gstreamer/ext/avtp/gstavtpsink.c

485 lines
14 KiB
C
Raw Normal View History

/*
* GStreamer AVTP Plugin
* Copyright (C) 2019 Intel Corporation
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 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
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with this library; if not, write to the
* Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
* Boston, MA 02110-1301 USA
*/
/**
* SECTION:element-avtpsink
* @see_also: avtpsrc
*
* avtpsink is a network sink that sends AVTPDUs to the network. It should be
* combined with AVTP payloaders to implement an AVTP talker. For more
* information see https://standards.ieee.org/standard/1722-2016.html.
*
* <note>
* This element opens an AF_PACKET socket which requires CAP_NET_RAW
* capability. Therefore, applications must have that capability in order to
* successfully use this element. The capability can be dropped by the
* application after the element transitions to PAUSED state if wanted.
* </note>
*
* <refsect2>
* <title>Example pipeline</title>
* |[
* gst-launch-1.0 audiotestsrc ! audioconvert ! avtpaafpay ! avtpsink
* ]| This example pipeline implements an AVTP talker that transmit an AAF
* stream.
* </refsect2>
*/
#include <arpa/inet.h>
#include <linux/errqueue.h>
#include <linux/if_packet.h>
avtpsink: Implement synchronization mechanism The avtpsink element is expected to transmit AVTPDUs at specific times, according to GstBuffer timestamps. Currently, the transmission time is controlled in software via the rendering synchronization mechanism provided by GstBaseSink class. However, that mechanism may not cope with some AVB use-cases such as Class A streams, where AVTPDUs are expected to be transmitted at every 125 us. Thus, this patch introduces avtpsink own mechanism which leverages the socket transmission scheduling infrastructure introduced in Linux kernel 4.19. When supported by the NIC, the transmission scheduling is offloaded to the hardware, improving transmission time accuracy considerably. To illustrate that, a before-after experiment was carried out. The experimental setup consisted in 2 PCs with Intel i210 card connected back-to-back running an up-to-date Archlinux with kernel 5.3.1. In one host gst-launch-1.0 was used to generate a 2-minute Class A stream while the other host captured the packets. The metric under evaluation is the transmission interval and it is measured by checking the 'time_delta' information from ethernet frames captured at the receiving side. The table below shows the outcome for a 48 kHz, 16-bit sample, stereo audio stream. The unit is nanoseconds. | Mean | Stdev | Min | Max | Range | -------+--------+---------+---------+---------+---------+ Before | 125000 │ 2401 │ 110056 │ 288432 │ 178376 | After | 125000 │ 18 │ 124943 │ 125055 │ 112 | Before this patch, the transmission interval mean is equal to the optimal value (Class A stream -> 125 us interval), and it is kept the same after the patch. The dispersion measurements, however, had improved considerably, meaning the system is now consistently transmitting AVTPDUs at the correct time. Finally, the socket transmission scheduling infrastructure requires the system clock to be synchronized with PTP clock so this patches modifies the AVTP plugin documentation to cover how to achieve that.
2019-10-04 18:39:10 +00:00
#include <linux/net_tstamp.h>
#include <net/ethernet.h>
#include <net/if.h>
#include <stdio.h>
#include <string.h>
#include <sys/ioctl.h>
#include <sys/socket.h>
#include <unistd.h>
#include "gstavtpsink.h"
GST_DEBUG_CATEGORY_STATIC (avtpsink_debug);
#define GST_CAT_DEFAULT (avtpsink_debug)
#define DEFAULT_IFNAME "eth0"
#define DEFAULT_ADDRESS "01:AA:AA:AA:AA:AA"
#define DEFAULT_PRIORITY 0
avtpsink: Implement synchronization mechanism The avtpsink element is expected to transmit AVTPDUs at specific times, according to GstBuffer timestamps. Currently, the transmission time is controlled in software via the rendering synchronization mechanism provided by GstBaseSink class. However, that mechanism may not cope with some AVB use-cases such as Class A streams, where AVTPDUs are expected to be transmitted at every 125 us. Thus, this patch introduces avtpsink own mechanism which leverages the socket transmission scheduling infrastructure introduced in Linux kernel 4.19. When supported by the NIC, the transmission scheduling is offloaded to the hardware, improving transmission time accuracy considerably. To illustrate that, a before-after experiment was carried out. The experimental setup consisted in 2 PCs with Intel i210 card connected back-to-back running an up-to-date Archlinux with kernel 5.3.1. In one host gst-launch-1.0 was used to generate a 2-minute Class A stream while the other host captured the packets. The metric under evaluation is the transmission interval and it is measured by checking the 'time_delta' information from ethernet frames captured at the receiving side. The table below shows the outcome for a 48 kHz, 16-bit sample, stereo audio stream. The unit is nanoseconds. | Mean | Stdev | Min | Max | Range | -------+--------+---------+---------+---------+---------+ Before | 125000 │ 2401 │ 110056 │ 288432 │ 178376 | After | 125000 │ 18 │ 124943 │ 125055 │ 112 | Before this patch, the transmission interval mean is equal to the optimal value (Class A stream -> 125 us interval), and it is kept the same after the patch. The dispersion measurements, however, had improved considerably, meaning the system is now consistently transmitting AVTPDUs at the correct time. Finally, the socket transmission scheduling infrastructure requires the system clock to be synchronized with PTP clock so this patches modifies the AVTP plugin documentation to cover how to achieve that.
