Merge remote-tracking branch 'qatar/master'

* qatar/master: (29 commits)
  lavfi: reclassify showfiltfmts as a TESTPROG
  graph2dot: fix printf format specifier
  swscale: yuv2planeX 8bit >=sse2 functions need aligned stack on x86-32.
  vp8: loopfilter >=sse2 functions need aligned stack on x86-32.
  amr: remove shift out of the AMR_BIT() macro.
  dsputilenc: group yasm and inline asm function pointer assignment.
  mov: use forward declaration of a function instead of a table.
  Clarify Doxygen comment for FF_API_* #defines.
  configure: simplify get_version()
  Create version.h headers for libraries that lack them
  gitignore: Use full path instead of relative path to specify patterns
  mpegvideo: remove VLAs
  Add XTEA encryption support in libavutil
  Add Blowfish encryption support in libavutil
  eval: Add the isinf() function and tests for it
  flacdec: move lpc filter to flacdsp
  flacdec: split off channel decorrelation as flacdsp
  avplay: Add an option for not limiting the input buffer size
  FATE: add a test for WMA cover art.
  FATE: add a test for apetag cover art
  ...

Conflicts:
	.gitignore
	configure
	ffplay.c
	libavcodec/Makefile
	libavcodec/error_resilience.c
	libavcodec/mpegvideo.c
	libavcodec/ratecontrol.c
	libavdevice/avdevice.h
	libavfilter/Makefile
	libavfilter/filtfmts.c
	libavfilter/version.h
	libavformat/mov.c
	libavformat/version.h
	libavutil/Makefile
	libavutil/avutil.h
	libavutil/version.h
	libswscale/swscale.h
	libswscale/x86/swscale_mmx.c
	tests/fate/libavutil.mak
	tests/lavfi-regression.sh
	tools/graph2dot.c

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

18 files changed, 317 insertions, 139 deletions

diff --git a/libavcodec/Makefile b/libavcodec/Makefile
index a3891c2da4..3cc0124a93 100644
--- a/libavcodec/Makefile
+++ b/libavcodec/Makefile
@@ -179,7 +179,7 @@ OBJS-$(CONFIG_FFV1_ENCODER)            += ffv1.o rangecoder.o
 OBJS-$(CONFIG_FFVHUFF_DECODER)         += huffyuv.o
 OBJS-$(CONFIG_FFVHUFF_ENCODER)         += huffyuv.o
 OBJS-$(CONFIG_FFWAVESYNTH_DECODER)     += ffwavesynth.o
-OBJS-$(CONFIG_FLAC_DECODER)            += flacdec.o flacdata.o flac.o
+OBJS-$(CONFIG_FLAC_DECODER)            += flacdec.o flacdata.o flac.o flacdsp.o
 OBJS-$(CONFIG_FLAC_ENCODER)            += flacenc.o flacdata.o flac.o vorbis_data.o
 OBJS-$(CONFIG_FLASHSV_DECODER)         += flashsv.o
 OBJS-$(CONFIG_FLASHSV_ENCODER)         += flashsvenc.o
diff --git a/libavcodec/amr.h b/libavcodec/amr.h
index 7e5a4dce5b..6b24002145 100644
--- a/libavcodec/amr.h
+++ b/libavcodec/amr.h
@@ -61,7 +61,7 @@ static inline void ff_amr_bit_reorder(uint16_t *out, int size,
            field <<= 1;
            field |= data[bit >> 3] >> (bit & 7) & 1;
         }
-        out[field_offset] = field;
+        out[field_offset >> 1] = field;
     }
 }
 
diff --git a/libavcodec/amrnbdata.h b/libavcodec/amrnbdata.h
index 2745d33ffb..6e555bfdb4 100644
--- a/libavcodec/amrnbdata.h
+++ b/libavcodec/amrnbdata.h
@@ -71,7 +71,7 @@ typedef struct {
 } AMRNBFrame;
 
 /** The index of a frame parameter */
-#define AMR_BIT(field)                  (offsetof(AMRNBFrame, field) >> 1)
+#define AMR_BIT(field)                  (offsetof(AMRNBFrame, field))
 /** The index of a subframe-specific parameter */
 #define AMR_OF(frame_num, variable)     AMR_BIT(subframe[frame_num].variable)
 
diff --git a/libavcodec/amrwbdata.h b/libavcodec/amrwbdata.h
index 972c257d2a..83390fc393 100644
--- a/libavcodec/amrwbdata.h
+++ b/libavcodec/amrwbdata.h
@@ -82,7 +82,7 @@ typedef struct {
 } AMRWBFrame;
 
