Merge remote-tracking branch 'qatar/master'

* qatar/master: (21 commits)
  build: simplify commands for clean target
  swscale: split swscale.c in unscaled and generic conversion routines.
  swscale: cosmetics.
  swscale: integrate (literally) swscale_template.c in swscale.c.
  swscale: split out x86/swscale_template.c from swscale.c.
  swscale: enable hScale_altivec_real.
  swscale: split out ppc _template.c files from main swscale.c.
  swscale: remove indirections in ppc/swscale_template.c.
  swscale: split out unscaled altivec YUV converters in their own file.
  mpegvideoenc: fix multislice fate tests with threading disabled.
  mpegts: Wrap #ifdef DEBUG and av_hex_dump_log() combination in a macro.
  build: Simplify texi2html invocation through the --output option.
  Mark some variables with av_unused
  Replace avcodec_get_pix_fmt_name() by av_get_pix_fmt_name().
  svq3: Check negative mb_type to fix potential crash.
  svq3: Move svq3-specific fields to their own context.
  rawdec: initialize return value to 0.
  Remove unused get_psnr() prototype
  rawdec: don't leak option strings.
  bktr: get default framerate from video standard.
  ...

Merged-by: Michael Niedermayer <michaelni@gmx.at>

5 files changed, 260 insertions, 231 deletions

diff --git a/libswscale/ppc/swscale_altivec_template.c b/libswscale/ppc/swscale_altivec.c
index 5d4eac5bde..c3b52ac2cc 100644
--- a/libswscale/ppc/swscale_altivec_template.c
+++ b/libswscale/ppc/swscale_altivec.c
@@ -21,6 +21,13 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 
+#include <inttypes.h>
+#include "config.h"
+#include "libswscale/swscale.h"
+#include "libswscale/swscale_internal.h"
+#include "libavutil/cpu.h"
+#include "yuv2rgb_altivec.h"
+
 #define vzero vec_splat_s32(0)
 
 static inline void
@@ -85,12 +92,15 @@ altivec_packIntArrayToCharArray(int *val, uint8_t* dest, int dstW)
     }
 }
 
-static inline void
-yuv2yuvX_altivec_real(const int16_t *lumFilter, const int16_t **lumSrc,
+static void
+yuv2yuvX_altivec_real(SwsContext *c,
+                      const int16_t *lumFilter, const int16_t **lumSrc,
                       int lumFilterSize, const int16_t *chrFilter,
                       const int16_t **chrUSrc, const int16_t **chrVSrc,
-                      int chrFilterSize, uint8_t *dest, uint8_t *uDest,
-                      uint8_t *vDest, int dstW, int chrDstW)
+                      int chrFilterSize, const int16_t **alpSrc,
+                      uint8_t *dest, uint8_t *uDest,
+                      uint8_t *vDest, uint8_t *aDest,
+                      int dstW, int chrDstW)
 {
     const vector signed int vini = {(1 << 18), (1 << 18), (1 << 18), (1 << 18)};
     register int i, j;
@@ -211,10 +221,10 @@ yuv2yuvX_altivec_real(const int16_t *lumFilter, const int16_t **lumSrc,
     }
 }
 
-static inline void hScale_altivec_real(int16_t *dst, int dstW,
-                                       const uint8_t *src, int srcW,
-                                       int xInc, const int16_t *filter,
-                                       const int16_t *filterPos, int filterSize)
+static void hScale_altivec_real(int16_t *dst, int dstW,
+                                const uint8_t *src, int srcW,
+                                int xInc, const int16_t *filter,
+                                const int16_t *filterPos, int filterSize)
 {
     register int i;
     DECLARE_ALIGNED(16, int, tempo)[4];
@@ -391,157 +401,20 @@ static inline void hScale_altivec_real(int16_t *dst, int dstW,
     }
 }
 
-static inline int yv12toyuy2_unscaled_altivec(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
-                                              int srcSliceH, uint8_t* dstParam[], int dstStride_a[])
+void ff_sws_init_swScale_altivec(SwsContext *c)
 {
