Merge branch 'master' into oldabi

* master: (22 commits)
  ffmpeg:Daemon mode, add -d as first option to try it. Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
  ffmpeg:Fix negative verbositiy Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
  Include authorship information from ffmpeg-mt at Ronald S. Bultjes request.
  In mov and flv muxer, check aac bitstream validity.
  Added key_frame and pict_type to vsrc_movie
  Allow h264pred_init_arm.c to compile.
  anm decoder: move buffer allocation from decode_init() to decode_frame()
  vsrc_movie: fix leak in request_frame()
  Replace mplayerhq.hu URLs by libav.org.
  asfdec: Remove dead code from asf_read_close().
  ptx: Use av_log_ask_for_sample() where appropriate.
  Merge remote-tracking branch 'ffmpeg-mt/master'
  10l, commit that should have been stashed into the merge. Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
  Update regtest checksums after revision 6001dad.
  Replace more FFmpeg references by Libav.
  ac3dec: fix processing of delta bit allocation information.
  vc1: fix fate-vc1 after previous commit.
  wmv3dec: fix playback of complex WMV3 files using simple_idct.
  Replace references to ffmpeg-devel with libav-devel; fix roundup URL.
  make av_dup_packet() more cautious on allocation failures
  ...

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

10 files changed, 93 insertions, 65 deletions

diff --git a/libavcodec/ac3.c b/libavcodec/ac3.c
index dc7cbb3dcc..29e132f5d1 100644
--- a/libavcodec/ac3.c
+++ b/libavcodec/ac3.c
@@ -192,9 +192,9 @@ int ff_ac3_bit_alloc_calc_mask(AC3BitAllocParameters *s, int16_t *band_psd,
 
     if (dba_mode == DBA_REUSE || dba_mode == DBA_NEW) {
         int i, seg, delta;
-        if (dba_nsegs >= 8)
+        if (dba_nsegs > 8)
             return -1;
-        band = 0;
+        band = band_start;
         for (seg = 0; seg < dba_nsegs; seg++) {
             band += dba_offsets[seg];
             if (band >= AC3_CRITICAL_BANDS || dba_lengths[seg] > AC3_CRITICAL_BANDS-band)
diff --git a/libavcodec/ac3dec.c b/libavcodec/ac3dec.c
index 5ef576ae0d..b2e4f81704 100644
--- a/libavcodec/ac3dec.c
+++ b/libavcodec/ac3dec.c
@@ -1175,8 +1175,8 @@ static int decode_audio_block(AC3DecodeContext *s, int blk)
         /* channel delta offset, len and bit allocation */
         for (ch = !cpl_in_use; ch <= fbw_channels; ch++) {
             if (s->dba_mode[ch] == DBA_NEW) {
-                s->dba_nsegs[ch] = get_bits(gbc, 3);
-                for (seg = 0; seg <= s->dba_nsegs[ch]; seg++) {
+                s->dba_nsegs[ch] = get_bits(gbc, 3) + 1;
+                for (seg = 0; seg < s->dba_nsegs[ch]; seg++) {
                     s->dba_offsets[ch][seg] = get_bits(gbc, 5);
                     s->dba_lengths[ch][seg] = get_bits(gbc, 4);
                     s->dba_values[ch][seg] = get_bits(gbc, 3);
diff --git a/libavcodec/anm.c b/libavcodec/anm.c
index 5b2b9df8ea..e216c08441 100644
--- a/libavcodec/anm.c
+++ b/libavcodec/anm.c
@@ -29,6 +29,7 @@
 
 typedef struct AnmContext {
     AVFrame frame;
+    int palette[AVPALETTE_COUNT];
     int x;  ///< x coordinate position
 } AnmContext;
 
