From mboxrd@z Thu Jan 1 00:00:00 1970 From: David Laight Date: Wed, 11 Jul 2018 11:09:33 +0000 Subject: RE: [PATCH net] sctp: Free connecting association if there is a pending signal Message-Id: List-Id: References: <20180709101142.GA1820@bordel.klfree.net> In-Reply-To: <20180709101142.GA1820@bordel.klfree.net> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: linux-sctp@vger.kernel.org From: Petr Malat > Sent: 11 July 2018 10:13 ... > I can look further and try to change the patch to free the association > only for blocking sockets, but that leads me to a question what is > actually the intended behavior for these cases, should SCTP be as close > to TCP as possible (to make s/IPPROTO_TCP/IPPROTO_SCTP/ working for > most applications)? Most TCP applications will fail to work properly over SCTP for other reasons. At least one problem is that unless the application enables receipt of 'sctp_association_event' it won't be told about connection restarts and will get very confused by the missing data. It is also worth noting that TCP is likely to be considerably faster than SCTP. Partially because the SCTP code is significantly more complicated, and partially because it is difficult for an application to get multiple data chunks into a single ethernet frame (unless the traffic pattern allows Nagle be enabled). Even some of the protocols which (I think) were co-developed with SCTP (I'm thinking about SIGTRAN M3UA and M2PA) have mis-understandings about the way SCTP behaves and probably work better over TCP! David