From patchwork Thu Nov 5 16:38:11 2015 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Jerin Jacob X-Patchwork-Id: 8718 Return-Path: X-Original-To: patchwork@dpdk.org Delivered-To: patchwork@dpdk.org Received: from [92.243.14.124] (localhost [IPv6:::1]) by dpdk.org (Postfix) with ESMTP id 09EC891B8; Thu, 5 Nov 2015 17:39:44 +0100 (CET) Received: from na01-bn1-obe.outbound.protection.outlook.com (mail-bn1on0078.outbound.protection.outlook.com [157.56.110.78]) by dpdk.org (Postfix) with ESMTP id 9D27691C1 for ; Thu, 5 Nov 2015 17:39:41 +0100 (CET) Authentication-Results: spf=none (sender IP is ) smtp.mailfrom=Jerin.Jacob@caviumnetworks.com; Received: from jerin.caveonetworks.com (111.93.218.67) by CY1PR0701MB1977.namprd07.prod.outlook.com (10.163.141.19) with Microsoft SMTP Server (TLS) id 15.1.312.18; Thu, 5 Nov 2015 16:39:38 +0000 From: Jerin Jacob To: Date: Thu, 5 Nov 2015 22:08:11 +0530 Message-ID: <1446741498-3096-9-git-send-email-jerin.jacob@caviumnetworks.com> X-Mailer: git-send-email 1.9.3 In-Reply-To: <1446741498-3096-8-git-send-email-jerin.jacob@caviumnetworks.com> References: <1446741498-3096-1-git-send-email-jerin.jacob@caviumnetworks.com> <1446741498-3096-2-git-send-email-jerin.jacob@caviumnetworks.com> <1446741498-3096-3-git-send-email-jerin.jacob@caviumnetworks.com> <1446741498-3096-4-git-send-email-jerin.jacob@caviumnetworks.com> <1446741498-3096-5-git-send-email-jerin.jacob@caviumnetworks.com> <1446741498-3096-6-git-send-email-jerin.jacob@caviumnetworks.com> <1446741498-3096-7-git-send-email-jerin.jacob@caviumnetworks.com> <1446741498-3096-8-git-send-email-jerin.jacob@caviumnetworks.com> MIME-Version: 1.0 X-Originating-IP: [111.93.218.67] X-ClientProxiedBy: MA1PR01CA0067.INDPRD01.PROD.OUTLOOK.COM (25.164.116.167) To CY1PR0701MB1977.namprd07.prod.outlook.com (25.163.141.19) X-Microsoft-Exchange-Diagnostics: 1; CY1PR0701MB1977; 2:mSMdcdwxW6XO0q8xdpY5ESc84hlO5I+PP4pVIlpy69Oo0Tor91fFeN+dpAC5GVoE63HGqWfQBOfohodeFA3zmwkPG8JHOiO/UTOFZERWxjWfA6nBwyK4rfGvxfuuuPoksP+s0hLaHK31tHIp4TI7BWZa3+9nuHOwPz/BLPpvA54=; 3:O4y7m2nWM5i7H7NuXB8mVOr6oBiwOjszdv5vZYkMmmp4qRndsAQp+M4BqQxGYv5zy6XLHVIWU5T4u204l5ow1BUrKfjrkkMX8Ckak53F9H9tmj0j+XL+tU/XKxEmQrs37UrNpN78QSC8zGNd8KCSpA==; 25:HPb5kPeWvyEFKjsrRJjJtQkV5KrYzYZY/yteEXK+Y+g/yYpLTVrIzGxw/u3rxm1ucg9hVRIHXp4xIyBjcqxfMxJVVY6kRHuCw/Mk8H5UkzOiqPPywXpu9h3t2L9ag93rasRMZSiIe5k2l8NqiLXst1u0myiXPOYeRDntv+QxFC/cXmMwCILw4EkW5A7FryaVrkaM/2qaqqFaIwJqwXvU2kcV28VE+a7NBiKRT471RRNB9pZGifeuHMYelV65dD7eM/Au0927Jkr4BfdQinuu8Q== X-Microsoft-Antispam: UriScan:;BCL:0;PCL:0;RULEID:;SRVR:CY1PR0701MB1977; X-Microsoft-Exchange-Diagnostics: 1; CY1PR0701MB1977; 20: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 X-Microsoft-Antispam-PRVS: X-Exchange-Antispam-Report-Test: UriScan:(236414709691187); X-Exchange-Antispam-Report-CFA-Test: BCL:0; PCL:0; RULEID:(601004)(2401047)(520078)(5005006)(8121501046)(3002001)(10201501046); SRVR:CY1PR0701MB1977; BCL:0; PCL:0; RULEID:; SRVR:CY1PR0701MB1977; X-Microsoft-Exchange-Diagnostics: 