Moved some assembly optimizations to the target tree

git-svn-id: svn://svn.rockbox.org/rockbox/trunk@10318 a1c6a512-1295-4272-9138-f99709370657

9 files changed, 2376 insertions, 0 deletions

diff --git a/firmware/target/arm/memset-arm.S b/firmware/target/arm/memset-arm.S
new file mode 100755
index 0000000000..b3faafcb37
--- /dev/null
+++ b/firmware/target/arm/memset-arm.S
@@ -0,0 +1,96 @@
+/***************************************************************************
+ *             __________               __   ___.
+ *   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
+
+    .align      2
+
+/*  The following code is based on code found in Linux kernel version 2.6.15.3
+ *  linux/arch/arm/lib/memset.S
+ *
+ *  Copyright (C) 1995-2000 Russell King
+ */
+
+/* This code will align a pointer for memset, if needed */
+1:      cmp     r2, #4                  @ 1 do we have enough
+        blt     5f                      @ 1 bytes to align with?
+        cmp     r3, #2                  @ 1
+        strgtb  r1, [r0, #-1]!          @ 1
+        strgeb  r1, [r0, #-1]!          @ 1
+        strb    r1, [r0, #-1]!          @ 1
+        sub     r2, r2, r3              @ 1 r2 = r2 - r3
+        b 2f
+
+        .global     memset
+        .type       memset,%function
+memset:
+        add     r0, r0, r2              @ we'll write backwards in memory
+        ands    r3, r0, #3              @ 1 unaligned?
+        bne     1b                      @ 1
+2:
+/*
+ * we know that the pointer in r0 is aligned to a word boundary.
+ */
+        orr     r1, r1, r1, lsl #8
+        orr     r1, r1, r1, lsl #16
+        mov     r3, r1
+        cmp     r2, #16
+        blt     5f
+/*
+ * We need an extra register for this loop - save the return address and
+ * use the LR
+ */
+        str     lr, [sp, #-4]!
+        mov     ip, r1
+        mov     lr, r1
+
+3:      subs    r2, r2, #64
+        stmgedb r0!, {r1, r3, ip, lr}   @ 64 bytes at a time.
+        stmgedb r0!, {r1, r3, ip, lr}
+        stmgedb r0!, {r1, r3, ip, lr}
+        stmgedb r0!, {r1, r3, ip, lr}
+        bgt     3b
+        ldmeqfd sp!, {pc}               @ Now <64 bytes to go.
+/*
+ * No need to correct the count; we're only testing bits from now on
+ */
+        tst     r2, #32
+        stmnedb r0!, {r1, r3, ip, lr}
+        stmnedb r0!, {r1, r3, ip, lr}
+        tst     r2, #16
+        stmnedb r0!, {r1, r3, ip, lr}
+        ldr     lr, [sp], #4
+
+5:      tst     r2, #8
+        stmnedb r0!, {r1, r3}
+        tst     r2, #4
+        strne   r1, [r0, #-4]!
+/*
+ * When we get here, we've got less than 4 bytes to zero.  We
+ * may have an unaligned pointer as well.
+ */
+6:      tst     r2, #2
+        strneb  r1, [r0, #-1]!
+        strneb  r1, [r0, #-1]!
+        tst     r2, #1
+        strneb  r1, [r0, #-1]!
+        mov     pc, lr
+end:
+        .size   memset,.end-memset
diff --git a/firmware/target/coldfire/memcpy-coldfire.S b/firmware/target/coldfire/memcpy-coldfire.S
new file mode 100755
index 0000000000..523e1f5ed9
--- /dev/null
+++ b/firmware/target/coldfire/memcpy-coldfire.S
@@ -0,0 +1,680 @@
+/***************************************************************************
+ *             __________               __   ___.
+ *   Open      \______   \ ____   ____ |  | _\_ |__   _______  ___
+ *   Source     |       _//  _ \_/ ___\|  |/ /| __ \ /  _ \  \/  /
+ *   Jukebox    |    |   (  <_> )  \___|    < | \_\ (  <_> > <  <
+ *   Firmware   |____|_  /\____/ \___  >__|_ \|___  /\____/__/\_ \
+ *                     \/            \/     \/    \/            \/
+ * $Id$
+ *
+ * Copyright (C) 2004-2005 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
+
+#define FULLSPEED /* use burst writing for word aligned destinations */
+    .align  2
+    .global memcpy
+    .global __memcpy_fwd_entry
+    .type   memcpy,@function
+
+/* Copies <length> bytes of data in memory from <source> to <dest>
+ * This version is optimized for speed
+ *
+ * arguments:
+ *  (4,%sp)  - destination address
+ *  (8,%sp)  - source address
+ *  (12,%sp) - length
+ *
+ * return value:
+ *  %d0 - destination address (like ANSI version)
+ *
+ * register usage:
+ *  %a0 - current source address
+ *  %a1 - current dest address
+ *  %a2 - source end address (in line-copy loops)
+ *  %d0 - data / scratch
+ *  %d1 - source end address (byte and longword copy) / data / scratch
+ *  %d2 - data / scratch
+ *  %d3..%d7 - data
+ *
+ * For maximum speed this routine reads and writes whole lines using burst
+ * move (movem.l) where possible. For byte aligned destinations (long+1 and
+ * long+3) it writes longwords only. Same goes for word aligned destinations
+ * if FULLSPEED is undefined.
+ */
+memcpy:
+    move.l  (4,%sp),%a1     /* Destination */
+    move.l  (8,%sp),%a0     /* Source */
+    move.l  (12,%sp),%d1    /* Length */
+
+__memcpy_fwd_entry:
+    add.l   %a0,%d1         /* %d1 = source end */
+
+    move.l  %a0,%d0
+    addq.l  #7,%d0
+    and.l   #0xFFFFFFFC,%d0 /* %d0 = first source long bound + 4 */
+    cmp.l   %d0,%d1         /* at least one aligned longword to copy? */
+    blo.w   .bytes2_start   /* no, jump directly to trailing byte loop */
+    
+    subq.l  #4,%d0          /* %d0 = first source long bound */
+    cmp.l   %a0,%d0         /* any bytes to copy? */
+    jls     .bytes1_end     /* no: skip byte loop */
+
+    /* leading byte loop: copies 0..3 bytes */
+.bytes1_loop:
+    move.b  (%a0)+,(%a1)+   /* copy byte */
+    cmp.l   %a0,%d0         /* runs %a0 up to first long bound */
+    jhi     .bytes1_loop
+
+.bytes1_end:
+    moveq.l #31,%d0
+    add.l   %a0,%d0
+    and.l   #0xFFFFFFF0,%d0 /* %d0 = first source line bound + 16 */
+    cmp.l   %d0,%d1         /* at least one aligned line to copy? */
+    blo.w   .long_start     /* no: jump to longword copy loop */
+    
+    lea.l   (-28,%sp),%sp   /* free up some registers */
+    movem.l %d2-%d7/%a2,(%sp)
+
+    moveq.l #16,%d2
+    sub.l   %d2,%d0         /* %d0 = first source line bound */
+    move.l  %d1,%a2         /* %a2 = end address */
+    lea.l   (-15,%a2),%a2   /* adjust end address for loops doing 16 bytes/ pass */
+    move.l  %a1,%d1
+    moveq.l #3,%d2          /* mask */
+    and.l   %d2,%d1
+    jmp.l   (2,%pc,%d1.l*4) /* switch (dest_addr & 3) */
+    bra.w   .lines_do0_start
+    bra.w   .lines_do1_start
+    bra.w   .lines_do2_start
+ /* bra.w   .lines_do3_start   implicit */
+
+    /* byte aligned destination (long + 3): use line burst reads in main loop */
+.lines_do3_start:
+    moveq.l #24,%d1         /* shift count for shifting by 3 bytes */
+    cmp.l   %a0,%d0         /* any leading longwords? */
+    jhi     .lines_do3_head_start  /* yes: leading longword copy */
+
+    movem.l (%a0),%d4-%d7   /* load first line */
+    lea.l   (16,%a0),%a0
+    move.l  %d4,%d2
+    lsr.l   %d1,%d2         /* get high byte of first longword */
+    move.b  %d2,(%a1)+      /* store byte */
+    jra     .lines_do3_entry       /* jump into main loop */
+    
+.lines_do3_head_start:
+    move.l  (%a0)+,%d7      /* load first longword */
+    move.l  %d7,%d2
+    lsr.l   %d1,%d2         /* get high byte */
+    move.b  %d2,(%a1)+      /* store byte */
+    jra     .lines_do3_head_entry  /* jump into leading longword loop */
+
+.lines_do3_head_loop:
+    move.l  %d7,%d6         /* move old longword away */
+    move.l  (%a0)+,%d7      /* load new longword */
+    move.l  %d7,%d2
+    lsr.l   %d1,%d2         /* get high byte */
+    or.l    %d2,%d6         /* combine with old lower 3 bytes */
+    move.l  %d6,(%a1)+      /* store longword */
+.lines_do3_head_entry:
+    lsl.l   #8,%d7          /* shift up lower 3 bytes */
+    cmp.l   %a0,%d0         /* runs %a0 up to first line bound */
+    jhi     .lines_do3_head_loop
+
+.lines_do3_loop:
+    move.l  %d7,%d3         /* move last longword of old line away */
+    movem.l (%a0),%d4-%d7   /* load new line */
+    lea.l   (16,%a0),%a0
+    move.l  %d4,%d2
+    lsr.l   %d1,%d2         /* get high byte of 1st longword */
+    or.l    %d2,%d3         /* combine with old lower 3 bytes */
+    move.l  %d3,(%a1)+      /* store longword */
+.lines_do3_entry:
+    lsl.l   #8,%d4          /* shift up lower 3 bytes */
+    move.l  %d5,%d2
+    lsr.l   %d1,%d2         /* get high byte of 2nd longword */
+    or.l    %d2,%d4         /* combine with 1st lower 3 bytes */
+    move.l  %d4,(%a1)+      /* store longword */
+    lsl.l   #8,%d5          /* shift up lower 3 bytes */
+    move.l  %d6,%d2
+    lsr.l   %d1,%d2         /* get high byte of 3rd longword */
+    or.l    %d2,%d5         /* combine with 2nd lower 3 bytes */
+    move.l  %d5,(%a1)+      /* store longword */
+    lsl.l   #8,%d6          /* shift up lower 3 bytes */
+    move.l  %d7,%d2
+    lsr.l   %d1,%d2         /* get high byte of 4th longword */
+    or.l    %d2,%d6         /* combine with 3rd lower 3 bytes */
+    move.l  %d6,(%a1)+      /* store longword */
+    lsl.l   #8,%d7          /* shift up lower 3 bytes */
+    cmp.l   %a0,%a2         /* runs %a0 up to last line bound */
+    jhi     .lines_do3_loop
+
+    lea.l   (12,%a2),%a2    /* readjust end address for doing longwords */
+    cmp.l   %a0,%a2         /* any trailing longwords? */
+    jls     .lines_do3_tail_end    /* no: just store last lower 3 bytes */
+
+.lines_do3_tail_loop:
+    move.l  %d7,%d6         /* move old longword away */
+    move.l  (%a0)+,%d7      /* load new longword */
+    move.l  %d7,%d2
+    lsr.l   %d1,%d2         /* get high byte */
+    or.l    %d2,%d6         /* combine with old lower 3 bytes */
+    move.l  %d6,(%a1)+      /* store longword */
+    lsl.l   #8,%d7          /* shift up lower 3 bytes */
+    cmp.l   %a0,%a2         /* runs %a0 up to last long bound */
+    jhi     .lines_do3_tail_loop
+
+.lines_do3_tail_end:
+    swap    %d7             /* get high word */
+    move.w  %d7,(%a1)+      /* store word */
+    lsr.l   %d1,%d7         /* get moved-up low byte */
+    move.b  %d7,(%a1)+      /* store byte */
+    jra     .lines_end
+
+    /* byte aligned destination (long + 1): use line burst reads in main loop */
+.lines_do1_start:
+    moveq.l #24,%d1         /* shift count for shifting by 3 bytes */
+    cmp.l   %a0,%d0         /* any leading longwords? */
+    jhi     .lines_do1_head_start  /* yes: leading longword copy */
+
+    movem.l (%a0),%d4-%d7   /* load first line */
+    lea.l   (16,%a0),%a0
+    move.l  %d4,%d2         /* first longword, bytes 3210 */
+    lsr.l   #8,%d2          /* first longword, bytes .321 */
+    swap    %d2             /* first longword, bytes 21.3 */
+    move.b  %d2,(%a1)+      /* store byte */
+    swap    %d2             /* first longword, bytes .321 */
+    move.w  %d2,(%a1)+      /* store word */
+    jra     .lines_do1_entry
+
+.lines_do1_head_start:
+    move.l  (%a0)+,%d7      /* load first longword */
+    move.l  %d7,%d2         /* first longword, bytes 3210 */
