2007-03-28 18:38:11 +00:00
|
|
|
/* GStreamer unit tests for multiqueue
|
|
|
|
*
|
|
|
|
* Copyright (C) 2007 Tim-Philipp Müller <tim centricular net>
|
|
|
|
*
|
|
|
|
* This library is free software; you can redistribute it and/or
|
|
|
|
* modify it under the terms of the GNU Library General Public
|
|
|
|
* License as published by the Free Software Foundation; either
|
|
|
|
* version 2 of the License, or (at your option) any later version.
|
|
|
|
*
|
|
|
|
* This library is distributed in the hope that it will be useful,
|
|
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
|
|
|
* Library General Public License for more details.
|
|
|
|
*
|
|
|
|
* You should have received a copy of the GNU Library General Public
|
|
|
|
* License along with this library; if not, write to the
|
|
|
|
* Free Software Foundation, Inc., 59 Temple Place - Suite 330,
|
|
|
|
* Boston, MA 02111-1307, USA.
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include <unistd.h>
|
|
|
|
|
|
|
|
#include <gst/check/gstcheck.h>
|
|
|
|
|
2007-07-16 16:04:49 +00:00
|
|
|
static GStaticMutex _check_lock = G_STATIC_MUTEX_INIT;
|
|
|
|
|
2007-03-28 18:38:11 +00:00
|
|
|
static GstElement *
|
|
|
|
setup_multiqueue (GstElement * pipe, GstElement * inputs[],
|
|
|
|
GstElement * outputs[], guint num)
|
|
|
|
{
|
|
|
|
GstElement *mq;
|
|
|
|
guint i;
|
|
|
|
|
|
|
|
mq = gst_element_factory_make ("multiqueue", NULL);
|
|
|
|
fail_unless (mq != NULL, "failed to create 'multiqueue' element");
|
|
|
|
|
|
|
|
gst_bin_add (GST_BIN (pipe), mq);
|
|
|
|
|
|
|
|
for (i = 0; i < num; ++i) {
|
|
|
|
GstPad *sinkpad = NULL;
|
|
|
|
GstPad *srcpad = NULL;
|
|
|
|
|
|
|
|
/* create multiqueue sink (and source) pad */
|
|
|
|
sinkpad = gst_element_get_request_pad (mq, "sink%d");
|
|
|
|
fail_unless (sinkpad != NULL,
|
|
|
|
"failed to create multiqueue request pad #%u", i);
|
|
|
|
|
|
|
|
/* link input element N to the N-th multiqueue sink pad we just created */
|
|
|
|
if (inputs != NULL && inputs[i] != NULL) {
|
|
|
|
gst_bin_add (GST_BIN (pipe), inputs[i]);
|
|
|
|
|
2008-05-21 15:57:52 +00:00
|
|
|
srcpad = gst_element_get_static_pad (inputs[i], "src");
|
2007-03-28 18:38:11 +00:00
|
|
|
fail_unless (srcpad != NULL, "failed to find src pad for input #%u", i);
|
|
|
|
|
|
|
|
fail_unless_equals_int (GST_PAD_LINK_OK, gst_pad_link (srcpad, sinkpad));
|
|
|
|
|
|
|
|
gst_object_unref (srcpad);
|
|
|
|
srcpad = NULL;
|
|
|
|
}
|
|
|
|
gst_object_unref (sinkpad);
|
|
|
|
sinkpad = NULL;
|
|
|
|
|
|
|
|
/* link output element N to the N-th multiqueue src pad */
|
|
|
|
if (outputs != NULL && outputs[i] != NULL) {
|
|
|
|
gchar padname[10];
|
|
|
|
|
|
|
|
/* only the sink pads are by request, the source pads are sometimes pads,
|
|
|
|
* so this should return NULL */
|
|
|
|
srcpad = gst_element_get_request_pad (mq, "src%d");
|
|
|
|
fail_unless (srcpad == NULL);
|
|
|
|
|
|
|
|
g_snprintf (padname, sizeof (padname), "src%d", i);
|
2008-05-21 15:57:52 +00:00
|
|
|
srcpad = gst_element_get_static_pad (mq, padname);
|
2007-03-28 18:38:11 +00:00
|
|
|
