[ltt-dev] [UST] Trace locking
mathieu.desnoyers at efficios.com
Thu Oct 28 17:33:18 EDT 2010
* Nils Carlson (nils.carlson at ericsson.com) wrote:
> Hi there,
> I've been looking through the ust trace locking mechanism,
> ltt_lock_traces, and am wondering: is there any real point to it?
> Assuming only one single tracectl thread in the process the traces will
> only ever be accessed or changed by that single thread. There is
> therefore never a possibility of a lock contention, so why bother?
> Mathieu, do you have any insight?
If we ever move to a scheme where we have one thread that controls each trace
session, we will need the mutex.
Operating System Efficiency R&D Consultant
More information about the lttng-dev