From: "Nicholas A. Bellinger" Subject: Re: [PATCH 0/2] Add struct crypto_alg->cra_check_optimized for crc32c_intel Date: Thu, 10 Mar 2011 02:05:17 -0800 Message-ID: <1299751517.5263.60.camel@haakon2.linux-iscsi.org> References: <1299745272-25477-1-git-send-email-nab@linux-iscsi.org> <20110310084346.GA10650@gondor.apana.org.au> <1299747264.5263.35.camel@haakon2.linux-iscsi.org> <20110310090929.GA10919@gondor.apana.org.au> <1299748387.5263.45.camel@haakon2.linux-iscsi.org> Mime-Version: 1.0 Content-Type: text/plain Content-Transfer-Encoding: 7bit Cc: linux-kernel , linux-crypto , James Bottomley , Christoph Hellwig , Randy Dunlap , linux-scsi To: Herbert Xu Return-path: Received: from mail.linux-iscsi.org ([67.23.28.174]:43854 "EHLO linux-iscsi.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750921Ab1CJKMG (ORCPT ); Thu, 10 Mar 2011 05:12:06 -0500 In-Reply-To: <1299748387.5263.45.camel@haakon2.linux-iscsi.org> Sender: linux-crypto-owner@vger.kernel.org List-ID: On Thu, 2011-03-10 at 01:13 -0800, Nicholas A. Bellinger wrote: > On Thu, 2011-03-10 at 17:09 +0800, Herbert Xu wrote: > > On Thu, Mar 10, 2011 at 12:54:24AM -0800, Nicholas A. Bellinger wrote: > > > > > > OK, so you mean each struct crypto_alg should define something like a > > > 'cra_optimized_name' for which request_module(alg->cra_optimized_name) > > > is called somewhere in libcrypto code..? > > > > No, what I mean is that whenever we look up an algorithm through > > crypto_alg_mod_lookup, we should conditionally call modprobe if > > we havn't done so already. > > > > So you just need to record one bit of info in each crypto_alg > > object to indicate whether we have invoked modprobe. I suggest > > adding a CRYPTO_ALG_* bit. > > > > Mmmm, now I am really confused, and please let me apologize in advance > for my lack of experience with libcrypto internals.. ;) > > I thought the problem was that CONFIG_CRYPTO_ALGFOO=y and > CONFIG_CRYPTO_ALGFOO_ARCH_HW_OFFLOAD=m would cause the latter to not > explictly call request_module() for this HW offload case.. > > So what I don't understand how adding a request_module() call to a list > of known modules works when crc32c_intel.ko has not been loaded yet..? > > Am I missing something obvious wrt to how crc32c.ko can tell libcrypto > about which architecture dependent optimized modules it should load..? > Just to clarify a bit on my previous comments.. We are still expecting the libcrypto consumer (iscsi_target_mod.ko) to call the arch independent crypto_alloc_hash("crc32c", ...) in order to have libcrypto backend logic perform a request_module() upon architecture dependent offload modules (like crc32c_intel.ko) that libcrypto consumers are not (and should not) be calling directly via crypto_alloc_host("crc32c_intel", ...), correct..? Where I am getting confused is wrt to a new crypto_alg_mod_lookup() -> request_module() call for a struct shash_alg that has not yet be loaded via arch/x86/crypto/crc32c-intel.c:crc32c_intel_mod_init() -> crypto_register_shash(). Thanks, --nab