From patchwork Thu Jun 21 07:25:02 2018 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: =?utf-8?q?N=C3=A9lio_Laranjeiro?= X-Patchwork-Id: 41343 X-Patchwork-Delegate: ferruh.yigit@amd.com Return-Path: X-Original-To: patchwork@dpdk.org Delivered-To: patchwork@dpdk.org Received: from [92.243.14.124] (localhost [127.0.0.1]) by dpdk.org (Postfix) with ESMTP id 014951BB20; Thu, 21 Jun 2018 09:24:59 +0200 (CEST) Received: from mail-wm0-f67.google.com (mail-wm0-f67.google.com [74.125.82.67]) by dpdk.org (Postfix) with ESMTP id 0A0DA1BB1C for ; Thu, 21 Jun 2018 09:24:58 +0200 (CEST) Received: by mail-wm0-f67.google.com with SMTP id p11-v6so3881125wmc.4 for ; Thu, 21 Jun 2018 00:24:57 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=6wind-com.20150623.gappssmtp.com; s=20150623; h=from:to:subject:date:message-id:in-reply-to:references; bh=3gYti5iutHVY+p+b2eG4iwWU0XQ4qaqGQnECd1BhKEU=; b=zITNVj9Dx55mDGZEz1XPnUnG+Yp3z44z4agobX7L4iYc3hfAfZ2IQ0ZUf3K9FEFCx0 vJFlFhbuQmZXrOTQbF3vNG4wbWhErNh0JiexLI9S7L0NPDAIJuC4NJWaBCWK3Vc5IYAP wixyU8G/HkOwqyhzsSGFojGZ+PwJL95PVcHam2zA53nMNOrgPKoXbwWQbRFG4hkADu4b 58m4AICzxtqUHotPyK71a8q0w4ZtWmewLz2kkF5R/VS+0z74oYQ5LM34fjMEpQssbLJo bhTza8Qta9Ew2Mo8z8ZRubmmVO1cXi3mcXDKPt94v74qVJTgH1S5rc1YFL7QC93TBjwe tfBw== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:from:to:subject:date:message-id:in-reply-to :references; bh=3gYti5iutHVY+p+b2eG4iwWU0XQ4qaqGQnECd1BhKEU=; b=daHezhApH4Csl+anmmZDEq1FyQyNaAAKh2Conlf2/Hh+kJJw+dXx83sQb+lzG+cnPZ ML7ICuPpp4faafuCoEvEKzP4QxHTQsMao6XQJGy0TWCVRl4oUgw9hiX0BUzhetXtKaA+ CfHCCWcIT38S7KIr3SS8wjNLWDlnrzwbw3TQlitaBv8EJXJAWsL8Y+5aeAsXgTh/0wDz 6+teT4FQ3PtIRDbM8X9qLRA42EY/iRsgMcIQQ85R3xsUDFZ1Aw1IMofDaag36ATeDYpY /lUWdt9VUHufoI7VNbFRyee0PZPS57QkOL7ucnQuSsoFAupsUSMUvYdSWT0GVcDo3ucF TrDw== X-Gm-Message-State: APt69E1Tz9dZP8+Q0M1wvnFyj9GSQQswv6NRE9BiXTOX8bNrGsMG/bmq pi6273Q8nQk0IE0bK/2tUGcBsHYbdQ== X-Google-Smtp-Source: ADUXVKI3//fywxCUD8icuqTCccvYgt8ZkeCwO3QSIE98KHvG1LG1wqnbY9bOltrAB8ZboUT5N5d+2A== X-Received: by 2002:a1c:8c55:: with SMTP id o82-v6mr3843304wmd.60.1529565897283; Thu, 21 Jun 2018 00:24:57 -0700 (PDT) Received: from laranjeiro-vm.dev.6wind.com (host.78.145.23.62.rev.coltfrance.com. [62.23.145.78]) by smtp.gmail.com with ESMTPSA id m10-v6sm5821907wrq.56.2018.06.21.00.24.56 (version=TLS1_2 cipher=ECDHE-RSA-AES128-GCM-SHA256 bits=128/128); Thu, 21 Jun 2018 00:24:56 -0700 (PDT) From: Nelio Laranjeiro To: dev@dpdk.org, Adrien Mazarguil Date: Thu, 21 Jun 2018 09:25:02 +0200 Message-Id: <15e22f72e9b4c56e79809c413ce3001e4f6067d8.1529565844.git.nelio.laranjeiro@6wind.com> X-Mailer: git-send-email 2.18.0.rc2 In-Reply-To: <3c31112367e8d44d1568afbf52de20c6c9ff0c9c.1528187101.git.nelio.laranjeiro@6wind.com> References: <3c31112367e8d44d1568afbf52de20c6c9ff0c9c.1528187101.git.nelio.laranjeiro@6wind.com> Subject: [dpdk-dev] [PATCH v3] ethdev: add flow API to expand RSS flows 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" Introduce an helper for PMD to expand easily flows items list with RSS action into multiple flow items lists with priority information. For instance a user items list being "eth / end" with rss action types "ipv4-udp ipv6-udp end" needs to be expanded into three items lists: - eth - eth / ipv4 / udp - eth / ipv6 / udp to match the user request. Some drivers are unable to reach such request without this expansion, this API is there to help those. Only PMD should use such API for their internal cooking, the application will still handle a single flow. Signed-off-by: Nelio Laranjeiro --- Changes in v3: - Fix a segmentation fault due to an uninitialized pointer. Changes in v2: - Fix expansion for UDP/TCP layers where L3 may not be in the original items list and thus is missing in the expansion. - Fix size verification for some layers causing a segfault --- lib/librte_ethdev/rte_flow.c | 408 ++++++++++++++++++++++++++++ lib/librte_ethdev/rte_flow_driver.h | 32 +++ 2 files changed, 440 insertions(+) diff --git a/lib/librte_ethdev/rte_flow.c b/lib/librte_ethdev/rte_flow.c index b2afba089..9d86c20cb 100644 --- a/lib/librte_ethdev/rte_flow.c +++ b/lib/librte_ethdev/rte_flow.c @@ -526,3 +526,411 @@ rte_flow_copy(struct rte_flow_desc *desc, size_t len, } return 0; } + +/* Copy the existing items list and expand with new items. */ +static int +rte_flow_expand_rss_item(void *buf, size_t size, + const struct rte_flow_item *items, + const struct rte_flow_item *newitems) +{ + void *data = buf; + const struct rte_flow_item *item; + struct rte_flow_item *dst; + size_t data_size = 0; + + dst = data; + /* Copy Item structure into buffer. */ + for (item = items; item->type != RTE_FLOW_ITEM_TYPE_END; ++item) { + if (item->type == RTE_FLOW_ITEM_TYPE_VOID) + continue; + if (data_size + sizeof(*item) <= size) { + memcpy(dst, item, sizeof(*item)); + ++dst; + } + data_size += sizeof(*item); + } + item = newitems; + do { + if (item->type == RTE_FLOW_ITEM_TYPE_VOID) { + ++item; + continue; + } + if (data_size + sizeof(*item) <= size) { + memcpy(dst, item, sizeof(*item)); + ++dst; + } + data_size += sizeof(*item); + ++item; + } while ((item - 1)->type != RTE_FLOW_ITEM_TYPE_END); + /** + * Copy Item spec, last, mask into buffer and set pointers + * accordingly. + */ + dst = data; + for (item = items; item->type != RTE_FLOW_ITEM_TYPE_END; ++item) { + if (item->type == RTE_FLOW_ITEM_TYPE_VOID) + continue; + if (item->spec) { + size_t s = flow_item_spec_copy(NULL, item, ITEM_SPEC); + void *addr = (data_size + s) <= size ? + (void *)((uintptr_t)data + data_size) : + NULL; + + data_size += flow_item_spec_copy(addr, item, ITEM_SPEC); + if (addr) + dst->spec = addr; + } + if (item->last) { + size_t s = flow_item_spec_copy(NULL, item, ITEM_LAST); + void *addr = (data_size + s) <= size ? + (void *)((uintptr_t)data + data_size) : + NULL; + + data_size += flow_item_spec_copy(addr, item, ITEM_LAST); + if (addr) + dst->last = addr; + } + if (item->mask) { + size_t s = flow_item_spec_copy(NULL, item, ITEM_MASK); + void *addr = (data_size + s) <= size ? + (void *)((uintptr_t)data + data_size) : + NULL; + + data_size += flow_item_spec_copy(addr, item, ITEM_MASK); + if (addr) + dst->mask = addr; + } + if (data_size <= size) + ++dst; + } + return data_size; +} + +/** Verify the expansion is supported by the device. */ +static int +rte_flow_expand_rss_is_supported(const enum rte_flow_item_type **supported, + const enum rte_flow_item_type *expand) +{ + unsigned int i; + unsigned int sidx; + unsigned int eidx; + + for (i = 0; supported[i]; ++i) { + sidx = 0; + eidx = 0; + while (1) { + if (expand[eidx] != supported[i][sidx]) { + break; + } else if ((expand[eidx] == RTE_FLOW_ITEM_TYPE_END) && + (supported[i][sidx] == + RTE_FLOW_ITEM_TYPE_END)) { + return 1; + } else if ((expand[eidx] == RTE_FLOW_ITEM_TYPE_END) || + (supported[i][sidx] == + RTE_FLOW_ITEM_TYPE_END)) { + break; + } else if (expand[eidx] == RTE_FLOW_ITEM_TYPE_VOID) { + ++eidx; + continue; + } else if (supported[i][sidx] == + RTE_FLOW_ITEM_TYPE_VOID) { + ++sidx; + continue; + } + ++sidx; + ++eidx; + } + } + return 0; +} + +/** Update internal buffer. */ +static inline void +rte_flow_expand_rss_update(struct rte_flow_expand_rss *buf, void *addr, + uint32_t priority) +{ + buf->priority[buf->entries] = priority; + buf->patterns[buf->entries] = addr; + buf->entries++; +} + +int +rte_flow_expand_rss(struct rte_flow_expand_rss *buf, size_t size, + const struct rte_flow_item *pat, uint64_t types, + const enum rte_flow_item_type **supported) +{ + const struct rte_flow_item *item; + uint32_t priority = 0; + struct { + uint32_t eth:1; /**< Ethernet item is present. */ + uint32_t ipv4:1; /**< IPv4 item is present. */ + uint32_t ipv6:1; /**< IPv6 item is present. */ + uint32_t ipv6_ex:1; /**< IPv6 EXT item is present. */ + uint32_t udp:1; /**< UDP item is present. */ + uint32_t tcp:1; /**< TCP item is present. */ + uint32_t sctp:1; /**< STCP item is present. */ + uint32_t vxlan:1; /**< VXLAN item is present. */ + uint32_t geneve:1; /**< GENEVE item is present. */ + uint32_t nvgre:1; /**< NVGRE item is present. */ + } layer = { .eth = 0 }; + const struct rte_flow_item end[] = { + { .type = RTE_FLOW_ITEM_TYPE_END }, + }; + void *addr = NULL; + uint32_t off; /**< Offset to write new items data starting from *buf. */ + uint32_t max_entries; + + for (max_entries = 0; supported[max_entries]; ++max_entries) + ; + off = sizeof(*buf) + + /* Size for the list of patterns. */ + sizeof(*buf->patterns) + + RTE_ALIGN_CEIL(max_entries * sizeof(struct rte_flow_item *), + sizeof(void *)) + + /* Size for priorities. */ + sizeof(*buf->priority) + + RTE_ALIGN_CEIL(max_entries * sizeof(uint32_t), sizeof(void *)); + if (off < size) { + buf->priority = (void *)(buf + 1); + buf->patterns = (void *)&buf->priority[max_entries]; + buf->patterns[0] = (void *)&buf->patterns[max_entries]; + addr = buf->patterns[0]; + buf->entries = 0; + } + /** + * Parse the pattern and deactivate the bit-field in RSS which cannot + * match anymore the pattern. + */ + for (item = pat; item->type != RTE_FLOW_ITEM_TYPE_END; ++item) { + switch (item->type) { + case RTE_FLOW_ITEM_TYPE_ETH: + layer.eth = 1; + break; + case RTE_FLOW_ITEM_TYPE_IPV4: + layer.ipv4 = 1; + break; + case RTE_FLOW_ITEM_TYPE_IPV6: + layer.ipv6 = 1; + break; + case RTE_FLOW_ITEM_TYPE_IPV6_EXT: + layer.ipv6_ex = 1; + break; + case RTE_FLOW_ITEM_TYPE_UDP: + layer.udp = 1; + break; + case RTE_FLOW_ITEM_TYPE_TCP: + layer.tcp = 1; + break; + case RTE_FLOW_ITEM_TYPE_VXLAN: + layer.vxlan = 1; + break; + case RTE_FLOW_ITEM_TYPE_GENEVE: + layer.geneve = 1; + break; + case RTE_FLOW_ITEM_TYPE_NVGRE: + layer.nvgre = 1; + break; + default: + break; + } + } + off += rte_flow_expand_rss_item(addr, (off < size) ? size - off : 0, + pat, end); + if (off <= size) { + rte_flow_expand_rss_update(buf, addr, priority); + addr = (void *)((uintptr_t)buf + off); + } + if ((types & ETH_RSS_IP) && + (!(layer.ipv4 || layer.ipv6 || layer.ipv6_ex))) { + ++priority; + if (types & (ETH_RSS_IPV4 | ETH_RSS_FRAG_IPV4 | + ETH_RSS_NONFRAG_IPV4_OTHER)) { + const struct rte_flow_item new[] = { + { .type = RTE_FLOW_ITEM_TYPE_IPV4 }, + { .type = RTE_FLOW_ITEM_TYPE_END }, + }; + const enum rte_flow_item_type list[] = { + RTE_FLOW_ITEM_TYPE_ETH, + RTE_FLOW_ITEM_TYPE_IPV4, + RTE_FLOW_ITEM_TYPE_END, + }; + int ret; + + ret = rte_flow_expand_rss_is_supported(supported, list); + if (ret) { + off += rte_flow_expand_rss_item + (addr, (off <= size) ? size - off : 0, + pat, new); + if (off <= size) { + rte_flow_expand_rss_update(buf, addr, + priority); + addr = (void *)((uintptr_t)buf + off); + } + } + } + if (types & (ETH_RSS_IPV6 | ETH_RSS_FRAG_IPV6 | + ETH_RSS_NONFRAG_IPV6_OTHER)) { + const struct rte_flow_item new[] = { + { .type = RTE_FLOW_ITEM_TYPE_IPV6 }, + { .type = RTE_FLOW_ITEM_TYPE_END }, + }; + const enum rte_flow_item_type list[] = { + RTE_FLOW_ITEM_TYPE_ETH, + RTE_FLOW_ITEM_TYPE_IPV6, + RTE_FLOW_ITEM_TYPE_END, + }; + int ret; + + ret = rte_flow_expand_rss_is_supported(supported, list); + if (ret) { + off += rte_flow_expand_rss_item + (addr, (off <= size) ? size - off : 0, + pat, new); + if (off <= size) { + rte_flow_expand_rss_update(buf, addr, + priority); + addr = (void *)((uintptr_t)buf + off); + } + } + } + if (types & ETH_RSS_IPV6_EX) { + const struct rte_flow_item new[] = { + { .type = RTE_FLOW_ITEM_TYPE_IPV6_EXT }, + { .type = RTE_FLOW_ITEM_TYPE_END }, + }; + const enum rte_flow_item_type list[] = { + RTE_FLOW_ITEM_TYPE_ETH, + RTE_FLOW_ITEM_TYPE_IPV6_EXT, + RTE_FLOW_ITEM_TYPE_END, + }; + int ret; + + ret = rte_flow_expand_rss_is_supported(supported, list); + if (ret) { + off += rte_flow_expand_rss_item + (addr, (off <= size) ? size - off : 0, + pat, new); + if (off <= size) { + rte_flow_expand_rss_update(buf, addr, + priority); + addr = (void *)((uintptr_t)buf + off); + } + } + } + } + if (types & (ETH_RSS_TCP | ETH_RSS_UDP)) { + ++priority; + if ((types & ETH_RSS_NONFRAG_IPV4_UDP) && + !(layer.ipv6 || layer.ipv6_ex || layer.tcp || layer.udp)) { + const struct rte_flow_item new[] = { + { .type = RTE_FLOW_ITEM_TYPE_IPV4 }, + { .type = RTE_FLOW_ITEM_TYPE_UDP }, + { .type = RTE_FLOW_ITEM_TYPE_END }, + }; + const enum rte_flow_item_type list[] = { + RTE_FLOW_ITEM_TYPE_ETH, + RTE_FLOW_ITEM_TYPE_IPV4, + RTE_FLOW_ITEM_TYPE_UDP, + RTE_FLOW_ITEM_TYPE_END, + }; + int ret; + + ret = rte_flow_expand_rss_is_supported(supported, list); + if (ret) { + off += rte_flow_expand_rss_item + (addr, (off <= size) ? size - off : 0, + pat, layer.ipv4 ? &new[1] : new); + if (off <= size) { + rte_flow_expand_rss_update(buf, addr, + priority); + addr = (void *)((uintptr_t)buf + off); + } + } + } + if ((types & ETH_RSS_NONFRAG_IPV4_TCP) && + !(layer.ipv6 || layer.ipv6_ex || layer.tcp || layer.udp)) { + const struct rte_flow_item new[] = { + { .type = RTE_FLOW_ITEM_TYPE_IPV4 }, + { .type = RTE_FLOW_ITEM_TYPE_TCP }, + { .type = RTE_FLOW_ITEM_TYPE_END }, + }; + const enum rte_flow_item_type list[] = { + RTE_FLOW_ITEM_TYPE_ETH, + RTE_FLOW_ITEM_TYPE_IPV4, + RTE_FLOW_ITEM_TYPE_TCP, + RTE_FLOW_ITEM_TYPE_END, + }; + int ret; + + ret = rte_flow_expand_rss_is_supported(supported, list); + if (ret) { + off += rte_flow_expand_rss_item + (addr, (off <= size) ? size - off : 0, + pat, layer.ipv4 ? &new[1] : new); + if (off <= size) { + rte_flow_expand_rss_update(buf, addr, + priority); + addr = (void *)((uintptr_t)buf + off); + } + } + } + if ((types & ETH_RSS_NONFRAG_IPV6_UDP) && + !(layer.ipv4 || layer.tcp || layer.udp)) { + const struct rte_flow_item new[] = { + { .type = RTE_FLOW_ITEM_TYPE_IPV6 }, + { .type = RTE_FLOW_ITEM_TYPE_UDP }, + { .type = RTE_FLOW_ITEM_TYPE_END }, + }; + const enum rte_flow_item_type list[] = { + RTE_FLOW_ITEM_TYPE_ETH, + RTE_FLOW_ITEM_TYPE_IPV6, + RTE_FLOW_ITEM_TYPE_UDP, + RTE_FLOW_ITEM_TYPE_END, + }; + int ret; + + ret = rte_flow_expand_rss_is_supported(supported, list); + if (ret) { + off += rte_flow_expand_rss_item + (addr, (off <= size) ? size - off : 0, + pat, layer.ipv6 ? &new[1] : new); + if (off <= size) { + rte_flow_expand_rss_update(buf, addr, + priority); + addr = (void *)((uintptr_t)buf + off); + } + } + } + if ((types & (ETH_RSS_NONFRAG_IPV6_TCP | + ETH_RSS_IPV6_TCP_EX)) && + !(layer.ipv4 || layer.tcp || layer.udp)) { + const struct rte_flow_item new[] = { + { .type = RTE_FLOW_ITEM_TYPE_IPV6 }, + { .type = RTE_FLOW_ITEM_TYPE_TCP }, + { .type = RTE_FLOW_ITEM_TYPE_END }, + }; + const enum rte_flow_item_type list[] = { + RTE_FLOW_ITEM_TYPE_ETH, + (layer.ipv6_ex ? + RTE_FLOW_ITEM_TYPE_IPV6_EXT : + RTE_FLOW_ITEM_TYPE_IPV6), + RTE_FLOW_ITEM_TYPE_UDP, + RTE_FLOW_ITEM_TYPE_END, + }; + int ret; + + ret = rte_flow_expand_rss_is_supported(supported, list); + if (ret) { + off += rte_flow_expand_rss_item + (addr, (off <= size) ? size - off : 0, + pat, layer.ipv6 ? &new[1] : new); + if (off <= size) { + rte_flow_expand_rss_update(buf, addr, + priority); + addr = (void *)((uintptr_t)buf + off); + } + } + } + } + return off; +} diff --git a/lib/librte_ethdev/rte_flow_driver.h b/lib/librte_ethdev/rte_flow_driver.h index 1c90c600d..9058a8715 100644 --- a/lib/librte_ethdev/rte_flow_driver.h +++ b/lib/librte_ethdev/rte_flow_driver.h @@ -114,6 +114,38 @@ struct rte_flow_ops { const struct rte_flow_ops * rte_flow_ops_get(uint16_t port_id, struct rte_flow_error *error); +/** + * Expansion structure for RSS flows. + */ +struct rte_flow_expand_rss { + uint32_t entries; /**< Number of entries in the following arrays. */ + struct rte_flow_item **patterns; /**< Expanded pattern array. */ + uint32_t *priority; /**< Priority offset for each expansion. */ +}; + +/** + * Expand RSS flows into several possible flows according to the RSS hash + * fields requested and the driver capabilities. + * + * @param[in,out] buf + * Buffer to store the result expansion. + * @param[in] size + * Size in octets of the buffer. + * @param[in] pat + * User flow pattern. + * @param[in] types + * RSS types expected (see ETH_RSS_*). + * @param[in] supported. + * List of support expansion pattern from the device. + * + * @return + * The size in octets used to expand. + */ +int +rte_flow_expand_rss(struct rte_flow_expand_rss *buf, size_t size, + const struct rte_flow_item *pat, uint64_t types, + const enum rte_flow_item_type **supported); + #ifdef __cplusplus } #endif