From patchwork Wed Jul 4 18:06:22 2018 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: "Lam, Tiago" X-Patchwork-Id: 939494 X-Patchwork-Delegate: ian.stokes@intel.com Return-Path: X-Original-To: incoming@patchwork.ozlabs.org Delivered-To: patchwork-incoming@bilbo.ozlabs.org Authentication-Results: ozlabs.org; spf=pass (mailfrom) smtp.mailfrom=openvswitch.org (client-ip=140.211.169.12; helo=mail.linuxfoundation.org; envelope-from=ovs-dev-bounces@openvswitch.org; receiver=) Authentication-Results: ozlabs.org; dmarc=fail (p=none dis=none) header.from=intel.com Received: from mail.linuxfoundation.org (mail.linuxfoundation.org [140.211.169.12]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by ozlabs.org (Postfix) with ESMTPS id 41LTVd45Rlz9s4c for ; Thu, 5 Jul 2018 04:07:53 +1000 (AEST) Received: from mail.linux-foundation.org (localhost [127.0.0.1]) by mail.linuxfoundation.org (Postfix) with ESMTP id 9C3DECEB; Wed, 4 Jul 2018 18:06:51 +0000 (UTC) X-Original-To: ovs-dev@openvswitch.org Delivered-To: ovs-dev@mail.linuxfoundation.org Received: from smtp1.linuxfoundation.org (smtp1.linux-foundation.org [172.17.192.35]) by mail.linuxfoundation.org (Postfix) with ESMTPS id 81A92CE2 for ; Wed, 4 Jul 2018 18:06:50 +0000 (UTC) X-Greylist: domain auto-whitelisted by SQLgrey-1.7.6 Received: from mga05.intel.com (mga05.intel.com [192.55.52.43]) by smtp1.linuxfoundation.org (Postfix) with ESMTPS id 9892F780 for ; Wed, 4 Jul 2018 18:06:48 +0000 (UTC) X-Amp-Result: SKIPPED(no attachment in message) X-Amp-File-Uploaded: False Received: from orsmga001.jf.intel.com ([10.7.209.18]) by fmsmga105.fm.intel.com with ESMTP/TLS/DHE-RSA-AES256-GCM-SHA384; 04 Jul 2018 11:06:46 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.51,308,1526367600"; d="scan'208";a="70109051" Received: from silpixa00399125.ir.intel.com ([10.237.223.34]) by orsmga001.jf.intel.com with ESMTP; 04 Jul 2018 11:06:42 -0700 From: Tiago Lam To: ovs-dev@openvswitch.org Date: Wed, 4 Jul 2018 19:06:22 +0100 Message-Id: <1530727595-217860-2-git-send-email-tiago.lam@intel.com> X-Mailer: git-send-email 2.7.4 In-Reply-To: <1530727595-217860-1-git-send-email-tiago.lam@intel.com> References: <1530727595-217860-1-git-send-email-tiago.lam@intel.com> X-Spam-Status: No, score=-4.2 required=5.0 tests=BAYES_00, RCVD_IN_DNSWL_MED autolearn=ham version=3.3.1 X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on smtp1.linux-foundation.org Cc: i.maximets@samsung.com Subject: [ovs-dev] [PATCH v2 01/14] netdev-dpdk: fix mbuf sizing X-BeenThere: ovs-dev@openvswitch.org X-Mailman-Version: 2.1.12 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , MIME-Version: 1.0 Sender: ovs-dev-bounces@openvswitch.org Errors-To: ovs-dev-bounces@openvswitch.org From: Mark Kavanagh There are numerous factors that must be considered when calculating the size of an mbuf: - the data portion of the mbuf must be sized in accordance With Rx buffer alignment (typically 1024B). So, for example, in order to successfully receive and capture a 1500B packet, mbufs with a data portion of size 2048B must be used. - in OvS, the elements that comprise an mbuf are: * the dp packet, which includes a struct rte mbuf (704B) * RTE_PKTMBUF_HEADROOM (128B) * packet data (aligned to 1k, as previously described) * RTE_PKTMBUF_TAILROOM (typically 0) Some PMDs require that the total mbuf size (i.e. the total sum of all of the above-listed components' lengths) is cache-aligned. To satisfy this requirement, it may be necessary to round up the total mbuf size with respect to cacheline size. In doing so, it's possible that the dp_packet's data portion is inadvertently increased in size, such that it no longer adheres to Rx buffer alignment. Consequently, the following property of the mbuf no longer holds true: mbuf.data_len == mbuf.buf_len - mbuf.data_off This creates a problem in the case of multi-segment mbufs, where that assumption is assumed to be true for all but the final segment in an mbuf chain. Resolve this issue by adjusting the size of the mbuf's private data portion, as opposed to the packet data portion when aligning mbuf size to cachelines. Fixes: 4be4d22 ("netdev-dpdk: clean up mbuf initialization") Fixes: 31b88c9 ("netdev-dpdk: round up mbuf_size to cache_line_size") CC: Santosh Shukla Signed-off-by: Mark Kavanagh Acked-by: Santosh Shukla Acked-by: Eelco Chaudron --- lib/netdev-dpdk.c | 53 +++++++++++++++++++++++++++++++++++------------------ 1 file changed, 35 insertions(+), 18 deletions(-) diff --git a/lib/netdev-dpdk.c b/lib/netdev-dpdk.c index c223e5b..8fb859e 100644 --- a/lib/netdev-dpdk.c +++ b/lib/netdev-dpdk.c @@ -82,12 +82,6 @@ static struct vlog_rate_limit rl = VLOG_RATE_LIMIT_INIT(5, 20); + (2 * VLAN_HEADER_LEN)) #define MTU_TO_FRAME_LEN(mtu) ((mtu) + ETHER_HDR_LEN + ETHER_CRC_LEN) #define MTU_TO_MAX_FRAME_LEN(mtu) ((mtu) + ETHER_HDR_MAX_LEN) -#define FRAME_LEN_TO_MTU(frame_len) ((frame_len) \ - - ETHER_HDR_LEN - ETHER_CRC_LEN) -#define MBUF_SIZE(mtu) ROUND_UP((MTU_TO_MAX_FRAME_LEN(mtu) \ - + sizeof(struct dp_packet) \ - + RTE_PKTMBUF_HEADROOM), \ - RTE_CACHE_LINE_SIZE) #define NETDEV_DPDK_MBUF_ALIGN 1024 #define NETDEV_DPDK_MAX_PKT_LEN 9728 @@ -494,7 +488,7 @@ is_dpdk_class(const struct netdev_class *class) * behaviour, which reduces performance. To prevent this, use a buffer size * that is closest to 'mtu', but which satisfies the aforementioned criteria. */ -static uint32_t +static uint16_t dpdk_buf_size(int mtu) { return ROUND_UP((MTU_TO_MAX_FRAME_LEN(mtu) + RTE_PKTMBUF_HEADROOM), @@ -586,7 +580,7 @@ dpdk_mp_do_not_free(struct rte_mempool *mp) OVS_REQUIRES(dpdk_mp_mutex) * - a new mempool was just created; * - a matching mempool already exists. */ static struct rte_mempool * -dpdk_mp_create(struct netdev_dpdk *dev, int mtu) +dpdk_mp_create(struct netdev_dpdk *dev, uint16_t mbuf_pkt_data_len) { char mp_name[RTE_MEMPOOL_NAMESIZE]; const char *netdev_name = netdev_get_name(&dev->up); @@ -594,6 +588,7 @@ dpdk_mp_create(struct netdev_dpdk *dev, int mtu) uint32_t n_mbufs; uint32_t hash = hash_string(netdev_name, 0); struct rte_mempool *mp = NULL; + uint16_t mbuf_size, aligned_mbuf_size, mbuf_priv_data_len; /* * XXX: rough estimation of number of mbufs required for this port: @@ -612,24 +607,46 @@ dpdk_mp_create(struct netdev_dpdk *dev, int mtu) /* Full DPDK memory pool name must be unique and cannot be * longer than RTE_MEMPOOL_NAMESIZE. */ int ret = snprintf(mp_name, RTE_MEMPOOL_NAMESIZE, - "ovs%08x%02d%05d%07u", - hash, socket_id, mtu, n_mbufs); + "ovs%08x%02d%05u%07u", + hash, socket_id, mbuf_pkt_data_len, n_mbufs); if (ret < 0 || ret >= RTE_MEMPOOL_NAMESIZE) { VLOG_DBG("snprintf returned %d. " "Failed to generate a mempool name for \"%s\". " - "Hash:0x%x, socket_id: %d, mtu:%d, mbufs:%u.", - ret, netdev_name, hash, socket_id, mtu, n_mbufs); + "Hash:0x%x, socket_id: %d, pkt data room:%u, mbufs:%u.", + ret, netdev_name, hash, socket_id, mbuf_pkt_data_len, + n_mbufs); break; } VLOG_DBG("Port %s: Requesting a mempool of %u mbufs " - "on socket %d for %d Rx and %d Tx queues.", + "on socket %d for %d Rx and %d Tx queues, " + "cache line size of %u", netdev_name, n_mbufs, socket_id, - dev->requested_n_rxq, dev->requested_n_txq); + dev->requested_n_rxq, dev->requested_n_txq, + RTE_CACHE_LINE_SIZE); + + mbuf_priv_data_len = sizeof(struct dp_packet) - + sizeof(struct rte_mbuf); + /* The size of the entire mbuf. */ + mbuf_size = sizeof (struct dp_packet) + + mbuf_pkt_data_len + RTE_PKTMBUF_HEADROOM; + /* mbuf size, rounded up to cacheline size. */ + aligned_mbuf_size = ROUND_UP(mbuf_size, RTE_CACHE_LINE_SIZE); + /* If there is a size discrepancy, add padding to mbuf_priv_data_len. + * This maintains mbuf size cache alignment, while also honoring RX + * buffer alignment in the data portion of the mbuf. If this adjustment + * is not made, there is a possiblity later on that for an element of + * the mempool, buf, buf->data_len < (buf->buf_len - buf->data_off). + * This is problematic in the case of multi-segment mbufs, particularly + * when an mbuf segment needs to be resized (when [push|popp]ing a VLAN + * header, for example. + */ + mbuf_priv_data_len += (aligned_mbuf_size - mbuf_size); mp = rte_pktmbuf_pool_create(mp_name, n_mbufs, MP_CACHE_SZ, - sizeof (struct dp_packet) - sizeof (struct rte_mbuf), - MBUF_SIZE(mtu) - sizeof(struct dp_packet), socket_id); + mbuf_priv_data_len, + mbuf_pkt_data_len + RTE_PKTMBUF_HEADROOM, + socket_id); if (mp) { VLOG_DBG("Allocated \"%s\" mempool with %u mbufs", @@ -694,13 +711,13 @@ static int netdev_dpdk_mempool_configure(struct netdev_dpdk *dev) OVS_REQUIRES(dev->mutex) { - uint32_t buf_size = dpdk_buf_size(dev->requested_mtu); + uint16_t buf_size = dpdk_buf_size(dev->requested_mtu); struct rte_mempool *mp; int ret = 0; dpdk_mp_sweep(); - mp = dpdk_mp_create(dev, FRAME_LEN_TO_MTU(buf_size)); + mp = dpdk_mp_create(dev, buf_size); if (!mp) { VLOG_ERR("Failed to create memory pool for netdev " "%s, with MTU %d on socket %d: %s\n", From patchwork Wed Jul 4 18:06:23 2018 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: "Lam, Tiago" X-Patchwork-Id: 939493 X-Patchwork-Delegate: ian.stokes@intel.com Return-Path: X-Original-To: incoming@patchwork.ozlabs.org Delivered-To: patchwork-incoming@bilbo.ozlabs.org Authentication-Results: ozlabs.org; spf=pass (mailfrom) smtp.mailfrom=openvswitch.org (client-ip=140.211.169.12; helo=mail.linuxfoundation.org; envelope-from=ovs-dev-bounces@openvswitch.org; receiver=) Authentication-Results: ozlabs.org; dmarc=fail (p=none dis=none) header.from=intel.com Received: from mail.linuxfoundation.org (mail.linuxfoundation.org [140.211.169.12]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by ozlabs.org (Postfix) with ESMTPS id 41LTV83KJzz9s29 for ; Thu, 5 Jul 2018 04:07:28 +1000 (AEST) Received: from mail.linux-foundation.org (localhost [127.0.0.1]) by mail.linuxfoundation.org (Postfix) with ESMTP id DD803CCC; Wed, 4 Jul 2018 18:06:50 +0000 (UTC) X-Original-To: ovs-dev@openvswitch.org Delivered-To: ovs-dev@mail.linuxfoundation.org Received: from smtp1.linuxfoundation.org (smtp1.linux-foundation.org [172.17.192.35]) by mail.linuxfoundation.org (Postfix) with ESMTPS id 435B6CA1 for ; Wed, 4 Jul 2018 18:06:49 +0000 (UTC) X-Greylist: domain auto-whitelisted by SQLgrey-1.7.6 Received: from mga05.intel.com (mga05.intel.com [192.55.52.43]) by smtp1.linuxfoundation.org (Postfix) with ESMTPS id D0338782 for ; Wed, 4 Jul 2018 18:06:48 +0000 (UTC) X-Amp-Result: SKIPPED(no attachment in message) X-Amp-File-Uploaded: False Received: from orsmga001.jf.intel.com ([10.7.209.18]) by fmsmga105.fm.intel.com with ESMTP/TLS/DHE-RSA-AES256-GCM-SHA384; 04 Jul 2018 11:06:47 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.51,308,1526367600"; d="scan'208";a="70109055" Received: from silpixa00399125.ir.intel.com ([10.237.223.34]) by orsmga001.jf.intel.com with ESMTP; 04 Jul 2018 11:06:45 -0700 From: Tiago Lam To: ovs-dev@openvswitch.org Date: Wed, 4 Jul 2018 19:06:23 +0100 Message-Id: <1530727595-217860-3-git-send-email-tiago.lam@intel.com> X-Mailer: git-send-email 2.7.4 In-Reply-To: <1530727595-217860-1-git-send-email-tiago.lam@intel.com> References: <1530727595-217860-1-git-send-email-tiago.lam@intel.com> X-Spam-Status: No, score=-4.2 required=5.0 tests=BAYES_00, RCVD_IN_DNSWL_MED autolearn=ham version=3.3.1 X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on smtp1.linux-foundation.org Cc: i.maximets@samsung.com Subject: [ovs-dev] [PATCH v2 02/14] dp-packet: Init specific mbuf fields. X-BeenThere: ovs-dev@openvswitch.org X-Mailman-Version: 2.1.12 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , MIME-Version: 1.0 Sender: ovs-dev-bounces@openvswitch.org Errors-To: ovs-dev-bounces@openvswitch.org From: Mark Kavanagh dp_packets are created using xmalloc(); in the case of OvS-DPDK, it's possible the the resultant mbuf portion of the dp_packet contains random data. For some mbuf fields, specifically those related to multi-segment mbufs and/or offload features, random values may cause unexpected behaviour, should the dp_packet's contents be later copied to a DPDK mbuf. It is critical therefore, that these fields should be initialized to 0. This patch ensures that the following mbuf fields are initialized to appropriate values on creation of a new dp_packet: - ol_flags=0 - nb_segs=1 - tx_offload=0 - packet_type=0 - next=NULL Adapted from an idea by Michael Qiu : https://patchwork.ozlabs.org/patch/777570/ Co-authored-by: Tiago Lam Signed-off-by: Mark Kavanagh Signed-off-by: Tiago Lam Acked-by: Eelco Chaudron --- lib/dp-packet.h | 9 +++++---- 1 file changed, 5 insertions(+), 4 deletions(-) diff --git a/lib/dp-packet.h b/lib/dp-packet.h index 596cfe6..cfe0824 100644 --- a/lib/dp-packet.h +++ b/lib/dp-packet.h @@ -625,14 +625,15 @@ dp_packet_mbuf_rss_flag_reset(struct dp_packet *p OVS_UNUSED) } /* This initialization is needed for packets that do not come - * from DPDK interfaces, when vswitchd is built with --with-dpdk. - * The DPDK rte library will still otherwise manage the mbuf. - * We only need to initialize the mbuf ol_flags. */ + * from DPDK interfaces, when vswitchd is built with --with-dpdk. */ static inline void dp_packet_mbuf_init(struct dp_packet *p OVS_UNUSED) { #ifdef DPDK_NETDEV - p->mbuf.ol_flags = 0; + struct rte_mbuf *mbuf = &(p->mbuf); + mbuf->ol_flags = mbuf->tx_offload = mbuf->packet_type = 0; + mbuf->nb_segs = 1; + mbuf->next = NULL; #endif } From patchwork Wed Jul 4 18:06:24 2018 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: "Lam, Tiago" X-Patchwork-Id: 939497 X-Patchwork-Delegate: ian.stokes@intel.com Return-Path: X-Original-To: incoming@patchwork.ozlabs.org Delivered-To: patchwork-incoming@bilbo.ozlabs.org Authentication-Results: ozlabs.org; spf=pass (mailfrom) smtp.mailfrom=openvswitch.org (client-ip=140.211.169.12; helo=mail.linuxfoundation.org; envelope-from=ovs-dev-bounces@openvswitch.org; receiver=) Authentication-Results: ozlabs.org; dmarc=fail (p=none dis=none) header.from=intel.com Received: from mail.linuxfoundation.org (mail.linuxfoundation.org [140.211.169.12]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by ozlabs.org (Postfix) with ESMTPS id 41LTWq2Fdwz9s29 for ; Thu, 5 Jul 2018 04:08:55 +1000 (AEST) Received: from mail.linux-foundation.org (localhost [127.0.0.1]) by mail.linuxfoundation.org (Postfix) with ESMTP id 5F58AD0B; Wed, 4 Jul 2018 18:07:04 +0000 (UTC) X-Original-To: ovs-dev@openvswitch.org Delivered-To: ovs-dev@mail.linuxfoundation.org Received: from smtp1.linuxfoundation.org (smtp1.linux-foundation.org [172.17.192.35]) by mail.linuxfoundation.org (Postfix) with ESMTPS id 0C5E3CF5 for ; Wed, 4 Jul 2018 18:07:03 +0000 (UTC) X-Greylist: domain auto-whitelisted by SQLgrey-1.7.6 Received: from mga12.intel.com (mga12.intel.com [192.55.52.136]) by smtp1.linuxfoundation.org (Postfix) with ESMTPS id 9AAE3780 for ; Wed, 4 Jul 2018 18:07:02 +0000 (UTC) X-Amp-Result: SKIPPED(no attachment in message) X-Amp-File-Uploaded: False Received: from orsmga001.jf.intel.com ([10.7.209.18]) by fmsmga106.fm.intel.com with ESMTP/TLS/DHE-RSA-AES256-GCM-SHA384; 04 Jul 2018 11:07:01 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.51,308,1526367600"; d="scan'208";a="70109060" Received: from silpixa00399125.ir.intel.com ([10.237.223.34]) by orsmga001.jf.intel.com with ESMTP; 04 Jul 2018 11:06:47 -0700 From: Tiago Lam To: ovs-dev@openvswitch.org Date: Wed, 4 Jul 2018 19:06:24 +0100 Message-Id: <1530727595-217860-4-git-send-email-tiago.lam@intel.com> X-Mailer: git-send-email 2.7.4 In-Reply-To: <1530727595-217860-1-git-send-email-tiago.lam@intel.com> References: <1530727595-217860-1-git-send-email-tiago.lam@intel.com> X-Spam-Status: No, score=-4.2 required=5.0 tests=BAYES_00, RCVD_IN_DNSWL_MED autolearn=ham version=3.3.1 X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on smtp1.linux-foundation.org Cc: i.maximets@samsung.com Subject: [ovs-dev] [PATCH v2 03/14] dp-packet: Fix allocated size on DPDK init. X-BeenThere: ovs-dev@openvswitch.org X-Mailman-Version: 2.1.12 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , MIME-Version: 1.0 Sender: ovs-dev-bounces@openvswitch.org Errors-To: ovs-dev-bounces@openvswitch.org When enabled with DPDK OvS deals with two types of packets, the ones coming from the mempool and the ones locally created by OvS - which are copied to mempool mbufs before output. In the latter, the space is allocated from the system, while in the former the mbufs are allocated from a mempool, which takes care of initialising them appropriately. In the current implementation, during mempool's initialisation of mbufs, dp_packet_set_allocated() is called from dp_packet_init_dpdk() without considering that the allocated space, in the case of multi-segment mbufs, might be greater than a single mbuf. Furthermore, given that dp_packet_init_dpdk() is on the code path that's called upon mempool's initialisation, a call to dp_packet_set_allocated() is redundant, since mempool takes care of initialising it. To fix this, dp_packet_set_allocated() is no longer called after initialisation of a mempool, only in dp_packet_init__(), which is still called by OvS when initialising locally created packets. Signed-off-by: Tiago Lam Acked-by: Eelco Chaudron --- lib/dp-packet.c | 3 +-- lib/dp-packet.h | 2 +- lib/netdev-dpdk.c | 2 +- 3 files changed, 3 insertions(+), 4 deletions(-) diff --git a/lib/dp-packet.c b/lib/dp-packet.c index 443c225..782e7c2 100644 --- a/lib/dp-packet.c +++ b/lib/dp-packet.c @@ -99,9 +99,8 @@ dp_packet_use_const(struct dp_packet *b, const void *data, size_t size) * buffer. Here, non-transient ovs dp-packet fields are initialized for * packets that are part of a DPDK memory pool. */ void -dp_packet_init_dpdk(struct dp_packet *b, size_t allocated) +dp_packet_init_dpdk(struct dp_packet *b) { - dp_packet_set_allocated(b, allocated); b->source = DPBUF_DPDK; } diff --git a/lib/dp-packet.h b/lib/dp-packet.h index cfe0824..c612ad4 100644 --- a/lib/dp-packet.h +++ b/lib/dp-packet.h @@ -114,7 +114,7 @@ void dp_packet_use(struct dp_packet *, void *, size_t); void dp_packet_use_stub(struct dp_packet *, void *, size_t); void dp_packet_use_const(struct dp_packet *, const void *, size_t); -void dp_packet_init_dpdk(struct dp_packet *, size_t allocated); +void dp_packet_init_dpdk(struct dp_packet *); void dp_packet_init(struct dp_packet *, size_t); void dp_packet_uninit(struct dp_packet *); diff --git a/lib/netdev-dpdk.c b/lib/netdev-dpdk.c index 8fb859e..675dd48 100644 --- a/lib/netdev-dpdk.c +++ b/lib/netdev-dpdk.c @@ -520,7 +520,7 @@ ovs_rte_pktmbuf_init(struct rte_mempool *mp OVS_UNUSED, { struct rte_mbuf *pkt = _p; - dp_packet_init_dpdk((struct dp_packet *) pkt, pkt->buf_len); + dp_packet_init_dpdk((struct dp_packet *) pkt); } static int From patchwork Wed Jul 4 18:06:25 2018 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: "Lam, Tiago" X-Patchwork-Id: 939496 X-Patchwork-Delegate: ian.stokes@intel.com Return-Path: X-Original-To: incoming@patchwork.ozlabs.org Delivered-To: patchwork-incoming@bilbo.ozlabs.org Authentication-Results: ozlabs.org; spf=pass (mailfrom) smtp.mailfrom=openvswitch.org (client-ip=140.211.169.12; helo=mail.linuxfoundation.org; envelope-from=ovs-dev-bounces@openvswitch.org; receiver=) Authentication-Results: ozlabs.org; dmarc=fail (p=none dis=none) header.from=intel.com Received: from mail.linuxfoundation.org (mail.linuxfoundation.org [140.211.169.12]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by ozlabs.org (Postfix) with ESMTPS id 41LTW71CZSz9s29 for ; Thu, 5 Jul 2018 04:08:19 +1000 (AEST) Received: from mail.linux-foundation.org (localhost [127.0.0.1]) by mail.linuxfoundation.org (Postfix) with ESMTP id 56D44CF5; Wed, 4 Jul 2018 18:07:03 +0000 (UTC) X-Original-To: ovs-dev@openvswitch.org Delivered-To: ovs-dev@mail.linuxfoundation.org Received: from smtp1.linuxfoundation.org (smtp1.linux-foundation.org [172.17.192.35]) by mail.linuxfoundation.org (Postfix) with ESMTPS id 930A4CF2 for ; Wed, 4 Jul 2018 18:07:02 +0000 (UTC) X-Greylist: domain auto-whitelisted by SQLgrey-1.7.6 Received: from mga12.intel.com (mga12.intel.com [192.55.52.136]) by smtp1.linuxfoundation.org (Postfix) with ESMTPS id 35E41780 for ; Wed, 4 Jul 2018 18:07:02 +0000 (UTC) X-Amp-Result: SKIPPED(no attachment in message) X-Amp-File-Uploaded: False Received: from orsmga001.jf.intel.com ([10.7.209.18]) by fmsmga106.fm.intel.com with ESMTP/TLS/DHE-RSA-AES256-GCM-SHA384; 04 Jul 2018 11:07:01 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.51,308,1526367600"; d="scan'208";a="70109064" Received: from silpixa00399125.ir.intel.com ([10.237.223.34]) by orsmga001.jf.intel.com with ESMTP; 04 Jul 2018 11:06:49 -0700 From: Tiago Lam To: ovs-dev@openvswitch.org Date: Wed, 4 Jul 2018 19:06:25 +0100 Message-Id: <1530727595-217860-5-git-send-email-tiago.lam@intel.com> X-Mailer: git-send-email 2.7.4 In-Reply-To: <1530727595-217860-1-git-send-email-tiago.lam@intel.com> References: <1530727595-217860-1-git-send-email-tiago.lam@intel.com> X-Spam-Status: No, score=-4.2 required=5.0 tests=BAYES_00, RCVD_IN_DNSWL_MED autolearn=ham version=3.3.1 X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on smtp1.linux-foundation.org Cc: i.maximets@samsung.com Subject: [ovs-dev] [PATCH v2 04/14] netdev-dpdk: Serialise non-pmds mbufs' alloc/free. X-BeenThere: ovs-dev@openvswitch.org X-Mailman-Version: 2.1.12 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , MIME-Version: 1.0 Sender: ovs-dev-bounces@openvswitch.org Errors-To: ovs-dev-bounces@openvswitch.org A new mutex, 'nonpmd_mp_mutex', has been introduced to serialise allocation and free operations by non-pmd threads on a given mempool. free_dpdk_buf() has been modified to make use of the introduced mutex. Signed-off-by: Tiago Lam Acked-by: Eelco Chaudron --- lib/netdev-dpdk.c | 30 +++++++++++++++++++++++++++--- 1 file changed, 27 insertions(+), 3 deletions(-) diff --git a/lib/netdev-dpdk.c b/lib/netdev-dpdk.c index 675dd48..6c9f0e7 100644 --- a/lib/netdev-dpdk.c +++ b/lib/netdev-dpdk.c @@ -294,6 +294,16 @@ static struct ovs_mutex dpdk_mp_mutex OVS_ACQ_AFTER(dpdk_mutex) static struct ovs_list dpdk_mp_free_list OVS_GUARDED_BY(dpdk_mp_mutex) = OVS_LIST_INITIALIZER(&dpdk_mp_free_list); +/* This mutex must be used by non pmd threads when allocating or freeing + * mbufs through mempools, when outside of the `non_pmd_mutex` mutex, in struct + * dp_netdev. + * The reason, as pointed out in the "Known Issues" section in DPDK's EAL docs, + * is that the implementation on which mempool is based off is non-preemptable. + * Since non-pmds may end up not being pinned this could lead to the preemption + * between non-pmds performing operations on the same mempool, which could lead + * to memory corruption. */ +static struct ovs_mutex nonpmd_mp_mutex = OVS_MUTEX_INITIALIZER; + /* Wrapper for a mempool released but not yet freed. */ struct dpdk_mp { struct rte_mempool *mp; @@ -462,6 +472,8 @@ struct netdev_rxq_dpdk { dpdk_port_t port_id; }; +static bool dpdk_thread_is_pmd(void); + static void netdev_dpdk_destruct(struct netdev *netdev); static void netdev_dpdk_vhost_destruct(struct netdev *netdev); @@ -495,6 +507,12 @@ dpdk_buf_size(int mtu) NETDEV_DPDK_MBUF_ALIGN); } +static bool +dpdk_thread_is_pmd(void) +{ + return rte_lcore_id() != NON_PMD_CORE_ID; +} + /* Allocates an area of 'sz' bytes from DPDK. The memory is zero'ed. * * Unlike xmalloc(), this function can return NULL on failure. */ @@ -505,11 +523,17 @@ dpdk_rte_mzalloc(size_t sz) } void -free_dpdk_buf(struct dp_packet *p) +free_dpdk_buf(struct dp_packet *packet) { - struct rte_mbuf *pkt = (struct rte_mbuf *) p; + if (!dpdk_thread_is_pmd()) { + ovs_mutex_lock(&nonpmd_mp_mutex); + } - rte_pktmbuf_free(pkt); + rte_pktmbuf_free(&packet->mbuf); + + if (!dpdk_thread_is_pmd()) { + ovs_mutex_unlock(&nonpmd_mp_mutex); + } } static void From patchwork Wed Jul 4 18:06:26 2018 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: "Lam, Tiago" X-Patchwork-Id: 939498 X-Patchwork-Delegate: ian.stokes@intel.com Return-Path: X-Original-To: incoming@patchwork.ozlabs.org Delivered-To: patchwork-incoming@bilbo.ozlabs.org Authentication-Results: ozlabs.org; spf=pass (mailfrom) smtp.mailfrom=openvswitch.org (client-ip=140.211.169.12; helo=mail.linuxfoundation.org; envelope-from=ovs-dev-bounces@openvswitch.org; receiver=) Authentication-Results: ozlabs.org; dmarc=fail (p=none dis=none) header.from=intel.com Received: from mail.linuxfoundation.org (mail.linuxfoundation.org [140.211.169.12]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by ozlabs.org (Postfix) with ESMTPS id 41LTXL1X0zz9s29 for ; Thu, 5 Jul 2018 04:09:22 +1000 (AEST) Received: from mail.linux-foundation.org (localhost [127.0.0.1]) by mail.linuxfoundation.org (Postfix) with ESMTP id 2F857CFF; Wed, 4 Jul 2018 18:07:05 +0000 (UTC) X-Original-To: ovs-dev@openvswitch.org Delivered-To: ovs-dev@mail.linuxfoundation.org Received: from smtp1.linuxfoundation.org (smtp1.linux-foundation.org [172.17.192.35]) by mail.linuxfoundation.org (Postfix) with ESMTPS id 94175CF9 for ; Wed, 4 Jul 2018 18:07:03 +0000 (UTC) X-Greylist: domain auto-whitelisted by SQLgrey-1.7.6 Received: from mga12.intel.com (mga12.intel.com [192.55.52.136]) by smtp1.linuxfoundation.org (Postfix) with ESMTPS id 14494784 for ; Wed, 4 Jul 2018 18:07:03 +0000 (UTC) X-Amp-Result: SKIPPED(no attachment in message) X-Amp-File-Uploaded: False Received: from orsmga001.jf.intel.com ([10.7.209.18]) by fmsmga106.fm.intel.com with ESMTP/TLS/DHE-RSA-AES256-GCM-SHA384; 04 Jul 2018 11:07:02 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.51,308,1526367600"; d="scan'208";a="70109069" Received: from silpixa00399125.ir.intel.com ([10.237.223.34]) by orsmga001.jf.intel.com with ESMTP; 04 Jul 2018 11:06:50 -0700 From: Tiago Lam To: ovs-dev@openvswitch.org Date: Wed, 4 Jul 2018 19:06:26 +0100 Message-Id: <1530727595-217860-6-git-send-email-tiago.lam@intel.com> X-Mailer: git-send-email 2.7.4 In-Reply-To: <1530727595-217860-1-git-send-email-tiago.lam@intel.com> References: <1530727595-217860-1-git-send-email-tiago.lam@intel.com> X-Spam-Status: No, score=-4.2 required=5.0 tests=BAYES_00, RCVD_IN_DNSWL_MED autolearn=ham version=3.3.1 X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on smtp1.linux-foundation.org Cc: Marcin Ksiadz , Przemyslaw Lal , Michael Qiu , i.maximets@samsung.com Subject: [ovs-dev] [PATCH v2 05/14] dp-packet: Fix data_len handling multi-seg mbufs. X-BeenThere: ovs-dev@openvswitch.org X-Mailman-Version: 2.1.12 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , MIME-Version: 1.0 Sender: ovs-dev-bounces@openvswitch.org Errors-To: ovs-dev-bounces@openvswitch.org When a dp_packet is from a DPDK source, and it contains multi-segment mbufs, the data_len is not equal to the packet size, pkt_len. Instead, the data_len of each mbuf in the chain should be considered while distributing the new (provided) size. To account for the above dp_packet_set_size() has been changed so that, in the multi-segment mbufs case, only the data_len on the last mbuf of the chain and the total size of the packet, pkt_len, are changed. The data_len on the intermediate mbufs preceeding the last mbuf is not changed by dp_packet_set_size(). Furthermore, in some cases dp_packet_set_size() may be used to set a smaller size than the current packet size, thus effectively trimming the end of the packet. In the multi-segment mbufs case this may lead to lingering mbufs that may need freeing. __dp_packet_set_data() now also updates an mbufs' data_len after setting the data offset. This is so that both fields are always in sync for each mbuf in a chain. Co-authored-by: Michael Qiu Co-authored-by: Mark Kavanagh Co-authored-by: Przemyslaw Lal Co-authored-by: Marcin Ksiadz Co-authored-by: Yuanhan Liu Signed-off-by: Michael Qiu Signed-off-by: Mark Kavanagh Signed-off-by: Przemyslaw Lal Signed-off-by: Marcin Ksiadz Signed-off-by: Yuanhan Liu Signed-off-by: Tiago Lam --- lib/dp-packet.h | 76 ++++++++++++++++++++++++++++++++++++++++++++++++--------- 1 file changed, 64 insertions(+), 12 deletions(-) diff --git a/lib/dp-packet.h b/lib/dp-packet.h index c612ad4..fcbaf60 100644 --- a/lib/dp-packet.h +++ b/lib/dp-packet.h @@ -429,17 +429,49 @@ dp_packet_size(const struct dp_packet *b) static inline void dp_packet_set_size(struct dp_packet *b, uint32_t v) { - /* netdev-dpdk does not currently support segmentation; consequently, for - * all intents and purposes, 'data_len' (16 bit) and 'pkt_len' (32 bit) may - * be used interchangably. - * - * On the datapath, it is expected that the size of packets - * (and thus 'v') will always be <= UINT16_MAX; this means that there is no - * loss of accuracy in assigning 'v' to 'data_len'. - */ - b->mbuf.data_len = (uint16_t)v; /* Current seg