atlflight / atlflightdocs Goto Github PK

I am getting the following error when I try to run the HelloWorld programme. I tried building with both methods.

Asking DSP to say hello
/media/intrinsyc/BigDrive/Project/APQ8074Flight/LNX.LER.1.0-68037/APQ8074Flight/oe-core/build/tmp-eglibc/work/cortexa8hf-vfp-neon-linux-gnueabi/adsprpc/1.0-r0/adsprpc-1.0/src/fastrpc_apps_user.c:136:failed to create tls key/media/intrinsyc/BigDrive/Project/APQ8074Flight/LNX.LER.1.0-68037/APQ8074Flight/oe-core/build/tmp-eglibc/work/cortexa8hf-vfp-neon-linux-gnueabi/adsprpc/1.0-r0/adsprpc-1.0/src/listener_android.c:112:listener using ion heap: -1
Segmentation fault

When: executing "make posix_eagle_default"
What: Make error: arm-linux-gnueabihf-g++ not found even through path is set correctly
Why: Missing multi-architecture (64bit Host - 32bit binaries) compatibility.
Fix: Install support through a combination of lib32stdc++6 lib32z1 lib32z1-dev gcc-multilib, then rerun cross_toolchain/install.sh

This section of ALTFlight Documentation says

To build code for this configuration, please SKIP the rest of this page and follow the instructions in the official PX4 documentation at http://dev.px4.io/starting-building.html.

However the PX4 documentation at the link above points back to ALTFlightDocs:

If you use the Qualcomm ESC board, please follow their instructions here.

I already know that one needs to build the PX4 with eagle_legacy_default. Is there anything else that needs to be changed while following PX4 docs to use Qualcomm ESC board?

The qcom_flight_controller_hexagon_sdk_add_on.zip contains different static image files:

images/8074-eagle/normal/adsp_proc/obj/qdsp6v5_ReleaseG
├── LA
│   └── system
│       └── etc
│           └── firmware
│               ├── ...
├── non_reloc
│   └── LA
│       └── system
│           └── etc
│               └── firmware
│                   ├── ...
└── unsigned
    ├── ...

It is not clear what each of them are for and which ones need to be flashed. I would propose only to include what is actually needed and working. I (as a stupid user) for instance ended up flashing the wrong ones because I was desperate and ended up loosing almost a week of development because it bricked my board.

@mcharleb @jywilson

I was wondering whether the GPIO controller directly supports open drain outputs, or whether one needs to emulate this with first:
DSPAL_GPIO_DIRECTION_OUTPUT, DSPAL_GPIO_LOW_VALUE
and then
DSPAL_GPIO_DIRECTION_INPUT, DSPAL_GPIO_PULL_DOWN/DSPAL_GPIO_NO_PULL
as described here: https://www.kernel.org/doc/Documentation/gpio/gpio-legacy.txt

because I couldn't find any setting in dspal/include/dev_fs_lib_gpio.h. Any suggestion?

Hi all,
I'm new to snapdragon , I use ubuntu 64 bit to compile dspal test.
If I use make only function. The program built successful. However if I push to the host ,
I get these error
/home/cberg/Desktop/r00015/oe-core/build/tmp-eglibc/work/cortexa8hf-vfp-neon-linux-gnueabi/adsprpc/1.0-r0/adsprpc-1.0/src/fastrpc_apps_user.c:126:/dev/adsprpc-smd device initialization error: Operation not permitted
/home/cberg/Desktop/r00015/oe-core/build/tmp-eglibc/work/cortexa8hf-vfp-neon-linux-gnueabi/adsprpc/1.0-r0/adsprpc-1.0/src/fastrpc_apps_user.c:179::error: -1: dev != -1
/home/cberg/Desktop/r00015/oe-core/build/tmp-eglibc/work/cortexa8hf-vfp-neon-linux-gnueabi/adsprpc/1.0-r0/adsprpc-1.0/src/fastrpc_apps_user.c:224::error: -1: 0 == (nErr = remotectl_open(name, (int_)ph, dlerrstr, sizeof(dlerrstr), &dlerr))
/home/cberg/Desktop/r00015/oe-core/build/tmp-eglibc/work/cortexa8hf-vfp-neon-linux-gnueabi/adsprpc/1.0-r0/adsprpc-1.0/src/fastrpc_apps_user.c:126:/dev/adsprpc-smd device initialization error: Operation not permitted
/home/cberg/Desktop/r00015/oe-core/build/tmp-eglibc/work/cortexa8hf-vfp-neon-linux-gnueabi/adsprpc/1.0-r0/adsprpc-1.0/src/fastrpc_apps_user.c:179::error: -1: dev != -1
_Failed to get DSP image version information.*

if I use make load, The program is not compiled . and here is the error
error: unknown host service
error: unknown host service
dspal_tester/CMakeFiles/dspal_tester_app-load.dir/build.make:57: recipe for target 'dspal_tester/CMakeFiles/dspal_tester_app-load' failed
......
I also upload new adsp image but I am still get the same error
Should I move to ubuntu x86 to compile the program

@mcharleb @jywilson
By default the clock is high when inactive, CPOL = 1, as also noted here:
https://developer.qualcomm.com/forum/qdn-forums/software/hexagon-dsp-sdk/hardware-support/29739

Is it possible to extend the functionality of the dspal abstraction layer such that one can set that?

