librealsense-2.56.1 (source build) relocation R_X86_64_32S against symbol `malloc@@GLIBC_2.2.5' can not be used when making a PIE object error

[HariKrishnan06082k]

Issue Description

Building error when trying to install librealsense SDK from source code when building with graphical examples enabled. I'm able to build successfully with RSUSB backend method but I would like to ideally use realsense-viewer for my ROS2 wrapper. The error occurs at a point when it has to generate executable for the realsense-viewer. I have had success building realsense-viewer before from other versions of SDK. The other issue I have is when using D455 with ROS2 wrapper version 4.56.1. For my work I would require the infra stream to be enabled at 60Hz for both IR1 and IR2 along with metadata when I do that I get no frames received error. But that I would create another issue ig for now I just need the rs-viewer to work because that would give me and idea about what resolutions and what streams can be enable simultaneously. Cuz recently the default profile resolution is 848x480 or something .

Camera Model - Realsense D455 camera
Firmware Version - 5.16.0.1
OS - Linux 22.04 LTS
Platform - PC AMD CPU (cuda 11.8)
Kernel Version - 6.8.0-48-generic
Librealsense version - 2.56.1 (Build from source code)
ldd --version = 2.35
cmake --version = 3.22.1
Kernel Patch Done - Yes

[HariKrishnan06082k]

Device info:
Name : Intel RealSense D455
Serial Number : 246322300557
Firmware Version : 5.16.0.1
Recommended Firmware Version : 5.16.0.1
Physical Port : /sys/devices/pci0000:80/0000:80:01.3/0000:82:00.0/0000:83:0c.0/0000:84:00.0/usb4/4-2/4-2.1/4-2.1:1.0/video4linux/video24
Debug Op Code : 15
Advanced Mode : YES
Product Id : 0B5C
Camera Locked : YES
Usb Type Descriptor : 3.2
Product Line : D400
Asic Serial Number : 322743060552
Firmware Update Id : 322743060552

[MartyG-RealSense]

Hi @HariKrishnan06082k Rather than building realsense-viewer just for checking resolutions, it may be easier if I provide D455-relevant information below.

You can enable depth, RGB and both infrared streams simultaneously. Each stream enabled will add a further processing burden to the computer's CPU though, which could lead to the No Frames Received problem.

Whilst RealSense should be able to work with any Intel or ARM processor, there are no guarantees made about AMD. Past support cases have shown that RealSense cameras should be able to work with AMD Ryzen CPUs but non-Ryzen AMD processors may have a lesser chance of working correctly.

The default depth and infrared resolution is indeed 848x480. The default RGB resolution is 1280x720.

Whilst depth and infrared resolution can be set to 1280x720, it is limited to 30 FPS rather than 60 and so would not be suitable for your 60 FPS / Hz requirement. 60 FPS for depth and infrared is supported on 848x480.

Infrared 2 (the right-side infrared) will be unavailable on USB 2.1 connections and requires full-speed USB 3.

Building librealsense with the RSUSB backend method automatically enables metadata support. I note though that you applied a kernel patch. You do not need to patch the kernel on an RSUSB build and I would recommend against doing so because of the risk of breaking the RSUSB = true librealsense build.

[HariKrishnan06082k]

Hi @MartyG-RealSense thanks for the quick response my goal was to get rs-viewer as a tool , but i was successful building with RSUSB backend. The other question I have is using realsense param for enabling IR streams. Couple of questions to clarify.

1. To enable both IR streams should I set depth_module.depth_profile and depth_module.infra_profile both to 640x480x60 and also enable the streams separately like enable_ir1 and enable_ir2 , I want enable rosparam emitter_on_off to be true as well as I'm interested in using the meta_data generated by IR frames to get only those images when the emitter is turned off.

2. I'm trying to launch the realsense driver from a composable node in ros2 humble using RealsenseNodeFactory Plugin and a config file realsense.yaml but for some reason the parameters which I give in .yaml file is not being reflected when the driver node is launched.

