From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mails.dpdk.org (mails.dpdk.org [217.70.189.124]) by inbox.dpdk.org (Postfix) with ESMTP id 65EE6A0C4D; Fri, 15 Oct 2021 22:12:16 +0200 (CEST) Received: from [217.70.189.124] (localhost [127.0.0.1]) by mails.dpdk.org (Postfix) with ESMTP id DD22441195; Fri, 15 Oct 2021 22:11:47 +0200 (CEST) Received: from mail-pl1-f179.google.com (mail-pl1-f179.google.com [209.85.214.179]) by mails.dpdk.org (Postfix) with ESMTP id CE21941185 for ; Fri, 15 Oct 2021 22:11:45 +0200 (CEST) Received: by mail-pl1-f179.google.com with SMTP id w14so7071170pll.2 for ; Fri, 15 Oct 2021 13:11:45 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=networkplumber-org.20210112.gappssmtp.com; s=20210112; h=from:to:cc:subject:date:message-id:in-reply-to:references :mime-version:content-transfer-encoding; bh=73Uw2WJ8IHkweyVKecnSaBpIOk+YlqeaMiwRgK9+ayg=; b=UsUOYNx/zyXet0BWDlaA2/TgahZmJGxOgTAKQu1u5rd6m1JR1pf/FAh2mrIAkAR6Ef wiDbggU0O3EBhFc2G0VeLEfYpXBOrp3HeUSq9x3PULp81W9sRlIyPwpOZpLhw/ae6gUS Axw7Hhu0IME5H3paEe3OFfOfMyyaEP/oWSxCsVYzxIxuvhJZPNNAt1kvN0/tjI5bESe0 grsbn8Y2tSKP2pjnLG8JlnoYNMO8kDNtIkRtaqFDVKAOfMwkg6K3nKOV+1XRIHM3FYqv b1hcbkC10z7SEV5Zeuw68thu2LAPI7Dl+K8LmoUJ0z8K3aLckoV/yzW9rCevq5q1HWgs LiIA== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20210112; h=x-gm-message-state:from:to:cc:subject:date:message-id:in-reply-to :references:mime-version:content-transfer-encoding; bh=73Uw2WJ8IHkweyVKecnSaBpIOk+YlqeaMiwRgK9+ayg=; b=qmCYNnzzmhvZAhk5T6vOegstoUqu7L61PM6oesC7mEAgXyP0gNvMV9a+JvRODh9MIO ihMcbqB6J6FIwxeS9iTbxspOtqyS1idmjsAn6z0pAv8JTBegZO10LBHrI0f1SK9dWqcG aF7eISgmGfljKaLLtsD/DSm81aa2iynfXoA9fMES+OepqK+GFlj4Hq5CcCkOrzF924vn gt6UKpng07rDbY5HwRN38gQmbi5EjAthNaaOCRlQod265kDAzxWruf8Har8TMhRB1s3J HzzEfS6Cy76hTm8yqUwmCYU+HPHV5yQ+1eikIfb8m+ogOEVfxMZgA3GX62OEL2KIwncg K3OA== X-Gm-Message-State: AOAM5326jGbZyH/O0YHcTpR9t3HyQ+hUTdDQ8Yao6t0qiRqWrKlpUCK5 ZKXUtWP86Qzdpe83KBsHDaD8neQ6QDNvJA== X-Google-Smtp-Source: ABdhPJwbzJWh59w4zm5WDZ1pcPu6WsFwlP6SB8y+D6hIz1LzhDLSUxzynFNqCQaqWbxGDiTT+Bx3xQ== X-Received: by 2002:a17:902:7797:b0:13f:974a:db07 with SMTP id o23-20020a170902779700b0013f974adb07mr2858514pll.64.1634328704129; Fri, 15 Oct 2021 13:11:44 -0700 (PDT) Received: from hermes.local (204-195-33-123.wavecable.com. [204.195.33.123]) by smtp.gmail.com with ESMTPSA id g4sm5250456pgs.42.2021.10.15.13.11.42 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Fri, 15 Oct 2021 13:11:43 -0700 (PDT) From: Stephen Hemminger To: dev@dpdk.org Cc: Stephen Hemminger , Reshma Pattan , Ray Kinsella , Anatoly Burakov Date: Fri, 15 Oct 2021 13:11:23 -0700 Message-Id: <20211015201129.63220-7-stephen@networkplumber.org> X-Mailer: git-send-email 2.30.2 In-Reply-To: <20211015201129.63220-1-stephen@networkplumber.org> References: <20210903004732.109023-1-stephen@networkplumber.org> <20211015201129.63220-1-stephen@networkplumber.org> MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Subject: [dpdk-dev] [PATCH v14 06/12] pdump: support pcapng and filtering X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: DPDK patches and discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: dev-bounces@dpdk.org Sender: "dev" This enhances the DPDK pdump library to support new pcapng format and filtering via BPF. The internal client/server protocol is changed to support two versions: the original pdump basic version and a new pcapng version. The internal version number (not part of exposed API or ABI) is intentionally increased to cause any attempt to try mismatched primary/secondary