Exe crashes on first cuda call when remote debugging

Hi, I have a Tx2 and I’m trying to debug a cuda example project using NSight EE and its remote debuging capability. I have the remote session connecting as root and I have the LD library path set. When I execute the code it breaks in main and I can single step until a cuda call happens. Once I step over a cuda call ex ( cudaGetDeviceProperties ) I get a _dl_find_dso_for_object exception and the program exits. I can launch cuda-gdb and debug the exact same executable on the remote system as the user I am remote debugging as. BTW I’ve factory reset the TX2 with Jetpack 4.4 and continue to see the issue. My host system is Ubunutu 18.04 configured via the nvidia sdkmanager.

My search skills must have been lacking. Found the answer here after I posted. The short … just click resume its a known issue and will be fixed in a later release.

cuda-api-calls-throw-exceptions-in-remote-debugger 77088