* [PATCH net-next 08/10] IB/mlx4: Add port aggregation support
From: Or Gerlitz @ 2015-02-03 14:48 UTC (permalink / raw)
To: David S. Miller
Cc: netdev, Roland Dreier, Amir Vadai, Tal Alon, Moni Shoua,
Or Gerlitz
In-Reply-To: <1422974919-28084-1-git-send-email-ogerlitz@mellanox.com>
From: Moni Shoua <monis@mellanox.com>
Register the interface with the mlx4 core driver with port aggregation support
and check for port aggregation mode when the 'add' function is called.
In this mode, only one physical port is reported to the upper layer
(RoCE/IB core stack and ULPs).
Signed-off-by: Moni Shoua <monis@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
---
drivers/infiniband/hw/mlx4/main.c | 76 ++++++++++++++++++++++++++++++++++---
1 files changed, 70 insertions(+), 6 deletions(-)
diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c
index 9db258f..ed21ae6 100644
--- a/drivers/infiniband/hw/mlx4/main.c
+++ b/drivers/infiniband/hw/mlx4/main.c
@@ -351,6 +351,7 @@ static int eth_link_query_port(struct ib_device *ibdev, u8 port,
enum ib_mtu tmp;
struct mlx4_cmd_mailbox *mailbox;
int err = 0;
+ int is_bonded = mlx4_is_bonded(mdev->dev);
mailbox = mlx4_alloc_cmd_mailbox(mdev->dev);
if (IS_ERR(mailbox))
@@ -374,8 +375,12 @@ static int eth_link_query_port(struct ib_device *ibdev, u8 port,
props->state = IB_PORT_DOWN;
props->phys_state = state_to_phys_state(props->state);
props->active_mtu = IB_MTU_256;
+ if (is_bonded)
+ rtnl_lock(); /* required to get upper dev */
spin_lock_bh(&iboe->lock);
ndev = iboe->netdevs[port - 1];
+ if (ndev && is_bonded)
+ ndev = netdev_master_upper_dev_get(ndev);
if (!ndev)
goto out_unlock;
@@ -387,6 +392,8 @@ static int eth_link_query_port(struct ib_device *ibdev, u8 port,
props->phys_state = state_to_phys_state(props->state);
out_unlock:
spin_unlock_bh(&iboe->lock);
+ if (is_bonded)
+ rtnl_unlock();
out:
mlx4_free_cmd_mailbox(mdev->dev, mailbox);
return err;
@@ -1440,6 +1447,7 @@ static void update_gids_task(struct work_struct *work)
union ib_gid *gids;
int err;
struct mlx4_dev *dev = gw->dev->dev;
+ int is_bonded = mlx4_is_bonded(dev);
if (!gw->dev->ib_active)
return;
@@ -1459,7 +1467,10 @@ static void update_gids_task(struct work_struct *work)
if (err)
pr_warn("set port command failed\n");
else
- mlx4_ib_dispatch_event(gw->dev, gw->port, IB_EVENT_GID_CHANGE);
+ if ((gw->port == 1) || !is_bonded)
+ mlx4_ib_dispatch_event(gw->dev,
+ is_bonded ? 1 : gw->port,
+ IB_EVENT_GID_CHANGE);
mlx4_free_cmd_mailbox(dev, mailbox);
kfree(gw);
@@ -1875,7 +1886,8 @@ static void mlx4_ib_scan_netdevs(struct mlx4_ib_dev *ibdev,
* don't want the bond IP based gids in the table since
* flows that select port by gid may get the down port.
*/
- if (port_state == IB_PORT_DOWN) {
+ if (port_state == IB_PORT_DOWN &&
+ !mlx4_is_bonded(ibdev->dev)) {
reset_gid_table(ibdev, port);
mlx4_ib_set_default_gid(ibdev,
curr_netdev,
@@ -2047,6 +2059,7 @@ static void *mlx4_ib_add(struct mlx4_dev *dev)
int err;
struct mlx4_ib_iboe *iboe;
int ib_num_ports = 0;
+ int num_req_counters;
pr_info_once("%s", mlx4_ib_version);
@@ -2086,7 +2099,8 @@ static void *mlx4_ib_add(struct mlx4_dev *dev)
ibdev->ib_dev.node_type = RDMA_NODE_IB_CA;
ibdev->ib_dev.local_dma_lkey = dev->caps.reserved_lkey;
ibdev->num_ports = num_ports;
- ibdev->ib_dev.phys_port_cnt = ibdev->num_ports;
+ ibdev->ib_dev.phys_port_cnt = mlx4_is_bonded(dev) ?
+ 1 : ibdev->num_ports;
ibdev->ib_dev.num_comp_vectors = dev->caps.num_comp_vectors;
ibdev->ib_dev.dma_device = &dev->persist->pdev->dev;
@@ -2207,7 +2221,8 @@ static void *mlx4_ib_add(struct mlx4_dev *dev)
if (init_node_data(ibdev))
goto err_map;
- for (i = 0; i < ibdev->num_ports; ++i) {
+ num_req_counters = mlx4_is_bonded(dev) ? 1 : ibdev->num_ports;
+ for (i = 0; i < num_req_counters; ++i) {
mutex_init(&ibdev->qp1_proxy_lock[i]);
if (mlx4_ib_port_link_layer(&ibdev->ib_dev, i + 1) ==
IB_LINK_LAYER_ETHERNET) {
@@ -2218,6 +2233,10 @@ static void *mlx4_ib_add(struct mlx4_dev *dev)
ibdev->counters[i] = -1;
}
}
+ if (mlx4_is_bonded(dev))
+ for (i = 1; i < ibdev->num_ports ; ++i)
+ ibdev->counters[i] = ibdev->counters[0];
+
mlx4_foreach_port(i, dev, MLX4_PORT_TYPE_IB)
ib_num_ports++;
@@ -2538,6 +2557,38 @@ out:
return;
}
+static void handle_bonded_port_state_event(struct work_struct *work)
+{
+ struct ib_event_work *ew =
+ container_of(work, struct ib_event_work, work);
+ struct mlx4_ib_dev *ibdev = ew->ib_dev;
+ enum ib_port_state bonded_port_state = IB_PORT_NOP;
+ int i;
+ struct ib_event ibev;
+
+ kfree(ew);
+ spin_lock_bh(&ibdev->iboe.lock);
+ for (i = 0; i < MLX4_MAX_PORTS; ++i) {
+ struct net_device *curr_netdev = ibdev->iboe.netdevs[i];
+
+ enum ib_port_state curr_port_state =
+ (netif_running(curr_netdev) &&
+ netif_carrier_ok(curr_netdev)) ?
+ IB_PORT_ACTIVE : IB_PORT_DOWN;
+
+ bonded_port_state = (bonded_port_state != IB_PORT_ACTIVE) ?
+ curr_port_state : IB_PORT_ACTIVE;
+ }
+ spin_unlock_bh(&ibdev->iboe.lock);
+
+ ibev.device = &ibdev->ib_dev;
+ ibev.element.port_num = 1;
+ ibev.event = (bonded_port_state == IB_PORT_ACTIVE) ?
+ IB_EVENT_PORT_ACTIVE : IB_EVENT_PORT_ERR;
+
+ ib_dispatch_event(&ibev);
+}
+
static void mlx4_ib_event(struct mlx4_dev *dev, void *ibdev_ptr,
enum mlx4_dev_event event, unsigned long param)
{
@@ -2547,6 +2598,18 @@ static void mlx4_ib_event(struct mlx4_dev *dev, void *ibdev_ptr,
struct ib_event_work *ew;
int p = 0;
+ if (mlx4_is_bonded(dev) &&
+ ((event == MLX4_DEV_EVENT_PORT_UP) ||
+ (event == MLX4_DEV_EVENT_PORT_DOWN))) {
+ ew = kmalloc(sizeof(*ew), GFP_ATOMIC);
+ if (!ew)
+ return;
+ INIT_WORK(&ew->work, handle_bonded_port_state_event);
+ ew->ib_dev = ibdev;
+ queue_work(wq, &ew->work);
+ return;
+ }
+
if (event == MLX4_DEV_EVENT_PORT_MGMT_CHANGE)
eqe = (struct mlx4_eqe *)param;
else
@@ -2607,7 +2670,7 @@ static void mlx4_ib_event(struct mlx4_dev *dev, void *ibdev_ptr,
}
ibev.device = ibdev_ptr;
- ibev.element.port_num = (u8) p;
+ ibev.element.port_num = mlx4_is_bonded(ibdev->dev) ? 1 : (u8)p;
ib_dispatch_event(&ibev);
}
@@ -2616,7 +2679,8 @@ static struct mlx4_interface mlx4_ib_interface = {
.add = mlx4_ib_add,
.remove = mlx4_ib_remove,
.event = mlx4_ib_event,
- .protocol = MLX4_PROT_IB_IPV6
+ .protocol = MLX4_PROT_IB_IPV6,
+ .flags = MLX4_INTFF_BONDING
};
static int __init mlx4_ib_init(void)
--
1.7.1
^ permalink raw reply related
* [PATCH net-next 09/10] IB/mlx4: Create mirror flows in port aggregation mode
From: Or Gerlitz @ 2015-02-03 14:48 UTC (permalink / raw)
To: David S. Miller
Cc: netdev, Roland Dreier, Amir Vadai, Tal Alon, Moni Shoua,
Or Gerlitz
In-Reply-To: <1422974919-28084-1-git-send-email-ogerlitz@mellanox.com>
From: Moni Shoua <monis@mellanox.com>
In port aggregation mode flows for port #1 (the only port) should be mirrored
on port #2. This is because packets can arrive from either physical ports.