process to fail. Add new API to do allow filtering of captured packets with DPDK BPF (eBPF) filter program. It keeps statistics on packets captured, filtered, and missed (because ring was full). Signed-off-by: Stephen Hemminger Acked-by: Reshma Pattan --- lib/meson.build | 4 +- lib/pdump/meson.build | 2 +- lib/pdump/rte_pdump.c | 432 ++++++++++++++++++++++++++++++------------ lib/pdump/rte_pdump.h | 113 ++++++++++- lib/pdump/version.map | 8 + 5 files changed, 433 insertions(+), 126 deletions(-) diff --git a/lib/meson.build b/lib/meson.build index 15150efa19a7..c71c6917dbb7 100644 --- a/lib/meson.build +++ b/lib/meson.build @@ -27,6 +27,7 @@ libraries = [ 'acl', 'bbdev', 'bitratestats', + 'bpf', 'cfgfile', 'compressdev', 'cryptodev', @@ -43,7 +44,6 @@ libraries = [ 'member', 'pcapng', 'power', - 'pdump', 'rawdev', 'regexdev', 'rib', @@ -55,10 +55,10 @@ libraries = [ 'ipsec', # ipsec lib depends on net, crypto and security 'fib', #fib lib depends on rib 'port', # pkt framework libs which use other libs from above + 'pdump', # pdump lib depends on bpf 'table', 'pipeline', 'flow_classify', # flow_classify lib depends on pkt framework table lib - 'bpf', 'graph', 'node', ] diff --git a/lib/pdump/meson.build b/lib/pdump/meson.build index 3a95eabde6a6..51ceb2afdec5 100644 --- a/lib/pdump/meson.build +++ b/lib/pdump/meson.build @@ -3,4 +3,4 @@ sources = files('rte_pdump.c') headers = files('rte_pdump.h') -deps += ['ethdev'] +deps += ['ethdev', 'bpf', 'pcapng'] diff --git a/lib/pdump/rte_pdump.c b/lib/pdump/rte_pdump.c index 382217bc1564..2636a216994b 100644 --- a/lib/pdump/rte_pdump.c +++ b/lib/pdump/rte_pdump.c @@ -7,8 +7,10 @@ #include #include #include +#include #include #include +#include #include "rte_pdump.h" @@ -27,30 +29,23 @@ enum pdump_operation { ENABLE = 2 }; +/* Internal version number in request */ enum pdump_version { - V1 = 1 + V1 = 1, /* no filtering or snap */ + V2 = 2, }; struct pdump_request { uint16_t ver; uint16_t op; uint32_t flags; - union pdump_data { - struct enable_v1 { - char device[RTE_DEV_NAME_MAX_LEN]; - uint16_t queue; - struct rte_ring *ring; - struct rte_mempool *mp; - void *filter; - } en_v1; - struct disable_v1 { - char device[RTE_DEV_NAME_MAX_LEN]; - uint16_t queue; - struct rte_ring *ring; - struct rte_mempool *mp; - void *filter; - } dis_v1; - } data; + char device[RTE_DEV_NAME_MAX_LEN]; + uint16_t queue; + struct rte_ring *ring; + struct rte_mempool *mp; + + const struct rte_bpf_prm *prm; + uint32_t snaplen; }; struct pdump_response { @@ -63,80 +58,140 @@ static struct pdump_rxtx_cbs { struct rte_ring *ring; struct rte_mempool *mp; const struct rte_eth_rxtx_callback *cb; - void *filter; + const struct rte_bpf *filter; + enum pdump_version ver; + uint32_t snaplen; } rx_cbs[RTE_MAX_ETHPORTS][RTE_MAX_QUEUES_PER_PORT], tx_cbs[RTE_MAX_ETHPORTS][RTE_MAX_QUEUES_PER_PORT]; -static inline void -pdump_copy(struct rte_mbuf **pkts, uint16_t nb_pkts, void *user_params) +/* + * The packet capture statistics keep track of packets + * accepted, filtered and dropped. These are per-queue + * and in memory between primary and secondary processes. + */ +static const char MZ_RTE_PDUMP_STATS[] = "rte_pdump_stats"; +static struct { + struct rte_pdump_stats rx[RTE_MAX_ETHPORTS][RTE_MAX_QUEUES_PER_PORT]; + struct rte_pdump_stats tx[RTE_MAX_ETHPORTS][RTE_MAX_QUEUES_PER_PORT]; +} *pdump_stats; + +/* Create a clone of mbuf to be placed into ring. */ +static void +pdump_copy(uint16_t port_id, uint16_t queue, + enum rte_pcapng_direction direction, + struct rte_mbuf **pkts, uint16_t nb_pkts, + const struct pdump_rxtx_cbs *cbs, + struct rte_pdump_stats *stats) { unsigned int i; int ring_enq; uint16_t d_pkts = 0; struct rte_mbuf *dup_bufs[nb_pkts]; - struct pdump_rxtx_cbs *cbs; + uint64_t ts; struct rte_ring *ring; struct rte_mempool *mp; struct rte_mbuf *p; + uint64_t rcs[nb_pkts]; + + if (cbs->filter) + rte_bpf_exec_burst(cbs->filter, (void **)pkts, rcs, nb_pkts); - cbs = user_params; + ts = rte_get_tsc_cycles(); ring = cbs->ring; mp = cbs->mp; for (i = 0; i < nb_pkts; i++) { - p = rte_pktmbuf_copy(pkts[i], mp, 0, UINT32_MAX); - if (p) + /* + * This uses same BPF return value convention as socket filter + * and pcap_offline_filter. + * if program returns zero + * then packet doesn't match the filter (will be ignored). + */ + if (cbs->filter && rcs[i] == 0) { + __atomic_fetch_add(&stats->filtered, + 1, __ATOMIC_RELAXED); + continue; + } + + /* + * If using pcapng then want to wrap packets + * otherwise a simple copy. + */ + if (cbs->ver == V2) + p = rte_pcapng_copy(port_id, queue, + pkts[i], mp, cbs->snaplen, + ts, direction); + else + p = rte_pktmbuf_copy(pkts[i], mp, 0, cbs->snaplen); + + if (unlikely(p == NULL)) + __atomic_fetch_add(&stats->nombuf, 1, __ATOMIC_RELAXED); + else dup_bufs[d_pkts++] = p; } + __atomic_fetch_add(&stats->accepted, d_pkts, __ATOMIC_RELAXED); + ring_enq = rte_ring_enqueue_burst(ring, (void *)dup_bufs, d_pkts, NULL); if (unlikely(ring_enq < d_pkts)) { unsigned int drops = d_pkts - ring_enq; - PDUMP_LOG(DEBUG, - "only %d of packets enqueued to ring\n", ring_enq); + __atomic_fetch_add(&stats->ringfull, drops, __ATOMIC_RELAXED); rte_pktmbuf_free_bulk(&dup_bufs[ring_enq], drops); } } static uint16_t -pdump_rx(uint16_t port __rte_unused, uint16_t qidx __rte_unused, +pdump_rx(uint16_t port, uint16_t queue, struct rte_mbuf **pkts, uint16_t nb_pkts, - uint16_t max_pkts __rte_unused, - void *user_params) + uint16_t max_pkts __rte_unused, void *user_params) { - pdump_copy(pkts, nb_pkts, user_params); + const struct pdump_rxtx_cbs *cbs = user_params; + struct rte_pdump_stats *stats = &pdump_stats->rx[port][queue]; + + pdump_copy(port, queue, RTE_PCAPNG_DIRECTION_IN, + pkts, nb_pkts, cbs, stats); return nb_pkts; } static uint16_t -pdump_tx(uint16_t port __rte_unused, uint16_t qidx __rte_unused, +pdump_tx(uint16_t port, uint16_t queue, struct rte_mbuf **pkts, uint16_t nb_pkts, void *user_params) { - pdump_copy(pkts, nb_pkts, user_params); + const struct pdump_rxtx_cbs *cbs = user_params; + struct rte_pdump_stats *stats = &pdump_stats->tx[port][queue]; + + pdump_copy(port, queue, RTE_PCAPNG_DIRECTION_OUT, + pkts, nb_pkts, cbs, stats); return nb_pkts; } static int -pdump_register_rx_callbacks(uint16_t end_q, uint16_t port, uint16_t queue, - struct rte_ring *ring, struct rte_mempool *mp, - uint16_t operation) +pdump_register_rx_callbacks(enum pdump_version ver, + uint16_t end_q, uint16_t port, uint16_t queue, + struct rte_ring *ring, struct rte_mempool *mp, + struct rte_bpf *filter, + uint16_t operation, uint32_t snaplen) { uint16_t qid; - struct pdump_rxtx_cbs *cbs = NULL; qid = (queue == RTE_PDUMP_ALL_QUEUES) ? 