adapting existing mmx/mmx2/sse/3dnow optimizations so they work on x86_64

git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@13721 b3059339-0415-0410-9bf9-f77b7e298cf2

12 files changed, 208 insertions, 186 deletions

diff --git a/libmpcodecs/pullup.c b/libmpcodecs/pullup.c
index 2abaa158a7..3627084cb5 100644
--- a/libmpcodecs/pullup.c
+++ b/libmpcodecs/pullup.c
@@ -8,6 +8,7 @@
 
 
 
+#ifdef ARCH_X86
 #ifdef HAVE_MMX
 static int diff_y_mmx(unsigned char *a, unsigned char *b, int s)
 {
@@ -147,6 +148,7 @@ static int licomb_y_mmx(unsigned char *a, unsigned char *b, int s)
 	return ret;
 }
 #endif
+#endif
 
 #define ABS(a) (((a)^((a)>>31))-((a)>>31))
 
@@ -682,12 +684,14 @@ void pullup_init_context(struct pullup_context *c)
 	case PULLUP_FMT_Y:
 		c->diff = diff_y;
 		c->comb = licomb_y;
+#ifdef ARCH_X86
 #ifdef HAVE_MMX
 		if (c->cpu & PULLUP_CPU_MMX) {
 			c->diff = diff_y_mmx;
 			c->comb = licomb_y_mmx;
 		}
 #endif
+#endif
 		/* c->comb = qpcomb_y; */
 		break;
 #if 0
diff --git a/libmpcodecs/vf_decimate.c b/libmpcodecs/vf_decimate.c
index 7cc8d2af76..1a80e9dd07 100644
--- a/libmpcodecs/vf_decimate.c
+++ b/libmpcodecs/vf_decimate.c
@@ -31,11 +31,11 @@ static int diff_MMX(unsigned char *old, unsigned char *new, int os, int ns)
 		".balign 16 \n\t"
 		"1: \n\t"
 		
-		"movq (%%esi), %%mm0 \n\t"
-		"movq (%%esi), %%mm2 \n\t"
-		"addl %%eax, %%esi \n\t"
-		"movq (%%edi), %%mm1 \n\t"
-		"addl %%ebx, %%edi \n\t"
+		"movq (%%"REG_S"), %%mm0 \n\t"
+		"movq (%%"REG_S"), %%mm2 \n\t"
+		"add %%"REG_a", %%"REG_S" \n\t"
+		"movq (%%"REG_D"), %%mm1 \n\t"
+		"add %%"REG_b", %%"REG_D" \n\t"
 		"psubusb %%mm1, %%mm2 \n\t"
 		"psubusb %%mm0, %%mm1 \n\t"
 		"movq %%mm2, %%mm0 \n\t"
@@ -51,10 +51,10 @@ static int diff_MMX(unsigned char *old, unsigned char *new, int os, int ns)
 		
 		"decl %%ecx \n\t"
 		"jnz 1b \n\t"
-		"movq %%mm4, (%%edx) \n\t"
+		"movq %%mm4, (%%"REG_d") \n\t"
 		"emms \n\t"
 		: 
-		: "S" (old), "D" (new), "a" (os), "b" (ns), "d" (out)
+		: "S" (old), "D" (new), "a" ((long)os), "b" ((long)ns), "d" (out)
 		: "memory"
 		);
 	return out[0]+out[1]+out[2]+out[3];
diff --git a/libmpcodecs/vf_divtc.c b/libmpcodecs/vf_divtc.c
index e17600edbd..d3f287a73d 100644
--- a/libmpcodecs/vf_divtc.c
+++ b/libmpcodecs/vf_divtc.c
@@ -44,11 +44,11 @@ static int diff_MMX(unsigned char *old, unsigned char *new, int os, int ns)
 	".balign 16 \n\t"
 	"1: \n\t"
 
-	"movq (%%esi), %%mm0 \n\t"
-	"movq (%%esi), %%mm2 \n\t"
-	"addl %%eax, %%esi \n\t"
-	"movq (%%edi), %%mm1 \n\t"
-	"addl %%ebx, %%edi \n\t"
+	"movq (%%"REG_S"), %%mm0 \n\t"
+	"movq (%%"REG_S"), %%mm2 \n\t"
+	"add %%"REG_a", %%"REG_S" \n\t"
+	"movq (%%"REG_D"), %%mm1 \n\t"
+	"add %%"REG_b", %%"REG_D" \n\t"
 	"psubusb %%mm1, %%mm2 \n\t"
 	"psubusb %%mm0, %%mm1 \n\t"
 	"movq %%mm2, %%mm0 \n\t"
@@ -64,10 +64,10 @@ static int diff_MMX(unsigned char *old, unsigned char *new, int os, int ns)
 
