spice-gtk updates the mm-time based on the audio playback channel and then uses it to compute the video delay of the video streams if any. More precisely whenever spice_playback_channel_set_delay() is called it takes the mm-time of the last audio packet it received, and substracts the delay from it to it to compute the new mm-time. There are two issues: 1) First if a video plays with no sound, e.g. 'mplayer -nosound', then the playback only receives a single SpiceMsgPlaybackStart packet when the client connects. This means any call to spice_playback_channel_set_delay() will reset the mm-time to a constant value. 2) Second, for an as yet unknown reason, if you start playing a video, disconnect the client without stopping the video, and reconnect the client, the audio stream does not seem to resume as expected, resulting, again, in a constant mm-time. In both cases the video delay quickly gets over the 5 second limit which freezes the video. Issue 2 can probably be fixed within the current framework but issue 1 points to a much larger issue. Any ideas? -- Francois Gouget <fgouget@xxxxxxxxxxxxxxx> _______________________________________________ Spice-devel mailing list Spice-devel@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/spice-devel