OpenSolaris_b135/lib/libc/sparcv9/gen/strncpy.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 "strncpy.s"
/*
* strncpy(s1, s2)
*
* Copy string s2 to s1, truncating or null-padding to always copy n bytes
* return s1.
*
* Fast assembler language version of the following C-program for strncpy
* which represents the `standard' for the C-library.
*
* char *
* strncpy(char *s1, const char *s2, size_t n)
* {
* char *os1 = s1;
*
* n++;
* while ((--n != 0) && ((*s1++ = *s2++) != '\0'))
* ;
* if (n != 0)
* while (--n != 0)
* *s1++ = '\0';
* return (os1);
* }
*/
#include <sys/asm_linkage.h>
! strncpy works similarly to strcpy, except that n bytes of s2
! are copied to s1. If a null character is reached in s2 yet more
! bytes remain to be copied, strncpy will copy null bytes into
! the destination string.
!
! This implementation works by first aligning the src ptr and
! performing small copies until it is aligned. Then, the string
! is copied based upon destination alignment. (byte, half-word,
! word, etc.)
ENTRY(strncpy)
.align 32
nop ! pad to align loop on 16-byte boundary
subcc %g0, %o2, %g4 ! n = -n, n == 0 ?
bz,pn %ncc, .done ! n == 0, done
add %o1, %o2, %o3 ! src = src + n
andcc %o1, 7, %o4 ! dword aligned ?
bz,pn %ncc, .dwordaligned ! yup
add %o0, %o2, %o2 ! dst = dst + n
sub %o4, 8, %o4 ! bytes until src aligned
.alignsrc:
ldub [%o3 + %g4], %o1 ! src[]
stb %o1, [%o2 + %g4] ! dst[] = src[]
addcc %g4, 1, %g4 ! src++, dst++, n--
bz,pn %ncc, .done ! n == 0, done
tst %o1 ! end of src reached (null byte) ?
bz,a %ncc, .bytepad ! yes, at least one byte to pad here
add %o2, %g4, %o3 ! need single dest pointer for fill
addcc %o4, 1, %o4 ! src aligned now?
bnz,a %ncc, .alignsrc ! no, copy another byte
nop ! pad
nop ! pad
.dwordaligned:
sethi %hi(0x01010101), %o4 ! Alan Mycroft's magic1
add %o2, %g4, %g5 ! dst
or %o4, %lo(0x01010101),%o4! finish loading magic1
and %g5, 3, %g1 ! dst<1:0> to examine offset
sllx %o4, 32, %o1 ! spread magic1
cmp %g1, 1 ! dst offset of 1 or 5
or %o4, %o1, %o4 ! to all 64 bits
sub %o2, 8, %o2 ! adjust for dest pre-incr in cpy loops
be,pn %ncc, .storebyte1241 ! store 1, 2, 4, 1 bytes
sllx %o4, 7, %o5 ! Alan Mycroft's magic2
cmp %g1, 3 ! dst offset of 3 or 7
be,pn %ncc, .storebyte1421 ! store 1, 4, 2, 1 bytes
cmp %g1, 2 ! dst halfword aligned ?
be,pn %ncc, .storehalfword ! yup, store half-word wise
andcc %g5, 7, %g0 ! dst word aligned ?
bnz,pn %ncc, .storeword2 ! yup, store word wise
nop ! ensure loop is 16-byte aligned
.storedword:
ldx [%o3 + %g4], %o1 ! src dword
addcc %g4, 8, %g4 ! n += 8, src += 8, dst += 8
bcs,pn %ncc,.lastword ! if counter wraps, last word
andn %o5, %o1, %g1 ! ~dword & 0x8080808080808080
sub %o1, %o4, %g5 ! dword - 0x0101010101010101
andcc %g5, %g1, %g0 ! ((dword - 0x0101010101010101) & ~dword & 0x8080808080808080)
bz,a,pt %ncc, .storedword ! no zero byte if magic expression == 0
stx %o1, [%o2 + %g4] ! store word to dst (address pre-incremented)
! n has not expired, but src is at the end. we need to push out the
! remaining src bytes and then start padding with null bytes
.zerobyte:
add %o2, %g4, %o3 ! pointer to dest string
srlx %o1, 56, %g1 ! first byte
stb %g1, [%o3] ! store it
andcc %g1, 0xff, %g0 ! end of string ?
