I am looking into some of the tools and mechanisms that can be used in order to trace events on the Linux-kernel. I was interested in particular in the LTTng tool. I understand that the tool allows you to add extra context to traced events, for instance perf counters like cache-misses, branch-misses etc. Does anyone know whether or not LTTng allows sampling of performance counters ? And if so, can you configure the perf counters, that is, choose which events are to be monitored ?
No comments:
Post a Comment