On Mon, Jun 25, 2001 at 08:09:01AM +0200, Dag-Erling Smorgrav wrote:
> Sean Chittenden <[EMAIL PROTECTED]> writes:
> > Quick question. Anyone know how gracefully the kernel handles a
> > socket connection that is killed by the client between a select and
> > accept call? I don't expect any p
Sean Chittenden <[EMAIL PROTECTED]> writes:
> It is possible to select(2) a socket for the purposes of doing an
> accept() by selecting it for read.
This simply means that select(2) will consider a listen socket
readable when there's at least one incoming connection in that
socket's lis
> There was one in FreeBSD too. It's been fixed; accept(2) will return
> -1 and set errno to ECONNABORTED, which you'd know if you'd RTFM.
Already RTFM'd. The following was a tad vague and it led me to
be a skeptic.
It is possible to select(2) a socket for the purposes of doing a
Sean Chittenden <[EMAIL PROTECTED]> writes:
> Quick question. Anyone know how gracefully the kernel handles a
> socket connection that is killed by the client between a select and
> accept call? I don't expect any problems, but I know there was a race
> condition in Linux that caused all k
4 matches
Mail list logo
