From patchwork Sat Jan 1 00:12:44 2022 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: "Ding, Xuan" X-Patchwork-Id: 105538 X-Patchwork-Delegate: maxime.coquelin@redhat.com Return-Path: X-Original-To: patchwork@inbox.dpdk.org Delivered-To: patchwork@inbox.dpdk.org Received: from mails.dpdk.org (mails.dpdk.org [217.70.189.124]) by inbox.dpdk.org (Postfix) with ESMTP id 55A97A00C2; Fri, 31 Dec 2021 16:36:47 +0100 (CET) Received: from [217.70.189.124] (localhost [127.0.0.1]) by mails.dpdk.org (Postfix) with ESMTP id 5E29E4113D; Fri, 31 Dec 2021 16:36:44 +0100 (CET) Received: from mga02.intel.com (mga02.intel.com [134.134.136.20]) by mails.dpdk.org (Postfix) with ESMTP id BF7074067E for ; Fri, 31 Dec 2021 16:36:41 +0100 (CET) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1640965001; x=1672501001; h=from:to:cc:subject:date:message-id:in-reply-to: references; bh=iW3cxIJfKPSz9hWHpblRTBkAEnMvE4qYf89CyNvbOOY=; b=TvbL+MIXTnF4Df76w7Y1rpKUgjUU5Ob9xb3DTildQAyVNFMao7fbckQk Fx544iaA+yNOLqPfmrVf1akm5Q43cCGlv6ek7Coxg/ZV5JHya9fOmopo+ HPwYMsnWhK3wrhe0qFI2lOeguzgiv5Q6bW/GqXwP90YN6AVgbq/m0QJwo ZhXwpc/SZ8XdKlDnRZfa73UBlz3oSGJfUvM20dPa1p2Vd0a/M39wkfYal Gt/9/1+mNw/riQSAn+zXYNXq9wKXTEPBny+KUuEGIsDVlu78RYDPWH05F I1APGZJOUz41o9GcpWb/cOENzsnvWUOj9OEIwvCIYGPGowsdq/6KvP/U9 w==; X-IronPort-AV: E=McAfee;i="6200,9189,10213"; a="229136260" X-IronPort-AV: E=Sophos;i="5.88,251,1635231600"; d="scan'208";a="229136260" Received: from orsmga008.jf.intel.com ([10.7.209.65]) by orsmga101.jf.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 31 Dec 2021 07:36:41 -0800 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.88,251,1635231600"; d="scan'208";a="524765519" Received: from npg-dpdk-xuan-cbdma.sh.intel.com ([10.67.110.228]) by orsmga008.jf.intel.com with ESMTP; 31 Dec 2021 07:36:38 -0800 From: xuan.ding@intel.com To: maxime.coquelin@redhat.com, chenbo.xia@intel.com Cc: dev@dpdk.org, jiayu.hu@intel.com, sunil.pai.g@intel.com, liangma@liangbit.com, yuanx.wang@intel.com, cheng1.jiang@intel.com, wenwux.ma@intel.com, Xuan Ding , Yuan Wang Subject: [RFC 2/2] examples/vhost: support async dequeue data path Date: Sat, 1 Jan 2022 00:12:44 +0000 Message-Id: <20220101001244.90147-3-xuan.ding@intel.com> X-Mailer: git-send-email 2.17.1 In-Reply-To: <20220101001244.90147-1-xuan.ding@intel.com> References: <20220101001244.90147-1-xuan.ding@intel.com> 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 From: Xuan Ding This patch adds the use case for async dequeue API. Vswitch can leverage DMA device to accelerate vhost async dequeue path. Signed-off-by: Wenwu Ma Signed-off-by: Yuan Wang Signed-off-by: Xuan Ding --- doc/guides/sample_app_ug/vhost.rst | 9 +- examples/vhost/main.c | 301 +++++++++++++++++++---------- examples/vhost/main.h | 35 +++- examples/vhost/virtio_net.c | 16 +- 4 files changed, 255 insertions(+), 106 deletions(-) diff --git a/doc/guides/sample_app_ug/vhost.rst b/doc/guides/sample_app_ug/vhost.rst index a6ce4bc8ac..09db965e70 100644 --- a/doc/guides/sample_app_ug/vhost.rst +++ b/doc/guides/sample_app_ug/vhost.rst @@ -169,9 +169,12 @@ demonstrates how to use the async vhost APIs. It's used in combination with dmas **--dmas** This parameter is used to specify the assigned DMA device of a vhost device. Async vhost-user net driver will be used if --dmas is set. For example ---dmas [txd0@00:04.0,txd1@00:04.1] means use DMA channel 00:04.0 for vhost -device 0 enqueue operation and use DMA channel 00:04.1 for vhost device 1 -enqueue operation. +--dmas [txd0@00:04.0,txd1@00:04.1,rxd0@00:04.2,rxd1@00:04.3] means use +DMA channel 00:04.0/00:04.2 for vhost device 0 enqueue/dequeue operation +and use DMA channel 00:04.1/00:04.3 for vhost device 1 enqueue/dequeue +operation. The index of the device corresponds to the socket file in order, +that means vhost device 0 is created through the first socket file, vhost +device 1 is created through the second socket file, and so on. Common Issues ------------- diff --git a/examples/vhost/main.c b/examples/vhost/main.c index 44073499bc..8c952d3874 100644 --- a/examples/vhost/main.c +++ b/examples/vhost/main.c @@ -62,6 +62,9 @@ #define MAX_VHOST_DEVICE 1024 #define DMA_RING_SIZE 4096 +#define ASYNC_ENQUEUE_VHOST 1 +#define ASYNC_DEQUEUE_VHOST 2 + struct dma_for_vhost dma_bind[MAX_VHOST_DEVICE]; struct rte_vhost_async_dma_info dma_config[RTE_DMADEV_DEFAULT_MAX]; static int dma_count; @@ -103,8 +106,6 @@ static int client_mode; static int builtin_net_driver; -static int async_vhost_driver; - /* Specify timeout (in useconds) between retries on RX. */ static uint32_t burst_rx_delay_time = BURST_RX_WAIT_US; /* Specify the number of retries on RX. */ @@ -114,6 +115,8 @@ static uint32_t burst_rx_retry_num = BURST_RX_RETRIES; static char *socket_files; static int nb_sockets; +static struct vhost_queue_ops vdev_queue_ops[MAX_VHOST_DEVICE]; + /* empty vmdq configuration structure. Filled in programatically */ static struct rte_eth_conf vmdq_conf_default = { .rxmode = { @@ -203,6 +206,18 @@ struct vhost_bufftable *vhost_txbuff[RTE_MAX_LCORE * MAX_VHOST_DEVICE]; #define MBUF_TABLE_DRAIN_TSC ((rte_get_tsc_hz() + US_PER_S - 1) \ / US_PER_S * BURST_TX_DRAIN_US) +static int vid2socketid[MAX_VHOST_DEVICE]; + +static uint32_t get_async_flag_by_socketid(int socketid) +{ + return dma_bind[socketid].async_flag; +} + +static void init_vid2socketid_array(int vid, int socketid) +{ + vid2socketid[vid] = socketid; +} + static inline bool is_dma_configured(int16_t dev_id) { @@ -224,7 +239,7 @@ open_dma(const