2019-10-04 18:39:10 +00:00
#define NSEC_PER_SEC 1000000000
#define TAI_OFFSET (37ULL * NSEC_PER_SEC)
#define UTC_TO_TAI(t) (t + TAI_OFFSET)
enum
{
PROP_0,
PROP_IFNAME,
PROP_ADDRESS,
PROP_PRIORITY,
};
static GstStaticPadTemplate sink_template = GST_STATIC_PAD_TEMPLATE ("sink",
GST_PAD_SINK,
GST_PAD_ALWAYS,
GST_STATIC_CAPS ("application/x-avtp")
);
#define gst_avtp_sink_parent_class parent_class
G_DEFINE_TYPE (GstAvtpSink, gst_avtp_sink, GST_TYPE_BASE_SINK);
static void gst_avtp_sink_finalize (GObject * gobject);
static void gst_avtp_sink_set_property (GObject * object, guint prop_id,
const GValue * value, GParamSpec * pspec);
static void gst_avtp_sink_get_property (GObject * object, guint prop_id,
GValue * value, GParamSpec * pspec);
static gboolean gst_avtp_sink_start (GstBaseSink * basesink);
static gboolean gst_avtp_sink_stop (GstBaseSink * basesink);
static GstFlowReturn gst_avtp_sink_render (GstBaseSink * basesink, GstBuffer *
buffer);
avtpsink: Implement synchronization mechanism The avtpsink element is expected to transmit AVTPDUs at specific times, according to GstBuffer timestamps. Currently, the transmission time is controlled in software via the rendering synchronization mechanism provided by GstBaseSink class. However, that mechanism may not cope with some AVB use-cases such as Class A streams, where AVTPDUs are expected to be transmitted at every 125 us. Thus, this patch introduces avtpsink own mechanism which leverages the socket transmission scheduling infrastructure introduced in Linux kernel 4.19. When supported by the NIC, the transmission scheduling is offloaded to the hardware, improving transmission time accuracy considerably. To illustrate that, a before-after experiment was carried out. The experimental setup consisted in 2 PCs with Intel i210 card connected back-to-back running an up-to-date Archlinux with kernel 5.3.1. In one host gst-launch-1.0 was used to generate a 2-minute Class A stream while the other host captured the packets. The metric under evaluation is the transmission interval and it is measured by checking the 'time_delta' information from ethernet frames captured at the receiving side. The table below shows the outcome for a 48 kHz, 16-bit sample, stereo audio stream. The unit is nanoseconds. | Mean | Stdev | Min | Max | Range | -------+--------+---------+---------+---------+---------+ Before | 125000 │ 2401 │ 110056 │ 288432 │ 178376 | After | 125000 │ 18 │ 124943 │ 125055 │ 112 | Before this patch, the transmission interval mean is equal to the optimal value (Class A stream -> 125 us interval), and it is kept the same after the patch. The dispersion measurements, however, had improved considerably, meaning the system is now consistently transmitting AVTPDUs at the correct time. Finally, the socket transmission scheduling infrastructure requires the system clock to be synchronized with PTP clock so this patches modifies the AVTP plugin documentation to cover how to achieve that.
2019-10-04 18:39:10 +00:00
static void gst_avtp_sink_get_times (GstBaseSink * bsink, GstBuffer * buffer,
GstClockTime * start, GstClockTime * end);
static void
gst_avtp_sink_class_init (GstAvtpSinkClass * klass)
{
GObjectClass *object_class = G_OBJECT_CLASS (klass);
GstElementClass *element_class = GST_ELEMENT_CLASS (klass);
GstBaseSinkClass *basesink_class = GST_BASE_SINK_CLASS (klass);
object_class->finalize = gst_avtp_sink_finalize;
object_class->get_property = gst_avtp_sink_get_property;
object_class->set_property = gst_avtp_sink_set_property;
g_object_class_install_property (object_class, PROP_IFNAME,
g_param_spec_string ("ifname", "Interface Name",
"Network interface utilized to transmit AVTPDUs",
DEFAULT_IFNAME, G_PARAM_READWRITE | G_PARAM_STATIC_STRINGS |
GST_PARAM_MUTABLE_READY));
g_object_class_install_property (object_class, PROP_ADDRESS,
g_param_spec_string ("address", "Destination MAC address",
"Destination MAC address from Ethernet frames",
DEFAULT_ADDRESS, G_PARAM_READWRITE | G_PARAM_STATIC_STRINGS |
GST_PARAM_MUTABLE_READY));
g_object_class_install_property (object_class, PROP_PRIORITY,
g_param_spec_int ("priority", "Socket priority",
"Priority configured into socket (SO_PRIORITY)", 0, G_MAXINT,
DEFAULT_PRIORITY, G_PARAM_READWRITE | G_PARAM_STATIC_STRINGS |
GST_PARAM_MUTABLE_READY));
gst_element_class_add_static_pad_template (element_class, &sink_template);
gst_element_class_set_static_metadata (element_class,
"Audio/Video Transport Protocol (AVTP) Sink",
"Sink/Network", "Send AVTPDUs over the network",
"Andre Guedes <andre.guedes@intel.com>");
basesink_class->start = GST_DEBUG_FUNCPTR (gst_avtp_sink_start);
basesink_class->stop = GST_DEBUG_FUNCPTR (gst_avtp_sink_stop);
basesink_class->render = GST_DEBUG_FUNCPTR (gst_avtp_sink_render);
avtpsink: Implement synchronization mechanism The avtpsink element is expected to transmit AVTPDUs at specific times, according to GstBuffer timestamps. Currently, the transmission time is controlled in software via the rendering synchronization mechanism provided by GstBaseSink class. However, that mechanism may not cope with some AVB use-cases such as Class A streams, where AVTPDUs are expected to be transmitted at every 125 us. Thus, this patch introduces avtpsink own mechanism which leverages the socket transmission scheduling infrastructure introduced in Linux kernel 4.19. When supported by the NIC, the transmission scheduling is offloaded to the hardware, improving transmission time accuracy considerably. To illustrate that, a before-after experiment was carried out. The experimental setup consisted in 2 PCs with Intel i210 card connected back-to-back running an up-to-date Archlinux with kernel 5.3.1. In one host gst-launch-1.0 was used to generate a 2-minute Class A stream while the other host captured the packets. The metric under evaluation is the transmission interval and it is measured by checking the 'time_delta' information from ethernet frames captured at the receiving side. The table below shows the outcome for a 48 kHz, 16-bit sample, stereo audio stream. The unit is nanoseconds. | Mean | Stdev | Min | Max | Range | -------+--------+---------+---------+---------+---------+ Before | 125000 │ 2401 │ 110056 │ 288432 │ 178376 | After | 125000 │ 18 │ 124943 │ 125055 │ 112 | Before this patch, the transmission interval mean is equal to the optimal value (Class A stream -> 125 us interval), and it is kept the same after the patch. The dispersion measurements, however, had improved considerably, meaning the system is now consistently transmitting AVTPDUs at the correct time. Finally, the socket transmission scheduling infrastructure requires the system clock to be synchronized with PTP clock so this patches modifies the AVTP plugin documentation to cover how to achieve that.