fail_unless (srcpad != NULL, "failed to get multiqueue src pad #%u", i);
|
|
|
|
fail_unless (GST_PAD_IS_SRC (srcpad),
|
|
|
|
"%s:%s is not a source pad?!", GST_DEBUG_PAD_NAME (srcpad));
|
|
|
|
|
|
|
|
gst_bin_add (GST_BIN (pipe), outputs[i]);
|
|
|
|
|
2008-05-21 15:57:52 +00:00
|
|
|
sinkpad = gst_element_get_static_pad (outputs[i], "sink");
|
2007-03-28 18:38:11 +00:00
|
|
|
fail_unless (sinkpad != NULL, "failed to find sink pad of output #%u", i);
|
|
|
|
fail_unless (GST_PAD_IS_SINK (sinkpad));
|
|
|
|
|
|
|
|
fail_unless_equals_int (GST_PAD_LINK_OK, gst_pad_link (srcpad, sinkpad));
|
|
|
|
|
|
|
|
gst_object_unref (srcpad);
|
|
|
|
gst_object_unref (sinkpad);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return mq;
|
|
|
|
}
|
|
|
|
|
|
|
|
GST_START_TEST (test_simple_pipeline)
|
|
|
|
{
|
|
|
|
GstElement *pipe, *mq;
|
|
|
|
GstElement *inputs[1];
|
|
|
|
GstElement *outputs[1];
|
|
|
|
GstMessage *msg;
|
|
|
|
|
|
|
|
pipe = gst_pipeline_new ("pipeline");
|
|
|
|
|
|
|
|
inputs[0] = gst_element_factory_make ("fakesrc", NULL);
|
|
|
|
fail_unless (inputs[0] != NULL, "failed to create 'fakesrc' element");
|
|
|
|
g_object_set (inputs[0], "num-buffers", 256, NULL);
|
|
|
|
|
|
|
|
outputs[0] = gst_element_factory_make ("fakesink", NULL);
|
|
|
|
fail_unless (outputs[0] != NULL, "failed to create 'fakesink' element");
|
|
|
|
|
|
|
|
mq = setup_multiqueue (pipe, inputs, outputs, 1);
|
|
|
|
|
|
|
|
gst_element_set_state (pipe, GST_STATE_PLAYING);
|
|
|
|
|
|
|
|
msg = gst_bus_poll (GST_ELEMENT_BUS (pipe),
|
|
|
|
GST_MESSAGE_EOS | GST_MESSAGE_ERROR, -1);
|
|
|
|
|
|
|
|
fail_if (GST_MESSAGE_TYPE (msg) == GST_MESSAGE_ERROR,
|
|
|
|
"Expected EOS message, got ERROR message");
|
|
|
|
gst_message_unref (msg);
|
|
|
|
|
|
|
|
GST_LOG ("Got EOS, cleaning up");
|
|
|
|
|
|
|
|
gst_element_set_state (pipe, GST_STATE_NULL);
|
|
|
|
gst_object_unref (pipe);
|
|
|
|
}
|
|
|
|
|
|
|
|
GST_END_TEST;
|
|
|
|
|
|
|
|
GST_START_TEST (test_simple_shutdown_while_running)
|
|
|
|
{
|
|
|
|
GstElement *pipe, *mq;
|
|
|
|
GstElement *inputs[1];
|
|
|
|
GstElement *outputs[1];
|
|
|
|
GstMessage *msg;
|
|
|
|
|
|
|
|
pipe = gst_pipeline_new ("pipeline");
|
|
|
|
|
|
|
|
inputs[0] = gst_element_factory_make ("fakesrc", NULL);
|
|
|
|
fail_unless (inputs[0] != NULL, "failed to create 'fakesrc' element");
|
|
|
|
|
|
|
|
outputs[0] = gst_element_factory_make ("fakesink", NULL);
|
|
|
|
fail_unless (outputs[0] != NULL, "failed to create 'fakesink' element");
|
|
|
|
|
|
|
|
mq = setup_multiqueue (pipe, inputs, outputs, 1);
|
|
|
|
|
|
|
|
gst_element_set_state (pipe, GST_STATE_PAUSED);
|
|
|
|
|
|
|
|
/* wait until pipeline is up and running */
|
|
|
|
msg = gst_bus_poll (GST_ELEMENT_BUS (pipe),
|
|
|
|
GST_MESSAGE_ERROR | GST_MESSAGE_ASYNC_DONE, -1);
|
|
|
|
fail_if (GST_MESSAGE_TYPE (msg) == GST_MESSAGE_ERROR, "Got ERROR message");
|
|
|
|
gst_message_unref (msg);
|
|
|
|
|
|
|
|
GST_LOG ("pipeline is running now");
|
|
|
|
gst_element_set_state (pipe, GST_STATE_PAUSED);
|
|
|
|
|
|
|
|
/* wait a bit to accumulate some buffers in the queue (while it's blocking
|
|
|
|
* in the sink) */
|
|
