Python3 ToF Bindings Seg Fault on NXP
pclass-sensonix opened this issue · comments
Hello - I am trying to use the Python3 ToF bindings on the NXP i.MX8. I can successfully capture frames from the ADSD3500 Camera with V4L2 and the following script on the i.MX8:
v4l2-ctl --set-ctrl=operating_mode=0 -d /dev/v4l-subdev1
v4l2-ctl --set-ctrl=phase_depth_bits=4 -d /dev/v4l-subdev1
v4l2-ctl --set-ctrl=ab_bits=6 -d /dev/v4l-subdev1
v4l2-ctl --set-ctrl=confidence_bits=0 -d /dev/v4l-subdev1
v4l2-ctl --set-ctrl=ab_averaging=0 -d /dev/v4l-subdev1
v4l2-ctl --set-ctrl=depth_enable=0 -d /dev/v4l-subdev1
v4l2-ctl --device /dev/video0 --set-fmt-video=width=2048,height=2560,pixelformat=BA81 --stream-mmap --stream-to=mode0.bin --stream-count=$nr_frames
--
[ 673.848452] bypass csc
[ 673.848462] input fmt BA81
[ 673.848464] output fmt BA81
[ 674.262957] isi-capture 32e00000.isi:cap_device: ISI channel[0] has stopped
However, when I try to run the python3 example script below:
import aditofpython as tof
import numpy as np
system = tof.System()
cameras = []
status = system.getCameraList(cameras)
print("system.getCameraList()", status)
camera1 = cameras[0]
modes = []
status = camera1.getAvailableModes(modes)
print("system.getAvailableModes()", status)
print(modes)
camera1.setControl("initialization_config", "./tof-viewer_config.json")
status = camera1.initialize()
print("camera1.initialize()", status)
types = []
status = camera1.getAvailableFrameTypes(types)
print("system.getAvailableFrameTypes()", status)
print(types)
camDetails = tof.CameraDetails()
status = camera1.getDetails(camDetails)
print("system.getDetails()", status)
print("camera1 details:", "id:", camDetails.cameraId, "connection:", camDetails.connection)
status = camera1.setFrameType(types[0])
print("camera1.setFrameType()", status)
status = camera1.setMode(modes[0])
print("camera1.setMode()", status)
I get the following error:
WARNING: Logging before InitGoogleLogging() is written to STDERR
I0324 10:38:33.923457 658 system_impl.cpp:90] SDK built with websockets version:4.1.6-v4.1.6
I0324 10:38:33.923589 658 sensor_enumerator_imx.cpp:116] Looking for sensors on the target
I0324 10:38:33.923907 658 sensor_enumerator_imx.cpp:137] Looking at: /dev/media0 for an eligible TOF camera
I0324 10:38:33.964524 658 sensor_enumerator_imx.cpp:153] Considering: /dev/media0 an eligible TOF camera
system.getCameraList() Status.Ok
system.getAvailableModes() Status.Ok
['pcmmp', 'srqmp', 'lrqmp', 'srmp', 'lrmp']
I0324 10:38:33.965276 658 camera_itof.cpp:146] Initializing camera
I0324 10:38:33.965310 658 adsd3500_sensor.cpp:207] Opening device
I0324 10:38:33.965328 658 adsd3500_sensor.cpp:225] Looking for the following cards:
I0324 10:38:33.965343 658 adsd3500_sensor.cpp:227] mxc-isi-cap
I0324 10:38:33.965356 658 adsd3500_sensor.cpp:239] device: /dev/video0 subdevice: /dev/v4l-subdev1
W0324 10:38:33.966929 658 adsd3500_sensor.cpp:990] Reading Adsd3500 error errno: 5 error: Input/output error
I0324 10:38:39.818653 658 adsd3500_interrupt_notifier.cpp:33] Received signal 0 from kernel
I0324 10:38:39.822163 658 adsd3500_sensor.cpp:1554] statusRegister:0(Adsd3500Status::OK)
I0324 10:38:39.822237 658 adsd3500_interrupt_notifier.cpp:33] Received signal 0 from kernel
I0324 10:38:39.825736 658 adsd3500_sensor.cpp:1554] statusRegister:0(Adsd3500Status::OK)
I0324 10:38:39.838490 658 adsd3500_sensor.cpp:335] ADSD3500 is ready to communicate with.
I0324 10:38:39.852028 658 mode_info.cpp:148] Using new modes table for adsd3500.
I0324 10:38:39.860694 658 mode_info.cpp:144] Using new mixed modes table for adsd3500.