2019-10-04 18:39:10 +00:00
basesink_class->get_times = GST_DEBUG_FUNCPTR (gst_avtp_sink_get_times);
GST_DEBUG_CATEGORY_INIT (avtpsink_debug, "avtpsink", 0, "AVTP Sink");
}
static void
gst_avtp_sink_init (GstAvtpSink * avtpsink)
{
gst_base_sink_set_sync (GST_BASE_SINK (avtpsink), TRUE);
avtpsink->ifname = g_strdup (DEFAULT_IFNAME);
avtpsink->address = g_strdup (DEFAULT_ADDRESS);
avtpsink->priority = DEFAULT_PRIORITY;
avtpsink->sk_fd = -1;
memset (&avtpsink->sk_addr, 0, sizeof (avtpsink->sk_addr));
}
static void
gst_avtp_sink_finalize (GObject * object)
{
GstAvtpSink *avtpsink = GST_AVTP_SINK (object);
g_free (avtpsink->ifname);
g_free (avtpsink->address);
G_OBJECT_CLASS (parent_class)->finalize (object);
}
static void
gst_avtp_sink_set_property (GObject * object, guint prop_id,
const GValue * value, GParamSpec * pspec)
{
GstAvtpSink *avtpsink = GST_AVTP_SINK (object);
GST_DEBUG_OBJECT (avtpsink, "prop_id %u", prop_id);
switch (prop_id) {
case PROP_IFNAME:
g_free (avtpsink->ifname);
avtpsink->ifname = g_value_dup_string (value);
break;
case PROP_ADDRESS:
g_free (avtpsink->address);
avtpsink->address = g_value_dup_string (value);
break;
case PROP_PRIORITY:
avtpsink->priority = g_value_get_int (value);
break;
default:
G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
break;
}
}
static void
gst_avtp_sink_get_property (GObject * object, guint prop_id,
GValue * value, GParamSpec * pspec)
{
GstAvtpSink *avtpsink = GST_AVTP_SINK (object);
GST_DEBUG_OBJECT (avtpsink, "prop_id %u", prop_id);
switch (prop_id) {
case PROP_IFNAME:
g_value_set_string (value, avtpsink->ifname);
break;
case PROP_ADDRESS:
g_value_set_string (value, avtpsink->address);
break;
case PROP_PRIORITY:
g_value_set_int (value, avtpsink->priority);
break;
default:
G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
break;
}
}
static gboolean
gst_avtp_sink_init_socket (GstAvtpSink * avtpsink)
{
int fd, res;
unsigned int index;
guint8 addr[ETH_ALEN];
struct sockaddr_ll sk_addr;
avtpsink: Implement synchronization mechanism The avtpsink element is expected to transmit AVTPDUs at specific times, according to GstBuffer timestamps. Currently, the transmission time is controlled in software via the rendering synchronization mechanism provided by GstBaseSink class. However, that mechanism may not cope with some AVB use-cases such as Class A streams, where AVTPDUs are expected to be transmitted at every 125 us. Thus, this patch introduces avtpsink own mechanism which leverages the socket transmission scheduling infrastructure introduced in Linux kernel 4.19. When supported by the NIC, the transmission scheduling is offloaded to the hardware, improving transmission time accuracy considerably. To illustrate that, a before-after experiment was carried out. The experimental setup consisted in 2 PCs with Intel i210 card connected back-to-back running an up-to-date Archlinux with kernel 5.3.1. In one host gst-launch-1.0 was used to generate a 2-minute Class A stream while the other host captured the packets. The metric under evaluation is the transmission interval and it is measured by checking the 'time_delta' information from ethernet frames captured at the receiving side. The table below shows the outcome for a 48 kHz, 16-bit sample, stereo audio stream. The unit is nanoseconds. | Mean | Stdev | Min | Max | Range | -------+--------+---------+---------+---------+---------+ Before | 125000 │ 2401 │ 110056 │ 288432 │ 178376 | After | 125000 │ 18 │ 124943 │ 125055 │ 112 | Before this patch, the transmission interval mean is equal to the optimal value (Class A stream -> 125 us interval), and it is kept the same after the patch. The dispersion measurements, however, had improved considerably, meaning the system is now consistently transmitting AVTPDUs at the correct time. Finally, the socket transmission scheduling infrastructure requires the system clock to be synchronized with PTP clock so this patches modifies the AVTP plugin documentation to cover how to achieve that.