1; CY1PR0701MB1977; 4:sX8cwspyTrBeCaSaW2FExFibfsf1QnRAO8+3u7qHeKtu+sIeMg75E4fu2fLWoC9xbrtjorXOweQCQl8i4wsd0AWCW6Ytf51+8Tbyt7KRWFwGSc/7kekgMl4OmNm+m1GdIc94dM8GczznhDGc3TUUtYLIVXUH9dNIfmu5E+afDuzUKDG/1c5CIAnqSxIDOIopIBKoNI8iLYADDXS5y55WnKOxe+QJkv+7P+k8YcYkV1qpEAyrx4rOto1IQpjJ1wAz9reL+/bvNpF5g5xzSO1ZTG5XRX1/XGMXTsVEKbiyVR2U64blVjE7dQLk4LFgEp78pA3ffBqNw6Hhdr+k+IT0k2Lc2rsFgSQqLNdpT3cF4eAIo/XX41Y72dncCknxQgKx X-Forefront-PRVS: 0751474A44 X-Forefront-Antispam-Report: SFV:NSPM; SFS:(10009020)(6009001)(189002)(199003)(5007970100001)(86362001)(5001960100002)(107886002)(36756003)(122386002)(40100003)(189998001)(42186005)(5008740100001)(575784001)(87976001)(110136002)(97736004)(66066001)(101416001)(92566002)(2950100001)(5009440100003)(4001430100002)(19580405001)(76176999)(5003940100001)(50986999)(50226001)(50466002)(106356001)(77096005)(105586002)(5004730100002)(2351001)(81156007)(229853001)(48376002)(33646002)(19580395003)(69596002)(53416004)(93886004)(47776003)(7099028); DIR:OUT; SFP:1101; SCL:1; SRVR:CY1PR0701MB1977; H:jerin.caveonetworks.com; FPR:; SPF:None; PTR:InfoNoRecords; MX:1; A:1; LANG:en; Received-SPF: None (protection.outlook.com: caviumnetworks.com does not designate permitted sender hosts) X-Microsoft-Exchange-Diagnostics: =?us-ascii?Q?1; CY1PR0701MB1977; 23:taBharJm1zfhC0VIgCZdvaaOcIyyIATWkoahVpn?= =?us-ascii?Q?qxM7B4vYxxsqDUX3q8W52pFFa8KnsYiDcJGTQIVCPkNrvftHeyi6klTyxjLu?= =?us-ascii?Q?tWAfk33wZFAlZbxChdbj9y9fv1VRP9JkwAXt/S8UWtK74j7dx++Q404evgQO?= =?us-ascii?Q?D6xEi6RBak1+7of00fvAGh7tt0eCvEiIYFPdFY4cWwuzpAuzMwk54NikoQGn?= =?us-ascii?Q?sd//ZnvHCIXGwJWOdYAXpfcbxcDv5AQkPh5FegrDM30xZTr0DiLduyTDrehg?= =?us-ascii?Q?Kw1nCENp6wUN/ngE52/ndJbKS5bRxM/nJ7urxnMHIjuBjiTR+RrnPCRptRaQ?= =?us-ascii?Q?NLsNhvesJYGZv4+gqBFQsS6S4QUXvz91iXcfdn2PKfRPBU54sbUy22uRuqah?= =?us-ascii?Q?AYq4/V+BE046R6P+DkFqZyD4O9OGKTmbq1OnrltObKRXFEAJHm33HIuZJ5aO?= =?us-ascii?Q?1thRQ3ayTXz4z0i/KCEheF1kBIXohom3N19VGGmdOB5yAzFHjRi33+bN7oBo?= =?us-ascii?Q?8m1IwBbUvhavtV4dZEIhRKc+HV7RxHldOPaUUnzyhTo0wyheVA7eMpRc04Nh?= =?us-ascii?Q?9ZL+G8D9XrOcrElIV9kU0bbbnrSsMAfRC9ZCu9VsDXkTKhi8DxgACMh3+9ez?= =?us-ascii?Q?g9O7hCoq1nWlUSYvbF0s4CMsMg+dIp6WBEF6iYFoXTTDYOT9LSeAxSYO0Zuk?= =?us-ascii?Q?Cfxyub0fIfZsbjm/8qLZWUzt36xki4zQTezXAdLbygqiFrqQYke0tp85Eqz4?= =?us-ascii?Q?AXZt7DG1Wa6ZTp26o83THpDVxHD0334FghC8BbZu16NcsIVYuH5jMnSd5brt?= =?us-ascii?Q?PE1eIHo4vGyjKyX4wSKbYKI40A8D6guAnCvYb/XYCYef0oDsWemv7rOBlTfm?= =?us-ascii?Q?1c+mG48VrN7Rfa628ibTNcM2hjQHL1GOdKHvFfoqPLcQbQ7K1SS8/N/nkdNX?= =?us-ascii?Q?7iqX9B+yvk3kDef3l+cKwM7GvKSfN0TqKaiewD6zIP2LUv4jiziapuzLb9Jz?= =?us-ascii?Q?xtK2IS2VZ2dL8k87a5pH4l2LTJ4k4u7lOHTIyU918UY/zKMkWcoePJUxrFAl?= =?us-ascii?Q?NTw8eEp35gVCrJXuE6vNCbUvzHe1elqqXgoXxLQt40h7Z1T+7XvCHG8dqui7?= =?us-ascii?Q?DS2Grv1OOJIr05vALFlNF1IJXtQCRR9vEBkNodX+aVtpQ2l2xt2qgpuvZHaS?= =?us-ascii?Q?psrQuUxGuf7xVTfE=3D?