 	"decl %%ecx \n\t"
 	"jnz 1b \n\t"
-	"movq %%mm4, (%%edx) \n\t"
+	"movq %%mm4, (%%"REG_d") \n\t"
 	"emms \n\t"
 	:
-	: "S" (old), "D" (new), "a" (os), "b" (ns), "d" (out)
+	: "S" (old), "D" (new), "a" ((long)os), "b" ((long)ns), "d" (out)
 	: "memory"
 	);
    return out[0]+out[1]+out[2]+out[3];
diff --git a/libmpcodecs/vf_eq.c b/libmpcodecs/vf_eq.c
index 74395f61c6..d7adeea1ff 100644
--- a/libmpcodecs/vf_eq.c
+++ b/libmpcodecs/vf_eq.c
@@ -64,9 +64,9 @@ static void process_MMX(unsigned char *dest, int dstride, unsigned char *src, in
 			"paddw %%mm3, %%mm1 \n\t"
 			"paddw %%mm3, %%mm2 \n\t"
 			"packuswb %%mm2, %%mm1 \n\t"
-			"addl $8, %0 \n\t"
+			"add $8, %0 \n\t"
 			"movq %%mm1, (%1) \n\t"
-			"addl $8, %1 \n\t"
+			"add $8, %1 \n\t"
 			"decl %%eax \n\t"
 			"jnz 1b \n\t"
 			: "=r" (src), "=r" (dest)
diff --git a/libmpcodecs/vf_eq2.c b/libmpcodecs/vf_eq2.c
index 123bcea55e..f2641f9c0b 100644
--- a/libmpcodecs/vf_eq2.c
+++ b/libmpcodecs/vf_eq2.c
@@ -152,9 +152,9 @@ void affine_1d_MMX (eq2_param_t *par, unsigned char *dst, unsigned char *src,
       "paddw %%mm3, %%mm1 \n\t"
       "paddw %%mm3, %%mm2 \n\t"
       "packuswb %%mm2, %%mm1 \n\t"
-      "addl $8, %0 \n\t"
+      "add $8, %0 \n\t"
       "movq %%mm1, (%1) \n\t"
-      "addl $8, %1 \n\t"
+      "add $8, %1 \n\t"
       "decl %%eax \n\t"
       "jnz 1b \n\t"
       : "=r" (src), "=r" (dst)
diff --git a/libmpcodecs/vf_filmdint.c b/libmpcodecs/vf_filmdint.c
index 27a527cac3..90e25b8423 100644
--- a/libmpcodecs/vf_filmdint.c
+++ b/libmpcodecs/vf_filmdint.c
@@ -406,8 +406,8 @@ block_metrics_faster_c(unsigned char *a, unsigned char *b, int as, int bs,
 	    "psllq $16, %%mm0\n\t"					     \
 	    "paddusw %%mm0, %%mm7\n\t"					     \
 	    "movq (%1), %%mm4\n\t"					     \
-	    "leal (%0,%2,2), %0\n\t"					     \
-	    "leal (%1,%3,2), %1\n\t"					     \
+	    "lea (%0,%2,2), %0\n\t"					     \
+	    "lea (%1,%3,2), %1\n\t"					     \
 	    "psubusb %4, %%mm4\n\t"					     \
 	    PAVGB(%%mm2, %%mm4)						     \
 	    PAVGB(%%mm2, %%mm4)    /* mm4 = qup odd */			     \
@@ -440,7 +440,7 @@ block_metrics_faster_c(unsigned char *a, unsigned char *b, int as, int bs,
 	    "paddusw %%mm2, %%mm7\n\t"					     \
 	    "paddusw %%mm1, %%mm7\n\t"					     \
 	    : "=r" (a), "=r" (b)					     \
-	    : "r"(as), "r"(bs), "m" (ones), "0"(a), "1"(b), "X"(*a), "X"(*b) \
+	    : "r"((long)as), "r"((long)bs), "m" (ones), "0"(a), "1"(b), "X"(*a), "X"(*b) \
 	    );								     \
     } while (--lines);
 
@@ -650,7 +650,7 @@ dint_copy_line_mmx2(unsigned char *dst, unsigned char *a, long bos,
 	    "por %%mm3, %%mm1 \n\t"     /* avg if >= threshold */
 	    "movq %%mm1, (%2,%4) \n\t"
 	    : /* no output */
-	    : "r" (a), "r" (bos), "r" (dst), "r" (ss), "r" (ds), "r" (cos)
+	    : "r" (a), "r" (bos), "r" (dst), "r" ((long)ss), "r" ((long)ds), "r" (cos)
 	    );
 	a += 8;
 	dst += 8;
diff --git a/libmpcodecs/vf_halfpack.c b/libmpcodecs/vf_halfpack.c
index b4fc0e648f..900aed6a9c 100644
--- a/libmpcodecs/vf_halfpack.c
+++ b/libmpcodecs/vf_halfpack.c
@@ -75,13 +75,13 @@ static void halfpack_MMX(unsigned char *dst, unsigned char *src[3],
 			"por %%mm5, %%mm1 \n\t"
 			"por %%mm6, %%mm2 \n\t"
 
-			"addl $8, %0 \n\t"
-			"addl $8, %1 \n\t"
-			"addl $4, %2 \n\t"
-			"addl $4, %3 \n\t"
+			"add $8, %0 \n\t"
+			"add $8, %1 \n\t"
+			"add $4, %2 \n\t"
+			"add $4, %3 \n\t"
 			"movq %%mm1, (%8) \n\t"
 			"movq %%mm2, 8(%8) \n\t"
-			"addl $16, %8 \n\t"
+			"add $16, %8 \n\t"
 			"decl %9 \n\t"
 			"jnz 1b \n\t"
 			: "=r" (y1), "=r" (y2), "=r" (u), "=r" (v)
diff --git a/libmpcodecs/vf_ilpack.c b/libmpcodecs/vf_ilpack.c
index 66bad26de4..43c6bad218 100644
--- a/libmpcodecs/vf_ilpack.c
+++ b/libmpcodecs/vf_ilpack.c
@@ -76,12 +76,12 @@ static void pack_nn_MMX(unsigned char *dst, unsigned char *y,
 		"punpcklbw %%mm4, %%mm1 \n\t"
 		"punpckhbw %%mm4, %%mm2 \n\t"
 		
-		"addl $8, %0 \n\t"
-		"addl $4, %1 \n\t"
-		"addl $4, %2 \n\t"
+		"add $8, %0 \n\t"
+		"add $4, %1 \n\t"
+		"add $4, %2 \n\t"
 		"movq %%mm1, (%3) \n\t"
 		"movq %%mm2, 8(%3) \n\t"
-		"addl $16, %3 \n\t"
+		"add $16, %3 \n\t"
 		"decl %4 \n\t"
 		"jnz 1b \n\t"
 		"emms \n\t"
@@ -96,22 +96,26 @@ static void pack_li_0_MMX(unsigned char *dst, unsigned char *y,
 	unsigned char *u, unsigned char *v, int w, int us, int vs)
 {
 	asm volatile (""
-		"pushl %%ebp \n\t"
-		"movl 4(%%edx), %%ebp \n\t"
-		"movl (%%edx), %%edx \n\t"
+		"push %%"REG_BP" \n\t"
+#ifdef ARCH_X86_64
+		"mov %6, %%"REG_BP" \n\t"
+#else
+		"movl 4(%%"REG_d"), %%"REG_BP" \n\t"
+		"movl (%%"REG_d"), %%"REG_d" \n\t"
+#endif
 		"pxor %%mm0, %%mm0 \n\t"
 		