0 : queue; for (; qid < end_q; qid++) { - cbs = &rx_cbs[port][qid]; - if (cbs && operation == ENABLE) { + struct pdump_rxtx_cbs *cbs = &rx_cbs[port][qid]; + + if (operation == ENABLE) { if (cbs->cb) { PDUMP_LOG(ERR, "rx callback for port=%d queue=%d, already exists\n", port, qid); return -EEXIST; } + cbs->ver = ver; cbs->ring = ring; cbs->mp = mp; + cbs->snaplen = snaplen; + cbs->filter = filter; + cbs->cb = rte_eth_add_first_rx_callback(port, qid, pdump_rx, cbs); if (cbs->cb == NULL) { @@ -145,8 +200,7 @@ pdump_register_rx_callbacks(uint16_t end_q, uint16_t port, uint16_t queue, rte_errno); return rte_errno; } - } - if (cbs && operation == DISABLE) { + } else if (operation == DISABLE) { int ret; if (cbs->cb == NULL) { @@ -170,26 +224,32 @@ pdump_register_rx_callbacks(uint16_t end_q, uint16_t port, uint16_t queue, } static int -pdump_register_tx_callbacks(uint16_t end_q, uint16_t port, uint16_t queue, - struct rte_ring *ring, struct rte_mempool *mp, - uint16_t operation) +pdump_register_tx_callbacks(enum pdump_version ver, + uint16_t end_q, uint16_t port, uint16_t queue, + struct rte_ring *ring, struct rte_mempool *mp, + struct rte_bpf *filter, + uint16_t operation, uint32_t snaplen) { uint16_t qid; - struct pdump_rxtx_cbs *cbs = NULL; qid = (queue == RTE_PDUMP_ALL_QUEUES) ? 0 : queue; for (; qid < end_q; qid++) { - cbs = &tx_cbs[port][qid]; - if (cbs && operation == ENABLE) { + struct pdump_rxtx_cbs *cbs = &tx_cbs[port][qid]; + + if (operation == ENABLE) { if (cbs->cb) { PDUMP_LOG(ERR, "tx callback for port=%d queue=%d, already exists\n", port, qid); return -EEXIST; } + cbs->ver = ver; cbs->ring = ring; cbs->mp = mp; + cbs->snaplen = snaplen; + cbs->filter = filter; + cbs->cb = rte_eth_add_tx_callback(port, qid, pdump_tx, cbs); if (cbs->cb == NULL) { @@ -198,8 +258,7 @@ pdump_register_tx_callbacks(uint16_t end_q, uint16_t port, uint16_t queue, rte_errno); return rte_errno; } - } - if (cbs && operation == DISABLE) { + } else if (operation == DISABLE) { int ret; if (cbs->cb == NULL) { @@ -228,37 +287,47 @@ set_pdump_rxtx_cbs(const struct pdump_request *p) uint16_t nb_rx_q = 0, nb_tx_q = 0, end_q, queue; uint16_t port; int ret = 0; + struct rte_bpf *filter = NULL; uint32_t flags; uint16_t operation; struct rte_ring *ring; struct rte_mempool *mp; - flags = p->flags; - operation = p->op; - if (operation == ENABLE) { - ret = rte_eth_dev_get_port_by_name(p->data.en_v1.device, - &port); - if (ret < 0) { + /* Check for possible DPDK version mismatch */ + if (!(p->ver == V1 || p->ver == V2)) { + PDUMP_LOG(ERR, + "incorrect client version %u\n", p->ver); + return -EINVAL; + } + + if (p->prm) { + if (p->prm->prog_arg.type != RTE_BPF_ARG_PTR_MBUF) { PDUMP_LOG(ERR, - "failed to get port id for device id=%s\n", - p->data.en_v1.device); + "invalid BPF program type: %u\n", + p->prm->prog_arg.type); return -EINVAL; } - queue = p->data.en_v1.queue; - ring = p->data.en_v1.ring; - mp = p->data.en_v1.mp; - } else { - ret = rte_eth_dev_get_port_by_name(p->data.dis_v1.device, - &port); - if (ret < 0) { - PDUMP_LOG(ERR, - "failed to get port id for device id=%s\n", - p->data.dis_v1.device); - return -EINVAL; + + filter = rte_bpf_load(p->prm); + if (filter == NULL) { + PDUMP_LOG(ERR, "cannot load BPF filter: %s\n", + rte_strerror(rte_errno)); + return -rte_errno; } - queue = p->data.dis_v1.queue; - ring = p->data.dis_v1.ring; - mp = p->data.dis_v1.mp; + } + + flags = p->flags; + operation = p->op; + queue = p->queue; + ring = p->ring; + mp = p->mp; + + ret = rte_eth_dev_get_port_by_name(p->device, &port); + if (ret < 0) { + PDUMP_LOG(ERR, + "failed to get port id for device id=%s\n", + p->device); + return -EINVAL; } /* validation if packet capture is for all queues */ @@ -296,8 +365,9 @@ set_pdump_rxtx_cbs(const struct pdump_request *p) /* register RX callback */ if (flags & RTE_PDUMP_FLAG_RX) { end_q = (queue == RTE_PDUMP_ALL_QUEUES) ? nb_rx_q : queue + 1; - ret = pdump_register_rx_callbacks(end_q, port, queue, ring, mp, - operation); + ret = pdump_register_rx_callbacks(p->ver, end_q, port, queue, + ring, mp, filter, + operation, p->snaplen); if (ret < 0) return ret; } @@ -305,8 +375,9 @@ set_pdump_rxtx_cbs(const struct pdump_request *p) /* register TX callback */ if (flags & RTE_PDUMP_FLAG_TX) { end_q = (queue == RTE_PDUMP_ALL_QUEUES) ? nb_tx_q : queue + 1; - ret = pdump_register_tx_callbacks(end_q, port, queue, ring, mp, - operation); + ret = pdump_register_tx_callbacks(p->ver, end_q, port, queue, + ring, mp, filter, + operation, p->snaplen); if (ret < 0) return ret; } @@ -332,7 +403,7 @@ pdump_server(const struct rte_mp_msg *mp_msg, const void *peer) resp->err_value = set_pdump_rxtx_cbs(cli_req); } - strlcpy(mp_resp.name, PDUMP_MP, RTE_MP_MAX_NAME_LEN); + rte_strscpy(mp_resp.name, PDUMP_MP, RTE_MP_MAX_NAME_LEN); mp_resp.len_param = sizeof(*resp); mp_resp.num_fds = 0; if (rte_mp_reply(&mp_resp, peer) < 0) { @@ -347,8 +418,18 @@ pdump_server(const struct rte_mp_msg *mp_msg, const void *peer) int rte_pdump_init(void) { + const struct rte_memzone *mz; int ret; + mz = rte_memzone_reserve(MZ_RTE_PDUMP_STATS, sizeof(*pdump_stats), + rte_socket_id(), 0); + if (mz == NULL) { + PDUMP_LOG(ERR, "cannot allocate pdump statistics\n"); + rte_errno = ENOMEM; + return -1; + } + pdump_stats = mz->addr; + ret = rte_mp_action_register(PDUMP_MP, pdump_server); if (ret && rte_errno != ENOTSUP) return -1; @@ -392,14 +473,21 @@ pdump_validate_ring_mp(struct rte_ring *ring, struct rte_mempool *mp) static int pdump_validate_flags(uint32_t flags) { - if (flags != RTE_PDUMP_FLAG_RX && flags != RTE_PDUMP_FLAG_TX && - flags != RTE_PDUMP_FLAG_RXTX) { + if ((flags & RTE_PDUMP_FLAG_RXTX) == 0) { PDUMP_LOG(ERR, "invalid flags, should be either rx/tx/rxtx\n"); rte_errno = EINVAL; return -1; } + /* mask off the flags we know about */ + if (flags & ~(RTE_PDUMP_FLAG_RXTX | RTE_PDUMP_FLAG_PCAPNG)) { + PDUMP_LOG(ERR, + "unknown flags: %#x\n", flags); + rte_errno = ENOTSUP; + return -1; + } + return 0; } @@ -426,12 +514,12 @@ pdump_validate_port(uint16_t port, char *name) } static int -pdump_prepare_client_request(char *device, uint16_t queue, - uint32_t flags, - uint16_t operation, - struct rte_ring *ring, - struct rte_mempool *mp, - void *filter) +pdump_prepare_client_request(const char *device, uint16_t queue, + uint32_t flags, uint32_t snaplen, + uint16_t operation, + struct rte_ring *ring, + struct rte_mempool *mp, + const struct rte_bpf_prm *prm) { int ret = -1; struct rte_mp_msg mp_req, *mp_rep; @@ -440,26 +528,22 @@ pdump_prepare_client_request(char *device, uint16_t queue, struct pdump_request *req = (struct pdump_request *)mp_req.param; struct pdump_response *resp; - req->ver = 1; - req->flags = flags; + memset(req, 0, sizeof(*req)); + + req->ver = (flags & RTE_PDUMP_FLAG_PCAPNG) ? V2 : V1; + req->flags = flags & RTE_PDUMP_FLAG_RXTX; req->op = operation; + req->queue = queue; + rte_strscpy(req->device, device, sizeof(req->device)); + if ((operation & ENABLE) != 0) { - strlcpy(req->data.en_v1.device, device, - sizeof(req->data.en_v1.device)); - req->data.en_v1.queue = queue; - req->data.en_v1.ring = ring; - req->data.en_v1.mp = mp; - req->data.en_v1.filter = filter; - } else { - strlcpy(req->data.dis_v1.device, device, - sizeof(req->data.dis_v1.device)); - req->data.dis_v1.queue = queue; - req->data.dis_v1.ring = NULL; - req->data.dis_v1.mp = NULL; - req->data.dis_v1.filter = NULL; + req->ring = ring; + req->mp = mp; + req->prm = prm; + req->snaplen = snaplen; } - strlcpy(mp_req.name, PDUMP_MP, RTE_MP_MAX_NAME_LEN); + rte_strscpy(mp_req.name, PDUMP_MP, RTE_MP_MAX_NAME_LEN); mp_req.len_param = sizeof(*req); mp_req.num_fds = 0; if (rte_mp_request_sync(&mp_req, &mp_reply, &ts) == 0) { @@ -477,11 +561,17 @@ pdump_prepare_client_request(char *device, uint16_t queue, return ret; } -int -rte_pdump_enable(uint16_t port, uint16_t queue, uint32_t flags, - struct rte_ring *ring, - struct rte_mempool *mp, - void *filter) +/* + * There are two versions of this function, because although original API + * left place holder for future filter, it never checked the value. + * Therefore the API can't depend on application passing a non + * bogus value. + */ +static int +pdump_enable(uint16_t port, uint16_t queue, + uint32_t flags, uint32_t snaplen, + struct rte_ring *ring, struct rte_mempool *mp, + const struct rte_bpf_prm *prm) { int ret; char name[RTE_DEV_NAME_MAX_LEN]; @@ -496,20 +586,42 @@ rte_pdump_enable(uint16_t port, uint16_t queue, uint32_t flags, if (ret < 0) return ret; - ret = pdump_prepare_client_request(name, queue, flags, - ENABLE, ring, mp, filter); + if (snaplen == 0) + snaplen = UINT32_MAX; - return ret; + return pdump_prepare_client_request(name, queue, flags, snaplen, + ENABLE, ring, mp, prm); } int -rte_pdump_enable_by_deviceid(char *device_id, uint16_t queue, - uint32_t flags, - struct rte_ring *ring, - struct rte_mempool *mp, - void *filter) +rte_pdump_enable(uint16_t port, uint16_t queue, uint32_t flags, + struct rte_ring *ring, + struct rte_mempool *mp, + void *filter __rte_unused) { - int ret = 0; + return pdump_enable(port, queue, flags, 0, + ring, mp, NULL); +} + +int +rte_pdump_enable_bpf(uint16_t port, uint16_t queue, + uint32_t flags, uint32_t snaplen, + struct rte_ring *ring, + struct rte_mempool *mp, + const struct rte_bpf_prm *prm) +{ + return pdump_enable(port, queue, flags, snaplen, + ring, mp, prm); +} + +static int +pdump_enable_by_deviceid(const char *device_id, uint16_t queue, + uint32_t flags, uint32_t snaplen, + struct rte_ring *ring, + struct rte_mempool *mp, + const struct rte_bpf_prm *prm) +{ + int ret; ret = pdump_validate_ring_mp(ring, mp); if (ret < 0) @@ -518,10 +630,30 @@ rte_pdump_enable_by_deviceid(char *device_id, uint16_t queue, if (ret < 0) return ret; - ret = pdump_prepare_client_request(device_id, queue, flags, - ENABLE, ring, mp, filter); + return pdump_prepare_client_request(device_id, queue, flags, snaplen, + ENABLE, ring, mp, prm); +} - return ret; +int +rte_pdump_enable_by_deviceid(char *device_id, uint16_t queue, + uint32_t flags, + struct rte_ring *ring, + struct rte_mempool *mp, + void *filter __rte_unused) +{ + return pdump_enable_by_deviceid(device_id, queue, flags, 0, + ring, mp, NULL); +} + +int +rte_pdump_enable_bpf_by_deviceid(const char *device_id, uint16_t queue, + uint32_t flags, uint32_t snaplen, + struct rte_ring *ring, + struct rte_mempool *mp, + const struct rte_bpf_prm *prm) +{ + return pdump_enable_by_deviceid(device_id, queue, flags, snaplen, + ring, mp, prm); } int @@ -537,8 +669,8 @@ rte_pdump_disable(uint16_t port, uint16_t queue, uint32_t flags) if (ret < 0) return ret; - ret = pdump_prepare_client_request(name, queue, flags, - DISABLE, NULL, NULL, NULL); + ret = pdump_prepare_client_request(name, queue, flags, 0, + DISABLE, NULL, NULL, NULL); return ret; } @@ -553,8 +685,68 @@ rte_pdump_disable_by_deviceid(char *device_id, uint16_t queue, if (ret < 0) return ret; - ret = pdump_prepare_client_request(device_id, queue, flags, - DISABLE, NULL, NULL, NULL); + ret = pdump_prepare_client_request(device_id, queue, flags, 