mirror of
https://github.com/autc04/Retro68.git
synced 2024-11-28 05:51:04 +00:00
290 lines
10 KiB
ArmAsm
290 lines
10 KiB
ArmAsm
/*******************************************************************************
|
|
*
|
|
* Copyright (c) 1993 Intel Corporation
|
|
*
|
|
* Intel hereby grants you permission to copy, modify, and distribute this
|
|
* software and its documentation. Intel grants this permission provided
|
|
* that the above copyright notice appears in all copies and that both the
|
|
* copyright notice and this permission notice appear in supporting
|
|
* documentation. In addition, Intel grants this permission provided that
|
|
* you prominently mark as "not part of the original" any modifications
|
|
* made to this software or documentation, and that the name of Intel
|
|
* Corporation not be used in advertising or publicity pertaining to
|
|
* distribution of the software or the documentation without specific,
|
|
* written prior permission.
|
|
*
|
|
* Intel Corporation provides this AS IS, WITHOUT ANY WARRANTY, EXPRESS OR
|
|
* IMPLIED, INCLUDING, WITHOUT LIMITATION, ANY WARRANTY OF MERCHANTABILITY
|
|
* OR FITNESS FOR A PARTICULAR PURPOSE. Intel makes no guarantee or
|
|
* representations regarding the use of, or the results of the use of,
|
|
* the software and documentation in terms of correctness, accuracy,
|
|
* reliability, currentness, or otherwise; and you rely on the software,
|
|
* documentation and results solely at your own risk.
|
|
*
|
|
* IN NO EVENT SHALL INTEL BE LIABLE FOR ANY LOSS OF USE, LOSS OF BUSINESS,
|
|
* LOSS OF PROFITS, INDIRECT, INCIDENTAL, SPECIAL OR CONSEQUENTIAL DAMAGES
|
|
* OF ANY KIND. IN NO EVENT SHALL INTEL'S TOTAL LIABILITY EXCEED THE SUM
|
|
* PAID TO INTEL FOR THE PRODUCT LICENSED HEREUNDER.
|
|
*
|
|
******************************************************************************/
|
|
|
|
.file "sncpy_ca.s"
|
|
#ifdef __PIC
|
|
.pic
|
|
#endif
|
|
#ifdef __PID
|
|
.pid
|
|
#endif
|
|
/*
|
|
* (c) copyright 1988,1993 Intel Corp., all rights reserved
|
|
*/
|
|
|
|
/*
|
|
procedure strncpy (optimized assembler version for the CA)
|
|
|
|
dest_addr = strncpy (dest_addr, src_addr, max_bytes)
|
|
|
|
copy the null terminated string pointed to by src_addr to
|
|
the string space pointed to by dest_addr. Return the original
|
|
dest_addr. If the source string is shorter than max_bytes,
|
|
then null-pad the destination string.
|
|
|
|
This routine will fail if the source and destination string
|
|
overlap (in particular, if the end of the source is overlapped
|
|
by the beginning of the destination). The behavior is undefined.
|
|
This is acceptable according to the draft C standard.
|
|
|
|
Undefined behavior will also occur if the end of the source string
|
|
(i.e. the terminating null byte) is in the last word of the program's
|
|
allocated memory space. This is so because, in several cases, strcpy
|
|
will fetch ahead one word. Disallowing the fetch ahead would impose
|
|
a severe performance penalty.
|
|
|
|
This program handles five cases:
|
|
|
|
1) both arguments start on a word boundary
|
|
2) neither are word aligned, but they are offset by the same amount
|
|
3) source is word aligned, destination is not
|
|
4) destination is word aligned, source is not
|
|
5) neither is word aligned, and they are offset by differing amounts
|
|
|
|
At the time of this writing, only g0 thru g7 and g13 are available
|
|
for use in this leafproc; other registers would have to be saved and
|
|
restored. These nine registers, plus tricky use of g14 are sufficient
|
|
to implement the routine. The registers are used as follows:
|
|
|
|
g0 original dest ptr; not modified, so that it may be returned.
|
|
g1 src ptr; shift count
|
|
g2 max_bytes
|
|
g3 src ptr (word aligned)
|
|
g4 dest ptr (word aligned)
|
|
g5 0xff -- byte extraction mask
|
|
Little endian:
|
|
g6 lsw of double word for extraction of 4 bytes
|
|
g7 msw of double word for extraction of 4 bytes
|
|
Big endian:
|
|
g6 msw of double word for extraction of 4 bytes
|
|
g7 lsw of double word for extraction of 4 bytes
|
|
g13 return address
|
|
g14 byte extracted.