-    uint8_t *dst=dstParam[0] + dstStride_a[0]*srcSliceY;
-    // yv12toyuy2(src[0], src[1], src[2], dst, c->srcW, srcSliceH, srcStride[0], srcStride[1], dstStride[0]);
-    const uint8_t *ysrc = src[0];
-    const uint8_t *usrc = src[1];
-    const uint8_t *vsrc = src[2];
-    const int width = c->srcW;
-    const int height = srcSliceH;
-    const int lumStride = srcStride[0];
-    const int chromStride = srcStride[1];
-    const int dstStride = dstStride_a[0];
-    const vector unsigned char yperm = vec_lvsl(0, ysrc);
-    const int vertLumPerChroma = 2;
-    register unsigned int y;
-
-    if (width&15) {
-        yv12toyuy2(ysrc, usrc, vsrc, dst, c->srcW, srcSliceH, lumStride, chromStride, dstStride);
-        return srcSliceH;
-    }
-
-    /* This code assumes:
-
-    1) dst is 16 bytes-aligned
-    2) dstStride is a multiple of 16
-    3) width is a multiple of 16
-    4) lum & chrom stride are multiples of 8
-    */
-
-    for (y=0; y<height; y++) {
-        int i;
-        for (i = 0; i < width - 31; i+= 32) {
-            const unsigned int j = i >> 1;
-            vector unsigned char v_yA = vec_ld(i, ysrc);
-            vector unsigned char v_yB = vec_ld(i + 16, ysrc);
-            vector unsigned char v_yC = vec_ld(i + 32, ysrc);
-            vector unsigned char v_y1 = vec_perm(v_yA, v_yB, yperm);
-            vector unsigned char v_y2 = vec_perm(v_yB, v_yC, yperm);
-            vector unsigned char v_uA = vec_ld(j, usrc);
-            vector unsigned char v_uB = vec_ld(j + 16, usrc);
-            vector unsigned char v_u = vec_perm(v_uA, v_uB, vec_lvsl(j, usrc));
-            vector unsigned char v_vA = vec_ld(j, vsrc);
-            vector unsigned char v_vB = vec_ld(j + 16, vsrc);
-            vector unsigned char v_v = vec_perm(v_vA, v_vB, vec_lvsl(j, vsrc));
-            vector unsigned char v_uv_a = vec_mergeh(v_u, v_v);
-            vector unsigned char v_uv_b = vec_mergel(v_u, v_v);
-            vector unsigned char v_yuy2_0 = vec_mergeh(v_y1, v_uv_a);
-            vector unsigned char v_yuy2_1 = vec_mergel(v_y1, v_uv_a);
-            vector unsigned char v_yuy2_2 = vec_mergeh(v_y2, v_uv_b);
-            vector unsigned char v_yuy2_3 = vec_mergel(v_y2, v_uv_b);
-            vec_st(v_yuy2_0, (i << 1), dst);
-            vec_st(v_yuy2_1, (i << 1) + 16, dst);
-            vec_st(v_yuy2_2, (i << 1) + 32, dst);
-            vec_st(v_yuy2_3, (i << 1) + 48, dst);
-        }
-        if (i < width) {
-            const unsigned int j = i >> 1;
-            vector unsigned char v_y1 = vec_ld(i, ysrc);
-            vector unsigned char v_u = vec_ld(j, usrc);
-            vector unsigned char v_v = vec_ld(j, vsrc);
-            vector unsigned char v_uv_a = vec_mergeh(v_u, v_v);
-            vector unsigned char v_yuy2_0 = vec_mergeh(v_y1, v_uv_a);
-            vector unsigned char v_yuy2_1 = vec_mergel(v_y1, v_uv_a);
-            vec_st(v_yuy2_0, (i << 1), dst);
-            vec_st(v_yuy2_1, (i << 1) + 16, dst);
+    if (!(av_get_cpu_flags() & AV_CPU_FLAG_ALTIVEC))
+        return;
+
+    c->hScale       = hScale_altivec_real;
+    c->yuv2yuvX     = yuv2yuvX_altivec_real;
+
+    /* The following list of supported dstFormat values should
+     * match what's found in the body of ff_yuv2packedX_altivec() */
+    if (!(c->flags & SWS_BITEXACT) && !c->alpPixBuf &&
+        (c->dstFormat==PIX_FMT_ABGR  || c->dstFormat==PIX_FMT_BGRA  ||
+         c->dstFormat==PIX_FMT_BGR24 || c->dstFormat==PIX_FMT_RGB24 ||
+         c->dstFormat==PIX_FMT_RGBA  || c->dstFormat==PIX_FMT_ARGB)) {
+            c->yuv2packedX  = ff_yuv2packedX_altivec;
         }
-        if ((y&(vertLumPerChroma-1)) == vertLumPerChroma-1) {
-            usrc += chromStride;
-            vsrc += chromStride;
-        }
-        ysrc += lumStride;
-        dst += dstStride;