char *value) char *addrs = input; char *ptrs[2]; char *start, *end, *substr; - int64_t vid, vring_id; + int64_t socketid, vring_id; struct rte_dma_info info; struct rte_dma_conf dev_config = { .nb_vchans = 1 }; @@ -263,7 +278,9 @@ open_dma(const char *value) while (i < args_nr) { char *arg_temp = dma_arg[i]; + char *txd, *rxd; uint8_t sub_nr; + int async_flag; sub_nr = rte_strsplit(arg_temp, strlen(arg_temp), ptrs, 2, '@'); if (sub_nr != 2) { @@ -271,21 +288,28 @@ open_dma(const char *value) goto out; } - start = strstr(ptrs[0], "txd"); - if (start == NULL) { + txd = strstr(ptrs[0], "txd"); + rxd = strstr(ptrs[0], "rxd"); + if (txd) { + start = txd; + vring_id = VIRTIO_RXQ; + async_flag = ASYNC_ENQUEUE_VHOST; + } else if (rxd) { + start = rxd; + vring_id = VIRTIO_TXQ; + async_flag = ASYNC_DEQUEUE_VHOST; + } else { ret = -1; goto out; } start += 3; - vid = strtol(start, &end, 0); + socketid = strtol(start, &end, 0); if (end == start) { ret = -1; goto out; } - vring_id = 0 + VIRTIO_RXQ; - dev_id = rte_dma_get_dev_id_by_name(ptrs[1]); if (dev_id < 0) { RTE_LOG(ERR, VHOST_CONFIG, "Fail to find DMA %s.\n", ptrs[1]); @@ -325,7 +349,8 @@ open_dma(const char *value) dma_config[dma_count++].max_desc = DMA_RING_SIZE; done: - (dma_info + vid)->dmas[vring_id].dev_id = dev_id; + (dma_info + socketid)->dmas[vring_id].dev_id = dev_id; + (dma_info + socketid)->async_flag |= async_flag; i++; } out: @@ -792,7 +817,6 @@ us_vhost_parse_args(int argc, char **argv) us_vhost_usage(prgname); return -1; } - async_vhost_driver = 1; break; case OPT_CLIENT_NUM: @@ -961,13 +985,13 @@ complete_async_pkts(struct vhost_dev *vdev) { struct rte_mbuf *p_cpl[MAX_PKT_BURST]; uint16_t complete_count; - int16_t dma_id = dma_bind[vdev->vid].dmas[VIRTIO_RXQ].dev_id; + int16_t dma_id = dma_bind[vid2socketid[vdev->vid]].dmas[VIRTIO_RXQ].dev_id; complete_count = rte_vhost_poll_enqueue_completed(vdev->vid, VIRTIO_RXQ, p_cpl, MAX_PKT_BURST, dma_id, 0); if (complete_count) { free_pkts(p_cpl, complete_count); - __atomic_sub_fetch(&vdev->pkts_inflight, complete_count, __ATOMIC_SEQ_CST); + __atomic_sub_fetch(&vdev->pkts_enq_inflight, complete_count, __ATOMIC_SEQ_CST); } } @@ -1002,23 +1026,7 @@ drain_vhost(struct vhost_dev *vdev) uint16_t nr_xmit = vhost_txbuff[buff_idx]->len; struct rte_mbuf **m = vhost_txbuff[buff_idx]->m_table; - if (builtin_net_driver) { - ret = vs_enqueue_pkts(vdev, VIRTIO_RXQ, m, nr_xmit); - } else if (dma_bind[vdev->vid].dmas[VIRTIO_RXQ].async_enabled) { - uint16_t enqueue_fail = 0; - int16_t dma_id = dma_bind[vdev->vid].dmas[VIRTIO_RXQ].dev_id; - - complete_async_pkts(vdev); - ret = rte_vhost_submit_enqueue_burst(vdev->vid, VIRTIO_RXQ, m, nr_xmit, dma_id, 0); - __atomic_add_fetch(&vdev->pkts_inflight, ret, __ATOMIC_SEQ_CST); - - enqueue_fail = nr_xmit - ret; - if (enqueue_fail) - free_pkts(&m[ret], nr_xmit - ret); - } else { - ret = rte_vhost_enqueue_burst(vdev->vid, VIRTIO_RXQ, - m, nr_xmit); - } + ret = vdev_queue_ops[vdev->vid].enqueue_pkt_burst(vdev, VIRTIO_RXQ, m, nr_xmit); if (enable_stats) { __atomic_add_fetch(&vdev->stats.rx_total_atomic, nr_xmit, @@ -1027,7 +1035,7 @@ drain_vhost(struct vhost_dev *vdev) __ATOMIC_SEQ_CST); } - if (!dma_bind[vdev->vid].dmas[VIRTIO_RXQ].async_enabled) + if (!dma_bind[vid2socketid[vdev->vid]].dmas[VIRTIO_RXQ].async_enabled) free_pkts(m, nr_xmit); } @@ -1300,6 +1308,33 @@ drain_mbuf_table(struct mbuf_table *tx_q) } } +uint16_t +async_enqueue_pkts(struct vhost_dev *dev, uint16_t queue_id, + struct rte_mbuf **pkts, uint32_t rx_count) +{ + uint16_t enqueue_count; + uint16_t enqueue_fail = 0; + uint16_t dma_id = dma_bind[vid2socketid[dev->vid]].dmas[VIRTIO_RXQ].dev_id; + + complete_async_pkts(dev); + enqueue_count = rte_vhost_submit_enqueue_burst(dev->vid, queue_id, + pkts, rx_count, dma_id, 0); + __atomic_add_fetch(&dev->pkts_enq_inflight, enqueue_count, __ATOMIC_SEQ_CST); + + enqueue_fail = rx_count - enqueue_count; + if (enqueue_fail) + free_pkts(&pkts[enqueue_count], enqueue_fail); + + return enqueue_count; +} + +uint16_t +sync_enqueue_pkts(struct vhost_dev *dev, uint16_t queue_id, + struct rte_mbuf **pkts, uint32_t rx_count) +{ + return rte_vhost_enqueue_burst(dev->vid, queue_id, pkts, rx_count); +} + static __rte_always_inline void drain_eth_rx(struct vhost_dev *vdev) { @@ -1330,26 +1365,8 @@ drain_eth_rx(struct vhost_dev *vdev) } } - if (builtin_net_driver) { - enqueue_count = vs_enqueue_pkts(vdev, VIRTIO_RXQ, - pkts, rx_count); - } else if (dma_bind[vdev->vid].dmas[VIRTIO_RXQ].async_enabled) { - uint16_t enqueue_fail = 0; - int16_t dma_id = dma_bind[vdev->vid].dmas[VIRTIO_RXQ].dev_id; - - complete_async_pkts(vdev); - enqueue_count = rte_vhost_submit_enqueue_burst(vdev->vid, - VIRTIO_RXQ, pkts, rx_count, dma_id, 0); - __atomic_add_fetch(&vdev->pkts_inflight, enqueue_count, __ATOMIC_SEQ_CST); - - enqueue_fail = rx_count - enqueue_count; - if (enqueue_fail) - free_pkts(&pkts[enqueue_count], enqueue_fail); - - } else { - enqueue_count = rte_vhost_enqueue_burst(vdev->vid, VIRTIO_RXQ, - pkts, rx_count); - } + enqueue_count = vdev_queue_ops[vdev->vid].enqueue_pkt_burst(vdev, + VIRTIO_RXQ, pkts, rx_count); if (enable_stats) { __atomic_add_fetch(&vdev->stats.rx_total_atomic, rx_count, @@ -1358,10 +1375,33 @@ drain_eth_rx(struct vhost_dev *vdev) __ATOMIC_SEQ_CST); } - if (!dma_bind[vdev->vid].dmas[VIRTIO_RXQ].async_enabled) + if (!dma_bind[vid2socketid[vdev->vid]].dmas[VIRTIO_RXQ].async_enabled) free_pkts(pkts, rx_count); } +uint16_t async_dequeue_pkts(struct vhost_dev *dev, uint16_t