Hi Mathieu,
Thank you for the reply and my apologies for the confusion. Let me try and
clarify.
I understood that the use of the class and instance concepts in the tpl file
was to be able to declare a class with as you say a number of fields and then
create instances of that class that provided a way of defining specific
variants of that class.
For example
TRACEPOINT_EVENT_CLASS(
/* Tracepoint provider name */
core_event_trace,
/* Tracepoint class name */
core_event_tracepoint_class,
/* Input arguments */
TP_ARGS(
int, event_type
int, second_field
int, third_field
int, fourth_field
),
/* Output event fields */
TP_FIELDS(
ctf_integer(int, event_type, event_type)
ctf_integer(int, second_field, second_field)
ctf_integer(int, third_field, third_field)
)
)
Then we declare an instance
TRACEPOINT_EVENT_INSTANCE(
core_event_trace,
core_event_tracepoint_class,
first_event_tracepoint,
TP_ARGS(
int, event_type
Int, first_field
)
)
And a second one
TRACEPOINT_EVENT_INSTANCE(
core_event_trace,
core_event_tracepoint_class,
second_event_tracepoint,
TP_ARGS(
int, event_type
Int, second_field
)
)
So the two instances do not define all of the fields in the class definition,
the first only defines first_field and the second only defines second_field,
will this work or do the field in the instance always have to match the fields
in the class?
From: Mathieu Desnoyers [mailto:[email protected]]
Sent: 30 November 2018 17:13
To: John O'Sullivan
Cc: lttng-dev
Subject: [EXTERNAL] Re: [lttng-dev] Correct way to use the tpl file
----- On Nov 30, 2018, at 11:09 AM, John OSullivan
<[email protected]> wrote:
Hi,
I am looking for some advice on the best way to manage the tracing of some user
space events.
I have a series of events that are fed to queues, they look something like the
following sample
EventEnd* event_end_new()
{
EventEnd* event_end = bb_xnew(EventEnd,__func__);
if (event_end != NULL)
{
event_end->type = EQ_EVENT_END;
. .. other event specific fields here
..
}
#ifdef LTTNG_TRACE
tracepoint(core_event_trace, end_event_tracepoint, EQ_EVENT_END);
#endif
return event_end;
}
I want to try and trace these so created the following template shown below. My
question is, does the class have to have all of the possible members that the
different instances will have, as each unique event is likely to have different
members will the class have to have all of these members defines, and if this
is the case is it okay for an instance to just implement a subset of the
members the class defines?
I am a bit confused by your use of the word "members" here. A tracepoint class
has a provider name,
a class name, input argument, and output fields.
The event instance refers to the class by provider and class name, defines its
own name, and needs to
have a compatible input argument prototype (same number of arguments, same
types). It will serialize
the output fields in the exact layout described by the event class.
I know it does not directly answer your question, but I'm really unsure about
what you mean
by "members".
Thanks,
Mathieu
TRACEPOINT_EVENT_CLASS(
/* Tracepoint provider name */
core_event_trace,
/* Tracepoint class name */
core_event_tracepoint_class,
/* Input arguments */
TP_ARGS(
int, event_type
),
/* Output event fields */
TP_FIELDS(
ctf_integer(int, event_type, event_type)
)
)
TRACEPOINT_EVENT_INSTANCE(
core_event_trace,
core_event_tracepoint_class,
end_event_tracepoint,
TP_ARGS(
int, event_type
)
)
TRACEPOINT_EVENT_INSTANCE(
core_event_trace,
core_event_tracepoint_class,
start_event_tracepoint,
TP_ARGS(
int, event_type
)
)
TRACEPOINT_EVENT_INSTANCE(
core_event_trace,
core_event_tracepoint_class,
foo_event_tracepoint,
TP_ARGS(
int, event_type
)
)
________________________________
This email and any files transmitted with it are confidential and are intended
for the sole use of the individual to whom they are addressed. Black Box
Corporation reserves the right to scan all e-mail traffic for restricted
content and to monitor all e-mail in general. If you are not the intended
recipient or you have received this email in error, any use, dissemination or
forwarding of this email is strictly prohibited. If you have received this
email in error, please notify the sender by replying to this email.
_______________________________________________
lttng-dev mailing list
[email protected]
https://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev
--
Mathieu Desnoyers
EfficiOS Inc.
http://www.efficios.com
_______________________________________________
lttng-dev mailing list
[email protected]
https://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev
