)        \
 113         ba,a,pt                 %xcc, 93f;
 114
 115         .register       %g2,#scratch
 116         .register       %g3,#scratch
 117
 118         .text
 119         .align          64
 120
 121         .globl          FUNC_NAME
 122         .type           FUNC_NAME,#function
 123 FUNC_NAME:              /* %o0=dst, %o1=src, %o2=len */
 124         srlx            %o2, 31, %g2
 125         cmp             %g2, 0
 126         tne             %xcc, 5
 127         PREAMBLE
 128         mov             %o0, %o4
 129         cmp             %o2, 0
 130         be,pn           %XCC, 85f
 131          or             %o0, %o1, %o3
 132         cmp             %o2, 16
 133         blu,a,pn        %XCC, 80f
 134          or             %o3, %o2, %o3
 135
 136         cmp             %o2, (5 * 64)
 137         blu,pt          %XCC, 70f
 138          andcc          %o3, 0x7, %g0
 139
 140         /* Clobbers o5/g1/g2/g3/g7/icc/xcc.  */
 141         VISEntry
 142
 143         /* Is 'dst' already aligned on an 64-byte boundary? */
 144         andcc           %o0, 0x3f, %g2
 145         be,pt           %XCC, 2f
 146
 147         /* Compute abs((dst & 0x3f) - 0x40) into %g2.  This is the number
 148          * of bytes to copy to make 'dst' 64-byte aligned.  We pre-
 149          * subtract this from 'len'.
 150          */
 151          sub            %o0, %o1, %GLOBAL_SPARE
 152         sub             %g2, 0x40, %g2
 153         sub             %g0, %g2, %g2
 154         sub             %o2, %g2, %o2
 155         andcc           %g2, 0x7, %g1
 156         be,pt           %icc, 2f
 157          and            %g2, 0x38, %g2
 158
 159 1:      subcc           %g1, 0x1, %g1
 160         EX_LD(LOAD(ldub, %o1 + 0x00, %o3))
 161         EX_ST(STORE(stb, %o3, %o1 + %GLOBAL_SPARE))
 162         bgu,pt          %XCC, 1b
 163          add            %o1, 0x1, %o1
 164
 165         add             %o1, %GLOBAL_SPARE, %o0
 166
 167 2:      cmp             %g2, 0x0
 168         and             %o1, 0x7, %g1
 169         be,pt           %icc, 3f
 170          alignaddr      %o1, %g0, %o1
 171
 172         EX_LD(LOAD(ldd, %o1, %f4))
 173 1:      EX_LD(LOAD(ldd, %o1 + 0x8, %f6))
 174         add             %o1, 0x8, %o1
 175         subcc           %g2, 0x8, %g2
 176         faligndata      %f4, %f6, %f0
 177         EX_ST(STORE(std, %f0, %o0))
 178         be,pn           %icc, 3f
 179          add            %o0, 0x8, %o0
 180
 181         EX_LD(LOAD(ldd, %o1 + 0x8, %f4))
 182         add             %o1, 0x8, %o1
 183         subcc           %g2, 0x8, %g2
 184         faligndata      %f6, %f4, %f0
 185         EX_ST(STORE(std, %f0, %o0))
 186         bne,pt          %icc, 1b
 187          add            %o0, 0x8, %o0
 188
 189         /* Destination is 64-byte aligned.  */
 190 3:
 191         membar            #LoadStore | #StoreStore | #StoreLoad
 192
 193         subcc           %o2, 0x40, %GLOBAL_SPARE
 194         add             %o1, %g1, %g1
 195         andncc          %GLOBAL_SPARE, (0x40 - 1), %GLOBAL_SPARE
 196         srl             %g1, 3, %g2
 197         sub             %o2, %GLOBAL_SPARE, %g3
 198         andn            %o1, (0x40 - 1), %o1
 199         and             %g2, 7, %g2
 200         andncc          %g3, 0x7, %g3
 201         fmovd           %f0, %f2
 202         sub             %g3, 0x8, %g3
 203         sub             %o2, %GLOBAL_SPARE, %o2
 204
 205         add             %g1, %GLOBAL_SPARE, %g1
 206         subcc           %o2, %g3, %o2
 207
 208         EX_LD(LOAD_BLK(%o1, %f0))
 209         add             %o1, 0x40, %o1
 210         add             %g1, %g3, %g1
 211         EX_LD(LOAD_BLK(%o1, %f16))
 212         add             %o1, 0x40, %o1
 213         sub             %GLOBAL_SPARE, 0x80, %GLOBAL_SPARE
 214         EX_LD(LOAD_BLK(%o1, %f32))
 215         add             %o1, 0x40, %o1
 216
 217         /* There are 8 instances of the unrolled loop,
 218          * one for each possible alignment of the
 219          * source buffer.  Each loop instance is 452
 220          * bytes.
