Hi Nick,
Thanks for the reply.
Since you are unaware of such a case and it's supposed to work. Then I'm
starting to think it's a bug in my code ("double delete" ==> reference count
getting decremented prematurely - which Valgrind will not be able to catch).
The software I'm writing is for production. Therefore stability is very
important to me. How stable is the Alpha. In your opinion should I stay with
2.0.21 or move to the Alpha?
Thanks,
Tomer.
________________________________________
From: [email protected] <[email protected]>
on behalf of Nick Mathewson <[email protected]>
Sent: Tuesday, July 30, 2013 6:06 PM
To: [email protected]
Subject: Re: [Libevent-users] calling bufferevent_free within it's own event
(version 2.0.21).
On Mon, Jul 29, 2013 at 2:55 AM, Tomer Heber <[email protected]> wrote:
> Hi,
>
>
>
> I’ve encountered a strange behavior with ssl bufferevents:
>
> 1. My program calls bufferevent_write(),
>
> 2. (Some error occurs – which is ok) bufferevent_write calls my
> event_cb.
>
> 3. My event_cb releases the bufferevent.
>
> 4. Program segfaults.
>
>
>
> I’ve worked with Valgrind and checked the libevent source code – it seems
> that after bufferevent_write is calling my event_cb (which releases the
> bufferevent) it then modifies a value in the ssl_bufferevent (which is
> already released).
>
>
>
> So what’s the deal? Is it allowed to release a bufferevent with in its own
> events? Or is it not allowed?
It's *supposed* to work. Not sure why it isn't in your case. Is
there a reference count getting decremented prematurely or something?
Have you tried the latest alpha? It refactors how bufferevent cleanup
works significantly.
--
Nick
***********************************************************************
To unsubscribe, send an e-mail to [email protected] with
unsubscribe libevent-users in the
body.***********************************************************************
To unsubscribe, send an e-mail to [email protected] with
unsubscribe libevent-users in the body.
