2019-01-14 18:18:42 +00:00
|
|
|
/*
|
|
|
|
* 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
|
|
|
|
*/
|
|
|
|
|
|
|
|
/**
|
|
|
|
* plugin-avtp:
|
|
|
|
*
|
|
|
|
* ## Audio Video Transport Protocol (AVTP) Plugin
|
|
|
|
*
|
|
|
|
* The AVTP plugin implements typical Talker and Listener functionalities that
|
|
|
|
* can be leveraged by GStreamer-based applications in order to implement TSN
|
|
|
|
* audio/video applications.
|
|
|
|
*
|
|
|
|
* ### Dependencies
|
|
|
|
*
|
|
|
|
* The plugin uses libavtp to handle AVTP packetization. Libavtp source code can
|
|
|
|
* be found in https://github.com/AVnu/libavtp as well as instructions to build
|
|
|
|
* and install it.
|
|
|
|
*
|
|
|
|
* If libavtp isn't detected by configure, the plugin isn't built.
|
|
|
|
*
|
avtp: Introduce AAF payloader element
This patch introduces the AVTP Audio Format (AAF) payloader element from
the AVTP plugin. The element inputs audio raw data and outputs AVTP
packets (aka AVTPDUs), implementing a typical protocol payloader element
from GStreamer.
AAF is one of the available formats to transport audio data in an AVTP
system. AAF is specified in IEEE 1722-2016 section 7 and provides two
encapsulation mode: PCM and AES3. This patch implements PCM
encapsulation mode only.
The AAF payloader working mechanism consists of building the AAF header,
prepending it to the GstBuffer received on the sink pad, and pushing the
buffer downstream. Payloader parameters such as stream ID, maximum
transit time, time uncertainty, and timestamping mode are passed via
element properties. AAF doesn't support all possible sample format and
sampling rate values so the sink pad caps template from the payloader is
a subset of audio/x-raw. Additionally, this patch implements only
"normal" timestamping mode from AAF. "Sparse" mode should be implemented
in future.
Upcoming patches will introduce other AVTP payloader elements that will
have some common code. For that reason, this patch introduces the
GstAvtpBasePayload abstract class that implements common payloader
functionalities, and the GstAvtpAafPay class that extends the
GstAvtpBasePayload class, implementing AAF-specific functionalities.
The AAF payloader element is most likely to be used with the AVTP sink
element (to be introduced by a later patch) but it could also be used
with UDP sink element to implement AVTP over UDP as described in IEEE
1722-2016 Annex J.
This element was inspired by RTP payloader elements.
2019-01-17 01:16:59 +00:00
|
|
|
* ### The application/x-avtp mime type
|
|
|
|
*
|
|
|
|
* For valid AVTPDUs encapsulated in GstBuffers, we use the caps with mime type
|
|
|
|
* application/x-avtp.
|
|
|
|
*
|
|
|
|
* AVTP mime type is pretty simple and has no fields.
|
|
|
|
*
|
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
|
|
|
* ### gPTP Setup
|
2019-05-17 23:00:24 +00:00
|
|
|
*
|
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
|
|
|
* The Linuxptp project provides the ptp4l daemon, which synchronizes the PTP
|
|
|
|
* clock from NIC, and the pmc tool which communicates with ptp4l to get/set
|
|
|
|
* some runtime settings. The project also provides the phc2sys daemon which
|
|
|
|
* synchronizes the PTP clock and system clock.
|
2019-05-17 23:00:24 +00:00
|
|
|
*
|
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
|
|
|
* The AVTP plugin requires system clock is synchronized with PTP clock and
|
|
|
|
* TAI offset is properly set in the kernel. ptp4l and phc2sys can be set up
|
|
|
|
* in many different ways, below we provide an example that fullfils the plugin
|
|
|
|
* requirements. For further information check ptp4l(8) and phc2sys(8).
|
2019-05-17 23:00:24 +00:00
|
|
|
*
|
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
|
|
|
* In the following instructions, replace $IFNAME by your PTP capable NIC
|
|
|
|
* interface. The gPTP.cfg file mentioned below can be found in /usr/share/
|
|
|
|
* doc/linuxptp/ (depending on your distro).
