summaryrefslogtreecommitdiff
path: root/drivers/infiniband/ulp/ipoib
diff options
context:
space:
mode:
Diffstat (limited to 'drivers/infiniband/ulp/ipoib')
-rw-r--r--drivers/infiniband/ulp/ipoib/ipoib.h29
-rw-r--r--drivers/infiniband/ulp/ipoib/ipoib_cm.c33
-rw-r--r--drivers/infiniband/ulp/ipoib/ipoib_ib.c49
-rw-r--r--drivers/infiniband/ulp/ipoib/ipoib_main.c40
-rw-r--r--drivers/infiniband/ulp/ipoib/ipoib_verbs.c8
5 files changed, 84 insertions, 75 deletions
diff --git a/drivers/infiniband/ulp/ipoib/ipoib.h b/drivers/infiniband/ulp/ipoib/ipoib.h
index bd94b0a6e..79859c4d4 100644
--- a/drivers/infiniband/ulp/ipoib/ipoib.h
+++ b/drivers/infiniband/ulp/ipoib/ipoib.h
@@ -239,7 +239,7 @@ struct ipoib_cm_tx {
struct net_device *dev;
struct ipoib_neigh *neigh;
struct ipoib_path *path;
- struct ipoib_cm_tx_buf *tx_ring;
+ struct ipoib_tx_buf *tx_ring;
unsigned tx_head;
unsigned tx_tail;
unsigned long flags;
@@ -504,6 +504,33 @@ int ipoib_mcast_stop_thread(struct net_device *dev);
void ipoib_mcast_dev_down(struct net_device *dev);
void ipoib_mcast_dev_flush(struct net_device *dev);
+int ipoib_dma_map_tx(struct ib_device *ca, struct ipoib_tx_buf *tx_req);
+void ipoib_dma_unmap_tx(struct ipoib_dev_priv *priv,
+ struct ipoib_tx_buf *tx_req);
+
+static inline void ipoib_build_sge(struct ipoib_dev_priv *priv,
+ struct ipoib_tx_buf *tx_req)
+{
+ int i, off;
+ struct sk_buff *skb = tx_req->skb;
+ skb_frag_t *frags = skb_shinfo(skb)->frags;
+ int nr_frags = skb_shinfo(skb)->nr_frags;
+ u64 *mapping = tx_req->mapping;
+
+ if (skb_headlen(skb)) {
+ priv->tx_sge[0].addr = mapping[0];
+ priv->tx_sge[0].length = skb_headlen(skb);
+ off = 1;
+ } else
+ off = 0;
+
+ for (i = 0; i < nr_frags; ++i) {
+ priv->tx_sge[i + off].addr = mapping[i + off];
+ priv->tx_sge[i + off].length = skb_frag_size(&frags[i]);
+ }
+ priv->tx_wr.num_sge = nr_frags + off;
+}
+
#ifdef CONFIG_INFINIBAND_IPOIB_DEBUG
struct ipoib_mcast_iter *ipoib_mcast_iter_init(struct net_device *dev);
int ipoib_mcast_iter_next(struct ipoib_mcast_iter *iter);
diff --git a/drivers/infiniband/ulp/ipoib/ipoib_cm.c b/drivers/infiniband/ulp/ipoib/ipoib_cm.c
index cf32a778e..ee39be6cc 100644
--- a/drivers/infiniband/ulp/ipoib/ipoib_cm.c
+++ b/drivers/infiniband/ulp/ipoib/ipoib_cm.c
@@ -694,14 +694,12 @@ repost:
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;
return ib_post_send(tx->qp, &priv->tx_wr, &bad_wr);
@@ -710,8 +708,7 @@ static inline int post_send(struct ipoib_dev_priv *priv,
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)) {
@@ -735,24 +732,21 @@ void ipoib_cm_send(struct net_device *dev, struct sk_buff *skb, struct ipoib_cm_
*/
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;
@@ -777,7 +771,7 @@ void ipoib_cm_handle_tx_wc(struct net_device *dev, struct ib_wc *wc)
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",
@@ -791,7 +785,7 @@ void ipoib_cm_handle_tx_wc(struct net_device *dev, struct ib_wc *wc)
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;
@@ -1036,6 +1030,9 @@ static struct ib_qp *ipoib_cm_create_tx_qp(struct net_device *dev, struct ipoib_
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) {
ipoib_warn(priv, "can't use GFP_NOIO for QPs on device %s, using GFP_KERNEL\n",
@@ -1170,7 +1167,7 @@ err_tx:
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",
@@ -1197,8 +1194,7 @@ timeout:
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);
@@ -1455,7 +1451,6 @@ static void ipoib_cm_stale_task(struct work_struct *work)
spin_unlock_irq(&priv->lock);
}
-
static ssize_t show_mode(struct device *d, struct device_attribute *attr,
char *buf)
{