0, + DISABLE, NULL, NULL, NULL); return ret; } + +static void +pdump_sum_stats(uint16_t port, uint16_t nq, + struct rte_pdump_stats stats[RTE_MAX_ETHPORTS][RTE_MAX_QUEUES_PER_PORT], + struct rte_pdump_stats *total) +{ + uint64_t *sum = (uint64_t *)total; + unsigned int i; + uint64_t val; + uint16_t qid; + + for (qid = 0; qid < nq; qid++) { + const uint64_t *perq = (const uint64_t *)&stats[port][qid]; + + for (i = 0; i < sizeof(*total) / sizeof(uint64_t); i++) { + val = __atomic_load_n(&perq[i], __ATOMIC_RELAXED); + sum[i] += val; + } + } +} + +int +rte_pdump_stats(uint16_t port, struct rte_pdump_stats *stats) +{ + struct rte_eth_dev_info dev_info; + const struct rte_memzone *mz; + int ret; + + memset(stats, 0, sizeof(*stats)); + ret = rte_eth_dev_info_get(port, &dev_info); + if (ret != 0) { + PDUMP_LOG(ERR, + "Error during getting device (port %u) info: %s\n", + port, strerror(-ret)); + return ret; + } + + if (pdump_stats == NULL) { + if (rte_eal_process_type() == RTE_PROC_PRIMARY) { + /* rte_pdump_init was not called */ + PDUMP_LOG(ERR, "pdump stats not initialized\n"); + rte_errno = EINVAL; + return -1; + } + + /* secondary process looks up the memzone */ + mz = rte_memzone_lookup(MZ_RTE_PDUMP_STATS); + if (mz == NULL) { + /* rte_pdump_init was not called in primary process?? */ + PDUMP_LOG(ERR, "can not find pdump stats\n"); + rte_errno = EINVAL; + return -1; + } + pdump_stats = mz->addr; + } + + pdump_sum_stats(port, dev_info.nb_rx_queues, pdump_stats->rx, stats); + pdump_sum_stats(port, dev_info.nb_tx_queues, pdump_stats->tx, stats); + return 0; +} diff --git a/lib/pdump/rte_pdump.h b/lib/pdump/rte_pdump.h index 6b00fc17aeb2..6efa0274f2ce 100644 --- a/lib/pdump/rte_pdump.h +++ b/lib/pdump/rte_pdump.h @@ -15,6 +15,7 @@ #include #include #include +#include #ifdef __cplusplus extern "C" { @@ -26,7 +27,9 @@ enum { RTE_PDUMP_FLAG_RX = 1, /* receive direction */ RTE_PDUMP_FLAG_TX = 2, /* transmit direction */ /* both receive and transmit directions */ - RTE_PDUMP_FLAG_RXTX = (RTE_PDUMP_FLAG_RX|RTE_PDUMP_FLAG_TX) + RTE_PDUMP_FLAG_RXTX = (RTE_PDUMP_FLAG_RX|RTE_PDUMP_FLAG_TX), + + RTE_PDUMP_FLAG_PCAPNG = 4, /* format for pcapng */ }; /** @@ -68,7 +71,7 @@ rte_pdump_uninit(void); * @param mp * mempool on to which original packets will be mirrored or duplicated. * @param filter - * place holder for packet filtering. + * Unused should be NULL. * * @return * 0 on success, -1 on error, rte_errno is set accordingly. @@ -80,6 +83,41 @@ rte_pdump_enable(uint16_t port, uint16_t queue, uint32_t flags, struct rte_mempool *mp, void *filter); +/** + * @warning + * @b EXPERIMENTAL: this API may change, or be removed, without prior notice + * + * Enables packet capturing on given port and queue with filtering. + * + * @param port_id + * The Ethernet port on which packet capturing should be enabled. + * @param queue + * The queue on the Ethernet port which packet capturing + * should be enabled. Pass UINT16_MAX to enable packet capturing on all + * queues of a given port. + * @param flags + * Pdump library flags that specify direction and packet format. + * @param snaplen + * The upper limit on bytes to copy. + * Passing UINT32_MAX means capture all the possible data. + * @param ring + * The ring on which captured packets will be enqueued for user. + * @param mp + * The mempool on to which original packets will be mirrored or duplicated. + * @param prm + * Use BPF program to run to filter packes (can be NULL) + * + * @return + * 0 on success, -1 on error, rte_errno