= X-Microsoft-Exchange-Diagnostics: 1; CY1PR0701MB1977; 5:L3ooiCqxXpS8f/9gbgIzgF8yg8wn3UdMGJ+bnO5N6kd84cQ4F6K3OqKlzrjF0I3p3Y4Rc995uMP/P5ccMHkoKMWEJ46/OpmdyJskQcAPSOBC5wHAVetF10pNwBDLk1Tryp666KqMXCAdkLhO+VDUhQ==; 24:Mo8Ublv2yGmOmLh2VhSmXDNGrupTtlE8d64eoNg74alsamYObEezVwyOeVL2R+J6b4+qSnUYE854IR4stE+2wsup3DDgSWW528eKIlJgcng=; 20:Mvj81TfmAiz4yRkKXpYIcwB53t17LOui4MGVgZOfjqaQhHszr24R6OCMkfQCYgCM7A6QL//ZlgGpuu20HTO02g== SpamDiagnosticOutput: 1:23 SpamDiagnosticMetadata: NSPM X-OriginatorOrg: caviumnetworks.com X-MS-Exchange-CrossTenant-OriginalArrivalTime: 05 Nov 2015 16:39:38.3201 (UTC) X-MS-Exchange-CrossTenant-FromEntityHeader: Hosted X-MS-Exchange-Transport-CrossTenantHeadersStamped: CY1PR0701MB1977 Subject: [dpdk-dev] [PATCH 08/15] acl: arm64: acl implementation using NEON gcc intrinsic X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: patches and discussions about DPDK List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: dev-bounces@dpdk.org Sender: "dev" verified with testacl and acl_autotest applications on arm64 architecture. Signed-off-by: Jerin Jacob --- app/test-acl/main.c | 4 + lib/librte_acl/Makefile | 5 + lib/librte_acl/acl.h | 4 + lib/librte_acl/acl_run_neon.c | 46 +++++++ lib/librte_acl/acl_run_neon.h | 289 ++++++++++++++++++++++++++++++++++++++++++ lib/librte_acl/rte_acl.c | 25 ++++ lib/librte_acl/rte_acl.h | 1 + 7 files changed, 374 insertions(+) create mode 100644 lib/librte_acl/acl_run_neon.c create mode 100644 lib/librte_acl/acl_run_neon.h diff --git a/app/test-acl/main.c b/app/test-acl/main.c index 72ce83c..0b0c093 100644 --- a/app/test-acl/main.c +++ b/app/test-acl/main.c @@ -101,6 +101,10 @@ static const struct acl_alg acl_alg[] = { .name = "avx2", .alg = RTE_ACL_CLASSIFY_AVX2, }, + { + .name = "neon", + .alg = RTE_ACL_CLASSIFY_NEON, + }, }; static struct { diff --git a/lib/librte_acl/Makefile b/lib/librte_acl/Makefile index 7a1cf8a..27f91d5 100644 --- a/lib/librte_acl/Makefile +++ b/lib/librte_acl/Makefile @@ -48,9 +48,14 @@ SRCS-$(CONFIG_RTE_LIBRTE_ACL) += rte_acl.c SRCS-$(CONFIG_RTE_LIBRTE_ACL) += acl_bld.c SRCS-$(CONFIG_RTE_LIBRTE_ACL) += acl_gen.c SRCS-$(CONFIG_RTE_LIBRTE_ACL) += acl_run_scalar.c +ifeq ($(CONFIG_RTE_ARCH_ARM64),y) +SRCS-$(CONFIG_RTE_LIBRTE_ACL) += acl_run_neon.c +else SRCS-$(CONFIG_RTE_LIBRTE_ACL) += acl_run_sse.c +endif CFLAGS_acl_run_sse.o += -msse4.1 +CFLAGS_acl_run_neon.o += -flax-vector-conversions -Wno-maybe-uninitialized # # If the compiler supports AVX2 instructions, diff --git a/lib/librte_acl/acl.h b/lib/librte_acl/acl.h index eb4930c..09d6784 100644 --- a/lib/librte_acl/acl.h +++ b/lib/librte_acl/acl.h @@ -230,6 +230,10 @@ int rte_acl_classify_avx2(const struct rte_acl_ctx *ctx, const uint8_t **data, uint32_t *results, uint32_t num, uint32_t categories); +int +rte_acl_classify_neon(const struct rte_acl_ctx *ctx, const uint8_t **data, + uint32_t *results, uint32_t num, uint32_t categories); + #ifdef __cplusplus } #endif /* __cplusplus */ diff --git a/lib/librte_acl/acl_run_neon.c b/lib/librte_acl/acl_run_neon.c new file mode 100644 index 0000000..b014451 --- /dev/null +++ b/lib/librte_acl/acl_run_neon.c @@ -0,0 +1,46 @@ +/* + * BSD LICENSE + * + * Copyright (C) Cavium networks Ltd. 2015. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * * Neither the name of Cavium networks nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#include "acl_run_neon.h" + +int +rte_acl_classify_neon(const struct rte_acl_ctx *ctx, const uint8_t **data, + uint32_t *results, uint32_t num, uint32_t categories) +{ + if (likely(num >= 8)) + return search_neon_8(ctx, data, results, num, categories); + else if (num >= 4) + return search_neon_4(ctx, data, results, num, categories); + else + return rte_acl_classify_scalar(ctx, data, results, num, + categories); +} diff --git a/lib/librte_acl/acl_run_neon.h b/lib/librte_acl/acl_run_neon.h new file mode 100644 index 0000000..cf7c57f --- /dev/null +++ b/lib/librte_acl/acl_run_neon.h @@ -0,0 +1,289 @@ +/* + * BSD LICENSE + * + * Copyright (C) Cavium networks Ltd. 2015. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * * Neither the name of Cavium networks nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#include "acl_run.h" +#include "acl_vect.h" + +struct _neon_acl_const { + rte_xmm_t xmm_shuffle_input; + rte_xmm_t xmm_index_mask; + rte_xmm_t range_base; +} neon_acl_const __attribute__((aligned(RTE_CACHE_LINE_SIZE))) = { + { + .u32 = {0x00000000, 0x04040404, 0x08080808, 0x0c0c0c0c} + }, + { + .u32 = {RTE_ACL_NODE_INDEX, RTE_ACL_NODE_INDEX, + RTE_ACL_NODE_INDEX, RTE_ACL_NODE_INDEX} + }, + { + .u32 = {0xffffff00, 0xffffff04, 0xffffff08, 0xffffff0c} + }, +}; + +/* + * Resolve priority for multiple results (neon version). + * This consists comparing the priority of the current traversal with the + * running set of results for the packet. + * For each result, keep a running array of the result (rule number) and + * its priority for each category. + */ +static inline void +resolve_priority_neon(uint64_t transition, int n, const struct rte_acl_ctx *ctx, + struct parms *parms, + const struct rte_acl_match_results *p, + uint32_t categories) +{ + uint32_t x; + int32x4_t results, priority, results1, priority1; + uint32x4_t selector; + int32_t *saved_results, *saved_priority; + + for (x = 0; x < categories; x += RTE_ACL_RESULTS_MULTIPLIER) { + saved_results = (int32_t *)(&parms[n].cmplt->results[x]); + saved_priority = (int32_t *)(&parms[n].cmplt->priority[x]); + + /* get results and priorities for completed trie */ + results = vld1q_s32( + (const int32_t *)&p[transition].results[x]); + priority = vld1q_s32( + (const int32_t *)&p[transition].priority[x]); + + /* if this is not the first completed trie */ + if (parms[n].cmplt->count != ctx->num_tries) { + /* get running best results and their priorities */ + results1 = vld1q_s32(saved_results); + priority1 = vld1q_s32(saved_priority); + + /* select results that are highest priority */ + selector = vcgtq_s32(priority1, priority); + results = vbslq_s32(selector, results1, results); + priority = vbslq_s32(selector, priority1, priority); + } + + /* save running best results and their priorities */ + vst1q_s32(saved_results, results); + vst1q_s32(saved_priority, priority); + } +} + +/* + * Check for any match in 4 transitions + */ +static inline __attribute__((always_inline)) uint32_t +check_any_match_x4(uint64_t val[]) +{ + return ((val[0] | val[1] | val[2] | val[3]) & RTE_ACL_NODE_MATCH); +} + +static inline __attribute__((always_inline)) void +acl_match_check_x4(int slot, const struct rte_acl_ctx *ctx, struct parms *parms, + struct acl_flow_data *flows, uint64_t transitions[]) +{ + while (check_any_match_x4(transitions)) { + transitions[0] = acl_match_check(transitions[0], slot, ctx, + parms, flows, resolve_priority_neon); + transitions[1] = acl_match_check(transitions[1], slot + 1, ctx, + parms, flows, resolve_priority_neon); + transitions[2] = acl_match_check(transitions[2], slot + 2, ctx, + parms, flows, resolve_priority_neon); + transitions[3] = acl_match_check(transitions[3], slot + 3, ctx, + parms, flows, resolve_priority_neon); + } +} + +/* + * Process 4 transitions (in 2 NEON Q registers) in parallel + */ +static inline __attribute__((always_inline)) int32x4_t +transition4(int32x4_t next_input, const uint64_t *trans, uint64_t transitions[]) +{ + int32x4x2_t tr_hi_lo; + int32x4_t t, in, r; + uint32x4_t index_msk, node_type, addr; + uint32x4_t dfa_msk, mask, quad_ofs, dfa_ofs; + + /* Move low 32 into tr_hi_lo.val[0] and high 32 into tr_hi_lo.val[1] */ + tr_hi_lo = vld2q_s32((const int32_t *)transitions); + + /* Calculate the address (array index) for all 4 transitions. */ + + index_msk = vld1q_u32((const uint32_t *)&neon_acl_const.xmm_index_mask); + + /* Calc node type and node addr */ + node_type = vbicq_s32(tr_hi_lo.val[0], index_msk); + addr = vandq_s32(tr_hi_lo.val[0], index_msk); + + /* t = 0 */ + t = veorq_s32(node_type, node_type); + + /* mask for DFA type(0) nodes */ + dfa_msk = vceqq_u32(node_type, t); + + mask = vld1q_s32((const int32_t *)&neon_acl_const.xmm_shuffle_input); + in = vqtbl1q_u8((uint8x16_t)next_input, (uint8x16_t)mask); + + /* DFA calculations. */ + r = vshrq_n_u32(in, 30); /* div by 64 */ + mask = vld1q_s32((const int32_t *)&neon_acl_const.range_base); + r = vaddq_u8(r, mask); + t = vshrq_n_u32(in, 24); + r = vqtbl1q_u8((uint8x16_t)tr_hi_lo.val[1], (uint8x16_t)r); + dfa_ofs = vsubq_s32(t, r); + + /* QUAD/SINGLE calculations. */ + t = vcgtq_s8(in, tr_hi_lo.val[1]); + t = vabsq_s8(t); + t = vpaddlq_u8(t); + quad_ofs = vpaddlq_u16(t); + + /* blend DFA and QUAD/SINGLE. */ + t = vbslq_u8(dfa_msk, dfa_ofs, quad_ofs); + + /* calculate address for next transitions */ + addr = vaddq_u32(addr, t); + + /* Fill next