From: Maor Gottlieb <maorg@mellanox.com>
To: davem@davemloft.net, jgg@mellanox.com, dledford@redhat.com,
j.vosburgh@gmail.com, vfalico@gmail.com, andy@greyhouse.net,
kuba@kernel.org
Cc: leonro@mellanox.com, saeedm@mellanox.com, jiri@mellanox.com,
linux-rdma@vger.kernel.org, netdev@vger.kernel.org,
alexr@mellanox.com, Maor Gottlieb <maorg@mellanox.com>
Subject: [PATCH V2 mlx5-next 06/10] RDMA/core: Get xmit slave for LAG
Date: Mon, 20 Apr 2020 10:54:22 +0300 [thread overview]
Message-ID: <20200420075426.31462-7-maorg@mellanox.com> (raw)
In-Reply-To: <20200420075426.31462-1-maorg@mellanox.com>
Add a call to rdma_lag_get_ah_roce_slave when
Address handle is created.
Low driver can use it to select the QP's affinity port.
Signed-off-by: Maor Gottlieb <maorg@mellanox.com>
---
drivers/infiniband/core/verbs.c | 44 ++++++++++++++++++++++-----------
1 file changed, 30 insertions(+), 14 deletions(-)
diff --git a/drivers/infiniband/core/verbs.c b/drivers/infiniband/core/verbs.c
index 56a71337112c..a0d60376ba6b 100644
--- a/drivers/infiniband/core/verbs.c
+++ b/drivers/infiniband/core/verbs.c
@@ -50,6 +50,7 @@
#include <rdma/ib_cache.h>
#include <rdma/ib_addr.h>
#include <rdma/rw.h>
+#include <rdma/lag.h>
#include "core_priv.h"
#include <trace/events/rdma_core.h>
@@ -554,8 +555,14 @@ struct ib_ah *rdma_create_ah(struct ib_pd *pd, struct rdma_ah_attr *ah_attr,
if (ret)
return ERR_PTR(ret);
- ah = _rdma_create_ah(pd, ah_attr, flags, NULL);
+ ret = rdma_lag_get_ah_roce_slave(pd->device, ah_attr);
+ if (ret) {
+ rdma_unfill_sgid_attr(ah_attr, old_sgid_attr);
+ return ERR_PTR(ret);
+ }
+ ah = _rdma_create_ah(pd, ah_attr, flags, NULL);
+ rdma_lag_put_ah_roce_slave(ah_attr);
rdma_unfill_sgid_attr(ah_attr, old_sgid_attr);
return ah;
}
@@ -1638,6 +1645,25 @@ static int _ib_modify_qp(struct ib_qp *qp, struct ib_qp_attr *attr,
&old_sgid_attr_av);
if (ret)
return ret;
+
+ if (attr->ah_attr.type == RDMA_AH_ATTR_TYPE_ROCE &&
+ is_qp_type_connected(qp)) {
+ /*
+ * If the user provided the qp_attr then we have to
+ * resolve it. Kerne users have to provide already
+ * resolved rdma_ah_attr's.
