From owner-svn-soc-all@freebsd.org Mon Jun 27 09:35:37 2016 Return-Path: Delivered-To: svn-soc-all@mailman.ysv.freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:1900:2254:206a::19:1]) by mailman.ysv.freebsd.org (Postfix) with ESMTP id BBA32B81020 for ; Mon, 27 Jun 2016 09:35:37 +0000 (UTC) (envelope-from vincenzo@FreeBSD.org) Received: from socsvn.freebsd.org (socsvn.freebsd.org [IPv6:2001:1900:2254:206a::50:2]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (Client did not present a certificate) by mx1.freebsd.org (Postfix) with ESMTPS id AF0EB2234 for ; Mon, 27 Jun 2016 09:35:37 +0000 (UTC) (envelope-from vincenzo@FreeBSD.org) Received: from socsvn.freebsd.org ([127.0.1.124]) by socsvn.freebsd.org (8.15.2/8.15.2) with ESMTP id u5R9ZbOh029981 for ; Mon, 27 Jun 2016 09:35:37 GMT (envelope-from vincenzo@FreeBSD.org) Received: (from www@localhost) by socsvn.freebsd.org (8.15.2/8.15.2/Submit) id u5R9ZbeE029972 for svn-soc-all@FreeBSD.org; Mon, 27 Jun 2016 09:35:37 GMT (envelope-from vincenzo@FreeBSD.org) Date: Mon, 27 Jun 2016 09:35:37 GMT Message-Id: <201606270935.u5R9ZbeE029972@socsvn.freebsd.org> X-Authentication-Warning: socsvn.freebsd.org: www set sender to vincenzo@FreeBSD.org using -f From: vincenzo@FreeBSD.org To: svn-soc-all@FreeBSD.org Subject: socsvn commit: r305584 - soc2016/vincenzo/head/sys/dev/netmap MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit X-BeenThere: svn-soc-all@freebsd.org X-Mailman-Version: 2.1.22 Precedence: list List-Id: SVN commit messages for the entire Summer of Code repository List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 27 Jun 2016 09:35:37 -0000 Author: vincenzo Date: Mon Jun 27 09:35:36 2016 New Revision: 305584 URL: http://svnweb.FreeBSD.org/socsvn/?view=rev&rev=305584 Log: freebsd: ptnet: move transmit functionality at the bottom Modified: soc2016/vincenzo/head/sys/dev/netmap/if_ptnet.c Modified: soc2016/vincenzo/head/sys/dev/netmap/if_ptnet.c ============================================================================== --- soc2016/vincenzo/head/sys/dev/netmap/if_ptnet.c Mon Jun 27 09:35:16 2016 (r305583) +++ soc2016/vincenzo/head/sys/dev/netmap/if_ptnet.c Mon Jun 27 09:35:36 2016 (r305584) @@ -784,195 +784,6 @@ return 0; } -static inline void -ptnet_sync_tail(struct ptnet_ring *ptring, struct netmap_kring *kring) -{ - struct netmap_ring *ring = kring->ring; - - /* Update hwcur and hwtail as known by the host. */ - ptnetmap_guest_read_kring_csb(ptring, kring); - - /* nm_sync_finalize */ - ring->tail = kring->rtail = kring->nr_hwtail; -} - -static void -ptnet_ring_update(struct ptnet_queue *pq, struct netmap_kring *kring, - unsigned int head) -{ - struct netmap_ring *ring = kring->ring; - struct ptnet_ring *ptring = pq->ptring; - - /* Some packets have been pushed to the netmap ring. We have - * to tell the host to process the new packets, updating cur - * and head in the CSB. */ - ring->head = ring->cur = head; - - /* nm_txsync_prologue */ - kring->rcur = kring->rhead = ring->head; - - ptnetmap_guest_write_kring_csb(ptring, kring->rcur, kring->rhead); - - /* Kick the host if needed. */ - if (NM_ACCESS_ONCE(ptring->host_need_kick)) { - ptring->sync_flags = NAF_FORCE_RECLAIM; - bus_write_4(pq->sc->iomem, pq->kick, 0); - } -} - -static int -ptnet_transmit(struct ifnet *ifp, struct mbuf *m) -{ - struct ptnet_softc *sc = ifp->if_softc; - struct netmap_adapter *na = &sc->ptna_dr.hwup.up; - unsigned int batch_count = 0; - struct ptnet_ring *ptring; - struct netmap_kring *kring; - struct netmap_ring *ring; - struct netmap_slot *slot; - struct ptnet_queue *pq; - unsigned int prev_head; - unsigned int head; - unsigned int lim; - struct mbuf *mf; - int nmbuf_bytes; - uint8_t *nmbuf; - - DBG(device_printf(sc->dev, "transmit %p\n", m)); - - pq = sc->queues + 0; - - if (m) { - int err; - - /* Here we are called by the network stack, and not by - * by the taskqueue thread. */ - err = drbr_enqueue(ifp, pq->bufring, m); - m = NULL; /* just to stay safe */ - if (unlikely(err)) { - device_printf(sc->dev, "%s: drbr_enqueue() failed %d\n", - __func__, err); - return err; - } - } - - if (unlikely(!