is set accordingly. + */ +__rte_experimental +int +rte_pdump_enable_bpf(uint16_t port_id, uint16_t queue, + uint32_t flags, uint32_t snaplen, + struct rte_ring *ring, + struct rte_mempool *mp, + const struct rte_bpf_prm *prm); + /** * Disables packet capturing on given port and queue. * @@ -118,7 +156,7 @@ rte_pdump_disable(uint16_t port, uint16_t queue, uint32_t flags); * @param mp * mempool on to which original packets will be mirrored or duplicated. * @param filter - * place holder for packet filtering. + * unused should be NULL * * @return * 0 on success, -1 on error, rte_errno is set accordingly. @@ -131,6 +169,43 @@ rte_pdump_enable_by_deviceid(char *device_id, uint16_t queue, struct rte_mempool *mp, void *filter); +/** + * @warning + * @b EXPERIMENTAL: this API may change, or be removed, without prior notice + * + * Enables packet capturing on given device id and queue with filtering. + * device_id can be name or pci address of device. + * + * @param device_id + * device id on which packet capturing should be enabled. + * @param queue + * The queue on the Ethernet port which packet capturing + * should be enabled. Pass UINT16_MAX to enable packet capturing on all + * queues of a given port. + * @param flags + * Pdump library flags that specify direction and packet format. + * @param snaplen + * The upper limit on bytes to copy. + * Passing UINT32_MAX means capture all the possible data. + * @param ring + * The ring on which captured packets will be enqueued for user. + * @param mp + * The mempool on to which original packets will be mirrored or duplicated. + * @param filter + * Use BPF program to run to filter packes (can be NULL) + * + * @return + * 0 on success, -1 on error, rte_errno is set accordingly. + */ +__rte_experimental +int +rte_pdump_enable_bpf_by_deviceid(const char *device_id, uint16_t queue, + uint32_t flags, uint32_t snaplen, + struct rte_ring *ring, + struct rte_mempool *mp, + const struct rte_bpf_prm *filter); + + /** * Disables packet capturing on given device_id and queue. * device_id can be name or pci address of device. @@ -153,6 +228,38 @@ int rte_pdump_disable_by_deviceid(char *device_id, uint16_t queue, uint32_t flags); + +/** + * A structure used to retrieve statistics from packet capture. + * The statistics are sum of both receive and transmit queues. + */ +struct rte_pdump_stats { + uint64_t accepted; /**< Number of packets accepted by filter. */ + uint64_t filtered; /**< Number of packets rejected by filter. */ + uint64_t nombuf; /**< Number of mbuf allocation failures. */ + uint64_t ringfull; /**< Number of missed packets due to ring full. */ + + uint64_t reserved[4]; /**< Reserved and pad to cache line */ +}; + +/** + * @warning + * @b EXPERIMENTAL: this API may change, or be removed, without prior notice + * + * Retrieve the packet capture statistics for a queue. + * + * @param port_id + * The port identifier of the Ethernet device. + * @param stats + * A pointer to structure of type *rte_pdump_stats* to be filled in. + * @return + * Zero if successful. -1 on error and rte_errno is set. + */ +__rte_experimental +int +rte_pdump_stats(uint16_t port_id, struct rte_pdump_stats *stats); + + #ifdef __cplusplus } #endif diff --git a/lib/pdump/version.map b/lib/pdump/version.map index f0a9d12c9a9e..ce5502d9cdf4 100644 --- a/lib/pdump/version.map +++ b/lib/pdump/version.map @@ -10,3 +10,11 @@ DPDK_22 { local: *; }; + +EXPERIMENTAL { + global: + + rte_pdump_enable_bpf; + rte_pdump_enable_bpf_by_deviceid; + rte_pdump_stats; +}; -- 2.30.2