From e8b45f1de423a83e414d7b917fcd4eea2148fd3b Mon Sep 17 00:00:00 2001
From: Matthias Clasen <mclasen@fedoraproject.org>
Date: Oct 03 2008 16:03:15 +0000
Subject: add missing source file


---

diff --git a/dct_decode_mmx.c b/dct_decode_mmx.c
new file mode 100644
index 0000000..295703f
--- /dev/null
+++ b/dct_decode_mmx.c
@@ -0,0 +1,409 @@
+/********************************************************************
+ *                                                                  *
+ * THIS FILE IS PART OF THE OggTheora SOFTWARE CODEC SOURCE CODE.   *
+ * USE, DISTRIBUTION AND REPRODUCTION OF THIS LIBRARY SOURCE IS     *
+ * GOVERNED BY A BSD-STYLE SOURCE LICENSE INCLUDED WITH THIS SOURCE *
+ * IN 'COPYING'. PLEASE READ THESE TERMS BEFORE DISTRIBUTING.       *
+ *                                                                  *
+ * THE Theora SOURCE CODE IS COPYRIGHT (C) 2002-2008                *
+ * by the Xiph.Org Foundation http://www.xiph.org/                  *
+ *                                                                  *
+ ********************************************************************
+
+  function:
+  last mod: $Id$
+
+ ********************************************************************/
+
+#include <stdlib.h>
+
+#include "codec_internal.h"
+
+#if defined(USE_ASM)
+
+static const __attribute__((aligned(8),used)) ogg_int64_t OC_V3=
+ 0x0003000300030003LL;
+static const __attribute__((aligned(8),used)) ogg_int64_t OC_V4=
+ 0x0004000400040004LL;
+
+static void loop_filter_v(unsigned char *_pix,int _ystride,
+                          const ogg_int16_t *_ll){
+  long esi;
+  _pix-=_ystride*2;
+  __asm__ __volatile__(
+    /*mm0=0*/
+    "pxor %%mm0,%%mm0\n\t"
+    /*esi=_ystride*3*/
+    "lea (%[ystride],%[ystride],2),%[s]\n\t"
+    /*mm7=_pix[0...8]*/
+    "movq (%[pix]),%%mm7\n\t"
+    /*mm4=_pix[0...8+_ystride*3]*/
+    "movq (%[pix],%[s]),%%mm4\n\t"
+    /*mm6=_pix[0...8]*/
+    "movq %%mm7,%%mm6\n\t"
+    /*Expand unsigned _pix[0...3] to 16 bits.*/
+    "punpcklbw %%mm0,%%mm6\n\t"
+    "movq %%mm4,%%mm5\n\t"
+    /*Expand unsigned _pix[4...8] to 16 bits.*/
+    "punpckhbw %%mm0,%%mm7\n\t"
+    /*Expand other arrays too.*/
+    "punpcklbw %%mm0,%%mm4\n\t"
+    "punpckhbw %%mm0,%%mm5\n\t"
+    /*mm7:mm6=_p[0...8]-_p[0...8+_ystride*3]:*/
+    "psubw %%mm4,%%mm6\n\t"
+    "psubw %%mm5,%%mm7\n\t"
