On 3/27/2020 12:49 PM, Jason Gunthorpe wrote:
On Mon, Mar 23, 2020 at 07:15:12PM -0400, Dennis Dalessandro wrote:
@@ -240,13 +241,11 @@ static int ipoib_mcast_join_finish(struct ipoib_mcast *mcast,
priv->broadcast->mcmember.flow_label = mcmember->flow_label;
priv->broadcast->mcmember.hop_limit = mcmember->hop_limit;
/* assume if the admin and the mcast are the same both can be changed */
+ mtu = rdma_mtu_enum_to_int(priv->ca, priv->port,
+ priv->broadcast->mcmember.mtu);
if (priv->mcast_mtu == priv->admin_mtu)
- priv->admin_mtu =
- priv->mcast_mtu =
- IPOIB_UD_MTU(ib_mtu_enum_to_int(priv->broadcast->mcmember.mtu));
- else
- priv->mcast_mtu =
- IPOIB_UD_MTU(ib_mtu_enum_to_int(priv->broadcast->mcmember.mtu));
+ priv->admin_mtu = IPOIB_UD_MTU(mtu);
+ priv->mcast_mtu = IPOIB_UD_MTU(mtu);
Er, how did this ever work? Does the OPA SM not use the 6 & 7 values
for the mtu in the path record? Why is it being changed now?
Prior to this patch series, we can only run AIP at a max mtu of 4K, even
on OPA devices. Therefore, we need a way to get the max physical mtu for
the underlying device.
+/**
+ * rdma_mtu_from_attr - Return the mtu of the port from the port attribute.
+ * @device: Device
+ * @port_num: Port number
+ * @attr: port attribute
+ *
+ * Return the MTU size supported by the port as an integer value.
+ */
+static inline int rdma_mtu_from_attr(struct ib_device *device, u8 port,
+ struct ib_port_attr *attr)
+{
+ if (rdma_core_cap_opa_port(device, port))
+ return attr->phys_mtu;
Why not just always set this?
Because this is a new field and other vendor devices does not set it at all.
Sorry for the delayed response this got lost in the shuffle.
-Denny
