Piotr Caban : msvcrt: Optimize memmove implementation.

[Alexandre Julliard]

Module: wine
Branch: master
Commit: dd90fdda0e8b8f4b0c669d637193ffdddd1de820
URL:    https://source.winehq.org/git/wine.git/?a=commit;h=dd90fdda0e8b8f4b0c669d637...

Author: Piotr Caban 
Date:   Thu Aug 20 20:00:42 2020 +0200

msvcrt: Optimize memmove implementation.

Based on Torbjorn Granlund code from Glibc.

Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=49663
Signed-off-by: Piotr Caban 
Signed-off-by: Alexandre Julliard

---

dlls/msvcrt/string.c | 91 +++++++++++++++++++++++++++++++++++++++++++++++++---
 1 file changed, 87 insertions(+), 4 deletions(-)

diff --git a/dlls/msvcrt/string.c b/dlls/msvcrt/string.c
index ab6747db5c..3343f1e3cc 100644
--- a/dlls/msvcrt/string.c
+++ b/dlls/msvcrt/string.c
@@ -2483,23 +2483,106 @@ int __cdecl MSVCRT_memcmp(const void *ptr1, const void *ptr2, MSVCRT_size_t n)
 /*********************************************************************
  *                  memmove (MSVCRT.@)
  */
+#ifdef WORDS_BIGENDIAN
+# define MERGE(w1, sh1, w2, sh2) ((w1 << sh1) | (w2 >> sh2))
+#else
+# define MERGE(w1, sh1, w2, sh2) ((w1 >> sh1) | (w2 << sh2))
+#endif
 void * __cdecl MSVCRT_memmove(void *dst, const void *src, MSVCRT_size_t n)
 {
-    volatile unsigned char *d = dst;  /* avoid gcc optimizations */
+    unsigned char *d = dst;
     const unsigned char *s = src;
+    int sh1;
+
+    if (!n) return dst;

if ((MSVCRT_size_t)dst - (MSVCRT_size_t)src >= n)
     {
+        for (; (MSVCRT_size_t)d % sizeof(MSVCRT_size_t) && n; n--) *d++ = *s++;
+
+        sh1 = 8 * ((MSVCRT_size_t)s % sizeof(MSVCRT_size_t));
+        if (!sh1)
+        {
+            while (n >= sizeof(MSVCRT_size_t))
+            {
+                *(MSVCRT_size_t*)d = *(MSVCRT_size_t*)s;
+                s += sizeof(MSVCRT_size_t);
+                d += sizeof(MSVCRT_size_t);
+                n -= sizeof(MSVCRT_size_t);
+            }
+        }
+        else if (n >= 2 * sizeof(MSVCRT_size_t))
+        {
+            int sh2 = 8 * sizeof(MSVCRT_size_t) - sh1;
+            MSVCRT_size_t x, y;
+
+            s -= sh1 / 8;
+            x = *(MSVCRT_size_t*)s;
+            do
+            {
+                s += sizeof(MSVCRT_size_t);
+                y = *(MSVCRT_size_t*)s;
+                *(MSVCRT_size_t*)d = MERGE(x, sh1, y, sh2);
+                d += sizeof(MSVCRT_size_t);
+
+                s += sizeof(MSVCRT_size_t);
+                x = *(MSVCRT_size_t*)s;
+                *(MSVCRT_size_t*)d = MERGE(y, sh1, x, sh2);
+                d += sizeof(MSVCRT_size_t);
+
+                n -= 2 * sizeof(MSVCRT_size_t);
+            } while (n >= 2 * sizeof(MSVCRT_size_t));
+            s += sh1 / 8;
+        }
         while (n--) *d++ = *s++;
+        return dst;
     }
     else
     {
-        d += n - 1;
-        s += n - 1;
-        while (n--) *d-- = *s--;
+        d += n;
+        s += n;
+
+        for (; (MSVCRT_size_t)d % sizeof(MSVCRT_size_t) && n; n--) *--d = *--s;
+
+        sh1 = 8 * ((MSVCRT_size_t)s % sizeof(MSVCRT_size_t));
+        if (!sh1)
+        {
+            while (n >= sizeof(MSVCRT_size_t))
+            {
+                s -= sizeof(MSVCRT_size_t);
+                d -= sizeof(MSVCRT_size_t);
+                *(MSVCRT_size_t*)d = *(MSVCRT_size_t*)s;
+                n -= sizeof(MSVCRT_size_t);
+            }
+        }
+        else if (n >= 2 * sizeof(MSVCRT_size_t))
+        {
+            int sh2 = 8 * sizeof(MSVCRT_size_t) - sh1;
+            MSVCRT_size_t x, y;
+
+            s -= sh1 / 8;
+            x = *(MSVCRT_size_t*)s;
+            do
+            {
+                s -= sizeof(MSVCRT_size_t);
+                y = *(MSVCRT_size_t*)s;
+                d -= sizeof(MSVCRT_size_t);
+                *(MSVCRT_size_t*)d = MERGE(y, sh1, x, sh2);
+
+                s -= sizeof(MSVCRT_size_t);
+                x = *(MSVCRT_size_t*)s;
+                d -= sizeof(MSVCRT_size_t);
+                *(MSVCRT_size_t*)d = MERGE(x, sh1, y, sh2);
+
+                n -= 2 * sizeof(MSVCRT_size_t);
+            } while (n >= 2 * sizeof(MSVCRT_size_t));
+            s += sh1 / 8;
+        }
+        while (n--) *--d = *--s;
     }
     return dst;
 }
+#undef MERGE

/*********************************************************************
  *                  memcpy   (MSVCRT.@)