From patchwork Mon Oct 14 16:16:44 2019 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Ard Biesheuvel X-Patchwork-Id: 11189033 X-Patchwork-Delegate: herbert@gondor.apana.org.au Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id C3B1413BD for ; Mon, 14 Oct 2019 16:16:55 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id 95E23217D9 for ; Mon, 14 Oct 2019 16:16:55 +0000 (UTC) Authentication-Results: mail.kernel.org; dkim=pass (2048-bit key) header.d=linaro.org header.i=@linaro.org header.b="sEUdGrbs" Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S2387548AbfJNQQz (ORCPT ); Mon, 14 Oct 2019 12:16:55 -0400 Received: from mail-wr1-f65.google.com ([209.85.221.65]:36599 "EHLO mail-wr1-f65.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1728931AbfJNQQy (ORCPT ); Mon, 14 Oct 2019 12:16:54 -0400 Received: by mail-wr1-f65.google.com with SMTP id y19so20444862wrd.3 for ; Mon, 14 Oct 2019 09:16:52 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=linaro.org; s=google; h=from:to:cc:subject:date:message-id:in-reply-to:references :mime-version:content-transfer-encoding; bh=3dT5xeg+W/abI4zH3wzED14fHVMhPnQCtSYtaVYmbAk=; b=sEUdGrbsRLd3EwtlPYCvYLobAS8DboVsmiF74W6HbzLDftwr/2u8ncB4S560bifRGE cgbT3SSMn4mBGZYNQlsumpBlqLzuBeXOocIkr78gGAArNgCtTR6L83FZknC34c7E2Eq6 1izW8TIiSloeHIfl1/lf75NpZ/DMXnGXV99UXpV35b1ZAgENScr8YRTmLPIjf0SyzMML xsP3qL/yg0YgPu2VG8n5w8mexdZGGLVHotqUtDEyaiSpQJDrolkI4cJGdhH7XCPQ1BH+ jyqgrEH3iQ6sim9TauqewOO+ZUW9WFsPtNuvQoXGJvLz1DT/PEwcORhdNGx6e3Vdk5yQ pMLg== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:from:to:cc:subject:date:message-id:in-reply-to :references:mime-version:content-transfer-encoding; bh=3dT5xeg+W/abI4zH3wzED14fHVMhPnQCtSYtaVYmbAk=; b=fQKV9zisX/9RjThU0y9KA7PIgiGfI1sXP2r7LBvxksw/KPwOl/SiNgwvkLsOLjNUA/ 5Y7RlOpHCDYg5QYRuFWeFDvh1pCHQoA27Kdnu6Lp+JSe5lMYvw15cME5RkbwMrELV7pR 6r9BLP+NzeHmsMdn1ezcn5nNZn6Im+TlX1hHceOf4r3ZpZI2AV4P2n33NVeDFi/xJymx P5bwth9gosdTqk/oB+O/7X07wIISmundQ7SRAxWVVJGvC9UL/Kd7xg6l6rDBuE3NxlNN m7FIq3+WXOP1Un2FQDV3psBrstv7jie2+WfD0IDJiWBsE3upFwqIzlOPQyHjDZFw7gFq Z1BQ== X-Gm-Message-State: APjAAAWyKc+PNw/Txx4oUJsRoefSe/66rFOGN3kX5IPhO5zQGJ4u6iOl wUovp31uxrB5q5AJr0d7yTa0SD66QF+ckA== X-Google-Smtp-Source: APXvYqx+NUcHnWHkKMY5rggcvxBJ1qKMdlWkT6RMALJ5Vt7QVdMcaeRNmrOxXRby/i41dhpL0T4n9g== X-Received: by 2002:a5d:4286:: with SMTP id k6mr14780034wrq.192.1571069811593; Mon, 14 Oct 2019 09:16:51 -0700 (PDT) Received: from localhost.localdomain (aaubervilliers-681-1-23-27.w90-88.abo.wanadoo.fr. [90.88.143.27]) by smtp.gmail.com with ESMTPSA id a14sm17308655wmm.44.2019.10.14.09.16.50 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Mon, 14 Oct 2019 09:16:50 -0700 (PDT) From: Ard Biesheuvel To: linux-crypto@vger.kernel.org Cc: linux-arm-kernel@lists.infradead.org, herbert@gondor.apana.org.au, Ard Biesheuvel Subject: [PATCH v2 1/2] crypto: aegis128 - avoid function pointers for parameterization Date: Mon, 14 Oct 2019 18:16:44 +0200 Message-Id: <20191014161645.1961-2-ard.biesheuvel@linaro.org> X-Mailer: git-send-email 2.20.1 In-Reply-To: <20191014161645.1961-1-ard.biesheuvel@linaro.org> References: <20191014161645.1961-1-ard.biesheuvel@linaro.org> MIME-Version: 1.0 Sender: linux-crypto-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-crypto@vger.kernel.org