|
|
msg =
|
|
|
|
gst_bus_poll (GST_ELEMENT_BUS (pipe), GST_MESSAGE_ERROR, GST_SECOND / 4);
|
|
|
|
if (msg)
|
|
|
|
g_error ("Got ERROR message");
|
|
|
|
|
|
|
|
/* now shut down only the sink, so the queue gets a wrong-state flow return */
|
|
|
|
gst_element_set_state (outputs[0], GST_STATE_NULL);
|
|
|
|
msg =
|
|
|
|
gst_bus_poll (GST_ELEMENT_BUS (pipe), GST_MESSAGE_ERROR, GST_SECOND / 2);
|
|
|
|
if (msg)
|
|
|
|
g_error ("Got ERROR message");
|
|
|
|
|
|
|
|
GST_LOG ("Cleaning up");
|
|
|
|
|
|
|
|
gst_element_set_state (pipe, GST_STATE_NULL);
|
|
|
|
gst_object_unref (pipe);
|
|
|
|
}
|
|
|
|
|
|
|
|
GST_END_TEST;
|
|
|
|
|
|
|
|
GST_START_TEST (test_simple_create_destroy)
|
|
|
|
{
|
|
|
|
GstElement *mq;
|
|
|
|
|
|
|
|
mq = gst_element_factory_make ("multiqueue", NULL);
|
|
|
|
gst_object_unref (mq);
|
|
|
|
}
|
|
|
|
|
|
|
|
GST_END_TEST;
|
|
|
|
|
|
|
|
GST_START_TEST (test_request_pads)
|
|
|
|
{
|
|
|
|
gboolean change_state_before_cleanup = TRUE;
|
|
|
|
GstElement *mq;
|
|
|
|
GstPad *sink1, *sink2;
|
|
|
|
|
|
|
|
again:
|
|
|
|
|
|
|
|
mq = gst_element_factory_make ("multiqueue", NULL);
|
|
|
|
|
|
|
|
sink1 = gst_element_get_request_pad (mq, "foo%d");
|
|
|
|
fail_unless (sink1 == NULL,
|
|
|
|
"Expected NULL pad, as there is no request pad template for 'foo%%d'");
|
|
|
|
|
|
|
|
sink1 = gst_element_get_request_pad (mq, "src%d");
|
|
|
|
fail_unless (sink1 == NULL,
|
|
|
|
"Expected NULL pad, as there is no request pad template for 'src%%d'");
|
|
|
|
|
|
|
|
sink1 = gst_element_get_request_pad (mq, "sink%d");
|
|
|
|
fail_unless (sink1 != NULL);
|
|
|
|
fail_unless (GST_IS_PAD (sink1));
|
|
|
|
fail_unless (GST_PAD_IS_SINK (sink1));
|
|
|
|
GST_LOG ("Got pad %s:%s", GST_DEBUG_PAD_NAME (sink1));
|
|
|
|
|
|
|
|
sink2 = gst_element_get_request_pad (mq, "sink%d");
|
|
|
|
fail_unless (sink2 != NULL);
|
|
|
|
fail_unless (GST_IS_PAD (sink2));
|
|
|
|
fail_unless (GST_PAD_IS_SINK (sink2));
|
|
|
|
GST_LOG ("Got pad %s:%s", GST_DEBUG_PAD_NAME (sink2));
|
|
|
|
|
|
|
|
fail_unless (sink1 != sink2);
|
|
|
|
|
|
|
|
if (change_state_before_cleanup) {
|
|
|
|
/* FIXME: if we don't change state, it will deadlock forever when unref'ing
|
|
|
|
* the queue (waiting for pad tasks to join) */
|
|
|
|
GST_LOG ("Changing state to PLAYING");
|
|
|
|
gst_element_set_state (mq, GST_STATE_PLAYING);
|
|
|
|
g_usleep (G_USEC_PER_SEC);
|
|
|
|
GST_LOG ("Changing state to NULL");
|
|
|
|
gst_element_set_state (mq, GST_STATE_NULL);
|
|
|
|
}
|
|
|
|
|
|
|
|
GST_LOG ("Cleaning up");
|
|
|
|
gst_object_unref (sink1);
|
|
|
|
gst_object_unref (sink2);
|
|
|
|
gst_object_unref (mq);
|
|
|
|
|
|
|
|
/* FIXME: this should work without state change before cleanup as well,
|
|
|
|
* but currently doesn't, see above, so disable this for now */
|
|
|
|
if (change_state_before_cleanup && 0) {
|
|
|
|
change_state_before_cleanup = FALSE;
|
|
|
|
goto again;
|
|
|
|
}
|
|
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
GST_END_TEST;
|
|
|
|
|
plugins/elements/gstmultiqueue.*: Take the multiqueue lock when updating the fill level so we don't get confused.
