From mboxrd@z Thu Jan 1 00:00:00 1970 From: Neil Horman Subject: Re: [RFC 1/2] net: Introduce recvmmsg socket syscall Date: Fri, 22 May 2009 16:06:11 -0400 Message-ID: <20090522200611.GB967@hmsreliant.think-freely.org> References: <20090520230652.GB5956@ghostprotocols.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: David Miller , netdev@vger.kernel.org, Chris Van Hoof , Clark Williams To: Arnaldo Carvalho de Melo Return-path: Received: from charlotte.tuxdriver.com ([70.61.120.58]:58904 "EHLO smtp.tuxdriver.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1756939AbZEVUG0 (ORCPT ); Fri, 22 May 2009 16:06:26 -0400 Content-Disposition: inline In-Reply-To: <20090520230652.GB5956@ghostprotocols.net> Sender: netdev-owner@vger.kernel.org List-ID: On Wed, May 20, 2009 at 08:06:52PM -0300, Arnaldo Carvalho de Melo wrote: > Meaning receive multiple messages, reducing the number of syscalls and > net stack entry/exit operations. > > Next patches will introduce mechanisms where protocols that want to > optimize this operation will provide an unlocked_recvmsg operation. > Not to throw more questions into the mix again, but didn't Ingo write a batching syscall a while back, which let you issue several syscalls in one trap to kernel space? I understand that your approach has some efficiency gains over that, but did that ever get accepted upstream? Is the overlap there sufficient to make this approach redundant? Or are the gains in performance here sufficient to warrant this new call? Regards Neil