+    lsr.l   #8,%d2          /* first longword, bytes .321 */
+    swap    %d2             /* first longword, bytes 21.3 */
+    move.b  %d2,(%a1)+      /* store byte */
+    swap    %d2             /* first longword, bytes .321 */
+    move.w  %d2,(%a1)+      /* store word */
+    jra     .lines_do1_head_entry
+
+.lines_do1_head_loop:
+    move.l  %d7,%d6         /* move old longword away */
+    move.l  (%a0)+,%d7      /* load new longword */
+    move.l  %d7,%d2
+    lsr.l   #8,%d2          /* get upper 3 bytes */
+    or.l    %d2,%d6         /* combine with old low byte */
+    move.l  %d6,(%a1)+      /* store longword */
+.lines_do1_head_entry:
+    lsl.l   %d1,%d7         /* shift up low byte */
+    cmp.l   %a0,%d0         /* runs %a0 up to first line bound */
+    jhi     .lines_do1_head_loop
+
+.lines_do1_loop:
+    move.l  %d7,%d3         /* move last longword of old line away */
+    movem.l (%a0),%d4-%d7   /* load new line */
+    lea.l   (16,%a0),%a0
+    move.l  %d4,%d2
+    lsr.l   #8,%d2          /* get upper 3 bytes of 1st longword */
+    or.l    %d2,%d3         /* combine with low byte of old longword */
+    move.l  %d3,(%a1)+      /* store longword */
+.lines_do1_entry:
+    lsl.l   %d1,%d4         /* shift up low byte */
+    move.l  %d5,%d2
+    lsr.l   #8,%d2          /* get upper 3 bytes of 2nd longword */
+    or.l    %d2,%d4         /* combine with low byte of 1st longword */
+    move.l  %d4,(%a1)+      /* store longword */
+    lsl.l   %d1,%d5         /* shift up low byte */
+    move.l  %d6,%d2
+    lsr.l   #8,%d2          /* get upper 3 bytes of 3rd longword */
+    or.l    %d2,%d5         /* combine with low byte of 2nd longword */
+    move.l  %d5,(%a1)+      /* store longword */
+    lsl.l   %d1,%d6         /* shift up low byte */
+    move.l  %d7,%d2
+    lsr.l   #8,%d2          /* get upper 3 bytes of 4th longword */
+    or.l    %d2,%d6         /* combine with low byte of 4th longword */
+    move.l  %d6,(%a1)+      /* store longword */
+    lsl.l   %d1,%d7         /* shift up low byte */
+    cmp.l   %a0,%a2         /* runs %a0 up to last line bound */
+    jhi     .lines_do1_loop
+    
+    lea.l   (12,%a2),%a2    /* readjust end address for doing longwords */
+    cmp.l   %a0,%a2         /* any trailing longwords? */
+    jls     .lines_do1_tail_end    /* no: just store last low byte */
+    
+.lines_do1_tail_loop:
+    move.l  %d7,%d6         /* move old longword away */
+    move.l  (%a0)+,%d7      /* load new longword */
+    move.l  %d7,%d2
+    lsr.l   #8,%d2          /* get upper 3 bytes */
+    or.l    %d2,%d6         /* combine with old low byte */
+    move.l  %d6,(%a1)+      /* store longword */
+    lsl.l   %d1,%d7         /* shift up low byte */
+    cmp.l   %a0,%a2         /* runs %a0 up to last long bound */
+    jhi     .lines_do1_tail_loop
+
+.lines_do1_tail_end:
+    lsr.l   %d1,%d7         /* get shifted-up low byte */
+    move.b  %d7,(%a1)+      /* store byte */
+    jra     .lines_end
+
+    /* long aligned destination (line + 0/4/8/12): head */
+.lines_do0_head_loop:
+    move.l  (%a0)+,(%a1)+   /* copy longword */
+.lines_do0_start:
+    cmp.l   %a0,%d0         /* runs %a0 up to first line bound */
+    jhi     .lines_do0_head_loop
+
+.lines_do0_head_end:
+    move.l  %a1,%d1
+    lsr.l   #2,%d1
+    moveq.l #3,%d0          /* mask */
+    and.l   %d0,%d1
+    moveq.l #16,%d0         /* address increment for one main loop pass */
+    jmp.l   (2,%pc,%d1.l*2) /* switch ((dest_addr >> 2) & 3) */
+    bra.b   .lines_lo0_start
+    bra.b   .lines_lo4_start
+    bra.b   .lines_lo8_start
+ /* bra.b   .lines_lo12_start   implicit */
+
+    /* long aligned destination (line + 12): use line bursts in the loop */
+.lines_lo12_start:
+    movem.l (%a0),%d4-%d7   /* load first line */
+    add.l   %d0,%a0
+    move.l  %d4,(%a1)+      /* store 1st longword */
+    cmp.l   %a0,%a2         /* any full lines? */
+    jls     .lines_lo12_end /* no: skip main loop */
+
+.lines_lo12_loop:
+    move.l  %d5,%d1         /* move last 3 longwords of old line away */
+    move.l  %d6,%d2
+    move.l  %d7,%d3
+    movem.l (%a0),%d4-%d7   /* load new line */
+    add.l   %d0,%a0
+    movem.l %d1-%d4,(%a1)   /* store line (3 old + 1 new longwords) */
+    add.l   %d0,%a1
+    cmp.l   %a0,%a2         /* runs %a0 up to last line bound */
+    jhi     .lines_lo12_loop
+
+    /* long aligned destination (line + 0/4/8/12): tail */
+.lines_lo12_end:
+    move.l  %d5,(%a1)+      /* store 3rd last longword */
+.lines_lo8_end:
+    move.l  %d6,(%a1)+      /* store 2nd last longword */
+.lines_lo4_end:
+    move.l  %d7,(%a1)+      /* store last longword */
+.lines_lo0_end:
+    lea.l   (12,%a2),%a2    /* readjust end address for doing longwords */
+    cmp.l   %a0,%a2         /* any trailing longwords? */
+    jls     .lines_end      /* no: get outta here */
+
+.lines_do0_tail_loop:
+    move.l  (%a0)+,(%a1)+   /* copy longword */
+    cmp.l   %a0,%a2         /* runs %a0 up to last long bound */
+    jhi     .lines_do0_tail_loop
+
+    jra     .lines_end 
+    
+    /* line aligned destination: use line bursts in the loop */
+.lines_lo0_start:
+.lines_lo0_loop:
+    movem.l (%a0),%d4-%d7   /* load line */
+    add.l   %d0,%a0
+    movem.l %d4-%d7,(%a1)   /* store line */
+    add.l   %d0,%a1
+    cmp.l   %a0,%a2         /* runs %a0 up to last line bound */
+    jhi     .lines_lo0_loop
+
+    jra     .lines_lo0_end  /* handle trailing longwords */
+
+    /* long aligned destination (line + 4): use line bursts in the loop */
+.lines_lo4_start:
+    movem.l (%a0),%d4-%d7   /* load first line */
+    add.l   %d0,%a0
+    move.l  %d4,(%a1)+      /* store 1st longword */
+    move.l  %d5,(%a1)+      /* store 2nd longword */
+    move.l  %d6,(%a1)+      /* store 3rd longword */
+    cmp.l   %a0,%a2         /* any full lines? */
+    jls     .lines_lo4_end  /* no: skip main loop */
+
+.lines_lo4_loop:
+    move.l  %d7,%d3         /* move last longword of old line away */
+    movem.l (%a0),%d4-%d7   /* load new line */
+    add.l   %d0,%a0
+    movem.l %d3-%d6,(%a1)   /* store line (1 old + 3 new longwords) */
+    add.l   %d0,%a1
+    cmp.l   %a0,%a2         /* runs %a0 up to last line bound */
+    jhi     .lines_lo4_loop
+
+    jra     .lines_lo4_end  /* handle trailing longwords */
+
+    /* long aligned destination (line + 8): use line bursts in the loop */
+.lines_lo8_start:
+    movem.l (%a0),%d4-%d7   /* load first line */
+    add.l   %d0,%a0
+    move.l  %d4,(%a1)+      /* store 1st longword */
+    move.l  %d5,(%a1)+      /* store 2nd longword */
+    cmp.l   %a0,%a2
+    jls     .lines_lo8_end
+
+.lines_lo8_loop:
+    move.l  %d6,%d2         /* move last 2 longwords of old line away */
+    move.l  %d7,%d3
+    movem.l (%a0),%d4-%d7   /* load new line */
+    add.l   %d0,%a0
+    movem.l %d2-%d5,(%a1)   /* store line (2 old + 2 new longwords) */
+    add.l   %d0,%a1
+    cmp.l   %a0,%a2         /* runs %a0 up to last line bound */
+    jhi     .lines_lo8_loop
+
+    jra     .lines_lo8_end  /* handle trailing longwords */
+
+#ifdef FULLSPEED
+
+    /* word aligned destination (line + 2/6/10/14): head */
+.lines_do2_start:
+    cmp.l   %a0,%d0         /* any leading longwords? */
+    jls     .lines_do2_selector    /* no: jump to mainloop selector */
+
+    move.l  (%a0)+,%d7      /* load first longword */
+    swap    %d7             /* swap words */
+    move.w  %d7,(%a1)+      /* store high word */
+    cmp.l   %a0,%d0         /* any more longword? */
+    jls     .lines_do2_head_end    /* no: skip head loop */
+
+.lines_do2_head_loop:
+    move.l  %d7,%d6         /* move old longword away */
+    move.l  (%a0)+,%d7      /* load new longword */
+    swap    %d7             /* swap words */
+    move.w  %d7,%d6         /* combine high word with old low word */
+    move.l  %d6,(%a1)+      /* store longword */
+    cmp.l   %a0,%d0         /* runs %a0 up to first line bound */
+    jhi     .lines_do2_head_loop
+
+.lines_do2_head_end:
+    swap    %d7             /* undo swap */
+    move.w  %d7,(%a1)+      /* store word */
+
+.lines_do2_selector:
+    move.l  %a1,%d1
+    lsr.l   #2,%d1
+    moveq.l #3,%d0          /* mask */
+    and.l   %d0,%d1
+    moveq.l #16,%d0         /* address increment for one main loop pass */
+    jmp.l   (2,%pc,%d1.l*4) /* switch ((dest_addr >> 2) & 3) */
+    bra.w   .lines_lo2_start
+    bra.w   .lines_lo6_start
+    bra.w   .lines_lo10_start
+ /* bra.w   .lines_lo14_start   implicit */
+
+    /* word aligned destination (line + 14): use line bursts in the loop */
+.lines_lo14_start:
+    movem.l (%a0),%d4-%d7   /* load first line */
+    add.l   %d0,%a0
+    swap    %d4             /* swap words of 1st long */
+    move.w  %d4,(%a1)+      /* store word */
+    jra     .lines_lo14_entry      /* jump into main loop */
+
+.lines_lo14_loop:
+    move.l  %d4,%d0         /* move old line away */
+    move.l  %d5,%d1
+    move.l  %d6,%d2
+    move.l  %d7,%d3
+    movem.l (%a0),%d4-%d7   /* load new line */
+    lea.l   (16,%a0),%a0
+    swap    %d4             /* swap words of 1st long */
+    move.w  %d4,%d3         /* combine 1st high word with old low word */
+    movem.l %d0-%d3,(%a1)   /* store line */
+    lea.l   (16,%a1),%a1
+.lines_lo14_entry:
+    swap    %d5             /* swap words of 2nd long */
+    move.w  %d5,%d4         /* combine 2nd high word with 1st low word */
+    swap    %d6             /* swap words of 3rd long */
+    move.w  %d6,%d5         /* combine 3rd high word with 2nd low word */
+    swap    %d7             /* swap words of 4th long */
+    move.w  %d7,%d6         /* combine 4th high word with 3rd low word */
+    cmp.l   %a0,%a2         /* runs %a0 up to last line bound */
+    jhi     .lines_lo14_loop
+
+    /* word aligned destination (line + 2/6/10/14): tail */
+.lines_lo14_end:
+    move.l  %d4,(%a1)+      /* store third last longword */
+.lines_lo10_end:
+    move.l  %d5,(%a1)+      /* store second last longword */
+.lines_lo6_end:
+    move.l  %d6,(%a1)+      /* store last longword */
+.lines_lo2_end:
+    lea.l   (12,%a2),%a2    /* readjust end address for doing longwords */
+    cmp.l   %a0,%a2         /* any trailing longwords? */
+    jls     .lines_do2_tail_end    /* no: skip tail loop */
+    
+.lines_do2_tail_loop:
+    move.l  %d7,%d6         /* move old longword away */
+    move.l  (%a0)+,%d7      /* load new longword */
+    swap    %d7             /* swap words */
+    move.w  %d7,%d6         /* combine high word with old low word */
+    move.l  %d6,(%a1)+      /* store longword */
+    cmp.l   %a0,%a2         /* runs %a0 up to last long bound */
+    jhi     .lines_do2_tail_loop
+
+.lines_do2_tail_end:
+    swap    %d7             /* undo swap */
+    move.w  %d7,(%a1)+      /* store last word */
+    jra     .lines_end  
+
+    /* word aligned destination (line + 2): use line bursts in the loop */
+.lines_lo2_start:
+    movem.l (%a0),%d4-%d7   /* load first line */
+    add.l   %d0,%a0
+    swap    %d4             /* swap words of 1st long */
+    move.w  %d4,(%a1)+      /* store high word */
+    swap    %d5             /* swap words of 2nd long */
+    move.w  %d5,%d4         /* combine 2nd high word with 1st low word */
+    swap    %d6             /* swap words of 3rd long */
+    move.w  %d6,%d5         /* combine 3nd high word with 2nd low word */
+    swap    %d7             /* swap words of 4th long */
+    move.w  %d7,%d6         /* combine 4th high word with 3rd low word */
+    move.l  %d4,(%a1)+      /* store 1st longword */
+    move.l  %d5,(%a1)+      /* store 2nd longword */
+    move.l  %d6,(%a1)+      /* store 3rd longword */
+    cmp.l   %a0,%a2         /* any full lines? */
+    jls     .lines_lo2_end  /* no: skip main loop */
+
+.lines_lo2_loop:
+    move.l  %d7,%d3         /* move last longword of old line away */
+    movem.l (%a0),%d4-%d7   /* load line */
+    add.l   %d0,%a0
+    swap    %d4             /* swap words of 1st long */
+    move.w  %d4,%d3         /* combine 1st high word with old low word */
+    swap    %d5             /* swap words of 2nd long */
+    move.w  %d5,%d4         /* combine 2nd high word with 1st low word */
+    swap    %d6             /* swap words of 3rd long */
+    move.w  %d6,%d5         /* combine 3rd high word with 2nd low word */
+    swap    %d7             /* swap words of 4th long */
+    move.w  %d7,%d6         /* combine 4th high word with 3rd low word */
+    movem.l %d3-%d6,(%a1)   /* store line */
+    add.l   %d0,%a1
+    cmp.l   %a0,%a2         /* runs %a0 up to last line bound */
+    jhi     .lines_lo2_loop
+
+    jra     .lines_lo2_end  /* handle trailing longwords */
+
+    /* word aligned destination (line + 6): use line bursts in the loop */
+.lines_lo6_start:
+    movem.l (%a0),%d4-%d7   /* load first line */
+    add.l   %d0,%a0
+    swap    %d4             /* swap words of 1st long */
+    move.w  %d4,(%a1)+      /* store high word */
+    swap    %d5             /* swap words of 2nd long */
+    move.w  %d5,%d4         /* combine 2nd high word with 1st low word */
+    swap    %d6             /* swap words of 3rd long */
+    move.w  %d6,%d5         /* combine 3rd high word with 2nd low word */
+    move.l  %d4,(%a1)+      /* store 1st longword */
+    move.l  %d5,(%a1)+      /* store 2nd longword */
+    jra     .lines_lo6_entry       /* jump into main loop */
+
+.lines_lo6_loop:
+    move.l  %d6,%d2         /* move last 2 longwords of old line away */
+    move.l  %d7,%d3
+    movem.l (%a0),%d4-%d7   /* load line */
+    add.l   %d0,%a0
+    swap    %d4             /* swap words of 1st long */
+    move.w  %d4,%d3         /* combine 1st high word with old low word */
+    swap    %d5             /* swap words of 2nd long */
+    move.w  %d5,%d4         /* combine 2nd high word with 1st low word */
+    swap    %d6             /* swap words of 3rd long */
+    move.w  %d6,%d5         /* combine 3rd high word with 2nd low word */
+    movem.l %d2-%d5,(%a1)   /* store line */
+    add.l   %d0,%a1
+.lines_lo6_entry:
+    swap    %d7             /* swap words of 4th long */
+    move.w  %d7,%d6         /* combine 4th high word with 3rd low word */
+    cmp.l   %a0,%a2         /* runs %a0 up to last line bound */
+    jhi     .lines_lo6_loop
+
+    jra     .lines_lo6_end  /* handle trailing longwords */
+
+    /* word aligned destination (line + 10): use line bursts in the loop */
+.lines_lo10_start:
+    movem.l (%a0),%d4-%d7   /* load first line */
+    add.l   %d0,%a0
+    swap    %d4             /* swap words of 1st long */
+    move.w  %d4,(%a1)+      /* store high word */
+    swap    %d5             /* swap words of 2nd long */
+    move.w  %d5,%d4         /* combine 2nd high word with 1st low word */
+    move.l  %d4,(%a1)+      /* store 1st longword */
+    jra     .lines_lo10_entry      /* jump into main loop */
+
+.lines_lo10_loop:
+    move.l  %d5,%d1         /* move last 3 longwords of old line away */
+    move.l  %d6,%d2
+    move.l  %d7,%d3
+    movem.l (%a0),%d4-%d7   /* load line */
+    add.l   %d0,%a0
+    swap    %d4             /* swap words of 1st long */
+    move.w  %d4,%d3         /* combine 1st high word with old low word */
+    swap    %d5             /* swap words of 2nd long */
+    move.w  %d5,%d4         /* combine 2nd high word with 1st low word */
+    movem.l %d1-%d4,(%a1)   /* store line */
+    add.l   %d0,%a1
+.lines_lo10_entry:
+    swap    %d6             /* swap words of 3rd long */
+    move.w  %d6,%d5         /* combine 3rd high word with 2nd low word */
+    swap    %d7             /* swap words of 4th long */
+    move.w  %d7,%d6         /* combine 4th high word with 3rd low word */
+    cmp.l   %a0,%a2         /* runs %a0 up to last line bound */
+    jhi     .lines_lo10_loop
+
+    jra     .lines_lo10_end /* handle trailing longwords */
+
+#else /* !FULLSPEED */
+
+    /* word aligned destination (long + 2): use line burst reads in the loop */
+.lines_do2_start:
+    cmp.l   %a0,%d0         /* any leading longwords? */
+    jhi     .lines_do2_head_start  /* yes: leading longword copy */
+
+    movem.l (%a0),%d4-%d7   /* load first line */
+    lea.l   (16,%a0),%a0
+    swap    %d4             /* swap words of 1st long */
+    move.w  %d4,(%a1)+      /* store high word */
+    jra     .lines_do2_entry       /* jump into main loop */
+
+.lines_do2_head_start:
+    move.l  (%a0)+,%d7      /* load first longword */
+    swap    %d7             /* swap words */
+    move.w  %d7,(%a1)+      /* store high word */
+    cmp.l   %a0,%d0         /* any full longword? */
+    jls     .lines_do2_loop /* no: skip head loop */
+
+.lines_do2_head_loop:
+    move.l  %d7,%d6         /* move old longword away */
+    move.l  (%a0)+,%d7      /* load new longword */
+    swap    %d7             /* swap words */
+    move.w  %d7,%d6         /* combine high word with old low word */
+    move.l  %d6,(%a1)+      /* store longword */
+    cmp.l   %a0,%d0         /* runs %a0 up to first line bound */
+    jhi     .lines_do2_head_loop
+
+.lines_do2_loop:
+    move.l  %d7,%d3         /* move last longword of old line away */
+    movem.l (%a0),%d4-%d7   /* load line */
+    lea.l   (16,%a0),%a0
+    swap    %d4             /* swap words of 1st long */
+    move.w  %d4,%d3         /* combine 1st high word with old low word */
+    move.l  %d3,(%a1)+      /* store 1st longword */
+.lines_do2_entry:
+    swap    %d5             /* swap words of 2nd long */
+    move.w  %d5,%d4         /* combine 2nd high word with 1st low word */
+    move.l  %d4,(%a1)+      /* store 2nd longword */
+    swap    %d6             /* swap words of 3rd long */
+    move.w  %d6,%d5         /* combine 3rd high word with 2nd low word */
+    move.l  %d5,(%a1)+      /* store 3rd longword */
+    swap    %d7             /* swap words of 4th long */
+    move.w  %d7,%d6         /* combine 4th high word with 3rd low word */
+    move.l  %d6,(%a1)+      /* store 4th longword */
+    cmp.l   %a0,%a2         /* runs %a0 up to last line bound */
+    jhi     .lines_do2_loop
+
+.lines_do2_end:
+    lea.l   (12,%a2),%a2    /* readjust end address for doing longwords */
+    cmp.l   %a0,%a2         /* any trailing longwords? */
+    jls     .lines_do2_tail_end    /* no: skip tail loop */
+    
+.lines_do2_tail_loop:
+    move.l  %d7,%d6         /* move old longword away */
+    move.l  (%a0)+,%d7      /* load new longword */
+    swap    %d7             /* swap words */
+    move.w  %d7,%d6         /* combine high word with old low word */
+    move.l  %d6,(%a1)+      /* store longword */
+    cmp.l   %a0,%a2         /* runs %a0 up to last long bound */
+    jhi     .lines_do2_tail_loop
+
+.lines_do2_tail_end:
+    swap    %d7             /* undo swap */
+    move.w  %d7,(%a1)+      /* store last word */
+ /* jra     .lines_end    implicit */
+
+#endif /* !FULLSPEED */
+
+.lines_end:
+    addq.l  #3,%a2          /* readjust end address */
+    move.l  %a2,%d1         /* end address in %d1 again */
+    movem.l (%sp),%d2-%d7/%a2      /* restore registers */
+    lea.l   (28,%sp),%sp
+    jra     .bytes2_start   /* jump to trailing byte loop */
+
+.long_start:
+    subq.l  #3,%d1          /* adjust end address for doing 4 bytes/ pass */
+
+    /* longword copy loop - no lines */
+.long_loop:
+    move.l  (%a0)+,(%a1)+   /* copy longword (write can be unaligned) */
+    cmp.l   %a0,%d1         /* runs %a0 up to last long bound */
+    jhi     .long_loop
+
+    addq.l  #3,%d1          /* readjust end address */
+    cmp.l   %a0,%d1         /* any bytes left? */
+    jls     .bytes2_end     /* no: skip trailing byte loop */
+
+    /* trailing byte loop */
+.bytes2_loop:
+    move.b  (%a0)+,(%a1)+   /* copy byte */
+.bytes2_start:
+    cmp.l   %a0,%d1         /* runs %a0 up to end address */
+    jhi     .bytes2_loop
+
+.bytes2_end:
+    move.l  (4,%sp),%d0     /* return destination */
+    rts
+
+.end:
+    .size   memcpy,.end-memcpy
diff --git a/firmware/target/coldfire/memmove-coldfire.S b/firmware/target/coldfire/memmove-coldfire.S
new file mode 100755
index 0000000000..bdd2e2e206
--- /dev/null
+++ b/firmware/target/coldfire/memmove-coldfire.S
@@ -0,0 +1,668 @@
+/***************************************************************************
+ *             __________               __   ___.
+ *   Open      \______   \ ____   ____ |  | _\_ |__   _______  ___
+ *   Source     |       _//  _ \_/ ___\|  |/ /| __ \ /  _ \  \/  /
+ *   Jukebox    |    |   (  <_> )  \___|    < | \_\ (  <_> > <  <
+ *   Firmware   |____|_  /\____/ \___  >__|_ \|___  /\____/__/\_ \
+ *                     \/            \/     \/    \/            \/
+ * $Id$
+ *
+ * Copyright (C) 2006 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
+
+#define FULLSPEED /* use burst writing for word aligned destinations */
+    .align  2
+    .global memmove
+    .type   memmove,@function
+
+/* Moves <length> bytes of data in memory from <source> to <dest>
+ * Regions may overlap.
+ * This version is optimized for speed, and needs the corresponding memcpy
+ * implementation for the forward copy branch.
+ *
+ * arguments:
+ *  (4,%sp)  - destination address
+ *  (8,%sp)  - source address
+ *  (12,%sp) - length
+ *
+ * return value:
+ *  %d0 - destination address (like ANSI version)
+ *
+ * register usage:
+ *  %a0 - current source address
+ *  %a1 - current dest address
+ *  %a2 - source start address (in line-copy loops)
+ *  %d0 - source start address (byte and longword copy) / data / scratch
+ *  %d1 - data / scratch
+ *  %d2 - data / scratch
+ *  %d3..%d7 - data
+ *
+ * For maximum speed this routine reads and writes whole lines using burst
+ * move (movem.l) where possible. For byte aligned destinations (long-1 and
+ * long-3) it writes longwords only. Same goes for word aligned destinations
+ * if FULLSPEED is undefined.