 		".balign 16 \n\t"
 		".Lli0: \n\t"
-		"movq (%%esi), %%mm1 \n\t"
-		"movq (%%esi), %%mm2 \n\t"
+		"movq (%%"REG_S"), %%mm1 \n\t"
+		"movq (%%"REG_S"), %%mm2 \n\t"
 		
-		"movq (%%eax,%%edx,2), %%mm4 \n\t"
-		"movq (%%ebx,%%ebp,2), %%mm6 \n\t"
+		"movq (%%"REG_a",%%"REG_d",2), %%mm4 \n\t"
+		"movq (%%"REG_b",%%"REG_BP",2), %%mm6 \n\t"
 		"punpcklbw %%mm0, %%mm4 \n\t"
 		"punpcklbw %%mm0, %%mm6 \n\t"
-		"movq (%%eax), %%mm3 \n\t"
-		"movq (%%ebx), %%mm5 \n\t"
+		"movq (%%"REG_a"), %%mm3 \n\t"
+		"movq (%%"REG_b"), %%mm5 \n\t"
 		"punpcklbw %%mm0, %%mm3 \n\t"
 		"punpcklbw %%mm0, %%mm5 \n\t"
 		"paddw %%mm3, %%mm4 \n\t"
@@ -136,18 +140,18 @@ static void pack_li_0_MMX(unsigned char *dst, unsigned char *y,
 		"punpcklbw %%mm4, %%mm1 \n\t"
 		"punpckhbw %%mm4, %%mm2 \n\t"
 		
-		"movq %%mm1, (%%edi) \n\t"
-		"movq %%mm2, 8(%%edi) \n\t"
+		"movq %%mm1, (%%"REG_D") \n\t"
+		"movq %%mm2, 8(%%"REG_D") \n\t"
 		
-		"movq 8(%%esi), %%mm1 \n\t"
-		"movq 8(%%esi), %%mm2 \n\t"
+		"movq 8(%%"REG_S"), %%mm1 \n\t"
+		"movq 8(%%"REG_S"), %%mm2 \n\t"
 		
-		"movq (%%eax,%%edx,2), %%mm4 \n\t"
-		"movq (%%ebx,%%ebp,2), %%mm6 \n\t"
+		"movq (%%"REG_a",%%"REG_d",2), %%mm4 \n\t"
+		"movq (%%"REG_b",%%"REG_BP",2), %%mm6 \n\t"
 		"punpckhbw %%mm0, %%mm4 \n\t"
 		"punpckhbw %%mm0, %%mm6 \n\t"
-		"movq (%%eax), %%mm3 \n\t"
-		"movq (%%ebx), %%mm5 \n\t"
+		"movq (%%"REG_a"), %%mm3 \n\t"
+		"movq (%%"REG_b"), %%mm5 \n\t"
 		"punpckhbw %%mm0, %%mm3 \n\t"
 		"punpckhbw %%mm0, %%mm5 \n\t"
 		"paddw %%mm3, %%mm4 \n\t"
@@ -172,20 +176,25 @@ static void pack_li_0_MMX(unsigned char *dst, unsigned char *y,
 		"punpcklbw %%mm4, %%mm1 \n\t"
 		"punpckhbw %%mm4, %%mm2 \n\t"
 		
-		"addl $16, %%esi \n\t"
-		"addl $8, %%eax \n\t"
-		"addl $8, %%ebx \n\t"
+		"add $16, %%"REG_S" \n\t"
+		"add $8, %%"REG_a" \n\t"
+		"add $8, %%"REG_b" \n\t"
 		
-		"movq %%mm1, 16(%%edi) \n\t"
-		"movq %%mm2, 24(%%edi) \n\t"
-		"addl $32, %%edi \n\t"
+		"movq %%mm1, 16(%%"REG_D") \n\t"
+		"movq %%mm2, 24(%%"REG_D") \n\t"
+		"add $32, %%"REG_D" \n\t"
 		
 		"decl %%ecx \n\t"
 		"jnz .Lli0 \n\t"
 		"emms \n\t"
-		"popl %%ebp \n\t"
+		"pop %%"REG_BP" \n\t"
 		: 
-		: "S" (y), "D" (dst), "a" (u), "b" (v), "d" (&us), "c" (w/16)
+		: "S" (y), "D" (dst), "a" (u), "b" (v), "c" (w/16),
+#ifdef ARCH_X86_64
+		"d" ((long)us), "r" ((long)vs)
+#else
+		"d" (&us)
+#endif
 		: "memory"
 		);
 	pack_li_0_C(dst, y, u, v, (w&15), us, vs);
@@ -195,22 +204,26 @@ static void pack_li_1_MMX(unsigned char *dst, unsigned char *y,
 	unsigned char *u, unsigned char *v, int w, int us, int vs)
 {
 	asm volatile (""
-		"pushl %%ebp \n\t"
-		"movl 4(%%edx), %%ebp \n\t"
-		"movl (%%edx), %%edx \n\t"
+		"push %%"REG_BP" \n\t"
+#ifdef ARCH_X86_64
+		"mov %6, %%"REG_BP" \n\t"
+#else
+		"movl 4(%%"REG_d"), %%"REG_BP" \n\t"
+		"movl (%%"REG_d"), %%"REG_d" \n\t"
+#endif
 		"pxor %%mm0, %%mm0 \n\t"
 		
 		".balign 16 \n\t"
 		".Lli1: \n\t"
-		"movq (%%esi), %%mm1 \n\t"
-		"movq (%%esi), %%mm2 \n\t"
+		"movq (%%"REG_S"), %%mm1 \n\t"
+		"movq (%%"REG_S"), %%mm2 \n\t"
 		