-    }
-
-    return srcSliceH;
-}
-
-static inline int yv12touyvy_unscaled_altivec(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
-                                              int srcSliceH, uint8_t* dstParam[], int dstStride_a[])
-{
-    uint8_t *dst=dstParam[0] + dstStride_a[0]*srcSliceY;
-    // yv12toyuy2(src[0], src[1], src[2], dst, c->srcW, srcSliceH, srcStride[0], srcStride[1], dstStride[0]);
-    const uint8_t *ysrc = src[0];
-    const uint8_t *usrc = src[1];
-    const uint8_t *vsrc = src[2];
-    const int width = c->srcW;
-    const int height = srcSliceH;
-    const int lumStride = srcStride[0];
-    const int chromStride = srcStride[1];
-    const int dstStride = dstStride_a[0];
-    const int vertLumPerChroma = 2;
-    const vector unsigned char yperm = vec_lvsl(0, ysrc);
-    register unsigned int y;
-
-    if (width&15) {
-        yv12touyvy(ysrc, usrc, vsrc, dst, c->srcW, srcSliceH, lumStride, chromStride, dstStride);
-        return srcSliceH;
-    }
-
-    /* This code assumes:
-
-    1) dst is 16 bytes-aligned
-    2) dstStride is a multiple of 16
-    3) width is a multiple of 16
-    4) lum & chrom stride are multiples of 8
-    */
-
-    for (y=0; y<height; y++) {
-        int i;
-        for (i = 0; i < width - 31; i+= 32) {
-            const unsigned int j = i >> 1;
-            vector unsigned char v_yA = vec_ld(i, ysrc);
-            vector unsigned char v_yB = vec_ld(i + 16, ysrc);
-            vector unsigned char v_yC = vec_ld(i + 32, ysrc);
-            vector unsigned char v_y1 = vec_perm(v_yA, v_yB, yperm);
-            vector unsigned char v_y2 = vec_perm(v_yB, v_yC, yperm);
-            vector unsigned char v_uA = vec_ld(j, usrc);
-            vector unsigned char v_uB = vec_ld(j + 16, usrc);
-            vector unsigned char v_u = vec_perm(v_uA, v_uB, vec_lvsl(j, usrc));
-            vector unsigned char v_vA = vec_ld(j, vsrc);
-            vector unsigned char v_vB = vec_ld(j + 16, vsrc);
-            vector unsigned char v_v = vec_perm(v_vA, v_vB, vec_lvsl(j, vsrc));
-            vector unsigned char v_uv_a = vec_mergeh(v_u, v_v);
-            vector unsigned char v_uv_b = vec_mergel(v_u, v_v);
-            vector unsigned char v_uyvy_0 = vec_mergeh(v_uv_a, v_y1);
-            vector unsigned char v_uyvy_1 = vec_mergel(v_uv_a, v_y1);
-            vector unsigned char v_uyvy_2 = vec_mergeh(v_uv_b, v_y2);
-            vector unsigned char v_uyvy_3 = vec_mergel(v_uv_b, v_y2);
-            vec_st(v_uyvy_0, (i << 1), dst);
-            vec_st(v_uyvy_1, (i << 1) + 16, dst);
-            vec_st(v_uyvy_2, (i << 1) + 32, dst);
-            vec_st(v_uyvy_3, (i << 1) + 48, dst);
-        }
-        if (i < width) {
-            const unsigned int j = i >> 1;
-            vector unsigned char v_y1 = vec_ld(i, ysrc);
-            vector unsigned char v_u = vec_ld(j, usrc);
-            vector unsigned char v_v = vec_ld(j, vsrc);
-            vector unsigned char v_uv_a = vec_mergeh(v_u, v_v);
-            vector unsigned char v_uyvy_0 = vec_mergeh(v_uv_a, v_y1);
-            vector unsigned char v_uyvy_1 = vec_mergel(v_uv_a, v_y1);
-            vec_st(v_uyvy_0, (i << 1), dst);
-            vec_st(v_uyvy_1, (i << 1) + 16, dst);
-        }
-        if ((y&(vertLumPerChroma-1)) == vertLumPerChroma-1) {
-            usrc += chromStride;
-            vsrc += chromStride;
-        }
-        ysrc += lumStride;
-        dst += dstStride;
-    }
-    return srcSliceH;
 }
diff --git a/libswscale/ppc/swscale_template.c b/libswscale/ppc/swscale_template.c
deleted file mode 100644
index 27351adca1..0000000000
--- a/libswscale/ppc/swscale_template.c
+++ /dev/null
@@ -1,71 +0,0 @@
-/*
- * Copyright (C) 2001-2003 Michael Niedermayer <michaelni@gmx.at>
- *
- * This file is part of FFmpeg.
