From: Saeed Mahameed <saeed@kernel.org>
To: "David S. Miller" <davem@davemloft.net>,
Jakub Kicinski <kuba@kernel.org>, Paolo Abeni <pabeni@redhat.com>,
Eric Dumazet <edumazet@google.com>
Cc: Saeed Mahameed <saeedm@nvidia.com>,
netdev@vger.kernel.org, Tariq Toukan <tariqt@nvidia.com>,
Mark Bloch <mbloch@nvidia.com>, Shay Drory <shayd@nvidia.com>,
Roi Dayan <roid@nvidia.com>
Subject: [net-next V2 03/14] net/mlx5e: rep, store send to vport rules per peer
Date: Fri, 2 Jun 2023 12:12:50 -0700 [thread overview]
Message-ID: <20230602191301.47004-4-saeed@kernel.org> (raw)
In-Reply-To: <20230602191301.47004-1-saeed@kernel.org>
From: Mark Bloch <mbloch@nvidia.com>
Each representor, for each send queue, is holding a
send_to_vport rule for the peer eswitch.
In order to support more than one peer, and to map between the peer
rules and peer eswitches, refactor representor to hold both the peer
rules and pointer to the peer eswitches.
This enables mlx5 to store send_to_vport rules per peer, where each
peer have dedicate index via mlx5_get_dev_index().
Signed-off-by: Mark Bloch <mbloch@nvidia.com>
Signed-off-by: Shay Drory <shayd@nvidia.com>
Reviewed-by: Roi Dayan <roid@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
.../net/ethernet/mellanox/mlx5/core/en_rep.c | 99 +++++++++++++++----
.../net/ethernet/mellanox/mlx5/core/en_rep.h | 7 +-
.../mellanox/mlx5/core/eswitch_offloads.c | 18 ++--
3 files changed, 98 insertions(+), 26 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c b/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c
index 3e7041bd5705..3fbb454f7228 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c
@@ -374,7 +374,9 @@ static void mlx5e_sqs2vport_stop(struct mlx5_eswitch *esw,
struct mlx5_eswitch_rep *rep)
{
struct mlx5e_rep_sq *rep_sq, *tmp;
+ struct mlx5e_rep_sq_peer *sq_peer;
struct mlx5e_rep_priv *rpriv;
+ unsigned long i;
if (esw->mode != MLX5_ESWITCH_OFFLOADS)
return;
@@ -382,8 +384,15 @@ static void mlx5e_sqs2vport_stop(struct mlx5_eswitch *esw,
rpriv = mlx5e_rep_to_rep_priv(rep);
list_for_each_entry_safe(rep_sq, tmp, &rpriv->vport_sqs_list, list) {
mlx5_eswitch_del_send_to_vport_rule(rep_sq->send_to_vport_rule);
- if (rep_sq->send_to_vport_rule_peer)
- mlx5_eswitch_del_send_to_vport_rule(rep_sq->send_to_vport_rule_peer);
+ xa_for_each(&rep_sq->sq_peer, i, sq_peer) {
+ if (sq_peer->rule)
+ mlx5_eswitch_del_send_to_vport_rule(sq_peer->rule);
+
+ xa_erase(&rep_sq->sq_peer, i);
+ kfree(sq_peer);
+ }
+
+ xa_destroy(&rep_sq->sq_peer);
list_del(&rep_sq->list);
kfree(rep_sq);
}
@@ -395,6 +404,7 @@ static int mlx5e_sqs2vport_start(struct mlx5_eswitch *esw,
{
struct mlx5_eswitch *peer_esw = NULL;
struct mlx5_flow_handle *flow_rule;
+ struct mlx5e_rep_sq_peer *sq_peer;
struct mlx5e_rep_priv *rpriv;
struct mlx5e_rep_sq *rep_sq;
int err;
@@ -414,6 +424,7 @@ static int mlx5e_sqs2vport_start(struct mlx5_eswitch *esw,
err = -ENOMEM;
goto out_err;
}
+ xa_init(&rep_sq->sq_peer);
/* Add re-inject rule to the PF/representor sqs */