 /** The index of a frame parameter */
-#define AMR_BIT(field)                  (offsetof(AMRWBFrame, field) >> 1)
+#define AMR_BIT(field)                  (offsetof(AMRWBFrame, field))
 /** The index of a subframe-specific parameter */
 #define AMR_OF(frame_num, variable)     AMR_BIT(subframe[frame_num].variable)
 
diff --git a/libavcodec/error_resilience.c b/libavcodec/error_resilience.c
index 63f996ccde..58e3754b7c 100644
--- a/libavcodec/error_resilience.c
+++ b/libavcodec/error_resilience.c
@@ -408,7 +408,7 @@ static void v_block_filter(MpegEncContext *s, uint8_t *dst, int w, int h,
 
 static void guess_mv(MpegEncContext *s)
 {
-    uint8_t *fixed = av_malloc(s->mb_stride * s->mb_height);
+    uint8_t *fixed = s->er_temp_buffer;
 #define MV_FROZEN    3
 #define MV_CHANGED   2
 #define MV_UNCHANGED 1
@@ -470,7 +470,7 @@ static void guess_mv(MpegEncContext *s)
                 decode_mb(s, 0);
             }
         }
-        goto end;
+        return;
     }
 
     for (depth = 0; ; depth++) {
@@ -722,7 +722,7 @@ skip_last_mv:
         }
 
         if (none_left)
-            goto end;
+            return;
 
         for (i = 0; i < s->mb_num; i++) {
             int mb_xy = s->mb_index2xy[i];
@@ -731,8 +731,6 @@ skip_last_mv:
         }
         // printf(":"); fflush(stdout);
     }
-end:
-    av_free(fixed);
 }
 
 static int is_intra_more_likely(MpegEncContext *s)
diff --git a/libavcodec/flac.c b/libavcodec/flac.c
index 484a44efb3..92268eac8c 100644
--- a/libavcodec/flac.c
+++ b/libavcodec/flac.c
@@ -55,8 +55,9 @@ int ff_flac_decode_frame_header(AVCodecContext *avctx, GetBitContext *gb,
     if (fi->ch_mode < FLAC_MAX_CHANNELS) {
         fi->channels = fi->ch_mode + 1;
         fi->ch_mode = FLAC_CHMODE_INDEPENDENT;
-    } else if (fi->ch_mode <= FLAC_CHMODE_MID_SIDE) {
+    } else if (fi->ch_mode < FLAC_MAX_CHANNELS + FLAC_CHMODE_MID_SIDE) {
         fi->channels = 2;
+        fi->ch_mode -= FLAC_MAX_CHANNELS - 1;
     } else {
         av_log(avctx, AV_LOG_ERROR + log_level_offset,
                "invalid channel mode: %d\n", fi->ch_mode);
diff --git a/libavcodec/flac.h b/libavcodec/flac.h
index 65965af6ad..94444f8a67 100644
--- a/libavcodec/flac.h
+++ b/libavcodec/flac.h
@@ -37,10 +37,10 @@
 #define FLAC_MIN_FRAME_SIZE    11
 
 enum {
-    FLAC_CHMODE_INDEPENDENT =  0,
-    FLAC_CHMODE_LEFT_SIDE   =  8,
-    FLAC_CHMODE_RIGHT_SIDE  =  9,
-    FLAC_CHMODE_MID_SIDE    = 10,
+    FLAC_CHMODE_INDEPENDENT = 0,
+    FLAC_CHMODE_LEFT_SIDE   = 1,
+    FLAC_CHMODE_RIGHT_SIDE  = 2,
+    FLAC_CHMODE_MID_SIDE    = 3,
 };
 
 enum {
diff --git a/libavcodec/flacdec.c b/libavcodec/flacdec.c
index 02ed463eb3..13ce95d86c 100644
--- a/libavcodec/flacdec.c
+++ b/libavcodec/flacdec.c
@@ -42,6 +42,7 @@
 #include "golomb.h"
 #include "flac.h"
 #include "flacdata.h"
+#include "flacdsp.h"
 
 #undef NDEBUG
 #include <assert.h>
@@ -54,13 +55,13 @@ typedef struct FLACContext {
     GetBitContext gb;                       ///< GetBitContext initialized to start at the current frame
 
     int blocksize;                          ///< number of samples in the current frame
-    int curr_bps;                           ///< bps for current subframe, adjusted for channel correlation and wasted bits
     int sample_shift;                       ///< shift required to make output samples 16-bit or 32-bit
-    int is32;                               ///< flag to indicate if output should be 32-bit instead of 16-bit
     int ch_mode;                            ///< channel decorrelation type in the current frame
     int got_streaminfo;                     ///< indicates if the STREAMINFO has been read
 
     int32_t *decoded[FLAC_MAX_CHANNELS];    ///< decoded samples
+
+    FLACDSPContext dsp;
 } FLACContext;
 
 static const int64_t flac_channel_layouts[6] = {
@@ -101,6 +102,17 @@ int avpriv_flac_is_extradata_valid(AVCodecContext *avctx,
     return 1;
 }
 
+static void flac_set_bps(FLACContext *s)
+{
+    if (s->bps > 16) {
+        s->avctx->sample_fmt = AV_SAMPLE_FMT_S32;
+        s->sample_shift = 32 - s->bps;
+    } else {
+        s->avctx->sample_fmt = AV_SAMPLE_FMT_S16;
+        s->sample_shift = 16 - s->bps;
+    }
+}
+
 static av_cold int flac_decode_init(AVCodecContext *avctx)
 {
     enum FLACExtradataFormat format;
@@ -118,11 +130,9 @@ static av_cold int flac_decode_init(AVCodecContext *avctx)
 