|
2019-05-17 23:00:24 +00:00
|
|
|
*
|
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
|
|
|
* Synchronize PTP clock with PTP time:
|
2019-05-17 23:00:24 +00:00
|
|
|
*
|
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
|
|
|
* $ ptp4l -f gPTP.cfg -i $IFNAME
|
2019-05-17 23:00:24 +00:00
|
|
|
*
|
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
|
|
|
* Enable TAI offset to be automatically set by phc2sys:
|
|
|
|
*
|
|
|
|
* $ pmc -u -t 1 -b 0 'SET GRANDMASTER_SETTINGS_NP \
|
|
|
|
* clockClass 248 clockAccuracy 0xfe \
|
|
|
|
* offsetScaledLogVariance 0xffff \
|
|
|
|
* currentUtcOffset 37 leap61 0 leap59 0 \
|
|
|
|
* currentUtcOffsetValid 1 ptpTimescale 1 \
|
|
|
|
* timeTraceable 1 frequencyTraceable 0 timeSource 0xa0'
|
|
|
|
*
|
|
|
|
* Synchronize system clock with PTP clock:
|
|
|
|
*
|
|
|
|
* $ phc2sys -f gPTP.cfg -s $IFNAME -c CLOCK_REALTIME -w
|
|
|
|
*
|
|
|
|
* The commands above should be run on both AVTP Talker and Listener hosts.
|
|
|
|
*
|
|
|
|
* With clocks properly synchronized, applications using the AVTP plugin
|
|
|
|
* should use GstSytemClock with GST_CLOCK_TYPE_REALTIME as the pipeline
|
|
|
|
* clock.
|
|
|
|
*
|
2020-02-06 00:17:39 +00:00
|
|
|
* ### Clock Reference Format (CRF)
|
|
|
|
*
|
|
|
|
* Even though the systems are synchronized by PTP, it is possible that
|
|
|
|
* different talkers can send media streams which are out of phase or the
|
|
|
|
* frequencies do not exactly match. This is partcularly important when there
|
|
|
|
* is a single listener processing data from multiple talkers. The systems in
|
|
|
|
* this scenario can benefit if a common clock is distributed among the
|
|
|
|
* systems.
|
|
|
|
*
|
|
|
|
* This can be achieved by using the avtpcrfsync element which implements CRF
|
2019-10-14 20:55:57 +00:00
|
|
|
* as described in Chapter 10 of IEEE 1722-2016. avtpcrfcheck can also be used
|
|
|
|
* to validate that the adjustment conforms to the criteria specified in the
|
|
|
|
* spec. For further details, look at the documentation for the respective
|
|
|
|
* elements.
|
2020-02-06 00:17:39 +00:00
|
|
|
*
|
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
|
|
|
* ### Traffic Control Setup
|
2019-05-17 23:00:24 +00:00
|
|
|
*
|
|
|
|
* FQTSS (Forwarding and Queuing Enhancements for Time-Sensitive Streams) can be
|
|
|
|
* enabled on Linux with the help of the mqprio and cbs qdiscs provided by the
|
|
|
|
* Linux Traffic Control. Below we provide an example to configure those qdiscs
|
|
|
|
* in order to transmit a CVF H.264 stream 1280x720@30fps. For further
|
|
|
|
* information on how to configure these qdiscs check tc-mqprio(8) and
|
|
|
|
* tc-cbs(8) man pages.
|
|
|
|
*
|
|
|
|
* On the host that will run as AVTP Talker (pipeline that generates the video
|
|
|
|
* stream), run the following commands:
|
|
|
|
*
|
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
|
|
|
* Configure mpqrio qdisc (replace $MQPRIO_HANDLE_ID by an unused handle ID):
|
2019-05-17 23:00:24 +00:00
|
|
|
*
|
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
|
|
|
* $ tc qdisc add dev $IFNAME parent root handle $MQPRIO_HANDLE_ID mqprio \
|
2019-05-17 23:00:24 +00:00
|
|
|
* num_tc 3 map 2 2 1 0 2 2 2 2 2 2 2 2 2 2 2 2 \
|
|
|
|
* queues 1@0 1@1 2@2 hw 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
|
|
|
* Configure cbs qdisc (replace $CBS_HANDLE_ID by an unused handle ID):
|
|
|
|
*
|
|
|
|
* $ tc qdisc replace dev $IFNAME parent $MQPRIO_HANDLE_ID:1 \
|
|
|
|
* handle $CBS_HANDLE_ID cbs idleslope 27756 sendslope -972244 \
|
|
|
|
* hicredit 42 locredit -1499 offload 1
|
|
|
|
*
|
|
|
|
* Also, the plugin implements a transmission scheduling mechanism that relies
|
|
|
|
* on ETF qdisc so make sure it is properly configured in your system. It could
|
|
|
|
* be configured in many ways, below follows an example.
