From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from dpdk.org (dpdk.org [92.243.14.124]) by inbox.dpdk.org (Postfix) with ESMTP id 56AC5A034F; Wed, 6 May 2020 17:24:16 +0200 (CEST) Received: from [92.243.14.124] (localhost [127.0.0.1]) by dpdk.org (Postfix) with ESMTP id CE05D1D910; Wed, 6 May 2020 17:24:15 +0200 (CEST) Received: from dispatch1-us1.ppe-hosted.com (dispatch1-us1.ppe-hosted.com [148.163.129.52]) by dpdk.org (Postfix) with ESMTP id 0FD141D904 for ; Wed, 6 May 2020 17:24:14 +0200 (CEST) Received: from mx1-us1.ppe-hosted.com (unknown [10.7.65.62]) by dispatch1-us1.ppe-hosted.com (PPE Hosted ESMTP Server) with ESMTP id 5125D60124; Wed, 6 May 2020 15:24:13 +0000 (UTC) Received: from us4-mdac16-65.ut7.mdlocal (unknown [10.7.66.64]) by mx1-us1.ppe-hosted.com (PPE Hosted ESMTP Server) with ESMTP id 4F36A8009B; Wed, 6 May 2020 15:24:13 +0000 (UTC) X-Virus-Scanned: Proofpoint Essentials engine Received: from mx1-us1.ppe-hosted.com (unknown [10.7.65.174]) by mx1-us1.ppe-hosted.com (PPE Hosted ESMTP Server) with ESMTPS id 52B04280085; Wed, 6 May 2020 15:24:12 +0000 (UTC) Received: from webmail.solarflare.com (uk.solarflare.com [193.34.186.16]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-SHA384 (256/256 bits)) (No client certificate requested) by mx1-us1.ppe-hosted.com (PPE Hosted ESMTP Server) with ESMTPS id 74D931C0094; Wed, 6 May 2020 15:24:11 +0000 (UTC) Received: from [192.168.38.17] (10.17.10.39) by ukex01.SolarFlarecom.com (10.17.10.4) with Microsoft SMTP Server (TLS) id 15.0.1395.4; Wed, 6 May 2020 16:24:02 +0100 To: Wisam Jaddo , , , , , , , References: <20200430103341.1845-2-wisamm@mellanox.com> <20200506123627.22340-1-wisamm@mellanox.com> <20200506123627.22340-3-wisamm@mellanox.com> From: Andrew Rybchenko Autocrypt: addr=arybchenko@solarflare.com; keydata= mQINBF2681gBEACbdTxu8eLL3UX2oAelsnK9GkeaJeUYSOHPJQpV7RL/iaIskqTwBRnhjXt7 j9UEwGA+omnOmqQMpeQTb/F9Ma2dYE+Hw4/t/1KVjxr3ehFaASvwR4fWJfO4e2l/Rk4rG6Yi 5r6CWU2y8su2654Fr8KFc+cMGOAgKoZTZHZsRy5lHpMlemeF+VZkv8L5sYJWPnsypgqlCG3h v6lbtfZs+QqYbFH6bqoZwBAl5irmxywGR7ZJr1GLUZZ1lfdazSY8r6Vz0/Ip/KVxGu2uxo81 QCsAj0ZsQtwji9Sds/prTiPrIjx8Fc/tfbnAuVuPcnPbczwCJACzQr4q26XATL3kVuZhSBWh 4XfO/EAUuEq5AemUG5DDTM87g7Lp4eT9gMZB6P+rJwWPNWTiV3L7Cn+fO+l9mTPnOqdzBgDe OaulKiNSft1o0DY4bGzOmM2ad2cZt0jfnbMPMTE9zsr6+RFa+M8Ct20o6U1MUE4vP6veErMK of4kZ8PdoMM+Sq1hxMPNtlcVBSP9xMmdSZPlfDYI5VWosOceEcz7XZdjBJKdwKuz70V7eac4 ITSxgNFCTbeJ03zL2MR5s0IvD9ghISAwZ6ieCjU5UATn5+63qpD0nVNLsAdb/UpfvQcKAmvj 0fKlxu/PMVkjBa7/4cfNogYOhWDKUO+1pMaFwvb6/XTo6uMpfQARAQABtCxBbmRyZXcgUnli Y2hlbmtvIDxhcnliY2hlbmtvQHNvbGFyZmxhcmUuY29tPokCVAQTAQoAPhYhBP6NPgcKRj/Y X0yXQahue0sAy4m+BQJduvNYAhsDBQkB4TOABQsJCAcDBRUKCQgLBRYCAwEAAh4BAheAAAoJ EKhue0sAy4m+t3gP/j1MNc63CEozZo1IZ2UpVPAVWTYbLdPjIRdFqhlwvZYIgGIgIBk3ezKL K0/oc4ZeIwL6wQ5+V24ahuXvvcxLlKxfbJ6lo2iQGC7GLGhsDG9Y2k6sW13/sTJB/XuR2yov k5FtIgJ+aHa1PDZnepnGGOt9ka9n/Jzrc9WKYapOIIyLRe9U26ikoVgyqsD37PVeq5tLWHHA NGTUKupe9G6DFWidxx0KzyMoWDTbW2AWYcEmV2eQsgRT094AZwLFN5ErfefYzsGdO8TAUU9X YTiQN2MvP1pBxY/r0/5UfwV4UKBcR0S3ZvzyvrPoYER2Kxdf/qurx0Mn7StiCQ/JlNZb/GWQ TQ7huduuZHNQKWm7ufbqvKSfbPYvfl3akj7Wl8/zXhYdLqb5mmK45HXrgYGEqPN53OnK2Ngx IgYKEWr05KNv09097jLT5ONgYvszflqlLIzC4dV245g7ucuf9fYmsvmM1p/gFnOJBJL18YE5 P1fuGYNfLP+qp4WMiDqXlzaJfB4JcinyU49BXUj3Utd6f6sNBsO8YWcLbKBV9WmA324S3+wj f4NPRp3A5E+6OmTVMLWire2ZvnYp3YvifUj1r8lhoZ2B2vKuWwiTlHOKYBEjnOQJQnqYZEF0 JQQ1xzVDBQKE01BPlA3vy6BGWe6I4psBVqMOB9lAev/H+xa4u6Z3uQINBF269JsBEAC2KB3W 8JES/fh74avN7LOSdK4QA7gFIUQ4egVL81KnxquLzzilABuOhmZf3Rq6rMHSM8xmUAWa7Dkt YtzXStjEBI/uF0mAR3mMz1RcL2Wp+WD/15HjVpA7hPjXSEsWY0K2ymPerK4yrLcfFTHdMonY JfuACCC9NtOZxrWHOJoUS+RT7AWk80q/6D2iwQ47/2dBTznVG+gSeHSes9l91TB09w6f9JX/ sT+Ud0NQfm7HJ7t2pmGI9O6Po/NLZsDogmnIpJp/WwYOZN9JK7u2FyX2UyRzR8jK42aJkRsh DXs16Cc2/eYGakjrdO3x9a+RoxN7EuFtYhGR1PzMXdUiB5i+FyddYXkYUyO43QE/3VPA5l1v TUOagzZq6aONsdNonGJkV3TIG3JmUNtM+D/+r6QKzmgoJ8w576JxEZI09I/ZFN+g7BnUmlMx 6Z3IUOXVX/SWfGFga0YajwajHz03IBhChEbYbbqndVhmshu2GFURxrfUPYWdDXEqkh+08a5U Didia9jm2Opv4oE1e1TXAePyYJl/Zyps4Cv00GObAxibvMBQCUZQ+IBnNldRBOwXXRQV2xpx P+9iO1VYA/QXn0KqRK+SH1JGRXbJYi42YFaW1gE0EU0fiR2Wb9pK+doNEjjOhlzUGuvOEAUS +4m0m3dlfEvpCV9GMr7ERRpZzh9QkQARAQABiQI8BBgBCgAmFiEE/o0+BwpGP9hfTJdBqG57 SwDLib4FAl269JsCGwwFCQlmAYAACgkQqG57SwDLib7x6g//e+eCtNnJz7qFGbjWRJYNLCe5 gQwkhdyEGk4omr3VmjGj3z9kNFy/muh4pmHUngSAnnpwZggx14N4hhKf9y8G4Dwvsqa6b1zB Jq/c4t/SBDtGW4M/E331N04PaQZpcrbTfp1KqHNknk2N7yOk4CcoLVuIZmA5tPguASV8aAfz ZwhWAwn6vUEw9552eXEAnGFGDTCbyryNwzB5jtVQOEEDjTxcCkpcXMB45Tb1QUslRTu/sBAe HhPCQSUcJHR+KOq+P6yKICGAr291PZd6Qc7C3UyE+A3pY/UfdEVWj0STBWx1qvYLaHLrI4O9 KXDgh7luLjZZafcueCaPYmNo4V2lmNb3+7S4TvqhoZS+wN+9ldRQ4gH3wmRZybN6Y/ZCqxol RaZpE3AqdWsGvIgAkD0FpmtZNii9s2pnrhw0K6S4t4tYgXGTossxNSJUltfFQZdXM1xkZhtv dBZuUEectbZWuviGvQXahOMuH2pM64mx2hpdZzPcI2beeJNHkAsGT2KcaMETgvtHUBFRlLVB YxsUYz3UZmi2JSua4tbcGd6iWVN90eb8CxszYtivfpz6o2nPSjNwg0NaVGSHXjAK0tdByZ9t SkwjC3tEPljVycRSDpbauogOiAkvjENfaPd/H26V5hY822kaclaKDAW6ZG9UKiMijcAgb9u5 CJoOyqE8aGS5Ag0EXbr1RwEQAMXZHbafqmZiu6Kudp+Filgdkj2/XJva5Elv3fLfpXvhVt0Y if5Rzds3RpffoLQZk9nPwK8TbZFqNXPu7HSgg9AY7UdCM94WRFTkUCGKzbgiqGdXZ7Vyc8cy teGW+BcdfQycDvjfy50T3fO4kJNVp2LDNdknPaZVe8HJ80Od63+9ksB6Ni+EijMkh6Uk3ulB CSLnT4iFV57KgU2IsxOQVLnm+0bcsWMcCnGfphkY0yKP+aJ6MfmZkEeaDa7kf24N14ktg50m vOGDitcxA/+XXQXOsOIDJx1VeidxYsQ2FfsKu1G8+G6ejuaLf4rV5MI/+B/tfLbbOdikM5PF pxZVgTir9q13qHumMxdme7w5c7hybW412yWAe9TsrlXktFmFjRSFzAAxQhQSQxArS6db4oBk yeYJ59mW52i4occkimPWSm/raSgdSM+0P6zdWUlxxj+r1qiLgCYvruzLNtp5Nts5tR/HRQjE /ohQYaWDSVJEsc/4eGmgwzHzmvHtXeKkasn01381A1Lv3xwtpnfwERMAhxBZ8EGKEkc5gNdk vIPhknnGgPXqKmE1aWu8LcHiY+RHAF8gYPCDMuwyzBYnbiosKcicuIUp0Fj8XIaPao6F+WTi In4UOrqrYhsaCUvhVjsTBbNphGih9xbFJ8E+lkTLL8P3umtTcMPnpsB4xqcDABEBAAGJBHIE