 221          */
 222         sll             %g2, 3, %o3
 223         sub             %o3, %g2, %o3
 224         sllx            %o3, 4, %o3
 225         add             %o3, %g2, %o3
 226         sllx            %o3, 2, %g2
 227 1:      rd              %pc, %o3
 228         add             %o3, %lo(1f - 1b), %o3
 229         jmpl            %o3 + %g2, %g0
 230          nop
 231
 232         .align          64
 233 1:      FREG_FROB(f0, f2, f4, f6, f8, f10,f12,f14,f16)
 234         LOOP_CHUNK1(o1, o0, GLOBAL_SPARE, 1f)
 235         FREG_FROB(f16,f18,f20,f22,f24,f26,f28,f30,f32)
 236         LOOP_CHUNK2(o1, o0, GLOBAL_SPARE, 2f)
 237         FREG_FROB(f32,f34,f36,f38,f40,f42,f44,f46,f0)
 238         LOOP_CHUNK3(o1, o0, GLOBAL_SPARE, 3f)
 239         ba,pt           %xcc, 1b+4
 240          faligndata     %f0, %f2, %f48
 241 1:      FREG_FROB(f16,f18,f20,f22,f24,f26,f28,f30,f32)
 242         STORE_SYNC(o0, f48) membar #Sync
 243         FREG_FROB(f32,f34,f36,f38,f40,f42,f44,f46,f0)
 244         STORE_JUMP(o0, f48, 40f) membar #Sync
 245 2:      FREG_FROB(f32,f34,f36,f38,f40,f42,f44,f46,f0)
 246         STORE_SYNC(o0, f48) membar #Sync
 247         FREG_FROB(f0, f2, f4, f6, f8, f10,f12,f14,f16)
 248         STORE_JUMP(o0, f48, 48f) membar #Sync
 249 3:      FREG_FROB(f0, f2, f4, f6, f8, f10,f12,f14,f16)
 250         STORE_SYNC(o0, f48) membar #Sync
 251         FREG_FROB(f16,f18,f20,f22,f24,f26,f28,f30,f32)
 252         STORE_JUMP(o0, f48, 56f) membar #Sync
 253
 254 1:      FREG_FROB(f2, f4, f6, f8, f10,f12,f14,f16,f18)
 255         LOOP_CHUNK1(o1, o0, GLOBAL_SPARE, 1f)
 256         FREG_FROB(f18,f20,f22,f24,f26,f28,f30,f32,f34)
 257         LOOP_CHUNK2(o1, o0, GLOBAL_SPARE, 2f)
 258         FREG_FROB(f34,f36,f38,f40,f42,f44,f46,f0, f2)
 259         LOOP_CHUNK3(o1, o0, GLOBAL_SPARE, 3f)
 260         ba,pt           %xcc, 1b+4
 261          faligndata     %f2, %f4, %f48
 262 1:      FREG_FROB(f18,f20,f22,f24,f26,f28,f30,f32,f34)
 263         STORE_SYNC(o0, f48) membar #Sync
 264         FREG_FROB(f34,f36,f38,f40,f42,f44,f46,f0, f2)
 265         STORE_JUMP(o0, f48, 41f) membar #Sync
 266 2:      FREG_FROB(f34,f36,f38,f40,f42,f44,f46,f0, f2)
 267         STORE_SYNC(o0, f48) membar #Sync
 268         FREG_FROB(f2, f4, f6, f8, f10,f12,f14,f16,f18)
 269         STORE_JUMP(o0, f48, 49f) membar #Sync
 270 3:      FREG_FROB(f2, f4, f6, f8, f10,f12,f14,f16,f18)
 271         STORE_SYNC(o0, f48) membar #Sync
 272         FREG_FROB(f18,f20,f22,f24,f26,f28,f30,f32,f34)
