| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <20251218-vf-bw-lag-mode-v1-10-7d8ed4368bea@nvidia.com>
Date: Thu, 18 Dec 2025 17:59:00 +0200
From: Edward Srouji <edwards@...dia.com>
To: <edwards@...dia.com>, Leon Romanovsky <leon@...nel.org>, Saeed Mahameed
<saeedm@...dia.com>, Tariq Toukan <tariqt@...dia.com>, Mark Bloch
<mbloch@...dia.com>, Andrew Lunn <andrew+netdev@...n.ch>, "David S. Miller"
<davem@...emloft.net>, Eric Dumazet <edumazet@...gle.com>, Jakub Kicinski
<kuba@...nel.org>, Paolo Abeni <pabeni@...hat.com>, Jason Gunthorpe
<jgg@...pe.ca>
CC: <netdev@...r.kernel.org>, <linux-rdma@...r.kernel.org>,
<linux-kernel@...r.kernel.org>, Or Har-Toov <ohartoov@...dia.com>
Subject: [PATCH rdma-next 10/10] RDMA/mlx5: Implement query_port_speed callback
From: Or Har-Toov <ohartoov@...dia.com>
Implement the query_port_speed callback for mlx5 driver to support
querying effective port bandwidth.
For LAG configurations, query the aggregated speed from the LAG layer
or from the modified vport max_tx_speed.
Signed-off-by: Or Har-Toov <ohartoov@...dia.com>
Reviewed-by: Mark Bloch <mbloch@...dia.com>
Signed-off-by: Edward Srouji <edwards@...dia.com>
---
drivers/infiniband/hw/mlx5/main.c | 124 +++++++++++++++++++++++++++++++++++
drivers/infiniband/hw/mlx5/mlx5_ib.h | 2 +
2 files changed, 126 insertions(+)
diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index bea42acbeaad..47c19d527fa2 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -1581,6 +1581,129 @@ static int mlx5_ib_rep_query_pkey(struct ib_device *ibdev, u32 port, u16 index,
return 0;
}
+static int mlx5_ib_query_port_speed_from_port(struct mlx5_ib_dev *dev,
+ u32 port_num, u64 *speed)
+{
+ struct ib_port_speed_info speed_info;
+ struct ib_port_attr attr = {};
+ int err;
+
+ err = mlx5_ib_query_port(&dev->ib_dev, port_num, &attr);
+ if (err)
+ return err;
+
+ if (attr.state == IB_PORT_DOWN) {
+ *speed = 0;
+ return 0;
+ }
+
+ err = ib_port_attr_to_speed_info(&attr, &speed_info);
+ if (err)
+ return err;
+
+ *speed = speed_info.rate;
+ return 0;
+}
+
+static int mlx5_ib_query_port_speed_from_vport(struct mlx5_core_dev *mdev,
+ u8 op_mod, u16 vport,
+ u8 other_vport, u64 *speed,
+ struct mlx5_ib_dev *dev,
+ u32 port_num)
+{
+ u32 max_tx_speed;
+ int err;
+
+ err = mlx5_query_vport_max_tx_speed(mdev, op_mod, vport, other_vport,
+ &max_tx_speed);
+ if (err)
+ return err;
+
+ if (max_tx_speed == 0)
+ /* Value 0 indicates field not supported, fallback */
+ return mlx5_ib_query_port_speed_from_port(dev, port_num,
+ speed);
+
+ *speed = max_tx_speed;
+ return 0;
+}
+
+static int mlx5_ib_query_port_speed_from_bond(struct mlx5_ib_dev *dev,
+ u32 port_num, u64 *speed)
+{
+ struct mlx5_core_dev *mdev = dev->mdev;
+ u32 bond_speed;
+ int err;
+
+ err = mlx5_lag_query_bond_speed(mdev, &bond_speed);
+ if (err)
+ return err;
+
+ *speed = bond_speed / MLX5_MAX_TX_SPEED_UNIT;
+
+ return 0;
+}
+
+static int mlx5_ib_query_port_speed_non_rep(struct mlx5_ib_dev *dev,
+ u32 port_num, u64 *speed)
+{
+ u16 op_mod = MLX5_VPORT_STATE_OP_MOD_VNIC_VPORT;
+
+ if (mlx5_lag_is_roce(dev->mdev))
+ return mlx5_ib_query_port_speed_from_bond(dev, port_num,
+ speed);
+
+ return mlx5_ib_query_port_speed_from_vport(dev->mdev, op_mod, 0, false,
+ speed, dev, port_num);
+}
+
+static int mlx5_ib_query_port_speed_rep(struct mlx5_ib_dev *dev, u32 port_num,
+ u64 *speed)
+{
+ struct mlx5_eswitch_rep *rep;
+ struct mlx5_core_dev *mdev;
+ u16 op_mod;
+
+ if (!dev->port[port_num - 1].rep) {
+ mlx5_ib_warn(dev, "Representor doesn't exist for port %u\n",
+ port_num);
+ return -EINVAL;
+ }
+
+ rep = dev->port[port_num - 1].rep;
+ mdev = mlx5_eswitch_get_core_dev(rep->esw);
+ if (!mdev)
+ return -ENODEV;
+
+ if (rep->vport == MLX5_VPORT_UPLINK) {
+ if (mlx5_lag_is_sriov(mdev))
+ return mlx5_ib_query_port_speed_from_bond(dev,
+ port_num,
+ speed);
+
+ return mlx5_ib_query_port_speed_from_port(dev, port_num,
+ speed);
+ }
+
+ op_mod = MLX5_VPORT_STATE_OP_MOD_ESW_VPORT;
+ return mlx5_ib_query_port_speed_from_vport(dev->mdev, op_mod,
+ rep->vport, true, speed, dev,
+ port_num);
+}
+
+int mlx5_ib_query_port_speed(struct ib_device *ibdev, u32 port_num, u64 *speed)
+{
+ struct mlx5_ib_dev *dev = to_mdev(ibdev);
+
+ if (mlx5_ib_port_link_layer(ibdev, port_num) ==
+ IB_LINK_LAYER_INFINIBAND || mlx5_core_mp_enabled(dev->mdev))
+ return mlx5_ib_query_port_speed_from_port(dev, port_num, speed);
+ else if (!dev->is_rep)
+ return mlx5_ib_query_port_speed_non_rep(dev, port_num, speed);
+ else
+ return mlx5_ib_query_port_speed_rep(dev, port_num, speed);
+}
+
static int mlx5_ib_query_gid(struct ib_device *ibdev, u32 port, int index,
union ib_gid *gid)
{
@@ -4305,6 +4428,7 @@ static const struct ib_device_ops mlx5_ib_dev_ops = {
.query_device = mlx5_ib_query_device,
.query_gid = mlx5_ib_query_gid,
.query_pkey = mlx5_ib_query_pkey,
+ .query_port_speed = mlx5_ib_query_port_speed,
.query_qp = mlx5_ib_query_qp,
.query_srq = mlx5_ib_query_srq,
.query_ucontext = mlx5_ib_query_ucontext,
diff --git a/drivers/infiniband/hw/mlx5/mlx5_ib.h b/drivers/infiniband/hw/mlx5/mlx5_ib.h
index 09d82d5f95e3..cc6b3b6c713c 100644
--- a/drivers/infiniband/hw/mlx5/mlx5_ib.h
+++ b/drivers/infiniband/hw/mlx5/mlx5_ib.h
@@ -1435,6 +1435,8 @@ int mlx5_query_mad_ifc_port(struct ib_device *ibdev, u32 port,
struct ib_port_attr *props);
int mlx5_ib_query_port(struct ib_device *ibdev, u32 port,
struct ib_port_attr *props);
+int mlx5_ib_query_port_speed(struct ib_device *ibdev, u32 port_num,
+ u64 *speed);
void mlx5_ib_populate_pas(struct ib_umem *umem, size_t page_size, __be64 *pas,
u64 access_flags);
int mlx5_ib_get_cqe_size(struct ib_cq *ibcq);
--
2.47.1
Powered by blists - more mailing lists