flow_rule = mlx5_eswitch_add_send_to_vport_rule(esw, esw, rep,
@@ -427,15 +438,26 @@ static int mlx5e_sqs2vport_start(struct mlx5_eswitch *esw,
rep_sq->sqn = sqns_array[i];
if (peer_esw) {
+ int peer_rule_idx = mlx5_get_dev_index(peer_esw->dev);
+
+ sq_peer = kzalloc(sizeof(*sq_peer), GFP_KERNEL);
+ if (!sq_peer) {
+ err = -ENOMEM;
+ goto out_sq_peer_err;
+ }
+
flow_rule = mlx5_eswitch_add_send_to_vport_rule(peer_esw, esw,
rep, sqns_array[i]);
if (IS_ERR(flow_rule)) {
err = PTR_ERR(flow_rule);
- mlx5_eswitch_del_send_to_vport_rule(rep_sq->send_to_vport_rule);
- kfree(rep_sq);
- goto out_err;
+ goto out_flow_rule_err;
}
- rep_sq->send_to_vport_rule_peer = flow_rule;
+
+ sq_peer->rule = flow_rule;
+ sq_peer->peer = peer_esw;
+ err = xa_insert(&rep_sq->sq_peer, peer_rule_idx, sq_peer, GFP_KERNEL);
+ if (err)
+ goto out_xa_err;
}
list_add(&rep_sq->list, &rpriv->vport_sqs_list);
@@ -446,6 +468,14 @@ static int mlx5e_sqs2vport_start(struct mlx5_eswitch *esw,
return 0;
+out_xa_err:
+ mlx5_eswitch_del_send_to_vport_rule(flow_rule);
+out_flow_rule_err:
+ kfree(sq_peer);
+out_sq_peer_err:
+ mlx5_eswitch_del_send_to_vport_rule(rep_sq->send_to_vport_rule);
+ xa_destroy(&rep_sq->sq_peer);
+ kfree(rep_sq);
out_err:
mlx5e_sqs2vport_stop(esw, rep);
@@ -1530,17 +1560,24 @@ static void *mlx5e_vport_rep_get_proto_dev(struct mlx5_eswitch_rep *rep)
return rpriv->netdev;
}
-static void mlx5e_vport_rep_event_unpair(struct mlx5_eswitch_rep *rep)
+static void mlx5e_vport_rep_event_unpair(struct mlx5_eswitch_rep *rep,
+ struct mlx5_eswitch *peer_esw)
{
+ int i = mlx5_get_dev_index(peer_esw->dev);
struct mlx5e_rep_priv *rpriv;
struct mlx5e_rep_sq *rep_sq;
+ WARN_ON_ONCE(!peer_esw);
rpriv = mlx5e_rep_to_rep_priv(rep);
list_for_each_entry(rep_sq, &rpriv->vport_sqs_list, list) {
- if (!rep_sq->send_to_vport_rule_peer)
+ struct mlx5e_rep_sq_peer *sq_peer = xa_load(&rep_sq->sq_peer, i);
+
+ if (!sq_peer || sq_peer->peer != peer_esw)
continue;
- mlx5_eswitch_del_send_to_vport_rule(rep_sq->send_to_vport_rule_peer);
- rep_sq->send_to_vport_rule_peer = NULL;
+
+ mlx5_eswitch_del_send_to_vport_rule(sq_peer->rule);
+ xa_erase(&rep_sq->sq_peer, i);
+ kfree(sq_peer);
}
}
@@ -1548,24 +1585,52 @@ static int mlx5e_vport_rep_event_pair(struct mlx5_eswitch *esw,
struct mlx5_eswitch_rep *rep,
struct mlx5_eswitch *peer_esw)
{
+ int i = mlx5_get_dev_index(peer_esw->dev);
struct mlx5_flow_handle *flow_rule;
+ struct mlx5e_rep_sq_peer *sq_peer;
struct mlx5e_rep_priv *rpriv;
struct mlx5e_rep_sq *rep_sq;
+ int err;
rpriv = mlx5e_rep_to_rep_priv(rep);
list_for_each_entry(rep_sq, &rpriv->vport_sqs_list, list) {
- if (rep_sq->send_to_vport_rule_peer)
+ sq_peer = xa_load(&rep_sq->sq_peer, i);
+
+ if (sq_peer && sq_peer->peer)
continue;
- flow_rule = mlx5_eswitch_add_send_to_vport_rule(peer_esw, esw, rep, rep_sq->sqn);
- if (IS_ERR(flow_rule))
+
+ flow_rule = mlx5_eswitch_add_send_to_vport_rule(peer_esw, esw, rep,
+ rep_sq->sqn);
+ if (IS_ERR(flow_rule)) {
+ err = PTR_ERR(flow_rule);
goto err_out;
- rep_sq->send_to_vport_rule_peer = flow_rule;
+ }
+
+ if (sq_peer) {
+ sq_peer->rule = flow_rule;
+ sq_peer->peer = peer_esw;
+ continue;
+ }
+ sq_peer = kzalloc(sizeof(*sq_peer), GFP_KERNEL);
+ if (!sq_peer) {
+ err = -ENOMEM;
+ goto err_sq_alloc;
+ }
+ err = xa_insert(&rep_sq->sq_peer, i, sq_peer, GFP_KERNEL);
+ if (err)
+ goto err_xa;
+ sq_peer->rule = flow_rule;
+ sq_peer->peer = peer_esw;
}
return 0;
+err_xa:
+ kfree(sq_peer);
+err_sq_alloc:
+ mlx5_eswitch_del_send_to_vport_rule(flow_rule);
err_out:
- mlx5e_vport_rep_event_unpair(rep);
- return PTR_ERR(flow_rule);
+ mlx5e_vport_rep_event_unpair(rep, peer_esw);
+ return err;
}
static int mlx5e_vport_rep_event(struct mlx5_eswitch *esw,
@@ -1578,7 +1643,7 @@ static int mlx5e_vport_rep_event(struct mlx5_eswitch *esw,
if (event == MLX5_SWITCHDEV_EVENT_PAIR)
err = mlx5e_vport_rep_event_pair(esw, rep, data);
else if (event == MLX5_SWITCHDEV_EVENT_UNPAIR)
- mlx5e_vport_rep_event_unpair(rep);
+ mlx5e_vport_rep_event_unpair(rep, data);
return err;
}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_rep.h b/drivers/net/ethernet/mellanox/mlx5/core/en_rep.h
index 80b7f5079a5a..70640fa1ad7b 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_rep.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_rep.h
@@ -225,9 +225,14 @@ struct mlx5e_encap_entry {
struct rcu_head rcu;
};
+struct mlx5e_rep_sq_peer {
+ struct mlx5_flow_handle *rule;
+ void *peer;
+};
+
struct mlx5e_rep_sq {
struct mlx5_flow_handle *send_to_vport_rule;
- struct mlx5_flow_handle *send_to_vport_rule_peer;
+ struct xarray sq_peer;
u32 sqn;
struct list_head list;
};
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
index 1b2f5e273525..9526382f1573 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
@@ -2673,7 +2673,8 @@ void mlx5_eswitch_offloads_destroy_single_fdb(struct mlx5_eswitch *master_esw,
#define ESW_OFFLOADS_DEVCOM_PAIR (0)
#define ESW_OFFLOADS_DEVCOM_UNPAIR (1)
-static void mlx5_esw_offloads_rep_event_unpair(struct mlx5_eswitch *esw)
+static void mlx5_esw_offloads_rep_event_unpair(struct mlx5_eswitch *esw,
+ struct mlx5_eswitch *peer_esw)
{
const struct mlx5_eswitch_rep_ops *ops;
struct mlx5_eswitch_rep *rep;
@@ -2686,17 +2687,18 @@ static void mlx5_esw_offloads_rep_event_unpair(struct mlx5_eswitch *esw)
ops = esw->offloads.rep_ops[rep_type];
if (atomic_read(&rep->rep_data[rep_type].state) == REP_LOADED &&
ops->event)
- ops->event(esw, rep, MLX5_SWITCHDEV_EVENT_UNPAIR, NULL);
+ ops->event(esw, rep, MLX5_SWITCHDEV_EVENT_UNPAIR, peer_esw);
}
}
}
-static void mlx5_esw_offloads_unpair(struct mlx5_eswitch *esw)
+static void mlx5_esw_offloads_unpair(struct mlx5_eswitch *esw,
+ struct mlx5_eswitch *peer_esw)
{
#if IS_ENABLED(CONFIG_MLX5_CLS_ACT)
mlx5e_tc_clean_fdb_peer_flows(esw);
#endif
- mlx5_esw_offloads_rep_event_unpair(esw);
+ mlx5_esw_offloads_rep_event_unpair(esw, peer_esw);
esw_del_fdb_peer_miss_rules(esw);
}
@@ -2728,7 +2730,7 @@ static int mlx5_esw_offloads_pair(struct mlx5_eswitch *esw,
return 0;
err_out:
- mlx5_esw_offloads_unpair(esw);