+    /*mm5=mm4=_pix[0...8+_ystride]*/
+    "movq (%[pix],%[ystride]),%%mm4\n\t"
+    /*mm1=mm3=mm2=_pix[0..8]+_ystride*2]*/
+    "movq (%[pix],%[ystride],2),%%mm2\n\t"
+    "movq %%mm4,%%mm5\n\t"
+    "movq %%mm2,%%mm3\n\t"
+    "movq %%mm2,%%mm1\n\t"
+    /*Expand these arrays.*/
+    "punpckhbw %%mm0,%%mm5\n\t"
+    "punpcklbw %%mm0,%%mm4\n\t"
+    "punpckhbw %%mm0,%%mm3\n\t"
+    "punpcklbw %%mm0,%%mm2\n\t"
+    /*Preload...*/
+    "movq %[OC_V3],%%mm0\n\t"
+    /*mm3:mm2=_pix[0...8+_ystride*2]-_pix[0...8+_ystride]*/
+    "psubw %%mm5,%%mm3\n\t"
+    "psubw %%mm4,%%mm2\n\t"
+    /*Scale by 3.*/
+    "pmullw %%mm0,%%mm3\n\t"
+    "pmullw %%mm0,%%mm2\n\t"
+    /*Preload...*/
+    "movq %[OC_V4],%%mm0\n\t"
+    /*f=mm3:mm2==_pix[0...8]-_pix[0...8+_ystride*3]+
+       3*(_pix[0...8+_ystride*2]-_pix[0...8+_ystride])*/
+    "paddw %%mm7,%%mm3\n\t"
+    "paddw %%mm6,%%mm2\n\t"
+    /*Add 4.*/
+    "paddw %%mm0,%%mm3\n\t"
+    "paddw %%mm0,%%mm2\n\t"
+    /*"Divide" by 8.*/
+    "psraw $3,%%mm3\n\t"
+    "psraw $3,%%mm2\n\t"
+    /*Now compute lflim of mm3:mm2 cf. Section 7.10 of the sepc.*/
+    /*Free up mm5.*/
+    "packuswb %%mm5,%%mm4\n\t"
+    /*mm0=L L L L*/
+    "movq (%[ll]),%%mm0\n\t"
+    /*if(R_i<-2L||R_i>2L)R_i=0:*/
+    "movq %%mm2,%%mm5\n\t"
+    "pxor %%mm6,%%mm6\n\t"
+    "movq %%mm0,%%mm7\n\t"
+    "psubw %%mm0,%%mm6\n\t"
+    "psllw $1,%%mm7\n\t"
+    "psllw $1,%%mm6\n\t"
+    /*mm2==R_3 R_2 R_1 R_0*/
+    /*mm5==R_3 R_2 R_1 R_0*/
+    /*mm6==-2L -2L -2L -2L*/
+    /*mm7==2L 2L 2L 2L*/
+    "pcmpgtw %%mm2,%%mm7\n\t"
+    "pcmpgtw %%mm6,%%mm5\n\t"
+    "pand %%mm7,%%mm2\n\t"
+    "movq %%mm0,%%mm7\n\t"
+    "pand %%mm5,%%mm2\n\t"
+    "psllw $1,%%mm7\n\t"
+    "movq %%mm3,%%mm5\n\t"
+    /*mm3==R_7 R_6 R_5 R_4*/
+    /*mm5==R_7 R_6 R_5 R_4*/
+    /*mm6==-2L -2L -2L -2L*/
+    /*mm7==2L 2L 2L 2L*/
+    "pcmpgtw %%mm3,%%mm7\n\t"
+    "pcmpgtw %%mm6,%%mm5\n\t"
+    "pand %%mm7,%%mm3\n\t"
+    "movq %%mm0,%%mm7\n\t"
+    "pand %%mm5,%%mm3\n\t"
+    /*if(R_i<-L)R_i'=R_i+2L;
+      if(R_i>L)R_i'=R_i-2L;
+      if(R_i<-L||R_i>L)R_i=-R_i':*/
+    "psraw $1,%%mm6\n\t"
+    "movq %%mm2,%%mm5\n\t"
+    "psllw $1,%%mm7\n\t"
+    /*mm2==R_3 R_2 R_1 R_0*/
+    /*mm5==R_3 R_2 R_1 R_0*/
+    /*mm6==-L -L -L -L*/
+    /*mm0==L L L L*/
+    /*mm5=R_i>L?FF:00*/
+    "pcmpgtw %%mm0,%%mm5\n\t"
+    /*mm6=-L>R_i?FF:00*/