- *
- * FFmpeg is free software; you can redistribute it and/or
- * modify it under the terms of the GNU Lesser General Public
- * License as published by the Free Software Foundation; either
- * version 2.1 of the License, or (at your option) any later version.
- *
- * FFmpeg is distributed in the hope that it will be useful,
- * but WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
- * Lesser General Public License for more details.
- *
- * You should have received a copy of the GNU Lesser General Public
- * License along with FFmpeg; if not, write to the Free Software
- * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
- */
-
-
-#if COMPILE_TEMPLATE_ALTIVEC
-#include "swscale_altivec_template.c"
-#endif
-
-#if COMPILE_TEMPLATE_ALTIVEC
-static inline void RENAME(yuv2yuvX)(SwsContext *c, const int16_t *lumFilter,
-                                    const int16_t **lumSrc, int lumFilterSize,
-                                    const int16_t *chrFilter, const int16_t **chrUSrc,
-                                    const int16_t **chrVSrc, int chrFilterSize,
-                                    const int16_t **alpSrc,
-                                    uint8_t *dest, uint8_t *uDest, uint8_t *vDest,
-                                    uint8_t *aDest, int dstW, int chrDstW)
-{
-    yuv2yuvX_altivec_real(lumFilter, lumSrc, lumFilterSize,
-                          chrFilter, chrUSrc, chrVSrc, chrFilterSize,
-                          dest, uDest, vDest, dstW, chrDstW);
-}
-
-/**
- * vertical scale YV12 to RGB
- */
-static inline void RENAME(yuv2packedX)(SwsContext *c, const int16_t *lumFilter,
-                                       const int16_t **lumSrc, int lumFilterSize,
-                                       const int16_t *chrFilter, const int16_t **chrUSrc,
-                                       const int16_t **chrVSrc, int chrFilterSize,
-                                       const int16_t **alpSrc, uint8_t *dest,
-                                       int dstW, int dstY)
-{
-    /* The following list of supported dstFormat values should
-       match what's found in the body of ff_yuv2packedX_altivec() */
-    if (!(c->flags & SWS_BITEXACT) && !c->alpPixBuf &&
-         (c->dstFormat==PIX_FMT_ABGR  || c->dstFormat==PIX_FMT_BGRA  ||
-          c->dstFormat==PIX_FMT_BGR24 || c->dstFormat==PIX_FMT_RGB24 ||
-          c->dstFormat==PIX_FMT_RGBA  || c->dstFormat==PIX_FMT_ARGB))
-            ff_yuv2packedX_altivec(c, lumFilter, lumSrc, lumFilterSize,
-                                   chrFilter, chrUSrc, chrVSrc, chrFilterSize,
-                                   dest, dstW, dstY);
-    else
-        yuv2packedXinC(c, lumFilter, lumSrc, lumFilterSize,
-                       chrFilter, chrUSrc, chrVSrc, chrFilterSize,
-                       alpSrc, dest, dstW, dstY);
-}
-#endif
-
-
-static void RENAME(sws_init_swScale)(SwsContext *c)
-{
-    c->yuv2yuvX     = RENAME(yuv2yuvX    );
-    c->yuv2packedX  = RENAME(yuv2packedX );
-}
diff --git a/libswscale/ppc/yuv2rgb_altivec.c b/libswscale/ppc/yuv2rgb_altivec.c
index 8aaa987e62..e13702b100 100644
--- a/libswscale/ppc/yuv2rgb_altivec.c
+++ b/libswscale/ppc/yuv2rgb_altivec.c
@@ -96,6 +96,7 @@ adjustment.
