--- zzzz-none-000/linux-3.10.107/drivers/infiniband/ulp/ipoib/ipoib_cm.c 2017-06-27 09:49:32.000000000 +0000 +++ scorpion-7490-727/linux-3.10.107/drivers/infiniband/ulp/ipoib/ipoib_cm.c 2021-02-04 17:41:59.000000000 +0000 @@ -63,6 +63,8 @@ #define IPOIB_CM_RX_DELAY (3 * 256 * HZ) #define IPOIB_CM_RX_UPDATE_MASK (0x3) +#define IPOIB_CM_RX_RESERVE (ALIGN(IPOIB_HARD_LEN, 16) - IPOIB_ENCAP_LEN) + static struct ib_qp_attr ipoib_cm_err_attr = { .qp_state = IB_QPS_ERR }; @@ -140,21 +142,22 @@ static struct sk_buff *ipoib_cm_alloc_rx_skb(struct net_device *dev, struct ipoib_cm_rx_buf *rx_ring, int id, int frags, - u64 mapping[IPOIB_CM_RX_SG]) + u64 mapping[IPOIB_CM_RX_SG], + gfp_t gfp) { struct ipoib_dev_priv *priv = netdev_priv(dev); struct sk_buff *skb; int i; - skb = dev_alloc_skb(IPOIB_CM_HEAD_SIZE + 12); + skb = dev_alloc_skb(ALIGN(IPOIB_CM_HEAD_SIZE + IPOIB_PSEUDO_LEN, 16)); if (unlikely(!skb)) return NULL; /* - * IPoIB adds a 4 byte header. So we need 12 more bytes to align the + * IPoIB adds a IPOIB_ENCAP_LEN byte header, this will align the * IP header to a multiple of 16. */ - skb_reserve(skb, 12); + skb_reserve(skb, IPOIB_CM_RX_RESERVE); mapping[0] = ib_dma_map_single(priv->ca, skb->data, IPOIB_CM_HEAD_SIZE, DMA_FROM_DEVICE); @@ -164,7 +167,7 @@ } for (i = 0; i < frags; i++) { - struct page *page = alloc_page(GFP_ATOMIC); + struct page *page = alloc_page(gfp); if (!page) goto partial_error; @@ -331,7 +334,7 @@ int i; for (i = 0; i < priv->cm.num_frags; ++i) - sge[i].lkey = priv->mr->lkey; + sge[i].lkey = priv->pd->local_dma_lkey; sge[0].length = IPOIB_CM_HEAD_SIZE; for (i = 1; i < priv->cm.num_frags; ++i) @@ -382,10 +385,11 @@ for (i = 0; i < ipoib_recvq_size; ++i) { if (!ipoib_cm_alloc_rx_skb(dev, rx->rx_ring, i, IPOIB_CM_RX_SG - 1, - rx->rx_ring[i].mapping)) { + rx->rx_ring[i].mapping, + GFP_KERNEL)) { ipoib_warn(priv, "failed to allocate receive buffer %d\n", i); - ret = -ENOMEM; - goto err_count; + ret = -ENOMEM; + goto err_count; } ret = ipoib_cm_post_receive_nonsrq(dev, rx, &t->wr, t->sge, i); if (ret) { @@ -472,7 +476,7 @@ } spin_lock_irq(&priv->lock); - queue_delayed_work(ipoib_workqueue, + queue_delayed_work(priv->wq, &priv->cm.stale_task, IPOIB_CM_RX_DELAY); /* Add this entry to passive ids list head, but do not re-add it * if IB_EVENT_QP_LAST_WQE_REACHED has moved it to flush list. */ @@ -574,7 +578,7 @@ spin_lock_irqsave(&priv->lock, flags); list_splice_init(&priv->cm.rx_drain_list, &priv->cm.rx_reap_list); ipoib_cm_start_rx_drain(priv); - queue_work(ipoib_workqueue, &priv->cm.rx_reap_task); + queue_work(priv->wq, &priv->cm.rx_reap_task); spin_unlock_irqrestore(&priv->lock, flags); } else