schmorp at schmorp.de
Tue Jan 1 18:36:48 CET 2008
On Mon, Dec 31, 2007 at 04:45:06PM -0700, Matt Tolton <matt at tolton.com> wrote:
> >From what I can determine, it should not be possible for a watcher to
> be pending within its callback. This should mean that if I stop the
> watcher, it will be ok to free the memory associated with it.
> I'd like to verify that the above is true...is it?
I don't understand the reasoning. A watcher can be pending within its own
callback (because it can become pending while executing the callback, e.g.
when recursing into ev_loop).
stopping, however, will be clearing the pending state, so you can indeed
free its memory after calling stop (even when the watcher already *is*
stopped, ev_xxx_stop will clear any pending state).
if you make sure in your callback that:
a) the watcher is stopped (e.g. because it was a one-shot timer or
you called ev_xxx_sotp)
b) it did not become pending again (e.g. by avoiding calls to ev_loop
or submittign events)
then you can safely free the weatcher memory, it will no longer be accessed
by libev afterwards.
if you do "interesting" stuff in your callback and are not sure wether you
recurse again into the event loop or so, the safe route is to always call
ev_xxx_stop first, but this is unliekly to be necessary.
The choice of a Deliantra, the free code+content MORPG
-----==- _GNU_ http://www.deliantra.net
----==-- _ generation
---==---(_)__ __ ____ __ Marc Lehmann
--==---/ / _ \/ // /\ \/ / pcg at goof.com
More information about the libev