movz %ncc, %g0, %o1 ! if so, start padding with null bytes
srlx %o1, 48, %g1 ! second byte
stb %g1, [%o3 + 1] ! store it
andcc %g1, 0xff, %g0 ! end of string ?
movz %ncc, %g0, %o1 ! if so, start padding with null bytes
srlx %o1, 40, %g1 ! third byte
stb %g1, [%o3 + 2] ! store it
andcc %g1, 0xff, %g0 ! end of string ?
movz %ncc, %g0, %o1 ! if so, start padding with null bytes
srlx %o1, 32, %g1 ! fourth byte
stb %g1, [%o3 + 3] ! store it
andcc %g1, 0xff, %g0 ! end of string ?
movz %ncc, %g0, %o1 ! if so, start padding with null bytes
srlx %o1, 24, %g1 ! fifth byte
stb %g1, [%o3 + 4] ! store it
andcc %g1, 0xff, %g0 ! end of string ?
movz %ncc, %g0, %o1 ! if so, start padding with null bytes
srlx %o1, 16, %g1 ! sixth byte
stb %g1, [%o3 + 5] ! store it
andcc %g1, 0xff, %g0 ! end of string ?
movz %ncc, %g0, %o1 ! if so, start padding with null bytes
srlx %o1, 8, %g1 ! seventh byte
stb %g1, [%o3 + 6] ! store it
andcc %g1, 0xff, %g0 ! end of string ?
movz %ncc, %g0, %o1 ! if so, start padding with null bytes
stb %o1, [%o3 + 7] ! store eighth byte
addcc %g4, 16, %g0 ! number of pad bytes < 16 ?
bcs,pn %ncc, .bytepad ! yes, do simple byte wise fill
add %o3, 8, %o3 ! dst += 8
andcc %o3, 7, %o4 ! dst offset relative to dword boundary
bz,pn %ncc, .fillaligned ! dst already dword aligned
! here there is a least one more byte to zero out: otherwise we would
! have exited through label .lastword
sub %o4, 8, %o4 ! bytes to align dst to dword boundary
.makealigned:
stb %g0, [%o3] ! dst[] = 0
addcc %g4, 1, %g4 ! n--
bz,pt %ncc, .done ! n == 0, we are done
addcc %o4, 1, %o4 ! any more byte needed to align
bnz,pt %ncc, .makealigned ! yup, pad another byte
add %o3, 1, %o3 ! dst++
nop ! pad to align copy loop below
nop ! pad to align copy loop below
! here we know that there at least another 8 bytes to pad, since
! we don't get here unless there were >= 16 bytes to pad to begin
! with, and we have padded at most 7 bytes suring dst aligning
.fillaligned:
add %g4, 7, %o2 ! round up to next dword boundary
and %o2, -8, %o4 ! pointer to next dword boundary
and %o2, 8, %o2 ! dword count odd ? 8 : 0
stx %g0, [%o3] ! store first dword
addcc %o4, %o2, %o4 ! dword count == 1 ?
add %g4, %o2, %g4 ! if dword count odd, n -= 8
bz,pt %ncc, .bytepad ! if dword count == 1, pad leftover bytes
add %o3, %o2, %o3 ! bump dst if dword count odd
.filldword:
addcc %o4, 16, %o4 ! count -= 16
stx %g0, [%o3] ! dst[n] = 0
stx %g0, [%o3 + 8] ! dst[n+8] = 0
add %o3, 16, %o3 ! dst += 16
bcc,pt %ncc, .filldword ! fill dwords until count == 0
addcc %g4, 16, %g4 ! n -= 16
bz,pn %ncc, .done ! if n == 0, we are done
.bytepad:
and %g4, 1, %o2 ! byte count odd ? 1 : 0
stb %g0, [%o3] ! store first byte
addcc %g4, %o2, %g4 ! byte count == 1 ?