diff --git a/drivers/infiniband/ulp/ipoib/ipoib_ib.c b/drivers/infiniband/ulp/ipoib/ipoib_ib.c
index 63b92cbb2..d266667ca 100644
--- a/drivers/infiniband/ulp/ipoib/ipoib_ib.c
+++ b/drivers/infiniband/ulp/ipoib/ipoib_ib.c
@@ -263,8 +263,7 @@ repost:
"for buf %d\n", wr_id);
}
-static int ipoib_dma_map_tx(struct ib_device *ca,
- struct ipoib_tx_buf *tx_req)
+int ipoib_dma_map_tx(struct ib_device *ca, struct ipoib_tx_buf *tx_req)
{
struct sk_buff *skb = tx_req->skb;
u64 *mapping = tx_req->mapping;
@@ -305,8 +304,8 @@ partial_error:
return -EIO;
}
-static void ipoib_dma_unmap_tx(struct ib_device *ca,
- struct ipoib_tx_buf *tx_req)
+void ipoib_dma_unmap_tx(struct ipoib_dev_priv *priv,
+ struct ipoib_tx_buf *tx_req)
{
struct sk_buff *skb = tx_req->skb;
u64 *mapping = tx_req->mapping;
@@ -314,7 +313,8 @@ static void ipoib_dma_unmap_tx(struct ib_device *ca,
int off;
if (skb_headlen(skb)) {
- ib_dma_unmap_single(ca, mapping[0], skb_headlen(skb), DMA_TO_DEVICE);
+ ib_dma_unmap_single(priv->ca, mapping[0], skb_headlen(skb),
+ DMA_TO_DEVICE);
off = 1;
} else
off = 0;
@@ -322,8 +322,8 @@ static void ipoib_dma_unmap_tx(struct ib_device *ca,
for (i = 0; i < skb_shinfo(skb)->nr_frags; ++i) {
const skb_frag_t *frag = &skb_shinfo(skb)->frags[i];
- ib_dma_unmap_page(ca, mapping[i + off], skb_frag_size(frag),
- DMA_TO_DEVICE);
+ ib_dma_unmap_page(priv->ca, mapping[i + off],
+ skb_frag_size(frag), DMA_TO_DEVICE);
}
}
@@ -389,7 +389,7 @@ static void ipoib_ib_handle_tx_wc(struct net_device *dev, struct ib_wc *wc)
tx_req = &priv->tx_ring[wr_id];
- ipoib_dma_unmap_tx(priv->ca, tx_req);
+ ipoib_dma_unmap_tx(priv, tx_req);
++dev->stats.tx_packets;
dev->stats.tx_bytes += tx_req->skb->len;
@@ -514,24 +514,10 @@ static inline int post_send(struct ipoib_dev_priv *priv,
void *head, int hlen)
{
struct ib_send_wr *bad_wr;
- int i, off;
struct sk_buff *skb = tx_req->skb;
- skb_frag_t *frags = skb_shinfo(skb)->frags;
- int nr_frags = skb_shinfo(skb)->nr_frags;
- u64 *mapping = tx_req->mapping;
- if (skb_headlen(skb)) {
- priv->tx_sge[0].addr = mapping[0];
- priv->tx_sge[0].length = skb_headlen(skb);
- off = 1;
- } else
- off = 0;
+ ipoib_build_sge(priv, tx_req);
- for (i = 0; i < nr_frags; ++i) {
- priv->tx_sge[i + off].addr = mapping[i + off];
- priv->tx_sge[i + off].length = skb_frag_size(&frags[i]);
- }
- priv->tx_wr.num_sge = nr_frags + off;
priv->tx_wr.wr_id = wr_id;
priv->tx_wr.wr.ud.remote_qpn = qpn;
priv->tx_wr.wr.ud.ah = address;
@@ -617,7 +603,7 @@ void ipoib_send(struct net_device *dev, struct sk_buff *skb,
ipoib_warn(priv, "post_send failed, error %d\n", rc);
++dev->stats.tx_errors;
--priv->tx_outstanding;
- ipoib_dma_unmap_tx(priv->ca, tx_req);
+ ipoib_dma_unmap_tx(priv, tx_req);
dev_kfree_skb_any(skb);
if (netif_queue_stopped(dev))
netif_wake_queue(dev);
@@ -868,7 +854,7 @@ int ipoib_ib_dev_stop(struct net_device *dev)
while ((int) priv->tx_tail - (int) priv->tx_head < 0) {
tx_req = &priv->tx_ring[priv->tx_tail &
(ipoib_sendq_size - 1)];
- ipoib_dma_unmap_tx(priv->ca, tx_req);
+ ipoib_dma_unmap_tx(priv, tx_req);
dev_kfree_skb_any(tx_req->skb);
++priv->tx_tail;
--priv->tx_outstanding;
@@ -985,20 +971,21 @@ static inline int update_child_pkey(struct ipoib_dev_priv *priv)
}
static void __ipoib_ib_dev_flush(struct ipoib_dev_priv *priv,
- enum ipoib_flush_level level)
+ enum ipoib_flush_level level,
+ int nesting)
{
struct ipoib_dev_priv *cpriv;
struct net_device *dev = priv->dev;
int result;
- down_read(&priv->vlan_rwsem);
+ down_read_nested(&priv->vlan_rwsem, nesting);
/*
* Flush any child interfaces too -- they might be up even if
* the parent is down.