from isaac_ros_launch_utils.all_types import *
import isaac_ros_launch_utils as lu
from nvblox_ros_python_utils.nvblox_constants import NVBLOX_CONTAINER_NAME

def generate_launch_description() -> LaunchDescription:
    args = lu.ArgumentContainer()
    args.add_arg('container_name', NVBLOX_CONTAINER_NAME)
    args.add_arg('run_standalone', 'False')

    config_file = lu.get_path('nvblox_examples_bringup', 'config/sensors/realsense.yaml')

    realsense_splitter_node = ComposableNode(
        namespace='camera',
        name='realsense_splitter_node',
        package='realsense_splitter',
        plugin='nvblox::RealsenseSplitterNode',
        parameters=[{
            'input_qos': 'SYSTEM_DEFAULT',
            'output_qos': 'SYSTEM_DEFAULT'
        }],
        remappings=[
            ('input/infra_1', 'camera/infra1/image_rect_raw'),
            ('input/infra_1_metadata', 'camera/infra1/metadata'),
            ('input/infra_2', 'camera/infra2/image_rect_raw'),
            ('input/infra_2_metadata', 'camera/infra2/metadata'),
            ('input/depth', 'camera/depth/image_rect_raw'),
            ('input/depth_metadata', 'camera/depth/metadata'),
            ('input/pointcloud', 'camera/depth/color/points'),
            ('input/pointcloud_metadata', 'camera/depth/metadata'),
        ])

    realsense_node = ComposableNode(
        namespace='camera',
        package='realsense2_camera',
        plugin='realsense2_camera::RealSenseNodeFactory',
        parameters=[config_file])

    actions = args.get_launch_actions()
    actions.append(
        lu.component_container(
            args.container_name, condition=IfCondition(lu.is_true(args.run_standalone))))
    actions.append(
        lu.load_composable_nodes(
            args.container_name,
            [realsense_splitter_node, realsense_node],
        ))

    return LaunchDescription(actions)

and my yaml file description is as follows:

device_type: ''
serial_no: ''
usb_port_id: ''

rgb_camera:
  profile: '640x480x15'
  color_info_qos: "SYSTEM_DEFAULT"
  color_qos: "SYSTEM_DEFAULT"

depth_module:
  profile: '640x480x60'
  emitter_enabled: 1
  emitter_on_off: true
  depth_qos: "SYSTEM_DEFAULT"
  depth_info_qos: "SYSTEM_DEFAULT"

infra_qos: "SYSTEM_DEFAULT"

enable_accel: false
enable_color: true
enable_depth: true
enable_gyro: false
enable_infra1: true
enable_infra2: true

pointcloud:
  enable: false

pointcloud_texture_index: 0
pointcloud_texture_stream: RS2_STREAM_ANY

enable_sync: false
align_depth: false

3. If I had previously done a patching process how do I properly undo or clean it, I know I can clean the residue from librealsense by going into the build folder and doing sudo make uninstall && make clean.

4. We are planning to use D455 for industrial vision solution, is it ok to buy a USB 3.2 gen 2 cable to work with realsense seamlessly as I've had issues using a USB extender from previous cameras. Sorry to bother you with too many questions. Thanks

[MartyG-RealSense]

Multiple questions are no problem at all. :)

1. It is not necessary to enable the depth topic in order to publish the infra1 and infra2 topics.

The infra topics can be enabled in the launch instruction with enable_infra1:=true enable_infra2:=true or by setting infra1_enabled and infra2_enabled to true in the launch file.

IntelRealSense/realsense-ros#2760 (comment) may be a helpful reference for your interest in emitter_on_off and metadata.

2. At Trying to load custom config_file fails realsense-ros#2561 (comment) one of my Intel RealSense colleagues suggests putting " " quotes around the .yaml path in the config_file instruction.

3. Changing to a different kernel version on your computer should clear the kernel patching. So you could perhaps update to a slightly newer kernel and then reinstall the previous kernel version.

4. High quality, lengthy USB cables that are RealSense-validated are supplied by a company called Newnex. They also offer USB cables with 'active repeater' signal booster components for 10 meters + use and fiber-optic cabling solutions that will work with RealSense cameras up to 100 meters.

https://newnex.com/realsense-3d-camera-connectivity.php

https://www.youtube.com/watch?v=GLQgR1jT04M

[HariKrishnan06082k]

Thank you, I managed to solve the Composable node problem by modifying the launch file and using a custom .yaml file to load it into a runtime container. As for the librealsense 2.56.1 the working solution for me is to use RSUSB method to make it work with current latest kernel of Ubuntu Kernel Version - 6.8.0-48-generic. I think a lot of hassle around this kernel issue would be solved with binary release of librealsense but ig I'll have to wait for the official release. Is there any way to build rs-viewer separately because It would be helpful for my application. For now , in Ubuntu system it looks like you build it only along with the entire source code with DBUILD_GRAPHICAL_EXAMPLES set to true in cmake flags. Or if there's a way to fix the error caused for the specific .o file which builds the graphical tools in source code that also would help. Pretty sure my computer supports X11 forwarding and OpenGL as I've had success before building rs-viewer with other versions of librealsense.

