filter/pytorch,caffe2: they have redundant decls.

Add -Wredundant-decl only when !(pytorch or caffe2), not !(pytorch and caffe2).
Even when only one of the two are included, they have redundant decl warnings.
It appears that we haven't tested cases where only one of pytorch and caffe2 is
enabled.

Fixes #3693

Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>

doc: update README and products

- Update paper link
- Update support hardware
- Update product lists

Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>

[Build/PyUtil] dependency to py-util

Add link to python util library in meson.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>

[CodeClean/Python] possible leak in python subbplugin

Code clean,
1. possible leak and exception to load python module.
2. remove unnecessary macro.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>

[TEST/cpp] Wait until pipeline save the files

Wait until pipeline save the files.

Signed-off-by: gichan <gichan2.jang@samsung.com>

[build/tflite] Check each delegate symbol in build time

- Check each tflite delegate symbol by linking with tflite dep in build time

Signed-off-by: Yongjoo Ahn <yongjoo1.ahn@samsung.com>

[Log] exporting logger

Log util is available only in nnstreamer repo.
Remove exporting the header for logging.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>

[SSAT] Enable filter_python3 test to run on target

 - remove exit if build folder not exists
 - bypass pushd failure if build folder not exists

Signed-off-by: Aymen Sghaier <aymen.sghaier@nxp.com>

[Pkg] add filter-single header

dev-internal pkg should include filter-single header for single-shot API.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>

[TEST] Disable python converter tests.

TEMPORAL SOLUTIONS: PLEASE REVERT THIS COMMIT AFTER FIXING THE ISSUE.

When the Python tensor converter test is completed, a deadlock occurs.

Signed-off-by: gichan <gichan2.jang@samsung.com>

[Util] fix typo and license

Fix typo and license in util functions and header.
Update internal function description and remove unnecessary param.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>

[tensor_filter] Modify the order of caps_intersect_full

- The pipeline reported in #3636 has nego fail issue because the caps is
   fixated with flexible right after tensor_filter.
- To resolve this, change the order of `gst_caps_intersect_full` in transform_caps.
   This makes static format to be chosen.
- This resolves #3636

Signed-off-by: Yongjoo Ahn <yongjoo1.ahn@samsung.com>

[tests] Add a ssat testcase to check proper negotiation

- Test whether tensor_filter - tensor_transform pipeline can be
  negotiated with static format, not flexible format.

Signed-off-by: Yongjoo Ahn <yongjoo1.ahn@samsung.com>

[transform] Fix transform_caps in direction GST_PAD_SRC

- Since tensor_transform cannot specify the sinkpad's type caps in some modes,
  let the `transform_caps` remove the field in that case

Signed-off-by: Yongjoo Ahn <yongjoo1.ahn@samsung.com>

[fix/common] Fix calling g_string_append with null

- Check null before call g_string_append
- This removes the logs `g_string_insert_len: assertion 'len == 0 || val != NULL' failed`

REF: #3666
Signed-off-by: Yongjoo Ahn <yongjoo1.ahn@samsung.com>

[Util] util function to compare tensor-config

remove gst-dependency to compare tensor config.
add return val of invalid condition.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>

[fix/tests] Fix mem-leak issues in some unittests

- Fix minor mem-leak in lua, snpe, tvm unittests

Signed-off-by: Yongjoo Ahn <yongjoo1.ahn@samsung.com>

[fix/lua] Fix mem-leak of lua filter

- Use smart pointer for lua script
- Fix minor typo

Signed-off-by: Yongjoo Ahn <yongjoo1.ahn@samsung.com>

[fix/filter] Add check for model file in some filters

- Add check for `G_FILE_TEST_IS_REGULAR` in some cpp class based filters

Signed-off-by: Yongjoo Ahn <yongjoo1.ahn@samsung.com>

[Single] Rename single-shot files

 - Rename the files related to tensor common functions.
   renamed:    gst/nnstreamer/include/nnstreamer_plugin_api_single.h -> gst/nnstreamer/include/nnstreamer_plugin_api_util.h
   renamed:    gst/nnstreamer/tensor_common.c -> gst/nnstreamer/nnstreamer_plugin_api_impl.c
   renamed:    gst/nnstreamer/tensor_common_single.c -> gst/nnstreamer/nnstreamer_plugin_api_util_impl.c
 - Remove unnecessary pipeline utils from nnstreamer-internal header.