+ */
+memmove:
+    move.l  (4,%sp),%a1     /* Destination */
+    move.l  (8,%sp),%a0     /* Source */
+    move.l  (12,%sp),%d1    /* Length */
+
+    cmp.l   %a0,%a1
+    bhi.b   .backward       /* dest > src -> backward copy */
+    jmp     __memcpy_fwd_entry
+
+.backward:
+    move.l  %a0,%d0         /* %d0 = source start */
+    add.l   %d1,%a0         /* %a0 = source end */
+    add.l   %d1,%a1         /* %a1 = destination end */
+
+    move.l  %a0,%d1
+    and.l   #0xFFFFFFFC,%d1 /* %d1 = last source long bound */
+    subq.l  #4,%d1
+    cmp.l   %d0,%d1         /* at least one aligned longword to copy? */
+    blo.w   .bytes2r_start
+
+    addq.l  #4,%d1          /* %d1 = last source long bound */
+    cmp.l   %d1,%a0         /* any bytes to copy */
+    jls     .bytes1r_end    /* no: skip byte loop */
+    
+    /* leading byte loop: copies 0..3 bytes */
+.bytes1r_loop:
+    move.b  -(%a0),-(%a1)   /* copy byte */
+    cmp.l   %d1,%a0         /* runs %a0 down to last long bound */
+    jhi     .bytes1r_loop
+
+.bytes1r_end:
+    moveq.l #-16,%d1
+    add.l   %a0,%d1
+    and.l   #0xFFFFFFF0,%d1 /* %d1 = last source line bound - 16 */
+    cmp.l   %d0,%d1         /* at least one aligned line to copy? */
+    blo.w   .longr_start    /* no: jump to longword copy loop */
+    
+    lea.l   (-28,%sp),%sp   /* free up some registers */
+    movem.l %d2-%d7/%a2,(%sp)
+
+    moveq.l #16,%d2
+    add.l   %d2,%d1         /* %d1 = last source line bound */
+    move.l  %d0,%a2         /* %a2 = start address */
+    lea.l   (15,%a2),%a2    /* adjust start address for loops doing 16 bytes/pass */
+    move.l  %a1,%d0
+    moveq.l #3,%d2          /* mask */
+    and.l   %d2,%d0
+    jmp.l   (2,%pc,%d0.l*4) /* switch (dest_addr & 3) */
+    bra.w   .lines_do0r_start
+    bra.w   .lines_do1r_start
+    bra.w   .lines_do2r_start
+ /* bra.w   .lines_do3r_start   implicit */
+
+    /* byte aligned destination (long - 1): use line burst reads in main loop */
+.lines_do3r_start:
+    moveq.l #24,%d0         /* shift count for shifting by 3 bytes */
+    cmp.l   %d1,%a0         /* any leading longwords? */
+    jhi     .lines_do3r_head_start  /* yes: leading longword copy */
+    
+    lea.l   (-16,%a0),%a0
+    movem.l (%a0),%d3-%d6   /* load initial line */
+    move.l  %d6,%d2         /* last longword, bytes 3210 */
+    move.b  %d2,-(%a1)      /* store byte */
+    lsr.l   #8,%d2          /* last longword, bytes .321 */
+    move.w  %d2,-(%a1)      /* store word */
+    jra     .lines_do3r_entry
+    
+.lines_do3r_head_start:
+    move.l  -(%a0),%d3      /* load initial longword */
+    move.l  %d3,%d2         /* bytes 3210 */
+    move.b  %d2,-(%a1)      /* store byte */
+    lsr.l   #8,%d2          /* bytes .321 */
+    move.w  %d2,-(%a1)      /* store word */
+    jra     .lines_do3r_head_entry
+    
+.lines_do3r_head_loop:
+    move.l  %d3,%d4         /* move old longword away */
+    move.l  -(%a0),%d3      /* load new longword */
+    move.l  %d3,%d2     
+    lsl.l   #8,%d2          /* get bytes 210. */
+    or.l    %d2,%d4         /* combine with old high byte */
+    move.l  %d4,-(%a1)      /* store longword */
+.lines_do3r_head_entry:
+    lsr.l   %d0,%d3         /* shift down high byte */
+    cmp.l   %d1,%a0         /* run %a0 down to last line bound */
+    jhi     .lines_do3r_head_loop
+
+.lines_do3r_loop:
+    move.l  %d3,%d7         /* move first longword of last line away */
+    lea.l   (-16,%a0),%a0
+    movem.l (%a0),%d3-%d6   /* load new line */
+    move.l  %d6,%d2
+    lsl.l   #8,%d2          /* get bytes 210. of 4th longword */
+    or.l    %d2,%d7         /* combine with high byte of old longword */
+    move.l  %d7,-(%a1)      /* store longword */
+.lines_do3r_entry:
+    lsr.l   %d0,%d6         /* shift down high byte */
+    move.l  %d5,%d2
+    lsl.l   #8,%d2          /* get bytes 210. of 3rd longword */
+    or.l    %d2,%d6         /* combine with high byte of 4th longword */
+    move.l  %d6,-(%a1)      /* store longword */
+    lsr.l   %d0,%d5         /* shift down high byte */
+    move.l  %d4,%d2
+    lsl.l   #8,%d2          /* get bytes 210. of 2nd longword */
+    or.l    %d2,%d5         /* combine with high byte or 3rd longword */
+    move.l  %d5,-(%a1)      /* store longword */
+    lsr.l   %d0,%d4         /* shift down high byte */
+    move.l  %d3,%d2
+    lsl.l   #8,%d2          /* get bytes 210. of 1st longword */
+    or.l    %d2,%d4         /* combine with high byte of 2nd longword */
+    move.l  %d4,-(%a1)      /* store longword */
+    lsr.l   %d0,%d3         /* shift down high byte */
+    cmp.l   %a2,%a0         /* run %a0 down to first line bound */
+    jhi     .lines_do3r_loop
+
+    lea.l   (-12,%a2),%a2   /* readjust start address for doing longwords */
+    cmp.l   %a2,%a0         /* any trailing longwords? */
+    jls     .lines_do3r_tail_end    /* no: just store last high byte */
+
+.lines_do3r_tail_loop:
+    move.l  %d3,%d4         /* move old longword away */
+    move.l  -(%a0),%d3      /* load new longword */
+    move.l  %d3,%d2   
+    lsl.l   #8,%d2          /* get bytes 210. */
+    or.l    %d2,%d4         /* combine with old high byte */
+    move.l  %d4,-(%a1)      /* store longword */
+    lsr.l   %d0,%d3         /* shift down high byte */
+    cmp.l   %a2,%a0         /* run %a0 down to first long bound */
+    jhi     .lines_do3r_tail_loop
+    
+.lines_do3r_tail_end:
+    move.b  %d3,-(%a1)      /* store shifted-down high byte */
+    jra     .linesr_end
+    
+    /* byte aligned destination (long - 3): use line burst reads in main loop */
+.lines_do1r_start:
+    moveq.l #24,%d0         /* shift count for shifting by 3 bytes */
+    cmp.l   %d1,%a0         /* any leading longwords? */
+    jhi     .lines_do1r_head_start  /* yes: leading longword copy */
+
+    lea.l   (-16,%a0),%a0
+    movem.l (%a0),%d3-%d6   /* load initial line */
+    move.b  %d6,-(%a1)      /* store low byte of last longword */
+    jra     .lines_do1r_entry
+    
+.lines_do1r_head_start:
+    move.l  -(%a0),%d3      /* load initial longword */
+    move.b  %d3,-(%a1)      /* store low byte */
+    jra     .lines_do1r_head_entry
+
+.lines_do1r_head_loop:
+    move.l  %d3,%d4         /* move old longword away */
+    move.l  -(%a0),%d3      /* load new longword */
+    move.l  %d3,%d2
+    lsl.l   %d0,%d2         /* get low byte */
+    or.l    %d2,%d4         /* combine with old bytes .321 */
+    move.l  %d4,-(%a1)      /* store longword */
+.lines_do1r_head_entry:
+    lsr.l   #8,%d3          /* get bytes .321 */
+    cmp.l   %d1,%a0         /* run %a0 down to last line bound */
+    jhi     .lines_do1r_head_loop
+
+.lines_do1r_loop:
+    move.l  %d3,%d7         /* move first longword of old line away */
+    lea.l   (-16,%a0),%a0
+    movem.l (%a0),%d3-%d6   /* load new line */
+    move.l  %d6,%d2
+    lsl.l   %d0,%d2         /* get low byte of 4th longword */
+    or.l    %d2,%d7         /* combine with bytes .321 of old longword */
+    move.l  %d7,-(%a1)      /* store longword */
+.lines_do1r_entry:
+    lsr.l   #8,%d6          /* get bytes .321 */
+    move.l  %d5,%d2
+    lsl.l   %d0,%d2         /* get low byte of 3rd longword */
+    or.l    %d2,%d6         /* combine with bytes .321 of 4th longword */
+    move.l  %d6,-(%a1)      /* store longword */
+    lsr.l   #8,%d5          /* get bytes .321 */
+    move.l  %d4,%d2
+    lsl.l   %d0,%d2         /* get low byte of 2nd longword */
+    or.l    %d2,%d5         /* combine with bytes .321 of 3rd longword */
+    move.l  %d5,-(%a1)      /* store longword */
+    lsr.l   #8,%d4          /* get bytes .321 */
+    move.l  %d3,%d2
+    lsl.l   %d0,%d2         /* get low byte of 1st longword */
+    or.l    %d2,%d4         /* combine with bytes .321 of 2nd longword */
+    move.l  %d4,-(%a1)      /* store longword */
+    lsr.l   #8,%d3          /* get bytes .321 */
+    cmp.l   %a2,%a0         /* run %a0 down to first line bound */
+    jhi     .lines_do1r_loop
+
+    lea.l   (-12,%a2),%a2   /* readjust start address for doing longwords */
+    cmp.l   %a2,%a0         /* any trailing longwords? */
+    jls     .lines_do1r_tail_end    /* no: just store last high byte */
+
+.lines_do1r_tail_loop:
+    move.l  %d3,%d4         /* move old longword away */
+    move.l  -(%a0),%d3      /* load new longword */
+    move.l  %d3,%d2
+    lsl.l   %d0,%d2         /* get low byte */
+    or.l    %d2,%d4         /* combine with old bytes .321 */
+    move.l  %d4,-(%a1)      /* store longword */
+    lsr.l   #8,%d3          /* get bytes .321 */
+    cmp.l   %a2,%a0         /* run %a0 down to first long bound */
+    jhi     .lines_do1r_tail_loop
+    
+.lines_do1r_tail_end:
+    move.w  %d3,-(%a1)      /* store word 21 */
+    swap    %d3
+    move.b  %d3,-(%a1)      /* store byte 3 */
+    jra     .linesr_end
+
+    /* long aligned destination (line - 0/4/8/12): head */
+.lines_do0r_head_loop:
+    move.l  -(%a0),-(%a1)   /* copy longword */
+.lines_do0r_start:
+    cmp.l   %d1,%a0         /* run %a0 down to last line bound */
+    jhi     .lines_do0r_head_loop
+
+.lines_do0r_head_end:
+    move.l  %a1,%d1
+    lsr.l   #2,%d1
+    moveq.l #3,%d0          /* mask */
+    and.l   %d0,%d1
+    moveq.l #16,%d0         /* address decrement for one main loop pass */
+    jmp.l   (2,%pc,%d1.l*2) /* switch ((dest_addr >> 2) & 3) */
+    bra.b   .lines_lo0r_start
+    bra.b   .lines_lo4r_start
+    bra.b   .lines_lo8r_start
+ /* bra.b   .lines_lo12r_start   implicit */
+
+    /* long aligned destination (line - 4): use line bursts in the loop */
+.lines_lo12r_start:
+    sub.l   %d0,%a0
+    movem.l (%a0),%d1-%d4   /* load initial line */
+    move.l  %d4,-(%a1)      /* store 4th longword */
+    move.l  %d3,-(%a1)      /* store 3rd longword */
+    move.l  %d2,-(%a1)      /* store 2nd longword */
+    cmp.l   %a2,%a0         /* any full lines? */
+    jls     .lines_lo12r_end   /* no: skip main loop */
+    
+.lines_lo12r_loop:
+    move.l  %d1,%d5         /* move first longword of old line away */
+    sub.l   %d0,%a0
+    movem.l (%a0),%d1-%d4   /* load new line */
+    sub.l   %d0,%a1
+    movem.l %d2-%d5,(%a1)   /* store line (1 old + 3 new longwords */
+    cmp.l   %a2,%a0         /* run %a0 down to first line bound */
+    jhi     .lines_lo12r_loop
+
+    jra     .lines_lo12r_end   /* handle trailing longwords */
+
+    /* line aligned destination: use line bursts in the loop */
+.lines_lo0r_start:
+.lines_lo0r_loop:
+    sub.l   %d0,%a0
+    movem.l (%a0),%d1-%d4   /* load line */
+    sub.l   %d0,%a1
+    movem.l %d1-%d4,(%a1)   /* store line */
+    cmp.l   %a2,%a0         /* run %a0 down to first line bound */
+    jhi     .lines_lo0r_loop
+