 273         STORE_JUMP(o0, f48, 57f) membar #Sync
 274
 275 1:      FREG_FROB(f4, f6, f8, f10,f12,f14,f16,f18,f20)
 276         LOOP_CHUNK1(o1, o0, GLOBAL_SPARE, 1f)
 277         FREG_FROB(f20,f22,f24,f26,f28,f30,f32,f34,f36)
 278         LOOP_CHUNK2(o1, o0, GLOBAL_SPARE, 2f)
 279         FREG_FROB(f36,f38,f40,f42,f44,f46,f0, f2, f4)
 280         LOOP_CHUNK3(o1, o0, GLOBAL_SPARE, 3f)
 281         ba,pt           %xcc, 1b+4
 282          faligndata     %f4, %f6, %f48
 283 1:      FREG_FROB(f20,f22,f24,f26,f28,f30,f32,f34,f36)
 284         STORE_SYNC(o0, f48) membar #Sync
 285         FREG_FROB(f36,f38,f40,f42,f44,f46,f0, f2, f4)
 286         STORE_JUMP(o0, f48, 42f) membar #Sync
 287 2:      FREG_FROB(f36,f38,f40,f42,f44,f46,f0, f2, f4)
 288         STORE_SYNC(o0, f48) membar #Sync
 289         FREG_FROB(f4, f6, f8, f10,f12,f14,f16,f18,f20)
 290         STORE_JUMP(o0, f48, 50f) membar #Sync
 291 3:      FREG_FROB(f4, f6, f8, f10,f12,f14,f16,f18,f20)
 292         STORE_SYNC(o0, f48) membar #Sync
 293         FREG_FROB(f20,f22,f24,f26,f28,f30,f32,f34,f36)
 294         STORE_JUMP(o0, f48, 58f) membar #Sync
 295
 296 1:      FREG_FROB(f6, f8, f10,f12,f14,f16,f18,f20,f22)
 297         LOOP_CHUNK1(o1, o0, GLOBAL_SPARE, 1f)
 298         FREG_FROB(f22,f24,f26,f28,f30,f32,f34,f36,f38)
 299         LOOP_CHUNK2(o1, o0, GLOBAL_SPARE, 2f)
 300         FREG_FROB(f38,f40,f42,f44,f46,f0, f2, f4, f6)
 301         LOOP_CHUNK3(o1, o0, GLOBAL_SPARE, 3f)
 302         ba,pt           %xcc, 1b+4
 303          faligndata     %f6, %f8, %f48
 304 1:      FREG_FROB(f22,f24,f26,f28,f30,f32,f34,f36,f38)
 305         STORE_SYNC(o0, f48) membar #Sync
 306         FREG_FROB(f38,f40,f42,f44,f46,f0, f2, f4, f6)
 307         STORE_JUMP(o0, f48, 43f) membar #Sync
 308 2:      FREG_FROB(f38,f40,f42,f44,f46,f0, f2, f4, f6)
 309         STORE_SYNC(o0, f48) membar #Sync
 310         FREG_FROB(f6, f8, f10,f12,f14,f16,f18,f20,f22)
 311         STORE_JUMP(o0, f48, 51f) membar #Sync
 312 3:      FREG_FROB(f6, f8, f10,f12,f14,f16,f18,f20,f22)
 313         STORE_SYNC(o0, f48) membar #Sync
 314         FREG_FROB(f22,f24,f26,f28,f30,f32,f34,f36,f38)
 315         STORE_JUMP(o0, f48, 59f) membar #Sync
 316
 317 1:      FREG_FROB(f8, f10,f12,f14,f16,f18,f20,f22,f24)
 318         LOOP_CHUNK1(o1, o0, GLOBAL_SPARE, 1f)
 319         FREG_FROB(f24,f26,f28,f30,f32,f34,f36,f38,f40)
 320         LOOP_CHUNK2(o1, o0, GLOBAL_SPARE, 2f)
 321         FREG_FROB(f40,f42,f44,f46,f0, f2, f4, f6, f8)
 322         LOOP_CHUNK3(o1, o0, GLOBAL_SPARE, 3f)
