Forum Replies Created
-
Posts
-
Hi,
Most likely one of your files ends up including the core_cmX.h file that checks for the _FPU_PRESENT macro before including stm32f407xx.h file that defines it.
First of all, please identify the source file that is showing the error (you can remove the -j<number of CPUs> flag from Makefile arguments and check the last built file in the output log). Then try including the stm32f407xx.h file before any other files.
If this does not help, please try reducing the source file to just one include statement that still causes the error and then post the source file a build log showing the include directories here.
Hi,
Please check the OpenOCD window for errors. There was a bug in our recent OpenOCD release that could have caused it and it was fixed in VisualGDB 5.1r6.
Please upgrade to that version or alternatively re-download the latest OpenOCD debug package via Embedded Tools Manager. Then open the Debug Settings page, change any setting there and press “OK”. This should fix the problem.
Not sure what you meant. Adding a non-existent path to IntelliSense should have no effect and be the same as not adding any path there.
Hi,
Good to know it works. Let us know if you encounter further problems.
Note that we do recommend using Debian-based Linux distros like Ubuntu as they have convenient package managers that can install most of the commonly used software.
Hi,
Thanks for letting us know. Yes, the after-initialization commands are indeed executed after ALL of the commands from the debug method.
As a better workaround, you can define a variable like “com.yourname.init_mcu” in your custom device definition and then add $$com.yourname.init_mcu$$ to the startup commands in the EDP. Then you won’t have to hardcode the device-specific commands in the EDP file.
We currently don’t have any plans for changing the current behavior as it looks like a fairly rare scenario and the workaround is also very simple, however if you believe it’s a big annoyance (or anyone else encounters a similar issue), please let us know so that we could reconsider this.
Hi,
It looks like Nordic has discontinued the S110 softdevice in the latest SDK. Please use S130 instead or install the previous version of the SDK.
Hi,
Most likely the new image does not have the OpenSSH server installed. You can install it by logging on via a COM port and running “sudo apt-get install openssh-server”.
OK, this confirms that you can connect from your Windows machine. Could you run “telnet 192.168.0.100 2000” on the Linux machine and see if gdbserver reports an incoming connection like it does when connecting from Windows?
Please check the actual output of OpenOCD. It should contain information about the specific error. If you are not sure, please post the output here so that we could help you.
OK, so you can telnet to 192.168.0.100:2000 from the target itself, but not from Windows, right? If yes, please try disabling the firewall on the target.
OK, this means that the connection between the gdb machine and the target machine cannot be established. Please try the following:
- Manually launch “gdbserver :2000 <deployed binary>” on the target
- Try connecting to 192.168.0.100:2000 via telnet from the gdb machine. If this does not work, try connecting from the Windows machine. If this does not work either, try connecting from the target machine itself.
If the only machine you can connect from is the target machine itself, please check your network & firewall settings. If you can connect from Windows, but not from Linux, check that both machines are on the same network and that no firewall between them is blocking connections.
If VisualGDB asks for a process to attach, it means that your debug settings have “Attach to a process” selected. Please open VisualGDB Project Properties and change gdb launch mode to “Debug a new instance”. This should resolve the problem.