+    "pcmpgtw %%mm2,%%mm6\n\t"
+    /*mm7=R_i>L?2L:0*/
+    "pand %%mm5,%%mm7\n\t"
+    /*mm2=R_i>L?R_i-2L:R_i*/
+    "psubw %%mm7,%%mm2\n\t"
+    "movq %%mm0,%%mm7\n\t"
+    /*mm5=-L>R_i||R_i>L*/
+    "por %%mm6,%%mm5\n\t"
+    "psllw $1,%%mm7\n\t"
+    /*mm7=-L>R_i?2L:0*/
+    "pand %%mm6,%%mm7\n\t"
+    "pxor %%mm6,%%mm6\n\t"
+    /*mm2=-L>R_i?R_i+2L:R_i*/
+    "paddw %%mm7,%%mm2\n\t"
+    "psubw %%mm0,%%mm6\n\t"
+    /*mm5=-L>R_i||R_i>L?-R_i':0*/
+    "pand %%mm2,%%mm5\n\t"
+    "movq %%mm0,%%mm7\n\t"
+    /*mm2=-L>R_i||R_i>L?0:R_i*/
+    "psubw %%mm5,%%mm2\n\t"
+    "psllw $1,%%mm7\n\t"
+    /*mm2=-L>R_i||R_i>L?-R_i':R_i*/
+    "psubw %%mm5,%%mm2\n\t"
+    "movq %%mm3,%%mm5\n\t"
+    /*mm3==R_7 R_6 R_5 R_4*/
+    /*mm5==R_7 R_6 R_5 R_4*/
+    /*mm6==-L -L -L -L*/
+    /*mm0==L L L L*/
+    /*mm6=-L>R_i?FF:00*/
+    "pcmpgtw %%mm3,%%mm6\n\t"
+    /*mm5=R_i>L?FF:00*/
+    "pcmpgtw %%mm0,%%mm5\n\t"
+    /*mm7=R_i>L?2L:0*/
+    "pand %%mm5,%%mm7\n\t"
+    /*mm2=R_i>L?R_i-2L:R_i*/
+    "psubw %%mm7,%%mm3\n\t"
+    "psllw $1,%%mm0\n\t"
+    /*mm5=-L>R_i||R_i>L*/
+    "por %%mm6,%%mm5\n\t"
+    /*mm0=-L>R_i?2L:0*/
+    "pand %%mm6,%%mm0\n\t"
+    /*mm3=-L>R_i?R_i+2L:R_i*/
+    "paddw %%mm0,%%mm3\n\t"
+    /*mm5=-L>R_i||R_i>L?-R_i':0*/
+    "pand %%mm3,%%mm5\n\t"
+    /*mm2=-L>R_i||R_i>L?0:R_i*/
+    "psubw %%mm5,%%mm3\n\t"
+    /*mm2=-L>R_i||R_i>L?-R_i':R_i*/
+    "psubw %%mm5,%%mm3\n\t"
+    /*Unfortunately, there's no unsigned byte+signed byte with unsigned
+       saturation op code, so we have to promote things back 16 bits.*/
+    "pxor %%mm0,%%mm0\n\t"
+    "movq %%mm4,%%mm5\n\t"
+    "punpcklbw %%mm0,%%mm4\n\t"
+    "punpckhbw %%mm0,%%mm5\n\t"
+    "movq %%mm1,%%mm6\n\t"
+    "punpcklbw %%mm0,%%mm1\n\t"
+    "punpckhbw %%mm0,%%mm6\n\t"
+    /*_pix[0...8+_ystride]+=R_i*/
+    "paddw %%mm2,%%mm4\n\t"
+    "paddw %%mm3,%%mm5\n\t"
+    /*_pix[0...8+_ystride*2]-=R_i*/
+    "psubw %%mm2,%%mm1\n\t"
+    "psubw %%mm3,%%mm6\n\t"
+    "packuswb %%mm5,%%mm4\n\t"
+    "packuswb %%mm6,%%mm1\n\t"
+    /*Write it back out.*/
+    "movq %%mm4,(%[pix],%[ystride])\n\t"
+    "movq %%mm1,(%[pix],%[ystride],2)\n\t"
+    :[s]"=&S"(esi)
+    :[pix]"r"(_pix),[ystride]"r"((long)_ystride),[ll]"r"(_ll),
+     [OC_V3]"m"(OC_V3),[OC_V4]"m"(OC_V4)
+    :"memory"
+  );
+}
+
+/*This code implements the bulk of loop_filter_h().
+  Data are striped p0 p1 p2 p3 ... p0 p1 p2 p3 ..., so in order to load all
+   four p0's to one register we must transpose the values in four mmx regs.
+  When half is done we repeat this for the rest.*/
+static void loop_filter_h4(unsigned char *_pix,long _ystride,
+                           const ogg_int16_t *_ll){
+  long esi;
+  long edi;
+  __asm__ __volatile__(
+    /*x x x x 3 2 1 0*/
+    "movd (%[pix]),%%mm0\n\t"
+    /*esi=_ystride*3*/
+    "lea (%[ystride],%[ystride],2),%[s]\n\t"
+    /*x x x x 7 6 5 4*/
+    "movd (%[pix],%[ystride]),%%mm1\n\t"
+    /*x x x x B A 9 8*/
+    "movd (%[pix],%[ystride],2),%%mm2\n\t"
+    /*x x x x F E D C*/
+    "movd (%[pix],%[s]),%%mm3\n\t"
+    /*mm0=7 3 6 2 5 1 4 0*/
+    "punpcklbw %%mm1,%%mm0\n\t"
+    /*mm2=F B E A D 9 C 8*/
+    "punpcklbw %%mm3,%%mm2\n\t"
+    /*mm1=7 3 6 2 5 1 4 0*/
+    "movq %%mm0,%%mm1\n\t"
+    /*mm0=F B 7 3 E A 6 2*/
+    "punpckhwd %%mm2,%%mm0\n\t"
+    /*mm1=D 9 5 1 C 8 4 0*/
+    "punpcklwd %%mm2,%%mm1\n\t"
+    "pxor %%mm7,%%mm7\n\t"
+    /*mm5=D 9 5 1 C 8 4 0*/
+    "movq %%mm1,%%mm5\n\t"
+    /*mm1=x C x 8 x 4 x 0==pix[0]*/
+    "punpcklbw %%mm7,%%mm1\n\t"
+    /*mm5=x D x 9 x 5 x 1==pix[1]*/
+    "punpckhbw %%mm7,%%mm5\n\t"
+    /*mm3=F B 7 3 E A 6 2*/
+    "movq %%mm0,%%mm3\n\t"
+    /*mm0=x E x A x 6 x 2==pix[2]*/
+    "punpcklbw %%mm7,%%mm0\n\t"
+    /*mm3=x F x B x 7 x 3==pix[3]*/
+    "punpckhbw %%mm7,%%mm3\n\t"
+    /*mm1=mm1-mm3==pix[0]-pix[3]*/
+    "psubw %%mm3,%%mm1\n\t"
+    /*Save a copy of pix[2] for later.*/
+    "movq %%mm0,%%mm4\n\t"
+    /*mm0=mm0-mm5==pix[2]-pix[1]*/
+    "psubw %%mm5,%%mm0\n\t"
+    /*Scale by 3.*/
+    "pmullw %[OC_V3],%%mm0\n\t"
+    /*f=mm1==_pix[0]-_pix[3]+ 3*(_pix[2]-_pix[1])*/
+    "paddw %%mm1,%%mm0\n\t"
+    /*Add 4.*/
+    "paddw %[OC_V4],%%mm0\n\t"
+    /*"Divide" by 8, producing the residuals R_i.*/
+    "psraw $3,%%mm0\n\t"
+    /*Now compute lflim of mm0 cf. Section 7.10 of the sepc.*/
+    /*mm6=L L L L*/
+    "movq (%[ll]),%%mm6\n\t"
+    /*if(R_i<-2L||R_i>2L)R_i=0:*/
+    "movq %%mm0,%%mm1\n\t"
+    "pxor %%mm2,%%mm2\n\t"
+    "movq %%mm6,%%mm3\n\t"
+    "psubw %%mm6,%%mm2\n\t"
+    "psllw $1,%%mm3\n\t"
+    "psllw $1,%%mm2\n\t"
+    /*mm0==R_3 R_2 R_1 R_0*/
+    /*mm1==R_3 R_2 R_1 R_0*/
+    /*mm2==-2L -2L -2L -2L*/
+    /*mm3==2L 2L 2L 2L*/
+    "pcmpgtw %%mm0,%%mm3\n\t"
+    "pcmpgtw %%mm2,%%mm1\n\t"
+    "pand %%mm3,%%mm0\n\t"
+    "pand %%mm1,%%mm0\n\t"
+    /*if(R_i<-L)R_i'=R_i+2L;
+      if(R_i>L)R_i'=R_i-2L;
+      if(R_i<-L||R_i>L)R_i=-R_i':*/
+    "psraw $1,%%mm2\n\t"
+    "movq %%mm0,%%mm1\n\t"
+    "movq %%mm6,%%mm3\n\t"
+    /*mm0==R_3 R_2 R_1 R_0*/
+    /*mm1==R_3 R_2 R_1 R_0*/
+    /*mm2==-L -L -L -L*/
+    /*mm6==L L L L*/
+    /*mm2=-L>R_i?FF:00*/
+    "pcmpgtw %%mm0,%%mm2\n\t"
+    /*mm1=R_i>L?FF:00*/
+    "pcmpgtw %%mm6,%%mm1\n\t"
+    /*mm3=2L 2L 2L 2L*/
+    "psllw $1,%%mm3\n\t"
+    /*mm6=2L 2L 2L 2L*/
+    "psllw $1,%%mm6\n\t"
+    /*mm3=R_i>L?2L:0*/
+    "pand %%mm1,%%mm3\n\t"
+    /*mm6=-L>R_i?2L:0*/
+    "pand %%mm2,%%mm6\n\t"
+    /*mm0=R_i>L?R_i-2L:R_i*/
+    "psubw %%mm3,%%mm0\n\t"
+    /*mm1=-L>R_i||R_i>L*/
+    "por %%mm2,%%mm1\n\t"
+    /*mm0=-L>R_i?R_i+2L:R_i*/
+    "paddw %%mm6,%%mm0\n\t"
+    /*mm1=-L>R_i||R_i>L?R_i':0*/
+    "pand %%mm0,%%mm1\n\t"
+    /*mm0=-L>R_i||R_i>L?0:R_i*/
+    "psubw %%mm1,%%mm0\n\t"
+    /*mm0=-L>R_i||R_i>L?-R_i':R_i*/
+    "psubw %%mm1,%%mm0\n\t"
+    /*_pix[1]+=R_i;*/
+    "paddw %%mm0,%%mm5\n\t"
+    /*_pix[2]-=R_i;*/
+    "psubw %%mm0,%%mm4\n\t"
+    /*mm5=x x x x D 9 5 1*/
+    "packuswb %%mm7,%%mm5\n\t"
+    /*mm4=x x x x E A 6 2*/
+    "packuswb %%mm7,%%mm4\n\t"
+    /*mm5=E D A 9 6 5 2 1*/
+    "punpcklbw %%mm4,%%mm5\n\t"
+    /*edi=6 5 2 1*/
+    "movd %%mm5,%%edi\n\t"
+    "movw %%di,1(%[pix])\n\t"
+    /*Why is there such a big stall here?*/
+    "psrlq $32,%%mm5\n\t"
+    "shrl $16,%%edi\n\t"
+    "movw %%di,1(%[pix],%[ystride])\n\t"
+    /*edi=E D A 9*/
+    "movd %%mm5,%%edi\n\t"
+    "movw %%di,1(%[pix],%[ystride],2)\n\t"
+    "shrl $16,%%edi\n\t"
+    "movw %%di,1(%[pix],%[s])\n\t"
+    :[s]"=&S"(esi),[d]"=&D"(edi),
+     [pix]"+r"(_pix),[ystride]"+r"(_ystride),[ll]"+r"(_ll)
+    :[OC_V3]"m"(OC_V3),[OC_V4]"m"(OC_V4)
+    :"memory"
+  );
+}
+
+static void loop_filter_h(unsigned char *_pix,int _ystride,
+                          const ogg_int16_t *_ll){
+  _pix-=2;
+  loop_filter_h4(_pix,_ystride,_ll);
+  loop_filter_h4(_pix+(_ystride<<2),_ystride,_ll);
+}
+
+static void loop_filter_mmx(PB_INSTANCE *pbi, int FLimit){
+  int j;
+  ogg_int16_t __attribute__((aligned(8)))  ll[4];
+  unsigned char *cp = pbi->display_fragments;
+  ogg_uint32_t *bp = pbi->recon_pixel_index_table;
+
+  if ( FLimit == 0 ) return;
+  ll[0]=ll[1]=ll[2]=ll[3]=FLimit;
+
+  for ( j = 0; j < 3 ; j++){
+    ogg_uint32_t *bp_begin = bp;
+    ogg_uint32_t *bp_end;
+    int stride;
+    int h;
+
+    switch(j) {
+    case 0: /* y */
+      bp_end = bp + pbi->YPlaneFragments;
+      h = pbi->HFragments;
+      stride = pbi->YStride;
+      break;
+    default: /* u,v, 4:20 specific */
+      bp_end = bp + pbi->UVPlaneFragments;
+      h = pbi->HFragments >> 1;
+      stride = pbi->UVStride;
+      break;
+    }
+
+    while(bp<bp_end){
+      ogg_uint32_t *bp_left = bp;
+      ogg_uint32_t *bp_right = bp + h;
+      while(bp<bp_right){
+        if(cp[0]){
+          if(bp>bp_left)
+            loop_filter_h(&pbi->LastFrameRecon[bp[0]],stride,ll);
+          if(bp_left>bp_begin)
+            loop_filter_v(&pbi->LastFrameRecon[bp[0]],stride,ll);
+          if(bp+1<bp_right && !cp[1])
+            loop_filter_h(&pbi->LastFrameRecon[bp[0]]+8,stride,ll);
+          if(bp+h<bp_end && !cp[h])
+            loop_filter_v(&pbi->LastFrameRecon[bp[h]],stride,ll);
+        }
+        bp++;
+        cp++;
+      }
+    }
+  }
+
+  __asm__ __volatile__("emms\n\t");
+}
+
+/* install our implementation in the function table */
+void dsp_mmx_dct_decode_init(DspFunctions *funcs)
+{
+  funcs->LoopFilter = loop_filter_mmx;
+}
+
+#endif /* USE_ASM */
diff --git a/libtheora-x86_64.patch b/libtheora-x86_64.patch
deleted file mode 100644
index 2c17bb0..0000000
--- a/libtheora-x86_64.patch
+++ /dev/null
@@ -1,98 +0,0 @@
-diff -up libtheora-1.0RC1/lib/Makefile.am.x86_64 libtheora-1.0RC1/lib/Makefile.am
---- libtheora-1.0RC1/lib/Makefile.am.x86_64	2008-10-03 11:52:51.000000000 -0400
-+++ libtheora-1.0RC1/lib/Makefile.am	2008-10-03 11:53:08.000000000 -0400
-@@ -52,7 +52,6 @@ encoder_sources = \
- if CPU_x86_64
- enc_arch_dir = enc/x86_64
- encoder_arch_sources= \
--	$(enc_arch_dir)/dct_decode_mmx.c \
- 	$(enc_arch_dir)/dsp_mmx.c \
- 	$(enc_arch_dir)/dsp_mmxext.c \
- 	$(enc_arch_dir)/recon_mmx.c \
-diff -up libtheora-1.0RC1/lib/Makefile.in.x86_64 libtheora-1.0RC1/lib/Makefile.in
---- libtheora-1.0RC1/lib/Makefile.in.x86_64	2008-10-03 11:54:42.000000000 -0400
