From: Chuck Lever Subject: nfs_utils: O_NONBLOCK sockets in support/nfs/svc_socket.c Date: Thu, 7 Aug 2008 15:58:09 -0400 Message-ID: Mime-Version: 1.0 (Apple Message framework v928.1) Content-Type: text/plain; charset=US-ASCII; format=flowed; delsp=yes Cc: Linux NFS Mailing List To: Neil Brown , Greg Banks Return-path: Received: from agminet02.oracle.com ([141.146.126.229]:43382 "EHLO agminet02.oracle.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752858AbYHGVGb (ORCPT ); Thu, 7 Aug 2008 17:06:31 -0400 Received: from agminet01.oracle.com (agminet01.oracle.com [141.146.126.228]) by agminet02.oracle.com (Switch-3.2.4/Switch-3.1.7) with ESMTP id m77K3AKC002609 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO) for ; Thu, 7 Aug 2008 15:03:10 -0500 Sender: linux-nfs-owner@vger.kernel.org List-ID: Hi guys- I'm looking at svc_socket() and friends to add IPv6 support. Got a question. I see in commits 11d34d11 and 58e58ffb there was a set of fcntl() calls added to make RPC service listener sockets non-blocking. How come this support wasn't added in support/nfs/rpcmisc.c:makesock() as well? I'm probably misreading the logic in rpc_init(), but it seems to me there are possibilities for the listener socket to be created by makesock() and not by svc_socket(), and then that listener would not be O_NONBLOCK. For instance, if there was a port value specified on the mountd command line...? Please straighten me out. Thanks. -- Chuck Lever chuck[dot]lever[at]oracle[dot]com