When: following the instruction [(https://github.com/ATLFlight/dspal)]

What:

nanfly@nanfly-T440p:~/dspal/test/dspal_tester$ ls
adsp_proc  cmake_hexagon   common            include
apps_proc  CMakeLists.txt  dspal_tester.idl  Makefile
nanfly@nanfly-T440p:~/dspal/test/dspal_tester$ make
make: *** No rule to make target `cmake_hexagon/toolchain/Toolchain-qurt.cmake, needed by `dspal_tester'.  Stop.

Why: Don't know. Can anyone help?

I am using the latest pull and can make the HelloWorld program with no problem.

Thanks in advance.

The USB input in my snapdragon flight is not working anymore. Is it possible to use the USB/serial interface to access the board?

I use libsensor_imu.so in Flight_3.1.3.1_qcom_flight_controller_hexagon_sdk_add_on.zip ,and Flight_3.1.3_release_notes.txt said Updated the imu api version (1.2.3) and refactored the IMU server ,but when I test sensor_imu_tester, it's wrong, like this

root@linaro-developer:~# sensor_imu_tester 2
The apps <--> dsp offset = 1530621443843763266 Apps ClockType[CLOCK_REALTIME]
Error: Imu Lirary version mis-match. expected[sensor-imu-1.2.3] got[sensor-imu-1.2.2]

for the same reason , I can't pass snap_imu. hope for help, Thank u

Hi
Can I prototype a MV app on my Pixel2 XL phone which also has the latest 835 SOC ? I noticed that the libs are Debian packages. Would love to use VSLAM on the phone and use the Hexagon DSP as well.

Thanks
Sanjit

Trying to use the bottom facing cam on snapdragon flight for optic flow outside of mvSDK. I'm not sure what the parameters in the fisheye model calibration represent or how to use them in the undistortion. I skimmed thru S.Shah's paper but couldn't figure it out. Specifically, if the first four params are k1-4, then what are the four trailing zeros?

Secondly, would it be correct to call the following to retrieve the flat plane image?

cv::fisheye::undistortPoints( distorted=capturedImagePts, undistored=flatImagePts, K=KFromCamMgrParams D=DFromCamMgrParams, R=noArray(), P=noArray());

Last, to verify:
Is the image from qcamvid optic flow cam original image or undistorted image?

Hi all
I do exactly like the tutorial. but I can not run "make load-bundle".
The terminal returns

ake[4]: Entering directory '/home/khoi/cross_toolchain/HelloWorld/bundle/build_bundle'
error: unknown host service
error: unknown host service
CMakeFiles/helloworld_app-load.dir/build.make:57: recipe for target 'CMakeFiles/helloworld_app-load' failed
make[4]: *** [CMakeFiles/helloworld_app-load] Error 1

How can I fix the host service?
Thank all

When make in the HelloWorld directory, error comes up showing

arm-linux-gnueabihf-gcc: error: /home/nanfly/Qualcomm/Hexagon_SDK/2.0/lib/common/rpcmem/UbuntuARM_Debug/rpcmem.a: No such file or directory

Pull and run the latest croo-toolchain, every seems normal but the rpcmem.a is still not there.

Try directly generate the rpemem.a in the directory as indicated in the install.sh and error comes up as

nanfly@nanfly-XPS:~/Qualcomm/Hexagon_SDK/2.0/lib/common/rpcmem$ make V=UbuntuARM_Debug ../../../tools/make.d.ext/UbuntuARM/../../../gcc-linaro-arm-linux-gnueabihf-4.8-2013.08_linux/bin/arm-linux-gnueabihf-gcc -c -fPIC -Wall -Wno-missing-braces -mword-relocations -mthumb-interwork -mfloat-abi=hard -mfpu=neon -mtune=cortex-a8 -march=armv7-a -Werror -g -isystem ../../../tools/make.d.ext/UbuntuARM/../../../gcc-linaro-arm-linux-gnueabihf-4.8-2013.08_linux/arm-linux-gnueabihf/libc/usr/include -DARM_ARCH_7A -DUSE_SYSLOG -std=gnu99 -D__FILENAME__=\"rpcmem_android.c\" -D_DEBUG -I. -I../../../inc/a1std -I../../../inc/qlist -I../../../inc/stddef -IUbuntuARM_Debug -oUbuntuARM_Debug/rpcmem_android.o rpcmem_android.c make: ../../../tools/make.d.ext/UbuntuARM/../../../gcc-linaro-arm-linux-gnueabihf-4.8-2013.08_linux/bin/arm-linux-gnueabihf-gcc: Command not found make: *** [UbuntuARM_Debug/rpcmem_android.o] Error 127

I have been having this problem since I got the flight dragon board a few weeks ago and tried every time the source code on the github is updated. Can anyone help me finally solve this problem?