length. */ - b->mbuf.pkt_len = v; /* Total length of all segments linked to - * this segment. */ + if (b->source == DPBUF_DPDK) { + struct rte_mbuf *mbuf = &b->mbuf; + uint16_t new_len = v; + uint16_t data_len; + uint16_t nb_segs = 0; + uint16_t pkt_len = 0; + + /* Trim 'v' length bytes from the end of the chained buffers, freeing + any buffers that may be left floating */ + while (mbuf) { + data_len = MIN(new_len, mbuf->data_len); + mbuf->data_len = data_len; + + if (new_len - data_len <= 0) { + /* Free the rest of chained mbufs */ + free_dpdk_buf(CONTAINER_OF(mbuf->next, struct dp_packet, + mbuf)); + mbuf->next = NULL; + } else if (!mbuf->next) { + /* Don't assign more than what we have available */ + mbuf->data_len = MIN(new_len, + mbuf->buf_len - mbuf->data_off); + } + + new_len -= data_len; + nb_segs += 1; + pkt_len += mbuf->data_len; + mbuf = mbuf->next; + } + + /* pkt_len != v would effectively mean that pkt_len < than 'v' (as + * being bigger is logically impossible). Being < than 'v' would mean + * the 'v' provided was bigger than the available room, which is the + * responsibility of the caller to make sure there is enough room */ + ovs_assert(pkt_len == v); + + b->mbuf.nb_segs = nb_segs; + b->mbuf.pkt_len = pkt_len; + } else { + b->mbuf.data_len = v; + /* Total length of all segments linked to this segment. */ + b->mbuf.pkt_len = v; + } } static inline uint16_t @@ -451,7 +483,27 @@ __packet_data(const struct dp_packet *b) static inline void __packet_set_data(struct dp_packet *b, uint16_t v) { - b->mbuf.data_off = v; + if (b->source == DPBUF_DPDK) { + /* Moving data_off away from the first mbuf in the chain is not a + * possibility using DPBUF_DPDK dp_packets */ + ovs_assert(v == UINT16_MAX || v <= b->mbuf.buf_len - b->mbuf.data_off); + + uint16_t prev_ofs = b->mbuf.data_off; + b->mbuf.data_off = v; + int16_t ofs_diff = prev_ofs - b->mbuf.data_off; + + /* When dealing with DPDK mbufs, keep data_off and data_len in sync. + * Thus, update data_len if the length changes with the move of + * data_off. However, if data_len is 0, there's no data to move and + * data_len should remain 0. */ + + if (b->mbuf.data_len != 0) { + b->mbuf.data_len = MIN(b->mbuf.data_len + ofs_diff, + b->mbuf.buf_len - b->mbuf.data_off); + } + } else { + b->mbuf.data_off = v; + } } static inline uint16_t From patchwork Wed Jul 4 18:06:27 2018 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: "Lam, Tiago" X-Patchwork-Id: 939508 X-Patchwork-Delegate: ian.stokes@intel.com Return-Path: X-Original-To: incoming@patchwork.ozlabs.org Delivered-To: patchwork-incoming@bilbo.ozlabs.org Authentication-Results: ozlabs.org; spf=pass (mailfrom) smtp.mailfrom=openvswitch.org (client-ip=140.211.169.12; helo=mail.linuxfoundation.org; envelope-from=ovs-dev-bounces@openvswitch.org; receiver=) Authentication-Results: ozlabs.org; dmarc=fail (p=none dis=none) header.from=intel.com Received: from mail.linuxfoundation.org (mail.linuxfoundation.org [140.211.169.12]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by ozlabs.org (Postfix) with ESMTPS id 41LTZj4xv0z9s29 for ; Thu, 5 Jul 2018 04:11:25 +1000 (AEST) Received: from mail.linux-foundation.org (localhost [127.0.0.1]) by mail.linuxfoundation.org (Postfix) with ESMTP id D89C4D48; Wed, 4 Jul 2018 18:07:08 +0000 (UTC) X-Original-To: ovs-dev@openvswitch.org Delivered-To: ovs-dev@mail.linuxfoundation.org Received: from smtp1.linuxfoundation.org (smtp1.linux-foundation.org [172.17.192.35]) by mail.linuxfoundation.org (Postfix) with ESMTPS id E45D1D0D for ; Wed, 4 Jul 2018 18:07:04 +0000 (UTC) X-Greylist: domain auto-whitelisted by SQLgrey-1.7.6 Received: from mga12.intel.com (mga12.intel.com [192.55.52.136]) by smtp1.linuxfoundation.org (Postfix) with ESMTPS id 320E4784 for ; Wed, 4 Jul 2018 18:07:04 +0000 (UTC) X-Amp-Result: SKIPPED(no attachment in message) X-Amp-File-Uploaded: False Received: from orsmga001.jf.intel.com ([10.7.209.18]) by fmsmga106.fm.intel.com with ESMTP/TLS/DHE-RSA-AES256-GCM-SHA384; 04 Jul 2018 11:07:02 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.51,308,1526367600"; d="scan'208";a="70109073" Received: from silpixa00399125.ir.intel.com ([10.237.223.34]) by orsmga001.jf.intel.com with ESMTP; 04 Jul 2018 11:06:53 -0700 From: Tiago Lam To: ovs-dev@openvswitch.org Date: Wed, 4 Jul 2018 19:06:27 +0100 Message-Id: <1530727595-217860-7-git-send-email-tiago.lam@intel.com> X-Mailer: git-send-email 2.7.4 In-Reply-To: <1530727595-217860-1-git-send-email-tiago.lam@intel.com> References: <1530727595-217860-1-git-send-email-tiago.lam@intel.com> X-Spam-Status: No, score=-4.2 required=5.0 tests=BAYES_00, RCVD_IN_DNSWL_MED autolearn=ham version=3.3.1 X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on smtp1.linux-foundation.org Cc: i.maximets@samsung.com Subject: [ovs-dev] [PATCH v2 06/14] dp-packet: Handle multi-seg mbufs in helper funcs. X-BeenThere: ovs-dev@openvswitch.org X-Mailman-Version: 2.1.12 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , MIME-Version: 1.0 Sender: ovs-dev-bounces@openvswitch.org Errors-To: ovs-dev-bounces@openvswitch.org Most helper functions in dp-packet assume that the data held by a dp_packet is contiguous, and perform operations such as pointer arithmetic under that assumption. However, with the introduction of multi-segment mbufs, where data is non-contiguous, such assumptions are no longer possible. Some examples of Such helper functions are dp_packet_tail(), dp_packet_tailroom(), dp_packet_end(), dp_packet_get_allocated() and dp_packet_at(). Thus, instead of assuming contiguous data in dp_packet, they now iterate over the (non-contiguous) data in mbufs to perform their calculations. Finally, dp_packet_use__() has also been modified to perform the initialisation of the packet (and setting the source) before continuing to set its size and data length, which now depends on the type of packet. Co-authored-by: Mark Kavanagh Signed-off-by: Mark Kavanagh Signed-off-by: Tiago Lam Acked-by: Eelco Chaudron --- lib/dp-packet.c | 4 +- lib/dp-packet.h | 114 +++++++++++++++++++++++++++++++++++++++++++++++++++++--- 2 files changed, 111 insertions(+), 7 deletions(-) diff --git a/lib/dp-packet.c b/lib/dp-packet.c index 782e7c2..2aaeaae 100644 --- a/lib/dp-packet.c +++ b/lib/dp-packet.c @@ -41,11 +41,11 @@ static void dp_packet_use__(struct dp_packet *b, void *base, size_t allocated, enum dp_packet_source source) { + dp_packet_init__(b, allocated, source); + dp_packet_set_base(b, base); dp_packet_set_data(b, base); dp_packet_set_size(b, 0); - - dp_packet_init__(b, allocated, source); } /* Initializes 'b' as an empty dp_packet that contains the 'allocated' bytes of diff --git a/lib/dp-packet.h b/lib/dp-packet.h index fcbaf60..4880fed 100644 --- a/lib/dp-packet.h +++ b/lib/dp-packet.h @@ -185,9 +185,25 @@ dp_packet_delete(struct dp_packet *b) static inline void * dp_packet_at(const struct dp_packet *b, size_t offset, size_t size) { - return offset + size <= dp_packet_size(b) - ? (char *) dp_packet_data(b) + offset - : NULL; + if (offset + size > dp_packet_size(b)) { + return NULL; + } + +#ifdef DPDK_NETDEV + if (b->source == DPBUF_DPDK) { + struct rte_mbuf *buf = CONST_CAST(struct rte_mbuf *, &b->mbuf); + + while (buf && offset > buf->data_len) { + offset -= buf->data_len; + + buf = buf->next; + } + + return buf ? rte_pktmbuf_mtod_offset(buf, char *, offset) : NULL; + } +#endif + + return (char *) dp_packet_data(b) + offset; } /* Returns a pointer to byte 'offset' in 'b', which must contain at least @@ -196,13 +212,23 @@ static inline void * dp_packet_at_assert(const struct dp_packet *b, size_t offset, size_t size) { ovs_assert(offset + size <= dp_packet_size(b)); - return ((char *) dp_packet_data(b)) + offset; + return dp_packet_at(b, offset, size); } /* Returns a pointer to byte following the last byte of data in use in 'b'. */ static inline void * dp_packet_tail(const struct dp_packet *b) { +#ifdef DPDK_NETDEV + if (b->source == DPBUF_DPDK) { + struct rte_mbuf *buf = CONST_CAST(struct rte_mbuf *, &b->mbuf); + /* Find last segment where data ends, meaning the tail of the chained + * mbufs must be there */ + buf = rte_pktmbuf_lastseg(buf); + + return rte_pktmbuf_mtod_offset(buf, void *, buf->data_len); + } +#endif return (char *) dp_packet_data(b) + dp_packet_size(b); } @@ -211,6 +237,15 @@ dp_packet_tail(const struct dp_packet *b) static inline void * dp_packet_end(const struct dp_packet *b) { +#ifdef DPDK_NETDEV + if (b->source == DPBUF_DPDK) { + struct rte_mbuf *buf = CONST_CAST(struct rte_mbuf *, &(b->mbuf)); + + buf = rte_pktmbuf_lastseg(buf); + + return (char *) buf->buf_addr + buf->buf_len; + } +#endif return (char *) dp_packet_base(b) + dp_packet_get_allocated(b); } @@ -236,6 +271,15 @@ dp_packet_tailroom(const struct dp_packet *b) static inline void dp_packet_clear(struct dp_packet *b) { +#ifdef DPDK_NETDEV + if (b->source == DPBUF_DPDK) { + /* sets pkt_len and data_len to zero and frees unused mbufs */ + dp_packet_set_size(b, 0); + rte_pktmbuf_reset(&b->mbuf); + + return; + } +#endif dp_packet_set_data(b, dp_packet_base(b)); dp_packet_set_size(b, 0); } @@ -252,12 +296,33 @@ dp_packet_pull(struct dp_packet *b, size_t size) return data; } +#ifdef DPDK_NETDEV +/* Similar to dp_packet_try_pull() but doesn't actually pull any data, only + * checks if it could and returns true or false accordingly. + * + * Valid for dp_packets carrying mbufs only. */ +static inline bool +dp_packet_mbuf_may_pull(const struct dp_packet *b, size_t size) { + if (size > b->mbuf.data_len) { + return false; + } + + return true; +} +#endif + /* If 'b' has at least 'size' bytes of data, removes that many bytes from the * head end of 'b' and returns the first byte removed. Otherwise, returns a * null pointer without modifying 'b'. */ static inline void * dp_packet_try_pull(struct dp_packet *b, size_t size) { +#ifdef DPDK_NETDEV + if (!dp_packet_mbuf_may_pull(b, size)) { + return NULL; + } +#endif + return dp_packet_size(b) - dp_packet_l2_pad_size(b) >= size ? dp_packet_pull(b, size) : NULL; } @@ -311,6 +376,12 @@ dp_packet_set_l2_pad_size(struct dp_packet *b, uint8_t pad_size) static inline void * dp_packet_l2_5(const struct dp_packet *b) { +#ifdef DPDK_NETDEV + if (!dp_packet_mbuf_may_pull(b, b->l2_5_ofs)) { + return NULL; + } +#endif + return b->l2_5_ofs != UINT16_MAX ? (char *) dp_packet_data(b) + b->l2_5_ofs : NULL; @@ -327,6 +398,12 @@ dp_packet_set_l2_5(struct dp_packet *b, void *l2_5) static inline void * dp_packet_l3(const struct dp_packet *b) { +#ifdef DPDK_NETDEV + if (!dp_packet_mbuf_may_pull(b, b->l3_ofs)) { + return NULL; + } +#endif + return b->l3_ofs != UINT16_MAX ? (char *) dp_packet_data(b) + b->l3_ofs : NULL; @@ -341,6 +418,12 @@ dp_packet_set_l3(struct dp_packet *b, void *l3) static inline void * dp_packet_l4(const struct dp_packet *b) { +#ifdef DPDK_NETDEV + if (!dp_packet_mbuf_may_pull(b, b->l4_ofs)) { + return NULL; + } +#endif + return b->l4_ofs != UINT16_MAX ? (char *) dp_packet_data(b) + b->l4_ofs : NULL; @@ -355,6 +438,27 @@ dp_packet_set_l4(struct dp_packet *b, void *l4) static inline size_t dp_packet_l4_size(const struct dp_packet *b) { +#ifdef DPDK_NETDEV + if (b->source == DPBUF_DPDK) { + if (!dp_packet_mbuf_may_pull(b, b->l4_ofs)) { + return 0; + } + + struct rte_mbuf *mbuf = CONST_CAST(struct rte_mbuf *, &b->mbuf); + size_t l4_size = mbuf->data_len - b->l4_ofs; + + mbuf = mbuf->next; + while (mbuf) { + l4_size += mbuf->data_len; + + mbuf = mbuf->next; + } + + l4_size -= dp_packet_l2_pad_size(b); + + return l4_size; + } +#endif return b->l4_ofs != UINT16_MAX ? (const char *)dp_packet_tail(b) - (const char *)dp_packet_l4(b) - dp_packet_l2_pad_size(b) @@ -509,7 +613,7 @@ __packet_set_data(struct dp_packet *b, uint16_t v) static inline uint16_t dp_packet_get_allocated(const struct dp_packet *b) { - return b->mbuf.buf_len; + return b->mbuf.nb_segs * b->mbuf.buf_len; } static inline void From patchwork Wed Jul 4 18:06:28 2018 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: "Lam, Tiago" X-Patchwork-Id: 939499 X-Patchwork-Delegate: ian.stokes@intel.com Return-Path: X-Original-To: incoming@patchwork.ozlabs.org Delivered-To: patchwork-incoming@bilbo.ozlabs.org Authentication-Results: ozlabs.org; spf=pass (mailfrom) smtp.mailfrom=openvswitch.org (client-ip=140.211.169.12; helo=mail.linuxfoundation.org; envelope-from=ovs-dev-bounces@openvswitch.org; receiver=) Authentication-Results: ozlabs.org; dmarc=fail (p=none dis=none) header.from=intel.com Received: from mail.linuxfoundation.org (mail.linuxfoundation.org [140.211.169.12]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by ozlabs.org (Postfix) with ESMTPS id 41LTXq5PY7z9s4r for ; Thu, 5 Jul 2018 04:09:47 +1000 (AEST) Received: from mail.linux-foundation.org (localhost [127.0.0.1]) by mail.linuxfoundation.org (Postfix) with ESMTP id E6207D23; Wed, 4 Jul 2018 18:07:05 +0000 (UTC) X-Original-To: ovs-dev@openvswitch.org Delivered-To: ovs-dev@mail.linuxfoundation.org Received: from smtp1.linuxfoundation.org (smtp1.linux-foundation.org [172.17.192.35]) by mail.linuxfoundation.org (Postfix) with ESMTPS id CF440CF9 for ; Wed, 4 Jul 2018 18:07:03 +0000 (UTC) X-Greylist: domain auto-whitelisted by SQLgrey-1.7.6 Received: from mga12.intel.com (mga12.intel.com [192.55.52.136]) by smtp1.linuxfoundation.org (Postfix) with ESMTPS id 478C2780 for ; Wed, 4 Jul 2018 18:07:03 +0000 (UTC) X-Amp-Result: SKIPPED(no attachment in message) X-Amp-File-Uploaded: False Received: from orsmga001.jf.intel.com ([10.7.209.18]) by fmsmga106.fm.intel.com with ESMTP/TLS/DHE-RSA-AES256-GCM-SHA384; 04 Jul 2018 11:07:02 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.51,308,1526367600"; d="scan'208";a="70109078" Received: from silpixa00399125.ir.intel.com ([10.237.223.34]) by orsmga001.jf.intel.com with ESMTP; 04 Jul 2018 11:06:54 -0700 From: Tiago Lam To: ovs-dev@openvswitch.org Date: Wed, 4 Jul 2018 19:06:28 +0100 Message-Id: <1530727595-217860-8-git-send-email-tiago.lam@intel.com> X-Mailer: git-send-email 2.7.4 In-Reply-To: <1530727595-217860-1-git-send-email-tiago.lam@intel.com> References: <1530727595-217860-1-git-send-email-tiago.lam@intel.com> X-Spam-Status: No, score=-4.2 required=5.0 tests=BAYES_00, RCVD_IN_DNSWL_MED autolearn=ham version=3.3.1 X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on smtp1.linux-foundation.org Cc: i.maximets@samsung.com Subject: [ovs-dev] [PATCH