2019-10-04 18:39:10 +00:00
struct sock_txtime txtime_cfg;
index = if_nametoindex (avtpsink->ifname);
if (!index) {
GST_ERROR_OBJECT (avtpsink, "Failed to get if_index: %s",
g_strerror (errno));
return FALSE;
}
fd = socket (AF_PACKET, SOCK_DGRAM, htons (ETH_P_TSN));
if (fd < 0) {
GST_ERROR_OBJECT (avtpsink, "Failed to open socket: %s",
g_strerror (errno));
return FALSE;
}
res = setsockopt (fd, SOL_SOCKET, SO_PRIORITY, &avtpsink->priority,
sizeof (avtpsink->priority));
if (res < 0) {
GST_ERROR_OBJECT (avtpsink, "Failed to socket priority: %s", g_strerror
(errno));
goto err;
}
avtpsink: Implement synchronization mechanism The avtpsink element is expected to transmit AVTPDUs at specific times, according to GstBuffer timestamps. Currently, the transmission time is controlled in software via the rendering synchronization mechanism provided by GstBaseSink class. However, that mechanism may not cope with some AVB use-cases such as Class A streams, where AVTPDUs are expected to be transmitted at every 125 us. Thus, this patch introduces avtpsink own mechanism which leverages the socket transmission scheduling infrastructure introduced in Linux kernel 4.19. When supported by the NIC, the transmission scheduling is offloaded to the hardware, improving transmission time accuracy considerably. To illustrate that, a before-after experiment was carried out. The experimental setup consisted in 2 PCs with Intel i210 card connected back-to-back running an up-to-date Archlinux with kernel 5.3.1. In one host gst-launch-1.0 was used to generate a 2-minute Class A stream while the other host captured the packets. The metric under evaluation is the transmission interval and it is measured by checking the 'time_delta' information from ethernet frames captured at the receiving side. The table below shows the outcome for a 48 kHz, 16-bit sample, stereo audio stream. The unit is nanoseconds. | Mean | Stdev | Min | Max | Range | -------+--------+---------+---------+---------+---------+ Before | 125000 │ 2401 │ 110056 │ 288432 │ 178376 | After | 125000 │ 18 │ 124943 │ 125055 │ 112 | Before this patch, the transmission interval mean is equal to the optimal value (Class A stream -> 125 us interval), and it is kept the same after the patch. The dispersion measurements, however, had improved considerably, meaning the system is now consistently transmitting AVTPDUs at the correct time. Finally, the socket transmission scheduling infrastructure requires the system clock to be synchronized with PTP clock so this patches modifies the AVTP plugin documentation to cover how to achieve that.
2019-10-04 18:39:10 +00:00
txtime_cfg.clockid = CLOCK_TAI;
txtime_cfg.flags = SOF_TXTIME_REPORT_ERRORS;
avtpsink: Implement synchronization mechanism The avtpsink element is expected to transmit AVTPDUs at specific times, according to GstBuffer timestamps. Currently, the transmission time is controlled in software via the rendering synchronization mechanism provided by GstBaseSink class. However, that mechanism may not cope with some AVB use-cases such as Class A streams, where AVTPDUs are expected to be transmitted at every 125 us. Thus, this patch introduces avtpsink own mechanism which leverages the socket transmission scheduling infrastructure introduced in Linux kernel 4.19. When supported by the NIC, the transmission scheduling is offloaded to the hardware, improving transmission time accuracy considerably. To illustrate that, a before-after experiment was carried out. The experimental setup consisted in 2 PCs with Intel i210 card connected back-to-back running an up-to-date Archlinux with kernel 5.3.1. In one host gst-launch-1.0 was used to generate a 2-minute Class A stream while the other host captured the packets. The metric under evaluation is the transmission interval and it is measured by checking the 'time_delta' information from ethernet frames captured at the receiving side. The table below shows the outcome for a 48 kHz, 16-bit sample, stereo audio stream. The unit is nanoseconds. | Mean | Stdev | Min | Max | Range | -------+--------+---------+---------+---------+---------+ Before | 125000 │ 2401 │ 110056 │ 288432 │ 178376 | After | 125000 │ 18 │ 124943 │ 125055 │ 112 | Before this patch, the transmission interval mean is equal to the optimal value (Class A stream -> 125 us interval), and it is kept the same after the patch. The dispersion measurements, however, had improved considerably, meaning the system is now consistently transmitting AVTPDUs at the correct time. Finally, the socket transmission scheduling infrastructure requires the system clock to be synchronized with PTP clock so this patches modifies the AVTP plugin documentation to cover how to achieve that.
2019-10-04 18:39:10 +00:00
res = setsockopt (fd, SOL_SOCKET, SO_TXTIME, &txtime_cfg,
sizeof (txtime_cfg));
if (res < 0) {
GST_ERROR_OBJECT (avtpsink, "Failed to set SO_TXTIME: %s", g_strerror
avtpsink: Implement synchronization mechanism The avtpsink element is expected to transmit AVTPDUs at specific times, according to GstBuffer timestamps. Currently, the transmission time is controlled in software via the rendering synchronization mechanism provided by GstBaseSink class. However, that mechanism may not cope with some AVB use-cases such as Class A streams, where AVTPDUs are expected to be transmitted at every 125 us. Thus, this patch introduces avtpsink own mechanism which leverages the socket transmission scheduling infrastructure introduced in Linux kernel 4.19. When supported by the NIC, the transmission scheduling is offloaded to the hardware, improving transmission time accuracy considerably. To illustrate that, a before-after experiment was carried out. The experimental setup consisted in 2 PCs with Intel i210 card connected back-to-back running an up-to-date Archlinux with kernel 5.3.1. In one host gst-launch-1.0 was used to generate a 2-minute Class A stream while the other host captured the packets. The metric under evaluation is the transmission interval and it is measured by checking the 'time_delta' information from ethernet frames captured at the receiving side. The table below shows the outcome for a 48 kHz, 16-bit sample, stereo audio stream. The unit is nanoseconds. | Mean | Stdev | Min | Max | Range | -------+--------+---------+---------+---------+---------+ Before | 125000 │ 2401 │ 110056 │ 288432 │ 178376 | After | 125000 │ 18 │ 124943 │ 125055 │ 112 | Before this patch, the transmission interval mean is equal to the optimal value (Class A stream -> 125 us interval), and it is kept the same after the patch. The dispersion measurements, however, had improved considerably, meaning the system is now consistently transmitting AVTPDUs at the correct time. Finally, the socket transmission scheduling infrastructure requires the system clock to be synchronized with PTP clock so this patches modifies the AVTP plugin documentation to cover how to achieve that.