     /* initialize based on the demuxer-supplied streamdata header */
     avpriv_flac_parse_streaminfo(avctx, (FLACStreaminfo *)s, streaminfo);
-    if (s->bps > 16)
-        avctx->sample_fmt = AV_SAMPLE_FMT_S32;
-    else
-        avctx->sample_fmt = AV_SAMPLE_FMT_S16;
     allocate_buffers(s);
+    flac_set_bps(s);
+    ff_flacdsp_init(&s->dsp, avctx->sample_fmt);
     s->got_streaminfo = 1;
 
     avcodec_get_frame_defaults(&s->frame);
@@ -150,8 +160,7 @@ static void allocate_buffers(FLACContext *s)
     assert(s->max_blocksize);
 
     for (i = 0; i < s->channels; i++) {
-        s->decoded[i] = av_realloc(s->decoded[i],
-                                   sizeof(int32_t)*s->max_blocksize);
+        s->decoded[i] = av_malloc(sizeof(int32_t)*s->max_blocksize);
     }
 }
 
@@ -223,6 +232,8 @@ static int parse_streaminfo(FLACContext *s, const uint8_t *buf, int buf_size)
     }
     avpriv_flac_parse_streaminfo(s->avctx, (FLACStreaminfo *)s, &buf[8]);
     allocate_buffers(s);
+    flac_set_bps(s);
+    ff_flacdsp_init(&s->dsp, s->avctx->sample_fmt);
     s->got_streaminfo = 1;
 
     return 0;
@@ -295,7 +306,8 @@ static int decode_residuals(FLACContext *s, int channel, int pred_order)
     return 0;
 }
 
-static int decode_subframe_fixed(FLACContext *s, int channel, int pred_order)
+static int decode_subframe_fixed(FLACContext *s, int channel, int pred_order,
+                                 int bps)
 {
     const int blocksize = s->blocksize;
     int32_t *decoded = s->decoded[channel];
@@ -303,7 +315,7 @@ static int decode_subframe_fixed(FLACContext *s, int channel, int pred_order)
 
     /* warm up samples */
     for (i = 0; i < pred_order; i++) {
-        decoded[i] = get_sbits_long(&s->gb, s->curr_bps);
+        decoded[i] = get_sbits_long(&s->gb, bps);
     }
 
     if (decode_residuals(s, channel, pred_order) < 0)
@@ -345,16 +357,17 @@ static int decode_subframe_fixed(FLACContext *s, int channel, int pred_order)
     return 0;
 }
 
-static int decode_subframe_lpc(FLACContext *s, int channel, int pred_order)
+static int decode_subframe_lpc(FLACContext *s, int channel, int pred_order,
+                               int bps)
 {
-    int i, j;
+    int i;
     int coeff_prec, qlevel;
     int coeffs[32];
     int32_t *decoded = s->decoded[channel];
 
     /* warm up samples */
     for (i = 0; i < pred_order; i++) {
-        decoded[i] = get_sbits_long(&s->gb, s->curr_bps);
+        decoded[i] = get_sbits_long(&s->gb, bps);
     }
 
     coeff_prec = get_bits(&s->gb, 4) + 1;
@@ -376,38 +389,7 @@ static int decode_subframe_lpc(FLACContext *s, int channel, int pred_order)
     if (decode_residuals(s, channel, pred_order) < 0)
         return -1;
 
-    if (s->bps > 16) {
-        int64_t sum;
-        for (i = pred_order; i < s->blocksize; i++) {
-            sum = 0;
-            for (j = 0; j < pred_order; j++)
-                sum += (int64_t)coeffs[j] * decoded[i-j-1];
-            decoded[i] += sum >> qlevel;
-        }
-    } else {
-        for (i = pred_order; i < s->blocksize-1; i += 2) {
-            int c;
-            int d = decoded[i-pred_order];
-            int s0 = 0, s1 = 0;
-            for (j = pred_order-1; j > 0; j--) {
-                c = coeffs[j];
-                s0 += c*d;
-                d = decoded[i-j];
-                s1 += c*d;
-            }
-            c = coeffs[0];
-            s0 += c*d;
-            d = decoded[i] += s0 >> qlevel;
-            s1 += c*d;
-            decoded[i+1] += s1 >> qlevel;
-        }
-        if (i < s->blocksize) {
-            int sum = 0;
-            for (j = 0; j < pred_order; j++)
-                sum += coeffs[j] * decoded[i-j-1];
-            decoded[i] += sum >> qlevel;
-        }
-    }
+    s->dsp.lpc(decoded, coeffs, pred_order, qlevel, s->blocksize);
 
     return 0;
 }
@@ -415,15 +397,15 @@ static int decode_subframe_lpc(FLACContext *s, int channel, int pred_order)
 static inline int decode_subframe(FLACContext *s, int channel)
 {
     int type, wasted = 0;
+    int bps = s->bps;
     int i, tmp;
 