Original commit message from CVS:
* 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.
2007-06-26 14:45:15 +00:00
|
|
|
static GstPad *
|
|
|
|
mq_sinkpad_to_srcpad (GstElement * mq, GstPad * sink)
|
|
|
|
{
|
|
|
|
GstPad *srcpad = NULL;
|
|
|
|
|
|
|
|
gchar *mq_sinkpad_name;
|
|
|
|
gchar *mq_srcpad_name;
|
|
|
|
|
|
|
|
mq_sinkpad_name = gst_pad_get_name (sink);
|
|
|
|
fail_unless (g_str_has_prefix (mq_sinkpad_name, "sink"));
|
|
|
|
mq_srcpad_name = g_strdup_printf ("src%s", mq_sinkpad_name + 4);
|
2008-05-21 15:57:52 +00:00
|
|
|
srcpad = gst_element_get_static_pad (mq, mq_srcpad_name);
|
plugins/elements/gstmultiqueue.*: Take the multiqueue lock when updating the fill level so we don't get confused.
Original commit message from CVS:
* 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.
2007-06-26 14:45:15 +00:00
|
|
|
fail_unless (srcpad != NULL);
|
|
|
|
|
|
|
|
g_free (mq_sinkpad_name);
|
|
|
|
g_free (mq_srcpad_name);
|
|
|
|
|
|
|
|
return srcpad;
|
|
|
|
}
|
|
|
|
|
|
|
|
static GstCaps *
|
|
|
|
mq_dummypad_getcaps (GstPad * sinkpad)
|
|
|
|
{
|
|
|
|
return gst_caps_new_any ();
|
|
|
|
}
|
|
|
|
|
|
|
|
struct PadData
|
|
|
|
{
|
|
|
|
guint8 pad_num;
|
|
|
|
guint32 *max_linked_id_ptr;
|
|
|
|
guint32 *eos_count_ptr;
|
|
|
|
gboolean is_linked;
|
2007-10-25 15:14:02 +00:00
|
|
|
gboolean first_buf;
|
plugins/elements/gstmultiqueue.*: Take the multiqueue lock when updating the fill level so we don't get confused.
Original commit message from CVS:
* 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.
2007-06-26 14:45:15 +00:00
|
|
|
gint n_linked;
|
|
|
|
|
|
|
|
GMutex *mutex;
|
|
|
|
GCond *cond;
|
|
|
|
};
|
|
|
|
|
|
|
|
static GstFlowReturn
|
|
|
|
mq_dummypad_chain (GstPad * sinkpad, GstBuffer * buf)
|
|
|
|
{
|
|
|
|
guint32 cur_id;
|
|
|
|
struct PadData *pad_data;
|
|
|
|
|
|
|
|
pad_data = gst_pad_get_element_private (sinkpad);
|
|
|
|
|
2007-07-16 16:04:49 +00:00
|
|
|
g_static_mutex_lock (&_check_lock);
|
|
|
|
fail_if (pad_data == NULL);
|
plugins/elements/gstmultiqueue.*: Take the multiqueue lock when updating the fill level so we don't get confused.
Original commit message from CVS:
* 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.
2007-06-26 14:45:15 +00:00
|
|
|
/* Read an ID from the first 4 bytes of the buffer data and check it's
|
|
|
|
* what we expect */
|
|
|
|
fail_unless (GST_BUFFER_SIZE (buf) >= 4);
|
2007-07-16 16:04:49 +00:00
|
|
|
g_static_mutex_unlock (&_check_lock);
|
plugins/elements/gstmultiqueue.*: Take the multiqueue lock when updating the fill level so we don't get confused.