@@ -44,14 +45,10 @@ static av_cold int decode_init(AVCodecContext *avctx)
         return -1;
 
     s->frame.reference = 1;
-    if (avctx->get_buffer(avctx, &s->frame) < 0) {
-        av_log(avctx, AV_LOG_ERROR, "get_buffer() failed\n");
-        return -1;
-    }
 
     buf = avctx->extradata + 16*8;
     for (i = 0; i < 256; i++)
-        ((uint32_t*)s->frame.data[1])[i] = bytestream_get_le32(&buf);
+        s->palette[i] = bytestream_get_le32(&buf);
 
     return 0;
 }
@@ -170,6 +167,8 @@ static int decode_frame(AVCodecContext *avctx,
         }
     } while (buf + 1 < buf_end);
 
+    memcpy(s->frame.data[1], s->palette, AVPALETTE_SIZE);
+
     *data_size = sizeof(AVFrame);
     *(AVFrame*)data = s->frame;
     return buf_size;
diff --git a/libavcodec/arm/h264pred_init_arm.c b/libavcodec/arm/h264pred_init_arm.c
index 142f4b0177..5b11b7da88 100644
--- a/libavcodec/arm/h264pred_init_arm.c
+++ b/libavcodec/arm/h264pred_init_arm.c
@@ -74,7 +74,7 @@ static void ff_h264_pred_init_neon(H264PredContext *h, int codec_id, const int b
         h->pred16x16[PLANE_PRED8x8  ] = ff_pred16x16_plane_neon;
 }
 
-void ff_h264_pred_init_arm(H264PredContext *h, int codec_id, bit_depth)
+void ff_h264_pred_init_arm(H264PredContext *h, int codec_id, const int bit_depth)
 {
     if (HAVE_NEON)    ff_h264_pred_init_neon(h, codec_id, bit_depth);
 }
diff --git a/libavcodec/h264.c b/libavcodec/h264.c
index 48a2455b22..c64e6fb3f9 100644
--- a/libavcodec/h264.c
+++ b/libavcodec/h264.c
@@ -715,7 +715,7 @@ static int decode_update_thread_context(AVCodecContext *dst, const AVCodecContex
     copy_fields(h, h1, short_ref, cabac_init_idc);
 
     copy_picture_range(h->short_ref,   h1->short_ref,   32, s, s1);
-    copy_picture_range(h->long_ref,    h1->long_ref,    32,  s, s1);
+    copy_picture_range(h->long_ref,    h1->long_ref,    32, s, s1);
     copy_picture_range(h->delayed_pic, h1->delayed_pic, MAX_DELAYED_PIC_COUNT+2, s, s1);
 
     h->last_slice_type = h1->last_slice_type;
@@ -930,6 +930,8 @@ static void decode_postinit(H264Context *h){
 
     if(out_of_order || pics > s->avctx->has_b_frames){
         out->reference &= ~DELAYED_PIC_REF;
+        out->owner2 = s; // for frame threading, the owner must be the second field's thread
+                         // or else the first thread can release the picture and reuse it unsafely
         for(i=out_idx; h->delayed_pic[i]; i++)
             h->delayed_pic[i] = h->delayed_pic[i+1];
     }
@@ -2049,9 +2051,13 @@ static int decode_slice_header(H264Context *h, H264Context *h0){
             s0->first_field = FIELD_PICTURE;
         }
 
-        if((!FIELD_PICTURE || s0->first_field) && ff_h264_frame_start(h) < 0) {
-            s0->first_field = 0;
-            return -1;
+        if(!FIELD_PICTURE || s0->first_field) {
+            if (ff_h264_frame_start(h) < 0) {
+                s0->first_field = 0;
+                return -1;
+            }
+        } else {
+            ff_release_unused_pictures(s, 0);
         }
     }
     if(h != h0)
diff --git a/libavcodec/mpegvideo.c b/libavcodec/mpegvideo.c
index 676c3ebe8a..819f58dc95 100644
--- a/libavcodec/mpegvideo.c
+++ b/libavcodec/mpegvideo.c
@@ -316,6 +316,7 @@ int ff_alloc_picture(MpegEncContext *s, Picture *pic, int shared){
     s->prev_pict_types[0]= s->dropable ? FF_B_TYPE : s->pict_type;
     if(pic->age < PREV_PICT_TYPES_BUFFER_SIZE && s->prev_pict_types[pic->age] == FF_B_TYPE)
         pic->age= INT_MAX; // Skipped MBs in B-frames are quite rare in MPEG-1/2 and it is a bit tricky to skip them anyway.
+    pic->owner2 = s;
 
     return 0;
 fail: //for the FF_ALLOCZ_OR_GOTO macro
@@ -946,6 +947,21 @@ void init_vlc_rl(RLTable *rl)
     }
 }
 
+void ff_release_unused_pictures(MpegEncContext *s, int remove_current)
+{
+    int i;
+
+    /* release non reference frames */
+    for(i=0; i<s->picture_count; i++){
+        if(s->picture[i].data[0] && !s->picture[i].reference
+           && s->picture[i].owner2 == s
+           && (remove_current || &s->picture[i] != s->current_picture_ptr)
+           /*&& s->picture[i].type!=FF_BUFFER_TYPE_SHARED*/){
+            free_frame_buffer(s, &s->picture[i]);
+        }
+    }
+}
+
 int ff_find_unused_picture(MpegEncContext *s, int shared){
     int i;
 
@@ -1025,12 +1041,7 @@ int MPV_frame_start(MpegEncContext *s, AVCodecContext *avctx)
     }
 
     if(!s->encoding){
-        /* release non reference frames */
-        for(i=0; i<s->picture_count; i++){
-            if(s->picture[i].data[0] && !s->picture[i].reference /*&& s->picture[i].type!=FF_BUFFER_TYPE_SHARED*/){
-                free_frame_buffer(s, &s->picture[i]);
-            }
-        }
+        ff_release_unused_pictures(s, 1);
 
         if(s->current_picture_ptr && s->current_picture_ptr->data[0]==NULL)
             pic= s->current_picture_ptr; //we already have a unused image (maybe it was set before reading the header)
diff --git a/libavcodec/mpegvideo.h b/libavcodec/mpegvideo.h
index fee75b248d..3836a6ef38 100644
--- a/libavcodec/mpegvideo.h
+++ b/libavcodec/mpegvideo.h
@@ -76,6 +76,8 @@ enum OutputFormat {
 #define EXT_START_CODE          0x000001b5
 #define USER_START_CODE         0x000001b2
 
+struct MpegEncContext;
+
 /**
  * Picture.
  */
@@ -132,10 +134,9 @@ typedef struct Picture{
     uint8_t *mb_mean;           ///< Table for MB luminance
     int32_t *mb_cmp_score;      ///< Table for MB cmp scores, for mb decision FIXME remove
     int b_frame_score;          /* */
+    struct MpegEncContext *owner2; ///< pointer to the MpegEncContext that allocated this picture
 } Picture;
 
-struct MpegEncContext;
-
 /**
  * Motion estimation context.
  */
@@ -712,6 +713,7 @@ void ff_draw_horiz_band(MpegEncContext *s, int y, int h);
 void ff_mpeg_flush(AVCodecContext *avctx);
 void ff_print_debug_info(MpegEncContext *s, AVFrame *pict);
 void ff_write_quant_matrix(PutBitContext *pb, uint16_t *matrix);
+void ff_release_unused_pictures(MpegEncContext *s, int remove_current);
 int ff_find_unused_picture(MpegEncContext *s, int shared);
 void ff_denoise_dct(MpegEncContext *s, DCTELEM *block);
 void ff_update_duplicate_context(MpegEncContext *dst, MpegEncContext *src);
diff --git a/libavcodec/ptx.c b/libavcodec/ptx.c
index 28df0b6f2e..426f46ce6c 100644
--- a/libavcodec/ptx.c
+++ b/libavcodec/ptx.c
@@ -51,14 +51,14 @@ static int ptx_decode_frame(AVCodecContext *avctx, void *data, int *data_size,
     bytes_per_pixel = AV_RL16(buf+12) >> 3;
 
     if (bytes_per_pixel != 2) {
-        av_log(avctx, AV_LOG_ERROR, "image format is not rgb15, please report on ffmpeg-users mailing list\n");
+        av_log_ask_for_sample(avctx, "Image format is not RGB15.\n");
         return -1;
     }
 
     avctx->pix_fmt = PIX_FMT_RGB555;
 
     if (offset != 0x2c)
-        av_log(avctx, AV_LOG_WARNING, "offset != 0x2c, untested due to lack of sample files\n");
+        av_log_ask_for_sample(avctx, "offset != 0x2c\n");
 
     buf += offset;
 
diff --git a/libavcodec/vc1.h b/libavcodec/vc1.h
index 261ac541f9..19be3c3452 100644
--- a/libavcodec/vc1.h
+++ b/libavcodec/vc1.h
@@ -218,6 +218,7 @@ typedef struct VC1Context{
     int range_x, range_y; ///< MV range
     uint8_t pq, altpq;    ///< Current/alternate frame quantizer scale
     uint8_t zz_8x8[4][64];///< Zigzag table for TT_8x8, permuted for IDCT
+    int left_blk_sh, top_blk_sh; ///< Either 3 or 0, positions of l/t in blk[]
     const uint8_t* zz_8x4;///< Zigzag scan table for TT_8x4 coding mode
     const uint8_t* zz_4x8;///< Zigzag scan table for TT_4x8 coding mode
     /** pquant parameters */
diff --git a/libavcodec/vc1dec.c b/libavcodec/vc1dec.c
index 41322ef726..27695f184a 100644
--- a/libavcodec/vc1dec.c
+++ b/libavcodec/vc1dec.c
@@ -1499,16 +1499,16 @@ static int vc1_decode_i_block(VC1Context *v, DCTELEM block[64], int n, int coded
         if(s->ac_pred) {
             if(dc_pred_dir) { //left
                 for(k = 1; k < 8; k++)
-                    block[k] += ac_val[k];
+                    block[k << v->left_blk_sh] += ac_val[k];
             } else { //top
                 for(k = 1; k < 8; k++)
-                    block[k << 3] += ac_val[k + 8];
+                    block[k << v->top_blk_sh] += ac_val[k + 8];
             }
         }
         /* save AC coeffs for further prediction */
         for(k = 1; k < 8; k++) {
-            ac_val2[k] = block[k];
-            ac_val2[k + 8] = block[k << 3];
+            ac_val2[k]     = block[k << v->left_blk_sh];
+            ac_val2[k + 8] = block[k << v->top_blk_sh];
         }
 
         /* scale AC coeffs */
@@ -1545,15 +1545,15 @@ not_coded:
         if(s->ac_pred) {
             if(dc_pred_dir) { //left
                 for(k = 1; k < 8; k++) {
-                    block[k] = ac_val[k] * scale;
-                    if(!v->pquantizer && block[k])
-                        block[k] += (block[k] < 0) ? -v->pq : v->pq;
+                    block[k << v->left_blk_sh] = ac_val[k] * scale;
+                    if(!v->pquantizer && block[k << v->left_blk_sh])
+                        block[k << v->left_blk_sh] += (block[k << v->left_blk_sh] < 0) ? -v->pq : v->pq;
                 }
             } else { //top
                 for(k = 1; k < 8; k++) {
-                    block[k << 3] = ac_val[k + 8] * scale;
-                    if(!v->pquantizer && block[k << 3])
-                        block[k << 3] += (block[k << 3] < 0) ? -v->pq : v->pq;
+                    block[k << v->top_blk_sh] = ac_val[k + 8] * scale;
+                    if(!v->pquantizer && block[k << v->top_blk_sh])
+                        block[k << v->top_blk_sh] += (block[k << v->top_blk_sh] < 0) ? -v->pq : v->pq;
                 }
             }
             i = 63;
@@ -1680,25 +1680,25 @@ static int vc1_decode_i_block_adv(VC1Context *v, DCTELEM block[64], int n, int c
 
                 if(dc_pred_dir) { //left
                     for(k = 1; k < 8; k++)
-                        block[k] += (ac_val[k] * q2 * ff_vc1_dqscale[q1 - 1] + 0x20000) >> 18;
+                        block[k << v->left_blk_sh] += (ac_val[k] * q2 * ff_vc1_dqscale[q1 - 1] + 0x20000) >> 18;
                 } else { //top
                     for(k = 1; k < 8; k++)
-                        block[k << 3] += (ac_val[k + 8] * q2 * ff_vc1_dqscale[q1 - 1] + 0x20000) >> 18;
+                        block[k << v->top_blk_sh] += (ac_val[k + 8] * q2 * ff_vc1_dqscale[q1 - 1] + 0x20000) >> 18;
                 }
             } else {
                 if(dc_pred_dir) { //left
                     for(k = 1; k < 8; k++)
-                        block[k] += ac_val[k];
+                        block[k << v->left_blk_sh] += ac_val[k];
                 } else { //top
                     for(k = 1; k < 8; k++)
-                        block[k << 3] += ac_val[k + 8];
+                        block[k << v->top_blk_sh] += ac_val[k + 8];
                 }
             }
         }
         /* save AC coeffs for further prediction */
         for(k = 1; k < 8; k++) {
-            ac_val2[k] = block[k];
-            ac_val2[k + 8] = block[k << 3];
+            ac_val2[k    ] = block[k << v->left_blk_sh];
+            ac_val2[k + 8] = block[k << v->top_blk_sh];
         }
 
         /* scale AC coeffs */
@@ -1740,15 +1740,15 @@ static int vc1_decode_i_block_adv(VC1Context *v, DCTELEM block[64], int n, int c
         if(use_pred) {
             if(dc_pred_dir) { //left
                 for(k = 1; k < 8; k++) {
-                    block[k] = ac_val2[k] * scale;
-                    if(!v->pquantizer && block[k])
-                        block[k] += (block[k] < 0) ? -mquant : mquant;
+                    block[k << v->left_blk_sh] = ac_val2[k] * scale;
+                    if(!v->pquantizer && block[k << v->left_blk_sh])
+                        block[k << v->left_blk_sh] += (block[k << v->left_blk_sh] < 0) ? -mquant : mquant;
                 }
             } else { //top
                 for(k = 1; k < 8; k++) {
-                    block[k << 3] = ac_val2[k + 8] * scale;
-                    if(!v->pquantizer && block[k << 3])
-                        block[k << 3] += (block[k << 3] < 0) ? -mquant : mquant;
+                    block[k << v->top_blk_sh] = ac_val2[k + 8] * scale;
+                    if(!v->pquantizer && block[k << v->top_blk_sh])
+                        block[k << v->top_blk_sh] += (block[k << v->top_blk_sh] < 0) ? -mquant : mquant;
                 }
             }
             i = 63;
@@ -1878,25 +1878,25 @@ static int vc1_decode_intra_block(VC1Context *v, DCTELEM block[64], int n, int c
 
                 if(dc_pred_dir) { //left
                     for(k = 1; k < 8; k++)
-                        block[k] += (ac_val[k] * q2 * ff_vc1_dqscale[q1 - 1] + 0x20000) >> 18;
+                        block[k << v->left_blk_sh] += (ac_val[k] * q2 * ff_vc1_dqscale[q1 - 1] + 0x20000) >> 18;
                 } else { //top
                     for(k = 1; k < 8; k++)
-                        block[k << 3] += (ac_val[k + 8] * q2 * ff_vc1_dqscale[q1 - 1] + 0x20000) >> 18;
+                        block[k << v->top_blk_sh] += (ac_val[k + 8] * q2 * ff_vc1_dqscale[q1 - 1] + 0x20000) >> 18;
                 }
             } else {
                 if(dc_pred_dir) { //left
                     for(k = 1; k < 8; k++)
-                        block[k] += ac_val[k];
+                        block[k << v->left_blk_sh] += ac_val[k];
                 } else { //top
                     for(k = 1; k < 8; k++)
-                        block[k << 3] += ac_val[k + 8];
+                        block[k << v->top_blk_sh] += ac_val[k + 8];
                 }
             }
         }
         /* save AC coeffs for further prediction */
         for(k = 1; k < 8; k++) {
-            ac_val2[k] = block[k];
-            ac_val2[k + 8] = block[k << 3];
+            ac_val2[k    ] = block[k << v->left_blk_sh];
+            ac_val2[k + 8] = block[k << v->top_blk_sh];
         }
 
         /* scale AC coeffs */
@@ -1938,15 +1938,15 @@ static int vc1_decode_intra_block(VC1Context *v, DCTELEM block[64], int n, int c
         if(use_pred) {
             if(dc_pred_dir) { //left
                 for(k = 1; k < 8; k++) {
-                    block[k] = ac_val2[k] * scale;
-                    if(!v->pquantizer && block[k])
-                        block[k] += (block[k] < 0) ? -mquant : mquant;
+                    block[k << v->left_blk_sh] = ac_val2[k] * scale;
+                    if(!v->pquantizer && block[k << v->left_blk_sh])
+                        block[k << v->left_blk_sh] += (block[k << v->left_blk_sh] < 0) ? -mquant : mquant;
                 }
             } else { //top
                 for(k = 1; k < 8; k++) {
-                    block[k << 3] = ac_val2[k + 8] * scale;
-                    if(!v->pquantizer && block[k << 3])
-                        block[k << 3] += (block[k << 3] < 0) ? -mquant : mquant;
+                    block[k << v->top_blk_sh] = ac_val2[k + 8] * scale;
+                    if(!v->pquantizer && block[k << v->top_blk_sh])
+                        block[k << v->top_blk_sh] += (block[k << v->top_blk_sh] < 0) ? -mquant : mquant;
                 }
             }
             i = 63;
@@ -3236,13 +3236,6 @@ static av_cold int vc1_decode_init(AVCodecContext *avctx)
         return -1;
     if (vc1_init_common(v) < 0) return -1;
     ff_vc1dsp_init(&v->vc1dsp);
-    for (i = 0; i < 64;  i++) {
-#define transpose(x) ((x>>3) | ((x&7)<<3))
-        v->zz_8x8[0][i] = transpose(wmv1_scantable[0][i]);
-        v->zz_8x8[1][i] = transpose(wmv1_scantable[1][i]);
-        v->zz_8x8[2][i] = transpose(wmv1_scantable[2][i]);
-        v->zz_8x8[3][i] = transpose(wmv1_scantable[3][i]);
-    }
 
     avctx->coded_width = avctx->width;
     avctx->coded_height = avctx->height;
@@ -3326,6 +3319,22 @@ static av_cold int vc1_decode_init(AVCodecContext *avctx)
     s->mb_width = (avctx->coded_width+15)>>4;
     s->mb_height = (avctx->coded_height+15)>>4;
 
+    if (v->profile == PROFILE_ADVANCED || v->res_fasttx) {
+        for (i = 0; i < 64;  i++) {
+#define transpose(x) ((x>>3) | ((x&7)<<3))
+            v->zz_8x8[0][i] = transpose(wmv1_scantable[0][i]);
+            v->zz_8x8[1][i] = transpose(wmv1_scantable[1][i]);
+            v->zz_8x8[2][i] = transpose(wmv1_scantable[2][i]);
+            v->zz_8x8[3][i] = transpose(wmv1_scantable[3][i]);
+        }
+        v->left_blk_sh = 0;
+        v->top_blk_sh  = 3;
+    } else {
+        memcpy(v->zz_8x8, wmv1_scantable, 4*64);
+        v->left_blk_sh = 3;
+        v->top_blk_sh  = 0;
+    }
+
     /* Allocate mb bitplanes */
     v->mv_type_mb_plane = av_malloc(s->mb_stride * s->mb_height);
     v->direct_mb_plane = av_malloc(s->mb_stride * s->mb_height);