-+++ libtheora-1.0RC1/lib/Makefile.in	2008-10-03 11:55:05.000000000 -0400
-@@ -38,10 +38,7 @@ subdir = lib
- DIST_COMMON = $(noinst_HEADERS) $(srcdir)/Makefile.am \
- 	$(srcdir)/Makefile.in
- ACLOCAL_M4 = $(top_srcdir)/aclocal.m4
--am__aclocal_m4_deps = $(top_srcdir)/m4/as-ac-expand.m4 \
--	$(top_srcdir)/m4/ogg.m4 $(top_srcdir)/m4/pkg.m4 \
--	$(top_srcdir)/m4/sdl.m4 $(top_srcdir)/m4/vorbis.m4 \
--	$(top_srcdir)/configure.ac
-+am__aclocal_m4_deps = $(top_srcdir)/configure.ac
- am__configure_deps = $(am__aclocal_m4_deps) $(CONFIGURE_DEPENDENCIES) \
- 	$(ACLOCAL_M4)
- mkinstalldirs = $(install_sh) -d
-@@ -84,7 +81,7 @@ am__objects_3 = apiwrapper.lo bitpack.lo
- @CPU_x86_32_TRUE@@CPU_x86_64_FALSE@@THEORA_DISABLE_ENCODE_FALSE@	recon_mmx.lo \
- @CPU_x86_32_TRUE@@CPU_x86_64_FALSE@@THEORA_DISABLE_ENCODE_FALSE@	idct_mmx.lo \
- @CPU_x86_32_TRUE@@CPU_x86_64_FALSE@@THEORA_DISABLE_ENCODE_FALSE@	fdct_mmx.lo
--@CPU_x86_64_TRUE@@THEORA_DISABLE_ENCODE_FALSE@am__objects_4 = dct_decode_mmx.lo \
-+@CPU_x86_64_TRUE@@THEORA_DISABLE_ENCODE_FALSE@am__objects_4 =  \
- @CPU_x86_64_TRUE@@THEORA_DISABLE_ENCODE_FALSE@	dsp_mmx.lo \
- @CPU_x86_64_TRUE@@THEORA_DISABLE_ENCODE_FALSE@	dsp_mmxext.lo \
- @CPU_x86_64_TRUE@@THEORA_DISABLE_ENCODE_FALSE@	recon_mmx.lo \
-@@ -119,7 +116,6 @@ libtheoradec_la_OBJECTS = $(am_libtheora
- libtheoradec_la_LINK = $(LIBTOOL) --tag=CC $(AM_LIBTOOLFLAGS) \
- 	$(LIBTOOLFLAGS) --mode=link $(CCLD) $(AM_CFLAGS) $(CFLAGS) \
- 	$(libtheoradec_la_LDFLAGS) $(LDFLAGS) -o $@
--libtheoraenc_la_LIBADD = libtheoradec.la
- libtheoraenc_la_DEPENDENCIES = libtheoradec.la
- am__libtheoraenc_la_SOURCES_DIST = cpu.c \
- 	$(enc_arch_dir)/dct_decode_mmx.c $(enc_arch_dir)/dsp_mmx.c \
-@@ -166,7 +162,6 @@ AUTOCONF = @AUTOCONF@
- AUTOHEADER = @AUTOHEADER@
- AUTOMAKE = @AUTOMAKE@
- AWK = @AWK@
--BINDIR = @BINDIR@
- BUILDABLE_EXAMPLES = @BUILDABLE_EXAMPLES@
- CC = @CC@
- CCDEPMODE = @CCDEPMODE@
-@@ -182,7 +177,6 @@ DEBUG = @DEBUG@
- DEFS = @DEFS@
- DEPDIR = @DEPDIR@
- DLLTOOL = @DLLTOOL@
--DOCDIR = @DOCDIR@
- DSYMUTIL = @DSYMUTIL@
- ECHO = @ECHO@
- ECHO_C = @ECHO_C@
-@@ -200,14 +194,12 @@ HAVE_PDFLATEX = @HAVE_PDFLATEX@
- HAVE_PKG_CONFIG = @HAVE_PKG_CONFIG@
- HAVE_TRANSFIG = @HAVE_TRANSFIG@
- HAVE_VALGRIND = @HAVE_VALGRIND@
--INCLUDEDIR = @INCLUDEDIR@
