memcpy.s

glibc 2.9,最新版的C语言库函数

    mtcrf   0x01,11
    cmpldi  cr6,9,4
    mtctr   8
    ld      7,8(5)   /* pre load 2nd full doubleword.  */
    bge     cr0, L(du4_do)
    blt     cr5, L(du1_do)
    beq     cr5, L(du2_do)
    b       L(du3_do) 
       
    .align 4
L(du1_do):
    bf      30,L(du1_1dw)

    /* there are at least two DWs to copy */
    sldi     0,6, 8
    srdi     8,7, 64-8
    or      0,0,8
    ld      6,16(5)
    std     0,0(4)
    sldi     0,7, 8
    srdi     8,6, 64-8
    or      0,0,8
    ld      7,24(5)
    std     0,8(4)
    addi    4,4,16
    addi    5,5,32
    blt     cr6,L(du1_fini)  /* if total DWs = 3, then bypass loop */
    bf      31,L(du1_loop)
    /* there is a third DW to copy */
    sldi     0,6, 8
    srdi     8,7, 64-8
    or      0,0,8
    std     0,0(4)
    mr      6,7
    ld      7,0(5)
    addi    5,5,8
    addi    4,4,8
    beq     cr6,L(du1_fini)  /* if total DWs = 4, then bypass loop */
    b       L(du1_loop)
    .align 4
L(du1_1dw):
    sldi     0,6, 8
    srdi     8,7, 64-8
    addi    5,5,16
    or      0,0,8
    bf      31,L(du1_loop)
    mr      6,7
    ld      7,0(5)
    addi    5,5,8
    std     0,0(4)
    addi    4,4,8
    .align 4
/* copy 32 bytes at a time */
L(du1_loop):
    sldi   0,6, 8
    srdi   8,7, 64-8
    or    0,0,8
    ld    6,0(5)
    std   0,0(4)
    sldi   0,7, 8
    srdi   8,6, 64-8
    or    0,0,8
    ld    7,8(5)
    std   0,8(4)
    sldi   0,6, 8
    srdi   8,7, 64-8
    or    0,0,8
    ld    6,16(5)
    std   0,16(4)
    sldi   0,7, 8
    srdi   8,6, 64-8
    or    0,0,8
    ld    7,24(5)
    std   0,24(4)
    addi  5,5,32
    addi  4,4,32
    bdnz+ L(du1_loop)
    .align 4
L(du1_fini):
    /* calculate and store the final DW */
    sldi   0,6, 8
    srdi   8,7, 64-8
    or    0,0,8  
    std   0,0(4)
    b     L(du_done)

    .align 4
L(du2_do):
    bf      30,L(du2_1dw)

    /* there are at least two DWs to copy */
    sldi     0,6, 16
    srdi     8,7, 64-16
    or      0,0,8
    ld      6,16(5)
    std     0,0(4)
    sldi     0,7, 16
    srdi     8,6, 64-16
    or      0,0,8
    ld      7,24(5)
    std     0,8(4)
    addi    4,4,16
    addi    5,5,32
    blt     cr6,L(du2_fini)  /* if total DWs = 3, then bypass loop */
    bf      31,L(du2_loop)
    /* there is a third DW to copy */
    sldi     0,6, 16
    srdi     8,7, 64-16
    or      0,0,8
    std     0,0(4)
    mr      6,7
    ld      7,0(5)
    addi    5,5,8
    addi    4,4,8
    beq     cr6,L(du2_fini)  /* if total DWs = 4, then bypass loop */
    b       L(du2_loop)
    .align 4
L(du2_1dw):
    sldi     0,6, 16
    srdi     8,7, 64-16
    addi    5,5,16
    or      0,0,8
    bf      31,L(du2_loop)
    mr      6,7
    ld      7,0(5)
    addi    5,5,8
    std     0,0(4)
    addi    4,4,8
    .align 4
/* copy 32 bytes at a time */
L(du2_loop):
    sldi   0,6, 16
    srdi   8,7, 64-16
    or    0,0,8
    ld    6,0(5)
    std   0,0(4)
    sldi   0,7, 16
    srdi   8,6, 64-16
    or    0,0,8
    ld    7,8(5)
    std   0,8(4)
    sldi   0,6, 16
    srdi   8,7, 64-16
    or    0,0,8
    ld    6,16(5)
    std   0,16(4)
    sldi   0,7, 16
    srdi   8,6, 64-16
    or    0,0,8
    ld    7,24(5)
    std   0,24(4)
    addi  5,5,32
    addi  4,4,32
    bdnz+ L(du2_loop)
    .align 4
L(du2_fini):
    /* calculate and store the final DW */
    sldi   0,6, 16
    srdi   8,7, 64-16
    or    0,0,8  
    std   0,0(4)
    b     L(du_done)

