From a035261089403de259e74ce4dd196e2715138ed2 Mon Sep 17 00:00:00 2001
From: Thomas Martitz <kugel@rockbox.org>
Date: Sat, 7 Jan 2012 19:56:09 +0100
Subject: Move optimized memcpy and friends and strlen to firmware/asm, using
 the new automatic-asm-picking infrastructure.

---
 firmware/asm/memmove.c | 147 +++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 147 insertions(+)
 create mode 100644 firmware/asm/memmove.c

(limited to 'firmware/asm/memmove.c')

diff --git a/firmware/asm/memmove.c b/firmware/asm/memmove.c
new file mode 100644
index 0000000000..5f423964bb
--- /dev/null
+++ b/firmware/asm/memmove.c
@@ -0,0 +1,147 @@
+/*
+FUNCTION
+        <<memmove>>---move possibly overlapping memory
+
+INDEX
+        memmove
+
+ANSI_SYNOPSIS
+        #include <string.h>
+        void *memmove(void *<[dst]>, const void *<[src]>, size_t <[length]>);
+
+TRAD_SYNOPSIS
+        #include <string.h>
+        void *memmove(<[dst]>, <[src]>, <[length]>)
+        void *<[dst]>;
+        void *<[src]>;
+        size_t <[length]>;
+
+DESCRIPTION
+        This function moves <[length]> characters from the block of
+        memory starting at <<*<[src]>>> to the memory starting at
+        <<*<[dst]>>>. <<memmove>> reproduces the characters correctly
+        at <<*<[dst]>>> even if the two areas overlap.
+
+
+RETURNS
+        The function returns <[dst]> as passed.
+
+PORTABILITY
+<<memmove>> is ANSI C.
+
+<<memmove>> requires no supporting OS subroutines.
+
+QUICKREF
+        memmove ansi pure
+*/
+
+#include "config.h"
+#include <_ansi.h>
+#include <string.h>
+
+/* Nonzero if either X or Y is not aligned on a "long" boundary.  */
+#define UNALIGNED(X, Y) \
+  (((long)X & (sizeof (long) - 1)) | ((long)Y & (sizeof (long) - 1)))
+
+/* How many bytes are copied each iteration of the 4X unrolled loop.  */
+#define BIGBLOCKSIZE    (sizeof (long) << 2)
+
+/* How many bytes are copied each iteration of the word copy loop.  */
+#define LITTLEBLOCKSIZE (sizeof (long))
+
+/* Threshhold for punting to the byte copier.  */
+#define TOO_SMALL(LEN)  ((LEN) < BIGBLOCKSIZE)
+
+_PTR
+_DEFUN (memmove, (dst_void, src_void, length),
+        _PTR dst_void _AND
+        _CONST _PTR src_void _AND
+        size_t length) ICODE_ATTR;
+
+_PTR
+_DEFUN (memmove, (dst_void, src_void, length),
+        _PTR dst_void _AND
+        _CONST _PTR src_void _AND
+        size_t length)
+{
+#if defined(PREFER_SIZE_OVER_SPEED) || defined(__OPTIMIZE_SIZE__)
+  char *dst = dst_void;
+  _CONST char *src = src_void;
+
+  if (src < dst && dst < src + length)
+    {
+      /* Have to copy backwards */
+      src += length;
+      dst += length;
+      while (length--)
+        {
+          *--dst = *--src;
+        }
+    }
+  else
+    {
+      while (length--)
+        {
+          *dst++ = *src++;
+        }
+    }
+
+  return dst_void;
+#else
+  char *dst = dst_void;
+  _CONST char *src = src_void;
+  long *aligned_dst;
+  _CONST long *aligned_src;
+  unsigned int   len =  length;
+
+  if (src < dst && dst < src + len)
+    {
+      /* Destructive overlap...have to copy backwards */
+      src += len;
+      dst += len;
+      while (len--)
+        {
+          *--dst = *--src;
+        }
+    }
+  else
+    {
+      /* Use optimizing algorithm for a non-destructive copy to closely 
+         match memcpy. If the size is small or either SRC or DST is unaligned,
+         then punt into the byte copy loop.  This should be rare.  */
+      if (!TOO_SMALL(len) && !UNALIGNED (src, dst))
+        {
+          aligned_dst = (long*)dst;
+          aligned_src = (long*)src;
+
+          /* Copy 4X long words at a time if possible.  */
+          while (len >= BIGBLOCKSIZE)
+            {
+              *aligned_dst++ = *aligned_src++;
+              *aligned_dst++ = *aligned_src++;
+              *aligned_dst++ = *aligned_src++;
+              *aligned_dst++ = *aligned_src++;
+              len -= BIGBLOCKSIZE;
+            }
+
+          /* Copy one long word at a time if possible.  */
+          while (len >= LITTLEBLOCKSIZE)
+            {
+              *aligned_dst++ = *aligned_src++;
+              len -= LITTLEBLOCKSIZE;
+            }
+
+          /* Pick up any residual with a byte copier.  */
+          dst = (char*)aligned_dst;
+          src = (char*)aligned_src;
+        }
+
+      while (len--)
+        {
+          *dst++ = *src++;
+        }
+    }
+
+  return dst_void;
+#endif /* not PREFER_SIZE_OVER_SPEED */
+}
-- 
cgit v1.2.3