GAEKACYWIQT+jT4HCkY/2F9Ml0GobntLAMuJvgUCXbr1RwIbAgUJCWYBgAJACRCobntLAMuJ vsF0IAQZAQoAHRYhBNTYjdjWgdaEN5MrAN+9UR5r/4d3BQJduvVHAAoJEN+9UR5r/4d3EiQP /3lyby6v49HTU94Q2Fn2Xat6uifR7kWE5SO/1pUwYzx6v+z5K2jqPgqUYmuNoejcGl0CTNhg LbsxzUmAuf1OTAdE+ZYvOAjjKQhY4haxHc4enby/ltnHfWJYWJZ9UN5SsIQLvITvYu6rqthO CYjpXJhwkj3ODmC9H1TrvjrBGc6i7CTnR8RCjMEwCs2LI2frHa4R6imViEr9ScMfUnzdABMQ B0T5MOg8NX92/FRjTldU2KovG0ML9mSveSvVHAoEBLy4UIs5nEDdNiO1opJgKb5CXvWQugub 7AR52phNdKVdEB0S4tigJT4NalyTaPiUhFEm+CzZpMQDJ5E+/OowaPRfN4HeJX+c8sB+vUAZ mkAaG75N+IEk5JKFK9Z+bBYgPgaBDFZYdWDB/TMH0ANt+KI5uYg0i12TB4M8pwKG1DEPUmWc F2YpvB3jnbwzsOpSFiJOOlSs6nOB0Sb5GRtPOO3h6XGj+6mzQd6tcL63c9TrrUkjq7LDkxCz SJ2hTYRC8WNX8Uw9skWo5728JNrXdazEYCenUWmYiKLNKLslXCFodUCRDh/sUiyqRwS7PHEA LYC/UIWLMomI0Yvju3KA5v3RQVXhL+Gx2CzSj3GDz9xxGhJB2LfRfjzPbTR/Z27UpjCkd8z0 Ro3Ypmi1FLQwnRgoOKDbetTAIhugEShaLTITzJAP/iRDJCQsrZah5tE8oIl81qKEmBJEGcdt HYikbpQe7ydcXhqTj7+IECa3O7azI5OhCxUH2jNyonJ/phUslHH2G1TTBZK8y4Hrx5RpuRNS esn3P9uKu9DHqBAL7DMsCPwb2p1VNnapD72DBmRhzS/e6zS2R4+r9yNv03Hv7VCxKkmtE63H qpS//qpjfrtsIcHAjnKDaDtL1LYCtHoweI+DOpKKULSAYp/JE6F8LNibPQ0/P3S5ZIJNC4QZ uESjFOalJwFIqGQdkQB7ltRNJENLrHc+2jKGOuyFHm/Sbvp5EMGdaeQ0+u8CY0P+y6oXenwx 7WrJz/GvbNoFhJoJ6RzxCMQrFgxrssVZ7w5HcUj94lbnJ6osdYE/WpSd50B6jet6LKh5revg u9XI9CoqsPQ1V4wKYYdllPuogCye7KNYNKuiiuSNpaF4gHq1ZWGArwZtWHjgc2v3LegOpRQF SwOskMKmWsUyHIRMG1p8RpkBQTqY2rGSeUqPSvaqjT0nq+SUEM6qxEXD/2Wqri/X6bamuPDb S0PkBvFD2+0zr5Bc2YkMGPBYPNGZiTp3UjmZlLfn3TiBKIC92jherY563CULjSsiBEJCOSvv 4VPLn5aAcfbCXJnE3IGCp/hPl50iQqu7BPOYBbWXeb9ptDjGCAThNxSz0WAXkmcjAFE8gdE6 Znk9 CC: Stephen Hemminger , Message-ID: Date: Wed, 6 May 2020 18:23:58 +0300 User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101 Thunderbird/68.7.0 MIME-Version: 1.0 In-Reply-To: <20200506123627.22340-3-wisamm@mellanox.com> Content-Type: text/plain; charset="utf-8" Content-Language: en-US Content-Transfer-Encoding: 7bit X-Originating-IP: [10.17.10.39] X-ClientProxiedBy: ocex03.SolarFlarecom.com (10.20.40.36) To ukex01.SolarFlarecom.com (10.17.10.4) X-TM-AS-Product-Ver: SMEX-12.5.0.1300-8.5.1020-25402.001 X-TM-AS-Result: No-22.857500-8.000000-10 X-TMASE-MatchedRID: IeZYkn8zfFrRbY5xvhIMEH4neC0h7SAD5CSGAyjxe9/fBejl4AK2sZj2 D83PVPcpnXHMFZLyU3fnTSz7KvF5qsOQecnAhEpbv1vPAPz22FlRvgR0hkbG4OTWTJAOkSPwg/5 gNr8OG72S/UW1f1LbKQOzZM/MayD94X8HWMSNX7o1VHP4fCovgtLi4HRGFLBZy5JfHvVu9IuEve sJDpTT5GzTlV5U3k20TAQ4HXbD0o1zGfR/zlVmWqbtlVLKF4zpj87/LK+2sqPieGQLRIaTJDEiN rkVm9a8B+X4plgLh5sTKs060FUxYRy/DZ8tz1CLPwKTD1v8YV5MkOX0UoduuZGPHiE2kiT4JWTQ sTqHabyyf5C/G25MxcOfkwJIS8CpPPoJUXE2kSZGjDwRSl8gFFZaRojBPvj6a0TOsL14A2k0NSC 223LgVf9cYufMqxyubErGlCjYMMXmJW5VyRS67nYZxYoZm58F6qG5M9QNAO2MUViaYYbK3FWh9a RY+Gu1mcUZ3An8FZHzDYaiaqURtBWVVkGuW6Jrh2VzUlo4HVNc1jwHBugxQEcq9Sq5BrsynoiqT apU5Zq/bQ5VOF8E6+HhTaBhYIHgMgpiuaPdDzmcnm0v4tsY43EONw67gSVc9J5avMd3ezAf6aBS hjcUYKyNAXivXV0YnaoElBh3N4jD6egsHKBhON1+No0HEURtAp+UH372RZVTq//rncsNWUSM1Cj LAhUsaz87imoHjqOHpdIJtbLnJu6uy19UVDsHGG99bHRMT9u8xE2H2EuMWcNWm4M1rA2moqRTS3 ju9kgwyVZUEw+R8+MnX/BNijnqFC1+JwI1dIeeAiCmPx4NwLTrdaH1ZWqC1B0Hk1Q1KyLgfCfWl nNb/1cppCzPq+1UkGUtrowrXLg= X-TM-AS-User-Approved-Sender: Yes X-TM-AS-User-Blocked-Sender: No X-TMASE-Result: 10--22.857500-8.000000 X-TMASE-Version: SMEX-12.5.0.1300-8.5.1020-25402.001 X-MDID: 1588778652-Z4x5fAJ-bGEC Subject: Re: [dpdk-dev] [PATCH v5 2/5] app/flow-perf: add insertion rate calculation X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: DPDK patches and discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: dev-bounces@dpdk.org Sender: "dev" My biggest concert with the patch is usage of huge number of global variables which makes the code hard to read, understand and maintain. See my notes below. Please, share your thoughts. On 5/6/20 3:36 PM, Wisam Jaddo wrote: > Add insertion rate calculation feature into flow > performance application. > > The application now provide the ability to test > insertion rate of specific rte_flow rule, by > stressing it to the NIC, and calculate the > insertion rate. > > The application offers some options in the command > line, to configure which rule to apply. > > After that the application will start producing > rules with same pattern but increasing the outer IP > source address by 1 each time, thus it will give > different flow each time, and all other items will > have open masks. > > The current design have single core insertion rate. > In the future we may have a multi core insertion > rate measurement support in the app. > > Signed-off-by: Wisam Jaddo > --- [snip] > diff --git a/app/test-flow-perf/actions_gen.c b/app/test-flow-perf/actions_gen.c > new file mode 100644 > index 000000000..fa60084cf > --- /dev/null > +++ b/app/test-flow-perf/actions_gen.c > @@ -0,0 +1,88 @@ > +/* SPDX-License-Identifier: BSD-3-Clause > + * Copyright 2020 Mellanox Technologies, Ltd > + * > + * The file contains the implementations of actions generators. > + * Each generator is responsible for preparing it's action instance > + * and initializing it with needed data. > + **/ > + > +#include > +#include > +#include > +#include > + > +#define ALLOCATE_ACTION_VARS > +#include "actions_gen.h" > +#include "config.h" > + > +void > +gen_mark(void) > +{ > + mark_action.id = MARK_ID; > +} > + > +void > +gen_queue(uint16_t queue) > +{ > + queue_action.index = queue; > +} > + > +void > +gen_jump(uint16_t next_table) > +{ > + jump_action.group = next_table; > +} > + > +void > +gen_rss(uint16_t *queues, uint16_t queues_number) > +{ > + uint16_t queue; > + struct action_rss_data *rss_data; > + > + rss_data = rte_malloc("rss_data", > + sizeof(struct action_rss_data), 0); > + > + if (rss_data == NULL) > + rte_exit(EXIT_FAILURE, "No Memory available!"); > + > + *rss_data = (struct action_rss_data){ > + .conf = (struct rte_flow_action_rss){ > + .func = RTE_ETH_HASH_FUNCTION_DEFAULT, > + .level = 0, > + .types = ETH_RSS_IP | > + ETH_RSS_TCP, Duplicating it in two places suggests addition of a macro with the value and usage in two places. > + .key_len = sizeof(rss_data->key), > + .queue_num = queues_number, > + .key = rss_data->key, > + .queue = rss_data->queue, > + }, > + .key = { 1 }, > + .queue = { 0 }, > + }; > + > + for (queue = 0; queue < queues_number; queue++) > + rss_data->queue[queue] = queues[queue]; > + > + rss_action = &rss_data->conf; > +} > + > +void > +gen_set_meta(void) > +{ > + meta_action.data = RTE_BE32(META_DATA); > + meta_action.mask = RTE_BE32(0xffffffff); > +} > + > +void > +gen_set_tag(void) > +{ > + tag_action.data = RTE_BE32(META_DATA); > + tag_action.mask = RTE_BE32(0xffffffff); > + tag_action.index = TAG_INDEX; > +} > + > +void > +gen_port_id(void) > +{ > + port_id.id = PORT_ID_DST; > +} > [snip] > diff --git a/app/test-flow-perf/flow_gen.c b/app/test-flow-perf/flow_gen.c > new file mode 100644 > index 000000000..cf5453586 > --- /dev/null > +++ b/app/test-flow-perf/flow_gen.c > @@ -0,0 +1,179 @@ > +/* SPDX-License-Identifier: BSD-3-Clause > + * Copyright 2020 Mellanox Technologies, Ltd > + * > + * The file contains the implementations of the method to > + * fill items, actions & attributes in their corresponding > + * arrays, and then generate rte_flow rule. > + * > + * After the generation. The rule goes to validation then > + * creation state and then return the results. > + */ > + > +#include > + > +#include "flow_gen.h" > +#include "items_gen.h" > +#include "actions_gen.h" > +#include "config.h" > + > +static void > +fill_attributes(struct rte_flow_attr *attr, > + uint8_t flow_attrs, uint16_t group) > +{ > + if (flow_attrs & INGRESS) > + attr->ingress = 1; > + if (flow_attrs & EGRESS) > + attr->egress = 1; > + if (flow_attrs & TRANSFER) > + attr->transfer = 1; > + attr->group = group; > +} > + > +static void > +fill_items(struct rte_flow_item items[MAX_ITEMS_NUM], > + uint16_t flow_items, uint32_t outer_ip_src) > +{ > + uint8_t items_counter = 0; > + > + if (flow_items & META_ITEM) > + add_meta_data(items, items_counter++); > + if (flow_items & TAG_ITEM) > + add_meta_tag(items, items_counter++); > + if (flow_items & ETH_ITEM) > + add_ether(items, items_counter++); > + if (flow_items & VLAN_ITEM) > + add_vlan(items, items_counter++); > + if (flow_items & IPV4_ITEM) > + add_ipv4(items, items_counter++, outer_ip_src); > + if (flow_items & IPV6_ITEM) > + add_ipv6(items, items_counter++, outer_ip_src); > + if (flow_items & TCP_ITEM) > + add_tcp(items, items_counter++); > + if (flow_items & UDP_ITEM) > + add_udp(items, items_counter++); > + if (flow_items & VXLAN_ITEM) > + add_vxlan(items, items_counter++); > + if (flow_items & VXLAN_GPE_ITEM) > + add_vxlan_gpe(items, items_counter++); > + if (flow_items & GRE_ITEM) > + add_gre(items, items_counter++); > + if (flow_items & GENEVE_ITEM) > + add_geneve(items, items_counter++); > + if (flow_items & GTP_ITEM) > + add_gtp(items, items_counter++); > + > + items[items_counter].type = RTE_FLOW_ITEM_TYPE_END; > +} > + > +static void > +fill_actions(struct rte_flow_action actions[MAX_ACTIONS_NUM], > + uint16_t flow_actions, uint32_t counter, uint16_t next_table, > + uint16_t hairpinq) > +{ > + struct rte_flow_action_count count_action; > + uint8_t actions_counter = 0; > + uint16_t queues[RXQ_NUM]; > + uint16_t i; > + > + /* None-fate actions */ > + if (flow_actions & MARK_ACTION) { > + if (!counter) DPDK coding style says compare to 0 [1]. Many similar comparisons below. [1] https://doc.dpdk.org/guides/contributing/coding_style.html#null-pointers > + gen_mark(); > + actions[actions_counter].type = RTE_FLOW_ACTION_TYPE_MARK; > + actions[actions_counter++].conf = &mark_action; > + } > + if (flow_actions & COUNT_ACTION) { > + actions[actions_counter].type = RTE_FLOW_ACTION_TYPE_COUNT; > + actions[actions_counter++].conf = &count_action; > + } > + if (flow_actions & META_ACTION) { > + if (!counter) > + gen_set_meta(); > + actions[actions_counter].type = RTE_FLOW_ACTION_TYPE_SET_META; > + actions[actions_counter++].conf = &meta_action; > + } > + if (flow_actions & TAG_ACTION) { > + if (!counter) > + gen_set_tag(); > + actions[actions_counter].type = RTE_FLOW_ACTION_TYPE_SET_TAG; > + actions[actions_counter++].conf = &tag_action; > + } > + > + /* Fate actions */ > + if (flow_actions & QUEUE_ACTION) { > + gen_queue(counter % RXQ_NUM); > + actions[actions_counter].type = RTE_FLOW_ACTION_TYPE_QUEUE; > + actions[actions_counter++].conf = &queue_action; > + } > + if (flow_actions & RSS_ACTION) { > + if (!counter) { > + for (i = 0; i < RXQ_NUM; i++) > + queues[i] = i; > + gen_rss(queues, RXQ_NUM); > + } > + actions[actions_counter].type = RTE_FLOW_ACTION_TYPE_RSS; > + actions[actions_counter++].conf = rss_action; > + } > + if (flow_actions & JUMP_ACTION) { > + if (!counter) > + gen_jump(next_table); > + actions[actions_counter].type = RTE_FLOW_ACTION_TYPE_JUMP; > + actions[actions_counter++].conf = &jump_action; > + } > + if (flow_actions & PORT_ID_ACTION) { > + if (!counter) > + gen_port_id(); > + actions[actions_counter].type = RTE_FLOW_ACTION_TYPE_PORT_ID; > + actions[actions_counter++].conf = &port_id; > + } > + if (flow_actions & DROP_ACTION) > + actions[actions_counter++].type = RTE_FLOW_ACTION_TYPE_DROP; > + if (flow_actions & HAIRPIN_QUEUE_ACTION) { > + gen_queue((counter % hairpinq) + RXQ_NUM); > + actions[actions_counter].type = RTE_FLOW_ACTION_TYPE_QUEUE; > + actions[actions_counter++].conf = &queue_action; > + } > + if (flow_actions & HAIRPIN_RSS_ACTION) { > + if (!counter) { > + uint16_t hairpin_queues[hairpinq]; > + for (i = 0; i < hairpinq; i++) > + hairpin_queues[i] = i + RXQ_NUM; > + gen_rss(hairpin_queues, hairpinq); > + } > + actions[actions_counter].type = RTE_FLOW_ACTION_TYPE_RSS; > + actions[actions_counter++].conf = rss_action; > + } > + > + actions[actions_counter].type = RTE_FLOW_ACTION_TYPE_END; > +} > + > +struct rte_flow * > +generate_flow(uint16_t port_id, > + uint16_t group, > + uint8_t flow_attrs, > + uint16_t flow_items, > + uint16_t flow_actions, > + uint16_t next_table, > + uint32_t outer_ip_src, > + uint16_t hairpinq, > + struct rte_flow_error *error) > +{ > + struct rte_flow_attr attr; > + struct rte_flow_item items[MAX_ITEMS_NUM]; > + struct rte_flow_action actions[MAX_ACTIONS_NUM]; > + struct rte_flow *flow = NULL; > + > + memset(items, 0, sizeof(items)); > + memset(actions, 0, sizeof(actions)); > + memset(&attr, 0, sizeof(struct rte_flow_attr)); > + > + fill_attributes(&attr, flow_attrs, group); > + > + fill_actions(actions, flow_actions, > + outer_ip_src, next_table, hairpinq); > + > + fill_items(items, flow_items, outer_ip_src); > + > + flow = rte_flow_create(port_id, &attr, items, actions, error); > + return flow; > +} > diff --git a/app/test-flow-perf/flow_gen.h b/app/test-flow-perf/flow_gen.h > new file mode 100644 > index 000000000..43d9e7cfe > --- /dev/null > +++ b/app/test-flow-perf/flow_gen.h > @@ -0,0 +1,63 @@ > +/* SPDX-License-Identifier: BSD-3-Clause > + * Copyright 2020 Mellanox Technologies, Ltd > + * > + * This file contains the items, actions and attributes > + * definition. And the methods to prepare and fill items, > + * actions and attributes to generate rte_flow rule. > + */ > + > +#ifndef FLOW_PERF_FLOW_GEN > +#define FLOW_PERF_FLOW_GEN > + > +#include > +#include > + > +#include "config.h" > + > +#define FLOW_ITEM_MASK(_x) (UINT64_C(1) << _x) > + > +/* Items */ > +#define ETH_ITEM FLOW_ITEM_MASK(0) I don't understand why you don't use enum rte_flow_item_type members as offsets. In this case don't need these defines at all, just use FLOW_ITEM_MASK(ETH). It will make it easier to support new items. > +#define IPV4_ITEM FLOW_ITEM_MASK(1) > +#define IPV6_ITEM FLOW_ITEM_MASK(2) > +#define VLAN_ITEM FLOW_ITEM_MASK(3) > +#define TCP_ITEM FLOW_ITEM_MASK(4) > +#define UDP_ITEM FLOW_ITEM_MASK(5) > +#define VXLAN_ITEM FLOW_ITEM_MASK(6) > +#define VXLAN_GPE_ITEM FLOW_ITEM_MASK(7) > +#define GRE_ITEM FLOW_ITEM_MASK(8) > +#define GENEVE_ITEM FLOW_ITEM_MASK(9) > +#define GTP_ITEM FLOW_ITEM_MASK(10) > +#define META_ITEM FLOW_ITEM_MASK(11) > +#define TAG_ITEM FLOW_ITEM_MASK(12) > + > +/* Actions */ > +#define QUEUE_ACTION FLOW_ITEM_MASK(0) I don't understand why you don't use enum rte_flow_action_type members as offsets. > +#define MARK_ACTION FLOW_ITEM_MASK(1) > +#define JUMP_ACTION FLOW_ITEM_MASK(2) > +#define RSS_ACTION FLOW_ITEM_MASK(3) > +#define COUNT_ACTION FLOW_ITEM_MASK(4) > +#define META_ACTION FLOW_ITEM_MASK(5) > +#define TAG_ACTION FLOW_ITEM_MASK(6) > +#define DROP_ACTION FLOW_ITEM_MASK(7) > +#define PORT_ID_ACTION FLOW_ITEM_MASK(8) > +#define HAIRPIN_QUEUE_ACTION FLOW_ITEM_MASK(9) > +#define HAIRPIN_RSS_ACTION FLOW_ITEM_MASK(10) > + > +/* Attributes */ > +#define INGRESS FLOW_ITEM_MASK(0) > +#define EGRESS FLOW_ITEM_MASK(1) > +#define TRANSFER FLOW_ITEM_MASK(2) > + > +struct rte_flow * > +generate_flow(uint16_t port_id, > + uint16_t group, > + uint8_t flow_attrs, > + uint16_t flow_items, > + uint16_t flow_actions, > + uint16_t next_table, > + uint32_t outer_ip_src, > + uint16_t hairpinq, > + struct rte_flow_error *error); > + > +#endif /* FLOW_PERF_FLOW_GEN */ > diff --git a/app/test-flow-perf/items_gen.c b/app/test-flow-perf/items_gen.c > new file mode 100644 > index 000000000..1e9479fb8 > --- /dev/null > +++ b/app/test-flow-perf/items_gen.c > @@ -0,0 +1,265 @@ > +/* SPDX-License-Identifier: BSD-3-Clause > + * Copyright 2020 Mellanox Technologies, Ltd > + * > + * This file contain the implementations of the items > + * related methods. Each Item have a method to prepare > + * the item and add it into items array in given index. > + */ > + > +#include > +#include > + > +#include "items_gen.h" > +#include "config.h" > + > +/* Current design is single threaded. */ > +static struct rte_flow_item_eth eth_spec; > +static struct rte_flow_item_eth eth_mask; It looks like that the design has problems with: eth / ip4 / udp / vxlan /eth / end patterns. > +static struct rte_flow_item_vlan vlan_spec; > +static struct rte_flow_item_vlan vlan_mask; > +static struct rte_flow_item_ipv4 ipv4_spec; > +static struct rte_flow_item_ipv4 ipv4_mask; > +static struct rte_flow_item_ipv6 ipv6_spec; > +static struct rte_flow_item_ipv6 ipv6_mask; > +static struct rte_flow_item_udp udp_spec; > +static struct rte_flow_item_udp udp_mask; > +static struct rte_flow_item_tcp tcp_spec; > +static struct rte_flow_item_tcp tcp_mask; > +static struct rte_flow_item_vxlan vxlan_spec; > +static struct rte_flow_item_vxlan vxlan_mask; > +static struct rte_flow_item_vxlan_gpe vxlan_gpe_spec; > +static struct rte_flow_item_vxlan_gpe vxlan_gpe_mask; > +static struct rte_flow_item_gre gre_spec; > +static struct rte_flow_item_gre gre_mask; > +static struct rte_flow_item_geneve geneve_spec; > +static struct rte_flow_item_geneve geneve_mask; > +static struct rte_flow_item_gtp gtp_spec; > +static struct rte_flow_item_gtp gtp_mask; > +static struct rte_flow_item_meta