 323         ba,pt           %xcc, 1b+4
 324          faligndata     %f8, %f10, %f48
 325 1:      FREG_FROB(f24,f26,f28,f30,f32,f34,f36,f38,f40)
 326         STORE_SYNC(o0, f48) membar #Sync
 327         FREG_FROB(f40,f42,f44,f46,f0, f2, f4, f6, f8)
 328         STORE_JUMP(o0, f48, 44f) membar #Sync
 329 2:      FREG_FROB(f40,f42,f44,f46,f0, f2, f4, f6, f8)
 330         STORE_SYNC(o0, f48) membar #Sync
 331         FREG_FROB(f8, f10,f12,f14,f16,f18,f20,f22,f24)
 332         STORE_JUMP(o0, f48, 52f) membar #Sync
 333 3:      FREG_FROB(f8, f10,f12,f14,f16,f18,f20,f22,f24)
 334         STORE_SYNC(o0, f48) membar #Sync
 335         FREG_FROB(f24,f26,f28,f30,f32,f34,f36,f38,f40)
 336         STORE_JUMP(o0, f48, 60f) membar #Sync
 337
 338 1:      FREG_FROB(f10,f12,f14,f16,f18,f20,f22,f24,f26)
 339         LOOP_CHUNK1(o1, o0, GLOBAL_SPARE, 1f)
 340         FREG_FROB(f26,f28,f30,f32,f34,f36,f38,f40,f42)
 341         LOOP_CHUNK2(o1, o0, GLOBAL_SPARE, 2f)
 342         FREG_FROB(f42,f44,f46,f0, f2, f4, f6, f8, f10)
 343         LOOP_CHUNK3(o1, o0, GLOBAL_SPARE, 3f)
 344         ba,pt           %xcc, 1b+4
 345          faligndata     %f10, %f12, %f48
 346 1:      FREG_FROB(f26,f28,f30,f32,f34,f36,f38,f40,f42)
 347         STORE_SYNC(o0, f48) membar #Sync
 348         FREG_FROB(f42,f44,f46,f0, f2, f4, f6, f8, f10)
 349         STORE_JUMP(o0, f48, 45f) membar #Sync
 350 2:      FREG_FROB(f42,f44,f46,f0, f2, f4, f6, f8, f10)
 351         STORE_SYNC(o0, f48) membar #Sync
 352         FREG_FROB(f10,f12,f14,f16,f18,f20,f22,f24,f26)
 353         STORE_JUMP(o0, f48, 53f) membar #Sync
 354 3:      FREG_FROB(f10,f12,f14,f16,f18,f20,f22,f24,f26)
 355         STORE_SYNC(o0, f48) membar #Sync
 356         FREG_FROB(f26,f28,f30,f32,f34,f36,f38,f40,f42)
 357         STORE_JUMP(o0, f48, 61f) membar #Sync
 358
 359 1:      FREG_FROB(f12,f14,f16,f18,f20,f22,f24,f26,f28)
 360         LOOP_CHUNK1(o1, o0, GLOBAL_SPARE, 1f)
 361         FREG_FROB(f28,f30,f32,f34,f36,f38,f40,f42,f44)
 362         LOOP_CHUNK2(o1, o0, GLOBAL_SPARE, 2f)
 363         FREG_FROB(f44,f46,f0, f2, f4, f6, f8, f10,f12)
 364         LOOP_CHUNK3(o1, o0, GLOBAL_SPARE, 3f)
 365         ba,pt           %xcc, 1b+4
 366          faligndata     %f12, %f14, %f48
 367 1:      FREG_FROB(f28,f30,f32,f34,f36,f38,f40,f42,f44)
 368         STORE_SYNC(o0, f48) membar #Sync
 369         FREG_FROB(f44,f46,f0, f2, f4, f6, f8, f10,f12)
 370         STORE_JUMP(o0, f48, 46f) membar #Sync