 #include "libswscale/swscale_internal.h"
 #include "libavutil/cpu.h"
 #include "libavutil/pixdesc.h"
+#include "yuv2rgb_altivec.h"
 
 #undef PROFILE_THE_BEAST
 #undef INC_SCALING
@@ -631,7 +632,8 @@ ff_yuv2packedX_altivec(SwsContext *c, const int16_t *lumFilter,
                        const int16_t **lumSrc, int lumFilterSize,
                        const int16_t *chrFilter, const int16_t **chrUSrc,
                        const int16_t **chrVSrc, int chrFilterSize,
-                       uint8_t *dest, int dstW, int dstY)
+                       const int16_t **alpSrc, uint8_t *dest,
+                       int dstW, int dstY)
 {
     int i,j;
     vector signed short X,X0,X1,Y0,U0,V0,Y1,U1,V1,U,V;
diff --git a/libswscale/ppc/yuv2rgb_altivec.h b/libswscale/ppc/yuv2rgb_altivec.h
new file mode 100644
index 0000000000..15385b1d3b
--- /dev/null
+++ b/libswscale/ppc/yuv2rgb_altivec.h
@@ -0,0 +1,34 @@
+/*
+ * AltiVec-enhanced yuv2yuvX
+ *
+ * Copyright (C) 2004 Romain Dolbeau <romain@dolbeau.org>
+ * based on the equivalent C code in swscale.c
+ *
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * FFmpeg is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with FFmpeg; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#ifndef PPC_YUV2RGB_ALTIVEC_H
+#define PPC_YUV2RGB_ALTIVEC_H 1
+
+void ff_yuv2packedX_altivec(SwsContext *c, const int16_t *lumFilter,
+                            const int16_t **lumSrc, int lumFilterSize,
+                            const int16_t *chrFilter, const int16_t **chrUSrc,
+                            const int16_t **chrVSrc, int chrFilterSize,
+                            const int16_t **alpSrc, uint8_t *dest,
+                            int dstW, int dstY);
+
+#endif /* PPC_YUV2RGB_ALTIVEC_H */
diff --git a/libswscale/ppc/yuv2yuv_altivec.c b/libswscale/ppc/yuv2yuv_altivec.c
new file mode 100644
index 0000000000..82c265afd2
--- /dev/null
+++ b/libswscale/ppc/yuv2yuv_altivec.c
@@ -0,0 +1,191 @@
+/*
+ * AltiVec-enhanced yuv-to-yuv convertion routines.
