IB/mlx4: Fix incorrect MC join state bit-masking on SR-IOV
authorAlex Vesker <valex@mellanox.com>
Mon, 12 Sep 2016 16:16:18 +0000 (19:16 +0300)
committerDoug Ledford <dledford@redhat.com>
Fri, 16 Sep 2016 18:14:08 +0000 (14:14 -0400)
Because of an incorrect bit-masking done on the join state bits, when
handling a join request we failed to detect a difference between the
group join state and the request join state when joining as send only
full member (0x8). This caused the MC join request not to be sent.
This issue is relevant only when SRIOV is enabled and SM supports
send only full member.

This fix separates scope bits and join states bits a nibble each.

Fixes: b9c5d6a64358 ('IB/mlx4: Add multicast group (MCG) paravirtualization for SR-IOV')
Signed-off-by: Alex Vesker <valex@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Doug Ledford <dledford@redhat.com>
drivers/infiniband/hw/mlx4/mcg.c

index 8f7ad07915b07e27ed7f512eabc07757269332c8..097bfcc4ee997eaab178f8d8353efe79fa00d0a7 100644 (file)
@@ -489,7 +489,7 @@ static u8 get_leave_state(struct mcast_group *group)
                if (!group->members[i])
                        leave_state |= (1 << i);
 
-       return leave_state & (group->rec.scope_join_state & 7);
+       return leave_state & (group->rec.scope_join_state & 0xf);
 }
 
 static int join_group(struct mcast_group *group, int slave, u8 join_mask)
@@ -564,8 +564,8 @@ static void mlx4_ib_mcg_timeout_handler(struct work_struct *work)
                } else
                        mcg_warn_group(group, "DRIVER BUG\n");
        } else if (group->state == MCAST_LEAVE_SENT) {
-               if (group->rec.scope_join_state & 7)
-                       group->rec.scope_join_state &= 0xf8;
+               if (group->rec.scope_join_state & 0xf)
+                       group->rec.scope_join_state &= 0xf0;
                group->state = MCAST_IDLE;
                mutex_unlock(&group->lock);
                if (release_group(group, 1))
@@ -605,7 +605,7 @@ static int handle_leave_req(struct mcast_group *group, u8 leave_mask,
 static int handle_join_req(struct mcast_group *group, u8 join_mask,
                           struct mcast_req *req)
 {
-       u8 group_join_state = group->rec.scope_join_state & 7;
+       u8 group_join_state = group->rec.scope_join_state & 0xf;
        int ref = 0;
        u16 status;
        struct ib_sa_mcmember_data *sa_data = (struct ib_sa_mcmember_data *)req->sa_mad.data;
@@ -690,8 +690,8 @@ static void mlx4_ib_mcg_work_handler(struct work_struct *work)
                        u8 cur_join_state;
 
                        resp_join_state = ((struct ib_sa_mcmember_data *)
-                                               group->response_sa_mad.data)->scope_join_state & 7;
-                       cur_join_state = group->rec.scope_join_state & 7;
+                                               group->response_sa_mad.data)->scope_join_state & 0xf;
+                       cur_join_state = group->rec.scope_join_state & 0xf;
 
                        if (method == IB_MGMT_METHOD_GET_RESP) {
                                /* successfull join */
@@ -710,7 +710,7 @@ process_requests:
                req = list_first_entry(&group->pending_list, struct mcast_req,
                                       group_list);
                sa_data = (struct ib_sa_mcmember_data *)req->sa_mad.data;
-               req_join_state = sa_data->scope_join_state & 0x7;
+               req_join_state = sa_data->scope_join_state & 0xf;
 
                /* For a leave request, we will immediately answer the VF, and
                 * update our internal counters. The actual leave will be sent