I0324 10:38:39.860736 658 mode_info.cpp:144] Using new mixed modes table for adsd3500.
I0324 10:38:40.028141 658 camera_itof.cpp:355] Current adsd3500 firmware version is: 4.2.1.0
I0324 10:38:40.028206 658 camera_itof.cpp:357] Current adsd3500 firmware git hash is: a59afad6d6c8e7fb153a2db8cd8840359f17195a
I0324 10:38:40.032168 658 camera_itof.cpp:1777] Current Depth ini file is: ./RawToDepthAdsd3500_sr-native.ini
I0324 10:38:40.034569 658 camera_itof.cpp:450] Camera FPS set from Json file at: 10
I0324 10:38:40.035817 658 camera_itof.cpp:462] Camera initialized
camera1.initialize() Status.Ok
system.getAvailableFrameTypes() Status.Ok
['sr-native', 'lr-native', 'sr-qnative', 'lr-qnative', 'pcm-native', 'sr-mixed', 'lr-mixed']
system.getDetails() Status.Ok
camera1 details: id: connection: ConnectionType.???
I0324 10:38:40.041918 658 camera_itof.cpp:539] Chosen mode: sr-native
I0324 10:38:40.041970 658 camera_itof.cpp:593] Using ini file: ./RawToDepthAdsd3500_sr-native.ini
I0324 10:38:40.047096 658 adsd3500_sensor.cpp:207] Opening device
I0324 10:38:40.047125 658 adsd3500_sensor.cpp:225] Looking for the following cards:
I0324 10:38:40.047142 658 adsd3500_sensor.cpp:227] mxc-isi-cap
I0324 10:38:40.047160 658 adsd3500_sensor.cpp:239] device: /dev/video0 subdevice: /dev/v4l-subdev1
I0324 10:38:40.047346 658 adsd3500_sensor.cpp:463] Setting camera mode to sr-native
I0324 10:38:40.058508 658 camera_itof.cpp:924] initComputeLibrary
Segmentation fault
The working directory looks like this:
ls -rlt
total 20
-rwxr-x--- 1 root root 912 Mar 24 10:26 test.py
-rw-r----- 1 root root 117 Mar 24 10:28 tof-viewer_config.json
-rw-r----- 1 root root 406 Mar 24 10:30 RawToDepthAdsd3500_sr-native.ini
-rw-r----- 1 root root 8136 Mar 24 10:32 libtofi_compute.so
The config files look like this:
root@imx8-gateway:/test# cat tof-viewer_config.json
{
"skip_network_cameras": "off",
"DEPTH_INI": "./RawToDepthAdsd3500_sr-native.ini",
"FPS": "10",
"FSYNC_MODE": "1"
}
root@imx8-gateway:/test# cat RawToDepthAdsd3500_sr-native.ini
abThreshMin=1.0
abSumThresh=25.0
confThresh=25.0
radialThreshMin=30.0
radialThreshMax=4200.0
jblfApplyFlag=1
jblfWindowSize=5
jblfGaussianSigma=10.0
jblfExponentialTerm=5.0
jblfMaxEdge=12.0
jblfABThreshold=10.0
headerSize=0
deltaCompEnable=0
inputFormat=mipiRaw12_8
depthComputeIspEnable=0
partialDepthEnable=0
interleavingEnable=0
bitsInPhaseOrDepth=12
bitsInConf=0
bitsInAB=16
phaseInvalid=0
xyzEnable=1
Please tell me how to fix the seg fault with the Python bindings. Thank you.