+ */
+ if (udata) {
+ ret = ib_resolve_eth_dmac(qp->device,
+ &attr->ah_attr);
+ if (ret)
+ goto out_av;
+ }
+ ret = rdma_lag_get_ah_roce_slave(qp->device,
+ &attr->ah_attr);
+ if (ret)
+ goto out_av;
+ }
}
if (attr_mask & IB_QP_ALT_PATH) {
/*
@@ -1664,18 +1690,6 @@ static int _ib_modify_qp(struct ib_qp *qp, struct ib_qp_attr *attr,
}
}
- /*
- * If the user provided the qp_attr then we have to resolve it. Kernel
- * users have to provide already resolved rdma_ah_attr's
- */
- if (udata && (attr_mask & IB_QP_AV) &&
- attr->ah_attr.type == RDMA_AH_ATTR_TYPE_ROCE &&
- is_qp_type_connected(qp)) {
- ret = ib_resolve_eth_dmac(qp->device, &attr->ah_attr);
- if (ret)
- goto out;
- }
-
if (rdma_ib_or_roce(qp->device, port)) {
if (attr_mask & IB_QP_RQ_PSN && attr->rq_psn & ~0xffffff) {
dev_warn(&qp->device->dev,
@@ -1717,8 +1731,10 @@ static int _ib_modify_qp(struct ib_qp *qp, struct ib_qp_attr *attr,
if (attr_mask & IB_QP_ALT_PATH)
rdma_unfill_sgid_attr(&attr->alt_ah_attr, old_sgid_attr_alt_av);
out_av:
- if (attr_mask & IB_QP_AV)
+ if (attr_mask & IB_QP_AV) {
+ rdma_lag_put_ah_roce_slave(&attr->ah_attr);
rdma_unfill_sgid_attr(&attr->ah_attr, old_sgid_attr_av);
+ }
return ret;
}
--
2.17.2
next prev parent reply other threads:[~2020-04-20 7:54 UTC|newest]
Thread overview: 31+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-04-20 7:54 [PATCH V2 mlx5-next 00/10] Add support to get xmit slave Maor Gottlieb
2020-04-20 7:54 ` [PATCH V2 mlx5-next 01/10] net/core: Introduce master_xmit_slave_get Maor Gottlieb
2020-04-20 14:01 ` Jiri Pirko
2020-04-20 17:29 ` David Ahern
2020-04-20 17:41 ` Jiri Pirko
2020-04-20 17:43 ` David Ahern
2020-04-20 17:53 ` Jiri Pirko
2020-04-20 17:54 ` Jiri Pirko
2020-04-20 17:56 ` David Ahern
2020-04-20 18:01 ` Jiri Pirko
2020-04-20 18:04 ` David Ahern
2020-04-20 18:48 ` Jiri Pirko
2020-04-20 18:56 ` Maor Gottlieb
2020-04-20 19:02 ` David Ahern
2020-04-21 5:37 ` Jiri Pirko
2020-04-21 5:43 ` Maor Gottlieb
2020-04-20 7:54 ` [PATCH V2 mlx5-next 02/10] bonding: Rename slave_arr to usable_slaves Maor Gottlieb
2020-04-20 14:17 ` Jiri Pirko
2020-04-20 7:54 ` [PATCH V2 mlx5-next 03/10] bonding: Add helpers to get xmit slave Maor Gottlieb
2020-04-20 14:27 ` Jiri Pirko
2020-04-20 18:46 ` Jay Vosburgh
2020-04-20 18:59 ` Maor Gottlieb
2020-04-20 7:54 ` [PATCH V2 mlx5-next 04/10] bonding: Implement ndo_xmit_slave_get Maor Gottlieb
2020-04-20 15:04 ` Jiri Pirko
2020-04-20 15:36 ` David Ahern
2020-04-20 7:54 ` [PATCH V2 mlx5-next 05/10] RDMA/core: Add LAG functionality Maor Gottlieb
2020-04-20 7:54 ` Maor Gottlieb [this message]
2020-04-20 7:54 ` [PATCH V2 mlx5-next 07/10] net/mlx5: Change lag mutex lock to spin lock Maor Gottlieb
2020-04-20 7:54 ` [PATCH V2 mlx5-next 08/10] net/mlx5: Add support to get lag physical port Maor Gottlieb
2020-04-20 7:54 ` [PATCH V2 mlx5-next 09/10] RDMA/mlx5: Refactor affinity related code Maor Gottlieb
2020-04-20 7:54 ` [PATCH V2 mlx5-next 10/10] RDMA/mlx5: Set lag tx affinity according to slave Maor Gottlieb
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20200420075426.31462-7-maorg@mellanox.com \
--to=maorg@mellanox.com \
--cc=alexr@mellanox.com \
--cc=andy@greyhouse.net \
--cc=davem@davemloft.net \
--cc=dledford@redhat.com \
--cc=j.vosburgh@gmail.com \
--cc=jgg@mellanox.com \
--cc=jiri@mellanox.com \
--cc=kuba@kernel.org \
--cc=leonro@mellanox.com \
--cc=linux-rdma@vger.kernel.org \
--cc=netdev@vger.kernel.org \
--cc=saeedm@mellanox.com \
--cc=vfalico@gmail.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).