Quoting r. Roland Dreier <[EMAIL PROTECTED]>:
> Subject: Re: [PATCH] IB/ipoib: use appropriate path selector
>
> > + path->pathrec.mtu = priv->broadcast->mcmember.mtu;
> > + path->pathrec.mtu_selector = IB_SA_GTE;
>
> Does this do anything without setting the component mask of the actual
> request??
>
Ugh. Correct of course. The SA I was testing against seems to have a bug.
The following is yet untested - any more obvious gotchas you can see?
diff --git a/drivers/infiniband/ulp/ipoib/ipoib_main.c
b/drivers/infiniband/ulp/ipoib/ipoib_main.c
index cf71d2a..c8e8dd3 100644
--- a/drivers/infiniband/ulp/ipoib/ipoib_main.c
+++ b/drivers/infiniband/ulp/ipoib/ipoib_main.c
@@ -441,9 +441,11 @@ static struct ipoib_path *path_rec_creat
INIT_LIST_HEAD(&path->neigh_list);
memcpy(path->pathrec.dgid.raw, gid, sizeof (union ib_gid));
- path->pathrec.sgid = priv->local_gid;
- path->pathrec.pkey = cpu_to_be16(priv->pkey);
- path->pathrec.numb_path = 1;
+ path->pathrec.sgid = priv->local_gid;
+ path->pathrec.pkey = cpu_to_be16(priv->pkey);
+ path->pathrec.numb_path = 1;
+ path->pathrec.mtu = priv->broadcast->mcmember.mtu;
+ path->pathrec.mtu_selector = IB_SA_GTE;
return path;
}
@@ -464,7 +466,8 @@ static int path_rec_start(struct net_dev
IB_SA_PATH_REC_DGID |
IB_SA_PATH_REC_SGID |
IB_SA_PATH_REC_NUMB_PATH |
- IB_SA_PATH_REC_PKEY,
+ IB_SA_PATH_REC_PKEY |
+ IB_SA_PATH_REC_MTU_SELECTOR,
1000, GFP_ATOMIC,
path_rec_completion,
path, &path->query);
--
MST
_______________________________________________
openib-general mailing list
[email protected]
http://openib.org/mailman/listinfo/openib-general
To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
