Re: [Qemu-devel] [RFC] tracing: consistent usage of "disable" in "trace-events"

[Daniel P. Berrange]

On Mon, Nov 08, 2010 at 03:42:15PM +0100, Lluís wrote:
> On the current implementation, the "disable" keyword in "trace-events"
> has different semantics, depending on the backend:
> 
> * nop    : ignored (not a problem)
> * simple : enables tracing, but sets dynamic state to disable
> * ust    : disables tracing (uses nop backend)
> * dtrace : same as simple
> 
> Would it be possible to just use nop whenever the event is disabled in
> trace-events? If you agree I can cook the patch, as it's pretty simple.

I don't particularly see the point of the 'disable' keyword existing at
all, unless there are performance implications for a particular trace 
backend. For the DTrace backend I strip & ignore the disable keyword
because probes that are compiled in, reduce to a inline conditional 
check that has no serious overhead when no trace client is active.

Regards,
Daniel
-- 
|: Red Hat, Engineering, London    -o-   http://people.redhat.com/berrange/ :|
|: http://libvirt.org -o- http://virt-manager.org -o- http://deltacloud.org :|
|: http://autobuild.org        -o-         http://search.cpan.org/~danberr/ :|
|: GnuPG: 7D3B9505  -o-   F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 :|