*/
list_for_each_entry(cpriv, &priv->child_intfs, list)
- __ipoib_ib_dev_flush(cpriv, level);
+ __ipoib_ib_dev_flush(cpriv, level, nesting + 1);
up_read(&priv->vlan_rwsem);
@@ -1076,7 +1063,7 @@ void ipoib_ib_dev_flush_light(struct work_struct *work)
struct ipoib_dev_priv *priv =
container_of(work, struct ipoib_dev_priv, flush_light);
- __ipoib_ib_dev_flush(priv, IPOIB_FLUSH_LIGHT);
+ __ipoib_ib_dev_flush(priv, IPOIB_FLUSH_LIGHT, 0);
}
void ipoib_ib_dev_flush_normal(struct work_struct *work)
@@ -1084,7 +1071,7 @@ void ipoib_ib_dev_flush_normal(struct work_struct *work)
struct ipoib_dev_priv *priv =
container_of(work, struct ipoib_dev_priv, flush_normal);
- __ipoib_ib_dev_flush(priv, IPOIB_FLUSH_NORMAL);
+ __ipoib_ib_dev_flush(priv, IPOIB_FLUSH_NORMAL, 0);
}
void ipoib_ib_dev_flush_heavy(struct work_struct *work)
@@ -1092,7 +1079,7 @@ void ipoib_ib_dev_flush_heavy(struct work_struct *work)
struct ipoib_dev_priv *priv =
container_of(work, struct ipoib_dev_priv, flush_heavy);
- __ipoib_ib_dev_flush(priv, IPOIB_FLUSH_HEAVY);
+ __ipoib_ib_dev_flush(priv, IPOIB_FLUSH_HEAVY, 0);
}
void ipoib_ib_dev_cleanup(struct net_device *dev)
diff --git a/drivers/infiniband/ulp/ipoib/ipoib_main.c b/drivers/infiniband/ulp/ipoib/ipoib_main.c
index 9e1b203d7..b2943c84a 100644
--- a/drivers/infiniband/ulp/ipoib/ipoib_main.c
+++ b/drivers/infiniband/ulp/ipoib/ipoib_main.c
@@ -190,7 +190,7 @@ static netdev_features_t ipoib_fix_features(struct net_device *dev, netdev_featu
struct ipoib_dev_priv *priv = netdev_priv(dev);
if (test_bit(IPOIB_FLAG_ADMIN_CM, &priv->flags))
- features &= ~(NETIF_F_SG | NETIF_F_IP_CSUM | NETIF_F_TSO);
+ features &= ~(NETIF_F_IP_CSUM | NETIF_F_TSO);
return features;
}
@@ -232,6 +232,7 @@ int ipoib_set_mode(struct net_device *dev, const char *buf)
ipoib_warn(priv, "enabling connected mode "
"will cause multicast packet drops\n");
netdev_update_features(dev);
+ dev_set_mtu(dev, ipoib_cm_max_mtu(dev));
rtnl_unlock();
priv->tx_wr.send_flags &= ~IB_SEND_IP_CSUM;
@@ -1128,7 +1129,7 @@ static int ipoib_neigh_hash_init(struct ipoib_dev_priv *priv)
{
struct ipoib_neigh_table *ntbl = &priv->ntbl;
struct ipoib_neigh_hash *htbl;
- struct ipoib_neigh **buckets;
+ struct ipoib_neigh __rcu **buckets;
u32 size;
clear_bit(IPOIB_NEIGH_TBL_FLUSH, &priv->flags);
@@ -1146,7 +1147,7 @@ static int ipoib_neigh_hash_init(struct ipoib_dev_priv *priv)
htbl->size = size;
htbl->mask = (size - 1);
htbl->buckets = buckets;
- ntbl->htbl = htbl;
+ RCU_INIT_POINTER(ntbl->htbl, htbl);
htbl->ntbl = ntbl;
atomic_set(&ntbl->entries, 0);
@@ -1577,7 +1578,8 @@ static struct net_device *ipoib_add_port(const char *format,
SET_NETDEV_DEV(priv->dev, hca->dma_device);
priv->dev->dev_id = port - 1;
- if (!ib_query_port(hca, port, &attr))
+ result = ib_query_port(hca, port, &attr);
+ if (!result)
