Fix decoder handling of intra-only frames

This patch fixes bug 633:
https://code.google.com/p/webm/issues/detail?id=633

The first decoded frame does not have to be a keyframe,
it could be an inter-frame that is coded intra-only.

This patch fixes the handling of intra-only frames.

A test vector has also been added that encodes 3
intra-only frames at the start of the clip. The
test vector was generated using the code in the
following patch:
https://gerrit.chromium.org/gerrit/#/c/70680/

Change-Id: Ib40b1dbf91aae2bc047e23c626eaef09d1860147

diff --git a/test/test-data.sha1 b/test/test-data.sha1
index f9c09c6..8940027 100644 (file)
--- a/test/test-data.sha1
+++ b/test/test-data.sha1
@@ -653,6 +653,8 @@ e615575ded499ea1d992f3b38e3baa434509cdcd  vp90-2-15-segkey.webm
 e3ab35d4316c5e81325c50f5236ceca4bc0d35df  vp90-2-15-segkey.webm.md5
 9b7ca2cac09d34c4a5d296c1900f93b1e2f69d0d  vp90-2-15-segkey_adpq.webm
 8f46ba5f785d0c2170591a153e0d0d146a7c8090  vp90-2-15-segkey_adpq.webm.md5
+698a6910a97486b833073ef0c0b18d75dce57ee8  vp90-2-16-intra-only.webm
+5661b0168752969f055eec37b05fa9fa947dc7eb  vp90-2-16-intra-only.webm.md5
 0321d507ce62dedc8a51b4e9011f7a19aed9c3dc  vp91-2-04-yuv444.webm
 367e423dd41fdb49aa028574a2cfec5c2f325c5c  vp91-2-04-yuv444.webm.md5
 76024eb753cdac6a5e5703aaea189d35c3c30ac7  invalid-vp90-2-00-quantizer-00.webm.ivf.s5861_r01-05_b6-.ivf

diff --git a/test/test.mk b/test/test.mk
index 85212d9..ef81ab1 100644 (file)
--- a/test/test.mk
+++ b/test/test.mk
@@ -771,6 +771,8 @@ LIBVPX_TEST_DATA-$(CONFIG_VP9_DECODER) += vp90-2-15-segkey.webm
 LIBVPX_TEST_DATA-$(CONFIG_VP9_DECODER) += vp90-2-15-segkey.webm.md5
 LIBVPX_TEST_DATA-$(CONFIG_VP9_DECODER) += vp90-2-15-segkey_adpq.webm
 LIBVPX_TEST_DATA-$(CONFIG_VP9_DECODER) += vp90-2-15-segkey_adpq.webm.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP9_DECODER) += vp90-2-16-intra-only.webm
+LIBVPX_TEST_DATA-$(CONFIG_VP9_DECODER) += vp90-2-16-intra-only.webm.md5
 LIBVPX_TEST_DATA-$(CONFIG_VP9_DECODER) += vp91-2-04-yuv444.webm
 LIBVPX_TEST_DATA-$(CONFIG_VP9_DECODER) += vp91-2-04-yuv444.webm.md5
 

diff --git a/test/test_vectors.cc b/test/test_vectors.cc
index 41c9e26..4955887 100644 (file)
--- a/test/test_vectors.cc
+++ b/test/test_vectors.cc
@@ -180,6 +180,7 @@ const char *const kVP9TestVectors[] = {
   "vp90-2-14-resize-fp-tiles-8-16.webm", "vp90-2-14-resize-fp-tiles-8-1.webm",
   "vp90-2-14-resize-fp-tiles-8-2.webm", "vp90-2-14-resize-fp-tiles-8-4.webm",
   "vp90-2-15-segkey.webm", "vp90-2-15-segkey_adpq.webm",
+  "vp90-2-16-intra-only.webm",
   "vp91-2-04-yuv444.webm",
 };
 const int kNumVP9TestVectors = NELEMENTS(kVP9TestVectors);

diff --git a/vp9/common/vp9_onyxc_int.h b/vp9/common/vp9_onyxc_int.h
index e1753a1..afe831a 100644 (file)
--- a/vp9/common/vp9_onyxc_int.h
+++ b/vp9/common/vp9_onyxc_int.h
@@ -257,10 +257,14 @@ static INLINE void init_macroblockd(VP9_COMMON *cm, MACROBLOCKD *xd) {
   xd->mi_stride = cm->mi_stride;
 }
 