-    s->curr_bps = s->bps;
     if (channel == 0) {
         if (s->ch_mode == FLAC_CHMODE_RIGHT_SIDE)
-            s->curr_bps++;
+            bps++;
     } else {
         if (s->ch_mode == FLAC_CHMODE_LEFT_SIDE || s->ch_mode == FLAC_CHMODE_MID_SIDE)
-            s->curr_bps++;
+            bps++;
     }
 
     if (get_bits1(&s->gb)) {
@@ -436,35 +418,35 @@ static inline int decode_subframe(FLACContext *s, int channel)
         int left = get_bits_left(&s->gb);
         wasted = 1;
         if ( left < 0 ||
-            (left < s->curr_bps && !show_bits_long(&s->gb, left)) ||
-                                   !show_bits_long(&s->gb, s->curr_bps)) {
+            (left < bps && !show_bits_long(&s->gb, left)) ||
+                           !show_bits_long(&s->gb, bps)) {
             av_log(s->avctx, AV_LOG_ERROR,
                    "Invalid number of wasted bits > available bits (%d) - left=%d\n",
-                   s->curr_bps, left);
+                   bps, left);
             return AVERROR_INVALIDDATA;
         }
         while (!get_bits1(&s->gb))
             wasted++;
-        s->curr_bps -= wasted;
+        bps -= wasted;
     }
-    if (s->curr_bps > 32) {
+    if (bps > 32) {
         av_log_missing_feature(s->avctx, "decorrelated bit depth > 32", 0);
         return -1;
     }
 
 //FIXME use av_log2 for types
     if (type == 0) {
-        tmp = get_sbits_long(&s->gb, s->curr_bps);
+        tmp = get_sbits_long(&s->gb, bps);
         for (i = 0; i < s->blocksize; i++)
             s->decoded[channel][i] = tmp;
     } else if (type == 1) {
         for (i = 0; i < s->blocksize; i++)
-            s->decoded[channel][i] = get_sbits_long(&s->gb, s->curr_bps);
+            s->decoded[channel][i] = get_sbits_long(&s->gb, bps);
     } else if ((type >= 8) && (type <= 12)) {
-        if (decode_subframe_fixed(s, channel, type & ~0x8) < 0)
+        if (decode_subframe_fixed(s, channel, type & ~0x8, bps) < 0)
             return -1;
     } else if (type >= 32) {
-        if (decode_subframe_lpc(s, channel, (type & ~0x20)+1) < 0)
+        if (decode_subframe_lpc(s, channel, (type & ~0x20)+1, bps) < 0)
             return -1;
     } else {
         av_log(s->avctx, AV_LOG_ERROR, "invalid coding type\n");
@@ -512,15 +494,7 @@ static int decode_frame(FLACContext *s)
     }
     s->bps = s->avctx->bits_per_raw_sample = fi.bps;
 
-    if (s->bps > 16) {
-        s->avctx->sample_fmt = AV_SAMPLE_FMT_S32;
-        s->sample_shift = 32 - s->bps;
-        s->is32 = 1;
-    } else {
-        s->avctx->sample_fmt = AV_SAMPLE_FMT_S16;
-        s->sample_shift = 16 - s->bps;
-        s->is32 = 0;
-    }
+    flac_set_bps(s);
 
     if (!s->max_blocksize)
         s->max_blocksize = FLAC_MAX_BLOCKSIZE;
@@ -546,6 +520,7 @@ static int decode_frame(FLACContext *s)
 
     if (!s->got_streaminfo) {
         allocate_buffers(s);
+        ff_flacdsp_init(&s->dsp, s->avctx->sample_fmt);
         s->got_streaminfo = 1;
         dump_headers(s->avctx, (FLACStreaminfo *)s);
     }
@@ -572,9 +547,7 @@ static int flac_decode_frame(AVCodecContext *avctx, void *data,
     const uint8_t *buf = avpkt->data;
     int buf_size = avpkt->size;
     FLACContext *s = avctx->priv_data;
-    int i, j = 0, bytes_read = 0;
-    int16_t *samples_16;
-    int32_t *samples_32;
+    int bytes_read = 0;
     int ret;
 
     *got_frame_ptr = 0;
@@ -614,42 +587,9 @@ static int flac_decode_frame(AVCodecContext *avctx, void *data,
         av_log(avctx, AV_LOG_ERROR, "get_buffer() failed\n");
         return ret;
     }
-    samples_16 = (int16_t *)s->frame.data[0];
-    samples_32 = (int32_t *)s->frame.data[0];
-
-#define DECORRELATE(left, right)\
-            assert(s->channels == 2);\
-            for (i = 0; i < s->blocksize; i++) {\
-                int a= s->decoded[0][i];\
-                int b= s->decoded[1][i];\
-                if (s->is32) {\
-                    *samples_32++ = (left)  << s->sample_shift;\
-                    *samples_32++ = (right) << s->sample_shift;\
-                } else {\
-                    *samples_16++ = (left)  << s->sample_shift;\
-                    *samples_16++ = (right) << s->sample_shift;\
-                }\
-            }\
-            break;
-
-    switch (s->ch_mode) {
-    case FLAC_CHMODE_INDEPENDENT:
-        for (j = 0; j < s->blocksize; j++) {
-            for (i = 0; i < s->channels; i++) {
-                if (s->is32)
-                    *samples_32++ = s->decoded[i][j] << s->sample_shift;
-                else
-                    *samples_16++ = s->decoded[i][j] << s->sample_shift;
-            }
-        }
-        break;
-    case FLAC_CHMODE_LEFT_SIDE:
-        DECORRELATE(a,a-b)
-    case FLAC_CHMODE_RIGHT_SIDE:
-        DECORRELATE(a+b,b)
-    case FLAC_CHMODE_MID_SIDE:
-        DECORRELATE( (a-=b>>1) + b, a)
-    }
+
+    s->dsp.decorrelate[s->ch_mode](s->frame.data, s->decoded, s->channels,
+                                   s->blocksize, s->sample_shift);
 
     if (bytes_read > buf_size) {
         av_log(s->avctx, AV_LOG_ERROR, "overread: %d\n", bytes_read - buf_size);
diff --git a/libavcodec/flacdsp.c b/libavcodec/flacdsp.c
new file mode 100644
index 0000000000..6c90e89d9b
--- /dev/null
+++ b/libavcodec/flacdsp.c
@@ -0,0 +1,94 @@
+/*
+ * Copyright (c) 2012 Mans Rullgard <mans@mansr.com>
+ *
+ * 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 "libavutil/attributes.h"
+#include "libavutil/samplefmt.h"
+#include "flacdsp.h"
+
+#define SAMPLE_SIZE 16
+#include "flacdsp_template.c"
+
+#undef  SAMPLE_SIZE
+#define SAMPLE_SIZE 32
+#include "flacdsp_template.c"
+
+static void flac_lpc_16_c(int32_t *decoded, const int coeffs[32],
+                          int pred_order, int qlevel, int len)
+{
+    int i, j;
+
+    for (i = pred_order; i < len - 1; i += 2) {
+        int c;
+        int d = decoded[i-pred_order];
+        int s0 = 0, s1 = 0;
+        for (j = pred_order-1; j > 0; j--) {
+            c = coeffs[j];
+            s0 += c*d;
+            d = decoded[i-j];
+            s1 += c*d;
+        }
+        c = coeffs[0];
+        s0 += c*d;
+        d = decoded[i] += s0 >> qlevel;
+        s1 += c*d;
+        decoded[i+1] += s1 >> qlevel;
+    }
+    if (i < len) {
+        int sum = 0;
+        for (j = 0; j < pred_order; j++)
+            sum += coeffs[j] * decoded[i-j-1];
+        decoded[i] += sum >> qlevel;
+    }
+}
+
+static void flac_lpc_32_c(int32_t *decoded, const int coeffs[32],
+                          int pred_order, int qlevel, int len)
+{
+    int i, j;
+
+    for (i = pred_order; i < len; i++) {
+        int64_t sum = 0;
+        for (j = 0; j < pred_order; j++)
+            sum += (int64_t)coeffs[j] * decoded[i-j-1];
+        decoded[i] += sum >> qlevel;
+    }
+
+}
+
+av_cold void ff_flacdsp_init(FLACDSPContext *c, enum AVSampleFormat fmt)
+{
+    switch (fmt) {
+    case AV_SAMPLE_FMT_S32:
+        c->decorrelate[0] = flac_decorrelate_indep_c_32;
+        c->decorrelate[1] = flac_decorrelate_ls_c_32;
+        c->decorrelate[2] = flac_decorrelate_rs_c_32;
+        c->decorrelate[3] = flac_decorrelate_ms_c_32;
+        c->lpc            = flac_lpc_32_c;
+        break;
+
+    case AV_SAMPLE_FMT_S16:
+        c->decorrelate[0] = flac_decorrelate_indep_c_16;
+        c->decorrelate[1] = flac_decorrelate_ls_c_16;
+        c->decorrelate[2] = flac_decorrelate_rs_c_16;
+        c->decorrelate[3] = flac_decorrelate_ms_c_16;
+        c->lpc            = flac_lpc_16_c;
+        break;
+    }
+}
diff --git a/libavcodec/flacdsp.h b/libavcodec/flacdsp.h
new file mode 100644
index 0000000000..e2abefb7f5
--- /dev/null
+++ b/libavcodec/flacdsp.h
@@ -0,0 +1,34 @@
+/*
+ * 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 AVCODEC_FLACDSP_H
+#define AVCODEC_FLACDSP_H
+
+#include <stdint.h>
+#include "libavutil/samplefmt.h"
+
+typedef struct FLACDSPContext {
+    void (*decorrelate[4])(uint8_t **out, int32_t **in, int channels,
+                           int len, int shift);
+    void (*lpc)(int32_t *samples, const int coeffs[32], int order,
+                int qlevel, int len);
+} FLACDSPContext;
+
+void ff_flacdsp_init(FLACDSPContext *c, enum AVSampleFormat fmt);
+
+#endif /* AVCODEC_FLACDSP_H */
diff --git a/libavcodec/flacdsp_template.c b/libavcodec/flacdsp_template.c
new file mode 100644
index 0000000000..808b20d41d
--- /dev/null
+++ b/libavcodec/flacdsp_template.c
@@ -0,0 +1,86 @@
+/*
+ * Copyright (c) 2012 Mans Rullgard <mans@mansr.com>
+ *
+ * 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 <stdint.h>
+
+#undef FUNC
+#undef sample
+
+#if SAMPLE_SIZE == 32
+#   define FUNC(n) n ## _32
+#   define sample  int32_t
+#else
+#   define FUNC(n) n ## _16
+#   define sample  int16_t
+#endif
+
+static void FUNC(flac_decorrelate_indep_c)(uint8_t **out, int32_t **in,
+                                           int channels, int len, int shift)
+{
+    sample *samples = (sample *) out[0];
+    int i, j;
+
+    for (j = 0; j < len; j++)
+        for (i = 0; i < channels; i++)
+            *samples++ = in[i][j] << shift;
+}
+
+static void FUNC(flac_decorrelate_ls_c)(uint8_t **out, int32_t **in,
+                                        int channels, int len, int shift)
+{
+    sample *samples = (sample *) out[0];
+    int i;
+
+    for (i = 0; i < len; i++) {
+        int a = in[0][i];
+        int b = in[1][i];
+        *samples++ =  a      << shift;
+        *samples++ = (a - b) << shift;
+    }
+}
+
+static void FUNC(flac_decorrelate_rs_c)(uint8_t **out, int32_t **in,
+                                        int channels, int len, int shift)
+{
+    sample *samples = (sample *) out[0];
+    int i;
+
+    for (i = 0; i < len; i++) {
+        int a = in[0][i];
+        int b = in[1][i];
+        *samples++ = (a + b) << shift;
+        *samples++ =  b      << shift;
+    }
+}
+
+static void FUNC(flac_decorrelate_ms_c)(uint8_t **out, int32_t **in,
+                                        int channels, int len, int shift)
+{
+    sample *samples = (sample *) out[0];
+    int i;
+
+    for (i = 0; i < len; i++) {
+        int a = in[0][i];
+        int b = in[1][i];
+        a -= b >> 1;
+        *samples++ = (a + b) << shift;
+        *samples++ =  a      << shift;
+    }
+}
diff --git a/libavcodec/flacenc.c b/libavcodec/flacenc.c
index b900cc9998..31ab740b59 100644
--- a/libavcodec/flacenc.c
+++ b/libavcodec/flacenc.c
@@ -53,6 +53,7 @@ typedef struct CompressionOptions {
     int prediction_order_method;
     int min_partition_order;
     int max_partition_order;
+    int ch_mode;
 } CompressionOptions;
 
