Hi,

Mathieu Desnoyers wrote:
> Oestman, Fredrik ([email protected]) wrote:
> > Is there any way to know from within the instrumented application
> > when tracepoint registration is finished?
> 
> No.

OK. Does any internal knowledge about this exist somewhere, which could be 
externalized?


> > On ARM, this messes up LTTng UST and faulty events are recorded.
> What do you mean by faulty events ?

1. Wrong trace data type:
1.a) A tracepoint for one type produces events of another type, for which there 
is a tracepoint.
1.b) A tracepoint produces a non-existing type, which makes babeltrace choke on 
the data, index out of range
2. Wrong field: Data belonging to field #2 sent in field #1, in combination 
with 1.a)


> Well, if you trace _all_ function calls, then you get all function
> calls

Yes, this was no surprise, just one more reason to try to find out when 
tracepoint registration is finished.


Cheers,

Fredrik Östman


_______________________________________________
lttng-dev mailing list
[email protected]
http://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev

Reply via email to