Original commit message from CVS:
* 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.
2007-06-26 14:45:15 +00:00
|
|
|
cur_id = GST_READ_UINT32_BE (GST_BUFFER_DATA (buf));
|
|
|
|
|
|
|
|
g_mutex_lock (pad_data->mutex);
|
|
|
|
|
|
|
|
/* For not-linked pads, ensure that we're not running ahead of the 'linked'
|
2007-10-25 15:14:02 +00:00
|
|
|
* pads. The first buffer is allowed to get ahead, because otherwise things can't
|
|
|
|
* always pre-roll correctly */
|
plugins/elements/gstmultiqueue.*: Take the multiqueue lock when updating the fill level so we don't get confused.
Original commit message from CVS:
* 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.
2007-06-26 14:45:15 +00:00
|
|
|
if (!pad_data->is_linked) {
|
|
|
|
/* If there are no linked pads, we can't track a max_id for them :) */
|
2007-10-25 15:14:02 +00:00
|
|
|
if (pad_data->n_linked > 0 && !pad_data->first_buf) {
|
2007-07-16 16:04:49 +00:00
|
|
|
g_static_mutex_lock (&_check_lock);
|
plugins/elements/gstmultiqueue.*: Take the multiqueue lock when updating the fill level so we don't get confused.
Original commit message from CVS:
* 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.
2007-06-26 14:45:15 +00:00
|
|
|
fail_unless (cur_id <= *(pad_data->max_linked_id_ptr) + 1,
|
|
|
|
"Got buffer %u on pad %u before buffer %u was seen on a "
|
|
|
|
"linked pad (max: %u)", cur_id, pad_data->pad_num, cur_id - 1,
|
|
|
|
*(pad_data->max_linked_id_ptr));
|
2007-07-16 16:04:49 +00:00
|
|
|
g_static_mutex_unlock (&_check_lock);
|
plugins/elements/gstmultiqueue.*: Take the multiqueue lock when updating the fill level so we don't get confused.
Original commit message from CVS:
* 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.
2007-06-26 14:45:15 +00:00
|
|
|
}
|
|
|
|
} else {
|
|
|
|
/* Update the max_id value */
|
|
|
|
if (cur_id > *(pad_data->max_linked_id_ptr))
|
|
|
|
*(pad_data->max_linked_id_ptr) = cur_id;
|
|
|
|
}
|
2007-10-25 15:14:02 +00:00
|
|
|
pad_data->first_buf = FALSE;
|
plugins/elements/gstmultiqueue.*: Take the multiqueue lock when updating the fill level so we don't get confused.
Original commit message from CVS:
* 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.
2007-06-26 14:45:15 +00:00
|
|
|
|
|
|
|
g_mutex_unlock (pad_data->mutex);
|
|
|
|
|
|
|
|
/* Unref the buffer */
|
|
|
|
gst_buffer_unref (buf);
|
|
|
|
|
|
|
|
/* Return OK or not-linked as indicated */
|
|
|
|
return pad_data->is_linked ? GST_FLOW_OK : GST_FLOW_NOT_LINKED;
|
|
|
|
}
|
|
|
|
|
|
|
|
static gboolean
|
|
|
|
mq_dummypad_event (GstPad * sinkpad, GstEvent * event)
|
|
|
|
{
|
|
|
|
struct PadData *pad_data;
|
|
|
|
|
|
|
|
pad_data = gst_pad_get_element_private (sinkpad);
|
2007-07-16 16:04:49 +00:00
|
|
|
g_static_mutex_lock (&_check_lock);
|
plugins/elements/gstmultiqueue.*: Take the multiqueue lock when updating the fill level so we don't get confused.
Original commit message from CVS:
* 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.
2007-06-26 14:45:15 +00:00
|
|
|
fail_if (pad_data == NULL);
|
2007-07-16 16:04:49 +00:00
|
|
|
g_static_mutex_unlock (&_check_lock);
|
plugins/elements/gstmultiqueue.*: Take the multiqueue lock when updating the fill level so we don't get confused.
Original commit message from CVS:
* 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.