 typedef struct RiceContext {
@@ -1022,15 +1023,8 @@ static int estimate_stereo_mode(int32_t *left_ch, int32_t *right_ch, int n)
     for (i = 1; i < 4; i++)
         if (score[i] < score[best])
             best = i;
-    if (best == 0) {
-        return FLAC_CHMODE_INDEPENDENT;
-    } else if (best == 1) {
-        return FLAC_CHMODE_LEFT_SIDE;
-    } else if (best == 2) {
-        return FLAC_CHMODE_RIGHT_SIDE;
-    } else {
-        return FLAC_CHMODE_MID_SIDE;
-    }
+
+    return best;
 }
 
 
@@ -1053,7 +1047,10 @@ static void channel_decorrelation(FlacEncodeContext *s)
         return;
     }
 
-    frame->ch_mode = estimate_stereo_mode(left, right, n);
+    if (s->options.ch_mode < 0)
+        frame->ch_mode = estimate_stereo_mode(left, right, n);
+    else
+        frame->ch_mode = s->options.ch_mode;
 
     /* perform decorrelation and adjust bits-per-sample */
     if (frame->ch_mode == FLAC_CHMODE_INDEPENDENT)
@@ -1099,7 +1096,7 @@ static void write_frame_header(FlacEncodeContext *s)
     if (frame->ch_mode == FLAC_CHMODE_INDEPENDENT)
         put_bits(&s->pb, 4, s->channels-1);
     else
-        put_bits(&s->pb, 4, frame->ch_mode);
+        put_bits(&s->pb, 4, frame->ch_mode + FLAC_MAX_CHANNELS - 1);
 
     put_bits(&s->pb, 3, 4); /* bits-per-sample code */
     put_bits(&s->pb, 1, 0);
@@ -1308,6 +1305,12 @@ static const AVOption options[] = {
 { "8level",     NULL, 0, AV_OPT_TYPE_CONST, {.dbl = ORDER_METHOD_8LEVEL }, INT_MIN, INT_MAX, FLAGS, "predm" },
 { "search",     NULL, 0, AV_OPT_TYPE_CONST, {.dbl = ORDER_METHOD_SEARCH }, INT_MIN, INT_MAX, FLAGS, "predm" },
 { "log",        NULL, 0, AV_OPT_TYPE_CONST, {.dbl = ORDER_METHOD_LOG },    INT_MIN, INT_MAX, FLAGS, "predm" },
+{ "ch_mode", "Stereo decorrelation mode", offsetof(FlacEncodeContext, options.ch_mode), AV_OPT_TYPE_INT, { .dbl = -1 }, -1, FLAC_CHMODE_MID_SIDE, FLAGS, "ch_mode" },
+{ "auto",       NULL, 0, AV_OPT_TYPE_CONST, { .dbl = -1                      }, INT_MIN, INT_MAX, FLAGS, "ch_mode" },
+{ "indep",      NULL, 0, AV_OPT_TYPE_CONST, { .dbl = FLAC_CHMODE_INDEPENDENT }, INT_MIN, INT_MAX, FLAGS, "ch_mode" },
+{ "left_side",  NULL, 0, AV_OPT_TYPE_CONST, { .dbl = FLAC_CHMODE_LEFT_SIDE   }, INT_MIN, INT_MAX, FLAGS, "ch_mode" },
+{ "right_side", NULL, 0, AV_OPT_TYPE_CONST, { .dbl = FLAC_CHMODE_RIGHT_SIDE  }, INT_MIN, INT_MAX, FLAGS, "ch_mode" },
+{ "mid_side",   NULL, 0, AV_OPT_TYPE_CONST, { .dbl = FLAC_CHMODE_MID_SIDE    }, INT_MIN, INT_MAX, FLAGS, "ch_mode" },
 { NULL },
 };
 
diff --git a/libavcodec/mpegvideo.c b/libavcodec/mpegvideo.c
index 8710c650d6..a3c55d048f 100644
--- a/libavcodec/mpegvideo.c
+++ b/libavcodec/mpegvideo.c
@@ -775,6 +775,11 @@ av_cold int ff_MPV_common_init(MpegEncContext *s)
         if(s->avctx->noise_reduction){
             FF_ALLOCZ_OR_GOTO(s->avctx, s->dct_offset, 2 * 64 * sizeof(uint16_t), fail)
         }
+
+            FF_ALLOC_OR_GOTO(s->avctx, s->cplx_tab,
+                             mb_array_size * sizeof(float), fail);
+            FF_ALLOC_OR_GOTO(s->avctx, s->bits_tab,
+                             mb_array_size * sizeof(float), fail);
     }
 
     s->picture_count = MAX_PICTURE_COUNT * FFMAX(1, s->avctx->thread_count);
@@ -784,7 +789,10 @@ av_cold int ff_MPV_common_init(MpegEncContext *s)
         avcodec_get_frame_defaults(&s->picture[i].f);
     }
 
-    FF_ALLOCZ_OR_GOTO(s->avctx, s->error_status_table, mb_array_size*sizeof(uint8_t), fail)
+        FF_ALLOC_OR_GOTO(s->avctx, s->er_temp_buffer,
+                         mb_array_size * sizeof(uint8_t), fail);
+        FF_ALLOCZ_OR_GOTO(s->avctx, s->error_status_table,
+                          mb_array_size * sizeof(uint8_t), fail);
 
         if(s->codec_id==CODEC_ID_MPEG4 || (s->flags & CODEC_FLAG_INTERLACED_ME)){
             /* interlaced direct mode decoding tables */
@@ -923,6 +931,7 @@ void ff_MPV_common_end(MpegEncContext *s)
     av_freep(&s->avctx->stats_out);
     av_freep(&s->ac_stats);
     av_freep(&s->error_status_table);
+    av_freep(&s->er_temp_buffer);
     av_freep(&s->mb_index2xy);
     av_freep(&s->lambda_table);
     if(s->q_chroma_intra_matrix   != s->q_intra_matrix  ) av_freep(&s->q_chroma_intra_matrix);
@@ -936,6 +945,8 @@ void ff_MPV_common_end(MpegEncContext *s)
     av_freep(&s->input_picture);
     av_freep(&s->reordered_input_picture);
     av_freep(&s->dct_offset);
+    av_freep(&s->cplx_tab);
+    av_freep(&s->bits_tab);
 
     if (s->picture && !s->avctx->internal->is_copy) {
         for (i = 0; i < s->picture_count; i++) {
diff --git a/libavcodec/mpegvideo.h b/libavcodec/mpegvideo.h
index 50b86f5b99..89160d2412 100644
--- a/libavcodec/mpegvideo.h
+++ b/libavcodec/mpegvideo.h
@@ -705,6 +705,12 @@ typedef struct MpegEncContext {
 
     int mpv_flags;      ///< flags set by private options
     int quantizer_noise_shaping;
+
+    /* error resilience stuff */
+    uint8_t *er_temp_buffer;
+
+    /* temp buffers for rate control */
+    float *cplx_tab, *bits_tab;
 } MpegEncContext;
 
 #define REBASE_PICTURE(pic, new_ctx, old_ctx) (pic ? \
diff --git a/libavcodec/ratecontrol.c b/libavcodec/ratecontrol.c
index c4624f9e3c..197c807ef2 100644
--- a/libavcodec/ratecontrol.c
+++ b/libavcodec/ratecontrol.c
@@ -535,8 +535,8 @@ static void adaptive_quantization(MpegEncContext *s, double q){
     const float border_masking = s->avctx->border_masking;
     float bits_sum= 0.0;
     float cplx_sum= 0.0;
-    float *cplx_tab = av_malloc(s->mb_num * sizeof(*cplx_tab));
-    float *bits_tab = av_malloc(s->mb_num * sizeof(*bits_tab));
+    float *cplx_tab = s->cplx_tab;
+    float *bits_tab = s->bits_tab;
     const int qmin= s->avctx->mb_lmin;
     const int qmax= s->avctx->mb_lmax;
     Picture * const pic= &s->current_picture;
@@ -633,9 +633,6 @@ static void adaptive_quantization(MpegEncContext *s, double q){
 //printf("%2d%3d ", intq, ff_sqrt(s->mc_mb_var[i]));
         s->lambda_table[mb_xy]= intq;
     }
-
-    av_free(cplx_tab);
-    av_free(bits_tab);
 }
 
 void ff_get_2pass_fcode(MpegEncContext *s){
diff --git a/libavcodec/version.h b/libavcodec/version.h
index 8709230c24..5cbd41bac8 100644
--- a/libavcodec/version.h
+++ b/libavcodec/version.h
@@ -41,9 +41,11 @@
 #define LIBAVCODEC_IDENT        "Lavc" AV_STRINGIFY(LIBAVCODEC_VERSION)
 