+    jra     .lines_lo0r_end /* handle trailing longwords */
+
+    /* long aligned destination (line - 8): use line bursts in the loop */
+.lines_lo8r_start:
+    sub.l   %d0,%a0
+    movem.l (%a0),%d1-%d4   /* load initial line */
+    move.l  %d4,-(%a1)      /* store 4th longword */
+    move.l  %d3,-(%a1)      /* store 3rd longword */
+    cmp.l   %a2,%a0         /* any full lines? */
+    jls     .lines_lo8r_end /* no: skip main loop */
+    
+.lines_lo8r_loop:
+    move.l  %d2,%d6         /* move first 2 longwords of old line away */
+    move.l  %d1,%d5         
+    sub.l   %d0,%a0
+    movem.l (%a0),%d1-%d4   /* load new line */
+    sub.l   %d0,%a1
+    movem.l %d3-%d6,(%a1)   /* store line (2 old + 2 new longwords */
+    cmp.l   %a2,%a0         /* run %a0 down to first line bound */
+    jhi     .lines_lo8r_loop
+
+    jra     .lines_lo8r_end /* handle trailing longwords */
+
+    /* long aligned destination (line - 12): use line bursts in the loop */
+.lines_lo4r_start:
+    sub.l   %d0,%a0
+    movem.l (%a0),%d1-%d4   /* load initial line */
+    move.l  %d4,-(%a1)      /* store 4th longword */
+    cmp.l   %a2,%a0         /* any full lines? */
+    jls     .lines_lo4r_end /* no: skip main loop */
+    
+.lines_lo4r_loop:
+    move.l  %d3,%d7         /* move first 3 longwords of old line away */
+    move.l  %d2,%d6         
+    move.l  %d1,%d5         
+    sub.l   %d0,%a0
+    movem.l (%a0),%d1-%d4   /* load new line */
+    sub.l   %d0,%a1
+    movem.l %d4-%d7,(%a1)   /* store line (3 old + 1 new longwords */
+    cmp.l   %a2,%a0         /* run %a0 down to first line bound */
+    jhi     .lines_lo4r_loop
+
+    /* long aligned destination (line - 0/4/8/12): tail */
+.lines_lo4r_end:
+    move.l  %d3,-(%a1)      /* store 3rd last longword */
+.lines_lo8r_end:
+    move.l  %d2,-(%a1)      /* store 2nd last longword */
+.lines_lo12r_end:
+    move.l  %d1,-(%a1)      /* store last longword */
+.lines_lo0r_end:
+    lea.l   (-12,%a2),%a2   /* readjust end address for doing longwords */
+    cmp.l   %a2,%a0         /* any trailing longwords? */
+    jls     .linesr_end     /* no: get outta here */
+
+.lines_do0r_tail_loop:
+    move.l  -(%a0),-(%a1)   /* copy longword */
+    cmp.l   %a2,%a0         /* run %a0 down to first long bound */
+    jhi     .lines_do0r_tail_loop
+
+    jra     .linesr_end
+    
+#ifdef FULLSPEED
+    /* word aligned destination (line - 2/6/10/14): head */
+.lines_do2r_start:
+    cmp.l   %d1,%a0         /* any leading longwords? */
+    jls     .lines_do2r_selector    /* no: jump to mainloop selector */
+    
+    move.l  -(%a0),%d3      /* load initial longword */
+    move.w  %d3,-(%a1)      /* store low word */
+    cmp.l   %d1,%a0         /* any more longwords? */
+    jls     .lines_do2r_head_end    /* no: skip head loop */
+    
+.lines_do2r_head_loop:
+    move.l  %d3,%d4         /* move old longword away */
+    move.l  -(%a0),%d3      /* load new longword */
+    move.w  %d3,%d4         /* combine low word with old high word */
+    swap    %d4             /* swap words */
+    move.l  %d4,-(%a1)      /* store longword */
+    cmp.l   %d1,%a0         /* run %a0 down to last line bound */
+    jhi     .lines_do2r_head_loop
+    
+.lines_do2r_head_end:
+    swap    %d3             /* get high word */
+    move.w  %d3,-(%a1)      /* and store it */
+    
+.lines_do2r_selector:
+    move.l  %a1,%d1
+    lsr.l   #2,%d1
+    moveq.l #3,%d0          /* mask */
+    and.l   %d0,%d1
+    moveq.l #16,%d7         /* address decrement for one main loop pass */
+    jmp.l   (2,%pc,%d1.l*4) /* switch ((dest_addr >> 2) & 3) */
+    bra.w   .lines_lo2r_start
+    bra.w   .lines_lo6r_start
+    bra.w   .lines_lo10r_start
+ /* bra.w   .lines_lo14r_start   implicit */
+
+    /* word aligned destination (line - 2): use line bursts in the loop */
+.lines_lo14r_start:
+    sub.l   %d7,%a0
+    movem.l (%a0),%d0-%d3   /* load initial line */
+    move.w  %d3,-(%a1)      /* store last low word */
+    move.w  %d2,%d3         /* combine 3rd low word with 4th high word */
+    swap    %d3             /* swap words of 3rd long */
+    move.w  %d1,%d2         /* combine 2nd low word with 3rd high word */
+    swap    %d2             /* swap words of 2nd long */
+    move.w  %d0,%d1         /* combine 1st low word with 2nd high word */
+    swap    %d1             /* swap words of 1st long */
+    move.l  %d3,-(%a1)      /* store 3rd longword */
+    move.l  %d2,-(%a1)      /* store 2nd longword */
+    move.l  %d1,-(%a1)      /* store 1st longword */
+    cmp.l   %a2,%a0         /* any full lines? */
+    jls     .lines_lo14r_end   /* no: skip main loop */
+
+.lines_lo14r_loop:
+    move.l  %d0,%d4         /* move first longword of old line away */
+    sub.l   %d7,%a0
+    movem.l (%a0),%d0-%d3   /* load line */
+    move.w  %d3,%d4         /* combine 4th low word with old high word */
+    swap    %d4             /* swap words of 4th long */
+    move.w  %d2,%d3         /* combine 3rd low word with 4th high word */
+    swap    %d3             /* swap words of 3rd long */
+    move.w  %d1,%d2         /* combine 2nd low word with 3rd high word */
+    swap    %d2             /* swap words of 2nd long */
+    move.w  %d0,%d1         /* combine 1st low word with 2nd high word */
+    swap    %d1             /* swap words of 1st long */
+    sub.l   %d7,%a1
+    movem.l %d1-%d4,(%a1)   /* store line */
+    cmp.l   %a2,%a0         /* run %a0 down to first line bound */
+    jhi     .lines_lo14r_loop
+    
+    jra     .lines_lo14r_end   /* handle trailing longwords */
+
+    /* word aligned destination (line - 6): use line bursts in the loop */
+.lines_lo10r_start:
+    sub.l   %d7,%a0
+    movem.l (%a0),%d0-%d3   /* load initial line */
+    move.w  %d3,-(%a1)      /* store last low word */
+    move.w  %d2,%d3         /* combine 3rd low word with 4th high word */
+    swap    %d3             /* swap words of 3rd long */
+    move.w  %d1,%d2         /* combine 2nd low word with 3rd high word */
+    swap    %d2             /* swap words of 2nd long */
+    move.l  %d3,-(%a1)      /* store 3rd longword */
+    move.l  %d2,-(%a1)      /* store 2nd longword */
+    jra     .lines_lo10r_entry /* jump into main loop */
+    
+.lines_lo10r_loop:
+    move.l  %d0,%d4         /* move first 2 longwords of old line away */
+    move.l  %d1,%d5
+    sub.l   %d7,%a0
+    movem.l (%a0),%d0-%d3   /* load line */
+    move.w  %d3,%d4         /* combine 4th low word with old high word */
+    swap    %d4             /* swap words of 4th long */
+    move.w  %d2,%d3         /* combine 3rd low word with 4th high word */
+    swap    %d3             /* swap words of 3rd long */
+    move.w  %d1,%d2         /* combine 2nd low word with 3rd high word */
+    swap    %d2             /* swap words of 2nd long */
+    sub.l   %d7,%a1
+    movem.l %d2-%d5,(%a1)   /* store line */
+.lines_lo10r_entry:
+    move.w  %d0,%d1         /* combine 1st low word with 2nd high word */
+    swap    %d1             /* swap words of 1st long */
+    cmp.l   %a2,%a0         /* run %a0 down to first line bound */
+    jhi     .lines_lo10r_loop
+    
+    jra     .lines_lo10r_end   /* handle trailing longwords */
+
+    /* word aligned destination (line - 10): use line bursts in the loop */
+.lines_lo6r_start:
+    sub.l   %d7,%a0
+    movem.l (%a0),%d0-%d3   /* load initial line */
+    move.w  %d3,-(%a1)      /* store last low word */
+    move.w  %d2,%d3         /* combine 3rd low word with 4th high word */
+    swap    %d3             /* swap words of 3rd long */
+    move.l  %d3,-(%a1)      /* store 3rd longword */
+    jra     .lines_lo6r_entry  /* jump into main loop */
+    
+.lines_lo6r_loop:
+    move.l  %d0,%d4         /* move first 3 longwords of old line away */
+    move.l  %d1,%d5
+    move.l  %d2,%d6
+    sub.l   %d7,%a0
+    movem.l (%a0),%d0-%d3   /* load line */
+    move.w  %d3,%d4         /* combine 4th low word with old high word */
+    swap    %d4             /* swap words of 4th long */
+    move.w  %d2,%d3         /* combine 3rd low word with 4th high word */
+    swap    %d3             /* swap words of 3rd long */
+    sub.l   %d7,%a1
+    movem.l %d3-%d6,(%a1)   /* store line */
+.lines_lo6r_entry:
+    move.w  %d1,%d2         /* combine 2nd low word with 3rd high word */
+    swap    %d2             /* swap words of 2nd long */
+    move.w  %d0,%d1         /* combine 1st low word with 2nd high word */
+    swap    %d1             /* swap words of 1st long */
+    cmp.l   %a2,%a0         /* run %a0 down to first line bound */
+    jhi     .lines_lo6r_loop
+    
+    jra     .lines_lo6r_end /* handle trailing longwords */
+
+    /* word aligned destination (line - 14): use line bursts in the loop */
+.lines_lo2r_start:
+    sub.l   %d7,%a0
+    movem.l (%a0),%d0-%d3   /* load initial line */
+    move.w  %d3,-(%a1)      /* store last low word */
+    jra     .lines_lo2r_entry  /* jump into main loop */
+    
+.lines_lo2r_loop:
+    move.l  %d0,%d4         /* move old line away */
+    move.l  %d1,%d5
+    move.l  %d2,%d6
+    move.l  %d3,%d7
+    lea.l   (-16,%a0),%a0
+    movem.l (%a0),%d0-%d3   /* load line */
+    move.w  %d3,%d4         /* combine 4th low word with old high word */
+    swap    %d4             /* swap words of 4th long */
+    lea.l   (-16,%a1),%a1
+    movem.l %d4-%d7,(%a1)   /* store line */
+.lines_lo2r_entry:
+    move.w  %d2,%d3         /* combine 3rd low word with 4th high word */
+    swap    %d3             /* swap words of 3rd long */
+    move.w  %d1,%d2         /* combine 2nd low word with 3rd high word */
+    swap    %d2             /* swap words of 2nd long */
+    move.w  %d0,%d1         /* combine 1st low word with 2nd high word */
+    swap    %d1             /* swap words of 1st long */
+    cmp.l   %a2,%a0         /* run %a0 down to first line bound */
+    jhi     .lines_lo2r_loop
+    
+    /* word aligned destination (line - 2/6/10/14): tail */
+.lines_lo2r_end:
+    move.l  %d3,-(%a1)      /* store third last longword */
+.lines_lo6r_end:
+    move.l  %d2,-(%a1)      /* store second last longword */
+.lines_lo10r_end:
+    move.l  %d1,-(%a1)      /* store last longword */
+.lines_lo14r_end:
+    lea.l   (-12,%a2),%a2   /* readjust start address for doing longwords */
+    cmp.l   %a2,%a0         /* any trailing longwords? */
+    jls     .lines_do2r_tail_end    /* no: skip tail loop */
+
+.lines_do2r_tail_loop:
+    move.l  %d0,%d1         /* move old longword away */
+    move.l  -(%a0),%d0      /* load new longword */
+    move.w  %d0,%d1         /* combine low word with old high word */
+    swap    %d1             /* swap words */
+    move.l  %d1,-(%a1)      /* store longword */