 371 2:      FREG_FROB(f44,f46,f0, f2, f4, f6, f8, f10,f12)
 372         STORE_SYNC(o0, f48) membar #Sync
 373         FREG_FROB(f12,f14,f16,f18,f20,f22,f24,f26,f28)
 374         STORE_JUMP(o0, f48, 54f) membar #Sync
 375 3:      FREG_FROB(f12,f14,f16,f18,f20,f22,f24,f26,f28)
 376         STORE_SYNC(o0, f48) membar #Sync
 377         FREG_FROB(f28,f30,f32,f34,f36,f38,f40,f42,f44)
 378         STORE_JUMP(o0, f48, 62f) membar #Sync
 379
 380 1:      FREG_FROB(f14,f16,f18,f20,f22,f24,f26,f28,f30)
 381         LOOP_CHUNK1(o1, o0, GLOBAL_SPARE, 1f)
 382         FREG_FROB(f30,f32,f34,f36,f38,f40,f42,f44,f46)
 383         LOOP_CHUNK2(o1, o0, GLOBAL_SPARE, 2f)
 384         FREG_FROB(f46,f0, f2, f4, f6, f8, f10,f12,f14)
 385         LOOP_CHUNK3(o1, o0, GLOBAL_SPARE, 3f)
 386         ba,pt           %xcc, 1b+4
 387          faligndata     %f14, %f16, %f48
 388 1:      FREG_FROB(f30,f32,f34,f36,f38,f40,f42,f44,f46)
 389         STORE_SYNC(o0, f48) membar #Sync
 390         FREG_FROB(f46,f0, f2, f4, f6, f8, f10,f12,f14)
 391         STORE_JUMP(o0, f48, 47f) membar #Sync
 392 2:      FREG_FROB(f46,f0, f2, f4, f6, f8, f10,f12,f14)
 393         STORE_SYNC(o0, f48) membar #Sync
 394         FREG_FROB(f14,f16,f18,f20,f22,f24,f26,f28,f30)
 395         STORE_JUMP(o0, f48, 55f) membar #Sync
 396 3:      FREG_FROB(f14,f16,f18,f20,f22,f24,f26,f28,f30)
 397         STORE_SYNC(o0, f48) membar #Sync
 398         FREG_FROB(f30,f32,f34,f36,f38,f40,f42,f44,f46)
 399         STORE_JUMP(o0, f48, 63f) membar #Sync
 400
 401 40:     FINISH_VISCHUNK(o0, f0,  f2,  g3)
 402 41:     FINISH_VISCHUNK(o0, f2,  f4,  g3)
 403 42:     FINISH_VISCHUNK(o0, f4,  f6,  g3)
 404 43:     FINISH_VISCHUNK(o0, f6,  f8,  g3)
 405 44:     FINISH_VISCHUNK(o0, f8,  f10, g3)
 406 45:     FINISH_VISCHUNK(o0, f10, f12, g3)
 407 46:     FINISH_VISCHUNK(o0, f12, f14, g3)
 408 47:     UNEVEN_VISCHUNK(o0, f14, f0,  g3)
 409 48:     FINISH_VISCHUNK(o0, f16, f18, g3)
 410 49:     FINISH_VISCHUNK(o0, f18, f20, g3)
 411 50:     FINISH_VISCHUNK(o0, f20, f22, g3)
 412 51:     FINISH_VISCHUNK(o0, f22, f24, g3)
 413 52:     FINISH_VISCHUNK(o0, f24, f26, g3)
 414 53:     FINISH_VISCHUNK(o0, f26, f28, g3)
 415 54:     FINISH_VISCHUNK(o0, f28, f30, g3)
 416 55:     UNEVEN_VISCHUNK(o0, f30, f0,  g3)
 417 56:     FINISH_VISCHUNK(o0, f32, f34, g3)
 418 57:     FINISH_VISCHUNK(o0, f34, f36, g3)
 419 58:     FINISH_VISCHUNK(o0, f36, f38, g3)
 420 59:     FINISH_VISCHUNK(o0, f38, f40, g3)
 421 60:     FINISH_VISCHUNK(o0, f40, f42, g3)
 422 61:     FINISH_VISCHUNK(o0, f42, f44, g3)
