home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Atari FTP
/
ATARI_FTP_0693.zip
/
ATARI_FTP_0693
/
Mint
/
mntlib32.zoo
/
purec
/
bcopy.s
< prev
next >
Wrap
Text File
|
1993-05-23
|
4KB
|
201 lines
; ***** VERSION ONLY FOR TURBO/PUREC *****
; new version of bcopy, memcpy and memmove
; handles overlap, odd/even alignment
; uses movem to copy 256 bytes blocks faster.
; Alexander Lehmann alexlehm@iti.informatik.th-darmstadt.de
; sortof inspired by jrbs bcopy
.text
.even
.globl bcopy,_bcopy
.globl memcpy
.globl memmove
; void *memcpy( void *dest, const void *src, size_t len );
; void *memmove( void *dest, const void *src, size_t len );
; returns dest
; functions are aliased
memcpy:
memmove:
exg a0,a1 ; dest <-> src
; void bcopy( const void *src, void *dest, size_t length );
; return value not used (returns dest)
_bcopy:
bcopy:
pea (a1) ; push dest
tst.l d0 ; length
beq exit ; length==0? (size_t)
; a0 src, a1 dest, d0.l length
; overlay ?
cmp.l a0,a1
bgt top_down
move.w a0,d1 ; test for alignment
move.w a1,d2
eor.w d2,d1
btst #0,d1 ; one odd one even ?
bne slow_copy
btst #0,d2 ; both even ?
beq.b both_even
move.b (a0)+,(a1)+ ; copy one byte, now we are both even
subq.l #1,d0
both_even:
move.b d0,d1 ; save length less 256
lsr.l #8,d0 ; number of 256 bytes blocks
beq.b less256
movem.l d1/d3-d7/a2-a6,-(sp) ; d2 is scratch reg
copy256:
movem.l (a0)+,d1-d7/a2-a6 ; copy 5*48+16=256 bytes
movem.l d1-d7/a2-a6,(a1)
movem.l (a0)+,d1-d7/a2-a6
movem.l d1-d7/a2-a6,48(a1)
movem.l (a0)+,d1-d7/a2-a6
movem.l d1-d7/a2-a6,96(a1)
movem.l (a0)+,d1-d7/a2-a6
movem.l d1-d7/a2-a6,144(a1)
movem.l (a0)+,d1-d7/a2-a6
movem.l d1-d7/a2-a6,192(a1)
movem.l (a0)+,d1-d4
movem.l d1-d4,240(a1)
lea 256(a1),a1 ; increment dest, src is already
subq.l #1,d0
bne.b copy256 ; next, please
movem.l (sp)+,d1/d3-d7/a2-a6
less256: ; copy 16 bytes blocks
move.b d1,d0 ; d0.w always 0
lsr.b #2,d0 ; number of 4 bytes blocks
beq.b less4 ; less that 4 bytes left
move.w d0,d2
subq.b #1,d0
lsr.b #2,d0 ; number of 16 bytes blocks minus 1, if d2==0
neg.w d2
and.w #3,d2 ; d2 = number of bytes below 16 (-n)&3
add.w d2,d2 ; offset in code (movl two bytes)
jmp 2(pc,d2.w) ; jmp into loop
copy16:
move.l (a0)+,(a1)+
move.l (a0)+,(a1)+
move.l (a0)+,(a1)+
move.l (a0)+,(a1)+
dbra d0,copy16
less4:
btst #1,d1
beq.b less2
move.w (a0)+,(a1)+
less2:
btst #0,d1
beq.b exit
move.b (a0),(a1)
exit:
move.l (sp)+,a0 ; return dest (for memcpy only)
rts
slow_copy: ; byte by bytes copy
move.w d0,d1
neg.w d1
and.w #7,d1 ; d1 = number of bytes blow 8 (-n)&7
addq.l #7,d0
lsr.l #3,d0 ; number of 8 bytes block plus 1, if d1!=0
add.w d1,d1 ; offset in code (movb two bytes)
jmp 2(pc,d1.w) ; jump into loop
scopy:
move.b (a0)+,(a1)+
move.b (a0)+,(a1)+
move.b (a0)+,(a1)+
move.b (a0)+,(a1)+
move.b (a0)+,(a1)+
move.b (a0)+,(a1)+
move.b (a0)+,(a1)+
move.b (a0)+,(a1)+
subq.l #1,d0
bne.b scopy
bra.b exit
top_down:
add.l d0,a0 ; a0 byte after end of src
add.l d0,a1 ; a1 byte after end of dest
move.w a0,d1 ; exact the same as above, only with predec
move.w a1,d2
eor.w d2,d1
btst #0,d1
bne slow_copy_d
btst #0,d2
beq.b both_even_d
move.b -(a0),-(a1)
subq.l #1,d0
both_even_d:
move.b d0,d1
lsr.l #8,d0
beq.b less256_d
movem.l d1/d3-d7/a2-a6,-(sp) ; d2 is scratch reg
copy256_d:
movem.l -48(a0),d1-d7/a2-a6 ; copy 5*48+16=256 bytes
movem.l d1-d7/a2-a6,-(a1)
movem.l -96(a0),d1-d7/a2-a6
movem.l d1-d7/a2-a6,-(a1)
movem.l -144(a0),d1-d7/a2-a6
movem.l d1-d7/a2-a6,-(a1)
movem.l -192(a0),d1-d7/a2-a6
movem.l d1-d7/a2-a6,-(a1)
movem.l -240(a0),d1-d7/a2-a6
movem.l d1-d7/a2-a6,-(a1)
movem.l -256(a0),d1-d4
movem.l d1-d4,-(a1)
lea -256(a0),a0
subq.l #1,d0
bne.b copy256_d
movem.l (sp)+,d1/d3-d7/a2-a6
less256_d:
move.b d1,d0
lsr.b #2,d0
beq.b less4_d
move.w d0,d2
subq.b #1,d0
lsr.b #2,d0
neg.w d2
and.w #3,d2
add.w d2,d2
jmp 2(pc,d2.w)
copy16_d:
move.l -(a0),-(a1)
move.l -(a0),-(a1)
move.l -(a0),-(a1)
move.l -(a0),-(a1)
dbra d0,copy16_d
less4_d:
btst #1,d1
beq.b less2_d
move.w -(a0),-(a1)
less2_d:
btst #0,d1
beq exit
move.b -(a0),-(a1)
bra exit
slow_copy_d:
move.w d0,d1
neg.w d1
and.w #7,d1
addq.l #7,d0
lsr.l #3,d0
add.w d1,d1
jmp 2(pc,d1.w)
scopy_d:
move.b -(a0),-(a1)
move.b -(a0),-(a1)
move.b -(a0),-(a1)
move.b -(a0),-(a1)
move.b -(a0),-(a1)
move.b -(a0),-(a1)
move.b -(a0),-(a1)
move.b -(a0),-(a1)
subq.l #1,d0
bne.b scopy_d
bra exit