> When the tracee resumes from TASK_TRACED, context->options can be changed. > > Introduce ptrace_event->ev_options. If a callback sets ->ev_options != 0, > do_ptrace_resume() checks that context->options matches or discards the > event.
IMHO this hair is a pretty clear indication that the "queued reports" model is just not the right model. You have all this new complexity because of the model that you queue the details of the report at event time, but in reality much of that determination belongs at report time rather than event time. I'm merging everything just to let you make progress how you like. But I think you need to answer my earlier question about the motivation for this whole plan. I think the reality of what is actually required is substantially simpler than the complexities entailed in this abstraction. Thanks, Roland
