From: Moni Shoua <mo...@mellanox.com>
Since RoCEv2 is a protocol over IP header it is required to send IGMP
join and leave requests to the network when joining and leaving
multicast groups.
Signed-off-by: Moni Shoua <mo...@mellanox.com>
Signed-off-by: Somnath Kotur <somnath.ko...@emulex.com>
---
drivers/infiniband/core/cma.c | 78 ++++++++++++++++++++++++++++++++++---
drivers/infiniband/core/multicast.c | 18 ++++++++-
include/rdma/ib_sa.h | 3 ++
3 files changed, 92 insertions(+), 7 deletions(-)
diff --git a/drivers/infiniband/core/cma.c b/drivers/infiniband/core/cma.c
index 6f345e2..8f997d7 100644
--- a/drivers/infiniband/core/cma.c
+++ b/drivers/infiniband/core/cma.c
@@ -38,6 +38,7 @@
#include <linux/in6.h>
#include <linux/mutex.h>
#include <linux/random.h>
+#include <linux/igmp.h>
#include <linux/idr.h>
#include <linux/inetdevice.h>
#include <linux/slab.h>
@@ -196,6 +197,7 @@ struct cma_multicast {
void *context;
struct sockaddr_storage addr;
struct kref mcref;
+ bool igmp_joined;
};
struct cma_work {
@@ -283,6 +285,26 @@ static inline void cma_set_ip_ver(struct cma_hdr *hdr, u8
ip_ver)
hdr->ip_version = (ip_ver << 4) | (hdr->ip_version & 0xF);
}
+static int cma_igmp_send(struct net_device *ndev, union ib_gid *mgid, bool
join)
+{
+ struct in_device *in_dev = NULL;
+
+ if (ndev) {
+ rtnl_lock();
+ in_dev = __in_dev_get_rtnl(ndev);
+ if (in_dev) {
+ if (join)
+ ip_mc_inc_group(in_dev,
+ *(__be32 *)(mgid->raw+12));
+ else
+ ip_mc_dec_group(in_dev,
+ *(__be32 *)(mgid->raw+12));
+ }
+ rtnl_unlock();
+ }
+ return (in_dev) ? 0 : -ENODEV;
+}
+
static void cma_attach_to_dev(struct rdma_id_private *id_priv,
struct cma_device *cma_dev)
{
@@ -1076,6 +1098,20 @@ static void cma_leave_mc_groups(struct rdma_id_private
*id_priv)
kfree(mc);
break;
case IB_LINK_LAYER_ETHERNET:
+ if (mc->igmp_joined) {
+ struct rdma_dev_addr *dev_addr =
&id_priv->id.route.addr.dev_addr;
+ struct net_device *ndev = NULL;
+
+ if (dev_addr->bound_dev_if)
+ ndev = dev_get_by_index(&init_net,
+
dev_addr->bound_dev_if);
+ if (ndev) {
+ cma_igmp_send(ndev,
+
&mc->multicast.ib->rec.mgid,
+ false);
+ dev_put(ndev);
+ }
+ }
kref_put(&mc->mcref, release_mc);
break;
default:
@@ -3356,7 +3392,7 @@ static int cma_iboe_join_multicast(struct rdma_id_private
*id_priv,
{
struct iboe_mcast_work *work;
struct rdma_dev_addr *dev_addr = &id_priv->id.route.addr.dev_addr;
- int err;
+ int err = 0;
struct sockaddr *addr = (struct sockaddr *)&mc->addr;
struct net_device *ndev = NULL;
@@ -3388,13 +3424,30 @@ static int cma_iboe_join_multicast(struct
rdma_id_private *id_priv,
mc->multicast.ib->rec.rate = iboe_get_rate(ndev);
mc->multicast.ib->rec.hop_limit = 1;
mc->multicast.ib->rec.mtu = iboe_get_mtu(ndev->mtu);
+ mc->multicast.ib->rec.ifindex = dev_addr->bound_dev_if;
+ mc->multicast.ib->rec.net = &init_net;
+ rdma_ip2gid((struct sockaddr *)&id_priv->id.route.addr.src_addr,