The strace command gives this information:
write(2, "I0324 10:33:35.942015 670 adsd"..., 86I0324 10:33:35.942015 670 adsd3500_sensor.cpp:463] Setting camera mode to sr-native
) = 86
ioctl(5, VIDIOC_S_CTRL, {id=V4L2_CTRL_CLASS_USER+0x19e0, value=0 => 0}) = 0
ioctl(4, VIDIOC_REQBUFS, {type=V4L2_BUF_TYPE_VIDEO_CAPTURE, memory=V4L2_MEMORY_MMAP, count=0 => 0}) = 0
ioctl(4, VIDIOC_S_FMT, {type=V4L2_BUF_TYPE_VIDEO_CAPTURE, fmt.pix={width=2048, height=2560, pixelformat=v4l2_fourcc('B', 'A', '8', '1') /* V4L2_PIX_FMT_SBGGR8 */, field=V4L2_FIELD_ANY, bytesperline=0, sizeimage=0, colorspace=V4L2_COLORSPACE_DEFAULT}} => {fmt.pix={width=2048, height=2560, pixelformat=v4l2_fourcc('B', 'A', '8', '1') /* V4L2_PIX_FMT_SBGGR8 */, field=V4L2_FIELD_NONE, bytesperline=2048, sizeimage=5242880, colorspace=V4L2_COLORSPACE_SRGB}}) = 0
ioctl(4, VIDIOC_REQBUFS, {type=V4L2_BUF_TYPE_VIDEO_CAPTURE, memory=V4L2_MEMORY_MMAP, count=4 => 4}) = 0
ioctl(4, VIDIOC_QUERYBUF, {type=V4L2_BUF_TYPE_VIDEO_CAPTURE, index=0, memory=V4L2_MEMORY_MMAP, m.offset=0, length=5242880, bytesused=0, flags=V4L2_BUF_FLAG_TIMESTAMP_MONOTONIC|V4L2_BUF_FLAG_TSTAMP_SRC_EOF, ...}) = 0
mmap(NULL, 5242880, PROT_READ|PROT_WRITE, MAP_SHARED, 4, 0) = 0xffff914a2000
ioctl(4, VIDIOC_QUERYBUF, {type=V4L2_BUF_TYPE_VIDEO_CAPTURE, index=1, memory=V4L2_MEMORY_MMAP, m.offset=0x500000, length=5242880, bytesused=0, flags=V4L2_BUF_FLAG_TIMESTAMP_MONOTONIC|V4L2_BUF_FLAG_TSTAMP_SRC_EOF, ...}) = 0
mmap(NULL, 5242880, PROT_READ|PROT_WRITE, MAP_SHARED, 4, 0x500000) = 0xffff90fa2000
ioctl(4, VIDIOC_QUERYBUF, {type=V4L2_BUF_TYPE_VIDEO_CAPTURE, index=2, memory=V4L2_MEMORY_MMAP, m.offset=0xa00000, length=5242880, bytesused=0, flags=V4L2_BUF_FLAG_TIMESTAMP_MONOTONIC|V4L2_BUF_FLAG_TSTAMP_SRC_EOF, ...}) = 0
mmap(NULL, 5242880, PROT_READ|PROT_WRITE, MAP_SHARED, 4, 0xa00000) = 0xffff90aa2000
ioctl(4, VIDIOC_QUERYBUF, {type=V4L2_BUF_TYPE_VIDEO_CAPTURE, index=3, memory=V4L2_MEMORY_MMAP, m.offset=0xf00000, length=5242880, bytesused=0, flags=V4L2_BUF_FLAG_TIMESTAMP_MONOTONIC|V4L2_BUF_FLAG_TSTAMP_SRC_EOF, ...}) = 0
mmap(NULL, 5242880, PROT_READ|PROT_WRITE, MAP_SHARED, 4, 0xf00000) = 0xffff905a2000
gettid() = 670
write(2, "I0324 10:33:35.967603 670 came"..., 68I0324 10:33:35.967603 670 camera_itof.cpp:924] initComputeLibrary
) = 68
--- SIGSEGV {si_signo=SIGSEGV, si_code=SEGV_MAPERR, si_addr=0x88e60bfabe104} ---
+++ killed by SIGSEGV +++
Segmentation fault
Clinfo shows the OpenCL library and support as well:
/usr/bin/camera# clinfo
Number of platforms 1
Platform Name Vivante OpenCL Platform
Platform Vendor Vivante Corporation
Platform Version OpenCL 3.0 V6.4.3.p2.336687
Platform Profile FULL_PROFILE
Platform Extensions cl_khr_byte_addressable_store cl_khr_fp16
Platform Host timer resolution 0ns
Hello,
Please go to the build folder and check if you have the USE_DEPTH_COMPUTE_STUBS flag enabled. You can do this by running cmake -LA in the build folder.
If it's set to 1/on you need to set it to off and rebuild the project.
Before you rebuild the project you need to copy the libtofi_compute.so and libtofi_config.so from the installer onto the nxp
in a folder called libs right next to the /home/analog/Workspace/ToF folder.
After you copy the libraries go to the build folder and run:
cmake -DUSE_DEPTH_COMPUTE_STUBS=0 -DWITH_PYTHON=1 ..
make -j4
You can get all the files required to run a python script if you go to the build/bindings/python/examples/first_frame folder.
You need the libaditof.so* files, the libtofi_config.so, libtofi_compute.so and the config folder.
