c - Cross compilation doesn't work for my Beaglebone Black because of dynamic linker - Stack Overflow

admin管理员组

文章数量:1329891

I have begun to work on a Beaglebone Black for a school project, and we have been using Buildroot in order to build our custom images. Right now we are working with a basic 5.3 image.

We use Buildroot-2022.02.9.

For the image, we go from a beaglebone_defconfig using a make beaglebone_defconfig we use a custom "Kernel version" in buildroot -5.3 version. And in the Toolchain menu, in the Kernel Headers we choose it to be the same as the kernel being built, and in Custom kernel headers series we choose 5.3.x.

The objective is to cross-compile a C program which will utilize the LEDs of the Beaglebone, but I am running tests with only a helloworld program and it doesn't work.

Also, the helloworld file is transfered using tftp and I see it in my directory using ls, and I've added the executable flag, just to tell you that it exists on my beaglebone.

Here's the issue:

When I run the helloworld file on my beaglebone after I've compiled it on my VM - , I get this error message :

-sh: ./helloworld: not found

At first, I thought I was using the wrong compiler, so I've used many. I can list :

arm-linux-gnueabihf
arm-linux-gnueabi

And I've tried many versions of those two toolchains. The prompt I give is simply:

arm-linux-gnueabihf-gcc helloworld.c –o helloworld

So I've done my research, and I've found that the message -sh: ./helloworld: not found was misleading - see the verified answer on this post : Why can't I run custom application on my Beaglebone board?

So it is a dynamic linker problem, which in my case for the file when I look for it using readelf -a is: /lib/ld-linux.so.3.

So, considering that it wasn't being found, I proceeded to search it on the /lib/ directory on my Beagleboard.

Turns out I don't even have it, not any single version of it.

My guess is that I should include it in my configuration when customizing it with buildroot, but I don't know how to set it and which one to choose being new on the topic.

Thus, what I need to do is use the correct cross-toolchain (which uses the same C library version) for the rootfs I am using or build a new rootfs with the same toolchain.

In buildroot, the C-library used when I go to the toolchain menu is uClibc-ng ,so I've tried to set it to glibc but all I get is an error during the make process.

I have begun to work on a Beaglebone Black for a school project, and we have been using Buildroot in order to build our custom images. Right now we are working with a basic 5.3 image.

We use Buildroot-2022.02.9.

For the image, we go from a beaglebone_defconfig using a make beaglebone_defconfig we use a custom "Kernel version" in buildroot -5.3 version. And in the Toolchain menu, in the Kernel Headers we choose it to be the same as the kernel being built, and in Custom kernel headers series we choose 5.3.x.

The objective is to cross-compile a C program which will utilize the LEDs of the Beaglebone, but I am running tests with only a helloworld program and it doesn't work.

Also, the helloworld file is transfered using tftp and I see it in my directory using ls, and I've added the executable flag, just to tell you that it exists on my beaglebone.

Here's the issue:

When I run the helloworld file on my beaglebone after I've compiled it on my VM - , I get this error message :

-sh: ./helloworld: not found

At first, I thought I was using the wrong compiler, so I've used many. I can list :

arm-linux-gnueabihf
arm-linux-gnueabi

And I've tried many versions of those two toolchains. The prompt I give is simply:

arm-linux-gnueabihf-gcc helloworld.c –o helloworld

So I've done my research, and I've found that the message -sh: ./helloworld: not found was misleading - see the verified answer on this post : Why can't I run custom application on my Beaglebone board?

So it is a dynamic linker problem, which in my case for the file when I look for it using readelf -a is: /lib/ld-linux.so.3.

So, considering that it wasn't being found, I proceeded to search it on the /lib/ directory on my Beagleboard.

Turns out I don't even have it, not any single version of it.

My guess is that I should include it in my configuration when customizing it with buildroot, but I don't know how to set it and which one to choose being new on the topic.

Thus, what I need to do is use the correct cross-toolchain (which uses the same C library version) for the rootfs I am using or build a new rootfs with the same toolchain.

In buildroot, the C-library used when I go to the toolchain menu is uClibc-ng ,so I've tried to set it to glibc but all I get is an error during the make process.

• c
• embedded-linux
• cross-compiling
• beagleboneblack
• buildroot

Share Follow edited Feb 18 at 8:18 sawdust 17.1k33 gold badges4242 silver badges5555 bronze badges asked Nov 28, 2024 at 14:44 minimus_maximusminimus_maximus 4577 bronze badges

Add a comment  | 

1 Answer 1

Sorted by: Reset to default 2

I found where the adapted toolchain compiler is :

~/Documents/buildroot/output/host/bin/arm-buildroot-linux-uclibcgnueabihf-gcc 

The command line is then very simple:

~/Documents/buildroot/output/host/bin/arm-buildroot-linux-uclibcgnueabihf-gcc -o hello hello.c

You can obviously add it to PATH environment variable.

本文标签： cCross compilation doesn39t work for my Beaglebone Black because of dynamic linkerStack Overflow