On 7/16/2018 9:47 AM, Parav Pandit wrote:
-----Original Message-----
From: Santosh Shilimkar [mailto:santosh.shilimkar@xxxxxxxxxx]
Sent: Monday, July 16, 2018 11:38 AM
[...]
diff --git a/include/rdma/ib_cm.h b/include/rdma/ib_cm.h index
c98d603c0b63..568708a87239 100644
--- a/include/rdma/ib_cm.h
+++ b/include/rdma/ib_cm.h
@@ -120,6 +120,13 @@ struct ib_cm_req_event_param {
struct sa_path_rec *primary_path;
struct sa_path_rec *alternate_path;
+ /*
+ * SGID attribute of the primary path. Currently only
+ * useful for RoCE. Alternate path GID attributes
+ * are not yet supported.
+ */
+ const struct ib_gid_attr *ppath_sgid_attr;
+
__be64 remote_ca_guid;
u32 remote_qkey;
u32 remote_qpn;
@@ -226,6 +233,12 @@ struct ib_cm_apr_event_param {
struct ib_cm_sidr_req_event_param {
struct ib_cm_id *listen_id;
__be64 service_id;
+
+ /*
+ * SGID attribute of the request. Currently only
+ * useful for RoCE.
+ */
+ const struct ib_gid_attr *sgid_attr;
/* P_Key that was used by the GMP's BTH header */
u16 bth_pkey;
u8 port;
How does this interop with older CM request packets which doesn't have
ppath_sgid_attr ? Am I missing something ?
This is upstream kernel on top of previous patches. So CM requests will have ppath_sgid_attr.
OK. The question is still generic though. What happens to the CM request
sent from endpoints which doesn't have the patched struct and hence no
gid_attr.
Regards,
Santosh
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