Hi - thanks for the input.
I made sure to make my directory structure look like the first_frame example. Now I am getting a segmentation fault in a different spot.
Here is the output of the first_frame script:
python3 ./first_frame.py ./config.json
Looking for camera on UVC. Will use ./config.json.
WARNING: Logging before InitGoogleLogging() is written to STDERR
I0324 11:04:33.411160 801 system_impl.cpp:90] SDK built with websockets version:4.1.6-v4.1.6
I0324 11:04:33.411204 801 sensor_enumerator_imx.cpp:116] Looking for sensors on the target
I0324 11:04:33.411525 801 sensor_enumerator_imx.cpp:137] Looking at: /dev/media0 for an eligible TOF camera
I0324 11:04:33.438766 801 sensor_enumerator_imx.cpp:153] Considering: /dev/media0 an eligible TOF camera
system.getCameraList() Status.Ok
camera1.setControl() Status.Ok
I0324 11:04:33.439373 801 camera_itof.cpp:146] Initializing camera
I0324 11:04:33.439400 801 adsd3500_sensor.cpp:207] Opening device
I0324 11:04:33.439416 801 adsd3500_sensor.cpp:225] Looking for the following cards:
I0324 11:04:33.439429 801 adsd3500_sensor.cpp:227] mxc-isi-cap
I0324 11:04:33.439442 801 adsd3500_sensor.cpp:239] device: /dev/video0 subdevice: /dev/v4l-subdev1
I0324 11:04:33.449576 801 adsd3500_sensor.cpp:335] ADSD3500 is ready to communicate with.
I0324 11:04:33.461863 801 mode_info.cpp:148] Using new modes table for adsd3500.
I0324 11:04:33.471195 801 mode_info.cpp:144] Using new mixed modes table for adsd3500.
I0324 11:04:33.471231 801 mode_info.cpp:144] Using new mixed modes table for adsd3500.
I0324 11:04:33.830659 801 camera_itof.cpp:355] Current adsd3500 firmware version is: 4.2.1.0
I0324 11:04:33.830741 801 camera_itof.cpp:357] Current adsd3500 firmware git hash is: a59afad6d6c8e7fb153a2db8cd8840359f17195a
I0324 11:04:33.830858 801 camera_itof.cpp:1833] Current Depth ini file is: RawToDepthAdsd3500_lr-qnative.ini
I0324 11:04:33.832592 801 camera_itof.cpp:450] Camera FPS set from Json file at: 10
I0324 11:04:33.833477 801 camera_itof.cpp:462] Camera initialized
camera1.initialize() Status.Ok
camera1.getAvailableFrameTypes() Status.Ok
['sr-native', 'lr-native', 'sr-qnative', 'lr-qnative', 'pcm-native', 'sr-mixed', 'lr-mixed']
camera1.getDetails() Status.Ok
camera1 details: id: connection: ConnectionType.???
I0324 11:04:33.839605 801 camera_itof.cpp:539] Chosen mode: lr-qnative
I0324 11:04:33.839637 801 camera_itof.cpp:593] Using ini file: RawToDepthAdsd3500_lr-qnative.ini
I0324 11:04:33.844734 801 adsd3500_sensor.cpp:207] Opening device
I0324 11:04:33.844756 801 adsd3500_sensor.cpp:225] Looking for the following cards:
I0324 11:04:33.844772 801 adsd3500_sensor.cpp:227] mxc-isi-cap
I0324 11:04:33.844790 801 adsd3500_sensor.cpp:239] device: /dev/video0 subdevice: /dev/v4l-subdev1
I0324 11:04:33.845203 801 adsd3500_sensor.cpp:463] Setting camera mode to lr-qnative
I0324 11:04:33.855988 801 camera_itof.cpp:924] initComputeLibrary
Segmentation fault
The line that it seg faults on is this one:
m_tofi_config = InitTofiConfig_isp((ConfigFileData *)&depth_ini,
convertedMode, &status,
m_xyz_dealias_data);
My config.json file looks like this:
cat ./config.json
{
"VAUX_POWER_VOLTAGE": "18",
"skip_network_cameras": "off",
"camera_ip": "10.42.0.1",
"DEPTH_INI": "RawToDepthAdsd3500_lr-qnative.ini",
"FPS": "10",
"FSYNC_MODE": "1"
}
Can you please help me figure out what is wrong? Thanks.