queue_id, + struct rte_mempool *mbuf_pool, + struct rte_mbuf **pkts, uint16_t count) +{ + int nr_inflight; + uint16_t dequeue_count; + uint16_t dma_id = dma_bind[vid2socketid[dev->vid]].dmas[VIRTIO_TXQ].dev_id; + + dequeue_count = rte_vhost_async_try_dequeue_burst(dev->vid, queue_id, + mbuf_pool, pkts, count, &nr_inflight, dma_id, 0); + if (likely(nr_inflight != -1)) + dev->pkts_deq_inflight = nr_inflight; + + return dequeue_count; +} + +uint16_t sync_dequeue_pkts(struct vhost_dev *dev, uint16_t queue_id, + struct rte_mempool *mbuf_pool, + struct rte_mbuf **pkts, uint16_t count) +{ + return rte_vhost_dequeue_burst(dev->vid, queue_id, mbuf_pool, pkts, count); +} + static __rte_always_inline void drain_virtio_tx(struct vhost_dev *vdev) { @@ -1369,13 +1409,8 @@ drain_virtio_tx(struct vhost_dev *vdev) uint16_t count; uint16_t i; - if (builtin_net_driver) { - count = vs_dequeue_pkts(vdev, VIRTIO_TXQ, mbuf_pool, - pkts, MAX_PKT_BURST); - } else { - count = rte_vhost_dequeue_burst(vdev->vid, VIRTIO_TXQ, - mbuf_pool, pkts, MAX_PKT_BURST); - } + count = vdev_queue_ops[vdev->vid].dequeue_pkt_burst(vdev, + VIRTIO_TXQ, mbuf_pool, pkts, MAX_PKT_BURST); /* setup VMDq for the first packet */ if (unlikely(vdev->ready == DEVICE_MAC_LEARNING) && count) { @@ -1454,6 +1489,31 @@ switch_worker(void *arg __rte_unused) return 0; } +static void +vhost_clear_queue_thread_unsafe(struct vhost_dev *vdev, uint16_t queue_id) +{ + uint16_t n_pkt = 0; + uint16_t dma_id = dma_bind[vid2socketid[vdev->vid]].dmas[queue_id].dev_id; + struct rte_mbuf *m_enq_cpl[vdev->pkts_enq_inflight]; + struct rte_mbuf *m_deq_cpl[vdev->pkts_deq_inflight]; + + if (queue_id % 2 == 0) { + while (vdev->pkts_enq_inflight) { + n_pkt = rte_vhost_clear_queue_thread_unsafe(vdev->vid, + queue_id, m_enq_cpl, vdev->pkts_enq_inflight, dma_id, 0); + free_pkts(m_enq_cpl, n_pkt); + __atomic_sub_fetch(&vdev->pkts_enq_inflight, n_pkt, __ATOMIC_SEQ_CST); + } + } else { + while (vdev->pkts_deq_inflight) { + n_pkt = rte_vhost_clear_queue_thread_unsafe(vdev->vid, + queue_id, m_deq_cpl, vdev->pkts_deq_inflight, dma_id, 0); + free_pkts(m_deq_cpl, n_pkt); + __atomic_sub_fetch(&vdev->pkts_deq_inflight, n_pkt, __ATOMIC_SEQ_CST); + } + } +} + /* * Remove a device from the specific data core linked list and from the * main linked list. Synchonization occurs through the use of the @@ -1510,25 +1570,83 @@ destroy_device(int vid) "(%d) device has been removed from data core\n", vdev->vid); - if (dma_bind[vid].dmas[VIRTIO_RXQ].async_enabled) { - uint16_t n_pkt = 0; - int16_t dma_id = dma_bind[vid].dmas[VIRTIO_RXQ].dev_id; - struct rte_mbuf *m_cpl[vdev->pkts_inflight]; - - while (vdev->pkts_inflight) { - n_pkt = rte_vhost_clear_queue_thread_unsafe(vid, VIRTIO_RXQ, - m_cpl, vdev->pkts_inflight, dma_id, 