cancel
Showing results for 
Search instead for 
Did you mean: 

Headless build fail on linux STM32CubeIDE v1.13.1

Marshall
Associate

OS: Ubuntu 20.04.6 LTS

STM32CubeIDE version 1.13.0 or 1.13.1

My project builds fine in the IDE and with headless builds on version 1.12.1. With version 1.13.X builds within the IDE work fine, but attempting a headless build from command line fails nearly instantly with "An error has occurred. See the log file" (attached .log file). 

One line I found particularly disturbing in the .log was "!MESSAGE Only a single IDE contribution is supported. Your STM32Cube tools installation is corrupted." As a result I have attempted reinstalling the IDE from the installer 3 times (each for 1.13.0 and 1.13.1), when I first upgraded in the tool itself.

I've tried downgrading the MCU toolchain to the same toolchain used in v1.12.1.

I've tried changing the runtime library from reduced C/C++ to standard C/C++.

I've tried the newlib workaround (add linker file GROUP) described by @Peter BENSCH here.

I've tried another newlib workaround (add weak function defines to main.c) also described by @Peter BENSCH  here.

I've tried turning off CDT from available software sites.

I've reviewed this post that appears related to my problem. Was the long-term fix recently released that inadvertently reintroduced the problem?

Any help is greatly appreciated!

Marshall

1 ACCEPTED SOLUTION

Accepted Solutions
Rim LANDOLSI
ST Employee

Hello @Marshall ,

Thank you for your contributions and for your detailed explanations,

I confirm the issue on my side, it has been reported to STM32CubeIDE development team .

It will be fixed in the next releases.

Hope that this answered your question.

Regards,

Rim

View solution in original post

3 REPLIES 3
peterbtg
Associate

Hello,

I have no help to offer, but I think I have the same problem. I can run headless build just fine locally (windows) but it fails every time in my azure pipeline (Ubuntu 22.04). I have only tested 1.13.1, not any earlier versions of the IDE.

I have attached a log file with the output from a run. All info I can find online indicates that it is a race-condition bug in the STM32 eclipse plugins whereas the "workbench" (not the workspace!) has not been created yet but the plugin tries to do something with it.

Peter

Rim LANDOLSI
ST Employee

Hello @Marshall ,

Thank you for your contributions and for your detailed explanations,

I confirm the issue on my side, it has been reported to STM32CubeIDE development team .

It will be fixed in the next releases.

Hope that this answered your question.

Regards,

Rim

peterbtg
Associate

Great! Thank you very much Rim!