I did make sure that USE_DEPTH_COMPUTE_STUBS was off.
grep -R "USE_DEPTH_COMPUTE_STUBS" *
:USE_DEPTH_COMPUTE_STUBS:BOOL=offCMakeCache.txt:USE_DEPTH_COMPUTE_STUBS:BOOL=off
Here is the directory structure that I have:
root@imx8-gateway:/usr/bin/camera/test# ls -rlt
total 38580
-rw-r----- 1 root root 457 Mar 24 10:31 tof_tembin_adsd3500_new_modes_config.json
-rw-r----- 1 root root 457 Mar 24 10:31 tof_crosby_adsd3500_new_modes_config.json
drwxr-x--- 2 root root 4096 Mar 24 10:38 config
-rwxr-xr-x 1 root root 912 Mar 24 10:42 python-test.py
-rwxr-x--- 1 root root 1899 Mar 24 10:45 first_frame.py
-rw-r----- 1 root root 405 Mar 24 10:48 RawToDepthAdsd3500_sr-native.ini
-rw-r----- 1 root root 399 Mar 24 10:49 RawToDepthAdsd3500_lr-qnative.ini
-rw-r----- 1 root root 195 Mar 24 11:10 config.json
-rw-r----- 1 root root 13151608 Mar 24 11:12 libaditof.so.4.2.0
-rw-r----- 1 root root 13151608 Mar 24 11:12 libaditof.so.1.0
-rw-r----- 1 root root 13151608 Mar 24 11:12 libaditof.so
-rw-r----- 1 root root 8192 Mar 24 11:12 libtofi_config.so
-rw-r----- 1 root root 8136 Mar 24 11:12 libtofi_compute.so
Here is the configs directory:
root@imx8-gateway:/usr/bin/camera/test# ls -R config
config:
RawToDepthAdsd3030_lr-mixed.ini RawToDepthAdsd3030_sr-mixed.ini RawToDepthAdsd3500_lr-mixed.ini RawToDepthAdsd3500_mp.ini RawToDepthAdsd3500_sr-native.ini
RawToDepthAdsd3030_lr-native.ini RawToDepthAdsd3030_sr-native.ini RawToDepthAdsd3500_lr-native.ini RawToDepthAdsd3500_qmp.ini RawToDepthAdsd3500_sr-qnative.ini
RawToDepthAdsd3030_lr-qnative.ini RawToDepthAdsd3030_sr-qnative.ini RawToDepthAdsd3500_lr-qnative.ini RawToDepthAdsd3500_sr-mixed.ini RawToDepthAdsd_pcm-native.ini
It does not seem to matter what the camera mode is. If I change the mode, the ini is found, but the seg fault still occurs.
I0324 10:28:59.139642 604 camera_itof.cpp:1833] Current Depth ini file is: ./RawToDepthAdsd3500_mp.ini
I0324 10:28:59.141372 604 camera_itof.cpp:450] Camera FPS set from Json file at: 10
I0324 10:28:59.142261 604 camera_itof.cpp:462] Camera initialized
camera1.initialize() Status.Ok
camera1.getAvailableFrameTypes() Status.Ok
['sr-native', 'lr-native', 'sr-qnative', 'lr-qnative', 'pcm-native', 'sr-mixed', 'lr-mixed']
camera1.getDetails() Status.Ok
camera1 details: id: connection: ConnectionType.???
I0324 10:28:59.148275 604 camera_itof.cpp:539] Chosen mode: lr-qnative
I0324 10:28:59.148306 604 camera_itof.cpp:593] Using ini file: ./RawToDepthAdsd3500_mp.ini
I0324 10:28:59.153419 604 adsd3500_sensor.cpp:207] Opening device
I0324 10:28:59.153445 604 adsd3500_sensor.cpp:225] Looking for the following cards:
I0324 10:28:59.153465 604 adsd3500_sensor.cpp:227] mxc-isi-cap
I0324 10:28:59.153484 604 adsd3500_sensor.cpp:239] device: /dev/video0 subdevice: /dev/v4l-subdev1
I0324 10:28:59.153669 604 adsd3500_sensor.cpp:463] Setting camera mode to lr-qnative
I0324 10:28:59.167840 604 camera_itof.cpp:924] initComputeLibrary
I0324 10:28:59.167896 604 camera_itof.cpp:931] here 1
I0324 10:28:59.167907 604 camera_itof.cpp:938] here 2
I0324 10:28:59.167918 604 camera_itof.cpp:942] here 3
I0324 10:28:59.167929 604 camera_itof.cpp:949] here 4
I0324 10:28:59.167943 604 camera_itof.cpp:959] here 5
I0324 10:28:59.167956 604 camera_itof.cpp:965] here 6
I0324 10:28:59.167968 604 camera_itof.cpp:969] here 7
Segmentation fault
I have done another test. I downloaded your official NXP image, burned it onto a flash card, mounted that flash card, and extracted the following two directories:
/home/analog/Workspace/ToF/build/examples
/home/analog/Workspace/ToF/bindings/python/examples/first_frame
I copied these wholesale to my NXP dev kit. It doesn't matter if I run the Python bindings for 'first-frame' or the C++ version. They both seg fault.
