From mboxrd@z Thu Jan 1 00:00:00 1970 From: Tadeusz Struk Subject: Re: [PATCH RFC 4/4] crypto: qat - Add new algif interface for userspace Date: Thu, 06 Nov 2014 19:43:23 -0800 Message-ID: <545C3FDB.4010106@intel.com> References: <20141106195931.13328.90501.stgit@tstruk-mobl1> <20141106195953.13328.77022.stgit@tstruk-mobl1> <20141107025607.GA9474@gondor.apana.org.au> Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7BIT Cc: davem@davemloft.net, linux-crypto@vger.kernel.org, qat-linux@intel.com To: Herbert Xu Return-path: Received: from mga09.intel.com ([134.134.136.24]:32812 "EHLO mga09.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751216AbaKGDp4 (ORCPT ); Thu, 6 Nov 2014 22:45:56 -0500 In-Reply-To: <20141107025607.GA9474@gondor.apana.org.au> Sender: linux-crypto-owner@vger.kernel.org List-ID: Hi Herbert, On 11/06/2014 06:56 PM, Herbert Xu wrote: >> Add new socket algif interface for userspace for symmetric and asymmetric >> > crypto. >> > >> > Signed-off-by: Tadeusz Struk > No this is not acceptable. algif is meant to expose generic > algorithms to user-space not a specific hardware implementation. Yes, I know that PF_ALG wasn't designed to expose hardware specific implementation, but the problem is that currently there is no interface that allows using HW acceleration for algorithms such as RSA, DSA, or Diffie-Hellman. Using PF_ALG works perfectly for this purpose. I can get around 130K RSA decrypt operations per second with 1024 bits modulus. This also scales very well for more accelerators. PF_ALG meant to be crypto interface for user space and this is what I'm using it for here :) Is there any way I can make it (more) acceptable? Regards, Tadeusz