Created attachment 86502 [details] [review]
patch

This patch makes multiqueue recalculate the next_non_linked pad differently so that all non-linked pads get woken up in turn to deliver their data.

It also makes it so that all pads are considered not-linked by default. That way, data on them is held until preceding not-linked data is output, rather than having them immediately output data, which ends up being in the wrong order if the new pad turns out to be not-linked.

Actually, this patch still isn't quite right - it doesn't wake up not-linked pads correctly when they drain and get new data. (I'm pretty sure it's no worse than the old method in that respect though - it just wasn't as obvious there because pads weren't treated as initially not-linked)

Fixed in CVS:

2007-06-26  Jan Schmidt  <thaytan@noraisin.net>

        * plugins/elements/gstmultiqueue.c: (gst_multi_queue_init),
        (gst_single_queue_flush), (apply_segment), (apply_buffer),
        (gst_single_queue_push_one), (gst_multi_queue_loop),
        (gst_multi_queue_sink_activate_push), (gst_multi_queue_sink_event),
        (gst_multi_queue_src_activate_push), (wake_up_next_non_linked),
        (compute_high_id), (gst_single_queue_new):
        * plugins/elements/gstmultiqueue.h:
        Take the multiqueue lock when updating the fill level so we don't get
        confused.

        After applying a buffer or event on the src pad segment, make sure to
        call gst_data_queue_limits_changed() to get the data queue to unblock
        and check the filled state again.

        Rework the not-linked pad handling so the logic is that not-linked
        pads can push as fast as they like, but only so they never get
        ahead of any linked pads.

        * tests/check/elements/multiqueue.c: (mq_sinkpad_to_srcpad),
        (mq_dummypad_getcaps), (mq_dummypad_chain), (mq_dummypad_event),
        (run_output_order_test), (GST_START_TEST), (multiqueue_suite):

        Add a test to check that not-linked pads always stay behind
        linked pads.

        Fixes: #430682