dtrace discuss - Jan 2006 - cpu_sysinfo_xcalls Questions

I am making some changes to the sun4v version of send_mondo_set() and noticed
the following macro usage in the sun4v version of send_one_mondo().

  CPU_STATS_ADDQ(CPU, sys, xcalls, 1);

Which results in a call to __dtrace_probe___cpu_sysinfo_xcalls(). I wanted to
look at the implementation of this function, but couldn''t find it.
Where is it? I assume it results in the dtrace sysinfo xcalls probe being fired.

The sun4v send_mondo_set() code does not currently call the macro. This is a
bug, correct?

Thanks,
Haik
This message posted from opensolaris.org

On 1/6/06, Haik Aftandilian <haik.aftandilian at sun.com>
wrote:
> I am making some changes to the sun4v version of send_mondo_set() and
> noticed the following macro usage in the sun4v version of send_one_mondo().
>
>   CPU_STATS_ADDQ(CPU, sys, xcalls, 1);
>
> Which results in a call to __dtrace_probe___cpu_sysinfo_xcalls(). I wanted
to
> look at the implementation of this function, but couldn''t find it.
Where is it? I
> assume it results in the dtrace sysinfo xcalls probe being fired.
Yes;  krtld records the call location, converts it into nops, and the
sdt code uses the recorded location to instrument the location when
the probe is enabled.
> The sun4v send_mondo_set() code does not currently call the macro. This
> is a bug, correct?
Yes.  Please file a bug.

Cheers,
- jonathan

Thanks, Jonathan.
> Yes. Please file a bug.
I''ve filed the following bug and assigned it to myself.

6370233 sun4v send_mondo_set() is missing call to CPU_STATS_ADDQ()

Haik
This message posted from opensolaris.org