Signed-off-by: Moni Shoua <monis@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
---
drivers/infiniband/hw/mlx4/main.c | 84 +++++++++++++++++++++++++++++-----
drivers/infiniband/hw/mlx4/mlx4_ib.h | 9 +++-
2 files changed, 80 insertions(+), 13 deletions(-)
diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c
index ed21ae6..ca52238 100644
--- a/drivers/infiniband/hw/mlx4/main.c
+++ b/drivers/infiniband/hw/mlx4/main.c
@@ -851,7 +851,7 @@ int mlx4_ib_add_mc(struct mlx4_ib_dev *mdev, struct mlx4_ib_qp *mqp,
struct mlx4_ib_steering {
struct list_head list;
- u64 reg_id;
+ struct mlx4_flow_reg_id reg_id;
union ib_gid gid;
};
@@ -1142,9 +1142,11 @@ static struct ib_flow *mlx4_ib_create_flow(struct ib_qp *qp,
struct ib_flow_attr *flow_attr,
int domain)
{
- int err = 0, i = 0;
+ int err = 0, i = 0, j = 0;
struct mlx4_ib_flow *mflow;
enum mlx4_net_trans_promisc_mode type[2];
+ struct mlx4_dev *dev = (to_mdev(qp->device))->dev;
+ int is_bonded = mlx4_is_bonded(dev);
memset(type, 0, sizeof(type));
@@ -1179,26 +1181,55 @@ static struct ib_flow *mlx4_ib_create_flow(struct ib_qp *qp,
while (i < ARRAY_SIZE(type) && type[i]) {
err = __mlx4_ib_create_flow(qp, flow_attr, domain, type[i],
- &mflow->reg_id[i]);
+ &mflow->reg_id[i].id);
if (err)
goto err_create_flow;
i++;
+ if (is_bonded) {
+ flow_attr->port = 2;
+ err = __mlx4_ib_create_flow(qp, flow_attr,
+ domain, type[j],
+ &mflow->reg_id[j].mirror);
+ flow_attr->port = 1;
+ if (err)
+ goto err_create_flow;
+ j++;
+ }
+
}
if (i < ARRAY_SIZE(type) && flow_attr->type == IB_FLOW_ATTR_NORMAL) {
- err = mlx4_ib_tunnel_steer_add(qp, flow_attr, &mflow->reg_id[i]);
+ err = mlx4_ib_tunnel_steer_add(qp, flow_attr,
+ &mflow->reg_id[i].id);
if (err)
goto err_create_flow;
i++;
+ if (is_bonded) {
+ flow_attr->port = 2;
+ err = mlx4_ib_tunnel_steer_add(qp, flow_attr,
+ &mflow->reg_id[j].mirror);
+ flow_attr->port = 1;
+ if (err)
+ goto err_create_flow;
+ j++;
+ }
+ /* function to create mirror rule */
}
return &mflow->ibflow;
err_create_flow:
while (i) {
- (void)__mlx4_ib_destroy_flow(to_mdev(qp->device)->dev, mflow->reg_id[i]);
+ (void)__mlx4_ib_destroy_flow(to_mdev(qp->device)->dev,
+ mflow->reg_id[i].id);
i--;
}
+
+ while (j) {
+ (void)__mlx4_ib_destroy_flow(to_mdev(qp->device)->dev,
+ mflow->reg_id[j].mirror);
+ j--;
+ }
err_free:
kfree(mflow);
return ERR_PTR(err);
@@ -1211,10 +1242,16 @@ static int mlx4_ib_destroy_flow(struct ib_flow *flow_id)
struct mlx4_ib_dev *mdev = to_mdev(flow_id->qp->device);
struct mlx4_ib_flow *mflow = to_mflow(flow_id);
- while (i < ARRAY_SIZE(mflow->reg_id) && mflow->reg_id[i]) {
- err = __mlx4_ib_destroy_flow(mdev->dev, mflow->reg_id[i]);
+ while (i < ARRAY_SIZE(mflow->reg_id) && mflow->reg_id[i].id) {
+ err = __mlx4_ib_destroy_flow(mdev->dev, mflow->reg_id[i].id);
if (err)
ret = err;
+ if (mflow->reg_id[i].mirror) {
+ err = __mlx4_ib_destroy_flow(mdev->dev,
+ mflow->reg_id[i].mirror);
+ if (err)
+ ret = err;
+ }
i++;
}
@@ -1226,11 +1263,12 @@ static int mlx4_ib_mcg_attach(struct ib_qp *ibqp, union ib_gid *gid, u16 lid)
{
int err;
struct mlx4_ib_dev *mdev = to_mdev(ibqp->device);
+ struct mlx4_dev *dev = mdev->dev;
struct mlx4_ib_qp *mqp = to_mqp(ibqp);
- u64 reg_id;
struct mlx4_ib_steering *ib_steering = NULL;
enum mlx4_protocol prot = (gid->raw[1] == 0x0e) ?
MLX4_PROT_IB_IPV4 : MLX4_PROT_IB_IPV6;
+ struct mlx4_flow_reg_id reg_id;
if (mdev->dev->caps.steering_mode ==
MLX4_STEERING_MODE_DEVICE_MANAGED) {
@@ -1242,10 +1280,20 @@ static int mlx4_ib_mcg_attach(struct ib_qp *ibqp, union ib_gid *gid, u16 lid)
err = mlx4_multicast_attach(mdev->dev, &mqp->mqp, gid->raw, mqp->port,
!!(mqp->flags &
MLX4_IB_QP_BLOCK_MULTICAST_LOOPBACK),
- prot, ®_id);
+ prot, ®_id.id);
if (err)
goto err_malloc;
+ reg_id.mirror = 0;
+ if (mlx4_is_bonded(dev)) {
+ err = mlx4_multicast_attach(mdev->dev, &mqp->mqp, gid->raw, 2,
+ !!(mqp->flags &
+ MLX4_IB_QP_BLOCK_MULTICAST_LOOPBACK),
+ prot, ®_id.mirror);
+ if (err)
+ goto err_add;
+ }
+
err = add_gid_entry(ibqp, gid);
if (err)
goto err_add;
@@ -1261,7 +1309,10 @@ static int mlx4_ib_mcg_attach(struct ib_qp *ibqp, union ib_gid *gid, u16 lid)
err_add:
mlx4_multicast_detach(mdev->dev, &mqp->mqp, gid->raw,
- prot, reg_id);
+ prot, reg_id.id);
+ if (reg_id.mirror)
+ mlx4_multicast_detach(mdev->dev, &mqp->mqp, gid->raw,
+ prot, reg_id.mirror);
err_malloc:
kfree(ib_steering);
@@ -1288,10 +1339,12 @@ static int mlx4_ib_mcg_detach(struct ib_qp *ibqp, union ib_gid *gid, u16 lid)
{
int err;
struct mlx4_ib_dev *mdev = to_mdev(ibqp->device);
+ struct mlx4_dev *dev = mdev->dev;
struct mlx4_ib_qp *mqp = to_mqp(ibqp);
struct net_device *ndev;
struct mlx4_ib_gid_entry *ge;
- u64 reg_id = 0;
+ struct mlx4_flow_reg_id reg_id = {0, 0};
+
enum mlx4_protocol prot = (gid->raw[1] == 0x0e) ?
MLX4_PROT_IB_IPV4 : MLX4_PROT_IB_IPV6;
@@ -1316,10 +1369,17 @@ static int mlx4_ib_mcg_detach(struct ib_qp *ibqp, union ib_gid *gid, u16 lid)
}
err = mlx4_multicast_detach(mdev->dev, &mqp->mqp, gid->raw,
- prot, reg_id);
+ prot, reg_id.id);
if (err)
return err;
+ if (mlx4_is_bonded(dev)) {
+ err = mlx4_multicast_detach(mdev->dev, &mqp->mqp, gid->raw,
+ prot, reg_id.mirror);
+ if (err)
+ return err;
+ }
+
mutex_lock(&mqp->mutex);
ge = find_gid_entry(mqp, gid->raw);
if (ge) {
diff --git a/drivers/infiniband/hw/mlx4/mlx4_ib.h b/drivers/infiniband/hw/mlx4/mlx4_ib.h
index 6eb743f..2b49f9d 100644
--- a/drivers/infiniband/hw/mlx4/mlx4_ib.h
+++ b/drivers/infiniband/hw/mlx4/mlx4_ib.h
@@ -134,10 +134,17 @@ struct mlx4_ib_fmr {
struct mlx4_fmr mfmr;
};
+#define MAX_REGS_PER_FLOW 2
+
+struct mlx4_flow_reg_id {
+ u64 id;
+ u64 mirror;
+};
+
struct mlx4_ib_flow {
struct ib_flow ibflow;
/* translating DMFS verbs sniffer rule to FW API requires two reg IDs */
- u64 reg_id[2];
+ struct mlx4_flow_reg_id reg_id[MAX_REGS_PER_FLOW];
};
struct mlx4_ib_wq {
--
1.7.1
^ permalink raw reply related
* [PATCH net-next 10/10] IB/mlx4: Load balance ports in port aggregation mode
From: Or Gerlitz @ 2015-02-03 14:48 UTC (permalink / raw)
To: David S. Miller
Cc: netdev, Roland Dreier, Amir Vadai, Tal Alon, Moni Shoua,
Or Gerlitz
In-Reply-To: <1422974919-28084-1-git-send-email-ogerlitz@mellanox.com>
From: Moni Shoua <monis@mellanox.com>
When the mlx4 IB (RoCE) device works in link aggregation mode, it
exposes a single port to upper layers. Therefore, applications always
set '1' in port_num attribute when modifying a QP or creating an address handle.
To make sure that a node uses all available ports the mlx4 driver will
override the port_num attribute with a round robin policy.
Signed-off-by: Moni Shoua <monis@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
---
drivers/infiniband/hw/mlx4/ah.c | 1 +
drivers/infiniband/hw/mlx4/main.c | 1 +
drivers/infiniband/hw/mlx4/mlx4_ib.h | 8 ++++++++
drivers/infiniband/hw/mlx4/qp.c | 19 +++++++++++++++++++
4 files changed, 29 insertions(+), 0 deletions(-)
diff --git a/drivers/infiniband/hw/mlx4/ah.c b/drivers/infiniband/hw/mlx4/ah.c
index 2d8c339..f50a546 100644
--- a/drivers/infiniband/hw/mlx4/ah.c
+++ b/drivers/infiniband/hw/mlx4/ah.c
@@ -36,6 +36,7 @@
#include <linux/slab.h>
#include <linux/inet.h>
#include <linux/string.h>
+#include <linux/mlx4/driver.h>
#include "mlx4_ib.h"
diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c
index ca52238..2ed5b99 100644
--- a/drivers/infiniband/hw/mlx4/main.c
+++ b/drivers/infiniband/hw/mlx4/main.c
@@ -2153,6 +2153,7 @@ static void *mlx4_ib_add(struct mlx4_dev *dev)
MLX4_INIT_DOORBELL_LOCK(&ibdev->uar_lock);
ibdev->dev = dev;
+ ibdev->bond_next_port = 0;
strlcpy(ibdev->ib_dev.name, "mlx4_%d", IB_DEVICE_NAME_MAX);
ibdev->ib_dev.owner = THIS_MODULE;
diff --git a/drivers/infiniband/hw/mlx4/mlx4_ib.h b/drivers/infiniband/hw/mlx4/mlx4_ib.h
index 2b49f9d..721540c 100644
--- a/drivers/infiniband/hw/mlx4/mlx4_ib.h
+++ b/drivers/infiniband/hw/mlx4/mlx4_ib.h
@@ -534,6 +534,7 @@ struct mlx4_ib_dev {
struct mlx4_ib_qp *qp1_proxy[MLX4_MAX_PORTS];
/* lock when destroying qp1_proxy and getting netdev events */
struct mutex qp1_proxy_lock[MLX4_MAX_PORTS];
+ u8 bond_next_port;
};
struct ib_event_work {
@@ -629,6 +630,13 @@ static inline struct mlx4_ib_ah *to_mah(struct ib_ah *ibah)
return container_of(ibah, struct mlx4_ib_ah, ibah);
}
+static inline u8 mlx4_ib_bond_next_port(struct mlx4_ib_dev *dev)
+{
+ dev->bond_next_port = (dev->bond_next_port + 1) % dev->num_ports;
+
+ return dev->bond_next_port + 1;
+}
+
int mlx4_ib_init_sriov(struct mlx4_ib_dev *dev);
void mlx4_ib_close_sriov(struct mlx4_ib_dev *dev);
diff --git a/drivers/infiniband/hw/mlx4/qp.c b/drivers/infiniband/hw/mlx4/qp.c
index 2f85fc7..792f9dc 100644
--- a/drivers/infiniband/hw/mlx4/qp.c
+++ b/drivers/infiniband/hw/mlx4/qp.c
@@ -1905,6 +1905,22 @@ int mlx4_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr,
goto out;
}
+ if (mlx4_is_bonded(dev->dev) && (attr_mask & IB_QP_PORT)) {
+ if ((cur_state == IB_QPS_RESET) && (new_state == IB_QPS_INIT)) {
+ if ((ibqp->qp_type == IB_QPT_RC) ||
+ (ibqp->qp_type == IB_QPT_UD) ||
+ (ibqp->qp_type == IB_QPT_UC) ||
+ (ibqp->qp_type == IB_QPT_RAW_PACKET) ||
+ (ibqp->qp_type == IB_QPT_XRC_INI)) {
+ attr->port_num = mlx4_ib_bond_next_port(dev);
+ }
+ } else {
+ /* no sense in changing port_num
+ * when ports are bonded */
+ attr_mask &= ~IB_QP_PORT;
+ }
+ }
+
if ((attr_mask & IB_QP_PORT) &&
(attr->port_num == 0 || attr->port_num > dev->num_ports)) {
pr_debug("qpn 0x%x: invalid port number (%d) specified "
@@ -1955,6 +1971,9 @@ int mlx4_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr,
err = __mlx4_ib_modify_qp(ibqp, attr, attr_mask, cur_state, new_state);
+ if (mlx4_is_bonded(dev->dev) && (attr_mask & IB_QP_PORT))
+ attr->port_num = 1;
+
out:
mutex_unlock(&qp->mutex);
return err;
--
1.7.1
^ permalink raw reply related
* [PATCH 07/10] IB/mlx4: reuse mlx4_mac_to_u64()
From: Or Gerlitz @ 2015-02-03 14:48 UTC (permalink / raw)
To: David S. Miller; +Cc: netdev, Roland Dreier, Amir Vadai, Tal Alon, Moni Shoua
In-Reply-To: <1422974919-28084-1-git-send-email-ogerlitz@mellanox.com>
From: Moni Shoua <monis@mellanox.com>
This function is implemented twice... get rid of one copy.