Instead of passing around an ops structure with function pointers, which forces indirect calls to be used, refactor the code slightly so we can use ordinary function calls. At the same time, switch to a static key to decide whether or not the SIMD code path may be used. Signed-off-by: Ard Biesheuvel --- crypto/aegis128-core.c | 105 +++++++++----------- 1 file changed, 46 insertions(+), 59 deletions(-) diff --git a/crypto/aegis128-core.c b/crypto/aegis128-core.c index 80e73611bd5c..fe7ab66dd8f9 100644 --- a/crypto/aegis128-core.c +++ b/crypto/aegis128-core.c @@ -13,6 +13,7 @@ #include #include #include +#include #include #include #include @@ -35,15 +36,7 @@ struct aegis_ctx { union aegis_block key; }; -struct aegis128_ops { - int (*skcipher_walk_init)(struct skcipher_walk *walk, - struct aead_request *req, bool atomic); - - void (*crypt_chunk)(struct aegis_state *state, u8 *dst, - const u8 *src, unsigned int size); -}; - -static bool have_simd; +static __ro_after_init DEFINE_STATIC_KEY_FALSE(have_simd); static const union aegis_block crypto_aegis_const[2] = { { .words64 = { @@ -59,7 +52,7 @@ static const union aegis_block crypto_aegis_const[2] = { static bool aegis128_do_simd(void) { #ifdef CONFIG_CRYPTO_AEGIS128_SIMD - if (have_simd) + if (static_branch_likely(&have_simd)) return crypto_simd_usable(); #endif return false; @@ -323,25 +316,27 @@ static void crypto_aegis128_process_ad(struct aegis_state *state, } } -static void crypto_aegis128_process_crypt(struct aegis_state *state, - struct aead_request *req, - const struct aegis128_ops *ops) +static __always_inline +int crypto_aegis128_process_crypt(struct aegis_state *state, + struct aead_request *req, + struct skcipher_walk *walk, + void (*crypt)(struct aegis_state *state, + u8 *dst, const u8 *src, + unsigned int size)) { - struct skcipher_walk walk; + int err = 0; - ops->skcipher_walk_init(&walk, req, false); + while (walk->nbytes) { + unsigned int nbytes = walk->nbytes; - while (walk.nbytes) { - unsigned int nbytes = walk.nbytes; + if (nbytes < walk->total) + nbytes = round_down(nbytes, walk->stride); - if (nbytes < walk.total) - nbytes = round_down(nbytes, walk.stride); + crypt(state, walk->dst.virt.addr, walk->src.virt.addr, nbytes); - ops->crypt_chunk(state, walk.dst.virt.addr, walk.src.virt.addr, - nbytes); - - skcipher_walk_done(&walk, walk.nbytes - nbytes); + err = skcipher_walk_done(walk, walk->nbytes - nbytes); } + return err; } static void crypto_aegis128_final(struct aegis_state *state, @@ -390,39 +385,27 @@ static int crypto_aegis128_setauthsize(struct crypto_aead *tfm, return 0; } -static void crypto_aegis128_crypt(struct aead_request *req, - union aegis_block *tag_xor, - unsigned int cryptlen, - const struct aegis128_ops *ops) +static int crypto_aegis128_encrypt(struct aead_request *req) { struct crypto_aead *tfm = crypto_aead_reqtfm(req); + union aegis_block tag = {}; + unsigned int authsize = crypto_aead_authsize(tfm); struct aegis_ctx *ctx = crypto_aead_ctx(tfm); + unsigned int cryptlen = req->cryptlen; + struct skcipher_walk walk; struct aegis_state