-		"movq (%%eax,%%edx,2), %%mm4 \n\t"
-		"movq (%%ebx,%%ebp,2), %%mm6 \n\t"
+		"movq (%%"REG_a",%%"REG_d",2), %%mm4 \n\t"
+		"movq (%%"REG_b",%%"REG_BP",2), %%mm6 \n\t"
 		"punpcklbw %%mm0, %%mm4 \n\t"
 		"punpcklbw %%mm0, %%mm6 \n\t"
-		"movq (%%eax), %%mm3 \n\t"
-		"movq (%%ebx), %%mm5 \n\t"
+		"movq (%%"REG_a"), %%mm3 \n\t"
+		"movq (%%"REG_b"), %%mm5 \n\t"
 		"punpcklbw %%mm0, %%mm3 \n\t"
 		"punpcklbw %%mm0, %%mm5 \n\t"
 		"movq %%mm4, %%mm7 \n\t"
@@ -237,18 +250,18 @@ static void pack_li_1_MMX(unsigned char *dst, unsigned char *y,
 		"punpcklbw %%mm4, %%mm1 \n\t"
 		"punpckhbw %%mm4, %%mm2 \n\t"
 		
-		"movq %%mm1, (%%edi) \n\t"
-		"movq %%mm2, 8(%%edi) \n\t"
+		"movq %%mm1, (%%"REG_D") \n\t"
+		"movq %%mm2, 8(%%"REG_D") \n\t"
 		
-		"movq 8(%%esi), %%mm1 \n\t"
-		"movq 8(%%esi), %%mm2 \n\t"
+		"movq 8(%%"REG_S"), %%mm1 \n\t"
+		"movq 8(%%"REG_S"), %%mm2 \n\t"
 		
-		"movq (%%eax,%%edx,2), %%mm4 \n\t"
-		"movq (%%ebx,%%ebp,2), %%mm6 \n\t"
+		"movq (%%"REG_a",%%"REG_d",2), %%mm4 \n\t"
+		"movq (%%"REG_b",%%"REG_BP",2), %%mm6 \n\t"
 		"punpckhbw %%mm0, %%mm4 \n\t"
 		"punpckhbw %%mm0, %%mm6 \n\t"
-		"movq (%%eax), %%mm3 \n\t"
-		"movq (%%ebx), %%mm5 \n\t"
+		"movq (%%"REG_a"), %%mm3 \n\t"
+		"movq (%%"REG_b"), %%mm5 \n\t"
 		"punpckhbw %%mm0, %%mm3 \n\t"
 		"punpckhbw %%mm0, %%mm5 \n\t"
 		"movq %%mm4, %%mm7 \n\t"
@@ -275,20 +288,25 @@ static void pack_li_1_MMX(unsigned char *dst, unsigned char *y,
 		"punpcklbw %%mm4, %%mm1 \n\t"
 		"punpckhbw %%mm4, %%mm2 \n\t"
 		
-		"addl $16, %%esi \n\t"
-		"addl $8, %%eax \n\t"
-		"addl $8, %%ebx \n\t"
+		"add $16, %%"REG_S" \n\t"
+		"add $8, %%"REG_a" \n\t"
+		"add $8, %%"REG_b" \n\t"
 		
-		"movq %%mm1, 16(%%edi) \n\t"
-		"movq %%mm2, 24(%%edi) \n\t"
-		"addl $32, %%edi \n\t"
+		"movq %%mm1, 16(%%"REG_D") \n\t"
+		"movq %%mm2, 24(%%"REG_D") \n\t"
+		"add $32, %%"REG_D" \n\t"
 		
 		"decl %%ecx \n\t"
 		"jnz .Lli1 \n\t"
 		"emms \n\t"
-		"popl %%ebp \n\t"
+		"pop %%"REG_BP" \n\t"
 		: 
-		: "S" (y), "D" (dst), "a" (u), "b" (v), "d" (&us), "c" (w/16)
+		: "S" (y), "D" (dst), "a" (u), "b" (v), "c" (w/16),
+#ifdef ARCH_X86_64
+		"d" ((long)us), "r" ((long)vs)
+#else
+		"d" (&us)
+#endif
 		: "memory"
 		);
 	pack_li_1_C(dst, y, u, v, (w&15), us, vs);
diff --git a/libmpcodecs/vf_ivtc.c b/libmpcodecs/vf_ivtc.c
index 804f68a084..3fb00e5f7b 100644
--- a/libmpcodecs/vf_ivtc.c
+++ b/libmpcodecs/vf_ivtc.c
@@ -71,11 +71,11 @@ static void block_diffs_MMX(struct metrics *m, unsigned char *old, unsigned char
 		"1: \n\t"
 		
 		// Even difference
-		"movq (%%esi), %%mm0 \n\t"
-		"movq (%%esi), %%mm2 \n\t"
-		"addl %%eax, %%esi \n\t"
-		"movq (%%edi), %%mm1 \n\t"
-		"addl %%ebx, %%edi \n\t"
+		"movq (%%"REG_S"), %%mm0 \n\t"
+		"movq (%%"REG_S"), %%mm2 \n\t"
+		"add %%"REG_a", %%"REG_S" \n\t"
+		"movq (%%"REG_D"), %%mm1 \n\t"
+		"add %%"REG_b", %%"REG_D" \n\t"
 		"psubusb %%mm1, %%mm2 \n\t"
 		"psubusb %%mm0, %%mm1 \n\t"
 		"movq %%mm2, %%mm0 \n\t"
@@ -90,11 +90,11 @@ static void block_diffs_MMX(struct metrics *m, unsigned char *old, unsigned char
 		"paddw %%mm3, %%mm4 \n\t"
 		
 		// Odd difference
-		"movq (%%esi), %%mm0 \n\t"
-		"movq (%%esi), %%mm2 \n\t"
-		"addl %%eax, %%esi \n\t"
-		"movq (%%edi), %%mm1 \n\t"
-		"addl %%ebx, %%edi \n\t"
+		"movq (%%"REG_S"), %%mm0 \n\t"
+		"movq (%%"REG_S"), %%mm2 \n\t"
+		"add %%"REG_a", %%"REG_S" \n\t"
+		"movq (%%"REG_D"), %%mm1 \n\t"
+		"add %%"REG_b", %%"REG_D" \n\t"
 		"psubusb %%mm1, %%mm2 \n\t"
 		"psubusb %%mm0, %%mm1 \n\t"
 		"movq %%mm2, %%mm0 \n\t"
@@ -110,8 +110,8 @@ static void block_diffs_MMX(struct metrics *m, unsigned char *old, unsigned char
 			
