Can't deploy the executable on Raspberry Pi and Matlab Coder doesn't generate the .elf file

21 visualizzazioni (ultimi 30 giorni)

Pietro Ventrella il 28 Lug 2021

0
Link

Link diretto a questa domanda

https://it.mathworks.com/matlabcentral/answers/887539-can-t-deploy-the-executable-on-raspberry-pi-and-matlab-coder-doesn-t-generate-the-elf-file

Commentato: Pietro Ventrella il 19 Ott 2021

I've succesfully modified a deep learning matlab script to use it on a Raspberry Pi: i've done the code generation using Matlab Coder without any error and i've also tried the script directly on Matlab with the raspi connected to the pc. But if i try to deploy it as a standlone application on the raspi, i've got an error.

I've also tried to generate the code as an executable and even if the success message appears in the report saying that the output file is an .elf file, it doesn't appear in the folder or anywhere

I don't know if there is another possible way to deploy it on the raspi or to find out why the .elf file doesn't appear in the folder.

I post the options i used to generate the code and to deploy it, if they can be more useful.

Code Generation:

Deploy:

11 Commenti
Mostra 9 commenti meno recentiNascondi 9 commenti meno recenti

David Fink il 19 Ott 2021

Note on "I've also tried to generate the code as an executable and even if the success message appears in the report saying that the output file is an .elf file, it doesn't appear in the folder or anywhere":

The "Generate code only" checkbox on the code generation options page prevents the makefile from being invoked, so only the C++ sources are generated, and the C++ compiler is not invoked.

Based on later comments, it looks like you were able to invoke the makefile via the deployment workflow.

Pietro Ventrella il 19 Ott 2021

@David Fink

"The "Generate code only" checkbox on the code generation options page prevents the makefile from being invoked, so only the C++ sources are generated, and the C++ compiler is not invoked."

I've realized that shortly after. But even if I successfully generate code for an executable and run the "make -f filename.m" command I get errors.

Anyway i keep getting errors on the deployment phase even using different toolchains. In particular i have used "Microsoft Visual C++ 2019 v16.0 | nmake (64-bit Windows)" (which is also the specific compiler required on windows by functions that call opencv components, like the cascade object detector) but i got this error:

****************************************************************************************************************************************

cl /TP -c -nologo -GS -W4 -DWIN32 -D_MT -MT -EHs -D_CRT_SECURE_NO_WARNINGS /wd4251 /O2 /Oy- -std=c++11 -mfpu=neon -fopenmp /source-charset:utf-8 /openmp /wd4101 -D_MW_MATLABTGT_ -DUSE_19_05_LIBRARY -D__MW_TARGET_USE_HARDWARE_RESOURCES_H__ -DMW_MATLABTARGET -D__linux__ -DARM_PROJECT -D_USE_TARGET_UDP_ -D_RUNONTARGETHARDWARE_BUILD_ -DSTACK_SIZE=200000 -DMODEL=raspi_singleFaceRecognition @raspi_singleFaceRecognition_rtw_comp.rsp -Fo"MWAvgPoolingLayer.obj" D:\Politecnico\Tesi_Prof.Giorgio\Matlab\RaspberryPi\Riconoscimento_Singolo\codegen\exe\raspi_singleFaceRecognition\MWAvgPoolingLayer.cpp

cl : warning della riga di comando D9002 : l'opzione sconosciuta '-std=c++11' verrà ignorata

cl : warning della riga di comando D9002 : l'opzione sconosciuta '-mfpu=neon' verrà ignorata

cl : warning della riga di comando D9002 : l'opzione sconosciuta '-fopenmp' verrà ignorata

MWAvgPoolingLayer.cpp

D:\VisualStudio\VC\Tools\MSVC\14.29.30037\include\arm_neon.h(21): fatal error C1189: #error: This header is specific to ARM targets

NMAKE : fatal error U1077: 'D:\VisualStudio\VC\Tools\MSVC\14.29.30037\bin\HostX64\x64\cl.EXE' : codice restituito '0x2'

Stop.

The make command returned an error of 2

Error(s) encountered while building "raspi_singleFaceRecognition"

****************************************************************************************************************************************

Accedi per commentare.

Accedi per rispondere a questa domanda.

Risposte (1)

Walter Roberson il 8 Ott 2021

0
Link

Link diretto a questa risposta

https://it.mathworks.com/matlabcentral/answers/887539-can-t-deploy-the-executable-on-raspberry-pi-and-matlab-coder-doesn-t-generate-the-elf-file#answer_804551

The error:

https://stackoverflow.com/questions/61646912/using-soil-lib-with-gcc-error-adding-symbols-file-format-not-recognised

So opencv_core420.lib was not compiled with the same compiler.

You are compling with MinGW 7.1 or later, but you are using R2020b, which needs MinGW 6.3 https://www.mathworks.com/content/dam/mathworks/mathworks-dot-com/support/sysreq/files/system-requirements-release-2020b-supported-compilers.pdf

8 Commenti
Mostra 6 commenti meno recentiNascondi 6 commenti meno recenti

Pietro Ventrella il 12 Ott 2021

@Walter Roberson

so why did i get the same error while generating the code with a pc that has only visual studio 2019 compiler (and hasn't any version of minGW compiler installed) ?

With both PCs i have correctly generated this code https://it.mathworks.com/help/supportpkg/raspberrypiio/ref/identify-objects-within-video-using-resNet-50-on-raspberry-pi-hardware.html , so i don't get what is the problem with my code

Walter Roberson il 12 Ott 2021

Sorry, I do not have a Windows development system to test with.

(Microsoft considers me a commercial developer when I am volunteering to help other people, and the commercial licenses I would need are several thousand dollars per year.)

Accedi per commentare.

Accedi per rispondere a questa domanda.