state; crypto_aegis128_init(&state, &ctx->key, req->iv); crypto_aegis128_process_ad(&state, req->src, req->assoclen); - crypto_aegis128_process_crypt(&state, req, ops); - crypto_aegis128_final(&state, tag_xor, req->assoclen, cryptlen); -} - -static int crypto_aegis128_encrypt(struct aead_request *req) -{ - const struct aegis128_ops *ops = &(struct aegis128_ops){ - .skcipher_walk_init = skcipher_walk_aead_encrypt, - .crypt_chunk = crypto_aegis128_encrypt_chunk, - }; - - struct crypto_aead *tfm = crypto_aead_reqtfm(req); - union aegis_block tag = {}; - unsigned int authsize = crypto_aead_authsize(tfm); - unsigned int cryptlen = req->cryptlen; + skcipher_walk_aead_encrypt(&walk, req, false); if (aegis128_do_simd()) - ops = &(struct aegis128_ops){ - .skcipher_walk_init = skcipher_walk_aead_encrypt, - .crypt_chunk = crypto_aegis128_encrypt_chunk_simd }; - - crypto_aegis128_crypt(req, &tag, cryptlen, ops); + crypto_aegis128_process_crypt(&state, req, &walk, + crypto_aegis128_encrypt_chunk_simd); + else + crypto_aegis128_process_crypt(&state, req, &walk, + crypto_aegis128_encrypt_chunk); + crypto_aegis128_final(&state, &tag, req->assoclen, cryptlen); scatterwalk_map_and_copy(tag.bytes, req->dst, req->assoclen + cryptlen, authsize, 1); @@ -431,26 +414,29 @@ static int crypto_aegis128_encrypt(struct aead_request *req) static int crypto_aegis128_decrypt(struct aead_request *req) { - const struct aegis128_ops *ops = &(struct aegis128_ops){ - .skcipher_walk_init = skcipher_walk_aead_decrypt, - .crypt_chunk = crypto_aegis128_decrypt_chunk, - }; static const u8 zeros[AEGIS128_MAX_AUTH_SIZE] = {}; - struct crypto_aead *tfm = crypto_aead_reqtfm(req); union aegis_block tag; unsigned int authsize = crypto_aead_authsize(tfm); unsigned int cryptlen = req->cryptlen - authsize; + struct aegis_ctx *ctx = crypto_aead_ctx(tfm); + struct skcipher_walk walk; + struct aegis_state state; scatterwalk_map_and_copy(tag.bytes, req->src, req->assoclen + cryptlen, authsize, 0); - if (aegis128_do_simd()) - ops = &(struct aegis128_ops){ - .skcipher_walk_init = skcipher_walk_aead_decrypt, - .crypt_chunk = crypto_aegis128_decrypt_chunk_simd }; + crypto_aegis128_init(&state, &ctx->key, req->iv); + crypto_aegis128_process_ad(&state, req->src, req->assoclen); - crypto_aegis128_crypt(req, &tag, cryptlen, ops); + skcipher_walk_aead_decrypt(&walk, req, false); + if (aegis128_do_simd()) + crypto_aegis128_process_crypt(&state, req, &walk, + crypto_aegis128_decrypt_chunk_simd); + else + crypto_aegis128_process_crypt(&state, req, &walk, + crypto_aegis128_decrypt_chunk); + crypto_aegis128_final(&state, &tag, req->assoclen, cryptlen); return crypto_memneq(tag.bytes, zeros, authsize) ? -EBADMSG : 0; } @@ -481,8 +467,9 @@ static struct aead_alg crypto_aegis128_alg = { static int __init crypto_aegis128_module_init(void) { - if (IS_ENABLED(CONFIG_CRYPTO_AEGIS128_SIMD)) - have_simd = crypto_aegis128_have_simd(); + if (IS_ENABLED(CONFIG_CRYPTO_AEGIS128_SIMD) && + crypto_aegis128_have_simd()) + static_branch_enable(&have_simd); return crypto_register_aead(&crypto_aegis128_alg); } From patchwork Mon Oct 14 16:16:45 2019 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Ard Biesheuvel X-Patchwork-Id: 11189035 X-Patchwork-Delegate: herbert@gondor.apana.org.au Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id 9325914DB for ; Mon, 14 Oct 2019 16:16:56 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id 6B6BA217D9 for ; Mon, 14 Oct 2019 16:16:56 +0000 (UTC) Authentication-Results: mail.kernel.org; dkim=pass (2048-bit key) header.d=linaro.org header.i=@linaro.org header.b="ropoQQyZ" Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1728931AbfJNQQz (ORCPT ); Mon, 14 Oct 2019 12:16:55 -0400 Received: from mail-wm1-f66.google.com ([209.85.128.66]:55954 "EHLO mail-wm1-f66.