2019-10-04 18:39:10 +00:00
(errno));
goto err;
}
res = sscanf (avtpsink->address, "%hhx:%hhx:%hhx:%hhx:%hhx:%hhx",
&addr[0], &addr[1], &addr[2], &addr[3], &addr[4], &addr[5]);
if (res != 6) {
GST_ERROR_OBJECT (avtpsink, "Destination MAC address format not valid");
goto err;
}
sk_addr.sll_family = AF_PACKET;
sk_addr.sll_protocol = htons (ETH_P_TSN);
sk_addr.sll_halen = ETH_ALEN;
sk_addr.sll_ifindex = index;
sk_addr.sll_hatype = 0;
sk_addr.sll_pkttype = 0;
memcpy (sk_addr.sll_addr, addr, ETH_ALEN);
avtpsink->sk_fd = fd;
avtpsink->sk_addr = sk_addr;
return TRUE;
err:
close (fd);
return FALSE;
}
avtpsink: Implement synchronization mechanism The avtpsink element is expected to transmit AVTPDUs at specific times, according to GstBuffer timestamps. Currently, the transmission time is controlled in software via the rendering synchronization mechanism provided by GstBaseSink class. However, that mechanism may not cope with some AVB use-cases such as Class A streams, where AVTPDUs are expected to be transmitted at every 125 us. Thus, this patch introduces avtpsink own mechanism which leverages the socket transmission scheduling infrastructure introduced in Linux kernel 4.19. When supported by the NIC, the transmission scheduling is offloaded to the hardware, improving transmission time accuracy considerably. To illustrate that, a before-after experiment was carried out. The experimental setup consisted in 2 PCs with Intel i210 card connected back-to-back running an up-to-date Archlinux with kernel 5.3.1. In one host gst-launch-1.0 was used to generate a 2-minute Class A stream while the other host captured the packets. The metric under evaluation is the transmission interval and it is measured by checking the 'time_delta' information from ethernet frames captured at the receiving side. The table below shows the outcome for a 48 kHz, 16-bit sample, stereo audio stream. The unit is nanoseconds. | Mean | Stdev | Min | Max | Range | -------+--------+---------+---------+---------+---------+ Before | 125000 │ 2401 │ 110056 │ 288432 │ 178376 | After | 125000 │ 18 │ 124943 │ 125055 │ 112 | Before this patch, the transmission interval mean is equal to the optimal value (Class A stream -> 125 us interval), and it is kept the same after the patch. The dispersion measurements, however, had improved considerably, meaning the system is now consistently transmitting AVTPDUs at the correct time. Finally, the socket transmission scheduling infrastructure requires the system clock to be synchronized with PTP clock so this patches modifies the AVTP plugin documentation to cover how to achieve that.
2019-10-04 18:39:10 +00:00
static void
gst_avtp_sink_init_msghdr (GstAvtpSink * avtpsink)
{
struct msghdr *msg;
struct cmsghdr *cmsg;
msg = g_malloc0 (sizeof (struct msghdr));
msg->msg_name = &avtpsink->sk_addr;
msg->msg_namelen = sizeof (avtpsink->sk_addr);
msg->msg_iovlen = 1;
msg->msg_iov = g_malloc0 (sizeof (struct iovec));
msg->msg_controllen = CMSG_SPACE (sizeof (__u64));
msg->msg_control = g_malloc0 (msg->msg_controllen);
cmsg = CMSG_FIRSTHDR (msg);
cmsg->cmsg_level = SOL_SOCKET;
cmsg->cmsg_type = SCM_TXTIME;
cmsg->cmsg_len = CMSG_LEN (sizeof (__u64));
avtpsink->msg = msg;
}
static gboolean
gst_avtp_sink_start (GstBaseSink * basesink)
{
GstAvtpSink *avtpsink = GST_AVTP_SINK (basesink);
if (!gst_avtp_sink_init_socket (avtpsink))
return FALSE;
avtpsink: Implement synchronization mechanism The avtpsink element is expected to transmit AVTPDUs at specific times, according to GstBuffer timestamps. Currently, the transmission time is controlled in software via the rendering synchronization mechanism provided by GstBaseSink class. However, that mechanism may not cope with some AVB use-cases such as Class A streams, where AVTPDUs are expected to be transmitted at every 125 us. Thus, this patch introduces avtpsink own mechanism which leverages the socket transmission scheduling infrastructure introduced in Linux kernel 4.19. When supported by the NIC, the transmission scheduling is offloaded to the hardware, improving transmission time accuracy considerably. To illustrate that, a before-after experiment was carried out. The experimental setup consisted in 2 PCs with Intel i210 card connected back-to-back running an up-to-date Archlinux with kernel 5.3.1. In one host gst-launch-1.0 was used to generate a 2-minute Class A stream while the other host captured the packets. The metric under evaluation is the transmission interval and it is measured by checking the 'time_delta' information from ethernet frames captured at the receiving side. The table below shows the outcome for a 48 kHz, 16-bit sample, stereo audio stream. The unit is nanoseconds. | Mean | Stdev | Min | Max | Range | -------+--------+---------+---------+---------+---------+ Before | 125000 │ 2401 │ 110056 │ 288432 │ 178376 | After | 125000 │ 18 │ 124943 │ 125055 │ 112 | Before this patch, the transmission interval mean is equal to the optimal value (Class A stream -> 125 us interval), and it is kept the same after the patch. The dispersion measurements, however, had improved considerably, meaning the system is now consistently transmitting AVTPDUs at the correct time. Finally, the socket transmission scheduling infrastructure requires the system clock to be synchronized with PTP clock so this patches modifies the AVTP plugin documentation to cover how to achieve that.