2007-06-26 14:45:15 +00:00
|
|
|
|
|
|
|
if (GST_EVENT_TYPE (event) == GST_EVENT_EOS) {
|
|
|
|
g_mutex_lock (pad_data->mutex);
|
|
|
|
|
|
|
|
/* Accumulate that we've seen the EOS and signal the main thread */
|
|
|
|
*(pad_data->eos_count_ptr) += 1;
|
|
|
|
|
|
|
|
GST_DEBUG ("EOS on pad %u", pad_data->pad_num);
|
|
|
|
|
|
|
|
g_cond_broadcast (pad_data->cond);
|
|
|
|
g_mutex_unlock (pad_data->mutex);
|
|
|
|
}
|
|
|
|
|
|
|
|
gst_event_unref (event);
|
|
|
|
return TRUE;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
run_output_order_test (gint n_linked)
|
|
|
|
{
|
|
|
|
/* This test creates a multiqueue with 2 linked output, and 3 outputs that
|
|
|
|
* return 'not-linked' when data is pushed, then verifies that all buffers
|
|
|
|
* are received on not-linked pads only after earlier buffers on the
|
|
|
|
* 'linked' pads are made */
|
|
|
|
GstElement *pipe;
|
|
|
|
GstElement *mq;
|
|
|
|
GstPad *inputpads[5];
|
|
|
|
GstPad *sinkpads[5];
|
|
|
|
struct PadData pad_data[5];
|
|
|
|
guint32 max_linked_id;
|
|
|
|
guint32 eos_seen;
|
|
|
|
GMutex *mutex;
|
|
|
|
GCond *cond;
|
|
|
|
gint i;
|
|
|
|
const gint NPADS = 5;
|
|
|
|
const gint NBUFFERS = 1000;
|
|
|
|
|
|
|
|
mutex = g_mutex_new ();
|
|
|
|
cond = g_cond_new ();
|
|
|
|
|
|
|
|
pipe = gst_bin_new ("testbin");
|
|
|
|
|
|
|
|
mq = gst_element_factory_make ("multiqueue", NULL);
|
|
|
|
fail_unless (mq != NULL);
|
|
|
|
gst_bin_add (GST_BIN (pipe), mq);
|
|
|
|
|
2007-10-25 15:14:02 +00:00
|
|
|
/* No limits */
|
|
|
|
g_object_set (mq,
|
|
|
|
"max-size-bytes", (guint) 0,
|
|
|
|
"max-size-buffers", (guint) 0,
|
|
|
|
"max-size-time", (guint64) 0,
|
|
|
|
"extra-size-bytes", (guint) 0,
|
|
|
|
"extra-size-buffers", (guint) 0, "extra-size-time", (guint64) 0, NULL);
|
|
|
|
|
plugins/elements/gstmultiqueue.*: Take the multiqueue lock when updating the fill level so we don't get confused.
Original commit message from CVS:
* 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.
2007-06-26 14:45:15 +00:00
|
|
|
/* Construct NPADS dummy output pads. The first 'n_linked' return FLOW_OK, the rest
|
|
|
|
* return NOT_LINKED. The not-linked ones check the expected ordering of
|
|
|
|
* output buffers */
|
|
|
|
for (i = 0; i < NPADS; i++) {
|
|
|
|
GstPad *mq_srcpad, *mq_sinkpad;
|
|
|
|
gchar *name;
|
|
|
|
|
|
|
|
name = g_strdup_printf ("dummysrc%d", i);
|
|
|
|
inputpads[i] = gst_pad_new (name, GST_PAD_SRC);
|
|
|
|
g_free (name);
|
|
|
|
gst_pad_set_getcaps_function (inputpads[i], mq_dummypad_getcaps);
|
|
|
|
|
|
|
|
mq_sinkpad = gst_element_get_request_pad (mq, "sink%d");
|
|
|
|
fail_unless (mq_sinkpad != NULL);
|
|
|
|
gst_pad_link (inputpads[i], mq_sinkpad);
|
|
|
|
|
|
|
|
gst_pad_set_active (inputpads[i], TRUE);
|
|
|
|
|
|
|
|
mq_srcpad = mq_sinkpad_to_srcpad (mq, mq_sinkpad);
|
|
|
|
|
|
|
|
name = g_strdup_printf ("dummysink%d", i);
|
|
|
|
sinkpads[i] = gst_pad_new (name, GST_PAD_SINK);
|
|
|
|
g_free (name);
|
|
|
|
gst_pad_set_chain_function (sinkpads[i], mq_dummypad_chain);
|
|
|
|
gst_pad_set_event_function (sinkpads[i], mq_dummypad_event);
|
|
|
|
gst_pad_set_getcaps_function (sinkpads[i], mq_dummypad_getcaps);
|
|
|
|
|
|
|
|
pad_data[i].pad_num = i;
|
|
|
|
pad_data[i].max_linked_id_ptr = &max_linked_id;
|
|
|
|
pad_data[i].eos_count_ptr = &eos_seen;
|
|
|
|
pad_data[i].is_linked = (i < n_linked ? TRUE : FALSE);
|
|
|
|
pad_data[i].n_linked = n_linked;
|
|
|
|
pad_data[i].cond = cond;
|
|
|
|
pad_data[i].mutex = mutex;
|
2007-10-25 15:14:02 +00:00
|
|
|
pad_data[i].first_buf = TRUE;
|
plugins/elements/gstmultiqueue.*: Take the multiqueue lock when updating the fill level so we don't get confused.