Signed-off-by: Moni Shoua <monis@mellanox.com>
---
drivers/infiniband/hw/mlx4/qp.c | 12 +-----------
1 files changed, 1 insertions(+), 11 deletions(-)
diff --git a/drivers/infiniband/hw/mlx4/qp.c b/drivers/infiniband/hw/mlx4/qp.c
index cf000b7..2f85fc7 100644
--- a/drivers/infiniband/hw/mlx4/qp.c
+++ b/drivers/infiniband/hw/mlx4/qp.c
@@ -40,6 +40,7 @@
#include <rdma/ib_addr.h>
#include <rdma/ib_mad.h>
+#include <linux/mlx4/driver.h>
#include <linux/mlx4/qp.h>
#include "mlx4_ib.h"
@@ -93,17 +94,6 @@ enum {
#ifndef ETH_ALEN
#define ETH_ALEN 6
#endif
-static inline u64 mlx4_mac_to_u64(u8 *addr)
-{
- u64 mac = 0;
- int i;
-
- for (i = 0; i < ETH_ALEN; i++) {
- mac <<= 8;
- mac |= addr[i];
- }
- return mac;
-}
static const __be32 mlx4_ib_opcode[] = {
[IB_WR_SEND] = cpu_to_be32(MLX4_OPCODE_SEND),
--
1.7.1
^ permalink raw reply related
* Re: Per-connection tcp_retries2 and RFC 1122 compliance
From: Neal Cardwell @ 2015-02-03 14:50 UTC (permalink / raw)
To: John Eckersberg; +Cc: Netdev
In-Reply-To: <87a90w10tq.fsf@redhat.com>
On Mon, Feb 2, 2015 at 4:05 PM, John Eckersberg <jeckersb@redhat.com> wrote:
> Greetings,
>
> RFC 1122, section 4.2.3.5 "TCP Connection Failures", states:
>
> (d) An application MUST be able to set the value for R2 for
> a particular connection. For example, an interactive
> application might set R2 to "infinity," giving the user
> control over when to disconnect.
>
> The R2 value referenced above is implemented as the tcp_retries2 sysctl.
> However it seems that the only way to tune that value is via the global
> sysctl knob. In other words, there is no provided way to set it only
> for a particular connection as RFC 1122 requires.
>
> Could someone confirm that this is a legitimate bug/deficiency? Or am I
> just missing something?
I believe the functionality you are looking for is the
TCP_USER_TIMEOUT socket option:
commit dca43c75e7e545694a9dd6288553f55c53e2a3a3
Author: Jerry Chu <hkchu@google.com>
Date: Fri Aug 27 19:13:28 2010 +0000
tcp: Add TCP_USER_TIMEOUT socket option.
This patch provides a "user timeout" support as described in RFC793. The
socket option is also needed for the the local half of RFC5482 "TCP User
Timeout Option".
TCP_USER_TIMEOUT is a TCP level socket option that takes an unsigned int,
when > 0, to specify the maximum amount of time in ms that transmitted
data may remain unacknowledged before TCP will forcefully close the
corresponding connection and return ETIMEDOUT to the application. If
0 is given, TCP will continue to use the system default.
Increasing the user timeouts allows a TCP connection to survive extended
periods without end-to-end connectivity. Decreasing the user timeouts
allows applications to "fail fast" if so desired. Otherwise it may take
upto 20 minutes with the current system defaults in a normal WAN
environment.
....
Note how tcp_write_timeout() can pass in both sysctl_tcp_retries2 and
icsk->icsk_user_timeout to retransmits_timed_out(), and the
icsk->icsk_user_timeout value is used (if non-zero) in preference to
sysctl_tcp_retries2.
neal
^ permalink raw reply
* Re: [PATCH 00/10] Add HA and LAG support to mlx4 RoCE and SRIOV services
From: Or Gerlitz @ 2015-02-03 14:59 UTC (permalink / raw)
To: David S. Miller
Cc: netdev, Roland Dreier, Amir Vadai, Tal Alon, Jiri Pirko,
Jay Vosburgh
In-Reply-To: <1422974919-28084-1-git-send-email-ogerlitz@mellanox.com>
On 2/3/2015 4:48 PM, Or Gerlitz wrote:
> Hi Dave,
>
> This series takes advanges of bonding mlx4 Ethernet devices to support
> a model of High-Availability and Link Aggregation for more environments.
+Jay and Jiri, sorry for not adding you guys, guess you are subscribed...
>
> The mlx4 driver reacts on netdev events generated by bonding when
> slave state changes happen by programming a HW V2P (Virt-to-Phys)
> port table. Bonding was extended to expose these state changes
> through netdev events.
>
> When an mlx4 interface such as the mlx4 IB/RoCE driver is subject to
> this policy, QPs are created over virtual ports which are mapped
> to one of the two physical ports. When a failure happens, the
> re-programming of the V2P table allows traffic to keep flowing.
>
> The mlx4 Ethernet driver interfaces are not subject to this
> policy and act as usual.
>
> A 2nd use-case for this model would be to add HA and Link Aggregation
> support to single ported mlx4 Ethernet VFs. In this case, the PF Ethernet
> intrfaces are bonded, all the VFs see single port devices (which is
> supported already today), and VF QPs are subject to V2P.
>
> Moni and Or.
>
> Moni Shoua (10):
> net/core: Add event for a change in slave state
> net/bonding: Move slave state changes to a helper function
> net/bonding: Notify state change on slaves
> net/mlx4_core: Port aggregation low level interface
> net/mlx4_core: Port aggregation upper layer interface
> net/mlx4_en: Port aggregation configuration
> IB/mlx4: Reuse mlx4_mac_to_u64()
> IB/mlx4: Add port aggregation support
> IB/mlx4: Create mirror flows in port aggregation mode
> IB/mlx4: Load balance ports in port aggregation mode
>
> drivers/infiniband/hw/mlx4/ah.c | 1 +
> drivers/infiniband/hw/mlx4/main.c | 161 ++++++++++++++++--
> drivers/infiniband/hw/mlx4/mlx4_ib.h | 17 ++-
> drivers/infiniband/hw/mlx4/qp.c | 31 +++--
> drivers/net/bonding/bond_main.c | 106 +++++++++---
> drivers/net/ethernet/mellanox/mlx4/cmd.c | 9 +
> drivers/net/ethernet/mellanox/mlx4/en_main.c | 8 +
> drivers/net/ethernet/mellanox/mlx4/en_netdev.c | 176 ++++++++++++++++++++
> drivers/net/ethernet/mellanox/mlx4/en_resources.c | 8 +-
> drivers/net/ethernet/mellanox/mlx4/fw.c | 56 ++++++-
> drivers/net/ethernet/mellanox/mlx4/intf.c | 54 ++++++
> drivers/net/ethernet/mellanox/mlx4/main.c | 89 ++++++++++
> drivers/net/ethernet/mellanox/mlx4/mlx4.h | 3 +
> drivers/net/ethernet/mellanox/mlx4/mlx4_en.h | 5 +
> drivers/net/ethernet/mellanox/mlx4/qp.c | 2 +
> .../net/ethernet/mellanox/mlx4/resource_tracker.c | 3 +
> include/linux/mlx4/cmd.h | 7 +
> include/linux/mlx4/device.h | 11 +-
> include/linux/mlx4/driver.h | 19 ++
> include/linux/mlx4/qp.h | 1 +
> include/linux/netdevice.h | 15 ++
> include/net/bonding.h | 17 ++
> net/core/dev.c | 20 +++
> net/core/rtnetlink.c | 1 +
> 24 files changed, 756 insertions(+), 64 deletions(-)
>
^ permalink raw reply
* Re: Throughput regression with `tcp: refine TSO autosizing`
From: Eric Dumazet @ 2015-02-03 15:03 UTC (permalink / raw)
To: Michal Kazior
Cc: linux-wireless, Network Development,
eyalpe-LDSdmyG8hGV8YrgS2mwiifqBs+8SCbDb
In-Reply-To: <1422973660.907.10.camel-XN9IlZ5yJG9HTL0Zs8A6p/gx64E7kk8eUsxypvmhUTTZJqsBc5GL+g@public.gmane.org>
On Tue, 2015-02-03 at 06:27 -0800, Eric Dumazet wrote:
> Are packets TX completed after a timer or something ?
>
> Some very heavy stuff might run from tasklet (or other softirq triggered) event.
>
Right, commit 6c5151a9ffa9f796f2d707617cecb6b6b241dff8
("ath10k: batch htt tx/rx completions")
is very suspicious.
Please revert it.
BTW, ath10k_htt_txrx_compl_task() runs from softirq context, so the
_bh() prefixes are not really needed.
It seems lot of batching happens in wifi drivers, not necessarily at the
right places.
--
To unsubscribe from this list: send the line "unsubscribe linux-wireless" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply
* Re: [PATCH net-next] bridge: Let bridge not age 'externally' learnt FDB entries, they are removed when 'external' entity notifies the aging
From: roopa @ 2015-02-03 15:11 UTC (permalink / raw)
To: Siva Mannem; +Cc: netdev, Scott Feldman, Jiri Pirko
In-Reply-To: <1422897714-5956-1-git-send-email-siva.mannem.lnx@gmail.com>
On 2/2/15, 9:21 AM, Siva Mannem wrote:
> When 'learned_sync' flag is turned on, the offloaded switch
> port syncs learned MAC addresses to bridge's FDB via switchdev notifier
> (NETDEV_SWITCH_FDB_ADD). Currently, FDB entries learnt via this mechanism are
> wrongly being deleted by bridge aging logic. This patch ensures that FDB
> entries synced from offloaded switch ports are not deleted by bridging logic.