 		"decl %%ecx \n\t"
 		"jnz 1b \n\t"
-		"movq %%mm4, (%%edx) \n\t"
-		"movq %%mm5, 8(%%edx) \n\t"
+		"movq %%mm4, (%%"REG_d") \n\t"
+		"movq %%mm5, 8(%%"REG_d") \n\t"
 		: 
 		: "S" (old), "D" (new), "a" (os), "b" (ns), "d" (out)
 		: "memory"
@@ -130,14 +130,14 @@ static void block_diffs_MMX(struct metrics *m, unsigned char *old, unsigned char
 		".balign 16 \n\t"
 		"2: \n\t"
 		
-		"movq (%%esi), %%mm0 \n\t"
-		"movq (%%esi,%%eax), %%mm1 \n\t"
-		"addl %%eax, %%esi \n\t"
-		"addl %%eax, %%esi \n\t"
-		"movq (%%edi), %%mm2 \n\t"
-		"movq (%%edi,%%ebx), %%mm3 \n\t"
-		"addl %%ebx, %%edi \n\t"
-		"addl %%ebx, %%edi \n\t"
+		"movq (%%"REG_S"), %%mm0 \n\t"
+		"movq (%%"REG_S",%%"REG_a"), %%mm1 \n\t"
+		"add %%"REG_a", %%"REG_S" \n\t"
+		"add %%"REG_a", %%"REG_S" \n\t"
+		"movq (%%"REG_D"), %%mm2 \n\t"
+		"movq (%%"REG_D",%%"REG_b"), %%mm3 \n\t"
+		"add %%"REG_b", %%"REG_D" \n\t"
+		"add %%"REG_b", %%"REG_D" \n\t"
 		"punpcklbw %%mm7, %%mm0 \n\t"
 		"punpcklbw %%mm7, %%mm1 \n\t"
 		"punpcklbw %%mm7, %%mm2 \n\t"
@@ -164,16 +164,16 @@ static void block_diffs_MMX(struct metrics *m, unsigned char *old, unsigned char
 		"psubw %%mm1, %%mm4 \n\t"
 		"psubw %%mm2, %%mm5 \n\t"
 		"psubw %%mm3, %%mm6 \n\t"
-		"movq %%mm4, (%%edx) \n\t"
-		"movq %%mm5, 16(%%edx) \n\t"
-		"movq %%mm6, 32(%%edx) \n\t"
+		"movq %%mm4, (%%"REG_d") \n\t"
+		"movq %%mm5, 16(%%"REG_d") \n\t"
+		"movq %%mm6, 32(%%"REG_d") \n\t"
 
-		"movl %%eax, %%ecx \n\t"
-		"shll $3, %%ecx \n\t"
-		"subl %%ecx, %%esi \n\t"
-		"movl %%ebx, %%ecx \n\t"
-		"shll $3, %%ecx \n\t"
-		"subl %%ecx, %%edi \n\t"
+		"mov %%"REG_a", %%"REG_c" \n\t"
+		"shl $3, %%"REG_c" \n\t"
+		"sub %%"REG_c", %%"REG_S" \n\t"
+		"mov %%"REG_b", %%"REG_c" \n\t"
+		"shl $3, %%"REG_c" \n\t"
+		"sub %%"REG_c", %%"REG_D" \n\t"
 
 		// Second loop for the last four columns
 		"movl $4, %%ecx \n\t"
@@ -184,14 +184,14 @@ static void block_diffs_MMX(struct metrics *m, unsigned char *old, unsigned char
 		".balign 16 \n\t"
 		"3: \n\t"
 		
-		"movq (%%esi), %%mm0 \n\t"
-		"movq (%%esi,%%eax), %%mm1 \n\t"
-		"addl %%eax, %%esi \n\t"
-		"addl %%eax, %%esi \n\t"
-		"movq (%%edi), %%mm2 \n\t"
-		"movq (%%edi,%%ebx), %%mm3 \n\t"
-		"addl %%ebx, %%edi \n\t"
-		"addl %%ebx, %%edi \n\t"
+		"movq (%%"REG_S"), %%mm0 \n\t"
+		"movq (%%"REG_S",%%"REG_a"), %%mm1 \n\t"
+		"add %%"REG_a", %%"REG_S" \n\t"
+		"add %%"REG_a", %%"REG_S" \n\t"
+		"movq (%%"REG_D"), %%mm2 \n\t"
+		"movq (%%"REG_D",%%"REG_b"), %%mm3 \n\t"
+		"add %%"REG_b", %%"REG_D" \n\t"
+		"add %%"REG_b", %%"REG_D" \n\t"
 		"punpckhbw %%mm7, %%mm0 \n\t"
 		"punpckhbw %%mm7, %%mm1 \n\t"
 		"punpckhbw %%mm7, %%mm2 \n\t"
@@ -218,13 +218,13 @@ static void block_diffs_MMX(struct metrics *m, unsigned char *old, unsigned char
 		"psubw %%mm1, %%mm4 \n\t"
 		"psubw %%mm2, %%mm5 \n\t"
 		"psubw %%mm3, %%mm6 \n\t"
-		"movq %%mm4, 8(%%edx) \n\t"
-		"movq %%mm5, 24(%%edx) \n\t"
-		"movq %%mm6, 40(%%edx) \n\t"
+		"movq %%mm4, 8(%%"REG_d") \n\t"
+		"movq %%mm5, 24(%%"REG_d") \n\t"
+		"movq %%mm6, 40(%%"REG_d") \n\t"
 