v2 07/14] dp-packet: Handle multi-seg mubfs in shift() func. X-BeenThere: ovs-dev@openvswitch.org X-Mailman-Version: 2.1.12 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , MIME-Version: 1.0 Sender: ovs-dev-bounces@openvswitch.org Errors-To: ovs-dev-bounces@openvswitch.org In its current implementation dp_packet_shift() is also unaware of multi-seg mbufs (that holds data in memory non-contiguously) and assumes that data exists contiguously in memory, memmove'ing data to perform the shift. To add support for multi-seg mbuds a new set of functions was introduced, dp_packet_mbuf_shift() and dp_packet_mbuf_write(). These functions are used by dp_packet_shift(), when handling multi-seg mbufs, to shift and write data within a chain of mbufs. Signed-off-by: Tiago Lam Acked-by: Eelco Chaudron --- lib/dp-packet.c | 97 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++ lib/dp-packet.h | 10 ++++++ 2 files changed, 107 insertions(+) diff --git a/lib/dp-packet.c b/lib/dp-packet.c index 2aaeaae..d6e19eb 100644 --- a/lib/dp-packet.c +++ b/lib/dp-packet.c @@ -294,6 +294,97 @@ dp_packet_prealloc_headroom(struct dp_packet *b, size_t size) } } +#ifdef DPDK_NETDEV +/* Write len data bytes in a mbuf at specified offset. + * + * 'mbuf', pointer to the destination mbuf where 'ofs' is, and the mbuf where + * the data will first be written. + * 'ofs', the offset within the provided 'mbuf' where 'data' is to be written. + * 'len', the size of the to be written 'data'. + * 'data', pointer to the to be written bytes. + * + * XXX: This function is the counterpart of the `rte_pktmbuf_read()` function + * available with DPDK, in the rte_mbuf.h */ +void +dp_packet_mbuf_write(struct rte_mbuf *mbuf, int16_t ofs, uint32_t len, + const void *data) +{ + char *dst_addr; + uint16_t data_len; + int len_copy; + while (mbuf) { + if (len == 0) { + break; + } + + dst_addr = rte_pktmbuf_mtod_offset(mbuf, char *, ofs); + data_len = MBUF_BUF_END(mbuf->buf_addr, mbuf->buf_len) - dst_addr; + + len_copy = MIN(len, data_len); + /* We don't know if 'data' is the result of a rte_pktmbuf_read() call, + * in which case we may end up writing to the same region of memory we + * are reading from and overlapping. Hence the use of memmove() here */ + memmove(dst_addr, data, len_copy); + + data = ((char *) data) + len_copy; + len -= len_copy; + ofs = 0; + + mbuf = mbuf->next; + } +} + +static void +dp_packet_mbuf_shift_(struct rte_mbuf *dbuf, int16_t dst_ofs, + const struct rte_mbuf *sbuf, uint16_t src_ofs, int len) +{ + char rd[len]; + const char *wd = rte_pktmbuf_read(sbuf, src_ofs, len, rd); + + ovs_assert(wd); + + dp_packet_mbuf_write(dbuf, dst_ofs, len, wd); +} + +/* Similarly to dp_packet_shift(), shifts the data within the mbufs of a + * dp_packet of DPBUF_DPDK source by 'delta' bytes. + * Caller must make sure of the following conditions: + * - When shifting left, delta can't be bigger than the data_len available in + * the last mbuf; + * - When shifting right, delta can't be bigger than the space available in the + * first mbuf (buf_len - data_off). + * Both these conditions guarantee that a shift operation doesn't fall outside + * the bounds of the existing mbufs, so that the first and last mbufs (when + * using multi-segment mbufs), remain the same. */ +static void +dp_packet_mbuf_shift(struct dp_packet *b, int delta) +{ + uint16_t src_ofs; + int16_t dst_ofs; + + struct rte_mbuf *mbuf = CONST_CAST(struct rte_mbuf *, &b->mbuf); + struct rte_mbuf *tmbuf = rte_pktmbuf_lastseg(mbuf); + + if (delta < 0) { + ovs_assert(-delta <= tmbuf->data_len); + } else { + ovs_assert(delta < (mbuf->buf_len - mbuf->data_off)); + } + + /* Set the destination and source offsets to copy to */ + dst_ofs = delta; + src_ofs = 0; + + /* Shift data from src mbuf and offset to dst mbuf and offset */ + dp_packet_mbuf_shift_(mbuf, dst_ofs, mbuf, src_ofs, + rte_pktmbuf_pkt_len(mbuf)); + + /* Update mbufs' properties, and if using multi-segment mbufs, first and + * last mbuf's data_len also needs to be adjusted */ + mbuf->data_off = mbuf->data_off + dst_ofs; +} +#endif + /* Shifts all of the data within the allocated space in 'b' by 'delta' bytes. * For example, a 'delta' of 1 would cause each byte of data to move one byte * forward (from address 'p' to 'p+1'), and a 'delta' of -1 would cause each @@ -306,6 +397,12 @@ dp_packet_shift(struct dp_packet *b, int delta) : true); if (delta != 0) { +#ifdef DPDK_NETDEV + if (b->source == DPBUF_DPDK) { + dp_packet_mbuf_shift(b, delta); + return; + } +#endif char *dst = (char *) dp_packet_data(b) + delta; memmove(dst, dp_packet_data(b), dp_packet_size(b)); dp_packet_set_data(b, dst); diff --git a/lib/dp-packet.h b/lib/dp-packet.h index 4880fed..e46628d 100644 --- a/lib/dp-packet.h +++ b/lib/dp-packet.h @@ -80,6 +80,11 @@ struct dp_packet { }; }; +#ifdef DPDK_NETDEV +#define MBUF_BUF_END(BUF_ADDR, BUF_LEN) \ + (char *) (((char *) BUF_ADDR) + BUF_LEN) +#endif + static inline void *dp_packet_data(const struct dp_packet *); static inline void dp_packet_set_data(struct dp_packet *, void *); static inline void *dp_packet_base(const struct dp_packet *); @@ -133,6 +138,11 @@ static inline void *dp_packet_at(const struct dp_packet *, size_t offset, size_t size); static inline void *dp_packet_at_assert(const struct dp_packet *, size_t offset, size_t size); +#ifdef DPDK_NETDEV +void +dp_packet_mbuf_write(struct rte_mbuf *mbuf, int16_t ofs, uint32_t len, + const void *data); +#endif static inline void *dp_packet_tail(const struct dp_packet *); static inline void *dp_packet_end(const struct dp_packet *); From patchwork Wed Jul 4 18:06:29 2018 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: "Lam, Tiago" X-Patchwork-Id: 939506 X-Patchwork-Delegate: ian.stokes@intel.com Return-Path: X-Original-To: incoming@patchwork.ozlabs.org Delivered-To: patchwork-incoming@bilbo.ozlabs.org Authentication-Results: ozlabs.org; spf=pass (mailfrom) smtp.mailfrom=openvswitch.org (client-ip=140.211.169.12; helo=mail.linuxfoundation.org; envelope-from=ovs-dev-bounces@openvswitch.org; receiver=) Authentication-Results: ozlabs.org; dmarc=fail (p=none dis=none) header.from=intel.com Received: from mail.linuxfoundation.org (mail.linuxfoundation.org [140.211.169.12]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by ozlabs.org (Postfix) with ESMTPS id 41LTYn1F9vz9s29 for ; Thu, 5 Jul 2018 04:10:37 +1000 (AEST) Received: from mail.linux-foundation.org (localhost [127.0.0.1]) by mail.linuxfoundation.org (Postfix) with ESMTP id 66546D20; Wed, 4 Jul 2018 18:07:07 +0000 (UTC) X-Original-To: ovs-dev@openvswitch.org Delivered-To: ovs-dev@mail.linuxfoundation.org Received: from smtp1.linuxfoundation.org (smtp1.linux-foundation.org [172.17.192.35]) by mail.linuxfoundation.org (Postfix) with ESMTPS id 483A2CF9 for ; Wed, 4 Jul 2018 18:07:04 +0000 (UTC) X-Greylist: domain auto-whitelisted by SQLgrey-1.7.6 Received: from mga12.intel.com (mga12.intel.com [192.55.52.136]) by smtp1.linuxfoundation.org (Postfix) with ESMTPS id D79E3780 for ; Wed, 4 Jul 2018 18:07:03 +0000 (UTC) X-Amp-Result: SKIPPED(no attachment in message) X-Amp-File-Uploaded: False Received: from orsmga001.jf.intel.com ([10.7.209.18]) by fmsmga106.fm.intel.com with ESMTP/TLS/DHE-RSA-AES256-GCM-SHA384; 04 Jul 2018 11:07:02 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.51,308,1526367600"; d="scan'208";a="70109083" Received: from silpixa00399125.ir.intel.com ([10.237.223.34]) by orsmga001.jf.intel.com with ESMTP; 04 Jul 2018 11:06:56 -0700 From: Tiago Lam To: ovs-dev@openvswitch.org Date: Wed, 4 Jul 2018 19:06:29 +0100 Message-Id: <1530727595-217860-9-git-send-email-tiago.lam@intel.com> X-Mailer: git-send-email 2.7.4 In-Reply-To: <1530727595-217860-1-git-send-email-tiago.lam@intel.com> References: <1530727595-217860-1-git-send-email-tiago.lam@intel.com> X-Spam-Status: No, score=-4.2 required=5.0 tests=BAYES_00, RCVD_IN_DNSWL_MED autolearn=ham version=3.3.1 X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on smtp1.linux-foundation.org Cc: i.maximets@samsung.com Subject: [ovs-dev] [PATCH v2 08/14] dp-packet: Handle multi-seg mbufs in resize__(). X-BeenThere: ovs-dev@openvswitch.org X-Mailman-Version: 2.1.12 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , MIME-Version: 1.0 Sender: ovs-dev-bounces@openvswitch.org Errors-To: ovs-dev-bounces@openvswitch.org When enabled with DPDK OvS relies on mbufs allocated by mempools to receive and output data on DPDK ports. Until now, each OvS dp_packet has had only one mbuf associated, which is allocated with the maximum possible size, taking the MTU into account. This approach, however, doesn't allow us to increase the allocated size in an mbuf, if needed, since an mbuf is allocated and initialised upon mempool creation. Thus, in the current implementatin this is dealt with by calling OVS_NOT_REACHED() and terminating OvS. To avoid this, and allow the (already) allocated space to be better used, dp_packet_resize__() now tries to use the available room, both the tailroom and the headroom, to make enough space for the new data. Since this happens for packets of source DPBUF_DPDK, the single-segment mbuf case mentioned above is also covered by this new aproach in resize__(). Signed-off-by: Tiago Lam Acked-by: Eelco Chaudron --- lib/dp-packet.c | 48 ++++++++++++++++++++++++++++++++++++++++++++++-- 1 file changed, 46 insertions(+), 2 deletions(-) diff --git a/lib/dp-packet.c b/lib/dp-packet.c index d6e19eb..87af459 100644 --- a/lib/dp-packet.c +++ b/lib/dp-packet.c @@ -237,9 +237,51 @@ dp_packet_resize__(struct dp_packet *b, size_t new_headroom, size_t new_tailroom new_allocated = new_headroom + dp_packet_size(b) + new_tailroom; switch (b->source) { + /* When resizing mbufs, both a single mbuf and multi-segment mbufs (where + * data is not contigously held in memory), both the headroom and the + * tailroom available will be used to make more space for where data needs + * to be inserted. I.e if there's not enough headroom, data may be shifted + * right if there's enough tailroom. + * However, this is not bulletproof and in some cases the space available + * won't be enough - in those cases, an error should be returned and the + * packet dropped. */ case DPBUF_DPDK: - OVS_NOT_REACHED(); + { + size_t miss_len; + + if (new_headroom == dp_packet_headroom(b)) { + /* This is a tailroom adjustment. Since there's no tailroom space + * left, try and shift data towards the head to free up tail space, + * if there's enough headroom */ + + miss_len = new_tailroom - dp_packet_tailroom(b); + + if (miss_len <= new_headroom) { + dp_packet_shift(b, -miss_len); + } else { + /* XXX: Handle error case and report error to caller */ + OVS_NOT_REACHED(); + } + } else { + /* Otherwise, this is a headroom adjustment. Try to shift data + * towards the tail to free up head space, if there's enough + * tailroom */ + + miss_len = new_headroom - dp_packet_headroom(b); + + if (miss_len <= new_tailroom) { + dp_packet_shift(b, miss_len); + } else { + /* XXX: Handle error case and report error to caller */ + OVS_NOT_REACHED(); + } + } + + new_base = dp_packet_base(b); + + break; + } case DPBUF_MALLOC: if (new_headroom == dp_packet_headroom(b)) { new_base = xrealloc(dp_packet_base(b), new_allocated); @@ -263,7 +305,9 @@ dp_packet_resize__(struct dp_packet *b, size_t new_headroom, size_t new_tailroom OVS_NOT_REACHED(); } - dp_packet_set_allocated(b, new_allocated); + if (b->source != DPBUF_DPDK) { + dp_packet_set_allocated(b, new_allocated); + } dp_packet_set_base(b, new_base); new_data = (char *) new_base + new_headroom; From patchwork Wed Jul 4 18:06:30 2018 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: "Lam, Tiago" X-Patchwork-Id: 939505 X-Patchwork-Delegate: ian.stokes@intel.com Return-Path: X-Original-To: incoming@patchwork.ozlabs.org Delivered-To: patchwork-incoming@bilbo.ozlabs.org Authentication-Results: ozlabs.org; spf=pass (mailfrom) smtp.mailfrom=openvswitch.org (client-ip=140.211.169.12; helo=mail.linuxfoundation.org; envelope-from=ovs-dev-bounces@openvswitch.org; receiver=) Authentication-Results: ozlabs.org; dmarc=fail (p=none dis=none) header.from=intel.com Received: from mail.linuxfoundation.org (mail.linuxfoundation.org [140.211.169.12]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by ozlabs.org (Postfix) with ESMTPS id 41LTYJ3VvXz9s29 for ; Thu, 5 Jul 2018 04:10:12 +1000 (AEST) Received: from mail.linux-foundation.org (localhost [127.0.0.1]) by mail.linuxfoundation.org (Postfix) with ESMTP id A2F41D1C; Wed, 4 Jul 2018 18:07:06 +0000 (UTC) X-Original-To: ovs-dev@openvswitch.org Delivered-To: ovs-dev@mail.linuxfoundation.org Received: from smtp1.linuxfoundation.org (smtp1.linux-foundation.org [172.17.192.35]) by mail.linuxfoundation.org (Postfix) with ESMTPS id 2926BD01 for ; Wed, 4 Jul 2018 18:07:04 +0000 (UTC) X-Greylist: domain auto-whitelisted by SQLgrey-1.7.6 Received: from mga12.intel.com (mga12.intel.com [192.55.52.136]) by smtp1.linuxfoundation.org (Postfix) with ESMTPS id 9BEE9784 for ; Wed, 4 Jul 2018 18:07:03 +0000 (UTC) X-Amp-Result: SKIPPED(no attachment in message) X-Amp-File-Uploaded: False Received: from orsmga001.jf.intel.com ([10.7.209.18]) by fmsmga106.fm.intel.com with ESMTP/TLS/DHE-RSA-AES256-GCM-SHA384; 04 Jul 2018 11:07:02 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.51,308,1526367600"; d="scan'208";a="70109090" Received: from silpixa00399125.ir.intel.com ([10.237.223.34]) by orsmga001.jf.intel.com with ESMTP; 04 Jul 2018 11:06:58 -0700 From: Tiago Lam To: ovs-dev@openvswitch.org Date: Wed, 4 Jul 2018 19:06:30 +0100 Message-Id: <1530727595-217860-10-git-send-email-tiago.lam@intel.com> X-Mailer: git-send-email 2.7.4 In-Reply-To: <1530727595-217860-1-git-send-email-tiago.lam@intel.com> References: <1530727595-217860-1-git-send-email-tiago.lam@intel.com> X-Spam-Status: No, score=-4.2 required=5.0 tests=BAYES_00, RCVD_IN_DNSWL_MED autolearn=ham version=3.3.1 X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on smtp1.linux-foundation.org Cc: Michael Qiu , i.maximets@samsung.com Subject: [ovs-dev] [PATCH v2 09/14] dp-packet: copy data from multi-seg. DPDK mbuf X-BeenThere: ovs-dev@openvswitch.org X-Mailman-Version: 2.1.12 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , MIME-Version: 1.0 Sender: ovs-dev-bounces@openvswitch.org Errors-To: ovs-dev-bounces@openvswitch.org From: Michael Qiu When doing packet clone, if packet source is from DPDK driver, multi-segment must be considered, and copy the segment's data one by one. Also, lots of DPDK mbuf's info is missed during a copy, like packet type, ol_flags, etc. That information is very important for DPDK to do packets processing. Co-authored-by: Mark Kavanagh Co-authored-by: Tiago Lam Signed-off-by: Michael Qiu Signed-off-by: Mark Kavanagh Signed-off-by: Tiago Lam Acked-by: Eelco Chaudron --- lib/dp-packet.c | 71 ++++++++++++++++++++++++++++++++++++++++++++++--------- lib/dp-packet.h | 3 +++ lib/netdev-dpdk.c | 1 + 3 files changed, 64 insertions(+), 11 deletions(-) diff --git a/lib/dp-packet.c b/lib/dp-packet.c index 87af459..ae060e2 100644 --- a/lib/dp-packet.c +++ b/lib/dp-packet.c @@ -48,6 +48,23 @@ dp_packet_use__(struct dp_packet *b, void *base, size_t allocated, dp_packet_set_size(b, 0); } +#ifdef DPDK_NETDEV +void +dp_packet_copy_mbuf_flags(struct dp_packet *dst, const struct dp_packet *src) +{ + ovs_assert(dst != NULL && src != NULL); + struct rte_mbuf *buf_dst = &(dst->mbuf); + struct rte_mbuf buf_src = src->mbuf; + + buf_dst->nb_segs = buf_src.nb_segs; + buf_dst->ol_flags = buf_src.ol_flags; + buf_dst->packet_type = buf_src.packet_type; + buf_dst->tx_offload = buf_src.tx_offload; +} +#else +#define dp_packet_copy_mbuf_flags(arg1, arg2) +#endif + /* Initializes 'b' as an empty dp_packet that contains the 'allocated' bytes of * memory starting at 'base'. 