Thanks in advance.

https://github.com/ATLFlight/ATLFlightDocs/blob/master/TroubleshootFAQ.md#apt-get-does-not-work

on dpkg -i fastcv-internal_1.0-r0_armhf.deb:

(Reading database ... 51118 files and directories currently installed.)
Preparing to unpack fastcv-internal_1.0-r0_armhf.deb ...
Unpacking fastcv-internal (1.0-r0) over (1.0-r0) ...
dpkg: dependency problems prevent configuration of fastcv-internal:
fastcv-internal depends on fastcv-noship (>= 1.0); however:
Package fastcv-noship is not installed.

dpkg: error processing package fastcv-internal (--install):
dependency problems - leaving unconfigured
Errors were encountered while processing:
fastcv-internal

Hi,

I am using Windows environment trying to build this for SnapDragon Flight use. My last two commands are as below and was trying to follow steps to build the code.
git checkout tags/EAGLE_DRONES_1.2_FC_0
git submodule update --init --recursive

However, could not find "make eagle_legacy_default" useful, can someone advise how to continue?

Thanks.

Hill Chen

Simple thing... this tree for mini DM is different for me:

Github
host%> ${HEXAGON_SDK_ROOT}/tools/mini-dm/Linux_Debug/mini-dm

My Linux install using Hexagon SDK 3.0
host%> ${HEXAGON_SDK_ROOT}/tools/debug/mini-dm/Linux_Debug/mini-dm

@mcharleb, @jywilson
using the inline functions defined in dspal/include/endian.h, e.g. __htonl(x), leads to errors

/tmp/i2c_test_imp-2ab05a.s:78:2: error: invalid operand for instruction
eor r2, r16, r16, ror #16
^
/tmp/i2c_test_imp-2ab05a.s:79:1: error: invalid operand for instruction
bic r2, r2, #0x00ff0000
^
/tmp/i2c_test_imp-2ab05a.s:80:1: error: invalid operand for instruction
mov r16, r16, ror #8
^
/tmp/i2c_test_imp-2ab05a.s:81:1: error: invalid operand for instruction
eor r16, r16, r2, lsr #8

unfortunately I don't know anything about assembler instructions, so I don't know really how to fix it, but a workaround is to use something like this instead:

static __inline __uint32_t
__bswap32_var(__uint32_t val)
{
    return ((((val) >> 24) & 0x000000FF) | (((val) >>  8) & 0x0000FF00) | \
            (((val) <<  8) & 0x00FF0000) | (((val) << 24) & 0xFF000000));
}

can you confirm this issue?

Hi, Im using Spektrum DX8 RC. I followed all the instructions at https://github.com/ATLFlight/ATLFlightDocs/blob/master/PX4.md. I have my receiver connected to J15 of the snapdragon flight board(I tried J12 as well). The receiver is binded with the transmitter. My QGC doesnt show any response from RC (I tried many versions of QGC and mission planner as well to just see if it responds somewhere). It seems like the flight board doesnt recognize the RC. How can I solve it ? Is there any way to check whether it detects RC or not? Thanks

I did install snav to use snav_vehicles for sensor parameters. snav used to boot up automatically at startup causing the flight board to hang and shutdown itself after a while. I couldnt factory reset the board as well. I followed the guide and disabled autostart. I dont know if it has anything to do with RC parameters but I am just confused on whats the difference between calibrating using snav and qground control.

When I install Hexagon SDK 3.0,I met some questiones.

Both documents describe different ways of installing the qrlSDK.

cd media/camera-daemon 
make -f sdk_use.mk

This gives an error--

In file included from src/camerad.cpp:41:0:
src/camerad.h:33:20: fatal error: camera.h: No such file or directory
 #include "camera.h"
                    ^
compilation terminated.
make: *** [src/camerad.o] Error 1

How can I fix this?

Hello,

In the "Known issues and limitations" section it was mentioned that accelerometer calibration using QGC is not functional. Would somebody please let me know what tool should I use for sensor calibration?

Thank You
Ahmad

On Ubuntu 14.04, no clang++ installation available.
Is it really required?