Signed-off-by: gichan <gichan2.jang@samsung.com>

[TEST] Remove gst dep from singleshot test

Remove gst dependency from singleshot test

Signed-off-by: gichan <gichan2.jang@samsung.com>

[Filter] Remove gst dependency from filter subplugins.

Remove gst dependency from tensor filter subplugins.
 - tensorflow, tensorflow-lite, pytorch, nnfw,caffe2

TODO: Other subplugins later.

Signed-off-by: gichan <gichan2.jang@samsung.com>

[Filter] Remove gst dep from single-shot

 - Remove gst dependency from single-shot.
 - Separate single-shot package from nnstreamer(pipeline)
 - nnstreamer depends nnstreaner-single package.

Signed-off-by: gichan <gichan2.jang@samsung.com>

[tests/tflite] Add tests for tflite2

- Set tensor_filter_ext_enabled when tflite2_support is available
- Add a gtest based test to examine the XNNPACK delegate

Signed-off-by: Yongjoo Ahn <yongjoo1.ahn@samsung.com>

[tflite] Support tflite v2.7.0 for tizen and deian

- Remove unnecessary gles dependency
- Support newly introduced xnnpack delegate
- Remove unnecessary build options

Signed-off-by: Yongjoo Ahn <yongjoo1.ahn@samsung.com>

[Filter/trix-engine] Fix svace issue

 - Initialize class member.
 - Fix bad copy and paste

Signed-off-by: gichan <gichan2.jang@samsung.com>

[TEST] Increase sleep time of the query test

1 second sleep time is not enough to operate query server.
see: https://github.com/nnstreamer/nnstreamer/issues/3657#issuecomment-1038708557

Signed-off-by: gichan <gichan2.jang@samsung.com>

[filter] Support complex output types for pytorch

This patch adds support for complex output types for pytorch, where the
complex output is serialized before being passed out to the gstreamer
based element - tensorfilter.

V2:
- Added more comprehensive error messages
- Added indentation fix

Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>

[Build] Disable trix engine support for TV releases

This patch disables trix_engine_support for TV releases.

Signed-off-by: yelini-jeong <yelini.jeong@samsung.com>

Fix build error - unused param, sign comapre

 Fix build error caused by unused param.
  - Ignore auto generated files (flatbuf, protobuf)
 Fix different sign compare

Signed-off-by: gichan <gichan2.jang@samsung.com>

[MqttSrc/Sink] release mqtt handle

Destroy mqtt-async handle when the element instance is finalized.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>

[Svace] fix svace issues

Fix svace issues
- init local variable
- check allocation failure case

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>

[trivial/doc] Update android guide: gstreamer v1.16.2 -> v1.20.0

- Update android guide: gstreamer version v1.16.2 -> v1.20.0

Signed-off-by: Yongjoo Ahn <yongjoo1.ahn@samsung.com>

[test/ssat] Disable proto flat flex buf SSAT tests when there are no supports

- Disable protobuf, flatbuf, flexbuf SSAT tests when related libs did  not built

Signed-off-by: Yongjoo Ahn <yongjoo1.ahn@samsung.com>

[Docker] Remove the existing Dockerfiles for Ubuntu 16.04 and 18.04

This patch replaces the existing Dockerfiles corresponding to Ubuntu
16.04 and 18.04 each with the one Dockerfile that has adopted
multi-stage build and build-args concepts.

Signed-off-by: Wook Song <wook16.song@samsung.com>

[Docker/README] Add brief instructions on Docker support

This patch adds a README.md file to give brief instructions on getting
started NNStreamer with Docker.

Signed-off-by: Wook Song <wook16.song@samsung.com>

[Docker] Containerize the minimal running environment

This patch modifies the unified Dockerfile to containerize the minimal
running environment of NNStreamer. By using the build argument,
MIN_RUN_ENV, it is possible to build the image installed nnstreamer,
nnstreamer-core, and nnstreamer-configuration packages only.

Signed-off-by: Wook Song <wook16.song@samsung.com>

[Docker] Add another stage containing the runtime environment

This patch adds another stage to provide the runtime environment of
NNStreamer using the *.deb files created from the previous build stage.

Signed-off-by: Wook Song <wook16.song@samsung.com>

[Docker] Add an intermediate stage that runs debuild

This patch adds an intermediate stage which runs debuild to create
debian packages from the nnstreamer source code at the first development
environment stage. Note that a workaround has been applied in order to
temporarily avoid the unittest_mqtt failure issue.