|
2019-05-17 23:00:24 +00:00
|
|
|
*
|
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
|
|
|
* $ tc qdisc add dev $IFNAME parent $CBS_HANDLE_ID:1 etf \
|
|
|
|
* clockid CLOCK_TAI delta 500000 offload
|
2019-05-17 23:00:24 +00:00
|
|
|
*
|
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
|
|
|
* No Traffic Control configuration is required at the host running as AVTP
|
|
|
|
* Listener.
|
2019-05-17 23:00:24 +00:00
|
|
|
*
|
|
|
|
* ### Capabilities
|
|
|
|
*
|
|
|
|
* The `avtpsink` and `avtpsrc` elements open `AF_PACKET` sockets, which require
|
|
|
|
* `CAP_NET_RAW` capability. Therefore, applications must have that capability
|
|
|
|
* in order to successfully use this element. For instance, one can use:
|
|
|
|
*
|
|
|
|
* $ sudo setcap cap_net_raw+ep <application>
|
|
|
|
*
|
|
|
|
* Applications can drop this capability after the sockets are open, after
|
|
|
|
* `avtpsrc` or `avtpsink` elements transition to PAUSED state. See setcap(8)
|
|
|
|
* man page for more information.
|
|
|
|
*
|
|
|
|
* ### Elements configuration
|
|
|
|
*
|
|
|
|
* Each element has its own configuration properties, with some being common
|
|
|
|
* to several elements. Basic properties are:
|
|
|
|
*
|
2020-02-06 00:17:39 +00:00
|
|
|
* * streamid (avtpaafpay, avtpcvfpay, avtpaafdepay, avtpcvfdepay,
|
2019-10-14 20:55:57 +00:00
|
|
|
* avtpcrfsync, avtpcrfcheck): Stream ID associated with the stream.
|
2019-05-17 23:00:24 +00:00
|
|
|
*
|
2019-10-14 20:55:57 +00:00
|
|
|
* * ifname (avtpsink, avtpsrc, avtpcrfsync, avtpcrfcheck): Network interface
|
2020-02-06 00:17:39 +00:00
|
|
|
* used to send/receive AVTP packets.
|
2019-05-17 23:00:24 +00:00
|
|
|
*
|
|
|
|
* * dst-macaddr (avtpsink, avtpsrc): Destination MAC address for the stream.
|
|
|
|
*
|
|
|
|
* * priority (avtpsink): Priority used by the plugin to transmit AVTP
|
|
|
|
* traffic.
|
|
|
|
*
|
|
|
|
* * mtt (avtpaafpay, avtpcvfpay): Maximum Transit Time, in nanoseconds, as
|
|
|
|
* defined in AVTP spec.
|
|
|
|
*
|
|
|
|
* * tu (avtpaafpay, avtpcvfpay): Maximum Time Uncertainty, in nanoseconds, as
|
|
|
|
* defined in AVTP spec.
|
|
|
|
*
|
|
|
|
* * processing-deadline (avtpaafpay, avtpcvfpay, avtpsink): Maximum amount of
|
|
|
|
* time, in nanoseconds, that the pipeline is expected to process any
|
|
|
|
* buffer. This value should be in sync between the one used on the
|
|
|
|
* payloader and the sink, as this time is also taken into consideration to
|
|
|
|
* define the correct presentation time of the packets on the AVTP listener
|
|
|
|
* side. It should be as low as possible (zero if possible).
|
|
|
|
*
|
|
|
|
* * tstamp-mode (avtpaafpay): AAF timestamping mode, as defined in AVTP spec.
|
|
|
|
*
|
|
|
|
* * mtu (avtpcvfpay): Maximum Transmit Unit of the underlying network, used
|
|
|
|
* to determine when to fragment a CVF packet and how big it should be.