priv->max_ib_mtu = ib_mtu_enum_to_int(attr.max_mtu);
else {
printk(KERN_WARNING "%s: ib_query_port %d failed\n",
@@ -1598,7 +1600,8 @@ static struct net_device *ipoib_add_port(const char *format,
goto device_init_failed;
}
- if (ipoib_set_dev_features(priv, hca))
+ result = ipoib_set_dev_features(priv, hca);
+ if (result)
goto device_init_failed;
/*
@@ -1684,10 +1687,8 @@ static void ipoib_add_one(struct ib_device *device)
struct list_head *dev_list;
struct net_device *dev;
struct ipoib_dev_priv *priv;
- int s, e, p;
-
- if (rdma_node_get_transport(device->node_type) != RDMA_TRANSPORT_IB)
- return;
+ int p;
+ int count = 0;
dev_list = kmalloc(sizeof *dev_list, GFP_KERNEL);
if (!dev_list)
@@ -1695,24 +1696,22 @@ static void ipoib_add_one(struct ib_device *device)
INIT_LIST_HEAD(dev_list);
- if (device->node_type == RDMA_NODE_IB_SWITCH) {
- s = 0;
- e = 0;
- } else {
- s = 1;
- e = device->phys_port_cnt;
- }
-
- for (p = s; p <= e; ++p) {
- if (rdma_port_get_link_layer(device, p) != IB_LINK_LAYER_INFINIBAND)
+ for (p = rdma_start_port(device); p <= rdma_end_port(device); ++p) {
+ if (!rdma_protocol_ib(device, p))
continue;
dev = ipoib_add_port("ib%d", device, p);
if (!IS_ERR(dev)) {
priv = netdev_priv(dev);
list_add_tail(&priv->list, dev_list);
+ count++;
}
}
+ if (!count) {
+ kfree(dev_list);
+ return;
+ }
+
ib_set_client_data(device, &ipoib_client, dev_list);
}
@@ -1721,9 +1720,6 @@ static void ipoib_remove_one(struct ib_device *device)
struct ipoib_dev_priv *priv, *tmp;
struct list_head *dev_list;
- if (rdma_node_get_transport(device->node_type) != RDMA_TRANSPORT_IB)
- return;
-
dev_list = ib_get_client_data(device, &ipoib_client);
if (!dev_list)
return;
diff --git a/drivers/infiniband/ulp/ipoib/ipoib_verbs.c b/drivers/infiniband/ulp/ipoib/ipoib_verbs.c
index 2d13fd08c..851c8219d 100644
--- a/drivers/infiniband/ulp/ipoib/ipoib_verbs.c
+++ b/drivers/infiniband/ulp/ipoib/ipoib_verbs.c
@@ -141,6 +141,7 @@ int ipoib_transport_dev_init(struct net_device *dev, struct ib_device *ca)
.sq_sig_type = IB_SIGNAL_ALL_WR,
.qp_type = IB_QPT_UD
};
+ struct ib_cq_init_attr cq_attr = {};
int ret, size;
int i;
@@ -179,14 +180,17 @@ int ipoib_transport_dev_init(struct net_device *dev, struct ib_device *ca)
if (ret != -ENOSYS)
goto out_free_wq;
- priv->recv_cq = ib_create_cq(priv->ca, ipoib_ib_completion, NULL, dev, size, 0);
+ cq_attr.cqe = size;
+ priv->recv_cq = ib_create_cq(priv->ca, ipoib_ib_completion, NULL,
+ dev, &cq_attr);
if (IS_ERR(priv->recv_cq)) {
printk(KERN_WARNING "%s: failed to create receive CQ\n", ca->name);
goto out_cm_dev_cleanup;
}
+ cq_attr.cqe = ipoib_sendq_size;
priv->send_cq = ib_create_cq(priv->ca, ipoib_send_comp_handler, NULL,
- dev, ipoib_sendq_size, 0);
+ dev, &cq_attr);
if (IS_ERR(priv->send_cq)) {
printk(KERN_WARNING "%s: failed to create send CQ\n", ca->name);
goto out_free_recv_cq;