I saw another relevant issue for arm based processor here : https://github.com/IntelRealSense/librealsense/issues/11807
but the thread was closed with making the DBUILD_GRAPHICAL_EXAMPLES to false. I would ideally want to solve this issue so I don't compromise on losing out graphical tools when your processor actually supports it. Even though this error is probably specific to cmake I would like to keep the thread open to get a solution for making changes to respective CmakeFiles or flags that would enable the successful build of graphical tools. Thanks so much for your response.

Attaching a sample nodefactory_plugin ros launch file which worked seamlessly for testing to make it easy for others to use.

from launch import LaunchDescription
from launch.actions import DeclareLaunchArgument
from launch.substitutions import LaunchConfiguration
from launch_ros.actions import ComposableNodeContainer
from launch_ros.descriptions import ComposableNode

# Make sure required packages can be found
rs_node_class=  'RealSenseNodeFactory'

realsense_node_params = [{'name': 'serial_no',              'default': "''", 'description': 'choose device by serial number'},
                         {'name': 'usb_port_id',            'default': "''", 'description': 'choose device by usb port id'},
                         {'name': 'device_type',            'default': "''", 'description': 'choose device by type'},
                         {'name': 'log_level',              'default': 'info', 'description': 'debug log level [DEBUG|INFO|WARN|ERROR|FATAL]'},
                         {'name': 'rgb_camera.profile',     'default': '0,0,0', 'description': 'color image width'},
                         {'name': 'rgb_camera.color_profile',     'default': '640,480,15', 'description': 'color stream profile'}, #modified
                         {'name': 'depth_module.depth_profile',   'default': '640,480,60', 'description': 'depth stream profile'}, #modified
                         {'name': 'depth_module.depth_format',    'default': 'Z16', 'description': 'depth stream format'},#modified
                         {'name': 'depth_module.emitter_enabled',  'default': '1', 'description': 'enable/disable emitter'}, #modified
                         {'name': 'depth_module.emitter_on_off',  'default': 'true', 'description': 'enable/disable emitter'}, #modified
                         {'name': 'depth_module.infra_profile',   'default': '640,480,60', 'description': 'infra streams (0/1/2) profile'}, #modified
                         {'name': 'enable_color',           'default': 'true', 'description': 'enable color stream'},
                         {'name': 'enable_depth',           'default': 'true', 'description': 'enable depth stream'},
                         {'name': 'enable_infra',           'default': 'false', 'description': 'enable infra stream'},
                         {'name': 'enable_infra1',          'default': 'true', 'description': 'enable infra1 stream'},
                         {'name': 'enable_infra2',          'default': 'true', 'description': 'enable infra2 stream'},
                         {'name': 'enable_gyro',            'default': 'false', 'description': "enable gyro stream"},
                         {'name': 'enable_accel',           'default': 'false', 'description': "enable accel stream"},
                         {'name': 'unite_imu_method',       'default': "0", 'description': '[0-None, 1-copy, 2-linear_interpolation]'},
                         {'name': 'intra_process_comms',    'default': 'true', 'description': "enable intra-process communication"},
                         {'name': 'enable_sync',            'default': 'false', 'description': "'enable sync mode'"},
                         {'name': 'pointcloud.enable',      'default': 'false', 'description': ''},
                         {'name': 'enable_rgbd',            'default': 'false', 'description': "'enable rgbd topic'"},
                         {'name': 'align_depth.enable',     'default': 'false', 'description': "'enable align depth filter'"},
                         {'name': 'publish_tf',             'default': 'true', 'description': '[bool] enable/disable publishing static & dynamic TF'},
                         {'name': 'tf_publish_rate',        'default': '1.0', 'description': '[double] rate in HZ for publishing dynamic TF'},
                        ]

def declare_configurable_parameters(parameters):
    return [DeclareLaunchArgument(param['name'], default_value=param['default'], description=param['description']) for param in parameters]

def set_configurable_parameters(parameters):
    return dict([(param['name'], LaunchConfiguration(param['name'])) for param in parameters])