|
|
|
|
*
|
|
|
|
* Check each element documentation for more details.
|
|
|
|
*
|
|
|
|
*
|
|
|
|
* ### Running a sample pipeline
|
|
|
|
*
|
|
|
|
* The following pipelines assume a hypothetical `-k ptp` flag that forces the
|
|
|
|
* pipeline clock to be GstPtpClock. A real application would programmatically
|
|
|
|
* define GstPtpClock as the pipeline clock (see next section). It is also
|
|
|
|
* assumed that `gst-launch-1.0` has CAP_NET_RAW capability.
|
|
|
|
*
|
|
|
|
* On the AVTP talker, the following pipeline can be used to generate an H.264
|
|
|
|
* stream to be sent via network using AVTP:
|
|
|
|
*
|
|
|
|
* $ gst-launch-1.0 -k ptp videotestsrc is-live=true ! clockoverlay ! \
|
|
|
|
* x264enc ! avtpcvfpay processing-deadline=20000000 ! \
|
2020-02-06 00:17:39 +00:00
|
|
|
* avtpcrfsync ifname=$IFNAME ! avtpsink ifname=$IFNAME
|
2019-05-17 23:00:24 +00:00
|
|
|
*
|
|
|
|
* On the AVTP listener host, the following pipeline can be used to get the
|
|
|
|
* AVTP stream, depacketize it and show it on the screen:
|
|
|
|
*
|
2019-10-14 20:55:57 +00:00
|
|
|
* $ gst-launch-1.0 -k ptp avtpsrc ifname=$IFNAME ! \
|
|
|
|
* avtpcrfcheck ifname=$IFNAME ! avtpcvfdepay ! \
|
2019-05-17 23:00:24 +00:00
|
|
|
* vaapih264dec ! videoconvert ! clockoverlay halignment=right ! \
|
|
|
|
* queue ! autovideosink
|
|
|
|
*
|
|
|
|
* ### Pipeline clock
|
|
|
|
*
|
|
|
|
* The AVTP plugin elements require that the pipeline clock is in sync with
|
|
|
|
* the network PTP clock. As GStreamer has a GstPtpClock, using it should be
|
|
|
|
* the simplest way of achieving that.
|
|
|
|
*
|
|
|
|
* However, as there's no way of forcing a clock to a pipeline using
|
|
|
|
* gst-launch-1.0 application, even for quick tests, it's necessary to have
|
|
|
|
* an application. One can refer to GStreamer "hello world" application,
|
|
|
|
* remembering to set the pipeline clock to GstPtpClock before putting the
|
|
|
|
* pipeline on "PLAYING" state. Some code like:
|
|
|
|
*
|
|
|
|
* GstClock *clk = gst_ptp_clock_new("ptp-clock", 0);
|
|
|
|
* gst_clock_wait_for_sync(clk, GST_CLOCK_TIME_NONE);
|
|
|
|
* gst_pipeline_use_clock (GST_PIPELINE (pipeline), clk);
|
|
|
|
*
|
|
|
|
* Would do the trick.
|
|
|
|
*
|
|
|
|
* ### Disclaimer
|
|
|
|
*
|
|
|
|
* It's out of scope for the AVTP plugin to verify how it is invoked, should
|
|
|
|
* a malicious software do it for Denial of Service attempts, or other
|
|
|
|
* compromises attempts.
|
|
|
|
*
|
2019-01-14 18:18:42 +00:00
|
|
|
*/
|
|
|
|
#ifdef HAVE_CONFIG_H
|
|
|
|
#include <config.h>
|
|
|
|
#endif
|
|
|
|
|
|
|
|
#include <gst/gst.h>
|
|
|
|
|
2019-01-24 00:20:27 +00:00
|
|
|
#include "gstavtpaafdepay.h"
|
avtp: Introduce AAF payloader element
This patch introduces the AVTP Audio Format (AAF) payloader element from
the AVTP plugin. The element inputs audio raw data and outputs AVTP
packets (aka AVTPDUs), implementing a typical protocol payloader element
from GStreamer.
AAF is one of the available formats to transport audio data in an AVTP
system. AAF is specified in IEEE 1722-2016 section 7 and provides two
encapsulation mode: PCM and AES3. This patch implements PCM
encapsulation mode only.
