I’ve been having trouble running cuda-gdb on a second process that is forked from another process – cuda-gdb will not stop at a breakpoint in a global function in the second process. I attach to the second process (before the kernel is launched) with this command
cuda-gdb program <process_ID>
The cuda-gdb guide has this statement in the section Program Execution/Interrupting the Application, which discusses the use of to interrupt a kernel:
“This feature is limited to applications running within the debugger. It is not possible to
break into and debug applications that have been previously launched.”
I’m not sure if that is a general statement, or if it only applies to that little section of the manual. Does that mean that it is impossible to use cuda-gdb in my multiple process situation?