Log codec IDs for MediaSource content.

media/filters/ffmpeg_audio_decoder.cc

Index: media/filters/ffmpeg_audio_decoder.cc
diff --git a/media/filters/ffmpeg_audio_decoder.cc b/media/filters/ffmpeg_audio_decoder.cc
index 5c9a29c203177e9e46262d547a01c1bd726a8f7d..8af3fc2c4da45fc8fab81d2cdfa3b7c7442b8c37 100644
--- a/media/filters/ffmpeg_audio_decoder.cc
+++ b/media/filters/ffmpeg_audio_decoder.cc
@@ -126,8 +126,12 @@ static int GetAudioBuffer(struct AVCodecContext* s, AVFrame* frame, int flags) {
 }
 
 FFmpegAudioDecoder::FFmpegAudioDecoder(
-    const scoped_refptr<base::SingleThreadTaskRunner>& task_runner)
-    : task_runner_(task_runner), state_(kUninitialized), av_sample_format_(0) {
+    const scoped_refptr<base::SingleThreadTaskRunner>& task_runner,
+    const LogCB& log_cb)
+    : task_runner_(task_runner),
+      state_(kUninitialized),
+      av_sample_format_(0),
+      log_cb_(log_cb) {
 }
 
 FFmpegAudioDecoder::~FFmpegAudioDecoder() {
@@ -351,6 +355,12 @@ bool FFmpegAudioDecoder::FFmpegDecode(
                     << ", Sample Format: " << av_frame_->format << " vs "
                     << av_sample_format_;
 
+        if ((config_.codec() == kCodecAAC) &&
+            av_frame_->sample_rate == 2 * config_.samples_per_second()) {

[scherkus (not reviewing), 2014/05/08 00:26:14]

nit: it's funny you put the ()'s around the simple comparison to kCodecAAC but
not the 2 * config_.samples_per_second() bit :)

[acolwell GONE FROM CHROMIUM, 2014/05/08 00:57:01]

Yeah. Not sure why I did that. () removed.

+          MEDIA_LOG(log_cb_) << "Implicit HE-AAC signalling is being used."

[scherkus (not reviewing), 2014/05/07 20:25:53]

should this be moved closer to MSE code instead of being inside of the decoder?

[acolwell GONE FROM CHROMIUM, 2014/05/07 20:47:45]

We can't. Implict signalling of HE-AAC can only be detected by decoding the
packet. All the info in the header looks like AAC-LC. The only way to tell that
it is HE-AAC is to decode the AAC-LC data and notice that some HE-AAC follows
it. This is a backwards compatibility "feature". 

The REAL fix here is to defer audio device initialization until after the first
frame is decoded, but that is a non-trivial change. This is a stopgap measure to
help people understand what to do until we actually make the hard change.

+                             << " Please use mp4a.40.5 instead of mp4a.40.2 in"
+                             << " the mimetype.";
+        }
         // This is an unrecoverable error, so bail out.
         queued_audio_.clear();
         av_frame_unref(av_frame_.get());