2019-10-04 18:39:10 +00:00
gst_avtp_sink_init_msghdr (avtpsink);
GST_DEBUG_OBJECT (avtpsink, "AVTP sink started");
return TRUE;
}
static gboolean
gst_avtp_sink_stop (GstBaseSink * basesink)
{
GstAvtpSink *avtpsink = GST_AVTP_SINK (basesink);
avtpsink: Implement synchronization mechanism The avtpsink element is expected to transmit AVTPDUs at specific times, according to GstBuffer timestamps. Currently, the transmission time is controlled in software via the rendering synchronization mechanism provided by GstBaseSink class. However, that mechanism may not cope with some AVB use-cases such as Class A streams, where AVTPDUs are expected to be transmitted at every 125 us. Thus, this patch introduces avtpsink own mechanism which leverages the socket transmission scheduling infrastructure introduced in Linux kernel 4.19. When supported by the NIC, the transmission scheduling is offloaded to the hardware, improving transmission time accuracy considerably. To illustrate that, a before-after experiment was carried out. The experimental setup consisted in 2 PCs with Intel i210 card connected back-to-back running an up-to-date Archlinux with kernel 5.3.1. In one host gst-launch-1.0 was used to generate a 2-minute Class A stream while the other host captured the packets. The metric under evaluation is the transmission interval and it is measured by checking the 'time_delta' information from ethernet frames captured at the receiving side. The table below shows the outcome for a 48 kHz, 16-bit sample, stereo audio stream. The unit is nanoseconds. | Mean | Stdev | Min | Max | Range | -------+--------+---------+---------+---------+---------+ Before | 125000 │ 2401 │ 110056 │ 288432 │ 178376 | After | 125000 │ 18 │ 124943 │ 125055 │ 112 | Before this patch, the transmission interval mean is equal to the optimal value (Class A stream -> 125 us interval), and it is kept the same after the patch. The dispersion measurements, however, had improved considerably, meaning the system is now consistently transmitting AVTPDUs at the correct time. Finally, the socket transmission scheduling infrastructure requires the system clock to be synchronized with PTP clock so this patches modifies the AVTP plugin documentation to cover how to achieve that.
2019-10-04 18:39:10 +00:00
g_free (avtpsink->msg->msg_iov);
g_free (avtpsink->msg->msg_control);
g_free (avtpsink->msg);
close (avtpsink->sk_fd);
GST_DEBUG_OBJECT (avtpsink, "AVTP sink stopped");
return TRUE;
}
avtpsink: Implement synchronization mechanism The avtpsink element is expected to transmit AVTPDUs at specific times, according to GstBuffer timestamps. Currently, the transmission time is controlled in software via the rendering synchronization mechanism provided by GstBaseSink class. However, that mechanism may not cope with some AVB use-cases such as Class A streams, where AVTPDUs are expected to be transmitted at every 125 us. Thus, this patch introduces avtpsink own mechanism which leverages the socket transmission scheduling infrastructure introduced in Linux kernel 4.19. When supported by the NIC, the transmission scheduling is offloaded to the hardware, improving transmission time accuracy considerably. To illustrate that, a before-after experiment was carried out. The experimental setup consisted in 2 PCs with Intel i210 card connected back-to-back running an up-to-date Archlinux with kernel 5.3.1. In one host gst-launch-1.0 was used to generate a 2-minute Class A stream while the other host captured the packets. The metric under evaluation is the transmission interval and it is measured by checking the 'time_delta' information from ethernet frames captured at the receiving side. The table below shows the outcome for a 48 kHz, 16-bit sample, stereo audio stream. The unit is nanoseconds. | Mean | Stdev | Min | Max | Range | -------+--------+---------+---------+---------+---------+ Before | 125000 │ 2401 │ 110056 │ 288432 │ 178376 | After | 125000 │ 18 │ 124943 │ 125055 │ 112 | Before this patch, the transmission interval mean is equal to the optimal value (Class A stream -> 125 us interval), and it is kept the same after the patch. The dispersion measurements, however, had improved considerably, meaning the system is now consistently transmitting AVTPDUs at the correct time. Finally, the socket transmission scheduling infrastructure requires the system clock to be synchronized with PTP clock so this patches modifies the AVTP plugin documentation to cover how to achieve that.
2019-10-04 18:39:10 +00:00
/* This function was heavily inspired by gst_base_sink_adjust_time() from
* GstBaseSink.
*/
static GstClockTime
gst_avtp_sink_adjust_time (GstBaseSink * basesink, GstClockTime time)
{
GstClockTimeDiff ts_offset;
GstClockTime render_delay;
/* don't do anything funny with invalid timestamps */
if (G_UNLIKELY (!GST_CLOCK_TIME_IS_VALID (time)))
return time;
time += gst_base_sink_get_latency (basesink);
/* apply offset, be careful for underflows */
ts_offset = gst_base_sink_get_ts_offset (basesink);
if (ts_offset < 0) {
ts_offset = -ts_offset;
if (ts_offset < time)
time -= ts_offset;
else
time = 0;
} else
time += ts_offset;
/* subtract the render delay again, which was included in the latency */
render_delay = gst_base_sink_get_render_delay (basesink);
if (time > render_delay)
time -= render_delay;
else
time = 0;
return time;
}
static void
gst_avtp_sink_process_error_queue (GstAvtpSink * avtpsink, int fd)
{
uint8_t msg_control[CMSG_SPACE (sizeof (struct sock_extended_err))];
unsigned char err_buffer[256];
struct sock_extended_err *serr;
struct cmsghdr *cmsg;
struct iovec iov = {
.iov_base = err_buffer,
.iov_len = sizeof (err_buffer)
};
struct msghdr msg = {
.msg_iov = &iov,
.msg_iovlen = 1,
.msg_control = msg_control,
.msg_controllen = sizeof (msg_control)
};
if (recvmsg (fd, &msg, MSG_ERRQUEUE) == -1) {
GST_LOG_OBJECT (avtpsink, "Could not get socket errqueue: recvmsg failed");
return;
}
cmsg = CMSG_FIRSTHDR (&msg);
while (cmsg != NULL) {
serr = (void *) CMSG_DATA (cmsg);
if (serr->ee_origin == SO_EE_ORIGIN_TXTIME) {
switch (serr->ee_code) {
case SO_EE_CODE_TXTIME_INVALID_PARAM:
case SO_EE_CODE_TXTIME_MISSED:
GST_INFO_OBJECT (avtpsink, "AVTPDU dropped due to being late. "
"Check stream spec and pipeline settings.");
break;
default:
break;
}
return;
}
cmsg = CMSG_NXTHDR (&msg, cmsg);
}
}
static GstFlowReturn
gst_avtp_sink_render (GstBaseSink * basesink, GstBuffer * buffer)
{
ssize_t n;
GstMapInfo info;
GstAvtpSink *avtpsink = GST_AVTP_SINK (basesink);