> Such entries can only be deleted via switchdev notifier
> (NETDEV_SWITCH_FDB_DEL).
Your patch seems right and maintains symmetry for fdb add/del of
externally learnt entries.
However, this could be made configurable. I think some drivers may rely
on bridge driver aging these entries (The default setting needs more
thought).
I am not sure what rocker does (CC'ed rocker maintainers). But, our
driver does rely on the bridge driver aging these entries by default.
>
> Signed-off-by: Siva Mannem <siva.mannem.lnx@gmail.com>
> ---
> net/bridge/br_fdb.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/net/bridge/br_fdb.c b/net/bridge/br_fdb.c
> index 08bf04b..6eb94b5 100644
> --- a/net/bridge/br_fdb.c
> +++ b/net/bridge/br_fdb.c
> @@ -280,7 +280,7 @@ void br_fdb_cleanup(unsigned long _data)
>
> hlist_for_each_entry_safe(f, n, &br->hash[i], hlist) {
> unsigned long this_timer;
> - if (f->is_static)
> + if (f->is_static || f->added_by_external_learn)
> continue;
> this_timer = f->updated + delay;
> if (time_before_eq(this_timer, jiffies))
^ permalink raw reply
* Re: [PATCH v2 08/17] {macvtap,tun}_get_user(): switch to iov_iter
From: Michael S. Tsirkin @ 2015-02-03 15:19 UTC (permalink / raw)
To: Al Viro; +Cc: David Miller, netdev, linux-kernel
In-Reply-To: <20150203142728.GC29656@ZenIV.linux.org.uk>
On Tue, Feb 03, 2015 at 02:27:28PM +0000, Al Viro wrote:
> On Tue, Feb 03, 2015 at 12:10:44PM +0200, Michael S. Tsirkin wrote:
>
> > Hmm does copy_from_iter actually modify the iovec?
> > If so, won't this break aio on tun/macvtap, by
> > reversing the effect of
> > commit 6f26c9a7555e5bcca3560919db9b852015077dae
> > tun: fix tun_chr_aio_write so that aio works
> > ?
> >
> >
> > Maybe we should change iovec_iter to avoid modifying the
> > underlying iovec?
>
> iov_iter never changes the underlying iovec (or kvec, or bvec).
> iter->iov_offset changes as you go and once you have consumed an
> entire iovec element ->iov is incremented to point to the next one
> (and ->iov_offset is reset to 0 at that point). *Contents* of
> iter->iov is never modified.
I see, I think I misread the code.
Thanks for the clarification.
--
MST
^ permalink raw reply
* Re: [PATCH v2 17/18] vhost: don't bother copying iovecs in handle_rx(), kill memcpy_toiovecend()
From: Michael S. Tsirkin @ 2015-02-03 15:21 UTC (permalink / raw)
To: Al Viro; +Cc: David Miller, netdev, kvm
In-Reply-To: <20150203100412.GA4896@redhat.com>
> Hmm having second thoughts here.
> Will this modify the iov in vq->iov?
> If yes, how will recvmsg fill it?
OK that was just me misunderstanding what the
function does. As it doesn't modify the iovec itself,
I think there's no issue, my ack stands.
^ permalink raw reply
* Re: [PATCH net v3] net: ipv6: allow explicitly choosing optimistic addresses
From: Lorenzo Colitti @ 2015-02-03 15:40 UTC (permalink / raw)
To: Erik Kline; +Cc: netdev@vger.kernel.org, Hannes Frederic Sowa
In-Reply-To: <1422859156-30691-1-git-send-email-ek@google.com>
On Mon, Feb 2, 2015 at 3:39 PM, Erik Kline <ek@google.com> wrote:
> @@ -1526,8 +1534,13 @@ int ipv6_chk_addr(struct net *net, const struct in6_addr *addr,
> hlist_for_each_entry_rcu(ifp, &inet6_addr_lst[hash], addr_lst) {
> if (!net_eq(dev_net(ifp->idev->dev), net))
> continue;
> + /* Permit optimistic addresses, but only under explicitly
> + * defined circumstances.
> + */
I don't think this comment adds much of value, the code right below it
is pretty clear.
> + bool optimistic_ok = (ifp->flags & IFA_F_OPTIMISTIC) &&
> + (banned_flags == IFA_F_TENTATIVE);
Not sure if this can happen in any real use case, but I think that
technically this is incorrect if banned_flags contains both
IFA_F_TENTATIVE and other flags that aren't IFA_F_OPTIMISTIC. For
example, suppose banned_flags = IFA_F_TENTATIVE | IFA_F_PERMANENT. In
that case, I think the code would reject an address with
IFA_F_TENTATIVE | IFA_F_OPTIMISTIC. You might be able to fix that
using something like:
int ifp_flags;
...
ifp_flags = ifp->flags;
if (ifp_flags & IFA_F_OPTIMISTIC) ifp_flags &= ~IFA_F_TENTATIVE;
if (ipv6_addr_equal(&ifp->addr, addr) &&
!(ifp_flags & banned_flags) &&
Though I think that at this point your original formulation (the one
that treated IFA_F_TENTATIVE specially and did not pass it in via
banned_flags) might be faster/simpler/better.
Hannes, any better ideas?
^ permalink raw reply
* [PATCH net-next] veth: set iflink to the peer veth
From: Nicolas Dichtel @ 2015-02-03 15:41 UTC (permalink / raw)
To: netdev; +Cc: davem, Nicolas Dichtel
Now that the peer netns is advertised in rtnl messages, we can set this property
so that IFLA_LINK will advertise the peer ifindex. It allows the userland to get
the full veth configuration.
Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
---
drivers/net/veth.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/drivers/net/veth.c b/drivers/net/veth.c
index 4cca36ebc4fb..02d1f798a371 100644
--- a/drivers/net/veth.c
+++ b/drivers/net/veth.c
@@ -427,9 +427,11 @@ static int veth_newlink(struct net *src_net, struct net_device *dev,
priv = netdev_priv(dev);
rcu_assign_pointer(priv->peer, peer);
+ dev->iflink = peer->ifindex;
priv = netdev_priv(peer);
rcu_assign_pointer(priv->peer, dev);
+ peer->iflink = dev->ifindex;
return 0;
err_register_dev:
--
2.2.2
^ permalink raw reply related
* RE: [PATCH net] hyperv: Fix the error processing in netvsc_send()
From: Haiyang Zhang @ 2015-02-03 15:46 UTC (permalink / raw)
To: Jason Wang
Cc: olaf@aepfle.de, netdev@vger.kernel.org,
driverdev-devel@linuxdriverproject.org,
linux-kernel@vger.kernel.org, davem@davemloft.net
In-Reply-To: <1422859762.7028.2@smtp.corp.redhat.com>
> -----Original Message-----
> From: Jason Wang [mailto:jasowang@redhat.com]
> Sent: Monday, February 2, 2015 1:49 AM
> >> btw, I find during netvsc_start_xmit(), ret was change to -ENOSPC
> >> when
> >> queue_sends[q_idx] < 1. But non of the caller check -ENOSPC in fact?
> >
> > In this case, we don't request re-send, so set ret to a value other
> > than
> > -EAGAIN.
>
> Why not? We have available slots for it to be sent now. Dropping the
> packet in this case may cause out of order sending.
The EAGAIN error doesn't normally happen, because we set the hi water mark
to stop send queue. If in really rare case, the ring buffer is full and there
is no outstanding sends, we can't stop queue here because there will be no
send-completion msg to wake it up. And, the ring buffer is likely to be
occupied by other special msg, e.g. receive-completion msg (not a normal case),
so we can't assume there are available slots. We don't request retry from
the upper layer in this case to avoid possible busy retry.
Thanks,
- Haiyang
^ permalink raw reply
* low vxlan throughput with tso enabled
From: Reiner Herrmann @ 2015-02-03 15:50 UTC (permalink / raw)
To: netdev; +Cc: svens, Vittorio Curcio
Hi!
I have a vxlan tunnel established between two network interfaces, which
both have a MTU of 1500. The vxlan interfaces have the same MTU.
With TSO enabled, I observe low throughput with TCP connections (<100 kB/s).
Disabling TSO works around this issue and throughput is as expected.