Signed-off-by: Wook Song <wook16.song@samsung.com>

[Docker] Integrate the multiple dockerfiles into one

To integrate the multiple dockerfiles into universal one, this patch
introduces the concept of a build-time variable, ARG, and multi-stage
builds in the Dockerfile.

Signed-off-by: Wook Song <wook16.song@samsung.com>

[Meson/TRIx] target dependency for trix engine

Define target library to build filter-subplugin for trix-engine.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>

[MQTTSrc] default options before initializaion

Bug fix, set default value after initializing connection option.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>

[filter] Bug fix for pytorch filter with gpu

ptyorch filter when working with gpu backend had bug when converting
between the devices. The tensors returned by the conversion were not
catched. This patch provides the corresponding fix.

Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>

[test/pytorch] Add ssat test for multiple input/output tensor model

- Add a simple test for pytorch model whose input and output are multiple tensors

Signed-off-by: Yongjoo Ahn <yongjoo1.ahn@samsung.com>

[filter/pytorch] Add support for tuple type output

- Support torch script model whose output value is `Tuple`

Signed-off-by: Yongjoo Ahn <yongjoo1.ahn@samsung.com>

[Query] Fix invalid return value

Fix invalid return value when caps is same.

Signed-off-by: Gichan Jang <gichan2.jang@samsung.com>

[Doc] Fix and add new feature of the query.

 - Fix capability of the query server and client.
 - Add about mqtt-hybrid

Signed-off-by: Gichan Jang <gichan2.jang@samsung.com>
Signed-off-by: gichan <gichan2.jang@samsung.com>

[Test] Add test cases for extra configuration

This patch newly adds the test cases for extra configuration.

Signed-off-by: Sangjung Woo <sangjung.woo@samsung.com>

[INI] Add extra configuration for product support

The product team wants to update the nnstreamer configuration (i.e.
/etc/nnstreamer.ini). However, the partition is Read-Only so it is not
updatable. Because of this reason, they want another nnstreamer
configuration file to meet their request such as product-specific
allowed elements. This patch newly adds the nnstreamer extra
configuration file for product support.

Signed-off-by: Sangjung Woo <sangjung.woo@samsung.com>

[TRIx-Engine] Add TRIx-Engine plugin

This patch adds TRIx-Engine tensor filter subplugin implementation.
It requires TRIV2 NPU HW and devel packages from the upstream tizen repo.

Signed-off-by: Dongju Chae <dongju.chae@samsung.com>

[Build] unnecessary build dependency

Update unnecessary dependency of filter sub-plugins.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>

[decoder/bounding_box] Support decoder for yolov5

- Add feature for decoding yolov5 model's output
- Tested with tflite models from https://github.com/ultralytics/yolov5/releases/tag/v6.0

REF: https://github.com/ultralytics/yolov5
Signed-off-by: Yongjoo Ahn <yongjoo1.ahn@samsung.com>

[trivial] Fix some typos in tensordec-boundingbox.c

- Fix some trivial typo issues

Signed-off-by: Yongjoo Ahn <yongjoo1.ahn@samsung.com>

[Spec] Enable SNPE filter on DA profile

Tizen Robot Vacuum cleaner needs the SNPE filter. This patch enables it
on the DA profile.

Signed-off-by: Sangjung Woo <sangjung.woo@samsung.com>

[Test] Fix python custom converter test fail

Getting result from python json pareser, result string doesn't contain
null char, the test fails.
Add null char to the string from python json parser.

Signed-off-by: Gichan Jang <gichan2.jang@samsung.com>

[trivial] Fix CRLF endend files to be LF ended

- Replace CRLF with LF

Signed-off-by: Yongjoo Ahn <yongjoo1.ahn@samsung.com>

[Doc] Update debuild guide.

Add description for nnfw ppa.

Related issue: #3612

Signed-off-by: Gichan Jang <gichan2.jang@samsung.com>

[Conf] Update the name of 'restricted-elements'

The name 'restricted-elements' of nnstreamer configuration is not
matched as its real operation. This patch renames it to
'allowed_elements'.

Signed-off-by: Sangjung Woo <sangjung.woo@samsung.com>

[Test/Build] fix error on old machine

Fix implicit declaration on old machine.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>

[TEST] Changed the comparison method of the padded video.

If the test runs with the valgrind, the comparison of the padded video fails.
(Related issue: https://github.com/nnstreamer/nnstreamer/issues/1367)
Zero was padded when running without the valgrind, but trash value was padded when running with the valgrind.
The comparison of the padded data is meaningless, it is modified to compare only the actual data.