avtpsink: Implement synchronization mechanism The avtpsink element is expected to transmit AVTPDUs at specific times, according to GstBuffer timestamps. Currently, the transmission time is controlled in software via the rendering synchronization mechanism provided by GstBaseSink class. However, that mechanism may not cope with some AVB use-cases such as Class A streams, where AVTPDUs are expected to be transmitted at every 125 us. Thus, this patch introduces avtpsink own mechanism which leverages the socket transmission scheduling infrastructure introduced in Linux kernel 4.19. When supported by the NIC, the transmission scheduling is offloaded to the hardware, improving transmission time accuracy considerably. To illustrate that, a before-after experiment was carried out. The experimental setup consisted in 2 PCs with Intel i210 card connected back-to-back running an up-to-date Archlinux with kernel 5.3.1. In one host gst-launch-1.0 was used to generate a 2-minute Class A stream while the other host captured the packets. The metric under evaluation is the transmission interval and it is measured by checking the 'time_delta' information from ethernet frames captured at the receiving side. The table below shows the outcome for a 48 kHz, 16-bit sample, stereo audio stream. The unit is nanoseconds. | Mean | Stdev | Min | Max | Range | -------+--------+---------+---------+---------+---------+ Before | 125000 │ 2401 │ 110056 │ 288432 │ 178376 | After | 125000 │ 18 │ 124943 │ 125055 │ 112 | Before this patch, the transmission interval mean is equal to the optimal value (Class A stream -> 125 us interval), and it is kept the same after the patch. The dispersion measurements, however, had improved considerably, meaning the system is now consistently transmitting AVTPDUs at the correct time. Finally, the socket transmission scheduling infrastructure requires the system clock to be synchronized with PTP clock so this patches modifies the AVTP plugin documentation to cover how to achieve that.
2019-10-04 18:39:10 +00:00
struct iovec *iov = avtpsink->msg->msg_iov;
if (G_LIKELY (basesink->sync)) {
GstClockTime base_time, running_time;
struct cmsghdr *cmsg = CMSG_FIRSTHDR (avtpsink->msg);
gint ret;
avtpsink: Implement synchronization mechanism The avtpsink element is expected to transmit AVTPDUs at specific times, according to GstBuffer timestamps. Currently, the transmission time is controlled in software via the rendering synchronization mechanism provided by GstBaseSink class. However, that mechanism may not cope with some AVB use-cases such as Class A streams, where AVTPDUs are expected to be transmitted at every 125 us. Thus, this patch introduces avtpsink own mechanism which leverages the socket transmission scheduling infrastructure introduced in Linux kernel 4.19. When supported by the NIC, the transmission scheduling is offloaded to the hardware, improving transmission time accuracy considerably. To illustrate that, a before-after experiment was carried out. The experimental setup consisted in 2 PCs with Intel i210 card connected back-to-back running an up-to-date Archlinux with kernel 5.3.1. In one host gst-launch-1.0 was used to generate a 2-minute Class A stream while the other host captured the packets. The metric under evaluation is the transmission interval and it is measured by checking the 'time_delta' information from ethernet frames captured at the receiving side. The table below shows the outcome for a 48 kHz, 16-bit sample, stereo audio stream. The unit is nanoseconds. | Mean | Stdev | Min | Max | Range | -------+--------+---------+---------+---------+---------+ Before | 125000 │ 2401 │ 110056 │ 288432 │ 178376 | After | 125000 │ 18 │ 124943 │ 125055 │ 112 | Before this patch, the transmission interval mean is equal to the optimal value (Class A stream -> 125 us interval), and it is kept the same after the patch. The dispersion measurements, however, had improved considerably, meaning the system is now consistently transmitting AVTPDUs at the correct time. Finally, the socket transmission scheduling infrastructure requires the system clock to be synchronized with PTP clock so this patches modifies the AVTP plugin documentation to cover how to achieve that.
2019-10-04 18:39:10 +00:00
g_assert (GST_BUFFER_DTS_OR_PTS (buffer) != GST_CLOCK_TIME_NONE);
ret = gst_segment_to_running_time_full (&basesink->segment,
basesink->segment.format, GST_BUFFER_DTS_OR_PTS (buffer),
&running_time);
if (ret == -1)
running_time = -running_time;
avtpsink: Implement synchronization mechanism The avtpsink element is expected to transmit AVTPDUs at specific times, according to GstBuffer timestamps. Currently, the transmission time is controlled in software via the rendering synchronization mechanism provided by GstBaseSink class. However, that mechanism may not cope with some AVB use-cases such as Class A streams, where AVTPDUs are expected to be transmitted at every 125 us. Thus, this patch introduces avtpsink own mechanism which leverages the socket transmission scheduling infrastructure introduced in Linux kernel 4.19. When supported by the NIC, the transmission scheduling is offloaded to the hardware, improving transmission time accuracy considerably. To illustrate that, a before-after experiment was carried out. The experimental setup consisted in 2 PCs with Intel i210 card connected back-to-back running an up-to-date Archlinux with kernel 5.3.1. In one host gst-launch-1.0 was used to generate a 2-minute Class A stream while the other host captured the packets. The metric under evaluation is the transmission interval and it is measured by checking the 'time_delta' information from ethernet frames captured at the receiving side. The table below shows the outcome for a 48 kHz, 16-bit sample, stereo audio stream. The unit is nanoseconds. | Mean | Stdev | Min | Max | Range | -------+--------+---------+---------+---------+---------+ Before | 125000 │ 2401 │ 110056 │ 288432 │ 178376 | After | 125000 │ 18 │ 124943 │ 125055 │ 112 | Before this patch, the transmission interval mean is equal to the optimal value (Class A stream -> 125 us interval), and it is kept the same after the patch. The dispersion measurements, however, had improved considerably, meaning the system is now consistently transmitting AVTPDUs at the correct time. Finally, the socket transmission scheduling infrastructure requires the system clock to be synchronized with PTP clock so this patches modifies the AVTP plugin documentation to cover how to achieve that.