The AAF payloader working mechanism consists of building the AAF header,
prepending it to the GstBuffer received on the sink pad, and pushing the
buffer downstream. Payloader parameters such as stream ID, maximum
transit time, time uncertainty, and timestamping mode are passed via
element properties. AAF doesn't support all possible sample format and
sampling rate values so the sink pad caps template from the payloader is
a subset of audio/x-raw. Additionally, this patch implements only
"normal" timestamping mode from AAF. "Sparse" mode should be implemented
in future.
Upcoming patches will introduce other AVTP payloader elements that will
have some common code. For that reason, this patch introduces the
GstAvtpBasePayload abstract class that implements common payloader
functionalities, and the GstAvtpAafPay class that extends the
GstAvtpBasePayload class, implementing AAF-specific functionalities.
The AAF payloader element is most likely to be used with the AVTP sink
element (to be introduced by a later patch) but it could also be used
with UDP sink element to implement AVTP over UDP as described in IEEE
1722-2016 Annex J.
This element was inspired by RTP payloader elements.
2019-01-17 01:16:59 +00:00
|
|
|
#include "gstavtpaafpay.h"
|
2019-03-12 22:46:16 +00:00
|
|
|
#include "gstavtpcvfdepay.h"
|
2019-02-28 23:49:02 +00:00
|
|
|
#include "gstavtpcvfpay.h"
|
2019-01-23 18:56:10 +00:00
|
|
|
#include "gstavtpsink.h"
|
2019-01-23 23:17:48 +00:00
|
|
|
#include "gstavtpsrc.h"
|
2020-02-06 00:17:39 +00:00
|
|
|
#include "gstavtpcrfsync.h"
|
2019-10-14 20:55:57 +00:00
|
|
|
#include "gstavtpcrfcheck.h"
|
avtp: Introduce AAF payloader element
This patch introduces the AVTP Audio Format (AAF) payloader element from
the AVTP plugin. The element inputs audio raw data and outputs AVTP
packets (aka AVTPDUs), implementing a typical protocol payloader element
from GStreamer.
AAF is one of the available formats to transport audio data in an AVTP
system. AAF is specified in IEEE 1722-2016 section 7 and provides two
encapsulation mode: PCM and AES3. This patch implements PCM
encapsulation mode only.
The AAF payloader working mechanism consists of building the AAF header,
prepending it to the GstBuffer received on the sink pad, and pushing the
buffer downstream. Payloader parameters such as stream ID, maximum
transit time, time uncertainty, and timestamping mode are passed via
element properties. AAF doesn't support all possible sample format and
sampling rate values so the sink pad caps template from the payloader is
a subset of audio/x-raw. Additionally, this patch implements only
"normal" timestamping mode from AAF. "Sparse" mode should be implemented
in future.
Upcoming patches will introduce other AVTP payloader elements that will
have some common code. For that reason, this patch introduces the
GstAvtpBasePayload abstract class that implements common payloader
functionalities, and the GstAvtpAafPay class that extends the
GstAvtpBasePayload class, implementing AAF-specific functionalities.
The AAF payloader element is most likely to be used with the AVTP sink
element (to be introduced by a later patch) but it could also be used
with UDP sink element to implement AVTP over UDP as described in IEEE
1722-2016 Annex J.
This element was inspired by RTP payloader elements.
2019-01-17 01:16:59 +00:00
|
|
|
|
2019-01-14 18:18:42 +00:00
|
|
|
static gboolean
|
|
|
|
plugin_init (GstPlugin * plugin)
|
|
|
|
{
|
avtp: Introduce AAF payloader element
This patch introduces the AVTP Audio Format (AAF) payloader element from
the AVTP plugin. The element inputs audio raw data and outputs AVTP
packets (aka AVTPDUs), implementing a typical protocol payloader element
from GStreamer.
AAF is one of the available formats to transport audio data in an AVTP
system. AAF is specified in IEEE 1722-2016 section 7 and provides two
encapsulation mode: PCM and AES3. This patch implements PCM
encapsulation mode only.