root@imx8-gateway:/home/first_frame# ls
CMakeFiles aditofpython.cpython-38-aarch64-linux-gnu.so common libaditof.so.1.0 libtofi_config.so
Makefile buffer.pb.cc config libaditof.so.4.2.0 libwebsockets.so.14
aditof-config-version.cmake buffer.pb.h first_frame.py libglog.so.1 usb_buffer.pb.cc
aditof-config.cmake cmake_install.cmake libaditof.so libtofi_compute.so usb_buffer.pb.h
root@banner-fusion-imx8-gateway:/home/first_frame# python3 ./first_frame.py ./config/config_adsd3500_adsd3030_new_modes.json
WARNING: Logging before InitGoogleLogging() is written to STDERR
I20210324 13:20:43.599526 1513 system_impl.cpp:90] SDK built with websockets version:3.1.0
I20210324 13:20:43.599650 1513 sensor_enumerator_imx.cpp:116] Looking for sensors on the target
I20210324 13:20:43.600034 1513 sensor_enumerator_imx.cpp:137] Looking at: /dev/media0 for an eligible TOF camera
I20210324 13:20:43.609023 1513 sensor_enumerator_imx.cpp:153] Considering: /dev/media0 an eligible TOF camera
...
write(2, "I20210324 13:23:00.930301 1545 "..., 72I20210324 13:23:00.930301 1545 camera_itof.cpp:924] initComputeLibrary
) = 72
--- SIGSEGV {si_signo=SIGSEGV, si_code=SEGV_ACCERR, si_addr=0xffffa93ff778} ---
+++ killed by SIGSEGV +++
Segmentation fault
I only have the official libraries distributed in the official 4.2 release on the NXP.
Any ideas why this is seg faulting?
@SeptimiuVana - I used the official ADTF3175D - Release v4.2.0 image and flashed it to a new sd card. I booted the device and it still seg faults. Could the camera be causing this? The camera seems fine because I can take frames with v4l2-ctl, but I cannot use the ADI sdk:
analog@aditof:~/Workspace/ToF/build/examples/first-frame$ ./first-frame config/config_adsd3500_adsd3030_new_modes.json
I20230315 15:07:35.372468 582 main.cpp:78] SDK version: 4.2.0 | branch: HEAD | commit: b260ad46
I20230315 15:07:35.373270 582 system_impl.cpp:90] SDK built with websockets version:3.1.0
I20230315 15:07:35.373414 582 sensor_enumerator_imx.cpp:116] Looking for sensors on the target
I20230315 15:07:35.373764 582 sensor_enumerator_imx.cpp:137] Looking at: /dev/media0 for an eligible TOF camera
I20230315 15:07:35.379109 582 sensor_enumerator_imx.cpp:153] Considering: /dev/media0 an eligible TOF camera
W20230315 15:07:35.380399 582 adsd3500_interrupt_notifier.cpp:57] Failed to open the debug sysfs. Interrupts support will not be available!
I20230315 15:07:35.381754 582 camera_itof.cpp:146] Initializing camera
I20230315 15:07:35.381847 582 adsd3500_sensor.cpp:207] Opening device
I20230315 15:07:35.381908 582 adsd3500_sensor.cpp:225] Looking for the following cards:
I20230315 15:07:35.381948 582 adsd3500_sensor.cpp:227] mxc-isi-cap
I20230315 15:07:35.381997 582 adsd3500_sensor.cpp:239] device: /dev/video0 subdevice: /dev/v4l-subdev1
I20230315 15:07:35.391999 582 adsd3500_sensor.cpp:335] ADSD3500 is ready to communicate with.
I20230315 15:07:35.403398 582 mode_info.cpp:148] Using new modes table for adsd3500.
I20230315 15:07:35.412307 582 mode_info.cpp:144] Using new mixed modes table for adsd3500.
I20230315 15:07:35.412420 582 mode_info.cpp:144] Using new mixed modes table for adsd3500.
