Examine core dump of target device

[kurt]

My current setup:

• GDB compiled to run on Linux VM.
• gdbserver compiled to run on Linux Target device.
• Added a raise(SIGSEGV); to simulate an error.
• Target device produces core dump.  Core dump copied to Linux VM.

I followed this link (https://visualgdb.com/tutorials/linux/coredump/) to examine the core dump.  From the dialog, I selected the Linux VM, selected the core dump that was copied to the VM, and everything loads fine.  However, the Call Stack and Thread information is pretty limited, and is completely different from what I will get if I just deploy direction from VS.  It has no references to my code at all (I have ensured that gdb is reading the symbols correctly using symbol-file <path_to_debug file>

Am I doing it right?  Do I have to recompile my gdb to run on the target?  This is hard to do, considering my device has limited resources.

 

 

• This topic was modified 6 years, 10 months ago by kurt.

Attachments:

You must be logged in to view attached files.

[support]

Hi,

You do need to use the same target-specific gdb that you use for regular debugging. If you are using VisualGDB 5.3, you can actually open a core dump file by right-clicking on the project and selecting Debug->Debug a Crash Dump. This will automatically reuse the GDB and other settings from your project settings.

[kurt]

Yeah, I did point it to the gdb executable that I use for regular debugging.

Perhaps I still need to adjust the path to the libraries?  I think during regular debugging, it is using the .so files that reside in the target.

[support]

Hi,

Normally you don’t need to specify anything manually if you use the new “Debug a crash dump” command. If you still get the error when using this command, please let us know.

[Author]

Posts