 		"emms \n\t"
 		: 
-		: "S" (old), "D" (new), "a" (os), "b" (ns), "d" (out)
+		: "S" (old), "D" (new), "a" ((long)os), "b" ((long)ns), "d" (out)
 		: "memory"
 		);
 	m->p = m->t = m->s = 0;
diff --git a/libmpcodecs/vf_noise.c b/libmpcodecs/vf_noise.c
index 14ad8f9604..c8f669bffa 100644
--- a/libmpcodecs/vf_noise.c
+++ b/libmpcodecs/vf_noise.c
@@ -143,26 +143,26 @@ static int8_t *initNoise(FilterParam *fp){
 
 #ifdef HAVE_MMX
 static inline void lineNoise_MMX(uint8_t *dst, uint8_t *src, int8_t *noise, int len, int shift){
-	int mmx_len= len&(~7);
+	long mmx_len= len&(~7);
 	noise+=shift;
 
 	asm volatile(
-		"movl %3, %%eax			\n\t"
+		"mov %3, %%"REG_a"		\n\t"
 		"pcmpeqb %%mm7, %%mm7		\n\t"
 		"psllw $15, %%mm7		\n\t"
 		"packsswb %%mm7, %%mm7		\n\t"
 		".balign 16			\n\t"
 		"1:				\n\t"
-		"movq (%0, %%eax), %%mm0	\n\t"
-		"movq (%1, %%eax), %%mm1	\n\t"
+		"movq (%0, %%"REG_a"), %%mm0	\n\t"
+		"movq (%1, %%"REG_a"), %%mm1	\n\t"
 		"pxor %%mm7, %%mm0		\n\t"
 		"paddsb %%mm1, %%mm0		\n\t"
 		"pxor %%mm7, %%mm0		\n\t"
-		"movq %%mm0, (%2, %%eax)	\n\t"
-		"addl $8, %%eax			\n\t"
+		"movq %%mm0, (%2, %%"REG_a")	\n\t"
+		"add $8, %%"REG_a"		\n\t"
 		" js 1b				\n\t"
 		:: "r" (src+mmx_len), "r" (noise+mmx_len), "r" (dst+mmx_len), "g" (-mmx_len)
-		: "%eax"
+		: "%"REG_a
 	);
 	if(mmx_len!=len)
 		lineNoise_C(dst+mmx_len, src+mmx_len, noise+mmx_len, len-mmx_len, 0);
@@ -172,26 +172,26 @@ static inline void lineNoise_MMX(uint8_t *dst, uint8_t *src, int8_t *noise, int
 //duplicate of previous except movntq
 #ifdef HAVE_MMX2
 static inline void lineNoise_MMX2(uint8_t *dst, uint8_t *src, int8_t *noise, int len, int shift){
-	int mmx_len= len&(~7);
+	long mmx_len= len&(~7);
 	noise+=shift;
 
 	asm volatile(
-		"movl %3, %%eax			\n\t"
+		"mov %3, %%"REG_a"		\n\t"
 		"pcmpeqb %%mm7, %%mm7		\n\t"
 		"psllw $15, %%mm7		\n\t"
 		"packsswb %%mm7, %%mm7		\n\t"
 		".balign 16			\n\t"
 		"1:				\n\t"
-		"movq (%0, %%eax), %%mm0	\n\t"
-		"movq (%1, %%eax), %%mm1	\n\t"
+		"movq (%0, %%"REG_a"), %%mm0	\n\t"
+		"movq (%1, %%"REG_a"), %%mm1	\n\t"
 		"pxor %%mm7, %%mm0		\n\t"
 		"paddsb %%mm1, %%mm0		\n\t"
 		"pxor %%mm7, %%mm0		\n\t"
-		"movntq %%mm0, (%2, %%eax)	\n\t"
-		"addl $8, %%eax			\n\t"
+		"movntq %%mm0, (%2, %%"REG_a")	\n\t"
+		"add $8, %%"REG_a"		\n\t"
 		" js 1b				\n\t"
 		:: "r" (src+mmx_len), "r" (noise+mmx_len), "r" (dst+mmx_len), "g" (-mmx_len)
-		: "%eax"
+		: "%"REG_a
 	);
 	if(mmx_len!=len)
 		lineNoise_C(dst+mmx_len, src+mmx_len, noise+mmx_len, len-mmx_len, 0);
@@ -214,16 +214,16 @@ static inline void lineNoise_C(uint8_t *dst, uint8_t *src, int8_t *noise, int le
 
 #ifdef HAVE_MMX
 static inline void lineNoiseAvg_MMX(uint8_t *dst, uint8_t *src, int len, int8_t **shift){
-	int mmx_len= len&(~7);
+	long mmx_len= len&(~7);
 
 	asm volatile(
-		"movl %5, %%eax			\n\t"
+		"mov %5, %%"REG_a"		\n\t"
 		".balign 16			\n\t"
 		"1:				\n\t"
-		"movq (%1, %%eax), %%mm1	\n\t"
-		"movq (%0, %%eax), %%mm0	\n\t"
-		"paddb (%2, %%eax), %%mm1	\n\t"
-		"paddb (%3, %%eax), %%mm1	\n\t"
+		"movq (%1, %%"REG_a"), %%mm1	\n\t"
+		"movq (%0, %%"REG_a"), %%mm0	\n\t"
+		"paddb (%2, %%"REG_a"), %%mm1	\n\t"
+		"paddb (%3, %%"REG_a"), %%mm1	\n\t"
 		"movq %%mm0, %%mm2		\n\t"
 		"movq %%mm1, %%mm3		\n\t"
 		"punpcklbw %%mm0, %%mm0		\n\t"
@@ -239,12 +239,12 @@ static inline void lineNoiseAvg_MMX(uint8_t *dst, uint8_t *src, int len, int8_t
 		"psrlw $8, %%mm1		\n\t"
 		"psrlw $8, %%mm3		\n\t"
                 "packuswb %%mm3, %%mm1		\n\t"
-		"movq %%mm1, (%4, %%eax)	\n\t"
-		"addl $8, %%eax			\n\t"
+		"movq %%mm1, (%4, %%"REG_a")	\n\t"
+		"add $8, %%"REG_a"		\n\t"
 		" js 1b				\n\t"
 		:: "r" (src+mmx_len), "r" (shift[0]+mmx_len), "r" (shift[1]+mmx_len), "r" (shift[2]+mmx_len), 
                    "r" (dst+mmx_len), "g" (-mmx_len)
-		: "%eax"
+		: "%"REG_a
 	);
 