meta_spec; > +static struct rte_flow_item_meta meta_mask; > +static struct rte_flow_item_tag tag_spec; > +static struct rte_flow_item_tag tag_mask; I think that such amount of global variables makes the code hard to read and maintain. If that's only me, no problem. > + > + > +void > +add_ether(struct rte_flow_item items[MAX_ITEMS_NUM], > + uint8_t items_counter) > +{ > + memset(ð_spec, 0, sizeof(struct rte_flow_item_eth)); > + memset(ð_mask, 0, sizeof(struct rte_flow_item_eth)); > + eth_spec.type = 0; > + eth_mask.type = 0; What's the point to set type to 0 if you just memset the entire structure? > + > + items[items_counter].type = RTE_FLOW_ITEM_TYPE_ETH; > + items[items_counter].spec = ð_spec; > + items[items_counter].mask = ð_mask; > +} > + > +void > +add_vlan(struct rte_flow_item items[MAX_ITEMS_NUM], > + uint8_t items_counter) > +{ > + uint16_t vlan_value = VLAN_VALUE; > + memset(&vlan_spec, 0, sizeof(struct rte_flow_item_vlan)); > + memset(&vlan_mask, 0, sizeof(struct rte_flow_item_vlan)); > + > + vlan_spec.tci = RTE_BE16(vlan_value); > + vlan_mask.tci = RTE_BE16(0xffff); > + > + items[items_counter].type = RTE_FLOW_ITEM_TYPE_VLAN; > + items[items_counter].spec = &vlan_spec; > + items[items_counter].mask = &vlan_mask; > +} > + > +void > +add_ipv4(struct rte_flow_item items[MAX_ITEMS_NUM], > + uint8_t items_counter, uint32_t src_ipv4) Shouldn't src_ipv4 be rte_be32? > +{ > + memset(&ipv4_spec, 0, sizeof(struct rte_flow_item_ipv4)); > + memset(&ipv4_mask, 0, sizeof(struct rte_flow_item_ipv4)); > + > + ipv4_spec.hdr.src_addr = src_ipv4; > + ipv4_mask.hdr.src_addr = 0xffffffff; RTE_BE32() is missing above (at least to be consistent with RTE_BE16() few lines above. > + > + items[items_counter].type = RTE_FLOW_ITEM_TYPE_IPV4; > + items[items_counter].spec = &ipv4_spec; > + items[items_counter].mask = &ipv4_mask; > +} > + > + > +void > +add_ipv6(struct rte_flow_item items[MAX_ITEMS_NUM], > + uint8_t items_counter, int src_ipv6) I think such specification of the source IPv6 address is very confusing. If you really need, it would be nice to explain why comments. > +{ > + memset(&ipv6_spec, 0, sizeof(struct rte_flow_item_ipv6)); > + memset(&ipv6_mask, 0, sizeof(struct rte_flow_item_ipv6)); > + > + /** Set ipv6 src **/ > + memset(&ipv6_spec.hdr.src_addr, src_ipv6, > + sizeof(ipv6_spec.hdr.src_addr) / 2); > + > + /** Full mask **/ > + memset(&ipv6_mask.hdr.src_addr, 1, > + sizeof(ipv6_spec.hdr.src_addr)); Are you that 1 is what you really want here? May be 0xff? > + > + items[items_counter].type = RTE_FLOW_ITEM_TYPE_IPV6; > + items[items_counter].spec = &ipv6_spec; > + items[items_counter].mask = &ipv6_mask; > +} > + > +void > +add_tcp(struct rte_flow_item items[MAX_ITEMS_NUM], > + uint8_t items_counter) > +{ > + memset(&tcp_spec, 0, sizeof(struct rte_flow_item_tcp)); > + memset(&tcp_mask, 0, sizeof(struct rte_flow_item_tcp)); > + > + items[items_counter].type = RTE_FLOW_ITEM_TYPE_TCP; > + items[items_counter].spec = &tcp_spec; > + items[items_counter].mask = &tcp_mask; > +} > + > +void > +add_udp(struct rte_flow_item items[MAX_ITEMS_NUM], > + uint8_t items_counter) > +{ > + memset(&udp_spec, 0, sizeof(struct rte_flow_item_udp)); > + memset(&udp_mask, 0, sizeof(struct rte_flow_item_udp)); > + > + items[items_counter].type = RTE_FLOW_ITEM_TYPE_UDP; > + items[items_counter].spec = &udp_spec; > + items[items_counter].mask = &udp_mask; > +} > + > +void > +add_vxlan(struct rte_flow_item items[MAX_ITEMS_NUM], > + uint8_t items_counter) > +{ > + uint32_t vni_value = VNI_VALUE; > + uint8_t i; > + memset(&vxlan_spec, 0, sizeof(struct rte_flow_item_vxlan)); > + memset(&vxlan_mask, 0, sizeof(struct rte_flow_item_vxlan)); > + > + /* Set standard vxlan vni */ > + for (i = 0; i < 3; i++) { > + vxlan_spec.vni[2 - i] = vni_value >> (i * 8); > + vxlan_mask.vni[2 - i] = 0xff; > + } > + > + /* Standard vxlan flags */ > + vxlan_spec.flags = 0x8; > + > + items[items_counter].type = RTE_FLOW_ITEM_TYPE_VXLAN; > + items[items_counter].spec = &vxlan_spec; > + items[items_counter].mask = &vxlan_mask; > +} > + > +void > +add_vxlan_gpe(struct rte_flow_item items[MAX_ITEMS_NUM], > + uint8_t items_counter) > +{ > + uint32_t vni_value = VNI_VALUE; > + uint8_t i; > + memset(&vxlan_gpe_spec, 0, sizeof(struct rte_flow_item_vxlan_gpe)); > + memset(&vxlan_gpe_mask, 0, sizeof(struct rte_flow_item_vxlan_gpe)); > + > + /* Set vxlan-gpe vni */ > + for (i = 0; i < 3; i++) { > + vxlan_gpe_spec.vni[2 - i] = vni_value >> (i * 8); > + vxlan_gpe_mask.vni[2 - i] = 0xff; > + } > + > + /* vxlan-gpe flags */ > + vxlan_gpe_spec.flags = 0x0c; > + > + items[items_counter].type = RTE_FLOW_ITEM_TYPE_VXLAN_GPE; > + items[items_counter].spec = &vxlan_gpe_spec; > + items[items_counter].mask = &vxlan_gpe_mask; > +} > + > +void > +add_gre(struct rte_flow_item items[MAX_ITEMS_NUM], > + uint8_t items_counter) > +{ > + uint16_t proto = GRE_PROTO; > + memset(&gre_spec, 0, sizeof(struct rte_flow_item_gre)); > + memset(&gre_mask, 0, sizeof(struct rte_flow_item_gre)); > + > + gre_spec.protocol = RTE_BE16(proto); > + gre_mask.protocol = 0xffff; RTE_BE16(0xffff) to be consistent > + > + items[items_counter].type = RTE_FLOW_ITEM_TYPE_GRE; > + items[items_counter].spec = &gre_spec; > + items[items_counter].mask = &gre_mask; > +} > + > +void > +add_geneve(struct rte_flow_item items[MAX_ITEMS_NUM], > + uint8_t items_counter) > +{ > + uint32_t vni_value = VNI_VALUE; > + uint8_t i; > + memset(&geneve_spec, 0, sizeof(struct rte_flow_item_geneve)); > + memset(&geneve_mask, 0, sizeof(struct rte_flow_item_geneve)); > + > + for (i = 0; i < 3; i++) { > + geneve_spec.vni[2 - i] = vni_value >> (i * 8); > + geneve_mask.vni[2 - i] = 0xff; > + } > + > + items[items_counter].type = RTE_FLOW_ITEM_TYPE_GENEVE; > + items[items_counter].spec = &geneve_spec; > + items[items_counter].mask = &geneve_mask; > +} > + > +void > +add_gtp(struct rte_flow_item items[MAX_ITEMS_NUM], > + uint8_t items_counter) > +{ > + uint32_t teid_value = TEID_VALUE; > + memset(>p_spec, 0, sizeof(struct rte_flow_item_gtp)); > + memset(>p_mask, 0, sizeof(struct rte_flow_item_gtp)); > + > + gtp_spec.teid = RTE_BE32(teid_value); > + gtp_mask.teid = RTE_BE32(0xffffffff); > + > + items[items_counter].type = RTE_FLOW_ITEM_TYPE_GTP; > + items[items_counter].spec = >p_spec; > + items[items_counter].mask = >p_mask; > +} > + > +void > +add_meta_data(struct rte_flow_item items[MAX_ITEMS_NUM], > + uint8_t items_counter) > +{ > + uint32_t data = META_DATA; > + memset(&meta_spec, 0, sizeof(struct rte_flow_item_meta)); > + memset(&meta_mask, 0, sizeof(struct rte_flow_item_meta)); > + > + meta_spec.data = RTE_BE32(data); > + meta_mask.data = RTE_BE32(0xffffffff); > + > + items[items_counter].type = RTE_FLOW_ITEM_TYPE_META; > + items[items_counter].spec = &meta_spec; > + items[items_counter].mask = &meta_mask; > +} > + > + > +void > +add_meta_tag(struct rte_flow_item items[MAX_ITEMS_NUM], > + uint8_t items_counter) > +{ > + uint32_t data = META_DATA; > + uint8_t index = TAG_INDEX; > + memset(&tag_spec, 0, sizeof(struct rte_flow_item_tag)); > + memset(&tag_mask, 0, sizeof(struct rte_flow_item_tag)); > + > + tag_spec.data = RTE_BE32(data); > + tag_mask.data = RTE_BE32(0xffffffff); > + tag_spec.index = index; > + tag_mask.index = 0xff; > + > + items[items_counter].type = RTE_FLOW_ITEM_TYPE_TAG; > + items[items_counter].spec = &tag_spec; > + items[items_counter].mask = &tag_mask; > +} [snip] > diff --git a/app/test-flow-perf/main.c b/app/test-flow-perf/main.c > index 7a924cdb7..463e4a782 100644 > --- a/app/test-flow-perf/main.c > +++ b/app/test-flow-perf/main.c > @@ -35,29 +35,156 @@ > #include > > #include "config.h" > +#include "flow_gen.h" > > -static uint32_t nb_lcores; > +#define MAX_ITERATIONS 100 > +#define DEFAULT_RULES_COUNT 4000000 > +#define DEFAULT_ITERATION 100000 > + > +struct rte_flow *flow; > +static uint8_t flow_group; > + > +static uint16_t flow_items; > +static uint16_t flow_actions; > +static uint8_t flow_attrs; > +static volatile bool force_quit; > +static bool dump_iterations; > static struct rte_mempool *mbuf_mp; > +static uint32_t nb_lcores; > +static uint32_t flows_count; > +static uint32_t iterations_number; > +static uint32_t hairpinq; Global variables again. > static void > usage(char *progname) > { > printf("\nusage: %s\n", progname); > + printf("\nControl configurations:\n"); > + printf(" --flows-count=N: to set the number of needed" > + " flows to insert, default is 4,000,000\n"); > + printf(" --dump-iterations: To print rates for each" > + " iteration\n"); > + > + printf("To set flow attributes:\n"); > + printf(" --ingress: set ingress attribute in flows\n"); > + printf(" --egress: set egress attribute in flows\n"); > + printf(" --transfer: set transfer attribute in flows\n"); > + printf(" --group=N: set group for all flows," > + " default is 0\n"); > + > + printf("To set flow items:\n"); > + printf(" --ether: add ether layer in flow items\n"); > + printf(" --vlan: add vlan layer in flow items\n"); > + printf(" --ipv4: add ipv4 layer in flow items\n"); > + printf(" --ipv6: add ipv6 layer in flow items\n"); > + printf(" --tcp: add tcp layer in flow items\n"); > + printf(" --udp: add udp layer in flow items\n"); > + printf(" --vxlan: add vxlan layer in flow items\n"); > + printf(" --vxlan-gpe: add vxlan-gpe layer in flow items\n"); > + printf(" --gre: add gre layer in flow items\n"); > + printf(" --geneve: add geneve layer in flow items\n"); > + printf(" --gtp: add gtp layer in flow items\n"); > + printf(" --meta: add meta layer in flow items\n"); > + printf(" --tag: add tag layer in flow items\n"); > + > + printf("To set flow actions:\n"); > + printf(" --port-id: add port-id action in flow actions\n"); > + printf(" --rss: add rss action in flow actions\n"); > + printf(" --queue: add queue action in flow actions\n"); > + printf(" --jump: add jump action in flow actions\n"); > + printf(" --mark: add mark action in flow actions\n"); > + printf(" --count: add count action in flow actions\n"); > + printf(" --set-meta: add set meta action in flow actions\n"); > + printf(" --set-tag: add set tag action in flow actions\n"); > + printf(" --drop: add drop action in flow actions\n"); > + printf(" --hairpin-queue=N: add hairpin-queue action in flow actions\n"); > + printf(" --hairpin-rss=N: add hairping-rss action in flow actions\n"); > } > > static void > args_parse(int argc, char **argv) > { > char **argvopt; > - int opt; > + int n, opt, i; > int opt_idx; > + static const char * const items_str[] = { > + "ether", "vlan", "ipv4", "ipv6", Typicaly Ethernet pattern item is names "eth" (e.g. in testpmd if I'm not mistaken). Why does it differ here? > + "tcp", "udp", "vxlan", "vxlan-gpe", > + "gre", "geneve", "gtp", "meta", > + "tag" > + }; > + uint32_t items[] = { > + ETH_ITEM, VLAN_ITEM, IPV4_ITEM, > + IPV6_ITEM, TCP_ITEM, UDP_ITEM, > + VXLAN_ITEM, VXLAN_GPE_ITEM, GRE_ITEM, > + GENEVE_ITEM, GTP_ITEM, META_ITEM, > + TAG_ITEM > + }; I guess above two structures should be both const and must be in sync. If so, it would be better to create one array structures with name and mask members. It is more robust. > + static const char * const attributes_str[] = { > + "ingress", "egress", "transfer" > + }; > + uint32_t