I20230315 15:07:35.579528 582 camera_itof.cpp:355] Current adsd3500 firmware version is: 4.2.1.0
I20230315 15:07:35.579658 582 camera_itof.cpp:357] Current adsd3500 firmware git hash is: a59afad6d6c8e7fb153a2db8cd8840359f17195a
I20230315 15:07:35.580209 582 camera_itof.cpp:1766] Found Depth ini file: ./config/RawToDepthAdsd3030_sr-native.ini
I20230315 15:07:35.580366 582 camera_itof.cpp:1766] Found Depth ini file: ./config/RawToDepthAdsd3030_lr-native.ini
I20230315 15:07:35.580433 582 camera_itof.cpp:1766] Found Depth ini file: ./config/RawToDepthAdsd3030_sr-qnative.ini
I20230315 15:07:35.580550 582 camera_itof.cpp:1766] Found Depth ini file: ./config/RawToDepthAdsd3030_lr-qnative.ini
I20230315 15:07:35.580608 582 camera_itof.cpp:1766] Found Depth ini file: ./config/RawToDepthAdsd3030_sr-mixed.ini
I20230315 15:07:35.580670 582 camera_itof.cpp:1766] Found Depth ini file: ./config/RawToDepthAdsd3030_lr-mixed.ini
I20230315 15:07:35.580730 582 camera_itof.cpp:1766] Found Depth ini file: ./config/RawToDepthAdsd_pcm-native.ini
I20230315 15:07:35.580799 582 camera_itof.cpp:1777] Current Depth ini file is: ./config/RawToDepthAdsd3030_lr-mixed.ini
I20230315 15:07:35.583379 582 camera_itof.cpp:450] Camera FPS set from Json file at: 10
I20230315 15:07:35.584336 582 camera_itof.cpp:462] Camera initialized
I20230315 15:07:35.584461 582 main.cpp:117] SD card image version: microsd-4.2.0-b260ad46.img
I20230315 15:07:35.584515 582 main.cpp:118] Kernel version: lf-5.10.72-2.2.0
I20230315 15:07:35.584565 582 main.cpp:119] U-Boot version: imx_v2020.04_5.4.70_2.3.0
I20230315 15:07:35.590389 582 camera_itof.cpp:539] Chosen mode: lr-qnative
I20230315 15:07:35.590471 582 camera_itof.cpp:593] Using ini file: ./config/RawToDepthAdsd3030_lr-qnative.ini
I20230315 15:07:35.595618 582 adsd3500_sensor.cpp:207] Opening device
I20230315 15:07:35.595687 582 adsd3500_sensor.cpp:225] Looking for the following cards:
I20230315 15:07:35.595726 582 adsd3500_sensor.cpp:227] mxc-isi-cap
I20230315 15:07:35.595767 582 adsd3500_sensor.cpp:239] device: /dev/video0 subdevice: /dev/v4l-subdev1
I20230315 15:07:35.595984 582 adsd3500_sensor.cpp:463] Setting camera mode to lr-qnative
I20230315 15:07:35.602766 582 camera_itof.cpp:924] initComputeLibrary
Segmentation fault
analog@aditof:~/Workspace/ToF/build/examples/first-frame$
Hello, the nxp image contains by default those depth compute stub libraries. They are used only for building purposes. That's the cause of the seg fault.
Inside the install package, right next to the ADTF3175D - Release v4.2.0 image file there should be a folder called depth_compute.
In that folder you will find an installer for the depth compute library that you want to use. Please get the libtofi_compute.so and libtofi_config.so from there.
Hello @SeptimiuVana, thanks for the response. I have installed the depth_compute prebuilt libraries. They are the only ones under /usr/lib on the NXP and it still seg faults. I have pasted the md5sums below. Please see the screenshot also.
root@imx8-gateway:/# find . -name "libtofi*" | xargs md5sum
7519bde29e55ba3fa91a112c1f3279b8 ./usr/lib/libtofi_config.so
da70340cc2504ebc376dec82ee06dc32 ./usr/lib/libtofi_compute.so
root@imx8-gateway:/#
Test:
I0324 10:37:46.168463 663 adsd3500_sensor.cpp:227] mxc-isi-cap
I0324 10:37:46.168510 663 adsd3500_sensor.cpp:239] device: /dev/video0 subdevice: /dev/v4l-subdev1
I0324 10:37:46.168715 663 adsd3500_sensor.cpp:463] Setting camera mode to sr-native
I0324 10:37:46.179436 663 camera_itof.cpp:924] initComputeLibrary
Segmentation fault
Is there anything else that could be missing?