0); - free_pkts(m_cpl, n_pkt); - __atomic_sub_fetch(&vdev->pkts_inflight, n_pkt, __ATOMIC_SEQ_CST); - } - + if (dma_bind[vid2socketid[vid]].dmas[VIRTIO_RXQ].async_enabled) { + vhost_clear_queue_thread_unsafe(vdev, VIRTIO_RXQ); rte_vhost_async_channel_unregister(vid, VIRTIO_RXQ); - dma_bind[vid].dmas[VIRTIO_RXQ].async_enabled = false; + dma_bind[vid2socketid[vid]].dmas[VIRTIO_RXQ].async_enabled = false; + } + if (dma_bind[vid2socketid[vid]].dmas[VIRTIO_TXQ].async_enabled) { + vhost_clear_queue_thread_unsafe(vdev, VIRTIO_TXQ); + rte_vhost_async_channel_unregister(vid, VIRTIO_TXQ); + dma_bind[vid2socketid[vid]].dmas[VIRTIO_TXQ].async_enabled = false; } rte_free(vdev); } +static int +get_socketid_by_vid(int vid) +{ + int i; + char ifname[PATH_MAX]; + rte_vhost_get_ifname(vid, ifname, sizeof(ifname)); + + for (i = 0; i < nb_sockets; i++) { + char *file = socket_files + i * PATH_MAX; + if (strcmp(file, ifname) == 0) + return i; + } + + return -1; +} + +static int +init_vhost_queue_ops(int vid) +{ + int socketid = get_socketid_by_vid(vid); + if (socketid == -1) + return -1; + + init_vid2socketid_array(vid, socketid); + if (builtin_net_driver) { + vdev_queue_ops[vid].enqueue_pkt_burst = builtin_enqueue_pkts; + vdev_queue_ops[vid].dequeue_pkt_burst = builtin_dequeue_pkts; + } else { + if (dma_bind[vid2socketid[vid]].dmas[VIRTIO_RXQ].async_enabled) + vdev_queue_ops[vid].enqueue_pkt_burst = async_enqueue_pkts; + else + vdev_queue_ops[vid].enqueue_pkt_burst = sync_enqueue_pkts; + + if (dma_bind[vid2socketid[vid]].dmas[VIRTIO_TXQ].async_enabled) + vdev_queue_ops[vid].dequeue_pkt_burst = async_dequeue_pkts; + else + vdev_queue_ops[vid].dequeue_pkt_burst = sync_dequeue_pkts; + } + + return 0; +} + +static int +vhost_async_channel_register(int vid) +{ + int rx_ret = 0, tx_ret = 0; + + if (dma_bind[vid2socketid[vid]].dmas[VIRTIO_RXQ].dev_id != INVALID_DMA_ID) { + rx_ret = rte_vhost_async_channel_register(vid, VIRTIO_RXQ); + if (rx_ret == 0) + dma_bind[vid2socketid[vid]].dmas[VIRTIO_RXQ].async_enabled = true; + } + + if (dma_bind[vid2socketid[vid]].dmas[VIRTIO_TXQ].dev_id != INVALID_DMA_ID) { + tx_ret = rte_vhost_async_channel_register(vid, VIRTIO_TXQ); + if (tx_ret == 0) + dma_bind[vid2socketid[vid]].dmas[VIRTIO_TXQ].async_enabled = true; + } + + return rx_ret | tx_ret; +} + + /* * A new device is added to a data core. First the device is added to the main linked list * and then allocated to a specific data core. @@ -1540,6 +1658,8 @@ new_device(int vid) uint16_t i; uint32_t device_num_min = num_devices; struct vhost_dev *vdev; + int ret; + vdev = rte_zmalloc("vhost device", sizeof(*vdev), RTE_CACHE_LINE_SIZE); if (vdev == NULL) { RTE_LOG(INFO, VHOST_DATA, @@ -1593,17 +1713,12 @@ new_device(int