Is it possible to trace all exceptions in all user mode processes?

Would it be possible to trace all exceptions including stacktrace with such a filter or are there principal issues with such an approach?
I could imagine that recording all exceptions before a process crashes could be interesting to find the actual root cause much faster without a debugger.
Especially when the process is swallowing some fatal ones before.

No file, test, or entry point is named in the issue. Start by assessing BCC's ability to trace exceptions and capture stack traces across all user-mode processes; done would require a clear feasibility decision and defined implementation scope.

Written by the indexing model from the issue text.