+static INLINE int frame_is_intra_only(const VP9_COMMON *const cm) {
+  return cm->frame_type == KEY_FRAME || cm->intra_only;
+}
+
 static INLINE const vp9_prob* get_partition_probs(const VP9_COMMON *cm,
                                                   int ctx) {
-  return cm->frame_type == KEY_FRAME ? vp9_kf_partition_probs[ctx]
-                                     : cm->fc.partition_prob[ctx];
+  return frame_is_intra_only(cm) ? vp9_kf_partition_probs[ctx]
+                                 : cm->fc.partition_prob[ctx];
 }
 
 static INLINE void set_skip_context(MACROBLOCKD *xd, int mi_row, int mi_col) {
@@ -299,10 +303,6 @@ static INLINE void set_prev_mi(VP9_COMMON *cm) {
                   cm->prev_mip + cm->mi_stride + 1 : NULL;
 }
 
-static INLINE int frame_is_intra_only(const VP9_COMMON *const cm) {
-  return cm->frame_type == KEY_FRAME || cm->intra_only;
-}
-
 static INLINE void update_partition_context(MACROBLOCKD *xd,
                                             int mi_row, int mi_col,
                                             BLOCK_SIZE subsize,

diff --git a/vp9/decoder/vp9_decodeframe.c b/vp9/decoder/vp9_decodeframe.c
index 8b96abb..6fc1303 100644 (file)
--- a/vp9/decoder/vp9_decodeframe.c
+++ b/vp9/decoder/vp9_decodeframe.c
@@ -605,8 +605,8 @@ static INTERP_FILTER read_interp_filter(struct vp9_read_bit_buffer *rb) {
                              : literal_to_filter[vp9_rb_read_literal(rb, 2)];
 }
 
-static void read_frame_size(struct vp9_read_bit_buffer *rb,
-                            int *width, int *height) {
+void vp9_read_frame_size(struct vp9_read_bit_buffer *rb,
+                         int *width, int *height) {
   const int w = vp9_rb_read_literal(rb, 16) + 1;
   const int h = vp9_rb_read_literal(rb, 16) + 1;
   *width = w;
@@ -617,7 +617,7 @@ static void setup_display_size(VP9_COMMON *cm, struct vp9_read_bit_buffer *rb) {
   cm->display_width = cm->width;
   cm->display_height = cm->height;
   if (vp9_rb_read_bit(rb))
-    read_frame_size(rb, &cm->display_width, &cm->display_height);
+    vp9_read_frame_size(rb, &cm->display_width, &cm->display_height);
 }
 
 static void apply_frame_size(VP9_COMMON *cm, int width, int height) {
@@ -649,7 +649,7 @@ static void apply_frame_size(VP9_COMMON *cm, int width, int height) {
 
 static void setup_frame_size(VP9_COMMON *cm, struct vp9_read_bit_buffer *rb) {
   int width, height;
-  read_frame_size(rb, &width, &height);
+  vp9_read_frame_size(rb, &width, &height);
   apply_frame_size(cm, width, height);
   setup_display_size(cm, rb);
 }
@@ -669,7 +669,7 @@ static void setup_frame_size_with_refs(VP9_COMMON *cm,
   }
 
   if (!found)
-    read_frame_size(rb, &width, &height);
+    vp9_read_frame_size(rb, &width, &height);
 
   // Check that each of the frames that this frame references has valid
   // dimensions.
@@ -1053,20 +1053,17 @@ static const uint8_t *decode_tiles_mt(VP9Decoder *pbi,
   return bit_reader_end;
 }
 
-static void check_sync_code(VP9_COMMON *cm, struct vp9_read_bit_buffer *rb) {
-  if (vp9_rb_read_literal(rb, 8) != VP9_SYNC_CODE_0 ||
-      vp9_rb_read_literal(rb, 8) != VP9_SYNC_CODE_1 ||
-      vp9_rb_read_literal(rb, 8) != VP9_SYNC_CODE_2) {
-    vpx_internal_error(&cm->error, VPX_CODEC_UNSUP_BITSTREAM,
-                       "Invalid frame sync code");
-  }
-}
-
 static void error_handler(void *data) {
   VP9_COMMON *const cm = (VP9_COMMON *)data;
   vpx_internal_error(&cm->error, VPX_CODEC_CORRUPT_FRAME, "Truncated packet");
 }
 
+int vp9_read_sync_code(struct vp9_read_bit_buffer *const rb) {
+  return vp9_rb_read_literal(rb, 8) == VP9_SYNC_CODE_0 &&
+         vp9_rb_read_literal(rb, 8) == VP9_SYNC_CODE_1 &&
+         vp9_rb_read_literal(rb, 8) == VP9_SYNC_CODE_2;
+}
+
 static BITSTREAM_PROFILE read_profile(struct vp9_read_bit_buffer *rb) {
   int profile = vp9_rb_read_bit(rb);
   profile |= vp9_rb_read_bit(rb) << 1;
@@ -1112,7 +1109,9 @@ static size_t read_uncompressed_header(VP9Decoder *pbi,
   cm->error_resilient_mode = vp9_rb_read_bit(rb);
 
   if (cm->frame_type == KEY_FRAME) {
-    check_sync_code(cm, rb);
+    if (!vp9_read_sync_code(rb))
+      vpx_internal_error(&cm->error, VPX_CODEC_UNSUP_BITSTREAM,
+                         "Invalid frame sync code");
     if (cm->profile > PROFILE_1)
       cm->bit_depth = vp9_rb_read_bit(rb) ? BITS_12 : BITS_10;
     cm->color_space = (COLOR_SPACE)vp9_rb_read_literal(rb, 3);
@@ -1150,9 +1149,18 @@ static size_t read_uncompressed_header(VP9Decoder *pbi,
         0 : vp9_rb_read_literal(rb, 2);
 
     if (cm->intra_only) {
-      check_sync_code(cm, rb);
+      if (!vp9_read_sync_code(rb))
+        vpx_internal_error(&cm->error, VPX_CODEC_UNSUP_BITSTREAM,
+                           "Invalid frame sync code");
 
       pbi->refresh_frame_flags = vp9_rb_read_literal(rb, REF_FRAMES);
+
+      // NOTE: The intra-only frame header does not include the specification of
+      // either the color format or color sub-sampling. VP9 specifies that the
+      // default color space should be YUV 4:2:0 in this case (normative).
+      cm->color_space = BT_601;
+      cm->subsampling_y = cm->subsampling_x = 1;
+
       setup_frame_size(cm, rb);
     } else {
       pbi->refresh_frame_flags = vp9_rb_read_literal(rb, REF_FRAMES);

diff --git a/vp9/decoder/vp9_decodeframe.h b/vp9/decoder/vp9_decodeframe.h
index fb15645..e5d9d62 100644 (file)
--- a/vp9/decoder/vp9_decodeframe.h
+++ b/vp9/decoder/vp9_decodeframe.h
@@ -18,6 +18,7 @@ extern "C" {
 
 struct VP9Common;
 struct VP9Decoder;
+struct vp9_read_bit_buffer;
 
 void vp9_init_dequantizer(struct VP9Common *cm);
 
@@ -25,6 +26,10 @@ void vp9_decode_frame(struct VP9Decoder *pbi,
                       const uint8_t *data, const uint8_t *data_end,
                       const uint8_t **p_data_end);
 
+int vp9_read_sync_code(struct vp9_read_bit_buffer *const rb);
+void vp9_read_frame_size(struct vp9_read_bit_buffer *rb,
+                         int *width, int *height);
+
 #ifdef __cplusplus
 }  // extern "C"
 #endif

diff --git a/vp9/decoder/vp9_decoder.c b/vp9/decoder/vp9_decoder.c
index d154e9d..6f1f21f 100644 (file)
--- a/vp9/decoder/vp9_decoder.c
+++ b/vp9/decoder/vp9_decoder.c
@@ -267,7 +267,10 @@ int vp9_receive_compressed_data(VP9Decoder *pbi,
 
   vp9_decode_frame(pbi, source, source + size, psource);
 
-  swap_frame_buffers(pbi);
+  if (!cm->show_existing_frame)
+    swap_frame_buffers(pbi);
+  else
+    cm->frame_to_show = get_frame_new_buffer(cm);
 
   vp9_clear_system_state();
 
@@ -291,6 +294,7 @@ int vp9_receive_compressed_data(VP9Decoder *pbi,
 
 int vp9_get_raw_frame(VP9Decoder *pbi, YV12_BUFFER_CONFIG *sd,
                       vp9_ppflags_t *flags) {
+  VP9_COMMON *const cm = &pbi->common;
   int ret = -1;
 #if !CONFIG_VP9_POSTPROC
   (void)*flags;
@@ -300,15 +304,20 @@ int vp9_get_raw_frame(VP9Decoder *pbi, YV12_BUFFER_CONFIG *sd,
     return ret;
 
   /* no raw frame to show!!! */
-  if (pbi->common.show_frame == 0)
+  if (!cm->show_frame)
     return ret;
 
   pbi->ready_for_new_data = 1;
 
 #if CONFIG_VP9_POSTPROC
-  ret = vp9_post_proc_frame(&pbi->common, sd, flags);
+  if (!cm->show_existing_frame) {
+    ret = vp9_post_proc_frame(cm, sd, flags);
+  } else {
+    *sd = *cm->frame_to_show;
+    ret = 0;
+  }
 #else
-  *sd = *pbi->common.frame_to_show;
+  *sd = *cm->frame_to_show;
   ret = 0;
 #endif /*!CONFIG_POSTPROC*/
   vp9_clear_system_state();

diff --git a/vp9/decoder/vp9_decoder.h b/vp9/decoder/vp9_decoder.h
index ab4f9a2..2a618f9 100644 (file)
--- a/vp9/decoder/vp9_decoder.h
+++ b/vp9/decoder/vp9_decoder.h
@@ -19,7 +19,6 @@
 #include "vp9/common/vp9_onyxc_int.h"
 #include "vp9/common/vp9_ppflags.h"
 
-#include "vp9/decoder/vp9_decoder.h"
 #include "vp9/decoder/vp9_dthread.h"
 #include "vp9/decoder/vp9_thread.h"
 

diff --git a/vp9/vp9_dx_iface.c b/vp9/vp9_dx_iface.c
index c3ca7ee..2591852 100644 (file)
--- a/vp9/vp9_dx_iface.c
+++ b/vp9/vp9_dx_iface.c
@@ -20,6 +20,7 @@
 #include "vp9/common/vp9_frame_buffers.h"
 
 #include "vp9/decoder/vp9_decoder.h"
+#include "vp9/decoder/vp9_decodeframe.h"
 #include "vp9/decoder/vp9_read_bit_buffer.h"
 
 #include "vp9/vp9_iface_common.h"
@@ -98,8 +99,10 @@ static vpx_codec_err_t decoder_destroy(vpx_codec_alg_priv_t *ctx) {
 static vpx_codec_err_t decoder_peek_si_internal(const uint8_t *data,
                                                 unsigned int data_sz,
                                                 vpx_codec_stream_info_t *si,
+                                                int *is_intra_only,
                                                 vpx_decrypt_cb decrypt_cb,
                                                 void *decrypt_state) {
+  int intra_only_flag = 0;
   uint8_t clear_buffer[9];
 
   if (data + data_sz <= data)
@@ -115,6 +118,8 @@ static vpx_codec_err_t decoder_peek_si_internal(const uint8_t *data,
   }
 
   {
+    int show_frame;
+    int error_resilient;
     struct vp9_read_bit_buffer rb = { data, data + data_sz, 0, NULL, NULL };
     const int frame_marker = vp9_rb_read_literal(&rb, 2);
     const int version = vp9_rb_read_bit(&rb);
@@ -126,6 +131,7 @@ static vpx_codec_err_t decoder_peek_si_internal(const uint8_t *data,
     if (version > 1) return VPX_CODEC_UNSUP_BITSTREAM;
 
     if (vp9_rb_read_bit(&rb)) {  // show an existing frame
+      vp9_rb_read_literal(&rb, 3);  // Frame buffer to show.
       return VPX_CODEC_OK;
     }
 
@@ -133,18 +139,15 @@ static vpx_codec_err_t decoder_peek_si_internal(const uint8_t *data,
       return VPX_CODEC_UNSUP_BITSTREAM;
 
     si->is_kf = !vp9_rb_read_bit(&rb);
+    show_frame = vp9_rb_read_bit(&rb);
+    error_resilient = vp9_rb_read_bit(&rb);
+
     if (si->is_kf) {
       const int sRGB = 7;
       int colorspace;
 
-      rb.bit_offset += 1;  // show frame
-      rb.bit_offset += 1;  // error resilient
-
-      if (vp9_rb_read_literal(&rb, 8) != VP9_SYNC_CODE_0 ||
-          vp9_rb_read_literal(&rb, 8) != VP9_SYNC_CODE_1 ||
-          vp9_rb_read_literal(&rb, 8) != VP9_SYNC_CODE_2) {
+      if (!vp9_read_sync_code(&rb))
         return VPX_CODEC_UNSUP_BITSTREAM;
-      }
 
       colorspace = vp9_rb_read_literal(&rb, 3);
       if (colorspace != sRGB) {
@@ -161,20 +164,28 @@ static vpx_codec_err_t decoder_peek_si_internal(const uint8_t *data,
           return VPX_CODEC_UNSUP_BITSTREAM;
         }
       }
+      vp9_read_frame_size(&rb, (int *)&si->w, (int *)&si->h);
+    } else {
+      intra_only_flag = show_frame ? 0 : vp9_rb_read_bit(&rb);
+      rb.bit_offset += error_resilient ? 0 : 2;  // reset_frame_context
 
-      // TODO(jzern): these are available on non-keyframes in intra only mode.
-      si->w = vp9_rb_read_literal(&rb, 16) + 1;
-      si->h = vp9_rb_read_literal(&rb, 16) + 1;
+      if (intra_only_flag) {
+        if (!vp9_read_sync_code(&rb))
+          return VPX_CODEC_UNSUP_BITSTREAM;
+        rb.bit_offset += REF_FRAMES;  // refresh_frame_flags
+        vp9_read_frame_size(&rb, (int *)&si->w, (int *)&si->h);
+      }
     }
   }
-
+  if (is_intra_only != NULL)
+    *is_intra_only = intra_only_flag;
   return VPX_CODEC_OK;
 }
 
 static vpx_codec_err_t decoder_peek_si(const uint8_t *data,
                                        unsigned int data_sz,
                                        vpx_codec_stream_info_t *si) {
-  return decoder_peek_si_internal(data, data_sz, si, NULL, NULL);
+  return decoder_peek_si_internal(data, data_sz, si, NULL, NULL, NULL);
 }
 
 static vpx_codec_err_t decoder_get_si(vpx_codec_alg_priv_t *ctx,
@@ -266,13 +277,14 @@ static vpx_codec_err_t decode_one(vpx_codec_alg_priv_t *ctx,
   // validate that we have a buffer that does not wrap around the top
   // of the heap.
   if (!ctx->si.h) {
+    int is_intra_only = 0;
     const vpx_codec_err_t res =
-        decoder_peek_si_internal(*data, data_sz, &ctx->si, ctx->decrypt_cb,
-                                 ctx->decrypt_state);
+        decoder_peek_si_internal(*data, data_sz, &ctx->si, &is_intra_only,
+                                 ctx->decrypt_cb, ctx->decrypt_state);
     if (res != VPX_CODEC_OK)
       return res;
 
-    if (!ctx->si.is_kf)
+    if (!ctx->si.is_kf && !is_intra_only)
       return VPX_CODEC_ERROR;
   }