I tried the ones from PointCloud Float as well. Also, here is the strace:
gettid() = 669
write(2, "I0324 10:39:44.547245 669 came"..., 68I0324 10:39:44.547245 669 camera_itof.cpp:924] initComputeLibrary
) = 68
mmap(NULL, 4198400, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0xffff8d471000
mmap(NULL, 4198400, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0xffff8d070000
mmap(NULL, 4198400, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0xffff8cc6f000
mmap(NULL, 4198400, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0xffff8c86e000
mmap(NULL, 4198400, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0xffff8c46d000
mmap(NULL, 4198400, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0xffff8c06c000
mmap(NULL, 4198400, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0xffff8bc6b000
mmap(NULL, 4198400, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0xffff8b86a000
mmap(NULL, 4198400, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0xffff8b469000
munmap(0xffff8c86e000, 4198400) = 0
munmap(0xffff8c46d000, 4198400) = 0
munmap(0xffff8c06c000, 4198400) = 0
brk(0xaaab03bbb000) = 0xaaab03bbb000
brk(0xaaab03dbb000) = 0xaaab03dbb000
brk(0xaaab041bb000) = 0xaaab041bb000
mmap(NULL, 6295552, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0xffff8c66e000
brk(0xaaab041eb000) = 0xaaab041eb000
brk(0xaaab045eb000) = 0xaaab045eb000
brk(0xaaab049eb000) = 0xaaab049eb000
--- SIGSEGV {si_signo=SIGSEGV, si_code=SEGV_MAPERR, si_addr=NULL} ---
+++ killed by SIGSEGV +++
Segmentation fault
Hi,
Can you please open up a terminal and run the following?
dmesg
It should provide a log. It might help if you could send it for us to see it.
Also, if it's not too much trouble, could you run first-frame with the gdb debugger? Something like:
cd ~/Workspace/ToF/build/examples/first-frame
gdb --args ./first-frame config/config_adsd3500_adsd3100_new_modes.json
It is possible that gdb is not installed by default but it can be installed via apt-get.
Hello @dNechita -
I cleared the kernel log with 'dmesg -c' and then ran the app the create the seg fault. Nothing of note was logged to dmesg when it failed.
root@imx8-gateway:/first-frame# ./first-frame config/config_adsd3500_adsd3100_new_modes.json
I20210324 10:31:17.829505 676 main.cpp:78] SDK version: 4.2.0 | branch: HEAD | commit: b260ad46
I20210324 10:31:17.830018 676 system_impl.cpp:90] SDK built with websockets version:4.1.6-v4.1.6
I20210324 10:31:17.830049 676 sensor_enumerator_imx.cpp:116] Looking for sensors on the target
I20210324 10:31:17.830374 676 sensor_enumerator_imx.cpp:137] Looking at: /dev/media0 for an eligible TOF camera
I20210324 10:31:17.838143 676 sensor_enumerator_imx.cpp:153] Considering: /dev/media0 an eligible TOF camera
Segmentation fault
root@banner-fusion-imx8-gateway:/first-frame# dmesg
[ 370.520536] audit: type=1701 audit(1616581877.832:8): auid=4294967295 uid=0 gid=0 ses=4294967295 pid=676 comm="first-frame" exe="/first-frame/first-frame" sig=11 res=1
Here is the output from gdb also. It claims no debugging symbols:
root@imx8-gateway:/first-frame# gdb --args ./first-frame config/config_adsd3500_adsd3100_new_modes.json
GNU gdb (GDB) 10.1
Copyright (C) 2020 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law.
Type "show copying" and "show warranty" for details.
This GDB was configured as "aarch64-poky-linux".
Type "show configuration" for configuration details.
For bug reporting instructions, please see:
<https://www.gnu.org/software/gdb/bugs/>.
Find the GDB manual and other documentation resources online at:
<http://www.gnu.org/software/gdb/documentation/>.
For help, type "help".
Type "apropos word" to search for commands related to "word"...
Reading symbols from ./first-frame...
(No debugging symbols found in ./first-frame)
(gdb)
Hello,
Can you please check the library links of the first frame c++ example and the aditof.so file and send us the output?
You can run ldd ./first-frame and ldd aditof.so just to make sure the libraries are properly linked.
And please make sure the libtofi_config.so and libtofi_compute.so are from the arm64 package instead of the linux ones.
No response from user.