Assembler optimised memset16() for ARM, by Thom Johansen. Should speed up LCD clearing and solid rectangle drawing on colour iPods somewhat.

git-svn-id: svn://svn.rockbox.org/rockbox/trunk@10900 a1c6a512-1295-4272-9138-f99709370657

1 files changed, 80 insertions, 0 deletions

diff --git a/firmware/target/arm/memset16-arm.S b/firmware/target/arm/memset16-arm.S
new file mode 100755
index 0000000000..13213c57f0
--- /dev/null
+++ b/firmware/target/arm/memset16-arm.S
@@ -0,0 +1,80 @@
+/***************************************************************************
+ *             __________               __   ___.
+ *   Open      \______   \ ____   ____ |  | _\_ |__   _______  ___
+ *   Source     |       _//  _ \_/ ___\|  |/ /| __ \ /  _ \  \/  /
+ *   Jukebox    |    |   (  <_> )  \___|    < | \_\ (  <_> > <  <
+ *   Firmware   |____|_  /\____/ \___  >__|_ \|___  /\____/__/\_ \
+ *                     \/            \/     \/    \/            \/
+ * $Id$
+ *
+ * Copyright (C) 2006 by Thom Johansen
+ *
+ * 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 from the Linux kernel version 2.6.15.3,
+ *  linux/arch/arm/lib/memset.S
+ *
+ *  Copyright (C) 1995-2000 Russell King
+ */
+
+        .global     memset16
+        .type       memset16,%function
+memset16:
+        tst     r0, #2                  @ unaligned?
+        cmpne   r2, #0
+        strneh  r1, [r0], #2            @ store one halfword to align
+        subne   r2, r2, #1
+
+/*
+ * we know that the pointer in r0 is aligned to a word boundary.
+ */
+        orr     r1, r1, r1, lsl #16
+        mov     r3, r1
+        cmp     r2, #8
+        blt     4f
+/*
+ * 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
+
+2:      subs    r2, r2, #32
+        stmgeia r0!, {r1, r3, ip, lr}   @ 64 bytes at a time.
+        stmgeia r0!, {r1, r3, ip, lr}
+        stmgeia r0!, {r1, r3, ip, lr}
+        stmgeia r0!, {r1, r3, ip, lr}
+        bgt     2b
+        ldmeqfd sp!, {pc}               @ Now <64 bytes to go.
+/*
+ * No need to correct the count; we're only testing bits from now on
+ */
+        tst     r2, #16
+        stmneia r0!, {r1, r3, ip, lr}
+        stmneia r0!, {r1, r3, ip, lr}
+        tst     r2, #8
+        stmneia r0!, {r1, r3, ip, lr}
+        ldr     lr, [sp], #4
+
+4:      tst     r2, #4
+        stmneia r0!, {r1, r3}
+        tst     r2, #2
+        strne   r1, [r0], #4
+
+        tst     r2, #1
+        strneh  r1, [r0], #2
+        bx      lr
+.end:
+        .size   memset16,.end-memset16