transitions */ + transitions[0] = trans[vgetq_lane_u32(addr, 0)]; + transitions[1] = trans[vgetq_lane_u32(addr, 1)]; + transitions[2] = trans[vgetq_lane_u32(addr, 2)]; + transitions[3] = trans[vgetq_lane_u32(addr, 3)]; + + return vshrq_n_u32(next_input, CHAR_BIT); +} + +/* + * Execute trie traversal with 8 traversals in parallel + */ +static inline int +search_neon_8(const struct rte_acl_ctx *ctx, const uint8_t **data, + uint32_t *results, uint32_t total_packets, uint32_t categories) +{ + int n; + struct acl_flow_data flows; + uint64_t index_array[8]; + struct completion cmplt[8]; + struct parms parms[8]; + int32x4_t input0, input1; + + acl_set_flow(&flows, cmplt, RTE_DIM(cmplt), data, results, + total_packets, categories, ctx->trans_table); + + for (n = 0; n < 8; n++) { + cmplt[n].count = 0; + index_array[n] = acl_start_next_trie(&flows, parms, n, ctx); + } + + /* Check for any matches. */ + acl_match_check_x4(0, ctx, parms, &flows, &index_array[0]); + acl_match_check_x4(4, ctx, parms, &flows, &index_array[4]); + + while (flows.started > 0) { + /* Gather 4 bytes of input data for each stream. */ + input0 = vsetq_lane_s32(GET_NEXT_4BYTES(parms, 0), input0, 0); + input1 = vsetq_lane_s32(GET_NEXT_4BYTES(parms, 4), input1, 0); + + input0 = vsetq_lane_s32(GET_NEXT_4BYTES(parms, 1), input0, 1); + input1 = vsetq_lane_s32(GET_NEXT_4BYTES(parms, 5), input1, 1); + + input0 = vsetq_lane_s32(GET_NEXT_4BYTES(parms, 2), input0, 2); + input1 = vsetq_lane_s32(GET_NEXT_4BYTES(parms, 6), input1, 2); + + input0 = vsetq_lane_s32(GET_NEXT_4BYTES(parms, 3), input0, 3); + input1 = vsetq_lane_s32(GET_NEXT_4BYTES(parms, 7), input1, 3); + + /* Process the 4 bytes of input on each stream. */ + + input0 = transition4(input0, flows.trans, &index_array[0]); + input1 = transition4(input1, flows.trans, &index_array[4]); + + input0 = transition4(input0, flows.trans, &index_array[0]); + input1 = transition4(input1, flows.trans, &index_array[4]); + + input0 = transition4(input0, flows.trans, &index_array[0]); + input1 = transition4(input1, flows.trans, &index_array[4]); + + input0 = transition4(input0, flows.trans, &index_array[0]); + input1 = transition4(input1, flows.trans, &index_array[4]); + + /* Check for any matches. */ + acl_match_check_x4(0, ctx, parms, &flows, &index_array[0]); + acl_match_check_x4(4, ctx, parms, &flows, &index_array[4]); + } + + return 0; +} + +/* + * Execute trie traversal with 4 traversals in parallel + */ +static inline int +search_neon_4(const struct rte_acl_ctx *ctx, const uint8_t **data, + uint32_t *results, int total_packets, uint32_t categories) +{ + int n; + struct acl_flow_data flows; + uint64_t index_array[4]; + struct completion cmplt[4]; + struct parms parms[4]; + int32x4_t input; + + acl_set_flow(&flows, cmplt, RTE_DIM(cmplt), data, results, + total_packets, categories, ctx->trans_table); + + for (n = 0; n < 4; n++) { + cmplt[n].count = 0; + index_array[n] = acl_start_next_trie(&flows, parms, n, ctx); + } + + /* Check for any matches. */ + acl_match_check_x4(0, ctx, parms, &flows, index_array); + + while (flows.started > 0) { + /* Gather 4 bytes of input data for each stream. */ + input = vsetq_lane_s32(GET_NEXT_4BYTES(parms, 0), input, 0); + input = vsetq_lane_s32(GET_NEXT_4BYTES(parms, 1), input, 1); + input = vsetq_lane_s32(GET_NEXT_4BYTES(parms, 2), input, 2); + input = vsetq_lane_s32(GET_NEXT_4BYTES(parms, 3), input, 3); + + /* Process the 4 bytes of input on each stream. */ + input = transition4(input, flows.trans, index_array); + input = transition4(input, flows.trans, index_array); + input = transition4(input, flows.trans, index_array); + input = transition4(input, flows.trans, index_array); + + /* Check for any matches. */ + acl_match_check_x4(0, ctx, parms, &flows, index_array); + } + + return 0; +} diff --git a/lib/librte_acl/rte_acl.c b/lib/librte_acl/rte_acl.c index d60219f..e2fdebd 100644 --- a/lib/librte_acl/rte_acl.c +++ b/lib/librte_acl/rte_acl.c @@ -55,11 +55,32 @@ rte_acl_classify_avx2(__rte_unused const struct rte_acl_ctx *ctx, return -ENOTSUP; } +int __attribute__ ((weak)) +rte_acl_classify_sse(__rte_unused const struct rte_acl_ctx *ctx, + __rte_unused const uint8_t **data, + __rte_unused uint32_t *results, + __rte_unused uint32_t num, + __rte_unused uint32_t categories) +{ + return -ENOTSUP; +} + +int __attribute__ ((weak)) +rte_acl_classify_neon(__rte_unused const struct rte_acl_ctx *ctx, + __rte_unused const uint8_t **data, + __rte_unused uint32_t *results, + __rte_unused uint32_t num, + __rte_unused uint32_t categories) +{ + return -ENOTSUP; +} + static const rte_acl_classify_t classify_fns[] = { [RTE_ACL_CLASSIFY_DEFAULT] = rte_acl_classify_scalar, [RTE_ACL_CLASSIFY_SCALAR] = rte_acl_classify_scalar, [RTE_ACL_CLASSIFY_SSE] = rte_acl_classify_sse, [RTE_ACL_CLASSIFY_AVX2] = rte_acl_classify_avx2, + [RTE_ACL_CLASSIFY_NEON] = rte_acl_classify_neon, }; /* by default, use always available scalar code path. */ @@ -93,6 +114,9 @@ rte_acl_init(void) { enum rte_acl_classify_alg alg = RTE_ACL_CLASSIFY_DEFAULT; +#ifdef RTE_ARCH_ARM64 + alg = RTE_ACL_CLASSIFY_NEON; +#else #ifdef CC_AVX2_SUPPORT if (rte_cpu_get_flag_enabled(RTE_CPUFLAG_AVX2)) alg = RTE_ACL_CLASSIFY_AVX2; @@ -102,6 +126,7 @@ rte_acl_init(void) #endif alg = RTE_ACL_CLASSIFY_SSE; +#endif rte_acl_set_default_classify(alg); } diff --git a/lib/librte_acl/rte_acl.h b/lib/librte_acl/rte_acl.h index 98ef2fc..0979a09 100644 --- a/lib/librte_acl/rte_acl.h +++ b/lib/librte_acl/rte_acl.h @@ -270,6 +270,7 @@ enum rte_acl_classify_alg { RTE_ACL_CLASSIFY_SCALAR = 1, /**< generic implementation. */ RTE_ACL_CLASSIFY_SSE = 2, /**< requires SSE4.1 support. */ RTE_ACL_CLASSIFY_AVX2 = 3, /**< requires AVX2 support. */ + RTE_ACL_CLASSIFY_NEON = 4, /**< requires NEON support. */ RTE_ACL_CLASSIFY_NUM /* should always be the last one. */ };