How to do filesrc loop with gst-interpipe?

[pinjiang]

I use SEEK to loop the filesrc with success. But I use interpipe to have two pipeline branches after interpipe, it no longer works. I use seek in the pipeline before interpipe when EOS event, but it doesn't work anymore. The play pipeline get stuck. How to solve it?

#include <stdio.h>
#include <gst/gst.h>

gchar *base_stream = "/home/j00412711/git/pion_benchmark/server/videos/FaceTestVideo.mp4";
gchar *stream = NULL;

static gboolean gstreamer_send_bus_call(GstBus *bus, GstMessage *msg, gpointer data) {
  GstElement *pipeline = GST_ELEMENT(data);

  switch (GST_MESSAGE_TYPE(msg)) {
  case GST_MESSAGE_EOS:
    g_print("EOS");
    if (!gst_element_seek (pipeline, 1.0, GST_FORMAT_TIME, GST_SEEK_FLAG_FLUSH | GST_SEEK_FLAG_KEY_UNIT | GST_SEEK_FLAG_SKIP,
             GST_SEEK_TYPE_SET, 0,
             GST_SEEK_TYPE_NONE, GST_CLOCK_TIME_NONE)) {
        g_print ("EOS restart failed\n");
        exit(1);
    }
    break;

  case GST_MESSAGE_ERROR: {
    gchar *debug;
    GError *error;

    gst_message_parse_error(msg, &error, &debug);
    g_free(debug);

    g_printerr("GStreamer Error: %s\n", error->message);
    g_error_free(error);
    exit(1);
  }
  default:
    break;
  }

  return TRUE;
}

typedef struct
{
    GstPipeline *sink1;
    GstPipeline *src;
    GstElement *intersrc;
} gst_element;

int main(int argc, char *argv[])
{
    GstPipeline *src;
    GstPipeline *sink1;
    GstPipeline *sink2;

    GError *error = NULL;
    GMainLoop *loop;
    gst_init(&argc, &argv);

    gchar *str1 = (gchar *)"filesrc location=";
    gchar *str2 = (gchar *)" name=vtsrc1 ! decodebin name=demux ! queue ! interpipesink name=intersink sync=false async=false";
    stream = g_strconcat(str1, base_stream, str2, NULL);

    src = GST_PIPELINE(gst_parse_launch(stream, &error));
    // vtsrc1 = gst_bin_get_by_name(GST_BIN(sink1), "vtsrc1");

    sink1 =
        GST_PIPELINE(gst_parse_launch("interpipesrc name=intersrc1 listen-to=intersink is-live=true format=time ! "
                                      "autovideosink name=asink1",
                                      &error));
    //intersrc1 = gst_bin_get_by_name(GST_BIN(sink1), "intersrc1");

    sink2 =
        GST_PIPELINE(gst_parse_launch("interpipesrc name=intersrc2 listen-to=intersink is-live=true format=time ! "
                                      "autovideosink name=asink2",
                                      &error));
    //intersrc2 = gst_bin_get_by_name(GST_BIN(sink2), "intersrc2");

    GstBus *bus = gst_pipeline_get_bus(GST_PIPELINE(src));
    gst_bus_add_watch(bus, gstreamer_send_bus_call, src);
    gst_object_unref(bus);

    gst_element_set_state(GST_ELEMENT(src), GST_STATE_PLAYING);
    gst_element_set_state(GST_ELEMENT(sink1), GST_STATE_PLAYING);
    gst_element_set_state(GST_ELEMENT(sink2), GST_STATE_PLAYING);

    loop = g_main_loop_new(NULL, TRUE);

    g_main_loop_run(loop);
    g_main_loop_unref(loop);

    gst_element_set_state(GST_ELEMENT(src), GST_STATE_NULL);
    gst_element_set_state(GST_ELEMENT(sink1), GST_STATE_NULL);
    gst_element_set_state(GST_ELEMENT(sink2), GST_STATE_NULL);

    g_object_unref(src);
    g_object_unref(sink1);
    g_object_unref(sink2);

    return TRUE;

[clogwog]

can you run this with GST_DEBUG=3 ./yourapplication

when i do this i see the following after the EOS printout:

EOS0:00:58.801434302   460   0x5591f21b50 ERROR         v4l2bufferpool gstv4l2bufferpool.c:747:gst_v4l2_buffer_pool_alloc_buffer:<nvv4l2decoder0:pool:src> failed to allocate buffer
0:00:58.801523159   460   0x5591f21b50 WARN              bufferpool gstbufferpool.c:310:do_alloc_buffer:<nvv4l2decoder0:pool:src> alloc function failed
0:00:58.802511059   460   0x7f78008940 ERROR         v4l2bufferpool gstv4l2bufferpool.c:747:gst_v4l2_buffer_pool_alloc_buffer:<nvv4l2decoder0:pool:src> failed to allocate buffer
0:00:58.802552258   460   0x7f78008940 WARN              bufferpool gstbufferpool.c:310:do_alloc_buffer:<nvv4l2decoder0:pool:src> alloc function failed

so it looks like something with the autovideosink not being able to handle the restart ?