|
|
*/
|
|
|
|
#if __i960_BIG_ENDIAN__
|
|
#define MSW g6
|
|
#define LSW g7
|
|
#else
|
|
#define LSW g6
|
|
#define MSW g7
|
|
#endif
|
|
|
|
.globl _strncpy
|
|
.globl __strncpy
|
|
.leafproc _strncpy, __strncpy
|
|
.align 2
|
|
_strncpy:
|
|
#ifndef __PIC
|
|
lda Lrett,g14
|
|
#else
|
|
lda Lrett-(.+8)(ip),g14
|
|
#endif
|
|
__strncpy:
|
|
notand g1,3,g3 # extract word addr of start of src
|
|
lda (g14),g13 # preserve return address
|
|
cmpibge.f 0,g2,Lexit_code # Lexit if number of bytes to move is <= zero.
|
|
cmpo g3,g1 # check alignment of src
|
|
ld (g3),LSW # fetch word containing at least first byte
|
|
notand g0,3,g4 # extract word addr of start of dest
|
|
lda 4(g3),g3 # advance src word addr
|
|
bne.f Lcase_245 # branch if src is NOT word aligned
|
|
|
|
Lcase_13:
|
|
cmpo g0,g4 # check alignment of dest
|
|
lda 0xff,g5 # load mask for byte extraction
|
|
subo 4,g4,g4 # store is pre-incrementing; back up dest addr
|
|
bne.f Lcase_3 # branch if dest not word aligned
|
|
|
|
Lcase_1: # src and dest are word aligned
|
|
Lcase_1_wloop: # word copying loop
|
|
cmpi g2,4 # check for fewer than four bytes to move
|
|
lda (LSW),g1 # keep a copy of the src word
|
|
addo 4,g4,g4 # pre-increment dest addr
|
|
bl.f Lcase_1_cloop.a # branch if fewer than four bytes to copy
|
|
scanbyte 0,g1 # check for null byte in src word
|
|
ld (g3),LSW # pre-fetch next word of src
|
|
subi 4,g2,g2 # decrease max_byte count by the 4 bytes moved
|
|
bo.f Lcase_1_cloop.c # branch if word contains null byte
|
|
addo 4,g3,g3 # post-increment src addr
|
|
st g1,(g4) # store word in dest string
|
|
b Lcase_1_wloop
|
|
|
|
Lcase_3_cloop.a:
|
|
Lcase_1_cloop.a: # character copying loop (max_bytes <= 3)
|
|
#if __i960_BIG_ENDIAN__
|
|
rotate 8,g1,g1 # move next byte into position for extraction
|
|
#endif
|
|
and g5,g1,g14 # extract next char
|
|
Lcase_1_cloop.b:
|
|
cmpdeci 0,g2,g2 # is max_bytes exhausted?
|
|
be.f Lexit_code # Lexit if max_bytes is exhausted
|
|
cmpo 0,g14 # check for null byte
|
|
stob g14,(g4) # store the byte in dest
|
|
#if ! __i960_BIG_ENDIAN__
|
|
shro 8,g1,g1 # move next byte into position for extraction
|
|
#endif
|
|
lda 1(g4),g4 # post-increment dest byte addr
|
|
bne.t Lcase_1_cloop.a # branch if null not reached
|
|
b Lcase_1_cloop.b
|
|
|
|
Lexit_code:
|
|
mov 0,g14 # conform to register conventions
|
|
bx (g13) # g0 = addr of dest; g14 = 0
|
|
Lrett:
|
|
ret
|
|
|
|
Lcase_1_cloop.c:
|
|
Lcase_3_cloop.c:
|
|
#if __i960_BIG_ENDIAN__
|
|
rotate 24,g5,g5 # move mask into position for testing next byte
|
|
#endif
|
|
and g5,g1,g14 # extract next char
|
|
cmpo 0,g14 # check for null byte
|
|
#if ! __i960_BIG_ENDIAN__
|
|
lda (g5),LSW # keep a copy of the current mask
|
|
shlo 8,g5,g5 # move mask into position for testing next byte
|
|
#endif
|
|
bne.t Lcase_1_cloop.c # branch if null not reached
|
|
#if __i960_BIG_ENDIAN__
|
|
subo 1,g5,g5 # null pad.
|
|
andnot g5,g1,g1 # last bytes to copy, and null pad rest of word
|
|
#else
|
|
subo 1,LSW,g5 # mask to get last bytes to copy, and null pad
|
|
and g5,g1,g1 # last bytes to copy, and null pad rest of word
|
|
#endif
|
|
st g1,(g4)
|
|
|
|
Lcase_1_zwloop: # zero word loop
|
|
cmpi g2,4 # check for fewer than four bytes to move
|
|
addo 4,g4,g4 # pre-increment dest addr
|
|
bl.f Lcase_1_cloop.b # branch if fewer than four bytes to copy
|
|
subo 4,g2,g2 # decrease max_byte count by the 4 bytes moved
|
|
st g14,(g4) # store word in dest string
|
|
b Lcase_1_zwloop
|
|
|
|
Lcase_3: # src is word aligned; dest is not
|
|
addo 8,g4,g4 # move dest word ptr to first word boundary
|
|
lda (g0),g1 # copy dest byte ptr
|
|
mov LSW,MSW # make copy of first word of src
|
|
lda 32,g14 # initialize shift count to zero (mod 32)
|
|
|
|
Lcase_25:
|
|
Lcase_3_cloop_at_start: # character copying loop for start of dest str
|
|
cmpdeci 0,g2,g2 # is max_bytes exhausted?