attributes[] = { > + INGRESS, EGRESS, TRANSFER > + }; Same as above. > + static const char * const actions_str[] = { > + "port-id", "rss", "queue", "jump", > + "mark", "count", "set-meta", "set-tag", > + "drop", > + }; > + uint32_t actions[] = { > + PORT_ID_ACTION, RSS_ACTION, QUEUE_ACTION, > + JUMP_ACTION, MARK_ACTION, COUNT_ACTION, > + META_ACTION, TAG_ACTION, DROP_ACTION > + }; Same as above. > + int items_size = RTE_DIM(items); > + int attributes_size = RTE_DIM(attributes); > + int actions_size = RTE_DIM(actions); > + > static struct option lgopts[] = { > /* Control */ > { "help", 0, 0, 0 }, > + { "flows-count", 1, 0, 0 }, > + { "dump-iterations", 0, 0, 0 }, > + /* Attributes */ > + { "ingress", 0, 0, 0 }, > + { "egress", 0, 0, 0 }, > + { "transfer", 0, 0, 0 }, I think it should be possible to add these items in a loop by attributes_str-like array. > + { "group", 1, 0, 0 }, > + /* Items */ > + { "ether", 0, 0, 0 }, > + { "vlan", 0, 0, 0 }, > + { "ipv4", 0, 0, 0 }, > + { "ipv6", 0, 0, 0 }, > + { "tcp", 0, 0, 0 }, > + { "udp", 0, 0, 0 }, > + { "vxlan", 0, 0, 0 }, > + { "vxlan-gpe", 0, 0, 0 }, > + { "gre", 0, 0, 0 }, > + { "geneve", 0, 0, 0 }, > + { "gtp", 0, 0, 0 }, > + { "meta", 0, 0, 0 }, > + { "tag", 0, 0, 0 }, I think it should be possible to add these items in a loop by items_str-like array. It would allow to avoid copy-paste and simplify new items addition. > + /* Actions */ > + { "port-id", 0, 0, 0 }, > + { "rss", 0, 0, 0 }, > + { "queue", 0, 0, 0 }, > + { "jump", 0, 0, 0 }, > + { "mark", 0, 0, 0 }, > + { "count", 0, 0, 0 }, > + { "set-meta", 0, 0, 0 }, > + { "set-tag", 0, 0, 0 }, > + { "drop", 0, 0, 0 }, I think it should be possible to add these items in a loop by actions_str-like array. It would allow to avoid copy-paste and simplify new actions addition. > + { "hairpin-queue", 1, 0, 0 }, > + { "hairpin-rss", 1, 0, 0 }, > }; > > + flow_items = 0; > + flow_actions = 0; > + flow_attrs = 0; > + hairpinq = 0; > argvopt = argv; > > + printf(":: Flow -> "); > while ((opt = getopt_long(argc, argvopt, "", > lgopts, &opt_idx)) != EOF) { > switch (opt) { > @@ -66,6 +193,73 @@ args_parse(int argc, char **argv) > usage(argv[0]); > rte_exit(EXIT_SUCCESS, "Displayed help\n"); > } > + > + /* Attributes */ > + for (i = 0; i < attributes_size; i++) > + if (!strcmp(lgopts[opt_idx].name, > + attributes_str[i])) { Comparison vs 0. Many cases below. > + flow_attrs |= attributes[i]; > + printf("%s / ", attributes_str[i]); > + } > + if (!strcmp(lgopts[opt_idx].name, "group")) { > + n = atoi(optarg); > + if (n >= 0) > + flow_group = n; > + else > + rte_exit(EXIT_SUCCESS, > + "flow group should be >= 0"); > + printf("group %d ", flow_group); > + } > + > + /* Items */ > + for (i = 0; i < items_size; i++) > + if (!strcmp(lgopts[opt_idx].name, > + items_str[i])) { > + flow_items |= items[i]; > + printf("%s / ", items_str[i]); > + } > + > + /* Actions */ > + for (i = 0; i < actions_size; i++) > + if (!strcmp(lgopts[opt_idx].name, > + actions_str[i])) { > + flow_actions |= actions[i]; > + printf("%s / ", actions_str[i]); > + } > + if (!strcmp(lgopts[opt_idx].name, "hairpin-rss")) { > + n = atoi(optarg); > + if (n > 0) > + hairpinq = n; > + else > + rte_exit(EXIT_SUCCESS, "Hairpin queues should be > 0 "); > + > + flow_actions |= HAIRPIN_RSS_ACTION; > + printf("hairpin-rss / "); > + } > + if (!strcmp(lgopts[opt_idx].name, "hairpin-queue")) { > + n = atoi(optarg); > + if (n > 0) > + hairpinq = n; > + else > + rte_exit(EXIT_SUCCESS, "Hairpin queues should be > 0 "); > + > + flow_actions |= HAIRPIN_QUEUE_ACTION; > + printf("hairpin-queue / "); > + } > + > + /* Control */ > + if (!strcmp(lgopts[opt_idx].name, "flows-count")) { > + n = atoi(optarg); > + if (n > (int) iterations_number) > + flows_count = n; > + else { > + printf("\n\nflows_count should be > %d", > + iterations_number); > + rte_exit(EXIT_SUCCESS, " "); > + } > + } > + if (!strcmp(lgopts[opt_idx].name, "dump-iterations")) > + dump_iterations = true; > break; > default: > printf("Invalid option: %s\n", argv[optind]); > @@ -74,15 +268,141 @@ args_parse(int argc, char **argv) > break; > } > } > + printf("end_flow\n"); > +} > + > +static void > +print_flow_error(struct rte_flow_error error) > +{ > + printf("Flow can't be created %d message: %s\n", > + error.type, > + error.message ? error.message : "(no stated reason)"); > +} > + > +static inline void > +flows_handler(void) > +{ > + struct rte_flow_error error; > + clock_t start_iter, end_iter; > + double cpu_time_used; > + double flows_rate; > + double cpu_time_per_iter[MAX_ITERATIONS]; > + double delta; > + uint16_t nr_ports; > + uint32_t i; > + int port_id; > + int iter_id; > + uint32_t eagain_counter = 0; > + > + nr_ports = rte_eth_dev_count_avail(); > + > + for (i = 0; i < MAX_ITERATIONS; i++) > + cpu_time_per_iter[i] = -1; > + > + if (iterations_number > flows_count) > + iterations_number = flows_count; > + > + printf(":: Flows Count per port: %d\n", flows_count); > + > + for (port_id = 0; port_id < nr_ports; port_id++) { > + cpu_time_used = 0; > + if (flow_group > 0) { > + /* > + * Create global rule to jump into flow_group, > + * this way the app will avoid the default rules. > + * > + * Golbal rule: > + * group 0 eth / end actions jump group > + * > + */ > + flow = generate_flow(port_id, 0, flow_attrs, ETH_ITEM, > + JUMP_ACTION, flow_group, 0, 0, &error); > + > + if (!flow) { Comparison vs NULL > + print_flow_error(error); > + rte_exit(EXIT_FAILURE, "error in creating flow"); > + } > + } > + > + /* Insertion Rate */ > + printf("Flows insertion on port = %d\n", port_id); > + start_iter = clock(); > + for (i = 0; i < flows_count; i++) { > + do { > + rte_errno = 0; > + flow = generate_flow(port_id, flow_group, > + flow_attrs, flow_items, flow_actions, > + JUMP_ACTION_TABLE, i, hairpinq, &error); > + if (!flow) > + eagain_counter++; > + } while (rte_errno == EAGAIN); > + > + if (force_quit) > + i = flows_count; > + > + if (!flow) { > + print_flow_error(error); > + rte_exit(EXIT_FAILURE, "error in creating flow"); > + } > + > + if (i && !((i + 1) % iterations_number)) { > + /* Save the insertion rate of each iter */ > + end_iter = clock(); > + delta = (double) (end_iter - start_iter); > + iter_id = ((i + 1) / iterations_number) - 1; > + cpu_time_per_iter[iter_id] = > + delta / CLOCKS_PER_SEC; > + cpu_time_used += cpu_time_per_iter[iter_id]; > + start_iter = clock(); > + } > + } > + > + /* Iteration rate per iteration */ > + if (dump_iterations) > + for (i = 0; i < MAX_ITERATIONS; i++) { > + if (cpu_time_per_iter[i] == -1) > + continue; > + delta = (double)(iterations_number / > + cpu_time_per_iter[i]); > + flows_rate = delta / 1000; > + printf(":: Iteration #%d: %d flows " > + "in %f sec[ Rate = %f K/Sec ]\n", > + i, iterations_number, > + cpu_time_per_iter[i], flows_rate); > + } > + > + /* Insertion rate for all flows */ > + flows_rate = ((double) (flows_count / cpu_time_used) / 1000); > + printf("\n:: Total flow insertion rate -> %f K/Sec\n", > + flows_rate); > + printf(":: The time for creating %d in flows %f seconds\n", > + flows_count, cpu_time_used); > + printf(":: EAGAIN counter = %d\n", eagain_counter); > + } > +} > + > +static void > +signal_handler(int signum) > +{ > + if (signum == SIGINT || signum == SIGTERM) { > + printf("\n\nSignal %d received, preparing to exit...\n", > + signum); > + printf("Error: Stats are wrong due to sudden signal!\n\n"); > + force_quit = true; > + } It is the patch which starts to use sigint.h and it should be included in the patch, not the previous one. > } > > static void > init_port(void) > { > int ret; > - uint16_t i; > + uint16_t i, j; > uint16_t port_id; > uint16_t nr_ports; > + uint16_t nr_queues; > + struct rte_eth_hairpin_conf hairpin_conf = { > + .peer_count = 1, > + }; > struct rte_eth_conf port_conf = { > .rx_adv_conf = { > .rss_conf.rss_hf = > @@ -94,6 +414,10 @@ init_port(void) > struct rte_eth_rxconf rxq_conf; > struct rte_eth_dev_info dev_info; > > + nr_queues = RXQ_NUM; > + if (hairpinq) Comparison vs 0 > + nr_queues = RXQ_NUM + hairpinq; > + > nr_ports = rte_eth_dev_count_avail(); > if (nr_ports == 0) > rte_exit(EXIT_FAILURE, "Error: no port detected\n"); > @@ -118,8 +442,8 @@ init_port(void) > > printf(":: initializing port: %d\n", port_id); > > - ret = rte_eth_dev_configure(port_id, RXQ_NUM, > - TXQ_NUM, &port_conf); > + ret = rte_eth_dev_configure(port_id, nr_queues, > + nr_queues, &port_conf); > if (ret < 0) > rte_exit(EXIT_FAILURE, > ":: cannot configure device: err=%d, port=%u\n", > @@ -159,6 +483,30 @@ init_port(void) > ":: promiscuous mode enable failed: err=%s, port=%u\n", > rte_strerror(-ret), port_id); > > + if (hairpinq) { > + for (i = RXQ_NUM, j = 0; i < nr_queues; i++, j++) { > + hairpin_conf.peers[0].port = port_id; > + hairpin_conf.peers[0].queue = j + TXQ_NUM; > + ret = rte_eth_rx_hairpin_queue_setup(port_id, i, > + NR_RXD, &hairpin_conf); > + if (ret != 0) > + rte_exit(EXIT_FAILURE, > + ":: Hairpin rx queue setup failed: err=%d, port=%u\n", > + ret, port_id); > + } > + > + for (i = TXQ_NUM, j = 0; i < nr_queues; i++, j++) { > + hairpin_conf.peers[0].port = port_id; > + hairpin_conf.peers[0].queue = j + RXQ_NUM; > + ret = rte_eth_tx_hairpin_queue_setup(port_id, i, > + NR_TXD, &hairpin_conf); > + if (ret != 0) > + rte_exit(EXIT_FAILURE, > + ":: Hairpin tx queue setup failed: err=%d, port=%u\n", > + ret, port_id); > + } > + } > + > ret = rte_eth_dev_start(port_id); > if (ret < 0) > rte_exit(EXIT_FAILURE, > @@ -180,6 +528,15 @@ main(int argc, char **argv) > if (ret < 0) > rte_exit(EXIT_FAILURE, "EAL init failed\n"); > > + force_quit = false; > + dump_iterations = false; > + flows_count = DEFAULT_RULES_COUNT; > + iterations_number = DEFAULT_ITERATION; > + flow_group = 0; > + > + signal(SIGINT, signal_handler); > + signal(SIGTERM, signal_handler); > + > argc -= ret; > argv += ret; > if (argc > 1) > @@ -191,6 +548,8 @@ main(int argc, char **argv) > if (nb_lcores <= 1) > rte_exit(EXIT_FAILURE, "This app needs at least two cores\n"); > > + flows_handler(); > + > RTE_ETH_FOREACH_DEV(port) { > rte_flow_flush(port, &error); > rte_eth_dev_stop(port); [snip] > diff --git a/doc/guides/tools/flow-perf.rst b/doc/guides/tools/flow-perf.rst > index 49eb450ae..b45fccd69 100644 > --- a/doc/guides/tools/flow-perf.rst > +++ b/doc/guides/tools/flow-perf.rst > @@ -4,7 +4,18 @@ > Flow performance tool > ===================== > > -Application for rte_flow performance testing. > +Application for rte_flow performance testing. The application provide the As far as I remember documentaiton guidelines recommend to start new sentenses from a new line. > +ability to test insertion rate of specific rte_flow rule, by stressing it > +to the NIC, and calculate the insertion rate. > + > +The application offers some options in the command line, to configure > +which rule to apply. > + > +After that the application will start producing rules with same pattern > +but increasing the outer IP source address by 1 each time, thus it will > +give different flow each time, and all other items will have open masks. > + > +The current design have single core insertion rate. > > > Compiling the Application [snip]