+    cmp.l   %a2,%a0         /* run %a0 down to first long bound */
+    jhi     .lines_do2r_tail_loop
+    
+.lines_do2r_tail_end:
+    swap    %d0             /* get final high word */
+    move.w  %d0,-(%a1)      /* store it */
+ /* jra     .linesr_end    implicit */
+
+#else /* !FULLSPEED */
+
+    /* word aligned destination (long - 2): use line burst reads in the loop */
+.lines_do2r_start:
+    cmp.l   %d1,%a0         /* any leading longwords? */
+    jhi     .lines_do2r_head_start  /* yes: leading longword copy */
+
+    lea.l   (-16,%a0),%a0
+    movem.l (%a0),%d3-%d6   /* load initial line */
+    move.w  %d6,-(%a1)      /* store last low word */
+    jra     .lines_do2r_entry  /* jump into main loop */
+    
+.lines_do2r_head_start:
+    move.l  -(%a0),%d3      /* load initial longword */
+    move.w  %d3,-(%a1)      /* store low word */
+    cmp.l   %d1,%a0         /* any full longword? */
+    jls     .lines_do2r_loop   /* no: skip head loop */
+    
+.lines_do2r_head_loop:
+    move.l  %d3,%d4         /* move old longword away */
+    move.l  -(%a0),%d3      /* load new longword */
+    move.w  %d3,%d4         /* combine low word with old high word */
+    swap    %d4             /* swap words */
+    move.l  %d4,-(%a1)      /* store longword */
+    cmp.l   %d1,%a0         /* run %a0 down to last line bound */
+    jhi     .lines_do2r_head_loop
+    
+.lines_do2r_loop:
+    move.l  %d3,%d7         /* move first longword of old line away */
+    lea.l   (-16,%a0),%a0
+    movem.l (%a0),%d3-%d6   /* load line */
+    move.w  %d6,%d7         /* combine 4th low word with old high word */
+    swap    %d7             /* swap words of 4th long */
+    move.l  %d7,-(%a1)      /* store 4th longword */
+.lines_do2r_entry:
+    move.w  %d5,%d6         /* combine 3rd low word with 4th high word */
+    swap    %d6             /* swap words of 3rd long */
+    move.l  %d6,-(%a1)      /* store 3rd longword */
+    move.w  %d4,%d5         /* combine 2nd low word with 3rd high word */
+    swap    %d5             /* swap words of 2nd long */
+    move.l  %d5,-(%a1)      /* store 2nd longword */
+    move.w  %d3,%d4         /* combine 1st low word with 2nd high word */
+    swap    %d4             /* swap words of 1st long */
+    move.l  %d4,-(%a1)      /* store 1st longword */
+    cmp.l   %a2,%a0         /* run %a0 down to first line bound */
+    jhi     .lines_do2r_loop
+
+.lines_do2r_end:
+    lea.l   (-12,%a2),%a2   /* readjust start address for doing longwords */
+    cmp.l   %a2,%a0         /* any trailing longwords? */
+    jls     .lines_do2r_tail_end   /* no: skip tail loop */
+    
+.lines_do2r_tail_loop:
+    move.l  %d3,%d4         /* move old longword away */
+    move.l  -(%a0),%d3      /* load new longword */
+    move.w  %d3,%d4         /* combine low word with old high word */
+    swap    %d4             /* swap words */
+    move.l  %d4,-(%a1)      /* store longword */
+    cmp.l   %a2,%a0         /* run %a0 down to first long bound */
+    jhi     .lines_do2r_tail_loop
+
+.lines_do2r_tail_end:
+    swap    %d3             /* get final high word */
+    move.w  %d3,-(%a1)      /* store it */
+ /* jra     .linesr_end    implicit */
+
+#endif /* !FULLSPEED */
+
+.linesr_end:
+    subq.l  #3,%a2          /* readjust end address */
+    move.l  %a2,%d0         /* start address in %d0 again */
+    movem.l (%sp),%d2-%d7/%a2      /* restore registers */
+    lea.l   (28,%sp),%sp
+    jra     .bytes2r_start  /* jump to trailing byte loop */
+
+.longr_start:
+    addq.l  #3,%d0          /* adjust start address for doing 4 bytes/ pass */
+
+    /* longword copy loop - no lines */
+.longr_loop:
+    move.l  -(%a0),-(%a1)   /* copy longword (write can be unaligned) */
+    cmp.l   %d0,%a0         /* runs %a0 down to first long bound */
+    jhi     .longr_loop
+
+    subq.l  #3,%d0          /* readjust start address */
+    cmp.l   %d0,%a0         /* any bytes left? */
+    jls     .bytes2r_end    /* no: skip trailing byte loop */
+
+    /* trailing byte loop */
+.bytes2r_loop:
+    move.b  -(%a0),-(%a1)   /* copy byte */
+.bytes2r_start:
+    cmp.l   %d0,%a0         /* runs %a0 down to start address */
+    jhi     .bytes2r_loop
+
+.bytes2r_end:
+    rts                     /* returns start address */
+
+.end:
+    .size   memmove,.end-memmove
diff --git a/firmware/target/coldfire/memset-coldfire.S b/firmware/target/coldfire/memset-coldfire.S
new file mode 100755
index 0000000000..7c9fe88463
--- /dev/null
+++ b/firmware/target/coldfire/memset-coldfire.S
@@ -0,0 +1,150 @@
+/***************************************************************************
+ *             __________               __   ___.
+ *   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
+
+    .align      2
+    .global     memset
+    .type       memset,@function
+
+/* Fills a memory region with specified byte value
+ * This version is optimized for speed
+ *
+ * arguments:
+ *  (4,%sp)  - start address
+ *  (8,%sp)  - data
+ *  (12,%sp) - length
+ *
+ * return value:
+ *  %d0 - start address (like ANSI version)
+ *
+ * register usage:
+ *  %d0 - data (spread to all 4 bytes when using long stores)
+ *  %d1 - temporary / data (for burst transfer)
+ *  %d2 - data (for burst transfer)
+ *  %d3 - data (for burst transfer)
+ *  %a0 - start address
+ *  %a1 - current address (runs down from end to start)
+ *
+ * For maximum speed this routine uses both long stores and burst mode,
+ * storing whole lines with movem.l. The routine fills memory from end
+ * to start in order to ease returning the start address.
+ */
+memset:
+    move.l  (4,%sp),%a0     /* start address */
+    move.l  (8,%sp),%d0     /* data */
+    move.l  (12,%sp),%a1    /* length */
+    add.l   %a0,%a1         /* %a1 = end address */
+
+    move.l  %a0,%d1
+    addq.l  #7,%d1
+    and.l   #0xFFFFFFFC,%d1 /* %d1 = first long bound + 4 */
+    cmp.l   %d1,%a1         /* at least one aligned longword to fill? */
+    blo.b   .no_longs       /* no, jump directly to byte loop */
+
+    and.l   #0xFF,%d0       /* start: spread data to all 4 bytes */
+    move.l  %d0,%d1
+    lsl.l   #8,%d1
+    or.l    %d1,%d0         /* data now in 2 lower bytes of %d0 */
+    move.l  %d0,%d1
+    swap    %d0
+    or.l    %d1,%d0         /* data now in all 4 bytes of %d0 */
+
+    move.l  %a1,%d1
+    and.l   #0xFFFFFFFC,%d1 /* %d1 = last long bound */
+    cmp.l   %d1,%a1         /* any bytes to set? */
+    bls.b   .end_b1         /* no: skip byte loop */
+
+    /* leading byte loop: sets 0..3 bytes */
+.loop_b1:
+    move.b  %d0,-(%a1)      /* store byte */
+    cmp.l   %d1,%a1         /* runs %a1 down to last long bound */
+    bhi.b   .loop_b1
+    
+.end_b1:
+    moveq.l #31,%d1
+    add.l   %a0,%d1
+    and.l   #0xFFFFFFF0,%d1 /* %d1 = first line bound + 16 */
+    cmp.l   %d1,%a1         /* at least one full line to fill? */
+    blo.b   .no_lines       /* no, jump to longword loop */
+
+    mov.l   %a1,%d1
+    and.l   #0xFFFFFFF0,%d1 /* %d1 = last line bound */
+    cmp.l   %d1,%a1         /* any longwords to set? */
+    bls.b   .end_l1         /* no: skip longword loop */
+
+    /* leading longword loop: sets 0..3 longwords */
+.loop_l1:
+    move.l  %d0,-(%a1)      /* store longword */
+    cmp.l   %d1,%a1         /* runs %a1 down to last line bound */
+    bhi.b   .loop_l1
+
+.end_l1:
+    move.l  %d2,-(%sp)      /* free some registers */
+    move.l  %d3,-(%sp)
+
+    move.l  %d0,%d1         /* spread data to 4 data registers */
+    move.l  %d0,%d2
+    move.l  %d0,%d3
+    lea.l   (15,%a0),%a0    /* start address += 15, acct. for trl. data */
+    
+    /* main loop: set whole lines utilising burst mode */
+.loop_line:
+    lea.l   (-16,%a1),%a1   /* pre-decrement */
+    movem.l %d0-%d3,(%a1)   /* store line */
+    cmp.l   %a0,%a1         /* runs %a1 down to first line bound */
+    bhi.b   .loop_line
+
+    lea.l   (-15,%a0),%a0   /* correct start address */
+    move.l  (%sp)+,%d3      /* restore registers */
+    move.l  (%sp)+,%d2
+
+    move.l  %a0,%d1         /* %d1 = start address ... */
+    addq.l  #3,%d1          /* ... +3, account for possible trailing bytes */
+    cmp.l   %d1,%a1         /* any longwords left */
+    bhi.b   .loop_l2        /* yes: jump to longword loop */
+    bra.b   .no_longs       /* no: skip loop */
+
+.no_lines:
+    move.l  %a0,%d1         /* %d1 = start address ... */
+    addq.l  #3,%d1          /* ... +3, account for possible trailing bytes */
+
+    /* trailing longword loop */
+.loop_l2:
+    move.l  %d0,-(%a1)      /* store longword */
+    cmp.l   %d1,%a1         /* runs %a1 down to first long bound */
+    bhi.b   .loop_l2
+
+.no_longs:
+    cmp.l   %a0,%a1         /* any bytes left? */
+    bls.b   .end_b2         /* no: skip loop */
+
+    /* trailing byte loop */
+.loop_b2:
+    move.b  %d0,-(%a1)      /* store byte */
+    cmp.l   %a0,%a1         /* runs %a1 down to start address */
+    bhi.b   .loop_b2
+
+.end_b2:
+    move.l  %a0,%d0         /* return start address */
+    rts
+
+.end:
+    .size   memset,.end-memset
diff --git a/firmware/target/coldfire/memset16-coldfire.S b/firmware/target/coldfire/memset16-coldfire.S
new file mode 100755
index 0000000000..d9f72f683f
--- /dev/null
+++ b/firmware/target/coldfire/memset16-coldfire.S
@@ -0,0 +1,144 @@
+/***************************************************************************
+ *             __________               __   ___.
+ *   Open      \______   \ ____   ____ |  | _\_ |__   _______  ___
+ *   Source     |       _//  _ \_/ ___\|  |/ /| __ \ /  _ \  \/  /
+ *   Jukebox    |    |   (  <_> )  \___|    < | \_\ (  <_> > <  <
+ *   Firmware   |____|_  /\____/ \___  >__|_ \|___  /\____/__/\_ \
+ *                     \/            \/     \/    \/            \/
+ * $Id$
+ *
+ * Copyright (C) 2006 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
+
+    .global     memset16
+    .type       memset16,@function
+
+/* Fills a memory region with specified word value
+ * Start address must be word aligned, length is in words
+ * This version is optimized for speed
+ *
+ * arguments:
+ *  (4,%sp)  - start address
+ *  (8,%sp)  - data
+ *  (12,%sp) - length
+ *
+ * return value:
+ *  %d0 - start address
+ *
+ * register usage:
+ *  %d0 - data (spread to both words when using long stores)
+ *  %d1 - temporary / data (for burst transfer)
+ *  %d2 - data (for burst transfer)
+ *  %d3 - data (for burst transfer)
+ *  %a0 - start address
+ *  %a1 - current address (runs down from end to start)
+ *
+ * For maximum speed this routine uses both long stores and burst mode,
+ * storing whole lines with movem.l. The routine fills memory from end
+ * to start in order to ease returning the start address.