    .align 4
L(du3_do):
    bf      30,L(du3_1dw)

    /* there are at least two DWs to copy */
    sldi     0,6, 24
    srdi     8,7, 64-24
    or      0,0,8
    ld      6,16(5)
    std     0,0(4)
    sldi     0,7, 24
    srdi     8,6, 64-24
    or      0,0,8
    ld      7,24(5)
    std     0,8(4)
    addi    4,4,16
    addi    5,5,32
    blt     cr6,L(du3_fini)  /* if total DWs = 3, then bypass loop */
    bf      31,L(du3_loop)
    /* there is a third DW to copy */
    sldi     0,6, 24
    srdi     8,7, 64-24
    or      0,0,8
    std     0,0(4)
    mr      6,7
    ld      7,0(5)
    addi    5,5,8
    addi    4,4,8
    beq     cr6,L(du3_fini)  /* if total DWs = 4, then bypass loop */
    b       L(du3_loop)
    .align 4
L(du3_1dw):
    sldi     0,6, 24
    srdi     8,7, 64-24
    addi    5,5,16
    or      0,0,8
    bf      31,L(du3_loop)
    mr      6,7
    ld      7,0(5)
    addi    5,5,8
    std     0,0(4)
    addi    4,4,8
    .align 4
/* copy 32 bytes at a time */
L(du3_loop):
    sldi   0,6, 24
    srdi   8,7, 64-24
    or    0,0,8
    ld    6,0(5)
    std   0,0(4)
    sldi   0,7, 24
    srdi   8,6, 64-24
    or    0,0,8
    ld    7,8(5)
    std   0,8(4)
    sldi   0,6, 24
    srdi   8,7, 64-24
    or    0,0,8
    ld    6,16(5)
    std   0,16(4)
    sldi   0,7, 24
    srdi   8,6, 64-24
    or    0,0,8
    ld    7,24(5)
    std   0,24(4)
    addi  5,5,32
    addi  4,4,32
    bdnz+ L(du3_loop)
    .align 4
L(du3_fini):
    /* calculate and store the final DW */
    sldi   0,6, 24
    srdi   8,7, 64-24
    or    0,0,8  
    std   0,0(4)
    b     L(du_done)

    .align 4
L(du4_do):
    cmpldi  cr5, 10, 6
    beq     cr0, L(du4_dox)
    blt     cr5, L(du5_do)
    beq     cr5, L(du6_do)
    b       L(du7_do)
L(du4_dox):
    bf      30,L(du4_1dw)

    /* there are at least two DWs to copy */
    sldi     0,6, 32
    srdi     8,7, 64-32
    or      0,0,8
    ld      6,16(5)
    std     0,0(4)
    sldi     0,7, 32
    srdi     8,6, 64-32
    or      0,0,8
    ld      7,24(5)
    std     0,8(4)
    addi    4,4,16
    addi    5,5,32
    blt     cr6,L(du4_fini)  /* if total DWs = 3, then bypass loop */
    bf      31,L(du4_loop)
    /* there is a third DW to copy */
    sldi     0,6, 32
    srdi     8,7, 64-32
    or      0,0,8
    std     0,0(4)
    mr      6,7
    ld      7,0(5)
    addi    5,5,8
    addi    4,4,8
    beq     cr6,L(du4_fini)  /* if total DWs = 4, then bypass loop */
    b       L(du4_loop)
    .align 4
L(du4_1dw):
    sldi     0,6, 32
    srdi     8,7, 64-32
    addi    5,5,16
    or      0,0,8
    bf      31,L(du4_loop)
    mr      6,7
    ld      7,0(5)
    addi    5,5,8
    std     0,0(4)
    addi    4,4,8
    .align 4
/* copy 32 bytes at a time */
L(du4_loop):
    sldi   0,6, 32
    srdi   8,7, 64-32
    or    0,0,8
    ld    6,0(5)
    std   0,0(4)
    sldi   0,7, 32
    srdi   8,6, 64-32
    or    0,0,8
    ld    7,8(5)
    std   0,8(4)
    sldi   0,6, 32
    srdi   8,7, 64-32
    or    0,0,8
    ld    6,16(5)
    std   0,16(4)
    sldi   0,7, 32
    srdi   8,6, 64-32
    or    0,0,8
    ld    7,24(5)
    std   0,24(4)
    addi  5,5,32
    addi  4,4,32
    bdnz+ L(du4_loop)
    .align 4
L(du4_fini):
    /* calculate and store the final DW */
    sldi   0,6, 32
    srdi   8,7, 64-32
    or    0,0,8  
    std   0,0(4)
    b     L(du_done)