- INSTALL = @INSTALL@
- INSTALL_DATA = @INSTALL_DATA@
- INSTALL_PROGRAM = @INSTALL_PROGRAM@
- INSTALL_SCRIPT = @INSTALL_SCRIPT@
- INSTALL_STRIP_PROGRAM = @INSTALL_STRIP_PROGRAM@
- LDFLAGS = @LDFLAGS@
--LIBDIR = @LIBDIR@
- LIBOBJS = @LIBOBJS@
- LIBS = @LIBS@
- LIBTOOL = @LIBTOOL@
-@@ -311,6 +303,7 @@ target_alias = @target_alias@
- target_cpu = @target_cpu@
- target_os = @target_os@
- target_vendor = @target_vendor@
-+top_build_prefix = @top_build_prefix@
- top_builddir = @top_builddir@
- top_srcdir = @top_srcdir@
- INCLUDES = -I$(top_srcdir)/include -I$(top_srcdir)/lib -I$(top_srcdir)/lib/dec -I$(top_srcdir)/lib/enc
-@@ -372,7 +365,6 @@ lib_LTLIBRARIES = libtheoradec.la libthe
- @CPU_x86_32_TRUE@@CPU_x86_64_FALSE@@THEORA_DISABLE_ENCODE_FALSE@	$(enc_arch_dir)/fdct_mmx.c
- 
- @CPU_x86_64_TRUE@@THEORA_DISABLE_ENCODE_FALSE@encoder_arch_sources = \
--@CPU_x86_64_TRUE@@THEORA_DISABLE_ENCODE_FALSE@	$(enc_arch_dir)/dct_decode_mmx.c \
- @CPU_x86_64_TRUE@@THEORA_DISABLE_ENCODE_FALSE@	$(enc_arch_dir)/dsp_mmx.c \
- @CPU_x86_64_TRUE@@THEORA_DISABLE_ENCODE_FALSE@	$(enc_arch_dir)/dsp_mmxext.c \
- @CPU_x86_64_TRUE@@THEORA_DISABLE_ENCODE_FALSE@	$(enc_arch_dir)/recon_mmx.c \
-@@ -448,6 +440,7 @@ libtheoraenc_la_LDFLAGS = \
-   -version-info @THENC_LIB_CURRENT@:@THENC_LIB_REVISION@:@THENC_LIB_AGE@ \
-   @THEORAENC_LDFLAGS@ $(OGG_LIBS)
- 
-+libtheoraenc_la_LIBADD = libtheoradec.la
- libtheora_la_SOURCES = \
-   cpu.c \
- 	$(decoder_arch_sources) \
diff --git a/libtheora.spec b/libtheora.spec
index fc6a6f1..f333cf5 100644
--- a/libtheora.spec
+++ b/libtheora.spec
@@ -7,8 +7,8 @@ License: BSD
 Group: System Environment/Libraries
 URL: http://www.theora.org
 Source0: http://downloads.xiph.org/releases/theora/libtheora-1.0RC1.tar.bz2
-# fix the build on x86-64
-Patch0: libtheora-x86_64.patch
+# missing from tarball
+Source1: dct_decode_mmx.c
 Patch2: libtheora-1.0beta3-undef-nonweak-symbols.patch
 BuildRoot: %{_tmppath}/%{name}-%{version}-%{release}-root-%(%{__id_u} -n) 
 BuildRequires: libogg-devel >= 2:1.1
@@ -57,7 +57,7 @@ with theora bitstreams.
 
 %prep
 %setup -q -n libtheora-1.0RC1
-%patch0 -p1 -b .x86_64
+cp %{SOURCE1} lib/enc/x86_64
 %patch2 -p1
 # no custom CFLAGS please
 sed -i 's/CFLAGS="$CFLAGS $cflags_save"/CFLAGS="$cflags_save"/g' configure