+ */
+memset16:
+    move.l  (4,%sp),%a0     /* start address */
+    move.l  (8,%sp),%d0     /* data */
+    move.l  (12,%sp),%a1    /* length */
+    add.l   %a1,%a1
+    add.l   %a0,%a1         /* %a1 = end address */
+
+    move.l  %a0,%d1
+    addq.l  #6,%d1
+    and.l   #0xFFFFFFFC,%d1 /* %d1 = first long bound + 4 */
+    cmp.l   %d1,%a1         /* at least one aligned longword to fill? */
+    blo.b   .no_longs       /* no, jump directly to word loop */
+
+    and.l   #0xFFFF,%d0     /* start: spread data to both words */
+    move.l  %d0,%d1
+    swap    %d1
+    or.l    %d1,%d0         /* data now in both words */
+
+    move.l  %a1,%d1
+    and.l   #0xFFFFFFFC,%d1 /* %d1 = last long bound */
+    cmp.l   %d1,%a1         /* one extra word? */
+    bls.b   .end_w1         /* no: skip */
+
+    move.w  %d0,-(%a1)      /* set leading word */
+
+.end_w1:
+    moveq.l #30,%d1
+    add.l   %a0,%d1
+    and.l   #0xFFFFFFF0,%d1 /* %d1 = first line bound + 16 */
+    cmp.l   %d1,%a1         /* at least one full line to fill? */
+    blo.b   .no_lines       /* no, jump to longword loop */
+
+    mov.l   %a1,%d1
+    and.l   #0xFFFFFFF0,%d1 /* %d1 = last line bound */
+    cmp.l   %d1,%a1         /* any longwords to set? */
+    bls.b   .end_l1         /* no: skip longword loop */
+
+    /* leading longword loop: sets 0..3 longwords */
+.loop_l1:
+    move.l  %d0,-(%a1)      /* store longword */
+    cmp.l   %d1,%a1         /* runs %a1 down to last line bound */
+    bhi.b   .loop_l1
+
+.end_l1:
+    move.l  %d2,-(%sp)      /* free some registers */
+    move.l  %d3,-(%sp)
+
+    move.l  %d0,%d1         /* spread data to 4 data registers */
+    move.l  %d0,%d2
+    move.l  %d0,%d3
+    lea.l   (14,%a0),%a0    /* start address += 14, acct. for trl. data */
+    
+    /* main loop: set whole lines utilising burst mode */
+.loop_line:
+    lea.l   (-16,%a1),%a1   /* pre-decrement */
+    movem.l %d0-%d3,(%a1)   /* store line */
+    cmp.l   %a0,%a1         /* runs %a1 down to first line bound */
+    bhi.b   .loop_line
+
+    lea.l   (-14,%a0),%a0   /* correct start address */
+    move.l  (%sp)+,%d3      /* restore registers */
+    move.l  (%sp)+,%d2
+
+    move.l  %a0,%d1         /* %d1 = start address ... */
+    addq.l  #2,%d1          /* ... +2, account for possible trailing word */
+    cmp.l   %d1,%a1         /* any longwords left */
+    bhi.b   .loop_l2        /* yes: jump to longword loop */
+    bra.b   .no_longs       /* no: skip loop */
+
+.no_lines:
+    move.l  %a0,%d1         /* %d1 = start address ... */
+    addq.l  #2,%d1          /* ... +2, account for possible trailing word */
+
+    /* trailing longword loop */
+.loop_l2:
+    move.l  %d0,-(%a1)      /* store longword */
+    cmp.l   %d1,%a1         /* runs %a1 down to first long bound */
+    bhi.b   .loop_l2
+
+.no_longs:
+    cmp.l   %a0,%a1         /* any words left? */
+    bls.b   .end_w2         /* no: skip loop */
+
+    /* trailing word loop */
+.loop_w2:
+    move.w  %d0,-(%a1)      /* store word */
+    cmp.l   %a0,%a1         /* runs %a1 down to start address */
+    bhi.b   .loop_w2
+
+.end_w2:
+    move.l  %a0,%d0         /* return start address */
+    rts
+
+.end:
+    .size   memset16,.end-memset16
diff --git a/firmware/target/sh/memcpy-sh.S b/firmware/target/sh/memcpy-sh.S
new file mode 100755
index 0000000000..0b5e086be9
--- /dev/null
+++ b/firmware/target/sh/memcpy-sh.S
@@ -0,0 +1,217 @@
+/***************************************************************************
+ *             __________               __   ___.
+ *   Open      \______   \ ____   ____ |  | _\_ |__   _______  ___
+ *   Source     |       _//  _ \_/ ___\|  |/ /| __ \ /  _ \  \/  /
+ *   Jukebox    |    |   (  <_> )  \___|    < | \_\ (  <_> > <  <
+ *   Firmware   |____|_  /\____/ \___  >__|_ \|___  /\____/__/\_ \
+ *                     \/            \/     \/    \/            \/
+ * $Id$
+ *
+ * Copyright (C) 2004-2005 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
+
+    .align      2
+    .global     _memcpy
+    .global     ___memcpy_fwd_entry
+    .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 / scratch
+ *  r1 - 2nd data / scratch
+ *  r2 - scratch
+ *  r3 - first long bound / adjusted end address (only if >= 11 bytes)
+ *  r4 - current dest address
+ *  r5 - current source address
+ *  r6 - source end address
+ *  r7 - stored dest start address
+ *
+ * The instruction order 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:
+    mov     r4,r7       /* store dest for returning */
+___memcpy_fwd_entry:
+    add     #-8,r4      /* offset for early increment (max. 2 longs) */
+    mov     #11,r0
+    cmp/hs  r0,r6       /* at least 11 bytes to copy? (ensures 2 aligned longs) */
+    add     r5,r6       /* r6 = source_end */
+    bf      .start_b2   /* no: jump directly to byte loop */
+    
+    mov     #3,r0
+    neg     r5,r3
+    and     r0,r3       /* r3 = (4 - align_offset) % 4 */
+    tst     r3,r3       /* already aligned? */
+    bt      .end_b1     /* yes: skip leading byte loop */
+    
+    add     r5,r3       /* r3 = first source long bound */
+
+    /* leading byte loop: copies 0..3 bytes */
+.loop_b1:
+    mov.b   @r5+,r0     /* load byte & increment source addr */
+    add     #1,r4       /* increment dest addr */
+    mov.b   r0,@(7,r4)  /* store byte */
+    cmp/hi  r5,r3       /* 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 */
+.end_b1:
+    mov     #3,r1
+    and     r4,r1       /* r1 = dest alignment offset */
+    mova    .jmptab,r0
+    mov.b   @(r0,r1),r1 /* select appropriate main loop */
+    add     r0,r1
+    mov     r6,r3       /* move end address to r3 */
+    jmp     @r1         /* and jump to it */
+    add     #-7,r3      /* adjust end addr for main loops doing 2 longs/pass */
+    
+    /** main loops, copying 2 longs per pass to profit from fast page mode **/
+
+    /* long aligned destination (fastest) */
+    .align  2
+.loop_do0:
+    mov.l   @r5+,r1     /* load first long & increment source addr */
+    add     #16,r4      /* increment dest addr & account for decrementing stores */
+    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   r0,@-r4     /* store second long */
+    mov.l   r1,@-r4     /* store first long; NOT ALIGNED - no speed loss here! */
+    bt      .loop_do0
+           
+    add     #4,r3       /* readjust end address */
+    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 */
+
+    /* word aligned destination (long + 2) */
+    .align  2
+.loop_do2:
+    mov.l   @r5+,r1     /* load first long & increment source addr */
+    add     #16,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,@-r4     /* store low word of second long */
+    xtrct   r1,r0       /* extract low word of first long & high word of second long */
+    mov.l   r0,@-r4     /* and store as long */
+    swap.w  r1,r0       /* get high word of first long */
+    mov.w   r0,@-r4     /* and store it */
+    bt      .loop_do2
+
+    add     #4,r3       /* readjust end address */
+    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 */
+
+    /* jumptable for loop selector */
+    .align  2
+.jmptab:
+    .byte   .loop_do0 - .jmptab  /* placed in the middle because the SH1 */
+    .byte   .loop_do1 - .jmptab  /* loads bytes sign-extended. Otherwise */
+    .byte   .loop_do2 - .jmptab  /* the last loop would be out of reach */
+    .byte   .loop_do3 - .jmptab  /* of the offset range. */
+
+    /* byte aligned destination (long + 1) */
+    .align  2
+.loop_do1:
+    mov.l   @r5+,r1     /* load first long & increment source addr */
+    add     #16,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.b   r0,@-r4     /* store low byte of second long */
+    shlr8   r0          /* get upper 3 bytes */
+    mov     r1,r2       /* copy first long */
+    shll16  r2          /* move low byte of first long all the way up, .. */
+    shll8   r2
+    or      r2,r0       /* ..combine with the 3 bytes of second long.. */
+    mov.l   r0,@-r4     /* ..and store as long */
+    shlr8   r1          /* get middle 2 bytes */
+    mov.w   r1,@-r4     /* store as word */
+    shlr16  r1          /* get upper byte */
+    mov.b   r1,@-r4     /* and store */
+    bt      .loop_do1
+
+    add     #4,r3       /* readjust end address */
+.last_do13:
+    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     #12,r4      /* increment dest addr */
+    mov.b   r0,@-r4     /* store low byte */
+    shlr8   r0          /* get middle 2 bytes */
+    mov.w   r0,@-r4     /* store as word */
+    shlr16  r0          /* get upper byte */
+    mov.b   r0,@-r4     /* and store */
+    bra     .start_b2   /* jump to trailing byte loop */
+    add     #-4,r4      /* readjust destination */
+
+    /* byte aligned destination (long + 3) */
+    .align  2
+.loop_do3:
+    mov.l   @r5+,r1     /* load first long & increment source addr */
+    add     #16,r4      /* increment dest addr */
+    mov.l   @r5+,r0     /* load second long & increment source addr */
+    mov     r1,r2       /* copy first long */
+    mov.b   r0,@-r4     /* store low byte of second long */
+    shlr8   r0          /* get middle 2 bytes */
+    mov.w   r0,@-r4     /* store as word */
+    shlr16  r0          /* get upper byte */
+    shll8   r2          /* move lower 3 bytes of first long one up.. */
+    or      r2,r0       /* ..combine with the 1 byte of second long.. */
+    mov.l   r0,@-r4     /* ..and store as long */
+    shlr16  r1          /* get upper byte of first long.. */
+    shlr8   r1
+    cmp/hi  r5,r3       /* runs r5 up to last or second last long bound */
+    mov.b   r1,@-r4     /* ..and store */
+    bt      .loop_do3
+
+    bra     .last_do13  /* handle last longword: reuse routine for (long + 1) */
+    add     #4,r3       /* readjust end address */
+
+    /* trailing byte loop: copies 0..3 bytes (or all for < 11 in total) */
+    .align  2
+.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
diff --git a/firmware/target/sh/memmove-sh.S b/firmware/target/sh/memmove-sh.S
new file mode 100755
index 0000000000..9ae9ae5fa2
--- /dev/null
+++ b/firmware/target/sh/memmove-sh.S
@@ -0,0 +1,220 @@
+/***************************************************************************
+ *             __________               __   ___.
+ *   Open      \______   \ ____   ____ |  | _\_ |__   _______  ___
+ *   Source     |       _//  _ \_/ ___\|  |/ /| __ \ /  _ \  \/  /
+ *   Jukebox    |    |   (  <_> )  \___|    < | \_\ (  <_> > <  <
+ *   Firmware   |____|_  /\____/ \___  >__|_ \|___  /\____/__/\_ \
+ *                     \/            \/     \/    \/            \/
+ * $Id$
+ *
+ * Copyright (C) 2006 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
+
+    .align      2
+    .global     _memmove
+    .type       _memmove,@function
+
+/* Moves <length> bytes of data in memory from <source> to <dest>
+ * Regions may overlap.
+ * This version is optimized for speed, and needs the corresponding memcpy
+ * implementation for the forward copy branch.
+ *
+ * arguments:
+ *  r4 - destination address
+ *  r5 - source address
+ *  r6 - length
+ *
+ * return value:
+ *  r0 - destination address (like ANSI version)
+ *
+ * register usage:
+ *  r0 - data / scratch
+ *  r1 - 2nd data / scratch
+ *  r2 - scratch
+ *  r3 - last long bound / adjusted start address (only if >= 11 bytes)
+ *  r4 - current dest address
+ *  r5 - source start address
+ *  r6 - current source address
+ *
+ * The instruction order is devised in a way to utilize the pipelining
+ * of the SH1 to the max. The routine also tries to utilize fast page mode.