    .align 4
L(du5_do):
    bf      30,L(du5_1dw)

    /* there are at least two DWs to copy */
    sldi     0,6, 40
    srdi     8,7, 64-40
    or      0,0,8
    ld      6,16(5)
    std     0,0(4)
    sldi     0,7, 40
    srdi     8,6, 64-40
    or      0,0,8
    ld      7,24(5)
    std     0,8(4)
    addi    4,4,16
    addi    5,5,32
    blt     cr6,L(du5_fini)  /* if total DWs = 3, then bypass loop */
    bf      31,L(du5_loop)
    /* there is a third DW to copy */
    sldi     0,6, 40
    srdi     8,7, 64-40
    or      0,0,8
    std     0,0(4)
    mr      6,7
    ld      7,0(5)
    addi    5,5,8
    addi    4,4,8
    beq     cr6,L(du5_fini)  /* if total DWs = 4, then bypass loop */
    b       L(du5_loop)
    .align 4
L(du5_1dw):
    sldi     0,6, 40
    srdi     8,7, 64-40
    addi    5,5,16
    or      0,0,8
    bf      31,L(du5_loop)
    mr      6,7
    ld      7,0(5)
    addi    5,5,8
    std     0,0(4)
    addi    4,4,8
    .align 4
/* copy 32 bytes at a time */
L(du5_loop):
    sldi   0,6, 40
    srdi   8,7, 64-40
    or    0,0,8
    ld    6,0(5)
    std   0,0(4)
    sldi   0,7, 40
    srdi   8,6, 64-40
    or    0,0,8
    ld    7,8(5)
    std   0,8(4)
    sldi   0,6, 40
    srdi   8,7, 64-40
    or    0,0,8
    ld    6,16(5)
    std   0,16(4)
    sldi   0,7, 40
    srdi   8,6, 64-40
    or    0,0,8
    ld    7,24(5)
    std   0,24(4)
    addi  5,5,32
    addi  4,4,32
    bdnz+ L(du5_loop)
    .align 4
L(du5_fini):
    /* calculate and store the final DW */
    sldi   0,6, 40
    srdi   8,7, 64-40
    or    0,0,8  
    std   0,0(4)
    b     L(du_done)

    .align 4
L(du6_do):
    bf      30,L(du6_1dw)