The AAF payloader working mechanism consists of building the AAF header,
prepending it to the GstBuffer received on the sink pad, and pushing the
buffer downstream. Payloader parameters such as stream ID, maximum
transit time, time uncertainty, and timestamping mode are passed via
element properties. AAF doesn't support all possible sample format and
sampling rate values so the sink pad caps template from the payloader is
a subset of audio/x-raw. Additionally, this patch implements only
"normal" timestamping mode from AAF. "Sparse" mode should be implemented
in future.
Upcoming patches will introduce other AVTP payloader elements that will
have some common code. For that reason, this patch introduces the
GstAvtpBasePayload abstract class that implements common payloader
functionalities, and the GstAvtpAafPay class that extends the
GstAvtpBasePayload class, implementing AAF-specific functionalities.
The AAF payloader element is most likely to be used with the AVTP sink
element (to be introduced by a later patch) but it could also be used
with UDP sink element to implement AVTP over UDP as described in IEEE
1722-2016 Annex J.
This element was inspired by RTP payloader elements.
2019-01-17 01:16:59 +00:00
|
|
|
if (!gst_avtp_aaf_pay_plugin_init (plugin))
|
|
|
|
return FALSE;
|
2019-01-24 00:20:27 +00:00
|
|
|
if (!gst_avtp_aaf_depay_plugin_init (plugin))
|
|
|
|
return FALSE;
|
2019-01-23 18:56:10 +00:00
|
|
|
if (!gst_avtp_sink_plugin_init (plugin))
|
|
|
|
return FALSE;
|
2019-01-23 23:17:48 +00:00
|
|
|
if (!gst_avtp_src_plugin_init (plugin))
|
|
|
|
return FALSE;
|
2019-02-28 23:49:02 +00:00
|
|
|
if (!gst_avtp_cvf_pay_plugin_init (plugin))
|
|
|
|
return FALSE;
|
2019-03-12 22:46:16 +00:00
|
|
|
if (!gst_avtp_cvf_depay_plugin_init (plugin))
|
|
|
|
return FALSE;
|
2020-02-06 00:17:39 +00:00
|
|
|
if (!gst_avtp_crf_sync_plugin_init (plugin))
|
|
|
|
return FALSE;
|
2019-10-14 20:55:57 +00:00
|
|
|
if (!gst_avtp_crf_check_plugin_init (plugin))
|
|
|
|
return FALSE;
|
avtp: Introduce AAF payloader element
This patch introduces the AVTP Audio Format (AAF) payloader element from
the AVTP plugin. The element inputs audio raw data and outputs AVTP
packets (aka AVTPDUs), implementing a typical protocol payloader element
from GStreamer.
AAF is one of the available formats to transport audio data in an AVTP
system. AAF is specified in IEEE 1722-2016 section 7 and provides two
encapsulation mode: PCM and AES3. This patch implements PCM
encapsulation mode only.
The AAF payloader working mechanism consists of building the AAF header,
prepending it to the GstBuffer received on the sink pad, and pushing the
buffer downstream. Payloader parameters such as stream ID, maximum
transit time, time uncertainty, and timestamping mode are passed via
element properties. AAF doesn't support all possible sample format and
sampling rate values so the sink pad caps template from the payloader is
a subset of audio/x-raw. Additionally, this patch implements only
"normal" timestamping mode from AAF. "Sparse" mode should be implemented
in future.
Upcoming patches will introduce other AVTP payloader elements that will
have some common code. For that reason, this patch introduces the
GstAvtpBasePayload abstract class that implements common payloader
functionalities, and the GstAvtpAafPay class that extends the
GstAvtpBasePayload class, implementing AAF-specific functionalities.
The AAF payloader element is most likely to be used with the AVTP sink
element (to be introduced by a later patch) but it could also be used
with UDP sink element to implement AVTP over UDP as described in IEEE
1722-2016 Annex J.
This element was inspired by RTP payloader elements.
2019-01-17 01:16:59 +00:00
|
|
|
|
2019-01-14 18:18:42 +00:00
|
|
|
return TRUE;
|
|
|
|
}
|
|
|
|
|
|
|
|
GST_PLUGIN_DEFINE (GST_VERSION_MAJOR, GST_VERSION_MINOR,
|
|
|
|
avtp, "Audio/Video Transport Protocol (AVTP) plugin",
|
|
|
|
plugin_init, VERSION, GST_LICENSE, GST_PACKAGE_NAME, GST_PACKAGE_ORIGIN);
|