vid) "(%d) device has been added to data core %d\n", vid, vdev->coreid); - if (dma_bind[vid].dmas[VIRTIO_RXQ].dev_id != INVALID_DMA_ID) { - int ret; + ret = vhost_async_channel_register(vid); - ret = rte_vhost_async_channel_register(vid, VIRTIO_RXQ); - if (ret == 0) { - dma_bind[vid].dmas[VIRTIO_RXQ].async_enabled = true; - } - return ret; - } + if (init_vhost_queue_ops(vid) != 0) + return -1; - return 0; + return ret; } static int @@ -1621,19 +1736,9 @@ vring_state_changed(int vid, uint16_t queue_id, int enable) if (queue_id != VIRTIO_RXQ) return 0; - if (dma_bind[vid].dmas[queue_id].async_enabled) { - if (!enable) { - uint16_t n_pkt = 0; - int16_t dma_id = dma_bind[vid].dmas[VIRTIO_RXQ].dev_id; - struct rte_mbuf *m_cpl[vdev->pkts_inflight]; - - while (vdev->pkts_inflight) { - n_pkt = rte_vhost_clear_queue_thread_unsafe(vid, queue_id, - m_cpl, vdev->pkts_inflight, dma_id, 0); - free_pkts(m_cpl, n_pkt); - __atomic_sub_fetch(&vdev->pkts_inflight, n_pkt, __ATOMIC_SEQ_CST); - } - } + if (dma_bind[vid2socketid[vid]].dmas[queue_id].async_enabled) { + if (!enable) + vhost_clear_queue_thread_unsafe(vdev, queue_id); } return 0; @@ -1896,7 +2001,7 @@ main(int argc, char *argv[]) if (client_mode) flags |= RTE_VHOST_USER_CLIENT; - if (async_vhost_driver) { + if (dma_count > 0) { if (rte_vhost_async_dma_configure(dma_config, dma_count) < 0) { RTE_LOG(ERR, VHOST_PORT, "Failed to configure DMA in vhost.\n"); for (i = 0; i < dma_count; i++) { @@ -1906,18 +2011,18 @@ main(int argc, char *argv[]) } } dma_count = 0; - async_vhost_driver = false; } } /* Register vhost user driver to handle vhost messages. */ for (i = 0; i < nb_sockets; i++) { char *file = socket_files + i * PATH_MAX; + uint64_t flag = flags; - if (async_vhost_driver) - flags = flags | RTE_VHOST_USER_ASYNC_COPY; + if (dma_count > 0 && get_async_flag_by_socketid(i) != 0) + flag |= RTE_VHOST_USER_ASYNC_COPY; - ret = rte_vhost_driver_register(file, flags); + ret = rte_vhost_driver_register(file, flag); if (ret != 0) { unregister_drivers(i); rte_exit(EXIT_FAILURE, diff --git a/examples/vhost/main.h b/examples/vhost/main.h index b4a453e77e..40ac2841d1 100644 --- a/examples/vhost/main.h +++ b/examples/vhost/main.h @@ -52,7 +52,8 @@ struct vhost_dev { uint64_t features; size_t hdr_len; uint16_t nr_vrings; - uint16_t pkts_inflight; + uint16_t pkts_enq_inflight; + uint16_t pkts_deq_inflight; struct rte_vhost_memory *mem; struct device_statistics stats; TAILQ_ENTRY(vhost_dev) global_vdev_entry; @@ -62,6 +63,19 @@ struct vhost_dev { struct vhost_queue queues[MAX_QUEUE_PAIRS * 2]; } __rte_cache_aligned; +typedef uint16_t (*vhost_enqueue_burst_t)(struct vhost_dev *dev, + uint16_t queue_id, struct rte_mbuf **pkts, + uint32_t count); + +typedef uint16_t (*vhost_dequeue_burst_t)(struct