sparcv9/gen/memmove.s

/*
 * CDDL HEADER START
 *
 * The contents of this file are subject to the terms of the
 * Common Development and Distribution License (the "License").
 * You may not use this file except in compliance with the License.
 *
 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
 * or http://www.opensolaris.org/os/licensing.
 * See the License for the specific language governing permissions
 * and limitations under the License.
 *
 * When distributing Covered Code, include this CDDL HEADER in each
 * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
 * If applicable, add the following below this CDDL HEADER, with the
 * fields enclosed by brackets "[]" replaced with your own identifying
 * information: Portions Copyright [yyyy] [name of copyright owner]
 *
 * CDDL HEADER END
 */

/*
 * Copyright 2008 Sun Microsystems, Inc.  All rights reserved.
 * Use is subject to license terms.
 */

    .file   "memmove.s"

#include <sys/asm_linkage.h>

    ANSI_PRAGMA_WEAK(memmove,function)

/*
 * memmove(s1, s2, len)
 * Copy s2 to s1, always copy n bytes.
 * For overlapped copies it does the right thing.
 */
    ENTRY(memmove)
    save    %sp, -SA(MINFRAME), %sp     ! not a leaf routine any more
    mov %i0, %l6    ! Save pointer to destination
    cmp %i1, %i0    ! if from address is >= to use forward copy
    bgeu,a  %xcc, 2f    ! else use backward if ...
    cmp %i2, 17     ! delay slot, for small counts copy bytes

    sub %i0, %i1, %i4   ! get difference of two addresses
    cmp %i2, %i4    ! compare size and difference of addresses
    bgu %xcc, ovbc  ! if size is bigger, have to do overlapped copy
    cmp %i2, 17     ! delay slot, for small counts copy bytes
    !
    ! normal, copy forwards
    !
2:  ble %xcc, dbytecp
    andcc   %i1, 3, %i5     ! is src word aligned
    bz,pn   %icc, aldst
    cmp %i5, 2          ! is src half-word aligned
    be,pn   %icc, s2algn
    cmp %i5, 3          ! src is byte aligned
s1algn: ldub    [%i1], %i3      ! move 1 or 3 bytes to align it
    inc 1, %i1
    stb %i3, [%i0]      ! move a byte to align src
    inc 1, %i0
    bne,pn  %icc, s2algn
    dec %i2
    b   ald         ! now go align dest
    andcc   %i0, 3, %i5

s2algn: lduh    [%i1], %i3      ! know src is 2 byte alinged
    inc 2, %i1
    srl %i3, 8, %i4
    stb %i4, [%i0]      ! have to do bytes,
    stb %i3, [%i0 + 1]      ! don't know dst alingment
    inc 2, %i0
    dec 2, %i2

aldst:  andcc   %i0, 3, %i5     ! align the destination address
ald:    bz,pn   %icc, w4cp
    cmp %i5, 2
    bz,pn   %icc, w2cp
    cmp %i5, 3
w3cp:   lduw    [%i1], %i4
    inc 4, %i1
    srl %i4, 24, %i5
    stb %i5, [%i0]
    bne,pt  %icc, w1cp
    inc %i0
    dec 1, %i2
    andn    %i2, 3, %i3     ! i3 is aligned word count
    dec 4, %i3          ! avoid reading beyond tail of src
    sub %i1, %i0, %i1       ! i1 gets the difference

1:  sll %i4, 8, %g1     ! save residual bytes
    lduw    [%i1+%i0], %i4
    deccc   4, %i3
    srl %i4, 24, %i5        ! merge with residual
    or  %i5, %g1, %g1
    st  %g1, [%i0]
    bnz,pt  %xcc, 1b
    inc 4, %i0
    sub %i1, 3, %i1     ! used one byte of last word read
    and %i2, 3, %i2
    b   7f
    inc 4, %i2

w1cp:   srl %i4, 8, %i5
    sth %i5, [%i0]
    inc 2, %i0
    dec 3, %i2
    andn    %i2, 3, %i3
    dec 4, %i3          ! avoid reading beyond tail of src
    sub %i1, %i0, %i1       ! i1 gets the difference

2:  sll %i4, 24, %g1        ! save residual bytes
    lduw    [%i1+%i0], %i4
    deccc   4, %i3
    srl %i4, 8, %i5     ! merge with residual
    or  %i5, %g1, %g1
    st  %g1, [%i0]
    bnz,pt  %xcc, 2b
    inc 4, %i0
    sub %i1, 1, %i1     ! used three bytes of last word read
    and %i2, 3, %i2
    b   7f
    inc 4, %i2

w2cp:   lduw    [%i1], %i4
    inc 4, %i1
    srl %i4, 16, %i5
    sth %i5, [%i0]
    inc 2, %i0
    dec 2, %i2
    andn    %i2, 3, %i3     ! i3 is aligned word count
    dec 4, %i3          ! avoid reading beyond tail of src
    sub %i1, %i0, %i1       ! i1 gets the difference

3:  sll %i4, 16, %g1        ! save residual bytes
    lduw    [%i1+%i0], %i4
    deccc   4, %i3
    srl %i4, 16, %i5        ! merge with residual
    or  %i5, %g1, %g1
    st  %g1, [%i0]
    bnz,pt  %xcc, 3b
    inc 4, %i0
    sub %i1, 2, %i1     ! used two bytes of last word read
    and %i2, 3, %i2
    b   7f
    inc 4, %i2

w4cp:   andn    %i2, 3, %i3     ! i3 is aligned word count
    sub %i1, %i0, %i1       ! i1 gets the difference

1:  lduw    [%i1+%i0], %i4      ! read from address
    deccc   4, %i3          ! decrement count
    st  %i4, [%i0]      ! write at destination address
    bg,pt   %xcc, 1b
    inc 4, %i0          ! increment to address
    b   7f
    and %i2, 3, %i2     ! number of leftover bytes, if any

    !
    ! differenced byte copy, works with any alignment
    !
dbytecp:
    b   7f
    sub %i1, %i0, %i1       ! i1 gets the difference

4:  stb %i4, [%i0]      ! write to address
    inc %i0         ! inc to address
7:  deccc   %i2         ! decrement count
    bge,a   %xcc, 4b        ! loop till done
    ldub    [%i1+%i0], %i4      ! read from address
    ret
    restore %l6, %g0, %o0       ! return pointer to destination

    !
    ! an overlapped copy that must be done "backwards"
    !
ovbc:   add %i1, %i2, %i1       ! get to end of source space
    add %i0, %i2, %i0       ! get to end of destination space
    sub %i1, %i0, %i1       ! i1 gets the difference

5:  dec %i0         ! decrement to address
    ldub    [%i1+%i0], %i3      ! read a byte
    deccc   %i2         ! decrement count
    bg,pt   %xcc, 5b        ! loop until done
    stb %i3, [%i0]      ! write byte
    ret
    restore %l6, %g0, %o0       ! return pointer to destination

    SET_SIZE(memmove)