Can someone please explain how TSO is influencing the tunnel to cause
such a difference?
Kind regards,
Reiner
^ permalink raw reply
* [PATCH net-next 0/7] Mellanox drivers updates Feb-03-2015
From: Amir Vadai @ 2015-02-03 15:57 UTC (permalink / raw)
To: David S. Miller; +Cc: netdev, Amir Vadai, Or Gerlitz, Yevgeny Petrilin
Hi,
This patchset introduces some small bug fixes and code cleanups in mlx4_core,
mlx4_en and mlx5_core.
I am sending it in parallel to the patchset sent by Or Gerlitz today [1] because
this is the end of the time frame for 3.20. I also checked that there are no
conflicts between those two patchsets (Or's patchset is focused on the bonding
area while this on Mellanox drivers).
The patchset was applied on top of commit 7d37d0c ('net: sctp: Deletion of an
unnecessary check before the function call "kfree"')
[1] - [PATCH 00/10] Add HA and LAG support to mlx4 RoCE and SRIOV services
http://marc.info/?l=linux-netdev&m=142297582610254&w=2
Thanks,
Amir
Ido Shamay (3):
net/mlx4_en: Print page allocator information
net/mlx4_en: Adjust RX frag strides to frag sizes
net/mlx4_en: Notify TX Vlan offload change
Maor Gottlieb (2):
net/mlx4_core: Fix mpt_entry initialization in
mlx4_mr_rereg_mem_write()
net/mlx4: mlx4_config_dev_retrieval() - Initialize struct config_dev
before using
Or Gerlitz (2):
net/mlx4_core: Fix misleading debug print on CQE stride support
net/mlx5_core: Move to use hex PCI device IDs
drivers/net/ethernet/mellanox/mlx4/en_netdev.c | 4 ++++
drivers/net/ethernet/mellanox/mlx4/en_rx.c | 9 +++++++--
drivers/net/ethernet/mellanox/mlx4/fw.c | 2 +-
drivers/net/ethernet/mellanox/mlx4/main.c | 3 ++-
drivers/net/ethernet/mellanox/mlx4/mr.c | 13 +++++--------
drivers/net/ethernet/mellanox/mlx4/resource_tracker.c | 2 +-
drivers/net/ethernet/mellanox/mlx5/core/main.c | 12 ++++++------
7 files changed, 26 insertions(+), 19 deletions(-)
--
1.9.3
^ permalink raw reply
* [PATCH net-next 1/7] net/mlx4_core: Fix mpt_entry initialization in mlx4_mr_rereg_mem_write()
From: Amir Vadai @ 2015-02-03 15:57 UTC (permalink / raw)
To: David S. Miller
Cc: netdev, Amir Vadai, Or Gerlitz, Yevgeny Petrilin, Maor Gottlieb
In-Reply-To: <1422979041-25208-1-git-send-email-amirv@mellanox.com>
From: Maor Gottlieb <maorg@mellanox.com>
a) Previously, mlx4_mr_rereg_write filled the MPT's start
and length with the old MPT's values.
Fixing the initialization to take the new start and length.
b) In addition access flags in mpt_status were initialized instead of
status due to bad boolean operation. Fixing the operation.
c) Initialization of pd_slave caused a protection error.
Fix - removing this initialization.
d) In resource_tracker.c: Fixing vf encoding to be one-based.
Fixes: e630664c ('mlx4_core: Add helper functions to support MR re-registration')
Signed-off-by: Maor Gottlieb <maorg@mellanox.com>
Signed-off-by: Amir Vadai <amirv@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx4/mr.c | 13 +++++--------
drivers/net/ethernet/mellanox/mlx4/resource_tracker.c | 2 +-
2 files changed, 6 insertions(+), 9 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx4/mr.c b/drivers/net/ethernet/mellanox/mlx4/mr.c
index d21e884..78f51e1 100644
--- a/drivers/net/ethernet/mellanox/mlx4/mr.c
+++ b/drivers/net/ethernet/mellanox/mlx4/mr.c
@@ -598,14 +598,11 @@ int mlx4_mr_rereg_mem_write(struct mlx4_dev *dev, struct mlx4_mr *mr,
if (err)
return err;
- mpt_entry->start = cpu_to_be64(mr->iova);
- mpt_entry->length = cpu_to_be64(mr->size);
- mpt_entry->entity_size = cpu_to_be32(mr->mtt.page_shift);
-
- mpt_entry->pd_flags &= cpu_to_be32(MLX4_MPT_PD_MASK |
- MLX4_MPT_PD_FLAG_EN_INV);
- mpt_entry->flags &= cpu_to_be32(MLX4_MPT_FLAG_FREE |
- MLX4_MPT_FLAG_SW_OWNS);
+ mpt_entry->start = cpu_to_be64(iova);
+ mpt_entry->length = cpu_to_be64(size);
+ mpt_entry->entity_size = cpu_to_be32(page_shift);
+ mpt_entry->flags &= ~(cpu_to_be32(MLX4_MPT_FLAG_FREE |
+ MLX4_MPT_FLAG_SW_OWNS));
if (mr->mtt.order < 0) {
mpt_entry->flags |= cpu_to_be32(MLX4_MPT_FLAG_PHYSICAL);
mpt_entry->mtt_addr = 0;
diff --git a/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c b/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c
index 79feeb6..628c2e8 100644
--- a/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c
+++ b/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c
@@ -2541,7 +2541,7 @@ int mlx4_SW2HW_MPT_wrapper(struct mlx4_dev *dev, int slave,
/* Make sure that the PD bits related to the slave id are zeros. */
pd = mr_get_pd(inbox->buf);
pd_slave = (pd >> 17) & 0x7f;
- if (pd_slave != 0 && pd_slave != slave) {
+ if (pd_slave != 0 && --pd_slave != slave) {
err = -EPERM;
goto ex_abort;
}
--
1.9.3
^ permalink raw reply related
* [PATCH net-next 3/7] net/mlx4_core: Fix misleading debug print on CQE stride support
From: Amir Vadai @ 2015-02-03 15:57 UTC (permalink / raw)
To: David S. Miller; +Cc: netdev, Amir Vadai, Or Gerlitz, Yevgeny Petrilin
In-Reply-To: <1422979041-25208-1-git-send-email-amirv@mellanox.com>
From: Or Gerlitz <ogerlitz@mellanox.com>
We do support cache line sizes of 32 and 64 bytes without activating the
CQE stride feature. Fix a misleading print saying that these cache line
sizes aren't supported.
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Amir Vadai <amirv@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx4/main.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/drivers/net/ethernet/mellanox/mlx4/main.c b/drivers/net/ethernet/mellanox/mlx4/main.c
index cc9f484..e045562 100644
--- a/drivers/net/ethernet/mellanox/mlx4/main.c
+++ b/drivers/net/ethernet/mellanox/mlx4/main.c
@@ -251,7 +251,8 @@ static void mlx4_enable_cqe_eqe_stride(struct mlx4_dev *dev)
if (mlx4_is_master(dev))
dev_cap->function_caps |= MLX4_FUNC_CAP_EQE_CQE_STRIDE;
} else {
- mlx4_dbg(dev, "Disabling CQE stride cacheLine unsupported\n");
+ if (cache_line_size() != 32 && cache_line_size() != 64)
+ mlx4_dbg(dev, "Disabling CQE stride, cacheLine size unsupported\n");
dev_cap->flags2 &= ~MLX4_DEV_CAP_FLAG2_CQE_STRIDE;
dev_cap->flags2 &= ~MLX4_DEV_CAP_FLAG2_EQE_STRIDE;
}
--
1.9.3
^ permalink raw reply related
* [PATCH net-next 4/7] net/mlx5_core: Move to use hex PCI device IDs
From: Amir Vadai @ 2015-02-03 15:57 UTC (permalink / raw)
To: David S. Miller; +Cc: netdev, Amir Vadai, Or Gerlitz, Yevgeny Petrilin
In-Reply-To: <1422979041-25208-1-git-send-email-amirv@mellanox.com>
From: Or Gerlitz <ogerlitz@mellanox.com>
Align the IDs in the code with the modinfo, lspci -n, etc tools outputs.