vhost_dev *dev, + uint16_t queue_id, struct rte_mempool *mbuf_pool, + struct rte_mbuf **pkts, uint16_t count); + +struct vhost_queue_ops { + vhost_enqueue_burst_t enqueue_pkt_burst; + vhost_dequeue_burst_t dequeue_pkt_burst; +}; + TAILQ_HEAD(vhost_dev_tailq_list, vhost_dev); @@ -88,6 +102,7 @@ struct dma_info { struct dma_for_vhost { struct dma_info dmas[RTE_MAX_QUEUES_PER_PORT * 2]; + uint32_t async_flag; }; /* we implement non-extra virtio net features */ @@ -98,7 +113,19 @@ void vs_vhost_net_remove(struct vhost_dev *dev); uint16_t vs_enqueue_pkts(struct vhost_dev *dev, uint16_t queue_id, struct rte_mbuf **pkts, uint32_t count); -uint16_t vs_dequeue_pkts(struct vhost_dev *dev, uint16_t queue_id, - struct rte_mempool *mbuf_pool, - struct rte_mbuf **pkts, uint16_t count); +uint16_t builtin_enqueue_pkts(struct vhost_dev *dev, uint16_t queue_id, + struct rte_mbuf **pkts, uint32_t count); +uint16_t builtin_dequeue_pkts(struct vhost_dev *dev, uint16_t queue_id, + struct rte_mempool *mbuf_pool, + struct rte_mbuf **pkts, uint16_t count); +uint16_t sync_enqueue_pkts(struct vhost_dev *dev, uint16_t queue_id, + struct rte_mbuf **pkts, uint32_t count); +uint16_t sync_dequeue_pkts(struct vhost_dev *dev, uint16_t queue_id, + struct rte_mempool *mbuf_pool, + struct rte_mbuf **pkts, uint16_t count); +uint16_t async_enqueue_pkts(struct vhost_dev *dev, uint16_t queue_id, + struct rte_mbuf **pkts, uint32_t count); +uint16_t async_dequeue_pkts(struct vhost_dev *dev, uint16_t queue_id, + struct rte_mempool *mbuf_pool, + struct rte_mbuf **pkts, uint16_t count); #endif /* _MAIN_H_ */ diff --git a/examples/vhost/virtio_net.c b/examples/vhost/virtio_net.c index 9064fc3a82..2432a96566 100644 --- a/examples/vhost/virtio_net.c +++ b/examples/vhost/virtio_net.c @@ -238,6 +238,13 @@ vs_enqueue_pkts(struct vhost_dev *dev, uint16_t queue_id, return count; } +uint16_t +builtin_enqueue_pkts(struct vhost_dev *dev, uint16_t queue_id, + struct rte_mbuf **pkts, uint32_t count) +{ + return vs_enqueue_pkts(dev, queue_id, pkts, count); +} + static __rte_always_inline int dequeue_pkt(struct vhost_dev *dev, struct rte_vhost_vring *vr, struct rte_mbuf *m, uint16_t desc_idx, @@ -363,7 +370,7 @@ dequeue_pkt(struct vhost_dev *dev, struct rte_vhost_vring *vr, return 0; } -uint16_t +static uint16_t vs_dequeue_pkts(struct vhost_dev *dev, uint16_t queue_id, struct rte_mempool *mbuf_pool, struct rte_mbuf **pkts, uint16_t count) { @@ -440,3 +447,10 @@ vs_dequeue_pkts(struct vhost_dev *dev, uint16_t queue_id, return i; } + +uint16_t +builtin_dequeue_pkts(struct vhost_dev *dev, uint16_t queue_id, + struct rte_mempool *mbuf_pool, struct rte_mbuf **pkts, uint16_t count) +{ + return vs_dequeue_pkts(dev, queue_id, mbuf_pool, pkts, count); +}