Signed-off-by: gichan <gichan2.jang@samsung.com>
Signed-off-by: Gichan Jang <gichan2.jang@samsung.com>

[TEST] Move replace_string to unittest_util.

Since `repace_string` is copied to nnstreamer/api and `replace_string` is only used for unit test, moves it to unit test util.
And the test related to `replace_string` is moved to nnstreamer/api.

Test to replace string to check unintended change.

Signed-off-by: gichan <gichan2.jang@samsung.com>

[Filter] set default fw in common

Fix mem leak, init default framework in filter-common function.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>

[fix/bug] Fix destructor of cpp class subplugin

- Fix pointer for subplugin in `cpp_close`
- Fix edgetpu and snpe filter's model path free function

Signed-off-by: Yongjoo Ahn <yongjoo1.ahn@samsung.com>

[TEST] Wait for the server to start and fix comment.

 - Wait for the server to start.
    If client attempts to connect to the server when the server is not ready, it may  fails.
    Previous PR, sleep time was omitted only in Test 7.
 - Fix invalid comment.

Signed-off-by: Gichan Jang <gichan2.jang@samsung.com>

[Spec] Remove invalid elements from restricted elements.

Remove invalid elements from restricted elements.
  queryclient -> tensor_query_client
  queryserversrc -> tensor_query_serversrc
  queryserversink -> tensor_query_serversink

Signed-off-by: Gichan Jang <gichan2.jang@samsung.com>

[TEST/util] Change getTempFileName function

When creating a temp file name, it is set to create a unique name, but duplicated temp names may occasionally generated because we remove generated file and get the temp name only.
So sometimes duplicated temp filed names generated when creating temp filenames multiple times.

For example, generate tmp filename 5 times, tmp2 and tmp3 is duplicated.

[  200s] ** (unittest_cppfilter:132252): CRITICAL **: 05:07:52.416: tmp1: /tmp/nnstreamer_unittest_temp_UJOPE1
[  200s]
[  200s] ** (unittest_cppfilter:132252): CRITICAL **: 05:07:52.416: tmp2: /tmp/nnstreamer_unittest_temp_VIOPE1
[  200s]
[  200s] ** (unittest_cppfilter:132252): CRITICAL **: 05:07:52.416: tmp3: /tmp/nnstreamer_unittest_temp_VIOPE1
[  200s]
[  200s] ** (unittest_cppfilter:132252): CRITICAL **: 05:07:52.416: tmp4: /tmp/nnstreamer_unittest_temp_AJOPE1
[  200s]
[  200s] ** (unittest_cppfilter:132252): CRITICAL **: 05:07:52.416: tmp5: /tmp/nnstreamer_unittest_temp_ILOPE1

Signed-off-by: Gichan Jang <gichan2.jang@samsung.com>
Signed-off-by: gichan <gichan2.jang@samsung.com>

[Common] Add error message for tensor common functions

Add error message for tensor common functions.

Signed-off-by: gichan <gichan2.jang@samsung.com>

[Coverity] Fix use after free

This patch fixes coverity issues (WGID 131439, CID 1243725)

Signed-off-by: yelini-jeong <yelini.jeong@samsung.com>

[Filter/TFLite] Fix delegates deleter usage

On pipeline stop, crash in TFLite filter may be seen with some
delegates with log signature:

Setting pipeline to NULL ...
free(): invalid pointer

Pipeline termination procedure is then aborted.

TFlite delegates shall be released by specific deleter, provided
by their respective interface.
TFLite filter implementation manages delegates instance with
smart pointers for automatic disposal when interpreter goes out of
of scope.

Care should be taken to bind relevant delegate deleter to the
smart pointer, otherwise trivial deleter is used for the managed
pointer which may crash.

Signed-off-by: Julien Vuillaumier <julien.vuillaumier@nxp.com>

[Query] Check whether cap is fixed or not.

 - Check whether the cap is fixed or not before setting the cap.
 - If the tensor format is not fixed, static is set as default format.
 - Set the framerate of the source to the same as the sink.

Signed-off-by: gichan <gichan2.jang@samsung.com>

[Query] error handing in query-server init

Fix possible resource leak - handle error case in query-server init.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>

[Test/Query] fix invalid param

Fix invalid param - instance after launching test pipeline.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>

[Typo] Fix some typos in query serversink

This patch fixes some typos in query serversink.