(ifp->if_drv_flags & IFF_DRV_RUNNING))) { - RD(1, "Interface is down"); - return ENETDOWN; - } - - if (!PTNET_Q_TRYLOCK(pq)) { - /* We failed to acquire the lock, schedule the taskqueue. */ - RD(1, "Deferring TX work"); - taskqueue_enqueue(pq->taskq, &pq->task); - - return 0; - } - - ptring = pq->ptring; - kring = na->tx_rings + pq->kring_id; - ring = kring->ring; - lim = kring->nkr_num_slots - 1; - - /* Update hwcur and hwtail (completed TX slots) as known by the host, - * by reading from CSB. */ - ptnet_sync_tail(ptring, kring); - - head = ring->head; - slot = ring->slot + head; - nmbuf = NMB(na, slot); - nmbuf_bytes = 0; - - while (head != ring->tail) { - m = drbr_peek(ifp, pq->bufring); - if (!m) { - break; - } - - for (prev_head = head, mf = m; mf; mf = mf->m_next) { - uint8_t *mdata = mf->m_data; - int mlen = mf->m_len; - - for (;;) { - int copy = NETMAP_BUF_SIZE(na) - nmbuf_bytes; - - if (mlen < copy) { - copy = mlen; - } - memcpy(nmbuf, mdata, copy); - - mdata += copy; - mlen -= copy; - nmbuf += copy; - nmbuf_bytes += copy; - - if (!mlen) { - break; - } - - slot->len = nmbuf_bytes; - slot->flags = NS_MOREFRAG; - - head = nm_next(head, lim); - if (head == ring->tail) { - /* Run out of slots while processing - * a packet. Reset head to the previous - * position and requeue the mbuf. */ - device_printf(sc->dev, "%s: Drop, " - " no free slots\n", - __func__); - head = prev_head; - drbr_putback(ifp, pq->bufring, m); - goto escape; - } - slot = ring->slot + head; - nmbuf = NMB(na, slot); - nmbuf_bytes = 0; - } - } - - /* Complete last slot and update head. */ - slot->len = nmbuf_bytes; - slot->flags = 0; - head = nm_next(head, lim); - - /* Consume the packet just processed. */ - drbr_advance(ifp, pq->bufring); - m_freem(m); - - if (++batch_count == PTNET_TX_BATCH) { - batch_count = 0; - ptnet_ring_update(pq, kring, head); - } - } -escape: - if (batch_count) { - ptnet_ring_update(pq, kring, head); - } - - if (head == ring->tail) { - /* Reactivate the interrupts so that we can be notified - * when some free slots are made available by the host. */ - ptring->guest_need_kick = 1; - - /* Double-check. */ - ptnet_sync_tail(ptring, kring); - if (unlikely(head != ring->tail)) { - RD(1, "Found more slots by doublecheck"); - /* More slots were freed before reactivating - * the interrupts. */ - ptring->guest_need_kick = 0; - if (!drbr_empty(ifp, pq->bufring)) { - taskqueue_enqueue(pq->taskq, &pq->task); - } - } - } - - PTNET_Q_UNLOCK(pq); - - return 0; -} - static void ptnet_qflush(struct ifnet *ifp) { @@ -1314,6 +1125,195 @@ ptnet_rx_eof(pq); } +static inline void +ptnet_sync_tail(struct ptnet_ring *ptring, struct netmap_kring *kring) +{ + struct netmap_ring *ring = kring->ring; + + /* Update hwcur and hwtail as known by the host. */ + ptnetmap_guest_read_kring_csb(ptring, kring); + + /* nm_sync_finalize */ + ring->tail = kring->rtail = kring->nr_hwtail; +} + +static void +ptnet_ring_update(struct ptnet_queue *pq, struct netmap_kring *kring, + unsigned int head) +{ + struct netmap_ring *ring = kring->ring; + struct ptnet_ring *ptring = pq->ptring; + + /* Some packets have been pushed to the netmap ring. We have + * to tell the host to process the new packets, updating cur + * and head in the CSB. */ + ring->head = ring->cur = head; + + /* nm_txsync_prologue */ + kring->rcur = kring->rhead = ring->head; + + ptnetmap_guest_write_kring_csb(ptring, kring->rcur, kring->rhead); + + /* Kick the host if needed. */ + if (NM_ACCESS_ONCE(ptring->host_need_kick)) { + ptring->sync_flags = NAF_FORCE_RECLAIM; + bus_write_4(pq->sc->iomem, pq->kick, 0); + } +} + +static int +ptnet_transmit(struct ifnet *ifp, struct mbuf *m) +{ + struct ptnet_softc *sc = ifp->if_softc; + struct netmap_adapter *na = &sc->ptna_dr.hwup.up; + unsigned int batch_count = 0; + struct ptnet_ring *ptring; + struct netmap_kring *kring; + struct netmap_ring *ring; + struct netmap_slot *slot; + struct ptnet_queue *pq; + unsigned int prev_head; + unsigned int head; + unsigned int lim; + struct mbuf *mf; + int nmbuf_bytes; + uint8_t *nmbuf; + + DBG(device_printf(sc->dev, "transmit %p\n", m)); + + pq = sc->queues + 0; + + if (m) { + int err; + + /* Here we are called by the network stack, and not by + * by the taskqueue thread. */ + err = drbr_enqueue(ifp, pq->bufring, m); + m = NULL; /* just to stay safe */ + if (unlikely(err)) { + device_printf(sc->dev, "%s: drbr_enqueue() failed %d\n", + __func__, err); + return err; + } + } + + if (unlikely(!(ifp->if_drv_flags & IFF_DRV_RUNNING))) { + RD(1, "Interface is down"); + return ENETDOWN; + } + + if (!PTNET_Q_TRYLOCK(pq)) { + /* We failed to acquire the lock, schedule the taskqueue. */ + RD(1, "Deferring TX work"); + taskqueue_enqueue(pq->taskq, &pq->task); + + return 0; + } + + ptring = pq->ptring; + kring = na->tx_rings + pq->kring_id; + ring = kring->ring; + lim = kring->nkr_num_slots - 1; + + /* Update hwcur and hwtail (completed TX slots) as known by the host, + * by reading from CSB. */ + ptnet_sync_tail(ptring, kring); + + head = ring->head; + slot = ring->slot + head; + nmbuf = NMB(na, slot); + nmbuf_bytes = 0; + + while (head != ring->tail) { + m = drbr_peek(ifp, pq->bufring); + if (!m) { + break; + } + + for (prev_head = head, mf = m; mf; mf = mf->m_next) { + uint8_t *mdata = mf->m_data; + int mlen = mf->m_len; + + for (;;) { + int copy = NETMAP_BUF_SIZE(na) - nmbuf_bytes; + + if (mlen < copy) { + copy = mlen; + } + memcpy(nmbuf, mdata, copy); + + mdata += copy; + mlen -= copy; + nmbuf += copy; + nmbuf_bytes += copy; + + if (!mlen) { + break; + } + + slot->len = nmbuf_bytes; + slot->flags = NS_MOREFRAG; + + head = nm_next(head, lim); + if (head == ring->tail) { + /* Run out of slots while processing + * a packet. Reset head to the previous + * position and requeue the mbuf. */ + device_printf(sc->dev, "%s: Drop, " + " no free slots\n", + __func__); + head = prev_head; + drbr_putback(ifp, pq->bufring, m); + goto escape; + } + slot = ring->slot + head; + nmbuf = NMB(na, slot); + nmbuf_bytes = 0; + } + } + + /* Complete last slot and update head. */ + slot->len = nmbuf_bytes; + slot->flags = 0; + head = nm_next(head, lim); + + /* Consume the packet just processed. */ + drbr_advance(ifp, pq->bufring); + m_freem(m); + + if (++batch_count == PTNET_TX_BATCH) { + batch_count = 0; + ptnet_ring_update(pq, kring, head); + } + } +escape: + if (batch_count) { + ptnet_ring_update(pq, kring, head); + } + + if (head == ring->tail) { + /* Reactivate the interrupts so that we can be notified + * when some free slots are made available by the host. */ + ptring->guest_need_kick = 1; + + /* Double-check. */ + ptnet_sync_tail(ptring, kring); + if (unlikely(head != ring->tail)) { + RD(1, "Found more slots by doublecheck"); + /* More slots were freed before reactivating + * the interrupts. */ + ptring->guest_need_kick = 0; + if (!drbr_empty(ifp, pq->bufring)) { + taskqueue_enqueue(pq->taskq, &pq->task); + } + } + } + + PTNET_Q_UNLOCK(pq); + + return 0; +} + static int ptnet_rx_eof(struct ptnet_queue *pq) {