+ *
+ * Copyright (C) 2004 Romain Dolbeau <romain@dolbeau.org>
+ * based on the equivalent C code in swscale.c
+ *
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * FFmpeg is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with FFmpeg; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#include <inttypes.h>
+#include "config.h"
+#include "libswscale/swscale.h"
+#include "libswscale/swscale_internal.h"
+#include "libavutil/cpu.h"
+
+static int yv12toyuy2_unscaled_altivec(SwsContext *c, const uint8_t* src[],
+                                       int srcStride[], int srcSliceY,
+                                       int srcSliceH, uint8_t* dstParam[],
+                                       int dstStride_a[])
+{
+    uint8_t *dst=dstParam[0] + dstStride_a[0]*srcSliceY;
+    // yv12toyuy2(src[0], src[1], src[2], dst, c->srcW, srcSliceH, srcStride[0], srcStride[1], dstStride[0]);
+    const uint8_t *ysrc = src[0];
+    const uint8_t *usrc = src[1];
+    const uint8_t *vsrc = src[2];
+    const int width = c->srcW;
+    const int height = srcSliceH;
+    const int lumStride = srcStride[0];
+    const int chromStride = srcStride[1];
+    const int dstStride = dstStride_a[0];
+    const vector unsigned char yperm = vec_lvsl(0, ysrc);
+    const int vertLumPerChroma = 2;
+    register unsigned int y;
+
+    /* This code assumes:
+
+    1) dst is 16 bytes-aligned
+    2) dstStride is a multiple of 16
+    3) width is a multiple of 16
+    4) lum & chrom stride are multiples of 8
+    */
+
+    for (y=0; y<height; y++) {
+        int i;
+        for (i = 0; i < width - 31; i+= 32) {
+            const unsigned int j = i >> 1;
+            vector unsigned char v_yA = vec_ld(i, ysrc);
+            vector unsigned char v_yB = vec_ld(i + 16, ysrc);
+            vector unsigned char v_yC = vec_ld(i + 32, ysrc);
+            vector unsigned char v_y1 = vec_perm(v_yA, v_yB, yperm);
+            vector unsigned char v_y2 = vec_perm(v_yB, v_yC, yperm);
+            vector unsigned char v_uA = vec_ld(j, usrc);
+            vector unsigned char v_uB = vec_ld(j + 16, usrc);
+            vector unsigned char v_u = vec_perm(v_uA, v_uB, vec_lvsl(j, usrc));
+            vector unsigned char v_vA = vec_ld(j, vsrc);
+            vector unsigned char v_vB = vec_ld(j + 16, vsrc);
+            vector unsigned char v_v = vec_perm(v_vA, v_vB, vec_lvsl(j, vsrc));
+            vector unsigned char v_uv_a = vec_mergeh(v_u, v_v);
+            vector unsigned char v_uv_b = vec_mergel(v_u, v_v);
+            vector unsigned char v_yuy2_0 = vec_mergeh(v_y1, v_uv_a);
+            vector unsigned char v_yuy2_1 = vec_mergel(v_y1, v_uv_a);
+            vector unsigned char v_yuy2_2 = vec_mergeh(v_y2, v_uv_b);
+            vector unsigned char v_yuy2_3 = vec_mergel(v_y2, v_uv_b);
+            vec_st(v_yuy2_0, (i << 1), dst);
+            vec_st(v_yuy2_1, (i << 1) + 16, dst);
+            vec_st(v_yuy2_2, (i << 1) + 32, dst);
+            vec_st(v_yuy2_3, (i << 1) + 48, dst);
+        }
+        if (i < width) {
+            const unsigned int j = i >> 1;
+            vector unsigned char v_y1 = vec_ld(i, ysrc);
+            vector unsigned char v_u = vec_ld(j, usrc);
+            vector unsigned char v_v = vec_ld(j, vsrc);
+            vector unsigned char v_uv_a = vec_mergeh(v_u, v_v);
+            vector unsigned char v_yuy2_0 = vec_mergeh(v_y1, v_uv_a);
+            vector unsigned char v_yuy2_1 = vec_mergel(v_y1, v_uv_a);
+            vec_st(v_yuy2_0, (i << 1), dst);
+            vec_st(v_yuy2_1, (i << 1) + 16, dst);
+        }
+        if ((y&(vertLumPerChroma-1)) == vertLumPerChroma-1) {
+            usrc += chromStride;
+            vsrc += chromStride;
+        }
+        ysrc += lumStride;
+        dst += dstStride;
+    }
+
+    return srcSliceH;
+}
+
