Merge remote-tracking branch 'qatar/master'

* qatar/master:
  aacenc: Fix LONG_START windowing.
  aacenc: Fix a bug where deinterleaved samples were stored in the wrong place.
  avplay: use the correct array size for stride.
  lavc: extend doxy for avcodec_alloc_context3().
  APIchanges: mention avcodec_alloc_context()/2/3
  avcodec_align_dimensions2: set only 4 linesizes, not AV_NUM_DATA_POINTERS.
  aacsbr: ARM NEON optimised sbrdsp functions
  aacsbr: align some arrays
  aacsbr: move some simdable loops to function pointers
  cosmetics: Remove extra newlines at EOF

Conflicts:
	libavcodec/utils.c
	libavfilter/formats.c
	libavutil/mem.c

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

70 files changed, 847 insertions, 208 deletions

diff --git a/libavcodec/4xm.c b/libavcodec/4xm.c
index d665436844..9581f94344 100644
--- a/libavcodec/4xm.c
+++ b/libavcodec/4xm.c
@@ -935,4 +935,3 @@ AVCodec ff_fourxm_decoder = {
     .capabilities   = CODEC_CAP_DR1,
     .long_name = NULL_IF_CONFIG_SMALL("4X Movie"),
 };
-
diff --git a/libavcodec/Makefile b/libavcodec/Makefile
index 089b8668b3..8fd057d95a 100644
--- a/libavcodec/Makefile
+++ b/libavcodec/Makefile
@@ -56,7 +56,8 @@ OBJS-$(CONFIG_VDPAU)                   += vdpau.o
 OBJS-$(CONFIG_A64MULTI_ENCODER)        += a64multienc.o elbg.o
 OBJS-$(CONFIG_A64MULTI5_ENCODER)       += a64multienc.o elbg.o
 OBJS-$(CONFIG_AAC_DECODER)             += aacdec.o aactab.o aacsbr.o aacps.o \
-                                          aacadtsdec.o mpeg4audio.o kbdwin.o
+                                          aacadtsdec.o mpeg4audio.o kbdwin.o \
+                                          sbrdsp.o
 OBJS-$(CONFIG_AAC_ENCODER)             += aacenc.o aaccoder.o    \
                                           aacpsy.o aactab.o      \
                                           psymodel.o iirfilter.o \
diff --git a/libavcodec/aacenc.c b/libavcodec/aacenc.c
index 55f028687c..2ff6f9cc04 100644
--- a/libavcodec/aacenc.c
+++ b/libavcodec/aacenc.c
@@ -200,8 +200,8 @@ WINDOW_FUNC(long_start)
     float *out = sce->ret;
 
     dsp->vector_fmul(out, audio, lwindow, 1024);
-    memcpy(out + 1024, audio, sizeof(out[0]) * 448);
-    dsp->vector_fmul_reverse(out + 1024 + 448, audio, swindow, 128);
+    memcpy(out + 1024, audio + 1024, sizeof(out[0]) * 448);
+    dsp->vector_fmul_reverse(out + 1024 + 448, audio + 1024 + 448, swindow, 128);
     memset(out + 1024 + 576, 0, sizeof(out[0]) * 448);
 }
 
@@ -487,10 +487,10 @@ static void deinterleave_input_samples(AACEncContext *s,
         const float *sptr = samples + channel_map[ch];
 
         /* copy last 1024 samples of previous frame to the start of the current frame */
-        memcpy(&s->planar_samples[ch][0], &s->planar_samples[ch][1024], 1024 * sizeof(s->planar_samples[0][0]));
+        memcpy(&s->planar_samples[ch][1024], &s->planar_samples[ch][2048], 1024 * sizeof(s->planar_samples[0][0]));
 
         /* deinterleave */
-        for (i = 1024; i < 1024 * 2; i++) {
+        for (i = 2048; i < 3072; i++) {
             s->planar_samples[ch][i] = *sptr;
             sptr += sinc;
         }
diff --git a/libavcodec/aacsbr.c b/libavcodec/aacsbr.c
index 2f457b61cc..71bd2ceda5 100644
--- a/libavcodec/aacsbr.c
+++ b/libavcodec/aacsbr.c
@@ -32,6 +32,7 @@
 #include "aacsbrdata.h"
 #include "fft.h"
 #include "aacps.h"
+#include "sbrdsp.h"
 #include "libavutil/libm.h"
 #include "libavutil/avassert.h"
 
@@ -144,6 +145,7 @@ av_cold void ff_aac_sbr_ctx_init(AACContext *ac, SpectralBandReplication *sbr)
     ff_mdct_init(&sbr->mdct,     7, 1, 1.0 / (64 * mdct_scale));
     ff_mdct_init(&sbr->mdct_ana, 7, 1, -2.0 * mdct_scale);
     ff_ps_ctx_init(&sbr->ps);
+    ff_sbrdsp_init(&sbr->dsp);
 }
 
 av_cold void ff_aac_sbr_ctx_close(SpectralBandReplication *sbr)
@@ -1143,33 +1145,21 @@ static void sbr_dequant(SpectralBandReplication *sbr, int id_aac)
  * @param   x       pointer to the beginning of the first sample window
  * @param   W       array of complex-valued samples split into subbands
  */
-static void sbr_qmf_analysis(DSPContext *dsp, FFTContext *mdct, const float *in, float *x,
+static void sbr_qmf_analysis(DSPContext *dsp, FFTContext *mdct,
+                             SBRDSPContext *sbrdsp, const float *in, float *x,
                              float z[320], float W[2][32][32][2])
 {
-    int i, k;
+    int i;
     memcpy(W[0], W[1], sizeof(W[0]));
     memcpy(x    , x+1024, (320-32)*sizeof(x[0]));
     memcpy(x+288, in,         1024*sizeof(x[0]));
     for (i = 0; i < 32; i++) { // numTimeSlots*RATE = 16*2 as 960 sample frames
                                // are not supported
         dsp->vector_fmul_reverse(z, sbr_qmf_window_ds, x, 320);
-        for (k = 0; k < 64; k++) {
-            float f = z[k] + z[k + 64] + z[k + 128] + z[k + 192] + z[k + 256];
-            z[k] = f;
-        }
-        //Shuffle to IMDCT
-        z[64] = z[0];
-        for (k = 1; k < 32; k++) {
-            z[64+2*k-1] =  z[   k];
-            z[64+2*k  ] = -z[64-k];
-        }
-        z[64+63] = z[32];
-
+        sbrdsp->sum64x5(z);
+        sbrdsp->qmf_pre_shuffle(z);
         mdct->imdct_half(mdct, z, z+64);
-        for (k = 0; k < 32; k++) {
-            W[1][i][k][0] = -z[63-k];
-            W[1][i][k][1] = z[k];
-        }
+        sbrdsp->qmf_post_shuffle(W[1][i], z);
         x += 32;
     }
 }
@@ -1179,6 +1169,7 @@ static void sbr_qmf_analysis(DSPContext *dsp, FFTContext *mdct, const float *in,
  * (14496-3 sp04 p206)
  */
 static void sbr_qmf_synthesis(DSPContext *dsp, FFTContext *mdct,
+                              SBRDSPContext *sbrdsp,
                               float *out, float X[2][38][64],
                               float mdct_buf[2][64],
                               float *v0, int *v_off, const unsigned int div)
@@ -1202,20 +1193,12 @@ static void sbr_qmf_synthesis(DSPContext *dsp, FFTContext *mdct,
                 X[0][i][32+n] =  X[1][i][31-n];
             }
             mdct->imdct_half(mdct, mdct_buf[0], X[0][i]);
-            for (n = 0; n < 32; n++) {
-                v[     n] =  mdct_buf[0][63 - 2*n];
-                v[63 - n] = -mdct_buf[0][62 - 2*n];
-            }
+            sbrdsp->qmf_deint_neg(v, mdct_buf[0]);
         } else {
-            for (n = 1; n < 64; n+=2) {
-                X[1][i][n] = -X[1][i][n];
-            }
+            sbrdsp->neg_odd_64(X[1][i]);
             mdct->imdct_half(mdct, mdct_buf[0], X[0][i]);
             mdct->imdct_half(mdct, mdct_buf[1], X[1][i]);
-            for (n = 0; n < 64; n++) {
-                v[      n] = -mdct_buf[0][63 -   n] + mdct_buf[1][  n    ];
-                v[127 - n] =  mdct_buf[0][63 -   n] + mdct_buf[1][  n    ];
-            }
+            sbrdsp->qmf_deint_bfly(v, mdct_buf[1], mdct_buf[0]);
         }
         dsp->vector_fmul_add(out, v                , sbr_qmf_window               , zero64, 64 >> div);
         dsp->vector_fmul_add(out, v + ( 192 >> div), sbr_qmf_window + ( 64 >> div), out   , 64 >> div);
@@ -1231,45 +1214,20 @@ static void sbr_qmf_synthesis(DSPContext *dsp, FFTContext *mdct,
     }
 }
 
-static void autocorrelate(const float x[40][2], float phi[3][2][2], int lag)
-{
-    int i;
-    float real_sum = 0.0f;
-    float imag_sum = 0.0f;
-    if (lag) {
-        for (i = 1; i < 38; i++) {
-            real_sum += x[i][0] * x[i+lag][0] + x[i][1] * x[i+lag][1];
-            imag_sum += x[i][0] * x[i+lag][1] - x[i][1] * x[i+lag][0];
-        }
-        phi[2-lag][1][0] = real_sum + x[ 0][0] * x[lag][0] + x[ 0][1] * x[lag][1];
-        phi[2-lag][1][1] = imag_sum + x[ 0][0] * x[lag][1] - x[ 0][1] * x[lag][0];
-        if (lag == 1) {
-            phi[0][0][0] = real_sum + x[38][0] * x[39][0] + x[38][1] * x[39][1];
-            phi[0][0][1] = imag_sum + x[38][0] * x[39][1] - x[38][1] * x[39][0];
-        }
-    } else {
-        for (i = 1; i < 38; i++) {
-            real_sum += x[i][0] * x[i][0] + x[i][1] * x[i][1];
-        }
-        phi[2][1][0] = real_sum + x[ 0][0] * x[ 0][0] + x[ 0][1] * x[ 0][1];
-        phi[1][0][0] = real_sum + x[38][0] * x[38][0] + x[38][1] * x[38][1];
-    }
-}
-
 /** High Frequency Generation (14496-3 sp04 p214+) and Inverse Filtering
  * (14496-3 sp04 p214)
  * Warning: This routine does not seem numerically stable.
  */
-static void sbr_hf_inverse_filter(float (*alpha0)[2], float (*alpha1)[2],
+static void sbr_hf_inverse_filter(SBRDSPContext *dsp,
+                                  float (*alpha0)[2], float (*alpha1)[2],
                                   const float X_low[32][40][2], int k0)
 {
     int k;
     for (k = 0; k < k0; k++) {
-        float phi[3][2][2], dk;
+        LOCAL_ALIGNED_16(float, phi, [3], [2][2]);
+        float dk;
 
-        autocorrelate(X_low[k], phi, 0);
-        autocorrelate(X_low[k], phi, 1);
-        autocorrelate(X_low[k], phi, 2);
+        dsp->autocorrelate(X_low[k], phi);
 
         dk =  phi[2][1][0] * phi[1][0][0] -
              (phi[1][1][0] * phi[1][1][0] + phi[1][1][1] * phi[1][1][1]) / 1.000001f;
@@ -1365,12 +1323,11 @@ static int sbr_hf_gen(AACContext *ac, SpectralBandReplication *sbr,
                       const float bw_array[5], const uint8_t *t_env,
                       int bs_num_env)
 {
-    int i, j, x;
+    int j, x;
     int g = 0;
     int k = sbr->kx[1];
     for (j = 0; j < sbr->num_patches; j++) {
         for (x = 0; x < sbr->patch_num_subbands[j]; x++, k++) {
-            float alpha[4];
             const int p = sbr->patch_start_subband[j] + x;
             while (g <= sbr->n_q && k >= sbr->f_tablenoise[g])
                 g++;
@@ -1382,26 +1339,10 @@ static int sbr_hf_gen(AACContext *ac, SpectralBandReplication *sbr,
                 return -1;
             }
 
-            alpha[0] = alpha1[p][0] * bw_array[g] * bw_array[g];
-            alpha[1] = alpha1[p][1] * bw_array[g] * bw_array[g];
-            alpha[2] = alpha0[p][0] * bw_array[g];
-            alpha[3] = alpha0[p][1] * bw_array[g];
-
-            for (i = 2 * t_env[0]; i < 2 * t_env[bs_num_env]; i++) {
-                const int idx = i + ENVELOPE_ADJUSTMENT_OFFSET;
-                X_high[k][idx][0] =
-                    X_low[p][idx - 2][0] * alpha[0] -
-                    X_low[p][idx - 2][1] * alpha[1] +
-                    X_low[p][idx - 1][0] * alpha[2] -
-                    X_low[p][idx - 1][1] * alpha[3] +
-                    X_low[p][idx][0];
-                X_high[k][idx][1] =
-                    X_low[p][idx - 2][1] * alpha[0] +
-                    X_low[p][idx - 2][0] * alpha[1] +
-                    X_low[p][idx - 1][1] * alpha[2] +
-                    X_low[p][idx - 1][0] * alpha[3] +
-                    X_low[p][idx][1];
-            }
+            sbr->dsp.hf_gen(X_high[k] + ENVELOPE_ADJUSTMENT_OFFSET,
+                            X_low[p]  + ENVELOPE_ADJUSTMENT_OFFSET,
+                            alpha0[p], alpha1[p], bw_array[g],
+                            2 * t_env[0], 2 * t_env[bs_num_env]);
         }
     }
     if (k < sbr->m[1] + sbr->kx[1])
@@ -1502,7 +1443,8 @@ static void sbr_mapping(AACContext *ac, SpectralBandReplication *sbr,
 static void sbr_env_estimate(float (*e_curr)[48], float X_high[64][40][2],
                              SpectralBandReplication *sbr, SBRData *ch_data)
 {
-    int e, i, m;
+    int e, m;
+    int kx1 = sbr->kx[1];
 
     if (sbr->bs_interpol_freq) {
         for (e = 0; e < ch_data->bs_num_env; e++) {
@@ -1511,12 +1453,7 @@ static void sbr_env_estimate(float (*e_curr)[48], float X_high[64][40][2],
             int iub = ch_data->t_env[e + 1] * 2 + ENVELOPE_ADJUSTMENT_OFFSET;
 
             for (m = 0; m < sbr->m[1]; m++) {
-                float sum = 0.0f;
-
-                for (i = ilb; i < iub; i++) {
-                    sum += X_high[m + sbr->kx[1]][i][0] * X_high[m + sbr->kx[1]][i][0] +
-                           X_high[m + sbr->kx[1]][i][1] * X_high[m + sbr->kx[1]][i][1];
-                }
+                float sum = sbr->dsp.sum_square(X_high[m+kx1] + ilb, iub - ilb);
                 e_curr[e][m] = sum * recip_env_size;
             }
         }
@@ -1534,14 +1471,11 @@ static void sbr_env_estimate(float (*e_curr)[48], float X_high[64][40][2],
                 const int den = env_size * (table[p + 1] - table[p]);
 
                 for (k = table[p]; k < table[p + 1]; k++) {
-                    for (i = ilb; i < iub; i++) {
-                        sum += X_high[k][i][0] * X_high[k][i][0] +
-                               X_high[k][i][1] * X_high[k][i][1];
-                    }
+                    sum += sbr->dsp.sum_square(X_high[k] + ilb, iub - ilb);
                 }
                 sum /= den;
                 for (k = table[p]; k < table[p + 1]; k++) {
-                    e_curr[e][k - sbr->kx[1]] = sum;
+                    e_curr[e][k - kx1] = sum;
                 }
             }
         }
@@ -1652,55 +1586,35 @@ static void sbr_hf_assemble(float Y[2][38][64][2], const float X_high[64][40][2]
     for (e = 0; e < ch_data->bs_num_env; e++) {
         for (i = 2 * ch_data->t_env[e]; i < 2 * ch_data->t_env[e + 1]; i++) {
             int phi_sign = (1 - 2*(kx & 1));
+            LOCAL_ALIGNED_16(float, g_filt_tab, [48]);
+            LOCAL_ALIGNED_16(float, q_filt_tab, [48]);
+            float *g_filt, *q_filt;
 
             if (h_SL && e != e_a[0] && e != e_a[1]) {
+                g_filt = g_filt_tab;
+                q_filt = q_filt_tab;
                 for (m = 0; m < m_max; m++) {
                     const int idx1 = i + h_SL;
-                    float g_filt = 0.0f;
-                    for (j = 0; j <= h_SL; j++)
-                        g_filt += g_temp[idx1 - j][m] * h_smooth[j];
-                    Y[1][i][m + kx][0] =
-                        X_high[m + kx][i + ENVELOPE_ADJUSTMENT_OFFSET][0] * g_filt;
-                    Y[1][i][m + kx][1] =
-                        X_high[m + kx][i + ENVELOPE_ADJUSTMENT_OFFSET][1] * g_filt;
+                    g_filt[m] = 0.0f;
+                    q_filt[m] = 0.0f;
+                    for (j = 0; j <= h_SL; j++) {
+                        g_filt[m] += g_temp[idx1 - j][m] * h_smooth[j];
+                        q_filt[m] += q_temp[idx1 - j][m] * h_smooth[j];
+                    }
                 }
             } else {
-                for (m = 0; m < m_max; m++) {
-                    const float g_filt = g_temp[i + h_SL][m];
-                    Y[1][i][m + kx][0] =
-                        X_high[m + kx][i + ENVELOPE_ADJUSTMENT_OFFSET][0] * g_filt;
-                    Y[1][i][m + kx][1] =
-                        X_high[m + kx][i + ENVELOPE_ADJUSTMENT_OFFSET][1] * g_filt;
-                }
+                g_filt = g_temp[i + h_SL];
+                q_filt = q_temp[i];
             }
 
+            sbr->dsp.hf_g_filt(Y[1][i] + kx, X_high + kx, g_filt, m_max,
+                               i + ENVELOPE_ADJUSTMENT_OFFSET);
+
             if (e != e_a[0] && e != e_a[1]) {
-                for (m = 0; m < m_max; m++) {
-                    indexnoise = (indexnoise + 1) & 0x1ff;
-                    if (sbr->s_m[e][m]) {
-                        Y[1][i][m + kx][0] +=
-                            sbr->s_m[e][m] * phi[0][indexsine];
-                        Y[1][i][m + kx][1] +=
-                            sbr->s_m[e][m] * (phi[1][indexsine] * phi_sign);
-                    } else {
-                        float q_filt;
-                        if (h_SL) {
-                            const int idx1 = i + h_SL;
-                            q_filt = 0.0f;
-                            for (j = 0; j <= h_SL; j++)
-                                q_filt += q_temp[idx1 - j][m] * h_smooth[j];
-                        } else {
-                            q_filt = q_temp[i][m];
-                        }
-                        Y[1][i][m + kx][0] +=
-                            q_filt * sbr_noise_table[indexnoise][0];
-                        Y[1][i][m + kx][1] +=
-                            q_filt * sbr_noise_table[indexnoise][1];
-                    }
-                    phi_sign = -phi_sign;
-                }
+                sbr->dsp.hf_apply_noise[indexsine](Y[1][i] + kx, sbr->s_m[e],
+                                                   q_filt, indexnoise,
+                                                   kx, m_max);
             } else {
-                indexnoise = (indexnoise + m_max) & 0x1ff;
                 for (m = 0; m < m_max; m++) {
                     Y[1][i][m + kx][0] +=
                         sbr->s_m[e][m] * phi[0][indexsine];
@@ -1709,6 +1623,7 @@ static void sbr_hf_assemble(float Y[2][38][64][2], const float X_high[64][40][2]
                     phi_sign = -phi_sign;
                 }
             }
+            indexnoise = (indexnoise + m_max) & 0x1ff;
             indexsine = (indexsine + 1) & 3;
         }
     }
@@ -1728,12 +1643,12 @@ void ff_sbr_apply(AACContext *ac, SpectralBandReplication *sbr, int id_aac,
     }
     for (ch = 0; ch < nch; ch++) {
         /* decode channel */
-        sbr_qmf_analysis(&ac->dsp, &sbr->mdct_ana, ch ? R : L, sbr->data[ch].analysis_filterbank_samples,
+        sbr_qmf_analysis(&ac->dsp, &sbr->mdct_ana, &sbr->dsp, ch ? R : L, sbr->data[ch].analysis_filterbank_samples,
                          (float*)sbr->qmf_filter_scratch,
                          sbr->data[ch].W);
         sbr_lf_gen(ac, sbr, sbr->X_low, sbr->data[ch].W);
         if (sbr->start) {
-            sbr_hf_inverse_filter(sbr->alpha0, sbr->alpha1, sbr->X_low, sbr->k[0]);
+            sbr_hf_inverse_filter(&sbr->dsp, sbr->alpha0, sbr->alpha1, sbr->X_low, sbr->k[0]);
             sbr_chirp(sbr, &sbr->data[ch]);
             sbr_hf_gen(ac, sbr, sbr->X_high, sbr->X_low, sbr->alpha0, sbr->alpha1,
                        sbr->data[ch].bw_array, sbr->data[ch].t_env,
@@ -1760,12 +1675,12 @@ void ff_sbr_apply(AACContext *ac, SpectralBandReplication *sbr, int id_aac,
         nch = 2;
     }
 
-    sbr_qmf_synthesis(&ac->dsp, &sbr->mdct, L, sbr->X[0], sbr->qmf_filter_scratch,
+    sbr_qmf_synthesis(&ac->dsp, &sbr->mdct, &sbr->dsp, L, sbr->X[0], sbr->qmf_filter_scratch,
                       sbr->data[0].synthesis_filterbank_samples,
                       &sbr->data[0].synthesis_filterbank_samples_offset,
                       downsampled);
     if (nch == 2)
-        sbr_qmf_synthesis(&ac->dsp, &sbr->mdct, R, sbr->X[1], sbr->qmf_filter_scratch,
+        sbr_qmf_synthesis(&ac->dsp, &sbr->mdct, &sbr->dsp, R, sbr->X[1], sbr->qmf_filter_scratch,
                           sbr->data[1].synthesis_filterbank_samples,
                           &sbr->data[1].synthesis_filterbank_samples_offset,
                           downsampled);
diff --git a/libavcodec/aacsbrdata.h b/libavcodec/aacsbrdata.h
index 5d33a60888..a796e9deb7 100644
--- a/libavcodec/aacsbrdata.h
+++ b/libavcodec/aacsbrdata.h
@@ -352,7 +352,8 @@ static DECLARE_ALIGNED(16, float, sbr_qmf_window_us)[640] = {
      0.8537385600,
 };
 
-static const float sbr_noise_table[512][2] = {
+/* First two entries repeated at end to simplify SIMD implementations. */
+const DECLARE_ALIGNED(16, float, ff_sbr_noise_table)[][2] = {
 {-0.99948153278296, -0.59483417516607}, { 0.97113454393991, -0.67528515225647},
 { 0.14130051758487, -0.95090983575689}, {-0.47005496701697, -0.37340549728647},
 { 0.80705063769351,  0.29653668284408}, {-0.38981478896926,  0.89572605717087},
@@ -609,6 +610,7 @@ static const float sbr_noise_table[512][2] = {
 {-0.93412041758744,  0.41374052024363}, { 0.96063943315511,  0.93116709541280},
 { 0.97534253457837,  0.86150930812689}, { 0.99642466504163,  0.70190043427512},
 {-0.94705089665984, -0.29580042814306}, { 0.91599807087376, -0.98147830385781},
+{-0.99948153278296, -0.59483417516607}, { 0.97113454393991, -0.67528515225647},
 };
 
 #endif /* AVCODEC_AACSBRDATA_H */
diff --git a/libavcodec/acelp_filters.c b/libavcodec/acelp_filters.c
index 31f0e86f5b..1ce5eed5e2 100644
--- a/libavcodec/acelp_filters.c
+++ b/libavcodec/acelp_filters.c
@@ -142,4 +142,3 @@ void ff_tilt_compensation(float *mem, float tilt, float *samples, int size)
     samples[0] -= tilt * *mem;
     *mem = new_tilt_mem;
 }
-
diff --git a/libavcodec/allcodecs.c b/libavcodec/allcodecs.c
index 32f3f52942..f6fc15388a 100644
--- a/libavcodec/allcodecs.c
+++ b/libavcodec/allcodecs.c
@@ -465,4 +465,3 @@ void avcodec_register_all(void)
     REGISTER_BSF     (REMOVE_EXTRADATA, remove_extradata);
     REGISTER_BSF     (TEXT2MOVSUB, text2movsub);
 }
-
diff --git a/libavcodec/alsdec.c b/libavcodec/alsdec.c
index 9d371de0c8..6ce48de7de 100644
--- a/libavcodec/alsdec.c
+++ b/libavcodec/alsdec.c
@@ -1755,4 +1755,3 @@ AVCodec ff_als_decoder = {
     .capabilities = CODEC_CAP_SUBFRAMES | CODEC_CAP_DR1,
     .long_name = NULL_IF_CONFIG_SMALL("MPEG-4 Audio Lossless Coding (ALS)"),
 };
-
diff --git a/libavcodec/amrnbdata.h b/libavcodec/amrnbdata.h
index fab9f759ad..2745d33ffb 100644
--- a/libavcodec/amrnbdata.h
+++ b/libavcodec/amrnbdata.h
@@ -1669,4 +1669,3 @@ static const float highpass_poles[2] = { -1.933105469, 0.935913085 };
 static const float highpass_gain     = 0.939819335;
 
 #endif /* AVCODEC_AMRNBDATA_H */
-
diff --git a/libavcodec/arm/Makefile b/libavcodec/arm/Makefile
index 52709b6ae9..0e45bc0a52 100644
--- a/libavcodec/arm/Makefile
+++ b/libavcodec/arm/Makefile
@@ -1,6 +1,8 @@
 OBJS-$(CONFIG_AC3DSP)                  += arm/ac3dsp_init_arm.o         \
                                           arm/ac3dsp_arm.o
 
+OBJS-$(CONFIG_AAC_DECODER)             += arm/sbrdsp_init_arm.o
+
 OBJS-$(CONFIG_DCA_DECODER)             += arm/dcadsp_init_arm.o         \
 
 ARMV6-OBJS-$(CONFIG_AC3DSP)            += arm/ac3dsp_armv6.o
@@ -61,6 +63,8 @@ NEON-OBJS-$(CONFIG_H264PRED)           += arm/h264pred_neon.o           \
 
 NEON-OBJS-$(CONFIG_AC3DSP)             += arm/ac3dsp_neon.o
 
+NEON-OBJS-$(CONFIG_AAC_DECODER)        += arm/sbrdsp_neon.o
+
 NEON-OBJS-$(CONFIG_DCA_DECODER)        += arm/dcadsp_neon.o             \
                                           arm/synth_filter_neon.o       \
 
diff --git a/libavcodec/arm/sbrdsp_init_arm.c b/libavcodec/arm/sbrdsp_init_arm.c
new file mode 100644
index 0000000000..2ab0df829d
--- /dev/null
+++ b/libavcodec/arm/sbrdsp_init_arm.c
@@ -0,0 +1,70 @@
+/*
+ * Copyright (c) 2012 Mans Rullgard
+ *
+ * This file is part of Libav.
+ *
+ * Libav 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.
+ *
+ * Libav 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 Libav; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#include "config.h"
+#include "libavutil/attributes.h"
+#include "libavcodec/sbrdsp.h"
+
+void ff_sbr_sum64x5_neon(float *z);
+float ff_sbr_sum_square_neon(float (*x)[2], int n);
+void ff_sbr_neg_odd_64_neon(float *x);
+void ff_sbr_qmf_pre_shuffle_neon(float *z);
+void ff_sbr_qmf_post_shuffle_neon(float W[32][2], const float *z);
+void ff_sbr_qmf_deint_neg_neon(float *v, const float *src);
+void ff_sbr_qmf_deint_bfly_neon(float *v, const float *src0, const float *src1);
+void ff_sbr_hf_g_filt_neon(float (*Y)[2], const float (*X_high)[40][2],
+                           const float *g_filt, int m_max, int ixh);
+void ff_sbr_hf_gen_neon(float (*X_high)[2], const float (*X_low)[2],
+                        const float alpha0[2], const float alpha1[2],
+                        float bw, int start, int end);
+void ff_sbr_autocorrelate_neon(const float x[40][2], float phi[3][2][2]);
+
+void ff_sbr_hf_apply_noise_0_neon(float Y[64][2], const float *s_m,
+                                  const float *q_filt, int noise,
+                                  int kx, int m_max);
+void ff_sbr_hf_apply_noise_1_neon(float Y[64][2], const float *s_m,
+                                  const float *q_filt, int noise,
+                                  int kx, int m_max);
+void ff_sbr_hf_apply_noise_2_neon(float Y[64][2], const float *s_m,
+                                  const float *q_filt, int noise,
+                                  int kx, int m_max);
+void ff_sbr_hf_apply_noise_3_neon(float Y[64][2], const float *s_m,
+                                  const float *q_filt, int noise,
+                                  int kx, int m_max);
+
+av_cold void ff_sbrdsp_init_arm(SBRDSPContext *s)
+{
+    if (HAVE_NEON) {
+        s->sum64x5 = ff_sbr_sum64x5_neon;
+        s->sum_square = ff_sbr_sum_square_neon;
+        s->neg_odd_64 = ff_sbr_neg_odd_64_neon;
+        s->qmf_pre_shuffle = ff_sbr_qmf_pre_shuffle_neon;
+        s->qmf_post_shuffle = ff_sbr_qmf_post_shuffle_neon;
+        s->qmf_deint_neg = ff_sbr_qmf_deint_neg_neon;
+        s->qmf_deint_bfly = ff_sbr_qmf_deint_bfly_neon;
+        s->hf_g_filt = ff_sbr_hf_g_filt_neon;
+        s->hf_gen = ff_sbr_hf_gen_neon;
+        s->autocorrelate = ff_sbr_autocorrelate_neon;
+        s->hf_apply_noise[0] = ff_sbr_hf_apply_noise_0_neon;
+        s->hf_apply_noise[1] = ff_sbr_hf_apply_noise_1_neon;
+        s->hf_apply_noise[2] = ff_sbr_hf_apply_noise_2_neon;
+        s->hf_apply_noise[3] = ff_sbr_hf_apply_noise_3_neon;
+    }
+}
diff --git a/libavcodec/arm/sbrdsp_neon.S b/libavcodec/arm/sbrdsp_neon.S
new file mode 100644
index 0000000000..835c32caee
--- /dev/null
+++ b/libavcodec/arm/sbrdsp_neon.S
@@ -0,0 +1,411 @@
+/*
+ * Copyright (c) 2012 Mans Rullgard
+ *
+ * This file is part of Libav.
+ *
+ * Libav 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.
+ *
+ * Libav 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 Libav; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#include "asm.S"
+
+function ff_sbr_sum64x5_neon, export=1
+        push            {lr}
+        add             r1,  r0,  # 64*4
+        add             r2,  r0,  #128*4
+        add             r3,  r0,  #192*4
+        add             lr,  r0,  #256*4
+        mov             r12, #64
+1:
+        vld1.32         {q0},     [r0,:128]
+        vld1.32         {q1},     [r1,:128]!
+        vadd.f32        q0,  q0,  q1
+        vld1.32         {q2},     [r2,:128]!
+        vadd.f32        q0,  q0,  q2
+        vld1.32         {q3},     [r3,:128]!
+        vadd.f32        q0,  q0,  q3
+        vld1.32         {q8},     [lr,:128]!
+        vadd.f32        q0,  q0,  q8
+        vst1.32         {q0},     [r0,:128]!
+        subs            r12, #4
+        bgt             1b
+        pop             {pc}
+endfunc
+
+function ff_sbr_sum_square_neon, export=1
+        vmov.f32        q0,  #0.0
+1:
+        vld1.32         {q1},     [r0,:128]!
+        vmla.f32        q0,  q1,  q1
+        subs            r1,  r1,  #2
+        bgt             1b
+        vadd.f32        d0,  d0,  d1
+        vpadd.f32       d0,  d0,  d0
+NOVFP   vmov.32         r0,  d0[0]
+        bx              lr
+endfunc
+
+function ff_sbr_neg_odd_64_neon, export=1
+        mov             r1,  r0
+        vmov.i32        q8,  #1<<31
+        vld2.32         {q0,q1},  [r0,:128]!
+        veor            q1,  q1,  q8
+        vld2.32         {q2,q3},  [r0,:128]!
+    .rept 3
+        vst2.32         {q0,q1},  [r1,:128]!
+        veor            q3,  q3,  q8
+        vld2.32         {q0,q1},  [r0,:128]!
+        vst2.32         {q2,q3},  [r1,:128]!
+        veor            q1,  q1,  q8
+        vld2.32         {q2,q3},  [r0,:128]!
+    .endr
+        veor            q3,  q3,  q8
+        vst2.32         {q0,q1},  [r1,:128]!
+        vst2.32         {q2,q3},  [r1,:128]!
+        bx              lr
+endfunc
+
+function ff_sbr_qmf_pre_shuffle_neon, export=1
+        add             r1,  r0,  #60*4
+        add             r2,  r0,  #64*4
+        vld1.32         {d0},     [r0,:64]!
+        vst1.32         {d0},     [r2,:64]!
+        mov             r3,  #-16
+        mov             r12, #24
+        vmov.i32        q8,  #1<<31
+        vld1.32         {q0},     [r1,:128], r3
+        vld1.32         {d2},     [r0,:64]!
+1:
+        vld1.32         {d3,d4},  [r0,:128]!
+        vrev64.32       q0,  q0
+        vld1.32         {q9},     [r1,:128], r3
+        veor            q0,  q0,  q8
+        vld1.32         {d5,d6},  [r0,:128]!
+        vswp            d0,  d1
+        vrev64.32       q9,  q9
+        vst2.32         {q0,q1},  [r2,:64]!
+        vmov            q10, q2
+        veor            q9,  q9,  q8
+        vmov            d2,  d6
+        vswp            d18, d19
+        vld1.32         {q0},     [r1,:128], r3
+        vst2.32         {q9,q10}, [r2,:64]!
+        subs            r12, r12, #8
+        bgt             1b
+        vld1.32         {d3,d4},  [r0,:128]!
+        vrev64.32       q0,  q0
+        vld1.32         {q9},     [r1,:128], r3
+        veor            q0,  q0,  q8
+        vld1.32         {d5},     [r0,:64]!
+        vswp            d0,  d1
+        vrev64.32       q9,  q9
+        vst2.32         {q0,q1},  [r2,:64]!
+        vswp            d4,  d5
+        veor            q1,  q9,  q8
+        vst2.32         {d3,d5},  [r2,:64]!
+        vst2.32         {d2[0],d4[0]}, [r2,:64]!
+        bx              lr
+endfunc
+
+function ff_sbr_qmf_post_shuffle_neon, export=1
+        add             r2,  r1,  #60*4
+        mov             r3,  #-16
+        mov             r12, #32
+        vmov.i32        q8,  #1<<31
+        vld1.32         {q0},     [r2,:128], r3
+        vld1.32         {q1},     [r1,:128]!
+1:
+        pld             [r2, #-32]
+        vrev64.32       q0,  q0
+        vswp            d2,  d3
+        veor            q0,  q0,  q8
+        vld1.32         {q2},     [r2,:128], r3
+        vld1.32         {q3},     [r1,:128]!
+        vst2.32         {d1,d3},  [r0,:128]!
+        vst2.32         {d0,d2},  [r0,:128]!
+        pld             [r2, #-32]
+        vrev64.32       q2,  q2
+        vswp            d6,  d7
+        veor            q2,  q2,  q8
+        vld1.32         {q0},     [r2,:128], r3
+        vld1.32         {q1},     [r1,:128]!
+        vst2.32         {d5,d7},  [r0,:128]!
+        vst2.32         {d4,d6},  [r0,:128]!
+        subs            r12, r12, #8
+        bgt             1b
+        bx              lr
+endfunc
+
+function ff_sbr_qmf_deint_neg_neon, export=1
+        add             r1,  r1,  #60*4
+        add             r2,  r0,  #62*4
+        mov             r3,  #-16
+        mov             r12, #32
+        vmov.i32        d2,  #1<<31
+1:
+        vld2.32         {d0,d1},  [r1,:128], r3
+        veor            d0,  d0,  d2
+        vrev64.32       d1,  d1
+        vst1.32         {d0},     [r2,:64]
+        vst1.32         {d1},     [r0,:64]!
+        sub             r2,  r2,  #8
+        subs            r12, r12, #2
+        bgt             1b
+        bx              lr
+endfunc
+
+function ff_sbr_qmf_deint_bfly_neon, export=1
+        push            {lr}
+        add             r2,  r2,  #60*4
+        add             r3,  r0,  #124*4
+        mov             r12, #64
+        mov             lr,  #-16
+1:
+        vld1.32         {q0},     [r1,:128]!
+        vld1.32         {q1},     [r2,:128], lr
+        vrev64.32       q2,  q0
+        vrev64.32       q3,  q1
+        vadd.f32        d3,  d4,  d3
+        vadd.f32        d2,  d5,  d2
+        vsub.f32        d0,  d0,  d7
+        vsub.f32        d1,  d1,  d6
+        vst1.32         {q1},     [r3,:128], lr
+        vst1.32         {q0},     [r0,:128]!
+        subs            r12, r12, #4
+        bgt             1b
+        pop             {pc}
+endfunc
+
+function ff_sbr_hf_g_filt_neon, export=1
+        ldr             r12, [sp]
+        add             r1,  r1,  r12, lsl #3
+        mov             r12, #40*2*4
+        sub             r3,  r3,  #1
+        vld2.32         {d2[],d3[]},[r2,:64]!
+        vld1.32         {d0},     [r1,:64], r12
+1:
+        vld1.32         {d1},     [r1,:64], r12
+        vmul.f32        q3,  q0,  q1
+        vld2.32         {d2[],d3[]},[r2,:64]!
+        vld1.32         {d0},     [r1,:64], r12
+        vst1.32         {q3},     [r0,:64]!
+        subs            r3,  r3,  #2
+        bgt             1b
+        it              lt
+        bxlt            lr
+        vmul.f32        d0,  d0,  d2
+        vst1.32         {d0},     [r0,:64]!
+        bx              lr
+endfunc
+
+function ff_sbr_hf_gen_neon, export=1
+NOVFP   vld1.32         {d1[]},   [sp,:32]
+VFP     vdup.32         d1,  d0[0]
+        vmul.f32        d0,  d1,  d1
+        vld1.32         {d3},     [r2,:64]
+        vld1.32         {d2},     [r3,:64]
+        vmul.f32        q0,  q0,  q1
+        ldrd            r2,  r3,  [sp, #4*!HAVE_VFP_ARGS]
+        vtrn.32         d0,  d1
+        vneg.f32        d18, d1
+        vtrn.32         d18, d1
+        add             r0,  r0,  r2,  lsl #3
+        add             r1,  r1,  r2,  lsl #3
+        sub             r1,  r1,  #2*8
+        sub             r3,  r3,  r2
+        vld1.32         {q1},     [r1,:128]!
+1:
+        vld1.32         {q3},     [r1,:128]!
+        vrev64.32       q2,  q1
+        vmov            q8,  q3
+        vrev64.32       d20, d3
+        vrev64.32       d21, d6
+        vmla.f32        q3,  q1,  d0[0]
+        vmla.f32        d6,  d4,  d18
+        vmla.f32        d7,  d20, d18
+        vmla.f32        d6,  d3,  d0[1]
+        vmla.f32        d7,  d16, d0[1]
+        vmla.f32        d6,  d5,  d1
+        vmla.f32        d7,  d21, d1
+        vmov            q1,  q8
+        vst1.32         {q3},     [r0,:128]!
+        subs            r3,  r3,  #2
+        bgt             1b
+        bx              lr
+endfunc
+
+function ff_sbr_autocorrelate_neon, export=1
+        vld1.32         {q0},     [r0,:128]!
+        vmov.f32        q1,  #0.0
+        vmov.f32        q3,  #0.0
+        vmov.f32        d20, #0.0
+        vmul.f32        d21, d1,  d1
+        vmov            q8,  q0
+        vmov            q11, q0
+        mov             r12, #36
+1:
+        vld1.32         {q2},     [r0,:128]!
+        vrev64.32       q12, q2
+        vmla.f32        q10, q2,  q2
+        vmla.f32        d2,  d1,  d4
+        vmla.f32        d3,  d1,  d24
+        vmla.f32        d6,  d0,  d4
+        vmla.f32        d7,  d0,  d24
+        vmla.f32        d2,  d4,  d5
+        vmla.f32        d3,  d4,  d25
+        vmla.f32        d6,  d1,  d5
+        vmla.f32        d7,  d1,  d25
+        vmov            q0,  q2
+        subs            r12, r12, #2
+        bgt             1b
+        vld1.32         {q2},     [r0,:128]!
+        vrev64.32       q12, q2
+        vmla.f32        d2,  d1,  d4
+        vmla.f32        d3,  d1,  d24
+        vmla.f32        d6,  d0,  d4
+        vmla.f32        d7,  d0,  d24
+        vadd.f32        d20, d20, d21
+        vrev64.32       d18, d17
+        vmla.f32        d6,  d1,  d5
+        vmla.f32        d7,  d1,  d25
+        vmov            q0,  q1
+        vmla.f32        d0,  d16, d17
+        vmla.f32        d1,  d16, d18
+        vmla.f32        d2,  d4,  d5
+        vmla.f32        d3,  d4,  d25
+        vneg.f32        s15, s15
+        vmov            d21, d20
+        vpadd.f32       d0,  d0,  d2
+        vpadd.f32       d7,  d6,  d7
+        vtrn.32         d1,  d3
+        vsub.f32        d6,  d1,  d3
+        vmla.f32        d20, d22, d22
+        vmla.f32        d21, d4,  d4
+        vtrn.32         d0,  d6
+        vpadd.f32       d20, d20, d21
+        vst1.32         {q3},     [r1,:128]!
+        vst1.32         {d20[1]}, [r1,:32]
+        add             r1,  r1,  #2*4
+        vst1.32         {d0},     [r1,:64]
+        add             r1,  r1,  #4*4
+        vst1.32         {d20[0]}, [r1,:32]
+        bx              lr
+endfunc
+
+function ff_sbr_hf_apply_noise_0_neon, export=1
+        vmov.i32        d3,  #0
+.Lhf_apply_noise_0:
+        push            {r4,lr}
+        ldr             r12, [sp, #12]
+        movrel          r4,  X(ff_sbr_noise_table)
+        add             r3,  r3,  #1
+        bfc             r3,  #9,  #23
+        sub             r12, r12, #1
+1:
+        add             lr,  r4,  r3,  lsl #3
+        vld2.32         {q0},     [r0,:64]
+        vld2.32         {q3},     [lr,:64]
+        vld1.32         {d2},     [r1,:64]!
+        vld1.32         {d18},    [r2,:64]!
+        vceq.f32        d16, d2,  #0
+        veor            d2,  d2,  d3
+        vmov            q2,  q0
+        vmla.f32        d0,  d6,  d18
+        vmla.f32        d1,  d7,  d18
+        vadd.f32        d4,  d4,  d2
+        add             r3,  r3,  #2
+        bfc             r3,  #9,  #23
+        vbif            d0,  d4,  d16
+        vbif            d1,  d5,  d16
+        vst2.32         {q0},     [r0,:64]!
+        subs            r12, r12, #2
+        bgt             1b
+        blt             2f
+        add             lr,  r4,  r3,  lsl #3
+        vld1.32         {d0},     [r0,:64]
+        vld1.32         {d6},     [lr,:64]
+        vld1.32         {d2[]},   [r1,:32]!
+        vld1.32         {d3[]},   [r2,:32]!
+        vceq.f32        d4,  d2,  #0
+        veor            d2,  d2,  d3
+        vmov            d1,  d0
+        vmla.f32        d0,  d6,  d3
+        vadd.f32        s2,  s2,  s4
+        vbif            d0,  d1,  d4
+        vst1.32         {d0},     [r0,:64]!
+2:
+        pop             {r4,pc}
+endfunc
+
+function ff_sbr_hf_apply_noise_1_neon, export=1
+        ldr             r12, [sp]
+        push            {r4,lr}
+        lsl             r12, r12, #31
+        eor             lr,  r12, #1<<31
+        vmov            d3,  r12, lr
+.Lhf_apply_noise_1:
+        ldr             r12, [sp, #12]
+        movrel          r4,  X(ff_sbr_noise_table)
+        add             r3,  r3,  #1
+        bfc             r3,  #9,  #23
+        sub             r12, r12, #1
+1:
+        add             lr,  r4,  r3,  lsl #3
+        vld2.32         {q0},     [r0,:64]
+        vld2.32         {q3},     [lr,:64]
+        vld1.32         {d2},     [r1,:64]!
+        vld1.32         {d18},    [r2,:64]!
+        vceq.f32        d16, d2,  #0
+        veor            d2,  d2,  d3
+        vmov            q2,  q0
+        vmla.f32        d0,  d6,  d18
+        vmla.f32        d1,  d7,  d18
+        vadd.f32        d5,  d5,  d2
+        add             r3,  r3,  #2
+        bfc             r3,  #9,  #23
+        vbif            d0,  d4,  d16
+        vbif            d1,  d5,  d16
+        vst2.32         {q0},     [r0,:64]!
+        subs            r12, r12, #2
+        bgt             1b
+        blt             2f
+        add             lr,  r4,  r3,  lsl #3
+        vld1.32         {d0},     [r0,:64]
+        vld1.32         {d6},     [lr,:64]
+        vld1.32         {d2[]},   [r1,:32]!
+        vld1.32         {d18[]},  [r2,:32]!
+        vceq.f32        d4,  d2,  #0
+        veor            d2,  d2,  d3
+        vmov            d1,  d0
+        vmla.f32        d0,  d6,  d18
+        vadd.f32        s3,  s3,  s5
+        vbif            d0,  d1,  d4
+        vst1.32         {d0},     [r0,:64]!
+2:
+        pop             {r4,pc}
+endfunc
+
+function ff_sbr_hf_apply_noise_2_neon, export=1
+        vmov.i32        d3,  #1<<31
+        b               .Lhf_apply_noise_0
+endfunc
+
+function ff_sbr_hf_apply_noise_3_neon, export=1
+        ldr             r12, [sp]
+        push            {r4,lr}
+        lsl             r12, r12, #31
+        eor             lr,  r12, #1<<31
+        vmov            d3,  lr, r12
+        b               .Lhf_apply_noise_1
+endfunc
diff --git a/libavcodec/atrac.c b/libavcodec/atrac.c
index 6c1b1f19ce..1049df6281 100644
--- a/libavcodec/atrac.c
+++ b/libavcodec/atrac.c
@@ -117,4 +117,3 @@ void atrac_iqmf (float *inlo, float *inhi, unsigned int nIn, float *pOut, float
     /* Update the delay buffer. */
     memcpy(delayBuf, temp + nIn*2, 46*sizeof(float));
 }
-
diff --git a/libavcodec/aura.c b/libavcodec/aura.c
index e2c90b45f7..49694d55bf 100644
--- a/libavcodec/aura.c
+++ b/libavcodec/aura.c
@@ -134,4 +134,3 @@ AVCodec ff_aura2_decoder = {
     .capabilities   = CODEC_CAP_DR1,
     .long_name = NULL_IF_CONFIG_SMALL("Auravision Aura 2"),
 };
-
diff --git a/libavcodec/avcodec.h b/libavcodec/avcodec.h
index ca062eda91..4c5800b432 100644
--- a/libavcodec/avcodec.h
+++ b/libavcodec/avcodec.h
@@ -3566,6 +3566,9 @@ AVCodecContext *avcodec_alloc_context2(enum AVMediaType);
  * @param codec if non-NULL, allocate private data and initialize defaults
  *              for the given codec. It is illegal to then call avcodec_open2()
  *              with a different codec.
+ *              If NULL, then the codec-specific defaults won't be initialized,
+ *              which may result in suboptimal default settings (this is
+ *              important mainly for encoders, e.g. libx264).
  *
  * @return An AVCodecContext filled with default values or NULL on failure.
  * @see avcodec_get_context_defaults
diff --git a/libavcodec/bitstream.c b/libavcodec/bitstream.c
index 30aaf75fea..e1336438a0 100644
--- a/libavcodec/bitstream.c
+++ b/libavcodec/bitstream.c
@@ -322,4 +322,3 @@ void free_vlc(VLC *vlc)
 {
     av_freep(&vlc->table);
 }
-
diff --git a/libavcodec/cscd.c b/libavcodec/cscd.c
index 1e157edcbc..4f9b70c88b 100644
--- a/libavcodec/cscd.c
+++ b/libavcodec/cscd.c
@@ -266,4 +266,3 @@ AVCodec ff_cscd_decoder = {
     .capabilities   = CODEC_CAP_DR1,
     .long_name = NULL_IF_CONFIG_SMALL("CamStudio"),
 };
-
diff --git a/libavcodec/dvdata.c b/libavcodec/dvdata.c
index f292d4bf19..65cbff4b1c 100644
--- a/libavcodec/dvdata.c
+++ b/libavcodec/dvdata.c
@@ -294,4 +294,3 @@ const DVprofile* avpriv_dv_codec_profile(AVCodecContext* codec)
 
     return NULL;
 }
-
diff --git a/libavcodec/dxa.c b/libavcodec/dxa.c
index f32fdd33e5..28c6af9031 100644
--- a/libavcodec/dxa.c
+++ b/libavcodec/dxa.c
@@ -331,4 +331,3 @@ AVCodec ff_dxa_decoder = {
     .capabilities   = CODEC_CAP_DR1,
     .long_name = NULL_IF_CONFIG_SMALL("Feeble Files/ScummVM DXA"),
 };
-
diff --git a/libavcodec/dxva2.c b/libavcodec/dxva2.c
index b6f8aea429..9b94b58fbc 100644
--- a/libavcodec/dxva2.c
+++ b/libavcodec/dxva2.c
@@ -151,4 +151,3 @@ end:
         ff_draw_horiz_band(s, 0, s->avctx->height);
     return result;
 }
-
diff --git a/libavcodec/dxva2_h264.c b/libavcodec/dxva2_h264.c
index af91e02188..3bda097cc4 100644
--- a/libavcodec/dxva2_h264.c
+++ b/libavcodec/dxva2_h264.c
@@ -448,4 +448,3 @@ AVHWAccel ff_h264_dxva2_hwaccel = {
     .end_frame      = end_frame,
     .priv_data_size = sizeof(struct dxva2_picture_context),
 };
-
diff --git a/libavcodec/dxva2_mpeg2.c b/libavcodec/dxva2_mpeg2.c
index 951305d00f..e54eff09fb 100644
--- a/libavcodec/dxva2_mpeg2.c
+++ b/libavcodec/dxva2_mpeg2.c
@@ -270,4 +270,3 @@ AVHWAccel ff_mpeg2_dxva2_hwaccel = {
     .end_frame      = end_frame,
     .priv_data_size = sizeof(struct dxva2_picture_context),
 };
-
diff --git a/libavcodec/dxva2_vc1.c b/libavcodec/dxva2_vc1.c
index 0ae9685208..50da518d86 100644
--- a/libavcodec/dxva2_vc1.c
+++ b/libavcodec/dxva2_vc1.c
@@ -286,4 +286,3 @@ AVHWAccel ff_vc1_dxva2_hwaccel = {
     .end_frame      = end_frame,
     .priv_data_size = sizeof(struct dxva2_picture_context),
 };
-
diff --git a/libavcodec/escape124.c b/libavcodec/escape124.c
index c9025fddde..fd76e59fb2 100644
--- a/libavcodec/escape124.c
+++ b/libavcodec/escape124.c
@@ -376,4 +376,3 @@ AVCodec ff_escape124_decoder = {
     .capabilities   = CODEC_CAP_DR1,
     .long_name = NULL_IF_CONFIG_SMALL("Escape 124"),
 };
-
diff --git a/libavcodec/fft.c b/libavcodec/fft.c
index 4eb46f14c8..6b93a5cdf3 100644
--- a/libavcodec/fft.c
+++ b/libavcodec/fft.c
@@ -350,4 +350,3 @@ static void ff_fft_calc_c(FFTContext *s, FFTComplex *z)
 {
     fft_dispatch[s->nbits-2](z);
 }
-
diff --git a/libavcodec/flashsvenc.c b/libavcodec/flashsvenc.c
index 7e21e7d534..77290e866f 100644
--- a/libavcodec/flashsvenc.c
+++ b/libavcodec/flashsvenc.c
@@ -286,4 +286,3 @@ AVCodec ff_flashsv_encoder = {
     .pix_fmts       = (const enum PixelFormat[]){PIX_FMT_BGR24, PIX_FMT_NONE},
     .long_name      = NULL_IF_CONFIG_SMALL("Flash Screen Video"),
 };
-
diff --git a/libavcodec/h261.c b/libavcodec/h261.c
index 562a151e90..951997d65e 100644
--- a/libavcodec/h261.c
+++ b/libavcodec/h261.c
@@ -51,4 +51,3 @@ void ff_h261_loop_filter(MpegEncContext *s){
     s->dsp.h261_loop_filter(dest_cb, uvlinesize);
     s->dsp.h261_loop_filter(dest_cr, uvlinesize);
 }
-
diff --git a/libavcodec/h261enc.c b/libavcodec/h261enc.c
index 0110addf73..4385aee11b 100644
--- a/libavcodec/h261enc.c
+++ b/libavcodec/h261enc.c
@@ -332,4 +332,3 @@ AVCodec ff_h261_encoder = {
     .pix_fmts= (const enum PixelFormat[]){PIX_FMT_YUV420P, PIX_FMT_NONE},
     .long_name= NULL_IF_CONFIG_SMALL("H.261"),
 };
-
diff --git a/libavcodec/h264_cavlc.c b/libavcodec/h264_cavlc.c
index f9daeb29d9..a782f14698 100644
--- a/libavcodec/h264_cavlc.c
+++ b/libavcodec/h264_cavlc.c
@@ -1169,4 +1169,3 @@ decode_intra_mb:
 
     return 0;
 }
-
diff --git a/libavcodec/h264_mp4toannexb_bsf.c b/libavcodec/h264_mp4toannexb_bsf.c
index 5085ecbdfd..48e085e7c3 100644
--- a/libavcodec/h264_mp4toannexb_bsf.c
+++ b/libavcodec/h264_mp4toannexb_bsf.c
@@ -188,4 +188,3 @@ AVBitStreamFilter ff_h264_mp4toannexb_bsf = {
     sizeof(H264BSFContext),
     h264_mp4toannexb_filter,
 };
-
diff --git a/libavcodec/idcinvideo.c b/libavcodec/idcinvideo.c
index 0b22043365..7a7f76a8a0 100644
--- a/libavcodec/idcinvideo.c
+++ b/libavcodec/idcinvideo.c
@@ -265,4 +265,3 @@ AVCodec ff_idcin_decoder = {
     .capabilities   = CODEC_CAP_DR1,
     .long_name = NULL_IF_CONFIG_SMALL("id Quake II CIN video"),
 };
-
diff --git a/libavcodec/imgconvert.c b/libavcodec/imgconvert.c
index fac5f52107..9fcd94c7a8 100644
--- a/libavcodec/imgconvert.c
+++ b/libavcodec/imgconvert.c
@@ -904,4 +904,3 @@ int avpicture_deinterlace(AVPicture *dst, const AVPicture *src,
     emms_c();
     return 0;
 }
-
diff --git a/libavcodec/intelh263dec.c b/libavcodec/intelh263dec.c
index 836e98ee88..c97c1d7a1f 100644
--- a/libavcodec/intelh263dec.c
+++ b/libavcodec/intelh263dec.c
@@ -136,4 +136,3 @@ AVCodec ff_h263i_decoder = {
     .long_name = NULL_IF_CONFIG_SMALL("Intel H.263"),
     .pix_fmts= ff_pixfmt_list_420,
 };
-
diff --git a/libavcodec/kbdwin.c b/libavcodec/kbdwin.c
index 8b33861f71..aa4b9726f5 100644
--- a/libavcodec/kbdwin.c
+++ b/libavcodec/kbdwin.c
@@ -45,4 +45,3 @@ av_cold void ff_kbd_window_init(float *window, float alpha, int n)
    for (i = 0; i < n; i++)
        window[i] = sqrt(local_window[i] / sum);
 }
-
diff --git a/libavcodec/libvo-aacenc.c b/libavcodec/libvo-aacenc.c
index dcec664347..25e8a317f7 100644
--- a/libavcodec/libvo-aacenc.c
+++ b/libavcodec/libvo-aacenc.c
@@ -126,4 +126,3 @@ AVCodec ff_libvo_aacenc_encoder = {
     .sample_fmts = (const enum AVSampleFormat[]){AV_SAMPLE_FMT_S16,AV_SAMPLE_FMT_NONE},
     .long_name = NULL_IF_CONFIG_SMALL("Android VisualOn AAC"),
 };
-
diff --git a/libavcodec/libvo-amrwbenc.c b/libavcodec/libvo-amrwbenc.c
index b32fe4e1df..22c71b75f1 100644
--- a/libavcodec/libvo-amrwbenc.c
+++ b/libavcodec/libvo-amrwbenc.c
@@ -130,4 +130,3 @@ AVCodec ff_libvo_amrwbenc_encoder = {
                                       "(AMR) Wide-Band"),
     .priv_class = &class,
 };
-
diff --git a/libavcodec/libxavs.c b/libavcodec/libxavs.c
index acc674dc1a..d7027d1092 100644
--- a/libavcodec/libxavs.c
+++ b/libavcodec/libxavs.c
@@ -384,4 +384,3 @@ AVCodec ff_libxavs_encoder = {
     .priv_class     = &class,
     .defaults       = xavs_defaults,
 };
-
diff --git a/libavcodec/libxvid_rc.c b/libavcodec/libxvid_rc.c
index 8a2b487122..10a7c1cb3e 100644
--- a/libavcodec/libxvid_rc.c
+++ b/libavcodec/libxvid_rc.c
@@ -146,4 +146,3 @@ void ff_xvid_rate_control_uninit(MpegEncContext *s){
 
     xvid_plugin_2pass2(s->rc_context.non_lavc_opaque, XVID_PLG_DESTROY, &xvid_plg_destroy, NULL);
 }
-
diff --git a/libavcodec/mace.c b/libavcodec/mace.c
index ffa11ad80d..594ef6c795 100644
--- a/libavcodec/mace.c
+++ b/libavcodec/mace.c
@@ -310,4 +310,3 @@ AVCodec ff_mace6_decoder = {
     .capabilities   = CODEC_CAP_DR1,
     .long_name = NULL_IF_CONFIG_SMALL("MACE (Macintosh Audio Compression/Expansion) 6:1"),
 };
-
diff --git a/libavcodec/mathops.h b/libavcodec/mathops.h
index 1126ba6043..3ef7f8b660 100644
--- a/libavcodec/mathops.h
+++ b/libavcodec/mathops.h
@@ -179,4 +179,3 @@ if ((y) < (x)) {\
 #endif
 
 #endif /* AVCODEC_MATHOPS_H */
-
diff --git a/libavcodec/mdec.c b/libavcodec/mdec.c
index cf606935a6..2828417bcd 100644
--- a/libavcodec/mdec.c
+++ b/libavcodec/mdec.c
@@ -278,4 +278,3 @@ AVCodec ff_mdec_decoder = {
     .long_name= NULL_IF_CONFIG_SMALL("Sony PlayStation MDEC (Motion DECoder)"),
     .init_thread_copy= ONLY_IF_THREADS_ENABLED(decode_init_thread_copy)
 };
-
diff --git a/libavcodec/mlp_parser.h b/libavcodec/mlp_parser.h
index 3b81763d57..6aafed5086 100644
--- a/libavcodec/mlp_parser.h
+++ b/libavcodec/mlp_parser.h
@@ -59,4 +59,3 @@ uint64_t ff_truehd_layout(int chanmap);
 extern const uint64_t ff_mlp_layout[32];
 
 #endif /* AVCODEC_MLP_PARSER_H */
-
diff --git a/libavcodec/mpeg12.c b/libavcodec/mpeg12.c
index d34b54d76c..71c9a5893b 100644
--- a/libavcodec/mpeg12.c
+++ b/libavcodec/mpeg12.c
@@ -2668,4 +2668,3 @@ AVCodec ff_mpeg1_vdpau_decoder = {
     .long_name      = NULL_IF_CONFIG_SMALL("MPEG-1 video (VDPAU acceleration)"),
 };
 #endif
-
diff --git a/libavcodec/mpeg4video.c b/libavcodec/mpeg4video.c
index 9a093511d5..7444f26a9d 100644
--- a/libavcodec/mpeg4video.c
+++ b/libavcodec/mpeg4video.c
@@ -168,4 +168,3 @@ int ff_mpeg4_set_direct_mv(MpegEncContext *s, int mx, int my){
         return MB_TYPE_DIRECT2 | MB_TYPE_16x16 | MB_TYPE_L0L1; //Note see prev line
     }
 }
-
diff --git a/libavcodec/msrledec.c b/libavcodec/msrledec.c
index 129f0e0bc0..c0e7682689 100644
--- a/libavcodec/msrledec.c
+++ b/libavcodec/msrledec.c
@@ -260,4 +260,3 @@ int ff_msrle_decode(AVCodecContext *avctx, AVPicture *pic, int depth,
         return -1;
     }
 }
-
diff --git a/libavcodec/msrledec.h b/libavcodec/msrledec.h
index 2230162691..b45ea48bc1 100644
--- a/libavcodec/msrledec.h
+++ b/libavcodec/msrledec.h
@@ -37,4 +37,3 @@ int ff_msrle_decode(AVCodecContext *avctx, AVPicture *pic, int depth,
                     const uint8_t* data, int data_size);
 
 #endif /* AVCODEC_MSRLEDEC_H */
-
diff --git a/libavcodec/nellymoser.c b/libavcodec/nellymoser.c
index cbcc4f941b..195e2e80ae 100644
--- a/libavcodec/nellymoser.c
+++ b/libavcodec/nellymoser.c
@@ -223,4 +223,3 @@ void ff_nelly_get_sample_bits(const float *buf, int *bits)
             bits[i] = 0;
     }
 }
-
diff --git a/libavcodec/nellymoserdec.c b/libavcodec/nellymoserdec.c
index 2a1ec5b4cd..de26dc6a41 100644
--- a/libavcodec/nellymoserdec.c
+++ b/libavcodec/nellymoserdec.c
@@ -232,4 +232,3 @@ AVCodec ff_nellymoser_decoder = {
                                                       AV_SAMPLE_FMT_S16,
                                                       AV_SAMPLE_FMT_NONE },
 };
-
diff --git a/libavcodec/nuv.c b/libavcodec/nuv.c
index 7f8dc7501b..c82034a96b 100644
--- a/libavcodec/nuv.c
+++ b/libavcodec/nuv.c
@@ -297,4 +297,3 @@ AVCodec ff_nuv_decoder = {
     .capabilities   = CODEC_CAP_DR1,
     .long_name = NULL_IF_CONFIG_SMALL("NuppelVideo/RTJPEG"),
 };
-
diff --git a/libavcodec/pcm_tablegen.c b/libavcodec/pcm_tablegen.c
index b0fde936c7..bf8e7fb707 100644
--- a/libavcodec/pcm_tablegen.c
+++ b/libavcodec/pcm_tablegen.c
@@ -37,4 +37,3 @@ int main(void)
 
     return 0;
 }
-
diff --git a/libavcodec/psymodel.c b/libavcodec/psymodel.c
index b3cfb0e75f..a17aecc422 100644
--- a/libavcodec/psymodel.c
+++ b/libavcodec/psymodel.c
@@ -134,4 +134,3 @@ av_cold void ff_psy_preprocess_end(struct FFPsyPreprocessContext *ctx)
     av_freep(&ctx->fstate);
     av_free(ctx);
 }
-
diff --git a/libavcodec/qtrle.c b/libavcodec/qtrle.c
index 4ef529d5b2..606a2ea95d 100644
--- a/libavcodec/qtrle.c
+++ b/libavcodec/qtrle.c
@@ -565,4 +565,3 @@ AVCodec ff_qtrle_decoder = {
     .capabilities   = CODEC_CAP_DR1,
     .long_name = NULL_IF_CONFIG_SMALL("QuickTime Animation (RLE) video"),
 };
-
diff --git a/libavcodec/ratecontrol.h b/libavcodec/ratecontrol.h
index 32efe01d24..b4132bfcb6 100644
--- a/libavcodec/ratecontrol.h
+++ b/libavcodec/ratecontrol.h
@@ -102,4 +102,3 @@ void ff_xvid_rate_control_uninit(struct MpegEncContext *s);
 float ff_xvid_rate_estimate_qscale(struct MpegEncContext *s, int dry_run);
 
 #endif /* AVCODEC_RATECONTROL_H */
-
diff --git a/libavcodec/rl2.c b/libavcodec/rl2.c
index 07c6109c64..7eebddff6f 100644
--- a/libavcodec/rl2.c
+++ b/libavcodec/rl2.c
@@ -230,4 +230,3 @@ AVCodec ff_rl2_decoder = {
     .capabilities   = CODEC_CAP_DR1,
     .long_name = NULL_IF_CONFIG_SMALL("RL2 video"),
 };
-
diff --git a/libavcodec/sbr.h b/libavcodec/sbr.h
index 69c847ac15..1fe47fcebe 100644
--- a/libavcodec/sbr.h
+++ b/libavcodec/sbr.h
@@ -32,6 +32,7 @@
 #include <stdint.h>
 #include "fft.h"
 #include "aacps.h"
+#include "sbrdsp.h"
 
 /**
  * Spectral Band Replication header - spectrum parameters that invoke a reset if they differ from the previous header.
@@ -87,8 +88,8 @@ typedef struct {
     ///QMF values of the original signal
     float              W[2][32][32][2];
     ///QMF output of the HF adjustor
-    float              Y[2][38][64][2];
-    float              g_temp[42][48];
+    DECLARE_ALIGNED(16, float, Y)[2][38][64][2];
+    DECLARE_ALIGNED(16, float, g_temp)[42][48];
     float              q_temp[42][48];
     uint8_t            s_indexmapped[8][48];
     ///Envelope scalefactors
@@ -155,15 +156,15 @@ typedef struct {
     uint8_t            patch_num_subbands[6];
     uint8_t            patch_start_subband[6];
     ///QMF low frequency input to the HF generator
-    float              X_low[32][40][2];
+    DECLARE_ALIGNED(16, float, X_low)[32][40][2];
     ///QMF output of the HF generator
-    float              X_high[64][40][2];
+    DECLARE_ALIGNED(16, float, X_high)[64][40][2];
     ///QMF values of the reconstructed signal
     DECLARE_ALIGNED(16, float, X)[2][2][38][64];
     ///Zeroth coefficient used to filter the subband signals
-    float              alpha0[64][2];
+    DECLARE_ALIGNED(16, float, alpha0)[64][2];
     ///First coefficient used to filter the subband signals
-    float              alpha1[64][2];
+    DECLARE_ALIGNED(16, float, alpha1)[64][2];
     ///Dequantized envelope scalefactors, remapped
     float              e_origmapped[7][48];
     ///Dequantized noise scalefactors, remapped
@@ -180,6 +181,7 @@ typedef struct {
     DECLARE_ALIGNED(16, float, qmf_filter_scratch)[5][64];
     FFTContext         mdct_ana;
     FFTContext         mdct;
+    SBRDSPContext      dsp;
 } SpectralBandReplication;
 
 #endif /* AVCODEC_SBR_H */
diff --git a/libavcodec/sbrdsp.c b/libavcodec/sbrdsp.c
new file mode 100644
index 0000000000..2711e71338
--- /dev/null
+++ b/libavcodec/sbrdsp.c
@@ -0,0 +1,241 @@
+/*
+ * AAC Spectral Band Replication decoding functions
+ * Copyright (c) 2008-2009 Robert Swain ( rob opendot cl )
+ * Copyright (c) 2009-2010 Alex Converse <alex.converse@gmail.com>
+ *
+ * This file is part of Libav.
+ *
+ * Libav 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.
+ *
+ * Libav 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 Libav; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#include "config.h"
+#include "libavutil/attributes.h"
+#include "sbrdsp.h"
+
+static void sbr_sum64x5_c(float *z)
+{
+    int k;
+    for (k = 0; k < 64; k++) {
+        float f = z[k] + z[k + 64] + z[k + 128] + z[k + 192] + z[k + 256];
+        z[k] = f;
+    }
+}
+
+static float sbr_sum_square_c(float (*x)[2], int n)
+{
+    float sum = 0.0f;
+    int i;
+
+    for (i = 0; i < n; i++)
+        sum += x[i][0] * x[i][0] + x[i][1] * x[i][1];
+
+    return sum;
+}
+
+static void sbr_neg_odd_64_c(float *x)
+{
+    int i;
+    for (i = 1; i < 64; i += 2)
+        x[i] = -x[i];
+}
+
+static void sbr_qmf_pre_shuffle_c(float *z)
+{
+    int k;
+    z[64] = z[0];
+    z[65] = z[1];
+    for (k = 1; k < 32; k++) {
+        z[64+2*k  ] = -z[64 - k];
+        z[64+2*k+1] =  z[ k + 1];
+    }
+}
+
+static void sbr_qmf_post_shuffle_c(float W[32][2], const float *z)
+{
+    int k;
+    for (k = 0; k < 32; k++) {
+        W[k][0] = -z[63-k];
+        W[k][1] = z[k];
+    }
+}
+
+static void sbr_qmf_deint_neg_c(float *v, const float *src)
+{
+    int i;
+    for (i = 0; i < 32; i++) {
+        v[     i] =  src[63 - 2*i    ];
+        v[63 - i] = -src[63 - 2*i - 1];
+    }
+}
+
+static void sbr_qmf_deint_bfly_c(float *v, const float *src0, const float *src1)
+{
+    int i;
+    for (i = 0; i < 64; i++) {
+        v[      i] = src0[i] - src1[63 - i];
+        v[127 - i] = src0[i] + src1[63 - i];
+    }
+}
+
+static av_always_inline void autocorrelate(const float x[40][2],
+                                           float phi[3][2][2], int lag)
+{
+    int i;
+    float real_sum = 0.0f;
+    float imag_sum = 0.0f;
+    if (lag) {
+        for (i = 1; i < 38; i++) {
+            real_sum += x[i][0] * x[i+lag][0] + x[i][1] * x[i+lag][1];
+            imag_sum += x[i][0] * x[i+lag][1] - x[i][1] * x[i+lag][0];
+        }
+        phi[2-lag][1][0] = real_sum + x[ 0][0] * x[lag][0] + x[ 0][1] * x[lag][1];
+        phi[2-lag][1][1] = imag_sum + x[ 0][0] * x[lag][1] - x[ 0][1] * x[lag][0];
+        if (lag == 1) {
+            phi[0][0][0] = real_sum + x[38][0] * x[39][0] + x[38][1] * x[39][1];
+            phi[0][0][1] = imag_sum + x[38][0] * x[39][1] - x[38][1] * x[39][0];
+        }
+    } else {
+        for (i = 1; i < 38; i++) {
+            real_sum += x[i][0] * x[i][0] + x[i][1] * x[i][1];
+        }
+        phi[2][1][0] = real_sum + x[ 0][0] * x[ 0][0] + x[ 0][1] * x[ 0][1];
+        phi[1][0][0] = real_sum + x[38][0] * x[38][0] + x[38][1] * x[38][1];
+    }
+}
+
+static void sbr_autocorrelate_c(const float x[40][2], float phi[3][2][2])
+{
+    autocorrelate(x, phi, 0);
+    autocorrelate(x, phi, 1);
+    autocorrelate(x, phi, 2);
+}
+
+static void sbr_hf_gen_c(float (*X_high)[2], const float (*X_low)[2],
+                         const float alpha0[2], const float alpha1[2],
+                         float bw, int start, int end)
+{
+    float alpha[4];
+    int i;
+
+    alpha[0] = alpha1[0] * bw * bw;
+    alpha[1] = alpha1[1] * bw * bw;
+    alpha[2] = alpha0[0] * bw;
+    alpha[3] = alpha0[1] * bw;
+
+    for (i = start; i < end; i++) {
+        X_high[i][0] =
+            X_low[i - 2][0] * alpha[0] -
+            X_low[i - 2][1] * alpha[1] +
+            X_low[i - 1][0] * alpha[2] -
+            X_low[i - 1][1] * alpha[3] +
+            X_low[i][0];
+        X_high[i][1] =
+            X_low[i - 2][1] * alpha[0] +
+            X_low[i - 2][0] * alpha[1] +
+            X_low[i - 1][1] * alpha[2] +
+            X_low[i - 1][0] * alpha[3] +
+            X_low[i][1];
+    }
+}
+
+static void sbr_hf_g_filt_c(float (*Y)[2], const float (*X_high)[40][2],
+                            const float *g_filt, int m_max, int ixh)
+{
+    int m;
+
+    for (m = 0; m < m_max; m++) {
+        Y[m][0] = X_high[m][ixh][0] * g_filt[m];
+        Y[m][1] = X_high[m][ixh][1] * g_filt[m];
+    }
+}
+
+static av_always_inline void sbr_hf_apply_noise(float (*Y)[2],
+                                                const float *s_m,
+                                                const float *q_filt,
+                                                int noise,
+                                                float phi_sign0,
+                                                float phi_sign1,
+                                                int m_max)
+{
+    int m;
+
+    for (m = 0; m < m_max; m++) {
+        float y0 = Y[m][0];
+        float y1 = Y[m][1];
+        noise = (noise + 1) & 0x1ff;
+        if (s_m[m]) {
+            y0 += s_m[m] * phi_sign0;
+            y1 += s_m[m] * phi_sign1;
+        } else {
+            y0 += q_filt[m] * ff_sbr_noise_table[noise][0];
+            y1 += q_filt[m] * ff_sbr_noise_table[noise][1];
+        }
+        Y[m][0] = y0;
+        Y[m][1] = y1;
+        phi_sign1 = -phi_sign1;
+    }
+}
+
+static void sbr_hf_apply_noise_0(float (*Y)[2], const float *s_m,
+                                 const float *q_filt, int noise,
+                                 int kx, int m_max)
+{
+    sbr_hf_apply_noise(Y, s_m, q_filt, noise, 1.0, 0.0, m_max);
+}
+
+static void sbr_hf_apply_noise_1(float (*Y)[2], const float *s_m,
+                                 const float *q_filt, int noise,
+                                 int kx, int m_max)
+{
+    float phi_sign = 1 - 2 * (kx & 1);
+    sbr_hf_apply_noise(Y, s_m, q_filt, noise, 0.0, phi_sign, m_max);
+}
+
+static void sbr_hf_apply_noise_2(float (*Y)[2], const float *s_m,
+                                 const float *q_filt, int noise,
+                                 int kx, int m_max)
+{
+    sbr_hf_apply_noise(Y, s_m, q_filt, noise, -1.0, 0.0, m_max);
+}
+
+static void sbr_hf_apply_noise_3(float (*Y)[2], const float *s_m,
+                                 const float *q_filt, int noise,
+                                 int kx, int m_max)
+{
+    float phi_sign = 1 - 2 * (kx & 1);
+    sbr_hf_apply_noise(Y, s_m, q_filt, noise, 0.0, -phi_sign, m_max);
+}
+
+av_cold void ff_sbrdsp_init(SBRDSPContext *s)
+{
+    s->sum64x5 = sbr_sum64x5_c;
+    s->sum_square = sbr_sum_square_c;
+    s->neg_odd_64 = sbr_neg_odd_64_c;
+    s->qmf_pre_shuffle = sbr_qmf_pre_shuffle_c;
+    s->qmf_post_shuffle = sbr_qmf_post_shuffle_c;
+    s->qmf_deint_neg = sbr_qmf_deint_neg_c;
+    s->qmf_deint_bfly = sbr_qmf_deint_bfly_c;
+    s->autocorrelate = sbr_autocorrelate_c;
+    s->hf_gen = sbr_hf_gen_c;
+    s->hf_g_filt = sbr_hf_g_filt_c;
+
+    s->hf_apply_noise[0] = sbr_hf_apply_noise_0;
+    s->hf_apply_noise[1] = sbr_hf_apply_noise_1;
+    s->hf_apply_noise[2] = sbr_hf_apply_noise_2;
+    s->hf_apply_noise[3] = sbr_hf_apply_noise_3;
+
+    if (ARCH_ARM)
+        ff_sbrdsp_init_arm(s);
+}
diff --git a/libavcodec/sbrdsp.h b/libavcodec/sbrdsp.h
new file mode 100644
index 0000000000..88285b07ec
--- /dev/null
+++ b/libavcodec/sbrdsp.h
@@ -0,0 +1,48 @@
+/*
+ * Copyright (c) 2012 Mans Rullgard
+ *
+ * This file is part of Libav.
+ *
+ * Libav 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.
+ *
+ * Libav 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 Libav; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#ifndef LIBAVCODEC_SBRDSP_H
+#define LIBAVCODEC_SBRDSP_H
+
+typedef struct SBRDSPContext {
+    void (*sum64x5)(float *z);
+    float (*sum_square)(float (*x)[2], int n);
+    void (*neg_odd_64)(float *x);
+    void (*qmf_pre_shuffle)(float *z);
+    void (*qmf_post_shuffle)(float W[32][2], const float *z);
+    void (*qmf_deint_neg)(float *v, const float *src);
+    void (*qmf_deint_bfly)(float *v, const float *src0, const float *src1);
+    void (*autocorrelate)(const float x[40][2], float phi[3][2][2]);
+    void (*hf_gen)(float (*X_high)[2], const float (*X_low)[2],
+                   const float alpha0[2], const float alpha1[2],
+                   float bw, int start, int end);
+    void (*hf_g_filt)(float (*Y)[2], const float (*X_high)[40][2],
+                      const float *g_filt, int m_max, int ixh);
+    void (*hf_apply_noise[4])(float (*Y)[2], const float *s_m,
+                              const float *q_filt, int noise,
+                              int kx, int m_max);
+} SBRDSPContext;
+
+extern const float ff_sbr_noise_table[][2];
+
+void ff_sbrdsp_init(SBRDSPContext *s);
+void ff_sbrdsp_init_arm(SBRDSPContext *s);
+
+#endif
diff --git a/libavcodec/sgidec.c b/libavcodec/sgidec.c
index 6a98718131..56495c1d31 100644
--- a/libavcodec/sgidec.c
+++ b/libavcodec/sgidec.c
@@ -269,4 +269,3 @@ AVCodec ff_sgi_decoder = {
     .decode         = decode_frame,
     .long_name = NULL_IF_CONFIG_SMALL("SGI image"),
 };
-
diff --git a/libavcodec/smacker.c b/libavcodec/smacker.c
index a60a2b051b..593d409cb1 100644
--- a/libavcodec/smacker.c
+++ b/libavcodec/smacker.c
@@ -762,4 +762,3 @@ AVCodec ff_smackaud_decoder = {
     .capabilities   = CODEC_CAP_DR1,
     .long_name = NULL_IF_CONFIG_SMALL("Smacker audio"),
 };
-
diff --git a/libavcodec/snow.c b/libavcodec/snow.c
index a289ecfbce..88620dd28c 100644
--- a/libavcodec/snow.c
+++ b/libavcodec/snow.c
@@ -650,4 +650,3 @@ av_cold void ff_snow_common_end(SnowContext *s)
     if (s->current_picture.data[0])
         s->avctx->release_buffer(s->avctx, &s->current_picture);
 }
-
diff --git a/libavcodec/tscc.c b/libavcodec/tscc.c
index c07f3a87f7..e6c9492edc 100644
--- a/libavcodec/tscc.c
+++ b/libavcodec/tscc.c
@@ -210,4 +210,3 @@ AVCodec ff_tscc_decoder = {
     .capabilities   = CODEC_CAP_DR1,
     .long_name      = NULL_IF_CONFIG_SMALL("TechSmith Screen Capture Codec"),
 };
-
diff --git a/libavcodec/ulti.c b/libavcodec/ulti.c
index 6e7ba35f9d..cd6f2d24ab 100644
--- a/libavcodec/ulti.c
+++ b/libavcodec/ulti.c
@@ -429,4 +429,3 @@ AVCodec ff_ulti_decoder = {
     .capabilities   = CODEC_CAP_DR1,
     .long_name = NULL_IF_CONFIG_SMALL("IBM UltiMotion"),
 };
-
diff --git a/libavcodec/utils.c b/libavcodec/utils.c
index dec9003e1d..a41c8a5596 100644
--- a/libavcodec/utils.c
+++ b/libavcodec/utils.c
@@ -242,7 +242,7 @@ void avcodec_align_dimensions2(AVCodecContext *s, int *width, int *height,
         *height+=2; // some of the optimized chroma MC reads one line too much
                     // which is also done in mpeg decoders with lowres > 0
 
-    for (i = 0; i < AV_NUM_DATA_POINTERS; i++)
+    for (i = 0; i < 4; i++)
         linesize_align[i] = STRIDE_ALIGN;
 //STRIDE_ALIGN is 8 for SSE* but this does not work for SVQ1 chroma planes
 //we could change STRIDE_ALIGN to 16 for x86/sse but it would increase the
@@ -252,7 +252,7 @@ void avcodec_align_dimensions2(AVCodecContext *s, int *width, int *height,
     if(s->codec_id == CODEC_ID_SVQ1 || s->codec_id == CODEC_ID_VP5 ||
        s->codec_id == CODEC_ID_VP6 || s->codec_id == CODEC_ID_VP6F ||
        s->codec_id == CODEC_ID_VP6A || s->codec_id == CODEC_ID_DIRAC) {
-        for (i = 0; i < AV_NUM_DATA_POINTERS; i++)
+        for (i = 0; i < 4; i++)
             linesize_align[i] = 16;
     }
 #endif
diff --git a/libavcodec/utvideo.c b/libavcodec/utvideo.c
index a45f13fa54..ed11776a36 100644
--- a/libavcodec/utvideo.c
+++ b/libavcodec/utvideo.c
@@ -578,4 +578,3 @@ AVCodec ff_utvideo_decoder = {
     .capabilities   = CODEC_CAP_DR1 | CODEC_CAP_FRAME_THREADS,
     .long_name      = NULL_IF_CONFIG_SMALL("Ut Video"),
 };
-
diff --git a/libavcodec/vb.c b/libavcodec/vb.c
index 50e5d0b1a4..bcb26452b0 100644
--- a/libavcodec/vb.c
+++ b/libavcodec/vb.c
@@ -281,4 +281,3 @@ AVCodec ff_vb_decoder = {
     .decode         = decode_frame,
     .long_name = NULL_IF_CONFIG_SMALL("Beam Software VB"),
 };
-
diff --git a/libavcodec/vmnc.c b/libavcodec/vmnc.c
index 7655103b6a..21141fa052 100644
--- a/libavcodec/vmnc.c
+++ b/libavcodec/vmnc.c
@@ -520,4 +520,3 @@ AVCodec ff_vmnc_decoder = {
     .capabilities   = CODEC_CAP_DR1,
     .long_name = NULL_IF_CONFIG_SMALL("VMware Screen Codec / VMware Video"),
 };
-
diff --git a/libavcodec/vorbisdec.c b/libavcodec/vorbisdec.c
index 03e3d757cb..f1d9a79297 100644
--- a/libavcodec/vorbisdec.c
+++ b/libavcodec/vorbisdec.c
@@ -1725,4 +1725,3 @@ AVCodec ff_vorbis_decoder = {
         AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_NONE
     },
 };
-
diff --git a/libavcodec/wma.c b/libavcodec/wma.c
index feb121b591..562abf3812 100644
--- a/libavcodec/wma.c
+++ b/libavcodec/wma.c
@@ -525,4 +525,3 @@ int ff_wma_run_level_decode(AVCodecContext* avctx, GetBitContext* gb,
 
     return 0;
 }
-
diff --git a/libavcodec/xxan.c b/libavcodec/xxan.c
index 0bfebc8789..de55d082ae 100644
--- a/libavcodec/xxan.c
+++ b/libavcodec/xxan.c
@@ -432,4 +432,3 @@ AVCodec ff_xan_wc4_decoder = {
     .capabilities   = CODEC_CAP_DR1,
     .long_name = NULL_IF_CONFIG_SMALL("Wing Commander IV / Xxan"),
 };
-
diff --git a/libavcodec/zmbv.c b/libavcodec/zmbv.c
index a68e42d789..22a3272dee 100644
--- a/libavcodec/zmbv.c
+++ b/libavcodec/zmbv.c
@@ -677,4 +677,3 @@ AVCodec ff_zmbv_decoder = {
     .capabilities   = CODEC_CAP_DR1,
     .long_name = NULL_IF_CONFIG_SMALL("Zip Motion Blocks Video"),
 };
-