bz,pt %ncc, .done ! yup, we are done
add %o3, %o2, %o3 ! bump pointer if odd
.fillbyte:
addcc %g4, 2, %g4 ! n -= 2
stb %g0, [%o3] ! dst[n] = 0
stb %g0, [%o3 + 1] ! dst[n+1] = 0
bnz,pt %ncc, .fillbyte ! fill until n == 0
add %o3, 2, %o3 ! dst += 2
.done:
retl ! done
nop ! pad to align loops below
nop ! pad to align loops below
! this is the last word. It may contain null bytes. store bytes
! until n == 0. if null byte encountered, continue
.lastword:
sub %g4, 8, %g4 ! undo counter pre-increment
add %o2, 8, %o2 ! adjust dst for counter un-bumping
srlx %o1, 56, %g1 ! first byte
stb %g1, [%o2 + %g4] ! store it
inccc %g4 ! n--
bz .done ! if n == 0, we're done
andcc %g1, 0xff, %g0 ! end of src reached ?
movz %ncc, %g0, %o1 ! if so, start padding with null bytes
srlx %o1, 48, %g1 ! second byte
stb %g1, [%o2 + %g4] ! store it
inccc %g4 ! n--
bz .done ! if n == 0, we're done
andcc %g1, 0xff, %g0 ! end of src reached ?
movz %ncc, %g0, %o1 ! if so, start padding with null bytes
srlx %o1, 40, %g1 ! third byte
stb %g1, [%o2 + %g4] ! store it
inccc %g4 ! n--
bz .done ! if n == 0, we're done
andcc %g1, 0xff, %g0 ! end of src reached ?
movz %ncc, %g0, %o1 ! if so, start padding with null bytes
srlx %o1, 32, %g1 ! fourth byte
stb %g1, [%o2 + %g4] ! store it
inccc %g4 ! n--
bz .done ! if n == 0, we're done
andcc %g1, 0xff, %g0 ! end of src reached ?
movz %ncc, %g0, %o1 ! if so, start padding with null bytes
srlx %o1, 24, %g1 ! fifth byte
stb %g1, [%o2 + %g4] ! store it
inccc %g4 ! n--
bz .done ! if n == 0, we're done
andcc %g1, 0xff, %g0 ! end of src reached ?
movz %ncc, %g0, %o1 ! if so, start padding with null bytes
srlx %o1, 16, %g1 ! sixth byte
stb %g1, [%o2 + %g4] ! store it
inccc %g4 ! n--
bz .done ! if n == 0, we're done
andcc %g1, 0xff, %g0 ! end of src reached ?
movz %ncc, %g0, %o1 ! if so, start padding with null bytes
srlx %o1, 8, %g1 ! seventh byte
stb %g1, [%o2 + %g4] ! store it
inccc %g4 ! n--
bz .done ! if n == 0, we're done
andcc %g1, 0xff, %g0 ! end of src reached ?
movz %ncc, %g0, %o1 ! if so, start padding with null bytes
ba .done ! here n must be zero, we are done
stb %o1, [%o2 + %g4] ! store eigth byte
nop ! pad to align loops below
nop ! pad to align loops below
.storebyte1421:
ldx [%o3 + %g4], %o1 ! x = src[]
addcc %g4, 8, %g4 ! src += 8, dst += 8
bcs,pn %ncc,.lastword ! if counter wraps, last word
andn %o5, %o1, %g1 ! ~x & 0x8080808080808080
sub %o1, %o4, %g5 ! x - 0x0101010101010101
andcc %g5, %g1, %g0 ! ((x - 0x0101010101010101) & ~x & 0x8080808080808080)
bnz,pn %ncc, .zerobyte ! end of src found, may need to pad
add %o2, %g4, %g5 ! dst (in pointer form)
srlx %o1, 56, %g1 ! %g1<7:0> = first byte; word aligned now
stb %g1, [%g5] ! store first byte
srlx %o1, 24, %g1 ! %g1<31:0> = bytes 2, 3, 4, 5
stw %g1, [%g5 + 1] ! store bytes 2, 3, 4, 5
srlx %o1, 8, %g1 ! %g1<15:0> = bytes 6, 7
sth %g1, [%g5 + 5] ! store bytes 6, 7
ba .storebyte1421 ! next dword
stb %o1, [%g5 + 7] ! store eigth byte
.storebyte1241:
ldx [%o3 + %g4], %o1 ! x = src[]
addcc %g4, 8, %g4 ! src += 8, dst += 8
bcs,pn %ncc,.lastword ! if counter wraps, last word
andn %o5, %o1, %g1 ! ~x & 0x8080808080808080
sub %o1, %o4, %g5 ! x - 0x0101010101010101
andcc %g5, %g1, %g0 ! ((x - 0x0101010101010101) & ~x & 0x8080808080808080)
bnz,pn %ncc, .zerobyte ! x has zero byte, handle end cases
add %o2, %g4, %g5 ! dst (in pointer form)
srlx %o1, 56, %g1 ! %g1<7:0> = first byte; half-word aligned now
stb %g1, [%g5] ! store first byte
srlx %o1, 40, %g1 ! %g1<15:0> = bytes 2, 3
sth %g1, [%g5 + 1] ! store bytes 2, 3
srlx %o1, 8, %g1 ! %g1<31:0> = bytes 4, 5, 6, 7
stw %g1, [%g5 + 3] ! store bytes 4, 5, 6, 7
ba .storebyte1241 ! next dword
stb %o1, [%g5 + 7] ! store eigth byte
.storehalfword:
ldx [%o3 + %g4], %o1 ! x = src[]
addcc %g4, 8, %g4 ! src += 8, dst += 8
bcs,pn %ncc,.lastword ! if counter wraps, last word
andn %o5, %o1, %g1 ! ~x & 0x8080808080808080
sub %o1, %o4, %g5 ! x - 0x0101010101010101
andcc %g5, %g1, %g0 ! ((x - 0x0101010101010101) & ~x & 0x8080808080808080)
bnz,pn %ncc, .zerobyte ! x has zero byte, handle end cases
add %o2, %g4, %g5 ! dst (in pointer form)
srlx %o1, 48, %g1 ! %g1<15:0> = bytes 1, 2; word aligned now
sth %g1, [%g5] ! store bytes 1, 2
srlx %o1, 16, %g1 ! %g1<31:0> = bytes 3, 4, 5, 6
stw %g1, [%g5 + 2] ! store bytes 3, 4, 5, 6
ba .storehalfword ! next dword
sth %o1, [%g5 + 6] ! store bytes 7, 8
nop ! align next loop to 16-byte boundary
nop ! align next loop to 16-byte boundary
.storeword2:
ldx [%o3 + %g4], %o1 ! x = src[]
addcc %g4, 8, %g4 ! src += 8, dst += 8
bcs,pn %ncc,.lastword ! if counter wraps, last word
andn %o5, %o1, %g1 ! ~x & 0x8080808080808080
sub %o1, %o4, %g5 ! x - 0x0101010101010101
andcc %g5, %g1, %g0 ! ((x - 0x0101010101010101) & ~x & 0x8080808080808080)
bnz,pn %ncc, .zerobyte ! x has zero byte, handle end cases
add %o2, %g4, %g5 ! dst (in pointer form)
srlx %o1, 32, %g1 ! %g1<31:0> = bytes 1, 2, 3, 4
stw %g1, [%g5] ! store bytes 1, 2, 3, 4
ba .storeword2 ! next dword
stw %o1, [%g5 + 4] ! store bytes 5, 6, 7, 8
! do not remove these pads, loop above may slow down otherwise
nop ! pad
nop ! pad
SET_SIZE(strncpy)