+static int yv12touyvy_unscaled_altivec(SwsContext *c, const uint8_t* src[],
+                                       int srcStride[], int srcSliceY,
+                                       int srcSliceH, uint8_t* dstParam[],
+                                       int dstStride_a[])
+{
+    uint8_t *dst=dstParam[0] + dstStride_a[0]*srcSliceY;
+    // yv12toyuy2(src[0], src[1], src[2], dst, c->srcW, srcSliceH, srcStride[0], srcStride[1], dstStride[0]);
+    const uint8_t *ysrc = src[0];
+    const uint8_t *usrc = src[1];
+    const uint8_t *vsrc = src[2];
+    const int width = c->srcW;
+    const int height = srcSliceH;
+    const int lumStride = srcStride[0];
+    const int chromStride = srcStride[1];
+    const int dstStride = dstStride_a[0];
+    const int vertLumPerChroma = 2;
+    const vector unsigned char yperm = vec_lvsl(0, ysrc);
+    register unsigned int y;
+
+    /* This code assumes:
+
+    1) dst is 16 bytes-aligned
+    2) dstStride is a multiple of 16
+    3) width is a multiple of 16
+    4) lum & chrom stride are multiples of 8
+    */
+
+    for (y=0; y<height; y++) {
+        int i;
+        for (i = 0; i < width - 31; i+= 32) {
+            const unsigned int j = i >> 1;
+            vector unsigned char v_yA = vec_ld(i, ysrc);
+            vector unsigned char v_yB = vec_ld(i + 16, ysrc);
+            vector unsigned char v_yC = vec_ld(i + 32, ysrc);
+            vector unsigned char v_y1 = vec_perm(v_yA, v_yB, yperm);
+            vector unsigned char v_y2 = vec_perm(v_yB, v_yC, yperm);
+            vector unsigned char v_uA = vec_ld(j, usrc);
+            vector unsigned char v_uB = vec_ld(j + 16, usrc);
+            vector unsigned char v_u = vec_perm(v_uA, v_uB, vec_lvsl(j, usrc));
+            vector unsigned char v_vA = vec_ld(j, vsrc);
+            vector unsigned char v_vB = vec_ld(j + 16, vsrc);
+            vector unsigned char v_v = vec_perm(v_vA, v_vB, vec_lvsl(j, vsrc));
+            vector unsigned char v_uv_a = vec_mergeh(v_u, v_v);
+            vector unsigned char v_uv_b = vec_mergel(v_u, v_v);
+            vector unsigned char v_uyvy_0 = vec_mergeh(v_uv_a, v_y1);
+            vector unsigned char v_uyvy_1 = vec_mergel(v_uv_a, v_y1);
+            vector unsigned char v_uyvy_2 = vec_mergeh(v_uv_b, v_y2);
+            vector unsigned char v_uyvy_3 = vec_mergel(v_uv_b, v_y2);
+            vec_st(v_uyvy_0, (i << 1), dst);
+            vec_st(v_uyvy_1, (i << 1) + 16, dst);
+            vec_st(v_uyvy_2, (i << 1) + 32, dst);
+            vec_st(v_uyvy_3, (i << 1) + 48, dst);
+        }
+        if (i < width) {
+            const unsigned int j = i >> 1;
+            vector unsigned char v_y1 = vec_ld(i, ysrc);
+            vector unsigned char v_u = vec_ld(j, usrc);
+            vector unsigned char v_v = vec_ld(j, vsrc);
+            vector unsigned char v_uv_a = vec_mergeh(v_u, v_v);
+            vector unsigned char v_uyvy_0 = vec_mergeh(v_uv_a, v_y1);
+            vector unsigned char v_uyvy_1 = vec_mergel(v_uv_a, v_y1);
+            vec_st(v_uyvy_0, (i << 1), dst);
+            vec_st(v_uyvy_1, (i << 1) + 16, dst);
+        }
+        if ((y&(vertLumPerChroma-1)) == vertLumPerChroma-1) {
+            usrc += chromStride;
+            vsrc += chromStride;
+        }
+        ysrc += lumStride;
+        dst += dstStride;
+    }
+    return srcSliceH;
+}
+
+void ff_swscale_get_unscaled_altivec(SwsContext *c)
+{
+    if ((av_get_cpu_flags() & AV_CPU_FLAG_ALTIVEC) && !(c->srcW & 15) &&
+        !(c->flags & SWS_BITEXACT) && c->srcFormat == PIX_FMT_YUV420P) {
+        enum PixelFormat dstFormat = c->dstFormat;
+
+        // unscaled YV12 -> packed YUV, we want speed
+        if (dstFormat == PIX_FMT_YUYV422)
+            c->swScale= yv12toyuy2_unscaled_altivec;
+        else if (dstFormat == PIX_FMT_UYVY422)
+            c->swScale= yv12touyvy_unscaled_altivec;
+    }
+}