Re: [Libevent-users] event_add fails with ENOENT when on epoll

[William Ahern <william@xxxxxxxxxxxxxxxxxx>]

On Tue, Apr 20, 2010 at 01:20:21PM -0400, Nick Mathewson wrote:
> On Mon, Apr 19, 2010 at 11:24 PM, Denis Bilenko <denis.bilenko@xxxxxxxxx> wrote:
> > Hi,
> >
> > I've run into an issue where event_add() fails with ENOENT.
> > This happens when a socket reuses a descriptor that was recently
> > closed and was used before with another event.
> >
> > The details are below. The question I have is - what are the ways to
> > work around this?
> >
> 
> You must never modify an event that is currently pending.  If you want
> to reassign it, you need to call event_del(), then modify it, then
> call event_add().
> 
> For some backends (like epoll and kqueue) you need to call event_del()
> on the event when you close the socket, and then event_add() again,

More specifically, _before_ you close the socket, because the act of closing
the socket will also implicitly remove it from the epoll queue, a state
change that libevent can't track. In this case event_del() will return a
failure when it tries to remove the non-existent descriptor from the epoll
queue. Fortunately the order of operations in event_del() is such that the
Right Thing(tm) happens anyhow. Thus, transgressions will usually go
unpunished--it's rare that people check the return value of event_del(), and
there are no other ill effects.

***********************************************************************
To unsubscribe, send an e-mail to majordomo@xxxxxxxxxxxxx with
unsubscribe libevent-users    in the body.