Attachment 82132 Details for Bug 65703 – 0001-protocol-http-Add-missing-rewind-callback-for-source.patch

[patch] 0001-protocol-http-Add-missing-rewind-callback-for-source.patch

0001-protocol-http-Add-missing-rewind-callback-for-source.patch (text/plain), 3.41 KB, created by Wang Xingchao on 2013-07-07 12:00:58 UTC

(hide)

Description:

Filename:

MIME Type:

Creator: Wang Xingchao

Created: 2013-07-07 12:00:58 UTC

Size: 3.41 KB

patch

obsolete

>From 5cfeda9077946c21ea2766adf6a17e087644c10a Mon Sep 17 00:00:00 2001
>From: Wang xingchao <wangxingchao2011@gmail.com>
>Date: Sun, 7 Jul 2013 19:31:27 +0800
>Subject: [PATCH 1/2] protocol-http: Add missing rewind callback for
> source-output
>
>Ths missing process_rewind callback would cause remain stream
>continous playing out even user started to "Pause" or "Stop" the stream.
>This patch tried to flush the memblockq instead of only rewinding
>for specific bytes.
>
>Test background:
>
>load module-http-protocol-tcp, and you would see the sinks/sources through
>browser at address: http://localhost:4714.
>
>step 1) in first console
>gst-launch-0.10 souphttpsrc
>location=http://localhost:4714/listen/source/upnp.monitor ! decodebin2 !
>pulsesink device=alsa_output.pci-0000_00_1b.0.analog-stereo
>
>gst-launch fetch data in http-protocol from data source upnp.monitor and send
>back to sink through new pipeline.
>
>The data map would be like:
>source unpn.monitor --> http-protocol --> gst-launch client --> sink-input 0
>--> PA --> sink alsa_output.pci-0000_00_1b.0.analog-stereo
>
>step 2) in second console
>gst-launch-0.10 filesrc location=/home/tangtang/Downloads/b.mp3 ! mad !
> audioconvert ! audioresample ! pulsesink device=upnp
>
>Then you would hear music after step 2.
>
>The data map would be like:
>mp3 data stream --> sink-input 1 --> PA --> sink upnp --> source unpn.monitor
>
>Terminating commands in step 2) would cause rewind request.
>
>Besides of current implementation, i would prefer two more proposals:
>- Ask for rewind request to sink-inputs in next stage.
>In above scenario, it should be sink_input 0.
>- Stop do_work() calling after rewind request.
>Currently do_work() would continue to write the whole memblockq data to client
>even the steam stopped already. It's better as running state of pipeline and
>pause do_work().
>
>Bug fix for 65703:
>https://bugs.freedesktop.org/show_bug.cgi?id=65703
>
>Signed-off-by: Wang xingchao <wangxingchao2011@gmail.com>
>---
> src/pulsecore/protocol-http.c |   15 +++++++++++++++
> 1 file changed, 15 insertions(+)
>
>diff --git a/src/pulsecore/protocol-http.c b/src/pulsecore/protocol-http.c
>index 7d49bf8..c1dd90b 100644
>--- a/src/pulsecore/protocol-http.c
>+++ b/src/pulsecore/protocol-http.c
>@@ -229,6 +229,20 @@ static void source_output_push_cb(pa_source_output *o, const pa_memchunk *chunk)
>     pa_asyncmsgq_post(pa_thread_mq_get()->outq, PA_MSGOBJECT(o), SOURCE_OUTPUT_MESSAGE_POST_DATA, NULL, 0, chunk, NULL);
> }
> 
>+static void source_output_process_rewind_cb(pa_source_output *o, size_t nbytes) {
>+    struct connection *c;
>+
>+    pa_source_output_assert_ref(o);
>+    pa_source_output_assert_io_context(o);
>+    pa_assert_se(c = o->userdata);
>+
>+    /* Stop sending out data for client */
>+    pa_memblockq_flush_write(c->output_memblockq, TRUE);
>+
>+    pa_log_debug("Http source-output rewind (%lld) %lld", (long long) nbytes,
>+        (long long) pa_memblockq_get_length (c->output_memblockq));
>+}
>+
> /* Called from main context */
> static void source_output_kill_cb(pa_source_output *o) {
>     struct connection*c;
>@@ -575,6 +589,7 @@ static void handle_listen_prefix(struct connection *c, const char *source_name)
>     }
> 
>     c->source_output->parent.process_msg = source_output_process_msg;
>+    c->source_output->process_rewind = source_output_process_rewind_cb;
>     c->source_output->push = source_output_push_cb;
>     c->source_output->kill = source_output_kill_cb;
>     c->source_output->get_latency = source_output_get_latency_cb;
>-- 
>1.7.9.5
>

Actions: View | Splinter Review

Attachments on bug 65703: 82062 | 82081 | 82082 | 82117 | 82132 | 82133

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.