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1732278AbfJNQQz (ORCPT ); Mon, 14 Oct 2019 12:16:55 -0400 Received: by mail-wm1-f66.google.com with SMTP id a6so17873324wma.5 for ; Mon, 14 Oct 2019 09:16:53 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=linaro.org; s=google; h=from:to:cc:subject:date:message-id:in-reply-to:references :mime-version:content-transfer-encoding; bh=IpqwZs59nyi92LNM6KIZI+WaiWKBSPMKiEsrKFRiIIQ=; b=ropoQQyZtGHuwo63cXLCu0UHmNi3J3G0O3/6hGn/jCZ8k2yn5blG9AbNTOgLWawfN/ gnyYYtlbMMcYEqrdCPnifs3H4TaPZtx6MrMc6m6MZ2vh/sqvCueoWAgTivlT39idODkK MJjSxRiF0ZHO1SuAS9qGJcarbxYxUx2dRTat7JQ3mcOjWI/XP+0Zup6u4PNnUJ2Nw+Xe 5bPgcg4OlLhXxVT+a+usv3V3UI98oQHNKSgcVGca/AQuYDDpItaUmBaxfKxQlJkoO2uw OrcLjnI7oI+fYWuZgyVMAxwLrdv+LATzz2E9LlPVSpy3ZRSnDJVU5vR1b8UtPyULeQGA ngWw== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:from:to:cc:subject:date:message-id:in-reply-to :references:mime-version:content-transfer-encoding; bh=IpqwZs59nyi92LNM6KIZI+WaiWKBSPMKiEsrKFRiIIQ=; b=sHFZkTl9+zcGWFuC+VM/MR8aAinekZX+53FV87LqNQniO2ugnhjLGf2GZ6Gf+KiDAF hYpH2Ax/O5Eylx5CFPi+WbsnJXCFOiMlBR8qce5YOXcLy0eMiN2D9wxcKslQ3jehVgex kmNIobnaG7C/WnidEd4jSt4xQ1Gedh3zS9Qqe6dNlH23uegNynZ8ZnoAGiM/iQfbv8+z 2BopQmM3rgQ/eqF+foLloFreaJx4koqIjXymOxP8LT/X7rvsYLRBhMjTdksVzop4CV5k mlH+I12+wPEEMmYpyskk0vkTzthOJOnWYqwUeiuejdS3j1hi7n2IiyUOUgqZ6RDeuIeM wS1Q== X-Gm-Message-State: APjAAAUG9rq+rfL5yBco7iihqY6Jd9pfUZq4Bc/TWAUq6gU7pTla9NiI cSj72r5RuB/I5Er0Ja/lsQmJ2oJdHD5UbA== X-Google-Smtp-Source: APXvYqxZxQhFWBsmtL0WslRvh3I5uRBH3fe6rYKEQkXGdDmX+4c9gnRs80V9LP2Ughujf0e7h9vdmg== X-Received: by 2002:a1c:740a:: with SMTP id p10mr16573764wmc.90.1571069812884; Mon, 14 Oct 2019 09:16:52 -0700 (PDT) Received: from localhost.localdomain (aaubervilliers-681-1-23-27.w90-88.abo.wanadoo.fr. [90.88.143.27]) by smtp.gmail.com with ESMTPSA id a14sm17308655wmm.44.2019.10.14.09.16.51 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Mon, 14 Oct 2019 09:16:51 -0700 (PDT) From: Ard Biesheuvel To: linux-crypto@vger.kernel.org Cc: linux-arm-kernel@lists.infradead.org, herbert@gondor.apana.org.au, Ard Biesheuvel Subject: [PATCH v2 2/2] crypto: aegis128 - duplicate init() and final() hooks in SIMD code Date: Mon, 14 Oct 2019 18:16:45 +0200 Message-Id: <20191014161645.1961-3-ard.biesheuvel@linaro.org> X-Mailer: git-send-email 2.20.1 In-Reply-To: <20191014161645.1961-1-ard.biesheuvel@linaro.org> References: <20191014161645.1961-1-ard.biesheuvel@linaro.org> MIME-Version: 1.0 Sender: linux-crypto-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-crypto@vger.kernel.org In order to speed up aegis128 processing even more, duplicate the init() and final() routines as SIMD versions in their entirety. This results in a 2x speedup on ARM Cortex-A57 for ~1500 byte packets (using AES instructions). Signed-off-by: Ard Biesheuvel --- crypto/aegis128-core.c | 38 ++++++++++----- crypto/aegis128-neon-inner.c | 50 ++++++++++++++++++++ crypto/aegis128-neon.c | 21 ++++++++ 3 files changed, 97 insertions(+), 12 deletions(-) diff --git a/crypto/aegis128-core.c b/crypto/aegis128-core.c index fe7ab66dd8f9..71c11cb5bad1 100644 --- a/crypto/aegis128-core.c +++ b/crypto/aegis128-core.c @@ -60,10 +60,16 @@ static bool aegis128_do_simd(void) bool crypto_aegis128_have_simd(void); void crypto_aegis128_update_simd(struct aegis_state *state, const void *msg); +void crypto_aegis128_init_simd(struct aegis_state *state, + const union aegis_block *key, + const u8 *iv); void crypto_aegis128_encrypt_chunk_simd(struct aegis_state *state, u8 *dst, const u8 *src, unsigned int size); void crypto_aegis128_decrypt_chunk_simd(struct aegis_state *state, u8 *dst, const u8 *src, unsigned int size); +void crypto_aegis128_final_simd(struct aegis_state *state, + union aegis_block *tag_xor, + u64 assoclen, u64 cryptlen); static void crypto_aegis128_update(struct aegis_state *state) { @@ -395,17 +401,21 @@ static int crypto_aegis128_encrypt(struct aead_request *req) struct skcipher_walk walk; struct aegis_state state; - crypto_aegis128_init(&state, &ctx->key, req->iv); - crypto_aegis128_process_ad(&state, req->src, req->assoclen); - skcipher_walk_aead_encrypt(&walk, req, false); - if (aegis128_do_simd()) + if (aegis128_do_simd()) { + crypto_aegis128_init_simd(&state, &ctx->key, req->iv); + crypto_aegis128_process_ad(&state, req->src, req->assoclen); crypto_aegis128_process_crypt(&state, req, &walk, crypto_aegis128_encrypt_chunk_simd); - else + crypto_aegis128_final_simd(&state, &tag, req->assoclen, + cryptlen); + } else { + crypto_aegis128_init(&state, &ctx->key, req->iv); + crypto_aegis128_process_ad(&state, req->src, req->assoclen); crypto_aegis128_process_crypt(&state, req, &walk, crypto_aegis128_encrypt_chunk); - crypto_aegis128_final(&state, &tag, req->assoclen, cryptlen); + crypto_aegis128_final(&state, &tag, req->assoclen, cryptlen); + } scatterwalk_map_and_copy(tag.bytes, req->dst, req->assoclen + cryptlen, authsize, 1); @@ -426,17 +436,21 @@ static int crypto_aegis128_decrypt(struct aead_request *req) scatterwalk_map_and_copy(tag.bytes, req->src, req->assoclen + cryptlen, authsize, 0); - crypto_aegis128_init(&state, &ctx->key, req->iv); - crypto_aegis128_process_ad(&state, req->src, req->assoclen); - skcipher_walk_aead_decrypt(&walk, req, false); - if (aegis128_do_simd()) + if (aegis128_do_simd()) { + crypto_aegis128_init_simd(&state, &ctx->key, req->iv); + crypto_aegis128_process_ad(&state, req->src, req->assoclen); crypto_aegis128_process_crypt(&state, req, &walk, crypto_aegis128_decrypt_chunk_simd); - else + crypto_aegis128_final_simd(&state, &tag, req->assoclen, + cryptlen); + } else { + crypto_aegis128_init(&state, &ctx->key, req->iv); + crypto_aegis128_process_ad(&state, req->src, req->assoclen); crypto_aegis128_process_crypt(&state, req, &walk, crypto_aegis128_decrypt_chunk); - crypto_aegis128_final(&state, &tag, req->assoclen, cryptlen); + crypto_aegis128_final(&state, &tag, req->assoclen, cryptlen); + } return crypto_memneq(tag.bytes, zeros, authsize) ? -EBADMSG : 0; } diff --git