+ mlx5_esw_offloads_unpair(esw, peer_esw);
return err;
}
@@ -2802,8 +2804,8 @@ static int mlx5_esw_offloads_devcom_event(int event,
mlx5_devcom_set_paired(devcom, MLX5_DEVCOM_ESW_OFFLOADS, false);
esw->paired[mlx5_get_dev_index(peer_esw->dev)] = false;
peer_esw->paired[mlx5_get_dev_index(esw->dev)] = false;
- mlx5_esw_offloads_unpair(peer_esw);
- mlx5_esw_offloads_unpair(esw);
+ mlx5_esw_offloads_unpair(peer_esw, esw);
+ mlx5_esw_offloads_unpair(esw, peer_esw);
mlx5_esw_offloads_set_ns_peer(esw, peer_esw, false);
break;
}
@@ -2811,7 +2813,7 @@ static int mlx5_esw_offloads_devcom_event(int event,
return 0;
err_pair:
- mlx5_esw_offloads_unpair(esw);
+ mlx5_esw_offloads_unpair(esw, peer_esw);
err_peer:
mlx5_esw_offloads_set_ns_peer(esw, peer_esw, false);
err_out:
--
2.40.1
next prev parent reply other threads:[~2023-06-02 19:13 UTC|newest]
Thread overview: 17+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-06-02 19:12 [pull request][net-next V2 00/14] mlx5 updates 2023-05-31 Saeed Mahameed
2023-06-02 19:12 ` [net-next V2 01/14] net/mlx5e: en_tc, Extend peer flows to a list Saeed Mahameed
2023-06-05 23:00 ` patchwork-bot+netdevbpf
2023-06-02 19:12 ` [net-next V2 02/14] net/mlx5e: tc, Refactor peer add/del flow Saeed Mahameed
2023-06-02 19:12 ` Saeed Mahameed [this message]
2023-06-02 19:12 ` [net-next V2 04/14] net/mlx5e: en_tc, re-factor query route port Saeed Mahameed
2023-06-02 19:12 ` [net-next V2 05/14] net/mlx5e: Handle offloads flows per peer Saeed Mahameed
2023-06-02 19:12 ` [net-next V2 06/14] net/mlx5: E-switch, enlarge peer miss group table Saeed Mahameed
2023-06-02 19:12 ` [net-next V2 07/14] net/mlx5: E-switch, refactor FDB miss rule add/remove Saeed Mahameed
2023-06-02 19:12 ` [net-next V2 08/14] net/mlx5: E-switch, Handle multiple master egress rules Saeed Mahameed
2023-06-02 19:12 ` [net-next V2 09/14] net/mlx5: E-switch, generalize shared FDB creation Saeed Mahameed
2023-06-02 19:12 ` [net-next V2 10/14] net/mlx5: DR, handle more than one peer domain Saeed Mahameed
2023-06-02 19:12 ` [net-next V2 11/14] net/mlx5: Devcom, Rename paired to ready Saeed Mahameed
2023-06-02 19:12 ` [net-next V2 12/14] net/mlx5: E-switch, mark devcom as not ready when all eswitches are unpaired Saeed Mahameed
2023-06-02 19:13 ` [net-next V2 13/14] net/mlx5: Devcom, introduce devcom_for_each_peer_entry Saeed Mahameed
2023-06-02 19:13 ` [net-next V2 14/14] net/mlx5: Devcom, extend mlx5_devcom_send_event to work with more than two devices Saeed Mahameed
2023-06-03 7:26 ` [pull request][net-next V2 00/14] mlx5 updates 2023-05-31 Simon Horman
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=20230602191301.47004-4-saeed@kernel.org \
--to=saeed@kernel.org \
--cc=davem@davemloft.net \
--cc=edumazet@google.com \
--cc=kuba@kernel.org \
--cc=mbloch@nvidia.com \
--cc=netdev@vger.kernel.org \
--cc=pabeni@redhat.com \
--cc=roid@nvidia.com \
--cc=saeedm@nvidia.com \
--cc=shayd@nvidia.com \
--cc=tariqt@nvidia.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.