def generate_launch_description():
        return LaunchDescription(declare_configurable_parameters(realsense_node_params) + [
            ComposableNodeContainer(
                name='realsense_container',
                namespace='',
                package='rclcpp_components',
                executable='component_container',
                composable_node_descriptions=[
                    ComposableNode(
                        package='realsense2_camera',
                        namespace='',
                        plugin='realsense2_camera::' + rs_node_class,
                        name="camera",
                        parameters=[set_configurable_parameters(realsense_node_params)],
                        extra_arguments=[{'use_intra_process_comms': LaunchConfiguration("intra_process_comms")}])
                ],
                output='screen',
                emulate_tty=True, # needed for display of logs
                arguments=['--ros-args', '--log-level', LaunchConfiguration('log_level')],
            )
        ])

One warning which I couldn't figure out how to fix is this ,
" Failed to set value to sensor: depth_module.emitter_on_off = 0[0, 1]
[component_container-1] hwmon command 0x7b( 17 0 0 0 ) failed (response -9= No expected user action)"

coming from the configurable ros2 param depth_module.emitter_on_off true.

Also if you could provide a quick timeline update for upcoming dkms binary release for latest kernel of Noble and Jammy (Kernel Version - 6.8.0-48-generic) that would be appreciated.

[MartyG-RealSense]

I'm pleased to hear that you have made significant progress.

The source code of the realsense-viewer tool is located at the link below.

https://github.com/IntelRealSense/librealsense/tree/master/tools/realsense-viewer

I do not have information about a time estimate for release of a librealsense version that has dkms packages. Usually, there is a gap of around 6 months between major releases, and 2.55.1 was released in April 2024. So if the roughly 6 month gap is maintained then it might be November or December when the next 'final' release equipped with packages is released, though that is only speculation and not official advice.

Regarding the DBUILD_GRAPHICAL_EXAMPLES=true parameter not working on your computer: as you have an AMD CPU rather than an Intel or Arm one, there are not guarantees that I can provide about how well librealsense will work with an AMD chip even if changes were made to CMake, unfortunately.

It is possible to build the SDK examples individually with CMake if a specific example's C++ script has its code copied into a new project and then you create a special CMakeLists.txt file that contains the necessary linkages to librealsense. The link below has a project that demonstrates this. The example's .cpp C++ script file will be unchanged from the original when copied into the new project and only the CMakeLists.txt file of the project needs to be changed to enable it to run on its own as a standalone application.

https://github.com/zivsha/librealsense/tree/getting_started_example/examples/getting-started

Regarding setting emitter_on_off in ROS, does entering the command below into the ROS terminal work for you?

ros2 param set /camera/camera depth_module.emitter_on_off true

[HariKrishnan06082k]

Hi, yes the command ros2 param set seems to work on command line after the node is launched. But was wondering if there's a way to launch it through launch file or while starting a ros driver node class. As for the rs-viewer tool if you think AMD might be the issue then don't worry about it but I find it weird that it is able to build inside a docker container without any error , must be using the same kernel modules of my host system in that case.

[MartyG-RealSense]

You could try defining the setting in a .yaml configuration file and import it into the ROS2 launch file with the config_file parameter. Please see IntelRealSense/realsense-ros#1640 and IntelRealSense/realsense-ros#2561 for examples of doing so.

[HariKrishnan06082k]

Hi, thanks for replying patiently to all the questions. Feel free to close the issue if you are sure it might be an AMD issue for building graphical tools. Have a nice rest of your day. I will figure out the ros2 param issue :)

[MartyG-RealSense]

There is no way to confirm that the AMD CPU is related to the graphical tools issue. I can only speculate about the possibility of problems occurring based on AMD CPUs not being officially validated for RealSense cameras.

[HariKrishnan06082k]

Hi sorry to open this issue again but it looks like the emitter_on and emitter_on_off is not taking effect in config file also as it pretty much does the same as rs_launch.py there is no concept of rosparam tag in ros2 I think. My intention is to get infrared images without projector on (basically dots) . To do this the only thing I figured out which worked is to launch the IR streams then once the realsense_node is up then do runtime ros2 param set /camera/camera depth_module.emitter_enabled 1 followed by ros2 param set /camera/camera depth_module.emitter_on_off true . Now i get clean IR images without projector (to emulate stereo pair).

Is there a way to make it work in Launch Description as in rs_launch.launch.py

[MartyG-RealSense]

Another way to turn the projector off is to set the laser_power parameter to a value of '0'.