 423 62:     FINISH_VISCHUNK(o0, f44, f46, g3)
 424 63:     UNEVEN_VISCHUNK_LAST(o0, f46, f0,  g3)
 425
 426 93:     EX_LD(LOAD(ldd, %o1, %f2))
 427         add             %o1, 8, %o1
 428         subcc           %g3, 8, %g3
 429         faligndata      %f0, %f2, %f8
 430         EX_ST(STORE(std, %f8, %o0))
 431         bl,pn           %xcc, 95f
 432          add            %o0, 8, %o0
 433         EX_LD(LOAD(ldd, %o1, %f0))
 434         add             %o1, 8, %o1
 435         subcc           %g3, 8, %g3
 436         faligndata      %f2, %f0, %f8
 437         EX_ST(STORE(std, %f8, %o0))
 438         bge,pt          %xcc, 93b
 439          add            %o0, 8, %o0
 440
 441 95:     brz,pt          %o2, 2f
 442          mov            %g1, %o1
 443
 444 1:      EX_LD(LOAD(ldub, %o1, %o3))
 445         add             %o1, 1, %o1
 446         subcc           %o2, 1, %o2
 447         EX_ST(STORE(stb, %o3, %o0))
 448         bne,pt          %xcc, 1b
 449          add            %o0, 1, %o0
 450
 451 2:      membar          #StoreLoad | #StoreStore
 452         VISExit
 453         retl
 454          mov            EX_RETVAL(%o4), %o0
 455
 456         .align          64
 457 70:     /* 16 < len <= (5 * 64) */
 458         bne,pn          %XCC, 75f
 459          sub            %o0, %o1, %o3
 460
 461 72:     andn            %o2, 0xf, %GLOBAL_SPARE
 462         and             %o2, 0xf, %o2
 463 1:      EX_LD(LOAD(ldx, %o1 + 0x00, %o5))
 464         EX_LD(LOAD(ldx, %o1 + 0x08, %g1))
 465         subcc           %GLOBAL_SPARE, 0x10, %GLOBAL_SPARE
 466         EX_ST(STORE(stx, %o5, %o1 + %o3))
 467         add             %o1, 0x8, %o1
 468         EX_ST(STORE(stx, %g1, %o1 + %o3))
 469         bgu,pt          %XCC, 1b
 470          add            %o1, 0x8, %o1
 471 73:     andcc           %o2, 0x8, %g0
 472         be,pt           %XCC, 1f
 473          nop
 474         EX_LD(LOAD(ldx, %o1, %o5))
 475         sub             %o2, 0x8, %o2
 476         EX_ST(STORE(stx, %o5, %o1 + %o3))
 477         add             %o1, 0x8, %o1
 478 1:      andcc           %o2, 0x4, %g0
 479         be,pt           %XCC, 1f
 480          nop
 481         EX_LD(LOAD(lduw, %o1, %o5))
 482         sub             %o2, 0x4, %o2
 483         EX_ST(STORE(stw, %o5, %o1 + %o3))
 484         add             %o1, 0x4, %o1
 485 1:      cmp             %o2, 0
 486         be,pt           %XCC, 85f
 487          nop
 488         ba,pt           %xcc, 90f
 489          nop
 490