a/crypto/aegis128-neon-inner.c b/crypto/aegis128-neon-inner.c index f05310ca22aa..2a660ac1bc3a 100644 --- a/crypto/aegis128-neon-inner.c +++ b/crypto/aegis128-neon-inner.c @@ -132,6 +132,36 @@ void preload_sbox(void) :: "r"(crypto_aes_sbox)); } +void crypto_aegis128_init_neon(void *state, const void *key, const void *iv) +{ + static const uint8_t const0[] = { + 0x00, 0x01, 0x01, 0x02, 0x03, 0x05, 0x08, 0x0d, + 0x15, 0x22, 0x37, 0x59, 0x90, 0xe9, 0x79, 0x62, + }; + static const uint8_t const1[] = { + 0xdb, 0x3d, 0x18, 0x55, 0x6d, 0xc2, 0x2f, 0xf1, + 0x20, 0x11, 0x31, 0x42, 0x73, 0xb5, 0x28, 0xdd, + }; + uint8x16_t k = vld1q_u8(key); + uint8x16_t kiv = k ^ vld1q_u8(iv); + struct aegis128_state st = {{ + kiv, + vld1q_u8(const1), + vld1q_u8(const0), + k ^ vld1q_u8(const0), + k ^ vld1q_u8(const1), + }}; + int i; + + preload_sbox(); + + for (i = 0; i < 5; i++) { + st = aegis128_update_neon(st, k); + st = aegis128_update_neon(st, kiv); + } + aegis128_save_state_neon(st, state); +} + void crypto_aegis128_update_neon(void *state, const void *msg) { struct aegis128_state st = aegis128_load_state_neon(state); @@ -210,3 +240,23 @@ void crypto_aegis128_decrypt_chunk_neon(void *state, void *dst, const void *src, aegis128_save_state_neon(st, state); } + +void crypto_aegis128_final_neon(void *state, void *tag_xor, uint64_t assoclen, + uint64_t cryptlen) +{ + struct aegis128_state st = aegis128_load_state_neon(state); + uint8x16_t v; + int i; + + preload_sbox(); + + v = st.v[3] ^ (uint8x16_t)vcombine_u64(vmov_n_u64(8 * assoclen), + vmov_n_u64(8 * cryptlen)); + + for (i = 0; i < 7; i++) + st = aegis128_update_neon(st, v); + + v = vld1q_u8(tag_xor); + v ^= st.v[0] ^ st.v[1] ^ st.v[2] ^ st.v[3] ^ st.v[4]; + vst1q_u8(tag_xor, v); +} diff --git a/crypto/aegis128-neon.c b/crypto/aegis128-neon.c index 751f9c195aa4..8271b1fa0fbc 100644 --- a/crypto/aegis128-neon.c +++ b/crypto/aegis128-neon.c @@ -8,11 +8,14 @@ #include "aegis.h" +void crypto_aegis128_init_neon(void *state, const void *key, const void *iv); void crypto_aegis128_update_neon(void *state, const void *msg); void crypto_aegis128_encrypt_chunk_neon(void *state, void *dst, const void *src, unsigned int size); void crypto_aegis128_decrypt_chunk_neon(void *state, void *dst, const void *src, unsigned int size); +void crypto_aegis128_final_neon(void *state, void *tag_xor, uint64_t assoclen, + uint64_t cryptlen); int aegis128_have_aes_insn __ro_after_init; @@ -25,6 +28,15 @@ bool crypto_aegis128_have_simd(void) return IS_ENABLED(CONFIG_ARM64); } +void crypto_aegis128_init_simd(union aegis_block *state, + const union aegis_block *key, + const u8 *iv) +{ + kernel_neon_begin(); + crypto_aegis128_init_neon(state, key, iv); + kernel_neon_end(); +} + void crypto_aegis128_update_simd(union aegis_block *state, const void *msg) { kernel_neon_begin(); @@ -47,3 +59,12 @@ void crypto_aegis128_decrypt_chunk_simd(union aegis_block *state, u8 *dst, crypto_aegis128_decrypt_chunk_neon(state, dst, src, size); kernel_neon_end(); } + +void crypto_aegis128_final_simd(union aegis_block *state, + union aegis_block *tag_xor, + u64 assoclen, u64 cryptlen) +{ + kernel_neon_begin(); + crypto_aegis128_final_neon(state, tag_xor, assoclen, cryptlen); + kernel_neon_end(); +}