'base' should be the first byte of a region * obtained from malloc(). It will be freed (with free()) if 'b' is resized or @@ -158,6 +175,44 @@ dp_packet_clone(const struct dp_packet *buffer) return dp_packet_clone_with_headroom(buffer, 0); } +#ifdef DPDK_NETDEV +struct dp_packet * +dp_packet_clone_with_headroom(const struct dp_packet *b, size_t headroom) { + struct dp_packet *new_buffer; + uint32_t pkt_len = dp_packet_size(b); + + /* copy multi-seg data */ + if (b->source == DPBUF_DPDK && b->mbuf.nb_segs > 1) { + void *dst = NULL; + struct rte_mbuf *mbuf = CONST_CAST(struct rte_mbuf *, &b->mbuf); + + new_buffer = dp_packet_new_with_headroom(pkt_len, headroom); + dst = dp_packet_data(new_buffer); + dp_packet_set_size(new_buffer, pkt_len); + + if (!rte_pktmbuf_read(mbuf, 0, pkt_len, dst)) { + return NULL; + } + } else { + new_buffer = dp_packet_clone_data_with_headroom(dp_packet_data(b), + dp_packet_size(b), + headroom); + } + + /* Copy the following fields into the returned buffer: l2_pad_size, + * l2_5_ofs, l3_ofs, l4_ofs, cutlen, packet_type and md. */ + memcpy(&new_buffer->l2_pad_size, &b->l2_pad_size, + sizeof(struct dp_packet) - + offsetof(struct dp_packet, l2_pad_size)); + + dp_packet_copy_mbuf_flags(new_buffer, b); + if (dp_packet_rss_valid(new_buffer)) { + new_buffer->mbuf.hash.rss = b->mbuf.hash.rss; + } + + return new_buffer; +} +#else /* Creates and returns a new dp_packet whose data are copied from 'buffer'. * The returned dp_packet will additionally have 'headroom' bytes of * headroom. */ @@ -165,32 +220,25 @@ struct dp_packet * dp_packet_clone_with_headroom(const struct dp_packet *buffer, size_t headroom) { struct dp_packet *new_buffer; + uint32_t pkt_len = dp_packet_size(buffer); new_buffer = dp_packet_clone_data_with_headroom(dp_packet_data(buffer), - dp_packet_size(buffer), - headroom); + pkt_len, headroom); + /* Copy the following fields into the returned buffer: l2_pad_size, * l2_5_ofs, l3_ofs, l4_ofs, cutlen, packet_type and md. */ memcpy(&new_buffer->l2_pad_size, &buffer->l2_pad_size, sizeof(struct dp_packet) - offsetof(struct dp_packet, l2_pad_size)); -#ifdef DPDK_NETDEV - new_buffer->mbuf.ol_flags = buffer->mbuf.ol_flags; -#else new_buffer->rss_hash_valid = buffer->rss_hash_valid; -#endif - if (dp_packet_rss_valid(new_buffer)) { -#ifdef DPDK_NETDEV - new_buffer->mbuf.hash.rss = buffer->mbuf.hash.rss; -#else new_buffer->rss_hash = buffer->rss_hash; -#endif } return new_buffer; } +#endif /* Creates and returns a new dp_packet that initially contains a copy of the * 'size' bytes of data starting at 'data' with no headroom or tailroom. */ @@ -374,6 +422,7 @@ dp_packet_mbuf_write(struct rte_mbuf *mbuf, int16_t ofs, uint32_t len, len -= len_copy; ofs = 0; + mbuf->data_len = len_copy; mbuf = mbuf->next; } } diff --git a/lib/dp-packet.h b/lib/dp-packet.h index e46628d..595be5d 100644 --- a/lib/dp-packet.h +++ b/lib/dp-packet.h @@ -124,6 +124,9 @@ void dp_packet_init_dpdk(struct dp_packet *); void dp_packet_init(struct dp_packet *, size_t); void dp_packet_uninit(struct dp_packet *); +void dp_packet_copy_mbuf_flags(struct dp_packet *dst, + const struct dp_packet *src); + struct dp_packet *dp_packet_new(size_t); struct dp_packet *dp_packet_new_with_headroom(size_t, size_t headroom); struct dp_packet *dp_packet_clone(const struct dp_packet *); diff --git a/lib/netdev-dpdk.c b/lib/netdev-dpdk.c index 6c9f0e7..b9003fa 100644 --- a/lib/netdev-dpdk.c +++ b/lib/netdev-dpdk.c @@ -2247,6 +2247,7 @@ dpdk_do_tx_copy(struct netdev *netdev, int qid, struct dp_packet_batch *batch) memcpy(rte_pktmbuf_mtod(pkts[txcnt], void *), dp_packet_data(packet), size); dp_packet_set_size((struct dp_packet *)pkts[txcnt], size); + dp_packet_copy_mbuf_flags((struct dp_packet *)pkts[txcnt], packet); txcnt++; } From patchwork Wed Jul 4 18:06:31 2018 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: "Lam, Tiago" X-Patchwork-Id: 939507 X-Patchwork-Delegate: ian.stokes@intel.com Return-Path: X-Original-To: incoming@patchwork.ozlabs.org Delivered-To: patchwork-incoming@bilbo.ozlabs.org Authentication-Results: ozlabs.org; spf=pass (mailfrom) smtp.mailfrom=openvswitch.org (client-ip=140.211.169.12; helo=mail.linuxfoundation.org; envelope-from=ovs-dev-bounces@openvswitch.org; receiver=) Authentication-Results: ozlabs.org; dmarc=fail (p=none dis=none) header.from=intel.com Received: from mail.linuxfoundation.org (mail.linuxfoundation.org [140.211.169.12]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by ozlabs.org (Postfix) with ESMTPS id 41LTZF3Hp5z9s29 for ; Thu, 5 Jul 2018 04:11:01 +1000 (AEST) Received: from mail.linux-foundation.org (localhost [127.0.0.1]) by mail.linuxfoundation.org (Postfix) with ESMTP id 1C31CD1D; Wed, 4 Jul 2018 18:07:08 +0000 (UTC) X-Original-To: ovs-dev@openvswitch.org Delivered-To: ovs-dev@mail.linuxfoundation.org Received: from smtp1.linuxfoundation.org (smtp1.linux-foundation.org [172.17.192.35]) by mail.linuxfoundation.org (Postfix) with ESMTPS id C4317D0D for ; Wed, 4 Jul 2018 18:07:04 +0000 (UTC) X-Greylist: domain auto-whitelisted by SQLgrey-1.7.6 Received: from mga12.intel.com (mga12.intel.com [192.55.52.136]) by smtp1.linuxfoundation.org (Postfix) with ESMTPS id 4CE4B791 for ; Wed, 4 Jul 2018 18:07:04 +0000 (UTC) X-Amp-Result: SKIPPED(no attachment in message) X-Amp-File-Uploaded: False Received: from orsmga001.jf.intel.com ([10.7.209.18]) by fmsmga106.fm.intel.com with ESMTP/TLS/DHE-RSA-AES256-GCM-SHA384; 04 Jul 2018 11:07:02 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.51,308,1526367600"; d="scan'208";a="70109096" Received: from silpixa00399125.ir.intel.com ([10.237.223.34]) by orsmga001.jf.intel.com with ESMTP; 04 Jul 2018 11:07:00 -0700 From: Tiago Lam To: ovs-dev@openvswitch.org Date: Wed, 4 Jul 2018 19:06:31 +0100 Message-Id: <1530727595-217860-11-git-send-email-tiago.lam@intel.com> X-Mailer: git-send-email 2.7.4 In-Reply-To: <1530727595-217860-1-git-send-email-tiago.lam@intel.com> References: <1530727595-217860-1-git-send-email-tiago.lam@intel.com> X-Spam-Status: No, score=-4.2 required=5.0 tests=BAYES_00, RCVD_IN_DNSWL_MED autolearn=ham version=3.3.1 X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on smtp1.linux-foundation.org Cc: Michael Qiu , i.maximets@samsung.com Subject: [ovs-dev] [PATCH v2 10/14] netdev-dpdk: copy large packet to multi-seg. mbufs X-BeenThere: ovs-dev@openvswitch.org X-Mailman-Version: 2.1.12 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , MIME-Version: 1.0 Sender: ovs-dev-bounces@openvswitch.org Errors-To: ovs-dev-bounces@openvswitch.org From: Mark Kavanagh Currently, packets are only copied to a single segment in the function dpdk_do_tx_copy(). This could be an issue in the case of jumbo frames, particularly when multi-segment mbufs are involved. This patch calculates the number of segments needed by a packet and copies the data to each segment. A new function, dpdk_buf_alloc(), has also been introduced as a wrapper around the nonpmd_mp_mutex to serialise allocations from a non-pmd context. Co-authored-by: Michael Qiu Co-authored-by: Tiago Lam Signed-off-by: Mark Kavanagh Signed-off-by: Michael Qiu Signed-off-by: Tiago Lam Acked-by: Eelco Chaudron --- lib/netdev-dpdk.c | 78 ++++++++++++++++++++++++++++++++++++++++++++++++------- 1 file changed, 69 insertions(+), 9 deletions(-) diff --git a/lib/netdev-dpdk.c b/lib/netdev-dpdk.c index b9003fa..1b59123 100644 --- a/lib/netdev-dpdk.c +++ b/lib/netdev-dpdk.c @@ -522,6 +522,22 @@ dpdk_rte_mzalloc(size_t sz) return rte_zmalloc(OVS_VPORT_DPDK, sz, OVS_CACHE_LINE_SIZE); } +static struct rte_mbuf * +dpdk_buf_alloc(struct rte_mempool *mp) +{ + if (!dpdk_thread_is_pmd()) { + ovs_mutex_lock(&nonpmd_mp_mutex); + } + + struct rte_mbuf *mbuf = rte_pktmbuf_alloc(mp); + + if (!dpdk_thread_is_pmd()) { + ovs_mutex_unlock(&nonpmd_mp_mutex); + } + + return mbuf; +} + void free_dpdk_buf(struct dp_packet *packet) { @@ -2199,6 +2215,49 @@ out: } } +static int +dpdk_copy_dp_packet_to_mbuf(struct dp_packet *packet, struct rte_mbuf **head, + struct rte_mempool *mp) +{ + struct rte_mbuf *mbuf, *fmbuf; + uint32_t size = dp_packet_size(packet); + uint16_t max_data_len; + uint32_t nb_segs = 0; + + /* Allocate first mbuf to know the size of data available */ + fmbuf = mbuf = *head = dpdk_buf_alloc(mp); + if (OVS_UNLIKELY(!mbuf)) { + return ENOMEM; + } + + /* All new allocated mbuf's max data len is the same */ + max_data_len = mbuf->buf_len - mbuf->data_off; + + /* Calculate # of output mbufs. */ + nb_segs = size / max_data_len; + if (size % max_data_len) { + nb_segs = nb_segs + 1; + } + + /* Allocate additional mbufs, less the one alredy allocated above */ + for (int i = 1; i < nb_segs; i++) { + mbuf->next = dpdk_buf_alloc(mp); + if (!mbuf->next) { + free_dpdk_buf(CONTAINER_OF(fmbuf, struct dp_packet, mbuf)); + fmbuf = NULL; + return ENOMEM; + } + mbuf = mbuf->next; + } + + fmbuf->nb_segs = nb_segs; + fmbuf->pkt_len = size; + + dp_packet_mbuf_write(fmbuf, 0, size, dp_packet_data(packet)); + + return 0; +} + /* Tx function. Transmit packets indefinitely */ static void dpdk_do_tx_copy(struct netdev *netdev, int qid, struct dp_packet_batch *batch) @@ -2215,6 +2274,7 @@ dpdk_do_tx_copy(struct netdev *netdev, int qid, struct dp_packet_batch *batch) struct rte_mbuf *pkts[PKT_ARRAY_SIZE]; uint32_t cnt = batch_cnt; uint32_t dropped = 0; + uint32_t i; if (dev->type != DPDK_DEV_VHOST) { /* Check if QoS has been configured for this netdev. */ @@ -2225,28 +2285,28 @@ dpdk_do_tx_copy(struct netdev *netdev, int qid, struct dp_packet_batch *batch) uint32_t txcnt = 0; - for (uint32_t i = 0; i < cnt; i++) { + for (i = 0; i < cnt; i++) { struct dp_packet *packet = batch->packets[i]; uint32_t size = dp_packet_size(packet); + int err = 0; if (OVS_UNLIKELY(size > dev->max_packet_len)) { VLOG_WARN_RL(&rl, "Too big size %u max_packet_len %d", size, dev->max_packet_len); - dropped++; continue; } - pkts[txcnt] = rte_pktmbuf_alloc(dev->mp); - if (OVS_UNLIKELY(!pkts[txcnt])) { + err = dpdk_copy_dp_packet_to_mbuf(packet, &pkts[txcnt], dev->mp); + if (err != 0) { + if (err == ENOMEM) { + VLOG_ERR_RL(&rl, "Failed to alloc mbufs! %u packets dropped", + cnt - i); + } + dropped += cnt - i; break; } - - /* We have to do a copy for now */ - memcpy(rte_pktmbuf_mtod(pkts[txcnt], void *), - dp_packet_data(packet), size); - dp_packet_set_size((struct dp_packet *)pkts[txcnt], size); dp_packet_copy_mbuf_flags((struct dp_packet *)pkts[txcnt], packet); txcnt++; From patchwork Wed Jul 4 18:06:32 2018 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: "Lam, Tiago" X-Patchwork-Id: 939509 X-Patchwork-Delegate: ian.stokes@intel.com Return-Path: X-Original-To: incoming@patchwork.ozlabs.org Delivered-To: patchwork-incoming@bilbo.ozlabs.org Authentication-Results: ozlabs.org; spf=pass (mailfrom) smtp.mailfrom=openvswitch.org (client-ip=140.211.169.12; helo=mail.linuxfoundation.org; envelope-from=ovs-dev-bounces@openvswitch.org; receiver=) Authentication-Results: ozlabs.org; dmarc=fail (p=none dis=none) header.from=intel.com Received: from mail.linuxfoundation.org (mail.linuxfoundation.org [140.211.169.12]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by ozlabs.org (Postfix) with ESMTPS id 41LTbN3LMkz9s29 for ; Thu, 5 Jul 2018 04:12:00 +1000 (AEST) Received: from mail.linux-foundation.org (localhost [127.0.0.1]) by mail.linuxfoundation.org (Postfix) with ESMTP id E118AD53; Wed, 4 Jul 2018 18:07:09 +0000 (UTC) X-Original-To: ovs-dev@openvswitch.org Delivered-To: ovs-dev@mail.linuxfoundation.org Received: from smtp1.linuxfoundation.org (smtp1.linux-foundation.org [172.17.192.35]) by mail.linuxfoundation.org (Postfix) with ESMTPS id 7F2B3D08 for ; Wed, 4 Jul 2018 18:07:05 +0000 (UTC) X-Greylist: domain auto-whitelisted by SQLgrey-1.7.6 Received: from mga12.intel.com (mga12.intel.com [192.55.52.136]) by smtp1.linuxfoundation.org (Postfix) with ESMTPS id 516AC792 for ; Wed, 4 Jul 2018 18:07:04 +0000 (UTC) X-Amp-Result: SKIPPED(no attachment in message) X-Amp-File-Uploaded: False Received: from orsmga001.jf.intel.com ([10.7.209.18]) by fmsmga106.fm.intel.com with ESMTP/TLS/DHE-RSA-AES256-GCM-SHA384; 04 Jul 2018 11:07:03 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.51,308,1526367600"; d="scan'208";a="70109106" Received: from silpixa00399125.ir.intel.com ([10.237.223.34]) by orsmga001.jf.intel.com with ESMTP; 04 Jul 2018 11:07:01 -0700 From: Tiago Lam To: ovs-dev@openvswitch.org Date: Wed, 4 Jul 2018 19:06:32 +0100 Message-Id: <1530727595-217860-12-git-send-email-tiago.lam@intel.com> X-Mailer: git-send-email 2.7.4 In-Reply-To: <1530727595-217860-1-git-send-email-tiago.lam@intel.com> References: <1530727595-217860-1-git-send-email-tiago.lam@intel.com> X-Spam-Status: No, score=-4.2 required=5.0 tests=BAYES_00, RCVD_IN_DNSWL_MED autolearn=ham version=3.3.1 X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on smtp1.linux-foundation.org Cc: i.maximets@samsung.com Subject: [ovs-dev] [PATCH v2 