 	if(mmx_len!=len){
diff --git a/libmpcodecs/vf_spp.c b/libmpcodecs/vf_spp.c
index 0447ab9838..6cc147b8f6 100644
--- a/libmpcodecs/vf_spp.c
+++ b/libmpcodecs/vf_spp.c
@@ -357,9 +357,9 @@ static void store_slice_mmx(uint8_t *dst, int16_t *src, int dst_stride, int src_
 			"psraw %%mm2, %%mm1	\n\t"
 			"packuswb %%mm1, %%mm0	\n\t"
 			"movq %%mm0, (%1) 	\n\t"
-			"addl $16, %0		\n\t"
-			"addl $8, %1		\n\t"
-			"cmpl %2, %1		\n\t"
+			"add $16, %0		\n\t"
+			"add $8, %1		\n\t"
+			"cmp %2, %1		\n\t"
 			" jb 1b			\n\t"
 			: "+r" (src1), "+r"(dst1)
 			: "r"(dst + width), "r"(dither[y]), "g"(log2_scale), "g"(6-log2_scale)
diff --git a/libmpcodecs/vf_tfields.c b/libmpcodecs/vf_tfields.c
index 779fb6744d..76211d92cf 100644
--- a/libmpcodecs/vf_tfields.c
+++ b/libmpcodecs/vf_tfields.c
@@ -61,7 +61,7 @@ static void deint(unsigned char *dest, int ds, unsigned char *src, int ss, int w
 static void qpel_li_3DNOW(unsigned char *d, unsigned char *s, int w, int h, int ds, int ss, int up)
 {
 	int i, j, ssd=ss;
-	int crap1, crap2;
+	long crap1, crap2;
 	if (up) {
 		ssd = -ss;
 		memcpy(d, s, w);
@@ -71,17 +71,17 @@ static void qpel_li_3DNOW(unsigned char *d, unsigned char *s, int w, int h, int
 	for (i=h-1; i; i--) {
 		asm volatile(
 			"1: \n\t"
-			"movq (%%esi), %%mm0 \n\t"
-			"movq (%%esi,%%eax), %%mm1 \n\t"
+			"movq (%%"REG_S"), %%mm0 \n\t"
+			"movq (%%"REG_S",%%"REG_a"), %%mm1 \n\t"
 			"pavgusb %%mm0, %%mm1 \n\t"
-			"addl $8, %%esi \n\t"
+			"add $8, %%"REG_S" \n\t"
 			"pavgusb %%mm0, %%mm1 \n\t"
-			"movq %%mm1, (%%edi) \n\t"
-			"addl $8, %%edi \n\t"
+			"movq %%mm1, (%%"REG_D") \n\t"
+			"add $8, %%"REG_D" \n\t"
 			"decl %%ecx \n\t"
 			"jnz 1b \n\t"
 			: "=S"(crap1), "=D"(crap2)
-			: "c"(w>>3), "S"(s), "D"(d), "a"(ssd)
+			: "c"(w>>3), "S"(s), "D"(d), "a"((long)ssd)
 		);
 		for (j=w-(w&7); j<w; j++)
 			d[j] = (s[j+ssd] + 3*s[j])>>2;
@@ -97,7 +97,7 @@ static void qpel_li_3DNOW(unsigned char *d, unsigned char *s, int w, int h, int
 static void qpel_li_MMX2(unsigned char *d, unsigned char *s, int w, int h, int ds, int ss, int up)
 {
 	int i, j, ssd=ss;
-	int crap1, crap2;
+	long crap1, crap2;
 	if (up) {
 		ssd = -ss;
 		memcpy(d, s, w);
@@ -108,17 +108,17 @@ static void qpel_li_MMX2(unsigned char *d, unsigned char *s, int w, int h, int d
 		asm volatile(
 			"pxor %%mm7, %%mm7 \n\t"
 			"2: \n\t"
-			"movq (%%esi), %%mm0 \n\t"
-			"movq (%%esi,%%eax), %%mm1 \n\t"
+			"movq (%%"REG_S"), %%mm0 \n\t"
+			"movq (%%"REG_S",%%"REG_a"), %%mm1 \n\t"
 			"pavgb %%mm0, %%mm1 \n\t"
-			"addl $8, %%esi \n\t"
+			"add $8, %%"REG_S" \n\t"
 			"pavgb %%mm0, %%mm1 \n\t"
-			"movq %%mm1, (%%edi) \n\t"
-			"addl $8, %%edi \n\t"
+			"movq %%mm1, (%%"REG_D") \n\t"
+			"add $8, %%"REG_D" \n\t"
 			"decl %%ecx \n\t"
 			"jnz 2b \n\t"
 			: "=S"(crap1), "=D"(crap2)
-			: "c"(w>>3), "S"(s), "D"(d), "a"(ssd)
+			: "c"(w>>3), "S"(s), "D"(d), "a"((long)ssd)
 		);
 		for (j=w-(w&7); j<w; j++)
 			d[j] = (s[j+ssd] + 3*s[j])>>2;
@@ -145,11 +145,11 @@ static void qpel_li_MMX(unsigned char *d, unsigned char *s, int w, int h, int ds
 		asm volatile(
 			"pxor %%mm7, %%mm7 \n\t"
 			"3: \n\t"
-			"movq (%%esi), %%mm0 \n\t"
-			"movq (%%esi), %%mm1 \n\t"
-			"movq (%%esi,%%eax), %%mm2 \n\t"
-			"movq (%%esi,%%eax), %%mm3 \n\t"
-			"addl $8, %%esi \n\t"
+			"movq (%%"REG_S"), %%mm0 \n\t"
+			"movq (%%"REG_S"), %%mm1 \n\t"
+			"movq (%%"REG_S",%%"REG_a"), %%mm2 \n\t"
+			"movq (%%"REG_S",%%"REG_a"), %%mm3 \n\t"
+			"add $8, %%"REG_S" \n\t"
 			"punpcklbw %%mm7, %%mm0 \n\t"
 			"punpckhbw %%mm7, %%mm1 \n\t"
 			"punpcklbw %%mm7, %%mm2 \n\t"
@@ -163,12 +163,12 @@ static void qpel_li_MMX(unsigned char *d, unsigned char *s, int w, int h, int ds
 			"psrlw $2, %%mm2 \n\t"
 			"psrlw $2, %%mm3 \n\t"
 			"packsswb %%mm3, %%mm2 \n\t"
-			"movq %%mm2, (%%edi) \n\t"
-			"addl $8, %%edi \n\t"
+			"movq %%mm2, (%%"REG_D") \n\t"
+			"add $8, %%"REG_D" \n\t"
 			"decl %%ecx \n\t"
 			"jnz 3b \n\t"
 			: "=S"(crap1), "=D"(crap2)
-			: "c"(w>>3), "S"(s), "D"(d), "a"(ssd)
+			: "c"(w>>3), "S"(s), "D"(d), "a"((long)ssd)
 		);
 		for (j=w-(w&7); j<w; j++)
 			d[j] = (s[j+ssd] + 3*s[j])>>2;
@@ -198,15 +198,15 @@ static void qpel_4tap_MMX(unsigned char *d, unsigned char *s, int w, int h, int
 	for (i=h-3; i; i--) {
 		asm volatile(
 			"pxor %%mm0, %%mm0 \n\t"
-			"movq (%%edx), %%mm4 \n\t"
-			"movq 8(%%edx), %%mm5 \n\t"
-			"movq 16(%%edx), %%mm6 \n\t"
-			"movq 24(%%edx), %%mm7 \n\t"
+			"movq (%%"REG_d"), %%mm4 \n\t"
+			"movq 8(%%"REG_d"), %%mm5 \n\t"
+			"movq 16(%%"REG_d"), %%mm6 \n\t"
+			"movq 24(%%"REG_d"), %%mm7 \n\t"
 			"4: \n\t"
 