+ */
+
+_memmove:
+    cmp/hi  r4,r5       /* source > destination */
+    bf     .backward    /* no: backward copy */
+    mov.l  .memcpy_fwd,r0
+    jmp     @r0
+    mov     r4,r7       /* store dest for returning */
+
+    .align  2
+.memcpy_fwd:
+    .long   ___memcpy_fwd_entry
+
+.backward:
+    add     r6,r4       /* r4 = destination end */
+    mov     #11,r0
+    cmp/hs  r0,r6       /* at least 11 bytes to copy? (ensures 2 aligned longs) */
+    add     #-8,r5      /* adjust for late decrement (max. 2 longs) */
+    add     r5,r6       /* r6 = source end - 8 */
+    bf      .start_b2r  /* no: jump directly to byte loop */
+    
+    mov     #-4,r3      /* r3 = 0xfffffffc */
+    and     r6,r3       /* r3 = last source long bound */
+    cmp/hi  r3,r6       /* already aligned? */
+    bf      .end_b1r    /* yes: skip leading byte loop */
+    
+.loop_b1r:
+    mov.b   @(7,r6),r0  /* load byte */
+    add     #-1,r6      /* decrement source addr */
+    mov.b   r0,@-r4     /* store byte */
+    cmp/hi  r3,r6       /* runs r6 down to last long bound */
+    bt      .loop_b1r
+    
+.end_b1r:
+    mov     #3,r1
+    and     r4,r1       /* r1 = dest alignment offset */
+    mova    .jmptab_r,r0
+    mov.b   @(r0,r1),r1 /* select appropriate main loop.. */
+    add     r0,r1
+    mov     r5,r3       /* copy start adress to r3 */
+    jmp     @r1         /* ..and jump to it */
+    add     #7,r3       /* adjust end addr for main loops doing 2 longs/pass */
+
+    /** main loops, copying 2 longs per pass to profit from fast page mode **/
+
+    /* long aligned destination (fastest) */
+    .align  2
+.loop_do0r:
+    mov.l   @r6,r1      /* load first long */
+    add     #-8,r6      /* decrement source addr */
+    mov.l   @(12,r6),r0 /* load second long */
+    cmp/hi  r3,r6       /* runs r6 down to first or second long bound */
+    mov.l   r0,@-r4     /* store second long */
+    mov.l   r1,@-r4     /* store first long; NOT ALIGNED - no speed loss here! */
+    bt      .loop_do0r
+    
+    add     #-4,r3      /* readjust end address */
+    cmp/hi  r3,r6       /* first long left? */
+    bf      .start_b2r  /* no, jump to trailing byte loop */
+    
+    mov.l   @(4,r6),r0  /* load first long */
+    add     #-4,r6      /* decrement source addr */
+    bra     .start_b2r  /* jump to trailing byte loop */
+    mov.l   r0,@-r4     /* store first long */
+    
+    /* word aligned destination (long + 2) */
+    .align  2
+.loop_do2r:
+    mov.l   @r6,r1      /* load first long */
+    add     #-8,r6      /* decrement source addr */
+    mov.l   @(12,r6),r0 /* load second long */
+    cmp/hi  r3,r6       /* runs r6 down to first or second long bound */
+    mov.w   r0,@-r4     /* store low word of second long */
+    xtrct   r1,r0       /* extract low word of first long & high word of second long */
+    mov.l   r0,@-r4     /* and store as long */
+    shlr16  r1          /* get high word of first long */
+    mov.w   r1,@-r4     /* and store it */
+    bt      .loop_do2r
+
+    add     #-4,r3      /* readjust end address */
+    cmp/hi  r3,r6       /* first long left? */
+    bf      .start_b2r  /* no, jump to trailing byte loop */
+
+    mov.l   @(4,r6),r0  /* load first long & decrement source addr */
+    add     #-4,r6      /* decrement source addr */
+    mov.w   r0,@-r4     /* store low word */
+    shlr16  r0          /* get high word */
+    bra     .start_b2r  /* jump to trailing byte loop */
+    mov.w   r0,@-r4     /* and store it */
+
+    /* jumptable for loop selector */
+    .align  2
+.jmptab_r:
+    .byte   .loop_do0r - .jmptab_r  /* placed in the middle because the SH1 */
+    .byte   .loop_do1r - .jmptab_r  /* loads bytes sign-extended. Otherwise */
+    .byte   .loop_do2r - .jmptab_r  /* the last loop would be out of reach */
+    .byte   .loop_do3r - .jmptab_r  /* of the offset range. */
+    
+    /* byte aligned destination (long + 1) */
+    .align  2
+.loop_do1r:
+    mov.l   @r6,r1      /* load first long */
+    add     #-8,r6      /* decrement source addr */
+    mov.l   @(12,r6),r0 /* load second long */
+    cmp/hi  r3,r6       /* runs r6 down to first or second long bound */
+    mov.b   r0,@-r4     /* store low byte of second long */
+    shlr8   r0          /* get upper 3 bytes */
+    mov     r1,r2       /* copy first long */
+    shll16  r2          /* move low byte of first long all the way up, .. */
+    shll8   r2
+    or      r2,r0       /* ..combine with the 3 bytes of second long.. */
+    mov.l   r0,@-r4     /* ..and store as long */
+    shlr8   r1          /* get middle 2 bytes */
+    mov.w   r1,@-r4     /* store as word */
+    shlr16  r1          /* get upper byte */
+    mov.b   r1,@-r4     /* and store */
+    bt      .loop_do1r
+    
+    add     #-4,r3      /* readjust end address */
+.last_do13r:
+    cmp/hi  r3,r6       /* first long left? */
+    bf      .start_b2r  /* no, jump to trailing byte loop */
+
+    nop                 /* alignment */
+    mov.l   @(4,r6),r0  /* load first long */
+    add     #-4,r6      /* decrement source addr */
+    mov.b   r0,@-r4     /* store low byte */
+    shlr8   r0          /* get middle 2 bytes */
+    mov.w   r0,@-r4     /* store as word */
+    shlr16  r0          /* get upper byte */
+    bra     .start_b2r  /* jump to trailing byte loop */
+    mov.b   r0,@-r4     /* and store */
+
+    /* byte aligned destination (long + 3) */
+    .align  2
+.loop_do3r:
+    mov.l   @r6,r1      /* load first long */
+    add     #-8,r6      /* decrement source addr */
+    mov.l   @(12,r6),r0 /* load second long */
+    mov     r1,r2       /* copy first long */
+    mov.b   r0,@-r4     /* store low byte of second long */
+    shlr8   r0          /* get middle 2 bytes */
+    mov.w   r0,@-r4     /* store as word */
+    shlr16  r0          /* get upper byte */
+    shll8   r2          /* move lower 3 bytes of first long one up.. */
+    or      r2,r0       /* ..combine with the 1 byte of second long.. */
+    mov.l   r0,@-r4     /* ..and store as long */
+    shlr16  r1          /* get upper byte of first long */
+    shlr8   r1
+    cmp/hi  r3,r6       /* runs r6 down to first or second long bound */
+    mov.b   r1,@-r4     /* ..and store */
+    bt      .loop_do3r
+
+    bra     .last_do13r /* handle first longword: reuse routine for (long + 1) */
+    add     #-4,r3      /* readjust end address */
+
+    /* trailing byte loop: copies 0..3 bytes (or all for < 11 in total) */
+    .align  2
+.loop_b2r:
+    mov.b   @(7,r6),r0  /* load byte */
+    add     #-1,r6      /* decrement source addr */
+    mov.b   r0,@-r4     /* store byte */
+.start_b2r:
+    cmp/hi  r5,r6       /* runs r6 down to start address */
+    bt      .loop_b2r
+
+    rts
+    mov     r4,r0       /* return dest start address */
+.end:
+    .size   _memmove,.end-_memmove
diff --git a/firmware/target/sh/memset-sh.S b/firmware/target/sh/memset-sh.S
new file mode 100755
index 0000000000..9b96b93f27
--- /dev/null
+++ b/firmware/target/sh/memset-sh.S
@@ -0,0 +1,107 @@
+/***************************************************************************
+ *             __________               __   ___.
+ *   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
+
+    .align      2
+    .global     _memset
+    .type       _memset,@function
+
+/* Fills a memory region with specified byte value
+ * This version is optimized for speed
+ *
+ * arguments:
+ *  r4 - start address
+ *  r5 - data
+ *  r6 - length
+ *
+ * return value:
+ *  r0 - start address (like ANSI version)
+ *
+ * register usage:
+ *  r0 - temporary
+ *  r1 - start address +11 for main loop
+ *  r4 - start address
+ *  r5 - data (spread to all 4 bytes when using long stores)
+ *  r6 - current address (runs down from end to start)
+ *
+ * The instruction order below is devised in a way to utilize the pipelining
+ * of the SH1 to the max. The routine fills memory from end to start in
+ * order to utilize the auto-decrementing store instructions.
+ */
+
+_memset:
+    neg     r4,r0
+    and     #3,r0       /* r0 = (4 - align_offset) % 4 */
+    add     #4,r0
+    cmp/hs  r0,r6       /* at least one aligned longword to fill? */
+    add     r4,r6       /* r6 = end_address */
+    bf      .no_longs   /* no, jump directly to byte loop */
+
+    extu.b  r5,r5       /* start: spread data to all 4 bytes */
+    swap.b  r5,r0
+    or      r0,r5       /* data now in 2 lower bytes of r5 */
+    swap.w  r5,r0
+    or      r0,r5       /* data now in all 4 bytes of r5 */
+    
+    mov     r6,r0
+    tst     #3,r0       /* r0 already long aligned? */
+    bt      .end_b1     /* yes: skip loop */
+
+    /* leading byte loop: sets 0..3 bytes */
+.loop_b1:
+    mov.b   r5,@-r0     /* store byte */
+    tst     #3,r0       /* r0 long aligned? */
+    bf      .loop_b1    /* runs r0 down until long aligned */
+    
+    mov     r0,r6       /* r6 = last long bound */
+    nop                 /* keep alignment */
+
+.end_b1:
+    mov     r4,r1       /* r1 = start_address... */
+    add     #11,r1      /* ... + 11, combined for rounding and offset */
+    xor     r1,r0
+    tst     #4,r0       /* bit 2 tells whether an even or odd number of */
+    bf      .loop_odd   /* longwords to set */
+
+    /* main loop: set 2 longs per pass */
+.loop_2l:
+    mov.l   r5,@-r6     /* store first long */
+.loop_odd:
+    cmp/hi  r1,r6       /* runs r6 down to first long bound */
+    mov.l   r5,@-r6     /* store second long */
+    bt      .loop_2l
+
+.no_longs:
+    cmp/hi  r4,r6       /* any bytes left? */
+    bf      .end_b2     /* no: skip loop */
+
+    /* trailing byte loop */
+.loop_b2:
+    mov.b   r5,@-r6     /* store byte */
+    cmp/hi  r4,r6       /* runs r6 down to the start address */
+    bt      .loop_b2
+
+.end_b2:
+    rts
+    mov     r4,r0       /* return start address */
+
+.end:
+    .size   _memset,.end-_memset
diff --git a/firmware/target/sh/strlen-sh.S b/firmware/target/sh/strlen-sh.S
new file mode 100755
index 0000000000..34837605ac
--- /dev/null
+++ b/firmware/target/sh/strlen-sh.S
@@ -0,0 +1,94 @@
+/***************************************************************************
+ *             __________               __   ___.
+ *   Open      \______   \ ____   ____ |  | _\_ |__   _______  ___
+ *   Source     |       _//  _ \_/ ___\|  |/ /| __ \ /  _ \  \/  /
+ *   Jukebox    |    |   (  <_> )  \___|    < | \_\ (  <_> > <  <
+ *   Firmware   |____|_  /\____/ \___  >__|_ \|___  /\____/__/\_ \
+ *                     \/            \/     \/    \/            \/
+ * $Id$
+ *
+ * Copyright (C) 2005 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
+
+    .align      2
+    .global     _strlen
+    .type       _strlen,@function
+
+/* Works out the length of a string
+ * This version is optimized for speed
+ *
+ * arguments:
+ *  r4 - start address
+ *
+ * return value:
+ *  r0 - string length
+ *
+ * register usage:
+ *  r0 - current address
+ *  r1 - current value (byte/long)
+ *  r2 - mask for alignment / zero (for cmp/str)
+ *  r4 - start address
+ *
+ */
+
+_strlen:
+    mov     r4,r0       /* r0 = start address */
+    tst     #3,r0       /* long aligned? */
+    bt      .start_l    /* yes, jump directly to the longword loop */
+
+    /* not long aligned: check the first 3 bytes */
+    mov.b   @r0+,r1     /* fetch first byte */
+    tst     r1,r1       /* byte == 0 ? */
+    bt      .hitzero    /* yes, string end found */
+    mov.b   @r0+,r1     /* fetch second byte */
+    mov     #3,r2       /* prepare mask: r2 = 0..00000011b */
+    tst     r1,r1       /* byte == 0 ? */
+    bt      .hitzero    /* yes, string end found */
+    mov.b   @r0+,r1     /* fetch third byte */
+    not     r2,r2       /* prepare mask: r2 = 1..11111100b */
+    tst     r1,r1       /* byte == 0 ? */
+    bt      .hitzero    /* yes, string end found */
+
+    /* not yet found, fall through into longword loop */
+    and     r2,r0       /* align down to long bound */
+
+    /* main loop: check longwords */
+.start_l:
+    mov     #0,r2       /* zero longword for cmp/str */
+.loop_l:
+    mov.l   @r0+,r1     /* fetch long word */
+    cmp/str r1,r2       /* any zero byte within? */
+    bf      .loop_l     /* no, loop */
+    add     #-4,r0      /* set address back to start of this longword */
+
+    /* the last longword contains the string end: figure out the byte */
+    mov.b   @r0+,r1     /* fetch first byte */    
+    tst     r1,r1       /* byte == 0 ? */
+    bt      .hitzero    /* yes, string end found */
+    mov.b   @r0+,r1     /* fetch second byte */
+    tst     r1,r1       /* byte == 0 ? */
+    bt      .hitzero    /* yes, string end found */
+    mov.b   @r0+,r1     /* fetch third byte */
+    tst     r1,r1       /* byte == 0 ? */
+    bt      .hitzero    /* yes, string end found */
+    rts                 /* must be the fourth byte */
+    sub     r4,r0       /* len = string_end - string_start */
+
+.hitzero:
+    add     #-1,r0      /* undo address increment */
+    rts
+    sub     r4,r0       /* len = string_end - string_start */
+
+.end:
+    .size   _strlen,.end-_strlen
+