Original commit message from CVS:
* 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.
2007-06-26 14:45:15 +00:00
|
|
|
gst_pad_set_element_private (sinkpads[i], pad_data + i);
|
|
|
|
|
|
|
|
gst_pad_link (mq_srcpad, sinkpads[i]);
|
|
|
|
gst_pad_set_active (sinkpads[i], TRUE);
|
|
|
|
|
|
|
|
gst_object_unref (mq_sinkpad);
|
|
|
|
gst_object_unref (mq_srcpad);
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Run the test. Push 1000 buffers through the multiqueue in a pattern */
|
|
|
|
|
|
|
|
max_linked_id = 0;
|
|
|
|
eos_seen = 0;
|
|
|
|
gst_element_set_state (pipe, GST_STATE_PLAYING);
|
|
|
|
|
|
|
|
for (i = 0; i < NBUFFERS; i++) {
|
|
|
|
const guint8 pad_pattern[] =
|
|
|
|
{ 0, 0, 0, 0, 1, 1, 2, 1, 0, 2, 3, 2, 3, 1, 4 };
|
|
|
|
const guint n = sizeof (pad_pattern) / sizeof (guint8);
|
|
|
|
guint8 cur_pad;
|
|
|
|
GstBuffer *buf;
|
|
|
|
GstFlowReturn ret;
|
|
|
|
|
|
|
|
cur_pad = pad_pattern[i % n];
|
|
|
|
|
|
|
|
buf = gst_buffer_new_and_alloc (4);
|
2007-07-16 16:04:49 +00:00
|
|
|
g_static_mutex_lock (&_check_lock);
|
plugins/elements/gstmultiqueue.*: Take the multiqueue lock when updating the fill level so we don't get confused.
Original commit message from CVS:
* 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.
2007-06-26 14:45:15 +00:00
|
|
|
fail_if (buf == NULL);
|
2007-07-16 16:04:49 +00:00
|
|
|
g_static_mutex_unlock (&_check_lock);
|
2007-10-25 15:14:02 +00:00
|
|
|
GST_WRITE_UINT32_BE (GST_BUFFER_DATA (buf), i + 1);
|
|
|
|
GST_BUFFER_TIMESTAMP (buf) = (i + 1) * GST_SECOND;
|
plugins/elements/gstmultiqueue.*: Take the multiqueue lock when updating the fill level so we don't get confused.
Original commit message from CVS:
* 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.
2007-06-26 14:45:15 +00:00
|
|
|
|
|
|
|
ret = gst_pad_push (inputpads[cur_pad], buf);
|
2007-07-16 16:04:49 +00:00
|
|
|
g_static_mutex_lock (&_check_lock);
|
plugins/elements/gstmultiqueue.*: Take the multiqueue lock when updating the fill level so we don't get confused.
Original commit message from CVS:
* 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.
