Hi,
I wrote this message a view days ago but nobody answered so i dug into
the code of event_iocp.c. As far as I can figure out, concurrent calls
to the set callback _can_ occur because in event_iocp.c:80
handle_entry is called without port->lock beeing locked.
Is this correct? Moreover is it correct to assume that if i lock a
buffer for writing in an bufferevent callback, the amount of data to
be read won't change until i unlock the buffer?
On Wed, Jun 24, 2009 at 9:33 AM, q6Yr7e0o
nIJDVMjC wrote:
> Hi,
>
>
> I have some questions regarding the IOCP backend when used with
> bufferevents. How much locking is required in the callbacks? For
> example if I'm waiting for read events, will it be possible that one
> read event triggers a read callback and while this is running another
> read callback is triggered on the same socket because there is new
> data avaiable?
>
> Or is there any garantee that a read callback isn't called while it's
> still running?
>
___
Libevent-users mailing list
[email protected]
http://monkeymail.org/mailman/listinfo/libevent-users