    /* there are at least two DWs to copy */
    sldi     0,6, 48
    srdi     8,7, 64-48
    or      0,0,8
    ld      6,16(5)
    std     0,0(4)
    sldi     0,7, 48
    srdi     8,6, 64-48
    or      0,0,8
    ld      7,24(5)
    std     0,8(4)
    addi    4,4,16
    addi    5,5,32
    blt     cr6,L(du6_fini)  /* if total DWs = 3, then bypass loop */
    bf      31,L(du6_loop)
    /* there is a third DW to copy */
    sldi     0,6, 48
    srdi     8,7, 64-48
    or      0,0,8
    std     0,0(4)
    mr      6,7
    ld      7,0(5)
    addi    5,5,8
    addi    4,4,8
    beq     cr6,L(du6_fini)  /* if total DWs = 4, then bypass loop */
    b       L(du6_loop)
    .align 4
L(du6_1dw):
    sldi     0,6, 48
    srdi     8,7, 64-48
    addi    5,5,16
    or      0,0,8
    bf      31,L(du6_loop)
    mr      6,7
    ld      7,0(5)
    addi    5,5,8
    std     0,0(4)
    addi    4,4,8
    .align 4
/* copy 32 bytes at a time */
L(du6_loop):
    sldi   0,6, 48
    srdi   8,7, 64-48
    or    0,0,8
    ld    6,0(5)
    std   0,0(4)
    sldi   0,7, 48
    srdi   8,6, 64-48
    or    0,0,8
    ld    7,8(5)
    std   0,8(4)
    sldi   0,6, 48
    srdi   8,7, 64-48
    or    0,0,8
    ld    6,16(5)
    std   0,16(4)
    sldi   0,7, 48
    srdi   8,6, 64-48
    or    0,0,8
    ld    7,24(5)
    std   0,24(4)
    addi  5,5,32
    addi  4,4,32
    bdnz+ L(du6_loop)
    .align 4
L(du6_fini):
    /* calculate and store the final DW */
    sldi   0,6, 48
    srdi   8,7, 64-48
    or    0,0,8  
    std   0,0(4)
    b     L(du_done)

    .align 4
L(du7_do):
    bf      30,L(du7_1dw)

    /* there are at least two DWs to copy */
    sldi     0,6, 56
    srdi     8,7, 64-56
    or      0,0,8
    ld      6,16(5)
    std     0,0(4)
    sldi     0,7, 56
    srdi     8,6, 64-56
    or      0,0,8
    ld      7,24(5)
    std     0,8(4)
    addi    4,4,16
    addi    5,5,32
    blt     cr6,L(du7_fini)  /* if total DWs = 3, then bypass loop */
    bf      31,L(du7_loop)
    /* there is a third DW to copy */
    sldi     0,6, 56
    srdi     8,7, 64-56
    or      0,0,8
    std     0,0(4)
    mr      6,7
    ld      7,0(5)
    addi    5,5,8
    addi    4,4,8
    beq     cr6,L(du7_fini)  /* if total DWs = 4, then bypass loop */
    b       L(du7_loop)
    .align 4
L(du7_1dw):
    sldi     0,6, 56
    srdi     8,7, 64-56
    addi    5,5,16
    or      0,0,8
    bf      31,L(du7_loop)
    mr      6,7
    ld      7,0(5)
    addi    5,5,8
    std     0,0(4)
    addi    4,4,8
    .align 4
/* copy 32 bytes at a time */
L(du7_loop):
    sldi   0,6, 56
    srdi   8,7, 64-56
    or    0,0,8
    ld    6,0(5)
    std   0,0(4)
    sldi   0,7, 56
    srdi   8,6, 64-56
    or    0,0,8
    ld    7,8(5)
    std   0,8(4)
    sldi   0,6, 56
    srdi   8,7, 64-56
    or    0,0,8
    ld    6,16(5)
    std   0,16(4)
    sldi   0,7, 56
    srdi   8,6, 64-56
    or    0,0,8
    ld    7,24(5)
    std   0,24(4)
    addi  5,5,32
    addi  4,4,32
    bdnz+ L(du7_loop)
    .align 4
L(du7_fini):
    /* calculate and store the final DW */
    sldi   0,6, 56
    srdi   8,7, 64-56
    or    0,0,8  
    std   0,0(4)
    b     L(du_done)
    
    .align 4
L(du_done):
    rldicr 0,31,0,60
    mtcrf 0x01,31
    beq   cr1,0f	/* If the tail is 0 bytes we are done!  */

    add   3,3,0
    add   12,12,0    
/*  At this point we have a tail of 0-7 bytes and we know that the
    destiniation is double word aligned.  */
4:  bf    29,2f
    lwz   6,0(12)
    addi  12,12,4
    stw   6,0(3)
    addi  3,3,4
2:  bf    30,1f
    lhz   6,0(12)
    addi  12,12,2
    sth   6,0(3)
    addi  3,3,2
1:  bf    31,0f
    lbz   6,0(12)
    stb   6,0(3)
0:
  /* Return original dst pointer.  */
    ld 31,-8(1)
    ld 3,-16(1)
    blr
END_GEN_TB (BP_SYM (memcpy),TB_TOCLESS)
libc_hidden_builtin_def (memcpy)