+ &mc->multicast.ib->rec.port_gid);
+
+ if (addr->sa_family == AF_INET) {
+ mc->multicast.ib->rec.gid_type =
+ id_priv->cma_dev->default_gid_type;
+ if (mc->multicast.ib->rec.gid_type == IB_GID_TYPE_ROCE_V2)
+ err = cma_igmp_send(ndev, &mc->multicast.ib->rec.mgid,
+ true);
+ if (!err) {
+ mc->igmp_joined = true;
+ mc->multicast.ib->rec.hop_limit = IPV6_DEFAULT_HOPLIMIT;
+ }
+ } else {
+ mc->multicast.ib->rec.gid_type = IB_GID_TYPE_IB;
+ }
dev_put(ndev);
- if (!mc->multicast.ib->rec.mtu) {
+ if (err || !mc->multicast.ib->rec.mtu) {
err = -EINVAL;
goto out2;
}
- rdma_ip2gid((struct sockaddr *)&id_priv->id.route.addr.src_addr,
- &mc->multicast.ib->rec.port_gid);
+
work->id = id_priv;
work->mc = mc;
INIT_WORK(&work->work, iboe_mcast_work_handler);
@@ -3429,7 +3482,7 @@ int rdma_join_multicast(struct rdma_cm_id *id, struct
sockaddr *addr,
memcpy(&mc->addr, addr, rdma_addr_size(addr));
mc->context = context;
mc->id_priv = id_priv;
-
+ mc->igmp_joined = false;
spin_lock(&id_priv->lock);
list_add(&mc->list, &id_priv->mc_list);
spin_unlock(&id_priv->lock);
@@ -3486,6 +3539,21 @@ void rdma_leave_multicast(struct rdma_cm_id *id, struct
sockaddr *addr)
kfree(mc);
break;
case IB_LINK_LAYER_ETHERNET:
+ if (mc->igmp_joined) {
+ struct rdma_dev_addr *dev_addr
= &id->route.addr.dev_addr;
+ struct net_device *ndev = NULL;
+
+ if (dev_addr->bound_dev_if)
+ ndev =
dev_get_by_index(&init_net,
+
dev_addr->bound_dev_if);
+ if (ndev) {
+ cma_igmp_send(ndev,
+
&mc->multicast.ib->rec.mgid,
+ false);
+ dev_put(ndev);
+ }
+ mc->igmp_joined = false;
+ }
kref_put(&mc->mcref, release_mc);
break;
default:
diff --git a/drivers/infiniband/core/multicast.c
b/drivers/infiniband/core/multicast.c
index f1927f1..9cbee6c 100644
--- a/drivers/infiniband/core/multicast.c
+++ b/drivers/infiniband/core/multicast.c
@@ -729,8 +729,22 @@ int ib_init_ah_from_mcmember(struct ib_device *device, u8
port_num,
u16 gid_index;
u8 p;
- ret = ib_find_cached_gid(device, &rec->port_gid, IB_GID_TYPE_IB,
- NULL, 0, &p, &gid_index);
+ switch (rdma_port_get_link_layer(device, port_num)) {
+ case IB_LINK_LAYER_ETHERNET:
+ ret = ib_find_cached_gid_by_port(device, &rec->port_gid,
+ rec->gid_type, port_num,
+ rec->net, rec->ifindex,
+ &gid_index);
+ break;
+ case IB_LINK_LAYER_INFINIBAND:
+ ret = ib_find_cached_gid(device, &rec->port_gid,
+ IB_GID_TYPE_IB, NULL, 0, &p,
+ &gid_index);
+ break;
+ default:
+ ret = -EINVAL;
+ }
+
if (ret)
return ret;
diff --git a/include/rdma/ib_sa.h b/include/rdma/ib_sa.h
index 61bc231..653d538 100644
--- a/include/rdma/ib_sa.h
+++ b/include/rdma/ib_sa.h
@@ -198,6 +198,9 @@ struct ib_sa_mcmember_rec {
u8 scope;
u8 join_state;
int proxy_join;
+ int ifindex;
+ struct net *net;
+ enum ib_gid_type gid_type;
};
/* Service Record Component Mask Sec 15.2.5.14 Ver 1.1 */
--
2.1.0
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
