From: Tadeusz Struk Subject: Re: [PATCH 2/3] crypto: algif_akcipher user space interface Date: Wed, 22 Jul 2015 15:04:37 -0700 Message-ID: <55B01375.3040007@intel.com> References: <3908391.JkjAozYBaW@tachyon.chronox.de> <1537904.OY6iC43nQX@tachyon.chronox.de> <55AFBE4B.8080805@intel.com> <7336837.Ud0PiFKhfA@tachyon.chronox.de> Mime-Version: 1.0 Content-Type: text/plain; charset=windows-1252 Content-Transfer-Encoding: 7bit Cc: herbert@gondor.apana.org.au, linux-crypto@vger.kernel.org To: Stephan Mueller Return-path: Received: from mga14.intel.com ([192.55.52.115]:64531 "EHLO mga14.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753496AbbGVWFN (ORCPT ); Wed, 22 Jul 2015 18:05:13 -0400 In-Reply-To: <7336837.Ud0PiFKhfA@tachyon.chronox.de> Sender: linux-crypto-owner@vger.kernel.org List-ID: On 07/22/2015 11:55 AM, Stephan Mueller wrote: > I concur with you. But we have only two options: > > - either use SGLs which the current akcipher API does not do > > - or do a memcpy of the sendpage data into the internal buffer I would memcpy it here. Both the current software and QAT implementations do not support SGL so we'll end up memcpy it anyway. > > As the sendpage already has a speed penalty, I did not like the latter one. > Based on my measurements for AEAD, Hash and skicpher, sendpage starts to > become faster than sendmsg with input buffers > 8 to 16 kBytes (sendpage as at > least 4 syscalls where sendmsg uses only two). > > As our current akcipher API does not reach the mentioned limit, I opted to > require one sendpage call. > > But if we change the akcipher API to SGLs, I will lift that limit. As you said - the preferred way of invoking this would be sendmsg with data instead of splice. This is only to make splice happy in case the data is not in a contiguous buffer. Thanks, T