|
|
#if __i960_BIG_ENDIAN__
|
|
shro 24,MSW,g5 # extract next char
|
|
#else
|
|
and g5,MSW,g5 # extract next char
|
|
#endif
|
|
be.f Lexit_code # Lexit if max_bytes is exhausted
|
|
cmpo 0,g5 # check for null byte
|
|
stob g5,(g1) # store the byte in dest
|
|
addo 1,g1,g1 # post-increment dest ptr
|
|
lda 0xff,g5 # re-initialize byte extraction mask
|
|
bne.t 1f # drop thru if null byte reached (to pad)
|
|
movl 0,g6 # blank out remainder of input buffer
|
|
1:
|
|
cmpo g1,g4 # have we reached word boundary in dest yet?
|
|
#if __i960_BIG_ENDIAN__
|
|
lda -8(g14),g14 # augment the shift counter
|
|
rotate 8,MSW,MSW # move next byte into position for extraction
|
|
#else
|
|
lda 8(g14),g14 # augment the shift counter
|
|
shro 8,MSW,MSW # move next byte into position for extraction
|
|
#endif
|
|
bne.t Lcase_3_cloop_at_start # branch if reached word boundary?
|
|
|
|
ld (g3),MSW # fetch msw of operand for double shift
|
|
|
|
Lcase_4:
|
|
|
|
#if __i960_BIG_ENDIAN__
|
|
cmpobne 0,g14,Lcase_3_wloop # branch if src is still unaligned.
|
|
|
|
Lcase_3_wloop2:
|
|
cmpi g2,4 # less than four bytes to move?
|
|
lda (LSW),g1 # extract 4 bytes of src
|
|
lda 4(g3),g3 # post-increment src word addr
|
|
bl.f Lcase_3_cloop.a # branch if < four bytes left to move
|
|
scanbyte 0,g1 # check for null byte
|
|
lda (MSW),LSW # move msw to lsw
|
|
subi 4,g2,g2 # decrease max_byte count by the 4 bytes moved
|
|
ld (g3),MSW # pre-fetch msw of operand for double shift
|
|
bo.f Lcase_3_cloop.c # branch if word contains null byte
|
|
st g1,(g4) # store 4 bytes to dest
|
|
addo 4,g4,g4 # post-increment dest ptr
|
|
b Lcase_3_wloop2
|
|
|
|
#endif
|
|
|
|
Lcase_3_wloop:
|
|
cmpi g2,4 # less than four bytes to move?
|
|
eshro g14,g6,g1 # extract 4 bytes of src
|
|
lda 4(g3),g3 # post-increment src word addr
|
|
bl.f Lcase_3_cloop.a # branch if < four bytes left to move
|
|
scanbyte 0,g1 # check for null byte
|
|
lda (MSW),LSW # move msw to lsw
|
|
subi 4,g2,g2 # decrease max_byte count by the 4 bytes moved
|
|
ld (g3),MSW # pre-fetch msw of operand for double shift
|
|
bo.f Lcase_3_cloop.c # branch if word contains null byte
|
|
st g1,(g4) # store 4 bytes to dest
|
|
addo 4,g4,g4 # post-increment dest ptr
|
|
b Lcase_3_wloop
|
|
|
|
Lcase_245:
|
|
cmpo g0,g4 # check alignment of dest
|
|
ld (g3),MSW # pre-fetch second half
|
|
and 3,g1,g1 # compute shift count
|
|
lda 0xff,g5 # load mask for byte extraction
|
|
#if __i960_BIG_ENDIAN__
|
|
subo g1,4,g14 # adjust shift count for big endian.
|
|
shlo 3,g14,g14
|
|
#else
|
|
shlo 3,g1,g14
|
|
#endif
|
|
be.t Lcase_4 # branch if dest is word aligned
|
|
or g4,g1,g1 # is src earlier in word, later, or sync w/ dst
|
|
cmpo g0,g1 # < indicates first word of dest has more bytes
|
|
/* than first word of source. */
|
|
lda 4(g4),g4 # move dest word addr to first word boundary
|
|
eshro g14,g6,g5 # extract four bytes
|
|
lda (g0),g1
|
|
bg.f 1f
|
|
mov MSW,LSW
|
|
lda 4(g3),g3 # move src word addr to second word boundary
|
|
1:
|
|
mov g5,MSW
|
|
lda 0xff,g5
|
|
b Lcase_25
|
|
|
|
/* end of strncpy */
|
|
|