rockbox/firmware/common/memcpy_a.S

196 lines
7 KiB
ArmAsm

/***************************************************************************
* __________ __ ___.
* Open \______ \ ____ ____ | | _\_ |__ _______ ___
* Source | _// _ \_/ ___\| |/ /| __ \ / _ \ \/ /
* Jukebox | | ( <_> ) \___| < | \_\ ( <_> > < <
* Firmware |____|_ /\____/ \___ >__|_ \|___ /\____/__/\_ \
* \/ \/ \/ \/ \/
* $Id$
*
* Copyright (C) 2004 by Jens Arnold
*
* All files in this archive are subject to the GNU General Public License.
* See the file COPYING in the source tree root for full license agreement.
*
* This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY
* KIND, either express or implied.
*
****************************************************************************/
#include "config.h"
.section .icode,"ax",@progbits
#if CONFIG_CPU == SH7034
.align 2
.global _memcpy
.type _memcpy,@function
/* Copies <length> bytes of data in memory from <source> to <dest>
* This version is optimized for speed
*
* arguments:
* r4 - destination address
* r5 - source address
* r6 - length
*
* return value:
* r0 - destination address (like ANSI version)
*
* register usage:
* r0 - data / temporary
* r1 - bit mask for rounding to long bounds / 2nd data
* r2 - first long bound (only if >= 12 bytes)
* r3 - last long bound (-4) (only if >= 12 bytes)
* r4 - current dest address
* r5 - current source address
* r6 - source end address
* r7 - stored dest start address
*
* The instruction order below is devised in a way to utilize the pipelining
* of the SH1 to the max. The routine also tries to utilize fast page mode.
*/
_memcpy:
add r5,r6 /* r6 = source_end */
mov r4,r7 /* store for returning */
add #-8,r4 /* adjust for early increments (max. 2 longs) */
mov r6,r0
add #-12,r0 /* r0 = r6 - 12; don't go below 12 here! */
cmp/hs r5,r0 /* >= 12 bytes to copy? */
bf .start_b2 /* no, jump into byte loop */
mov #-4,r1 /* r1 = 0xFFFFFFFC */
mov r5,r2
add #3,r2
and r1,r2 /* r2 = first source long bound */
mov r6,r3
add #-4,r3 /* end offset for copying 2 longs per pass */
bra .start_b1 /* jump into leading byte loop */
and r1,r3 /* r3 = last source long bound - 4 */
/* leading byte loop: copies 0..3 bytes */
.align 2
.loop_b1:
mov.b @r5+,r0 /* load byte & increment source addr */
add #1,r4 /* increment dest addr */
mov.b r0,@(7,r4) /* store byte */
.start_b1:
cmp/hi r5,r2 /* runs r5 up to first long bound */
bt .loop_b1
/* now r5 is always at a long boundary */
/* -> memory reading is done in longs for all dest alignments */
/* selector for main copy loop */
mov r4,r0
tst #3,r0 /* dest now also at long bound? */
bt .loop2_l /* yes, do long copy */
tst #1,r0 /* dest now at least at word bound? */
bt .start4_w /* yes, do word copy */
/* main loop for byte aligned destination (fast) */
/* copies 1 long per pass */
add #4,r3 /* reset end offset */
add #-1,r4 /* adjust to word alignment for word write+ */
.loop4_b:
mov.l @r5+,r0 /* load a long & increment source addr */
add #4,r4 /* increment dest addr */
mov.b r0,@(8,r4) /* store low byte */
shlr8 r0 /* get middle 2 bytes */
mov.w r0,@(6,r4) /* store as word+ */
shlr16 r0 /* get upper byte */
mov.b r0,@(5,r4) /* and store */
cmp/hi r5,r3 /* runs r5 up to last long bound */
bt .loop4_b
bra .start_b2 /* jump to trailing byte loop */
add #1,r4 /* readjust */
/* main loop for word aligned destination (faster) */
/* copies 2 longs per pass, utilizing fast page mode */
.start4_w:
add #-2,r4 /* adjust to long alignment for long write+ */
.loop4_w:
mov.l @r5+,r1 /* load first long & increment source addr */
add #8,r4 /* increment dest addr */
mov.l @r5+,r0 /* load second long & increment source addr */
cmp/hi r5,r3 /* runs r5 up to last or second last long bound */
mov.w r0,@(8,r4) /* store low word of second long */
xtrct r1,r0 /* extract low word of first long & high word of second long */
mov.l r0,@(4,r4) /* and store as long+ */
swap.w r1,r0 /* get high word of first long */
mov.w r0,@(2,r4) /* and store it */
bt .loop4_w
add #2,r4 /* readjust destination */
add #4,r3 /* reset end offset */
cmp/hi r5,r3 /* one long left? */
bf .start_b2 /* no, jump to trailing byte loop */
mov.l @r5+,r0 /* load last long & increment source addr */
add #4,r4 /* increment dest addr */
mov.w r0,@(6,r4) /* store low word */
shlr16 r0 /* get high word */
bra .start_b2 /* jump to trailing byte loop */
mov.w r0,@(4,r4) /* and store it */
/* main loop for long aligned destination (fastest) */
/* copies 2 longs per pass, utilizing fast page mode */
.loop2_l:
mov.l @r5+,r1 /* load first long & increment source addr */
add #8,r4 /* increment dest addr */
mov.l @r5+,r0 /* load second long & increment source addr */
cmp/hi r5,r3 /* runs r5 up to last or second last long bound */
mov.l r1,@r4 /* store first long */
mov.l r0,@(4,r4) /* store second long; NOT ALIGNED - no speed loss here! */
bt .loop2_l
add #4,r3 /* reset end offset */
cmp/hi r5,r3 /* one long left? */
bf .start_b2 /* no, jump to trailing byte loop */
mov.l @r5+,r0 /* load last long & increment source addr */
add #4,r4 /* increment dest addr */
bra .start_b2 /* jump to trailing byte loop */
mov.l r0,@(4,r4) /* store last long */
/* trailing byte loop: copies 0..3 bytes (or all for < 12 in total) */
.loop_b2:
mov.b @r5+,r0 /* load byte & increment source addr */
add #1,r4 /* increment dest addr */
mov.b r0,@(7,r4) /* store byte */
.start_b2:
cmp/hi r5,r6 /* runs r5 up to end address */
bt .loop_b2
rts
mov r7,r0 /* return dest start address */
.end:
.size _memcpy,.end-_memcpy
#elif CONFIG_CPU == MCF5249
.align 2
.global memcpy
.type memcpy,@function
/* Copies <length> bytes of data in memory from <source> to <dest>
* This version is not optimized at all
*/
memcpy:
move.l (4,%sp),%a1 /* Destination */
move.l (8,%sp),%a0 /* Source */
move.l (12,%sp),%d1 /* Length */
cmp.l #0,%d1
bra.b .byteloopend
.byteloop:
move.b (%a0)+,(%a1)+
subq.l #1,%d1
.byteloopend:
bne.b .byteloop
rts
#endif