Project

General

Profile

Actions

Bug #1273

closed

Cannot use a #defined TRACEPOINT_PROVIDER name with TRACEPOINT_EVENT_{CLASS,INSTANCE}() macros

Added by Christophe Bedard almost 4 years ago. Updated over 3 years ago.

Status:
Resolved
Priority:
Normal
Assignee:
-
Target version:
-
Start date:
06/14/2020
Due date:
% Done:

100%

Estimated time:

Description

It doesn't seem to be possible to use a provider name that was #defined if using the TRACEPOINT_EVENT_{CLASS,INSTANCE}() macros, e.g.:

// + all the boilerplate code

#define TRACEPOINT_PROVIDER my_provider

TRACEPOINT_EVENT_CLASS(
  TRACEPOINT_PROVIDER,
  my_class,
  TP_ARGS(
    const void *, some_arg
  ),
  TP_FIELDS(
    ctf_integer_hex(const void *, whatever, some_arg)
  )
)

TRACEPOINT_EVENT_INSTANCE(
  TRACEPOINT_PROVIDER,
  my_class,
  my_tracepoint_does_not_work,
  TP_ARGS(
    const void *, some_arg
  )
)

In my case, I had:

  • a shared lib providing the tracepoints
  • an application linking against the above lib

When compiling, it fails on the app and not the lib.

It expected this to work because it works when using the simpler TRACEPOINT_EVENT() macro. However, if I use my_provider directly instead of TRACEPOINT_PROVIDER, it works.

I've reproduced this on master.

Here's a MWE: https://github.com/christophebedard/lttng-tracepoint-provider-name-define-mwe

Actions

Also available in: Atom PDF