11/14] netdev-dpdk: support multi-segment jumbo frames X-BeenThere: ovs-dev@openvswitch.org X-Mailman-Version: 2.1.12 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , MIME-Version: 1.0 Sender: ovs-dev-bounces@openvswitch.org Errors-To: ovs-dev-bounces@openvswitch.org From: Mark Kavanagh Currently, jumbo frame support for OvS-DPDK is implemented by increasing the size of mbufs within a mempool, such that each mbuf within the pool is large enough to contain an entire jumbo frame of a user-defined size. Typically, for each user-defined MTU, 'requested_mtu', a new mempool is created, containing mbufs of size ~requested_mtu. With the multi-segment approach, a port uses a single mempool, (containing standard/default-sized mbufs of ~2k bytes), irrespective of the user-requested MTU value. To accommodate jumbo frames, mbufs are chained together, where each mbuf in the chain stores a portion of the jumbo frame. Each mbuf in the chain is termed a segment, hence the name. == Enabling multi-segment mbufs == Multi-segment and single-segment mbufs are mutually exclusive, and the user must decide on which approach to adopt on init. The introduction of a new OVSDB field, 'dpdk-multi-seg-mbufs', facilitates this. This is a global boolean value, which determines how jumbo frames are represented across all DPDK ports. In the absence of a user-supplied value, 'dpdk-multi-seg-mbufs' defaults to false, i.e. multi-segment mbufs must be explicitly enabled / single-segment mbufs remain the default. Setting the field is identical to setting existing DPDK-specific OVSDB fields: ovs-vsctl set Open_vSwitch . other_config:dpdk-init=true ovs-vsctl set Open_vSwitch . other_config:dpdk-lcore-mask=0x10 ovs-vsctl set Open_vSwitch . other_config:dpdk-socket-mem=4096,0 ==> ovs-vsctl set Open_vSwitch . other_config:dpdk-multi-seg-mbufs=true Co-authored-by: Tiago Lam Signed-off-by: Mark Kavanagh Signed-off-by: Tiago Lam Acked-by: Eelco Chaudron --- Documentation/topics/dpdk/jumbo-frames.rst | 51 +++++++++++++++++++++++ NEWS | 1 + lib/dpdk.c | 8 ++++ lib/netdev-dpdk.c | 65 ++++++++++++++++++++++++++++-- lib/netdev-dpdk.h | 2 + vswitchd/vswitch.xml | 23 +++++++++++ 6 files changed, 146 insertions(+), 4 deletions(-) diff --git a/Documentation/topics/dpdk/jumbo-frames.rst b/Documentation/topics/dpdk/jumbo-frames.rst index 00360b4..d1fb111 100644 --- a/Documentation/topics/dpdk/jumbo-frames.rst +++ b/Documentation/topics/dpdk/jumbo-frames.rst @@ -71,3 +71,54 @@ Jumbo frame support has been validated against 9728B frames, which is the largest frame size supported by Fortville NIC using the DPDK i40e driver, but larger frames and other DPDK NIC drivers may be supported. These cases are common for use cases involving East-West traffic only. + +------------------- +Multi-segment mbufs +------------------- + +Instead of increasing the size of mbufs within a mempool, such that each mbuf +within the pool is large enough to contain an entire jumbo frame of a +user-defined size, mbufs can be chained together instead. In this approach each +mbuf in the chain stores a portion of the jumbo frame, by default ~2K bytes, +irrespective of the user-requested MTU value. Since each mbuf in the chain is +termed a segment, this approach is named "multi-segment mbufs". + +This approach may bring more flexibility in use cases where the maximum packet +length may be hard to guess. For example, in cases where packets originate from +sources marked for oflload (such as TSO), each packet may be larger than the +MTU, and as such, when forwarding it to a DPDK port a single mbuf may not be +enough to hold all of the packet's data. + +Multi-segment and single-segment mbufs are mutually exclusive, and the user +must decide on which approach to adopt on initialisation. If multi-segment +mbufs is to be enabled, it can be done so with the following command:: + + $ ovs-vsctl set Open_vSwitch . other_config:dpdk-multi-seg-mbufs=true + +Single-segment mbufs still remain the default when using OvS-DPDK, and the +above option `dpdk-multi-seg-mbufs` must be explicitly set to `true` if +multi-segment mbufs are to be used. + +~~~~~~~~~~~~~~~~~ +Performance notes +~~~~~~~~~~~~~~~~~ + +When using multi-segment mbufs some PMDs may not support vectorized Tx +functions, due to its non-contiguous nature. As a result this can hit +performance for smaller packet sizes. For example, on a setup sending 64B +packets at line rate, a decrease of ~20% has been observed. The performance +impact stops being noticeable for larger packet sizes, although the exact size +will between PMDs, and depending on the architecture one's using. + +Tests performed with the i40e PMD driver only showed this limitation for 64B +packets, and the same rate was observed when comparing multi-segment mbufs and +single-segment mbuf for 128B packets. In other words, the 20% drop in +performance was not observed for packets >= 128B during this test case. + +Because of this, multi-segment mbufs is not advised to be used with smaller +packet sizes, such as 64B. + +Also, note that using multi-segment mbufs won't improve memory usage. For a +packet of 9000B, for example, which would be stored on a single mbuf when using +the single-segment approach, 5 mbufs (9000/2048) of 2048B would be needed to +store the same data using the multi-segment mbufs approach. diff --git a/NEWS b/NEWS index cd15a33..1771dd1 100644 --- a/NEWS +++ b/NEWS @@ -111,6 +111,7 @@ v2.9.0 - 19 Feb 2018 pmd assignments. * Add rxq utilization of pmd to appctl 'dpif-netdev/pmd-rxq-show'. * Add support for vHost dequeue zero copy (experimental) + * Add support for multi-segment mbufs - Userspace datapath: * Output packet batching support. - vswitchd: diff --git a/lib/dpdk.c b/lib/dpdk.c index 5c68ce4..e49a8cb 100644 --- a/lib/dpdk.c +++ b/lib/dpdk.c @@ -491,6 +491,14 @@ dpdk_init__(const struct smap *ovs_other_config) /* Finally, register the dpdk classes */ netdev_dpdk_register(); + + bool multi_seg_mbufs_enable = smap_get_bool(ovs_other_config, + "dpdk-multi-seg-mbufs", false); + if (multi_seg_mbufs_enable) { + VLOG_INFO("DPDK multi-segment mbufs enabled\n"); + netdev_dpdk_multi_segment_mbufs_enable(); + } + return true; } diff --git a/lib/netdev-dpdk.c b/lib/netdev-dpdk.c index 1b59123..f767858 100644 --- a/lib/netdev-dpdk.c +++ b/lib/netdev-dpdk.c @@ -66,6 +66,7 @@ enum {VIRTIO_RXQ, VIRTIO_TXQ, VIRTIO_QNUM}; VLOG_DEFINE_THIS_MODULE(netdev_dpdk); static struct vlog_rate_limit rl = VLOG_RATE_LIMIT_INIT(5, 20); +static bool dpdk_multi_segment_mbufs = false; #define DPDK_PORT_WATCHDOG_INTERVAL 5 @@ -628,7 +629,7 @@ dpdk_mp_create(struct netdev_dpdk *dev, uint16_t mbuf_pkt_data_len) uint32_t n_mbufs; uint32_t hash = hash_string(netdev_name, 0); struct rte_mempool *mp = NULL; - uint16_t mbuf_size, aligned_mbuf_size, mbuf_priv_data_len; + uint16_t mbuf_size, aligned_mbuf_size, mbuf_priv_data_len, max_frame_len; /* * XXX: rough estimation of number of mbufs required for this port: @@ -642,6 +643,21 @@ dpdk_mp_create(struct netdev_dpdk *dev, uint16_t mbuf_pkt_data_len) + MIN(RTE_MAX_LCORE, dev->requested_n_rxq) * NETDEV_MAX_BURST + MIN_NB_MBUF; + /* If multi-segment mbufs are used, we also increase the number of mbufs + * used. This is done by calculating how many mbufs are needed to hold the + * data on a single packet of MTU size. For example, for a received packet + * of 9000B, 5 mbufs (9000 / 2048) are needed to hold the data - 4 more + * than with single-mbufs (as mbufs' size is extended to hold all data) */ + max_frame_len = MTU_TO_MAX_FRAME_LEN(dev->requested_mtu); + if (dpdk_multi_segment_mbufs && mbuf_pkt_data_len < max_frame_len) { + uint16_t nb_segs = max_frame_len / mbuf_pkt_data_len; + if (max_frame_len % mbuf_pkt_data_len) { + nb_segs += 1; + } + + n_mbufs *= nb_segs; + } + ovs_mutex_lock(&dpdk_mp_mutex); do { /* Full DPDK memory pool name must be unique and cannot be @@ -743,7 +759,13 @@ dpdk_mp_release(struct rte_mempool *mp) /* Tries to allocate a new mempool - or re-use an existing one where * appropriate - on requested_socket_id with a size determined by - * requested_mtu and requested Rx/Tx queues. + * requested_mtu and requested Rx/Tx queues. Some properties of the mempool's + * elements are dependent on the value of 'dpdk_multi_segment_mbufs': + * - if 'true', then the mempool contains standard-sized mbufs that are chained + * together to accommodate packets of size 'requested_mtu'. + * - if 'false', then the members of the allocated mempool are + * non-standard-sized mbufs. Each mbuf in the mempool is large enough to + * fully accomdate packets of size 'requested_mtu'. * On success - or when re-using an existing mempool - the new configuration * will be applied. * On error, device will be left unchanged. */ @@ -751,10 +773,18 @@ static int netdev_dpdk_mempool_configure(struct netdev_dpdk *dev) OVS_REQUIRES(dev->mutex) { - uint16_t buf_size = dpdk_buf_size(dev->requested_mtu); + uint16_t buf_size = 0; struct rte_mempool *mp; int ret = 0; + /* Contiguous mbufs in use - permit oversized mbufs */ + if (!dpdk_multi_segment_mbufs) { + buf_size = dpdk_buf_size(dev->requested_mtu); + } else { + /* multi-segment mbufs - use standard mbuf size */ + buf_size = dpdk_buf_size(ETHER_MTU); + } + dpdk_mp_sweep(); mp = dpdk_mp_create(dev, buf_size); @@ -836,6 +866,7 @@ dpdk_eth_dev_port_config(struct netdev_dpdk *dev, int n_rxq, int n_txq) int diag = 0; int i; struct rte_eth_conf conf = port_conf; + struct rte_eth_txconf txconf; struct rte_eth_dev_info info; uint16_t conf_mtu; @@ -852,6 +883,18 @@ dpdk_eth_dev_port_config(struct netdev_dpdk *dev, int n_rxq, int n_txq) } } + /* Multi-segment-mbuf-specific setup. */ + if (dpdk_multi_segment_mbufs) { + /* DPDK PMDs typically attempt to use simple or vectorized + * transmit functions, neither of which are compatible with + * multi-segment mbufs. Ensure that these are disabled when + * multi-segment mbufs are enabled. + */ + rte_eth_dev_info_get(dev->port_id, &info); + txconf = info.default_txconf; + txconf.txq_flags &= ~ETH_TXQ_FLAGS_NOMULTSEGS; + } + conf.intr_conf.lsc = dev->lsc_interrupt_mode; conf.rxmode.hw_ip_checksum = (dev->hw_ol_features & NETDEV_RX_CHECKSUM_OFFLOAD) != 0; @@ -896,7 +939,9 @@ dpdk_eth_dev_port_config(struct netdev_dpdk *dev, int n_rxq, int n_txq) for (i = 0; i < n_txq; i++) { diag = rte_eth_tx_queue_setup(dev->port_id, i, dev->txq_size, - dev->socket_id, NULL); + dev->socket_id, + dpdk_multi_segment_mbufs ? &txconf + : NULL); if (diag) { VLOG_INFO("Interface %s unable to setup txq(%d): %s", dev->up.name, i, rte_strerror(-diag)); @@ -3985,6 +4030,18 @@ unlock: return err; } +bool +netdev_dpdk_is_multi_segment_mbufs_enabled(void) +{ + return dpdk_multi_segment_mbufs == true; +} + +void +netdev_dpdk_multi_segment_mbufs_enable(void) +{ + dpdk_multi_segment_mbufs = true; +} + #define NETDEV_DPDK_CLASS(NAME, INIT, CONSTRUCT, DESTRUCT, \ SET_CONFIG, SET_TX_MULTIQ, SEND, \ GET_CARRIER, GET_STATS, \ diff --git a/lib/netdev-dpdk.h b/lib/netdev-dpdk.h index b7d02a7..19aa5c6 100644 --- a/lib/netdev-dpdk.h +++ b/lib/netdev-dpdk.h @@ -25,6 +25,8 @@ struct dp_packet; #ifdef DPDK_NETDEV +bool netdev_dpdk_is_multi_segment_mbufs_enabled(void); +void netdev_dpdk_multi_segment_mbufs_enable(void); void netdev_dpdk_register(void); void free_dpdk_buf(struct dp_packet *); diff --git a/vswitchd/vswitch.xml b/vswitchd/vswitch.xml index 7609485..2af6170 100644 --- a/vswitchd/vswitch.xml +++ b/vswitchd/vswitch.xml @@ -332,6 +332,29 @@

+ +

+ Specifies if DPDK uses multi-segment mbufs for handling jumbo frames. +

+

+ If true, DPDK allocates a single mempool per port, irrespective of + the ports' requested MTU sizes. The elements of this mempool are + 'standard'-sized mbufs (typically 2k MB), which may be chained + together to accommodate jumbo frames. In this approach, each mbuf + typically stores a fragment of the overall jumbo frame. +

+

+ If not specified, defaults to false, in which case, the + size of each mbuf within a DPDK port's mempool will be grown to + accommodate jumbo frames within a single mbuf. +

+

+ Changing this value requires restarting the daemon. +

+ + +

From patchwork Wed Jul 4 18:06:33 2018 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: "Lam, Tiago" X-Patchwork-Id: 939511 X-Patchwork-Delegate: ian.stokes@intel.com Return-Path: X-Original-To: incoming@patchwork.ozlabs.org Delivered-To: patchwork-incoming@bilbo.ozlabs.org Authentication-Results: ozlabs.org; spf=pass (mailfrom) smtp.mailfrom=openvswitch.org (client-ip=140.211.169.12; helo=mail.linuxfoundation.org; envelope-from=ovs-dev-bounces@openvswitch.org; receiver=) Authentication-Results: ozlabs.org; dmarc=fail (p=none dis=none) header.from=intel.com Received: from mail.linuxfoundation.org (mail.linuxfoundation.org [140.211.169.12]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by ozlabs.org (Postfix) with ESMTPS id 41LTc00t04z9s29 for ; Thu, 5 Jul 2018 04:12:32 +1000 (AEST) Received: from mail.linux-foundation.org (localhost [127.0.0.1]) by mail.linuxfoundation.org (Postfix) with ESMTP id CB053D3A; Wed, 4 Jul 2018 18:07:22 +0000 (UTC) X-Original-To: ovs-dev@openvswitch.org Delivered-To: ovs-dev@mail.linuxfoundation.org Received: from smtp1.linuxfoundation.org (smtp1.linux-foundation.org [172.17.192.35]) by mail.linuxfoundation.org (Postfix) with ESMTPS id 85C24CB8 for ; Wed, 4 Jul 2018 18:07:21 +0000 (UTC) X-Greylist: domain auto-whitelisted by SQLgrey-1.7.6 Received: from mga11.intel.com (mga11.intel.com [192.55.52.93]) by smtp1.linuxfoundation.org (Postfix) with ESMTPS id 2ED98782 for ; Wed, 4 Jul 2018 18:07:20 +0000 (UTC) X-Amp-Result: SKIPPED(no attachment in message) X-Amp-File-Uploaded: False Received: from orsmga001.jf.intel.com ([10.7.209.18]) by fmsmga102.fm.intel.com with ESMTP/TLS/DHE-RSA-AES256-GCM-SHA384; 04 Jul 2018 11:07:19 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.51,308,1526367600"; d="scan'208";a="70109112" Received: from silpixa00399125.ir.intel.com ([10.237.223.34]) by orsmga001.jf.intel.com with ESMTP; 04 Jul 2018 11:07:04 -0700 From: Tiago Lam To: ovs-dev@openvswitch.org Date: Wed, 4 Jul 2018 19:06:33 +0100 Message-Id: <1530727595-217860-13-git-send-email-tiago.lam@intel.com> X-Mailer: git-send-email 2.7.4 In-Reply-To: <1530727595-217860-1-git-send-email-tiago.lam@intel.com> References: <1530727595-217860-1-git-send-email-tiago.lam@intel.com> X-Spam-Status: No, score=-6.9 required=5.0 tests=BAYES_00,RCVD_IN_DNSWL_HI autolearn=ham version=3.3.1 X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on smtp1.linux-foundation.org Cc: i.maximets@samsung.com Subject: [ovs-dev] [PATCH v2 12/14] dpdk-tests: Add uni-tests for multi-seg mbufs. X-BeenThere: ovs-dev@openvswitch.org X-Mailman-Version: 2.1.12 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , MIME-Version: 1.0 Sender: ovs-dev-bounces@openvswitch.org Errors-To: ovs-dev-bounces@openvswitch.org In order to create a minimal environment that allows the tests to get mbufs from an existing mempool, the following approach is taken: - EAL is initialised (by using the main dpdk_init()) and a (very) small mempool is instantiated (mimicking the logic in dpdk_mp_create()). This mempool instance is global and used by all the tests; - Packets are then allocated from the instantiated mempool, and tested on, by running some operations on them and manipulating data. The tests introduced focus on testing DPDK dp_packets (where source=DPBUF_DPDK), linked with a single or multiple mbufs, across several operations, such as: - dp_packet_put(); - dp_packet_shift(); - dp_packet_reserve(); - dp_packet_push_uninit(); - dp_packet_clear(); - And as a consequence of some of these, dp_packet_put_uninit() and dp_packet_resize__(). Finally, this has also been integrated with the new DPDK testsuite. Thus, when running `$sudo make check-dpdk` one will also be running these tests. Signed-off-by: Tiago Lam Acked-by: Eelco Chaudron --- tests/automake.mk | 10 +- tests/dpdk-packet-mbufs.at | 7 + tests/system-dpdk-testsuite.at | 1 + tests/test-dpdk-mbufs.c | 518 +++++++++++++++++++++++++++++++++++++++++ 4 files changed, 535 insertions(+), 1 deletion(-) create mode 100644 tests/dpdk-packet-mbufs.at create mode 100644 tests/test-dpdk-mbufs.c diff --git a/tests/automake.mk b/tests/automake.mk index 8224e5a..b3941d0 100644 --- a/tests/automake.mk +++ b/tests/automake.mk @@ -134,7 +134,8 @@ SYSTEM_DPDK_TESTSUITE_AT = \ tests/system-common-macros.at \ tests/system-dpdk-macros.at \ tests/system-dpdk-testsuite.at \ - tests/system-dpdk.at + tests/system-dpdk.at \ + tests/dpdk-packet-mbufs.at check_SCRIPTS += tests/atlocal @@ -391,6 +392,10 @@ tests_ovstest_SOURCES = \ tests/test-vconn.c \ tests/test-aa.c \ tests/test-stopwatch.c +if DPDK_NETDEV +tests_ovstest_SOURCES = \ + tests/test-dpdk-mbufs.c +endif if !WIN32 tests_ovstest_SOURCES += \ @@ -403,6 +408,9 @@ tests_ovstest_SOURCES += \ endif tests_ovstest_LDADD = lib/libopenvswitch.la ovn/lib/libovn.la +if DPDK_NETDEV +tests_ovstest_LDFLAGS = $(AM_LDFLAGS) $(DPDK_vswitchd_LDFLAGS) +endif noinst_PROGRAMS += tests/test-strtok_r tests_test_strtok_r_SOURCES = tests/test-strtok_r.c diff --git a/tests/dpdk-packet-mbufs.at b/tests/dpdk-packet-mbufs.at new file mode 100644 index 0000000..f28e4fc --- /dev/null +++ b/tests/dpdk-packet-mbufs.at @@ -0,0 +1,7 @@ +AT_BANNER([OVS-DPDK dp_packet unit tests]) + +AT_SETUP([OVS-DPDK dp_packet - mbufs allocation]) +AT_KEYWORDS([dp_packet, multi-seg, mbufs]) +AT_CHECK(ovstest test-dpdk-packet, [], [ignore], [ignore]) + +AT_CLEANUP diff --git a/tests/system-dpdk-testsuite.at b/tests/system-dpdk-testsuite.at index 382f09e..f5edf58 100644 --- a/tests/system-dpdk-testsuite.at +++ b/tests/system-dpdk-testsuite.at @@ -23,3 +23,4 @@ m4_include([tests/system-common-macros.at]) m4_include([tests/system-dpdk-macros.at]) m4_include([tests/system-dpdk.at]) +m4_include([tests/dpdk-packet-mbufs.at]) diff --git a/tests/test-dpdk-mbufs.c b/tests/test-dpdk-mbufs.c new file mode 100644 index 0000000..1c77038 --- /dev/null +++ b/tests/test-dpdk-mbufs.c @@ -0,0 +1,518 @@ +/* + * Copyright (c) 2018 Intel Corporation + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at: + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include "dp-packet.h" +#include "ovstest.h" +#include "dpdk.h" +#include "smap.h" + +#define N_MBUFS 1024 +#define MBUF_DATA_LEN 2048 + +int num_tests = 0; + +/* Global var to hold a mempool instance, "test-mp", used in all of the tests + * below. This instance is instantiated in dpdk_setup_eal_with_mp(). */ +static struct rte_mempool *mp; + +/* Test data used to fill the packets with data. Note that this isn't a string + * that repsents a valid packet, by any means. The pattern is generated in set_ + * testing_pattern_str() and the sole purpose is to verify the data remains the + * same after inserting and operating on multi-segment mbufs. */ +static char *test_str; + +/* Asserts a dp_packet that holds a single mbuf, where: + * - nb_segs must be 1; + * - pkt_len must be equal to data_len which in turn must equal the provided + * 'pkt_len'; + * - data_off must start at the provided 'data_ofs'; + * - next must be NULL. */ +static void +assert_single_mbuf(struct dp_packet *pkt, uint16_t data_ofs, + uint32_t pkt_len) { + struct rte_mbuf *mbuf = CONST_CAST(struct rte_mbuf *, &pkt->mbuf); + ovs_assert(mbuf->nb_segs == 1); + ovs_assert(mbuf->data_off == data_ofs); + ovs_assert(mbuf->pkt_len == mbuf->data_len); + ovs_assert(mbuf->pkt_len == pkt_len); + ovs_assert(mbuf->next == NULL); +} + +/* Asserts a dp_packet that holds multiple mbufs, where: + * - nb_segs must be > 1 and equal to the provided 'nb_segs'; + * - data_off must start at the provided 'data_ofs'; + * - pkt_len must be equal to the provided 'pkt_len' and the some of each + * mbufs' 'data_len' must equal the pky_len; + * - next must not be NULL. */ +static void +assert_multiple_mbufs(struct dp_packet *pkt, uint16_t data_ofs, + uint32_t pkt_len, uint16_t nb_segs) { + struct rte_mbuf *mbuf = CONST_CAST(struct rte_mbuf *, &pkt->mbuf); + ovs_assert(mbuf->nb_segs > 1 && mbuf->nb_segs == nb_segs); + ovs_assert(mbuf->data_off == data_ofs); + ovs_assert(mbuf->pkt_len != mbuf->data_len); + ovs_assert(mbuf->next != NULL); + ovs_assert(mbuf->pkt_len == pkt_len); + /* Make sure pkt_len equals the sum of all segments data_len */ + while (mbuf) { + pkt_len -= rte_pktmbuf_data_len(mbuf); + mbuf = mbuf->next; + } + ovs_assert(pkt_len == 0); +} + +/* Asserts that the data existing in a packet, starting at 'data_ofs' of the + * first mbuf and of length 'data_len' matches the global test_str used, + * starting at index 0 and of the same length. */ +static void +assert_data(struct dp_packet *pkt, uint16_t data_ofs, uint16_t data_len) { + struct rte_mbuf *mbuf = CONST_CAST(struct rte_mbuf *, &pkt->mbuf); + + char data[data_len]; + const char *rd = rte_pktmbuf_read(mbuf, data_ofs, data_len, data); + + ovs_assert(rd != NULL); + ovs_assert(memcmp(rd, test_str, data_len) == 0); +} + +static void +set_testing_pattern_str(void) { + static const char *pattern = "1234567890"; + + /* Pattern will be of size 5000B */ + size_t test_str_len = 5000; + test_str = xmalloc(test_str_len * sizeof(*test_str) + 1); + + for (int i = 0; i < test_str_len; i += strlen(pattern)) { + memcpy(test_str + i, pattern, strlen(pattern)); + } + + test_str[test_str_len] = 0; +} + +static void +dpdk_eal_init(void) { + struct smap other_config; + smap_init(&other_config); + + printf("Initialising EAL...\n"); + smap_add(&other_config, "dpdk-init", "true"); + smap_add(&other_config, "dpdk-lcore-mask", "10"); + smap_add(&other_config, "dpdk-socket-mem", "2048,0"); + smap_add(&other_config, "dpdk-multi-seg-mbufs", "true"); + + dpdk_init(&other_config); +} + +/* The allocation of mbufs here mimics the logic in dpdk_mp_create in + * netdev-dpdk.c. */ +static struct rte_mempool * +dpdk_mp_create(char *mp_name) { + uint16_t mbuf_size, aligned_mbuf_size, mbuf_priv_data_len; + + mbuf_size = sizeof (struct dp_packet) + + MBUF_DATA_LEN + RTE_PKTMBUF_HEADROOM; + aligned_mbuf_size = ROUND_UP(mbuf_size, RTE_CACHE_LINE_SIZE); + mbuf_priv_data_len = sizeof(struct dp_packet) - sizeof(struct rte_mbuf) + + (aligned_mbuf_size - mbuf_size); + + struct rte_mempool *mpool = rte_pktmbuf_pool_create( + mp_name, N_MBUFS, + RTE_MEMPOOL_CACHE_MAX_SIZE, + mbuf_priv_data_len, + MBUF_DATA_LEN + + RTE_PKTMBUF_HEADROOM /* defaults 128B */, + SOCKET_ID_ANY); + if (mpool) { + printf("Allocated \"%s\" mempool with %u mbufs\n", mp_name, N_MBUFS); + } else { + printf("Failed mempool \"%s\" create request of %u mbufs: %s.\n", + mp_name, N_MBUFS, rte_strerror(rte_errno)); + + ovs_assert(mpool != NULL); + } + + return mpool; +} + +static void +dpdk_setup_eal_with_mp(void) { + dpdk_eal_init(); + + mp = dpdk_mp_create("test-mp"); + ovs_assert(mp != NULL); +} + +static struct dp_packet * +dpdk_mp_alloc_pkt(struct rte_mempool *mpool) { + struct rte_mbuf *mbuf = rte_pktmbuf_alloc(mpool); + + struct dp_packet *pkt = (struct dp_packet *) mbuf; + pkt->source = DPBUF_DPDK; + + return pkt; +} + +/* Similar to dp_packet_put() in dp-packet.c, appends the 'size' bytes of data + * in 'p' to the tail end of 'pkt', allocating new mbufs if needed. */ +static struct dp_packet * +dpdk_pkt_put(struct dp_packet *pkt, void *p, size_t size) { + uint16_t max_data_len, nb_segs; + struct rte_mbuf *mbuf, *fmbuf; + + mbuf = CONST_CAST(struct rte_mbuf *, &pkt->mbuf); + + /* All new allocated mbuf's max data len is the same */ + max_data_len = mbuf->buf_len - mbuf->data_off; + + /* Calculate # of needed mbufs to accomodate 'miss_len' */ + nb_segs = size / max_data_len; + if (size % max_data_len) { + nb_segs += 1; + } + + /* Proceed with the allocation of new mbufs */ + mp = mbuf->pool; + fmbuf = mbuf; + mbuf = rte_pktmbuf_lastseg(mbuf); + + for (int i = 0; i < nb_segs; i++) { + /* This takes care of initialising buf_len, data_len and other + * fields properly */ + mbuf->next = rte_pktmbuf_alloc(mp); + if (!mbuf->next) { + printf("Problem allocating more mbufs for tests.\n"); + rte_pktmbuf_free(mbuf); + fmbuf = NULL; + return NULL; + } + + fmbuf->nb_segs += 1; + + mbuf = mbuf->next; + } + + dp_packet_mbuf_write(fmbuf, 0, size, p); + + /* Adjust size of intermediate mbufs from current tail to end */ + /*size_t pkt_len = size; + while (fmbuf && pkt_len > 0) { + fmbuf->data_len = MIN(pkt_len, fmbuf->buf_len - fmbuf->data_off); + pkt_len -= fmbuf->data_len; + + fmbuf = fmbuf->next; + }*/ + + dp_packet_set_size(pkt, size); + + return pkt; +} + +static int +test_dpdk_packet_insert_headroom(void) { + struct dp_packet *pkt = dpdk_mp_alloc_pkt(mp); + ovs_assert(pkt != NULL); + + /* Reserve 256B of header */ + size_t str_len = 512; + dp_packet_reserve(pkt, str_len); + char *p = dp_packet_push_uninit(pkt, str_len); + ovs_assert(p != NULL); + /* Put the first 512B of "test_str" in the allocated header */ + memcpy(p, test_str, str_len); + + /* Check properties and data are as expected */ + assert_single_mbuf(pkt, RTE_PKTMBUF_HEADROOM, str_len); + assert_data(pkt, 0, str_len); + + dp_packet_uninit(pkt); + + return 0; +} + +static int +test_dpdk_packet_insert_tailroom_and_headroom(void) { + struct dp_packet *pkt = dpdk_mp_alloc_pkt(mp); + ovs_assert(pkt != NULL); + + /* Put the first 512B of "test_str" in the packet's header */ + size_t str_len = 512; + char *p = dp_packet_put(pkt, test_str, str_len); + ovs_assert(p != NULL); + /* Allocate extra 256B of header */ + size_t head_len = 256; + p = dp_packet_push_uninit(pkt, head_len); + ovs_assert(p != NULL); + + /* Check properties and data are as expected */ + assert_single_mbuf(pkt, 0, str_len + head_len); + + /* Check the data inserted in the packet is correct */ + char data[str_len + head_len]; + const char *rd = rte_pktmbuf_read(&pkt->mbuf, 0, str_len + head_len, data); + ovs_assert(rd != NULL); + /* Because of the headroom inserted, the data now begin at offset 256 */ + ovs_assert(memcmp(rd + head_len, test_str, str_len) == 0); + + dp_packet_uninit(pkt); + + return 0; +} + +static int +test_dpdk_packet_insert_tailroom_and_headroom_multiple_mbufs(void) { + struct dp_packet *pkt = dpdk_mp_alloc_pkt(mp); + ovs_assert(pkt != NULL); + + /* Put the first 2050B of "test_str" in the packet, just enought to + * allocate two mbufs */ + size_t str_len = MBUF_DATA_LEN + 2; + pkt = dpdk_pkt_put(pkt, test_str, str_len); + ovs_assert(pkt != NULL); + + /* Put the first 512B of "test_str" in the packet's header */ + size_t tail_len = 512; + char *p = dp_packet_put(pkt, test_str, tail_len); + ovs_assert(p != NULL); + + /* Allocate extra 256B of header */ + size_t head_len = 256; + p = dp_packet_push_uninit(pkt, head_len); + ovs_assert(p != NULL); + /* Copy the data to the reserved headroom */ + memcpy(p, test_str, head_len); + + /* Check properties and data are as expected */ + size_t pkt_len = head_len + str_len + tail_len; + uint16_t nb_segs = 2; + assert_multiple_mbufs(pkt, 0, pkt_len, nb_segs); + + /* Check the data inserted in the packet is correct */ + char data[pkt_len]; + const char *rd = rte_pktmbuf_read(&pkt->mbuf, 0, pkt_len, data); + ovs_assert(rd != NULL); + ovs_assert(memcmp(rd, test_str, head_len) == 0); + ovs_assert(memcmp(rd + head_len + str_len, test_str, tail_len) == 0); + + dp_packet_uninit(pkt); + + return 0; +} + +static int +test_dpdk_packet_insert_tailroom_multiple_mbufs(void) { + struct dp_packet *pkt = dpdk_mp_alloc_pkt(mp); + ovs_assert(pkt != NULL); + + /* Put the first 2050B of "test_str" in the packet, just enought to + * allocate two mbufs */ + size_t str_len = MBUF_DATA_LEN + 2; + pkt = dpdk_pkt_put(pkt, test_str, str_len); + ovs_assert(pkt != NULL); + + /* Put the first 2000B of "test_str" in the packet's end */ + size_t tail_len = 2000; + char *p = dp_packet_put(pkt, test_str, tail_len); + ovs_assert(p != NULL); + + /* Check properties and data are as expected */ + char data[str_len + tail_len]; + const char *rd = rte_pktmbuf_read(&pkt->mbuf, 0, str_len + tail_len, data); + ovs_assert(rd != NULL); + /* Because of the headroom inserted, the data now begin at offset 256 */ + ovs_assert(memcmp(rd + str_len, test_str, tail_len) == 0); + + dp_packet_uninit(pkt); + + return 0; +} + +static int +test_dpdk_packet_insert_headroom_multiple_mbufs(void) { + struct dp_packet *pkt = dpdk_mp_alloc_pkt(mp); + ovs_assert(pkt != NULL); + + /* Put the first 2050B of "test_str" in the packet, just enought to + * allocate two mbufs */ + size_t str_len = MBUF_DATA_LEN + 2; + pkt = dpdk_pkt_put(pkt, test_str, str_len); + + /* Push the first 2000B of "test_str" in the packet's header */ + size_t head_len = 2000; + char *p = dp_packet_push_uninit(pkt, head_len); + ovs_assert(p != NULL); + + /* Check properties and data are as expected */ + char data[str_len + head_len]; + const char *rd = rte_pktmbuf_read(&pkt->mbuf, 0, str_len + head_len, data); + ovs_assert(rd != NULL); + /* Because of the headroom inserted, the data is at offset 'head_len' */ + ovs_assert(memcmp(rd + head_len, test_str, str_len) == 0); + + dp_packet_uninit(pkt); + + return 0; +} + +static int +test_dpdk_packet_change_size(void) { + struct dp_packet *pkt = dpdk_mp_alloc_pkt(mp); + ovs_assert(pkt != NULL); + + /* Put enough data in the packet that spans three mbufs (5120B) */ + size_t str_len = MBUF_DATA_LEN * 2 + 1024; + pkt = dpdk_pkt_put(pkt, test_str, str_len); + ovs_assert(pkt != NULL); + + /* Check properties and data are as expected */ + uint16_t nb_segs = 3; + assert_multiple_mbufs(pkt, RTE_PKTMBUF_HEADROOM, str_len, nb_segs); + + /* Change the size of the packet to fit in a single mbuf */ + dp_packet_clear(pkt); + + assert_single_mbuf(pkt, RTE_PKTMBUF_HEADROOM, 0); + + dp_packet_uninit(pkt); + + return 0; +} + +/* Shift() tests */ + +static int +test_dpdk_packet_shift_single_mbuf(void) { + struct dp_packet *pkt = dpdk_mp_alloc_pkt(mp); + ovs_assert(pkt != NULL); + + /* Put the first 1024B of "test_str" in the packet */ + size_t str_len = 1024; + char *p = dp_packet_put(pkt, test_str, str_len); + ovs_assert(p != NULL); + + /* Shift data right by 512B */ + uint16_t shift_len = 512; + dp_packet_shift(pkt, shift_len); + + /* Check properties and data are as expected */ + assert_single_mbuf(pkt, RTE_PKTMBUF_HEADROOM + shift_len, str_len); + assert_data(pkt, 0, str_len); + + dp_packet_uninit(pkt); + + return 0; +} + +static int +test_dpdk_packet_shift_multiple_mbufs(void) { + struct dp_packet *pkt = dpdk_mp_alloc_pkt(mp); + ovs_assert(pkt != NULL); + + /* Put the data in "test_str" in the packet */ + size_t str_len = strlen(test_str); + pkt = dpdk_pkt_put(pkt, test_str, str_len); + ovs_assert(pkt != NULL); + + /* Check properties and data are as expected */ + uint16_t nb_segs = 3; + assert_multiple_mbufs(pkt, RTE_PKTMBUF_HEADROOM, str_len, nb_segs); + + /* Shift data right by 1024B */ + uint16_t shift_len = 1024; + dp_packet_shift(pkt, shift_len); + + /* Check the data has been inserted correctly */ + assert_multiple_mbufs(pkt, RTE_PKTMBUF_HEADROOM + shift_len, str_len, + nb_segs); + assert_data(pkt, 0, str_len); + + dp_packet_uninit(pkt); + + return 0; +} + +static int +test_dpdk_packet_shift_right_then_left(void) { + struct dp_packet *pkt = dpdk_mp_alloc_pkt(mp); + ovs_assert(pkt != NULL); + + /* Put the first 1024B of "test_str" in the packet */ + size_t str_len = strlen(test_str); + pkt = dpdk_pkt_put(pkt, test_str, str_len); + ovs_assert(pkt != NULL); + + /* Shift data right by 1024B */ + int16_t shift_len = 1024; + dp_packet_shift(pkt, 1024); + + /* Check properties and data are as expected */ + uint16_t nb_segs = 3; + assert_multiple_mbufs(pkt, RTE_PKTMBUF_HEADROOM + shift_len, str_len, + nb_segs); + + /* Shift data left by 512B */ + dp_packet_shift(pkt, -shift_len); + + /* We negative shift_len (-shift_len) since */ + assert_multiple_mbufs(pkt, RTE_PKTMBUF_HEADROOM, str_len, + nb_segs); + assert_data(pkt, 0, str_len); + + dp_packet_uninit(pkt); + + return 0; +} + +static int +test_dpdk_packet(int argc OVS_UNUSED, char *argv[] OVS_UNUSED) +{ + /* Setup environment for tests */ + dpdk_setup_eal_with_mp(); + set_testing_pattern_str(); + + test_dpdk_packet_insert_headroom(); + num_tests++; + test_dpdk_packet_insert_tailroom_and_headroom(); + num_tests++; + test_dpdk_packet_insert_tailroom_multiple_mbufs(); + num_tests++; + test_dpdk_packet_insert_headroom_multiple_mbufs(); + num_tests++; + test_dpdk_packet_insert_tailroom_and_headroom_multiple_mbufs(); + num_tests++; + test_dpdk_packet_change_size(); + num_tests++; + test_dpdk_packet_shift_single_mbuf(); + num_tests++; + test_dpdk_packet_shift_multiple_mbufs(); + num_tests++; + test_dpdk_packet_shift_right_then_left(); + num_tests++; + + printf("Executed %d tests\n", num_tests); + + exit(0); +} + +OVSTEST_REGISTER("test-dpdk-packet", test_dpdk_packet); From patchwork Wed Jul 4 18:06:34 2018 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: "Lam, Tiago" X-Patchwork-Id: 939516 X-Patchwork-Delegate: ian.stokes@intel.com Return-Path: X-Original-To: incoming@patchwork.ozlabs.org Delivered-To: patchwork-incoming@bilbo.ozlabs.org Authentication-Results: ozlabs.org; spf=pass (mailfrom) smtp.mailfrom=openvswitch.org (client-ip=140.211.169.12; helo=mail.linuxfoundation.org; envelope-from=ovs-dev-bounces@openvswitch.org; receiver=) Authentication-Results: ozlabs.org; dmarc=fail (p=none dis=none) header.from=intel.com Received: from mail.linuxfoundation.org (mail.linuxfoundation.org [140.211.169.12]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by ozlabs.org (Postfix) with ESMTPS id 41LTd30nxmz9s29 for ; Thu, 5 Jul 2018 04:13:27 +1000 (AEST) Received: from mail.linux-foundation.org (localhost [127.0.0.1]) by mail.linuxfoundation.org (Postfix) with ESMTP id 44F70D6A; Wed, 4 Jul 2018 18:07:25 +0000 (UTC) X-Original-To: ovs-dev@openvswitch.org Delivered-To: ovs-dev@mail.linuxfoundation.org Received: from smtp1.linuxfoundation.org (smtp1.linux-foundation.org [172.17.192.35]) by mail.linuxfoundation.org (Postfix) with ESMTPS id 0D3FAD06 for ; Wed, 4 Jul 2018 18:07:24 +0000 (UTC) X-Greylist: domain auto-whitelisted by SQLgrey-1.7.6 Received: from mga11.intel.com (mga11.intel.com [192.55.52.93]) by smtp1.linuxfoundation.org (Postfix) with ESMTPS id 8DD38782 for ; Wed, 4 Jul 2018 18:07:21 +0000 (UTC) X-Amp-Result: SKIPPED(no attachment in message) X-Amp-File-Uploaded: False Received: from orsmga001.jf.intel.com ([10.7.209.18]) by fmsmga102.fm.intel.com with ESMTP/TLS/DHE-RSA-AES256-GCM-SHA384; 04 Jul 2018 11:07:19 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.51,308,1526367600"; d="scan'208";a="70109117" Received: from silpixa00399125.ir.intel.com ([10.237.223.34]) by orsmga001.jf.intel.com with ESMTP; 04 Jul 2018 11:07:06 -0700 From: Tiago Lam To: ovs-dev@openvswitch.org Date: Wed, 4 Jul 2018 19:06:34 +0100 Message-Id: <1530727595-217860-14-git-send-email-tiago.lam@intel.com> X-Mailer: git-send-email 2.7.4 In-Reply-To: <1530727595-217860-1-git-send-email-tiago.lam@intel.com> References: <1530727595-217860-1-git-send-email-tiago.lam@intel.com> X-Spam-Status: No, score=-6.9 required=5.0 tests=BAYES_00,RCVD_IN_DNSWL_HI autolearn=ham version=3.3.1 X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on smtp1.linux-foundation.org Cc: i.maximets@samsung.com Subject: [ovs-dev] [PATCH v2 13/14] dpdk-tests: Accept other configs in OVS_DPDK_START X-BeenThere: ovs-dev@openvswitch.org X-Mailman-Version: 2.1.12 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , MIME-Version: 1.0 Sender: ovs-dev-bounces@openvswitch.org Errors-To: ovs-dev-bounces@openvswitch.org As it stands, OVS_DPDK_START() won't allow other configs to be set before starting the ovs-vswitchd daemon. This is a problem since some configs, such as the "dpdk-multi-seg-mbufs=true" for enabling the multi-segment mbufs, need to be set prior to start OvS. To support other options, OVS_DPDK_START() has been modified to accept extra configs in the form "$config_name=$config_value". It then uses ovs-vsctl to set the configs. Signed-off-by: Tiago Lam Acked-by: Eelco Chaudron --- tests/system-dpdk-macros.at | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) diff --git a/tests/system-dpdk-macros.at b/tests/system-dpdk-macros.at index 0762ee0..485a0f3 100644 --- a/tests/system-dpdk-macros.at +++ b/tests/system-dpdk-macros.at @@ -21,7 +21,7 @@ m4_define([OVS_DPDK_PRE_CHECK], ]) -# OVS_DPDK_START() +# OVS_DPDK_START([other-conf-args]) # # Create an empty database and start ovsdb-server. Add special configuration # dpdk-init to enable DPDK functionality. Start ovs-vswitchd connected to that @@ -48,6 +48,11 @@ m4_define([OVS_DPDK_START], AT_CHECK([lscpu], [], [stdout]) AT_CHECK([cat stdout | grep "NUMA node(s)" | awk '{c=1; while (c++<$(3)) {printf "1024,"}; print "1024"}' > SOCKET_MEM]) AT_CHECK([ovs-vsctl --no-wait set Open_vSwitch . other_config:dpdk-socket-mem="$(cat SOCKET_MEM)"]) + dnl Iterate through $other-conf-args list and include them + m4_foreach_w(opt, $1, [ + AT_CHECK([ovs-vsctl --no-wait set Open_vSwitch . other_config:opt]) + ]) + dnl m4_if([$1], [], [], [AT_CHECK(["$1"])]) dnl Start ovs-vswitchd. AT_CHECK([ovs-vswitchd --detach --no-chdir --pidfile --log-file -vvconn -vofproto_dpif -vunixctl], [0], [stdout], [stderr]) From patchwork Wed Jul 4 18:06:35 2018 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: "Lam, Tiago" X-Patchwork-Id: 939515 X-Patchwork-Delegate: ian.stokes@intel.com Return-Path: X-Original-To: incoming@patchwork.ozlabs.org Delivered-To: patchwork-incoming@bilbo.ozlabs.org Authentication-Results: ozlabs.org; spf=pass (mailfrom) smtp.mailfrom=openvswitch.org (client-ip=140.211.169.12; helo=mail.linuxfoundation.org; envelope-from=ovs-dev-bounces@openvswitch.org; receiver=) Authentication-Results: ozlabs.org; dmarc=fail (p=none dis=none) header.from=intel.com Received: from mail.linuxfoundation.org (mail.linuxfoundation.org [140.211.169.12]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by ozlabs.org (Postfix) with ESMTPS id 41LTcX0wG4z9s29 for ; Thu, 5 Jul 2018 04:13:00 +1000 (AEST) Received: from mail.linux-foundation.org (localhost [127.0.0.1]) by mail.linuxfoundation.org (Postfix) with ESMTP id 86EE9D5D; Wed, 4 Jul 2018 18:07:23 +0000 (UTC) X-Original-To: ovs-dev@openvswitch.org Delivered-To: ovs-dev@mail.linuxfoundation.org Received: from smtp1.linuxfoundation.org (smtp1.linux-foundation.org [172.17.192.35]) by mail.linuxfoundation.org (Postfix) with ESMTPS id E363FD60 for ; Wed, 4 Jul 2018 18:07:22 +0000 (UTC) X-Greylist: domain auto-whitelisted by SQLgrey-1.7.6 Received: from mga11.intel.com (mga11.intel.com [192.55.52.93]) by smtp1.linuxfoundation.org (Postfix) with ESMTPS id C14E878F for ; Wed, 4 Jul 2018 18:07:21 +0000 (UTC) X-Amp-Result: SKIPPED(no attachment in message) X-Amp-File-Uploaded: False Received: from orsmga001.jf.intel.com ([10.7.209.18]) by fmsmga102.fm.intel.com with ESMTP/TLS/DHE-RSA-AES256-GCM-SHA384; 04 Jul 2018 11:07:20 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.51,308,1526367600"; d="scan'208";a="70109122" Received: from silpixa00399125.ir.intel.com ([10.237.223.34]) by orsmga001.jf.intel.com with ESMTP; 04 Jul 2018 11:07:07 -0700 From: Tiago Lam To: ovs-dev@openvswitch.org Date: Wed, 4 Jul 2018 19:06:35 +0100 Message-Id: <1530727595-217860-15-git-send-email-tiago.lam@intel.com> X-Mailer: git-send-email 2.7.4 In-Reply-To: <1530727595-217860-1-git-send-email-tiago.lam@intel.com> References: <1530727595-217860-1-git-send-email-tiago.lam@intel.com> X-Spam-Status: No, score=-6.9 required=5.0 tests=BAYES_00,RCVD_IN_DNSWL_HI autolearn=ham version=3.3.1 X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on smtp1.linux-foundation.org Cc: i.maximets@samsung.com Subject: [ovs-dev] [PATCH v2 14/14] dpdk-tests: End-to-end tests for multi-seg mbufs. X-BeenThere: ovs-dev@openvswitch.org X-Mailman-Version: 2.1.12 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , MIME-Version: 1.0 Sender: ovs-dev-bounces@openvswitch.org Errors-To: ovs-dev-bounces@openvswitch.org The following tests are added to the DPDK testsuite to add some coverage for the multi-segment mbufs: - Check that multi-segment mbufs are disabled by default; - Check that providing `other_config:dpdk-multi-seg-mbufs=true` indeed enables mbufs; - Using a DPDK port, send a random packet out and check that `ofctl dump-flows` shows the correct amount of packets and bytes sent. Signed-off-by: Tiago Lam Acked-by: Eelco Chaudron --- tests/system-dpdk.at | 65 ++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 65 insertions(+) diff --git a/tests/system-dpdk.at b/tests/system-dpdk.at index 3d21b01..7468d49 100644 --- a/tests/system-dpdk.at +++ b/tests/system-dpdk.at @@ -71,3 +71,68 @@ OVS_VSWITCHD_STOP("/does not exist. The Open vSwitch kernel module is probably n ") AT_CLEANUP dnl -------------------------------------------------------------------------- + +AT_SETUP([Jumbo frames - Multi-segment disabled by default]) +OVS_DPDK_START() + +AT_CHECK([grep "multi-segment mbufs enabled" ovs-vswitchd.log], [1], []) +OVS_VSWITCHD_STOP("/Global register is changed during/d +/EAL: No free hugepages reported in hugepages-1048576kB/d +") +AT_CLEANUP + +AT_SETUP([Jumbo frames - Multi-segment enabled]) +OVS_DPDK_START([dpdk-multi-seg-mbufs=true]) +AT_CHECK([grep "multi-segment mbufs enabled" ovs-vswitchd.log], [], [stdout]) +OVS_VSWITCHD_STOP("/Global register is changed during/d +/EAL: No free hugepages reported in hugepages-1048576kB/d +") +AT_CLEANUP + +AT_SETUP([Jumbo frames - Multi-segment mbufs Tx]) +OVS_DPDK_PRE_CHECK() +OVS_DPDK_START([dpdk-multi-seg-mbufs=true]) + +dnl Add userspace bridge and attach it to OVS +AT_CHECK([ovs-vsctl add-br br10 -- set bridge br10 datapath_type=netdev]) +AT_CHECK([ovs-vsctl add-port br10 dpdk0 \ + -- set Interface dpdk0 type=dpdk options:dpdk-devargs=$(cat PCI_ADDR) \ + -- set Interface dpdk0 mtu_request=9000], [], [stdout], [stderr]) + +AT_CHECK([ovs-vsctl show], [], [stdout]) + +dnl Add flows to send packets out from the 'dpdk0' port +AT_CHECK([ +ovs-ofctl del-flows br10 +ovs-ofctl add-flow br10 in_port=LOCAL,actions=output:dpdk0 +], [], [stdout]) + +AT_CHECK([ovs-ofctl dump-flows br10], [], [stdout]) + +dnl Send packet out, of the 'dpdk0' port +AT_CHECK([ +ARP_HEADER="000009000B00000009000A00080600010800060400010000000000010A0000\ +010000000000020A000002" +dnl Build a random hex string to append to the ARP_HEADER +RANDOM_BODY=$(printf '0102030405%.0s' {1..1750}) +dnl 8792B ARP packet +RANDOM_ARP="$ARP_HEADER$RANDOM_BODY" + +ovs-ofctl packet-out br10 "packet=$RANDOM_ARP,action=resubmit:LOCAL" +], [], [stdout]) + +AT_CHECK([ovs-ofctl dump-flows br10], [0], [stdout]) + +dnl Confirm the single packet as been sent with correct size +AT_CHECK([ovs-ofctl dump-flows br10 | ofctl_strip | grep in_port], [0], [dnl + n_packets=1, n_bytes=8792, in_port=LOCAL actions=output:1 +]) + +dnl Clean up +OVS_VSWITCHD_STOP("/does not exist. The Open vSwitch kernel module is probably not loaded./d +/Failed to enable flow control/d +/failed to connect to \/tmp\/dpdkvhostclient0: No such file or directory/d +/Global register is changed during/d +/EAL: No free hugepages reported in hugepages-1048576kB/d +") +AT_CLEANUP