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Amir Vadai <amirv@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx5/core/main.c | 12 ++++++------
1 file changed, 6 insertions(+), 6 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/main.c b/drivers/net/ethernet/mellanox/mlx5/core/main.c
index 3f45256..d665193 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/main.c
@@ -903,12 +903,12 @@ static void remove_one(struct pci_dev *pdev)
}
static const struct pci_device_id mlx5_core_pci_table[] = {
- { PCI_VDEVICE(MELLANOX, 4113) }, /* Connect-IB */
- { PCI_VDEVICE(MELLANOX, 4114) }, /* Connect-IB VF */
- { PCI_VDEVICE(MELLANOX, 4115) }, /* ConnectX-4 */
- { PCI_VDEVICE(MELLANOX, 4116) }, /* ConnectX-4 VF */
- { PCI_VDEVICE(MELLANOX, 4117) }, /* ConnectX-4LX */
- { PCI_VDEVICE(MELLANOX, 4118) }, /* ConnectX-4LX VF */
+ { PCI_VDEVICE(MELLANOX, 0x1011) }, /* Connect-IB */
+ { PCI_VDEVICE(MELLANOX, 0x1012) }, /* Connect-IB VF */
+ { PCI_VDEVICE(MELLANOX, 0x1013) }, /* ConnectX-4 */
+ { PCI_VDEVICE(MELLANOX, 0x1014) }, /* ConnectX-4 VF */
+ { PCI_VDEVICE(MELLANOX, 0x1015) }, /* ConnectX-4LX */
+ { PCI_VDEVICE(MELLANOX, 0x1016) }, /* ConnectX-4LX VF */
{ 0, }
};
--
1.9.3
^ permalink raw reply related
* [PATCH net-next 5/7] net/mlx4_en: Print page allocator information
From: Amir Vadai @ 2015-02-03 15:57 UTC (permalink / raw)
To: David S. Miller
Cc: netdev, Amir Vadai, Or Gerlitz, Yevgeny Petrilin, Ido Shamay
In-Reply-To: <1422979041-25208-1-git-send-email-amirv@mellanox.com>
From: Ido Shamay <idos@mellanox.com>
After Initialization of page_alloc, print actual allocated page
size and number of frags it contains. prints is done only when drv
message level is set on the interface.
Signed-off-by: Ido Shamay <idos@mellanox.com>
Signed-off-by: Amir Vadai <amirv@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx4/en_rx.c | 4 ++++
1 file changed, 4 insertions(+)
diff --git a/drivers/net/ethernet/mellanox/mlx4/en_rx.c b/drivers/net/ethernet/mellanox/mlx4/en_rx.c
index 2ba5d36..30a2203 100644
--- a/drivers/net/ethernet/mellanox/mlx4/en_rx.c
+++ b/drivers/net/ethernet/mellanox/mlx4/en_rx.c
@@ -162,6 +162,10 @@ static int mlx4_en_init_allocator(struct mlx4_en_priv *priv,
if (mlx4_alloc_pages(priv, &ring->page_alloc[i],
frag_info, GFP_KERNEL | __GFP_COLD))
goto out;
+
+ en_dbg(DRV, priv, " frag %d allocator: - size:%d frags:%d\n",
+ i, ring->page_alloc[i].page_size,
+ atomic_read(&ring->page_alloc[i].page->_count));
}
return 0;
--
1.9.3
^ permalink raw reply related
* [PATCH net-next 2/7] net/mlx4: mlx4_config_dev_retrieval() - Initialize struct config_dev before using
From: Amir Vadai @ 2015-02-03 15:57 UTC (permalink / raw)
To: David S. Miller
Cc: netdev, Amir Vadai, Or Gerlitz, Yevgeny Petrilin, Maor Gottlieb
In-Reply-To: <1422979041-25208-1-git-send-email-amirv@mellanox.com>
From: Maor Gottlieb <maorg@mellanox.com>
Add Initialization to struct config_dev before filling and using it.
Fix to warning:
warning: config_dev.rx_checksum_val may be used uninitialized in this function
Signed-off-by: Maor Gottlieb <maorg@mellanox.com>
Signed-off-by: Amir Vadai <amirv@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx4/fw.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/net/ethernet/mellanox/mlx4/fw.c b/drivers/net/ethernet/mellanox/mlx4/fw.c
index dbabfae..241838f 100644
--- a/drivers/net/ethernet/mellanox/mlx4/fw.c
+++ b/drivers/net/ethernet/mellanox/mlx4/fw.c
@@ -2162,7 +2162,7 @@ static const u8 config_dev_csum_flags[] = {
int mlx4_config_dev_retrieval(struct mlx4_dev *dev,
struct mlx4_config_dev_params *params)
{
- struct mlx4_config_dev config_dev;
+ struct mlx4_config_dev config_dev = {0};
int err;
u8 csum_mask;
--
1.9.3
^ permalink raw reply related
* [PATCH net-next 6/7] net/mlx4_en: Adjust RX frag strides to frag sizes
From: Amir Vadai @ 2015-02-03 15:57 UTC (permalink / raw)
To: David S. Miller
Cc: netdev, Amir Vadai, Or Gerlitz, Yevgeny Petrilin, Ido Shamay
In-Reply-To: <1422979041-25208-1-git-send-email-amirv@mellanox.com>
From: Ido Shamay <idos@mellanox.com>
This patch improves memory utilization and therefore the packets rate
for special MTU's. Instead of setting the frag_stride to the maximal
hard coded frag_size, use the actual frag_size that is set according to
the MTU, when setting the stride of the last frag.
So, for example, for MTU 1600, where the frag_size of the 2nd frag is
86, the frag_size is set to 128 instead of 4096. See below:
Before:
frag:0 - size:1536 prefix:0 stride:1536
frag:1 - size:86 prefix:1536 stride:4096
frag 0 allocator: - size:32768 frags:21
frag 1 allocator: - size:32768 frags:8
After:
frag:0 - size:1536 prefix:0 stride:1536
frag:1 - size:86 prefix:1536 stride:128
frag 0 allocator: - size:32768 frags:21
frag 1 allocator: - size:32768 frags:256
Signed-off-by: Ido Shamay <idos@mellanox.com>
Signed-off-by: Amir Vadai <amirv@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx4/en_rx.c | 5 +++--
1 file changed, 3 insertions(+), 2 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx4/en_rx.c b/drivers/net/ethernet/mellanox/mlx4/en_rx.c
index 30a2203..698d60d 100644
--- a/drivers/net/ethernet/mellanox/mlx4/en_rx.c
+++ b/drivers/net/ethernet/mellanox/mlx4/en_rx.c
@@ -1063,8 +1063,9 @@ void mlx4_en_calc_rx_buf(struct net_device *dev)
(eff_mtu > buf_size + frag_sizes[i]) ?
frag_sizes[i] : eff_mtu - buf_size;
priv->frag_info[i].frag_prefix_size = buf_size;
- priv->frag_info[i].frag_stride = ALIGN(frag_sizes[i],
- SMP_CACHE_BYTES);
+ priv->frag_info[i].frag_stride =
+ ALIGN(priv->frag_info[i].frag_size,
+ SMP_CACHE_BYTES);
buf_size += priv->frag_info[i].frag_size;
i++;
}
--
1.9.3
^ permalink raw reply related
* [PATCH net-next 7/7] net/mlx4_en: Notify TX Vlan offload change
From: Amir Vadai @ 2015-02-03 15:57 UTC (permalink / raw)
To: David S. Miller
Cc: netdev, Amir Vadai, Or Gerlitz, Yevgeny Petrilin, Ido Shamay
In-Reply-To: <1422979041-25208-1-git-send-email-amirv@mellanox.com>
From: Ido Shamay <idos@mellanox.com>
Notify users when TX vlan offload feature changed with ethtool.
Relevant command - ethtool -K <eth> txvlan on/off.
Signed-off-by: Ido Shamay <idos@mellanox.com>
Signed-off-by: Amir Vadai <amirv@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx4/en_netdev.c | 4 ++++
1 file changed, 4 insertions(+)
diff --git a/drivers/net/ethernet/mellanox/mlx4/en_netdev.c b/drivers/net/ethernet/mellanox/mlx4/en_netdev.c
index e075ff1..0897274 100644
--- a/drivers/net/ethernet/mellanox/mlx4/en_netdev.c
+++ b/drivers/net/ethernet/mellanox/mlx4/en_netdev.c
@@ -2201,6 +2201,10 @@ static int mlx4_en_set_features(struct net_device *netdev,
return ret;
}
+ if (DEV_FEATURE_CHANGED(netdev, features, NETIF_F_HW_VLAN_CTAG_TX))
+ en_info(priv, "Turn %s TX vlan strip offload\n",
+ (features & NETIF_F_HW_VLAN_CTAG_TX) ? "ON" : "OFF");
+
if (features & NETIF_F_LOOPBACK)
priv->ctrl_flags |= cpu_to_be32(MLX4_WQE_CTRL_FORCE_LOOPBACK);
else
--
1.9.3
^ permalink raw reply related
* Re: [V2,for,3.19,3/7] rtlwifi: rtl8192ee: Fix adhoc fail
From: Larry Finger @ 2015-02-03 16:00 UTC (permalink / raw)
To: Kalle Valo; +Cc: linux-wireless, Troy Tan, netdev, Stable
In-Reply-To: <87r3u7w2l5.fsf@kamboji.qca.qualcomm.com>
On 02/03/2015 07:23 AM, Kalle Valo wrote:
> Kalle Valo <kvalo@codeaurora.org> writes:
>
>>> From: Troy Tan <troy_tan@realsil.com.cn>
>>>
>>> When the buffer descriptor index exceeds 2, then a TX HANG condition
>>> will result.