ipoib_warn(priv, "cm recv completion event with wrid %d (> %d)\n", @@ -601,7 +605,7 @@ spin_lock_irqsave(&priv->lock, flags); list_move(&p->list, &priv->cm.rx_reap_list); spin_unlock_irqrestore(&priv->lock, flags); - queue_work(ipoib_workqueue, &priv->cm.rx_reap_task); + queue_work(priv->wq, &priv->cm.rx_reap_task); } return; } @@ -622,9 +626,9 @@ if (wc->byte_len < IPOIB_CM_COPYBREAK) { int dlen = wc->byte_len; - small_skb = dev_alloc_skb(dlen + 12); + small_skb = dev_alloc_skb(dlen + IPOIB_CM_RX_RESERVE); if (small_skb) { - skb_reserve(small_skb, 12); + skb_reserve(small_skb, IPOIB_CM_RX_RESERVE); ib_dma_sync_single_for_cpu(priv->ca, rx_ring[wr_id].mapping[0], dlen, DMA_FROM_DEVICE); skb_copy_from_linear_data(skb, small_skb->data, dlen); @@ -639,7 +643,8 @@ frags = PAGE_ALIGN(wc->byte_len - min(wc->byte_len, (unsigned)IPOIB_CM_HEAD_SIZE)) / PAGE_SIZE; - newskb = ipoib_cm_alloc_rx_skb(dev, rx_ring, wr_id, frags, mapping); + newskb = ipoib_cm_alloc_rx_skb(dev, rx_ring, wr_id, frags, + mapping, GFP_ATOMIC); if (unlikely(!newskb)) { /* * If we can't allocate a new RX buffer, dump @@ -660,8 +665,7 @@ copied: skb->protocol = ((struct ipoib_header *) skb->data)->proto; - skb_reset_mac_header(skb); - skb_pull(skb, IPOIB_ENCAP_LEN); + skb_add_pseudo_hdr(skb); ++dev->stats.rx_packets; dev->stats.rx_bytes += skb->len; @@ -691,24 +695,21 @@ static inline int post_send(struct ipoib_dev_priv *priv, struct ipoib_cm_tx *tx, unsigned int wr_id, - u64 addr, int len) + struct ipoib_tx_buf *tx_req) { struct ib_send_wr *bad_wr; - priv->tx_sge[0].addr = addr; - priv->tx_sge[0].length = len; + ipoib_build_sge(priv, tx_req); - priv->tx_wr.num_sge = 1; - priv->tx_wr.wr_id = wr_id | IPOIB_OP_CM; + priv->tx_wr.wr.wr_id = wr_id | IPOIB_OP_CM; - return ib_post_send(tx->qp, &priv->tx_wr, &bad_wr); + return ib_post_send(tx->qp, &priv->tx_wr.wr, &bad_wr); } void ipoib_cm_send(struct net_device *dev, struct sk_buff *skb, struct ipoib_cm_tx *tx) { struct ipoib_dev_priv *priv = netdev_priv(dev); - struct ipoib_cm_tx_buf *tx_req; - u64 addr; + struct ipoib_tx_buf *tx_req; int rc; if (unlikely(skb->len > tx->mtu)) { @@ -732,24 +733,21 @@ */ tx_req = &tx->tx_ring[tx->tx_head & (ipoib_sendq_size - 1)]; tx_req->skb = skb; - addr = ib_dma_map_single(priv->ca, skb->data, skb->len, DMA_TO_DEVICE); - if (unlikely(ib_dma_mapping_error(priv->ca, addr))) { + + if (unlikely(ipoib_dma_map_tx(priv->ca, tx_req))) { ++dev->stats.tx_errors; dev_kfree_skb_any(skb); return; } - tx_req->mapping = addr; - skb_orphan(skb); skb_dst_drop(skb); - rc = post_send(priv, tx, tx->tx_head & (ipoib_sendq_size - 1), - addr, skb->len); + rc = post_send(priv, tx, tx->tx_head & (ipoib_sendq_size - 1), tx_req); if (unlikely(rc)) { ipoib_warn(priv, "post_send failed, error %d\n", rc); ++dev->stats.tx_errors; - ib_dma_unmap_single(priv->ca, addr, skb->len, DMA_TO_DEVICE); + ipoib_dma_unmap_tx(priv, tx_req); dev_kfree_skb_any(skb); } else { dev->trans_start = jiffies; @@ -774,7 +772,7 @@ struct ipoib_dev_priv *priv = netdev_priv(dev); struct ipoib_cm_tx *tx = wc->qp->qp_context; unsigned int wr_id = wc->wr_id & ~IPOIB_OP_CM; - struct ipoib_cm_tx_buf *tx_req; + struct ipoib_tx_buf *tx_req; unsigned long flags; ipoib_dbg_data(priv, "cm send completion: id %d, status: %d\n", @@ -788,7 +786,7 @@ tx_req = &tx->tx_ring[wr_id]; - ib_dma_unmap_single(priv->ca, tx_req->mapping, tx_req->skb->len, DMA_TO_DEVICE); + ipoib_dma_unmap_tx(priv, tx_req); /* FIXME: is this right? Shouldn't we only increment on success? */ ++dev->stats.tx_packets; @@ -817,7 +815,6 @@ if (neigh) { neigh->cm = NULL; - list_del(&neigh->list); ipoib_neigh_free(neigh); tx->neigh = NULL; @@ -825,7 +822,7 @@ if (test_and_clear_bit(IPOIB_FLAG_INITIALIZED, &tx->flags)) { list_move(&tx->list, &priv->cm.reap_list); - queue_work(ipoib_workqueue, &priv->cm.reap_task); + queue_work(priv->wq, &priv->cm.reap_task); } clear_bit(IPOIB_FLAG_OPER_UP, &tx->flags); @@ -852,7 +849,7 @@ } ret = ib_cm_listen(priv->cm.id, cpu_to_be64(IPOIB_CM_IETF_ID | priv->qp->qp_num), - 0, NULL); + 0); if (ret) { printk(KERN_WARNING "%s: failed to listen on ID 0x%llx\n", priv->ca->name, IPOIB_CM_IETF_ID | priv->qp->qp_num); @@ -1028,10 +1025,21 @@ .cap.max_send_sge = 1, .sq_sig_type = IB_SIGNAL_ALL_WR, .qp_type = IB_QPT_RC, - .qp_context = tx + .qp_context = tx, + .create_flags = IB_QP_CREATE_USE_GFP_NOIO }; - return ib_create_qp(priv->pd, &attr); + struct ib_qp *tx_qp; + + if (dev->features & NETIF_F_SG) + attr.cap.max_send_sge = MAX_SKB_FRAGS + 1; + + tx_qp = ib_create_qp(priv->pd, &attr); + if (PTR_ERR(tx_qp) == -EINVAL) { + attr.create_flags &= ~IB_QP_CREATE_USE_GFP_NOIO; + tx_qp = ib_create_qp(priv->pd, &attr); + } + return tx_qp; } static int ipoib_cm_send_req(struct net_device *dev, @@ -1102,12 +1110,14 @@ struct ipoib_dev_priv *priv = netdev_priv(p->dev); int ret; - p->tx_ring = vzalloc(ipoib_sendq_size * sizeof *p->tx_ring); + p->tx_ring = __vmalloc(ipoib_sendq_size * sizeof *p->tx_ring, + GFP_NOIO, PAGE_KERNEL); if (!p->tx_ring) { ipoib_warn(priv, "failed to allocate tx ring\n"); ret = -ENOMEM; goto err_tx; } + memset(p->tx_ring, 0, ipoib_sendq_size * sizeof *p->tx_ring); p->qp = ipoib_cm_create_tx_qp(p->dev, p); if (IS_ERR(p->qp)) { @@ -1156,7 +1166,7 @@ static void ipoib_cm_tx_destroy(struct ipoib_cm_tx *p) { struct ipoib_dev_priv *priv = netdev_priv(p->dev); - struct ipoib_cm_tx_buf *tx_req; + struct ipoib_tx_buf *tx_req; unsigned long begin; ipoib_dbg(priv, "Destroy active connection 0x%x head 0x%x tail 0x%x\n", @@ -1183,8 +1193,7 @@ while ((int) p->tx_tail - (int) p->tx_head < 0) { tx_req = &p->tx_ring[p->tx_tail & (ipoib_sendq_size - 1)]; - ib_dma_unmap_single(priv->ca, tx_req->mapping, tx_req->skb->len, - DMA_TO_DEVICE); + ipoib_dma_unmap_tx(priv, tx_req); dev_kfree_skb_any(tx_req->skb); ++p->tx_tail; netif_tx_lock_bh(p->dev); @@ -1234,7 +1243,6 @@ if (neigh) { neigh->cm = NULL; - list_del(&neigh->list); ipoib_neigh_free(neigh); tx->neigh = NULL; @@ -1242,7 +1250,7 @@ if (test_and_clear_bit(IPOIB_FLAG_INITIALIZED, &tx->flags)) { list_move(&tx->list, &priv->cm.reap_list); - queue_work(ipoib_workqueue, &priv->cm.reap_task); + queue_work(priv->wq, &priv->cm.reap_task); } spin_unlock_irqrestore(&priv->lock, flags); @@ -1271,7 +1279,7 @@ tx->dev = dev; list_add(&tx->list, &priv->cm.start_list); set_bit(IPOIB_FLAG_INITIALIZED, &tx->flags); - queue_work(ipoib_workqueue, &priv->cm.start_task); + queue_work(priv->wq, &priv->cm.start_task); return tx; } @@ -1282,7 +1290,7 @@ if (test_and_clear_bit(IPOIB_FLAG_INITIALIZED, &tx->flags)) { spin_lock_irqsave(&priv->lock, flags); list_move(&tx->list, &priv->cm.reap_list); - queue_work(ipoib_workqueue, &priv->cm.reap_task); + queue_work(priv->wq, &priv->cm.reap_task); ipoib_dbg(priv, "Reap connection for gid %pI6\n", tx->neigh->daddr + 4); tx->neigh = NULL; @@ -1341,7 +1349,6 @@ neigh = p->neigh; if (neigh) { neigh->cm = NULL; - list_del(&neigh->list); ipoib_neigh_free(neigh); } list_del(&p->list); @@ -1421,7 +1428,7 @@ skb_queue_tail(&priv->cm.skb_queue, skb); if (e) - queue_work(ipoib_workqueue, &priv->cm.skb_task); + queue_work(priv->wq, &priv->cm.skb_task); } static void ipoib_cm_rx_reap(struct work_struct *work) @@ -1454,12 +1461,11 @@ } if (!list_empty(&priv->cm.passive_ids)) - queue_delayed_work(ipoib_workqueue, + queue_delayed_work(priv->wq, &priv->cm.stale_task, IPOIB_CM_RX_DELAY); spin_unlock_irq(&priv->lock); } - static ssize_t show_mode(struct device *d, struct device_attribute *attr, char *buf) { @@ -1577,7 +1583,8 @@ for (i = 0; i < ipoib_recvq_size; ++i) { if (!ipoib_cm_alloc_rx_skb(dev, priv->cm.srq_ring, i, priv->cm.num_frags - 1, - priv->cm.srq_ring[i].mapping)) { + priv->cm.srq_ring[i].mapping, + GFP_KERNEL)) { ipoib_warn(priv, "failed to allocate " "receive buffer %d\n", i); ipoib_cm_dev_cleanup(dev);