Signed-off-by: Dongju Chae <dongju.chae@samsung.com>

filter/tensorflow-lite: loadModel error message

For developers, we need to elaborate error messages.
The error messages should have enough information to
understand what's wrong and what to fix.

We will be updating all error messages of nnstreamer.

This helps #3603
This is part of #3475

Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>

[Query] client id in connection handle

1. Add new function to set client ID.
2. Remove unnecessary value - protocol in cmd data. The connection handle includes this value.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>

[Query] Set timestamp for each buffer.

Set timestamp for each buffer.

Signed-off-by: Gichan Jang <gichan2.jang@samsung.com>

[Decoder] Decode tensors to octet stream.

Decode tensors to octet stream.
If tensor format is flexible, remove the header.

Signed-off-by: gichan <gichan2.jang@samsung.com>

[TEST] Change filesink buffer mode.

Change filesink buffer mode to ubnuffered.
Related issue: https://github.com/nnstreamer/nnstreamer/issues/3275

Signed-off-by: Gichan Jang <gichan2.jang@samsung.com>

[Query] Change error message when GstMeta is invalid.

Some GStreamer element don't copy GstMeta. So tensor query cannot handle
buffer.
Change error message when GstMeta is invalid and stop pipeline.
The available elements table is not completed yet and will change frequently, so I will write it on the wiki page first.

Signed-off-by: Gichan Jang <gichan2.jang@samsung.com>

[filter] Add a time duration log of filter open

- Some filter using hardware accelerator consume quite time for loading model file.
- Add log for what time it takes.

Signed-off-by: Yongjoo Ahn <yongjoo1.ahn@samsung.com>

[Query] fix invalid data size

1. Fix invalid data size - caps string length.
2. Clearly define client and server param and update log message.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>

Test: enhancing log behaviors for unittest

1. Add a dlog_print hijacker that makes dlog calls independent
from dlog daemon and drivers. As a result, developers can now
read the dlog log messages of unittests in gbs environment.

2. Clean up a bit of meson.build dependency declaration of
unit tests.

3. A unit test without direct dlog call is updated to call
it directly so that the hijacker is linked (edgetpu).

Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>

[Doc] Add meson build example.

Add a simple example of the meson build for a beginner.

Signed-off-by: Gichan Jang <gichan2.jang@samsung.com>

[Common] change log level to get tensor-config

Change log level and add proper message to set default tensor format.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>

[doc] query and image update

This patch updates query and image for tutorial 4

Signed-off-by: yelini-jeong <yelini.jeong@samsung.com>

meson: use version_copmare()

Do not compare version strings with general operators.
Use version_compare() API instead.

Reported by #3591

Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>

[Query] set default timeout

Define common timeout for tensor-query.

TODO: we should handle timeout in query-server elements.
Signed-off-by: Jaeyun <jy1210.jung@samsung.com>

[Query/CodeClean] remove unnecessary util functions

Code clean on tensor-query elements.
1. fix build issue on old machine (undefined ref).
2. remove unnecessary util functions.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>

[Query] util function to receive data

Add util function to receive data and generate gst-buffer.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>

meson.build: Fixup for multilib support

meson.build:
moved nnstreamer_prefix definition upward, before its first usage.

packaging/nnstreamer.spec:
corrected subplugindir definition to rely on actual %{_prefix} variable
and use the expected %{_prefix}/lib/nnstreamer path.

Signed-off-by: Julien Vuillaumier <julien.vuillaumier@nxp.com>

meson.build: Fix sideeffects of subplugin install fix

The PR #3554 breaks conventional non-multilib builds of x64/aarch64
systems (Tizen/Ubuntu). This fixes this side-effect.

Todo: fix Yocto build afterwards.

Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
Cherry-picked from https://github.com/panfir/nnstreamer/pull/1

Signed-off-by: Julien Vuillaumier <julien.vuillaumier@nxp.com>

meson.build: Fix subplugin install

Use 'libdir' for library path to cover a multilib envoironment.

These checks are usually done before release, for example in the scenario
where images are created for a 64-bit target (imx8 * boards) but which also
supports running 32-bit applications.