>>>
>>> Signed-off-by: Troy Tan <troy_tan@realsil.com.cn>
>>> Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>
>>> Cc: Stable <stable@vger.kernel.org> [V3.18]
>>
>> Thanks, 4 patches applied to wireless-drivers-next.git:
>>
>> b661a5da5776 rtlwifi: rtl8192ee: Fix adhoc fail
>> 6e5f44361628 rtlwifi: rtl8192ee: Fix TX hang due to failure to update TX write point
>> 92ff754240b8 rtlwifi: rtl8192ee: Fix parsing of received packet
>> 21b39ddb5bb2 rtlwifi: rtl8192ee: Fix DMA stalls
>>
>> 3 patches skipped:
>
> I had to skip these three patches because of my mistake. So what I did
> was that I had merged ("fast forwarded") the net tree into my
> wireless-drivers tree and not realising I should not do that. So now I
> can't merge wireless-drivers into wireless-drivers-next anymore, as it
> will pull unnecessary net changes.
>
>> [V2,for,3.19,1/7] rtlwifi: Remove logging statement that is no longer
>> needed
>
> I'll apply this after 3.20-rc1 is released, it should apply then without
> problems (or the conflicts are easy for me to fix). Luckily this is just
> a cosmetic error and can wait for 3.20-rc2, right?
That is correct. 3.20-rc2 will be OK. I'll probably get a bunch of "your ****
driver is spamming my logs, but I can ignore them".
>
>> [V2,for,3.19,2/7] rtlwifi: rtl8192ee: Fix handling of new style descriptors
>> [V2,for,3.19,6/7] rtlwifi: rtl8192ee: Fix problems with calculating free space in FIFO
>
> Not sure what to do with these one. Should you rebase and send them now?
These two fix real bugs and need to be in the kernel ASAP. Unfortunately, I saw
you pass them off to DaveM and I deleted them from my "Submitted Patch" list. I
can recreate them from my working copy of wireless-drivers, and I will resubmit.
When I do, please process them as quickly as possible.
Thanks,
Larry
^ permalink raw reply
* Re: [PATCH] net: rocker: Change netdev names to include slot number
From: David Ahern @ 2015-02-03 16:06 UTC (permalink / raw)
To: Scott Feldman; +Cc: Florian Fainelli, Netdev
In-Reply-To: <CAE4R7bArt7+C-kHJ7UXFo4w9feG0C_uZqQ7n8KC0brSXhqX3gg@mail.gmail.com>
On 2/2/15 10:46 PM, Scott Feldman wrote:
>> [root@f21 ~]# cat /sys/devices/virtual/net/sw5p0/phys_port_id
>> cat: /sys/devices/virtual/net/sw5p0/phys_port_id: Operation not supported
>
> We should implement .ndo_get_phys_port_id in rocker to return a unique
> name for the port, which can then be used in udev script to name
> interface. When rocker device is instantiated, a unique switch name
> is given on qemu cmd line, for example "sw1". Rocker device could
> return that string to the driver, or even combine that string with the
Right. That's what I went looking for at first and noticed that info is
not pushed to the driver. And then I know at least one vendor puts the
slot id in interface names which is the reason I went for the PCI slot id.
> port index to have "sw1p1". The second instantiated rocker device
> would give up "sw2px" port names.
And yes, adding the port index to the above name -- though ideally
starting at 1 and not 0.
>
> A real switch with a physical port than can be split, or when multiple
> ports are ganged to form one mega port, letting the device pick a
> unique name is probably best. For example, splitting a single
> physical 40Gb port into 4 10Gb ports would present 4 kernel
> interfaces, but we probably want to use the base port in the naming,
> so we'd have something like "sw1p1s1" to mean switch 1, front panel
> port 1, split port 1. Letting the device name the ports on current
> port configuration of the port will keep driver/kernel simple, and let
> udev be the final name chooser.
Exactly. When you add in break out ports to the mix having sane, logical
default names (without udev) makes life easier. That was my ultimate
intent here - logical default names that udev or a user could then
modify if so desired.
But anyways, since the request is to have udev do the naming I need to
come back to it. A quick search on the systemd web site did not stumble
onto examples I could use, so I need to put this on the back burner
until I have time to dig through systemd docs.
David
^ permalink raw reply
* Re: [PATCH net-next 03/10] net/bonding: Notify state change on slaves
From: Nikolay Aleksandrov @ 2015-02-03 16:09 UTC (permalink / raw)
To: Or Gerlitz, David S. Miller
Cc: netdev, Roland Dreier, Amir Vadai, Tal Alon, Moni Shoua
In-Reply-To: <1422974919-28084-4-git-send-email-ogerlitz@mellanox.com>
On 03/02/15 15:48, Or Gerlitz wrote:
> From: Moni Shoua <monis@mellanox.com>
>
> Use notifier chain to dispatch an event upon a change in slave state.
> Event is dispatched with slave specific info.
>
> Signed-off-by: Moni Shoua <monis@mellanox.com>
> Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
> ---
> drivers/net/bonding/bond_main.c | 42 +++++++++++++++++++++++++++++++++++++++
> include/net/bonding.h | 12 +++++++++++
> 2 files changed, 54 insertions(+), 0 deletions(-)
>
Hi Or,
A few questions below,
> diff --git a/drivers/net/bonding/bond_main.c b/drivers/net/bonding/bond_main.c
> index 0665608..c9771f3 100644
> --- a/drivers/net/bonding/bond_main.c
> +++ b/drivers/net/bonding/bond_main.c
> @@ -1191,6 +1191,47 @@ static void bond_fill_ifslave(struct slave *slave, struct ifslave *info)
> info->link_failure_count = slave->link_failure_count;
> }
>
> +static void bond_netdev_notify(struct slave *slave, struct net_device *dev)
> +{
> + struct bonding *bond = slave->bond;
^^^^^^^^^^^
What if the struct slave where "slave" points to gets freed before this execution ?
> + struct netdev_bonding_info bonding_info;
> +
> + rtnl_lock();
> + /* make sure that slave is still valid */
> + if (dev->priv_flags & IFF_BONDING) {
^^^^^^^
What if the slave is released, enslaved to a different bond and the old bond is
destroyed between the dereference up there and the rtnl_lock() ?
Or the bonding gets unloaded altogether ?
> + bond_fill_ifslave(slave, &bonding_info.slave);
> + bond_fill_ifbond(bond, &bonding_info.master);
> + netdev_bonding_info_change(slave->dev, &bonding_info);
> + }
> + rtnl_unlock();
> +}
> +
> +static void bond_netdev_notify_work(struct work_struct *_work)
> +{
> + struct netdev_notify_work *w =
> + container_of(_work, struct netdev_notify_work, work.work);
> +
> + bond_netdev_notify(w->slave, w->dev);
> + dev_put(w->dev);
> +}
> +
> +void bond_queue_slave_event(struct slave *slave)
> +{
> + struct netdev_notify_work *nnw = kzalloc(sizeof(*nnw), GFP_ATOMIC);
^^^^^^^^
Where's this freed after the work's done ?
> +
> + if (!nnw)
> + return;
> +
> + INIT_DELAYED_WORK(&nnw->work, bond_netdev_notify_work);
> + nnw->slave = slave;
> + nnw->dev = slave->dev;
> +
> + if (queue_delayed_work(slave->bond->wq, &nnw->work, 0))
> + dev_hold(slave->dev);
> + else
> + kfree(nnw);
> +}
> +
<snip>
Cheers,
Nik
^ permalink raw reply
page: next (older) | prev (newer) | latest
- recent:[subjects (threaded)|topics (new)|topics (active)]
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox