Memory leak when InbandEventStream is set in the AdaptationSet of the live stream manifest.

[grand719]

Environment

• [] The MPD passes the DASH-IF Conformance Tool on https://conformance.dashif.org/
• The stream has correct Access-Control-Allow-Origin headers (CORS)
• There are no network errors such as 404s in the browser console when trying to play the stream
• The issue observed is not mentioned on https://github.com/Dash-Industry-Forum/dash.js/wiki/FAQ
• The issue occurs in the latest reference client on http://reference.dashif.org/dash.js/ and not just on my page

• Link to playable MPD file:
  https://6180c994cb835402.mediapackage.eu-west-1.amazonaws.com/out/v1/3f772c2a59824b9c9901ca513f605da9/index.mpd
  If the stream is not available, please let me know so we can run it again.
  Also I have added example of manifest that was crated with usage AWS Elemental MediaPackage
• Dash.js version: 4.7.4
• Browser name/version: Chrome 128.0.6613.114, Chromium 111
• OS name/version: macOS Sonoma 14.5

Steps to reproduce

1. Prepare Stream with AWS Elemental MediaPackage
2. Run the prepared stream, which contains the <InbandEventStream schemeIdUri="https://aomedia.org/emsg/ID3" value="0"/> field in the AdaptationSet, in the referenced player."
3. Take a Heap Snapshot and Analyze JsArrayBuffer

Observed behavior

In the taken snapshot, we can observe that the JsArrayBuffer is consistently increasing, and the memory usage is not dropping. During the debug session, I noticed that unusual memory management behavior is caused by the <InbandEventStream schemeIdUri="https://aomedia.org/emsg/ID3" value="0"/> element, which is added to the video AdaptationSet. Additionally, after deleting the InbandEventStream and InbandEventStream_asArray properties from the adaptation set object in DashManifestModel.js within the getRealAdaptations function, everything works as intended.

Memory usage when Adaptation Set contains InbandEventStream:

Manifest

<?xml version="1.0" encoding="utf-8"?>
<MPD xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns="urn:mpeg:dash:schema:mpd:2011" xsi:schemaLocation="urn:mpeg:dash:schema:mpd:2011 http://standards.iso.org/ittf/PubliclyAvailableStandards/MPEG-DASH_schema_files/DASH-MPD.xsd" id="201" type="dynamic" publishTime="2024-08-30T10:09:25+00:00" minimumUpdatePeriod="PT2S" availabilityStartTime="2024-08-29T14:57:30+00:00" minBufferTime="PT10S" suggestedPresentationDelay="PT20.000S" timeShiftBufferDepth="PT37.000S" profiles="urn:mpeg:dash:profile:isoff-live:2011">
  <Period start="PT0.000S" id="1">
    <AdaptationSet id="1485523442" mimeType="video/mp4" segmentAlignment="true" startWithSAP="1" subsegmentAlignment="true" subsegmentStartsWithSAP="1" bitstreamSwitching="true">
      <InbandEventStream schemeIdUri="https://aomedia.org/emsg/ID3" value="0"/>
      <Representation id="1" width="1920" height="1080" frameRate="30/1" bandwidth="2000000" codecs="avc1.640028">
        <SegmentTemplate timescale="30000" media="index_video_3_0_$Time$.mp4?m=1724934975" initialization="index_video_3_0_init.mp4?m=1724934975" startNumber="15912">
          <SegmentTimeline>
            <S t="2072232040" d="180000" r="6"/>
          </SegmentTimeline>
        </SegmentTemplate>
      </Representation>
      <Representation id="2" width="1280" height="720" frameRate="30/1" bandwidth="1000000" codecs="avc1.64001F">
        <SegmentTemplate timescale="30000" media="index_video_5_0_$Time$.mp4?m=1724934975" initialization="index_video_5_0_init.mp4?m=1724934975" startNumber="15912">
          <SegmentTimeline>
            <S t="2072232040" d="180000" r="6"/>
          </SegmentTimeline>
        </SegmentTemplate>
      </Representation>
      <Representation id="3" width="853" height="480" frameRate="30/1" bandwidth="499968" codecs="avc1.4D401E">
        <SegmentTemplate timescale="30000" media="index_video_7_0_$Time$.mp4?m=1724934975" initialization="index_video_7_0_init.mp4?m=1724934975" startNumber="15912">
          <SegmentTimeline>
            <S t="2072232040" d="180000" r="6"/>
          </SegmentTimeline>
        </SegmentTemplate>
      </Representation>
      <Representation id="4" width="426" height="240" frameRate="30/1" bandwidth="249984" codecs="avc1.4D400D">
        <SegmentTemplate timescale="30000" media="index_video_9_0_$Time$.mp4?m=1724934975" initialization="index_video_9_0_init.mp4?m=1724934975" startNumber="15912">
          <SegmentTimeline>
            <S t="2072232040" d="180000" r="6"/>
          </SegmentTimeline>
        </SegmentTemplate>
      </Representation>
    </AdaptationSet>
    <AdaptationSet id="2083690061" mimeType="audio/mp4" segmentAlignment="true" startWithSAP="1" lang="und" bitstreamSwitching="true">
      <Label>und</Label>
      <Representation id="5" bandwidth="192357" audioSamplingRate="48000" codecs="mp4a.40.2">
        <AudioChannelConfiguration schemeIdUri="urn:mpeg:dash:23003:3:audio_channel_configuration:2011" value="2"/>
        <SegmentTemplate timescale="48000" media="index_audio_4_0_$Time$.mp4?m=1724934975" initialization="index_audio_4_0_init.mp4?m=1724934975" startNumber="15912">
          <SegmentTimeline>
            <S t="3315571968" d="287744" r="1"/>
            <S t="3316147456" d="288768"/>
            <S t="3316436224" d="287744" r="2"/>
            <S t="3317299456" d="288768"/>
          </SegmentTimeline>
        </SegmentTemplate>
      </Representation>
      <Representation id="6" bandwidth="192357" audioSamplingRate="48000" codecs="mp4a.40.2">
        <AudioChannelConfiguration schemeIdUri="urn:mpeg:dash:23003:3:audio_channel_configuration:2011" value="2"/>
        <SegmentTemplate timescale="48000" media="index_audio_6_0_$Time$.mp4?m=1724934975" initialization="index_audio_6_0_init.mp4?m=1724934975" startNumber="15912">
          <SegmentTimeline>
            <S t="3315571968" d="287744" r="1"/>
            <S t="3316147456" d="288768"/>
            <S t="3316436224" d="287744" r="2"/>
            <S t="3317299456" d="288768"/>
          </SegmentTimeline>
        </SegmentTemplate>
      </Representation>
      <Representation id="7" bandwidth="128651" audioSamplingRate="48000" codecs="mp4a.40.2">
        <AudioChannelConfiguration schemeIdUri="urn:mpeg:dash:23003:3:audio_channel_configuration:2011" value="2"/>
        <SegmentTemplate timescale="48000" media="index_audio_8_0_$Time$.mp4?m=1724934975" initialization="index_audio_8_0_init.mp4?m=1724934975" startNumber="15912">
          <SegmentTimeline>
            <S t="3315571968" d="287744" r="1"/>
            <S t="3316147456" d="288768"/>
            <S t="3316436224" d="287744" r="2"/>
            <S t="3317299456" d="288768"/>
          </SegmentTimeline>
        </SegmentTemplate>
      </Representation>
      <Representation id="8" bandwidth="128651" audioSamplingRate="48000" codecs="mp4a.40.2">
        <AudioChannelConfiguration schemeIdUri="urn:mpeg:dash:23003:3:audio_channel_configuration:2011" value="2"/>
        <SegmentTemplate timescale="48000" media="index_audio_10_0_$Time$.mp4?m=1724934975" initialization="index_audio_10_0_init.mp4?m=1724934975" startNumber="15912">
          <SegmentTimeline>
            <S t="3315571968" d="287744" r="1"/>
            <S t="3316147456" d="288768"/>
            <S t="3316436224" d="287744" r="2"/>
            <S t="3317299456" d="288768"/>
          </SegmentTimeline>
        </SegmentTemplate>
      </Representation>
    </AdaptationSet>
    <SupplementalProperty schemeIdUri="urn:scte:dash:utc-time" value="2024-08-29T14:57:25.400Z"/>
  </Period>
  <UTCTiming schemeIdUri="urn:mpeg:dash:utc:http-iso:2014" value="https://time.akamai.com/?iso&amp;ms"/>
</MPD>

Expected behavior

There should be no memory leaks when playing a live stream, even with InbandEventStream set in the AdaptationSet. Memory should be released correctly, and the size of the JsArrayBuffer should remain stable or decrease as unneeded resources are freed.

[grand719]

After further investigation, I managed to find out that the issue occurs due to the event_duration being set to 40 hours in IsoBox - emsg. How should this situation be approached, and how should the player react in such a case? We also managed to remove the emsg IsoBox using AWS MediaLive, but I believe the player should also have logic in place to prevent behavior like overloading the memory when events are too long.

[haudiobe]

I believe inband events should not have any duration that is longer than the containing segment. We should restrict it in MPEG to this. Also in case the event duration is longer than a segment, the event payload should be deleted on the client.

[ZmGorynych]

Events can legitimately be longer (e.g., duration of an ad break). They can appear before their presentation time.
I would rather do (maybe tunable) timeouts in the player.