2007-06-26 14:45:15 +00:00
|
|
|
if (pad_data[cur_pad].is_linked) {
|
|
|
|
fail_unless (ret == GST_FLOW_OK,
|
|
|
|
"Push on pad %d returned %d when FLOW_OK was expected", cur_pad, ret);
|
|
|
|
} else {
|
|
|
|
/* Expect OK initially, then NOT_LINKED when the srcpad starts pushing */
|
|
|
|
fail_unless (ret == GST_FLOW_OK || ret == GST_FLOW_NOT_LINKED,
|
|
|
|
"Push on pad %d returned %d when FLOW_OK or NOT_LINKED was expected",
|
|
|
|
cur_pad, ret);
|
|
|
|
}
|
2007-07-16 16:04:49 +00:00
|
|
|
g_static_mutex_unlock (&_check_lock);
|
plugins/elements/gstmultiqueue.*: Take the multiqueue lock when updating the fill level so we don't get confused.
Original commit message from CVS:
* 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.
2007-06-26 14:45:15 +00:00
|
|
|
}
|
|
|
|
for (i = 0; i < NPADS; i++) {
|
|
|
|
gst_pad_push_event (inputpads[i], gst_event_new_eos ());
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Wait while the buffers are processed */
|
|
|
|
g_mutex_lock (mutex);
|
2010-02-10 13:40:17 +00:00
|
|
|
/* We wait until EOS has been pushed on all linked pads */
|
|
|
|
while (eos_seen < n_linked) {
|
plugins/elements/gstmultiqueue.*: Take the multiqueue lock when updating the fill level so we don't get confused.
Original commit message from CVS:
* 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.
2007-06-26 14:45:15 +00:00
|
|
|
g_cond_wait (cond, mutex);
|
|
|
|
}
|
|
|
|
g_mutex_unlock (mutex);
|
|
|
|
|
|
|
|
/* Clean up */
|
2010-02-10 13:40:17 +00:00
|
|
|
for (i = 0; i < NPADS; i++) {
|
plugins/elements/gstmultiqueue.*: Take the multiqueue lock when updating the fill level so we don't get confused.
Original commit message from CVS:
* 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.
2007-06-26 14:45:15 +00:00
|
|
|
GstPad *mq_input = gst_pad_get_peer (inputpads[i]);
|
|
|
|
|
|
|
|
gst_pad_unlink (inputpads[i], mq_input);
|
|
|
|
gst_element_release_request_pad (mq, mq_input);
|
|
|
|
gst_object_unref (mq_input);
|
|
|
|
gst_object_unref (inputpads[i]);
|
|
|
|
|
|
|
|
gst_object_unref (sinkpads[i]);
|
|
|
|
}
|
|
|
|
|
|
|
|
gst_element_set_state (pipe, GST_STATE_NULL);
|
|
|
|
gst_object_unref (pipe);
|
|
|
|
|
|
|
|
g_cond_free (cond);
|
|
|
|
g_mutex_free (mutex);
|
|
|
|
}
|
|
|
|
|
|
|
|
GST_START_TEST (test_output_order)
|
|
|
|
{
|
|
|
|
run_output_order_test (2);
|
|
|
|
run_output_order_test (0);
|
|
|
|
}
|
|
|
|
|
|
|
|
GST_END_TEST;
|
|
|
|
|
2007-03-28 18:38:11 +00:00
|
|
|
static Suite *
|
|
|
|
multiqueue_suite (void)
|
|
|
|
{
|
|
|
|
Suite *s = suite_create ("multiqueue");
|
|
|
|
TCase *tc_chain = tcase_create ("general");
|
|
|
|
|
|
|
|
suite_add_tcase (s, tc_chain);
|
|
|
|
tcase_add_test (tc_chain, test_simple_create_destroy);
|
|
|
|
tcase_add_test (tc_chain, test_simple_pipeline);
|
2007-06-06 18:11:10 +00:00
|
|
|
tcase_add_test (tc_chain, test_simple_shutdown_while_running);
|
2007-03-28 18:38:11 +00:00
|
|
|
|
|
|
|
/* FIXME: test_request_pads() needs some more fixes, see comments there */
|
|
|
|
tcase_add_test (tc_chain, test_request_pads);
|
|
|
|
|
plugins/elements/gstmultiqueue.*: Take the multiqueue lock when updating the fill level so we don't get confused.
Original commit message from CVS:
* 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.
2007-06-26 14:45:15 +00:00
|
|
|
tcase_add_test (tc_chain, test_output_order);
|
2007-03-28 18:38:11 +00:00
|
|
|
return s;
|
|
|
|
}
|
|
|
|
|
|
|
|
GST_CHECK_MAIN (multiqueue)
|