[Pixman] [PATCH 1/3] mmx: use MMX inline assembly only when USE_MMX

[mattst88 at gmail.com]

From: Matt Turner <mattst88 at gmail.com>

Signed-off-by: Matt Turner <mattst88 at gmail.com>
---
 pixman/pixman-mmx.c |   88 ++------------------------------------------------
 1 files changed, 4 insertions(+), 84 deletions(-)

diff --git a/pixman/pixman-mmx.c b/pixman/pixman-mmx.c
index 697ec4c..71fa18e 100644
--- a/pixman/pixman-mmx.c
+++ b/pixman/pixman-mmx.c
@@ -1795,7 +1795,7 @@ pixman_fill_mmx (uint32_t *bits,
     uint32_t byte_width;
     uint8_t     *byte_line;
 
-#ifdef __GNUC__
+#if defined __GNUC__ && defined USE_MMX
     __m64 v1, v2, v3, v4, v5, v6, v7;
 #endif
 
@@ -1829,7 +1829,7 @@ pixman_fill_mmx (uint32_t *bits,
     fill = ((uint64_t)xor << 32) | xor;
     vfill = to_m64 (fill);
 
-#ifdef __GNUC__
+#if defined __GNUC__ && defined USE_MMX
     __asm__ (
         "movq		%7,	%0\n"
         "movq		%7,	%1\n"
@@ -1875,7 +1875,7 @@ pixman_fill_mmx (uint32_t *bits,
 
 	while (w >= 64)
 	{
-#ifdef __GNUC__
+#if defined __GNUC__ && defined USE_MMX
 	    __asm__ (
 	        "movq	%1,	  (%0)\n"
 	        "movq	%2,	 8(%0)\n"
@@ -2844,87 +2844,7 @@ pixman_blt_mmx (uint32_t *src_bits,
 	dst_bytes += dst_stride;
 	w = byte_width;
 
-	while (w >= 2 && ((unsigned long)d & 3))
-	{
-	    *(uint16_t *)d = *(uint16_t *)s;
-	    w -= 2;
-	    s += 2;
-	    d += 2;
-	}
-
-	while (w >= 4 && ((unsigned long)d & 7))
-	{
-	    *(uint32_t *)d = *(uint32_t *)s;
-
-	    w -= 4;
-	    s += 4;
-	    d += 4;
-	}
-
-	while (w >= 64)
-	{
-#if defined (__GNUC__) || (defined(__SUNPRO_C) && (__SUNPRO_C >= 0x590))
-	    __asm__ (
-	        "movq	  (%1),	  %%mm0\n"
-	        "movq	 8(%1),	  %%mm1\n"
-	        "movq	16(%1),	  %%mm2\n"
-	        "movq	24(%1),	  %%mm3\n"
-	        "movq	32(%1),	  %%mm4\n"
-	        "movq	40(%1),	  %%mm5\n"
-	        "movq	48(%1),	  %%mm6\n"
-	        "movq	56(%1),	  %%mm7\n"
-
-	        "movq	%%mm0,	  (%0)\n"
-	        "movq	%%mm1,	 8(%0)\n"
-	        "movq	%%mm2,	16(%0)\n"
-	        "movq	%%mm3,	24(%0)\n"
-	        "movq	%%mm4,	32(%0)\n"
-	        "movq	%%mm5,	40(%0)\n"
-	        "movq	%%mm6,	48(%0)\n"
-	        "movq	%%mm7,	56(%0)\n"
-		:
-		: "r" (d), "r" (s)
-		: "memory",
-		  "%mm0", "%mm1", "%mm2", "%mm3",
-		  "%mm4", "%mm5", "%mm6", "%mm7");
-#else
-	    __m64 v0 = *(__m64 *)(s + 0);
-	    __m64 v1 = *(__m64 *)(s + 8);
-	    __m64 v2 = *(__m64 *)(s + 16);
-	    __m64 v3 = *(__m64 *)(s + 24);
-	    __m64 v4 = *(__m64 *)(s + 32);
-	    __m64 v5 = *(__m64 *)(s + 40);
-	    __m64 v6 = *(__m64 *)(s + 48);
-	    __m64 v7 = *(__m64 *)(s + 56);
-	    *(__m64 *)(d + 0)  = v0;
-	    *(__m64 *)(d + 8)  = v1;
-	    *(__m64 *)(d + 16) = v2;
-	    *(__m64 *)(d + 24) = v3;
-	    *(__m64 *)(d + 32) = v4;
-	    *(__m64 *)(d + 40) = v5;
-	    *(__m64 *)(d + 48) = v6;
-	    *(__m64 *)(d + 56) = v7;
-#endif
-
-	    w -= 64;
-	    s += 64;
-	    d += 64;
-	}
-	while (w >= 4)
-	{
-	    *(uint32_t *)d = *(uint32_t *)s;
-
-	    w -= 4;
-	    s += 4;
-	    d += 4;
-	}
-	if (w >= 2)
-	{
-	    *(uint16_t *)d = *(uint16_t *)s;
-	    w -= 2;
-	    s += 2;
-	    d += 2;
-	}
+	memcpy(d, s, w);
     }
 
     _mm_empty ();
-- 
1.7.3.4