2019-10-04 18:39:10 +00:00
base_time = gst_element_get_base_time (GST_ELEMENT (avtpsink));
running_time = gst_avtp_sink_adjust_time (basesink, running_time);
*(__u64 *) CMSG_DATA (cmsg) = UTC_TO_TAI (base_time + running_time);
}
if (!gst_buffer_map (buffer, &info, GST_MAP_READ)) {
GST_ERROR_OBJECT (avtpsink, "Failed to map buffer");
return GST_FLOW_ERROR;
}
avtpsink: Implement synchronization mechanism The avtpsink element is expected to transmit AVTPDUs at specific times, according to GstBuffer timestamps. Currently, the transmission time is controlled in software via the rendering synchronization mechanism provided by GstBaseSink class. However, that mechanism may not cope with some AVB use-cases such as Class A streams, where AVTPDUs are expected to be transmitted at every 125 us. Thus, this patch introduces avtpsink own mechanism which leverages the socket transmission scheduling infrastructure introduced in Linux kernel 4.19. When supported by the NIC, the transmission scheduling is offloaded to the hardware, improving transmission time accuracy considerably. To illustrate that, a before-after experiment was carried out. The experimental setup consisted in 2 PCs with Intel i210 card connected back-to-back running an up-to-date Archlinux with kernel 5.3.1. In one host gst-launch-1.0 was used to generate a 2-minute Class A stream while the other host captured the packets. The metric under evaluation is the transmission interval and it is measured by checking the 'time_delta' information from ethernet frames captured at the receiving side. The table below shows the outcome for a 48 kHz, 16-bit sample, stereo audio stream. The unit is nanoseconds. | Mean | Stdev | Min | Max | Range | -------+--------+---------+---------+---------+---------+ Before | 125000 │ 2401 │ 110056 │ 288432 │ 178376 | After | 125000 │ 18 │ 124943 │ 125055 │ 112 | Before this patch, the transmission interval mean is equal to the optimal value (Class A stream -> 125 us interval), and it is kept the same after the patch. The dispersion measurements, however, had improved considerably, meaning the system is now consistently transmitting AVTPDUs at the correct time. Finally, the socket transmission scheduling infrastructure requires the system clock to be synchronized with PTP clock so this patches modifies the AVTP plugin documentation to cover how to achieve that.
2019-10-04 18:39:10 +00:00
iov->iov_base = info.data;
iov->iov_len = info.size;
n = sendmsg (avtpsink->sk_fd, avtpsink->msg, 0);
if (n < 0) {
GST_INFO_OBJECT (avtpsink, "Failed to send AVTPDU: %s", g_strerror (errno));
if (G_LIKELY (basesink->sync))
gst_avtp_sink_process_error_queue (avtpsink, avtpsink->sk_fd);
goto out;
}
if (n != info.size) {
GST_INFO_OBJECT (avtpsink, "Incomplete AVTPDU transmission");
goto out;
}
out:
gst_buffer_unmap (buffer, &info);
return GST_FLOW_OK;
}
avtpsink: Implement synchronization mechanism The avtpsink element is expected to transmit AVTPDUs at specific times, according to GstBuffer timestamps. Currently, the transmission time is controlled in software via the rendering synchronization mechanism provided by GstBaseSink class. However, that mechanism may not cope with some AVB use-cases such as Class A streams, where AVTPDUs are expected to be transmitted at every 125 us. Thus, this patch introduces avtpsink own mechanism which leverages the socket transmission scheduling infrastructure introduced in Linux kernel 4.19. When supported by the NIC, the transmission scheduling is offloaded to the hardware, improving transmission time accuracy considerably. To illustrate that, a before-after experiment was carried out. The experimental setup consisted in 2 PCs with Intel i210 card connected back-to-back running an up-to-date Archlinux with kernel 5.3.1. In one host gst-launch-1.0 was used to generate a 2-minute Class A stream while the other host captured the packets. The metric under evaluation is the transmission interval and it is measured by checking the 'time_delta' information from ethernet frames captured at the receiving side. The table below shows the outcome for a 48 kHz, 16-bit sample, stereo audio stream. The unit is nanoseconds. | Mean | Stdev | Min | Max | Range | -------+--------+---------+---------+---------+---------+ Before | 125000 │ 2401 │ 110056 │ 288432 │ 178376 | After | 125000 │ 18 │ 124943 │ 125055 │ 112 | Before this patch, the transmission interval mean is equal to the optimal value (Class A stream -> 125 us interval), and it is kept the same after the patch. The dispersion measurements, however, had improved considerably, meaning the system is now consistently transmitting AVTPDUs at the correct time. Finally, the socket transmission scheduling infrastructure requires the system clock to be synchronized with PTP clock so this patches modifies the AVTP plugin documentation to cover how to achieve that.
2019-10-04 18:39:10 +00:00
static void
gst_avtp_sink_get_times (GstBaseSink * bsink, GstBuffer * buffer,
GstClockTime * start, GstClockTime * end)
{
/* Rendering synchronization is handled by the GstAvtpSink class itself, not
* GstBaseSink so we set 'start' and 'end' to GST_CLOCK_TIME_NONE to signal
* that to the base class.
*/
*start = GST_CLOCK_TIME_NONE;
*end = GST_CLOCK_TIME_NONE;
}
gboolean
gst_avtp_sink_plugin_init (GstPlugin * plugin)
{
return gst_element_register (plugin, "avtpsink", GST_RANK_NONE,
GST_TYPE_AVTP_SINK);
}