 491 75:     andcc           %o0, 0x7, %g1
 492         sub             %g1, 0x8, %g1
 493         be,pn           %icc, 2f
 494          sub            %g0, %g1, %g1
 495         sub             %o2, %g1, %o2
 496
 497 1:      EX_LD(LOAD(ldub, %o1, %o5))
 498         subcc           %g1, 1, %g1
 499         EX_ST(STORE(stb, %o5, %o1 + %o3))
 500         bgu,pt          %icc, 1b
 501          add            %o1, 1, %o1
 502
 503 2:      add             %o1, %o3, %o0
 504         andcc           %o1, 0x7, %g1
 505         bne,pt          %icc, 8f
 506          sll            %g1, 3, %g1
 507
 508         cmp             %o2, 16
 509         bgeu,pt         %icc, 72b
 510          nop
 511         ba,a,pt         %xcc, 73b
 512
 513 8:      mov             64, %o3
 514         andn            %o1, 0x7, %o1
 515         EX_LD(LOAD(ldx, %o1, %g2))
 516         sub             %o3, %g1, %o3
 517         andn            %o2, 0x7, %GLOBAL_SPARE
 518         sllx            %g2, %g1, %g2
 519 1:      EX_LD(LOAD(ldx, %o1 + 0x8, %g3))
 520         subcc           %GLOBAL_SPARE, 0x8, %GLOBAL_SPARE
 521         add             %o1, 0x8, %o1
 522         srlx            %g3, %o3, %o5
 523         or              %o5, %g2, %o5
 524         EX_ST(STORE(stx, %o5, %o0))
 525         add             %o0, 0x8, %o0
 526         bgu,pt          %icc, 1b
 527          sllx           %g3, %g1, %g2
 528
 529         srl             %g1, 3, %g1
 530         andcc           %o2, 0x7, %o2
 531         be,pn           %icc, 85f
 532          add            %o1, %g1, %o1
 533         ba,pt           %xcc, 90f
 534          sub            %o0, %o1, %o3
 535
 536         .align          64
 537 80:     /* 0 < len <= 16 */
 538         andcc           %o3, 0x3, %g0
 539         bne,pn          %XCC, 90f
 540          sub            %o0, %o1, %o3
 541
 542 1:      EX_LD(LOAD(lduw, %o1, %g1))
 543         subcc           %o2, 4, %o2
 544         EX_ST(STORE(stw, %g1, %o1 + %o3))
 545         bgu,pt          %XCC, 1b
 546          add            %o1, 4, %o1
 547
 548 85:     retl
 549          mov            EX_RETVAL(%o4), %o0
 550
 551         .align          32
 552 90:     EX_LD(LOAD(ldub, %o1, %g1))
 553         subcc           %o2, 1, %o2
 554         EX_ST(STORE(stb, %g1, %o1 + %o3))
 555         bgu,pt          %XCC, 90b
 556          add            %o1, 1, %o1
 557         retl
 558          mov            EX_RETVAL(%o4), %o0
 559
 560         .size           FUNC_NAME, .-FUNC_NAME