This is the error in a linux/yocto build:

| ERROR: nnstreamer-2.0.0-r0 do_package: QA Issue: nnstreamer: Files/directories were installed but not shipped in any package:
|   /usr/lib/nnstreamer/customfilters/libnnstreamer_customfilter_passthrough_variable.so
|   /usr/lib/nnstreamer/customfilters/libdummyLSTM.so
|   /usr/lib/nnstreamer/customfilters/libnnstreamer_customfilter_scaler.so
|   /usr/lib/nnstreamer/customfilters/libnnstreamer_customfilter_average.so
|   /usr/lib/nnstreamer/customfilters/libnnstreamer_customfilter_scaler_allocator.so
|   /usr/lib/nnstreamer/customfilters/libnnscustom_drop_buffer.so
|   /usr/lib/nnstreamer/customfilters/libnnscustom_framecounter.so
|   /usr/lib/nnstreamer/customfilters/libdummyRNN.so
|   /usr/lib/nnstreamer/customfilters/libnnstreamer_customfilter_passthrough.so
|   /usr/lib/nnstreamer/converters/libnnstreamer_converter_protobuf.so
|   /usr/lib/nnstreamer/converters/libnnstreamer_converter_python3.so
|   /usr/lib/nnstreamer/decoders/libnnstreamer_decoder_bounding_boxes.so
|   /usr/lib/nnstreamer/decoders/libnnstreamer_decoder_direct_video.so
|   /usr/lib/nnstreamer/decoders/libnnstreamer_decoder_pose_estimation.so
|   /usr/lib/nnstreamer/decoders/libnnstreamer_decoder_image_labeling.so
|   /usr/lib/nnstreamer/decoders/libnnstreamer_decoder_image_segment.so
|   /usr/lib/nnstreamer/decoders/libnnstreamer_decoder_python3.so
|   /usr/lib/nnstreamer/decoders/libnnstreamer_decoder_protobuf.so
|   /usr/lib/nnstreamer/filters/libnnstreamer_filter_cpp.so
|   /usr/lib/nnstreamer/filters/libnnstreamer_filter_armnn.so
|   /usr/lib/nnstreamer/filters/libnnstreamer_filter_python3.so
|   /usr/lib/nnstreamer/filters/libnnstreamer_filter_tensorflow2-lite.so
|   /usr/lib/nnstreamer/extra/nnstreamer_python3.so
| Please set FILES such that these items are packaged. Alternatively if they are unneeded, avoid installing them or delete them within do_install.
| nnstreamer: 23 installed and not shipped files. [installed-vs-shipped]

Signed-off-by: Cristinel Panfir <cristinel.panfir@nxp.com>
Cherry-picked from https://github.com/nnstreamer/nnstreamer/pull/3554

Signed-off-by: Julien Vuillaumier <julien.vuillaumier@nxp.com>

[Mux/Merge] Change meta copy policy

Currently, GstMeta of the buffer from first pad is used.
Change to copy GstMeta same with sync mode of the tnesor mux and merge.

Signed-off-by: gichan <gichan2.jang@samsung.com>
Signed-off-by: Gichan Jang <gichan2.jang@samsung.com>

[TEST] Wait for the server to start

Wait for the server to start.
If client attempts to connect to the server when the server is not
ready, it may  fails.

Signed-off-by: gichan <gichan2.jang@samsung.com>

[Query] Change tensor query to accpet ANY caps

Change tensor query to accetp ANY caps.

Signed-off-by: gichan <gichan2.jang@samsung.com>

[Query] util function to send gst-buffer

Code clean, add util function to send gst-buffer to destination.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>

[Spec] Add restricted_element_vd for VD element

This patch newly adds the restricted_element_vd for VD specialized
elements.

Signed-off-by: Sangjung Woo <sangjung.woo@samsung.com>

[Query] fix invalid socket mode

Fix socket op mode, blocking IO is unnecessary with multi-threaded server node.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>

[Doc] Fix bad links

This patch fixes bad links in documentation.

Signed-off-by: yelini-jeong <yelini.jeong@samsung.com>
**Self evaluation:**
1. Build test: [ ]Passed [ ]Failed [X]Skipped
2. Run test: [ ]Passed [ ]Failed [X]Skipped

[fix/query] Fix the order of calling `g_socket_listener_set_backlog`

- backlog should be set before adding address or socket to listener.
REF:
https://people.gnome.org/~ebassi/docs/_build/Gio/method.SocketListener.set_backlog.html#description

- This commit should remove the following error logs:
```
...
GLib-GIO-CRITICAL **: 15:20:18.773: g_socket_set_listen_backlog: assertion '!socket->priv->listening' failed
...
```

Signed-off-by: Yongjoo Ahn <yongjoo1.ahn@samsung.com>