-			"movq (%%esi,%%eax), %%mm1 \n\t"
-			"movq (%%esi), %%mm2 \n\t"
-			"movq (%%esi,%%ebx), %%mm3 \n\t"
+			"movq (%%"REG_S",%%"REG_a"), %%mm1 \n\t"
+			"movq (%%"REG_S"), %%mm2 \n\t"
+			"movq (%%"REG_S",%%"REG_b"), %%mm3 \n\t"
 			"punpcklbw %%mm0, %%mm1 \n\t"
 			"punpcklbw %%mm0, %%mm2 \n\t"
 			"pmullw %%mm4, %%mm1 \n\t"
@@ -214,38 +214,38 @@ static void qpel_4tap_MMX(unsigned char *d, unsigned char *s, int w, int h, int
 			"pmullw %%mm5, %%mm2 \n\t"
 			"paddusw %%mm2, %%mm1 \n\t"
 			"pmullw %%mm6, %%mm3 \n\t"
-			"movq (%%esi,%%eax,2), %%mm2 \n\t"
+			"movq (%%"REG_S",%%"REG_a",2), %%mm2 \n\t"
 			"psubusw %%mm3, %%mm1 \n\t"
 			"punpcklbw %%mm0, %%mm2 \n\t"	
 			"pmullw %%mm7, %%mm2 \n\t"
 			"psubusw %%mm2, %%mm1 \n\t"
 			"psrlw $7, %%mm1 \n\t"
 
-			"movq (%%esi,%%eax), %%mm2 \n\t"
-			"movq (%%esi), %%mm3 \n\t"
+			"movq (%%"REG_S",%%"REG_a"), %%mm2 \n\t"
+			"movq (%%"REG_S"), %%mm3 \n\t"
 			"punpckhbw %%mm0, %%mm2 \n\t"
 			"punpckhbw %%mm0, %%mm3 \n\t"
 			"pmullw %%mm4, %%mm2 \n\t"
 			"pmullw %%mm5, %%mm3 \n\t"
 			"paddusw %%mm3, %%mm2 \n\t"
-			"movq (%%esi,%%ebx), %%mm3 \n\t"
+			"movq (%%"REG_S",%%"REG_b"), %%mm3 \n\t"
 			"punpckhbw %%mm0, %%mm3 \n\t"
 			"pmullw %%mm6, %%mm3 \n\t"
 			"psubusw %%mm3, %%mm2 \n\t"
-			"movq (%%esi,%%eax,2), %%mm3 \n\t"
+			"movq (%%"REG_S",%%"REG_a",2), %%mm3 \n\t"
 			"punpckhbw %%mm0, %%mm3 \n\t"	
-			"addl $8, %%esi \n\t"
+			"add $8, %%"REG_S" \n\t"
 			"pmullw %%mm7, %%mm3 \n\t"
 			"psubusw %%mm3, %%mm2 \n\t"
 			"psrlw $7, %%mm2 \n\t"
 			
 			"packuswb %%mm2, %%mm1 \n\t"
-			"movq %%mm1, (%%edi) \n\t"
-			"addl $8, %%edi \n\t"
+			"movq %%mm1, (%%"REG_D") \n\t"
+			"add $8, %%"REG_D" \n\t"
 			"decl %%ecx \n\t"
 			"jnz 4b \n\t"
 			: "=S"(crap1), "=D"(crap2)
-			: "c"(w>>3), "S"(s), "D"(d), "a"(ssd), "b"(-ssd), "d"(filter)
+			: "c"(w>>3), "S"(s), "D"(d), "a"((long)ssd), "b"((long)-ssd), "d"(filter)
 		);
 		for (j=w-(w&7); j<w; j++)
 			d[j] = (-9*s[j-ssd] + 111*s[j] + 29*s[j+ssd] - 3*s[j+ssd+ssd])>>7;