From a672530595e0a5cc0558d9e51a09b35a62c87fe7 Mon Sep 17 00:00:00 2001
From: "Ryan C. Gordon" <icculus@icculus.org>
Date: Sat, 29 Oct 2011 01:11:47 -0400
Subject: [PATCH] SDL_memcpyMMX(): Fixed handling of overflow bytes.

Thanks to Mason Wheeler for the fix!
---
 src/video/SDL_blit_copy.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/src/video/SDL_blit_copy.c b/src/video/SDL_blit_copy.c
index c0cdd23a1..a9f98eb73 100644
--- a/src/video/SDL_blit_copy.c
+++ b/src/video/SDL_blit_copy.c
@@ -59,6 +59,7 @@ SDL_memcpySSE(Uint8 * dst, const Uint8 * src, int len)
 static __inline__ void
 SDL_memcpyMMX(Uint8 * dst, const Uint8 * src, int len)
 {
+    const int remain = (len & 63);
     int i;
 
     __m64* d64 = (__m64*)dst;
@@ -78,8 +79,11 @@ SDL_memcpyMMX(Uint8 * dst, const Uint8 * src, int len)
         s64 += 8;
     }
 
-    if (len & 63)
-        SDL_memcpy(dst, src, len & 63);
+    if (remain)
+    {
+        const int skip = len - remain;
+        SDL_memcpy(dst + skip, src + skip, remain);
+    }
 }
 #endif /* __MMX__ */