 /**
- * These FF_API_* defines are not part of the public API.
- * They may change, break or disappear at any time.
+ * FF_API_* defines may be placed below to indicate public API that will be
+ * dropped at a future version bump. The defines themselves are not part of
+ * the public API and may change, break or disappear at any time.
  */
+
 #ifndef FF_API_REQUEST_CHANNELS
 #define FF_API_REQUEST_CHANNELS (LIBAVCODEC_VERSION_MAJOR < 55)
 #endif
diff --git a/libavcodec/x86/dsputilenc_mmx.c b/libavcodec/x86/dsputilenc_mmx.c
index 00e0a3fc37..de365ee400 100644
--- a/libavcodec/x86/dsputilenc_mmx.c
+++ b/libavcodec/x86/dsputilenc_mmx.c
@@ -1147,11 +1147,11 @@ void ff_dsputilenc_init_mmx(DSPContext* c, AVCodecContext *avctx)
 
 
         if (mm_flags & AV_CPU_FLAG_MMX2) {
-            c->sum_abs_dctelem= sum_abs_dctelem_mmx2;
 #if HAVE_YASM
             c->hadamard8_diff[0]= ff_hadamard8_diff16_mmx2;
             c->hadamard8_diff[1]= ff_hadamard8_diff_mmx2;
 #endif
+            c->sum_abs_dctelem= sum_abs_dctelem_mmx2;
             c->vsad[4]= vsad_intra16_mmx2;
 
             if(!(avctx->flags & CODEC_FLAG_BITEXACT)){
diff --git a/libavcodec/x86/vp8dsp-init.c b/libavcodec/x86/vp8dsp-init.c
index f4f6e92877..3e550b1543 100644
--- a/libavcodec/x86/vp8dsp-init.c
+++ b/libavcodec/x86/vp8dsp-init.c
@@ -401,11 +401,13 @@ av_cold void ff_vp8dsp_init_x86(VP8DSPContext* c)
 
         c->vp8_h_loop_filter_simple = ff_vp8_h_loop_filter_simple_sse2;
 
+#if ARCH_X86_64 || HAVE_ALIGNED_STACK
         c->vp8_h_loop_filter16y_inner = ff_vp8_h_loop_filter16y_inner_sse2;
         c->vp8_h_loop_filter8uv_inner = ff_vp8_h_loop_filter8uv_inner_sse2;
 
         c->vp8_h_loop_filter16y       = ff_vp8_h_loop_filter16y_mbedge_sse2;
         c->vp8_h_loop_filter8uv       = ff_vp8_h_loop_filter8uv_mbedge_sse2;
+#endif
     }
 
     if (mm_flags & AV_CPU_FLAG_SSSE3) {
@@ -419,6 +421,7 @@ av_cold void ff_vp8dsp_init_x86(VP8DSPContext* c)
         c->vp8_v_loop_filter_simple = ff_vp8_v_loop_filter_simple_ssse3;
         c->vp8_h_loop_filter_simple = ff_vp8_h_loop_filter_simple_ssse3;
 
+#if ARCH_X86_64 || HAVE_ALIGNED_STACK
         c->vp8_v_loop_filter16y_inner = ff_vp8_v_loop_filter16y_inner_ssse3;
         c->vp8_h_loop_filter16y_inner = ff_vp8_h_loop_filter16y_inner_ssse3;
         c->vp8_v_loop_filter8uv_inner = ff_vp8_v_loop_filter8uv_inner_ssse3;
@@ -428,14 +431,17 @@ av_cold void ff_vp8dsp_init_x86(VP8DSPContext* c)
         c->vp8_h_loop_filter16y       = ff_vp8_h_loop_filter16y_mbedge_ssse3;
         c->vp8_v_loop_filter8uv       = ff_vp8_v_loop_filter8uv_mbedge_ssse3;
         c->vp8_h_loop_filter8uv       = ff_vp8_h_loop_filter8uv_mbedge_ssse3;
+#endif
     }
 
     if (mm_flags & AV_CPU_FLAG_SSE4) {
         c->vp8_idct_dc_add                  = ff_vp8_idct_dc_add_sse4;
 
         c->vp8_h_loop_filter_simple   = ff_vp8_h_loop_filter_simple_sse4;
+#if ARCH_X86_64 || HAVE_ALIGNED_STACK
         c->vp8_h_loop_filter16y       = ff_vp8_h_loop_filter16y_mbedge_sse4;
         c->vp8_h_loop_filter8uv       = ff_vp8_h_loop_filter8uv_mbedge_sse4;
+#endif
     }
 #endif
 }